Changeset 6a490b2

Jenkinsfile

-              rb7d6a36
+              r6a490b2
+                        }
                         sh "${SrcDir}/configure CXX=${Settings.Compiler.CXX} CC=${Settings.Compiler.CC} ${Settings.Architecture.flags} ${targets} --quiet"
+                        sh "${SrcDir}/configure CXX=${Settings.Compiler.CXX} CC=${Settings.Compiler.CC} ${Settings.Architecture.flags} AR=gcc-ar RANLIB=gcc-ranlib ${targets} --quiet --prefix=${BuildDir}"
                         // Configure libcfa
 …
                 dir (BuildDir) {
                         sh "make -j 8 --no-print-directory -C libcfa/${Settings.Architecture.name}-nodebug"
+                }
+        }
+        build_stage('Build : install', true) {
+                // Build outside of the src tree to ease cleaning
+                dir (BuildDir) {
+                        sh "make -j 8 --no-print-directory install"
+                }
+        }
 …
                 echo "Archiving core dumps"
                 dir (BuildDir) {
                         archiveArtifacts artifacts: "tests/crashes/**/*", fingerprint: true
+                        archiveArtifacts artifacts: "tests/crashes/**/*,lib/**/lib*.so*", fingerprint: true
+                }
                 throw err
 …
         public String CXX
         public String CC
+        CC_Desc(String name, String CXX, String CC) {
+        public String lto
+        CC_Desc(String name, String CXX, String CC, String lto) {
                 this.name = name
                 this.CXX = CXX
+                this.CC = CC
+                this.CC  = CC
+                this.lto = lto
+        }
+}
 …
                 switch( param.Compiler ) {
                         case 'gcc-9':
                                 this.Compiler = new CC_Desc('gcc-9', 'g++-9', 'gcc-9')
+                                this.Compiler = new CC_Desc('gcc-9', 'g++-9', 'gcc-9', '-flto=auto')
                         break
                         case 'gcc-8':
                                 this.Compiler = new CC_Desc('gcc-8', 'g++-8', 'gcc-8')
+                                this.Compiler = new CC_Desc('gcc-8', 'g++-8', 'gcc-8', '-flto=auto')
                         break
                         case 'gcc-7':
                                 this.Compiler = new CC_Desc('gcc-7', 'g++-7', 'gcc-7')
+                                this.Compiler = new CC_Desc('gcc-7', 'g++-7', 'gcc-7', '-flto=auto')
                         break
                         case 'gcc-6':
                                 this.Compiler = new CC_Desc('gcc-6', 'g++-6', 'gcc-6')
+                                this.Compiler = new CC_Desc('gcc-6', 'g++-6', 'gcc-6', '-flto=auto')
                         break
                         case 'gcc-5':
                                 this.Compiler = new CC_Desc('gcc-5', 'g++-5', 'gcc-5')
+                                this.Compiler = new CC_Desc('gcc-5', 'g++-5', 'gcc-5', '-flto=auto')
                         break
                         case 'gcc-4.9':
                                 this.Compiler = new CC_Desc('gcc-4.9', 'g++-4.9', 'gcc-4.9')
+                                this.Compiler = new CC_Desc('gcc-4.9', 'g++-4.9', 'gcc-4.9', '-flto=auto')
                         break
                         case 'clang':
                                 this.Compiler = new CC_Desc('clang', 'clang++-6.0', 'gcc-6')
+                                this.Compiler = new CC_Desc('clang', 'clang++-6.0', 'gcc-6', '-flto=thin -flto-jobs=0')
                         break
                         default :
 …
         // prepare the properties
         properties ([                                                                                                   \
+                buildDiscarder(logRotator(                                                                              \
+                        artifactDaysToKeepStr: '',                                                                      \
+                        artifactNumToKeepStr: '',                                                                       \
+                        daysToKeepStr: '730',                                                                           \
+                        numToKeepStr: '1000'                                                                            \
+                )),                                                                                                             \
                 [$class: 'ParametersDefinitionProperty',                                                                \
                         parameterDefinitions: [                                                                         \

benchmark/Makefile.am

-              rb7d6a36
+              r6a490b2
 ## Created On       : Sun May 31 09:08:15 2015
 ## Last Modified By : Peter A. Buhr
 ## Last Modified On : Sat Jan 25 09:20:44 2020
 ## Update Count     : 255
+## Last Modified On : Tue Mar 10 11:41:18 2020
+## Update Count     : 258
 ###############################################################################
 …
 creation_cfa_coroutine_DURATION = 100000000
 creation_cfa_coroutine_eager_DURATION = 10000000
+creation_cfa_generator_DURATION = 1000000000
 creation_upp_coroutine_DURATION = ${creation_cfa_coroutine_eager_DURATION}
 creation_cfa_thread_DURATION = 10000000
 …
 compile-typeof$(EXEEXT):
         $(CFACOMPILE) -fsyntax-only -w $(testdir)/typeof.cfa
+## =========================================================================================================
+size$(EXEEXT) : size-cfa.runquiet
+size-cfa$(EXEEXT):
+        $(BENCH_V_CFA)$(CFACOMPILE) $(srcdir)/size/size.cfa

benchmark/Makefile.in

-              rb7d6a36
+              r6a490b2
 creation_cfa_coroutine_DURATION = 100000000
 creation_cfa_coroutine_eager_DURATION = 10000000
+creation_cfa_generator_DURATION = 1000000000
 creation_upp_coroutine_DURATION = ${creation_cfa_coroutine_eager_DURATION}
 creation_cfa_thread_DURATION = 10000000
 …
         $(CFACOMPILE) -fsyntax-only -w $(testdir)/typeof.cfa
+size$(EXEEXT) : size-cfa.runquiet
+size-cfa$(EXEEXT):
+        $(BENCH_V_CFA)$(CFACOMPILE) $(srcdir)/size/size.cfa
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.

benchmark/creation/cfa_gen.cfa

-              rb7d6a36
+              r6a490b2
 #include "bench.h"
+#include "../bench.h"
 struct C {
+generator G {
         volatile int restart; // ensure compiler does not optimize away all the code
 };
 void ?{}( C & c ) { c.restart = 0; }
 void main( C & ) {}
+void ?{}( G & g ) { g.restart = 0; }
+void main( G & ) {}
 int main( int argc, char * argv[] ) {
 …
         BENCH(
                 for ( times ) {
                          C c;
+                         G g;
                 },
                 result

benchmark/ctxswitch/cfa_cor.cfa

-              rb7d6a36
+              r6a490b2
 #include <thread.hfa>
 #include "bench.h"
+#include "../bench.h"
 coroutine C {} c;
+coroutine C {};
 void main( __attribute__((unused)) C & ) {
         while () {
                 suspend();
+        for () {
+                suspend;
+        }
+}
 int main( int argc, char * argv[] ) {
+        C c;
         BENCH_START()
         BENCH(

benchmark/ctxswitch/cfa_gen.cfa

-              rb7d6a36
+              r6a490b2
 #include "../bench.h"
+typedef struct {
+        void * next;
+} C;
+void comain( C * c ) {
+        if ( __builtin_expect(c->next != 0, 1) ) goto *(c->next);
+        c->next = &&s1;
+generator G {};
+void main( G & ) {
         for () {
+                return;
+          s1: ;
+                suspend;
+        }
+}
 int main( int argc, char * argv[] ) {
+        G g;
         BENCH_START()
-        C c = { 0 };
         BENCH(
                 for ( times ) {
                         comain( &c );
+                        resume( g );
                 },
                 result

configure

rb7d6a36	r6a490b2
2557	2557	# don't use the default CFLAGS as they unconditonnaly add -O2
2558	2558	: ${CFLAGS=""}
	2559	: ${CXXFLAGS=""}
2559	2560
2560	2561	am__api_version='1.15'

configure.ac

rb7d6a36	r6a490b2
14	14	# don't use the default CFLAGS as they unconditonnaly add -O2
15	15	: ${CFLAGS=""}
	16	: ${CXXFLAGS=""}
16	17
17	18	AM_INIT_AUTOMAKE([subdir-objects])

doc/bibliography/pl.bib

-              rb7d6a36
+              r6a490b2
 %    Predefined journal names:
 %  acmcs: Computing Surveys             acta: Acta Infomatica
-@string{acta="Acta Infomatica"}
 %  cacm: Communications of the ACM
 %  ibmjrd: IBM J. Research & Development ibmsj: IBM Systems Journal
 …
 %  tcs: Theoretical Computer Science
+@string{acta="Acta Infomatica"}
 string{ieeepds="IEEE Transactions on Parallel and Distributed Systems"}
 @string{ieeepds="IEEE Trans. Parallel Distrib. Syst."}
 …
     series      = {ACM Distinguished Dissertations},
     year        = 1983,
+}
+@article{Zhang19,
+    keywords    = {Algebraic effects, dynamic scoping, exceptions, parametricity, type systems},
+    author      = {Zhang, Yizhou and Myers, Andrew C.},
+    title       = {Abstraction-safe Effect Handlers via Tunneling},
+    journal     = {Proc. ACM Program. Lang.},
+    issue_date  = {January 2019},
+    volume      = {3},
+    number      = {POPL},
+    month       = jan,
+    year        = {2019},
+    issn        = {2475-1421},
+    pages       = {5:1--5:29},
+    articleno   = {5},
+    publisher   = {ACM},
+    address     = {New York, NY, USA},
+}
+@inproceedings{Zhang16,
+    keywords    = {Exception tunneling, Genus, exception handling},
+    author      = {Zhang, Yizhou and Salvaneschi, Guido and Beightol, Quinn and Liskov, Barbara and Myers, Andrew C.},
+    title       = {Accepting Blame for Safe Tunneled Exceptions},
+    booktitle   = {Proceedings of the 37th ACM SIGPLAN Conference on Programming Language Design and Implementation},
+    series      = {PLDI'16},
+    year        = {2016},
+    location    = {Santa Barbara, CA, USA},
+    pages       = {281--295},
+    publisher   = {ACM},
+    address     = {New York, NY, USA},
+}
 …
     journal     = sigplan,
     year        = 1981,
+    month       = feb, volume = 16, number = 2, pages = {48-52},
+    month       = feb,
+    volume      = 16,
+    number      = 2,
+    pages       = {48-52},
     comment     = {
         A one-pass, top-down algorithm for overload resolution.  Input is a
 …
     title       = {An Alternative to Subclassing},
     journal     = sigplan,
+    volume      = {21},    number = {11},
+    volume      = {21},
+    number      = {11},
     pages       = {424-428},
+    month       = nov, year = 1986,
+    month       = nov,
+    year        = 1986,
     comment     = {
         The Smalltalk class hierarchy has three uses: factoring out code;
 …
     isbn        = {3-540-66538-2},
     location    = {Toulouse, France},
-    doi         = {http://doi.acm.org/10.1145/318773.319251},
     publisher   = {Springer},
     address     = {London, UK},
 …
     year        = 2010,
     pages       = {39--50},
-    numpages    = {12},
     publisher   = {IEEE Computer Society},
     address     = {Washington, DC, USA},
 …
+}
+@manual{C99,
+    keywords    = {ISO/IEC C 9899},
+    contributer = {pabuhr@plg},
+    key         = {C99},
+    title       = {C Programming Language {ISO/IEC} 9899:1999(E)},
+    edition     = {2nd},
+    publisher   = {International Standard Organization},
+    address     = {\href{https://webstore.ansi.org/Standards/INCITS/INCITSISOIEC98991999R2005}{https://webstore.ansi.org/\-Standards/\-INCITS/\-INCITSISOIEC98991999R2005}},
+    year        = 1999,
+}
 @manual{C11,
     keywords    = {ISO/IEC C 11},
 …
     location    = {London, United Kingdom},
     pages       = {41--53},
-    numpages    = {13},
-    url         = {http://doi.acm.org/10.1145/360204.360207},
-    doi         = {10.1145/360204.360207},
-    acmid       = {360207},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     year        = 1993,
     pages       = {201--208},
-    url         = {http://doi.acm.org/10.1145/155360.155580},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     location    = {Boulder, Colorado, USA},
     pages       = {91--97},
-    numpages    = {7},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     issn        = {0004-5411},
     pages       = {215--225},
-    numpages    = {11},
-    url         = {http://doi.acm.org/10.1145/321879.321884},
-    doi         = {10.1145/321879.321884},
-    acmid       = {321884},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
+}
+@misc{Drepper13,
+    keywords    = {thread-local storage},
+    contributer = {pabuhr@plg},
+    author      = {Ulrich Drepper},
+    title       = {{ELF} Handling For Thread-Local Storage},
+    year        = 2013,
+    month       = aug,
+    note        = {WikipediA},
+    howpublished= {\href{http://www.akkadia.org/drepper/tls.pdf}
+                  {http://\-www.akkadia.org/\-drepper/\-tls.pdf}},
+}
 @misc{Turley99,
     keywords    = {embedded system, micrprocessor},
 …
     howpublished= {\href{https://www.eetimes.com/author.asp?sectionid=36&doc_id=1287712}
                   {https://\-www.eetimes.com/\-author.asp?sectionid=\-36&doc_id=1287712}},
+}
+@article{Xiao19,
+    keywords    = {bug classification, fault trigger, Linux operating system, regression bug},
+    contributer = {pabuhr@plg},
+    author      = {Guanping Xiao and Zheng Zheng and Beibei Yin and Kishor S. Trivedi and Xiaoting Du and Kai-Yuan Cai},
+    title       = {An Empirical Study of Fault Triggers in the Linux Operating System: An Evolutionary Perspective},
+    journal     = {IEEE Transactions on Reliability},
+    month       = dec,
+    year        = 2019,
+    volume      = 68,
+    number      = 4,
+    pages       = {1356-1383},
+}
 …
+}
+@inproceedings{Palix11,
+    keywords    = {Linux, fault-finding tools},
+    contributer = {pabuhr@plg},
+    author      = {Nicolas Palix and Ga\"el Thomas and Suman Saha and Christophe Calv\`es and Julia Lawall and Gilles Muller},
+    title       = {Faults in Linux: Ten Years Later},
+    booktitle   = {Proc. of the 16 International Conf. on Arch. Support for Prog. Lang. and Oper. Sys.},
+    series      = {ASPLOS'11},
+    month       = mar,
+    year        = 2011,
+    location    = {Newport Beach, California, USA},
+    pages       = {305-318},
+    publisher   = {ACM},
+    address     = {New York, NY, USA},
+}
 @article{Lamport87,
     keywords    = {software solutions, mutual exclusion, fast},
 …
     issn        = {0001-0782},
     pages       = {107--115},
-    numpages    = {9},
-    url         = {http://doi.acm.org/10.1145/1538788.1538814},
-    doi         = {10.1145/1538788.1538814},
-    acmid       = {1538814},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
+}
+@mastersthesis{Radhakrishnan19,
+    author      = {Srihari Radhakrishnan},
+    title       = {High Performance Web Servers: A Study In Concurrent Programming Models},
+    school      = {School of Computer Sc., University of Waterloo},
+    year        = 2019,
+    optaddress  = {Waterloo, Ontario, Canada, N2L 3G1},
+    note        = {\href{https://uwspace.uwaterloo.ca/handle/10012/14706}{https://\-uwspace.uwaterloo.ca/\-handle/\-10012/\-14706}},
+}
 @article{katzenelson83b,
     contributer = {gjditchfield@plg},
 …
     pages       = {115-138},
     year        = 1971,
+}
+@inproceedings{Hagersten03,
+    keywords    = {cache storage, parallel architectures, performance evaluation, shared memory systems},
+    author      = {Zoran Radovi\'{c} and Erik Hagersten},
+    title       = {Hierarchical backoff locks for nonuniform communication architectures},
+    booktitle   = {Proceedings of the Ninth International Symposium on High-Performance Computer Architecture},
+    year        = {2003},
+    location    = {Anaheim, CA, USA},
+    pages       = {241-252},
+    publisher   = {IEEE},
+}
 …
+}
+@misc{gccValueLabels,
+    keywords    = {gcc extension, value labels},
+    contributer = {pabuhr@plg},
+    key         = {Labels as Values},
+    author      = {{gcc Extension}},
+    title       = {Labels as Values},
+    year        = {since gcc-3},
+    howpublished= {\href{https://gcc.gnu.org/onlinedocs/gcc/Labels-as-Values.html}
+                  {https:\-//gcc.gnu.org/\-onlinedocs/\-gcc/\-Labels-as-Values.html}},
+}
 @mastersthesis{Clarke90,
     keywords    = {concurrency, postponing requests},
 …
+}
+@misc{libfibre,
+    key         = {libfibre},
+    author      = {Martin Karsten},
+    title       = {{libfibre:~User-Level Threading Runtime}},
+    howpublished= {\href{https://git.uwaterloo.ca/mkarsten/libfibre}
+                  {https://\-git.uwaterloo.ca/\-mkarsten/\-libfibre}},
+    note        = {[Online; accessed 2020-04-15]},
+}
 @article{Linda,
     keywords    = {Linda, concurrency},
 …
+}
+@inproceedings{Fang06,
+    author      = {Fang, Yi and McMillan, Kenneth L. and Pnueli, Amir and Zuck, Lenore D.},
+    editor      = {Najm, Elie and Pradat-Peyre, Jean-Fran{\c{c}}ois and Donzeau-Gouge, V{\'e}ronique Vigui{\'e}},
+    title       = {Liveness by Invisible Invariants},
+    booktitle   = {Formal Techniques for Networked and Distributed Systems - FORTE 2006},
+    year        = 2006,
+    publisher   = {Springer Berlin Heidelberg},
+    address     = {Berlin, Heidelberg},
+    pages       = {356--371},
+}
 @article{Pierce00,
     keywords    = {Scala},
+    keywords    = {Scala, polymorphism, subtyping, type inference},
     contributer = {a3moss@uwaterloo.ca},
     author      = {Pierce, Benjamin C. and Turner, David N.},
 …
     issn        = {0164-0925},
     pages       = {1--44},
-    numpages    = {44},
-    url         = {http://doi.acm.org/10.1145/345099.345100},
-    doi         = {10.1145/345099.345100},
-    acmid       = {345100},
     publisher   = {ACM},
     address     = {New York, NY, USA},
-    keywords    = {polymorphism, subtyping, type inference},
+}
+@article{Dice15,
+    keywords    = {Concurrency, NUMA, hierarchical locks, locks, multicore, mutex, mutual exclusion, spin locks},
+    author      = {Dice, David and Marathe, Virendra J. and Shavit, Nir},
+    title       = {Lock Cohorting: A General Technique for Designing NUMA Locks},
+    journal     = {ACM Trans. Parallel Comput.},
+    issue_date  = {January 2015},
+    volume      = 1,
+    number      = 2,
+    month       = feb,
+    year        = 2015,
+    pages       = {13:1--13:42},
+    publisher   = {ACM},
+    address     = {New York, NY, USA},
+}
 @article{Sundell08,
 …
     journal     = sigplan,
     year        = 1989,
+    month       = jun, volume = 24, number = 6, pages = {37-48},
+    month       = jun,
+    volume      = 24,
+    number      = 6,
+    pages       = {37-48},
     abstract    = {
         This paper describes a scheme we have used to manage a large
 …
     contributer = {pabuhr@plg},
     author      = {Gregory R. Andrews},
     title       = {A Method for Solving Synronization Problems},
+    title       = {A Method for Solving Synchronization Problems},
     journal     = scp,
     volume      = 13,
 …
     title       = {Multiple Inheritance for {C}{\kern-.1em\hbox{\large\texttt{+\kern-.25em+}}}},
     booktitle   = {Proceedings of the Spring '87 EUUG Conference},
+    month       = may, year = 1987
+    month       = may,
+    year        = 1987,
+}
 …
     year        = 1986,
     pages       = {313--326},
-    numpages    = {14},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     year        = 1986,
     pages       = {327--348},
-    numpages    = {22},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     year        = 2005,
     pages       = {146-196},
-    numpages    = {51},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     year        = 2000,
     pages       = {29-46},
     note        = {OOPSLA'00, Oct. 15--19, 2000, Minneapolis, Minnesota, U.S.A.},
+    note        = {OOPSLA'00, Oct. 15--19, 2000, Minneapolis, Minn., U.S.A.},
+}
 …
     location    = {San Diego, California, USA},
     pages       = {101--112},
-    numpages    = {12},
-    url         = {http://doi.acm.org/10.1145/2535838.2535878},
-    doi         = {10.1145/2535838.2535878},
-    acmid       = {2535878},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     issn        = {0362-1340},
     pages       = {30--42},
-    numpages    = {13},
-    url         = {http://doi.acm.org/10.1145/947586.947589},
-    doi         = {10.1145/947586.947589},
     publisher   = {ACM},
     address     = {New York, NY, USA}
 …
     month       = 9,
     year        = 2005,
+}
+@article{Bauer15,
+    keywords    = {resumption exceptions, theory},
+    contributer = {pabuhr@plg},
+    author      = {Andrej Bauer and Matija Pretnar},
+    title       = {Programming with Algebraic Effects and Handlers},
+    journal     = {Journal of Logical and Algebraic Methods in Programming},
+    publisher   = {Elsevier BV},
+    volume      = 84,
+    number      = 1,
+    month       = jan,
+    year        = 2015,
+    pages       = {108-123},
+}
 …
     issn        = {0164-0925},
     pages       = {429-475},
-    url         = {http://doi.acm.org/10.1145/1133651.1133653},
-    doi         = {10.1145/1133651.1133653},
-    acmid       = {1133653},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     address     = {\href{http://docs.paralleluniverse.co/quasar}{http://\-docs.paralleluniverse.co/\-quasar}},
     year        = 2018,
+}
+@article{Aravind09,
+    author      = {Alex A. Aravind and Wim H. Hesselink},
+    title       = {A Queue Based Mutual Exclusion Algorithm},
+    journal     = acta,
+    volume      = 46,
+    pages       = {73--86},
+    year        = 2009,
+}
 …
     issn        = {0001-0782},
     pages       = {565--569},
-    numpages    = {5},
-    url         = {http://doi.acm.org/10.1145/359545.359566},
-    doi         = {10.1145/359545.359566},
-    acmid       = {359566},
     publisher   = {ACM},
     address     = {New York, NY, USA}
 …
     issn        = {0362-1340},
     pages       = {145--147},
-    numpages    = {3},
-    url         = {http://doi.acm.org/10.1145/122598.122614},
-    doi         = {10.1145/122598.122614},
-    acmid       = {122614},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     issn        = {0362-1340},
     pages       = {82--87},
-    numpages    = {6},
-    url         = {http://doi.acm.org/10.1145/947680.947688},
-    doi         = {10.1145/947680.947688},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
+}
+@article{Cascaval08,
+    author      = {Cascaval, Calin and Blundell, Colin and Michael, Maged and Cain, Harold W. and Wu, Peng and Chiras, Stefanie and Chatterjee, Siddhartha},
+    title       = {Software Transactional Memory: Why Is It Only a Research Toy?},
+    journal     = {Queue},
+    volume      = {6},
+    number      = {5},
+    month       = sep,
+    year        = {2008},
+    pages       = {40:46--40:58},
+    publisher   = {ACM},
+    address     = {New York, NY, USA},
+}
 @article{Dijkstra65a,
     keywords    = {N-thread software-solution mutual exclusion},
 …
     year        = 1974,
     pages       = {261-301},
-    issn        = {0360-0300},
-    doi         = {http://doi.acm.org/10.1145/356635.356640},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     publisher   = {ACM Press},
     address     = {New York, NY, USA},
-    doi         = {http://doi.acm.org/10.1145/356586.356588},
+}
 …
     howpublished= {\href{https://projects.eclipse.org/proposals/trace-compass}{https://\-projects.eclipse.org/\-proposals/\-trace-compass}},
+}
+@inproceedings{Boehm09,
+    author      = {Boehm, Hans-J.},
+    title       = {Transactional Memory Should Be an Implementation Technique, Not a Programming Interface},
+    booktitle   = {Proceedings of the First USENIX Conference on Hot Topics in Parallelism},
+    series      = {HotPar'09},
+    year        = {2009},
+    location    = {Berkeley, California},
+    publisher   = {USENIX Association},
+    address     = {Berkeley, CA, USA},
+}
 @article{Leroy00,
     keywords    = {type-systems, exceptions},
 …
     number      = {2},
     pages       = {204-214},
+    month       = apr, year = 1988,
+    month       = apr,
+    year        = 1988,
     comment     = {
         Extended record types add fields to their base record.  Assignment
 …
+}
+@article{Karsten20,
+    author      = {Karsten, Martin and Barghi, Saman},
+    title       = {{User-level Threading: Have Your Cake and Eat It Too}},
+    year        = {2020},
+    issue_date  = {March 2020},
+    publisher   = {Association for Computing Machinery},
+    address     = {New York, NY, USA},
+    volume      = {4},
+    number      = {1},
+    url         = {https://doi.org/10.1145/3379483},
+    doi         = {10.1145/3379483},
+    journal     = {Proc. ACM Meas. Anal. Comput. Syst.},
+    month       = mar,
+    numpages    = {30},
+}
 @techreport{Harmony,
     keywords    = {messages, concurrency},
 …
     contributer = {gjditchfield@plg},
     author      = {Henry Lieverman},
+    title       = {Using Prototypical Objects to Implement Shared Behavior in
+                  Object Oriented Systems},
+    title       = {Using Prototypical Objects to Implement Shared Behavior in Object Oriented Systems},
     journal     = sigplan,
+    month       = nov, year = 1986,
+    volume      = 21, number = 11, pages = {214-223}
+    month       = nov,
+    year        = 1986,
+    volume      = 21,
+    number      = 11,
+    pages       = {214-223}
+}
 …
     issn        = {0004-5411},
     pages       = {245--281},
-    numpages    = {37},
-    url         = {http://doi.acm.org/10.1145/62.2160},
-    doi         = {10.1145/62.2160},
-    acmid       = {2160},
     publisher   = {ACM},
     address     = {New York, NY, USA},
 …
     contributer = {pabuhr@plg},
     author      = {Boehm, Hans-J. and Adve, Sarita V.},
     title       = {You Don'T Know Jack About Shared Variables or Memory Models},
+    title       = {You Don't Know Jack About Shared Variables or Memory Models},
     journal     = cacm,
     volume      = 55,

doc/papers/concurrency/Paper.tex

-              rb7d6a36
+              r6a490b2
 \newcommand{\CCseventeen}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}17\xspace} % C++17 symbolic name
 \newcommand{\CCtwenty}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}20\xspace} % C++20 symbolic name
 \newcommand{\Csharp}{C\raisebox{-0.7ex}{\Large$^\sharp$}\xspace} % C# symbolic name
+\newcommand{\Csharp}{C\raisebox{-0.7ex}{\large$^\sharp$}\xspace} % C# symbolic name
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 …
 \newcommand*{\etc}{%
         \@ifnextchar{.}{\ETC}%
         {\ETC.\xspace}%
+                {\ETC.\xspace}%
 }}{}%
 \@ifundefined{etal}{
 \newcommand{\ETAL}{\abbrevFont{et}~\abbrevFont{al}}
 \newcommand*{\etal}{%
         \@ifnextchar{.}{\protect\ETAL}%
                 {\protect\ETAL.\xspace}%
+        \@ifnextchar{.}{\ETAL}%
+                {\ETAL.\xspace}%
 }}{}%
 \@ifundefined{viz}{
 …
                 __float80, float80, __float128, float128, forall, ftype, generator, _Generic, _Imaginary, __imag, __imag__,
                 inline, __inline, __inline__, __int128, int128, __label__, monitor, mutex, _Noreturn, one_t, or,
                 otype, restrict, __restrict, __restrict__, __signed, __signed__, _Static_assert, thread,
+                otype, restrict, resume, __restrict, __restrict__, __signed, __signed__, _Static_assert, suspend, thread,
                 _Thread_local, throw, throwResume, timeout, trait, try, ttype, typeof, __typeof, __typeof__,
                 virtual, __volatile, __volatile__, waitfor, when, with, zero_t},
         moredirectives={defined,include_next},
         % replace/adjust listing characters that look bad in sanserif
         literate={-}{\makebox[1ex][c]{\raisebox{0.4ex}{\rule{0.8ex}{0.1ex}}}}1 {^}{\raisebox{0.6ex}{$\scriptstyle\land\,$}}1
+        literate={-}{\makebox[1ex][c]{\raisebox{0.5ex}{\rule{0.8ex}{0.1ex}}}}1 {^}{\raisebox{0.6ex}{$\scriptstyle\land\,$}}1
                 {~}{\raisebox{0.3ex}{$\scriptstyle\sim\,$}}1 % {`}{\ttfamily\upshape\hspace*{-0.1ex}`}1
                 {<}{\textrm{\textless}}1 {>}{\textrm{\textgreater}}1
 …
                 _Else, _Enable, _Event, _Finally, _Monitor, _Mutex, _Nomutex, _PeriodicTask, _RealTimeTask,
                 _Resume, _Select, _SporadicTask, _Task, _Timeout, _When, _With, _Throw},
+}
-\lstdefinelanguage{Golang}{
-        morekeywords=[1]{package,import,func,type,struct,return,defer,panic,recover,select,var,const,iota,},
-        morekeywords=[2]{string,uint,uint8,uint16,uint32,uint64,int,int8,int16,int32,int64,
-                bool,float32,float64,complex64,complex128,byte,rune,uintptr, error,interface},
-        morekeywords=[3]{map,slice,make,new,nil,len,cap,copy,close,true,false,delete,append,real,imag,complex,chan,},
-        morekeywords=[4]{for,break,continue,range,goto,switch,case,fallthrough,if,else,default,},
-        morekeywords=[5]{Println,Printf,Error,},
-        sensitive=true,
-        morecomment=[l]{//},
-        morecomment=[s]{/*}{*/},
-        morestring=[b]',
-        morestring=[b]",
-        morestring=[s]{`}{`},
+}
 …
 {}
 \lstnewenvironment{uC++}[1][]
 {\lstset{#1}}
+{\lstset{language=uC++,moredelim=**[is][\protect\color{red}]{`}{`},#1}\lstset{#1}}
 {}
 \lstnewenvironment{Go}[1][]
 …
+}
 \newbox\myboxA
 \newbox\myboxB
 \newbox\myboxC
 \newbox\myboxD
+\newsavebox{\myboxA}
+\newsavebox{\myboxB}
+\newsavebox{\myboxC}
+\newsavebox{\myboxD}
 \title{\texorpdfstring{Advanced Control-flow and Concurrency in \protect\CFA}{Advanced Control-flow in Cforall}}
 …
 \CFA is a polymorphic, non-object-oriented, concurrent, backwards-compatible extension of the C programming language.
 This paper discusses the design philosophy and implementation of its advanced control-flow and concurrent/parallel features, along with the supporting runtime written in \CFA.
 These features are created from scratch as ISO C has only low-level and/or unimplemented concurrency, so C programmers continue to rely on library features like pthreads.
+These features are created from scratch as ISO C has only low-level and/or unimplemented concurrency, so C programmers continue to rely on library approaches like pthreads.
 \CFA introduces modern language-level control-flow mechanisms, like generators, coroutines, user-level threading, and monitors for mutual exclusion and synchronization.
 % Library extension for executors, futures, and actors are built on these basic mechanisms.
 …
 \begin{document}
 \linenumbers                                            % comment out to turn off line numbering
+\linenumbers                            % comment out to turn off line numbering
 \maketitle
 …
 \section{Introduction}
+This paper discusses the design philosophy and implementation of advanced language-level control-flow and concurrent/parallel features in \CFA~\cite{Moss18,Cforall} and its runtime, which is written entirely in \CFA.
+\CFA is a modern, polymorphic, non-object-oriented\footnote{
+\CFA has features often associated with object-oriented programming languages, such as constructors, destructors, virtuals and simple inheritance.
+\CFA~\cite{Moss18,Cforall} is a modern, polymorphic, non-object-oriented\footnote{
+\CFA has object-oriented features, such as constructors, destructors, virtuals and simple trait/interface inheritance.
+% Go interfaces, Rust traits, Swift Protocols, Haskell Type Classes and Java Interfaces.
+% "Trait inheritance" works for me. "Interface inheritance" might also be a good choice, and distinguish clearly from implementation inheritance.
+% You'll want to be a little bit careful with terms like "structural" and "nominal" inheritance as well. CFA has structural inheritance (I think Go as well) -- it's inferred based on the structure of the code. Java, Rust, and Haskell (not sure about Swift) have nominal inheritance, where there needs to be a specific statement that "this type inherits from this type".
 However, functions \emph{cannot} be nested in structures, so there is no lexical binding between a structure and set of functions (member/method) implemented by an implicit \lstinline@this@ (receiver) parameter.},
 backwards-compatible extension of the C programming language.
+In many ways, \CFA is to C as Scala~\cite{Scala} is to Java, providing a \emph{research vehicle} for new typing and control-flow capabilities on top of a highly popular programming language allowing immediate dissemination.
+Within the \CFA framework, new control-flow features are created from scratch because ISO \Celeven defines only a subset of the \CFA extensions, where the overlapping features are concurrency~\cite[\S~7.26]{C11}.
+However, \Celeven concurrency is largely wrappers for a subset of the pthreads library~\cite{Butenhof97,Pthreads}, and \Celeven and pthreads concurrency is simple, based on thread fork/join in a function and mutex/condition locks, which is low-level and error-prone;
+no high-level language concurrency features are defined.
+Interestingly, almost a decade after publication of the \Celeven standard, neither gcc-8, clang-9 nor msvc-19 (most recent versions) support the \Celeven include @threads.h@, indicating little interest in the C11 concurrency approach (possibly because the effort to add concurrency to \CC).
+Finally, while the \Celeven standard does not state a threading model, the historical association with pthreads suggests implementations would adopt kernel-level threading (1:1)~\cite{ThreadModel}.
+In many ways, \CFA is to C as Scala~\cite{Scala} is to Java, providing a \emph{research vehicle} for new typing and control-flow capabilities on top of a highly popular programming language\footnote{
+The TIOBE index~\cite{TIOBE} for December 2019 ranks the top five \emph{popular} programming languages as Java 17\%, C 16\%, Python 10\%, and \CC 6\%, \Csharp 5\% = 54\%, and over the past 30 years, C has always ranked either first or second in popularity.}
+allowing immediate dissemination.
+This paper discusses the design philosophy and implementation of advanced language-level control-flow and concurrent/parallel features in \CFA and its runtime, which is written entirely in \CFA.
+The \CFA control-flow framework extends ISO \Celeven~\cite{C11} with new call/return and concurrent/parallel control-flow.
+% The call/return extensions retain state between callee and caller versus losing the callee's state on return;
+% the concurrency extensions allow high-level management of threads.
+Call/return control-flow with argument/parameter passing appeared in the first programming languages.
+Over the past 50 years, call/return has been augmented with features like static/dynamic call, exceptions (multi-level return) and generators/coroutines (retain state between calls).
+While \CFA has mechanisms for dynamic call (algebraic effects) and exceptions\footnote{
+\CFA exception handling will be presented in a separate paper.
+The key feature that dovetails with this paper is nonlocal exceptions allowing exceptions to be raised across stacks, with synchronous exceptions raised among coroutines and asynchronous exceptions raised among threads, similar to that in \uC~\cite[\S~5]{uC++}}, this work only discusses retaining state between calls via generators/coroutines.
+\newterm{Coroutining} was introduced by Conway~\cite{Conway63} (1963), discussed by Knuth~\cite[\S~1.4.2]{Knuth73V1}, implemented in Simula67~\cite{Simula67}, formalized by Marlin~\cite{Marlin80}, and is now popular and appears in old and new programming languages: CLU~\cite{CLU}, \Csharp~\cite{Csharp}, Ruby~\cite{Ruby}, Python~\cite{Python}, JavaScript~\cite{JavaScript}, Lua~\cite{Lua}, \CCtwenty~\cite{C++20Coroutine19}.
+Coroutining is sequential execution requiring direct handoff among coroutines, \ie only the programmer is controlling execution order.
+If coroutines transfer to an internal event-engine for scheduling the next coroutines, the program transitions into the realm of concurrency~\cite[\S~3]{Buhr05a}.
+Coroutines are only a stepping stone towards concurrency where the commonality is that coroutines and threads retain state between calls.
+\Celeven/\CCeleven define concurrency~\cite[\S~7.26]{C11}, but it is largely wrappers for a subset of the pthreads library~\cite{Pthreads}.\footnote{Pthreads concurrency is based on simple thread fork/join in a function and mutex/condition locks, which is low-level and error-prone}
+Interestingly, almost a decade after the \Celeven standard, neither gcc-9, clang-9 nor msvc-19 (most recent versions) support the \Celeven include @threads.h@, indicating no interest in the C11 concurrency approach (possibly because of the recent effort to add concurrency to \CC).
+While the \Celeven standard does not state a threading model, the historical association with pthreads suggests implementations would adopt kernel-level threading (1:1)~\cite{ThreadModel}, as for \CC.
 In contrast, there has been a renewed interest during the past decade in user-level (M:N, green) threading in old and new programming languages.
 As multi-core hardware became available in the 1980/90s, both user and kernel threading were examined.
 Kernel threading was chosen, largely because of its simplicity and fit with the simpler operating systems and hardware architectures at the time, which gave it a performance advantage~\cite{Drepper03}.
 Libraries like pthreads were developed for C, and the Solaris operating-system switched from user (JDK 1.1~\cite{JDK1.1}) to kernel threads.
 As a result, languages like Java, Scala, Objective-C~\cite{obj-c-book}, \CCeleven~\cite{C11}, and C\#~\cite{Csharp} adopt the 1:1 kernel-threading model, with a variety of presentation mechanisms.
 From 2000 onwards, languages like Go~\cite{Go}, Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, D~\cite{D}, and \uC~\cite{uC++,uC++book} have championed the M:N user-threading model, and many user-threading libraries have appeared~\cite{Qthreads,MPC,Marcel}, including putting green threads back into Java~\cite{Quasar}.
 The main argument for user-level threading is that it is lighter weight than kernel threading (locking and context switching do not cross the kernel boundary), so there is less restriction on programming styles that encourage large numbers of threads performing medium work units to facilitate load balancing by the runtime~\cite{Verch12}.
+As a result, many current languages implementations adopt the 1:1 kernel-threading model, like Java (Scala), Objective-C~\cite{obj-c-book}, \CCeleven~\cite{C11}, C\#~\cite{Csharp} and Rust~\cite{Rust}, with a variety of presentation mechanisms.
+From 2000 onwards, several language implementations have championed the M:N user-threading model, like Go~\cite{Go}, Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, D~\cite{D}, and \uC~\cite{uC++,uC++book}, including putting green threads back into Java~\cite{Quasar}, and many user-threading libraries have appeared~\cite{Qthreads,MPC,Marcel}.
+The main argument for user-level threading is that it is lighter weight than kernel threading (locking and context switching do not cross the kernel boundary), so there is less restriction on programming styles that encourages large numbers of threads performing medium-sized work to facilitate load balancing by the runtime~\cite{Verch12}.
 As well, user-threading facilitates a simpler concurrency approach using thread objects that leverage sequential patterns versus events with call-backs~\cite{Adya02,vonBehren03}.
 Finally, performant user-threading implementations (both time and space) meet or exceed direct kernel-threading implementations, while achieving the programming advantages of high concurrency levels and safety.
 A further effort over the past two decades is the development of language memory models to deal with the conflict between language features and compiler/hardware optimizations, \ie some language features are unsafe in the presence of aggressive sequential optimizations~\cite{Buhr95a,Boehm05}.
+A further effort over the past two decades is the development of language memory models to deal with the conflict between language features and compiler/hardware optimizations, \eg some language features are unsafe in the presence of aggressive sequential optimizations~\cite{Buhr95a,Boehm05}.
 The consequence is that a language must provide sufficient tools to program around safety issues, as inline and library code is all sequential to the compiler.
 One solution is low-level qualifiers and functions (\eg @volatile@ and atomics) allowing \emph{programmers} to explicitly write safe (race-free~\cite{Boehm12}) programs.
+A safer solution is high-level language constructs so the \emph{compiler} knows the optimization boundaries, and hence, provides implicit safety.
+This problem is best known with respect to concurrency, but applies to other complex control-flow, like exceptions\footnote{
+\CFA exception handling will be presented in a separate paper.
+The key feature that dovetails with this paper is nonlocal exceptions allowing exceptions to be raised across stacks, with synchronous exceptions raised among coroutines and asynchronous exceptions raised among threads, similar to that in \uC~\cite[\S~5]{uC++}
+} and coroutines.
+Finally, language solutions allow matching constructs with language paradigm, \ie imperative and functional languages often have different presentations of the same concept to fit their programming model.
+Finally, it is important for a language to provide safety over performance \emph{as the default}, allowing careful reduction of safety for performance when necessary.
+Two concurrency violations of this philosophy are \emph{spurious wakeup} (random wakeup~\cite[\S~8]{Buhr05a}) and \emph{barging}\footnote{
+The notion of competitive succession instead of direct handoff, \ie a lock owner releases the lock and an arriving thread acquires it ahead of preexisting waiter threads.
+A safer solution is high-level language constructs so the \emph{compiler} knows the concurrency boundaries (where mutual exclusion and synchronization are acquired/released) and provide implicit safety at and across these boundaries.
+While the optimization problem is best known with respect to concurrency, it applies to other complex control-flow, like exceptions and coroutines.
+As well, language solutions allow matching the language paradigm with the approach, \eg matching the functional paradigm with data-flow programming or the imperative paradigm with thread programming.
+Finally, it is important for a language to provide safety over performance \emph{as the default}, allowing careful reduction of safety (unsafe code) for performance when necessary.
+Two concurrency violations of this philosophy are \emph{spurious wakeup} (random wakeup~\cite[\S~9]{Buhr05a}) and \emph{barging}\footnote{
+Barging is competitive succession instead of direct handoff, \ie after a lock is released both arriving and preexisting waiter threads compete to acquire the lock.
+Hence, an arriving thread can temporally \emph{barge} ahead of threads already waiting for an event, which can repeat indefinitely leading to starvation of waiter threads.
 } (signals-as-hints~\cite[\S~8]{Buhr05a}), where one is a consequence of the other, \ie once there is spurious wakeup, signals-as-hints follow.
+However, spurious wakeup is \emph{not} a foundational concurrency property~\cite[\S~8]{Buhr05a}, it is a performance design choice.
+Similarly, signals-as-hints are often a performance decision.
+We argue removing spurious wakeup and signals-as-hints make concurrent programming significantly safer because it removes local non-determinism and matches with programmer expectation.
+(Author experience teaching concurrency is that students are highly confused by these semantics.)
+Clawing back performance, when local non-determinism is unimportant, should be an option not the default.
+\begin{comment}
+Most augmented traditional (Fortran 18~\cite{Fortran18}, Cobol 14~\cite{Cobol14}, Ada 12~\cite{Ada12}, Java 11~\cite{Java11}) and new languages (Go~\cite{Go}, Rust~\cite{Rust}, and D~\cite{D}), except \CC, diverge from C with different syntax and semantics, only interoperate indirectly with C, and are not systems languages, for those with managed memory.
+As a result, there is a significant learning curve to move to these languages, and C legacy-code must be rewritten.
+While \CC, like \CFA, takes an evolutionary approach to extend C, \CC's constantly growing complex and interdependent features-set (\eg objects, inheritance, templates, etc.) mean idiomatic \CC code is difficult to use from C, and C programmers must expend significant effort learning \CC.
+Hence, rewriting and retraining costs for these languages, even \CC, are prohibitive for companies with a large C software-base.
+\CFA with its orthogonal feature-set, its high-performance runtime, and direct access to all existing C libraries circumvents these problems.
+\end{comment}
+\CFA embraces user-level threading, language extensions for advanced control-flow, and safety as the default.
+We present comparative examples so the reader can judge if the \CFA control-flow extensions are better and safer than those in other concurrent, imperative programming languages, and perform experiments to show the \CFA runtime is competitive with other similar mechanisms.
+(Author experience teaching concurrency is that students are confused by these semantics.)
+However, spurious wakeup is \emph{not} a foundational concurrency property~\cite[\S~9]{Buhr05a};
+it is a performance design choice.
+We argue removing spurious wakeup and signals-as-hints make concurrent programming simpler and safer as there is less local non-determinism to manage.
+If barging acquisition is allowed, its specialized performance advantage should be available as an option not the default.
+\CFA embraces language extensions for advanced control-flow, user-level threading, and safety as the default.
+We present comparative examples to support our argument that the \CFA control-flow extensions are as expressive and safe as those in other concurrent imperative programming languages, and perform experiments to show the \CFA runtime is competitive with other similar mechanisms.
 The main contributions of this work are:
 \begin{itemize}[topsep=3pt,itemsep=1pt]
+\begin{itemize}[topsep=3pt,itemsep=0pt]
 \item
+language-level generators, coroutines and user-level threading, which respect the expectations of C programmers.
+a set of fundamental execution properties that dictate which language-level control-flow features need to be supported,
 \item
+monitor synchronization without barging, and the ability to safely acquiring multiple monitors \emph{simultaneously} (deadlock free), while seamlessly integrating these capabilities with all monitor synchronization mechanisms.
+integration of these language-level control-flow features, while respecting the style and expectations of C programmers,
 \item
+providing statically type-safe interfaces that integrate with the \CFA polymorphic type-system and other language features.
+monitor synchronization without barging, and the ability to safely acquiring multiple monitors \emph{simultaneously} (deadlock free), while seamlessly integrating these capabilities with all monitor synchronization mechanisms,
+\item
+providing statically type-safe interfaces that integrate with the \CFA polymorphic type-system and other language features,
 % \item
 % library extensions for executors, futures, and actors built on the basic mechanisms.
 \item
+a runtime system with no spurious wakeup.
+a runtime system without spurious wake-up and no performance loss,
 \item
+a dynamic partitioning mechanism to segregate the execution environment for specialized requirements.
+a dynamic partitioning mechanism to segregate groups of executing user and kernel threads performing specialized work (\eg web-server or compute engine) or requiring different scheduling (\eg NUMA or real-time).
 % \item
 % a non-blocking I/O library
 \item
 experimental results showing comparable performance of the new features with similar mechanisms in other programming languages.
+experimental results showing comparable performance of the \CFA features with similar mechanisms in other languages.
 \end{itemize}
+Section~\ref{s:StatefulFunction} begins advanced control by introducing sequential functions that retain data and execution state between calls, which produces constructs @generator@ and @coroutine@.
+Section~\ref{s:Concurrency} begins concurrency, or how to create (fork) and destroy (join) a thread, which produces the @thread@ construct.
+Section~\ref{s:FundamentalExecutionProperties} presents the compositional hierarchy of execution properties directing the design of control-flow features in \CFA.
+Section~\ref{s:StatefulFunction} begins advanced control by introducing sequential functions that retain data and execution state between calls producing constructs @generator@ and @coroutine@.
+Section~\ref{s:Concurrency} begins concurrency, or how to create (fork) and destroy (join) a thread producing the @thread@ construct.
 Section~\ref{s:MutualExclusionSynchronization} discusses the two mechanisms to restricted nondeterminism when controlling shared access to resources (mutual exclusion) and timing relationships among threads (synchronization).
 Section~\ref{s:Monitor} shows how both mutual exclusion and synchronization are safely embedded in the @monitor@ and @thread@ constructs.
 Section~\ref{s:CFARuntimeStructure} describes the large-scale mechanism to structure (cluster) threads and virtual processors (kernel threads).
+Section~\ref{s:Performance} uses a series of microbenchmarks to compare \CFA threading with pthreads, Java OpenJDK-9, Go 1.12.6 and \uC 7.0.0.
+Section~\ref{s:Performance} uses a series of microbenchmarks to compare \CFA threading with pthreads, Java 11.0.6, Go 1.12.6, Rust 1.37.0, Python 3.7.6, Node.js 12.14.1, and \uC 7.0.0.
+\section{Fundamental Execution Properties}
+\label{s:FundamentalExecutionProperties}
+The features in a programming language should be composed from a set of fundamental properties rather than an ad hoc collection chosen by the designers.
+To this end, the control-flow features created for \CFA are based on the fundamental properties of any language with function-stack control-flow (see also \uC~\cite[pp.~140-142]{uC++}).
+The fundamental properties are execution state, thread, and mutual-exclusion/synchronization (MES).
+These independent properties can be used alone, in pairs, or in triplets to compose different language features, forming a compositional hierarchy where the most advanced feature has all the properties (state/thread/MES).
+While it is possible for a language to only support the most advanced feature~\cite{Hermes90}, this unnecessarily complicates and makes inefficient solutions to certain classes of problems.
+As is shown, each of the (non-rejected) composed features solves a particular set of problems, and hence, has a defensible position in a programming language.
+If a compositional feature is missing, a programmer has too few/many fundamental properties resulting in a complex and/or is inefficient solution.
+In detail, the fundamental properties are:
+\begin{description}[leftmargin=\parindent,topsep=3pt,parsep=0pt]
+\item[\newterm{execution state}:]
+is the state information needed by a control-flow feature to initialize, manage compute data and execution location(s), and de-initialize.
+State is retained in fixed-sized aggregate structures and dynamic-sized stack(s), often allocated in the heap(s) managed by the runtime system.
+The lifetime of the state varies with the control-flow feature, where longer life-time and dynamic size provide greater power but also increase usage complexity and cost.
+Control-flow transfers among execution states occurs in multiple ways, such as function call, context switch, asynchronous await, etc.
+Because the programming language determines what constitutes an execution state, implicitly manages this state, and defines movement mechanisms among states, execution state is an elementary property of the semantics of a programming language.
+% An execution-state is related to the notion of a process continuation \cite{Hieb90}.
+\item[\newterm{threading}:]
+is execution of code that occurs independently of other execution, \ie the execution resulting from a thread is sequential.
+Multiple threads provide \emph{concurrent execution};
+concurrent execution becomes parallel when run on multiple processing units (hyper-threading, cores, sockets).
+There must be language mechanisms to create, block/unblock, and join with a thread.
+\item[\newterm{MES}:]
+is the concurrency mechanisms to perform an action without interruption and establish timing relationships among multiple threads.
+These two properties are independent, \ie mutual exclusion cannot provide synchronization and vice versa without introducing additional threads~\cite[\S~4]{Buhr05a}.
+Limiting MES, \eg no access to shared data, results in contrived solutions and inefficiency on multi-core von Neumann computers where shared memory is a foundational aspect of its design.
+\end{description}
+These properties are fundamental because they cannot be built from existing language features, \eg a basic programming language like C99~\cite{C99} cannot create new control-flow features, concurrency, or provide MES using atomic hardware mechanisms.
+\subsection{Execution Properties}
+Table~\ref{t:ExecutionPropertyComposition} shows how the three fundamental execution properties: state, thread, and mutual exclusion compose a hierarchy of control-flow features needed in a programming language.
+(When doing case analysis, not all combinations are meaningful.)
+Note, basic von Neumann execution requires at least one thread and an execution state providing some form of call stack.
+For table entries missing these minimal components, the property is borrowed from the invoker (caller).
+Case 1 is a function that borrows storage for its state (stack frame/activation) and a thread from its invoker and retains this state across \emph{callees}, \ie function local-variables are retained on the stack across calls.
+Case 2 is case 1 with access to shared state so callers are restricted during update (mutual exclusion) and scheduling for other threads (synchronization).
+Case 3 is a stateful function supporting resume/suspend along with call/return to retain state across \emph{callers}, but has some restrictions because the function's state is stackless.
+Note, stackless functions still borrow the caller's stack and thread, where the stack is used to preserve state across its callees.
+Case 4 is cases 2 and 3 with protection to shared state for stackless functions.
+Cases 5 and 6 are the same as 3 and 4 but only the thread is borrowed as the function state is stackful, so resume/suspend is a context switch from the caller's to the function's stack.
+Cases 7 and 8 are rejected because a function that is given a new thread must have its own stack where the thread begins and stack frames are stored for calls, \ie there is no stack to borrow.
+Cases 9 and 10 are rejected because a thread with a fixed state (no stack) cannot accept calls, make calls, block, or be preempted, all of which require an unknown amount of additional dynamic state.
+Hence, once started, this kind of thread must execute to completion, \ie computation only, which severely restricts runtime management.
+Cases 11 and 12 have a stackful thread with and without safe access to shared state.
+Execution properties increase the cost of creation and execution along with complexity of usage.
+\begin{table}
+\caption{Execution property composition}
+\centering
+\label{t:ExecutionPropertyComposition}
+\renewcommand{\arraystretch}{1.25}
+%\setlength{\tabcolsep}{5pt}
+\begin{tabular}{c|c||l|l}
+\multicolumn{2}{c||}{execution properties} & \multicolumn{2}{c}{mutual exclusion / synchronization} \\
+\hline
+stateful                        & thread        & \multicolumn{1}{c|}{No} & \multicolumn{1}{c}{Yes} \\
+\hline
+\hline
+No                                      & No            & \textbf{1}\ \ \ function                              & \textbf{2}\ \ \ @monitor@ function    \\
+\hline
+Yes (stackless)         & No            & \textbf{3}\ \ \ @generator@                   & \textbf{4}\ \ \ @monitor@ @generator@ \\
+\hline
+Yes (stackful)          & No            & \textbf{5}\ \ \ @coroutine@                   & \textbf{6}\ \ \ @monitor@ @coroutine@ \\
+\hline
+No                                      & Yes           & \textbf{7}\ \ \ {\color{red}rejected} & \textbf{8}\ \ \ {\color{red}rejected} \\
+\hline
+Yes (stackless)         & Yes           & \textbf{9}\ \ \ {\color{red}rejected} & \textbf{10}\ \ \ {\color{red}rejected} \\
+\hline
+Yes (stackful)          & Yes           & \textbf{11}\ \ \ @thread@                             & \textbf{12}\ \ @monitor@ @thread@             \\
+\end{tabular}
+\end{table}
+Given the execution-properties taxonomy, programmers can now answer three basic questions: is state necessary across calls and how much, is a separate thread necessary, is access to shared state necessary.
+The answers define the optimal language feature need for implementing a programming problem.
+The next sections discusses how \CFA fills in the table with language features, while other programming languages may only provide a subset of the table.
+\subsection{Design Requirements}
+The following design requirements largely stem from building \CFA on top of C.
+\begin{itemize}[topsep=3pt,parsep=0pt]
+\item
+All communication must be statically type checkable for early detection of errors and efficient code generation.
+This requirement is consistent with the fact that C is a statically-typed programming-language.
+\item
+Direct interaction among language features must be possible allowing any feature to be selected without restricting comm\-unication.
+For example, many concurrent languages do not provide direct communication (calls) among threads, \ie threads only communicate indirectly through monitors, channels, messages, and/or futures.
+Indirect communication increases the number of objects, consuming more resources, and require additional synchronization and possibly data transfer.
+\item
+All communication is performed using function calls, \ie data is transmitted from argument to parameter and results are returned from function calls.
+Alternative forms of communication, such as call-backs, message passing, channels, or communication ports, step outside of C's normal form of communication.
+\item
+All stateful features must follow the same declaration scopes and lifetimes as other language data.
+For C that means at program startup, during block and function activation, and on demand using dynamic allocation.
+\item
+MES must be available implicitly in language constructs as well as explicitly for specialized requirements, because requiring programmers to build MES using low-level locks often leads to incorrect programs.
+Furthermore, reducing synchronization scope by encapsulating it within language constructs further reduces errors in concurrent programs.
+\item
+Both synchronous and asynchronous communication are needed.
+However, we believe the best way to provide asynchrony, such as call-buffering/chaining and/or returning futures~\cite{multilisp}, is building it from expressive synchronous features.
+\item
+Synchronization must be able to control the service order of requests including prioritizing selection from different kinds of outstanding requests, and postponing a request for an unspecified time while continuing to accept new requests.
+Otherwise, certain concurrency problems are difficult, e.g.\ web server, disk scheduling, and the amount of concurrency is inhibited~\cite{Gentleman81}.
+\end{itemize}
+We have satisfied these requirements in \CFA while maintaining backwards compatibility with the huge body of legacy C programs.
+% In contrast, other new programming languages must still access C programs (\eg operating-system service routines), but do so through fragile C interfaces.
+\subsection{Asynchronous Await / Call}
+Asynchronous await/call is a caller mechanism for structuring programs and/or increasing concurrency, where the caller (client) postpones an action into the future, which is subsequently executed by a callee (server).
+The caller detects the action's completion through a \newterm{future}/\newterm{promise}.
+The benefit is asynchronous caller execution with respect to the callee until future resolution.
+For single-threaded languages like JavaScript, an asynchronous call passes a callee action, which is queued in the event-engine, and continues execution with a promise.
+When the caller needs the promise to be fulfilled, it executes @await@.
+A promise-completion call-back can be part of the callee action or the caller is rescheduled;
+in either case, the call back is executed after the promise is fulfilled.
+While asynchronous calls generate new callee (server) events, we content this mechanism is insufficient for advanced control-flow mechanisms like generators or coroutines (which are discussed next).
+Specifically, control between caller and callee occurs indirectly through the event-engine precluding direct handoff and cycling among events, and requires complex resolution of a control promise and data.
+Note, @async-await@ is just syntactic-sugar over the event engine so it does not solve these deficiencies.
+For multi-threaded languages like Java, the asynchronous call queues a callee action with an executor (server), which subsequently executes the work by a thread in the executor thread-pool.
+The problem is when concurrent work-units need to interact and/or block as this effects the executor, \eg stops threads.
+While it is possible to extend this approach to support the necessary mechanisms, \eg message passing in Actors, we show monitors and threads provide an equally competitive approach that does not deviate from normal call communication and can be used to build asynchronous call, as is done in Java.
 …
 \label{s:StatefulFunction}
+The stateful function is an old idea~\cite{Conway63,Marlin80} that is new again~\cite{C++20Coroutine19}, where execution is temporarily suspended and later resumed, \eg plugin, device driver, finite-state machine.
+Hence, a stateful function may not end when it returns to its caller, allowing it to be restarted with the data and execution location present at the point of suspension.
+This capability is accomplished by retaining a data/execution \emph{closure} between invocations.
+If the closure is fixed size, we call it a \emph{generator} (or \emph{stackless}), and its control flow is restricted, \eg suspending outside the generator is prohibited.
+If the closure is variable size, we call it a \emph{coroutine} (or \emph{stackful}), and as the names implies, often implemented with a separate stack with no programming restrictions.
+Hence, refactoring a stackless coroutine may require changing it to stackful.
+A foundational property of all \emph{stateful functions} is that resume/suspend \emph{do not} cause incremental stack growth, \ie resume/suspend operations are remembered through the closure not the stack.
+As well, activating a stateful function is \emph{asymmetric} or \emph{symmetric}, identified by resume/suspend (no cycles) and resume/resume (cycles).
+A fixed closure activated by modified call/return is faster than a variable closure activated by context switching.
+Additionally, any storage management for the closure (especially in unmanaged languages, \ie no garbage collection) must also be factored into design and performance.
+Therefore, selecting between stackless and stackful semantics is a tradeoff between programming requirements and performance, where stackless is faster and stackful is more general.
+Note, creation cost is amortized across usage, so activation cost is usually the dominant factor.
+A \emph{stateful function} has the ability to remember state between calls, where state can be either data or execution, \eg plugin, device driver, finite-state machine (FSM).
+A simple technique to retain data state between calls is @static@ declarations within a function, which is often implemented by hoisting the declarations to the global scope but hiding the names within the function using name mangling.
+However, each call starts the function at the top making it difficult to determine the last point of execution in an algorithm, and requiring multiple flag variables and testing to reestablish the continuation point.
+Hence, the next step of generalizing function state is implicitly remembering the return point between calls and reentering the function at this point rather than the top, called \emph{generators}\,/\,\emph{iterators} or \emph{stackless coroutines}.
+For example, a Fibonacci generator retains data and execution state allowing it to remember prior values needed to generate the next value and the location in the algorithm to compute that value.
+The next step of generalization is instantiating the function to allow multiple named instances, \eg multiple Fibonacci generators, where each instance has its own state, and hence, can generate an independent sequence of values.
+Note, a subset of generator state is a function \emph{closure}, \ie the technique of capturing lexical references when returning a nested function.
+A further generalization is adding a stack to a generator's state, called a \emph{coroutine}, so it can suspend outside of itself, \eg call helper functions to arbitrary depth before suspending back to its resumer without unwinding these calls.
+For example, a coroutine iterator for a binary tree can stop the traversal at the visit point (pre, infix, post traversal), return the node value to the caller, and then continue the recursive traversal from the current node on the next call.
+There are two styles of activating a stateful function, \emph{asymmetric} or \emph{symmetric}, identified by resume/suspend (no cycles) and resume/resume (cycles).
+These styles \emph{do not} cause incremental stack growth, \eg a million resume/suspend or resume/resume cycles do not remember each cycle just the last resumer for each cycle.
+Selecting between stackless/stackful semantics and asymmetric/symmetric style is a tradeoff between programming requirements, performance, and design, where stackless is faster and smaller (modified call/return between closures), stackful is more general but slower and larger (context switching between distinct stacks), and asymmetric is simpler control-flow than symmetric.
+Additionally, storage management for the closure/stack (especially in unmanaged languages, \ie no garbage collection) must be factored into design and performance.
+Note, creation cost (closure/stack) is amortized across usage, so activation cost (resume/suspend) is usually the dominant factor.
+% The stateful function is an old idea~\cite{Conway63,Marlin80} that is new again~\cite{C++20Coroutine19}, where execution is temporarily suspended and later resumed, \eg plugin, device driver, finite-state machine.
+% Hence, a stateful function may not end when it returns to its caller, allowing it to be restarted with the data and execution location present at the point of suspension.
+% If the closure is fixed size, we call it a \emph{generator} (or \emph{stackless}), and its control flow is restricted, \eg suspending outside the generator is prohibited.
+% If the closure is variable size, we call it a \emph{coroutine} (or \emph{stackful}), and as the names implies, often implemented with a separate stack with no programming restrictions.
+% Hence, refactoring a stackless coroutine may require changing it to stackful.
+% A foundational property of all \emph{stateful functions} is that resume/suspend \emph{do not} cause incremental stack growth, \ie resume/suspend operations are remembered through the closure not the stack.
+% As well, activating a stateful function is \emph{asymmetric} or \emph{symmetric}, identified by resume/suspend (no cycles) and resume/resume (cycles).
+% A fixed closure activated by modified call/return is faster than a variable closure activated by context switching.
+% Additionally, any storage management for the closure (especially in unmanaged languages, \ie no garbage collection) must also be factored into design and performance.
+% Therefore, selecting between stackless and stackful semantics is a tradeoff between programming requirements and performance, where stackless is faster and stackful is more general.
+% nppNote, creation cost is amortized across usage, so activation cost is usually the dominant factor.
+For example, Python presents asymmetric generators as a function object, \uC presents symmetric coroutines as a \lstinline[language=C++]|class|-like object, and many languages present threading using function pointers, @pthreads@~\cite{Butenhof97}, \Csharp~\cite{Csharp}, Go~\cite{Go}, and Scala~\cite{Scala}.
+\begin{center}
+\begin{tabular}{@{}l|l|l@{}}
+\multicolumn{1}{@{}c|}{Python asymmetric generator} & \multicolumn{1}{c|}{\uC symmetric coroutine} & \multicolumn{1}{c@{}}{Pthreads thread} \\
+\hline
+\begin{python}
+`def Gen():` $\LstCommentStyle{\color{red}// function}$
+        ... yield val ...
+gen = Gen()
+for i in range( 10 ):
+        print( next( gen ) )
+\end{python}
+&
+\begin{uC++}
+`_Coroutine Cycle {` $\LstCommentStyle{\color{red}// class}$
+        Cycle * p;
+        void main() { p->cycle(); }
+        void cycle() { resume(); }  `};`
+Cycle c1, c2; c1.p=&c2; c2.p=&c1; c1.cycle();
+\end{uC++}
+&
+\begin{cfa}
+void * rtn( void * arg ) { ... }
+int i = 3, rc;
+pthread_t t; $\C{// thread id}$
+$\LstCommentStyle{\color{red}// function pointer}$
+rc=pthread_create(&t, `rtn`, (void *)i);
+\end{cfa}
+\end{tabular}
+\end{center}
+\CFA's preferred presentation model for generators/coroutines/threads is a hybrid of functions and classes, giving an object-oriented flavour.
+Essentially, the generator/coroutine/thread function is semantically coupled with a generator/coroutine/thread custom type via the type's name.
+The custom type solves several issues, while accessing the underlying mechanisms used by the custom types is still allowed for flexibility reasons.
+Each custom type is discussed in detail in the following sections.
+\subsection{Generator}
+Stackless generators (Table~\ref{t:ExecutionPropertyComposition} case 3) have the potential to be very small and fast, \ie as small and fast as function call/return for both creation and execution.
+The \CFA goal is to achieve this performance target, possibly at the cost of some semantic complexity.
+A series of different kinds of generators and their implementation demonstrate how this goal is accomplished.\footnote{
+The \CFA operator syntax uses \lstinline|?| to denote operands, which allows precise definitions for pre, post, and infix operators, \eg \lstinline|?++|, \lstinline|++?|, and \lstinline|?+?|, in addition \lstinline|?\{\}| denotes a constructor, as in \lstinline|foo `f` = `\{`...`\}`|, \lstinline|^?\{\}| denotes a destructor, and \lstinline|?()| is \CC function call \lstinline|operator()|.
+Operator \lstinline+|+ is overloaded for printing, like bit-shift \lstinline|<<| in \CC.
+The \CFA \lstinline|with| clause opens an aggregate scope making its fields directly accessible, like Pascal \lstinline|with|, but using parallel semantics;
+multiple aggregates may be opened.
+\CFA has rebindable references \lstinline|int i, & ip = i, j; `&ip = &j;`| and non-rebindable references \lstinline|int i, & `const` ip = i, j; `&ip = &j;` // disallowed|.
+}%
 \begin{figure}
 …
         int fn = f->fn; f->fn = f->fn1;
                 f->fn1 = f->fn + fn;
         return fn;
+}
 int main() {
 …
 void `main(Fib & fib)` with(fib) {
         [fn1, fn] = [1, 0];
         for () {
 …
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
 typedef struct {
         int fn1, fn;  void * `next`;
+        int `restart`, fn1, fn;
 } Fib;
 #define FibCtor { 1, 0, NULL }
+#define FibCtor { `0`, 1, 0 }
 Fib * comain( Fib * f ) {
+        if ( f->next ) goto *f->next;
+        f->next = &&s1;
+        `static void * states[] = {&&s0, &&s1};`
+        `goto *states[f->restart];`
+  s0: f->`restart` = 1;
         for ( ;; ) {
                 return f;
           s1:; int fn = f->fn + f->fn1;
                         f->fn1 = f->fn; f->fn = fn;
+                f->fn1 = f->fn; f->fn = fn;
+        }
+}
 …
 \end{lrbox}
 \subfloat[C asymmetric generator]{\label{f:CFibonacci}\usebox\myboxA}
+\subfloat[C]{\label{f:CFibonacci}\usebox\myboxA}
 \hspace{3pt}
 \vrule
 \hspace{3pt}
 \subfloat[\CFA asymmetric generator]{\label{f:CFAFibonacciGen}\usebox\myboxB}
+\subfloat[\CFA]{\label{f:CFAFibonacciGen}\usebox\myboxB}
 \hspace{3pt}
 \vrule
 \hspace{3pt}
 \subfloat[C generator implementation]{\label{f:CFibonacciSim}\usebox\myboxC}
+\subfloat[C generated code for \CFA version]{\label{f:CFibonacciSim}\usebox\myboxC}
 \caption{Fibonacci (output) asymmetric generator}
 \label{f:FibonacciAsymmetricGenerator}
 …
 };
 void ?{}( Fmt & fmt ) { `resume(fmt);` } // constructor
 void ^?{}( Fmt & f ) with(f) { $\C[1.75in]{// destructor}$
+void ^?{}( Fmt & f ) with(f) { $\C[2.25in]{// destructor}$
         if ( g != 0 || b != 0 ) sout | nl; }
 void `main( Fmt & f )` with(f) {
 …
                 for ( ; g < 5; g += 1 ) { $\C{// groups}$
                         for ( ; b < 4; b += 1 ) { $\C{// blocks}$
                                 `suspend;` $\C{// wait for character}$
                                 while ( ch == '\n' ) `suspend;` // ignore
                                 sout | ch;                                              // newline
                         } sout | " ";  // block spacer
                 } sout | nl; // group newline
+                                do { `suspend;` $\C{// wait for character}$
+                                while ( ch == '\n' ); // ignore newline
+                                sout | ch;                      $\C{// print character}$
+                        } sout | " ";  $\C{// block separator}$
+                } sout | nl; $\C{// group separator}$
+        }
+}
 …
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
 typedef struct {
         void * next;
+        int `restart`, g, b;
         char ch;
-        int g, b;
 } Fmt;
 void comain( Fmt * f ) {
+        if ( f->next ) goto *f->next;
+        f->next = &&s1;
+        `static void * states[] = {&&s0, &&s1};`
+        `goto *states[f->restart];`
+  s0: f->`restart` = 1;
         for ( ;; ) {
                 for ( f->g = 0; f->g < 5; f->g += 1 ) {
                         for ( f->b = 0; f->b < 4; f->b += 1 ) {
                                 return;
                           s1:;  while ( f->ch == '\n' ) return;
+                                do { return;  s1: ;
+                                } while ( f->ch == '\n' );
                                 printf( "%c", f->ch );
                         } printf( " " );
 …
+}
 int main() {
         Fmt fmt = { NULL };  comain( &fmt ); // prime
+        Fmt fmt = { `0` };  comain( &fmt ); // prime
         for ( ;; ) {
                 scanf( "%c", &fmt.ch );
 …
 \end{lrbox}
 \subfloat[\CFA asymmetric generator]{\label{f:CFAFormatGen}\usebox\myboxA}
 \hspace{3pt}
+\subfloat[\CFA]{\label{f:CFAFormatGen}\usebox\myboxA}
+\hspace{35pt}
 \vrule
 \hspace{3pt}
 \subfloat[C generator simulation]{\label{f:CFormatSim}\usebox\myboxB}
+\subfloat[C generated code for \CFA version]{\label{f:CFormatGenImpl}\usebox\myboxB}
 \hspace{3pt}
 \caption{Formatter (input) asymmetric generator}
 …
 \end{figure}
+Stateful functions appear as generators, coroutines, and threads, where presentations are based on function objects or pointers~\cite{Butenhof97, C++14, MS:VisualC++, BoostCoroutines15}.
+For example, Python presents generators as a function object:
+\begin{python}
+def Gen():
+        ... `yield val` ...
+gen = Gen()
+for i in range( 10 ):
+        print( next( gen ) )
+\end{python}
+Boost presents coroutines in terms of four functor object-types:
+\begin{cfa}
+asymmetric_coroutine<>::pull_type
+asymmetric_coroutine<>::push_type
+symmetric_coroutine<>::call_type
+symmetric_coroutine<>::yield_type
+\end{cfa}
+and many languages present threading using function pointers, @pthreads@~\cite{Butenhof97}, \Csharp~\cite{Csharp}, Go~\cite{Go}, and Scala~\cite{Scala}, \eg pthreads:
+\begin{cfa}
+void * rtn( void * arg ) { ... }
+int i = 3, rc;
+pthread_t t; $\C{// thread id}$
+`rc = pthread_create( &t, rtn, (void *)i );` $\C{// create and initialized task, type-unsafe input parameter}$
+\end{cfa}
+% void mycor( pthread_t cid, void * arg ) {
+%       int * value = (int *)arg;                               $\C{// type unsafe, pointer-size only}$
+%       // thread body
+% }
+% int main() {
+%       int input = 0, output;
+%       coroutine_t cid = coroutine_create( &mycor, (void *)&input ); $\C{// type unsafe, pointer-size only}$
+%       coroutine_resume( cid, (void *)input, (void **)&output ); $\C{// type unsafe, pointer-size only}$
+% }
+\CFA's preferred presentation model for generators/coroutines/threads is a hybrid of objects and functions, with an object-oriented flavour.
+Essentially, the generator/coroutine/thread function is semantically coupled with a generator/coroutine/thread custom type.
+The custom type solves several issues, while accessing the underlying mechanisms used by the custom types is still allowed.
+\subsection{Generator}
+Stackless generators have the potential to be very small and fast, \ie as small and fast as function call/return for both creation and execution.
+The \CFA goal is to achieve this performance target, possibly at the cost of some semantic complexity.
+A series of different kinds of generators and their implementation demonstrate how this goal is accomplished.
+Figure~\ref{f:FibonacciAsymmetricGenerator} shows an unbounded asymmetric generator for an infinite sequence of Fibonacci numbers written in C and \CFA, with a simple C implementation for the \CFA version.
+Figure~\ref{f:FibonacciAsymmetricGenerator} shows an unbounded asymmetric generator for an infinite sequence of Fibonacci numbers written (left to right) in C, \CFA, and showing the underlying C implementation for the \CFA version.
 This generator is an \emph{output generator}, producing a new result on each resumption.
 To compute Fibonacci, the previous two values in the sequence are retained to generate the next value, \ie @fn1@ and @fn@, plus the execution location where control restarts when the generator is resumed, \ie top or middle.
 …
 The C version only has the middle execution state because the top execution state is declaration initialization.
 Figure~\ref{f:CFAFibonacciGen} shows the \CFA approach, which also has a manual closure, but replaces the structure with a custom \CFA @generator@ type.
+This generator type is then connected to a function that \emph{must be named \lstinline|main|},\footnote{
+The name \lstinline|main| has special meaning in C, specifically the function where a program starts execution.
+Hence, overloading this name for other starting points (generator/coroutine/thread) is a logical extension.}
+called a \emph{generator main},which takes as its only parameter a reference to the generator type.
+Each generator type must have a function named \lstinline|main|,
+% \footnote{
+% The name \lstinline|main| has special meaning in C, specifically the function where a program starts execution.
+% Leveraging starting semantics to this name for generator/coroutine/thread is a logical extension.}
+called a \emph{generator main} (leveraging the starting semantics for program @main@ in C), which is connected to the generator type via its single reference parameter.
 The generator main contains @suspend@ statements that suspend execution without ending the generator versus @return@.
+For the Fibonacci generator-main,\footnote{
+The \CFA \lstinline|with| opens an aggregate scope making its fields directly accessible, like Pascal \lstinline|with|, but using parallel semantics.
+Multiple aggregates may be opened.}
+For the Fibonacci generator-main,
 the top initialization state appears at the start and the middle execution state is denoted by statement @suspend@.
 Any local variables in @main@ \emph{are not retained} between calls;
 …
 Resuming an ended (returned) generator is undefined.
 Function @resume@ returns its argument generator so it can be cascaded in an expression, in this case to print the next Fibonacci value @fn@ computed in the generator instance.
+Figure~\ref{f:CFibonacciSim} shows the C implementation of the \CFA generator only needs one additional field, @next@, to handle retention of execution state.
+The computed @goto@ at the start of the generator main, which branches after the previous suspend, adds very little cost to the resume call.
+Finally, an explicit generator type provides both design and performance benefits, such as multiple type-safe interface functions taking and returning arbitrary types.\footnote{
+The \CFA operator syntax uses \lstinline|?| to denote operands, which allows precise definitions for pre, post, and infix operators, \eg \lstinline|++?|, \lstinline|?++|, and \lstinline|?+?|, in addition \lstinline|?\{\}| denotes a constructor, as in \lstinline|foo `f` = `\{`...`\}`|, \lstinline|^?\{\}| denotes a destructor, and \lstinline|?()| is \CC function call \lstinline|operator()|.
+}%
+Figure~\ref{f:CFibonacciSim} shows the C implementation of the \CFA asymmetric generator.
+Only one execution-state field, @restart@, is needed to subscript the suspension points in the generator.
+At the start of the generator main, the @static@ declaration, @states@, is initialized to the N suspend points in the generator (where operator @&&@ dereferences/references a label~\cite{gccValueLabels}).
+Next, the computed @goto@ selects the last suspend point and branches to it.
+The  cost of setting @restart@ and branching via the computed @goto@ adds very little cost to the suspend/resume calls.
+An advantage of the \CFA explicit generator type is the ability to allow multiple type-safe interface functions taking and returning arbitrary types.
 \begin{cfa}
 int ?()( Fib & fib ) { return `resume( fib )`.fn; } $\C[3.9in]{// function-call interface}$
+int ?()( Fib & fib, int N ) { for ( N - 1 ) `fib()`; return `fib()`; } $\C{// use function-call interface to skip N values}$
+double ?()( Fib & fib ) { return (int)`fib()` / 3.14159; } $\C{// different return type, cast prevents recursive call}\CRT$
+sout | (int)f1() | (double)f1() | f2( 2 ); // alternative interface, cast selects call based on return type, step 2 values
+int ?()( Fib & fib, int N ) { for ( N - 1 ) `fib()`; return `fib()`; } $\C{// add parameter to skip N values}$
+double ?()( Fib & fib ) { return (int)`fib()` / 3.14159; } $\C{// different return type, cast prevents recursive call}$
+Fib f;  int i;  double d;
+i = f();  i = f( 2 );  d = f();                                         $\C{// alternative interfaces}\CRT$
 \end{cfa}
 Now, the generator can be a separately compiled opaque-type only accessed through its interface functions.
 For contrast, Figure~\ref{f:PythonFibonacci} shows the equivalent Python Fibonacci generator, which does not use a generator type, and hence only has a single interface, but an implicit closure.
+Having to manually create the generator closure by moving local-state variables into the generator type is an additional programmer burden.
+(This restriction is removed by the coroutine in Section~\ref{s:Coroutine}.)
+This requirement follows from the generality of variable-size local-state, \eg local state with a variable-length array requires dynamic allocation because the array size is unknown at compile time.
+\begin{figure}
+%\centering
+\newbox\myboxA
+\begin{lrbox}{\myboxA}
+\begin{python}[aboveskip=0pt,belowskip=0pt]
+def Fib():
+        fn1, fn = 0, 1
+        while True:
+                `yield fn1`
+                fn1, fn = fn, fn1 + fn
+f1 = Fib()
+f2 = Fib()
+for i in range( 10 ):
+        print( next( f1 ), next( f2 ) )
+\end{python}
+\end{lrbox}
+\newbox\myboxB
+\begin{lrbox}{\myboxB}
+\begin{python}[aboveskip=0pt,belowskip=0pt]
+def Fmt():
+        try:
+                while True:                                             $\C[2.5in]{\# until destructor call}$
+                        for g in range( 5 ):            $\C{\# groups}$
+                                for b in range( 4 ):    $\C{\# blocks}$
+                                        while True:
+                                                ch = (yield)    $\C{\# receive from send}$
+                                                if '\n' not in ch: $\C{\# ignore newline}$
+                                                        break
+                                        print( ch, end='' )     $\C{\# print character}$
+                                print( '  ', end='' )   $\C{\# block separator}$
+                        print()                                         $\C{\# group separator}$
+        except GeneratorExit:                           $\C{\# destructor}$
+                if g != 0 | b != 0:                             $\C{\# special case}$
+                        print()
+fmt = Fmt()
+`next( fmt )`                                                   $\C{\# prime, next prewritten}$
+for i in range( 41 ):
+        `fmt.send( 'a' );`                                      $\C{\# send to yield}$
+\end{python}
+\end{lrbox}
+\hspace{30pt}
+\subfloat[Fibonacci]{\label{f:PythonFibonacci}\usebox\myboxA}
+\hspace{3pt}
+\vrule
+\hspace{3pt}
+\subfloat[Formatter]{\label{f:PythonFormatter}\usebox\myboxB}
+\caption{Python generator}
+\label{f:PythonGenerator}
+\end{figure}
+Having to manually create the generator closure by moving local-state variables into the generator type is an additional programmer burden (removed by the coroutine in Section~\ref{s:Coroutine}).
+This manual requirement follows from the generality of allowing variable-size local-state, \eg local state with a variable-length array requires dynamic allocation as the array size is unknown at compile time.
 However, dynamic allocation significantly increases the cost of generator creation/destruction and is a showstopper for embedded real-time programming.
 But more importantly, the size of the generator type is tied to the local state in the generator main, which precludes separate compilation of the generator main, \ie a generator must be inlined or local state must be dynamically allocated.
 With respect to safety, we believe static analysis can discriminate local state from temporary variables in a generator, \ie variable usage spanning @suspend@, and generate a compile-time error.
 Finally, our current experience is that most generator problems have simple data state, including local state, but complex execution state, so the burden of creating the generator type is small.
+With respect to safety, we believe static analysis can discriminate persistent generator state from temporary generator-main state and raise a compile-time error for temporary usage spanning suspend points.
+Our experience using generators is that the problems have simple data state, including local state, but complex execution state, so the burden of creating the generator type is small.
 As well, C programmers are not afraid of this kind of semantic programming requirement, if it results in very small, fast generators.
 …
 The example takes advantage of resuming a generator in the constructor to prime the loops so the first character sent for formatting appears inside the nested loops.
 The destructor provides a newline, if formatted text ends with a full line.
+Figure~\ref{f:CFormatSim} shows the C implementation of the \CFA input generator with one additional field and the computed @goto@.
+For contrast, Figure~\ref{f:PythonFormatter} shows the equivalent Python format generator with the same properties as the Fibonacci generator.
+Figure~\ref{f:DeviceDriverGen} shows a \emph{killer} asymmetric generator, a device-driver, because device drivers caused 70\%-85\% of failures in Windows/Linux~\cite{Swift05}.
+Device drives follow the pattern of simple data state but complex execution state, \ie finite state-machine (FSM) parsing a protocol.
+For example, the following protocol:
+Figure~\ref{f:CFormatGenImpl} shows the C implementation of the \CFA input generator with one additional field and the computed @goto@.
+For contrast, Figure~\ref{f:PythonFormatter} shows the equivalent Python format generator with the same properties as the format generator.
+% https://dl-acm-org.proxy.lib.uwaterloo.ca/
+Figure~\ref{f:DeviceDriverGen} shows an important application for an asymmetric generator, a device-driver, because device drivers are a significant source of operating-system errors: 85\% in Windows XP~\cite[p.~78]{Swift05} and 51.6\% in Linux~\cite[p.~1358,]{Xiao19}. %\cite{Palix11}
+Swift \etal~\cite[p.~86]{Swift05} restructure device drivers using the Extension Procedure Call (XPC) within the kernel via functions @nooks_driver_call@ and @nooks_kernel_call@, which have coroutine properties context switching to separate stacks with explicit hand-off calls;
+however, the calls do not retain execution state, and hence always start from the top.
+The alternative approach for implementing device drivers is using stack-ripping.
+However, Adya \etal~\cite{Adya02} argue against stack ripping in Section 3.2 and suggest a hybrid approach in Section 4 using cooperatively scheduled \emph{fibers}, which is coroutining.
+As an example, the following protocol:
 \begin{center}
 \ldots\, STX \ldots\, message \ldots\, ESC ETX \ldots\, message \ldots\, ETX 2-byte crc \ldots
 \end{center}
 is a network message beginning with the control character STX, ending with an ETX, and followed by a 2-byte cyclic-redundancy check.
+is for a simple network message beginning with the control character STX, ending with an ETX, and followed by a 2-byte cyclic-redundancy check.
 Control characters may appear in a message if preceded by an ESC.
 When a message byte arrives, it triggers an interrupt, and the operating system services the interrupt by calling the device driver with the byte read from a hardware register.
+The device driver returns a status code of its current state, and when a complete message is obtained, the operating system knows the message is in the message buffer.
+Hence, the device driver is an input/output generator.
+Note, the cost of creating and resuming the device-driver generator, @Driver@, is virtually identical to call/return, so performance in an operating-system kernel is excellent.
+As well, the data state is small, where variables @byte@ and @msg@ are communication variables for passing in message bytes and returning the message, and variables @lnth@, @crc@, and @sum@ are local variable that must be retained between calls and are manually hoisted into the generator type.
+% Manually, detecting and hoisting local-state variables is easy when the number is small.
+In contrast, the execution state is large, with one @resume@ and seven @suspend@s.
+Hence, the key benefits of the generator are correctness, safety, and maintenance because the execution states are transcribed directly into the programming language rather than using a table-driven approach.
+Because FSMs can be complex and frequently occur in important domains, direct generator support is important in a system programming language.
+The device driver returns a status code of its current state, and when a complete message is obtained, the operating system read the message accumulated in the supplied buffer.
+Hence, the device driver is an input/output generator, where the cost of resuming the device-driver generator is the same as call/return, so performance in an operating-system kernel is excellent.
+The key benefits of using a generator are correctness, safety, and maintenance because the execution states are transcribed directly into the programming language rather than table lookup or stack ripping.
+The conclusion is that FSMs are complex and occur in important domains, so direct generator support is important in a system programming language.
 \begin{figure}
 \centering
-\newbox\myboxA
-\begin{lrbox}{\myboxA}
-\begin{python}[aboveskip=0pt,belowskip=0pt]
-def Fib():
-        fn1, fn = 0, 1
-        while True:
-                `yield fn1`
-                fn1, fn = fn, fn1 + fn
-f1 = Fib()
-f2 = Fib()
-for i in range( 10 ):
-        print( next( f1 ), next( f2 ) )
-\end{python}
-\end{lrbox}
-\newbox\myboxB
-\begin{lrbox}{\myboxB}
-\begin{python}[aboveskip=0pt,belowskip=0pt]
-def Fmt():
-        try:
-                while True:
-                        for g in range( 5 ):
-                                for b in range( 4 ):
-                                        print( `(yield)`, end='' )
-                                print( '  ', end='' )
-                        print()
-        except GeneratorExit:
-                if g != 0 | b != 0:
-                        print()
-fmt = Fmt()
-`next( fmt )`                    # prime, next prewritten
-for i in range( 41 ):
-        `fmt.send( 'a' );`      # send to yield
-\end{python}
-\end{lrbox}
-\subfloat[Fibonacci]{\label{f:PythonFibonacci}\usebox\myboxA}
-\hspace{3pt}
-\vrule
-\hspace{3pt}
-\subfloat[Formatter]{\label{f:PythonFormatter}\usebox\myboxB}
-\caption{Python generator}
-\label{f:PythonGenerator}
-\bigskip
 \begin{tabular}{@{}l|l@{}}
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
 …
 `generator` Driver {
         Status status;
         unsigned char byte, * msg; // communication
         unsigned int lnth, sum;      // local state
         unsigned short int crc;
+        char byte, * msg; // communication
+        int lnth, sum;      // local state
+        short int crc;
 };
 void ?{}( Driver & d, char * m ) { d.msg = m; }
 …
 (The trivial cycle is a generator resuming itself.)
 This control flow is similar to recursion for functions but without stack growth.
 The steps for symmetric control-flow are creating, executing, and terminating the cycle.
+Figure~\ref{f:PingPongFullCoroutineSteps} shows the steps for symmetric control-flow are creating, executing, and terminating the cycle.
 Constructing the cycle must deal with definition-before-use to close the cycle, \ie, the first generator must know about the last generator, which is not within scope.
 (This issue occurs for any cyclic data structure.)
 % The example creates all the generators and then assigns the partners that form the cycle.
 % Alternatively, the constructor can assign the partners as they are declared, except the first, and the first-generator partner is set after the last generator declaration to close the cycle.
 Once the cycle is formed, the program main resumes one of the generators, and the generators can then traverse an arbitrary cycle using @resume@ to activate partner generator(s).
+The example creates the generators, @ping@/@pong@, and then assigns the partners that form the cycle.
+% (Alternatively, the constructor can assign the partners as they are declared, except the first, and the first-generator partner is set after the last generator declaration to close the cycle.)
+Once the cycle is formed, the program main resumes one of the generators, @ping@, and the generators can then traverse an arbitrary cycle using @resume@ to activate partner generator(s).
 Terminating the cycle is accomplished by @suspend@ or @return@, both of which go back to the stack frame that started the cycle (program main in the example).
+Note, the creator and starter may be different, \eg if the creator calls another function that starts the cycle.
 The starting stack-frame is below the last active generator because the resume/resume cycle does not grow the stack.
+Also, since local variables are not retained in the generator function, it does not contain any objects with destructors that must be called, so the  cost is the same as a function return.
+Destructor cost occurs when the generator instance is deallocated, which is easily controlled by the programmer.
+Figure~\ref{f:CPingPongSim} shows the implementation of the symmetric generator, where the complexity is the @resume@, which needs an extension to the calling convention to perform a forward rather than backward jump.
+This jump-starts at the top of the next generator main to re-execute the normal calling convention to make space on the stack for its local variables.
+However, before the jump, the caller must reset its stack (and any registers) equivalent to a @return@, but subsequently jump forward.
+This semantics is basically a tail-call optimization, which compilers already perform.
+The example shows the assembly code to undo the generator's entry code before the direct jump.
+This assembly code depends on what entry code is generated, specifically if there are local variables and the level of optimization.
+To provide this new calling convention requires a mechanism built into the compiler, which is beyond the scope of \CFA at this time.
+Nevertheless, it is possible to hand generate any symmetric generators for proof of concept and performance testing.
+A compiler could also eliminate other artifacts in the generator simulation to further increase performance, \eg LLVM has various coroutine support~\cite{CoroutineTS}, and \CFA can leverage this support should it fork @clang@.
+Also, since local variables are not retained in the generator function, there are no objects with destructors to be called, so the cost is the same as a function return.
+Destructor cost occurs when the generator instance is deallocated by the creator.
 \begin{figure}
 …
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
 `generator PingPong` {
+        int N, i;                               // local state
         const char * name;
-        int N;
-        int i;                          // local state
         PingPong & partner; // rebindable reference
 };
 void `main( PingPong & pp )` with(pp) {
         for ( ; i < N; i += 1 ) {
                 sout | name | i;
 …
 \begin{cfa}[escapechar={},aboveskip=0pt,belowskip=0pt]
 typedef struct PingPong {
+        int restart, N, i;
         const char * name;
-        int N, i;
         struct PingPong * partner;
-        void * next;
 } PingPong;
 #define PPCtor(name, N) {name,N,0,NULL,NULL}
+#define PPCtor(name, N) {0, N, 0, name, NULL}
 void comain( PingPong * pp ) {
+        if ( pp->next ) goto *pp->next;
+        pp->next = &&cycle;
+        static void * states[] = {&&s0, &&s1};
+        goto *states[pp->restart];
+  s0: pp->restart = 1;
         for ( ; pp->i < pp->N; pp->i += 1 ) {
                 printf( "%s %d\n", pp->name, pp->i );
                 asm( "mov  %0,%%rdi" : "=m" (pp->partner) );
                 asm( "mov  %rdi,%rax" );
+                asm( "popq %rbx" );
+                asm( "add  $16, %rsp" );
+                asm( "popq %rbp" );
                 asm( "jmp  comain" );
           cycle: ;
+          s1: ;
+        }
+}
 …
 \end{figure}
+Finally, part of this generator work was inspired by the recent \CCtwenty generator proposal~\cite{C++20Coroutine19} (which they call coroutines).
+\begin{figure}
+\centering
+\input{FullCoroutinePhases.pstex_t}
+\vspace*{-10pt}
+\caption{Symmetric coroutine steps: Ping / Pong}
+\label{f:PingPongFullCoroutineSteps}
+\end{figure}
+Figure~\ref{f:CPingPongSim} shows the C implementation of the \CFA symmetric generator, where there is still only one additional field, @restart@, but @resume@ is more complex because it does a forward rather than backward jump.
+Before the jump, the parameter for the next call @partner@ is placed into the register used for the first parameter, @rdi@, and the remaining registers are reset for a return.
+The @jmp comain@ restarts the function but with a different parameter, so the new call's behaviour depends on the state of the coroutine type, i.e., branch to restart location with different data state.
+While the semantics of call forward is a tail-call optimization, which compilers perform, the generator state is different on each call rather a common state for a tail-recursive function (i.e., the parameter to the function never changes during the forward calls.
+However, this assembler code depends on what entry code is generated, specifically if there are local variables and the level of optimization.
+Hence, internal compiler support is necessary for any forward call (or backwards return), \eg LLVM has various coroutine support~\cite{CoroutineTS}, and \CFA can leverage this support should it eventually fork @clang@.
+For this reason, \CFA does not support general symmetric generators at this time, but, it is possible to hand generate any symmetric generators (as in Figure~\ref{f:CPingPongSim}) for proof of concept and performance testing.
+Finally, part of this generator work was inspired by the recent \CCtwenty coroutine proposal~\cite{C++20Coroutine19}, which uses the general term coroutine to mean generator.
 Our work provides the same high-performance asymmetric generators as \CCtwenty, and extends their work with symmetric generators.
 An additional \CCtwenty generator feature allows @suspend@ and @resume@ to be followed by a restricted compound statement that is executed after the current generator has reset its stack but before calling the next generator, specified with \CFA syntax:
 …
 \label{s:Coroutine}
 Stackful coroutines extend generator semantics, \ie there is an implicit closure and @suspend@ may appear in a helper function called from the coroutine main.
+Stackful coroutines (Table~\ref{t:ExecutionPropertyComposition} case 5) extend generator semantics, \ie there is an implicit closure and @suspend@ may appear in a helper function called from the coroutine main.
 A coroutine is specified by replacing @generator@ with @coroutine@ for the type.
 Coroutine generality results in higher cost for creation, due to dynamic stack allocation, execution, due to context switching among stacks, and terminating, due to possible stack unwinding and dynamic stack deallocation.
+Coroutine generality results in higher cost for creation, due to dynamic stack allocation, for execution, due to context switching among stacks, and for terminating, due to possible stack unwinding and dynamic stack deallocation.
 A series of different kinds of coroutines and their implementations demonstrate how coroutines extend generators.
 First, the previous generator examples are converted to their coroutine counterparts, allowing local-state variables to be moved from the generator type into the coroutine main.
+\begin{description}
+\item[Fibonacci]
+Move the declaration of @fn1@ to the start of coroutine main.
+\begin{center}
+\begin{tabular}{@{}l|l|l|l@{}}
+\multicolumn{1}{c|}{Fibonacci} & \multicolumn{1}{c|}{Formatter} & \multicolumn{1}{c|}{Device Driver} & \multicolumn{1}{c}{PingPong} \\
+\hline
 \begin{cfa}[xleftmargin=0pt]
 void main( Fib & fib ) with(fib) {
+void main( Fib & fib ) ...
         `int fn1;`
+\end{cfa}
+\item[Formatter]
+Move the declaration of @g@ and @b@ to the for loops in the coroutine main.
+\end{cfa}
+&
 \begin{cfa}[xleftmargin=0pt]
 for ( `g`; 5 ) {
         for ( `b`; 4 ) {
+\end{cfa}
+\item[Device Driver]
+Move the declaration of @lnth@ and @sum@ to their points of initialization.
+\end{cfa}
+&
 \begin{cfa}[xleftmargin=0pt]
+        status = CONT;
+        `unsigned int lnth = 0, sum = 0;`
+        ...
+        `unsigned short int crc = byte << 8;`
+\end{cfa}
+\item[PingPong]
+Move the declaration of @i@ to the for loop in the coroutine main.
+status = CONT;
+`int lnth = 0, sum = 0;`
+...
+`short int crc = byte << 8;`
+\end{cfa}
+&
 \begin{cfa}[xleftmargin=0pt]
 void main( PingPong & pp ) with(pp) {
+void main( PingPong & pp ) ...
         for ( `i`; N ) {
+\end{cfa}
+\end{description}
+\end{cfa}
+\end{tabular}
+\end{center}
 It is also possible to refactor code containing local-state and @suspend@ statements into a helper function, like the computation of the CRC for the device driver.
 \begin{cfa}
 unsigned int Crc() {
+int Crc() {
         `suspend;`
         unsigned short int crc = byte << 8;
+        short int crc = byte << 8;
         `suspend;`
         status = (crc | byte) == sum ? MSG : ECRC;
 …
 \begin{comment}
 Figure~\ref{f:Coroutine3States} creates a @coroutine@ type, @`coroutine` Fib { int fn; }@, which provides communication, @fn@, for the \newterm{coroutine main}, @main@, which runs on the coroutine stack, and possibly multiple interface functions, \eg @next@.
+Figure~\ref{f:Coroutine3States} creates a @coroutine@ type, @`coroutine` Fib { int fn; }@, which provides communication, @fn@, for the \newterm{coroutine main}, @main@, which runs on the coroutine stack, and possibly multiple interface functions, \eg @restart@.
 Like the structure in Figure~\ref{f:ExternalState}, the coroutine type allows multiple instances, where instances of this type are passed to the (overloaded) coroutine main.
 The coroutine main's stack holds the state for the next generation, @f1@ and @f2@, and the code represents the three states in the Fibonacci formula via the three suspend points, to context switch back to the caller's @resume@.
 The interface function @next@, takes a Fibonacci instance and context switches to it using @resume@;
+The interface function @restart@, takes a Fibonacci instance and context switches to it using @resume@;
 on restart, the Fibonacci field, @fn@, contains the next value in the sequence, which is returned.
 The first @resume@ is special because it allocates the coroutine stack and cocalls its coroutine main on that stack;
 …
 \begin{figure}
 \centering
-\lstset{language=CFA,escapechar={},moredelim=**[is][\protect\color{red}]{`}{`}}% allow $
 \begin{tabular}{@{}l@{\hspace{2\parindentlnth}}l@{}}
 \begin{cfa}
 `coroutine` Prod {
         Cons & c;                       // communication
+        Cons & c;                       $\C[1.5in]{// communication}$
         int N, money, receipt;
 };
 void main( Prod & prod ) with( prod ) {
+        // 1st resume starts here
+        for ( i; N ) {
+        for ( i; N ) {          $\C{// 1st resume}\CRT$
                 int p1 = random( 100 ), p2 = random( 100 );
-                sout | p1 | " " | p2;
                 int status = delivery( c, p1, p2 );
-                sout | " $" | money | nl | status;
                 receipt += 1;
+        }
         stop( c );
-        sout | "prod stops";
+}
 int payment( Prod & prod, int money ) {
 …
 \begin{cfa}
 `coroutine` Cons {
         Prod & p;                       // communication
+        Prod & p;                       $\C[1.5in]{// communication}$
         int p1, p2, status;
         bool done;
 };
 void ?{}( Cons & cons, Prod & p ) {
         &cons.p = &p; // reassignable reference
+        &cons.p = &p;           $\C{// reassignable reference}$
         cons.[status, done ] = [0, false];
+}
 void main( Cons & cons ) with( cons ) {
+        // 1st resume starts here
+        int money = 1, receipt;
+        int money = 1, receipt; $\C{// 1st resume}\CRT$
         for ( ; ! done; ) {
-                sout | p1 | " " | p2 | nl | " $" | money;
                 status += 1;
                 receipt = payment( p, money );
-                sout | " #" | receipt;
                 money += 1;
+        }
-        sout | "cons stops";
+}
 int delivery( Cons & cons, int p1, int p2 ) {
 …
 This example is illustrative because both producer/consumer have two interface functions with @resume@s that suspend execution in these interface (helper) functions.
 The program main creates the producer coroutine, passes it to the consumer coroutine in its initialization, and closes the cycle at the call to @start@ along with the number of items to be produced.
+The first @resume@ of @prod@ creates @prod@'s stack with a frame for @prod@'s coroutine main at the top, and context switches to it.
+@prod@'s coroutine main starts, creates local-state variables that are retained between coroutine activations, and executes $N$ iterations, each generating two random values, calling the consumer to deliver the values, and printing the status returned from the consumer.
+The call to @start@ is the first @resume@ of @prod@, which remembers the program main as the starter and creates @prod@'s stack with a frame for @prod@'s coroutine main at the top, and context switches to it.
+@prod@'s coroutine main starts, creates local-state variables that are retained between coroutine activations, and executes $N$ iterations, each generating two random values, calling the consumer's @deliver@ function to transfer the values, and printing the status returned from the consumer.
 The producer call to @delivery@ transfers values into the consumer's communication variables, resumes the consumer, and returns the consumer status.
+On the first resume, @cons@'s stack is created and initialized, holding local-state variables retained between subsequent activations of the coroutine.
+The consumer iterates until the @done@ flag is set, prints the values delivered by the producer, increments status, and calls back to the producer via @payment@, and on return from @payment@, prints the receipt from the producer and increments @money@ (inflation).
+The call from the consumer to @payment@ introduces the cycle between producer and consumer.
+When @payment@ is called, the consumer copies values into the producer's communication variable and a resume is executed.
+The context switch restarts the producer at the point where it last context switched, so it continues in @delivery@ after the resume.
+@delivery@ returns the status value in @prod@'s coroutine main, where the status is printed.
+The loop then repeats calling @delivery@, where each call resumes the consumer coroutine.
+The context switch to the consumer continues in @payment@.
+The consumer increments and returns the receipt to the call in @cons@'s coroutine main.
+The loop then repeats calling @payment@, where each call resumes the producer coroutine.
+Similarly on the first resume, @cons@'s stack is created and initialized, holding local-state variables retained between subsequent activations of the coroutine.
+The symmetric coroutine cycle forms when the consumer calls the producer's @payment@ function, which resumes the producer in the consumer's delivery function.
+When the producer calls @delivery@ again, it resumes the consumer in the @payment@ function.
+Both interface function than return to the their corresponding coroutine-main functions for the next cycle.
 Figure~\ref{f:ProdConsRuntimeStacks} shows the runtime stacks of the program main, and the coroutine mains for @prod@ and @cons@ during the cycling.
+As a consequence of a coroutine retaining its last resumer for suspending back, these reverse pointers allow @suspend@ to cycle \emph{backwards} around a symmetric coroutine cycle.
 \begin{figure}
 …
 \caption{Producer / consumer runtime stacks}
 \label{f:ProdConsRuntimeStacks}
-\medskip
-\begin{center}
-\input{FullCoroutinePhases.pstex_t}
-\end{center}
-\vspace*{-10pt}
-\caption{Ping / Pong coroutine steps}
-\label{f:PingPongFullCoroutineSteps}
 \end{figure}
 Terminating a coroutine cycle is more complex than a generator cycle, because it requires context switching to the program main's \emph{stack} to shutdown the program, whereas generators started by the program main run on its stack.
+Furthermore, each deallocated coroutine must guarantee all destructors are run for object allocated in the coroutine type \emph{and} allocated on the coroutine's stack at the point of suspension, which can be arbitrarily deep.
+When a coroutine's main ends, its stack is already unwound so any stack allocated objects with destructors have been finalized.
+Furthermore, each deallocated coroutine must execute all destructors for object allocated in the coroutine type \emph{and} allocated on the coroutine's stack at the point of suspension, which can be arbitrarily deep.
+In the example, termination begins with the producer's loop stopping after N iterations and calling the consumer's @stop@ function, which sets the @done@ flag, resumes the consumer in function @payment@, terminating the call, and the consumer's loop in its coroutine main.
+% (Not shown is having @prod@ raise a nonlocal @stop@ exception at @cons@ after it finishes generating values and suspend back to @cons@, which catches the @stop@ exception to terminate its loop.)
+When the consumer's main ends, its stack is already unwound so any stack allocated objects with destructors are finalized.
+The question now is where does control continue?
 The na\"{i}ve semantics for coroutine-cycle termination is to context switch to the last resumer, like executing a @suspend@/@return@ in a generator.
 However, for coroutines, the last resumer is \emph{not} implicitly below the current stack frame, as for generators, because each coroutine's stack is independent.
 Unfortunately, it is impossible to determine statically if a coroutine is in a cycle and unrealistic to check dynamically (graph-cycle problem).
 Hence, a compromise solution is necessary that works for asymmetric (acyclic) and symmetric (cyclic) coroutines.
 Our solution is to context switch back to the first resumer (starter) once the coroutine ends.
+Our solution is to retain a coroutine's starter (first resumer), and context switch back to the starter when the coroutine ends.
+Hence, the consumer restarts its first resumer, @prod@, in @stop@, and when the producer ends, it restarts its first resumer, program main, in @start@ (see dashed lines from the end of the coroutine mains in Figure~\ref{f:ProdConsRuntimeStacks}).
 This semantics works well for the most common asymmetric and symmetric coroutine usage patterns.
+For asymmetric coroutines, it is common for the first resumer (starter) coroutine to be the only resumer.
+All previous generators converted to coroutines have this property.
+For symmetric coroutines, it is common for the cycle creator to persist for the lifetime of the cycle.
+Hence, the starter coroutine is remembered on the first resume and ending the coroutine resumes the starter.
+Figure~\ref{f:ProdConsRuntimeStacks} shows this semantic by the dashed lines from the end of the coroutine mains: @prod@ starts @cons@ so @cons@ resumes @prod@ at the end, and the program main starts @prod@ so @prod@ resumes the program main at the end.
+For asymmetric coroutines, it is common for the first resumer (starter) coroutine to be the only resumer;
+for symmetric coroutines, it is common for the cycle creator to persist for the lifetime of the cycle.
 For other scenarios, it is always possible to devise a solution with additional programming effort, such as forcing the cycle forward (backward) to a safe point before starting termination.
+The producer/consumer example does not illustrate the full power of the starter semantics because @cons@ always ends first.
+Assume generator @PingPong@ is converted to a coroutine.
+Figure~\ref{f:PingPongFullCoroutineSteps} shows the creation, starter, and cyclic execution steps of the coroutine version.
+The program main creates (declares) coroutine instances @ping@ and @pong@.
+Next, program main resumes @ping@, making it @ping@'s starter, and @ping@'s main resumes @pong@'s main, making it @pong@'s starter.
+Execution forms a cycle when @pong@ resumes @ping@, and cycles $N$ times.
+By adjusting $N$ for either @ping@/@pong@, it is possible to have either one finish first, instead of @pong@ always ending first.
+If @pong@ ends first, it resumes its starter @ping@ in its coroutine main, then @ping@ ends and resumes its starter the program main in function @start@.
+If @ping@ ends first, it resumes its starter the program main in function @start@.
+Regardless of the cycle complexity, the starter stack always leads back to the program main, but the stack can be entered at an arbitrary point.
+Once back at the program main, coroutines @ping@ and @pong@ are deallocated.
+For generators, deallocation runs the destructors for all objects in the generator type.
+For coroutines, deallocation deals with objects in the coroutine type and must also run the destructors for any objects pending on the coroutine's stack for any unterminated coroutine.
+Hence, if a coroutine's destructor detects the coroutine is not ended, it implicitly raises a cancellation exception (uncatchable exception) at the coroutine and resumes it so the cancellation exception can propagate to the root of the coroutine's stack destroying all local variable on the stack.
+So the \CFA semantics for the generator and coroutine, ensure both can be safely deallocated at any time, regardless of their current state, like any other aggregate object.
+Explicitly raising normal exceptions at another coroutine can replace flag variables, like @stop@, \eg @prod@ raises a @stop@ exception at @cons@ after it finishes generating values and resumes @cons@, which catches the @stop@ exception to terminate its loop.
+Finally, there is an interesting effect for @suspend@ with symmetric coroutines.
+A coroutine must retain its last resumer to suspend back because the resumer is on a different stack.
+These reverse pointers allow @suspend@ to cycle \emph{backwards}, which may be useful in certain cases.
+However, there is an anomaly if a coroutine resumes itself, because it overwrites its last resumer with itself, losing the ability to resume the last external resumer.
+To prevent losing this information, a self-resume does not overwrite the last resumer.
+Note, the producer/consumer example does not illustrate the full power of the starter semantics because @cons@ always ends first.
+Assume generator @PingPong@ in Figure~\ref{f:PingPongSymmetricGenerator} is converted to a coroutine.
+Unlike generators, coroutines have a starter structure with multiple levels, where the program main starts @ping@ and @ping@ starts @pong@.
+By adjusting $N$ for either @ping@/@pong@, it is possible to have either finish first.
+If @pong@ ends first, it resumes its starter @ping@ in its coroutine main, then @ping@ ends and resumes its starter the program main on return;
+if @ping@ ends first, it resumes its starter the program main on return.
+Regardless of the cycle complexity, the starter structure always leads back to the program main, but the path can be entered at an arbitrary point.
+Once back at the program main (creator), coroutines @ping@ and @pong@ are deallocated, runnning any destructors for objects within the coroutine and possibly deallocating any coroutine stacks for non-terminated coroutines, where stack deallocation implies stack unwinding to find destructors for allocated objects on the stack.
+Hence, the \CFA termination semantics for the generator and coroutine ensure correct deallocation semnatics, regardless of the coroutine's state (terminated or active), like any other aggregate object.
 …
 Users wanting to extend custom types or build their own can only do so in ways offered by the language.
 Furthermore, implementing custom types without language support may display the power of a programming language.
 \CFA blends the two approaches, providing custom type for idiomatic \CFA code, while extending and building new custom types is still possible, similar to Java concurrency with builtin and library.
+\CFA blends the two approaches, providing custom type for idiomatic \CFA code, while extending and building new custom types is still possible, similar to Java concurrency with builtin and library (@java.util.concurrent@) monitors.
 Part of the mechanism to generalize custom types is the \CFA trait~\cite[\S~2.3]{Moss18}, \eg the definition for custom-type @coroutine@ is anything satisfying the trait @is_coroutine@, and this trait both enforces and restricts the coroutine-interface functions.
 …
 forall( `dtype` T | is_coroutine(T) ) void $suspend$( T & ), resume( T & );
 \end{cfa}
+Note, copying generators/coroutines/threads is not meaningful.
+For example, both the resumer and suspender descriptors can have bidirectional pointers;
+copying these coroutines does not update the internal pointers so behaviour of both copies would be difficult to understand.
+Furthermore, two coroutines cannot logically execute on the same stack.
+A deep coroutine copy, which copies the stack, is also meaningless in an unmanaged language (no garbage collection), like C, because the stack may contain pointers to object within it that require updating for the copy.
+Note, copying generators/coroutines/threads is undefined because muliple objects cannot execute on a shared stack and stack copying does not work in unmanaged languages (no garbage collection), like C, because the stack may contain pointers to objects within it that require updating for the copy.
 The \CFA @dtype@ property provides no \emph{implicit} copying operations and the @is_coroutine@ trait provides no \emph{explicit} copying operations, so all coroutines must be passed by reference (pointer).
 The function definitions ensure there is a statically typed @main@ function that is the starting point (first stack frame) of a coroutine, and a mechanism to get (read) the coroutine descriptor from its handle.
 …
 The combination of custom types and fundamental @trait@ description of these types allows a concise specification for programmers and tools, while more advanced programmers can have tighter control over memory layout and initialization.
 Figure~\ref{f:CoroutineMemoryLayout} shows different memory-layout options for a coroutine (where a task is similar).
+Figure~\ref{f:CoroutineMemoryLayout} shows different memory-layout options for a coroutine (where a thread is similar).
 The coroutine handle is the @coroutine@ instance containing programmer specified type global/communication variables across interface functions.
 The coroutine descriptor contains all implicit declarations needed by the runtime, \eg @suspend@/@resume@, and can be part of the coroutine handle or separate.
 The coroutine stack can appear in a number of locations and be fixed or variable sized.
 Hence, the coroutine's stack could be a VLS\footnote{
 We are examining variable-sized structures (VLS), where fields can be variable-sized structures or arrays.
+Hence, the coroutine's stack could be a variable-length structure (VLS)\footnote{
+We are examining VLSs, where fields can be variable-sized structures or arrays.
 Once allocated, a VLS is fixed sized.}
 on the allocating stack, provided the allocating stack is large enough.
 For a VLS stack allocation/deallocation is an inexpensive adjustment of the stack pointer, modulo any stack constructor costs (\eg initial frame setup).
 For heap stack allocation, allocation/deallocation is an expensive heap allocation (where the heap can be a shared resource), modulo any stack constructor costs.
 With heap stack allocation, it is also possible to use a split (segmented) stack calling convention, available with gcc and clang, so the stack is variable sized.
+For stack allocation in the heap, allocation/deallocation is an expensive allocation, where the heap can be a shared resource, modulo any stack constructor costs.
+It is also possible to use a split (segmented) stack calling convention, available with gcc and clang, allowing a variable-sized stack via a set of connected blocks in the heap.
 Currently, \CFA supports stack/heap allocated descriptors but only fixed-sized heap allocated stacks.
 In \CFA debug-mode, the fixed-sized stack is terminated with a write-only page, which catches most stack overflows.
 Experience teaching concurrency with \uC~\cite{CS343} shows fixed-sized stacks are rarely an issue for students.
 Split-stack allocation is under development but requires recompilation of legacy code, which may be impossible.
+Split-stack allocation is under development but requires recompilation of legacy code, which is not always possible.
 \begin{figure}
 …
 Concurrency is nondeterministic scheduling of independent sequential execution paths (threads), where each thread has its own stack.
 A single thread with multiple call stacks, \newterm{coroutining}~\cite{Conway63,Marlin80}, does \emph{not} imply concurrency~\cite[\S~2]{Buhr05a}.
 In coroutining, coroutines self-schedule the thread across stacks so execution is deterministic.
+A single thread with multiple stacks, \ie coroutining, does \emph{not} imply concurrency~\cite[\S~3]{Buhr05a}.
+Coroutining self-schedule the thread across stacks so execution is deterministic.
 (It is \emph{impossible} to generate a concurrency error when coroutining.)
+However, coroutines are a stepping stone towards concurrency.
+The transition to concurrency, even for a single thread with multiple stacks, occurs when coroutines context switch to a \newterm{scheduling coroutine}, introducing non-determinism from the coroutine perspective~\cite[\S~3,]{Buhr05a}.
+The transition to concurrency, even for a single thread with multiple stacks, occurs when coroutines context switch to a \newterm{scheduling coroutine}, introducing non-determinism from the coroutine perspective~\cite[\S~3]{Buhr05a}.
 Therefore, a minimal concurrency system requires coroutines \emph{in conjunction with a nondeterministic scheduler}.
+The resulting execution system now follows a cooperative threading model~\cite{Adya02,libdill}, called \newterm{non-preemptive scheduling}.
+Adding \newterm{preemption} introduces non-cooperative scheduling, where context switching occurs randomly between any two instructions often based on a timer interrupt, called \newterm{preemptive scheduling}.
+While a scheduler introduces uncertain execution among explicit context switches, preemption introduces uncertainty by introducing implicit context switches.
+The resulting execution system now follows a cooperative threading-model~\cite{Adya02,libdill} because context-switching points to the scheduler (blocking) are known, but the next unblocking point is unknown due to the scheduler.
+Adding \newterm{preemption} introduces \newterm{non-cooperative} or \newterm{preemptive} scheduling, where context switching points to the scheduler are unknown as they can occur randomly between any two instructions often based on a timer interrupt.
 Uncertainty gives the illusion of parallelism on a single processor and provides a mechanism to access and increase performance on multiple processors.
 The reason is that the scheduler/runtime have complete knowledge about resources and how to best utilized them.
 However, the introduction of unrestricted nondeterminism results in the need for \newterm{mutual exclusion} and \newterm{synchronization}, which restrict nondeterminism for correctness;
+However, the introduction of unrestricted nondeterminism results in the need for \newterm{mutual exclusion} and \newterm{synchronization}~\cite[\S~4]{Buhr05a}, which restrict nondeterminism for correctness;
 otherwise, it is impossible to write meaningful concurrent programs.
 Optimal concurrent performance is often obtained by having as much nondeterminism as mutual exclusion and synchronization correctness allow.
 A scheduler can either be a stackless or stackful.
+A scheduler can also be stackless or stackful.
 For stackless, the scheduler performs scheduling on the stack of the current coroutine and switches directly to the next coroutine, so there is one context switch.
 For stackful, the current coroutine switches to the scheduler, which performs scheduling, and it then switches to the next coroutine, so there are two context switches.
 …
 \label{s:threads}
 Threading needs the ability to start a thread and wait for its completion.
+Threading (Table~\ref{t:ExecutionPropertyComposition} case 11) needs the ability to start a thread and wait for its completion.
 A common API for this ability is @fork@ and @join@.
+\begin{cquote}
+\begin{tabular}{@{}lll@{}}
+\multicolumn{1}{c}{\textbf{Java}} & \multicolumn{1}{c}{\textbf{\Celeven}} & \multicolumn{1}{c}{\textbf{pthreads}} \\
+\begin{cfa}
+class MyTask extends Thread {...}
+mytask t = new MyTask(...);
+\vspace{4pt}
+\par\noindent
+\begin{tabular}{@{}l|l|l@{}}
+\multicolumn{1}{c|}{\textbf{Java}} & \multicolumn{1}{c|}{\textbf{\Celeven}} & \multicolumn{1}{c}{\textbf{pthreads}} \\
+\hline
+\begin{cfa}
+class MyThread extends Thread {...}
+mythread t = new MyThread(...);
 `t.start();` // start
 // concurrency
 …
+&
 \begin{cfa}
 class MyTask { ... } // functor
 MyTask mytask;
 `thread t( mytask, ... );` // start
+class MyThread { ... } // functor
+MyThread mythread;
+`thread t( mythread, ... );` // start
 // concurrency
 `t.join();` // wait
 …
 \end{cfa}
 \end{tabular}
+\end{cquote}
+\vspace{1pt}
+\par\noindent
 \CFA has a simpler approach using a custom @thread@ type and leveraging declaration semantics (allocation/deallocation), where threads implicitly @fork@ after construction and @join@ before destruction.
 \begin{cfa}
 thread MyTask {};
 void main( MyTask & this ) { ... }
+thread MyThread {};
+void main( MyThread & this ) { ... }
 int main() {
         MyTask team`[10]`; $\C[2.5in]{// allocate stack-based threads, implicit start after construction}$
+        MyThread team`[10]`; $\C[2.5in]{// allocate stack-based threads, implicit start after construction}$
         // concurrency
 } $\C{// deallocate stack-based threads, implicit joins before destruction}$
 …
 Arbitrary topologies are possible using dynamic allocation, allowing threads to outlive their declaration scope, identical to normal dynamic allocation.
 \begin{cfa}
 MyTask * factory( int N ) { ... return `anew( N )`; } $\C{// allocate heap-based threads, implicit start after construction}$
+MyThread * factory( int N ) { ... return `anew( N )`; } $\C{// allocate heap-based threads, implicit start after construction}$
 int main() {
         MyTask * team = factory( 10 );
+        MyThread * team = factory( 10 );
         // concurrency
         `delete( team );` $\C{// deallocate heap-based threads, implicit joins before destruction}\CRT$
 …
 Threads in \CFA are user level run by runtime kernel threads (see Section~\ref{s:CFARuntimeStructure}), where user threads provide concurrency and kernel threads provide parallelism.
 Like coroutines, and for the same design reasons, \CFA provides a custom @thread@ type and a @trait@ to enforce and restrict the task-interface functions.
+Like coroutines, and for the same design reasons, \CFA provides a custom @thread@ type and a @trait@ to enforce and restrict the thread-interface functions.
 \begin{cquote}
 \begin{tabular}{@{}c@{\hspace{3\parindentlnth}}c@{}}
 …
 \label{s:MutualExclusionSynchronization}
 Unrestricted nondeterminism is meaningless as there is no way to know when the result is completed without synchronization.
+Unrestricted nondeterminism is meaningless as there is no way to know when a result is completed and safe to access.
 To produce meaningful execution requires clawing back some determinism using mutual exclusion and synchronization, where mutual exclusion provides access control for threads using shared data, and synchronization is a timing relationship among threads~\cite[\S~4]{Buhr05a}.
+Some concurrent systems eliminate mutable shared-state by switching to stateless communication like message passing~\cite{Thoth,Harmony,V-Kernel,MPI} (Erlang, MPI), channels~\cite{CSP} (CSP,Go), actors~\cite{Akka} (Akka, Scala), or functional techniques (Haskell).
+The shared data protected by mutual exlusion is called a \newterm{critical section}~\cite{Dijkstra65}, and the protection can be simple (only 1 thread) or complex (only N kinds of threads, \eg group~\cite{Joung00} or readers/writer~\cite{Courtois71}).
+Without synchronization control in a critical section, an arriving thread can barge ahead of preexisting waiter threads resulting in short/long-term starvation, staleness/freshness problems, and/or incorrect transfer of data.
+Preventing or detecting barging is a challenge with low-level locks, but made easier through higher-level constructs.
+This challenge is often split into two different approaches: barging \emph{avoidance} and \emph{prevention}.
+Approaches that unconditionally releasing a lock for competing threads to acquire must use barging avoidance with flag/counter variable(s) to force barging threads to wait;
+approaches that conditionally hold locks during synchronization, \eg baton-passing~\cite{Andrews89}, prevent barging completely.
+At the lowest level, concurrent control is provided by atomic operations, upon which different kinds of locking mechanisms are constructed, \eg spin locks, semaphores~\cite{Dijkstra68b}, barriers, and path expressions~\cite{Campbell74}.
+However, for productivity it is always desirable to use the highest-level construct that provides the necessary efficiency~\cite{Hochstein05}.
+A significant challenge with locks is composability because it takes careful organization for multiple locks to be used while preventing deadlock.
+Easing composability is another feature higher-level mutual-exclusion mechanisms can offer.
+Some concurrent systems eliminate mutable shared-state by switching to non-shared communication like message passing~\cite{Thoth,Harmony,V-Kernel,MPI} (Erlang, MPI), channels~\cite{CSP} (CSP,Go), actors~\cite{Akka} (Akka, Scala), or functional techniques (Haskell).
 However, these approaches introduce a new communication mechanism for concurrency different from the standard communication using function call/return.
 Hence, a programmer must learn and manipulate two sets of design/programming patterns.
 While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account.
+In contrast, approaches based on stateful models more closely resemble the standard call/return programming model, resulting in a single programming paradigm.
+At the lowest level, concurrent control is implemented by atomic operations, upon which different kinds of locking mechanisms are constructed, \eg semaphores~\cite{Dijkstra68b}, barriers, and path expressions~\cite{Campbell74}.
+However, for productivity it is always desirable to use the highest-level construct that provides the necessary efficiency~\cite{Hochstein05}.
+A newer approach for restricting non-determinism is transactional memory~\cite{Herlihy93}.
+While this approach is pursued in hardware~\cite{Nakaike15} and system languages, like \CC~\cite{Cpp-Transactions}, the performance and feature set is still too restrictive to be the main concurrency paradigm for system languages, which is why it is rejected as the core paradigm for concurrency in \CFA.
+One of the most natural, elegant, and efficient mechanisms for mutual exclusion and synchronization for shared-memory systems is the \emph{monitor}.
+First proposed by Brinch Hansen~\cite{Hansen73} and later described and extended by C.A.R.~Hoare~\cite{Hoare74}, many concurrent programming languages provide monitors as an explicit language construct: \eg Concurrent Pascal~\cite{ConcurrentPascal}, Mesa~\cite{Mesa}, Modula~\cite{Modula-2}, Turing~\cite{Turing:old}, Modula-3~\cite{Modula-3}, NeWS~\cite{NeWS}, Emerald~\cite{Emerald}, \uC~\cite{Buhr92a} and Java~\cite{Java}.
+In addition, operating-system kernels and device drivers have a monitor-like structure, although they often use lower-level primitives such as mutex locks or semaphores to simulate monitors.
+For these reasons, \CFA selected monitors as the core high-level concurrency construct, upon which higher-level approaches can be easily constructed.
+\subsection{Mutual Exclusion}
+A group of instructions manipulating a specific instance of shared data that must be performed atomically is called a \newterm{critical section}~\cite{Dijkstra65}, which is enforced by \newterm{simple mutual-exclusion}.
+The generalization is called a \newterm{group critical-section}~\cite{Joung00}, where multiple tasks with the same session use the resource simultaneously and different sessions are segregated, which is enforced by \newterm{complex mutual-exclusion} providing the correct kind and number of threads using a group critical-section.
+The readers/writer problem~\cite{Courtois71} is an instance of a group critical-section, where readers share a session but writers have a unique session.
+However, many solutions exist for mutual exclusion, which vary in terms of performance, flexibility and ease of use.
+Methods range from low-level locks, which are fast and flexible but require significant attention for correctness, to higher-level concurrency techniques, which sacrifice some performance to improve ease of use.
+Ease of use comes by either guaranteeing some problems cannot occur, \eg deadlock free, or by offering a more explicit coupling between shared data and critical section.
+For example, the \CC @std::atomic<T>@ offers an easy way to express mutual-exclusion on a restricted set of operations, \eg reading/writing, for numerical types.
+However, a significant challenge with locks is composability because it takes careful organization for multiple locks to be used while preventing deadlock.
+Easing composability is another feature higher-level mutual-exclusion mechanisms can offer.
+\subsection{Synchronization}
+Synchronization enforces relative ordering of execution, and synchronization tools provide numerous mechanisms to establish these timing relationships.
+Low-level synchronization primitives offer good performance and flexibility at the cost of ease of use;
+higher-level mechanisms often simplify usage by adding better coupling between synchronization and data, \eg receive-specific versus receive-any thread in message passing or offering specialized solutions, \eg barrier lock.
+Often synchronization is used to order access to a critical section, \eg ensuring a waiting writer thread enters the critical section before a calling reader thread.
+If the calling reader is scheduled before the waiting writer, the reader has barged.
+Barging can result in staleness/freshness problems, where a reader barges ahead of a writer and reads temporally stale data, or a writer barges ahead of another writer overwriting data with a fresh value preventing the previous value from ever being read (lost computation).
+Preventing or detecting barging is an involved challenge with low-level locks, which is made easier through higher-level constructs.
+This challenge is often split into two different approaches: barging avoidance and prevention.
+Algorithms that unconditionally releasing a lock for competing threads to acquire use barging avoidance during synchronization to force a barging thread to wait;
+algorithms that conditionally hold locks during synchronization, \eg baton-passing~\cite{Andrews89}, prevent barging completely.
+In contrast, approaches based on shared-state models more closely resemble the standard call/return programming model, resulting in a single programming paradigm.
+Finally, a newer approach for restricting non-determinism is transactional memory~\cite{Herlihy93}.
+While this approach is pursued in hardware~\cite{Nakaike15} and system languages, like \CC~\cite{Cpp-Transactions}, the performance and feature set is still too restrictive~\cite{Cascaval08,Boehm09} to be the main concurrency paradigm for system languages.
 …
 \label{s:Monitor}
+A \textbf{monitor} is a set of functions that ensure mutual exclusion when accessing shared state.
+More precisely, a monitor is a programming technique that implicitly binds mutual exclusion to static function scope, as opposed to locks, where mutual-exclusion is defined by acquire/release calls, independent of lexical context (analogous to block and heap storage allocation).
+One of the most natural, elegant, efficient, high-level mechanisms for mutual exclusion and synchronization for shared-memory systems is the \emph{monitor} (Table~\ref{t:ExecutionPropertyComposition} case 2).
+First proposed by Brinch Hansen~\cite{Hansen73} and later described and extended by C.A.R.~Hoare~\cite{Hoare74}, many concurrent programming languages provide monitors as an explicit language construct: \eg Concurrent Pascal~\cite{ConcurrentPascal}, Mesa~\cite{Mesa}, Modula~\cite{Modula-2}, Turing~\cite{Turing:old}, Modula-3~\cite{Modula-3}, NeWS~\cite{NeWS}, Emerald~\cite{Emerald}, \uC~\cite{Buhr92a} and Java~\cite{Java}.
+In addition, operating-system kernels and device drivers have a monitor-like structure, although they often use lower-level primitives such as mutex locks or semaphores to manually implement a monitor.
+For these reasons, \CFA selected monitors as the core high-level concurrency construct, upon which higher-level approaches can be easily constructed.
+Specifically, a \textbf{monitor} is a set of functions that ensure mutual exclusion when accessing shared state.
+More precisely, a monitor is a programming technique that implicitly binds mutual exclusion to static function scope by call/return, as opposed to locks, where mutual-exclusion is defined by acquire/release calls, independent of lexical context (analogous to block and heap storage allocation).
 Restricting acquire/release points eases programming, comprehension, and maintenance, at a slight cost in flexibility and efficiency.
 \CFA uses a custom @monitor@ type and leverages declaration semantics (deallocation) to protect active or waiting threads in a monitor.
 The following is a \CFA monitor implementation of an atomic counter.
 \begin{cfa}[morekeywords=nomutex]
+\begin{cfa}
 `monitor` Aint { int cnt; }; $\C[4.25in]{// atomic integer counter}$
+int ++?( Aint & `mutex`$\(_{opt}\)$ this ) with( this ) { return ++cnt; } $\C{// increment}$
+int ?=?( Aint & `mutex`$\(_{opt}\)$ lhs, int rhs ) with( lhs ) { cnt = rhs; } $\C{// conversions with int}\CRT$
+int ?=?( int & lhs, Aint & `mutex`$\(_{opt}\)$ rhs ) with( rhs ) { lhs = cnt; }
+\end{cfa}
+% The @Aint@ constructor, @?{}@, uses the \lstinline[morekeywords=nomutex]@nomutex@ qualifier indicating mutual exclusion is unnecessary during construction because an object is inaccessible (private) until after it is initialized.
+% (While a constructor may publish its address into a global variable, doing so generates a race-condition.)
+The prefix increment operation, @++?@, is normally @mutex@, indicating mutual exclusion is necessary during function execution, to protect the incrementing from race conditions, unless there is an atomic increment instruction for the implementation type.
+The assignment operators provide bidirectional conversion between an atomic and normal integer without accessing field @cnt@;
+these operations only need @mutex@, if reading/writing the implementation type is not atomic.
+The atomic counter is used without any explicit mutual-exclusion and provides thread-safe semantics, which is similar to the \CC template @std::atomic@.
+int ++?( Aint & `mutex` this ) with( this ) { return ++cnt; } $\C{// increment}$
+int ?=?( Aint & `mutex` lhs, int rhs ) with( lhs ) { cnt = rhs; } $\C{// conversions with int, mutex optional}\CRT$
+int ?=?( int & lhs, Aint & `mutex` rhs ) with( rhs ) { lhs = cnt; }
+\end{cfa}
+The operators use the parameter-only declaration type-qualifier @mutex@ to mark which parameters require locking during function execution to protect from race conditions.
+The assignment operators provide bidirectional conversion between an atomic and normal integer without accessing field @cnt@.
+(These operations only need @mutex@, if reading/writing the implementation type is not atomic.)
+The atomic counter is used without any explicit mutual-exclusion and provides thread-safe semantics.
 \begin{cfa}
 int i = 0, j = 0, k = 5;
 …
 i = x; j = y; k = z;
 \end{cfa}
+Note, like other concurrent programming languages, \CFA has specializations for the basic types using atomic instructions for performance and a general trait similar to the \CC template @std::atomic@.
 \CFA monitors have \newterm{multi-acquire} semantics so the thread in the monitor may acquire it multiple times without deadlock, allowing recursion and calling other interface functions.
+\newpage
 \begin{cfa}
 monitor M { ... } m;
 …
 \end{cfa}
 \CFA monitors also ensure the monitor lock is released regardless of how an acquiring function ends (normal or exceptional), and returning a shared variable is safe via copying before the lock is released.
+Similar safety is offered by \emph{explicit} mechanisms like \CC RAII;
+monitor \emph{implicit} safety ensures no programmer usage errors.
+Similar safety is offered by \emph{explicit} opt-in disciplines like \CC RAII versus the monitor \emph{implicit} language-enforced safety guarantee ensuring no programmer usage errors.
 Furthermore, RAII mechanisms cannot handle complex synchronization within a monitor, where the monitor lock may not be released on function exit because it is passed to an unblocking thread;
 RAII is purely a mutual-exclusion mechanism (see Section~\ref{s:Scheduling}).
 …
 \end{cquote}
 The @dtype@ property prevents \emph{implicit} copy operations and the @is_monitor@ trait provides no \emph{explicit} copy operations, so monitors must be passed by reference (pointer).
-% Copying a lock is insecure because it is possible to copy an open lock and then use the open copy when the original lock is closed to simultaneously access the shared data.
-% Copying a monitor is secure because both the lock and shared data are copies, but copying the shared data is meaningless because it no longer represents a unique entity.
 Similarly, the function definitions ensures there is a mechanism to get (read) the monitor descriptor from its handle, and a special destructor to prevent deallocation if a thread using the shared data.
 The custom monitor type also inserts any locks needed to implement the mutual exclusion semantics.
 …
 For example, a monitor may be passed through multiple helper functions before it is necessary to acquire the monitor's mutual exclusion.
+The benefit of mandatory monitor qualifiers is self-documentation, but requiring both @mutex@ and \lstinline[morekeywords=nomutex]@nomutex@ for all monitor parameters is redundant.
+Instead, the semantics has one qualifier as the default and the other required.
+For example, make the safe @mutex@ qualifier the default because assuming \lstinline[morekeywords=nomutex]@nomutex@ may cause subtle errors.
+Alternatively, make the unsafe \lstinline[morekeywords=nomutex]@nomutex@ qualifier the default because it is the \emph{normal} parameter semantics while @mutex@ parameters are rare.
+Providing a default qualifier implies knowing whether a parameter is a monitor.
+Since \CFA relies heavily on traits as an abstraction mechanism, types can coincidentally match the monitor trait but not be a monitor, similar to inheritance where a shape and playing card can both be drawable.
+For this reason, \CFA requires programmers to identify the kind of parameter with the @mutex@ keyword and uses no keyword to mean \lstinline[morekeywords=nomutex]@nomutex@.
+\CFA requires programmers to identify the kind of parameter with the @mutex@ keyword and uses no keyword to mean \lstinline[morekeywords=nomutex]@nomutex@, because @mutex@ parameters are rare and no keyword is the \emph{normal} parameter semantics.
+Hence, @mutex@ parameters are documentation, at the function and its prototype, to both programmer and compiler, without other redundant keywords.
+Furthermore, \CFA relies heavily on traits as an abstraction mechanism, so the @mutex@ qualifier prevents coincidentally matching of a monitor trait with a type that is not a monitor, similar to coincidental inheritance where a shape and playing card can both be drawable.
 The next semantic decision is establishing which parameter \emph{types} may be qualified with @mutex@.
 …
 Function @f3@ has a multiple object matrix, and @f4@ a multiple object data structure.
 While shown shortly, multiple object acquisition is possible, but the number of objects must be statically known.
 Therefore, \CFA only acquires one monitor per parameter with at most one level of indirection, excluding pointers as it is impossible to statically determine the size.
+Therefore, \CFA only acquires one monitor per parameter with exactly one level of indirection, and exclude pointer types to unknown sized arrays.
 For object-oriented monitors, \eg Java, calling a mutex member \emph{implicitly} acquires mutual exclusion of the receiver object, @`rec`.foo(...)@.
 …
 While object-oriented monitors can be extended with a mutex qualifier for multiple-monitor members, no prior example of this feature could be found.}
 called \newterm{bulk acquire}.
 \CFA guarantees acquisition order is consistent across calls to @mutex@ functions using the same monitors as arguments, so acquiring multiple monitors is safe from deadlock.
+\CFA guarantees bulk acquisition order is consistent across calls to @mutex@ functions using the same monitors as arguments, so acquiring multiple monitors in a bulk acquire is safe from deadlock.
 Figure~\ref{f:BankTransfer} shows a trivial solution to the bank transfer problem~\cite{BankTransfer}, where two resources must be locked simultaneously, using \CFA monitors with implicit locking and \CC with explicit locking.
 A \CFA programmer only has to manage when to acquire mutual exclusion;
 …
 void transfer( BankAccount & `mutex` my,
         BankAccount & `mutex` your, int me2you ) {
+        // bulk acquire
         deposit( my, -me2you ); // debit
         deposit( your, me2you ); // credit
 …
 void transfer( BankAccount & my,
                         BankAccount & your, int me2you ) {
         `scoped_lock lock( my.m, your.m );`
+        `scoped_lock lock( my.m, your.m );` // bulk acquire
         deposit( my, -me2you ); // debit
         deposit( your, me2you ); // credit
 …
 \end{figure}
 Users can still force the acquiring order by using @mutex@/\lstinline[morekeywords=nomutex]@nomutex@.
+Users can still force the acquiring order by using or not using @mutex@.
 \begin{cfa}
 void foo( M & mutex m1, M & mutex m2 ); $\C{// acquire m1 and m2}$
 void bar( M & mutex m1, M & /* nomutex */ m2 ) { $\C{// acquire m1}$
+void bar( M & mutex m1, M & m2 ) { $\C{// only acquire m1}$
         ... foo( m1, m2 ); ... $\C{// acquire m2}$
+}
 void baz( M & /* nomutex */ m1, M & mutex m2 ) { $\C{// acquire m2}$
+void baz( M & m1, M & mutex m2 ) { $\C{// only acquire m2}$
         ... foo( m1, m2 ); ... $\C{// acquire m1}$
+}
 …
 % There are many aspects of scheduling in a concurrency system, all related to resource utilization by waiting threads, \ie which thread gets the resource next.
 % Different forms of scheduling include access to processors by threads (see Section~\ref{s:RuntimeStructureCluster}), another is access to a shared resource by a lock or monitor.
+This section discusses monitor scheduling for waiting threads eligible for entry, \ie which thread gets the shared resource next. (See Section~\ref{s:RuntimeStructureCluster} for scheduling threads on virtual processors.)
+While monitor mutual-exclusion provides safe access to shared data, the monitor data may indicate that a thread accessing it cannot proceed, \eg a bounded buffer may be full/empty so produce/consumer threads must block.
+Leaving the monitor and trying again (busy waiting) is impractical for high-level programming.
+Monitors eliminate busy waiting by providing synchronization to schedule threads needing access to the shared data, where threads block versus spinning.
+This section discusses scheduling for waiting threads eligible for monitor entry, \ie which user thread gets the shared resource next. (See Section~\ref{s:RuntimeStructureCluster} for scheduling kernel threads on virtual processors.)
+While monitor mutual-exclusion provides safe access to its shared data, the data may indicate a thread cannot proceed, \eg a bounded buffer may be full/\-empty so produce/consumer threads must block.
+Leaving the monitor and retrying (busy waiting) is impractical for high-level programming.
+Monitors eliminate busy waiting by providing synchronization within the monitor critical-section to schedule threads needing access to the shared data, where threads block versus spin.
 Synchronization is generally achieved with internal~\cite{Hoare74} or external~\cite[\S~2.9.2]{uC++} scheduling.
+\newterm{Internal scheduling} is characterized by each thread entering the monitor and making an individual decision about proceeding or blocking, while \newterm{external scheduling} is characterized by an entering thread making a decision about proceeding for itself and on behalf of other threads attempting entry.
+Finally, \CFA monitors do not allow calling threads to barge ahead of signalled threads, which simplifies synchronization among threads in the monitor and increases correctness.
+If barging is allowed, synchronization between a signaller and signallee is difficult, often requiring additional flags and multiple unblock/block cycles.
+In fact, signals-as-hints is completely opposite from that proposed by Hoare in the seminal paper on monitors~\cite[p.~550]{Hoare74}.
+\newterm{Internal} (largely) schedules threads located \emph{inside} the monitor and is accomplished using condition variables with signal and wait.
+\newterm{External} (largely) schedules threads located \emph{outside} the monitor and is accomplished with the @waitfor@ statement.
+Note, internal scheduling has a small amount of external scheduling and vice versus, so the naming denotes where the majority of the block threads reside (inside or outside) for scheduling.
+For complex scheduling, the approaches can be combined, so there can be an equal number of threads waiting inside and outside.
+\CFA monitors do not allow calling threads to barge ahead of signalled threads (via barging prevention), which simplifies synchronization among threads in the monitor and increases correctness.
+A direct consequence of this semantics is that unblocked waiting threads are not required to recheck the waiting condition, \ie waits are not in a starvation-prone busy-loop as required by the signals-as-hints style with barging.
+Preventing barging comes directly from Hoare's semantics in the seminal paper on monitors~\cite[p.~550]{Hoare74}.
 % \begin{cquote}
 % However, we decree that a signal operation be followed immediately by resumption of a waiting program, without possibility of an intervening procedure call from yet a third program.
 % It is only in this way that a waiting program has an absolute guarantee that it can acquire the resource just released by the signalling program without any danger that a third program will interpose a monitor entry and seize the resource instead.~\cite[p.~550]{Hoare74}
 % \end{cquote}
+Furthermore, \CFA concurrency has no spurious wakeup~\cite[\S~9]{Buhr05a}, which eliminates an implicit form of self barging.
+Hence, a \CFA @wait@ statement is not enclosed in a @while@ loop retesting a blocking predicate, which can cause thread starvation due to barging.
+Figure~\ref{f:MonitorScheduling} shows general internal/external scheduling (for the bounded-buffer example in Figure~\ref{f:InternalExternalScheduling}).
+External calling threads block on the calling queue, if the monitor is occupied, otherwise they enter in FIFO order.
+Internal threads block on condition queues via @wait@ and reenter from the condition in FIFO order.
+Alternatively, internal threads block on urgent from the @signal_block@ or @waitfor@, and reenter implicitly when the monitor becomes empty, \ie, the thread in the monitor exits or waits.
+There are three signalling mechanisms to unblock waiting threads to enter the monitor.
+Note, signalling cannot have the signaller and signalled thread in the monitor simultaneously because of the mutual exclusion, so either the signaller or signallee can proceed.
+For internal scheduling, threads are unblocked from condition queues using @signal@, where the signallee is moved to urgent and the signaller continues (solid line).
+Multiple signals move multiple signallees to urgent until the condition is empty.
+When the signaller exits or waits, a thread blocked on urgent is processed before calling threads to prevent barging.
+Furthermore, \CFA concurrency has no spurious wakeup~\cite[\S~9]{Buhr05a}, which eliminates an implicit self barging.
+Monitor mutual-exclusion means signalling cannot have the signaller and signalled thread in the monitor simultaneously, so only the signaller or signallee can proceed.
+Figure~\ref{f:MonitorScheduling} shows internal/external scheduling for the bounded-buffer examples in Figure~\ref{f:GenericBoundedBuffer}.
+For internal scheduling in Figure~\ref{f:BBInt}, the @signal@ moves the signallee (front thread of the specified condition queue) to urgent and the signaller continues (solid line).
+Multiple signals move multiple signallees to urgent until the condition queue is empty.
+When the signaller exits or waits, a thread is implicitly unblocked from urgent (if available) before unblocking a calling thread to prevent barging.
 (Java conceptually moves the signalled thread to the calling queue, and hence, allows barging.)
+The alternative unblock is in the opposite order using @signal_block@, where the signaller is moved to urgent and the signallee continues (dashed line), and is implicitly unblocked from urgent when the signallee exits or waits.
+For external scheduling, the condition queues are not used;
+instead threads are unblocked directly from the calling queue using @waitfor@ based on function names requesting mutual exclusion.
+(The linear search through the calling queue to locate a particular call can be reduced to $O(1)$.)
+The @waitfor@ has the same semantics as @signal_block@, where the signalled thread executes before the signallee, which waits on urgent.
+Executing multiple @waitfor@s from different signalled functions causes the calling threads to move to urgent.
+External scheduling requires urgent to be a stack, because the signaller expects to execute immediately after the specified monitor call has exited or waited.
+Internal scheduling behaves the same for an urgent stack or queue, except for multiple signalling, where the threads unblock from urgent in reverse order from signalling.
+If the restart order is important, multiple signalling by a signal thread can be transformed into daisy-chain signalling among threads, where each thread signals the next thread.
+We tried both a stack for @waitfor@ and queue for signalling, but that resulted in complex semantics about which thread enters next.
+Hence, \CFA uses a single urgent stack to correctly handle @waitfor@ and adequately support both forms of signalling.
+Signal is used when the signaller is providing the cooperation needed by the signallee (\eg creating an empty slot in a buffer for a producer) and the signaller immediately exits the monitor to run concurrently (consume the buffer element) and passes control of the monitor to the signalled thread, which can immediately take advantage of the state change.
+Specifically, the @wait@ function atomically blocks the calling thread and implicitly releases the monitor lock(s) for all monitors in the function's parameter list.
+Signalling is unconditional because signalling an empty condition queue does nothing.
+It is common to declare condition queues as monitor fields to prevent shared access, hence no locking is required for access as the queues are protected by the monitor lock.
+In \CFA, a condition queue can be created/stored independently.
 \begin{figure}
 …
 \end{figure}
-Figure~\ref{f:BBInt} shows a \CFA generic bounded-buffer with internal scheduling, where producers/consumers enter the monitor, detect the buffer is full/empty, and block on an appropriate condition variable, @full@/@empty@.
-The @wait@ function atomically blocks the calling thread and implicitly releases the monitor lock(s) for all monitors in the function's parameter list.
-The appropriate condition variable is signalled to unblock an opposite kind of thread after an element is inserted/removed from the buffer.
-Signalling is unconditional, because signalling an empty condition variable does nothing.
-It is common to declare condition variables as monitor fields to prevent shared access, hence no locking is required for access as the conditions are protected by the monitor lock.
-In \CFA, a condition variable can be created/stored independently.
-% To still prevent expensive locking on access, a condition variable is tied to a \emph{group} of monitors on first use, called \newterm{branding}, resulting in a low-cost boolean test to detect sharing from other monitors.
-% Signalling semantics cannot have the signaller and signalled thread in the monitor simultaneously, which means:
-% \begin{enumerate}
-% \item
-% The signalling thread returns immediately and the signalled thread continues.
-% \item
-% The signalling thread continues and the signalled thread is marked for urgent unblocking at the next scheduling point (exit/wait).
-% \item
-% The signalling thread blocks but is marked for urgent unblocking at the next scheduling point and the signalled thread continues.
-% \end{enumerate}
-% The first approach is too restrictive, as it precludes solving a reasonable class of problems, \eg dating service (see Figure~\ref{f:DatingService}).
-% \CFA supports the next two semantics as both are useful.
 \begin{figure}
 \centering
 …
                 T elements[10];
         };
         void ?{}( Buffer(T) & buffer ) with(buffer) {
+        void ?{}( Buffer(T) & buf ) with(buf) {
                 front = back = count = 0;
+        }
+        void insert( Buffer(T) & mutex buffer, T elem )
                                 with(buffer) {
                 if ( count == 10 ) `wait( empty )`;
                 // insert elem into buffer
+        void insert(Buffer(T) & mutex buf, T elm) with(buf){
+                if ( count == 10 ) `wait( empty )`; // full ?
+                // insert elm into buf
                 `signal( full )`;
+        }
         T remove( Buffer(T) & mutex buffer ) with(buffer) {
                 if ( count == 0 ) `wait( full )`;
                 // remove elem from buffer
+        T remove( Buffer(T) & mutex buf ) with(buf) {
+                if ( count == 0 ) `wait( full )`; // empty ?
+                // remove elm from buf
                 `signal( empty )`;
                 return elem;
+                return elm;
+        }
+}
 \end{cfa}
 \end{lrbox}
-% \newbox\myboxB
-% \begin{lrbox}{\myboxB}
-% \begin{cfa}[aboveskip=0pt,belowskip=0pt]
-% forall( otype T ) { // distribute forall
-%       monitor Buffer {
+%
-%               int front, back, count;
-%               T elements[10];
-%       };
-%       void ?{}( Buffer(T) & buffer ) with(buffer) {
-%               [front, back, count] = 0;
-%       }
-%       T remove( Buffer(T) & mutex buffer ); // forward
-%       void insert( Buffer(T) & mutex buffer, T elem )
-%                               with(buffer) {
-%               if ( count == 10 ) `waitfor( remove, buffer )`;
-%               // insert elem into buffer
+%
-%       }
-%       T remove( Buffer(T) & mutex buffer ) with(buffer) {
-%               if ( count == 0 ) `waitfor( insert, buffer )`;
-%               // remove elem from buffer
+%
-%               return elem;
-%       }
-% }
-% \end{cfa}
-% \end{lrbox}
 \newbox\myboxB
 \begin{lrbox}{\myboxB}
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
+forall( otype T ) { // distribute forall
+        monitor Buffer {
+                int front, back, count;
+                T elements[10];
+        };
+        void ?{}( Buffer(T) & buf ) with(buf) {
+                front = back = count = 0;
+        }
+        T remove( Buffer(T) & mutex buf ); // forward
+        void insert(Buffer(T) & mutex buf, T elm) with(buf){
+                if ( count == 10 ) `waitfor( remove : buf )`;
+                // insert elm into buf
+        }
+        T remove( Buffer(T) & mutex buf ) with(buf) {
+                if ( count == 0 ) `waitfor( insert : buf )`;
+                // remove elm from buf
+                return elm;
+        }
+}
+\end{cfa}
+\end{lrbox}
+\subfloat[Internal scheduling]{\label{f:BBInt}\usebox\myboxA}
+\hspace{1pt}
+\vrule
+\hspace{3pt}
+\subfloat[External scheduling]{\label{f:BBExt}\usebox\myboxB}
+\caption{Generic bounded buffer}
+\label{f:GenericBoundedBuffer}
+\end{figure}
+The @signal_block@ provides the opposite unblocking order, where the signaller is moved to urgent and the signallee continues and a thread is implicitly unblocked from urgent when the signallee exits or waits (dashed line).
+Signal block is used when the signallee is providing the cooperation needed by the signaller (\eg if the buffer is removed and a producer hands off an item to a consumer, as in Figure~\ref{f:DatingSignalBlock}) so the signaller must wait until the signallee unblocks, provides the cooperation, exits the monitor to run concurrently, and passes control of the monitor to the signaller, which can immediately take advantage of the state change.
+Using @signal@ or @signal_block@ can be a dynamic decision based on whether the thread providing the cooperation arrives before or after the thread needing the cooperation.
+External scheduling in Figure~\ref{f:BBExt} simplifies internal scheduling by eliminating condition queues and @signal@/@wait@ (cases where it cannot are discussed shortly), and has existed in the programming language Ada for almost 40 years with variants in other languages~\cite{SR,ConcurrentC++,uC++}.
+While prior languages use external scheduling solely for thread interaction, \CFA generalizes it to both monitors and threads.
+External scheduling allows waiting for events from other threads while restricting unrelated events, that would otherwise have to wait on condition queues in the monitor.
+Scheduling is controlled by the @waitfor@ statement, which atomically blocks the calling thread, releases the monitor lock, and restricts the function calls that can next acquire mutual exclusion.
+Specifically, a thread calling the monitor is unblocked directly from the calling queue based on function names that can fulfill the cooperation required by the signaller.
+(The linear search through the calling queue to locate a particular call can be reduced to $O(1)$.)
+Hence, the @waitfor@ has the same semantics as @signal_block@, where the signallee thread from the calling queue executes before the signaller, which waits on urgent.
+Now when a producer/consumer detects a full/empty buffer, the necessary cooperation for continuation is specified by indicating the next function call that can occur.
+For example, a producer detecting a full buffer must have cooperation from a consumer to remove an item so function @remove@ is accepted, which prevents producers from entering the monitor, and after a consumer calls @remove@, the producer waiting on urgent is \emph{implicitly} unblocked because it can now continue its insert operation.
+Hence, this mechanism is done in terms of control flow, next call, versus in terms of data, channels, as in Go/Rust @select@.
+While both mechanisms have strengths and weaknesses, \CFA uses the control-flow mechanism to be consistent with other language features.
+Figure~\ref{f:ReadersWriterLock} shows internal/external scheduling for a readers/writer lock with no barging and threads are serviced in FIFO order to eliminate staleness/freshness among the reader/writer threads.
+For internal scheduling in Figure~\ref{f:RWInt}, the readers and writers wait on the same condition queue in FIFO order, making it impossible to tell if a waiting thread is a reader or writer.
+To clawback the kind of thread, a \CFA condition can store user data in the node for a blocking thread at the @wait@, \ie whether the thread is a @READER@ or @WRITER@.
+An unblocked reader thread checks if the thread at the front of the queue is a reader and unblock it, \ie the readers daisy-chain signal the next group of readers demarcated by the next writer or end of the queue.
+For external scheduling in Figure~\ref{f:RWExt}, a waiting reader checks if a writer is using the resource, and if so, restricts further calls until the writer exits by calling @EndWrite@.
+The writer does a similar action for each reader or writer using the resource.
+Note, no new calls to @StartRead@/@StartWrite@ may occur when waiting for the call to @EndRead@/@EndWrite@.
+\begin{figure}
+\centering
+\newbox\myboxA
+\begin{lrbox}{\myboxA}
+\begin{cfa}[aboveskip=0pt,belowskip=0pt]
+enum RW { READER, WRITER };
 monitor ReadersWriter {
+        int rcnt, wcnt; // readers/writer using resource
+        int rcnt, wcnt; // readers/writer using resource
+        `condition RWers;`
 };
 void ?{}( ReadersWriter & rw ) with(rw) {
 …
 void EndRead( ReadersWriter & mutex rw ) with(rw) {
         rcnt -= 1;
+        if ( rcnt == 0 ) `signal( RWers )`;
+}
 void EndWrite( ReadersWriter & mutex rw ) with(rw) {
         wcnt = 0;
+        `signal( RWers );`
+}
 void StartRead( ReadersWriter & mutex rw ) with(rw) {
+        if ( wcnt > 0 ) `waitfor( EndWrite, rw );`
+        if ( wcnt !=0 || ! empty( RWers ) )
+                `wait( RWers, READER )`;
         rcnt += 1;
+        if ( ! empty(RWers) && `front(RWers) == READER` )
+                `signal( RWers )`;  // daisy-chain signalling
+}
 void StartWrite( ReadersWriter & mutex rw ) with(rw) {
         if ( wcnt > 0 ) `waitfor( EndWrite, rw );`
+        else while ( rcnt > 0 ) `waitfor( EndRead, rw );`
+        if ( wcnt != 0 || rcnt != 0 ) `wait( RWers, WRITER )`;
         wcnt = 1;
+}
 \end{cfa}
 \end{lrbox}
+\subfloat[Generic bounded buffer, internal scheduling]{\label{f:BBInt}\usebox\myboxA}
+\hspace{3pt}
+\newbox\myboxB
+\begin{lrbox}{\myboxB}
+\begin{cfa}[aboveskip=0pt,belowskip=0pt]
+monitor ReadersWriter {
+        int rcnt, wcnt; // readers/writer using resource
+};
+void ?{}( ReadersWriter & rw ) with(rw) {
+        rcnt = wcnt = 0;
+}
+void EndRead( ReadersWriter & mutex rw ) with(rw) {
+        rcnt -= 1;
+}
+void EndWrite( ReadersWriter & mutex rw ) with(rw) {
+        wcnt = 0;
+}
+void StartRead( ReadersWriter & mutex rw ) with(rw) {
+        if ( wcnt > 0 ) `waitfor( EndWrite : rw );`
+        rcnt += 1;
+}
+void StartWrite( ReadersWriter & mutex rw ) with(rw) {
+        if ( wcnt > 0 ) `waitfor( EndWrite : rw );`
+        else while ( rcnt > 0 ) `waitfor( EndRead : rw );`
+        wcnt = 1;
+}
+\end{cfa}
+\end{lrbox}
+\subfloat[Internal scheduling]{\label{f:RWInt}\usebox\myboxA}
+\hspace{1pt}
 \vrule
 \hspace{3pt}
 \subfloat[Readers / writer lock, external scheduling]{\label{f:RWExt}\usebox\myboxB}
 \caption{Internal / external scheduling}
 \label{f:InternalExternalScheduling}
+\subfloat[External scheduling]{\label{f:RWExt}\usebox\myboxB}
+\caption{Readers / writer lock}
+\label{f:ReadersWriterLock}
 \end{figure}
+Figure~\ref{f:BBInt} can be transformed into external scheduling by removing the condition variables and signals/waits, and adding the following lines at the locations of the current @wait@s in @insert@/@remove@, respectively.
+\begin{cfa}[aboveskip=2pt,belowskip=1pt]
+if ( count == 10 ) `waitfor( remove, buffer )`;       |      if ( count == 0 ) `waitfor( insert, buffer )`;
+\end{cfa}
+Here, the producers/consumers detects a full/\-empty buffer and prevents more producers/consumers from entering the monitor until there is a free/empty slot in the buffer.
+External scheduling is controlled by the @waitfor@ statement, which atomically blocks the calling thread, releases the monitor lock, and restricts the function calls that can next acquire mutual exclusion.
+If the buffer is full, only calls to @remove@ can acquire the buffer, and if the buffer is empty, only calls to @insert@ can acquire the buffer.
+Threads calling excluded functions block outside of (external to) the monitor on the calling queue, versus blocking on condition queues inside of (internal to) the monitor.
+Figure~\ref{f:RWExt} shows a readers/writer lock written using external scheduling, where a waiting reader detects a writer using the resource and restricts further calls until the writer exits by calling @EndWrite@.
+The writer does a similar action for each reader or writer using the resource.
+Note, no new calls to @StarRead@/@StartWrite@ may occur when waiting for the call to @EndRead@/@EndWrite@.
+External scheduling allows waiting for events from other threads while restricting unrelated events, that would otherwise have to wait on conditions in the monitor.
+The mechnaism can be done in terms of control flow, \eg Ada @accept@ or \uC @_Accept@, or in terms of data, \eg Go @select@ on channels.
+While both mechanisms have strengths and weaknesses, this project uses the control-flow mechanism to be consistent with other language features.
+% Two challenges specific to \CFA for external scheduling are loose object-definitions (see Section~\ref{s:LooseObjectDefinitions}) and multiple-monitor functions (see Section~\ref{s:Multi-MonitorScheduling}).
+Figure~\ref{f:DatingService} shows a dating service demonstrating non-blocking and blocking signalling.
+The dating service matches girl and boy threads with matching compatibility codes so they can exchange phone numbers.
+A thread blocks until an appropriate partner arrives.
+The complexity is exchanging phone numbers in the monitor because of the mutual-exclusion property.
+For signal scheduling, the @exchange@ condition is necessary to block the thread finding the match, while the matcher unblocks to take the opposite number, post its phone number, and unblock the partner.
+For signal-block scheduling, the implicit urgent-queue replaces the explict @exchange@-condition and @signal_block@ puts the finding thread on the urgent condition and unblocks the matcher.
+The dating service is an example of a monitor that cannot be written using external scheduling because it requires knowledge of calling parameters to make scheduling decisions, and parameters of waiting threads are unavailable;
+as well, an arriving thread may not find a partner and must wait, which requires a condition variable, and condition variables imply internal scheduling.
+Furthermore, barging corrupts the dating service during an exchange because a barger may also match and change the phone numbers, invalidating the previous exchange phone number.
+Putting loops around the @wait@s does not correct the problem;
+the simple solution must be restructured to account for barging.
+Finally, external scheduling requires urgent to be a stack, because the signaller expects to execute immediately after the specified monitor call has exited or waited.
+Internal schedulling performing multiple signalling results in unblocking from urgent in the reverse order from signalling.
+It is rare for the unblocking order to be important as an unblocked thread can be time-sliced immediately after leaving the monitor.
+If the unblocking order is important, multiple signalling can be restructured into daisy-chain signalling, where each thread signals the next thread.
+Hence, \CFA uses a single urgent stack to correctly handle @waitfor@ and adequately support both forms of signalling.
+(Advanced @waitfor@ features are discussed in Section~\ref{s:ExtendedWaitfor}.)
 \begin{figure}
 …
 };
 int girl( DS & mutex ds, int phNo, int ccode ) {
         if ( is_empty( Boys[ccode] ) ) {
+        if ( empty( Boys[ccode] ) ) {
                 wait( Girls[ccode] );
                 GirlPhNo = phNo;
 …
 };
 int girl( DS & mutex ds, int phNo, int ccode ) {
         if ( is_empty( Boys[ccode] ) ) { // no compatible
+        if ( empty( Boys[ccode] ) ) { // no compatible
                 wait( Girls[ccode] ); // wait for boy
                 GirlPhNo = phNo; // make phone number available
 …
 \qquad
 \subfloat[\lstinline@signal_block@]{\label{f:DatingSignalBlock}\usebox\myboxB}
 \caption{Dating service}
 \label{f:DatingService}
+\caption{Dating service Monitor}
+\label{f:DatingServiceMonitor}
 \end{figure}
+In summation, for internal scheduling, non-blocking signalling (as in the producer/consumer example) is used when the signaller is providing the cooperation for a waiting thread;
+the signaller enters the monitor and changes state, detects a waiting threads that can use the state, performs a non-blocking signal on the condition queue for the waiting thread, and exits the monitor to run concurrently.
+The waiter unblocks next from the urgent queue, uses/takes the state, and exits the monitor.
+Blocking signal is the reverse, where the waiter is providing the cooperation for the signalling thread;
+the signaller enters the monitor, detects a waiting thread providing the necessary state, performs a blocking signal to place it on the urgent queue and unblock the waiter.
+The waiter changes state and exits the monitor, and the signaller unblocks next from the urgent queue to use/take the state.
+Figure~\ref{f:DatingServiceMonitor} shows a dating service demonstrating non-blocking and blocking signalling.
+The dating service matches girl and boy threads with matching compatibility codes so they can exchange phone numbers.
+A thread blocks until an appropriate partner arrives.
+The complexity is exchanging phone numbers in the monitor because of the mutual-exclusion property.
+For signal scheduling, the @exchange@ condition is necessary to block the thread finding the match, while the matcher unblocks to take the opposite number, post its phone number, and unblock the partner.
+For signal-block scheduling, the implicit urgent-queue replaces the explicit @exchange@-condition and @signal_block@ puts the finding thread on the urgent stack and unblocks the matcher.
+The dating service is an important example of a monitor that cannot be written using external scheduling.
+First, because scheduling requires knowledge of calling parameters to make matching decisions, and parameters of calling threads are unavailable within the monitor.
+For example, a girl thread within the monitor cannot examine the @ccode@ of boy threads waiting on the calling queue to determine if there is a matching partner.
+Second, because a scheduling decision may be delayed when there is no immediate match, which requires a condition queue for waiting, and condition queues imply internal scheduling.
+For example, if a girl thread could determine there is no calling boy with the same @ccode@, it must wait until a matching boy arrives.
+Finally, barging corrupts the dating service during an exchange because a barger may also match and change the phone numbers, invalidating the previous exchange phone number.
+This situation shows rechecking the waiting condition and waiting again (signals-as-hints) fails, requiring significant restructured to account for barging.
 Both internal and external scheduling extend to multiple monitors in a natural way.
 \begin{cquote}
 \begin{tabular}{@{}l@{\hspace{3\parindentlnth}}l@{}}
+\begin{tabular}{@{}l@{\hspace{2\parindentlnth}}l@{}}
 \begin{cfa}
 monitor M { `condition e`; ... };
 …
+&
 \begin{cfa}
 void rtn$\(_1\)$( M & mutex m1, M & mutex m2 );
+void rtn$\(_1\)$( M & mutex m1, M & mutex m2 ); // overload rtn
 void rtn$\(_2\)$( M & mutex m1 );
 void bar( M & mutex m1, M & mutex m2 ) {
         ... waitfor( `rtn` ); ...       // $\LstCommentStyle{waitfor( rtn\(_1\), m1, m2 )}$
         ... waitfor( `rtn, m1` ); ... // $\LstCommentStyle{waitfor( rtn\(_2\), m1 )}$
+        ... waitfor( `rtn`${\color{red}\(_1\)}$ ); ...       // $\LstCommentStyle{waitfor( rtn\(_1\) : m1, m2 )}$
+        ... waitfor( `rtn${\color{red}\(_2\)}$ : m1` ); ...
+}
 \end{cfa}
 …
 For @wait( e )@, the default semantics is to atomically block the signaller and release all acquired mutex parameters, \ie @wait( e, m1, m2 )@.
 To override the implicit multi-monitor wait, specific mutex parameter(s) can be specified, \eg @wait( e, m1 )@.
 Wait cannot statically verifies the released monitors are the acquired mutex-parameters without disallowing separately compiled helper functions calling @wait@.
 While \CC supports bulk locking, @wait@ only accepts a single lock for a condition variable, so bulk locking with condition variables is asymmetric.
+Wait cannot statically verify the released monitors are the acquired mutex-parameters without disallowing separately compiled helper functions calling @wait@.
+While \CC supports bulk locking, @wait@ only accepts a single lock for a condition queue, so bulk locking with condition queues is asymmetric.
 Finally, a signaller,
 \begin{cfa}
 …
 must have acquired at least the same locks as the waiting thread signalled from a condition queue to allow the locks to be passed, and hence, prevent barging.
 Similarly, for @waitfor( rtn )@, the default semantics is to atomically block the acceptor and release all acquired mutex parameters, \ie @waitfor( rtn, m1, m2 )@.
 To override the implicit multi-monitor wait, specific mutex parameter(s) can be specified, \eg @waitfor( rtn, m1 )@.
+Similarly, for @waitfor( rtn )@, the default semantics is to atomically block the acceptor and release all acquired mutex parameters, \ie @waitfor( rtn : m1, m2 )@.
+To override the implicit multi-monitor wait, specific mutex parameter(s) can be specified, \eg @waitfor( rtn : m1 )@.
 @waitfor@ does statically verify the monitor types passed are the same as the acquired mutex-parameters of the given function or function pointer, hence the function (pointer) prototype must be accessible.
 % When an overloaded function appears in an @waitfor@ statement, calls to any function with that name are accepted.
 …
 void rtn( M & mutex m );
 `int` rtn( M & mutex m );
+waitfor( (`int` (*)( M & mutex ))rtn, m );
+\end{cfa}
+The ability to release a subset of acquired monitors can result in a \newterm{nested monitor}~\cite{Lister77} deadlock.
+waitfor( (`int` (*)( M & mutex ))rtn : m );
+\end{cfa}
+The ability to release a subset of acquired monitors can result in a \newterm{nested monitor}~\cite{Lister77} deadlock (see Section~\ref{s:MutexAcquisition}).
+\newpage
 \begin{cfa}
 void foo( M & mutex m1, M & mutex m2 ) {
         ... wait( `e, m1` ); ...                                $\C{// release m1, keeping m2 acquired )}$
 void bar( M & mutex m1, M & mutex m2 ) {        $\C{// must acquire m1 and m2 )}$
+        ... wait( `e, m1` ); ...                                $\C{// release m1, keeping m2 acquired}$
+void bar( M & mutex m1, M & mutex m2 ) {        $\C{// must acquire m1 and m2}$
         ... signal( `e` ); ...
 \end{cfa}
 The @wait@ only releases @m1@ so the signalling thread cannot acquire @m1@ and @m2@ to enter @bar@ and @signal@ the condition.
+While deadlock can occur with multiple/nesting acquisition, this is a consequence of locks, and by extension monitors, not being perfectly composable.
+While deadlock can occur with multiple/nesting acquisition, this is a consequence of locks, and by extension monitor locking is not perfectly composable.
 \subsection{\texorpdfstring{Extended \protect\lstinline@waitfor@}{Extended waitfor}}
+\label{s:ExtendedWaitfor}
 Figure~\ref{f:ExtendedWaitfor} shows the extended form of the @waitfor@ statement to conditionally accept one of a group of mutex functions, with an optional statement to be performed \emph{after} the mutex function finishes.
 …
 Hence, the terminating @else@ clause allows a conditional attempt to accept a call without blocking.
 If both @timeout@ and @else@ clause are present, the @else@ must be conditional, or the @timeout@ is never triggered.
+There is also a traditional future wait queue (not shown) (\eg Microsoft (@WaitForMultipleObjects@)), to wait for a specified number of future elements in the queue.
+There is also a traditional future wait queue (not shown) (\eg Microsoft @WaitForMultipleObjects@), to wait for a specified number of future elements in the queue.
+Finally, there is a shorthand for specifying multiple functions using the same set of monitors: @waitfor( f, g, h : m1, m2, m3 )@.
 \begin{figure}
 …
 The right example accepts either @mem1@ or @mem2@ if @C1@ and @C2@ are true.
 An interesting use of @waitfor@ is accepting the @mutex@ destructor to know when an object is deallocated, \eg assume the bounded buffer is restructred from a monitor to a thread with the following @main@.
+An interesting use of @waitfor@ is accepting the @mutex@ destructor to know when an object is deallocated, \eg assume the bounded buffer is restructured from a monitor to a thread with the following @main@.
 \begin{cfa}
 void main( Buffer(T) & buffer ) with(buffer) {
         for () {
                 `waitfor( ^?{}, buffer )` break;
                 or when ( count != 20 ) waitfor( insert, buffer ) { ... }
                 or when ( count != 0 ) waitfor( remove, buffer ) { ... }
+                `waitfor( ^?{} : buffer )` break;
+                or when ( count != 20 ) waitfor( insert : buffer ) { ... }
+                or when ( count != 0 ) waitfor( remove : buffer ) { ... }
+        }
         // clean up
 …
 To support this efficient semantics (and prevent barging), the implementation maintains a list of monitors acquired for each blocked thread.
 When a signaller exits or waits in a monitor function/statement, the front waiter on urgent is unblocked if all its monitors are released.
 Implementing a fast subset check for the necessary released monitors is important.
+Implementing a fast subset check for the necessary released monitors is important and discussed in the following sections.
 % The benefit is encapsulating complexity into only two actions: passing monitors to the next owner when they should be released and conditionally waking threads if all conditions are met.
+\subsection{Loose Object Definitions}
+\label{s:LooseObjectDefinitions}
+In an object-oriented programming language, a class includes an exhaustive list of operations.
+A new class can add members via static inheritance but the subclass still has an exhaustive list of operations.
+(Dynamic member adding, \eg JavaScript~\cite{JavaScript}, is not considered.)
+In the object-oriented scenario, the type and all its operators are always present at compilation (even separate compilation), so it is possible to number the operations in a bit mask and use an $O(1)$ compare with a similar bit mask created for the operations specified in a @waitfor@.
+However, in \CFA, monitor functions can be statically added/removed in translation units, making a fast subset check difficult.
+\begin{cfa}
+        monitor M { ... }; // common type, included in .h file
+translation unit 1
+        void `f`( M & mutex m );
+        void g( M & mutex m ) { waitfor( `f`, m ); }
+translation unit 2
+        void `f`( M & mutex m ); $\C{// replacing f and g for type M in this translation unit}$
+        void `g`( M & mutex m );
+        void h( M & mutex m ) { waitfor( `f`, m ) or waitfor( `g`, m ); } $\C{// extending type M in this translation unit}$
+\end{cfa}
+The @waitfor@ statements in each translation unit cannot form a unique bit-mask because the monitor type does not carry that information.
+\subsection{\texorpdfstring{\protect\lstinline@waitfor@ Implementation}{waitfor Implementation}}
+\label{s:waitforImplementation}
+In a statically-typed object-oriented programming language, a class has an exhaustive list of members, even when members are added via static inheritance (see Figure~\ref{f:uCinheritance}).
+Knowing all members at compilation (even separate compilation) allows uniquely numbered them so the accept-statement implementation can use a fast/compact bit mask with $O(1)$ compare.
+\begin{figure}
+\centering
+\begin{lrbox}{\myboxA}
+\begin{uC++}[aboveskip=0pt,belowskip=0pt]
+$\emph{translation unit 1}$
+_Monitor B { // common type in .h file
+        _Mutex virtual void `f`( ... );
+        _Mutex virtual void `g`( ... );
+        _Mutex virtual void w1( ... ) { ... _Accept(`f`, `g`); ... }
+};
+$\emph{translation unit 2}$
+// include B
+_Monitor D : public B { // inherit
+        _Mutex void `h`( ... ); // add
+        _Mutex void w2( ... ) { ... _Accept(`f`, `h`); ... }
+};
+\end{uC++}
+\end{lrbox}
+\begin{lrbox}{\myboxB}
+\begin{cfa}[aboveskip=0pt,belowskip=0pt]
+$\emph{translation unit 1}$
+monitor M { ... }; // common type in .h file
+void `f`( M & mutex m, ... );
+void `g`( M & mutex m, ... );
+void w1( M & mutex m, ... ) { ... waitfor(`f`, `g` : m); ... }
+$\emph{translation unit 2}$
+// include M
+extern void `f`( M & mutex m, ... ); // import f but not g
+void `h`( M & mutex m ); // add
+void w2( M & mutex m, ... ) { ... waitfor(`f`, `h` : m); ... }
+\end{cfa}
+\end{lrbox}
+\subfloat[\uC]{\label{f:uCinheritance}\usebox\myboxA}
+\hspace{3pt}
+\vrule
+\hspace{3pt}
+\subfloat[\CFA]{\label{f:CFinheritance}\usebox\myboxB}
+\caption{Member / Function visibility}
+\label{f:MemberFunctionVisibility}
+\end{figure}
+However, the @waitfor@ statement in translation unit 2 (see Figure~\ref{f:CFinheritance}) cannot see function @g@ in translation unit 1 precluding a unique numbering for a bit-mask because the monitor type only carries the protected shared-data.
+(A possible way to construct a dense mapping is at link or load-time.)
 Hence, function pointers are used to identify the functions listed in the @waitfor@ statement, stored in a variable-sized array.
+Then, the same implementation approach used for the urgent stack is used for the calling queue.
+Each caller has a list of monitors acquired, and the @waitfor@ statement performs a (usually short) linear search matching functions in the @waitfor@ list with called functions, and then verifying the associated mutex locks can be transfers.
+(A possible way to construct a dense mapping is at link or load-time.)
+Then, the same implementation approach used for the urgent stack (see Section~\ref{s:Scheduling}) is used for the calling queue.
+Each caller has a list of monitors acquired, and the @waitfor@ statement performs a (short) linear search matching functions in the @waitfor@ list with called functions, and then verifying the associated mutex locks can be transfers.
 …
 The solution is for the programmer to disambiguate:
 \begin{cfa}
 waitfor( f, `m2` ); $\C{// wait for call to f with argument m2}$
+waitfor( f : `m2` ); $\C{// wait for call to f with argument m2}$
 \end{cfa}
 Both locks are acquired by function @g@, so when function @f@ is called, the lock for monitor @m2@ is passed from @g@ to @f@, while @g@ still holds lock @m1@.
 …
 monitor M { ... };
 void f( M & mutex m1, M & mutex m2 );
 void g( M & mutex m1, M & mutex m2 ) { waitfor( f, `m1, m2` ); $\C{// wait for call to f with arguments m1 and m2}$
+void g( M & mutex m1, M & mutex m2 ) { waitfor( f : `m1, m2` ); $\C{// wait for call to f with arguments m1 and m2}$
 \end{cfa}
 Again, the set of monitors passed to the @waitfor@ statement must be entirely contained in the set of monitors already acquired by the accepting function.
 Also, the order of the monitors in a @waitfor@ statement is unimportant.
 Figure~\ref{f:UnmatchedMutexSets} shows an example where, for internal and external scheduling with multiple monitors, a signalling or accepting thread must match exactly, \ie partial matching results in waiting.
 For both examples, the set of monitors is disjoint so unblocking is impossible.
+% Also, the order of the monitors in a @waitfor@ statement must match the order of the mutex parameters.
+Figure~\ref{f:UnmatchedMutexSets} shows internal and external scheduling with multiple monitors that must match exactly with a signalling or accepting thread, \ie partial matching results in waiting.
+In both cases, the set of monitors is disjoint so unblocking is impossible.
 \begin{figure}
 …
+}
 void g( M1 & mutex m1, M2 & mutex m2 ) {
         waitfor( f, m1, m2 );
+        waitfor( f : m1, m2 );
+}
 g( `m11`, m2 ); // block on accept
 …
 \end{figure}
-\subsection{\texorpdfstring{\protect\lstinline@mutex@ Threads}{mutex Threads}}
-Threads in \CFA can also be monitors to allow \emph{direct communication} among threads, \ie threads can have mutex functions that are called by other threads.
-Hence, all monitor features are available when using threads.
-Figure~\ref{f:DirectCommunication} shows a comparison of direct call communication in \CFA with direct channel communication in Go.
-(Ada provides a similar mechanism to the \CFA direct communication.)
-The program main in both programs communicates directly with the other thread versus indirect communication where two threads interact through a passive monitor.
-Both direct and indirection thread communication are valuable tools in structuring concurrent programs.
 \begin{figure}
 \centering
 …
 struct Msg { int i, j; };
 thread GoRtn { int i;  float f;  Msg m; };
+monitor thread GoRtn { int i;  float f;  Msg m; };
 void mem1( GoRtn & mutex gortn, int i ) { gortn.i = i; }
 void mem2( GoRtn & mutex gortn, float f ) { gortn.f = f; }
 …
         for () {
                 `waitfor( mem1, gortn )` sout | i;  // wait for calls
                 or `waitfor( mem2, gortn )` sout | f;
                 or `waitfor( mem3, gortn )` sout | m.i | m.j;
                 or `waitfor( ^?{}, gortn )` break;
+                `waitfor( mem1 : gortn )` sout | i;  // wait for calls
+                or `waitfor( mem2 : gortn )` sout | f;
+                or `waitfor( mem3 : gortn )` sout | m.i | m.j;
+                or `waitfor( ^?{} : gortn )` break; // low priority
+        }
 …
 \hspace{3pt}
 \subfloat[Go]{\label{f:Gochannel}\usebox\myboxB}
+\caption{Direct communication}
+\label{f:DirectCommunication}
+\caption{Direct versus indirect communication}
+\label{f:DirectCommunicationComparison}
+\medskip
+\begin{cfa}
+monitor thread DatingService {
+        condition Girls[CompCodes], Boys[CompCodes];
+        int girlPhoneNo, boyPhoneNo, ccode;
+};
+int girl( DatingService & mutex ds, int phoneno, int code ) with( ds ) {
+        girlPhoneNo = phoneno;  ccode = code;
+        `wait( Girls[ccode] );`                                                         $\C{// wait for boy}$
+        girlPhoneNo = phoneno;  return boyPhoneNo;
+}
+int boy( DatingService & mutex ds, int phoneno, int code ) with( ds ) {
+        boyPhoneNo = phoneno;  ccode = code;
+        `wait( Boys[ccode] );`                                                          $\C{// wait for girl}$
+        boyPhoneNo = phoneno;  return girlPhoneNo;
+}
+void main( DatingService & ds ) with( ds ) {                    $\C{// thread starts, ds defaults to mutex}$
+        for () {
+                waitfor( ^?{} ) break;                                                  $\C{// high priority}$
+                or waitfor( girl )                                                              $\C{// girl called, compatible boy ? restart boy then girl}$
+                        if ( ! is_empty( Boys[ccode] ) ) { `signal_block( Boys[ccode] );  signal_block( Girls[ccode] );` }
+                or waitfor( boy ) {                                                             $\C{// boy called, compatible girl ? restart girl then boy}$
+                        if ( ! is_empty( Girls[ccode] ) ) { `signal_block( Girls[ccode] );  signal_block( Boys[ccode] );` }
+        }
+}
+\end{cfa}
+\caption{Direct communication dating service}
+\label{f:DirectCommunicationDatingService}
 \end{figure}
 …
 void main( Ping & pi ) {
         for ( 10 ) {
                 `waitfor( ping, pi );`
+                `waitfor( ping : pi );`
                 `pong( po );`
+        }
 …
         for ( 10 ) {
                 `ping( pi );`
                 `waitfor( pong, po );`
+                `waitfor( pong : po );`
+        }
+}
 …
+\subsection{Execution Properties}
+Table~\ref{t:ObjectPropertyComposition} shows how the \CFA high-level constructs cover 3 fundamental execution properties: thread, stateful function, and mutual exclusion.
+Case 1 is a basic object, with none of the new execution properties.
+Case 2 allows @mutex@ calls to Case 1 to protect shared data.
+Case 3 allows stateful functions to suspend/resume but restricts operations because the state is stackless.
+Case 4 allows @mutex@ calls to Case 3 to protect shared data.
+Cases 5 and 6 are the same as 3 and 4 without restriction because the state is stackful.
+Cases 7 and 8 are rejected because a thread cannot execute without a stackful state in a preemptive environment when context switching from the signal handler.
+Cases 9 and 10 have a stackful thread without and with @mutex@ calls.
+For situations where threads do not require direct communication, case 9 provides faster creation/destruction by eliminating @mutex@ setup.
+\begin{table}
+\caption{Object property composition}
+\centering
+\label{t:ObjectPropertyComposition}
+\renewcommand{\arraystretch}{1.25}
+%\setlength{\tabcolsep}{5pt}
+\begin{tabular}{c|c||l|l}
+\multicolumn{2}{c||}{object properties} & \multicolumn{2}{c}{mutual exclusion} \\
+\hline
+thread  & stateful                              & \multicolumn{1}{c|}{No} & \multicolumn{1}{c}{Yes} \\
+\hline
+\hline
+No              & No                                    & \textbf{1}\ \ \ aggregate type                & \textbf{2}\ \ \ @monitor@ aggregate type \\
+\hline
+No              & Yes (stackless)               & \textbf{3}\ \ \ @generator@                   & \textbf{4}\ \ \ @monitor@ @generator@ \\
+\hline
+No              & Yes (stackful)                & \textbf{5}\ \ \ @coroutine@                   & \textbf{6}\ \ \ @monitor@ @coroutine@ \\
+\hline
+Yes             & No / Yes (stackless)  & \textbf{7}\ \ \ {\color{red}rejected} & \textbf{8}\ \ \ {\color{red}rejected} \\
+\hline
+Yes             & Yes (stackful)                & \textbf{9}\ \ \ @thread@                              & \textbf{10}\ \ @monitor@ @thread@ \\
+\end{tabular}
+\end{table}
+\subsection{\texorpdfstring{\protect\lstinline@monitor@ Generators / Coroutines / Threads}{monitor Generators / Coroutines / Threads}}
+\CFA generators, coroutines, and threads can also be monitors (Table~\ref{t:ExecutionPropertyComposition} cases 4, 6, 12) allowing safe \emph{direct communication} with threads, \ie the custom types can have mutex functions that are called by other threads.
+All monitor features are available within these mutex functions.
+For example, if the formatter generator (or coroutine equivalent) in Figure~\ref{f:CFAFormatGen} is extended with the monitor property and this interface function is used to communicate with the formatter:
+\begin{cfa}
+void fmt( Fmt & mutex fmt, char ch ) { fmt.ch = ch; resume( fmt ) }
+\end{cfa}
+multiple threads can safely pass characters for formatting.
+Figure~\ref{f:DirectCommunicationComparison} shows a comparison of direct call-communication in \CFA versus indirect channel-communication in Go.
+(Ada has a similar mechanism to \CFA direct communication.)
+The program thread in \CFA @main@ uses the call/return paradigm to directly communicate with the @GoRtn main@, whereas Go switches to the channel paradigm to indirectly communicate with the goroutine.
+Communication by multiple threads is safe for the @gortn@ thread via mutex calls in \CFA or channel assignment in Go.
+Figure~\ref{f:DirectCommunicationDatingService} shows the dating-service problem in Figure~\ref{f:DatingServiceMonitor} extended from indirect monitor communication to direct thread communication.
+When converting a monitor to a thread (server), the coding pattern is to move as much code as possible from the accepted members into the thread main so it does an much work as possible.
+Notice, the dating server is postponing requests for an unspecified time while continuing to accept new requests.
+For complex servers (web-servers), there can be hundreds of lines of code in the thread main and safe interaction with clients can be complex.
 …
 For completeness and efficiency, \CFA provides a standard set of low-level locks: recursive mutex, condition, semaphore, barrier, \etc, and atomic instructions: @fetchAssign@, @fetchAdd@, @testSet@, @compareSet@, \etc.
 Some of these low-level mechanism are used in the \CFA runtime, but we strongly advocate using high-level mechanisms whenever possible.
+Some of these low-level mechanism are used to build the \CFA runtime, but we always advocate using high-level mechanisms whenever possible.
 …
 \begin{cfa}
 struct Adder {
     int * row, cols;
+        int * row, cols;
 };
 int operator()() {
 …
 \label{s:RuntimeStructureCluster}
+A \newterm{cluster} is a collection of threads and virtual processors (abstract kernel-thread) that execute the (user) threads from its own ready queue (like an OS executing kernel threads).
+A \newterm{cluster} is a collection of user and kernel threads, where the kernel threads run the user threads from the cluster's ready queue, and the operating system runs the kernel threads on the processors from its ready queue.
+The term \newterm{virtual processor} is introduced as a synonym for kernel thread to disambiguate between user and kernel thread.
+From the language perspective, a virtual processor is an actual processor (core).
 The purpose of a cluster is to control the amount of parallelism that is possible among threads, plus scheduling and other execution defaults.
 The default cluster-scheduler is single-queue multi-server, which provides automatic load-balancing of threads on processors.
 …
 Programs may use more virtual processors than hardware processors.
 On a multiprocessor, kernel threads are distributed across the hardware processors resulting in virtual processors executing in parallel.
 (It is possible to use affinity to lock a virtual processor onto a particular hardware processor~\cite{affinityLinux, affinityWindows, affinityFreebsd, affinityNetbsd, affinityMacosx}, which is used when caching issues occur or for heterogeneous hardware processors.)
+(It is possible to use affinity to lock a virtual processor onto a particular hardware processor~\cite{affinityLinux,affinityWindows}, which is used when caching issues occur or for heterogeneous hardware processors.) %, affinityFreebsd, affinityNetbsd, affinityMacosx
 The \CFA runtime attempts to block unused processors and unblock processors as the system load increases;
 balancing the workload with processors is difficult because it requires future knowledge, \ie what will the applicaton workload do next.
+balancing the workload with processors is difficult because it requires future knowledge, \ie what will the application workload do next.
 Preemption occurs on virtual processors rather than user threads, via operating-system interrupts.
 Thus virtual processors execute user threads, where preemption frequency applies to a virtual processor, so preemption occurs randomly across the executed user threads.
 …
 Nondeterministic preemption provides fairness from long-running threads, and forces concurrent programmers to write more robust programs, rather than relying on code between cooperative scheduling to be atomic.
 This atomic reliance can fail on multi-core machines, because execution across cores is nondeterministic.
 A different reason for not supporting preemption is that it significantly complicates the runtime system, \eg Microsoft runtime does not support interrupts and on Linux systems, interrupts are complex (see below).
+A different reason for not supporting preemption is that it significantly complicates the runtime system, \eg Windows runtime does not support interrupts and on Linux systems, interrupts are complex (see below).
 Preemption is normally handled by setting a countdown timer on each virtual processor.
 When the timer expires, an interrupt is delivered, and the interrupt handler resets the countdown timer, and if the virtual processor is executing in user code, the signal handler performs a user-level context-switch, or if executing in the language runtime kernel, the preemption is ignored or rolled forward to the point where the runtime kernel context switches back to user code.
+When the timer expires, an interrupt is delivered, and its signal handler resets the countdown timer, and if the virtual processor is executing in user code, the signal handler performs a user-level context-switch, or if executing in the language runtime kernel, the preemption is ignored or rolled forward to the point where the runtime kernel context switches back to user code.
 Multiple signal handlers may be pending.
 When control eventually switches back to the signal handler, it returns normally, and execution continues in the interrupted user thread, even though the return from the signal handler may be on a different kernel thread than the one where the signal is delivered.
 The only issue with this approach is that signal masks from one kernel thread may be restored on another as part of returning from the signal handler;
 therefore, the same signal mask is required for all virtual processors in a cluster.
+Because preemption frequency is usually long (1 millisecond) performance cost is negligible.
+Linux switched a decade ago from specific to arbitrary process signal-delivery for applications with multiple kernel threads.
+\begin{cquote}
+A process-directed signal may be delivered to any one of the threads that does not currently have the signal blocked.
+If more than one of the threads has the signal unblocked, then the kernel chooses an arbitrary thread to which it will deliver the signal.
+SIGNAL(7) - Linux Programmer's Manual
+\end{cquote}
+Because preemption interval is usually long (1 millisecond) performance cost is negligible.
+Linux switched a decade ago from specific to arbitrary virtual-processor signal-delivery for applications with multiple kernel threads.
+In the new semantics, a virtual-processor directed signal may be delivered to any virtual processor created by the application that does not have the signal blocked.
 Hence, the timer-expiry signal, which is generated \emph{externally} by the Linux kernel to an application, is delivered to any of its Linux subprocesses (kernel threads).
 To ensure each virtual processor receives a preemption signal, a discrete-event simulation is run on a special virtual processor, and only it sets and receives timer events.
 …
 \label{s:Performance}
 To verify the implementation of the \CFA runtime, a series of microbenchmarks are performed comparing \CFA with pthreads, Java OpenJDK-9, Go 1.12.6 and \uC 7.0.0.
+To test the performance of the \CFA runtime, a series of microbenchmarks are used to compare \CFA with pthreads, Java 11.0.6, Go 1.12.6, Rust 1.37.0, Python 3.7.6, Node.js 12.14.1, and \uC 7.0.0.
 For comparison, the package must be multi-processor (M:N), which excludes libdill/libmil~\cite{libdill} (M:1)), and use a shared-memory programming model, \eg not message passing.
 The benchmark computer is an AMD Opteron\texttrademark\ 6380 NUMA 64-core, 8 socket, 2.5 GHz processor, running Ubuntu 16.04.6 LTS, and \CFA/\uC are compiled with gcc 6.5.
+The benchmark computer is an AMD Opteron\texttrademark\ 6380 NUMA 64-core, 8 socket, 2.5 GHz processor, running Ubuntu 16.04.6 LTS, and pthreads/\CFA/\uC are compiled with gcc 9.2.1.
 All benchmarks are run using the following harness. (The Java harness is augmented to circumvent JIT issues.)
 \begin{cfa}
+unsigned int N = 10_000_000;
+#define BENCH( `run` ) Time before = getTimeNsec();  `run;`  Duration result = (getTimeNsec() - before) / N;
+\end{cfa}
+The method used to get time is @clock_gettime( CLOCK_REALTIME )@.
+Each benchmark is performed @N@ times, where @N@ varies depending on the benchmark;
+the total time is divided by @N@ to obtain the average time for a benchmark.
+Each benchmark experiment is run 31 times.
+#define BENCH( `run` ) uint64_t start = cputime_ns();  `run;`  double result = (double)(cputime_ns() - start) / N;
+\end{cfa}
+where CPU time in nanoseconds is from the appropriate language clock.
+Each benchmark is performed @N@ times, where @N@ is selected so the benchmark runs in the range of 2--20 seconds for the specific programming language.
+The total time is divided by @N@ to obtain the average time for a benchmark.
+Each benchmark experiment is run 13 times and the average appears in the table.
 All omitted tests for other languages are functionally identical to the \CFA tests and available online~\cite{CforallBenchMarks}.
+% tar --exclude=.deps --exclude=Makefile --exclude=Makefile.in --exclude=c.c --exclude=cxx.cpp --exclude=fetch_add.c -cvhf benchmark.tar benchmark
+\paragraph{Object Creation}
+Object creation is measured by creating/deleting the specific kind of concurrent object.
+Figure~\ref{f:creation} shows the code for \CFA, with results in Table~\ref{tab:creation}.
+The only note here is that the call stacks of \CFA coroutines are lazily created, therefore without priming the coroutine to force stack creation, the creation cost is artificially low.
+\begin{multicols}{2}
+\lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}}
+\begin{cfa}
+@thread@ MyThread {};
+void @main@( MyThread & ) {}
+int main() {
+        BENCH( for ( N ) { @MyThread m;@ } )
+        sout | result`ns;
+}
+\end{cfa}
+\captionof{figure}{\CFA object-creation benchmark}
+\label{f:creation}
+\columnbreak
+\vspace*{-16pt}
+\captionof{table}{Object creation comparison (nanoseconds)}
+\label{tab:creation}
+\begin{tabular}[t]{@{}r*{3}{D{.}{.}{5.2}}@{}}
+\multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} & \multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+\CFA Coroutine Lazy             & 13.2          & 13.1          & 0.44          \\
+\CFA Coroutine Eager    & 531.3         & 536.0         & 26.54         \\
+\CFA Thread                             & 2074.9        & 2066.5        & 170.76        \\
+\uC Coroutine                   & 89.6          & 90.5          & 1.83          \\
+\uC Thread                              & 528.2         & 528.5         & 4.94          \\
+Goroutine                               & 4068.0        & 4113.1        & 414.55        \\
+Java Thread                             & 103848.5      & 104295.4      & 2637.57       \\
+Pthreads                                & 33112.6       & 33127.1       & 165.90
+\end{tabular}
+\end{multicols}
+\paragraph{Context-Switching}
+% tar --exclude-ignore=exclude -cvhf benchmark.tar benchmark
+\paragraph{Context Switching}
 In procedural programming, the cost of a function call is important as modularization (refactoring) increases.
 (In many cases, a compiler inlines function calls to eliminate this cost.)
 Similarly, when modularization extends to coroutines/tasks, the time for a context switch becomes a relevant factor.
+(In many cases, a compiler inlines function calls to increase the size and number of basic blocks for optimizing.)
+Similarly, when modularization extends to coroutines/threads, the time for a context switch becomes a relevant factor.
 The coroutine test is from resumer to suspender and from suspender to resumer, which is two context switches.
+%For async-await systems, the test is scheduling and fulfilling @N@ empty promises, where all promises are allocated before versus interleaved with fulfillment to avoid garbage collection.
+For async-await systems, the test measures the cost of the @await@ expression entering the event engine by awaiting @N@ promises, where each created promise is resolved by an immediate event in the engine (using Node.js @setImmediate@).
 The thread test is using yield to enter and return from the runtime kernel, which is two context switches.
 The difference in performance between coroutine and thread context-switch is the cost of scheduling for threads, whereas coroutines are self-scheduling.
+Figure~\ref{f:ctx-switch} only shows the \CFA code for coroutines/threads (other systems are similar) with all results in Table~\ref{tab:ctx-switch}.
+Figure~\ref{f:ctx-switch} shows the \CFA code for a coroutine/thread with results in Table~\ref{t:ctx-switch}.
+% From: Gregor Richards <gregor.richards@uwaterloo.ca>
+% To: "Peter A. Buhr" <pabuhr@plg2.cs.uwaterloo.ca>
+% Date: Fri, 24 Jan 2020 13:49:18 -0500
+%
+% I can also verify that the previous version, which just tied a bunch of promises together, *does not* go back to the
+% event loop at all in the current version of Node. Presumably they're taking advantage of the fact that the ordering of
+% events is intentionally undefined to just jump right to the next 'then' in the chain, bypassing event queueing
+% entirely. That's perfectly correct behavior insofar as its difference from the specified behavior isn't observable, but
+% it isn't typical or representative of much anything useful, because most programs wouldn't have whole chains of eager
+% promises. Also, it's not representative of *anything* you can do with async/await, as there's no way to encode such an
+% eager chain that way.
 \begin{multicols}{2}
 \lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}}
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
 @coroutine@ C {} c;
 void main( C & ) { for ( ;; ) { @suspend;@ } }
+@coroutine@ C {};
+void main( C & ) { for () { @suspend;@ } }
 int main() { // coroutine test
+        C c;
         BENCH( for ( N ) { @resume( c );@ } )
         sout | result`ns;
+}
 int main() { // task test
+        sout | result;
+}
+int main() { // thread test
         BENCH( for ( N ) { @yield();@ } )
         sout | result`ns;
+        sout | result;
+}
 \end{cfa}
 …
 \vspace*{-16pt}
 \captionof{table}{Context switch comparison (nanoseconds)}
 \label{tab:ctx-switch}
+\label{t:ctx-switch}
 \begin{tabular}{@{}r*{3}{D{.}{.}{3.2}}@{}}
 \multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} &\multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+C function              & 1.8   & 1.8   & 0.01  \\
+\CFA generator  & 2.4   & 2.2   & 0.25  \\
+\CFA Coroutine  & 36.2  & 36.2  & 0.25  \\
+\CFA Thread             & 93.2  & 93.5  & 2.09  \\
+\uC Coroutine   & 52.0  & 52.1  & 0.51  \\
+\uC Thread              & 96.2  & 96.3  & 0.58  \\
+Goroutine               & 141.0 & 141.3 & 3.39  \\
+Java Thread             & 374.0 & 375.8 & 10.38 \\
+Pthreads Thread & 361.0 & 365.3 & 13.19
+C function                      & 1.8           & 1.8           & 0.0   \\
+\CFA generator          & 1.8           & 2.0           & 0.3   \\
+\CFA coroutine          & 32.5          & 32.9          & 0.8   \\
+\CFA thread                     & 93.8          & 93.6          & 2.2   \\
+\uC coroutine           & 50.3          & 50.3          & 0.2   \\
+\uC thread                      & 97.3          & 97.4          & 1.0   \\
+Python generator        & 40.9          & 41.3          & 1.5   \\
+Node.js generator       & 32.6          & 32.2          & 1.0   \\
+Node.js await           & 1852.2        & 1854.7        & 16.4  \\
+Goroutine thread        & 143.0         & 143.3         & 1.1   \\
+Rust thread                     & 332.0         & 331.4         & 2.4   \\
+Java thread                     & 405.0         & 415.0         & 17.6  \\
+Pthreads thread         & 334.3         & 335.2         & 3.9
 \end{tabular}
 \end{multicols}
+\paragraph{Mutual-Exclusion}
+Uncontented mutual exclusion, which frequently occurs, is measured by entering/leaving a critical section.
+For monitors, entering and leaving a monitor function is measured.
+To put the results in context, the cost of entering a non-inline function and the cost of acquiring and releasing a @pthread_mutex@ lock is also measured.
+Figure~\ref{f:mutex} shows the code for \CFA with all results in Table~\ref{tab:mutex}.
+\paragraph{Internal Scheduling}
+Internal scheduling is measured using a cycle of two threads signalling and waiting.
+Figure~\ref{f:schedint} shows the code for \CFA, with results in Table~\ref{t:schedint}.
 Note, the incremental cost of bulk acquire for \CFA, which is largely a fixed cost for small numbers of mutex objects.
+Java scheduling is significantly greater because the benchmark explicitly creates multiple thread in order to prevent the JIT from making the program sequential, \ie removing all locking.
 \begin{multicols}{2}
 \lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}}
 \begin{cfa}
+volatile int go = 0;
+@condition c;@
 @monitor@ M {} m1/*, m2, m3, m4*/;
+void __attribute__((noinline))
+do_call( M & @mutex m/*, m2, m3, m4*/@ ) {}
+void call( M & @mutex p1/*, p2, p3, p4*/@ ) {
+        @signal( c );@
+}
+void wait( M & @mutex p1/*, p2, p3, p4*/@ ) {
+        go = 1; // continue other thread
+        for ( N ) { @wait( c );@ } );
+}
+thread T {};
+void main( T & ) {
+        while ( go == 0 ) { yield(); } // waiter must start first
+        BENCH( for ( N ) { call( m1/*, m2, m3, m4*/ ); } )
+        sout | result;
+}
 int main() {
+        BENCH(
+                for( N ) do_call( m1/*, m2, m3, m4*/ );
+        )
+        sout | result`ns;
+}
+\end{cfa}
+\captionof{figure}{\CFA acquire/release mutex benchmark}
+\label{f:mutex}
+        T t;
+        wait( m1/*, m2, m3, m4*/ );
+}
+\end{cfa}
+\captionof{figure}{\CFA Internal-scheduling benchmark}
+\label{f:schedint}
 \columnbreak
 \vspace*{-16pt}
+\captionof{table}{Mutex comparison (nanoseconds)}
+\label{tab:mutex}
+\begin{tabular}{@{}r*{3}{D{.}{.}{3.2}}@{}}
+\multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} &\multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+test and test-and-test lock             & 19.1  & 18.9  & 0.40  \\
+\CFA @mutex@ function, 1 arg.   & 45.9  & 46.6  & 1.45  \\
+\CFA @mutex@ function, 2 arg.   & 105.0 & 104.7 & 3.08  \\
+\CFA @mutex@ function, 4 arg.   & 165.0 & 167.6 & 5.65  \\
+\uC @monitor@ member rtn.               & 54.0  & 53.7  & 0.82  \\
+Java synchronized method                & 31.0  & 31.1  & 0.50  \\
+Pthreads Mutex Lock                             & 33.6  & 32.6  & 1.14
+\captionof{table}{Internal-scheduling comparison (nanoseconds)}
+\label{t:schedint}
+\bigskip
+\begin{tabular}{@{}r*{3}{D{.}{.}{5.2}}@{}}
+\multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} & \multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+\CFA @signal@, 1 monitor        & 364.4         & 364.2         & 4.4           \\
+\CFA @signal@, 2 monitor        & 484.4         & 483.9         & 8.8           \\
+\CFA @signal@, 4 monitor        & 709.1         & 707.7         & 15.0          \\
+\uC @signal@ monitor            & 328.3         & 327.4         & 2.4           \\
+Rust cond. variable                     & 7514.0        & 7437.4        & 397.2         \\
+Java @notify@ monitor           & 9623.0        & 9654.6        & 236.2         \\
+Pthreads cond. variable         & 5553.7        & 5576.1        & 345.6
 \end{tabular}
 \end{multicols}
 …
 External scheduling is measured using a cycle of two threads calling and accepting the call using the @waitfor@ statement.
 Figure~\ref{f:ext-sched} shows the code for \CFA, with results in Table~\ref{tab:ext-sched}.
+Figure~\ref{f:schedext} shows the code for \CFA with results in Table~\ref{t:schedext}.
 Note, the incremental cost of bulk acquire for \CFA, which is largely a fixed cost for small numbers of mutex objects.
 …
 \vspace*{-16pt}
 \begin{cfa}
+volatile int go = 0;
+@monitor@ M {} m;
+@monitor@ M {} m1/*, m2, m3, m4*/;
+void call( M & @mutex p1/*, p2, p3, p4*/@ ) {}
+void wait( M & @mutex p1/*, p2, p3, p4*/@ ) {
+        for ( N ) { @waitfor( call : p1/*, p2, p3, p4*/ );@ }
+}
 thread T {};
-void __attribute__((noinline))
-do_call( M & @mutex@ ) {}
 void main( T & ) {
+        while ( go == 0 ) { yield(); }
+        while ( go == 1 ) { do_call( m ); }
+}
+int __attribute__((noinline))
+do_wait( M & @mutex@ m ) {
+        go = 1; // continue other thread
+        BENCH( for ( N ) { @waitfor( do_call, m );@ } )
+        go = 0; // stop other thread
+        sout | result`ns;
+        BENCH( for ( N ) { call( m1/*, m2, m3, m4*/ ); } )
+        sout | result;
+}
 int main() {
         T t;
         do_wait( m );
+        wait( m1/*, m2, m3, m4*/ );
+}
 \end{cfa}
 \captionof{figure}{\CFA external-scheduling benchmark}
 \label{f:ext-sched}
+\label{f:schedext}
 \columnbreak
 …
 \vspace*{-16pt}
 \captionof{table}{External-scheduling comparison (nanoseconds)}
 \label{tab:ext-sched}
+\label{t:schedext}
 \begin{tabular}{@{}r*{3}{D{.}{.}{3.2}}@{}}
 \multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} &\multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+\CFA @waitfor@, 1 @monitor@     & 376.4 & 376.8 & 7.63  \\
+\CFA @waitfor@, 2 @monitor@     & 491.4 & 492.0 & 13.31 \\
+\CFA @waitfor@, 4 @monitor@     & 681.0 & 681.7 & 19.10 \\
+\uC @_Accept@                           & 331.1 & 331.4 & 2.66
+\CFA @waitfor@, 1 monitor       & 367.1 & 365.3 & 5.0   \\
+\CFA @waitfor@, 2 monitor       & 463.0 & 464.6 & 7.1   \\
+\CFA @waitfor@, 4 monitor       & 689.6 & 696.2 & 21.5  \\
+\uC \lstinline[language=uC++]|_Accept| monitor  & 328.2 & 329.1 & 3.4   \\
+Go \lstinline[language=Golang]|select| channel  & 365.0 & 365.5 & 1.2
 \end{tabular}
 \end{multicols}
+\paragraph{Internal Scheduling}
 Internal scheduling is measured using a cycle of two threads signalling and waiting.
 Figure~\ref{f:int-sched} shows the code for \CFA, with results in Table~\ref{tab:int-sched}.
 Note, the incremental cost of bulk acquire for \CFA, which is largely a fixed cost for small numbers of mutex objects.
 Java scheduling is significantly greater because the benchmark explicitly creates multiple thread in order to prevent the JIT from making the program sequential, \ie removing all locking.
+\paragraph{Mutual-Exclusion}
+Uncontented mutual exclusion, which frequently occurs, is measured by entering/leaving a critical section.
+For monitors, entering and leaving a monitor function is measured, otherwise the language-appropriate mutex-lock is measured.
+For comparison, a spinning (versus blocking) test-and-test-set lock is presented.
+Figure~\ref{f:mutex} shows the code for \CFA with results in Table~\ref{t:mutex}.
+Note the incremental cost of bulk acquire for \CFA, which is largely a fixed cost for small numbers of mutex objects.
 \begin{multicols}{2}
 \lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}}
 \begin{cfa}
+volatile int go = 0;
+@monitor@ M { @condition c;@ } m;
+void __attribute__((noinline))
+do_call( M & @mutex@ a1 ) { @signal( c );@ }
+thread T {};
+void main( T & this ) {
+        while ( go == 0 ) { yield(); }
+        while ( go == 1 ) { do_call( m ); }
+}
+int  __attribute__((noinline))
+do_wait( M & mutex m ) with(m) {
+        go = 1; // continue other thread
+        BENCH( for ( N ) { @wait( c );@ } );
+        go = 0; // stop other thread
+        sout | result`ns;
+}
+@monitor@ M {} m1/*, m2, m3, m4*/;
+call( M & @mutex p1/*, p2, p3, p4*/@ ) {}
 int main() {
         T t;
         do_wait( m );
+}
 \end{cfa}
 \captionof{figure}{\CFA Internal-scheduling benchmark}
 \label{f:int-sched}
+        BENCH( for( N ) call( m1/*, m2, m3, m4*/ ); )
+        sout | result;
+}
+\end{cfa}
+\captionof{figure}{\CFA acquire/release mutex benchmark}
+\label{f:mutex}
 \columnbreak
 \vspace*{-16pt}
+\captionof{table}{Internal-scheduling comparison (nanoseconds)}
+\label{tab:int-sched}
+\bigskip
+\begin{tabular}{@{}r*{3}{D{.}{.}{5.2}}@{}}
+\multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} & \multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+\CFA @signal@, 1 @monitor@      & 372.6         & 374.3         & 14.17         \\
+\CFA @signal@, 2 @monitor@      & 492.7         & 494.1         & 12.99         \\
+\CFA @signal@, 4 @monitor@      & 749.4         & 750.4         & 24.74         \\
+\uC @signal@                            & 320.5         & 321.0         & 3.36          \\
+Java @notify@                           & 10160.5       & 10169.4       & 267.71        \\
+Pthreads Cond. Variable         & 4949.6        & 5065.2        & 363
+\captionof{table}{Mutex comparison (nanoseconds)}
+\label{t:mutex}
+\begin{tabular}{@{}r*{3}{D{.}{.}{3.2}}@{}}
+\multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} &\multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+test-and-test-set lock                  & 19.1  & 18.9  & 0.4   \\
+\CFA @mutex@ function, 1 arg.   & 48.3  & 47.8  & 0.9   \\
+\CFA @mutex@ function, 2 arg.   & 86.7  & 87.6  & 1.9   \\
+\CFA @mutex@ function, 4 arg.   & 173.4 & 169.4 & 5.9   \\
+\uC @monitor@ member rtn.               & 54.8  & 54.8  & 0.1   \\
+Goroutine mutex lock                    & 34.0  & 34.0  & 0.0   \\
+Rust mutex lock                                 & 33.0  & 33.2  & 0.8   \\
+Java synchronized method                & 31.0  & 31.0  & 0.0   \\
+Pthreads mutex Lock                             & 31.0  & 31.1  & 0.4
 \end{tabular}
 \end{multicols}
+\paragraph{Creation}
+Creation is measured by creating/deleting a specific kind of control-flow object.
+Figure~\ref{f:creation} shows the code for \CFA with results in Table~\ref{t:creation}.
+Note, the call stacks of \CFA coroutines are lazily created on the first resume, therefore the cost of creation with and without a stack are presented.
+\begin{multicols}{2}
+\lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}}
+\begin{cfa}
+@coroutine@ MyCoroutine {};
+void ?{}( MyCoroutine & this ) {
+#ifdef EAGER
+        resume( this );
+#endif
+}
+void main( MyCoroutine & ) {}
+int main() {
+        BENCH( for ( N ) { @MyCoroutine c;@ } )
+        sout | result;
+}
+\end{cfa}
+\captionof{figure}{\CFA creation benchmark}
+\label{f:creation}
+\columnbreak
+\vspace*{-16pt}
+\captionof{table}{Creation comparison (nanoseconds)}
+\label{t:creation}
+\begin{tabular}[t]{@{}r*{3}{D{.}{.}{5.2}}@{}}
+\multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} & \multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\
+\CFA generator                  & 0.6           & 0.6           & 0.0           \\
+\CFA coroutine lazy             & 13.4          & 13.1          & 0.5           \\
+\CFA coroutine eager    & 144.7         & 143.9         & 1.5           \\
+\CFA thread                             & 466.4         & 468.0         & 11.3          \\
+\uC coroutine                   & 155.6         & 155.7         & 1.7           \\
+\uC thread                              & 523.4         & 523.9         & 7.7           \\
+Python generator                & 123.2         & 124.3         & 4.1           \\
+Node.js generator               & 32.3          & 32.2          & 0.3           \\
+Goroutine thread                & 751.0         & 750.5         & 3.1           \\
+Rust thread                             & 53801.0       & 53896.8       & 274.9         \\
+Java thread                             & 120274.0      & 120722.9      & 2356.7        \\
+Pthreads thread                 & 31465.5       & 31419.5       & 140.4
+\end{tabular}
+\end{multicols}
+\subsection{Discussion}
+Languages using 1:1 threading based on pthreads can at best meet or exceed (due to language overhead) the pthread results.
+Note, pthreads has a fast zero-contention mutex lock checked in user space.
+Languages with M:N threading have better performance than 1:1 because there is no operating-system interactions.
+Languages with stackful coroutines have higher cost than stackless coroutines because of stack allocation and context switching;
+however, stackful \uC and \CFA coroutines have approximately the same performance as stackless Python and Node.js generators.
+The \CFA stackless generator is approximately 25 times faster for suspend/resume and 200 times faster for creation than stackless Python and Node.js generators.
 \section{Conclusion}
 …
 Advanced control-flow will always be difficult, especially when there is temporal ordering and nondeterminism.
 However, many systems exacerbate the difficulty through their presentation mechanisms.
+This paper shows it is possible to present a hierarchy of control-flow features, generator, coroutine, thread, and monitor, providing an integrated set of high-level, efficient, and maintainable control-flow features.
+Eliminated from \CFA are spurious wakeup and barging, which are nonintuitive and lead to errors, and having to work with a bewildering set of low-level locks and acquisition techniques.
+\CFA high-level race-free monitors and tasks provide the core mechanisms for mutual exclusion and synchronization, without having to resort to magic qualifiers like @volatile@/@atomic@.
+This paper shows it is possible to understand high-level control-flow using three properties: statefulness, thread, mutual-exclusion/synchronization.
+Combining these properties creates a number of high-level, efficient, and maintainable control-flow types: generator, coroutine, thread, each of which can be a monitor.
+Eliminated from \CFA are barging and spurious wakeup, which are nonintuitive and lead to errors, and having to work with a bewildering set of low-level locks and acquisition techniques.
+\CFA high-level race-free monitors and threads provide the core mechanisms for mutual exclusion and synchronization, without having to resort to magic qualifiers like @volatile@/@atomic@.
 Extending these mechanisms to handle high-level deadlock-free bulk acquire across both mutual exclusion and synchronization is a unique contribution.
 The \CFA runtime provides concurrency based on a preemptive M:N user-level threading-system, executing in clusters, which encapsulate scheduling of work on multiple kernel threads providing parallelism.
 The M:N model is judged to be efficient and provide greater flexibility than a 1:1 threading model.
 These concepts and the \CFA runtime-system are written in the \CFA language, extensively leveraging the \CFA type-system, which demonstrates the expressiveness of the \CFA language.
 Performance comparisons with other concurrent systems/languages show the \CFA approach is competitive across all low-level operations, which translates directly into good performance in well-written concurrent applications.
 C programmers should feel comfortable using these mechanisms for developing complex control-flow in applications, with the ability to obtain maximum available performance by selecting mechanisms at the appropriate level of need.
+Performance comparisons with other concurrent systems/languages show the \CFA approach is competitive across all basic operations, which translates directly into good performance in well-written applications with advanced control-flow.
+C programmers should feel comfortable using these mechanisms for developing complex control-flow in applications, with the ability to obtain maximum available performance by selecting mechanisms at the appropriate level of need using only calling communication.
 …
 \label{futur:nbio}
 Many modern workloads are not bound by computation but IO operations, a common case being web servers and XaaS~\cite{XaaS} (anything as a service).
+Many modern workloads are not bound by computation but IO operations, common cases being web servers and XaaS~\cite{XaaS} (anything as a service).
 These types of workloads require significant engineering to amortizing costs of blocking IO-operations.
 At its core, non-blocking I/O is an operating-system level feature queuing IO operations, \eg network operations, and registering for notifications instead of waiting for requests to complete.
 …
 \section{Acknowledgements}
 The authors would like to recognize the design assistance of Aaron Moss, Rob Schluntz, Andrew Beach and Michael Brooks on the features described in this paper.
 Funding for this project has been provided by Huawei Ltd.\ (\url{http://www.huawei.com}). %, and Peter Buhr is partially funded by the Natural Sciences and Engineering Research Council of Canada.
+The authors recognize the design assistance of Aaron Moss, Rob Schluntz, Andrew Beach, and Michael Brooks; David Dice for commenting and helping with the Java benchmarks; and Gregor Richards for helping with the Node.js benchmarks.
+This research is funded by a grant from Waterloo-Huawei (\url{http://www.huawei.com}) Joint Innovation Lab. %, and Peter Buhr is partially funded by the Natural Sciences and Engineering Research Council of Canada.
 {%
 \fontsize{9bp}{12bp}\selectfont%
+\fontsize{9bp}{11.5bp}\selectfont%
 \bibliography{pl,local}
 }%

doc/papers/concurrency/examples/Fib.py

-              rb7d6a36
+              r6a490b2
         while True:
                 fn = fn1 + fn2; fn2 = fn1; fn1 = fn; yield fn
 f1 = Fib()
 …
 # Local Variables: #
 # tab-width: 4 #
 # compile-command: "python3.5 Fib.py" #
+# compile-command: "python3.7 Fib.py" #
 # End: #

doc/papers/concurrency/examples/Fib2.c

-              rb7d6a36
+              r6a490b2
 #include <stdio.h>
-void mary() {
-        printf( "MARY\n" );
+}
 #define FIB_INIT { 0 }
 typedef struct { int next; int fn1, fn2; } Fib;
+typedef struct { int restart; int fn1, fn2; } Fib;
 int fib( Fib * f ) {
+        static void * states[] = { &&s1, &&s2, &&s3 };
+        goto *states[f->next];
+        static void * states[] = { &&s0, &&s1, &&s2 };
+        goto *states[f->restart];
+  s0:
+        f->fn1 = 0;
+        f->restart = 1;
+        return f->fn1;
   s1:
-        mary();
-        f->fn1 = 0;
-        f->next = 1;
-        return f->fn1;
-  s2:
-        mary();
         f->fn2 = f->fn1;
         f->fn1 = 1;
         f->next = 2;
+        f->restart = 2;
         return f->fn1;
+  s3:;
+        mary();
+  s2:;
         int fn = f->fn1 + f->fn2;
         f->fn2 = f->fn1;

doc/papers/concurrency/examples/Fib2.py

-              rb7d6a36
+              r6a490b2
 def Fib():
     fn1, fn = 0, 1
+    fn1, fn = 1, 0
     while True:
         yield fn1
+        yield fn
         fn1, fn = fn, fn1 + fn
 …
 # Local Variables: #
 # tab-width: 4 #
 # compile-command: "python3.5 Fib2.py" #
+# compile-command: "python3.7 Fib2.py" #
 # End: #

doc/papers/concurrency/examples/Fib3.c

-              rb7d6a36
+              r6a490b2
 typedef struct {
+        int fn1, fn;
+        void * next;
+        int restart, fn1, fn;
 } Fib;
 #define FibCtor { 1, 0, NULL }
+#define FibCtor { 0, 1, 0 }
 Fib * comain( Fib * f ) {
+        if ( __builtin_expect(f->next != 0, 1) ) goto *f->next;
+        f->next = &&s1;
+        static void * states[] = {&&s0, &&s1};
+        goto *states[f->restart];
+  s0: f->restart = 1;
         for ( ;; ) {
                 return f;

doc/papers/concurrency/examples/FibRefactor.py

rb7d6a36	r6a490b2
22	22	# Local Variables: #
23	23	# tab-width: 4 #
24		# compile-command: "python3.5 FibRefactor.py" #
	24	# compile-command: "python3.7 FibRefactor.py" #
25	25	# End: #

doc/papers/concurrency/examples/Format.c

-              rb7d6a36
+              r6a490b2
 typedef struct {
         void * next;
+        int restart, g, b;
         char ch;
-        int g, b;
 } Fmt;
 void comain( Fmt * f ) {
+        if ( __builtin_expect(f->next != 0, 1) ) goto *f->next;
+        f->next = &&s1;
+        static void * states[] = {&&s0, &&s1};
+        goto *states[f->restart];
+  s0: f->restart = 1;
         for ( ;; ) {
                 for ( f->g = 0; f->g < 5; f->g += 1 ) {                 // groups
                         for ( f->b = 0; f->b < 4; f->b += 1 ) {         // blocks
+                                return;
+                          s1:;  while ( f->ch == '\n' ) return;         // ignore
+                                do {
+                                        return;  s1: ;
+                                } while ( f->ch == '\n' );                              // ignore
                                 printf( "%c", f->ch );                                  // print character
+                        }
 …
 int main() {
         Fmt fmt = { NULL };
+        Fmt fmt = { 0 };
         comain( &fmt );                                                                         // prime
         for ( ;; ) {

doc/papers/concurrency/examples/Format.cc

-              rb7d6a36
+              r6a490b2
                         for ( g = 0; g < 5; g += 1 ) { // groups of 5 blocks
                                 for ( b = 0; b < 4; b += 1 ) { // blocks of 4 characters
 //                                      for ( ;; ) { // for newline characters
+                                        for ( ;; ) { // for newline characters
                                                 suspend();
 //                                              if ( ch != '\n' ) break; // ignore newline
 //                                      }
+                                                if ( ch != '\n' ) break; // ignore newline
+                                        }
 //                                      cout << ch; // print character
+                                }
 …
 // Local Variables: //
 // tab-width: 4 //
 // compile-command: "u++-work -O2 -nodebubg Format.cc" //
+// compile-command: "u++-work -O2 -nodebug Format.cc" //
 // End: //

doc/papers/concurrency/examples/Format.cfa

-              rb7d6a36
+              r6a490b2
                 for ( g = 0; g < 5; g += 1 ) {          // groups of 5 blocks
                         for ( b = 0; b < 4; b += 1 ) {  // blocks of 4 characters
 //                              do {
+                                do {
                                         suspend();
 //                              } while ( ch == '\n' || ch == '\t' );
+                                } while ( ch == '\n' || ch == '\t' );
                                 sout | ch;                                      // print character
+                        }

doc/papers/concurrency/examples/Format.data

rb7d6a36	r6a490b2
1		abcdefghijklmnopqrstuvwxyzxxxxxxxxxxxxxx
	1	abcdefghijklmnop
	2	qrstuvwxyzx
	3	xxxxxxxxxxxxx

doc/papers/concurrency/examples/Format.py

-              rb7d6a36
+              r6a490b2
                         for g in range( 5 ):    # groups of 5 blocks
                                 for b in range( 4 ): # blocks of 4 characters
+                                        print( (yield), end='' ) # receive from send
+                                        while True:
+                                                ch = (yield) # receive from send
+                                                if '\n' not in ch:
+                                                        break
+                                        print( ch, end='' ) # receive from send
                                 print( '  ', end='' ) # block separator
                         print()                                 # group separator
 …
                         print()
+input = "abcdefghijklmnop\nqrstuvwx\nyzxxxxxxxxxxxxxx\n"
 fmt = Format()
 next( fmt )                                                     # prime generator
 for i in range( 41 ):
         fmt.send( 'a' );                                # send to yield
+for i in input:
+        fmt.send( i );                          # send to yield
 # Local Variables: #
 # tab-width: 4 #
 # compile-command: "python3.5 Format.py" #
+# compile-command: "python3.7 Format.py" #
 # End: #

doc/papers/concurrency/examples/Format1.c

-              rb7d6a36
+              r6a490b2
 typedef struct {
         void * next;
+        int restart, g, b;
         char ch;
-        int g, b;
 } Fmt;
 void format( Fmt * f ) {
+        if ( __builtin_expect(f->next != 0, 1) ) goto *f->next;
+        f->next = &&s1;
+        static void * states[] = {&&s0, &&s1};
+        goto *states[f->restart];
+  s0: f->restart = 1;
         for ( ;; ) {
                 for ( f->g = 0; f->g < 5; f->g += 1 ) {                 // groups
                         for ( f->b = 0; f->b < 4; f->b += 1 ) {         // blocks
                                 return;
+                          s1: ;
+                                if ( f->ch == '\0' ) goto fini;                 // EOF ?
+                          s1: if ( f->ch == '\0' ) goto fini;           // EOF ?
                                 while ( f->ch == '\n' ) return;                 // ignore
                                 printf( "%c", f->ch );                                  // print character
+//                              printf( "%c", f->ch );                                  // print character
+                        }
                         printf( " " );                                                          // block separator
+//                      printf( " " );                                                          // block separator
+                }
                 printf( "\n" );                                                                 // group separator
+//              printf( "\n" );                                                                 // group separator
+        }
   fini:
         if ( f->g != 0 || f->b != 0 ) printf( "\n" );
+  fini:;
+//      if ( f->g != 0 || f->b != 0 ) printf( "\n" );
+}
 int main() {
         Fmt fmt = { NULL };
+        Fmt fmt = { 0 };
         format( &fmt );                                                                         // prime
+        for ( ;; ) {
+                scanf( "%c", &fmt.ch );                                                 // direct read into communication variable
+          if ( feof( stdin ) ) break;
+        fmt.ch = 'a';
+        for ( long int i = 0; i < 1000000000; i += 1 ) {
+//              scanf( "%c", &fmt.ch );                                                 // direct read into communication variable
+//        if ( feof( stdin ) ) break;
                 format( &fmt );
+        }
         fmt.ch = '\0';
+        fmt.ch = '\0';                                                                          // sentential (EOF)
         format( &fmt );
+}

doc/papers/concurrency/examples/PingPong.c

-              rb7d6a36
+              r6a490b2
 typedef struct PingPong {
+        int restart;                                                                            // style 1
+        int N, i;
         const char * name;
-        int N, i;
         struct PingPong * partner;
         void * next;
+        void * next;                                                                            // style 2
 } PingPong;
+#define PPCtor( name, N ) { name, N, 0, NULL, NULL }
+#define PPCtor( name, N ) { 0, N, 0, name, NULL, NULL }
 void comain( PingPong * pp ) __attribute__(( noinline ));
 void comain( PingPong * pp ) {
+#if 0
         if ( __builtin_expect(pp->next != 0, 1) ) goto *pp->next;
-#if 0
-        pp->next = &&here;
-                asm( "mov  %0,%%rdi" : "=m" (pp) );
-                asm( "mov  %rdi,%rax" );
-#ifndef OPT
-#ifdef PRINT
-                asm( "add  $16, %rsp" );
-#endif // PRINT
-                asm( "popq %rbp" );
-#endif // ! OPT
-#ifdef OPT
-#ifdef PRINT
-                asm( "popq %rbx" );
-#endif // PRINT
-#endif // OPT
-                asm( "jmp  comain" );
-  here: ;
-#endif // 0
         pp->next = &&cycle;
         for ( ; pp->i < pp->N; pp->i += 1 ) {
 …
           cycle: ;
         } // for
+#endif // 0
+#if 1
+        static void * states[] = {&&s0, &&s1};
+        goto *states[pp->restart];
+  s0: pp->restart = 1;
+        for ( ; pp->i < pp->N; pp->i += 1 ) {
+#ifdef PRINT
+                printf( "%s %d\n", pp->name, pp->i );
+#endif // PRINT
+                asm( "mov  %0,%%rdi" : "=m" (pp->partner) );
+                asm( "mov  %rdi,%rax" );
+#ifndef OPT
+#ifdef PRINT
+                asm( "add  $16, %rsp" );
+#endif // PRINT
+                asm( "popq %rbp" );
+#endif // ! OPT
+#ifdef OPT
+#ifdef PRINT
+                asm( "popq %rbx" );
+#endif // PRINT
+#endif // OPT
+                asm( "jmp  comain" );
+          s1: ;
+        } // for
+#endif // 0
+}
 …
 // Local Variables: //
 // tab-width: 4 //
 // compile-command: "gcc-8 -g -DPRINT PingPong.c" //
+// compile-command: "gcc-9 -g -DPRINT PingPong.c" //
 // End: //

doc/papers/concurrency/examples/Pingpong.py

-              rb7d6a36
+              r6a490b2
 def PingPong( name, N ):
         partner = (yield)           # get partner
         yield                       # resume scheduler
+        partner = yield                         # get partner
+        yield                                           # resume scheduler
         for i in range( N ):
                 print( name )
                 yield partner           # execute next
+                yield partner                   # execute next
         print( "end", name )
 def Scheduler():
+        n = (yield)                 # starting coroutine
+        while True:
+                n = next( n )           # schedule coroutine
+        n = yield                                       # starting coroutine
+        try:
+                while True:
+                        n = next( n )           # schedule coroutine
+        except StopIteration:
+                pass
 pi = PingPong( "ping", 5 )
 po = PingPong( "pong", 5 )
 next( pi )                      # prime
 pi.send( po )                   # send partner
 next( po )                      # prime
 po.send( pi )                   # send partner
+next( pi )                                              # prime
+pi.send( po )                                   # send partner
+next( po )                                              # prime
+po.send( pi )                                   # send partner
 s = Scheduler();
 next( s )                       # prime
+next( s )                                               # prime
 try:
         s.send( pi )                            # start cycle
 except StopIteration:
         print( "scheduler stop" )
+except StopIteration:                   # scheduler stopped
+        pass
 print( "stop" )
 # Local Variables: #
 # tab-width: 4 #
 # compile-command: "python3.5 Pingpong.py" #
+# compile-command: "python3.7 Pingpong.py" #
 # End: #

doc/papers/concurrency/examples/ProdCons.py

-              rb7d6a36
+              r6a490b2
 def Prod( N ):
         cons = (yield)              # get cons
         yield                       # resume scheduler
+        cons = yield                            # get cons
+        yield                                           # resume scheduler
         for i in range( N ):
                 print( "prod" )
                 yield cons              # execute next
+                yield cons                              # execute next
         print( "end", "prod" )
 def Cons( N ):
         prod = (yield)              # get prod
         yield                       # resume scheduler
+        prod = yield                            # get prod
+        yield                                           # resume scheduler
         for i in range( N ):
                 print( "cons" )
                 yield prod              # execute next
+                yield prod                              # execute next
         print( "end", "cons" )
 def Scheduler():
+        n = (yield)                 # starting coroutine
+        while True:
+                n = next( n )           # schedule coroutine
+        n = yield                                       # starting coroutine
+        try:
+                while True:
+                        n = next( n )           # schedule coroutine
+        except StopIteration:
+                pass
 prod = Prod( 5 )
 cons = Cons( 5 )
 next( prod )                    # prime
 prod.send( cons )               # send cons
 next( cons )                    # prime
 cons.send( prod )               # send prod
+next( prod )                                    # prime
+prod.send( cons )                               # send cons
+next( cons )                                    # prime
+cons.send( prod )                               # send prod
 s = Scheduler();
 next( s )                       # prime
+next( s )                                               # prime
 try:
         s.send( prod )                          # start cycle
 except StopIteration:
         print( "scheduler stop" )
+except StopIteration:                   # scheduler stopped
+        pass
 print( "stop" )
 # Local Variables: #
 # tab-width: 4 #
 # compile-command: "python3.5 ProdCons.py" #
+# compile-command: "python3.7 ProdCons.py" #
 # End: #

doc/papers/concurrency/examples/RWMonitorEXT.cfa

-              rb7d6a36
+              r6a490b2
         int rcnt, wcnt;                                                                         // number of readers/writer using resource
 };
+void ?{}( ReadersWriter & rw ) with(rw) { rcnt = wcnt = 0; }
 void EndRead( ReadersWriter & mutex rw ) with(rw) { rcnt -= 1; }
 void EndWrite( ReadersWriter & mutex rw ) with(rw) { wcnt = 0; }
 void StartRead( ReadersWriter & mutex rw ) with(rw) {
         if ( wcnt > 0 ) waitfor( EndWrite, rw );
+        if ( wcnt > 0 ) waitfor( EndWrite : rw );
         rcnt += 1;
+}
 void StartWrite( ReadersWriter & mutex rw ) with(rw) {
         if ( wcnt > 0 ) waitfor( EndWrite, rw );
         else while ( rcnt > 0 ) waitfor( EndRead, rw );
+        if ( wcnt > 0 ) waitfor( EndWrite : rw );
+        else while ( rcnt > 0 ) waitfor( EndRead : rw );
         wcnt = 1;
+}
-void ?{}( ReadersWriter & rw ) with(rw) { rcnt = wcnt = 0; }
 int readers( ReadersWriter & rw ) { return rw.rcnt; }
 void Read( ReadersWriter & rw ) {
         StartRead( rw );
 …
         EndWrite( rw );
+}
 thread Worker {
         ReadersWriter &rw;
 …
         } // for
+}
 int main() {
         enum { MaxTask = 5 };
 …
 } // main
 // Local Variables: //
 // tab-width: 4 //
 // compile-command: "cfa -O2 RWMonitor.cfa" //
+// compile-command: "cfa -O2 RWMonitorEXT.cfa" //
 // End: //

doc/papers/concurrency/examples/Refactor.py

rb7d6a36	r6a490b2
26	26	# Local Variables: #
27	27	# tab-width: 4 #
28		# compile-command: "python3.5 Refactor.py" #
	28	# compile-command: "python3.7 Refactor.py" #
29	29	# End: #

doc/papers/concurrency/figures/FullCoroutinePhases.fig

-              rb7d6a36
+              r6a490b2
 -2
 2
 1 0 1 0 7 100 0 -1 0.000 0 0 1 0 4575.000 2437.500 4275 1875 4575 1800 4875 1875
+1 0 1 0 7 100 0 -1 0.000 0 0 1 0 5175.000 2437.500 4875 1875 5175 1800 5475 1875
 1 1.00 45.00 90.00
 1 0 1 0 7 100 0 -1 0.000 0 0 1 0 4575.000 1537.500 4875 2100 4575 2175 4275 2100
+1 0 1 0 7 100 0 -1 0.000 0 0 1 0 5175.000 1537.500 5475 2100 5175 2175 4875 2100
 1 1.00 45.00 90.00
 1 0 1 0 7 50 -1 -1 0.000 0 1 1 0 4207.500 1642.500 4125 1425 3975 1650 4200 1875
+1 0 1 0 7 50 -1 -1 0.000 0 1 1 0 4807.500 1642.500 4725 1425 4575 1650 4800 1875
 1 1.00 45.00 90.00
+1575 1575 2700 2025
 1 0 1 0 7 100 0 -1 0.000 0 0 -1 1 0 2
 1 1.00 45.00 90.00
 …
 1 1.00 45.00 90.00
 1575 2400 1800
+1 0 100 0 4 10 0.0000 2 165 300 1725 1950 ping\001
+1 0 100 0 4 10 0.0000 2 135 360 2475 1950 pong\001
+-6
+3075 1575 4200 2025
+3075 1575 4200 2025
 1 0 1 0 7 100 0 -1 0.000 0 0 -1 1 0 2
 1 1.00 45.00 90.00
 1575 3300 1800
+1575 3300 1800
 1 0 1 0 7 100 0 -1 0.000 0 0 -1 1 0 2
 1 1.00 45.00 90.00
+2025 3300 2250
+1 0 100 0 0 10 0.0000 2 105 555 2100 1200 creation\001
+1 0 100 0 4 10 0.0000 2 165 300 1725 1950 ping\001
+1 0 100 0 4 10 0.0000 2 135 360 2475 1950 pong\001
+1 0 100 0 4 10 0.0000 2 165 300 3300 1950 ping\001
+1 0 100 0 4 10 0.0000 2 135 360 3300 2400 pong\001
+1 0 100 0 0 10 0.0000 2 105 675 4575 1200 execution\001
+1 0 100 0 4 10 0.0000 2 165 300 4275 2025 ping\001
+1 0 100 0 4 10 0.0000 2 135 360 4875 2025 pong\001
+1 0 100 0 0 10 0.0000 2 90 420 3300 1200 starter\001
+1575 3900 1800
+1 0 100 0 4 10 0.0000 2 165 300 3225 1950 ping\001
+1 0 100 0 4 10 0.0000 2 135 360 3975 1950 pong\001
+-6
+-6
 1 0 100 0 4 10 0.0000 2 165 705 2100 1500 pgm main\001
+1 0 100 0 4 10 0.0000 2 165 705 3300 1500 pgm main\001
+1 0 100 0 4 10 0.0000 2 165 705 4500 1500 pgm main\001
+1 0 100 0 4 10 0.0000 2 165 705 3600 1500 pgm main\001
+1 0 100 0 4 10 0.0000 2 165 300 4875 2025 ping\001
+1 0 100 0 4 10 0.0000 2 135 360 5475 2025 pong\001
+1 0 100 0 4 10 0.0000 2 165 705 5100 1500 pgm main\001
+1 0 100 0 2 10 0.0000 2 105 540 2100 1275 creator\001
+1 0 100 0 2 10 0.0000 2 105 495 3600 1275 starter\001
+1 0 100 0 2 10 0.0000 2 105 690 5175 1275 execution\001

doc/papers/concurrency/figures/RunTimeStructure.fig

-              rb7d6a36
+              r6a490b2
 3 0 1 -1 -1 0 0 20 0.000 1 0.0000 4500 3600 15 15 4500 3600 4515 3615
 -6
+2175 4650 7050 4950
+3 0 1 0 0 0 0 0 0.000 1 0.0000 2250 4830 30 30 2250 4830 2280 4860
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4200 4800 150 75 4200 4800 4350 4875
+3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3275 4800 100 100 3275 4800 3375 4800
+3225 4125 4650 4425
+4350 4200 4650 4350
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 4425 4275 15 15 4425 4275 4440 4290
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 4500 4275 15 15 4500 4275 4515 4290
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 4575 4275 15 15 4575 4275 4590 4290
+-6
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3450 4275 225 150 3450 4275 3675 4425
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4050 4275 225 150 4050 4275 4275 4425
+-6
+6675 4125 7500 4425
+7200 4200 7500 4350
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 7275 4275 15 15 7275 4275 7290 4290
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 7350 4275 15 15 7350 4275 7365 4290
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 7425 4275 15 15 7425 4275 7440 4290
+-6
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 6900 4275 225 150 6900 4275 7125 4425
+-6
+6675 3525 8025 3975
+1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
+1 1.00 45.00 90.00
+3750 6975 3750
+1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
+1 1.00 45.00 90.00
+3750 7350 3750
 2 0 1 -1 -1 0 0 -1 0.000 0 0 0 0 0 5
+4950 5400 4725 5175 4725 5175 4950 5400 4950
+2 1 1 -1 -1 0 0 -1 3.000 0 0 0 0 0 5
+4950 6300 4950 6300 4725 6525 4725 6525 4950
+0 -1 0 0 0 10 0.0000 2 105 450 6600 4875 cluster\001
+0 -1 0 0 0 10 0.0000 2 105 660 5475 4875 processor\001
+0 -1 0 0 0 10 0.0000 2 105 555 4425 4875 monitor\001
+0 -1 0 0 0 10 0.0000 2 120 270 3450 4875 task\001
+0 -1 0 0 0 10 0.0000 2 105 660 2325 4875 coroutine\001
+-6
+3450 1275 3750 1425
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 3525 1350 15 15 3525 1350 3540 1365
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 3600 1350 15 15 3600 1350 3615 1365
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 3675 1350 15 15 3675 1350 3690 1365
+-6
+5550 1275 5850 1425
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 5625 1350 15 15 5625 1350 5640 1365
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 5700 1350 15 15 5700 1350 5715 1365
+3 0 1 -1 -1 0 0 20 0.000 1 0.0000 5775 1350 15 15 5775 1350 5790 1365
+3975 7800 3525 7350 3525 7350 3975 7800 3975
+1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
+1 1.00 45.00 90.00
+3750 8025 3750
 -6
 3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 5550 2625 150 150 5550 2625 5700 2625
 …
 3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4425 2850 150 150 4425 2850 4575 2850
 3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4650 2475 150 150 4650 2475 4800 2475
-3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3525 3600 150 150 3525 3600 3675 3600
 3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3975 3600 150 150 3975 3600 4125 3600
 3 0 1 0 0 0 0 0 0.000 1 0.0000 3525 3600 30 30 3525 3600 3555 3630
 …
 3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3975 2850 150 150 3975 2850 4125 2850
 3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 7200 2775 150 150 7200 2775 7350 2775
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4650 1350 225 150 4650 1350 4875 1500
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 5250 1350 225 150 5250 1350 5475 1500
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4050 1350 225 150 4050 1350 4275 1500
+3 0 1 0 0 0 0 0 0.000 1 0.0000 2250 4830 30 30 2250 4830 2280 4860
+3 0 1 0 0 0 0 0 0.000 1 0.0000 7200 2775 30 30 7200 2775 7230 2805
+3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3525 3600 150 150 3525 3600 3675 3600
+3 0 1 -1 -1 0 0 -1 0.000 1 0.0000 3875 4800 100 100 3875 4800 3975 4800
+1 0 1 -1 -1 0 0 -1 0.000 1 0.0000 4650 4800 150 75 4650 4800 4800 4875
 2 0 1 -1 -1 0 0 -1 0.000 0 0 0 0 0 5
 4200 2400 3750 1950 3750 1950 4200 2400 4200
 …
 1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
 1 1.00 45.00 90.00
-3975 6975 3975
-1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
-1 1.00 45.00 90.00
 2775 6825 2775
 1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 0 0 2
+2775 6825 3975
+1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
+1 1.00 45.00 90.00
+3975 7350 3975
+2 0 1 -1 -1 0 0 -1 0.000 0 0 0 0 0 5
+4200 7800 3750 7350 3750 7350 4200 7800 4200
+1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 2
+1 1.00 45.00 90.00
+3975 8025 3975
+2775 6825 3750
 1 0 1 -1 -1 0 0 -1 0.000 0 0 -1 1 0 4
 1 1.00 45.00 90.00
+3975 7875 2325 7200 2325 7200 2550
+3750 7875 2325 7200 2325 7200 2550
+2 0 1 -1 -1 0 0 -1 0.000 0 0 0 0 0 5
+4950 5850 4725 5625 4725 5625 4950 5850 4950
+2 1 1 -1 -1 0 0 -1 3.000 0 0 0 0 0 5
+4950 6750 4950 6750 4725 6975 4725 6975 4950
 1 -1 0 0 0 10 0.0000 2 105 720 5550 4425 Processors\001
 1 -1 0 0 0 10 0.0000 2 120 1005 4200 3225 Blocked Tasks\001
 …
 1 -1 0 0 0 10 0.0000 2 105 990 2175 3525 Discrete-event\001
 1 -1 0 0 0 10 0.0000 2 135 795 2175 4350 preemption\001
+0 -1 0 0 0 10 0.0000 2 150 1290 2325 4875 genrator/coroutine\001
+0 -1 0 0 0 10 0.0000 2 120 270 4050 4875 task\001
+0 -1 0 0 0 10 0.0000 2 105 450 7050 4875 cluster\001
+0 -1 0 0 0 10 0.0000 2 105 660 5925 4875 processor\001
+0 -1 0 0 0 10 0.0000 2 105 555 4875 4875 monitor\001

doc/papers/concurrency/mail2

-              rb7d6a36
+              r6a490b2
 Software: Practice and Experience Editorial Office
+Date: Tue, 12 Nov 2019 22:25:17 +0000
+From: Richard Jones <onbehalfof@manuscriptcentral.com>
+Reply-To: R.E.Jones@kent.ac.uk
+To: tdelisle@uwaterloo.ca, pabuhr@uwaterloo.ca
+Subject: Software: Practice and Experience - Decision on Manuscript ID
+ SPE-19-0219
+-Nov-2019
+Dear Dr Buhr,
+Many thanks for submitting SPE-19-0219 entitled "Advanced Control-flow and Concurrency in Cforall" to Software: Practice and Experience. The paper has now been reviewed and the comments of the referees are included at the bottom of this letter.
+The decision on this paper is that it requires substantial further work is required. The referees have a number of substantial concerns. All the reviewers found the submission very hard to read; two of the reviewers state that it needs very substantial restructuring. These concerns must be addressed before your submission can be considered further.
+A revised version of your manuscript that takes into account the comments of the referees will be reconsidered for publication.
+Please note that submitting a revision of your manuscript does not guarantee eventual acceptance, and that your revision will be subject to re-review by the referees before a decision is rendered.
+You have 90 days from the date of this email to submit your revision. If you are unable to complete the revision within this time, please contact me to request an extension.
+You can upload your revised manuscript and submit it through your Author Center. Log into https://mc.manuscriptcentral.com/spe  and enter your Author Center, where you will find your manuscript title listed under "Manuscripts with Decisions".
+When submitting your revised manuscript, you will be able to respond to the comments made by the referee(s) in the space provided.  You can use this space to document any changes you make to the original manuscript.
+If you feel that your paper could benefit from English language polishing, you may wish to consider having your paper professionally edited for English language by a service such as Wiley's at http://wileyeditingservices.com. Please note that while this service will greatly improve the readability of your paper, it does not guarantee acceptance of your paper by the journal.
+Once again, thank you for submitting your manuscript to Software: Practice and Experience and I look forward to receiving your revision.
+Sincerely,
+Prof. Richard Jones
+Software: Practice and Experience
+R.E.Jones@kent.ac.uk
+Referee(s)' Comments to Author:
+Reviewing: 1
+Comments to the Author
+This article presents the design and rationale behind the various
+threading and synchronization mechanisms of C-forall, a new low-level
+programming language.  This paper is very similar to a companion paper
+which I have also received: as the papers are similar, so will these
+reviews be --- in particular any general comments from the other
+review apply to this paper also.
+As far as I can tell, the article contains three main ideas: an
+asynchronous execution / threading model; a model for monitors to
+provide mutual exclusion; and an implementation.  The first two ideas
+are drawn together in Table 1: unfortunately this is on page 25 of 30
+pages of text. Implementation choices and descriptions are scattered
+throughout the paper - and the sectioning of the paper seems almost
+arbitrary.
+The article is about its contributions.  Simply adding feature X to
+language Y isn't by itself a contribution, (when feature X isn't
+already a contribution).  The contribution can be in the design: the
+motivation, the space of potential design options, the particular
+design chosen and the rationale for that choice, or the resulting
+performance.  For example: why support two kinds of generators as well
+as user-level threads?  Why support both low and high level
+synchronization constructs?  Similarly I would have found the article
+easier to follow if it was written top down, presenting the design
+principles, present the space of language features, justify chosen
+language features (and rationale) and those excluded, and then present
+implementation, and performance.
+Then the writing of the article is often hard to follow, to say the
+least. Two examples: section 3 "stateful functions" - I've some idea
+what that is (a function with Algol's "own" or C's "static" variables?
+but in fact the paper has a rather more specific idea than that. The
+top of page 3 throws a whole lot of defintions at the reader
+"generator" "coroutine" "stackful" "stackless" "symmetric"
+"asymmetric" without every stopping to define each one --- but then in
+footnote "C" takes the time to explain what C's "main" function is?  I
+cannot imagine a reader of this paper who doesn't know what "main" is
+in C; especially if they understand the other concepts already
+presented in the paper.  The start of section 3 then does the same
+thing: putting up a whole lot of definitions, making distinctions and
+comparisons, even talking about some runtime details, but the critical
+definition of a monitor doesn't appear until three pages later, at the
+start of section 5 on p15, lines 29-34 are a good, clear, description
+of what a monitor actually is.  That needs to come first, rather than
+being buried again after two sections of comparisons, discussions,
+implementations, and options that are ungrounded because they haven't
+told the reader what they are actually talking about.  First tell the
+reader what something is, then how they might use it (as programmers:
+what are the rules and restrictions) and only then start comparison
+with other things, other approaches, other languages, or
+implementations.
+The description of the implementation is similarly lost in the trees
+without ever really seeing the wood. Figure 19 is crucial here, but
+it's pretty much at the end of the paper, and comments about
+implementations are threaded throughout the paper without the context
+(fig 19) to understand what's going on.   The protocol for performance
+testing may just about suffice for C (although is N constantly ten
+million, or does it vary for each benchmark) but such evaluation isn't
+appropriate for garbage-collected or JITTed languages like Java or Go.
+other comments working through the paper - these are mostly low level
+and are certainly not comprehensive.
+p1 only a subset of C-forall extensions?
+p1 "has features often associated with object-oriented programming
+languages, such as constructors, destructors, virtuals and simple
+inheritance."   There's no need to quibble about this. Once a language
+has inheritance, it's hard to claim it's not object-oriented.
+p2 barging? signals-as-hints?
+p3 start your discussion of generations with a simple example of a
+C-forall generator.  Fig 1(b) might do: but put it inline instead of
+the python example - and explain the key rules and restrictions on the
+construct.  Then don't even start to compare with coroutines until
+you've presented, described and explained your coroutines...
+p3 I'd probably leave out the various "C" versions unless there are
+key points to make you can't make in C-forall. All the alternatives
+are just confusing.
+p4 but what's that "with" in Fig 1(B)
+p5 start with the high level features of C-forall generators...
+p5 why is the paper explaining networking protocols?
+p7 lines 1-9 (transforming generator to coroutine - why would I do any
+of this? Why would I want one instead of the other (do not use "stack"
+in your answer!)
+p10 last para "A coroutine must retain its last resumer to suspend
+back because the resumer is on a different stack. These reverse
+pointers allow suspend to cycle backwards, "  I've no idea what is
+going on here?  why should I care?  Shouldn't I just be using threads
+instead?  why not?
+p16 for the same reasons - what reasons?
+p17 if the multiple-monitor entry procedure really is novel, write a
+paper about that, and only about that.
+p23 "Loose Object Definitions" - no idea what that means.  in that
+section: you can't leave out JS-style dynamic properties.  Even in
+OOLs that (one way or another) allow separate definitions of methods
+(like Objective-C, Swift, Ruby, C#) at any time a runtime class has a
+fixed definition.  Quite why the detail about bit mask implementation
+is here anyway, I've no idea.
+p25 this cluster isn't a CLU cluster then?
+* conclusion should conclude the paper, not the related.
+Reviewing: 2
+Comments to the Author
+This paper describes the concurrency features of an extension of C (whose name I will write as "C\/" here, for convenience), including much design-level discussion of the coroutine- and monitor-based features and some microbenchmarks exploring the current implementation's performance. The key message of the latter is that the system's concurrency abstractions are much lighter-weight than the threading found in mainstream C or Java implementations.
+There is much description of the system and its details, but nothing about (non-artificial) uses of it. Although the microbenchmark data is encouraging, arguably not enough practical experience with the system has been reported here to say much about either its usability advantages or its performance.
+As such, the main contribution of the paper seem to be to document the existence of the described system and to provide a detailed design rationale and (partial) tutorial. I believe that could be of interest to some readers, so an acceptable manuscript is lurking in here somewhere.
+Unfortunately, at present the writing style is somewhere between unclear and infuriating. It omits to define terms; it uses needlessly many terms for what are apparently (but not clearly) the same things; it interrupts itself rather than deliver the natural consequent of whatever it has just said; and so on. Section 5 is particularly bad in these regards -- see my detailed comments below. Fairly major additional efforts will be needed to turn the present text into a digestible design-and-tutorial document. I suspect that a shorter paper could do this job better than the present manuscript, which is overwrought in parts.
+p2: lines 4--9 are a little sloppy. It is not the languages but their popular implementations which "adopt" the 1:1 kernel threading model.
+line 10: "medium work" -- "medium-sized work"?
+line 18: "is all sequential to the compiler" -- not true in modern compilers, and in 2004 H-J Boehm wrote a tech report describing exactly why ("Threads cannot be implemented as a library", HP Labs).
+line 20: "knows the optimization boundaries" -- I found this vague. What's an example?
+line 31: this paragraph has made a lot of claims. Perhaps forward-reference to the parts of the paper that discuss each one.
+line 33: "so the reader can judge if" -- this reads rather passive-aggressively. Perhaps better: "... to support our argument that..."
+line 41: "a dynamic partitioning mechanism" -- I couldn't tell what this meant
+p3. Presenting concept of a "stateful function" as a new language feature seems odd. In C, functions often have local state thanks to static local variables (or globals, indeed). Of course, that has several limitations. Can you perhaps present your contributions by enumerating these limitations? See also my suggestion below about a possible framing centred on a strawman.
+line 2: "an old idea that is new again" -- this is too oblique
+lines 2--15: I found this to be a word/concept soup. Stacks, closures, generators, stackless stackful, coroutine, symmetric, asymmetric, resume/suspend versus resume/resume... there needs to be a more gradual and structured way to introduce all this, and ideally one that minimises redundancy. Maybe present it as a series of "definitions" each with its own heading, e.g. "A closure is stackless if its local state has statically known fixed size"; "A generator simply means a stackless closure." And so on. Perhaps also strongly introduce the word "activate" as a direct contrast with resume and suspend. These are just a flavour of the sort of changes that might make this paragraph into something readable.
+Continuing the thought: I found it confusing that by these definitinos, a stackful closure is not a stack, even though logically the stack *is* a kind of closure (it is a representation of the current thread's continuation).
+lines 24--27: without explaining what the boost functor types mean, I don't think the point here comes across.
+line 34: "semantically coupled" -- I wasn't surew hat this meant
+p4: the point of Figure 1 (C) was not immediately clear. It seem to be showing how one might "compile down" Figure 1 (B). Or is that Figure 1 (A)?
+It's right that the incidental language features of the system are not front-and-centre, but I'd appreciate some brief glossing of non-C languages features as they appear. Examples are the square bracket notation, the pipe notation and the constructor syntax. These explanations could go in the caption of the figure which first uses them, perhaps. Overall I found the figure captions to be terse, and a missed opportunity to explain clearly what was going on.
+p5 line 23: "This restriction is removed..." -- give us some up-front summary of your contributions and the elements of the language design that will be talked about, so that this isn't an aside. This will reduce the "twisty passages" feeling that characterises much of the paper.
+line 40: "a killer asymmetric generator" -- this is stylistically odd, and the sentence about failures doesn't convincigly argue that C\/ will help with them. Have you any experience writing device drivers using C\/? Or any argument that the kinds of failures can be traced to the "stack-ripping" style that one is forced to use without coroutines? Also, a typo on line 41: "device drives". And saying "Windows/Linux" is sloppy... what does the cited paper actually say?
+p6 lines 13--23: this paragraph is difficult to understand. It seems to be talking about a control-flow pattern roughly equivalent to tail recursion. What is the high-level point, other than that this is possible?
+line 34: "which they call coroutines" -- a better way to make this point is presumably that the C++20 proposal only provides a specialised kind of coroutine, namely generators, despite its use of the more general word.
+line 47: "... due to dynamic stack allocation, execution..." -- this sentence doesn't scan. I suggest adding "and for" in the relevant places where currently there are only commas.
+p8 / Figure 5 (B) -- the GNU C extension of unary "&&" needs to be explained. The whole figure needs a better explanation, in fact.
+p9, lines 1--10: I wasn't sure this stepping-through really added much value. What are the truly important points to note about this code?
+p10: similarly, lines 3--27 again are somewhere between tedious and confusing. I'm sure the motivation and details of "starter semantics" can both be stated much more pithily.
+line 32: "a self-resume does not overwrite the last resumer" -- is this a hack or a defensible principled decision?
+p11: "a common source of errors" -- among beginners or among production code? Presumably the former.
+line 23: "with builtin and library" -- not sure what this means
+lines 31--36: these can be much briefer. The only important point here seems to be that coroutines cannot be copied.
+p12: line 1: what is a "task"? Does it matter?
+line 7: calling it "heap stack" seems to be a recipe for confusion. "Stack-and-heap" might be better, and contrast with "stack-and-VLS" perhaps. When "VLS" is glossed, suggest actually expanding its initials: say "length" not "size".
+line 21: are you saying "cooperative threading" is the same as "non-preemptive scheduling", or that one is a special case (kind) of the other? Both are defensible, but be clear.
+line 27: "mutual exclusion and synchronization" -- the former is a kind of the latter, so I suggest "and other forms of synchronization".
+line 30: "can either be a stackless or stackful" -- stray "a", but also, this seems to be switching from generic/background terminology to C\/-specific terminology.
+An expositional idea occurs: start the paper with a strawman naive/limited realisation of coroutines -- say, Simon Tatham's popular "Coroutines in C" web page -- and identify point by point what the limitations are and how C\/ overcomes them. Currently the presentation is often flat (lacking motivating contrasts) and backwards (stating solutions before problems). The foregoing approach might fix both of these.
+page 13: line 23: it seems a distraction to mention the Python feature here.
+p14 line 5: it seems odd to describe these as "stateless" just because they lack shared mutable state. It means the code itself is even more stateful. Maybe the "stack ripping" argument could usefully be given here.
+line 16: "too restrictive" -- would be good to have a reference to justify this, or at least give a sense of what the state-of-the-art performance in transactional memory systems is (both software and hardware)
+line 22: "simulate monitors" -- what about just *implementing* monitors? isn't that what these systems do? or is the point more about refining them somehow into something more specialised?
+p15: sections 4.1 and 4.2 seem adrift and misplaced. Split them into basic parts (which go earlier) and more advanced parts (e.g. barging, which can be explained later).
+line 31: "acquire/release" -- misses an opportunity to contrast the monitor's "enter/exit" abstraction with the less structured acquire/release of locks.
+p16 line 12: the "implicit" versus "explicit" point is unclear. Is it perhaps about the contract between an opt-in *discipline* and a language-enforced *guarantee*?
+line 28: no need to spend ages dithering about which one is default and which one is the explicit qualifier. Tell us what you decided, briefly justify it, and move on.
+p17: Figure 11: since the main point seems to be to highlight bulk acquire, include a comment which identifies the line where this is happening.
+line 2: "impossible to statically..." -- or dynamically. Doing it dynamically would be perfectly acceptable (locking is a dynamic operation after all)
+"guarantees acquisition order is consistent" -- assuming it's done in a single bulk acquire.
+p18: section 5.3: the text here is a mess. The explanations of "internal" versus "external" scheduling are unclear, and "signals as hints" is not explained. "... can cause thread starvation" -- means including a while loop, or not doing so? "There are three signalling mechanisms.." but the text does not follow that by telling us what they are. My own scribbled attempt at unpicking the internal/external thing: "threads already in the monitor, albeit waiting, have priority over those trying to enter".
+p19: line 3: "empty condition" -- explain that condition variables don't store anything. So being "empty" means that the queue of waiting threads (threads waiting to be signalled that the condition has become true) is empty.
+line 6: "... can be transformed into external scheduling..." -- OK, but give some motivation.
+p20: line 6: "mechnaism"
+lines 16--20: this is dense and can probably only be made clear with an example
+p21 line 21: clarify that nested monitor deadlock was describe earlier (in 5.2). (Is the repetition necessary?)
+line 27: "locks, and by extension monitors" -- this is true but the "by extension" argument is faulty. It is perfectly possible to use locks as a primitive and build a compositional mechanism out of them, e.g. transactions.
+p22 line 2: should say "restructured"
+line 33: "Implementing a fast subset check..." -- make clear that the following section explains how to do this. Restructuring the sections themselves could do this, or noting in the text.
+p23: line 3: "dynamic member adding, eg, JavaScript" -- needs to say "as permitted in JavaScript", and "dynamically adding members" is stylistically better
+p23: line 18: "urgent stack" -- back-reference to where this was explained before
+p24 line 7: I did not understand what was more "direct" about "direct communication". Also, what is a "passive monitor" -- just a monitor, given that monitors are passive by design?
+line 14 / section 5.9: this table was useful and it (or something like it) could be used much earlier on to set the structure of the rest of the paper. The explanation at present is too brief, e.g. I did not really understand the point about cases 7 and 8.
+p25 line 2: instead of casually dropping in a terse explanation for the newly intrdouced term "virtual processor", introduce it properly. Presumably the point is to give a less ambiguous meaning to "thread" by reserving it only for C\/'s green threads.
+Table 1: what does "No / Yes" mean?
+p26 line 15: "transforms user threads into fibres" -- a reference is needed to explain what "fibres" means... guessing it's in the sense of Adya et al.
+line 20: "Microsoft runtime" -- means Windows?
+lines 21--26: don't say "interrupt" to mean "signal", especially not without clear introduction. You can use "POSIX signal" to disambiguate from condition variables' "signal".
+p27 line 3: "frequency is usually long" -- that's a "time period" or "interval", not a frequency
+line 5: the lengthy quotation is not really necessary; just paraphrase the first sentence and move on.
+line 20: "to verify the implementation" -- I don't think that means what is intended
+Tables in section 7 -- too many significant figures. How many overall runs are described? What is N in each case?
+p29 line 2: "to eliminate this cost" -- arguably confusing since nowadays on commodity CPUs most of the benefits of inlining are not to do with call overheads, but from later optimizations enabled as a consequence of the inlining
+line 41: "a hierarchy" -- are they a hierarchy? If so, this could be explained earlier. Also, to say these make up "an integrated set... of control-flow features" verges on the tautologous.
+p30 line 15: "a common case being web servers and XaaS" -- that's two cases
+Reviewing: 3
+Comments to the Author
+# Cforall review
+Overall, I quite enjoyed reading the paper. Cforall has some very interesting ideas. I did have some suggestions that I think would be helpful before final publication. I also left notes on various parts of the paper that I find confusing when reading, in hopes that it may be useful to you.
+## Summary
+* Expand on the motivations for including both generator and coroutines, vs trying to build one atop the other
+* Expand on the motivations for having Why both symmetric and asymettric coroutines?
+* Comparison to async-await model adopted by other languages
+    * C#, JS
+    * Rust and its async/await model
+* Consider performance comparisons against node.js and Rust frameworks
+* Discuss performance of monitors vs finer-grained memory models and atomic operations found in other languages
+* Why both internal/external scheduling for synchronization?
+## Generator/coroutines
+In general, this section was clear, but I thought it would be useful to provide a somewhat deeper look into why Cforall opted for the particular combination of features that it offers. I see three main differences from other languages:
+* Generators are not exposed as a "function" that returns a generator object, but rather as a kind of struct, with communication happening via mutable state instead of "return values". That is, the generator must be manually resumed and (if I understood) it is expected to store values that can then later be read (perhaps via methods), instead of having a `yield <Expr>` statement that yields up a value explicitly.
+* Both "symmetric" and "asymmetric" generators are supported, instead of only asymmetric.
+* Coroutines (multi-frame generators) are an explicit mechanism.
+In most other languages, coroutines are rather built by layering single-frame generators atop one another (e.g., using a mechanism like async-await), and symmetric coroutines are basically not supported. I'd like to see a bit more justification for Cforall including all the above mechanisms -- it seemed like symmetric coroutines were a useful building block for some of the user-space threading and custom scheduler mechanisms that were briefly mentioned later in the paper.
+In the discussion of coroutines, I would have expected a bit more of a comparison to the async-await mechanism offered in other languages. Certainly the semantics of async-await in JavaScript implies significantly more overhead (because each async fn is a distinct heap object). [Rust's approach avoids this overhead][zc], however, and might be worthy of a comparison (see the Performance section).
+## Locks and threading
+### Comparison to atomics overlooks performance
+There are several sections in the paper that compare against atomics -- for example, on page 15, the paper shows a simple monitor that encapsulates an integer and compares that to C++ atomics. Later, the paper compares the simplicity of monitors against the `volatile` quantifier from Java. The conclusion in section 8 also revisits this point.
+While I agree that monitors are simpler, they are obviously also significantly different from a performance perspective -- the paper doesn't seem to address this at all. It's plausible that (e.g.) the `Aint` monitor type described in the paper can be compiled and mapped to the specialized instructions offered by hardware, but I didn't see any mention of how this would be done. There is also no mention of the more nuanced memory ordering relations offered by C++11 and how one might achieve similar performance characteristics in Cforall (perhaps the answer is that one simply doesn't need to; I think that's defensible, but worth stating explicitly).
+### Justification for external scheduling feels lacking
+Cforall includes both internal and external scheduling; I found the explanation for the external scheduling mechanism to be lacking in justification. Why include both mechanisms when most languages seem to make do with only internal scheduling? It would be useful to show some scenarios where external scheduling is truly more powerful.
+I would have liked to see some more discussion of external scheduling and how it  interacts with software engineering best practices. It seems somewhat similar to AOP in certain regards. It seems to add a bit of "extra semantics" to monitor methods, in that any method may now also become a kind of synchronization point. The "open-ended" nature of this feels like it could easily lead to subtle bugs, particularly when code refactoring occurs (which may e.g. split an existing method into two). This seems particularly true if external scheduling can occur across compilation units -- the paper suggested that this is true, but I wasn't entirely clear.
+I would have also appreciated a few more details on how external scheduling is implemented. It seems to me that there must be some sort of "hooks" on mutex methods so that they can detect whether some other function is waiting on them and awaken those blocked threads. I'm not sure how such hooks are inserted, particularly across compilation units. The material in Section 5.6 didn't quite clarify the matter for me. For example, it left me somewhat confused about whether the `f` and `g` functions declared were meant to be local to a translation unit, or shared with other unit.
+### Presentation of monitors is somewhat confusing
+I found myself confused fairly often in the section on monitors. I'm just going to leave some notes here on places that I got confused in how that it could be useful to you as feedback on writing that might want to be clarified.
+To start, I did not realize that the `mutex_opt` notation was a keyword, I thought it was a type annotation. I think this could be called out more explicitly.
+Later, in section 5.2, the paper discusses `nomutex` annotations, which initially threw me, as they had not been introduced (now I realize that this paragraph is there to justify why there is no such keyword). The paragraph might be rearranged to make that clearer, perhaps by leading with the choice that Cforall made.
+On page 17, the paper states that "acquiring multiple monitors is safe from deadlock", but this could be stated a bit more precisely: acquiring multiple monitors in a bulk-acquire is safe from deadlock (deadlock can still result from nested acquires).
+On page 18, the paper states that wait states do not have to be enclosed in loops, as there is no concern of barging. This seems true but there are also other reasons to use loops (e.g., if there are multiple reasons to notify on the same condition). Thus the statement initially surprised me, as barging is only one of many reasons that I typically employ loops around waits.
+I did not understand the diagram in Figure 12 for some time. Initially, I thought that it was generic to all monitors, and I could not understand the state space. It was only later that I realized it was specific to your example. Updating the caption from "Monitor scheduling to "Monitor scheduling in the example from Fig 13" might have helped me quite a bit.
+I spent quite some time reading the boy/girl dating example (\*) and I admit I found it somewhat confusing. For example, I couldn't tell whether there were supposed to be many "girl" threads executing at once, or if there was only supposed to be one girl and one boy thread executing in a loop. Are the girl/boy threads supposed to invoke the girl/boy methods or vice versa? Surely there is some easier way to set this up? I believe that when reading the paper I convinced myself of how it was supposed to be working, but I'm writing this review some days later, and I find myself confused all over again and not able to easily figure it out.
+(\*) as an aside, I would consider modifying the example to some other form of matching, like customers and support personnel.
+## Related work
+The paper offered a number of comparisons to Go, C#, Scala, and so forth, but seems to have overlooked another recent language, Rust. In many ways, Rust seems to be closest in philosophy to Cforall, so it seems like an odd omission. I already mentioned above that Rust is in the process of shipping [async-await syntax][aa], which is definitely an alternative to the generator/coroutine approach in Cforall (though one with clear pros/cons).
+## Performance
+In the performance section in particular, you might consider comparing against some of the Rust web servers and threading systems. For example, actix is top of the [single query TechEmpower Framework benchmarks], and tokio is near the top of the [plainthreading benchmarks][pt] (hyper, the top, is more of an HTTP framework, though it is also written in Rust). It would seem worth trying to compare their "context switching" costs as well -- I believe both actix and tokio have a notion of threads that could be readily compared.
+Another addition that might be worth considering is to compare against node.js promises, although I think the comparison to process creation is not as clean.
+That said, I think that the performance comparison is not a big focus of the paper, so it may not be necessary to add anything to it.
+## Authorship of this review
+I'm going to sign this review. This review was authored by Nicholas D. Matsakis. In the intrerest of full disclosure, I'm heavily involved in the Rust project, although I dont' think that influenced this review in particular. Feel free to reach out to me for clarifying questions.
+## Links
+[aa]: https://blog.rust-lang.org/2019/09/30/Async-await-hits-beta.html
+[zc]: https://aturon.github.io/blog/2016/08/11/futures/
+[sq]: https://www.techempower.com/benchmarks/#section=data-r18&hw=ph&test=db
+[pt]: https://www.techempower.com/benchmarks/#section=data-r18&hw=ph&test=plaintext
+Subject: Re: manuscript SPE-19-0219
+To: "Peter A. Buhr" <pabuhr@uwaterloo.ca>
+From: Richard Jones <R.E.Jones@kent.ac.uk>
+Date: Tue, 12 Nov 2019 22:43:55 +0000
+Dear Dr Buhr
+Your should have received a decision letter on this today. I am sorry that this
+has taken so long. Unfortunately SP&E receives a lot of submissions and getting
+reviewers is a perennial problem.
+Regards
+Richard
+Peter A. Buhr wrote on 11/11/2019 13:10:
+>     26-Jun-2019
+>     Your manuscript entitled "Advanced Control-flow and Concurrency in Cforall"
+>     has been received by Software: Practice and Experience. It will be given
+>     full consideration for publication in the journal.
+>
+> Hi, it has been over 4 months since submission of our manuscript SPE-19-0219
+> with no response.
+>
+> Currently, I am refereeing a paper for IEEE that already cites our prior SP&E
+> paper and the Master's thesis forming the bases of the SP&E paper under
+> review. Hence our work is apropos and we want to get it disseminates as soon as
+> possible.
+>
+> [3] A. Moss, R. Schluntz, and P. A. Buhr, "Cforall: Adding modern programming
+>      language features to C," Software - Practice and Experience, vol. 48,
+>      no. 12, pp. 2111-2146, 2018.
+>
+> [4] T. Delisle, "Concurrency in C for all," Master's thesis, University of
+>      Waterloo, 2018.  [Online].  Available:
+>      https://uwspace.uwaterloo.ca/bitstream/handle/10012/12888
+Date: Mon, 13 Jan 2020 05:33:15 +0000
+From: Richard Jones <onbehalfof@manuscriptcentral.com>
+Reply-To: R.E.Jones@kent.ac.uk
+To: pabuhr@uwaterloo.ca
+Subject: Revision reminder - SPE-19-0219
+-Jan-2020
+Dear Dr Buhr
+SPE-19-0219
+This is a reminder that your opportunity to revise and re-submit your
+manuscript will expire 28 days from now. If you require more time please
+contact me directly and I may grant an extension to this deadline, otherwise
+the option to submit a revision online, will not be available.
+I look forward to receiving your revision.
+Sincerely,
+Prof. Richard Jones
+Editor, Software: Practice and Experience
+https://mc.manuscriptcentral.com/spe
+Date: Wed, 5 Feb 2020 04:22:18 +0000
+From: Aaron Thomas <onbehalfof@manuscriptcentral.com>
+Reply-To: speoffice@wiley.com
+To: tdelisle@uwaterloo.ca, pabuhr@uwaterloo.ca
+Subject: SPE-19-0219.R1 successfully submitted
+-Feb-2020
+Dear Dr Buhr,
+Your manuscript entitled "Advanced Control-flow and Concurrency in Cforall" has
+been successfully submitted online and is presently being given full
+consideration for publication in Software: Practice and Experience.
+Your manuscript number is SPE-19-0219.R1.  Please mention this number in all
+future correspondence regarding this submission.
+You can view the status of your manuscript at any time by checking your Author
+Center after logging into https://mc.manuscriptcentral.com/spe.  If you have
+difficulty using this site, please click the 'Get Help Now' link at the top
+right corner of the site.
+Thank you for submitting your manuscript to Software: Practice and Experience.
+Sincerely,
+Software: Practice and Experience Editorial Office

doc/proposals/vtable.md

-              rb7d6a36
+              r6a490b2
 default is provided or not, the second syntax can be used to pick a
 parameter on instantiation.
+### Extension: Object Access
+This requires that the resolution scope (see below) is at the type level or
+has explicate points with names. These are the tables and table names used
+here.
+The system already knows where to find the virtual table and the object. If
+the tables have particular identities, or on the user side names, then it is
+meaningful to check if a binding virtual table is the same* as another. The
+main use of this is virtual table declarations also give the type they bind
+and if a binding table matches a known table then the underlyind object in the
+trait object must be of that type.
+* By identity, by value would work and in some senses be more flexiable. But
+  it would be slower and refering to further away functions would be harder.
+This gives one of the main new features of the hierarchical use of virtual
+tables (see below); the ability to recover the underlying object. Or a pointer
+of the approprate type it which both reflects the implementation and gives a
+convenent way to encode the boolean/conditional aspect of the operation which
+is that a different virtual table might be in use.
+There are two general ways to reperent this; a cast or a field access. The
+cast is traditional and would definitely fit if a single pointer repersents
+a trait object with the virtual table as part of the object. However for a
+double pointer field access might be more approprate. By this system though
+it is not the type that is used as the identifier but the virtual table. If
+there is one table per type than it becomes equivilant again. Otherwise the
+table has to be used as the identifier and the type is just a result of that
+which seems important for syntax.
 Hierarchy
 …
 be used in only some of the declarations.
     trait combiner fee = (summation_instance, sum);
+    trait combiner fee = {summation_instance, sum};
     trait combiner foe = summation_instance;

doc/theses/thierry_delisle_PhD/.gitignore

rb7d6a36	r6a490b2
8	8
9	9	comp_II/build/
	10	comp_II/img/*.fig.bak
10	11	comp_II/comp_II.pdf
11	12	comp_II/comp_II.ps

doc/theses/thierry_delisle_PhD/comp_II/Makefile

-              rb7d6a36
+              r6a490b2
 Build = build
 Figures = figures
+Figures = img
 Macros = ../../../LaTeXmacros
 TeXLIB = .:${Macros}:${Build}:../../../bibliography:
 …
 FIGURES = ${addsuffix .tex, \
+        base \
+        empty \
+        emptybit \
+        emptytree \
+        emptytls \
+        resize \
+        system \
+}
 …
         mkdir -p ${Build}
 %.tex : %.fig ${Build}
+%.tex : img/%.fig ${Build}
         fig2dev -L eepic $< > ${Build}/$@
 %.ps : %.fig | ${Build}
+%.ps : img/%.fig | ${Build}
         fig2dev -L ps $< > ${Build}/$@
 %.pstex : %.fig | ${Build}
+%.pstex : img/%.fig | ${Build}
         fig2dev -L pstex $< > ${Build}/$@
         fig2dev -L pstex_t -p ${Build}/$@ $< > ${Build}/$@_t

doc/theses/thierry_delisle_PhD/comp_II/comp_II.tex

-              rb7d6a36
+              r6a490b2
+\documentclass[11pt,fullpage]{article}
+\documentclass[11pt]{article}
+\usepackage{fullpage}
 \usepackage[T1]{fontenc}
 \usepackage[utf8]{inputenc}
-\usepackage{listings}           % for code listings
 \usepackage{xspace}
 \usepackage{xcolor}
 \usepackage{graphicx}
+\usepackage[hidelinks]{hyperref}
+\usepackage{epic,eepic}
+\usepackage{listings}                   % for code listings
 \usepackage{glossaries}
 \usepackage{textcomp}
-\usepackage{geometry}
 % cfa macros used in the document
 \input{common}
+\setlist{topsep=6pt,parsep=0pt}         % global reduce spacing between points
+\newcommand{\uC}{$\mu$\CC}
+\usepackage[hidelinks]{hyperref}
+\setlength{\abovecaptionskip}{5pt plus 3pt minus 2pt}
+\lstMakeShortInline$%                   % single-character for \lstinline
+%\usepackage[margin=1in]{geometry}
+%\usepackage{float}
 \input{glossary}
 …
 \author{
         \huge Thierry Delisle \\
         \Large \vspace*{0.1in} \texttt{tdelisle@uwaterloo.ca} \\
+        \huge Thierry Delisle \vspace*{5pt} \\
+        \Large \texttt{tdelisle@uwaterloo.ca} \vspace*{5pt} \\
         \Large Cheriton School of Computer Science \\
         \Large University of Waterloo
 …
 \newcommand{\cit}{\textsuperscript{[Citation Needed]}\xspace}
 \newcommand{\TODO}{~\newline{\large\bf\color{red} TODO :}\xspace}
+\newcommand{\TODO}{{\large\bf\color{red} TODO: }\xspace}
 % ===============================================================================
 …
 \section{Introduction}
 \subsection{\CFA and the \CFA concurrency package}
+\CFA\cit is a modern, polymorphic, non-object-oriented, backwards-compatible extension of the C programming language. It aims to add high productivity features while maintaning the predictible performance of C. As such concurrency in \CFA\cit aims to offer simple and safe high-level tools while still allowing performant code. Concurrent code is written in the syncrhonous programming paradigm but uses \glspl{uthrd} in order to achieve the simplicity and maintainability of synchronous programming without sacrificing the efficiency of asynchronous programing. As such the \CFA scheduler is a user-level scheduler that maps \glspl{uthrd} onto \glspl{kthrd}.
+The goal of this research is to produce a scheduler that is simple to use and offers acceptable performance in all cases. Here simplicity does not refer to the API but to how much scheduling concerns programmers need to take into account when using the \CFA concurrency package. Therefore, the main goal of this proposal is as follows :
+\CFA\cite{Moss18} is a modern, polymorphic, non-object-oriented, concurrent, backwards-compatible extension of the C programming language.
+It aims to add high-productivity features while maintaining the predictable performance of C.
+As such, concurrency in \CFA\cite{Delisle19} aims to offer simple and safe high-level tools while still allowing performant code.
+\CFA concurrent code is written in the synchronous programming paradigm but uses \glspl{uthrd} in order to achieve the simplicity and maintainability of synchronous programming without sacrificing the efficiency of asynchronous programing.
+As such, the \CFA \newterm{scheduler} is a preemptive user-level scheduler that maps \glspl{uthrd} onto \glspl{kthrd}.
+\newterm{Scheduling} occurs when execution switches from one thread to another, where the second thread is implicitly chosen by the scheduler.
+This scheduling is an indirect handoff, as opposed to generators and coroutines which explicitly switch to the next generator and coroutine respectively.
+The cost of switching between two threads for an indirect handoff has two components:
+\begin{enumerate}
+\item
+the cost of actually context-switching, \ie changing the relevant registers to move execution from one thread to the other,
+\item
+and the cost of scheduling, \ie deciding which thread to run next among all the threads ready to run.
+\end{enumerate}
+The first cost is generally constant and fixed\footnote{Affecting the constant context-switch cost is whether it is done in one step, after the scheduling, or in two steps, context-switching to a fixed third-thread before scheduling.}, while the scheduling cost can vary based on the system state.
+Adding multiple \glspl{kthrd} does not fundamentally change the scheduler semantics or requirements, it simply adds new correctness requirements, \ie \newterm{linearizability}\footnote{Meaning however fast the CPU threads run, there is an equivalent sequential order that gives the same result.}, and a new dimension to performance: scalability, where scheduling cost now also depends on contention.
+The more threads switch, the more the administration cost of scheduling becomes noticeable.
+It is therefore important to build a scheduler with the lowest possible cost and latency.
+Another important consideration is \newterm{fairness}.
+In principle, scheduling should give the illusion of perfect fairness, where all threads ready to run are running \emph{simultaneously}.
+While the illusion of simultaneity is easier to reason about, it can break down if the scheduler allows too much unfairness.
+Therefore, the scheduler should offer as much fairness as needed to guarantee eventual progress, but use unfairness to help performance.
+In practice, threads must wait in turn but there can be advantages to unfair scheduling, similar to the the express cash-register at a grocery store.
+The goal of this research is to produce a scheduler that is simple for programmers to understand and offers good performance.
+Here understandability does not refer to the API but to how much scheduling concerns programmers need to take into account when writing a \CFA concurrent package.
+Therefore, the main goal of this proposal is :
 \begin{quote}
 The \CFA scheduler should be \emph{viable} for any workload.
+The \CFA scheduler should be \emph{viable} for \emph{any} workload.
 \end{quote}
+This objective includes producing a scheduling strategy with minimal fairness guarantees, creating an abstraction layer over the operating system to handle kernel-threads spinning unnecessarily and hide blocking I/O operations and, writing sufficient library tools to allow developpers to properly use the scheduler.
+% ===============================================================================
+% ===============================================================================
+\section{Scheduling for \CFA}
+While the \CFA concurrency package doesn't have any particular scheduling needs beyond those of any concurrency package which uses \glspl{uthrd}, it is important that the default \CFA Scheduler be viable in general. Indeed, since the \CFA Scheduler does not target any specific workloads, it is unrealistic to demand that it use the best scheduling strategy in all cases. However, it should offer a viable ``out of the box'' solution for most scheduling problems so that programmers can quickly write performant concurrent without needed to think about which scheduling strategy is more appropriate for their workload. Indeed, only programmers with exceptionnaly high performance requirements should need to write their own scheduler. More specifically, two broad types of schedulering strategies should be avoided in order to avoid penalizing certain types of workloads : feedback-based and priority schedulers.
+For a general purpose scheduler, it is impossible to produce an optimal algorithm as it would require knowledge of the future behaviour of threads.
+As such, scheduling performance is generally either defined by the best case scenario, \ie a workload to which the scheduler is tailored, or the worst case scenario, \ie the scheduler behaves no worst than \emph{X}.
+For this proposal, the performance is evaluated using the second approach to allow \CFA programmers to rely on scheduling performance.
+Because there is no optimal scheduler, ultimately \CFA may allow programmers to write their own scheduler; but that is not the subject of this proposal, which considers only the default scheduler.
+As such, it is important that only programmers with exceptionally high performance requirements should need to write their own scheduler and replace the scheduler in this proposal.
+To achieve the \CFA scheduling goal includes:
+\begin{enumerate}
+\item
+producing a scheduling strategy with sufficient fairness guarantees,
+\item
+creating an abstraction layer over the operating system to handle kernel-threads spinning unnecessarily,
+\item
+scheduling blocking I/O operations,
+\item
+and writing sufficient library tools to allow developers to indirectly use the scheduler, either through tuning knobs or replacing the default scheduler.
+\end{enumerate}
+% ===============================================================================
+% ===============================================================================
+\section{\CFA Scheduling}
+To schedule user-level threads across all workloads, the scheduler has a number of requirements:
+\paragraph{Correctness} As with any other concurrent data structure or algorithm, the correctness requirement is paramount.
+The scheduler cannot allow threads to be dropped from the ready queue, \ie scheduled but never run, or be executed multiple times when only being scheduled once.
+Since \CFA concurrency has no spurious wakeup, this definition of correctness also means the scheduler should have no spurious wakeup.
+The \CFA scheduler must be correct.
+\paragraph{Performance} The performance of a scheduler can generally be measured in terms of scheduling cost, scalability and latency.
+\newterm{Scheduling cost} is the cost to switch from one thread to another, as mentioned above.
+For simple applications, where a single kernel thread does most of the scheduling, it is generally the dominating cost.
+\newterm{Scalability} is the cost of adding multiple kernel threads because it increases the time for context switching because of contention by multiple threads accessing shared resources, \eg the ready queue.
+Finally, \newterm{tail latency} is service delay and relates to thread fairness.
+Specifically, latency measures how long a thread waits to run once scheduled and is evaluated in the worst case.
+The \CFA scheduler should offer good performance for all three metrics.
+\paragraph{Fairness} Like performance, this requirement has several aspect : eventual progress, predictability and performance reliability.
+\newterm{Eventual progress} guarantees every scheduled thread is eventually run, \ie prevent starvation.
+As a hard requirement, the \CFA scheduler must guarantee eventual progress, otherwise the above mentioned illusion of simultaneous execution is broken and the scheduler becomes much more complex to reason about.
+\newterm{Predictability} and \newterm{reliability} means similar workloads achieve similar performance and programmer execution intuition is respected.
+For example, a thread that yields aggressively should not run more often then other tasks.
+While this is intuitive, it does not hold true for many work-stealing or feedback based schedulers.
+The \CFA scheduler must guarantee eventual progress and should be predictable and offer reliable performance.
+\paragraph{Efficiency} Finally, efficient usage of CPU resources is also an important requirement and is discussed in depth towards the end of the proposal.
+\newterm{Efficiency} means avoiding using CPU cycles when there are no threads to run, and conversely, use all CPUs available when the workload can benefit from it.
+Balancing these two states is where the complexity lies.
+The \CFA scheduler should be efficient with respect to the underlying (shared) computer.
+\bigskip To achieve these requirements, I can reject two broad types of scheduling strategies : feedback-based and priority schedulers.
 \subsection{Feedback-Based Schedulers}
+Many operating systems use schedulers based on feadback loops in some form, they measure how much CPU a particular thread has used\footnote{Different metrics can be used to here but it is not relevant to the discussion.} and schedule threads based on this metric. These strategies are sensible for operating systems but rely on two assumptions on the workload :
+\begin{enumerate}
+        \item Threads live long enough to be scheduled many times.
+        \item Cooperation among all threads is not simply infeasible, it is a security risk.
+\end{enumerate}
+While these two assumptions generally hold for operating systems, they may not for \CFA programs. In fact, \CFA uses \glspl{uthrd} which have the explicit goal of reducing the cost of threading primitives to allow many smaller threads. This can naturally lead to have threads with much shorter lifetime and only being scheduled a few times. Scheduling strategies based on feadback loops cannot be effective in these cases because they will not have the opportunity to measure the metrics that underlay the algorithm. Note that the problem of feadback loop convergence (reacting too slowly to scheduling events) is not specific to short lived threads but can also occur with threads that show drastic changes in scheduling event, e.g., threads running for long periods of time and then suddenly blocking and unblocking quickly and repeatedly.
+In the context of operating systems, these concerns can be overshadowed by a more pressing concern : security. When multiple users are involved, it is possible that some users are malevolent and try to exploit the scheduling strategy in order to achieve some nefarious objective. Security concerns mean that more precise and robust fairness metrics must be used. In the case of the \CFA scheduler, every thread runs in the same user-space and are controlled from the same user. It is then possible to safely ignore the possibility that threads are malevolent and assume that all threads will ignore or cooperate with each other. This allows for a much simpler fairness metric and in this proposal ``fairness'' will be considered as equal opportunities to run once scheduled.
+Since feadback is not necessarily feasible within the lifetime of all threads and a simple fairness metric can be used, the scheduling strategy proposed for the \CFA runtime does not user per-threads feedback. Feedback loops in general are not rejected for secondary concerns like idle sleep, but no feedback loop is used to decide which thread to run next.
+Many operating systems use schedulers based on feedback in some form, \eg measuring how much CPU a particular thread has used\footnote{Different metrics can be measured but it is not relevant to the discussion.} and schedule threads based on this metric.
+These strategies are sensible for operating systems but rely on two assumptions for the workload:
+\begin{enumerate}
+        \item Threads live long enough for useful feedback information to be to gathered.
+        \item Threads belong to multiple users so fairness across threads is insufficient.
+\end{enumerate}
+While these two assumptions generally hold for operating systems, they may not for user-level threading.
+Since \CFA has the explicit goal of allowing many smaller threads, this can naturally lead to threads with much shorter lifetimes that are only scheduled a few times.
+Scheduling strategies based on feedback cannot be effective in these cases because there is no opportunity to measure the metrics that underlie the algorithm.
+Note, the problem of \newterm{feedback convergence} (reacting too slowly to scheduling events) is not specific to short lived threads but can also occur with threads that show drastic changes in scheduling, \eg threads running for long periods of time and then suddenly blocking and unblocking quickly and repeatedly.
+In the context of operating systems, these concerns can be overshadowed by a more pressing concern : security.
+When multiple users are involved, it is possible some users are malevolent and try to exploit the scheduling strategy to achieve some nefarious objective.
+Security concerns mean more precise and robust fairness metrics must be used to guarantee fairness across processes created by users as well as threads created within a process.
+In the case of the \CFA scheduler, every thread runs in the same user space and is controlled by the same user.
+Fairness across users is therefore a given and it is then possible to safely ignore the possibility that threads are malevolent.
+This approach allows for a much simpler fairness metric and in this proposal \emph{fairness} is defined as: when multiple threads are cycling through the system, the total ordering of threads being scheduled, \ie pushed onto the ready-queue, should not differ much from the total ordering of threads being executed, \ie popped from the ready-queue.
+Since feedback is not necessarily feasible within the lifetime of all threads and a simple fairness metric can be used, the scheduling strategy proposed for the \CFA runtime does not use per-threads feedback.
+Feedback in general is not rejected for secondary concerns like idle sleep for kernel threads, but no feedback is used to decide which thread to run next.
 \subsection{Priority Schedulers}
+Another broad category of schedulers are priority schedulers. In these scheduling strategies threads have priorities and the runtime schedules the threads with the highest priority before scheduling other threads. Threads with equal priority are scheduled using a secondary strategy, often something simple like round-robin or FIFO. These priority mean that, as long as there is a thread with a higher priority that desires to run, a thread with a lower priority will not run. This possible starving of threads can dramatically increase programming complexity since starving threads and priority inversion (prioritising a lower priority thread) can both lead to serious problems, leaving programmers between a rock and a hard place.
+An important observation to make is that threads do not need to have explicit priorities for problems to be possible. Indeed, any system with multiple ready-queues and attempts to exhaust one queue before accessing the other queues, could encounter starvation problems. A popular scheduling strategy that suffers from implicit priorities is work-stealing. Work-stealing is generally presented as follows :
+Another broad category of schedulers are priority schedulers.
+In these scheduling strategies, threads have priorities and the runtime schedules the threads with the highest priority before scheduling other threads.
+Threads with equal priority are scheduled using a secondary strategy, often something simple like round-robin or FIFO.
+A consequence of priority is that, as long as there is a thread with a higher priority that desires to run, a thread with a lower priority does not run.
+This possible starving of threads can dramatically increase programming complexity since starving threads and priority inversion (prioritizing a lower priority thread) can both lead to serious problems.
+An important observation is that threads do not need to have explicit priorities for problems to occur.
+Indeed, any system with multiple ready-queues that attempts to exhaust one queue before accessing the other queues, essentially provide implicit priority, which can encounter starvation problems.
+For example, a popular scheduling strategy that suffers from implicit priorities is work stealing.
+\newterm{Work stealing} is generally presented as follows:
+\begin{enumerate}
+        \item Each processor has a list of ready threads.
+        \item Each processor runs threads from its ready queue first.
+        \item If a processor's ready queue is empty, attempt to run threads from some other processor's ready queue.
+\end{enumerate}
+In a loaded system\footnote{A \newterm{loaded system} is a system where threads are being run at the same rate they are scheduled.}, if a thread does not yield, block, or preempt for an extended period of time, threads on the same processor's list starve if no other processors exhaust their list.
+Since priorities can be complex for programmers to incorporate into their execution intuition, the scheduling strategy proposed for the \CFA runtime does not use a strategy with either implicit or explicit thread priorities.
+\subsection{Schedulers without feedback or priorities}
+This proposal conjectures that is is possible to construct a default scheduler for the \CFA runtime that offers good scalability and a simple fairness guarantee that is easy for programmers to reason about.
+The simplest fairness guarantee is FIFO ordering, \ie threads scheduled first run first.
+However, enforcing FIFO ordering generally conflicts with scalability across multiple processors because of the additional synchronization.
+Thankfully, strict FIFO is not needed for sufficient fairness.
+Since concurrency is inherently non-deterministic, fairness concerns in scheduling are only a problem if a thread repeatedly runs before another thread can run.
+Some relaxation is possible because non-determinism means programmers already handle ordering problems to produce correct code and hence rely on weak guarantees, \eg that a specific thread will \emph{eventually} run.
+Since some reordering does not break correctness, the FIFO fairness guarantee can be significantly relaxed without causing problems.
+For this proposal, the target guarantee is that the \CFA scheduler provides \emph{probable} FIFO ordering, which allows reordering but makes it improbable that threads are reordered far from their position in total ordering.
+The \CFA scheduler fairness is defined as follows:
 \begin{itemize}
         \item Each processor has a list of threads.
+        \item Given two threads $X$ and $Y$, the odds that thread $X$ runs $N$ times \emph{after} thread $Y$ is scheduled but \emph{before} it is run, decreases exponentially with regard to $N$.
 \end{itemize}
+\begin{enumerate}
+        \item Run threads from ``this'' processor's list.
+        \item If ``this'' processor's list is empty, run threads from some other processor's list.
+\end{enumerate}
+In a loaded system\footnote{A loaded system is a system where threads are being run at the same rate they are scheduled}, if a thread does not yield or block for an extended period of time, threads on the same processor list will starve if no other processors can exhaust their list.
+Since priorities can be complex to handle for programmers, the scheduling strategy proposed for the \CFA runtime does not use a strategy with either implicit or explicit thread priorities.
+\subsection{Schedulers without feadback or priorities}
+I claim that the ideal default scheduler for the \CFA runtime is a scheduler that offers good scalability and a simple fairness guarantee that is easy for programmers to reason about. The simplest fairness guarantee is to guarantee FIFO ordering, i.e., threads scheduled first will run first. However, enforcing FIFO ordering generally conflicts with scalability across multiple processors because of the additionnal synchronization. Thankfully, strict FIFO is not needed for scheduling. Since concurrency is inherently non-deterministic, fairness concerns in scheduling are only a problem if a thread repeatedly runs before another thread can run\footnote{This is because the non-determinism means that programmers must already handle ordering problems in order to produce correct code and already must rely on weak guarantees, for example that a specific thread will \emph{eventually} run.}. This need for unfairness to persist before problems occur means that the FIFO fairness guarantee can be significantly relaxed without causing problems. For this proposal, the target guarantee is that the \CFA scheduler guarantees \emph{probable} FIFO ordering, which is defined as follows :
+\begin{itemize}
+        \item Given two threads $X$ and $Y$, the odds that thread $X$ runs $N$ times \emph{after} thread $Y$ is scheduled but \emph{before} it is run, decreases exponentially with regards to $N$.
+\end{itemize}
+While this is not a strong guarantee, the probability that problems persist for long period of times decreases exponentially, making persisting problems virtually impossible.
+\subsection{Real-Time}
+While the objective of this proposed scheduler is similar to the objective of real-time scheduling, this proposal is not a proposal for real-time scheduler and as such makes no attempt to offer either soft or hard guarantees on scheduling delays.
+% ===============================================================================
+% ===============================================================================
+\section{Proposal}
+\subsection{Ready-Queue}
+Using trevor's paper\cit as basis, it is simple to build a relaxed FIFO list that is fast and scalable for loaded or overloaded systems. The described queue uses an array of underlying strictly FIFO queue. Pushing new data is done by selecting one of these underlying queues at random, recording a timestamp for the push and pushing to the selected queue. Popping is done by selecting two queues at random and popping from the queue for which the head has the oldest timestamp. In loaded or overloaded systems, it is higly likely that the queues is far from empty, e.i., several tasks are on each of the underlying queues. This means that selecting a queue at random to pop from is higly likely to yield a queue that is not empty.
+When the ready queue is "more empty", i.e., several of the inner queues are empty, selecting a random queue for popping is less likely to yield a valid selection and more attempts need to be made, resulting in a performance degradation. In cases, with few elements on the ready queue and few processors running, performance can be improved by adding information to help processors find which inner queues are used. Preliminary performance tests indicate that with few processors, a bitmask can be used to identify which inner queues are currently in use. This is especially effective in the single-thread case, where the bitmask will always be up-to-date. Furthermore, modern x86 CPUs have a BMI2 extension which allow using the bitmask with very little overhead over directly accessing the readyqueue offerring decent performance even in cases with many empty inner queues. This technique does not solve the problem completely, it randomly attempts to find a block of 64 queues where at least one is used, instead of attempting to find a used queue. For systems with a large number of cores this does not completely solve the problem, but it is a fixed improvement. The size of the blocks are limited by the maximum size atomic instruction can operate on, therefore atomic instructions on large words would increase the 64 queues per block limit.
+\TODO double check the next sentence
+Preliminary result indicate that the bitmask approach with the BMI2 extension can lead to multi-threaded performance that is contention agnostic in the worst case.
+This result suggests that the contention penalty and the increase performance for additionnal thread cancel each other exactly. This may indicate that a relatively small reduction in contention may tip the performance into positive scalling even for the worst case. It can be noted that in cases of high-contention, the use of the bitmask to find queues that are not empty is much less reliable. Indeed, if contention on the bitmask is high, it means it probably changes significantly between the moment it is read and the actual operation on the queues it represents. Furthermore, the objective of the bitmask is to avoid probing queues that are empty. Therefore, in cases where the bitmask is highly contented, it may be preferrable to probe queues randomly, either until contention decreases or until a prior prefetch of the bitmask completes. Ideally, the scheduler would be able to observe that the bitmask is highly contented and adjust its behaviour appropriately. However, I am not aware of any mechanism to query whether a cacheline is in cache or to run other instructions until a cacheline is fetch without blocking on the cacheline. As such, an alternative that may have a similar impact would be for each thread to have their own bitmask, which would be updated both after each scheduler action and after a certain number of failed probing. If the bitmask has little contention, the local bitmask will be mostly up-to-date and several threads won't need to contend as much on the global bitmask. If the bitmask has significant contention, then fetching it becomes more expensive and threads may as well probe randomly. This solution claims that probing randomly or against an out-of-date bitmask is equivalent.
+In cases where this is insufficient, another approach is to use a hiearchical data structure. Creating a tree of nodes to reduce contention has been shown to work in similar cases\cit(SNZI: Scalable NonZero Indicators)\footnote{This particular paper seems to be patented in the US. How does that affect \CFA? Can I use it in my work?}. However, this approach may lead to poorer single-threaded performance due to the inherent pointer chasing, as such, it was not considered as the first approach but as a fallback in case the bitmask approach does not satisfy the performance goals.
+Part of this performance relies on contention being low when there are few threads on the readyqueue. However, this can be assumed reliably if the system handles putting idle processors to sleep, which is addressed in section \ref{sleep}.
+While this is not a bounded guarantee, the probability that unfairness persist for long periods of times decreases exponentially, making persisting unfairness virtually impossible.
+% ===============================================================================
+% ===============================================================================
+\section{Proposal Details}
+\subsection{Central Ready Queue} \label{sec:queue}
+A central ready queue can be built from a FIFO queue, where user threads are pushed onto the queue when they are ready to run, and processors (kernel-threads acting as virtual processors) pop the user threads from the queue and execute them.
+Alistarh \etal~\cite{alistarh2018relaxed} show it is straightforward to build a relaxed FIFO list that is fast and scalable for loaded or overloaded systems.
+The described queue uses an array of underlying strictly FIFO queues as shown in Figure~\ref{fig:base}\footnote{For this section, the number of underlying queues is assumed to be constant.
+Section~\ref{sec:resize} discusses resizing the array.}.
+Pushing new data is done by selecting one of these underlying queues at random, recording a timestamp for the operation and pushing to the selected queue.
+Popping is done by selecting two queues at random and popping from the queue with the oldest timestamp.
+A higher number of underlying queues leads to less contention on each queue and therefore better performance.
+In a loaded system, it is highly likely the queues are non-empty, \ie several tasks are on each of the underlying queues.
+This means that selecting a queue at random to pop from is highly likely to yield a queue with available items.
+In Figure~\ref{fig:base}, ignoring the ellipsis, the chances of getting an empty queue is 2/7 per pick, meaning two random picks yield an item approximately 9 times out of 10.
+\begin{figure}
+        \begin{center}
+                \input{base}
+        \end{center}
+        \caption{Relaxed FIFO list at the base of the scheduler: an array of strictly FIFO lists.
+The timestamp is in all nodes and cell arrays.}
+        \label{fig:base}
+\end{figure}
+\begin{figure}
+        \begin{center}
+                \input{empty}
+        \end{center}
+        \caption{``More empty'' state of the queue: the array contains many empty cells.}
+        \label{fig:empty}
+\end{figure}
+When the ready queue is \emph{more empty}, \ie several of the queues are empty, selecting a random queue for popping is less likely to yield a successful selection and more attempts are needed, resulting in a performance degradation.
+Figure~\ref{fig:empty} shows an example with fewer elements, where the chances of getting an empty queue is 5/7 per pick, meaning two random picks yield an item only half the time.
+Since the ready queue is not empty, the pop operation \emph{must} find an element before returning and therefore must retry.
+Note, the popping kernel thread has no work to do, but CPU cycles are wasted both for available user and kernel threads during the pop operation as the popping thread is using a CPU.
+Overall performance is therefore influenced by the contention on the underlying queues and pop performance is influenced by the item density.
+This leads to four performance cases for the centralized ready-queue, as depicted in Table~\ref{tab:perfcases}.
+The number of processors (many or few) refers to the number of kernel threads \emph{actively} attempting to pop user threads from the queues, not the total number of kernel threads.
+The number of threads (many or few) refers to the number of user threads ready to be run.
+Many threads means they outnumber processors significantly and most underlying queues have items, few threads mean there are barely more threads than processors and most underlying queues are empty.
+Cases with fewer threads than processors are discussed in Section~\ref{sec:sleep}.
+\begin{table}
+        \begin{center}
+                \begin{tabular}{|r|l|l|}
+                        \cline{2-3}
+                        \multicolumn{1}{r|}{} & \multicolumn{1}{c|}{Many Processors} & \multicolumn{1}{c|}{Few Processors} \\
+                        \hline
+                        Many Threads & A: good performance & B: good performance \\
+                        \hline
+                        Few Threads  & C: worst performance & D: poor performance \\
+                        \hline
+                \end{tabular}
+        \end{center}
+        \caption{Expected performance of the relaxed FIFO list in different cases.}
+        \label{tab:perfcases}
+\end{table}
+Performance can be improved in case~D (Table~\ref{tab:perfcases}) by adding information to help processors find which inner queues are used.
+This addition aims to avoid the cost of retrying the pop operation but does not affect contention on the underlying queues and can incur some management cost for both push and pop operations.
+The approach used to encode this information can vary in density and be either global or local.
+\newterm{Density} means the information is either packed in a few cachelines or spread across several cachelines, and \newterm{local information} means each thread uses an independent copy instead of a single global, \ie common, source of information.
+For example, Figure~\ref{fig:emptybit} shows a dense bitmask to identify which inner queues are currently in use.
+This approach means processors can often find user threads in constant time, regardless of how many underlying queues are empty.
+Furthermore, modern x86 CPUs have extended bit manipulation instructions (BMI2) that allow using the bitmask with very little overhead compared to the randomized selection approach for a filled ready queue, offering good performance even in cases with many empty inner queues.
+However, this technique has its limits: with a single word\footnote{Word refers here to however many bits can be written atomically.} bitmask, the total number of underlying queues in the ready queue is limited to the number of bits in the word.
+With a multi-word bitmask, this maximum limit can be increased arbitrarily, but it is not possible to check if the queue is empty by reading the bitmask atomically.
+Finally, a dense bitmap, either single or multi-word, causes additional problems in case C (Table 1), because many processors are continuously scanning the bitmask to find the few available threads.
+This increased contention on the bitmask(s) reduces performance because of cache misses after updates and the bitmask is updated more frequently by the scanning processors racing to read and/or update that information.
+This increased update frequency means the information in the bitmask is more often stale before a processor can use it to find an item, \ie mask read says there are available user threads but none on queue.
+\begin{figure}
+        \begin{center}
+                {\resizebox{0.8\textwidth}{!}{\input{emptybit}}}
+        \end{center}
+        \caption{``More empty'' queue with added bitmask to indicate which array cells have items.}
+        \label{fig:emptybit}
+\end{figure}
+Figure~\ref{fig:emptytree} shows another approach using a hierarchical tree data-structure to reduce contention and has been shown to work in similar cases~\cite{ellen2007snzi}\footnote{This particular paper seems to be patented in the US.
+How does that affect \CFA? Can I use it in my work?}.
+However, this approach may lead to poorer performance in case~B (Table~\ref{tab:perfcases}) due to the inherent pointer chasing cost and already low contention cost in that case.
+\begin{figure}
+        \begin{center}
+                {\resizebox{0.8\textwidth}{!}{\input{emptytree}}}
+        \end{center}
+        \caption{``More empty'' queue with added binary search tree indicate which array cells have items.}
+        \label{fig:emptytree}
+\end{figure}
+Finally, a third approach is to use dense information, similar to the bitmap, but have each thread keep its own independent copy of it.
+While this approach can offer good scalability \emph{and} low latency, the liveliness of the information can become a problem.
+In the simple cases, local copies of which underlying queues are empty can become stale and end-up not being useful for the pop operation.
+A more serious problem is that reliable information is necessary for some parts of this algorithm to be correct.
+As mentioned in this section, processors must know \emph{reliably} whether the list is empty or not to decide if they can return \texttt{NULL} or if they must keep looking during a pop operation.
+Section~\ref{sec:sleep} discusses another case where reliable information is required for the algorithm to be correct.
+\begin{figure}
+        \begin{center}
+                \input{emptytls}
+        \end{center}
+        \caption{``More empty'' queue with added per processor bitmask to indicate which array cells have items.}
+        \label{fig:emptytls}
+\end{figure}
+There is a fundamental tradeoff among these approach.
+Dense global information about empty underlying queues helps zero-contention cases at the cost of high-contention case.
+Sparse global information helps high-contention cases but increases latency in zero-contention-cases, to read and ``aggregate'' the information\footnote{Hierarchical structures, \eg binary search tree, effectively aggregate information but follow pointer chains, learning information at each node.
+Similarly, other sparse schemes need to read multiple cachelines to acquire all the information needed.}.
+Finally, dense local information has both the advantages of low latency in zero-contention cases and scalability in high-contention cases, however the information can become stale making it difficult to use to ensure correctness.
+The fact that these solutions have these fundamental limits suggest to me a better solution that attempts to combine these properties in an interesting ways.
+Also, the lock discussed in Section~\ref{sec:resize} allows for solutions that adapt to the number of processors, which could also prove useful.
 \paragraph{Objectives and Existing Work}
+How much scalability is actually needed is highly debatable, libfibre\cit is has compared favorably to other schedulers in webserver tests\cit and uses a single atomic counter in its scheduling algorithm similarly to the proposed bitmask. As such the single atomic instruction on a shared cacheline may be sufficiently performant.
+I have built a prototype of this ready-queue (including the bitmask and BMI2 usage, but not the sharded bitmask) and ran performance experiments on it but it is difficult to compare this prototype to a thread scheduler as the prototype is used as a data-queue. I have also integrated this prototype into the \CFA runtime, but have not yet created performance experiments to compare results. I believe that the bitmask approach is currently one of the larger risks of the proposal, early tests lead me to believe it may work but it is not clear that the contention problem can be overcome. The worst-case scenario is a case where the number of processors and the number of ready threads are similar, yet scheduling events are very frequent. Fewer threads should lead to the Idle Sleep mechanism reducing contention while having many threads ready leads to optimal performance. It is difficult to evaluate the likeliness of this worst-case scenario in real workloads. I believe, frequent scheduling events suggest a more ``bursty'' workload where new work is finely divided among many threads which race to completion. This type of workload would only see a peek of contention close to the end of the work, but no sustained contention. Very fine-grained pipelines are less ``bursty'', these may lead to more sustained contention. However, they could also easily benefit from a direct hand-off strategy which would circumvent the problem entirely.
+\subsection{Dynamic Resizing}
+The \CFA runtime system currently handles dynamically adding and removing processors from clusters at any time. Since this is part of the existing design, the proposed scheduler must also support this behaviour. However, dynamicly resizing the clusters is considered a rare event associated with setup, teardown and major configuration changes. This assumptions is made both in the design of the proposed scheduler as well as in the original design of the \CFA runtime system. As such, the proposed scheduler must honor the correctness of these behaviour but does not have any performance objectives with regards to resizing a cluster. How long adding or removing processors take and how much this disrupts the performance of other threads is considered a secondary concern since it should be amortized over long period of times. This description effectively matches with te description of a Reader-Writer lock, in frequent but invasive updates among frequent (mostly) read operations. In the case of the Ready-Queue described above, read operations are operations that push or pop from the ready-queue but do not invalidate any references to the ready queue data structures. Writes on the other-hand would add or remove inner queues, invalidating references to the array of inner queues in the process. Therefore, the current proposed approach to this problem is the add a per-cluster Reader Writer lock around the ready queue to prevent restructuring of the ready-queue data structure while threads are being pushed or popped.
+There are possible alternatives to the Reader Writer lock solution. This problem is effectively a memory reclamation problem and as such there is a large body of research on the subject. However, the RWlock solution is simple and can be leveraged to solve other problems (e.g. processor ordering and memory reclamation of threads) which makes it an attractive solution.
+How much scalability is actually needed is highly debatable.
+\emph{libfibre}\cite{libfibre} has compared favorably to other schedulers in webserver tests\cite{karstenuser} and uses a single atomic counter in its scheduling algorithm similarly to the proposed bitmask.
+As such, the single atomic instruction on a shared cacheline may be sufficiently performant.
+I have built a prototype of this ready queue in the shape of a data queue, \ie nodes on the queue are structures with a single int representing a thread and intrusive data fields.
+Using this prototype I ran preliminary performance experiments that confirm the expected performance in Table~\ref{tab:perfcases}.
+However, these experiments only offer a hint at the actual performance of the scheduler since threads form more complex operations than simple integer nodes, \eg threads are not independent of each other, when a thread blocks some other thread must intervene to wake it.
+I have also integrated this prototype into the \CFA runtime, but have not yet created performance experiments to compare results, as creating one-to-one comparisons between the prototype and the \CFA runtime will be complex.
+\subsection{Dynamic Resizing} \label{sec:resize}
+\begin{figure}
+        \begin{center}
+                \input{system}
+        \end{center}
+        \caption{Global structure of the \CFA runtime system.}
+        \label{fig:system}
+\end{figure}
+The \CFA runtime system groups processors together as \newterm{clusters}, as shown in Figure~\ref{fig:system}.
+Threads on a cluster are always scheduled on one of the processors of the cluster.
+Currently, the runtime handles dynamically adding and removing processors from clusters at any time.
+Since this is part of the existing design, the proposed scheduler must also support this behaviour.
+However, dynamically resizing a cluster is considered a rare event associated with setup, tear down and major configuration changes.
+This assumption is made both in the design of the proposed scheduler as well as in the original design of the \CFA runtime system.
+As such, the proposed scheduler must honour the correctness of this behaviour but does not have any performance objectives with regard to resizing a cluster.
+How long adding or removing processors take and how much this disrupts the performance of other threads is considered a secondary concern since it should be amortized over long period of times.
+However, as mentioned in Section~\ref{sec:queue}, contention on the underlying queues can have a direct impact on performance.
+The number of underlying queues must therefore be adjusted as the number of processors grows or shrinks.
+Since the underlying queues are stored in a dense array, changing the number of queues requires resizing the array and expanding the array requires moving it, which can introduce memory reclamation problems if not done correctly.
+\begin{figure}
+        \begin{center}
+                \input{resize}
+        \end{center}
+        \caption{Copy of data structure shown in Figure~\ref{fig:base}.}
+        \label{fig:base2}
+\end{figure}
+It is important to note how the array is used in this case.
+While the array cells are modified by every push and pop operation, the array itself, \ie the pointer that would change when resized, is only read during these operations.
+Therefore the use of this pointer can be described as frequent reads and infrequent writes.
+This description effectively matches with the description of a reader-writer lock, infrequent but invasive updates among frequent read operations.
+In the case of the ready queue described above, read operations are operations that push or pop from the ready queue but do not invalidate any references to the ready queue data structures.
+Writes on the other hand would add or remove inner queues, invalidating references to the array of inner queues in a process.
+Therefore, the current proposed approach to this problem is to add a per-cluster reader-writer lock around the ready queue to prevent restructuring of the ready-queue data-structure while threads are being pushed or popped.
+There are possible alternatives to the reader-writer lock solution.
+This problem is effectively a memory reclamation problem and as such there is a large body of research on the subject\cite{michael2004hazard, brown2015reclaiming}.
+However, the reader-write lock-solution is simple and can be leveraged to solve other problems (\eg processor ordering and memory reclamation of threads), which makes it an attractive solution.
 \paragraph{Objectives and Existing Work}
+The lock must offer scalability and performance on par with the actual ready-queue in order not to introduce a new bottle neck. I have already built a lock that fits the desired requirements and preliminary testing show scalability and performance that exceed the target. As such, I do not consider this lock to be a risk on this project.
+\subsection{Idle Sleep} \label{sleep}
+As mentionned above, idle sleep is the process of putting processors to sleep while they do not have threads to execute. In this context processors are kernel-threads and sleeping refers to asking the kernel to block a thread. This can be achieved with either thread synchronization operations like pthread\_cond\_wait or using signal operations like sigsuspend.
+Support for idle sleep broadly involves calling the operating system to block the kernel thread but also handling the race between the sleeping and the waking up, and handling which kernel thread should sleep or wake-up.
+When a processor decides to sleep, there is a race that occurs between it signalling that it will go to sleep (so other processors can find sleeping processors) and actually blocking the kernel thread. This is equivalent to the classic problem of missing signals when using condition variables, the ``sleepy'' processor indicates that it will sleep but has not yet gone to sleep, if another processor attempts to wake it up, the waking-up operation may claim nothing needs to be done and the signal will have been missed. In cases where threads are scheduled from processors on the current cluster, loosing signals is not necessarily critical, because at least some processors on the cluster are awake. Individual processors always finish shceduling threads before looking for new work, which means that the last processor to go to sleep cannot miss threads scheduled from inside the cluster (if they do, that demonstrates the ready-queue is not linearizable). However, this guarantee does not hold if threads are shceduled from outside the cluster, either due to an external event like timers and I/O, or due to a thread migrating from a different cluster. In this case, missed signals can lead to the cluster deadlocking where it should not\footnote{Clusters ``should'' never deadlock, but for this proposal, cases where \CFA users \emph{actually} wrote \CFA code that leads to a deadlock it is considered as a deadlock that ``should'' happen. }. Therefore, it is important that the scheduling of threads include a mechanism where signals \emph{cannot} be missed. For performance reasons, it can be advantageous to have a secondary mechanism that allows signals to be missed in cases where it cannot lead to a deadlock. To be safe, this process must include a ``handshake'' where it is guaranteed that either~: the sleepy processor notices that a thread was scheduled after it signalled its intent to block or code scheduling threads well see the intent to sleep before scheduling and be able to wake-up the processor. This matter is complicated by the fact that pthread offers few tools to implement this solution and offers no guarantee of ordering of threads waking up for most of these tools.
+Another issues is trying to avoid kernel sleeping and waking frequently. A possible partial solution is to order the processors so that the one which most recently went to sleep is woken up. This allows other sleeping processors to reach deeper sleep state (when these are available) while keeping ``hot'' processors warmer. Note that while this generally means organising the processors in a stack, I believe that the unique index provided by the ReaderWriter lock can be reused to strictly order the waking order of processors, causing a LIFO like waking order. While a strict LIFO stack is probably better, using the processor index could proove useful and offer a sufficiently LIFO ordering.
+Finally, another important aspect of Idle Sleep is when should processors make the decision to sleep and when it is appropriate for sleeping processors to be woken up. Processors that are unnecessarily awake lead to unnecessary contention and power consumption, while too many sleeping processors can lead to sub-optimal throughput. Furthermore, transitions from sleeping to awake and vice-versa also add unnecessary latency. There is already a wealth of research on the subject and I do not plan to implement a novel idea for the Idle Sleep heuristic in this project.
+The lock must offer scalability and performance on par with the actual ready-queue in order not to introduce a new bottleneck.
+I have already built a lock that fits the desired requirements and preliminary testing show scalability and performance that exceed the target.
+As such, I do not consider this lock to be a risk for this project.
+\subsection{Idle Sleep} \label{sec:sleep}
+\newterm{Idle sleep} is the process of putting processors to sleep when they have no threads to execute.
+In this context, processors are kernel threads and sleeping refers to asking the kernel to block a thread.
+This operation can be achieved with either thread synchronization operations like $pthread_cond_wait$ or using signal operations like $sigsuspend$.
+The goal of putting idle processors to sleep is:
+\begin{enumerate}
+\item
+reduce contention on the ready queue, since the otherwise idle processors generally contend trying to pop items from the queue,
+\item
+give back unneeded CPU time associated with a process to other user processors executing on the computer,
+\item
+and reduce energy consumption in cases where more idle kernel-threads translate to idle CPUs, which can cycle down.
+\end{enumerate}
+Support for idle sleep broadly involves calling the operating system to block the kernel thread and handling the race between a blocking thread and the waking thread, and handling which kernel thread should sleep or wake up.
+When a processor decides to sleep, there is a race that occurs between it signalling that is going to sleep (so other processors can find sleeping processors) and actually blocking the kernel thread.
+This operation is equivalent to the classic problem of missing signals when using condition variables: the ``sleepy'' processor indicates its intention to block but has not yet gone to sleep when another processor attempts to wake it up.
+The waking-up operation sees the blocked process and signals it, but the blocking process is racing to sleep so the signal is missed.
+In cases where kernel threads are managed as processors on the current cluster, loosing signals is not necessarily critical, because at least some processors on the cluster are awake and may check for more processors eventually.
+Individual processors always finish scheduling user threads before looking for new work, which means that the last processor to go to sleep cannot miss threads scheduled from inside the cluster (if they do, that demonstrates the ready queue is not linearizable).
+However, this guarantee does not hold if threads are scheduled from outside the cluster, either due to an external event like timers and I/O, or due to a user (or kernel) thread migrating from a different cluster.
+In this case, missed signals can lead to the cluster deadlocking\footnote{Clusters should only deadlock in cases where a \CFA programmer \emph{actually} write \CFA code that leads to a deadlock.}.
+Therefore, it is important that the scheduling of threads include a mechanism where signals \emph{cannot} be missed.
+For performance reasons, it can be advantageous to have a secondary mechanism that allows signals to be missed in cases where it cannot lead to a deadlock.
+To be safe, this process must include a ``handshake'' where it is guaranteed that either~: the sleeping processor notices that a user thread is scheduled after the sleeping processor signalled its intent to block or code scheduling threads sees the intent to sleep before scheduling and be able to wake-up the processor.
+This matter is complicated by the fact that pthreads and Linux offer few tools to implement this solution and no guarantee of ordering of threads waking up for most of these tools.
+Another important issue is avoiding kernel threads sleeping and waking frequently because there is a significant operating-system cost.
+This scenario happens when a program oscillates between high and low activity, needing most and then less processors.
+A possible partial solution is to order the processors so that the one which most recently went to sleep is woken up.
+This allows other sleeping processors to reach deeper sleep state (when these are available) while keeping ``hot'' processors warmer.
+Note that while this generally means organizing the processors in a stack, I believe that the unique index provided in my reader-writer lock can be reused to strictly order the waking processors, causing a mostly LIFO order.
+While a strict LIFO stack is probably better, the processor index could prove useful for other reasons, while still offering a sufficiently LIFO ordering.
+A final important aspect of idle sleep is when should processors make the decision to sleep and when is it appropriate for sleeping processors to be woken up.
+Processors that are unnecessarily unblocked lead to unnecessary contention, CPU usage, and power consumption, while too many sleeping processors can lead to sub-optimal throughput.
+Furthermore, transitions from sleeping to awake and vice-versa also add unnecessary latency.
+There is already a wealth of research on the subject\cite{schillings1996engineering, wiki:thunderherd} and I may use an existing approach for the idle-sleep heuristic in this project, \eg\cite{karstenuser}.
 \subsection{Asynchronous I/O}
+The final aspect of this proposal is asynchronous I/O. Without it, user threads that execute I/O operations will block the underlying kernel thread. This leads to poor throughput, it would be preferrable to block the user-thread and reuse the underlying kernel-thread to run other ready threads. This requires intercepting the user-threads' calls to I/O operations, redirecting them to an asynchronous I/O interface and handling the multiplexing between the synchronous and asynchronous API. As such, these are the three components needed to implemented to support asynchronous I/O : an OS abstraction layer over the asynchronous interface, an event-engine to (de)multiplex the operations and a synchronous interface for users to use. None of these components currently exist in \CFA and I will need to build all three for this project.
+The final aspect of this proposal is asynchronous I/O.
+Without it, user threads that execute I/O operations block the underlying kernel thread, which leads to poor throughput.
+It is preferable to block the user thread performing the I/O and reuse the underlying kernel-thread to run other ready user threads.
+This approach requires intercepting user-thread calls to I/O operations, redirecting them to an asynchronous I/O interface, and handling the multiplexing/demultiplexing between the synchronous and asynchronous API.
+As such, there are three components needed to implemented support for asynchronous I/O:
+\begin{enumerate}
+\item
+an OS abstraction layer over the asynchronous interface,
+\item
+an event-engine to (de)multiplex the operations,
+\item
+and a synchronous interface for users to use.
+\end{enumerate}
+None of these components currently exist in \CFA and I will need to build all three for this project.
 \paragraph{OS Abstraction}
+One of the fundamental part of this converting blocking I/O operations into non-blocking ones. This relies on having an underlying asynchronous I/O interface to which to direct the I/O operations. While there exists many different APIs for asynchronous I/O, it is not part of this proposal to create a novel API, simply to use an existing one that is sufficient. uC++ uses the \texttt{select} as its interface, which handles pipes and sockets. It entails significant complexity and has performances problems which make it a less interesting alternative. Another interface which is becoming popular recently\cit is \texttt{epoll}. However, epoll also does not handle file system and seems to have problem to linux pipes and \texttt{TTY}s\cit. A very recent alternative that must still be investigated is \texttt{io\_uring}. It claims to address some of the issues with \texttt{epoll} but is too recent to be confident that it does. Finally, a popular cross-platform alternative is \texttt{libuv}, which offers asynchronous sockets and asynchronous file system operations (among other features). However, as a full-featured library it includes much more than what is needed and could conflict with other features of \CFA unless significant efforts are made to merge them together.
+\paragraph{Event-Engine}
+Laying on top of the asynchronous interface layer is the event-engine. This engine is responsible for multiplexing (batching) the synchronous I/O requests into an asynchronous I/O request and demultiplexing the results onto appropriate blocked threads. This can be straightforward for the simple cases, but can become quite complex. Decisions that will need to be made include : whether to poll from a seperate kernel thread or a regularly scheduled user thread, what should be the ordering used when results satisfy many requests, how to handle threads waiting for multiple operations, etc.
+One fundamental part for converting blocking I/O operations into non-blocking ones is having an underlying asynchronous I/O interface to direct the I/O operations.
+While there exists many different APIs for asynchronous I/O, it is not part of this proposal to create a novel API.
+It is sufficient to make one work in the complex context of the \CFA runtime.
+\uC uses the $select$\cite{select} as its interface, which handles ttys, pipes and sockets, but not disk.
+$select$ entails significant complexity and is being replaced in UNIX operating-systems, which make it a less interesting alternative.
+Another popular interface is $epoll$\cite{epoll}, which is supposed to be cheaper than $select$.
+However, $epoll$ also does not handle the file system and anectodal evidence suggest it has problem with linux pipes and $TTY$s.
+A popular cross-platform alternative is $libuv$\cite{libuv}, which offers asynchronous sockets and asynchronous file system operations (among other features).
+However, as a full-featured library it includes much more than I need and could conflict with other features of \CFA unless significant effort is made to merge them together.
+A very recent alternative that I am investigating is $io_uring$\cite{io_uring}.
+It claims to address some of the issues with $epoll$ and my early investigating suggest that the claim is accurate.
+$io_uring$ uses a much more general approach where system calls are register to a queue and later executed by the kernel, rather than relying on system calls to return an error instead of blocking and subsequently waiting for changes on file descriptors.
+I believe this approach allows for fewer problems, \eg the manpage for $open$\cite{open} states:
+\begin{quote}
+        Note that [the $O_NONBLOCK$ flag] has no effect for regular files and block devices;
+        that is, I/O operations will (briefly) block when device activity is required, regardless of whether $O_NONBLOCK$ is set.
+        Since $O_NONBLOCK$ semantics might eventually be implemented, applications should not depend upon blocking behavior when specifying this flag for regular files and block devices.
+\end{quote}
+This makes approach based on $epoll$/$select$ less reliable since they may not work for every file descriptors.
+For this reason, I plan to use $io_uring$ as the OS abstraction for the \CFA runtime, unless further work shows problems I haven't encountered yet.
+However, only a small subset of the features are available in Ubuntu as of April 2020\cite{wiki:ubuntu-linux}, which will limit performance comparisons.
+I do not believe this will affect the comparison result.
+\paragraph{Event Engine}
+Laying on top of the asynchronous interface layer is the event engine.
+This engine is responsible for multiplexing (batching) the synchronous I/O requests into asynchronous I/O requests and demultiplexing the results to appropriate blocked user threads.
+This step can be straightforward for simple cases, but becomes quite complex when there are thousands of user threads performing both reads and writes, possibly on overlapping file descriptors.
+Decisions that need to be made include:
+\begin{enumerate}
+\item
+whether to poll from a separate kernel thread or a regularly scheduled user thread,
+\item
+what should be the ordering used when results satisfy many requests,
+\item
+how to handle threads waiting for multiple operations, etc.
+\end{enumerate}
 \paragraph{Interface}
+Finally, for these components to be available, it is necessary to expose them through a synchronous interface. This can be a novel interface but it is preferrable to attempt to intercept the existing POSIX interface in order to be compatible with existing code. This will allow C programs written using this interface to be transparently converted to \CFA with minimal effeort. Where this is not applicable, a novel interface will be created to fill the gaps.
+Finally, for these non-blocking I/O components to be available, it is necessary to expose them through a synchronous interface because that is the \CFA concurrent programming style.
+The interface can be novel but it is preferable to match the existing POSIX interface when possible to be compatible with existing code.
+Matching allows C programs written using this interface to be transparently converted to \CFA with minimal effort.
+Where new functionality is needed, I will create a novel interface to fill gaps and provide advanced features.
 …
 % ===============================================================================
 \section{Discussion}
+I believe that runtime system and scheduling are still open topics.
+Many ``state of the art'' production frameworks still use single threaded event-loops because of performance considerations, \eg \cite{nginx-design}, and, to my knowledge, no wideyl available system language offers modern threading facilities.
+I believe the proposed work offers a novel runtime and scheduling package, where existing work only offers fragments that users must assemble themselves when possible.
 % ===============================================================================
 % ===============================================================================
 \section{Timeline}
+\cleardoublepage
+\begin{center}
+\begin{tabular}{ | r @{--} l | p{4in} | }
+\hline May 2020 & October 2020   & Creation of the performance benchmark. \\
+\hline November 2020 & March 2021   & Completion of the implementation. \\
+\hline March 2021 & April 2021  & Final Performance experiments. \\
+\hline May 2021 & August 2021 & Thesis writing and defense. \\
+\hline
+\end{tabular}
+\end{center}
 % B I B L I O G R A P H Y
 % -----------------------------
+\addcontentsline{toc}{chapter}{Bibliography}
+\cleardoublepage
+\phantomsection         % allows hyperref to link to the correct page
+\addcontentsline{toc}{section}{\refname}
 \bibliographystyle{plain}
 \bibliography{pl,local}
+% G L O S S A R Y
+% -----------------------------
 \cleardoublepage
 \phantomsection         % allows hyperref to link to the correct page
+% G L O S S A R Y
+% -----------------------------
+\addcontentsline{toc}{chapter}{Glossary}
+\addcontentsline{toc}{section}{Glossary}
 \printglossary
-\cleardoublepage
-\phantomsection         % allows hyperref to link to the correct page
 \end{document}

doc/theses/thierry_delisle_PhD/comp_II/local.bib

-              rb7d6a36
+              r6a490b2
 @article{finkel1987dib,
   title={DIB—a distributed implementation of backtracking},
+  title={DIB-a distributed implementation of backtracking},
   author={Finkel, Raphael and Manber, Udi},
   journal={ACM Transactions on Programming Languages and Systems (TOPLAS)},
 …
   organization={ACM}
+}
+% ===============================================================================
+% Algorithms
+% ===============================================================================
+@article{michael2004hazard,
+  title={Hazard pointers: Safe memory reclamation for lock-free objects},
+  author={Michael, Maged M},
+  journal={IEEE Transactions on Parallel and Distributed Systems},
+  volume={15},
+  number={6},
+  pages={491--504},
+  year={2004},
+  publisher={IEEE}
+}
+@inproceedings{brown2015reclaiming,
+  title={Reclaiming memory for lock-free data structures: There has to be a better way},
+  author={Brown, Trevor Alexander},
+  booktitle={Proceedings of the 2015 ACM Symposium on Principles of Distributed Computing},
+  pages={261--270},
+  year={2015}
+}
+% Trevor's relaxed FIFO list
+@inproceedings{alistarh2018relaxed,
+  title={Relaxed schedulers can efficiently parallelize iterative algorithms},
+  author={Alistarh, Dan and Brown, Trevor and Kopinsky, Justin and Nadiradze, Giorgi},
+  booktitle={Proceedings of the 2018 ACM Symposium on Principles of Distributed Computing},
+  pages={377--386},
+  year={2018}
+}
+% Scalable counters which only support is !0
+@inproceedings{ellen2007snzi,
+  title={SNZI: Scalable nonzero indicators},
+  author={Ellen, Faith and Lev, Yossi and Luchangco, Victor and Moir, Mark},
+  booktitle={Proceedings of the twenty-sixth annual ACM symposium on Principles of distributed computing},
+  pages={13--22},
+  year={2007}
+}
+% ===============================================================================
+% Linux Man Pages
+% ===============================================================================
+@manual{open,
+  key        = "open",
+  title      = "open(2) Linux User's Manual",
+  year       = "2020",
+  month      = "February",
+}
+@manual{epoll,
+  key        = "epoll",
+  title      = "epoll(7) Linux User's Manual",
+  year       = "2019",
+  month      = "March",
+}
+@manual{select,
+  key        = "select",
+  title      = "select(7) Linux User's Manual",
+  year       = "2019",
+  month      = "March",
+}
+@misc{io_uring,
+  title   = {Efficient IO with io\_uring},
+  author  = {Axboe, Jens},
+  year    = "2019",
+  month   = "March",
+  version = {0,4},
+  howpublished = {\url{https://kernel.dk/io_uring.pdf}}
+}
+@misc{libuv,
+  key   = "libuv",
+  title = {libuv},
+  howpublished = {\url{https://github.com/libuv/libuv}}
+}
+% ===============================================================================
+% MISC
+% ===============================================================================
+@misc{nginx-design,
+  key   = "nginx",
+  title={Inside {NGINX}: How We Designed for Performance \& Scale},
+  howpublished= {\href{https://www.nginx.com/blog/inside-nginx-how-we-designed-for-performance-scale}
+                {https://\-www.nginx.com/\-blog/\-inside\--nginx\--how\--we\--designed\--for\--performance\--scale}},
+}
+@article{schillings1996engineering,
+  title={Be engineering insights: Benaphores},
+  author={Schillings, Benoit},
+  journal={Be Newsletters},
+  volume={1},
+  number={26},
+  year={1996}
+}
+@misc{wiki:thunderherd,
+   author = "{Wikipedia contributors}",
+   title = "Thundering herd problem --- {W}ikipedia{,} The Free Encyclopedia",
+   year = "2020",
+   howpublished = {\href{https://en.wikipedia.org/wiki/Thundering_herd_problem}
+                  {https://\-en.wikipedia.org/\-wiki/\-Thundering\_herd\_problem}},},
+   note = "[Online; accessed 14-April-2020]"
+}
+@misc{wiki:ubuntu-linux,
+   author = "{Wikipedia contributors}",
+   title = "Ubuntu version history : Table of versions --- {W}ikipedia{,} The Free Encyclopedia",
+   year = "2020",
+   howpublished = {\href{https://en.wikipedia.org/wiki/Ubuntu_version_history\#Table_of_versions}
+                  {https://\-en.wikipedia.org/\-wiki/\-Ubuntu\_version\_history\#Table\_of\_versions}},
+   note = "[Online; accessed 15-April-2020]"
+}

doc/user/user.tex

-              rb7d6a36
+              r6a490b2
 %% Created On       : Wed Apr  6 14:53:29 2016
 %% Last Modified By : Peter A. Buhr
 %% Last Modified On : Sat Jul 13 18:36:18 2019
 %% Update Count     : 3876
+%% Last Modified On : Fri Mar  6 13:34:52 2020
+%% Update Count     : 3924
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 …
 Even with all its problems, C continues to be popular because it allows writing software at virtually any level in a computer system without restriction.
 For system programming, where direct access to hardware, storage management, and real-time issues are a requirement, C is usually the only language of choice.
 The TIOBE index~\cite{TIOBE} for July 2018 ranks the top five most \emph{popular} programming languages as \Index*{Java} 16\%, C 14\%, \Index*[C++]{\CC{}} 7.5\%, Python 6\%, Visual Basic 4\% = 47.5\%, where the next 50 languages are less than 4\% each, with a long tail.
 The top 3 rankings over the past 30 years are:
+The TIOBE index~\cite{TIOBE} for February 2020 ranks the top six most \emph{popular} programming languages as \Index*{Java} 17.4\%, C 16.8\%, Python 9.3\%, \Index*[C++]{\CC{}} 6.2\%, \Csharp 5.9\%, Visual Basic 5.9\% = 61.5\%, where the next 50 languages are less than 2\% each, with a long tail.
+The top 4 rankings over the past 35 years are:
 \begin{center}
 \setlength{\tabcolsep}{10pt}
+\begin{tabular}{@{}rccccccc@{}}
+                & 2018  & 2013  & 2008  & 2003  & 1998  & 1993  & 1988  \\ \hline
+Java    & 1             & 2             & 1             & 1             & 16    & -             & -             \\
+\R{C}   & \R{2} & \R{1} & \R{2} & \R{2} & \R{1} & \R{1} & \R{1} \\
+\CC             & 3             & 4             & 3             & 3             & 2             & 2             & 5             \\
+\begin{tabular}{@{}rcccccccc@{}}
+                & 2020  & 2015  & 2010  & 2005  & 2000  & 1995  & 1990  & 1985  \\ \hline
+Java    & 1             & 2             & 1             & 2             & 3             & -             & -             & -             \\
+\R{C}   & \R{2} & \R{1} & \R{2} & \R{1} & \R{1} & \R{2} & \R{1} & \R{1} \\
+Python  & 3             & 7             & 6             & 6             & 22    & 21    & -             & -             \\
+\CC             & 4             & 4             & 4             & 3             & 2             & 1             & 2             & 12    \\
 \end{tabular}
 \end{center}
 …
 Keyword clashes are accommodated by syntactic transformations using the \CFA backquote escape-mechanism:
 \begin{cfa}
 int ®`®otype®`® = 3; §\C{// make keyword an identifier}§
 double ®`®forall®`® = 3.5;
+int ®``®otype = 3; §\C{// make keyword an identifier}§
+double ®``®forall = 3.5;
 \end{cfa}
 …
 // include file uses the CFA keyword "with".
 #if ! defined( with ) §\C{// nesting ?}§
 #define with ®`®with®`® §\C{// make keyword an identifier}§
+#define with ®``®with §\C{// make keyword an identifier}§
 #define __CFA_BFD_H__
 #endif
+®#include_next <bfdlink.h> §\C{// must have internal check for multiple expansion}§
+®
+§{\color{red}\#\textbf{include\_next} <bfdlink.h>}§ §\C{// must have internal check for multiple expansion}§
 #if defined( with ) && defined( __CFA_BFD_H__ ) §\C{// reset only if set}§
 #undef with
 …
 \section{Exponentiation Operator}
 C, \CC, and Java (and many other programming languages) have no exponentiation operator\index{exponentiation!operator}\index{operator!exponentiation}, \ie $x^y$, and instead use a routine, like \Indexc{pow}, to perform the exponentiation operation.
 \CFA extends the basic operators with the exponentiation operator ©?\?©\index{?\\?@©?\?©} and ©?\=?©\index{?\\=?@©\=?©}, as in, ©x \ y© and ©x \= y©, which means $x^y$ and $x \leftarrow x^y$.
+C, \CC, and Java (and many other programming languages) have no exponentiation operator\index{exponentiation!operator}\index{operator!exponentiation}, \ie $x^y$, and instead use a routine, like \Indexc{pow(x,y)}, to perform the exponentiation operation.
+\CFA extends the basic operators with the exponentiation operator ©?®\®?©\index{?\\?@©?®\®?©} and ©?\=?©\index{?\\=?@©®\®=?©}, as in, ©x ®\® y© and ©x ®\®= y©, which means $x^y$ and $x \leftarrow x^y$.
 The priority of the exponentiation operator is between the cast and multiplicative operators, so that ©w * (int)x \ (int)y * z© is parenthesized as ©((w * (((int)x) \ ((int)y))) * z)©.
 As for \Index{division}, there are exponentiation operators for integral and floating types, including the builtin \Index{complex} types.
+There are exponentiation operators for integral and floating types, including the builtin \Index{complex} types.
 Integral exponentiation\index{exponentiation!unsigned integral} is performed with repeated multiplication\footnote{The multiplication computation is $O(\log y)$.} (or shifting if the exponent is 2).
 Overflow from large exponents or negative exponents return zero.
+Overflow for a large exponent or negative exponent returns zero.
 Floating exponentiation\index{exponentiation!floating} is performed using \Index{logarithm}s\index{exponentiation!logarithm}, so the exponent cannot be negative.
 \begin{cfa}
 …
 1 256 -64 125 ®0® 3273344365508751233 ®0® ®0® -0.015625 18.3791736799526 0.264715-1.1922i
 \end{cfa}
 Note, ©5 ®\® 32© and ©5L ®\® 64© overflow, and ©-4 ®\® -3© is a fraction but stored in an integer so all three computations generate an integral zero.
+Note, ©5 \ 32© and ©5L \ 64© overflow, and ©-4 \ -3© is a fraction but stored in an integer so all three computations generate an integral zero.
 Parenthesis are necessary for complex constants or the expression is parsed as ©1.0f+®(®2.0fi \ 3.0f®)®+2.0fi©.
 The exponentiation operator is available for all the basic types, but for user-defined types, only the integral-computation version is available.
 …
 OT ?®\®?( OT ep, unsigned long int y );
 \end{cfa}
 The user type ©T© must define multiplication, one, ©1©, and, ©*©.
+The user type ©T© must define multiplication, one (©1©), and ©*©.
 …
+\subsection{Loop Control}
+The ©for©/©while©/©do-while© loop-control allows empty or simplified ranges (see Figure~\ref{f:LoopControlExamples}).
+\begin{itemize}
+\item
+An empty conditional implies ©1©.
+\item
+The up-to range ©~©\index{~@©~©} means exclusive range [M,N).
+\item
+The up-to range ©~=©\index{~=@©~=©} means inclusive range [M,N].
+\item
+The down-to range ©-~©\index{-~@©-~©} means exclusive range [N,M).
+\item
+The down-to range ©-~=©\index{-~=@©-~=©} means inclusive range [N,M].
+\item
+©@© means put nothing in this field.
+\item
+©0© is the implicit start value;
+\item
+©1© is the implicit increment value.
+\item
+The up-to range uses ©+=© for increment;
+\item
+The down-to range uses ©-=© for decrement.
+\item
+The loop index is polymorphic in the type of the start value or comparison value when start is implicitly ©0©.
+\end{itemize}
+\begin{figure}
+%\section{\texorpdfstring{\protect\lstinline@case@ Clause}{case Clause}}
+\subsection{\texorpdfstring{\LstKeywordStyle{case} Clause}{case Clause}}
+C restricts the ©case© clause of a ©switch© statement to a single value.
+For multiple ©case© clauses associated with the same statement, it is necessary to have multiple ©case© clauses rather than multiple values.
+Requiring a ©case© clause for each value does not seem to be in the spirit of brevity normally associated with C.
+Therefore, the ©case© clause is extended with a list of values, as in:
 \begin{cquote}
+\begin{tabular}{@{}l|l@{}}
+\multicolumn{1}{c|}{loop control} & \multicolumn{1}{c}{output} \\
+\hline
+\begin{cfa}
+sout | nlOff;
+while ®()® { sout | "empty"; break; } sout | nl;
+do { sout | "empty"; break; } while ®()®; sout | nl;
+for ®()® { sout | "empty"; break; } sout | nl;
+for ( ®0® ) { sout | "A"; } sout | "zero" | nl;
+for ( ®1® ) { sout | "A"; } sout | nl;
+for ( ®10® ) { sout | "A"; } sout | nl;
+for ( ®1 ~= 10 ~ 2® ) { sout | "B"; } sout | nl;
+for ( ®10 -~= 1 ~ 2® ) { sout | "C"; } sout | nl;
+for ( ®0.5 ~ 5.5® ) { sout | "D"; } sout | nl;
+for ( ®5.5 -~ 0.5® ) { sout | "E"; } sout | nl;
+for ( ®i; 10® ) { sout | i; } sout | nl;
+for ( ®i; 1 ~= 10 ~ 2® ) { sout | i; } sout | nl;
+for ( ®i; 10 -~= 1 ~ 2® ) { sout | i; } sout | nl;
+for ( ®i; 0.5 ~ 5.5® ) { sout | i; } sout | nl;
+for ( ®i; 5.5 -~ 0.5® ) { sout | i; } sout | nl;
+for ( ®ui; 2u ~= 10u ~ 2u® ) { sout | ui; } sout | nl;
+for ( ®ui; 10u -~= 2u ~ 2u® ) { sout | ui; } sout | nl;
+enum { N = 10 };
+for ( ®N® ) { sout | "N"; } sout | nl;
+for ( ®i; N® ) { sout | i; } sout | nl;
+for ( ®i; N -~ 0® ) { sout | i; } sout | nl;
+const int start = 3, comp = 10, inc = 2;
+for ( ®i; start ~ comp ~ inc + 1® ) { sout | i; } sout | nl;
+for ( ®i; 1 ~ @® ) { if ( i > 10 ) break;
+        sout | i; } sout | nl;
+for ( ®i; 10 -~ @® ) { if ( i < 0 ) break;
+        sout | i; } sout | nl;
+for ( ®i; 2 ~ @ ~ 2® ) { if ( i > 10 ) break;
+        sout | i; } sout | nl;
+for ( ®i; 2.1 ~ @ ~ @® ) { if ( i > 10.5 ) break;
+        sout | i; i += 1.7; } sout | nl;
+for ( ®i; 10 -~ @ ~ 2® ) { if ( i < 0 ) break;
+        sout | i; } sout | nl;
+for ( ®i; 12.1 ~ @ ~ @® ) { if ( i < 2.5 ) break;
+        sout | i; i -= 1.7; } sout | nl;
+for ( ®i; 5 : j; -5 ~ @® ) { sout | i | j; } sout | nl;
+for ( ®i; 5 : j; -5 -~ @® ) { sout | i | j; } sout | nl;
+for ( ®i; 5 : j; -5 ~ @ ~ 2® ) { sout | i | j; } sout | nl;
+for ( ®i; 5 : j; -5 -~ @ ~ 2® ) { sout | i | j; } sout | nl;
+for ( ®j; -5 ~ @ : i; 5® ) { sout | i | j; } sout | nl;
+for ( ®j; -5 -~ @ : i; 5® ) { sout | i | j; } sout | nl;
+for ( ®j; -5 ~ @ ~ 2 : i; 5® ) { sout | i | j; } sout | nl;
+for ( ®j; -5 -~ @ ~ 2 : i; 5® ) { sout | i | j; } sout | nl;
+for ( ®j; -5 -~ @ ~ 2 : i; 5 : k; 1.5 ~ @® ) {
+        sout | i | j | k; } sout | nl;
+for ( ®j; -5 -~ @ ~ 2 : k; 1.5 ~ @ : i; 5® ) {
+        sout | i | j | k; } sout | nl;
+for ( ®k; 1.5 ~ @ : j; -5 -~ @ ~ 2 : i; 5® ) {
+        sout | i | j | k; } sout | nl;
+\begin{tabular}{@{}l@{\hspace{3em}}l@{\hspace{2em}}l@{}}
+\multicolumn{1}{c@{\hspace{3em}}}{\textbf{\CFA}}        & \multicolumn{1}{c@{\hspace{2em}}}{\textbf{C}} \\
+\begin{cfa}
+switch ( i ) {
+  case ®1, 3, 5®:
+        ...
+  case ®2, 4, 6®:
+        ...
+}
 \end{cfa}
+&
 \begin{cfa}
+empty
+empty
+empty
+zero
+A
+A A A A A A A A A A
+B B B B B
+C C C C C
+D D D D D
+E E E E E
+1 2 3 4 5 6 7 8 9
+3 5 7 9
+8 6 4 2
+.5 1.5 2.5 3.5 4.5
+.5 4.5 3.5 2.5 1.5
+4 6 8 10
+8 6 4 2
+N N N N N N N N N N
+1 2 3 4 5 6 7 8 9
+9 8 7 6 5 4 3 2 1
+6 9
+2 3 4 5 6 7 8 9 10
+9 8 7 6 5 4 3 2 1 0
+4 6 8 10
+.1 3.8 5.5 7.2 8.9
+8 6 4 2 0
+.1 10.4 8.7 7 5.3 3.6
+-5 1 -4 2 -3 3 -2 4 -1
+-5 1 -6 2 -7 3 -8 4 -9
+-5 1 -3 2 -1 3 1 4 3
+-5 1 -7 2 -9 3 -11 4 -13
+-5 1 -4 2 -3 3 -2 4 -1
+-5 1 -6 2 -7 3 -8 4 -9
+-5 1 -3 2 -1 3 1 4 3
+-5 1 -7 2 -9 3 -11 4 -13
+-5 1.5 1 -7 2.5 2 -9 3.5 3 -11 4.5 4 -13 5.5
+-5 1.5 1 -7 2.5 2 -9 3.5 3 -11 4.5 4 -13 5.5
+-5 1.5 1 -7 2.5 2 -9 3.5 3 -11 4.5 4 -13 5.5
+switch ( i ) {
+  case 1: case 3 : case 5:
+        ...
+  case 2: case 4 : case 6:
+        ...
+}
+\end{cfa}
+&
+\begin{cfa}
+// odd values
+// even values
 \end{cfa}
 \end{tabular}
 \end{cquote}
+\caption{Loop Control Examples}
+\label{f:LoopControlExamples}
+\end{figure}
+In addition, subranges are allowed to specify case values.\footnote{
+gcc has the same mechanism but awkward syntax, \lstinline@2 ...42@, because a space is required after a number, otherwise the period is a decimal point.}
+\begin{cfa}
+switch ( i ) {
+  case ®1~5:® §\C{// 1, 2, 3, 4, 5}§
+        ...
+  case ®10~15:® §\C{// 10, 11, 12, 13, 14, 15}§
+        ...
+}
+\end{cfa}
+Lists of subranges are also allowed.
+\begin{cfa}
+case ®1~5, 12~21, 35~42®:
+\end{cfa}
 …
+%\section{\texorpdfstring{\protect\lstinline@case@ Clause}{case Clause}}
+\subsection{\texorpdfstring{\LstKeywordStyle{case} Statement}{case Statement}}
+C restricts the ©case© clause of a ©switch© statement to a single value.
+For multiple ©case© clauses associated with the same statement, it is necessary to have multiple ©case© clauses rather than multiple values.
+Requiring a ©case© clause for each value does not seem to be in the spirit of brevity normally associated with C.
+Therefore, the ©case© clause is extended with a list of values, as in:
+\begin{cquote}
+\begin{tabular}{@{}l@{\hspace{3em}}l@{\hspace{2em}}l@{}}
+\multicolumn{1}{c@{\hspace{3em}}}{\textbf{\CFA}}        & \multicolumn{1}{c@{\hspace{2em}}}{\textbf{C}} \\
+\begin{cfa}
+switch ( i ) {
+  case ®1, 3, 5®:
+\subsection{Non-terminating and Labelled \texorpdfstring{\LstKeywordStyle{fallthrough}}{Non-terminating and Labelled fallthrough}}
+The ©fallthrough© clause may be non-terminating within a ©case© clause or have a target label to common code from multiple case clauses.
+\begin{center}
+\begin{tabular}{@{}lll@{}}
+\begin{cfa}
+choose ( ... ) {
+  case 3:
+        if ( ... ) {
+                ... ®fallthru;® // goto case 4
+        } else {
+                ...
+        }
+        // implicit break
+  case 4:
+\end{cfa}
+&
+\begin{cfa}
+choose ( ... ) {
+  case 3:
+        ... ®fallthrough common;®
+  case 4:
+        ... ®fallthrough common;®
+  ®common:® // below fallthrough
+                          // at case-clause level
+        ...     // common code for cases 3/4
+        // implicit break
+  case 4:
+\end{cfa}
+&
+\begin{cfa}
+choose ( ... ) {
+  case 3:
+        choose ( ... ) {
+          case 4:
+                for ( ... ) {
+                        // multi-level transfer
+                        ... ®fallthru common;®
+                }
+                ...
+        }
         ...
+  case ®2, 4, 6®:
+        ...
+}
+  ®common:® // below fallthrough
+                          // at case-clause level
+\end{cfa}
+\end{tabular}
+\end{center}
+The target label must be below the ©fallthrough© and may not be nested in a control structure, and
+the target label must be at the same or higher level as the containing ©case© clause and located at
+the same level as a ©case© clause; the target label may be case ©default©, but only associated
+with the current ©switch©/©choose© statement.
+\subsection{Loop Control}
+The ©for©/©while©/©do-while© loop-control allows empty or simplified ranges (see Figure~\ref{f:LoopControlExamples}).
+\begin{itemize}
+\item
+The loop index is polymorphic in the type of the comparison value N (when the start value is implicit) or the start value M.
+\item
+An empty conditional implies comparison value of ©1© (true).
+\item
+A comparison N is implicit up-to exclusive range [0,N©®)®©.
+\item
+A comparison ©=© N is implicit up-to inclusive range [0,N©®]®©.
+\item
+The up-to range M ©~©\index{~@©~©} N means exclusive range [M,N©®)®©.
+\item
+The up-to range M ©~=©\index{~=@©~=©} N means inclusive range [M,N©®]®©.
+\item
+The down-to range M ©-~©\index{-~@©-~©} N means exclusive range [N,M©®)®©.
+\item
+The down-to range M ©-~=©\index{-~=@©-~=©} N means inclusive range [N,M©®]®©.
+\item
+©0© is the implicit start value;
+\item
+©1© is the implicit increment value.
+\item
+The up-to range uses operator ©+=© for increment;
+\item
+The down-to range uses operator ©-=© for decrement.
+\item
+©@© means put nothing in this field.
+\item
+©:© means start another index.
+\end{itemize}
+\begin{figure}
+\begin{tabular}{@{}l|l@{}}
+\multicolumn{1}{c|}{loop control} & \multicolumn{1}{c}{output} \\
+\hline
+\begin{cfa}[xleftmargin=0pt]
+while ®()® { sout | "empty"; break; }
+do { sout | "empty"; break; } while ®()®;
+for ®()® { sout | "empty"; break; }
+for ( ®0® ) { sout | "A"; } sout | "zero";
+for ( ®1® ) { sout | "A"; }
+for ( ®10® ) { sout | "A"; }
+for ( ®= 10® ) { sout | "A"; }
+for ( ®1 ~= 10 ~ 2® ) { sout | "B"; }
+for ( ®10 -~= 1 ~ 2® ) { sout | "C"; }
+for ( ®0.5 ~ 5.5® ) { sout | "D"; }
+for ( ®5.5 -~ 0.5® ) { sout | "E"; }
+for ( ®i; 10® ) { sout | i; }
+for ( ®i; = 10® ) { sout | i; }
+for ( ®i; 1 ~= 10 ~ 2® ) { sout | i; }
+for ( ®i; 10 -~= 1 ~ 2® ) { sout | i; }
+for ( ®i; 0.5 ~ 5.5® ) { sout | i; }
+for ( ®i; 5.5 -~ 0.5® ) { sout | i; }
+for ( ®ui; 2u ~= 10u ~ 2u® ) { sout | ui; }
+for ( ®ui; 10u -~= 2u ~ 2u® ) { sout | ui; }
+enum { N = 10 };
+for ( ®N® ) { sout | "N"; }
+for ( ®i; N® ) { sout | i; }
+for ( ®i; N -~ 0® ) { sout | i; }
+const int start = 3, comp = 10, inc = 2;
+for ( ®i; start ~ comp ~ inc + 1® ) { sout | i; }
+for ( i; 1 ~ ®@® ) { if ( i > 10 ) break; sout | i; }
+for ( i; 10 -~ ®@® ) { if ( i < 0 ) break; sout | i; }
+for ( i; 2 ~ ®@® ~ 2 ) { if ( i > 10 ) break; sout | i; }
+for ( i; 2.1 ~ ®@® ~ ®@® ) { if ( i > 10.5 ) break; sout | i; i += 1.7; }
+for ( i; 10 -~ ®@® ~ 2 ) { if ( i < 0 ) break; sout | i; }
+for ( i; 12.1 ~ ®@® ~ ®@® ) { if ( i < 2.5 ) break; sout | i; i -= 1.7; }
+for ( i; 5 ®:® j; -5 ~ @ ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 -~ @ ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 ~ @ ~ 2 ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 -~ @ ~ 2 ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 ~ @ ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 -~ @ ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 ~ @ ~ 2 ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 -~ @ ~ 2 ) { sout | i | j; }
+for ( i; 5 ®:® j; -5 -~ @ ~ 2 ®:® k; 1.5 ~ @ ) { sout | i | j | k; }
+for ( i; 5 ®:® j; -5 -~ @ ~ 2 ®:® k; 1.5 ~ @ ) { sout | i | j | k; }
+for ( i; 5 ®:® k; 1.5 ~ @ ®:® j; -5 -~ @ ~ 2 ) { sout | i | j | k; }
 \end{cfa}
+&
 \begin{cfa}
+switch ( i ) {
+  case 1: case 3 : case 5:
+        ...
+  case 2: case 4 : case 6:
+        ...
+}
+\end{cfa}
+&
+\begin{cfa}
+// odd values
+// even values
+empty
+empty
+empty
+zero
+A
+A A A A A A A A A A
+A A A A A A A A A A A
+B B B B B
+C C C C C
+D D D D D
+E E E E E
+1 2 3 4 5 6 7 8 9
+1 2 3 4 5 6 7 8 9 10
+3 5 7 9
+8 6 4 2
+.5 1.5 2.5 3.5 4.5
+.5 4.5 3.5 2.5 1.5
+4 6 8 10
+8 6 4 2
+N N N N N N N N N N
+1 2 3 4 5 6 7 8 9
+9 8 7 6 5 4 3 2 1
+6 9
+2 3 4 5 6 7 8 9 10
+9 8 7 6 5 4 3 2 1 0
+4 6 8 10
+.1 3.8 5.5 7.2 8.9
+8 6 4 2 0
+.1 10.4 8.7 7. 5.3 3.6
+-5 1 -4 2 -3 3 -2 4 -1
+-5 1 -6 2 -7 3 -8 4 -9
+-5 1 -3 2 -1 3 1 4 3
+-5 1 -7 2 -9 3 -11 4 -13
+-5 1 -4 2 -3 3 -2 4 -1
+-5 1 -6 2 -7 3 -8 4 -9
+-5 1 -3 2 -1 3 1 4 3
+-5 1 -7 2 -9 3 -11 4 -13
+-5 1.5 1 -7 2.5 2 -9 3.5 3 -11 4.5 4 -13 5.5
+-5 1.5 1 -7 2.5 2 -9 3.5 3 -11 4.5 4 -13 5.5
+-5 1.5 1 -7 2.5 2 -9 3.5 3 -11 4.5 4 -13 5.5
 \end{cfa}
 \end{tabular}
+\end{cquote}
+In addition, subranges are allowed to specify case values.\footnote{
+gcc has the same mechanism but awkward syntax, \lstinline@2 ...42@, because a space is required after a number, otherwise the period is a decimal point.}
+\begin{cfa}
+switch ( i ) {
+  case ®1~5:® §\C{// 1, 2, 3, 4, 5}§
+        ...
+  case ®10~15:® §\C{// 10, 11, 12, 13, 14, 15}§
+        ...
+}
+\end{cfa}
+Lists of subranges are also allowed.
+\begin{cfa}
+case ®1~5, 12~21, 35~42®:
+\end{cfa}
+\caption{Loop Control Examples}
+\label{f:LoopControlExamples}
+\end{figure}
 % for ()  => for ( ;; )
 …
 hence, names in these include files are not mangled\index{mangling!name} (see~\VRef{s:Interoperability}).
 All other C header files must be explicitly wrapped in ©extern "C"© to prevent name mangling.
 For \Index*[C++]{\CC{}}, the name-mangling issue is often handled internally in many C header-files through checks for preprocessor variable ©__cplusplus©, which adds appropriate ©extern "C"© qualifiers.
+This approach is different from \Index*[C++]{\CC{}} where the name-mangling issue is handled internally in C header-files through checks for preprocessor variable ©__cplusplus©, which adds appropriate ©extern "C"© qualifiers.
 …
 The storage-management routines extend their C equivalents by overloading, alternate names, providing shallow type-safety, and removing the need to specify the allocation size for non-array types.
 Storage management provides the following capabilities:
+C storage management provides the following capabilities:
 \begin{description}
 \item[fill]
 after allocation the storage is filled with a specified character.
+\item[filled]
+after allocation with a specified character or value.
 \item[resize]
 an existing allocation is decreased or increased in size.
 In either case, new storage may or may not be allocated and, if there is a new allocation, as much data from the existing allocation is copied.
+an existing allocation to decreased or increased its size.
+In either case, new storage may or may not be allocated and, if there is a new allocation, as much data from the existing allocation is copied into the new allocation.
 For an increase in storage size, new storage after the copied data may be filled.
 \item[alignment]
 an allocation starts on a specified memory boundary, \eg, an address multiple of 64 or 128 for cache-line purposes.
+\item[align]
+an allocation on a specified memory boundary, \eg, an address multiple of 64 or 128 for cache-line purposes.
 \item[array]
 the allocation size is scaled to the specified number of array elements.
 An array may be filled, resized, or aligned.
 \end{description}
+The table shows allocation routines supporting different combinations of storage-management capabilities:
+\begin{center}
+\begin{tabular}{@{}r|r|l|l|l|l@{}}
+\VRef[Table]{t:AllocationVersusCapabilities} shows allocation routines supporting different combinations of storage-management capabilities.
+\begin{table}
+\centering
+\begin{minipage}{0.75\textwidth}
+\begin{tabular}{@{}r|l|l|l|l|l@{}}
 \multicolumn{1}{c}{}&           & \multicolumn{1}{c|}{fill}     & resize        & alignment     & array \\
 \hline
 C               & ©malloc©                      & no                    & no            & no            & no    \\
                 & ©calloc©                      & yes (0 only)  & no            & no            & yes   \\
                 & ©realloc©                     & no/copy               & yes           & no            & no    \\
+                & ©realloc©                     & copy                  & yes           & no            & no    \\
                 & ©memalign©            & no                    & no            & yes           & no    \\
+                & ©aligned_alloc©\footnote{Same as ©memalign© but size is an integral multiple of alignment, which is universally ignored.}
+                                                        & no                    & no            & yes           & no    \\
                 & ©posix_memalign©      & no                    & no            & yes           & no    \\
+                & ©valloc©                      & no                    & no            & yes (page size)& no   \\
+                & ©pvalloc©\footnote{Same as ©valloc© but rounds size to multiple of page size.}
+                                                        & no                    & no            & yes (page size)& no   \\
 \hline
+C11             & ©aligned_alloc©       & no                    & no            & yes           & no    \\
+\hline
+\CFA    & ©alloc©                       & no/copy/yes   & no/yes        & no            & yes   \\
+                & ©align_alloc©         & no/yes                & no            & yes           & yes   \\
+\CFA    & ©cmemalign©           & yes (0 only)  & no            & yes           & yes   \\
+                & ©realloc©                     & copy                  & yes           & yes           & no    \\
+                & ©alloc©                       & no                    & yes           & no            & yes   \\
+                & ©alloc_set©           & yes                   & yes           & no            & yes   \\
+                & ©alloc_align©         & no                    & yes           & yes           & yes   \\
+                & ©alloc_align_set©     & yes                   & yes           & yes           & yes   \\
 \end{tabular}
+\end{center}
+It is impossible to resize with alignment because the underlying ©realloc© allocates storage if more space is needed, and it does not honour alignment from the original allocation.
+\end{minipage}
+\caption{Allocation Routines versus Storage-Management Capabilities}
+\label{t:AllocationVersusCapabilities}
+\end{table}
+\CFA memory management extends the type safety of all allocations by using the type of the left-hand-side type to determine the allocation size and return a matching type for the new storage.
+Type-safe allocation is provided for all C allocation routines and new \CFA allocation routines, \eg in
+\begin{cfa}
+int * ip = (int *)malloc( sizeof(int) );                §\C{// C}§
+int * ip = malloc();                                                    §\C{// \CFA type-safe version of C malloc}§
+int * ip = alloc();                                                             §\C{// \CFA type-safe uniform alloc}§
+\end{cfa}
+the latter two allocations determine the allocation size from the type of ©p© (©int©) and cast the pointer to the allocated storage to ©int *©.
+\CFA memory management extends allocation safety by implicitly honouring all alignment requirements, \eg in
+\begin{cfa}
+struct S { int i; } __attribute__(( aligned( 128 ) )); // cache-line alignment
+S * sp = malloc();                                                              §\C{// honour type alignment}§
+\end{cfa}
+the storage allocation is implicitly aligned to 128 rather than the default 16.
+The alignment check is performed at compile time so there is no runtime cost.
+\CFA memory management extends the resize capability with the notion of \newterm{sticky properties}.
+Hence, initial allocation capabilities are remembered and maintained when resize requires copying.
+For example, an initial alignment and fill capability are preserved during a resize copy so the copy has the same alignment and extended storage is filled.
+Without sticky properties it is dangerous to use ©realloc©, resulting in an idiom of manually performing the reallocation to maintain correctness.
+\CFA memory management extends allocation to support constructors for initialization of allocated storage, \eg in
+\begin{cfa}
+struct S { int i; };                                                    §\C{// cache-line aglinment}§
+void ?{}( S & s, int i ) { s.i = i; }
+// assume ?|? operator for printing an S
+S & sp = *®new®( 3 );                                                   §\C{// call constructor after allocation}§
+sout | sp.i;
+®delete®( &sp );
+S * spa = ®anew®( 10, 5 );                                              §\C{// allocate array and initialize each array element}§
+for ( i; 10 ) sout | spa[i] | nonl;
+sout | nl;
+®adelete®( 10, spa );
+\end{cfa}
+Allocation routines ©new©/©anew© allocate a variable/array and initialize storage using the allocated type's constructor.
+Note, the matching deallocation routines ©delete©/©adelete©.
 \leavevmode
 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
-// C unsafe allocation
 extern "C" {
+void * malloc( size_t size );§\indexc{memset}§
+void * calloc( size_t dim, size_t size );§\indexc{calloc}§
+void * realloc( void * ptr, size_t size );§\indexc{realloc}§
+void * memalign( size_t align, size_t size );§\indexc{memalign}§
+int posix_memalign( void ** ptr, size_t align, size_t size );§\indexc{posix_memalign}§
+// C unsafe initialization/copy
+void * memset( void * dest, int c, size_t size );
+void * memcpy( void * dest, const void * src, size_t size );
+}
+        // C unsafe allocation
+        void * malloc( size_t size );§\indexc{malloc}§
+        void * calloc( size_t dim, size_t size );§\indexc{calloc}§
+        void * realloc( void * ptr, size_t size );§\indexc{realloc}§
+        void * memalign( size_t align, size_t size );§\indexc{memalign}§
+        void * aligned_alloc( size_t align, size_t size );§\indexc{aligned_alloc}§
+        int posix_memalign( void ** ptr, size_t align, size_t size );§\indexc{posix_memalign}§
+        void * cmemalign( size_t alignment, size_t noOfElems, size_t elemSize );§\indexc{cmemalign}§ // CFA
+        // C unsafe initialization/copy
+        void * memset( void * dest, int c, size_t size );§\indexc{memset}§
+        void * memcpy( void * dest, const void * src, size_t size );§\indexc{memcpy}§
+}
+void * realloc( void * oaddr, size_t nalign, size_t size ); // CFA heap
 forall( dtype T | sized(T) ) {
 // §\CFA§ safe equivalents, i.e., implicit size specification
+        // §\CFA§ safe equivalents, i.e., implicit size specification
         T * malloc( void );
         T * calloc( size_t dim );
         T * realloc( T * ptr, size_t size );
         T * memalign( size_t align );
+        T * cmemalign( size_t align, size_t dim  );
         T * aligned_alloc( size_t align );
         int posix_memalign( T ** ptr, size_t align );
 // §\CFA§ safe general allocation, fill, resize, array
+        // §\CFA§ safe general allocation, fill, resize, alignment, array
         T * alloc( void );§\indexc{alloc}§
-        T * alloc( char fill );
         T * alloc( size_t dim );
-        T * alloc( size_t dim, char fill );
         T * alloc( T ptr[], size_t dim );
+        T * alloc( T ptr[], size_t dim, char fill );
+// §\CFA§ safe general allocation, align, fill, array
+        T * align_alloc( size_t align );
+        T * align_alloc( size_t align, char fill );
+        T * align_alloc( size_t align, size_t dim );
+        T * align_alloc( size_t align, size_t dim, char fill );
+// §\CFA§ safe initialization/copy, i.e., implicit size specification
+        T * memset( T * dest, char c );§\indexc{memset}§
+        T * alloc_set( char fill );§\indexc{alloc_set}§
+        T * alloc_set( T fill );
+        T * alloc_set( size_t dim, char fill );
+        T * alloc_set( size_t dim, T fill );
+        T * alloc_set( size_t dim, const T fill[] );
+        T * alloc_set( T ptr[], size_t dim, char fill );
+        T * alloc_align( size_t align );
+        T * alloc_align( size_t align, size_t dim );
+        T * alloc_align( T ptr[], size_t align ); // aligned realloc array
+        T * alloc_align( T ptr[], size_t align, size_t dim ); // aligned realloc array
+        T * alloc_align_set( size_t align, char fill );
+        T * alloc_align_set( size_t align, T fill );
+        T * alloc_align_set( size_t align, size_t dim, char fill );
+        T * alloc_align_set( size_t align, size_t dim, T fill );
+        T * alloc_align_set( size_t align, size_t dim, const T fill[] );
+        T * alloc_align_set( T ptr[], size_t align, size_t dim, char fill );
+        // §\CFA§ safe initialization/copy, i.e., implicit size specification
+        T * memset( T * dest, char fill );§\indexc{memset}§
         T * memcpy( T * dest, const T * src );§\indexc{memcpy}§
+// §\CFA§ safe initialization/copy array
         T * amemset( T dest[], char c, size_t dim );
+        // §\CFA§ safe initialization/copy, i.e., implicit size specification, array types
+        T * amemset( T dest[], char fill, size_t dim );
         T * amemcpy( T dest[], const T src[], size_t dim );
+}
 // §\CFA§ allocation/deallocation and constructor/destructor
 forall( dtype T | sized(T), ttype Params | { void ?{}( T *, Params ); } ) T * new( Params p );§\indexc{new}§
 forall( dtype T | { void ^?{}( T * ); } ) void delete( T * ptr );§\indexc{delete}§
 forall( dtype T, ttype Params | { void ^?{}( T * ); void delete( Params ); } )
+// §\CFA§ allocation/deallocation and constructor/destructor, non-array types
+forall( dtype T | sized(T), ttype Params | { void ?{}( T &, Params ); } ) T * new( Params p );§\indexc{new}§
+forall( dtype T | sized(T) | { void ^?{}( T & ); } ) void delete( T * ptr );§\indexc{delete}§
+forall( dtype T, ttype Params | sized(T) | { void ^?{}( T & ); void delete( Params ); } )
   void delete( T * ptr, Params rest );
 // §\CFA§ allocation/deallocation and constructor/destructor, array
 forall( dtype T | sized(T), ttype Params | { void ?{}( T *, Params ); } ) T * anew( size_t dim, Params p );§\indexc{anew}§
 forall( dtype T | sized(T) | { void ^?{}( T * ); } ) void adelete( size_t dim, T arr[] );§\indexc{adelete}§
 forall( dtype T | sized(T) | { void ^?{}( T * ); }, ttype Params | { void adelete( Params ); } )
+// §\CFA§ allocation/deallocation and constructor/destructor, array types
+forall( dtype T | sized(T), ttype Params | { void ?{}( T &, Params ); } ) T * anew( size_t dim, Params p );§\indexc{anew}§
+forall( dtype T | sized(T) | { void ^?{}( T & ); } ) void adelete( size_t dim, T arr[] );§\indexc{adelete}§
+forall( dtype T | sized(T) | { void ^?{}( T & ); }, ttype Params | { void adelete( Params ); } )
   void adelete( size_t dim, T arr[], Params rest );
 \end{cfa}

driver/cfa.cc

-              rb7d6a36
+              r6a490b2
         } // if
+        string preludedir;
         switch(path) {
         case Installed   : Putenv( argv, "--prelude-dir=" + libdir ); break;
         case BuildTree   : Putenv( argv, "--prelude-dir=" + libdir + "/prelude" ); break;
         case Distributed : Putenv( argv, "--prelude-dir=" + dir(argv[0]) ); break;
+        case Installed   : preludedir = libdir; break;
+        case BuildTree   : preludedir = libdir + "/prelude"; break;
+        case Distributed : preludedir = dir(argv[0]); break;
+        }
+        Putenv( argv, "--prelude-dir=" + preludedir );
+        args[nargs++] = "-include";
+        args[nargs++] = (*new string(preludedir + "/defines.hfa")).c_str();
         for ( int i = 0; i < nlibs; i += 1 ) {                          // copy non-user libraries after all user libraries

libcfa/Makefile.in

rb7d6a36	r6a490b2
106	106	configure.lineno config.status.lineno
107	107	mkinstalldirs = $(install_sh) -d
	108	CONFIG_HEADER = $(top_builddir)/prelude/defines.hfa
108	109	CONFIG_CLEAN_FILES =
109	110	CONFIG_CLEAN_VPATH_FILES =

libcfa/configure

-              rb7d6a36
+              r6a490b2
 enable_distcc
 with_cfa_name
+enable_static
 enable_shared
-enable_static
 with_pic
 enable_fast_install
 …
   --disable-silent-rules  verbose build output (undo: "make V=0")
   --enable-distcc     whether or not to enable distributed compilation
+  --enable-static[=PKGS]  build static libraries [default=no]
   --enable-shared[=PKGS]  build shared libraries [default=yes]
-  --enable-static[=PKGS]  build static libraries [default=yes]
   --enable-fast-install[=PKGS]
                           optimize for fast installation [default=yes]
 …
 } # ac_fn_cxx_try_link
+# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_c_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+               { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+( $as_echo "## --------------------------------------- ##
+## Report this to cforall@plg.uwaterloo.ca ##
+## --------------------------------------- ##"
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+               { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+} # ac_fn_c_check_header_mongrel
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 …
 # Set options
+# Check whether --enable-static was given.
+if test "${enable_static+set}" = set; then :
+  enableval=$enable_static; p=${PACKAGE-default}
+    case $enableval in
+    yes) enable_static=yes ;;
+    no) enable_static=no ;;
+    *)
+     enable_static=no
+      # Look at the argument we got.  We use all the common list separators.
+      lt_save_ifs=$IFS; IFS=$IFS$PATH_SEPARATOR,
+      for pkg in $enableval; do
+        IFS=$lt_save_ifs
+        if test "X$pkg" = "X$p"; then
+          enable_static=yes
+        fi
+      done
+      IFS=$lt_save_ifs
+      ;;
+    esac
+else
+  enable_static=no
+fi
 …
 fi
-  # Check whether --enable-static was given.
-if test "${enable_static+set}" = set; then :
-  enableval=$enable_static; p=${PACKAGE-default}
-    case $enableval in
-    yes) enable_static=yes ;;
-    no) enable_static=no ;;
-    *)
-     enable_static=no
-      # Look at the argument we got.  We use all the common list separators.
-      lt_save_ifs=$IFS; IFS=$IFS$PATH_SEPARATOR,
-      for pkg in $enableval; do
-        IFS=$lt_save_ifs
-        if test "X$pkg" = "X$p"; then
-          enable_static=yes
-        fi
-      done
-      IFS=$lt_save_ifs
-      ;;
-    esac
-else
-  enable_static=yes
-fi
 …
+for ac_header in linux/io_uring.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "linux/io_uring.h" "ac_cv_header_linux_io_uring_h" "$ac_includes_default"
+if test "x$ac_cv_header_linux_io_uring_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_LINUX_IO_URING_H 1
+_ACEOF
+fi
+done
+for ac_func in preadv2 pwritev2
+do :
+  as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
+ac_fn_c_check_func "$LINENO" "$ac_func" "$as_ac_var"
+if eval test \"x\$"$as_ac_var"\" = x"yes"; then :
+  cat >>confdefs.h <<_ACEOF
+#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
+_ACEOF
+fi
+done
 ac_config_files="$ac_config_files Makefile src/Makefile prelude/Makefile"
+ac_config_headers="$ac_config_headers prelude/defines.hfa"
 …
 test "x$exec_prefix" = xNONE && exec_prefix='${prefix}'
+# Transform confdefs.h into DEFS.
+# Protect against shell expansion while executing Makefile rules.
+# Protect against Makefile macro expansion.
+#
+# If the first sed substitution is executed (which looks for macros that
+# take arguments), then branch to the quote section.  Otherwise,
+# look for a macro that doesn't take arguments.
+ac_script='
+:mline
+/\\$/{
+ N
+ s,\\\n,,
+ b mline
+}
+t clear
+:clear
+s/^[     ]*#[    ]*define[       ][      ]*\([^  (][^    (]*([^)]*)\)[   ]*\(.*\)/-D\1=\2/g
+t quote
+s/^[     ]*#[    ]*define[       ][      ]*\([^  ][^     ]*\)[   ]*\(.*\)/-D\1=\2/g
+t quote
+b any
+:quote
+s/[      `~#$^&*(){}\\|;'\''"<>?]/\\&/g
+s/\[/\\&/g
+s/\]/\\&/g
+s/\$/$$/g
+H
+:any
+${
+        g
+        s/^\n//
+        s/\n/ /g
+        p
+}
+'
+DEFS=`sed -n "$ac_script" confdefs.h`
+DEFS=-DHAVE_CONFIG_H
 ac_libobjs=
 …
 esac
+case $ac_config_headers in *"
+"*) set x $ac_config_headers; shift; ac_config_headers=$*;;
+esac
 …
 # Files that config.status was made for.
 config_files="$ac_config_files"
+config_headers="$ac_config_headers"
 config_commands="$ac_config_commands"
 …
       --file=FILE[:TEMPLATE]
                    instantiate the configuration file FILE
+      --header=FILE[:TEMPLATE]
+                   instantiate the configuration header FILE
 Configuration files:
 $config_files
+Configuration headers:
+$config_headers
 Configuration commands:
 …
     as_fn_append CONFIG_FILES " '$ac_optarg'"
     ac_need_defaults=false;;
+  --he | --h |  --help | --hel | -h )
+  --header | --heade | --head | --hea )
+    $ac_shift
+    case $ac_optarg in
+    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    esac
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    ac_need_defaults=false;;
+  --he | --h)
+    # Conflict between --help and --header
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
+  --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
 …
 macro_version='`$ECHO "$macro_version" | $SED "$delay_single_quote_subst"`'
 macro_revision='`$ECHO "$macro_revision" | $SED "$delay_single_quote_subst"`'
+enable_static='`$ECHO "$enable_static" | $SED "$delay_single_quote_subst"`'
 enable_shared='`$ECHO "$enable_shared" | $SED "$delay_single_quote_subst"`'
-enable_static='`$ECHO "$enable_static" | $SED "$delay_single_quote_subst"`'
 pic_mode='`$ECHO "$pic_mode" | $SED "$delay_single_quote_subst"`'
 enable_fast_install='`$ECHO "$enable_fast_install" | $SED "$delay_single_quote_subst"`'
 …
     "src/Makefile") CONFIG_FILES="$CONFIG_FILES src/Makefile" ;;
     "prelude/Makefile") CONFIG_FILES="$CONFIG_FILES prelude/Makefile" ;;
+    "prelude/defines.hfa") CONFIG_HEADERS="$CONFIG_HEADERS prelude/defines.hfa" ;;
   *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
 …
 if $ac_need_defaults; then
   test "${CONFIG_FILES+set}" = set || CONFIG_FILES=$config_files
+  test "${CONFIG_HEADERS+set}" = set || CONFIG_HEADERS=$config_headers
   test "${CONFIG_COMMANDS+set}" = set || CONFIG_COMMANDS=$config_commands
 fi
 …
 fi # test -n "$CONFIG_FILES"
+eval set X "  :F $CONFIG_FILES      :C $CONFIG_COMMANDS"
+# Set up the scripts for CONFIG_HEADERS section.
+# No need to generate them if there are no CONFIG_HEADERS.
+# This happens for instance with `./config.status Makefile'.
+if test -n "$CONFIG_HEADERS"; then
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+BEGIN {
+_ACEOF
+# Transform confdefs.h into an awk script `defines.awk', embedded as
+# here-document in config.status, that substitutes the proper values into
+# config.h.in to produce config.h.
+# Create a delimiter string that does not exist in confdefs.h, to ease
+# handling of long lines.
+ac_delim='%!_!# '
+for ac_last_try in false false :; do
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
+    break
+  elif $ac_last_try; then
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+  else
+    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
+  fi
+done
+# For the awk script, D is an array of macro values keyed by name,
+# likewise P contains macro parameters if any.  Preserve backslash
+# newline sequences.
+ac_word_re=[_$as_cr_Letters][_$as_cr_alnum]*
+sed -n '
+s/.\{148\}/&'"$ac_delim"'/g
+t rset
+:rset
+s/^[     ]*#[    ]*define[       ][      ]*/ /
+t def
+d
+:def
+s/\\$//
+t bsnl
+s/["\\]/\\&/g
+s/^ \('"$ac_word_re"'\)\(([^()]*)\)[     ]*\(.*\)/P["\1"]="\2"\
+D["\1"]=" \3"/p
+s/^ \('"$ac_word_re"'\)[         ]*\(.*\)/D["\1"]=" \2"/p
+d
+:bsnl
+s/["\\]/\\&/g
+s/^ \('"$ac_word_re"'\)\(([^()]*)\)[     ]*\(.*\)/P["\1"]="\2"\
+D["\1"]=" \3\\\\\\n"\\/p
+t cont
+s/^ \('"$ac_word_re"'\)[         ]*\(.*\)/D["\1"]=" \2\\\\\\n"\\/p
+t cont
+d
+:cont
+n
+s/.\{148\}/&'"$ac_delim"'/g
+t clear
+:clear
+s/\\$//
+t bsnlc
+s/["\\]/\\&/g; s/^/"/; s/$/"/p
+d
+:bsnlc
+s/["\\]/\\&/g; s/^/"/; s/$/\\\\\\n"\\/p
+b cont
+' <confdefs.h | sed '
+s/'"$ac_delim"'/"\\\
+"/g' >>$CONFIG_STATUS || ac_write_fail=1
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+  for (key in D) D_is_set[key] = 1
+  FS = ""
+}
+/^[\t ]*#[\t ]*(define|undef)[\t ]+$ac_word_re([\t (]|\$)/ {
+  line = \$ 0
+  split(line, arg, " ")
+  if (arg[1] == "#") {
+    defundef = arg[2]
+    mac1 = arg[3]
+  } else {
+    defundef = substr(arg[1], 2)
+    mac1 = arg[2]
+  }
+  split(mac1, mac2, "(") #)
+  macro = mac2[1]
+  prefix = substr(line, 1, index(line, defundef) - 1)
+  if (D_is_set[macro]) {
+    # Preserve the white space surrounding the "#".
+    print prefix "define", macro P[macro] D[macro]
+    next
+  } else {
+    # Replace #undef with comments.  This is necessary, for example,
+    # in the case of _POSIX_SOURCE, which is predefined and required
+    # on some systems where configure will not decide to define it.
+    if (defundef == "undef") {
+      print "/*", prefix defundef, macro, "*/"
+      next
+    }
+  }
+}
+{ print }
+_ACAWK
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+fi # test -n "$CONFIG_HEADERS"
+eval set X "  :F $CONFIG_FILES  :H $CONFIG_HEADERS    :C $CONFIG_COMMANDS"
 shift
 for ac_tag
 …
   || as_fn_error $? "could not create $ac_file" "$LINENO" 5
  ;;
+  :H)
+  #
+  # CONFIG_HEADER
+  #
+  if test x"$ac_file" != x-; then
+    {
+      $as_echo "/* $configure_input  */" \
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+$as_echo "$as_me: $ac_file is unchanged" >&6;}
+    else
+      rm -f "$ac_file"
+      mv "$ac_tmp/config.h" "$ac_file" \
+        || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    fi
+  else
+    $as_echo "/* $configure_input  */" \
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
+  fi
+# Compute "$ac_file"'s index in $config_headers.
+_am_arg="$ac_file"
+_am_stamp_count=1
+for _am_header in $config_headers :; do
+  case $_am_header in
+    $_am_arg | $_am_arg:* )
+      break ;;
+    * )
+      _am_stamp_count=`expr $_am_stamp_count + 1` ;;
+  esac
+done
+echo "timestamp for $_am_arg" >`$as_dirname -- "$_am_arg" ||
+$as_expr X"$_am_arg" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+         X"$_am_arg" : 'X\(//\)[^/]' \| \
+         X"$_am_arg" : 'X\(//\)$' \| \
+         X"$_am_arg" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$_am_arg" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+            s//\1/
+            q
+          }
+          /^X\(\/\/\)[^/].*/{
+            s//\1/
+            q
+          }
+          /^X\(\/\/\)$/{
+            s//\1/
+            q
+          }
+          /^X\(\/\).*/{
+            s//\1/
+            q
+          }
+          s/.*/./; q'`/stamp-h$_am_stamp_count
+ ;;
   :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
 …
 macro_revision=$macro_revision
+# Whether or not to build static libraries.
+build_old_libs=$enable_static
 # Whether or not to build shared libraries.
 build_libtool_libs=$enable_shared
-# Whether or not to build static libraries.
-build_old_libs=$enable_static
 # What type of objects to build.

libcfa/configure.ac

-              rb7d6a36
+              r6a490b2
 # Checks for programs.
 LT_INIT
+LT_INIT([disable-static])
 AC_PROG_CXX
 …
 AC_PROG_MAKE_SET
+AC_CHECK_HEADERS([linux/io_uring.h])
+AC_CHECK_FUNCS([preadv2 pwritev2])
 AC_CONFIG_FILES([
         Makefile
 …
         ])
+AC_CONFIG_HEADERS(prelude/defines.hfa)
 AC_OUTPUT()

libcfa/prelude/Makefile.am

rb7d6a36	r6a490b2
21	21	# put into lib for now
22	22	cfalibdir = ${CFA_LIBDIR}
23		cfalib_DATA = gcc-builtins.cf builtins.cf extras.cf prelude.cfa bootloader.c
	23	cfalib_DATA = gcc-builtins.cf builtins.cf extras.cf prelude.cfa bootloader.c defines.hfa
24	24
25	25	CC = @LOCAL_CFACC@

libcfa/prelude/Makefile.in

-              rb7d6a36
+              r6a490b2
 # Makefile.in generated by automake 1.16.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 # Copyright (C) 1994-2018 Free Software Foundation, Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
 # This Makefile.in is free software; the Free Software Foundation
 …
 DIST_COMMON = $(srcdir)/Makefile.am $(am__DIST_COMMON)
 mkinstalldirs = $(install_sh) -d
+CONFIG_HEADER = defines.hfa
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
 …
 am__installdirs = "$(DESTDIR)$(cfalibdir)"
 DATA = $(cfalib_DATA)
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
+am__DIST_COMMON = $(srcdir)/Makefile.in
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) \
+        $(LISP)defines.hfa.in
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
+ETAGS = etags
+CTAGS = ctags
+am__DIST_COMMON = $(srcdir)/Makefile.in $(srcdir)/defines.hfa.in
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 ACLOCAL = @ACLOCAL@
 …
 # put into lib for now
 cfalibdir = ${CFA_LIBDIR}
 cfalib_DATA = gcc-builtins.cf builtins.cf extras.cf prelude.cfa bootloader.c
+cfalib_DATA = gcc-builtins.cf builtins.cf extras.cf prelude.cfa bootloader.c defines.hfa
 AM_CFLAGS = -g -Wall -Wno-unused-function -fPIC @ARCH_FLAGS@ @CONFIG_CFLAGS@
 AM_CFAFLAGS = @CONFIG_CFAFLAGS@
 MOSTLYCLEANFILES = bootloader.c builtins.cf extras.cf gcc-builtins.c gcc-builtins.cf prelude.cfa
 MAINTAINERCLEANFILES = ${addprefix ${libdir}/,${cfalib_DATA}} ${addprefix ${libdir}/,${lib_LIBRARIES}}
+all: all-am
+all: defines.hfa
+        $(MAKE) $(AM_MAKEFLAGS) all-am
 .SUFFIXES:
 …
             cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
           *) \
             echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles)'; \
             cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles);; \
+            echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
+            cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
         esac;
 …
         cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
 $(am__aclocal_m4_deps):
+defines.hfa: stamp-h1
+        @test -f $@ || rm -f stamp-h1
+        @test -f $@ || $(MAKE) $(AM_MAKEFLAGS) stamp-h1
+stamp-h1: $(srcdir)/defines.hfa.in $(top_builddir)/config.status
+        @rm -f stamp-h1
+        cd $(top_builddir) && $(SHELL) ./config.status prelude/defines.hfa
+$(srcdir)/defines.hfa.in:  $(am__configure_deps)
+        ($(am__cd) $(top_srcdir) && $(AUTOHEADER))
+        rm -f stamp-h1
+        touch $@
+distclean-hdr:
+        -rm -f defines.hfa stamp-h1
 mostlyclean-libtool:
 …
         files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
         dir='$(DESTDIR)$(cfalibdir)'; $(am__uninstall_files_from_dir)
+tags TAGS:
+ctags CTAGS:
+cscope cscopelist:
+distdir: $(BUILT_SOURCES)
+        $(MAKE) $(AM_MAKEFLAGS) distdir-am
+distdir-am: $(DISTFILES)
+ID: $(am__tagged_files)
+        $(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-am
+TAGS: tags
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+        set x; \
+        here=`pwd`; \
+        $(am__define_uniq_tagged_files); \
+        shift; \
+        if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
+          test -n "$$unique" || unique=$$empty_fix; \
+          if test $$# -gt 0; then \
+            $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+              "$$@" $$unique; \
+          else \
+            $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+              $$unique; \
+          fi; \
+        fi
+ctags: ctags-am
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+        $(am__define_uniq_tagged_files); \
+        test -z "$(CTAGS_ARGS)$$unique" \
+          || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
+             $$unique
+GTAGS:
+        here=`$(am__cd) $(top_builddir) && pwd` \
+          && $(am__cd) $(top_srcdir) \
+          && gtags -i $(GTAGS_ARGS) "$$here"
+cscopelist: cscopelist-am
+cscopelist-am: $(am__tagged_files)
+        list='$(am__tagged_files)'; \
+        case "$(srcdir)" in \
+          [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+          *) sdir=$(subdir)/$(srcdir) ;; \
+        esac; \
+        for i in $$list; do \
+          if test -f "$$i"; then \
+            echo "$(subdir)/$$i"; \
+          else \
+            echo "$$sdir/$$i"; \
+          fi; \
+        done >> $(top_builddir)/cscope.files
+distclean-tags:
+        -rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+distdir: $(DISTFILES)
         @srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
         topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
 …
 check-am: all-am
 check: check-am
 all-am: Makefile $(DATA)
+all-am: Makefile $(DATA) defines.hfa
 installdirs:
         for dir in "$(DESTDIR)$(cfalibdir)"; do \
 …
 distclean: distclean-am
         -rm -f Makefile
 distclean-am: clean-am distclean-generic
+distclean-am: clean-am distclean-generic distclean-hdr distclean-tags
 dvi: dvi-am
 …
 uninstall-am: uninstall-cfalibDATA
+.MAKE: install-am install-strip
+.PHONY: all all-am check check-am clean clean-generic clean-libtool \
+        cscopelist-am ctags-am distclean distclean-generic \
+        distclean-libtool distdir dvi dvi-am html html-am info info-am \
+.MAKE: all install-am install-strip
+.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean clean-generic \
+        clean-libtool cscopelist-am ctags ctags-am distclean \
+        distclean-generic distclean-hdr distclean-libtool \
+        distclean-tags distdir dvi dvi-am html html-am info info-am \
         install install-am install-cfalibDATA install-data \
         install-data-am install-dvi install-dvi-am install-exec \
 …
         maintainer-clean-generic maintainer-clean-local mostlyclean \
         mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
         tags-am uninstall uninstall-am uninstall-cfalibDATA
+        tags tags-am uninstall uninstall-am uninstall-cfalibDATA
 .PRECIOUS: Makefile

libcfa/prelude/builtins.c

-              rb7d6a36
+              r6a490b2
 void exit( int status, const char fmt[], ... ) __attribute__ (( format(printf, 2, 3), __nothrow__, __leaf__, __noreturn__ ));
 void abort( const char fmt[], ... ) __attribute__ (( format(printf, 1, 2), __nothrow__, __leaf__, __noreturn__ ));
+forall(dtype T)
+static inline T & identity(T & i) {
+        return i;
+}
+// generator support
+struct $generator {
+        inline int;
+};
+static inline void  ?{}($generator & this) { ((int&)this) = 0; }
+static inline void ^?{}($generator &) {}
+trait is_generator(dtype T) {
+      void main(T & this);
+      $generator * get_generator(T & this);
+};
+forall(dtype T | is_generator(T))
+static inline T & resume(T & gen) {
+        main(gen);
+        return gen;
+}
 // implicit increment, decrement if += defined, and implicit not if != defined

libcfa/src/Makefile.am

-              rb7d6a36
+              r6a490b2
 ## Created On       : Sun May 31 08:54:01 2015
 ## Last Modified By : Peter A. Buhr
 ## Last Modified On : Mon Jul 15 22:43:27 2019
 ## Update Count     : 241
+## Last Modified On : Mon Mar 16 18:07:59 2020
+## Update Count     : 242
 ###############################################################################
 …
 # The built sources must not depend on the installed headers
 AM_CFAFLAGS = -quiet -cfalib -I$(srcdir)/stdhdr $(if $(findstring ${gdbwaittarget}, ${@}), -XCFA --gdb) @CONFIG_CFAFLAGS@
 AM_CFLAGS = -g -Wall -Wno-unused-function -fPIC -pthread @ARCH_FLAGS@ @CONFIG_CFLAGS@
+AM_CFLAGS = -g -Wall -Wno-unused-function -fPIC -fexceptions -pthread @ARCH_FLAGS@ @CONFIG_CFLAGS@
 AM_CCASFLAGS = -g -Wall -Wno-unused-function @ARCH_FLAGS@ @CONFIG_CFLAGS@
 CFACC = @CFACC@
 …
 #----------------------------------------------------------------------------------------------------------------
 if BUILDLIB
 headers_nosrc = math.hfa gmp.hfa time_t.hfa bits/align.hfa bits/containers.hfa bits/defs.hfa bits/debug.hfa bits/locks.hfa
+headers_nosrc = bitmanip.hfa math.hfa gmp.hfa time_t.hfa bits/align.hfa bits/containers.hfa bits/defs.hfa bits/debug.hfa bits/locks.hfa containers/list.hfa
 headers = fstream.hfa iostream.hfa iterator.hfa limits.hfa rational.hfa time.hfa stdlib.hfa common.hfa \
           containers/maybe.hfa containers/pair.hfa containers/result.hfa containers/vector.hfa
 …
 thread_headers_nosrc = concurrency/invoke.h
 thread_headers = concurrency/coroutine.hfa concurrency/thread.hfa concurrency/kernel.hfa concurrency/monitor.hfa concurrency/mutex.hfa
 thread_libsrc = concurrency/CtxSwitch-@ARCHITECTURE@.S concurrency/alarm.cfa concurrency/invoke.c concurrency/preemption.cfa concurrency/ready_queue.cfa ${thread_headers:.hfa=.cfa}
+thread_libsrc = concurrency/CtxSwitch-@ARCHITECTURE@.S concurrency/alarm.cfa concurrency/invoke.c concurrency/io.cfa concurrency/preemption.cfa concurrency/ready_queue.cfa ${thread_headers:.hfa=.cfa}
 else
 headers =

libcfa/src/Makefile.in

-              rb7d6a36
+              r6a490b2
         $(am__nobase_cfa_include_HEADERS_DIST) $(am__DIST_COMMON)
 mkinstalldirs = $(install_sh) -d
+CONFIG_HEADER = $(top_builddir)/prelude/defines.hfa
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
 …
 am__libcfathread_la_SOURCES_DIST =  \
         concurrency/CtxSwitch-@ARCHITECTURE@.S concurrency/alarm.cfa \
+        concurrency/invoke.c concurrency/preemption.cfa \
+        concurrency/ready_queue.cfa concurrency/coroutine.cfa \
+        concurrency/thread.cfa concurrency/kernel.cfa \
+        concurrency/monitor.cfa concurrency/mutex.cfa
+        concurrency/invoke.c concurrency/io.cfa \
+        concurrency/preemption.cfa concurrency/ready_queue.cfa \
+        concurrency/coroutine.cfa concurrency/thread.cfa \
+        concurrency/kernel.cfa concurrency/monitor.cfa \
+        concurrency/mutex.cfa
 @BUILDLIB_TRUE@am__objects_3 = concurrency/coroutine.lo \
 @BUILDLIB_TRUE@ concurrency/thread.lo concurrency/kernel.lo \
 …
 @BUILDLIB_TRUE@ concurrency/CtxSwitch-@ARCHITECTURE@.lo \
 @BUILDLIB_TRUE@ concurrency/alarm.lo concurrency/invoke.lo \
 @BUILDLIB_TRUE@ concurrency/preemption.lo \
+@BUILDLIB_TRUE@ concurrency/io.lo concurrency/preemption.lo \
 @BUILDLIB_TRUE@ concurrency/ready_queue.lo $(am__objects_3)
 am_libcfathread_la_OBJECTS = $(am__objects_4)
 …
 am__v_at_0 = @
 am__v_at_1 =
 DEFAULT_INCLUDES = -I.@am__isrc@
+DEFAULT_INCLUDES = -I.@am__isrc@ -I$(top_builddir)/prelude
 depcomp = $(SHELL) $(top_srcdir)/automake/depcomp
 am__depfiles_maybe = depfiles
 …
         limits.hfa rational.hfa time.hfa stdlib.hfa common.hfa \
         containers/maybe.hfa containers/pair.hfa containers/result.hfa \
         containers/vector.hfa math.hfa gmp.hfa time_t.hfa \
+        containers/vector.hfa bitmanip.hfa math.hfa gmp.hfa time_t.hfa \
         bits/align.hfa bits/containers.hfa bits/defs.hfa \
         bits/debug.hfa bits/locks.hfa concurrency/coroutine.hfa \
         concurrency/thread.hfa concurrency/kernel.hfa \
         concurrency/monitor.hfa concurrency/mutex.hfa \
         concurrency/invoke.h
+        bits/debug.hfa bits/locks.hfa containers/list.hfa \
+        concurrency/coroutine.hfa concurrency/thread.hfa \
+        concurrency/kernel.hfa concurrency/monitor.hfa \
+        concurrency/mutex.hfa concurrency/invoke.h
 HEADERS = $(nobase_cfa_include_HEADERS)
 am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
 …
 am__v_GOC_0 = @echo "  GOC     " $@;
 am__v_GOC_1 =
+AM_V_PY = $(am__v_PY_@AM_V@)
+am__v_PY_ = $(am__v_PY_@AM_DEFAULT_V@)
+am__v_PY_0 = @echo "  PYTHON  " $@;
+am__v_PY_1 =
 AM_V_RUST = $(am__v_RUST_@AM_V@)
 am__v_RUST_ = $(am__v_RUST_@AM_DEFAULT_V@)
 am__v_RUST_0 = @echo "  RUST     " $@;
+am__v_RUST_0 = @echo "  RUST    " $@;
 am__v_RUST_1 =
 AM_V_NODEJS = $(am__v_NODEJS_@AM_V@)
 am__v_NODEJS_ = $(am__v_NODEJS_@AM_DEFAULT_V@)
 am__v_NODEJS_0 = @echo "  NODEJS     " $@;
+am__v_NODEJS_0 = @echo "  NODEJS  " $@;
 am__v_NODEJS_1 =
 AM_V_JAVAC = $(am__v_JAVAC_@AM_V@)
 …
 # The built sources must not depend on the installed headers
 AM_CFAFLAGS = -quiet -cfalib -I$(srcdir)/stdhdr $(if $(findstring ${gdbwaittarget}, ${@}), -XCFA --gdb) @CONFIG_CFAFLAGS@
 AM_CFLAGS = -g -Wall -Wno-unused-function -fPIC -pthread @ARCH_FLAGS@ @CONFIG_CFLAGS@
+AM_CFLAGS = -g -Wall -Wno-unused-function -fPIC -fexceptions -pthread @ARCH_FLAGS@ @CONFIG_CFLAGS@
 AM_CCASFLAGS = -g -Wall -Wno-unused-function @ARCH_FLAGS@ @CONFIG_CFLAGS@
 @BUILDLIB_FALSE@headers_nosrc =
 #----------------------------------------------------------------------------------------------------------------
 @BUILDLIB_TRUE@headers_nosrc = math.hfa gmp.hfa time_t.hfa bits/align.hfa bits/containers.hfa bits/defs.hfa bits/debug.hfa bits/locks.hfa
+@BUILDLIB_TRUE@headers_nosrc = bitmanip.hfa math.hfa gmp.hfa time_t.hfa bits/align.hfa bits/containers.hfa bits/defs.hfa bits/debug.hfa bits/locks.hfa containers/list.hfa
 @BUILDLIB_FALSE@headers =
 @BUILDLIB_TRUE@headers = fstream.hfa iostream.hfa iterator.hfa limits.hfa rational.hfa time.hfa stdlib.hfa common.hfa \
 …
 @BUILDLIB_FALSE@thread_headers =
 @BUILDLIB_TRUE@thread_headers = concurrency/coroutine.hfa concurrency/thread.hfa concurrency/kernel.hfa concurrency/monitor.hfa concurrency/mutex.hfa
 @BUILDLIB_TRUE@thread_libsrc = concurrency/CtxSwitch-@ARCHITECTURE@.S concurrency/alarm.cfa concurrency/invoke.c concurrency/preemption.cfa concurrency/ready_queue.cfa ${thread_headers:.hfa=.cfa}
+@BUILDLIB_TRUE@thread_libsrc = concurrency/CtxSwitch-@ARCHITECTURE@.S concurrency/alarm.cfa concurrency/invoke.c concurrency/io.cfa concurrency/preemption.cfa concurrency/ready_queue.cfa ${thread_headers:.hfa=.cfa}
 #----------------------------------------------------------------------------------------------------------------
 …
         concurrency/$(DEPDIR)/$(am__dirstamp)
 concurrency/invoke.lo: concurrency/$(am__dirstamp) \
+        concurrency/$(DEPDIR)/$(am__dirstamp)
+concurrency/io.lo: concurrency/$(am__dirstamp) \
         concurrency/$(DEPDIR)/$(am__dirstamp)
 concurrency/preemption.lo: concurrency/$(am__dirstamp) \

libcfa/src/bits/containers.hfa

-              rb7d6a36
+              r6a490b2
         static inline forall( dtype T | is_node(T) ) {
                 void ?{}( __queue(T) & this ) with( this ) {
                         head{ 0p };
+                        head{ 1p };
                         tail{ &head };
+                        verify(*tail == 1p);
+                }
                 void append( __queue(T) & this, T * val ) with( this ) {
                         verify(tail != 0p);
+                        verify(*tail == 1p);
                         *tail = val;
                         tail = &get_next( *val );
+                        *tail = 1p;
+                }
                 T * pop_head( __queue(T) & this ) {
+                        verify(*this.tail == 1p);
                         T * head = this.head;
                         if( head ) {
+                        if( head != 1p ) {
                                 this.head = get_next( *head );
                                 if( !get_next( *head ) ) {
+                                if( get_next( *head ) == 1p ) {
                                         this.tail = &this.head;
+                                }
                                 get_next( *head ) = 0p;
+                        }
+                        return head;
+                                verify(*this.tail == 1p);
+                                verify( get_next(*head) == 0p );
+                                return head;
+                        }
+                        verify(*this.tail == 1p);
+                        return 0p;
+                }
 …
                         get_next( *val ) = 0p;
                         verify( (head == 0p) == (&head == tail) );
                         verify( *tail == 0p );
+                        verify( (head == 1p) == (&head == tail) );
+                        verify( *tail == 1p );
                         return val;
+                }
 …
                         return this.head != 0;
+                }
+                void move_to_front( __dllist(T) & src, __dllist(T) & dst, T & node ) {
+                        remove    (src, node);
+                        push_front(dst, node);
+                }
+        }
         #undef next

libcfa/src/bits/debug.hfa

-              rb7d6a36
+              r6a490b2
 // Author           : Thierry Delisle
 // Created On       : Mon Nov 28 12:27:26 2016
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Tue Feb  4 12:29:21 2020
 // Update Count     : 9
+// Last Modified By : Andrew Beach
+// Last Modified On : Mon Apr 27 10:15:00 2020
+// Update Count     : 10
 //
 …
         #define __cfaabi_dbg_ctx_param const char caller[]
         #define __cfaabi_dbg_ctx_param2 , const char caller[]
+        #define __cfaabi_dbg_ctx_fwd caller
+        #define __cfaabi_dbg_ctx_fwd2 , caller
 #else
         #define __cfaabi_dbg_debug_do(...)
 …
         #define __cfaabi_dbg_ctx_param
         #define __cfaabi_dbg_ctx_param2
+        #define __cfaabi_dbg_ctx_fwd
+        #define __cfaabi_dbg_ctx_fwd2
 #endif
 …
 #endif
         #include <stdarg.h>
-        #include <stdio.h>
-        #include <unistd.h>
         extern void __cfaabi_bits_write( int fd, const char buffer[], int len );
 …
         extern void __cfaabi_bits_print_vararg( int fd, const char fmt[], va_list arg );
         extern void __cfaabi_bits_print_buffer( int fd, char buffer[], int buffer_size, const char fmt[], ... ) __attribute__(( format(printf, 4, 5) ));
+#if defined(__CFA_DEBUG_PRINT__) \
+                || defined(__CFA_DEBUG_PRINT_IO__) || defined(__CFA_DEBUG_PRINT_IO_CORE__) \
+                || defined(__CFA_DEBUG_PRINT_MONITOR__) || defined(__CFA_DEBUG_PRINT_PREEMPTION__) \
+                || defined(__CFA_DEBUG_PRINT_RUNTIME_CORE__) || defined(__CFA_DEBUG_PRINT_EXCEPTION__)
+        #include <stdio.h>
+        #include <unistd.h>
+#endif
 #ifdef __cforall
+}
 #endif
+// #define __CFA_DEBUG_PRINT__
+// Deprecated: Use the versions with the new module names.
 #ifdef __CFA_DEBUG_PRINT__
         #define __cfaabi_dbg_write( buffer, len )         __cfaabi_bits_write( STDERR_FILENO, buffer, len )
         #define __cfaabi_dbg_acquire()                    __cfaabi_bits_acquire()
         #define __cfaabi_dbg_release()                    __cfaabi_bits_release()
         #define __cfaabi_dbg_print_safe(...)              __cfaabi_bits_print_safe  ( STDERR_FILENO, __VA_ARGS__ )
         #define __cfaabi_dbg_print_nolock(...)            __cfaabi_bits_print_nolock( STDERR_FILENO, __VA_ARGS__ )
         #define __cfaabi_dbg_print_buffer(...)            __cfaabi_bits_print_buffer( STDERR_FILENO, __VA_ARGS__ )
+        #define __cfaabi_dbg_print_safe(...)              __cfaabi_bits_print_safe   ( STDERR_FILENO, __VA_ARGS__ )
+        #define __cfaabi_dbg_print_nolock(...)            __cfaabi_bits_print_nolock ( STDERR_FILENO, __VA_ARGS__ )
+        #define __cfaabi_dbg_print_buffer(...)            __cfaabi_bits_print_buffer ( STDERR_FILENO, __VA_ARGS__ )
         #define __cfaabi_dbg_print_buffer_decl(...)       char __dbg_text[256]; int __dbg_len = snprintf( __dbg_text, 256, __VA_ARGS__ ); __cfaabi_bits_write( STDERR_FILENO, __dbg_text, __dbg_len );
         #define __cfaabi_dbg_print_buffer_local(...)      __dbg_len = snprintf( __dbg_text, 256, __VA_ARGS__ ); __cfaabi_bits_write( STDERR_FILENO, __dbg_text, __dbg_len );
+        #define __cfaabi_dbg_print_buffer_local(...)      __dbg_len = snprintf( __dbg_text, 256, __VA_ARGS__ ); __cfaabi_dbg_write( STDERR_FILENO, __dbg_text, __dbg_len );
 #else
         #define __cfaabi_dbg_write(...)               ((void)0)
 …
 #endif
+// Debug print functions and statements:
+// Most are wrappers around the bits printing function but are not always used.
+// If they are used depends if the group (first argument) is active or not. The group must be one
+// defined belowe. The other arguments depend on the wrapped function.
+#define __cfadbg_write(group, buffer, len) \
+        __CFADBG_PRINT_GROUP_##group(__cfaabi_bits_write(STDERR_FILENO, buffer, len))
+#define __cfadbg_acquire(group) \
+        __CFADBG_PRINT_GROUP_##group(__cfaabi_bits_acquire())
+#define __cfadbg_release(group) \
+        __CFADBG_PRINT_GROUP_##group(__cfaabi_bits_release())
+#define __cfadbg_print_safe(group, ...) \
+        __CFADBG_PRINT_GROUP_##group(__cfaabi_bits_print_safe(STDERR_FILENO, __VA_ARGS__))
+#define __cfadbg_print_nolock(group, ...) \
+        __CFADBG_PRINT_GROUP_##group(__cfaabi_bits_print_nolock(STDERR_FILENO, __VA_ARGS__))
+#define __cfadbg_print_buffer(group, ...) \
+        __CFADBG_PRINT_GROUP_##group(__cfaabi_bits_print_buffer(STDERR_FILENO, __VA_ARGS__))
+#define __cfadbg_print_buffer_decl(group, ...) \
+        __CFADBG_PRINT_GROUP_##group(char __dbg_text[256]; int __dbg_len = snprintf( __dbg_text, 256, __VA_ARGS__ ); __cfaabi_bits_write( __dbg_text, __dbg_len ))
+#define __cfadbg_print_buffer_local(group, ...) \
+        __CFADBG_PRINT_GROUP_##group(__dbg_len = snprintf( __dbg_text, 256, __VA_ARGS__ ); __cfaabi_bits_write(STDERR_FILENO, __dbg_text, __dbg_len))
+// The debug print groups:
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_IO__)
+#       define __CFADBG_PRINT_GROUP_io(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_io(...) ((void)0)
+#endif
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_IO__) || defined(__CFA_DEBUG_PRINT_IO_CORE__)
+#       define __CFADBG_PRINT_GROUP_io_core(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_io_core(...) ((void)0)
+#endif
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_MONITOR__)
+#       define __CFADBG_PRINT_GROUP_monitor(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_monitor(...) ((void)0)
+#endif
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_PREEMPTION__)
+#       define __CFADBG_PRINT_GROUP_preemption(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_preemption(...) ((void)0)
+#endif
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_RUNTIME_CORE__)
+#       define __CFADBG_PRINT_GROUP_runtime_core(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_runtime_core(...) ((void)0)
+#endif
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_READY_QUEUE__)
+#       define __CFADBG_PRINT_GROUP_ready_queue(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_ready_queue(...) ((void)0)
+#endif
+#if defined(__CFA_DEBUG_PRINT__) || defined(__CFA_DEBUG_PRINT_EXCEPTION__)
+#       define __CFADBG_PRINT_GROUP_exception(...) __VA_ARGS__
+#else
+#       define __CFADBG_PRINT_GROUP_exception(...) ((void)0)
+#endif
 // Local Variables: //
 // mode: c //

libcfa/src/bits/locks.hfa

-              rb7d6a36
+              r6a490b2
                 #ifdef __CFA_DEBUG__
                         void __cfaabi_dbg_record(__spinlock_t & this, const char prev_name[]);
+                        void __cfaabi_dbg_record_lock(__spinlock_t & this, const char prev_name[]);
                 #else
                         #define __cfaabi_dbg_record(x, y)
+                        #define __cfaabi_dbg_record_lock(x, y)
                 #endif
+        }
-        extern void yield( unsigned int );
         static inline void ?{}( __spinlock_t & this ) {
 …
         // Lock the spinlock, return false if already acquired
         static inline bool try_lock  ( __spinlock_t & this __cfaabi_dbg_ctx_param2 ) {
+                disable_interrupts();
                 bool result = (this.lock == 0) && (__atomic_test_and_set( &this.lock, __ATOMIC_ACQUIRE ) == 0);
                 if( result ) {
+                        disable_interrupts();
+                        __cfaabi_dbg_record( this, caller );
+                        __cfaabi_dbg_record_lock( this, caller );
+                } else {
+                        enable_interrupts_noPoll();
+                }
                 return result;
 …
                 #endif
+                disable_interrupts();
                 for ( unsigned int i = 1;; i += 1 ) {
                         if ( (this.lock == 0) && (__atomic_test_and_set( &this.lock, __ATOMIC_ACQUIRE ) == 0) ) break;
 …
                         #endif
+                }
+                disable_interrupts();
+                __cfaabi_dbg_record( this, caller );
+                __cfaabi_dbg_record_lock( this, caller );
+        }
         static inline void unlock( __spinlock_t & this ) {
+                __atomic_clear( &this.lock, __ATOMIC_RELEASE );
                 enable_interrupts_noPoll();
-                __atomic_clear( &this.lock, __ATOMIC_RELEASE );
+        }
 …
         #endif
+        extern "C" {
+                char * strerror(int);
+        }
+        #define CHECKED(x) { int err = x; if( err != 0 ) abort("KERNEL ERROR: Operation \"" #x "\" return error %d - %s\n", err, strerror(err)); }
         struct __bin_sem_t {
-                bool                    signaled;
                 pthread_mutex_t         lock;
                 pthread_cond_t          cond;
+                int                     val;
         };
         static inline void ?{}(__bin_sem_t & this) with( this ) {
+                signaled = false;
+                pthread_mutex_init(&lock, NULL);
+                pthread_cond_init (&cond, NULL);
+                // Create the mutex with error checking
+                pthread_mutexattr_t mattr;
+                pthread_mutexattr_init( &mattr );
+                pthread_mutexattr_settype( &mattr, PTHREAD_MUTEX_ERRORCHECK_NP);
+                pthread_mutex_init(&lock, &mattr);
+                pthread_cond_init (&cond, 0p);
+                val = 0;
+        }
         static inline void ^?{}(__bin_sem_t & this) with( this ) {
                 pthread_mutex_destroy(&lock);
                 pthread_cond_destroy (&cond);
+                CHECKED( pthread_mutex_destroy(&lock) );
+                CHECKED( pthread_cond_destroy (&cond) );
+        }
         static inline void wait(__bin_sem_t & this) with( this ) {
                 verify(__cfaabi_dbg_in_kernel());
                 pthread_mutex_lock(&lock);
                         if(!signaled) {   // this must be a loop, not if!
+                CHECKED( pthread_mutex_lock(&lock) );
+                        while(val < 1) {
                                 pthread_cond_wait(&cond, &lock);
+                        }
                         signaled = false;
                 pthread_mutex_unlock(&lock);
+                        val -= 1;
+                CHECKED( pthread_mutex_unlock(&lock) );
+        }
         static inline void post(__bin_sem_t & this) with( this ) {
                 verify(__cfaabi_dbg_in_kernel());
+        static inline bool post(__bin_sem_t & this) with( this ) {
+                bool needs_signal = false;
+                pthread_mutex_lock(&lock);
+                        bool needs_signal = !signaled;
+                        signaled = true;
+                pthread_mutex_unlock(&lock);
+                CHECKED( pthread_mutex_lock(&lock) );
+                        if(val < 1) {
+                                val += 1;
+                                pthread_cond_signal(&cond);
+                                needs_signal = true;
+                        }
+                CHECKED( pthread_mutex_unlock(&lock) );
+                if (needs_signal)
+                        pthread_cond_signal(&cond);
+                return needs_signal;
+        }
+        #undef CHECKED
 #endif

libcfa/src/bits/signal.hfa

rb7d6a36	r6a490b2
54	54	sig, handler, flags, errno, strerror( errno )
55	55	);
56		_exit( EXIT_FAILURE );
	56	_Exit( EXIT_FAILURE );
57	57	} // if
58	58	}

libcfa/src/concurrency/CtxSwitch-arm.S

-              rb7d6a36
+              r6a490b2
         .text
         .align  2
         .global CtxSwitch
         .type   CtxSwitch, %function
+        .global __cfactx_switch
+        .type   __cfactx_switch, %function
 CtxSwitch:
+__cfactx_switch:
         @ save callee-saved registers: r4-r8, r10, r11, r13(sp) (plus r9 depending on platform specification)
         @ I've seen reference to 31 registers on 64-bit, if this is the case, more need to be saved
 …
         mov r15, r14
         #endif // R9_SPECIAL
         .text
         .align  2
         .global CtxInvokeStub
         .type   CtxInvokeStub, %function
+        .global __cfactx_invoke_stub
+        .type   __cfactx_invoke_stub, %function
 CtxInvokeStub:
+__cfactx_invoke_stub:
         ldmfd r13!, {r0-r1}
         mov r15, r1

libcfa/src/concurrency/CtxSwitch-i386.S

-              rb7d6a36
+              r6a490b2
         .text
         .align 2
         .globl CtxSwitch
         .type  CtxSwitch, @function
 CtxSwitch:
+        .globl __cfactx_switch
+        .type  __cfactx_switch, @function
+__cfactx_switch:
         // Copy the "from" context argument from the stack to register eax
 …
         ret
         .size  CtxSwitch, .-CtxSwitch
+        .size  __cfactx_switch, .-__cfactx_switch
 // Local Variables: //

libcfa/src/concurrency/CtxSwitch-x86_64.S

-              rb7d6a36
+              r6a490b2
         .text
         .align 2
         .globl CtxSwitch
         .type  CtxSwitch, @function
 CtxSwitch:
+        .globl __cfactx_switch
+        .type  __cfactx_switch, @function
+__cfactx_switch:
         // Save volatile registers on the stack.
 …
         ret
         .size  CtxSwitch, .-CtxSwitch
+        .size  __cfactx_switch, .-__cfactx_switch
 //-----------------------------------------------------------------------------
 …
         .text
         .align 2
         .globl CtxInvokeStub
         .type    CtxInvokeStub, @function
 CtxInvokeStub:
+        .globl __cfactx_invoke_stub
+        .type    __cfactx_invoke_stub, @function
+__cfactx_invoke_stub:
         movq %rbx, %rdi
         movq %r12, %rsi
         jmp *%r13
         .size  CtxInvokeStub, .-CtxInvokeStub
+        .size  __cfactx_invoke_stub, .-__cfactx_invoke_stub
 // Local Variables: //

libcfa/src/concurrency/alarm.cfa

-              rb7d6a36
+              r6a490b2
 //=============================================================================================
 void ?{}( alarm_node_t & this, thread_desc * thrd, Time alarm, Duration period ) with( this ) {
+void ?{}( alarm_node_t & this, $thread * thrd, Time alarm, Duration period ) with( this ) {
         this.thrd = thrd;
         this.alarm = alarm;
         this.period = period;
-        next = 0;
         set = false;
         kernel_alarm = false;
 …
         this.alarm = alarm;
         this.period = period;
-        next = 0;
         set = false;
         kernel_alarm = true;
 …
+}
+#if !defined(NDEBUG) && (defined(__CFA_DEBUG__) || defined(__CFA_VERIFY__))
+bool validate( alarm_list_t * this ) {
+        alarm_node_t ** it = &this->head;
+        while( (*it) ) {
+                it = &(*it)->next;
+void insert( alarm_list_t * this, alarm_node_t * n ) {
+        alarm_node_t * it = & (*this)`first;
+        while( it && (n->alarm > it->alarm) ) {
+                it = & (*it)`next;
+        }
+        if ( it ) {
+                insert_before( *it, *n );
+        } else {
+                insert_last(*this, *n);
+        }
+        return it == this->tail;
+}
+#endif
+static inline void insert_at( alarm_list_t * this, alarm_node_t * n, __alarm_it_t p ) {
+        verify( !n->next );
+        if( p == this->tail ) {
+                this->tail = &n->next;
+        }
+        else {
+                n->next = *p;
+        }
+        *p = n;
+        verify( validate( this ) );
+}
+void insert( alarm_list_t * this, alarm_node_t * n ) {
+        alarm_node_t ** it = &this->head;
+        while( (*it) && (n->alarm > (*it)->alarm) ) {
+                it = &(*it)->next;
+        }
+        insert_at( this, n, it );
+        verify( validate( this ) );
+        verify( validate( *this ) );
+}
 alarm_node_t * pop( alarm_list_t * this ) {
+        alarm_node_t * head = this->head;
+        verify( validate( *this ) );
+        alarm_node_t * head = & (*this)`first;
         if( head ) {
+                this->head = head->next;
+                if( !head->next ) {
+                        this->tail = &this->head;
+                }
+                head->next = 0p;
+                remove(*head);
+        }
         verify( validate( this ) );
+        verify( validate( *this ) );
         return head;
+}
-static inline void remove_at( alarm_list_t * this, alarm_node_t * n, __alarm_it_t it ) {
-        verify( it );
-        verify( (*it) == n );
-        (*it) = n->next;
-        if( !n-> next ) {
-                this->tail = it;
+        }
-        n->next = 0p;
-        verify( validate( this ) );
+}
-static inline void remove( alarm_list_t * this, alarm_node_t * n ) {
-        alarm_node_t ** it = &this->head;
-        while( (*it) && (*it) != n ) {
-                it = &(*it)->next;
+        }
-        verify( validate( this ) );
-        if( *it ) { remove_at( this, n, it ); }
-        verify( validate( this ) );
+}
 void register_self( alarm_node_t * this ) {
         alarm_list_t * alarms = &event_kernel->alarms;
+        alarm_list_t & alarms = event_kernel->alarms;
         disable_interrupts();
 …
+        {
                 verify( validate( alarms ) );
                 bool first = !alarms->head;
+                bool first = ! & alarms`first;
                 insert( alarms, this );
+                insert( &alarms, this );
                 if( first ) {
                         __kernel_set_timer( alarms->head->alarm - __kernel_get_time() );
+                        __kernel_set_timer( alarms`first.alarm - __kernel_get_time() );
+                }
+        }
 …
         lock( event_kernel->lock __cfaabi_dbg_ctx2 );
+        {
                 verify( validate( &event_kernel->alarms ) );
                 remove( &event_kernel->alarms, this );
+                verify( validate( event_kernel->alarms ) );
+                remove( *this );
+        }
         unlock( event_kernel->lock );
 …
+}
+//=============================================================================================
+// Utilities
+//=============================================================================================
+void sleep( Duration duration ) {
+        alarm_node_t node = { active_thread(), __kernel_get_time() + duration, 0`s };
+        register_self( &node );
+        park( __cfaabi_dbg_ctx );
+        /* paranoid */ verify( !node.set );
+        /* paranoid */ verify( & node`next == 0p );
+        /* paranoid */ verify( & node`prev == 0p );
+}
 // Local Variables: //
 // mode: c //

libcfa/src/concurrency/alarm.hfa

-              rb7d6a36
+              r6a490b2
 #include "time.hfa"
+struct thread_desc;
+#include <containers/list.hfa>
+struct $thread;
 struct processor;
 …
         Time alarm;                             // time when alarm goes off
         Duration period;                        // if > 0 => period of alarm
+        alarm_node_t * next;            // intrusive link list field
+        DLISTED_MGD_IMPL_IN(alarm_node_t)
         union {
                 thread_desc * thrd;     // thrd who created event
+                $thread * thrd; // thrd who created event
                 processor * proc;               // proc who created event
         };
 …
         bool kernel_alarm       :1;             // true if this is not a user defined alarm
 };
+DLISTED_MGD_IMPL_OUT(alarm_node_t)
+typedef alarm_node_t ** __alarm_it_t;
+void ?{}( alarm_node_t & this, thread_desc * thrd, Time alarm, Duration period );
+void ?{}( alarm_node_t & this, $thread * thrd, Time alarm, Duration period );
 void ?{}( alarm_node_t & this, processor   * proc, Time alarm, Duration period );
 void ^?{}( alarm_node_t & this );
+struct alarm_list_t {
+        alarm_node_t * head;
+        __alarm_it_t tail;
+};
+static inline void ?{}( alarm_list_t & this ) with( this ) {
+        head = 0;
+        tail = &head;
+}
+typedef dlist(alarm_node_t, alarm_node_t) alarm_list_t;
 void insert( alarm_list_t * this, alarm_node_t * n );

libcfa/src/concurrency/coroutine.cfa

-              rb7d6a36
+              r6a490b2
 extern "C" {
         void _CtxCoroutine_Unwind(struct _Unwind_Exception * storage, struct coroutine_desc *) __attribute__ ((__noreturn__));
+        void _CtxCoroutine_Unwind(struct _Unwind_Exception * storage, struct $coroutine *) __attribute__ ((__noreturn__));
         static void _CtxCoroutine_UnwindCleanup(_Unwind_Reason_Code, struct _Unwind_Exception *) __attribute__ ((__noreturn__));
         static void _CtxCoroutine_UnwindCleanup(_Unwind_Reason_Code, struct _Unwind_Exception *) {
 …
+}
 void ?{}( coroutine_desc & this, const char name[], void * storage, size_t storageSize ) with( this ) {
+void ?{}( $coroutine & this, const char name[], void * storage, size_t storageSize ) with( this ) {
         (this.context){0p, 0p};
         (this.stack){storage, storageSize};
 …
+}
 void ^?{}(coroutine_desc& this) {
+void ^?{}($coroutine& this) {
         if(this.state != Halted && this.state != Start && this.state != Primed) {
                 coroutine_desc * src = TL_GET( this_thread )->curr_cor;
                 coroutine_desc * dst = &this;
+                $coroutine * src = TL_GET( this_thread )->curr_cor;
+                $coroutine * dst = &this;
                 struct _Unwind_Exception storage;
 …
+                }
                 CoroutineCtxSwitch( src, dst );
+                $ctx_switch( src, dst );
+        }
+}
 …
 forall(dtype T | is_coroutine(T))
 void prime(T& cor) {
         coroutine_desc* this = get_coroutine(cor);
+        $coroutine* this = get_coroutine(cor);
         assert(this->state == Start);
 …
 // is not inline (We can't inline Cforall in C)
 extern "C" {
         void __leave_coroutine( struct coroutine_desc * src ) {
                 coroutine_desc * starter = src->cancellation != 0 ? src->last : src->starter;
+        void __cfactx_cor_leave( struct $coroutine * src ) {
+                $coroutine * starter = src->cancellation != 0 ? src->last : src->starter;
                 src->state = Halted;
 …
                         src->name, src, starter->name, starter );
                 CoroutineCtxSwitch( src, starter );
+        }
         struct coroutine_desc * __finish_coroutine(void) {
                 struct coroutine_desc * cor = kernelTLS.this_thread->curr_cor;
+                $ctx_switch( src, starter );
+        }
+        struct $coroutine * __cfactx_cor_finish(void) {
+                struct $coroutine * cor = kernelTLS.this_thread->curr_cor;
                 if(cor->state == Primed) {
                         suspend();
+                        __cfactx_suspend();
+                }

libcfa/src/concurrency/coroutine.hfa

-              rb7d6a36
+              r6a490b2
 trait is_coroutine(dtype T) {
       void main(T & this);
       coroutine_desc * get_coroutine(T & this);
+      $coroutine * get_coroutine(T & this);
 };
 #define DECL_COROUTINE(X) static inline coroutine_desc* get_coroutine(X& this) { return &this.__cor; } void main(X& this)
+#define DECL_COROUTINE(X) static inline $coroutine* get_coroutine(X& this) { return &this.__cor; } void main(X& this)
 //-----------------------------------------------------------------------------
 …
 // void ^?{}( coStack_t & this );
 void ?{}( coroutine_desc & this, const char name[], void * storage, size_t storageSize );
 void ^?{}( coroutine_desc & this );
+void  ?{}( $coroutine & this, const char name[], void * storage, size_t storageSize );
+void ^?{}( $coroutine & this );
 static inline void ?{}( coroutine_desc & this)                                       { this{ "Anonymous Coroutine", 0p, 0 }; }
 static inline void ?{}( coroutine_desc & this, size_t stackSize)                     { this{ "Anonymous Coroutine", 0p, stackSize }; }
 static inline void ?{}( coroutine_desc & this, void * storage, size_t storageSize )  { this{ "Anonymous Coroutine", storage, storageSize }; }
 static inline void ?{}( coroutine_desc & this, const char name[])                    { this{ name, 0p, 0 }; }
 static inline void ?{}( coroutine_desc & this, const char name[], size_t stackSize ) { this{ name, 0p, stackSize }; }
+static inline void ?{}( $coroutine & this)                                       { this{ "Anonymous Coroutine", 0p, 0 }; }
+static inline void ?{}( $coroutine & this, size_t stackSize)                     { this{ "Anonymous Coroutine", 0p, stackSize }; }
+static inline void ?{}( $coroutine & this, void * storage, size_t storageSize )  { this{ "Anonymous Coroutine", storage, storageSize }; }
+static inline void ?{}( $coroutine & this, const char name[])                    { this{ name, 0p, 0 }; }
+static inline void ?{}( $coroutine & this, const char name[], size_t stackSize ) { this{ name, 0p, stackSize }; }
 //-----------------------------------------------------------------------------
 // Public coroutine API
-static inline void suspend(void);
-forall(dtype T | is_coroutine(T))
-static inline T & resume(T & cor);
 forall(dtype T | is_coroutine(T))
 void prime(T & cor);
 static inline struct coroutine_desc * active_coroutine() { return TL_GET( this_thread )->curr_cor; }
+static inline struct $coroutine * active_coroutine() { return TL_GET( this_thread )->curr_cor; }
 //-----------------------------------------------------------------------------
 …
 // Start coroutine routines
 extern "C" {
         void CtxInvokeCoroutine(void (*main)(void *), void * this);
+        void __cfactx_invoke_coroutine(void (*main)(void *), void * this);
         forall(dtype T)
         void CtxStart(void (*main)(T &), struct coroutine_desc * cor, T & this, void (*invoke)(void (*main)(void *), void *));
+        void __cfactx_start(void (*main)(T &), struct $coroutine * cor, T & this, void (*invoke)(void (*main)(void *), void *));
         extern void _CtxCoroutine_Unwind(struct _Unwind_Exception * storage, struct coroutine_desc *) __attribute__ ((__noreturn__));
+        extern void __cfactx_coroutine_unwind(struct _Unwind_Exception * storage, struct $coroutine *) __attribute__ ((__noreturn__));
         extern void CtxSwitch( struct __stack_context_t * from, struct __stack_context_t * to ) asm ("CtxSwitch");
+        extern void __cfactx_switch( struct __stack_context_t * from, struct __stack_context_t * to ) asm ("__cfactx_switch");
+}
 // Private wrappers for context switch and stack creation
 // Wrapper for co
 static inline void CoroutineCtxSwitch(coroutine_desc* src, coroutine_desc* dst) {
+static inline void $ctx_switch( $coroutine * src, $coroutine * dst ) __attribute__((nonnull (1, 2))) {
         // set state of current coroutine to inactive
         src->state = src->state == Halted ? Halted : Inactive;
+        src->state = src->state == Halted ? Halted : Blocked;
         // set new coroutine that task is executing
 …
         // context switch to specified coroutine
         verify( dst->context.SP );
         CtxSwitch( &src->context, &dst->context );
         // when CtxSwitch returns we are back in the src coroutine
+        __cfactx_switch( &src->context, &dst->context );
+        // when __cfactx_switch returns we are back in the src coroutine
         // set state of new coroutine to active
 …
         if( unlikely(src->cancellation != 0p) ) {
                 _CtxCoroutine_Unwind(src->cancellation, src);
+                __cfactx_coroutine_unwind(src->cancellation, src);
+        }
+}
 …
 // Suspend implementation inlined for performance
+static inline void suspend(void) {
+        // optimization : read TLS once and reuse it
+        // Safety note: this is preemption safe since if
+        // preemption occurs after this line, the pointer
+        // will also migrate which means this value will
+        // stay in syn with the TLS
+        coroutine_desc * src = TL_GET( this_thread )->curr_cor;
+extern "C" {
+        static inline void __cfactx_suspend(void) {
+                // optimization : read TLS once and reuse it
+                // Safety note: this is preemption safe since if
+                // preemption occurs after this line, the pointer
+                // will also migrate which means this value will
+                // stay in syn with the TLS
+                $coroutine * src = TL_GET( this_thread )->curr_cor;
         assertf( src->last != 0,
                 "Attempt to suspend coroutine \"%.256s\" (%p) that has never been resumed.\n"
                 "Possible cause is a suspend executed in a member called by a coroutine user rather than by the coroutine main.",
                 src->name, src );
         assertf( src->last->state != Halted,
                 "Attempt by coroutine \"%.256s\" (%p) to suspend back to terminated coroutine \"%.256s\" (%p).\n"
                 "Possible cause is terminated coroutine's main routine has already returned.",
                 src->name, src, src->last->name, src->last );
+                assertf( src->last != 0,
+                        "Attempt to suspend coroutine \"%.256s\" (%p) that has never been resumed.\n"
+                        "Possible cause is a suspend executed in a member called by a coroutine user rather than by the coroutine main.",
+                        src->name, src );
+                assertf( src->last->state != Halted,
+                        "Attempt by coroutine \"%.256s\" (%p) to suspend back to terminated coroutine \"%.256s\" (%p).\n"
+                        "Possible cause is terminated coroutine's main routine has already returned.",
+                        src->name, src, src->last->name, src->last );
+        CoroutineCtxSwitch( src, src->last );
+                $ctx_switch( src, src->last );
+        }
+}
 …
         // will also migrate which means this value will
         // stay in syn with the TLS
         coroutine_desc * src = TL_GET( this_thread )->curr_cor;
         coroutine_desc * dst = get_coroutine(cor);
+        $coroutine * src = TL_GET( this_thread )->curr_cor;
+        $coroutine * dst = get_coroutine(cor);
         if( unlikely(dst->context.SP == 0p) ) {
                 TL_GET( this_thread )->curr_cor = dst;
                 __stack_prepare(&dst->stack, 65000);
                 CtxStart(main, dst, cor, CtxInvokeCoroutine);
+                __cfactx_start(main, dst, cor, __cfactx_invoke_coroutine);
                 TL_GET( this_thread )->curr_cor = src;
+        }
 …
         // always done for performance testing
         CoroutineCtxSwitch( src, dst );
+        $ctx_switch( src, dst );
         return cor;
+}
 static inline void resume(coroutine_desc * dst) {
+static inline void resume( $coroutine * dst ) __attribute__((nonnull (1))) {
         // optimization : read TLS once and reuse it
         // Safety note: this is preemption safe since if
 …
         // will also migrate which means this value will
         // stay in syn with the TLS
         coroutine_desc * src = TL_GET( this_thread )->curr_cor;
+        $coroutine * src = TL_GET( this_thread )->curr_cor;
         // not resuming self ?
 …
         // always done for performance testing
         CoroutineCtxSwitch( src, dst );
+        $ctx_switch( src, dst );
+}

libcfa/src/concurrency/invoke.c

-              rb7d6a36
+              r6a490b2
 // Called from the kernel when starting a coroutine or task so must switch back to user mode.
+extern void __leave_coroutine ( struct coroutine_desc * );
+extern struct coroutine_desc * __finish_coroutine(void);
+extern void __leave_thread_monitor();
+extern struct $coroutine * __cfactx_cor_finish(void);
+extern void __cfactx_cor_leave ( struct $coroutine * );
+extern void __cfactx_thrd_leave();
 extern void disable_interrupts() OPTIONAL_THREAD;
 extern void enable_interrupts( __cfaabi_dbg_ctx_param );
 void CtxInvokeCoroutine(
+void __cfactx_invoke_coroutine(
         void (*main)(void *),
         void *this
 ) {
         // Finish setting up the coroutine by setting its state
         struct coroutine_desc * cor = __finish_coroutine();
+        struct $coroutine * cor = __cfactx_cor_finish();
         // Call the main of the coroutine
 …
         //Final suspend, should never return
         __leave_coroutine( cor );
+        __cfactx_cor_leave( cor );
         __cabi_abort( "Resumed dead coroutine" );
+}
 static _Unwind_Reason_Code _CtxCoroutine_UnwindStop(
+static _Unwind_Reason_Code __cfactx_coroutine_unwindstop(
         __attribute((__unused__)) int version,
         _Unwind_Action actions,
 …
                 // We finished unwinding the coroutine,
                 // leave it
                 __leave_coroutine( param );
+                __cfactx_cor_leave( param );
                 __cabi_abort( "Resumed dead coroutine" );
+        }
 …
+}
 void _CtxCoroutine_Unwind(struct _Unwind_Exception * storage, struct coroutine_desc * cor) __attribute__ ((__noreturn__));
 void _CtxCoroutine_Unwind(struct _Unwind_Exception * storage, struct coroutine_desc * cor) {
         _Unwind_Reason_Code ret = _Unwind_ForcedUnwind( storage, _CtxCoroutine_UnwindStop, cor );
+void __cfactx_coroutine_unwind(struct _Unwind_Exception * storage, struct $coroutine * cor) __attribute__ ((__noreturn__));
+void __cfactx_coroutine_unwind(struct _Unwind_Exception * storage, struct $coroutine * cor) {
+        _Unwind_Reason_Code ret = _Unwind_ForcedUnwind( storage, __cfactx_coroutine_unwindstop, cor );
         printf("UNWIND ERROR %d after force unwind\n", ret);
         abort();
+}
 void CtxInvokeThread(
+void __cfactx_invoke_thread(
         void (*main)(void *),
         void *this
 …
         // The order of these 4 operations is very important
         //Final suspend, should never return
         __leave_thread_monitor();
+        __cfactx_thrd_leave();
         __cabi_abort( "Resumed dead thread" );
+}
 void CtxStart(
+void __cfactx_start(
         void (*main)(void *),
         struct coroutine_desc * cor,
+        struct $coroutine * cor,
         void *this,
         void (*invoke)(void *)
 …
         fs->dummyReturn = NULL;
         fs->rturn = CtxInvokeStub;
+        fs->rturn = __cfactx_invoke_stub;
         fs->fixedRegisters[0] = main;
         fs->fixedRegisters[1] = this;
 …
         struct FakeStack *fs = (struct FakeStack *)cor->context.SP;
         fs->intRegs[8] = CtxInvokeStub;
+        fs->intRegs[8] = __cfactx_invoke_stub;
         fs->arg[0] = this;
         fs->arg[1] = invoke;

libcfa/src/concurrency/invoke.h

-              rb7d6a36
+              r6a490b2
         extern "Cforall" {
                 extern __attribute__((aligned(128))) thread_local struct KernelThreadData {
                         struct thread_desc    * volatile this_thread;
+                        struct $thread    * volatile this_thread;
                         struct processor      * volatile this_processor;
 …
         };
+        enum coroutine_state { Halted, Start, Inactive, Active, Primed };
+        struct coroutine_desc {
+                // context that is switch during a CtxSwitch
+        enum coroutine_state { Halted, Start, Primed, Blocked, Ready, Active, Rerun };
+        enum __Preemption_Reason { __NO_PREEMPTION, __ALARM_PREEMPTION, __POLL_PREEMPTION, __MANUAL_PREEMPTION };
+        struct $coroutine {
+                // context that is switch during a __cfactx_switch
                 struct __stack_context_t context;
 …
                 // first coroutine to resume this one
                 struct coroutine_desc * starter;
+                struct $coroutine * starter;
                 // last coroutine to resume this one
                 struct coroutine_desc * last;
+                struct $coroutine * last;
                 // If non-null stack must be unwound with this exception
 …
         };
+        static inline struct __stack_t * __get_stack( struct $coroutine * cor ) { return (struct __stack_t*)(((uintptr_t)cor->stack.storage) & ((uintptr_t)-2)); }
         // struct which calls the monitor is accepting
 …
         };
         struct monitor_desc {
+        struct $monitor {
                 // spinlock to protect internal data
                 struct __spinlock_t lock;
                 // current owner of the monitor
                 struct thread_desc * owner;
+                struct $thread * owner;
                 // queue of threads that are blocked waiting for the monitor
                 __queue_t(struct thread_desc) entry_queue;
+                __queue_t(struct $thread) entry_queue;
                 // stack of conditions to run next once we exit the monitor
 …
         struct __monitor_group_t {
                 // currently held monitors
                 __cfa_anonymous_object( __small_array_t(monitor_desc*) );
+                __cfa_anonymous_object( __small_array_t($monitor*) );
                 // last function that acquired monitors
 …
         // instrusive link field for threads
         struct __thread_desc_link {
                 struct thread_desc * next;
                 struct thread_desc * prev;
+                struct $thread * next;
+                struct $thread * prev;
                 unsigned long long ts;
         };
         struct thread_desc {
+        struct $thread {
                 // Core threading fields
                 // context that is switch during a CtxSwitch
+                // context that is switch during a __cfactx_switch
                 struct __stack_context_t context;
                 // current execution status for coroutine
+                enum coroutine_state state;
+                volatile int state;
+                enum __Preemption_Reason preempted;
                 //SKULLDUGGERY errno is not save in the thread data structure because returnToKernel appears to be the only function to require saving and restoring it
                 // coroutine body used to store context
                 struct coroutine_desc  self_cor;
+                struct $coroutine  self_cor;
                 // current active context
                 struct coroutine_desc * curr_cor;
+                struct $coroutine * curr_cor;
                 // monitor body used for mutual exclusion
                 struct monitor_desc    self_mon;
+                struct $monitor    self_mon;
                 // pointer to monitor with sufficient lifetime for current monitors
                 struct monitor_desc *  self_mon_p;
+                struct $monitor *  self_mon_p;
                 // pointer to the cluster on which the thread is running
 …
                 struct {
                         struct thread_desc * next;
                         struct thread_desc * prev;
+                        struct $thread * next;
+                        struct $thread * prev;
                 } node;
+        };
+                #ifdef __CFA_DEBUG__
+                        // previous function to park/unpark the thread
+                        const char * park_caller;
+                        enum coroutine_state park_result;
+                        bool park_stale;
+                        const char * unpark_caller;
+                        enum coroutine_state unpark_result;
+                        bool unpark_stale;
+                #endif
+        };
+        #ifdef __CFA_DEBUG__
+                void __cfaabi_dbg_record_thrd($thread & this, bool park, const char prev_name[]);
+        #else
+                #define __cfaabi_dbg_record_thrd(x, y, z)
+        #endif
         #ifdef __cforall
         extern "Cforall" {
+                static inline thread_desc *& get_next( thread_desc & this ) {
+                static inline $thread *& get_next( $thread & this ) __attribute__((const)) {
                         return this.link.next;
+                }
                 static inline [thread_desc *&, thread_desc *& ] __get( thread_desc & this ) {
+                static inline [$thread *&, $thread *& ] __get( $thread & this ) __attribute__((const)) {
                         return this.node.[next, prev];
+                }
 …
+                }
                 static inline void ?{}(__monitor_group_t & this, struct monitor_desc ** data, __lock_size_t size, fptr_t func) {
+                static inline void ?{}(__monitor_group_t & this, struct $monitor ** data, __lock_size_t size, fptr_t func) {
                         (this.data){data};
                         (this.size){size};
 …
+                }
                 static inline bool ?==?( const __monitor_group_t & lhs, const __monitor_group_t & rhs ) {
+                static inline bool ?==?( const __monitor_group_t & lhs, const __monitor_group_t & rhs ) __attribute__((const)) {
                         if( (lhs.data != 0) != (rhs.data != 0) ) return false;
                         if( lhs.size != rhs.size ) return false;
 …
         // assembler routines that performs the context switch
         extern void CtxInvokeStub( void );
         extern void CtxSwitch( struct __stack_context_t * from, struct __stack_context_t * to ) asm ("CtxSwitch");
+        extern void __cfactx_invoke_stub( void );
+        extern void __cfactx_switch( struct __stack_context_t * from, struct __stack_context_t * to ) asm ("__cfactx_switch");
         // void CtxStore ( void * this ) asm ("CtxStore");
         // void CtxRet   ( void * dst  ) asm ("CtxRet");

libcfa/src/concurrency/kernel.cfa

-              rb7d6a36
+              r6a490b2
 #define __cforall_thread__
+// #define __CFA_DEBUG_PRINT_RUNTIME_CORE__
 //C Includes
 …
 #include "invoke.h"
 //-----------------------------------------------------------------------------
 // Some assembly required
 …
 //-----------------------------------------------------------------------------
 //Start and stop routine for the kernel, declared first to make sure they run first
+static void kernel_startup(void)  __attribute__(( constructor( STARTUP_PRIORITY_KERNEL ) ));
+static void kernel_shutdown(void) __attribute__(( destructor ( STARTUP_PRIORITY_KERNEL ) ));
+static void __kernel_startup (void) __attribute__(( constructor( STARTUP_PRIORITY_KERNEL ) ));
+static void __kernel_shutdown(void) __attribute__(( destructor ( STARTUP_PRIORITY_KERNEL ) ));
+//-----------------------------------------------------------------------------
+// Kernel Scheduling logic
+static $thread * __next_thread(cluster * this);
+static void __run_thread(processor * this, $thread * dst);
+static $thread * __halt(processor * this);
+static bool __wake_one(cluster * cltr, bool was_empty);
+static bool __wake_proc(processor *);
 //-----------------------------------------------------------------------------
 …
 KERNEL_STORAGE(cluster,         mainCluster);
 KERNEL_STORAGE(processor,       mainProcessor);
 KERNEL_STORAGE(thread_desc,     mainThread);
+KERNEL_STORAGE($thread, mainThread);
 KERNEL_STORAGE(__stack_t,       mainThreadCtx);
 cluster     * mainCluster;
 processor   * mainProcessor;
 thread_desc * mainThread;
+$thread * mainThread;
 extern "C" {
 …
 // Main thread construction
 void ?{}( coroutine_desc & this, current_stack_info_t * info) with( this ) {
+void ?{}( $coroutine & this, current_stack_info_t * info) with( this ) {
         stack.storage = info->storage;
         with(*stack.storage) {
 …
+}
 void ?{}( thread_desc & this, current_stack_info_t * info) with( this ) {
+void ?{}( $thread & this, current_stack_info_t * info) with( this ) {
         state = Start;
         self_cor{ info };
 …
+}
+static void start(processor * this);
+static void * __invoke_processor(void * arg);
 void ?{}(processor & this, const char name[], cluster & cltr) with( this ) {
         this.name = name;
 …
         id = -1u;
         terminated{ 0 };
+        destroyer = 0p;
         do_terminate = false;
         preemption_alarm = 0p;
 …
         runner.proc = &this;
+        idleLock{};
+        start( &this );
+        idle{};
+        __cfadbg_print_safe(runtime_core, "Kernel : Starting core %p\n", &this);
+        this.stack = __create_pthread( &this.kernel_thread, __invoke_processor, (void *)&this );
+        __cfadbg_print_safe(runtime_core, "Kernel : core %p created\n", &this);
+}
 void ^?{}(processor & this) with( this ){
         if( ! __atomic_load_n(&do_terminate, __ATOMIC_ACQUIRE) ) {
                 __cfaabi_dbg_print_safe("Kernel : core %p signaling termination\n", &this);
+                __cfadbg_print_safe(runtime_core, "Kernel : core %p signaling termination\n", &this);
                 __atomic_store_n(&do_terminate, true, __ATOMIC_RELAXED);
                 wake( &this );
+                __wake_proc( &this );
                 P( terminated );
 …
+        }
+        pthread_join( kernel_thread, 0p );
+        int err = pthread_join( kernel_thread, 0p );
+        if( err != 0 ) abort("KERNEL ERROR: joining processor %p caused error %s\n", &this, strerror(err));
         free( this.stack );
+}
 void ?{}(cluster & this, const char name[], Duration preemption_rate) with( this ) {
+void ?{}(cluster & this, const char name[], Duration preemption_rate, int io_flags) with( this ) {
         this.name = name;
         this.preemption_rate = preemption_rate;
 …
         ready_lock{};
+        #if !defined(__CFA_NO_STATISTICS__)
+                print_stats = false;
+        #endif
+        procs{ __get };
         idles{ __get };
         threads{ __get };
+        __kernel_io_startup( this, io_flags, &this == mainCluster );
         doregister(this);
+}
 void ^?{}(cluster & this) {
+        __kernel_io_shutdown( this, &this == mainCluster );
         unregister(this);
+}
 …
 // Kernel Scheduling logic
 //=============================================================================================
-static void runThread(processor * this, thread_desc * dst);
-static void finishRunning(processor * this);
-static void halt(processor * this);
 //Main of the processor contexts
 void main(processorCtx_t & runner) {
 …
         verify(this);
         __cfaabi_dbg_print_safe("Kernel : core %p starting\n", this);
+        __cfadbg_print_safe(runtime_core, "Kernel : core %p starting\n", this);
         // register the processor unless it's the main thread which is handled in the boot sequence
 …
                 preemption_scope scope = { this };
                 __cfaabi_dbg_print_safe("Kernel : core %p started\n", this);
                 thread_desc * readyThread = 0p;
+                __cfadbg_print_safe(runtime_core, "Kernel : core %p started\n", this);
+                $thread * readyThread = 0p;
                 for( unsigned int spin_count = 0; ! __atomic_load_n(&this->do_terminate, __ATOMIC_SEQ_CST); spin_count++ ) {
+                        readyThread = nextThread( this->cltr );
+                        if(readyThread) {
+                                verify( ! kernelTLS.preemption_state.enabled );
+                                runThread(this, readyThread);
+                                verify( ! kernelTLS.preemption_state.enabled );
+                                //Some actions need to be taken from the kernel
+                                finishRunning(this);
+                                spin_count = 0;
+                        } else {
+                                // spin(this, &spin_count);
+                                halt(this);
+                        // Try to get the next thread
+                        readyThread = __next_thread( this->cltr );
+                        // If no ready thread
+                        if( readyThread == 0p ) {
+                                // Block until a thread is ready
+                                readyThread = __halt(this);
+                        }
+                        // Check if we actually found a thread
+                        if( readyThread ) {
+                                /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+                                /* paranoid */ verifyf( readyThread->state == Ready || readyThread->preempted != __NO_PREEMPTION, "state : %d, preempted %d\n", readyThread->state, readyThread->preempted);
+                                /* paranoid */ verifyf( readyThread->next == 0p, "Expected null got %p", readyThread->next );
+                                // We found a thread run it
+                                __run_thread(this, readyThread);
+                                /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+                        }
+                }
                 __cfaabi_dbg_print_safe("Kernel : core %p stopping\n", this);
+                __cfadbg_print_safe(runtime_core, "Kernel : core %p stopping\n", this);
+        }
         V( this->terminated );
         // unregister the processor unless it's the main thread which is handled in the boot sequence
 …
                 unregister(this->cltr, this);
+        }
+        __cfaabi_dbg_print_safe("Kernel : core %p terminated\n", this);
+        else {
+                // HACK : the coroutine context switch expects this_thread to be set
+                // and it make sense for it to be set in all other cases except here
+                // fake it
+                kernelTLS.this_thread = mainThread;
+        }
+        __cfadbg_print_safe(runtime_core, "Kernel : core %p terminated\n", this);
         stats_tls_tally(this->cltr);
 …
 // runThread runs a thread by context switching
 // from the processor coroutine to the target thread
+static void runThread(processor * this, thread_desc * thrd_dst) {
+        coroutine_desc * proc_cor = get_coroutine(this->runner);
+        // Reset the terminating actions here
+        this->finish.action_code = No_Action;
+static void __run_thread(processor * this, $thread * thrd_dst) {
+        $coroutine * proc_cor = get_coroutine(this->runner);
         // Update global state
         kernelTLS.this_thread = thrd_dst;
+        // set state of processor coroutine to inactive and the thread to active
+        proc_cor->state = proc_cor->state == Halted ? Halted : Inactive;
+        thrd_dst->state = Active;
+        // set context switch to the thread that the processor is executing
+        verify( thrd_dst->context.SP );
+        CtxSwitch( &proc_cor->context, &thrd_dst->context );
+        // when CtxSwitch returns we are back in the processor coroutine
+        // set state of processor coroutine to active and the thread to inactive
+        thrd_dst->state = thrd_dst->state == Halted ? Halted : Inactive;
+        // set state of processor coroutine to inactive
+        verify(proc_cor->state == Active);
+        proc_cor->state = Blocked;
+        // Actually run the thread
+        RUNNING:  while(true) {
+                if(unlikely(thrd_dst->preempted)) {
+                        thrd_dst->preempted = __NO_PREEMPTION;
+                        verify(thrd_dst->state == Active  || thrd_dst->state == Rerun);
+                } else {
+                        verify(thrd_dst->state == Blocked || thrd_dst->state == Ready); // Ready means scheduled normally, blocked means rerun
+                        thrd_dst->state = Active;
+                }
+                __cfaabi_dbg_debug_do(
+                        thrd_dst->park_stale   = true;
+                        thrd_dst->unpark_stale = true;
+                )
+                /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+                /* paranoid */ verify( kernelTLS.this_thread == thrd_dst );
+                /* paranoid */ verifyf( ((uintptr_t)thrd_dst->context.SP) < ((uintptr_t)__get_stack(thrd_dst->curr_cor)->base ) || thrd_dst->curr_cor == proc_cor, "ERROR : Destination $thread %p has been corrupted.\n StackPointer too small.\n", thrd_dst ); // add escape condition if we are setting up the processor
+                /* paranoid */ verifyf( ((uintptr_t)thrd_dst->context.SP) > ((uintptr_t)__get_stack(thrd_dst->curr_cor)->limit) || thrd_dst->curr_cor == proc_cor, "ERROR : Destination $thread %p has been corrupted.\n StackPointer too large.\n", thrd_dst ); // add escape condition if we are setting up the processor
+                // set context switch to the thread that the processor is executing
+                verify( thrd_dst->context.SP );
+                __cfactx_switch( &proc_cor->context, &thrd_dst->context );
+                // when __cfactx_switch returns we are back in the processor coroutine
+                /* paranoid */ verifyf( ((uintptr_t)thrd_dst->context.SP) > ((uintptr_t)__get_stack(thrd_dst->curr_cor)->limit), "ERROR : Destination $thread %p has been corrupted.\n StackPointer too large.\n", thrd_dst );
+                /* paranoid */ verifyf( ((uintptr_t)thrd_dst->context.SP) < ((uintptr_t)__get_stack(thrd_dst->curr_cor)->base ), "ERROR : Destination $thread %p has been corrupted.\n StackPointer too small.\n", thrd_dst );
+                /* paranoid */ verify( kernelTLS.this_thread == thrd_dst );
+                /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+                // We just finished running a thread, there are a few things that could have happened.
+                // 1 - Regular case : the thread has blocked and now one has scheduled it yet.
+                // 2 - Racy case    : the thread has blocked but someone has already tried to schedule it.
+                // 4 - Preempted
+                // In case 1, we may have won a race so we can't write to the state again.
+                // In case 2, we lost the race so we now own the thread.
+                if(unlikely(thrd_dst->preempted != __NO_PREEMPTION)) {
+                        // The thread was preempted, reschedule it and reset the flag
+                        __schedule_thread( thrd_dst );
+                        break RUNNING;
+                }
+                // set state of processor coroutine to active and the thread to inactive
+                static_assert(sizeof(thrd_dst->state) == sizeof(int));
+                enum coroutine_state old_state = __atomic_exchange_n(&thrd_dst->state, Blocked, __ATOMIC_SEQ_CST);
+                __cfaabi_dbg_debug_do( thrd_dst->park_result = old_state; )
+                switch(old_state) {
+                        case Halted:
+                                // The thread has halted, it should never be scheduled/run again, leave it back to Halted and move on
+                                thrd_dst->state = Halted;
+                                // We may need to wake someone up here since
+                                unpark( this->destroyer __cfaabi_dbg_ctx2 );
+                                this->destroyer = 0p;
+                                break RUNNING;
+                        case Active:
+                                // This is case 1, the regular case, nothing more is needed
+                                break RUNNING;
+                        case Rerun:
+                                // This is case 2, the racy case, someone tried to run this thread before it finished blocking
+                                // In this case, just run it again.
+                                continue RUNNING;
+                        default:
+                                // This makes no sense, something is wrong abort
+                                abort("Finished running a thread that was Blocked/Start/Primed %d\n", old_state);
+                }
+        }
+        // Just before returning to the processor, set the processor coroutine to active
         proc_cor->state = Active;
+        kernelTLS.this_thread = 0p;
+}
 // KERNEL_ONLY
+static void returnToKernel() {
+        coroutine_desc * proc_cor = get_coroutine(kernelTLS.this_processor->runner);
+        thread_desc * thrd_src = kernelTLS.this_thread;
+        // set state of current coroutine to inactive
+        thrd_src->state = thrd_src->state == Halted ? Halted : Inactive;
+        proc_cor->state = Active;
+        int local_errno = *__volatile_errno();
+        #if defined( __i386 ) || defined( __x86_64 )
+                __x87_store;
+        #endif
+        // set new coroutine that the processor is executing
+        // and context switch to it
+        verify( proc_cor->context.SP );
+        CtxSwitch( &thrd_src->context, &proc_cor->context );
+        // set state of new coroutine to active
+        proc_cor->state = proc_cor->state == Halted ? Halted : Inactive;
+        thrd_src->state = Active;
+        #if defined( __i386 ) || defined( __x86_64 )
+                __x87_load;
+        #endif
+        *__volatile_errno() = local_errno;
+}
+// KERNEL_ONLY
+// Once a thread has finished running, some of
+// its final actions must be executed from the kernel
+static void finishRunning(processor * this) with( this->finish ) {
+        verify( ! kernelTLS.preemption_state.enabled );
+        choose( action_code ) {
+        case No_Action:
+                break;
+        case Release:
+                unlock( *lock );
+        case Schedule:
+                ScheduleThread( thrd );
+        case Release_Schedule:
+                unlock( *lock );
+                ScheduleThread( thrd );
+        case Release_Multi:
+                for(int i = 0; i < lock_count; i++) {
+                        unlock( *locks[i] );
+                }
+        case Release_Multi_Schedule:
+                for(int i = 0; i < lock_count; i++) {
+                        unlock( *locks[i] );
+                }
+                for(int i = 0; i < thrd_count; i++) {
+                        ScheduleThread( thrds[i] );
+                }
+        case Callback:
+                callback();
+        default:
+                abort("KERNEL ERROR: Unexpected action to run after thread");
+        }
+void returnToKernel() {
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        $coroutine * proc_cor = get_coroutine(kernelTLS.this_processor->runner);
+        $thread * thrd_src = kernelTLS.this_thread;
+        // Run the thread on this processor
+        {
+                int local_errno = *__volatile_errno();
+                #if defined( __i386 ) || defined( __x86_64 )
+                        __x87_store;
+                #endif
+                verify( proc_cor->context.SP );
+                __cfactx_switch( &thrd_src->context, &proc_cor->context );
+                #if defined( __i386 ) || defined( __x86_64 )
+                        __x87_load;
+                #endif
+                *__volatile_errno() = local_errno;
+        }
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        /* paranoid */ verifyf( ((uintptr_t)thrd_src->context.SP) < ((uintptr_t)__get_stack(thrd_src->curr_cor)->base ), "ERROR : Returning $thread %p has been corrupted.\n StackPointer too small.\n", thrd_src );
+        /* paranoid */ verifyf( ((uintptr_t)thrd_src->context.SP) > ((uintptr_t)__get_stack(thrd_src->curr_cor)->limit), "ERROR : Returning $thread %p has been corrupted.\n StackPointer too large.\n", thrd_src );
+}
 …
 // This is the entry point for processors (kernel threads)
 // It effectively constructs a coroutine by stealing the pthread stack
 static void * CtxInvokeProcessor(void * arg) {
+static void * __invoke_processor(void * arg) {
         processor * proc = (processor *) arg;
         kernelTLS.this_processor = proc;
 …
         //We now have a proper context from which to schedule threads
         __cfaabi_dbg_print_safe("Kernel : core %p created (%p, %p)\n", proc, &proc->runner, &ctx);
+        __cfadbg_print_safe(runtime_core, "Kernel : core %p created (%p, %p)\n", proc, &proc->runner, &ctx);
         // SKULLDUGGERY: Since the coroutine doesn't have its own stack, we can't
 …
         // Main routine of the core returned, the core is now fully terminated
         __cfaabi_dbg_print_safe("Kernel : core %p main ended (%p)\n", proc, &proc->runner);
+        __cfadbg_print_safe(runtime_core, "Kernel : core %p main ended (%p)\n", proc, &proc->runner);
         return 0p;
 …
 } // Abort
 void * create_pthread( pthread_t * pthread, void * (*start)(void *), void * arg ) {
+void * __create_pthread( pthread_t * pthread, void * (*start)(void *), void * arg ) {
         pthread_attr_t attr;
 …
+}
-static void start(processor * this) {
-        __cfaabi_dbg_print_safe("Kernel : Starting core %p\n", this);
-        this->stack = create_pthread( &this->kernel_thread, CtxInvokeProcessor, (void *)this );
-        __cfaabi_dbg_print_safe("Kernel : core %p started\n", this);
+}
 // KERNEL_ONLY
 void kernel_first_resume( processor * this ) {
         thread_desc * src = mainThread;
         coroutine_desc * dst = get_coroutine(this->runner);
+static void __kernel_first_resume( processor * this ) {
+        $thread * src = mainThread;
+        $coroutine * dst = get_coroutine(this->runner);
         verify( ! kernelTLS.preemption_state.enabled );
 …
         kernelTLS.this_thread->curr_cor = dst;
         __stack_prepare( &dst->stack, 65000 );
         CtxStart(main, dst, this->runner, CtxInvokeCoroutine);
+        __cfactx_start(main, dst, this->runner, __cfactx_invoke_coroutine);
         verify( ! kernelTLS.preemption_state.enabled );
 …
         dst->starter = dst->starter ? dst->starter : &src->self_cor;
         // set state of current coroutine to inactive
         src->state = src->state == Halted ? Halted : Inactive;
+        // make sure the current state is still correct
+        /* paranoid */ verify(src->state == Ready);
         // context switch to specified coroutine
         verify( dst->context.SP );
         CtxSwitch( &src->context, &dst->context );
         // when CtxSwitch returns we are back in the src coroutine
+        __cfactx_switch( &src->context, &dst->context );
+        // when __cfactx_switch returns we are back in the src coroutine
         mainThread->curr_cor = &mainThread->self_cor;
         // set state of new coroutine to active
         src->state = Active;
+        // make sure the current state has been update
+        /* paranoid */ verify(src->state == Active);
         verify( ! kernelTLS.preemption_state.enabled );
 …
 // KERNEL_ONLY
 void kernel_last_resume( processor * this ) {
         coroutine_desc * src = &mainThread->self_cor;
         coroutine_desc * dst = get_coroutine(this->runner);
+static void __kernel_last_resume( processor * this ) {
+        $coroutine * src = &mainThread->self_cor;
+        $coroutine * dst = get_coroutine(this->runner);
         verify( ! kernelTLS.preemption_state.enabled );
 …
         verify( dst->context.SP );
+        // SKULLDUGGERY in debug the processors check that the
+        // stack is still within the limit of the stack limits after running a thread.
+        // that check doesn't make sense if we context switch to the processor using the
+        // coroutine semantics. Since this is a special case, use the current context
+        // info to populate these fields.
+        __cfaabi_dbg_debug_do(
+                __stack_context_t ctx;
+                CtxGet( ctx );
+                mainThread->context.SP = ctx.SP;
+                mainThread->context.FP = ctx.FP;
+        )
         // context switch to the processor
         CtxSwitch( &src->context, &dst->context );
+        __cfactx_switch( &src->context, &dst->context );
+}
 //-----------------------------------------------------------------------------
 // Scheduler routines
 // KERNEL ONLY
+void ScheduleThread( thread_desc * thrd ) {
+        verify( thrd );
+        verify( thrd->state != Halted );
+        verify( ! kernelTLS.preemption_state.enabled );
+        verifyf( thrd->link.next == 0p, "Expected null got %p", thrd->link.next );
+void __schedule_thread( $thread * thrd ) {
+        /* paranoid */ verify( thrd );
+        /* paranoid */ verify( thrd->state != Halted );
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        /* paranoid */ #if defined( __CFA_WITH_VERIFY__ )
+        /* paranoid */ if( thrd->state == Blocked || thrd->state == Start ) assertf( thrd->preempted == __NO_PREEMPTION,
+                          "Error inactive thread marked as preempted, state %d, preemption %d\n", thrd->state, thrd->preempted );
+        /* paranoid */ if( thrd->preempted != __NO_PREEMPTION ) assertf(thrd->state == Active || thrd->state == Rerun,
+                          "Error preempted thread marked as not currently running, state %d, preemption %d\n", thrd->state, thrd->preempted );
+        /* paranoid */ #endif
+        /* paranoid */ verifyf( thrd->link.next == 0p, "Expected null got %p", thrd->link.next );
+        if (thrd->preempted == __NO_PREEMPTION) thrd->state = Ready;
         ready_schedule_lock(thrd->curr_cluster, kernelTLS.this_processor);
 …
         ready_schedule_unlock(thrd->curr_cluster, kernelTLS.this_processor);
+        with( *thrd->curr_cluster ) {
+                // if(was_empty) {
+                //      lock      (proc_list_lock __cfaabi_dbg_ctx2);
+                //      if(idles) {
+                //              wake_fast(idles.head);
+                //      }
+                //      unlock    (proc_list_lock);
+                // }
+                // else if( struct processor * idle = idles.head ) {
+                //      wake_fast(idle);
+                // }
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        __wake_one(thrd->curr_cluster, was_empty);
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+}
 // KERNEL ONLY
 thread_desc * nextThread(cluster * this) with( *this ) {
         verify( ! kernelTLS.preemption_state.enabled );
+static $thread * __next_thread(cluster * this) with( *this ) {
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
         ready_schedule_lock(this, kernelTLS.this_processor);
                 thread_desc * head = pop( this );
+                $thread * head = pop( this );
         ready_schedule_unlock(this, kernelTLS.this_processor);
         verify( ! kernelTLS.preemption_state.enabled );
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
         return head;
+}
+void BlockInternal() {
+// KERNEL ONLY unpark with out disabling interrupts
+void __unpark( $thread * thrd __cfaabi_dbg_ctx_param2 ) {
+        static_assert(sizeof(thrd->state) == sizeof(int));
+        // record activity
+        __cfaabi_dbg_record_thrd( *thrd, false, caller );
+        enum coroutine_state old_state = __atomic_exchange_n(&thrd->state, Rerun, __ATOMIC_SEQ_CST);
+        __cfaabi_dbg_debug_do( thrd->unpark_result = old_state; )
+        switch(old_state) {
+                case Active:
+                        // Wake won the race, the thread will reschedule/rerun itself
+                        break;
+                case Blocked:
+                        /* paranoid */ verify( ! thrd->preempted != __NO_PREEMPTION );
+                        // Wake lost the race,
+                        thrd->state = Blocked;
+                        __schedule_thread( thrd );
+                        break;
+                case Rerun:
+                        abort("More than one thread attempted to schedule thread %p\n", thrd);
+                        break;
+                case Halted:
+                case Start:
+                case Primed:
+                default:
+                        // This makes no sense, something is wrong abort
+                        abort();
+        }
+}
+void unpark( $thread * thrd __cfaabi_dbg_ctx_param2 ) {
+        if( !thrd ) return;
         disable_interrupts();
+        verify( ! kernelTLS.preemption_state.enabled );
+        __unpark( thrd __cfaabi_dbg_ctx_fwd2 );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void park( __cfaabi_dbg_ctx_param ) {
+        /* paranoid */ verify( kernelTLS.preemption_state.enabled );
+        disable_interrupts();
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        /* paranoid */ verify( kernelTLS.this_thread->preempted == __NO_PREEMPTION );
+        // record activity
+        __cfaabi_dbg_record_thrd( *kernelTLS.this_thread, true, caller );
         returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
         enable_interrupts( __cfaabi_dbg_ctx );
+}
+void BlockInternal( __spinlock_t * lock ) {
+        /* paranoid */ verify( kernelTLS.preemption_state.enabled );
+}
+// KERNEL ONLY
+void __leave_thread() {
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        abort();
+}
+// KERNEL ONLY
+bool force_yield( __Preemption_Reason reason ) {
+        /* paranoid */ verify( kernelTLS.preemption_state.enabled );
         disable_interrupts();
+        with( *kernelTLS.this_processor ) {
+                finish.action_code = Release;
+                finish.lock        = lock;
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void BlockInternal( thread_desc * thrd ) {
+        disable_interrupts();
+        with( * kernelTLS.this_processor ) {
+                finish.action_code = Schedule;
+                finish.thrd        = thrd;
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void BlockInternal( __spinlock_t * lock, thread_desc * thrd ) {
+        assert(thrd);
+        disable_interrupts();
+        with( * kernelTLS.this_processor ) {
+                finish.action_code = Release_Schedule;
+                finish.lock        = lock;
+                finish.thrd        = thrd;
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void BlockInternal(__spinlock_t * locks [], unsigned short count) {
+        disable_interrupts();
+        with( * kernelTLS.this_processor ) {
+                finish.action_code = Release_Multi;
+                finish.locks       = locks;
+                finish.lock_count  = count;
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void BlockInternal(__spinlock_t * locks [], unsigned short lock_count, thread_desc * thrds [], unsigned short thrd_count) {
+        disable_interrupts();
+        with( *kernelTLS.this_processor ) {
+                finish.action_code = Release_Multi_Schedule;
+                finish.locks       = locks;
+                finish.lock_count  = lock_count;
+                finish.thrds       = thrds;
+                finish.thrd_count  = thrd_count;
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void BlockInternal(__finish_callback_fptr_t callback) {
+        disable_interrupts();
+        with( *kernelTLS.this_processor ) {
+                finish.action_code = Callback;
+                finish.callback    = callback;
+        }
+        verify( ! kernelTLS.preemption_state.enabled );
+        returnToKernel();
+        verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+// KERNEL ONLY
+void LeaveThread(__spinlock_t * lock, thread_desc * thrd) {
+        verify( ! kernelTLS.preemption_state.enabled );
+        with( * kernelTLS.this_processor ) {
+                finish.action_code = thrd ? Release_Schedule : Release;
+                finish.lock        = lock;
+                finish.thrd        = thrd;
+        }
+        returnToKernel();
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        $thread * thrd = kernelTLS.this_thread;
+        /* paranoid */ verify(thrd->state == Active || thrd->state == Rerun);
+        // SKULLDUGGERY: It is possible that we are preempting this thread just before
+        // it was going to park itself. If that is the case and it is already using the
+        // intrusive fields then we can't use them to preempt the thread
+        // If that is the case, abandon the preemption.
+        bool preempted = false;
+        if(thrd->next == 0p) {
+                preempted = true;
+                thrd->preempted = reason;
+                returnToKernel();
+        }
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        enable_interrupts_noPoll();
+        /* paranoid */ verify( kernelTLS.preemption_state.enabled );
+        return preempted;
+}
 …
 //-----------------------------------------------------------------------------
 // Kernel boot procedures
 static void kernel_startup(void) {
+static void __kernel_startup(void) {
         verify( ! kernelTLS.preemption_state.enabled );
         __cfaabi_dbg_print_safe("Kernel : Starting\n");
+        __cfadbg_print_safe(runtime_core, "Kernel : Starting\n");
         __page_size = sysconf( _SC_PAGESIZE );
 …
         (*mainCluster){"Main Cluster"};
         __cfaabi_dbg_print_safe("Kernel : Main cluster ready\n");
+        __cfadbg_print_safe(runtime_core, "Kernel : Main cluster ready\n");
         // Start by initializing the main thread
         // SKULLDUGGERY: the mainThread steals the process main thread
         // which will then be scheduled by the mainProcessor normally
         mainThread = (thread_desc *)&storage_mainThread;
+        mainThread = ($thread *)&storage_mainThread;
         current_stack_info_t info;
         info.storage = (__stack_t*)&storage_mainThreadCtx;
         (*mainThread){ &info };
         __cfaabi_dbg_print_safe("Kernel : Main thread ready\n");
+        __cfadbg_print_safe(runtime_core, "Kernel : Main thread ready\n");
 …
                 runner{ &this };
                 __cfaabi_dbg_print_safe("Kernel : constructed main processor context %p\n", &runner);
+                __cfadbg_print_safe(runtime_core, "Kernel : constructed main processor context %p\n", &runner);
+        }
 …
         // Add the main thread to the ready queue
         // once resume is called on mainProcessor->runner the mainThread needs to be scheduled like any normal thread
         ScheduleThread(mainThread);
+        __schedule_thread(mainThread);
         // SKULLDUGGERY: Force a context switch to the main processor to set the main thread's context to the current UNIX
         // context. Hence, the main thread does not begin through CtxInvokeThread, like all other threads. The trick here is that
+        // context. Hence, the main thread does not begin through __cfactx_invoke_thread, like all other threads. The trick here is that
         // mainThread is on the ready queue when this call is made.
+        kernel_first_resume( kernelTLS.this_processor );
+        __kernel_first_resume( kernelTLS.this_processor );
         // THE SYSTEM IS NOW COMPLETELY RUNNING
+        __cfaabi_dbg_print_safe("Kernel : Started\n--------------------------------------------------\n\n");
+        // Now that the system is up, finish creating systems that need threading
+        __kernel_io_finish_start( *mainCluster );
+        __cfadbg_print_safe(runtime_core, "Kernel : Started\n--------------------------------------------------\n\n");
         verify( ! kernelTLS.preemption_state.enabled );
 …
+}
+static void kernel_shutdown(void) {
+        __cfaabi_dbg_print_safe("\n--------------------------------------------------\nKernel : Shutting down\n");
+        verify( TL_GET( preemption_state.enabled ) );
+static void __kernel_shutdown(void) {
+        //Before we start shutting things down, wait for systems that need threading to shutdown
+        __kernel_io_prepare_stop( *mainCluster );
+        /* paranoid */ verify( TL_GET( preemption_state.enabled ) );
         disable_interrupts();
+        verify( ! kernelTLS.preemption_state.enabled );
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        __cfadbg_print_safe(runtime_core, "\n--------------------------------------------------\nKernel : Shutting down\n");
         // SKULLDUGGERY: Notify the mainProcessor it needs to terminates.
 …
         // which is currently here
         __atomic_store_n(&mainProcessor->do_terminate, true, __ATOMIC_RELEASE);
         kernel_last_resume( kernelTLS.this_processor );
+        __kernel_last_resume( kernelTLS.this_processor );
         mainThread->self_cor.state = Halted;
 …
         // Destroy the main processor and its context in reverse order of construction
         // These were manually constructed so we need manually destroy them
         void ^?{}(processor & this) with( this ) {
                 //don't join the main thread here, that wouldn't make any sense
+        void ^?{}(processor & this) with( this ){
+                /* paranoid */ verify( this.do_terminate == true );
                 __cfaabi_dbg_print_safe("Kernel : destroyed main processor context %p\n", &runner);
+        }
 …
         // Final step, destroy the main thread since it is no longer needed
+        // Since we provided a stack to this task it will not destroy anything
+        // Since we provided a stack to this taxk it will not destroy anything
+        /* paranoid */ verify(mainThread->self_cor.stack.storage == (__stack_t*)(((uintptr_t)&storage_mainThreadCtx)| 0x1));
         ^(*mainThread){};
 …
         ^(__cfa_dbg_global_clusters.lock){};
         __cfaabi_dbg_print_safe("Kernel : Shutdown complete\n");
+        __cfadbg_print_safe(runtime_core, "Kernel : Shutdown complete\n");
+}
 //=============================================================================================
 // Kernel Quiescing
+// Kernel Idle Sleep
 //=============================================================================================
+static void halt(processor * this) with( *this ) {
+        // // verify( ! __atomic_load_n(&do_terminate, __ATOMIC_SEQ_CST) );
+        // with( *cltr ) {
+        //      lock      (proc_list_lock __cfaabi_dbg_ctx2);
+        //      push_front(idles, *this);
+        //      unlock    (proc_list_lock);
+        // }
+        // __cfaabi_dbg_print_safe("Kernel : Processor %p ready to sleep\n", this);
+        // wait( idleLock );
+        // __cfaabi_dbg_print_safe("Kernel : Processor %p woke up and ready to run\n", this);
+        // with( *cltr ) {
+        //      lock      (proc_list_lock __cfaabi_dbg_ctx2);
+        //      remove    (idles, *this);
+        //      unlock    (proc_list_lock);
+        // }
+static $thread * __halt(processor * this) with( *this ) {
+        if( do_terminate ) return 0p;
+        // First, lock the cluster idle
+        lock( cltr->idle_lock __cfaabi_dbg_ctx2 );
+        // Check if we can find a thread
+        if( $thread * found = __next_thread( cltr ) ) {
+                unlock( cltr->idle_lock );
+                return found;
+        }
+        // Move this processor from the active list to the idle list
+        move_to_front(cltr->procs, cltr->idles, *this);
+        // Unlock the idle lock so we don't go to sleep with a lock
+        unlock    (cltr->idle_lock);
+        // We are ready to sleep
+        __cfadbg_print_safe(runtime_core, "Kernel : Processor %p ready to sleep\n", this);
+        wait( idle );
+        // We have woken up
+        __cfadbg_print_safe(runtime_core, "Kernel : Processor %p woke up and ready to run\n", this);
+        // Get ourself off the idle list
+        with( *cltr ) {
+                lock  (idle_lock __cfaabi_dbg_ctx2);
+                move_to_front(idles, procs, *this);
+                unlock(idle_lock);
+        }
+        // Don't check the ready queue again, we may not be in a position to run a thread
+        return 0p;
+}
+// Wake a thread from the front if there are any
+static bool __wake_one(cluster * this, __attribute__((unused)) bool force) {
+        // if we don't want to force check if we know it's false
+        // if( !this->idles.head && !force ) return false;
+        // First, lock the cluster idle
+        lock( this->idle_lock __cfaabi_dbg_ctx2 );
+        // Check if there is someone to wake up
+        if( !this->idles.head ) {
+                // Nope unlock and return false
+                unlock( this->idle_lock );
+                return false;
+        }
+        // Wake them up
+        __cfadbg_print_safe(runtime_core, "Kernel : waking Processor %p\n", this->idles.head);
+        /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+        post( this->idles.head->idle );
+        // Unlock and return true
+        unlock( this->idle_lock );
+        return true;
+}
+// Unconditionnaly wake a thread
+static bool __wake_proc(processor * this) {
+        __cfadbg_print_safe(runtime_core, "Kernel : waking Processor %p\n", this);
+        disable_interrupts();
+                /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+                bool ret = post( this->idle );
+        enable_interrupts( __cfaabi_dbg_ctx );
+        return ret;
+}
 …
 void kernel_abort_msg( void * kernel_data, char * abort_text, int abort_text_size ) {
         thread_desc * thrd = kernel_data;
+        $thread * thrd = kernel_data;
         if(thrd) {
 …
 void ^?{}(semaphore & this) {}
 void P(semaphore & this) with( this ){
+bool P(semaphore & this) with( this ){
         lock( lock __cfaabi_dbg_ctx2 );
         count -= 1;
 …
                 // atomically release spin lock and block
+                BlockInternal( &lock );
+                unlock( lock );
+                park( __cfaabi_dbg_ctx );
+                return true;
+        }
         else {
             unlock( lock );
+        }
+}
+void V(semaphore & this) with( this ) {
+        thread_desc * thrd = 0p;
+            return false;
+        }
+}
+bool V(semaphore & this) with( this ) {
+        $thread * thrd = 0p;
         lock( lock __cfaabi_dbg_ctx2 );
         count += 1;
 …
         // make new owner
+        WakeThread( thrd );
+        unpark( thrd __cfaabi_dbg_ctx2 );
+        return thrd != 0p;
+}
+bool V(semaphore & this, unsigned diff) with( this ) {
+        $thread * thrd = 0p;
+        lock( lock __cfaabi_dbg_ctx2 );
+        int release = max(-count, (int)diff);
+        count += diff;
+        for(release) {
+                unpark( pop_head( waiting ) __cfaabi_dbg_ctx2 );
+        }
+        unlock( lock );
+        return thrd != 0p;
+}
 …
+}
 void doregister( cluster * cltr, thread_desc & thrd ) {
+void doregister( cluster * cltr, $thread & thrd ) {
         lock      (cltr->thread_list_lock __cfaabi_dbg_ctx2);
         cltr->nthreads += 1;
 …
+}
 void unregister( cluster * cltr, thread_desc & thrd ) {
+void unregister( cluster * cltr, $thread & thrd ) {
         lock  (cltr->thread_list_lock __cfaabi_dbg_ctx2);
         remove(cltr->threads, thrd );
 …
 __cfaabi_dbg_debug_do(
         extern "C" {
                 void __cfaabi_dbg_record(__spinlock_t & this, const char prev_name[]) {
+                void __cfaabi_dbg_record_lock(__spinlock_t & this, const char prev_name[]) {
                         this.prev_name = prev_name;
                         this.prev_thrd = kernelTLS.this_thread;
+                }
+                void __cfaabi_dbg_record_thrd($thread & this, bool park, const char prev_name[]) {
+                        if(park) {
+                                this.park_caller   = prev_name;
+                                this.park_stale    = false;
+                        }
+                        else {
+                                this.unpark_caller = prev_name;
+                                this.unpark_stale  = false;
+                        }
+                }
+        }
+)
 …
 //-----------------------------------------------------------------------------
 // Debug
 bool threading_enabled(void) {
+bool threading_enabled(void) __attribute__((const)) {
         return true;
+}

libcfa/src/concurrency/kernel.hfa

-              rb7d6a36
+              r6a490b2
 #include <stdbool.h>
+#include <stdint.h>
 #include "invoke.h"
 …
         __spinlock_t lock;
         int count;
         __queue_t(thread_desc) waiting;
+        __queue_t($thread) waiting;
 };
 void  ?{}(semaphore & this, int count = 1);
 void ^?{}(semaphore & this);
+void   P (semaphore & this);
+void   V (semaphore & this);
+bool   P (semaphore & this);
+bool   V (semaphore & this);
+bool   V (semaphore & this, unsigned count);
 …
 // Processor
 extern struct cluster * mainCluster;
-enum FinishOpCode { No_Action, Release, Schedule, Release_Schedule, Release_Multi, Release_Multi_Schedule, Callback };
-typedef void (*__finish_callback_fptr_t)(void);
-//TODO use union, many of these fields are mutually exclusive (i.e. MULTI vs NOMULTI)
-struct FinishAction {
-        FinishOpCode action_code;
-        /*
-        // Union of possible actions
-        union {
-                // Option 1 : locks and threads
-                struct {
-                        // 1 thread or N thread
-                        union {
-                                thread_desc * thrd;
-                                struct {
-                                        thread_desc ** thrds;
-                                        unsigned short thrd_count;
-                                };
-                        };
-                        // 1 lock or N lock
-                        union {
-                                __spinlock_t * lock;
-                                struct {
-                                        __spinlock_t ** locks;
-                                        unsigned short lock_count;
-                                };
-                        };
-                };
-                // Option 2 : action pointer
-                __finish_callback_fptr_t callback;
-        };
-        /*/
-        thread_desc * thrd;
-        thread_desc ** thrds;
-        unsigned short thrd_count;
-        __spinlock_t * lock;
-        __spinlock_t ** locks;
-        unsigned short lock_count;
-        __finish_callback_fptr_t callback;
-        //*/
-};
-static inline void ?{}(FinishAction & this) {
-        this.action_code = No_Action;
-        this.thrd = 0p;
-        this.lock = 0p;
+}
-static inline void ^?{}(FinishAction &) {}
 // Processor
 …
         // RunThread data
         // Action to do after a thread is ran
         struct FinishAction finish;
+        $thread * destroyer;
         // Preemption data
 …
         bool pending_preemption;
         // Idle lock
         __bin_sem_t idleLock;
+        // Idle lock (kernel semaphore)
+        __bin_sem_t idle;
         // Termination
 …
         volatile bool do_terminate;
         // Termination synchronisation
+        // Termination synchronisation (user semaphore)
         semaphore terminated;
 …
 static inline void  ?{}(processor & this, const char name[]) { this{name, *mainCluster }; }
+static inline [processor *&, processor *& ] __get( processor & this ) {
+        return this.node.[next, prev];
+}
+static inline [processor *&, processor *& ] __get( processor & this ) __attribute__((const)) { return this.node.[next, prev]; }
+//-----------------------------------------------------------------------------
+// I/O
+struct __io_data;
+#define CFA_CLUSTER_IO_POLLER_USER_THREAD 1 << 0
+// #define CFA_CLUSTER_IO_POLLER_KERNEL_SIDE 1 << 1
 …
         // List of threads
         __spinlock_t thread_list_lock;
         __dllist_t(struct thread_desc) threads;
+        __dllist_t(struct $thread) threads;
         unsigned int nthreads;
 …
                 cluster * prev;
         } node;
+        struct __io_data * io;
+        #if !defined(__CFA_NO_STATISTICS__)
+                bool print_stats;
+        #endif
 };
 extern Duration default_preemption();
 void ?{} (cluster & this, const char name[], Duration preemption_rate);
+void ?{} (cluster & this, const char name[], Duration preemption_rate, int flags);
 void ^?{}(cluster & this);
+static inline void ?{} (cluster & this)                           { this{"Anonymous Cluster", default_preemption()}; }
+static inline void ?{} (cluster & this, Duration preemption_rate) { this{"Anonymous Cluster", preemption_rate}; }
+static inline void ?{} (cluster & this, const char name[])        { this{name, default_preemption()}; }
+static inline [cluster *&, cluster *& ] __get( cluster & this ) {
+        return this.node.[next, prev];
+}
+static inline void ?{} (cluster & this)                                      { this{"Anonymous Cluster", default_preemption(), 0}; }
+static inline void ?{} (cluster & this, Duration preemption_rate)            { this{"Anonymous Cluster", preemption_rate, 0}; }
+static inline void ?{} (cluster & this, const char name[])                   { this{name, default_preemption(), 0}; }
+static inline void ?{} (cluster & this, int flags)                           { this{"Anonymous Cluster", default_preemption(), flags}; }
+static inline void ?{} (cluster & this, Duration preemption_rate, int flags) { this{"Anonymous Cluster", preemption_rate, flags}; }
+static inline void ?{} (cluster & this, const char name[], int flags)        { this{name, default_preemption(), flags}; }
+static inline [cluster *&, cluster *& ] __get( cluster & this ) __attribute__((const)) { return this.node.[next, prev]; }
 static inline struct processor * active_processor() { return TL_GET( this_processor ); } // UNSAFE
 static inline struct cluster   * active_cluster  () { return TL_GET( this_processor )->cltr; }
+#if !defined(__CFA_NO_STATISTICS__)
+        static inline void print_stats_at_exit( cluster & this ) {
+                this.print_stats = true;
+        }
+#endif
 // Local Variables: //

libcfa/src/concurrency/kernel_private.hfa

-              rb7d6a36
+              r6a490b2
+}
+void ScheduleThread( thread_desc * );
+static inline void WakeThread( thread_desc * thrd ) {
+        if( !thrd ) return;
+        verify(thrd->state == Inactive);
+        disable_interrupts();
+        ScheduleThread( thrd );
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+thread_desc * nextThread(cluster * this);
+void __schedule_thread( $thread * ) __attribute__((nonnull (1)));
 //Block current thread and release/wake-up the following resources
+void BlockInternal(void);
+void BlockInternal(__spinlock_t * lock);
+void BlockInternal(thread_desc * thrd);
+void BlockInternal(__spinlock_t * lock, thread_desc * thrd);
+void BlockInternal(__spinlock_t * locks [], unsigned short count);
+void BlockInternal(__spinlock_t * locks [], unsigned short count, thread_desc * thrds [], unsigned short thrd_count);
+void BlockInternal(__finish_callback_fptr_t callback);
+void LeaveThread(__spinlock_t * lock, thread_desc * thrd);
+void __leave_thread() __attribute__((noreturn));
 //-----------------------------------------------------------------------------
 …
 void main(processorCtx_t *);
+void * create_pthread( pthread_t *, void * (*)(void *), void * );
+static inline void wake_fast(processor * this) {
+        __cfaabi_dbg_print_safe("Kernel : Waking up processor %p\n", this);
+        post( this->idleLock );
+}
+static inline void wake(processor * this) {
+        disable_interrupts();
+        wake_fast(this);
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+void * __create_pthread( pthread_t *, void * (*)(void *), void * );
 struct event_kernel_t {
 …
 extern volatile thread_local __cfa_kernel_preemption_state_t preemption_state __attribute__ ((tls_model ( "initial-exec" )));
+extern cluster * mainCluster;
 //-----------------------------------------------------------------------------
 // Threads
 extern "C" {
+      void CtxInvokeThread(void (*main)(void *), void * this);
+}
+extern void ThreadCtxSwitch(coroutine_desc * src, coroutine_desc * dst);
+      void __cfactx_invoke_thread(void (*main)(void *), void * this);
+}
 __cfaabi_dbg_debug_do(
         extern void __cfaabi_dbg_thread_register  ( thread_desc * thrd );
         extern void __cfaabi_dbg_thread_unregister( thread_desc * thrd );
+        extern void __cfaabi_dbg_thread_register  ( $thread * thrd );
+        extern void __cfaabi_dbg_thread_unregister( $thread * thrd );
+)
+// KERNEL ONLY unpark with out disabling interrupts
+void __unpark( $thread * thrd __cfaabi_dbg_ctx_param2 );
+//-----------------------------------------------------------------------------
+// I/O
+void __kernel_io_startup     ( cluster &, int, bool );
+void __kernel_io_finish_start( cluster & );
+void __kernel_io_prepare_stop( cluster & );
+void __kernel_io_shutdown    ( cluster &, bool );
 //-----------------------------------------------------------------------------
 …
 #define KERNEL_STORAGE(T,X) __attribute((aligned(__alignof__(T)))) static char storage_##X[sizeof(T)]
 static inline uint32_t tls_rand() {
+static inline uint32_t __tls_rand() {
         kernelTLS.rand_seed ^= kernelTLS.rand_seed << 6;
         kernelTLS.rand_seed ^= kernelTLS.rand_seed >> 21;
 …
 void unregister( struct cluster & cltr );
 void doregister( struct cluster * cltr, struct thread_desc & thrd );
 void unregister( struct cluster * cltr, struct thread_desc & thrd );
+void doregister( struct cluster * cltr, struct $thread & thrd );
+void unregister( struct cluster * cltr, struct $thread & thrd );
 //=======================================================================

libcfa/src/concurrency/monitor.cfa

-              rb7d6a36
+              r6a490b2
 // file "LICENCE" distributed with Cforall.
 //
 // monitor_desc.c --
+// $monitor.c --
 //
 // Author           : Thierry Delisle
 …
 //-----------------------------------------------------------------------------
 // Forward declarations
 static inline void set_owner ( monitor_desc * this, thread_desc * owner );
 static inline void set_owner ( monitor_desc * storage [], __lock_size_t count, thread_desc * owner );
 static inline void set_mask  ( monitor_desc * storage [], __lock_size_t count, const __waitfor_mask_t & mask );
 static inline void reset_mask( monitor_desc * this );
 static inline thread_desc * next_thread( monitor_desc * this );
 static inline bool is_accepted( monitor_desc * this, const __monitor_group_t & monitors );
+static inline void __set_owner ( $monitor * this, $thread * owner );
+static inline void __set_owner ( $monitor * storage [], __lock_size_t count, $thread * owner );
+static inline void set_mask  ( $monitor * storage [], __lock_size_t count, const __waitfor_mask_t & mask );
+static inline void reset_mask( $monitor * this );
+static inline $thread * next_thread( $monitor * this );
+static inline bool is_accepted( $monitor * this, const __monitor_group_t & monitors );
 static inline void lock_all  ( __spinlock_t * locks [], __lock_size_t count );
 static inline void lock_all  ( monitor_desc * source [], __spinlock_t * /*out*/ locks [], __lock_size_t count );
+static inline void lock_all  ( $monitor * source [], __spinlock_t * /*out*/ locks [], __lock_size_t count );
 static inline void unlock_all( __spinlock_t * locks [], __lock_size_t count );
 static inline void unlock_all( monitor_desc * locks [], __lock_size_t count );
 static inline void save   ( monitor_desc * ctx [], __lock_size_t count, __spinlock_t * locks [], unsigned int /*out*/ recursions [], __waitfor_mask_t /*out*/ masks [] );
 static inline void restore( monitor_desc * ctx [], __lock_size_t count, __spinlock_t * locks [], unsigned int /*in */ recursions [], __waitfor_mask_t /*in */ masks [] );
 static inline void init     ( __lock_size_t count, monitor_desc * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] );
 static inline void init_push( __lock_size_t count, monitor_desc * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] );
 static inline thread_desc *        check_condition   ( __condition_criterion_t * );
+static inline void unlock_all( $monitor * locks [], __lock_size_t count );
+static inline void save   ( $monitor * ctx [], __lock_size_t count, __spinlock_t * locks [], unsigned int /*out*/ recursions [], __waitfor_mask_t /*out*/ masks [] );
+static inline void restore( $monitor * ctx [], __lock_size_t count, __spinlock_t * locks [], unsigned int /*in */ recursions [], __waitfor_mask_t /*in */ masks [] );
+static inline void init     ( __lock_size_t count, $monitor * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] );
+static inline void init_push( __lock_size_t count, $monitor * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] );
+static inline $thread *        check_condition   ( __condition_criterion_t * );
 static inline void                 brand_condition   ( condition & );
 static inline [thread_desc *, int] search_entry_queue( const __waitfor_mask_t &, monitor_desc * monitors [], __lock_size_t count );
+static inline [$thread *, int] search_entry_queue( const __waitfor_mask_t &, $monitor * monitors [], __lock_size_t count );
 forall(dtype T | sized( T ))
 static inline __lock_size_t insert_unique( T * array [], __lock_size_t & size, T * val );
 static inline __lock_size_t count_max    ( const __waitfor_mask_t & mask );
 static inline __lock_size_t aggregate    ( monitor_desc * storage [], const __waitfor_mask_t & mask );
+static inline __lock_size_t aggregate    ( $monitor * storage [], const __waitfor_mask_t & mask );
 //-----------------------------------------------------------------------------
 …
 #define monitor_ctx( mons, cnt )                                /* Define that create the necessary struct for internal/external scheduling operations */ \
         monitor_desc ** monitors = mons;                          /* Save the targeted monitors                                                          */ \
+        $monitor ** monitors = mons;                          /* Save the targeted monitors                                                          */ \
         __lock_size_t count = cnt;                                /* Save the count to a local variable                                                  */ \
         unsigned int recursions[ count ];                         /* Save the current recursion levels to restore them later                             */ \
 …
 //-----------------------------------------------------------------------------
 // Enter/Leave routines
+extern "C" {
+        // Enter single monitor
+        static void __enter_monitor_desc( monitor_desc * this, const __monitor_group_t & group ) {
+                // Lock the monitor spinlock
+                lock( this->lock __cfaabi_dbg_ctx2 );
+                // Interrupts disable inside critical section
+                thread_desc * thrd = kernelTLS.this_thread;
+                __cfaabi_dbg_print_safe( "Kernel : %10p Entering mon %p (%p)\n", thrd, this, this->owner);
+                if( !this->owner ) {
+                        // No one has the monitor, just take it
+                        set_owner( this, thrd );
+                        __cfaabi_dbg_print_safe( "Kernel :  mon is free \n" );
+                }
+                else if( this->owner == thrd) {
+                        // We already have the monitor, just note how many times we took it
+                        this->recursion += 1;
+                        __cfaabi_dbg_print_safe( "Kernel :  mon already owned \n" );
+                }
+                else if( is_accepted( this, group) ) {
+                        // Some one was waiting for us, enter
+                        set_owner( this, thrd );
+                        // Reset mask
+                        reset_mask( this );
+                        __cfaabi_dbg_print_safe( "Kernel :  mon accepts \n" );
+                }
+                else {
+                        __cfaabi_dbg_print_safe( "Kernel :  blocking \n" );
+                        // Some one else has the monitor, wait in line for it
+                        append( this->entry_queue, thrd );
+                        BlockInternal( &this->lock );
+                        __cfaabi_dbg_print_safe( "Kernel : %10p Entered  mon %p\n", thrd, this);
+                        // BlockInternal will unlock spinlock, no need to unlock ourselves
+                        return;
+                }
+// Enter single monitor
+static void __enter( $monitor * this, const __monitor_group_t & group ) {
+        // Lock the monitor spinlock
+        lock( this->lock __cfaabi_dbg_ctx2 );
+        // Interrupts disable inside critical section
+        $thread * thrd = kernelTLS.this_thread;
+        __cfaabi_dbg_print_safe( "Kernel : %10p Entering mon %p (%p)\n", thrd, this, this->owner);
+        if( !this->owner ) {
+                // No one has the monitor, just take it
+                __set_owner( this, thrd );
+                __cfaabi_dbg_print_safe( "Kernel :  mon is free \n" );
+        }
+        else if( this->owner == thrd) {
+                // We already have the monitor, just note how many times we took it
+                this->recursion += 1;
+                __cfaabi_dbg_print_safe( "Kernel :  mon already owned \n" );
+        }
+        else if( is_accepted( this, group) ) {
+                // Some one was waiting for us, enter
+                __set_owner( this, thrd );
+                // Reset mask
+                reset_mask( this );
+                __cfaabi_dbg_print_safe( "Kernel :  mon accepts \n" );
+        }
+        else {
+                __cfaabi_dbg_print_safe( "Kernel :  blocking \n" );
+                // Some one else has the monitor, wait in line for it
+                /* paranoid */ verify( thrd->next == 0p );
+                append( this->entry_queue, thrd );
+                /* paranoid */ verify( thrd->next == 1p );
+                unlock( this->lock );
+                park( __cfaabi_dbg_ctx );
                 __cfaabi_dbg_print_safe( "Kernel : %10p Entered  mon %p\n", thrd, this);
+                // Release the lock and leave
+                /* paranoid */ verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+                return;
+        }
+        __cfaabi_dbg_print_safe( "Kernel : %10p Entered  mon %p\n", thrd, this);
+        /* paranoid */ verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+        /* paranoid */ verify( this->lock.lock );
+        // Release the lock and leave
+        unlock( this->lock );
+        return;
+}
+static void __dtor_enter( $monitor * this, fptr_t func ) {
+        // Lock the monitor spinlock
+        lock( this->lock __cfaabi_dbg_ctx2 );
+        // Interrupts disable inside critical section
+        $thread * thrd = kernelTLS.this_thread;
+        __cfaabi_dbg_print_safe( "Kernel : %10p Entering dtor for mon %p (%p)\n", thrd, this, this->owner);
+        if( !this->owner ) {
+                __cfaabi_dbg_print_safe( "Kernel : Destroying free mon %p\n", this);
+                // No one has the monitor, just take it
+                __set_owner( this, thrd );
+                verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
                 unlock( this->lock );
                 return;
+        }
+        static void __enter_monitor_dtor( monitor_desc * this, fptr_t func ) {
+                // Lock the monitor spinlock
+                lock( this->lock __cfaabi_dbg_ctx2 );
+                // Interrupts disable inside critical section
+                thread_desc * thrd = kernelTLS.this_thread;
+                __cfaabi_dbg_print_safe( "Kernel : %10p Entering dtor for mon %p (%p)\n", thrd, this, this->owner);
+                if( !this->owner ) {
+                        __cfaabi_dbg_print_safe( "Kernel : Destroying free mon %p\n", this);
+                        // No one has the monitor, just take it
+                        set_owner( this, thrd );
+                        unlock( this->lock );
+                        return;
+        else if( this->owner == thrd) {
+                // We already have the monitor... but where about to destroy it so the nesting will fail
+                // Abort!
+                abort( "Attempt to destroy monitor %p by thread \"%.256s\" (%p) in nested mutex.", this, thrd->self_cor.name, thrd );
+        }
+        __lock_size_t count = 1;
+        $monitor ** monitors = &this;
+        __monitor_group_t group = { &this, 1, func };
+        if( is_accepted( this, group) ) {
+                __cfaabi_dbg_print_safe( "Kernel :  mon accepts dtor, block and signal it \n" );
+                // Wake the thread that is waiting for this
+                __condition_criterion_t * urgent = pop( this->signal_stack );
+                /* paranoid */ verify( urgent );
+                // Reset mask
+                reset_mask( this );
+                // Create the node specific to this wait operation
+                wait_ctx_primed( thrd, 0 )
+                // Some one else has the monitor, wait for him to finish and then run
+                unlock( this->lock );
+                // Release the next thread
+                /* paranoid */ verifyf( urgent->owner->waiting_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+                unpark( urgent->owner->waiting_thread __cfaabi_dbg_ctx2 );
+                // Park current thread waiting
+                park( __cfaabi_dbg_ctx );
+                // Some one was waiting for us, enter
+                /* paranoid */ verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+        }
+        else {
+                __cfaabi_dbg_print_safe( "Kernel :  blocking \n" );
+                wait_ctx( thrd, 0 )
+                this->dtor_node = &waiter;
+                // Some one else has the monitor, wait in line for it
+                /* paranoid */ verify( thrd->next == 0p );
+                append( this->entry_queue, thrd );
+                /* paranoid */ verify( thrd->next == 1p );
+                unlock( this->lock );
+                // Park current thread waiting
+                park( __cfaabi_dbg_ctx );
+                /* paranoid */ verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+                return;
+        }
+        __cfaabi_dbg_print_safe( "Kernel : Destroying %p\n", this);
+}
+// Leave single monitor
+void __leave( $monitor * this ) {
+        // Lock the monitor spinlock
+        lock( this->lock __cfaabi_dbg_ctx2 );
+        __cfaabi_dbg_print_safe( "Kernel : %10p Leaving mon %p (%p)\n", kernelTLS.this_thread, this, this->owner);
+        /* paranoid */ verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+        // Leaving a recursion level, decrement the counter
+        this->recursion -= 1;
+        // If we haven't left the last level of recursion
+        // it means we don't need to do anything
+        if( this->recursion != 0) {
+                __cfaabi_dbg_print_safe( "Kernel :  recursion still %d\n", this->recursion);
+                unlock( this->lock );
+                return;
+        }
+        // Get the next thread, will be null on low contention monitor
+        $thread * new_owner = next_thread( this );
+        // Check the new owner is consistent with who we wake-up
+        // new_owner might be null even if someone owns the monitor when the owner is still waiting for another monitor
+        /* paranoid */ verifyf( !new_owner || new_owner == this->owner, "Expected owner to be %p, got %p (m: %p)", new_owner, this->owner, this );
+        // We can now let other threads in safely
+        unlock( this->lock );
+        //We need to wake-up the thread
+        /* paranoid */ verifyf( !new_owner || new_owner == this->owner, "Expected owner to be %p, got %p (m: %p)", new_owner, this->owner, this );
+        unpark( new_owner __cfaabi_dbg_ctx2 );
+}
+// Leave single monitor for the last time
+void __dtor_leave( $monitor * this ) {
+        __cfaabi_dbg_debug_do(
+                if( TL_GET( this_thread ) != this->owner ) {
+                        abort( "Destroyed monitor %p has inconsistent owner, expected %p got %p.\n", this, TL_GET( this_thread ), this->owner);
+                }
+                else if( this->owner == thrd) {
+                        // We already have the monitor... but where about to destroy it so the nesting will fail
+                        // Abort!
+                        abort( "Attempt to destroy monitor %p by thread \"%.256s\" (%p) in nested mutex.", this, thrd->self_cor.name, thrd );
+                if( this->recursion != 1 ) {
+                        abort( "Destroyed monitor %p has %d outstanding nested calls.\n", this, this->recursion - 1);
+                }
+                __lock_size_t count = 1;
+                monitor_desc ** monitors = &this;
+                __monitor_group_t group = { &this, 1, func };
+                if( is_accepted( this, group) ) {
+                        __cfaabi_dbg_print_safe( "Kernel :  mon accepts dtor, block and signal it \n" );
+                        // Wake the thread that is waiting for this
+                        __condition_criterion_t * urgent = pop( this->signal_stack );
+                        verify( urgent );
+                        // Reset mask
+                        reset_mask( this );
+                        // Create the node specific to this wait operation
+                        wait_ctx_primed( thrd, 0 )
+                        // Some one else has the monitor, wait for him to finish and then run
+                        BlockInternal( &this->lock, urgent->owner->waiting_thread );
+                        // Some one was waiting for us, enter
+                        set_owner( this, thrd );
+                }
+                else {
+                        __cfaabi_dbg_print_safe( "Kernel :  blocking \n" );
+                        wait_ctx( thrd, 0 )
+                        this->dtor_node = &waiter;
+                        // Some one else has the monitor, wait in line for it
+                        append( this->entry_queue, thrd );
+                        BlockInternal( &this->lock );
+                        // BlockInternal will unlock spinlock, no need to unlock ourselves
+                        return;
+                }
+                __cfaabi_dbg_print_safe( "Kernel : Destroying %p\n", this);
+        }
+        // Leave single monitor
+        void __leave_monitor_desc( monitor_desc * this ) {
+                // Lock the monitor spinlock
+                lock( this->lock __cfaabi_dbg_ctx2 );
+                __cfaabi_dbg_print_safe( "Kernel : %10p Leaving mon %p (%p)\n", kernelTLS.this_thread, this, this->owner);
+                verifyf( kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+                // Leaving a recursion level, decrement the counter
+                this->recursion -= 1;
+                // If we haven't left the last level of recursion
+                // it means we don't need to do anything
+                if( this->recursion != 0) {
+                        __cfaabi_dbg_print_safe( "Kernel :  recursion still %d\n", this->recursion);
+                        unlock( this->lock );
+                        return;
+                }
+                // Get the next thread, will be null on low contention monitor
+                thread_desc * new_owner = next_thread( this );
+                // We can now let other threads in safely
+                unlock( this->lock );
+                //We need to wake-up the thread
+                WakeThread( new_owner );
+        }
+        // Leave single monitor for the last time
+        void __leave_dtor_monitor_desc( monitor_desc * this ) {
+                __cfaabi_dbg_debug_do(
+                        if( TL_GET( this_thread ) != this->owner ) {
+                                abort( "Destroyed monitor %p has inconsistent owner, expected %p got %p.\n", this, TL_GET( this_thread ), this->owner);
+                        }
+                        if( this->recursion != 1 ) {
+                                abort( "Destroyed monitor %p has %d outstanding nested calls.\n", this, this->recursion - 1);
+                        }
+                )
+        }
+        )
+}
+extern "C" {
         // Leave the thread monitor
         // last routine called by a thread.
         // Should never return
         void __leave_thread_monitor() {
                 thread_desc * thrd = TL_GET( this_thread );
                 monitor_desc * this = &thrd->self_mon;
+        void __cfactx_thrd_leave() {
+                $thread * thrd = TL_GET( this_thread );
+                $monitor * this = &thrd->self_mon;
                 // Lock the monitor now
 …
                 disable_interrupts();
                 thrd->self_cor.state = Halted;
                 verifyf( thrd == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", thrd, this->owner, this->recursion, this );
+                thrd->state = Halted;
+                /* paranoid */ verifyf( thrd == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", thrd, this->owner, this->recursion, this );
                 // Leaving a recursion level, decrement the counter
 …
                 // Fetch the next thread, can be null
+                thread_desc * new_owner = next_thread( this );
+                // Leave the thread, this will unlock the spinlock
+                // Use leave thread instead of BlockInternal which is
+                // specialized for this case and supports null new_owner
+                LeaveThread( &this->lock, new_owner );
+                $thread * new_owner = next_thread( this );
+                // Release the monitor lock
+                unlock( this->lock );
+                // Unpark the next owner if needed
+                /* paranoid */ verifyf( !new_owner || new_owner == this->owner, "Expected owner to be %p, got %p (m: %p)", new_owner, this->owner, this );
+                /* paranoid */ verify( ! kernelTLS.preemption_state.enabled );
+                /* paranoid */ verify( ! kernelTLS.this_processor->destroyer );
+                /* paranoid */ verify( thrd->state == Halted );
+                kernelTLS.this_processor->destroyer = new_owner;
+                // Leave the thread
+                __leave_thread();
                 // Control flow should never reach here!
 …
 static inline void enter( __monitor_group_t monitors ) {
         for( __lock_size_t i = 0; i < monitors.size; i++) {
                 __enter_monitor_desc( monitors[i], monitors );
+                __enter( monitors[i], monitors );
+        }
+}
 …
 // Leave multiple monitor
 // relies on the monitor array being sorted
 static inline void leave(monitor_desc * monitors [], __lock_size_t count) {
+static inline void leave($monitor * monitors [], __lock_size_t count) {
         for( __lock_size_t i = count - 1; i >= 0; i--) {
                 __leave_monitor_desc( monitors[i] );
+                __leave( monitors[i] );
+        }
+}
 …
 // Ctor for monitor guard
 // Sorts monitors before entering
 void ?{}( monitor_guard_t & this, monitor_desc * m [], __lock_size_t count, fptr_t func ) {
         thread_desc * thrd = TL_GET( this_thread );
+void ?{}( monitor_guard_t & this, $monitor * m [], __lock_size_t count, fptr_t func ) {
+        $thread * thrd = TL_GET( this_thread );
         // Store current array
 …
 // Ctor for monitor guard
 // Sorts monitors before entering
 void ?{}( monitor_dtor_guard_t & this, monitor_desc * m [], fptr_t func ) {
+void ?{}( monitor_dtor_guard_t & this, $monitor * m [], fptr_t func ) {
         // optimization
         thread_desc * thrd = TL_GET( this_thread );
+        $thread * thrd = TL_GET( this_thread );
         // Store current array
 …
         (thrd->monitors){m, 1, func};
         __enter_monitor_dtor( this.m, func );
+        __dtor_enter( this.m, func );
+}
 …
 void ^?{}( monitor_dtor_guard_t & this ) {
         // Leave the monitors in order
         __leave_dtor_monitor_desc( this.m );
+        __dtor_leave( this.m );
         // Restore thread context
 …
 //-----------------------------------------------------------------------------
 // Internal scheduling types
 void ?{}(__condition_node_t & this, thread_desc * waiting_thread, __lock_size_t count, uintptr_t user_info ) {
+void ?{}(__condition_node_t & this, $thread * waiting_thread, __lock_size_t count, uintptr_t user_info ) {
         this.waiting_thread = waiting_thread;
         this.count = count;
 …
+}
 void ?{}(__condition_criterion_t & this, monitor_desc * target, __condition_node_t & owner ) {
+void ?{}(__condition_criterion_t & this, $monitor * target, __condition_node_t & owner ) {
         this.ready  = false;
         this.target = target;
 …
         // Append the current wait operation to the ones already queued on the condition
         // We don't need locks for that since conditions must always be waited on inside monitor mutual exclusion
+        /* paranoid */ verify( waiter.next == 0p );
         append( this.blocked, &waiter );
+        /* paranoid */ verify( waiter.next == 1p );
         // Lock all monitors (aggregates the locks as well)
 …
         // Find the next thread(s) to run
         __lock_size_t thread_count = 0;
         thread_desc * threads[ count ];
+        $thread * threads[ count ];
         __builtin_memset( threads, 0, sizeof( threads ) );
 …
         // Remove any duplicate threads
         for( __lock_size_t i = 0; i < count; i++) {
                 thread_desc * new_owner = next_thread( monitors[i] );
+                $thread * new_owner = next_thread( monitors[i] );
                 insert_unique( threads, thread_count, new_owner );
+        }
+        // Unlock the locks, we don't need them anymore
+        for(int i = 0; i < count; i++) {
+                unlock( *locks[i] );
+        }
+        // Wake the threads
+        for(int i = 0; i < thread_count; i++) {
+                unpark( threads[i] __cfaabi_dbg_ctx2 );
+        }
         // Everything is ready to go to sleep
         BlockInternal( locks, count, threads, thread_count );
+        park( __cfaabi_dbg_ctx );
         // We are back, restore the owners and recursions
 …
         //Some more checking in debug
         __cfaabi_dbg_debug_do(
                 thread_desc * this_thrd = TL_GET( this_thread );
+                $thread * this_thrd = TL_GET( this_thread );
                 if ( this.monitor_count != this_thrd->monitors.size ) {
                         abort( "Signal on condition %p made with different number of monitor(s), expected %zi got %zi", &this, this.monitor_count, this_thrd->monitors.size );
 …
         //Find the thread to run
         thread_desc * signallee = pop_head( this.blocked )->waiting_thread;
         set_owner( monitors, count, signallee );
+        $thread * signallee = pop_head( this.blocked )->waiting_thread;
+        __set_owner( monitors, count, signallee );
         __cfaabi_dbg_print_buffer_decl( "Kernel : signal_block condition %p (s: %p)\n", &this, signallee );
+        // unlock all the monitors
+        unlock_all( locks, count );
+        // unpark the thread we signalled
+        unpark( signallee __cfaabi_dbg_ctx2 );
         //Everything is ready to go to sleep
         BlockInternal( locks, count, &signallee, 1 );
+        park( __cfaabi_dbg_ctx );
 …
         // Create one!
         __lock_size_t max = count_max( mask );
         monitor_desc * mon_storage[max];
+        $monitor * mon_storage[max];
         __builtin_memset( mon_storage, 0, sizeof( mon_storage ) );
         __lock_size_t actual_count = aggregate( mon_storage, mask );
 …
+        {
                 // Check if the entry queue
                 thread_desc * next; int index;
+                $thread * next; int index;
                 [next, index] = search_entry_queue( mask, monitors, count );
 …
                                 verifyf( accepted.size == 1,  "ERROR: Accepted dtor has more than 1 mutex parameter." );
                                 monitor_desc * mon2dtor = accepted[0];
+                                $monitor * mon2dtor = accepted[0];
                                 verifyf( mon2dtor->dtor_node, "ERROR: Accepted monitor has no dtor_node." );
 …
                                 // Set the owners to be the next thread
+                                set_owner( monitors, count, next );
+                                // Everything is ready to go to sleep
+                                BlockInternal( locks, count, &next, 1 );
+                                __set_owner( monitors, count, next );
+                                // unlock all the monitors
+                                unlock_all( locks, count );
+                                // unpark the thread we signalled
+                                unpark( next __cfaabi_dbg_ctx2 );
+                                //Everything is ready to go to sleep
+                                park( __cfaabi_dbg_ctx );
                                 // We are back, restore the owners and recursions
 …
+        }
+        // unlock all the monitors
+        unlock_all( locks, count );
         //Everything is ready to go to sleep
         BlockInternal( locks, count );
+        park( __cfaabi_dbg_ctx );
 …
 // Utilities
 static inline void set_owner( monitor_desc * this, thread_desc * owner ) {
         // __cfaabi_dbg_print_safe( "Kernal :   Setting owner of %p to %p ( was %p)\n", this, owner, this->owner );
+static inline void __set_owner( $monitor * this, $thread * owner ) {
+        /* paranoid */ verify( this->lock.lock );
         //Pass the monitor appropriately
 …
+}
+static inline void set_owner( monitor_desc * monitors [], __lock_size_t count, thread_desc * owner ) {
+        monitors[0]->owner     = owner;
+        monitors[0]->recursion = 1;
+static inline void __set_owner( $monitor * monitors [], __lock_size_t count, $thread * owner ) {
+        /* paranoid */ verify ( monitors[0]->lock.lock );
+        /* paranoid */ verifyf( monitors[0]->owner == kernelTLS.this_thread, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, monitors[0]->owner, monitors[0]->recursion, monitors[0] );
+        monitors[0]->owner        = owner;
+        monitors[0]->recursion    = 1;
         for( __lock_size_t i = 1; i < count; i++ ) {
+                monitors[i]->owner     = owner;
+                monitors[i]->recursion = 0;
+        }
+}
+static inline void set_mask( monitor_desc * storage [], __lock_size_t count, const __waitfor_mask_t & mask ) {
+                /* paranoid */ verify ( monitors[i]->lock.lock );
+                /* paranoid */ verifyf( monitors[i]->owner == kernelTLS.this_thread, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, monitors[i]->owner, monitors[i]->recursion, monitors[i] );
+                monitors[i]->owner        = owner;
+                monitors[i]->recursion    = 0;
+        }
+}
+static inline void set_mask( $monitor * storage [], __lock_size_t count, const __waitfor_mask_t & mask ) {
         for( __lock_size_t i = 0; i < count; i++) {
                 storage[i]->mask = mask;
 …
+}
 static inline void reset_mask( monitor_desc * this ) {
+static inline void reset_mask( $monitor * this ) {
         this->mask.accepted = 0p;
         this->mask.data = 0p;
 …
+}
 static inline thread_desc * next_thread( monitor_desc * this ) {
+static inline $thread * next_thread( $monitor * this ) {
         //Check the signaller stack
         __cfaabi_dbg_print_safe( "Kernel :  mon %p AS-stack top %p\n", this, this->signal_stack.top);
 …
                 //regardless of if we are ready to baton pass,
                 //we need to set the monitor as in use
+                set_owner( this,  urgent->owner->waiting_thread );
+                /* paranoid */ verifyf( !this->owner || kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+                __set_owner( this,  urgent->owner->waiting_thread );
                 return check_condition( urgent );
 …
         // No signaller thread
         // Get the next thread in the entry_queue
+        thread_desc * new_owner = pop_head( this->entry_queue );
+        set_owner( this, new_owner );
+        $thread * new_owner = pop_head( this->entry_queue );
+        /* paranoid */ verifyf( !this->owner || kernelTLS.this_thread == this->owner, "Expected owner to be %p, got %p (r: %i, m: %p)", kernelTLS.this_thread, this->owner, this->recursion, this );
+        /* paranoid */ verify( !new_owner || new_owner->next == 0p );
+        __set_owner( this, new_owner );
         return new_owner;
+}
 static inline bool is_accepted( monitor_desc * this, const __monitor_group_t & group ) {
+static inline bool is_accepted( $monitor * this, const __monitor_group_t & group ) {
         __acceptable_t * it = this->mask.data; // Optim
         __lock_size_t count = this->mask.size;
 …
+}
 static inline void init( __lock_size_t count, monitor_desc * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] ) {
+static inline void init( __lock_size_t count, $monitor * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] ) {
         for( __lock_size_t i = 0; i < count; i++) {
                 (criteria[i]){ monitors[i], waiter };
 …
+}
 static inline void init_push( __lock_size_t count, monitor_desc * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] ) {
+static inline void init_push( __lock_size_t count, $monitor * monitors [], __condition_node_t & waiter, __condition_criterion_t criteria [] ) {
         for( __lock_size_t i = 0; i < count; i++) {
                 (criteria[i]){ monitors[i], waiter };
 …
+}
 static inline void lock_all( monitor_desc * source [], __spinlock_t * /*out*/ locks [], __lock_size_t count ) {
+static inline void lock_all( $monitor * source [], __spinlock_t * /*out*/ locks [], __lock_size_t count ) {
         for( __lock_size_t i = 0; i < count; i++ ) {
                 __spinlock_t * l = &source[i]->lock;
 …
+}
 static inline void unlock_all( monitor_desc * locks [], __lock_size_t count ) {
+static inline void unlock_all( $monitor * locks [], __lock_size_t count ) {
         for( __lock_size_t i = 0; i < count; i++ ) {
                 unlock( locks[i]->lock );
 …
 static inline void save(
         monitor_desc * ctx [],
+        $monitor * ctx [],
         __lock_size_t count,
         __attribute((unused)) __spinlock_t * locks [],
 …
 static inline void restore(
         monitor_desc * ctx [],
+        $monitor * ctx [],
         __lock_size_t count,
         __spinlock_t * locks [],
 …
 // 2 - Checks if all the monitors are ready to run
 //     if so return the thread to run
 static inline thread_desc * check_condition( __condition_criterion_t * target ) {
+static inline $thread * check_condition( __condition_criterion_t * target ) {
         __condition_node_t * node = target->owner;
         unsigned short count = node->count;
 …
 static inline void brand_condition( condition & this ) {
         thread_desc * thrd = TL_GET( this_thread );
+        $thread * thrd = TL_GET( this_thread );
         if( !this.monitors ) {
                 // __cfaabi_dbg_print_safe( "Branding\n" );
 …
                 this.monitor_count = thrd->monitors.size;
                 this.monitors = (monitor_desc **)malloc( this.monitor_count * sizeof( *this.monitors ) );
+                this.monitors = ($monitor **)malloc( this.monitor_count * sizeof( *this.monitors ) );
                 for( int i = 0; i < this.monitor_count; i++ ) {
                         this.monitors[i] = thrd->monitors[i];
 …
+}
 static inline [thread_desc *, int] search_entry_queue( const __waitfor_mask_t & mask, monitor_desc * monitors [], __lock_size_t count ) {
         __queue_t(thread_desc) & entry_queue = monitors[0]->entry_queue;
+static inline [$thread *, int] search_entry_queue( const __waitfor_mask_t & mask, $monitor * monitors [], __lock_size_t count ) {
+        __queue_t($thread) & entry_queue = monitors[0]->entry_queue;
         // For each thread in the entry-queue
         for(    thread_desc ** thrd_it = &entry_queue.head;
+        for(    $thread ** thrd_it = &entry_queue.head;
                 *thrd_it;
                 thrd_it = &(*thrd_it)->link.next
 …
+}
 static inline __lock_size_t aggregate( monitor_desc * storage [], const __waitfor_mask_t & mask ) {
+static inline __lock_size_t aggregate( $monitor * storage [], const __waitfor_mask_t & mask ) {
         __lock_size_t size = 0;
         for( __lock_size_t i = 0; i < mask.size; i++ ) {

libcfa/src/concurrency/monitor.hfa

-              rb7d6a36
+              r6a490b2
 trait is_monitor(dtype T) {
         monitor_desc * get_monitor( T & );
+        $monitor * get_monitor( T & );
         void ^?{}( T & mutex );
 };
 static inline void ?{}(monitor_desc & this) with( this ) {
+static inline void ?{}($monitor & this) with( this ) {
         lock{};
         entry_queue{};
 …
+}
 static inline void ^?{}(monitor_desc & ) {}
+static inline void ^?{}($monitor & ) {}
 struct monitor_guard_t {
         monitor_desc **         m;
+        $monitor **     m;
         __lock_size_t           count;
         __monitor_group_t prev;
 };
 void ?{}( monitor_guard_t & this, monitor_desc ** m, __lock_size_t count, void (*func)() );
+void ?{}( monitor_guard_t & this, $monitor ** m, __lock_size_t count, void (*func)() );
 void ^?{}( monitor_guard_t & this );
 struct monitor_dtor_guard_t {
         monitor_desc *    m;
+        $monitor *    m;
         __monitor_group_t prev;
 };
 void ?{}( monitor_dtor_guard_t & this, monitor_desc ** m, void (*func)() );
+void ?{}( monitor_dtor_guard_t & this, $monitor ** m, void (*func)() );
 void ^?{}( monitor_dtor_guard_t & this );
 …
         // The monitor this criterion concerns
         monitor_desc * target;
+        $monitor * target;
         // The parent node to which this criterion belongs
 …
 struct __condition_node_t {
         // Thread that needs to be woken when all criteria are met
         thread_desc * waiting_thread;
+        $thread * waiting_thread;
         // Array of criteria (Criterions are contiguous in memory)
 …
+}
 void ?{}(__condition_node_t & this, thread_desc * waiting_thread, __lock_size_t count, uintptr_t user_info );
+void ?{}(__condition_node_t & this, $thread * waiting_thread, __lock_size_t count, uintptr_t user_info );
 void ?{}(__condition_criterion_t & this );
 void ?{}(__condition_criterion_t & this, monitor_desc * target, __condition_node_t * owner );
+void ?{}(__condition_criterion_t & this, $monitor * target, __condition_node_t * owner );
 struct condition {
 …
         // Array of monitor pointers (Monitors are NOT contiguous in memory)
         monitor_desc ** monitors;
+        $monitor ** monitors;
         // Number of monitors in the array
 …
               bool signal      ( condition & this );
               bool signal_block( condition & this );
 static inline bool is_empty    ( condition & this ) { return !this.blocked.head; }
+static inline bool is_empty    ( condition & this ) { return this.blocked.head == 1p; }
          uintptr_t front       ( condition & this );

libcfa/src/concurrency/mutex.cfa

-              rb7d6a36
+              r6a490b2
         if( is_locked ) {
                 append( blocked_threads, kernelTLS.this_thread );
+                BlockInternal( &lock );
+                unlock( lock );
+                park( __cfaabi_dbg_ctx );
+        }
         else {
 …
         lock( this.lock __cfaabi_dbg_ctx2 );
         this.is_locked = (this.blocked_threads != 0);
         WakeThread(
                 pop_head( this.blocked_threads )
+        unpark(
+                pop_head( this.blocked_threads ) __cfaabi_dbg_ctx2
         );
         unlock( this.lock );
 …
         else {
                 append( blocked_threads, kernelTLS.this_thread );
+                BlockInternal( &lock );
+                unlock( lock );
+                park( __cfaabi_dbg_ctx );
+        }
+}
 …
         recursion_count--;
         if( recursion_count == 0 ) {
                 thread_desc * thrd = pop_head( blocked_threads );
+                $thread * thrd = pop_head( blocked_threads );
                 owner = thrd;
                 recursion_count = (thrd ? 1 : 0);
                 WakeThread( thrd );
+                unpark( thrd __cfaabi_dbg_ctx2 );
+        }
         unlock( lock );
 …
 void notify_one(condition_variable & this) with(this) {
         lock( lock __cfaabi_dbg_ctx2 );
         WakeThread(
                 pop_head( this.blocked_threads )
+        unpark(
+                pop_head( this.blocked_threads ) __cfaabi_dbg_ctx2
         );
         unlock( lock );
 …
         lock( lock __cfaabi_dbg_ctx2 );
         while(this.blocked_threads) {
                 WakeThread(
                         pop_head( this.blocked_threads )
+                unpark(
+                        pop_head( this.blocked_threads ) __cfaabi_dbg_ctx2
                 );
+        }
 …
         lock( this.lock __cfaabi_dbg_ctx2 );
         append( this.blocked_threads, kernelTLS.this_thread );
+        BlockInternal( &this.lock );
+        unlock( this.lock );
+        park( __cfaabi_dbg_ctx );
+}
 …
         lock( this.lock __cfaabi_dbg_ctx2 );
         append( this.blocked_threads, kernelTLS.this_thread );
+        void __unlock(void) {
+                unlock(l);
+                unlock(this.lock);
+        }
+        BlockInternal( __unlock );
+        unlock(l);
+        unlock(this.lock);
+        park( __cfaabi_dbg_ctx );
         lock(l);
+}

libcfa/src/concurrency/mutex.hfa

-              rb7d6a36
+              r6a490b2
         // List of blocked threads
         __queue_t(struct thread_desc) blocked_threads;
+        __queue_t(struct $thread) blocked_threads;
         // Locked flag
 …
         // List of blocked threads
         __queue_t(struct thread_desc) blocked_threads;
+        __queue_t(struct $thread) blocked_threads;
         // Current thread owning the lock
         struct thread_desc * owner;
+        struct $thread * owner;
         // Number of recursion level
 …
         // List of blocked threads
         __queue_t(struct thread_desc) blocked_threads;
+        __queue_t(struct $thread) blocked_threads;
 };

libcfa/src/concurrency/preemption.cfa

-              rb7d6a36
+              r6a490b2
 // FwdDeclarations : timeout handlers
 static void preempt( processor   * this );
 static void timeout( thread_desc * this );
+static void timeout( $thread * this );
 // FwdDeclarations : Signal handlers
 static void sigHandler_ctxSwitch( __CFA_SIGPARMS__ );
+static void sigHandler_alarm    ( __CFA_SIGPARMS__ );
 static void sigHandler_segv     ( __CFA_SIGPARMS__ );
 static void sigHandler_ill      ( __CFA_SIGPARMS__ );
 …
 // Get next expired node
 static inline alarm_node_t * get_expired( alarm_list_t * alarms, Time currtime ) {
         if( !alarms->head ) return 0p;                                          // If no alarms return null
         if( alarms->head->alarm >= currtime ) return 0p;        // If alarms head not expired return null
+        if( ! & (*alarms)`first ) return 0p;                                            // If no alarms return null
+        if( (*alarms)`first.alarm >= currtime ) return 0p;      // If alarms head not expired return null
         return pop(alarms);                                                                     // Otherwise just pop head
+}
 …
         while( node = get_expired( alarms, currtime ) ) {
                 // __cfaabi_dbg_print_buffer_decl( " KERNEL: preemption tick.\n" );
+                Duration period = node->period;
+                if( period == 0) {
+                        node->set = false;                  // Node is one-shot, just mark it as not pending
+                }
                 // Check if this is a kernel
 …
                 // Check if this is a periodic alarm
-                Duration period = node->period;
                 if( period > 0 ) {
                         // __cfaabi_dbg_print_buffer_local( " KERNEL: alarm period is %lu.\n", period.tv );
 …
                         insert( alarms, node );             // Reinsert the node for the next time it triggers
+                }
-                else {
-                        node->set = false;                  // Node is one-shot, just mark it as not pending
+                }
+        }
         // If there are still alarms pending, reset the timer
         if( alarms->head ) {
                 // __cfaabi_dbg_print_buffer_decl( " KERNEL: @%ju(%ju) resetting alarm to %ju.\n", currtime.tv, __kernel_get_time().tv, (alarms->head->alarm - currtime).tv);
                 Duration delta = alarms->head->alarm - currtime;
                 Duration caped = max(delta, 50`us);
+        if( & (*alarms)`first ) {
+                __cfadbg_print_buffer_decl(preemption, " KERNEL: @%ju(%ju) resetting alarm to %ju.\n", currtime.tv, __kernel_get_time().tv, (alarms->head->alarm - currtime).tv);
+                Duration delta = (*alarms)`first.alarm - currtime;
+                Duration capped = max(delta, 50`us);
                 // itimerval tim  = { caped };
                 // __cfaabi_dbg_print_buffer_local( "    Values are %lu, %lu, %lu %lu.\n", delta.tv, caped.tv, tim.it_value.tv_sec, tim.it_value.tv_usec);
                 __kernel_set_timer( caped );
+                __kernel_set_timer( capped );
+        }
+}
 …
         // Enable interrupts by decrementing the counter
         // If counter reaches 0, execute any pending CtxSwitch
+        // If counter reaches 0, execute any pending __cfactx_switch
         void enable_interrupts( __cfaabi_dbg_ctx_param ) {
                 processor   * proc = kernelTLS.this_processor; // Cache the processor now since interrupts can start happening after the atomic store
-                thread_desc * thrd = kernelTLS.this_thread;       // Cache the thread now since interrupts can start happening after the atomic store
                 with( kernelTLS.preemption_state ){
 …
                                 if( proc->pending_preemption ) {
                                         proc->pending_preemption = false;
                                         BlockInternal( thrd );
+                                        force_yield( __POLL_PREEMPTION );
+                                }
+                        }
 …
         // Disable interrupts by incrementint the counter
         // Don't execute any pending CtxSwitch even if counter reaches 0
+        // Don't execute any pending __cfactx_switch even if counter reaches 0
         void enable_interrupts_noPoll() {
                 unsigned short prev = kernelTLS.preemption_state.disable_count;
 …
         if ( pthread_sigmask( SIG_BLOCK, &mask, 0p ) == -1 ) {
             abort( "internal error, pthread_sigmask" );
+                abort( "internal error, pthread_sigmask" );
+        }
+}
 …
 // reserved for future use
 static void timeout( thread_desc * this ) {
         //TODO : implement waking threads
+static void timeout( $thread * this ) {
+        __unpark( this __cfaabi_dbg_ctx2 );
+}
 // KERNEL ONLY
 // Check if a CtxSwitch signal handler shoud defer
+// Check if a __cfactx_switch signal handler shoud defer
 // If true  : preemption is safe
 // If false : preemption is unsafe and marked as pending
 …
         // Setup proper signal handlers
+        __cfaabi_sigaction( SIGUSR1, sigHandler_ctxSwitch, SA_SIGINFO | SA_RESTART ); // CtxSwitch handler
+        __cfaabi_sigaction( SIGUSR1, sigHandler_ctxSwitch, SA_SIGINFO | SA_RESTART ); // __cfactx_switch handler
+        __cfaabi_sigaction( SIGALRM, sigHandler_alarm    , SA_SIGINFO | SA_RESTART ); // debug handler
         signal_block( SIGALRM );
         alarm_stack = create_pthread( &alarm_thread, alarm_loop, 0p );
+        alarm_stack = __create_pthread( &alarm_thread, alarm_loop, 0p );
+}
 …
         // Preemption can occur here
+        BlockInternal( kernelTLS.this_thread ); // Do the actual CtxSwitch
+        force_yield( __ALARM_PREEMPTION ); // Do the actual __cfactx_switch
+}
+static void sigHandler_alarm( __CFA_SIGPARMS__ ) {
+        abort("SIGALRM should never reach the signal handler");
+}

libcfa/src/concurrency/thread.cfa

-              rb7d6a36
+              r6a490b2
 #include "invoke.h"
-extern "C" {
-        #include <fenv.h>
-        #include <stddef.h>
+}
-//extern volatile thread_local processor * this_processor;
 //-----------------------------------------------------------------------------
 // Thread ctors and dtors
 void ?{}(thread_desc & this, const char * const name, cluster & cl, void * storage, size_t storageSize ) with( this ) {
+void ?{}($thread & this, const char * const name, cluster & cl, void * storage, size_t storageSize ) with( this ) {
         context{ 0p, 0p };
         self_cor{ name, storage, storageSize };
         state = Start;
+        preempted = __NO_PREEMPTION;
         curr_cor = &self_cor;
         self_mon.owner = &this;
 …
+}
 void ^?{}(thread_desc& this) with( this ) {
+void ^?{}($thread& this) with( this ) {
         unregister(curr_cluster, this);
         ^self_cor{};
+}
+//-----------------------------------------------------------------------------
+// Starting and stopping threads
+forall( dtype T | is_thread(T) )
+void __thrd_start( T & this, void (*main_p)(T &) ) {
+        $thread * this_thrd = get_thread(this);
+        disable_interrupts();
+        __cfactx_start(main_p, get_coroutine(this), this, __cfactx_invoke_thread);
+        this_thrd->context.[SP, FP] = this_thrd->self_cor.context.[SP, FP];
+        verify( this_thrd->context.SP );
+        __schedule_thread(this_thrd);
+        enable_interrupts( __cfaabi_dbg_ctx );
+}
+//-----------------------------------------------------------------------------
+// Support for threads that don't ues the thread keyword
 forall( dtype T | sized(T) | is_thread(T) | { void ?{}(T&); } )
 void ?{}( scoped(T)& this ) with( this ) {
 …
+}
-//-----------------------------------------------------------------------------
-// Starting and stopping threads
-forall( dtype T | is_thread(T) )
-void __thrd_start( T & this, void (*main_p)(T &) ) {
-        thread_desc * this_thrd = get_thread(this);
-        disable_interrupts();
-        CtxStart(main_p, get_coroutine(this), this, CtxInvokeThread);
-        this_thrd->context.[SP, FP] = this_thrd->self_cor.context.[SP, FP];
-        verify( this_thrd->context.SP );
-        ScheduleThread(this_thrd);
-        enable_interrupts( __cfaabi_dbg_ctx );
+}
-void yield( void ) {
-        // Safety note : This could cause some false positives due to preemption
-      verify( TL_GET( preemption_state.enabled ) );
-        BlockInternal( TL_GET( this_thread ) );
-        // Safety note : This could cause some false positives due to preemption
-      verify( TL_GET( preemption_state.enabled ) );
+}
-void yield( unsigned times ) {
-        for( unsigned i = 0; i < times; i++ ) {
-                yield();
+        }
+}
 // Local Variables: //
 // mode: c //

libcfa/src/concurrency/thread.hfa

-              rb7d6a36
+              r6a490b2
       void ^?{}(T& mutex this);
       void main(T& this);
       thread_desc* get_thread(T& this);
+      $thread* get_thread(T& this);
 };
+#define DECL_THREAD(X) thread_desc* get_thread(X& this) { return &this.__thrd; } void main(X& this)
+// define that satisfies the trait without using the thread keyword
+#define DECL_THREAD(X) $thread* get_thread(X& this) __attribute__((const)) { return &this.__thrd; } void main(X& this)
+// Inline getters for threads/coroutines/monitors
+forall( dtype T | is_thread(T) )
+static inline $coroutine* get_coroutine(T & this) __attribute__((const)) { return &get_thread(this)->self_cor; }
 forall( dtype T | is_thread(T) )
+static inline coroutine_desc* get_coroutine(T & this) {
+        return &get_thread(this)->self_cor;
+}
+static inline $monitor  * get_monitor  (T & this) __attribute__((const)) { return &get_thread(this)->self_mon; }
+forall( dtype T | is_thread(T) )
+static inline monitor_desc* get_monitor(T & this) {
+        return &get_thread(this)->self_mon;
+}
+static inline $coroutine* get_coroutine($thread * this) __attribute__((const)) { return &this->self_cor; }
+static inline $monitor  * get_monitor  ($thread * this) __attribute__((const)) { return &this->self_mon; }
+static inline coroutine_desc* get_coroutine(thread_desc * this) {
+        return &this->self_cor;
+}
+static inline monitor_desc* get_monitor(thread_desc * this) {
+        return &this->self_mon;
+}
+//-----------------------------------------------------------------------------
+// forward declarations needed for threads
 extern struct cluster * mainCluster;
 …
 //-----------------------------------------------------------------------------
 // Ctors and dtors
 void ?{}(thread_desc & this, const char * const name, struct cluster & cl, void * storage, size_t storageSize );
 void ^?{}(thread_desc & this);
+void ?{}($thread & this, const char * const name, struct cluster & cl, void * storage, size_t storageSize );
+void ^?{}($thread & this);
 static inline void ?{}(thread_desc & this)                                                                  { this{ "Anonymous Thread", *mainCluster, 0p, 65000 }; }
 static inline void ?{}(thread_desc & this, size_t stackSize )                                               { this{ "Anonymous Thread", *mainCluster, 0p, stackSize }; }
 static inline void ?{}(thread_desc & this, void * storage, size_t storageSize )                             { this{ "Anonymous Thread", *mainCluster, storage, storageSize }; }
 static inline void ?{}(thread_desc & this, struct cluster & cl )                                            { this{ "Anonymous Thread", cl, 0p, 65000 }; }
 static inline void ?{}(thread_desc & this, struct cluster & cl, size_t stackSize )                          { this{ "Anonymous Thread", cl, 0p, stackSize }; }
 static inline void ?{}(thread_desc & this, struct cluster & cl, void * storage, size_t storageSize )        { this{ "Anonymous Thread", cl, storage, storageSize }; }
 static inline void ?{}(thread_desc & this, const char * const name)                                         { this{ name, *mainCluster, 0p, 65000 }; }
 static inline void ?{}(thread_desc & this, const char * const name, struct cluster & cl )                   { this{ name, cl, 0p, 65000 }; }
 static inline void ?{}(thread_desc & this, const char * const name, struct cluster & cl, size_t stackSize ) { this{ name, cl, 0p, stackSize }; }
+static inline void ?{}($thread & this)                                                                  { this{ "Anonymous Thread", *mainCluster, 0p, 65000 }; }
+static inline void ?{}($thread & this, size_t stackSize )                                               { this{ "Anonymous Thread", *mainCluster, 0p, stackSize }; }
+static inline void ?{}($thread & this, void * storage, size_t storageSize )                             { this{ "Anonymous Thread", *mainCluster, storage, storageSize }; }
+static inline void ?{}($thread & this, struct cluster & cl )                                            { this{ "Anonymous Thread", cl, 0p, 65000 }; }
+static inline void ?{}($thread & this, struct cluster & cl, size_t stackSize )                          { this{ "Anonymous Thread", cl, 0p, stackSize }; }
+static inline void ?{}($thread & this, struct cluster & cl, void * storage, size_t storageSize )        { this{ "Anonymous Thread", cl, storage, storageSize }; }
+static inline void ?{}($thread & this, const char * const name)                                         { this{ name, *mainCluster, 0p, 65000 }; }
+static inline void ?{}($thread & this, const char * const name, struct cluster & cl )                   { this{ name, cl, 0p, 65000 }; }
+static inline void ?{}($thread & this, const char * const name, struct cluster & cl, size_t stackSize ) { this{ name, cl, 0p, stackSize }; }
 //-----------------------------------------------------------------------------
 …
 void ^?{}( scoped(T)& this );
+void yield();
+void yield( unsigned times );
+//-----------------------------------------------------------------------------
+// Thread getters
+static inline struct $thread * active_thread () { return TL_GET( this_thread ); }
+static inline struct thread_desc * active_thread () { return TL_GET( this_thread ); }
+//-----------------------------------------------------------------------------
+// Scheduler API
+//----------
+// Park thread: block until corresponding call to unpark, won't block if unpark is already called
+void park( __cfaabi_dbg_ctx_param );
+//----------
+// Unpark a thread, if the thread is already blocked, schedule it
+//                  if the thread is not yet block, signal that it should rerun immediately
+void unpark( $thread * this __cfaabi_dbg_ctx_param2 );
+forall( dtype T | is_thread(T) )
+static inline void unpark( T & this __cfaabi_dbg_ctx_param2 ) { if(!&this) return; unpark( get_thread( this ) __cfaabi_dbg_ctx_fwd2 );}
+//----------
+// Yield: force thread to block and be rescheduled
+bool force_yield( enum __Preemption_Reason );
+static inline void yield() {
+        force_yield(__MANUAL_PREEMPTION);
+}
+// Yield: yield N times
+static inline void yield( unsigned times ) {
+        for( times ) {
+                yield();
+        }
+}
+//----------
+// sleep: force thread to block and be rescheduled after Duration duration
+void sleep( Duration duration );
 // Local Variables: //

libcfa/src/exception.c

-              rb7d6a36
+              r6a490b2
 // Author           : Andrew Beach
 // Created On       : Mon Jun 26 15:13:00 2017
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Thu Feb 22 18:17:34 2018
 // Update Count     : 11
+// Last Modified By : Andrew Beach
+// Last Modified On : Tue Apr 14 12:01:00 2020
+// Update Count     : 18
 //
+// Normally we would get this from the CFA prelude.
 #include <stddef.h> // for size_t
 #include "exception.h"
+// Implementation of the secret header.
+// Implementation of the secret header is hardware dependent.
+#if !( defined( __x86_64 ) || defined( __i386 ) )
+#error Exception Handling: No known architecture detected.
+#endif
 #include <stdlib.h>
 …
 #include <unwind.h>
 #include <bits/debug.hfa>
+#include "stdhdr/assert.h"
 // FIX ME: temporary hack to keep ARM build working
 #ifndef _URC_FATAL_PHASE1_ERROR
 #define _URC_FATAL_PHASE1_ERROR 2
+#define _URC_FATAL_PHASE1_ERROR 3
 #endif // ! _URC_FATAL_PHASE1_ERROR
 #ifndef _URC_FATAL_PHASE2_ERROR
 …
 #include "lsda.h"
+/* The exception class for our exceptions. Because of the vendor component
+ * its value would not be standard.
+ * Vendor: UWPL
+ * Language: CFA\0
+ */
+const _Unwind_Exception_Class __cfaehm_exception_class = 0x4c50575500414643;
 // Base exception vtable is abstract, you should not have base exceptions.
 struct __cfaabi_ehm__base_exception_t_vtable
                 ___cfaabi_ehm__base_exception_t_vtable_instance = {
+struct __cfaehm_base_exception_t_vtable
+                ___cfaehm_base_exception_t_vtable_instance = {
         .parent = NULL,
         .size = 0,
 …
 // Temperary global exception context. Does not work with concurency.
 struct exception_context_t {
+    struct __cfaabi_ehm__try_resume_node * top_resume;
+    struct __cfaabi_ehm__try_resume_node * current_resume;
+    exception_t * current_exception;
+    int current_handler_index;
+} shared_stack = {NULL, NULL, 0, 0};
+        struct __cfaehm_try_resume_node * top_resume;
+        exception_t * current_exception;
+        int current_handler_index;
+} static shared_stack = {NULL, NULL, 0};
 // Get the current exception context.
 …
         return &shared_stack;
+}
-//#define SAVE_EXCEPTION_CONTEXT(to_name)
-//struct exception_context_t * to_name = this_exception_context();
-//exception * this_exception() {
-//    return this_exception_context()->current_exception;
-//}
-// This macro should be the only thing that needs to change across machines.
-// Used in the personality function, way down in termination.
-// struct _Unwind_Context * -> _Unwind_Reason_Code(*)(exception_t *)
-#define MATCHER_FROM_CONTEXT(ptr_to_context) \
-        (*(_Unwind_Reason_Code(**)(exception_t *))(_Unwind_GetCFA(ptr_to_context) + 8))
 // RESUMPTION ================================================================
+void __cfaabi_ehm__throw_resume(exception_t * except) {
+        __cfaabi_dbg_print_safe("Throwing resumption exception\n");
+        struct __cfaabi_ehm__try_resume_node * original_head = shared_stack.current_resume;
+        struct __cfaabi_ehm__try_resume_node * current =
+                (original_head) ? original_head->next : shared_stack.top_resume;
+static void reset_top_resume(struct __cfaehm_try_resume_node ** store) {
+        this_exception_context()->top_resume = *store;
+}
+void __cfaehm_throw_resume(exception_t * except) {
+        struct exception_context_t * context = this_exception_context();
+        __cfadbg_print_safe(exception, "Throwing resumption exception\n");
+        __attribute__((cleanup(reset_top_resume)))
+        struct __cfaehm_try_resume_node * original_head = context->top_resume;
+        struct __cfaehm_try_resume_node * current = context->top_resume;
         for ( ; current ; current = current->next) {
                 shared_stack.current_resume = current;
+                context->top_resume = current->next;
                 if (current->handler(except)) {
-                        shared_stack.current_resume = original_head;
                         return;
+                }
+        }
+        __cfaabi_dbg_print_safe("Unhandled exception\n");
+        shared_stack.current_resume = original_head;
+        __cfadbg_print_safe(exception, "Unhandled exception\n");
         // Fall back to termination:
         __cfaabi_ehm__throw_terminate(except);
+        __cfaehm_throw_terminate(except);
         // TODO: Default handler for resumption.
+}
 …
 // be added after the node is built but before it is made the top node.
 void __cfaabi_ehm__try_resume_setup(struct __cfaabi_ehm__try_resume_node * node,
+void __cfaehm_try_resume_setup(struct __cfaehm_try_resume_node * node,
                         _Bool (*handler)(exception_t * except)) {
+        node->next = shared_stack.top_resume;
+        struct exception_context_t * context = this_exception_context();
+        node->next = context->top_resume;
         node->handler = handler;
+        shared_stack.top_resume = node;
+}
+void __cfaabi_ehm__try_resume_cleanup(struct __cfaabi_ehm__try_resume_node * node) {
+        shared_stack.top_resume = node->next;
+        context->top_resume = node;
+}
+void __cfaehm_try_resume_cleanup(struct __cfaehm_try_resume_node * node) {
+        struct exception_context_t * context = this_exception_context();
+        context->top_resume = node->next;
+}
 …
 // May have to move to cfa for constructors and destructors (references).
+struct __cfaabi_ehm__node {
+        struct __cfaabi_ehm__node * next;
+// How to clean up an exception in various situations.
+static void __cfaehm_exception_cleanup(
+                _Unwind_Reason_Code reason,
+                struct _Unwind_Exception * exception) {
+        switch (reason) {
+        case _URC_FOREIGN_EXCEPTION_CAUGHT:
+                // This one we could clean-up to allow cross-language exceptions.
+        case _URC_FATAL_PHASE1_ERROR:
+        case _URC_FATAL_PHASE2_ERROR:
+        default:
+                abort();
+        }
+}
+// We need a piece of storage to raise the exception, for now its a single
+// piece.
+static struct _Unwind_Exception this_exception_storage;
+struct __cfaehm_node {
+        struct __cfaehm_node * next;
 };
 #define NODE_TO_EXCEPT(node) ((exception_t *)(1 + (node)))
 #define EXCEPT_TO_NODE(except) ((struct __cfaabi_ehm__node *)(except) - 1)
+#define EXCEPT_TO_NODE(except) ((struct __cfaehm_node *)(except) - 1)
 // Creates a copy of the indicated exception and sets current_exception to it.
 static void __cfaabi_ehm__allocate_exception( exception_t * except ) {
+static void __cfaehm_allocate_exception( exception_t * except ) {
         struct exception_context_t * context = this_exception_context();
         // Allocate memory for the exception.
         struct __cfaabi_ehm__node * store = malloc(
                 sizeof( struct __cfaabi_ehm__node ) + except->virtual_table->size );
+        struct __cfaehm_node * store = malloc(
+                sizeof( struct __cfaehm_node ) + except->virtual_table->size );
         if ( ! store ) {
 …
         // Copy the exception to storage.
         except->virtual_table->copy( context->current_exception, except );
+        // Set up the exception storage.
+        this_exception_storage.exception_class = __cfaehm_exception_class;
+        this_exception_storage.exception_cleanup = __cfaehm_exception_cleanup;
+}
 // Delete the provided exception, unsetting current_exception if relivant.
 static void __cfaabi_ehm__delete_exception( exception_t * except ) {
+static void __cfaehm_delete_exception( exception_t * except ) {
         struct exception_context_t * context = this_exception_context();
         __cfaabi_dbg_print_safe("Deleting Exception\n");
+        __cfadbg_print_safe(exception, "Deleting Exception\n");
         // Remove the exception from the list.
         struct __cfaabi_ehm__node * to_free = EXCEPT_TO_NODE(except);
         struct __cfaabi_ehm__node * node;
+        struct __cfaehm_node * to_free = EXCEPT_TO_NODE(except);
+        struct __cfaehm_node * node;
         if ( context->current_exception == except ) {
 …
                 node = EXCEPT_TO_NODE(context->current_exception);
                 // It may always be in the first or second position.
                 while( to_free != node->next ) {
+                while ( to_free != node->next ) {
                         node = node->next;
+                }
 …
 // If this isn't a rethrow (*except==0), delete the provided exception.
+void __cfaabi_ehm__cleanup_terminate( void * except ) {
+        if ( *(void**)except ) __cfaabi_ehm__delete_exception( *(exception_t **)except );
+}
+// We need a piece of storage to raise the exception
+struct _Unwind_Exception this_exception_storage;
+void __cfaehm_cleanup_terminate( void * except ) {
+        if ( *(void**)except ) __cfaehm_delete_exception( *(exception_t **)except );
+}
 // Function needed by force unwind
 …
                 int version,
                 _Unwind_Action actions,
                 _Unwind_Exception_Class exceptionClass,
+                _Unwind_Exception_Class exception_class,
                 struct _Unwind_Exception * unwind_exception,
+                struct _Unwind_Context * context,
+                void * some_param) {
+        if( actions & _UA_END_OF_STACK  ) exit(1);
+        if( actions & _UA_CLEANUP_PHASE ) return _URC_NO_REASON;
+        return _URC_FATAL_PHASE2_ERROR;
+                struct _Unwind_Context * unwind_context,
+                void * stop_param) {
+        // Verify actions follow the rules we expect.
+        verify((actions & _UA_CLEANUP_PHASE) && (actions & _UA_FORCE_UNWIND));
+        verify(!(actions & (_UA_SEARCH_PHASE | _UA_HANDER_FRAME)));
+        if ( actions & _UA_END_OF_STACK ) {
+                exit(1);
+        } else {
+                return _URC_NO_REASON;
+        }
+}
 // The exception that is being thrown must already be stored.
 __attribute__((noreturn)) void __cfaabi_ehm__begin_unwind(void) {
+static __attribute__((noreturn)) void __cfaehm_begin_unwind(void) {
         if ( ! this_exception_context()->current_exception ) {
                 printf("UNWIND ERROR missing exception in begin unwind\n");
                 abort();
+        }
         // Call stdlibc to raise the exception
 …
         // the whole stack.
         if( ret == _URC_END_OF_STACK ) {
+        if ( ret == _URC_END_OF_STACK ) {
                 // No proper handler was found. This can be handled in many ways, C++ calls std::terminate.
                 // Here we force unwind the stack, basically raising a cancellation.
 …
+}
+void __cfaabi_ehm__throw_terminate( exception_t * val ) {
+        __cfaabi_dbg_print_safe("Throwing termination exception\n");
+        __cfaabi_ehm__allocate_exception( val );
+        __cfaabi_ehm__begin_unwind();
+}
+void __cfaabi_ehm__rethrow_terminate(void) {
+        __cfaabi_dbg_print_safe("Rethrowing termination exception\n");
+        __cfaabi_ehm__begin_unwind();
+}
+#pragma GCC push_options
+#pragma GCC optimize("O0")
+void __cfaehm_throw_terminate( exception_t * val ) {
+        __cfadbg_print_safe(exception, "Throwing termination exception\n");
+        __cfaehm_allocate_exception( val );
+        __cfaehm_begin_unwind();
+}
+void __cfaehm_rethrow_terminate(void) {
+        __cfadbg_print_safe(exception, "Rethrowing termination exception\n");
+        __cfaehm_begin_unwind();
+}
 // This is our personality routine. For every stack frame annotated with
 // ".cfi_personality 0x3,__gcfa_personality_v0" this function will be called twice when unwinding.
 //  Once in the search phase and once in the cleanup phase.
+_Unwind_Reason_Code __gcfa_personality_v0 (
+                int version, _Unwind_Action actions, unsigned long long exceptionClass,
+                struct _Unwind_Exception* unwind_exception,
+                struct _Unwind_Context* context)
+_Unwind_Reason_Code __gcfa_personality_v0(
+                int version,
+                _Unwind_Action actions,
+                unsigned long long exception_class,
+                struct _Unwind_Exception * unwind_exception,
+                struct _Unwind_Context * unwind_context)
+{
+        //__cfaabi_dbg_print_safe("CFA: 0x%lx\n", _Unwind_GetCFA(context));
+        __cfaabi_dbg_print_safe("Personality function (%d, %x, %llu, %p, %p):",
+                        version, actions, exceptionClass, unwind_exception, context);
+        // If we've reached the end of the stack then there is nothing much we can do...
+        if( actions & _UA_END_OF_STACK ) return _URC_END_OF_STACK;
+        //__cfadbg_print_safe(exception, "CFA: 0x%lx\n", _Unwind_GetCFA(context));
+        __cfadbg_print_safe(exception, "Personality function (%d, %x, %llu, %p, %p):",
+                        version, actions, exception_class, unwind_exception, unwind_context);
+        // Verify that actions follow the rules we expect.
+        // This function should never be called at the end of the stack.
+        verify(!(actions & _UA_END_OF_STACK));
+        // Either only the search phase flag is set or...
         if (actions & _UA_SEARCH_PHASE) {
+                __cfaabi_dbg_print_safe(" lookup phase");
+        }
+        else if (actions & _UA_CLEANUP_PHASE) {
+                __cfaabi_dbg_print_safe(" cleanup phase");
+        }
+        // Just in case, probably can't actually happen
+        else {
+                printf(" error\n");
+                return _URC_FATAL_PHASE1_ERROR;
+                verify(actions == _UA_SEARCH_PHASE);
+                __cfadbg_print_safe(exception, " lookup phase");
+        // ... we are in clean-up phase.
+        } else {
+                verify(actions & _UA_CLEANUP_PHASE);
+                __cfadbg_print_safe(exception, " cleanup phase");
+                // We shouldn't be the handler frame during forced unwind.
+                if (actions & _UA_HANDLER_FRAME) {
+                        verify(!(actions & _UA_FORCE_UNWIND));
+                        __cfadbg_print_safe(exception, " (handler frame)");
+                } else if (actions & _UA_FORCE_UNWIND) {
+                        __cfadbg_print_safe(exception, " (force unwind)");
+                }
+        }
         // Get a pointer to the language specific data from which we will read what we need
         const unsigned char * lsd = (const unsigned char*) _Unwind_GetLanguageSpecificData( context );
         if( !lsd ) {    //Nothing to do, keep unwinding
+        const unsigned char * lsd = _Unwind_GetLanguageSpecificData( unwind_context );
+        if ( !lsd ) {   //Nothing to do, keep unwinding
                 printf(" no LSD");
                 goto UNWIND;
 …
         // Get the instuction pointer and a reading pointer into the exception table
         lsda_header_info lsd_info;
+        const unsigned char * cur_ptr = parse_lsda_header(context, lsd, &lsd_info);
+        _Unwind_Ptr instruction_ptr = _Unwind_GetIP( context );
+        const unsigned char * cur_ptr = parse_lsda_header(unwind_context, lsd, &lsd_info);
+        _Unwind_Ptr instruction_ptr = _Unwind_GetIP(unwind_context);
+        struct exception_context_t * context = this_exception_context();
         // Linearly search the table for stuff to do
         while( cur_ptr < lsd_info.action_table ) {
+        while ( cur_ptr < lsd_info.action_table ) {
                 _Unwind_Ptr callsite_start;
                 _Unwind_Ptr callsite_len;
 …
                 // Have we reach the correct frame info yet?
                 if( lsd_info.Start + callsite_start + callsite_len < instruction_ptr ) {
+                if ( lsd_info.Start + callsite_start + callsite_len < instruction_ptr ) {
 #ifdef __CFA_DEBUG_PRINT__
                         void * ls = (void*)lsd_info.Start;
 …
                         void * ep = (void*)lsd_info.Start + callsite_start + callsite_len;
                         void * ip = (void*)instruction_ptr;
                         __cfaabi_dbg_print_safe("\nfound %p - %p (%p, %p, %p), looking for %p\n",
+                        __cfadbg_print_safe(exception, "\nfound %p - %p (%p, %p, %p), looking for %p\n",
                                         bp, ep, ls, cs, cl, ip);
 #endif // __CFA_DEBUG_PRINT__
 …
                 // Have we gone too far?
                 if( lsd_info.Start + callsite_start > instruction_ptr ) {
+                if ( lsd_info.Start + callsite_start > instruction_ptr ) {
                         printf(" gone too far");
                         break;
+                }
+                // Something to do?
+                if( callsite_landing_pad ) {
+                        // Which phase are we in
+                        if (actions & _UA_SEARCH_PHASE) {
+                                // In search phase, these means we found a potential handler we must check.
+                                // We have arbitrarily decided that 0 means nothing to do and 1 means there is
+                                // a potential handler. This doesn't seem to conflict the gcc default behavior.
+                                if (callsite_action != 0) {
+                                        // Now we want to run some code to see if the handler matches
+                                        // This is the tricky part where we want to the power to run arbitrary code
+                                        // However, generating a new exception table entry and try routine every time
+                                        // is way more expansive than we might like
+                                        // The information we have is :
+                                        //  - The GR (Series of registers)
+                                        //    GR1=GP Global Pointer of frame ref by context
+                                        //  - The instruction pointer
+                                        //  - The instruction pointer info (???)
+                                        //  - The CFA (Canonical Frame Address)
+                                        //  - The BSP (Probably the base stack pointer)
+                                        // The current apprach uses one exception table entry per try block
+                                        _uleb128_t imatcher;
+                                        // Get the relative offset to the {...}?
+                                        cur_ptr = read_uleb128(cur_ptr, &imatcher);
+                                        _Unwind_Reason_Code (*matcher)(exception_t *) =
+                                                MATCHER_FROM_CONTEXT(context);
+                                        int index = matcher(shared_stack.current_exception);
+                                        _Unwind_Reason_Code ret = (0 == index)
+                                                ? _URC_CONTINUE_UNWIND : _URC_HANDLER_FOUND;
+                                        shared_stack.current_handler_index = index;
+                                        // Based on the return value, check if we matched the exception
+                                        if( ret == _URC_HANDLER_FOUND) {
+                                                __cfaabi_dbg_print_safe(" handler found\n");
+                                        } else {
+                                                __cfaabi_dbg_print_safe(" no handler\n");
+                                        }
+                                        return ret;
+                // Check for what we must do:
+                if ( 0 == callsite_landing_pad ) {
+                        // Nothing to do, move along
+                        __cfadbg_print_safe(exception, " no landing pad");
+                } else if (actions & _UA_SEARCH_PHASE) {
+                        // In search phase, these means we found a potential handler we must check.
+                        // We have arbitrarily decided that 0 means nothing to do and 1 means there is
+                        // a potential handler. This doesn't seem to conflict the gcc default behavior.
+                        if (callsite_action != 0) {
+                                // Now we want to run some code to see if the handler matches
+                                // This is the tricky part where we want to the power to run arbitrary code
+                                // However, generating a new exception table entry and try routine every time
+                                // is way more expansive than we might like
+                                // The information we have is :
+                                //  - The GR (Series of registers)
+                                //    GR1=GP Global Pointer of frame ref by context
+                                //  - The instruction pointer
+                                //  - The instruction pointer info (???)
+                                //  - The CFA (Canonical Frame Address)
+                                //  - The BSP (Probably the base stack pointer)
+                                // The current apprach uses one exception table entry per try block
+                                _uleb128_t imatcher;
+                                // Get the relative offset to the {...}?
+                                cur_ptr = read_uleb128(cur_ptr, &imatcher);
+#                               if defined( __x86_64 )
+                                _Unwind_Word match_pos = _Unwind_GetCFA(unwind_context) + 8;
+#                               elif defined( __i386 )
+                                _Unwind_Word match_pos = _Unwind_GetCFA(unwind_context) + 24;
+#                               endif
+                                int (*matcher)(exception_t *) = *(int(**)(exception_t *))match_pos;
+                                int index = matcher(context->current_exception);
+                                _Unwind_Reason_Code ret = (0 == index)
+                                        ? _URC_CONTINUE_UNWIND : _URC_HANDLER_FOUND;
+                                context->current_handler_index = index;
+                                // Based on the return value, check if we matched the exception
+                                if (ret == _URC_HANDLER_FOUND) {
+                                        __cfadbg_print_safe(exception, " handler found\n");
+                                } else {
+                                        __cfadbg_print_safe(exception, " no handler\n");
+                                }
+                                // This is only a cleanup handler, ignore it
+                                __cfaabi_dbg_print_safe(" no action");
+                                return ret;
+                        }
+                        else if (actions & _UA_CLEANUP_PHASE) {
+                                if( (callsite_action != 0) && !(actions & _UA_HANDLER_FRAME) ){
+                                        // If this is a potential exception handler
+                                        // but not the one that matched the exception in the seach phase,
+                                        // just ignore it
+                                        goto UNWIND;
+                                }
+                                // We need to run some clean-up or a handler
+                                // These statment do the right thing but I don't know any specifics at all
+                                _Unwind_SetGR( context, __builtin_eh_return_data_regno(0), (_Unwind_Ptr) unwind_exception );
+                                _Unwind_SetGR( context, __builtin_eh_return_data_regno(1), 0 );
+                                // I assume this sets the instruction pointer to the adress of the landing pad
+                                // It doesn't actually set it, it only state the value that needs to be set once we return _URC_INSTALL_CONTEXT
+                                _Unwind_SetIP( context, ((lsd_info.LPStart) + (callsite_landing_pad)) );
+                                __cfaabi_dbg_print_safe(" action\n");
+                                // Return have some action to run
+                                return _URC_INSTALL_CONTEXT;
+                        // This is only a cleanup handler, ignore it
+                        __cfadbg_print_safe(exception, " no action");
+                } else {
+                        // In clean-up phase, no destructors here but this could be the handler.
+                        if ( (callsite_action != 0) && !(actions & _UA_HANDLER_FRAME) ){
+                                // If this is a potential exception handler
+                                // but not the one that matched the exception in the seach phase,
+                                // just ignore it
+                                goto UNWIND;
+                        }
+                        // We need to run some clean-up or a handler
+                        // These statment do the right thing but I don't know any specifics at all
+                        _Unwind_SetGR( unwind_context, __builtin_eh_return_data_regno(0),
+                                (_Unwind_Ptr)unwind_exception );
+                        _Unwind_SetGR( unwind_context, __builtin_eh_return_data_regno(1), 0 );
+                        // I assume this sets the instruction pointer to the adress of the landing pad
+                        // It doesn't actually set it, it only state the value that needs to be set once we
+                        // return _URC_INSTALL_CONTEXT
+                        _Unwind_SetIP( unwind_context, ((lsd_info.LPStart) + (callsite_landing_pad)) );
+                        __cfadbg_print_safe(exception, " action\n");
+                        // Return have some action to run
+                        return _URC_INSTALL_CONTEXT;
+                }
-                // Nothing to do, move along
-                __cfaabi_dbg_print_safe(" no landing pad");
+        }
         // No handling found
         __cfaabi_dbg_print_safe(" table end reached\n");
+        __cfadbg_print_safe(exception, " table end reached");
         UNWIND:
         __cfaabi_dbg_print_safe(" unwind\n");
+        __cfadbg_print_safe(exception, " unwind\n");
         // Keep unwinding the stack
         return _URC_CONTINUE_UNWIND;
+}
+#pragma GCC push_options
+#pragma GCC optimize(0)
 // Try statements are hoisted out see comments for details. While this could probably be unique
 // and simply linked from libcfa but there is one problem left, see the exception table for details
 __attribute__((noinline))
 void __cfaabi_ehm__try_terminate(void (*try_block)(),
+void __cfaehm_try_terminate(void (*try_block)(),
                 void (*catch_block)(int index, exception_t * except),
                 __attribute__((unused)) int (*match_block)(exception_t * except)) {
 …
         //! printf("%p %p %p %p\n", &try_block, &catch_block, &match_block, &xy);
-        // Setup statments: These 2 statments won't actually result in any code, they only setup global tables.
-        // However, they clobber gcc cancellation support from gcc.  We can replace the personality routine but
-        // replacing the exception table gcc generates is not really doable, it generates labels based on how the
-        // assembly works.
         // Setup the personality routine and exception table.
+        // Unforturnately these clobber gcc cancellation support which means we can't get access to
+        // the attribute cleanup tables at the same time. We would have to inspect the assembly to
+        // create a new set ourselves.
 #ifdef __PIC__
         asm volatile (".cfi_personality 0x9b,CFA.ref.__gcfa_personality_v0");
 …
         // Label which defines the end of the area for which the handler is setup.
         asm volatile (".TRYEND:");
         // Label which defines the start of the exception landing pad.  Basically what is called when the exception is
         // caught.  Note, if multiple handlers are given, the multiplexing should be done by the generated code, not the
         // exception runtime.
+        // Label which defines the start of the exception landing pad. Basically what is called when
+        // the exception is caught. Note, if multiple handlers are given, the multiplexing should be
+        // done by the generated code, not the exception runtime.
         asm volatile (".CATCH:");
         // Exception handler
+        catch_block( shared_stack.current_handler_index,
+                     shared_stack.current_exception );
+        // Note: Saving the exception context on the stack breaks termination exceptions.
+        catch_block( this_exception_context()->current_handler_index,
+                     this_exception_context()->current_exception );
+}
 …
 #ifdef __PIC__
-#if defined( __i386 ) || defined( __x86_64 )
 asm (
         // HEADER
 …
         // handler landing pad offset and 1 (action code, gcc seems to use 0).
         ".LLSDACSBCFA2:\n"
         "       .uleb128 .TRYSTART-__cfaabi_ehm__try_terminate\n"
+        "       .uleb128 .TRYSTART-__cfaehm_try_terminate\n"
         "       .uleb128 .TRYEND-.TRYSTART\n"
         "       .uleb128 .CATCH-__cfaabi_ehm__try_terminate\n"
+        "       .uleb128 .CATCH-__cfaehm_try_terminate\n"
         "       .uleb128 1\n"
         ".LLSDACSECFA2:\n"
         // TABLE FOOTER
         "       .text\n"
         "       .size   __cfaabi_ehm__try_terminate, .-__cfaabi_ehm__try_terminate\n"
+        "       .size   __cfaehm_try_terminate, .-__cfaehm_try_terminate\n"
 );
 …
         "       .quad __gcfa_personality_v0\n"
 #else // then __i386
         "   .long __gcfa_personality_v0\n"
+        "       .long __gcfa_personality_v0\n"
 #endif
 );
-#else
-#error Exception Handling: unknown architecture for position independent code.
-#endif // __i386 || __x86_64
 #else // __PIC__
-#if defined( __i386 ) || defined( __x86_64 )
 asm (
         // HEADER
 …
         ".LLSDACSBCFA2:\n"
         //      Handled area start (relative to start of function)
         "       .uleb128 .TRYSTART-__cfaabi_ehm__try_terminate\n"
+        "       .uleb128 .TRYSTART-__cfaehm_try_terminate\n"
         //      Handled area length
         "       .uleb128 .TRYEND-.TRYSTART\n"
         //      Handler landing pad address (relative to start of function)
         "       .uleb128 .CATCH-__cfaabi_ehm__try_terminate\n"
+        "       .uleb128 .CATCH-__cfaehm_try_terminate\n"
         //      Action code, gcc seems to always use 0.
         "       .uleb128 1\n"
 …
         ".LLSDACSECFA2:\n"
         "       .text\n"
         "       .size   __cfaabi_ehm__try_terminate, .-__cfaabi_ehm__try_terminate\n"
+        "       .size   __cfaehm_try_terminate, .-__cfaehm_try_terminate\n"
         "       .ident  \"GCC: (Ubuntu 6.2.0-3ubuntu11~16.04) 6.2.0 20160901\"\n"
         "       .section        .note.GNU-stack,\"x\",@progbits\n"
 );
-#else
-#error Exception Handling: unknown architecture for position dependent code.
-#endif // __i386 || __x86_64
 #endif // __PIC__

libcfa/src/exception.h

-              rb7d6a36
+              r6a490b2
 // Author           : Andrew Beach
 // Created On       : Mon Jun 26 15:11:00 2017
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Thu Feb 22 18:11:15 2018
 // Update Count     : 8
+// Last Modified By : Andrew Beach
+// Last Modified On : Fri Mar 27 10:16:00 2020
+// Update Count     : 9
 //
 …
 #endif
 struct __cfaabi_ehm__base_exception_t;
 typedef struct __cfaabi_ehm__base_exception_t exception_t;
 struct __cfaabi_ehm__base_exception_t_vtable {
         const struct __cfaabi_ehm__base_exception_t_vtable * parent;
+struct __cfaehm_base_exception_t;
+typedef struct __cfaehm_base_exception_t exception_t;
+struct __cfaehm_base_exception_t_vtable {
+        const struct __cfaehm_base_exception_t_vtable * parent;
         size_t size;
         void (*copy)(struct __cfaabi_ehm__base_exception_t *this,
                      struct __cfaabi_ehm__base_exception_t * other);
         void (*free)(struct __cfaabi_ehm__base_exception_t *this);
         const char * (*msg)(struct __cfaabi_ehm__base_exception_t *this);
+        void (*copy)(struct __cfaehm_base_exception_t *this,
+                     struct __cfaehm_base_exception_t * other);
+        void (*free)(struct __cfaehm_base_exception_t *this);
+        const char * (*msg)(struct __cfaehm_base_exception_t *this);
 };
 struct __cfaabi_ehm__base_exception_t {
         struct __cfaabi_ehm__base_exception_t_vtable const * virtual_table;
+struct __cfaehm_base_exception_t {
+        struct __cfaehm_base_exception_t_vtable const * virtual_table;
 };
 extern struct __cfaabi_ehm__base_exception_t_vtable
         ___cfaabi_ehm__base_exception_t_vtable_instance;
+extern struct __cfaehm_base_exception_t_vtable
+        ___cfaehm_base_exception_t_vtable_instance;
 // Used in throw statement translation.
 void __cfaabi_ehm__throw_terminate(exception_t * except) __attribute__((noreturn));
 void __cfaabi_ehm__rethrow_terminate() __attribute__((noreturn));
 void __cfaabi_ehm__throw_resume(exception_t * except);
+void __cfaehm_throw_terminate(exception_t * except) __attribute__((noreturn));
+void __cfaehm_rethrow_terminate() __attribute__((noreturn));
+void __cfaehm_throw_resume(exception_t * except);
 // Function catches termination exceptions.
 void __cfaabi_ehm__try_terminate(
+void __cfaehm_try_terminate(
     void (*try_block)(),
     void (*catch_block)(int index, exception_t * except),
 …
 // Clean-up the exception in catch blocks.
 void __cfaabi_ehm__cleanup_terminate(void * except);
+void __cfaehm_cleanup_terminate(void * except);
 // Data structure creates a list of resume handlers.
 struct __cfaabi_ehm__try_resume_node {
     struct __cfaabi_ehm__try_resume_node * next;
+struct __cfaehm_try_resume_node {
+    struct __cfaehm_try_resume_node * next;
     _Bool (*handler)(exception_t * except);
 };
 // These act as constructor and destructor for the resume node.
 void __cfaabi_ehm__try_resume_setup(
     struct __cfaabi_ehm__try_resume_node * node,
+void __cfaehm_try_resume_setup(
+    struct __cfaehm_try_resume_node * node,
     _Bool (*handler)(exception_t * except));
 void __cfaabi_ehm__try_resume_cleanup(
     struct __cfaabi_ehm__try_resume_node * node);
+void __cfaehm_try_resume_cleanup(
+    struct __cfaehm_try_resume_node * node);
 // Check for a standard way to call fake deconstructors.
 struct __cfaabi_ehm__cleanup_hook {};
+struct __cfaehm_cleanup_hook {};
 #ifdef __cforall

libcfa/src/heap.cfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Tue Dec 19 21:58:35 2017
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Tue Feb  4 10:04:51 2020
 // Update Count     : 648
+// Last Modified On : Wed May  6 17:29:26 2020
+// Update Count     : 727
 //
 …
 #include <errno.h>                                                                              // errno
 #include <string.h>                                                                             // memset, memcpy
+#include <limits.h>                                                                             // ULONG_MAX
 extern "C" {
 #include <sys/mman.h>                                                                   // mmap, munmap
 } // extern "C"
-// #comment TD : Many of these should be merged into math I believe
 #include "bits/align.hfa"                                                               // libPow2
 #include "bits/defs.hfa"                                                                // likely, unlikely
 …
 //#include "stdlib.hfa"                                                                 // bsearchl
 #include "malloc.h"
+#include "bitmanip.hfa"                                                                 // ceiling
 #define MIN(x, y) (y > x ? x : y)
 …
 };
+size_t default_heap_expansion() __attribute__(( weak )) {
+        return __CFA_DEFAULT_HEAP_EXPANSION__;
+} // default_heap_expansion
 size_t default_mmap_start() __attribute__(( weak )) {
         return __CFA_DEFAULT_MMAP_START__;
 } // default_mmap_start
-size_t default_heap_expansion() __attribute__(( weak )) {
-        return __CFA_DEFAULT_HEAP_EXPANSION__;
-} // default_heap_expansion
 …
                                                         union {
 //                                                              FreeHeader * home;              // allocated block points back to home locations (must overlay alignment)
+                                                                // 2nd low-order bit => zero filled
                                                                 void * home;                    // allocated block points back to home locations (must overlay alignment)
                                                                 size_t blockSize;               // size for munmap (must overlay alignment)
 …
                                 struct FakeHeader {
                                         #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+                                        uint32_t alignment;                                     // low-order bits of home/blockSize used for tricks
+                                        // 1st low-order bit => fake header & alignment
+                                        uint32_t alignment;
                                         #endif // __ORDER_LITTLE_ENDIAN__
 …
                                 } fake; // FakeHeader
                         } kind; // Kind
+                        size_t size;                                                            // allocation size in bytes
                 } header; // Header
                 char pad[libAlign() - sizeof( Header )];
 …
 static unsigned long long int free_storage;
 static unsigned int free_calls;
+static unsigned long long int aalloc_storage;
+static unsigned int aalloc_calls;
 static unsigned long long int calloc_storage;
 static unsigned int calloc_calls;
 static unsigned long long int memalign_storage;
 static unsigned int memalign_calls;
+static unsigned long long int amemalign_storage;
+static unsigned int amemalign_calls;
 static unsigned long long int cmemalign_storage;
 static unsigned int cmemalign_calls;
+static unsigned long long int resize_storage;
+static unsigned int resize_calls;
 static unsigned long long int realloc_storage;
 static unsigned int realloc_calls;
 …
 // Use "write" because streams may be shutdown when calls are made.
 static void printStats() {
         char helpText[512];
+        char helpText[1024];
         __cfaabi_bits_print_buffer( STDERR_FILENO, helpText, sizeof(helpText),
                                                                         "\nHeap statistics:\n"
                                                                         "  malloc: calls %u / storage %llu\n"
+                                                                        "  aalloc: calls %u / storage %llu\n"
                                                                         "  calloc: calls %u / storage %llu\n"
                                                                         "  memalign: calls %u / storage %llu\n"
+                                                                        "  amemalign: calls %u / storage %llu\n"
                                                                         "  cmemalign: calls %u / storage %llu\n"
+                                                                        "  resize: calls %u / storage %llu\n"
                                                                         "  realloc: calls %u / storage %llu\n"
                                                                         "  free: calls %u / storage %llu\n"
 …
                                                                         "  sbrk: calls %u / storage %llu\n",
                                                                         malloc_calls, malloc_storage,
+                                                                        aalloc_calls, calloc_storage,
                                                                         calloc_calls, calloc_storage,
                                                                         memalign_calls, memalign_storage,
+                                                                        amemalign_calls, amemalign_storage,
                                                                         cmemalign_calls, cmemalign_storage,
+                                                                        resize_calls, resize_storage,
                                                                         realloc_calls, realloc_storage,
                                                                         free_calls, free_storage,
 …
 static int printStatsXML( FILE * stream ) {                             // see malloc_info
         char helpText[512];
+        char helpText[1024];
         int len = snprintf( helpText, sizeof(helpText),
                                                 "<malloc version=\"1\">\n"
 …
                                                 "</sizes>\n"
                                                 "<total type=\"malloc\" count=\"%u\" size=\"%llu\"/>\n"
+                                                "<total type=\"aalloc\" count=\"%u\" size=\"%llu\"/>\n"
                                                 "<total type=\"calloc\" count=\"%u\" size=\"%llu\"/>\n"
                                                 "<total type=\"memalign\" count=\"%u\" size=\"%llu\"/>\n"
+                                                "<total type=\"amemalign\" count=\"%u\" size=\"%llu\"/>\n"
                                                 "<total type=\"cmemalign\" count=\"%u\" size=\"%llu\"/>\n"
+                                                "<total type=\"resize\" count=\"%u\" size=\"%llu\"/>\n"
                                                 "<total type=\"realloc\" count=\"%u\" size=\"%llu\"/>\n"
                                                 "<total type=\"free\" count=\"%u\" size=\"%llu\"/>\n"
 …
                                                 "</malloc>",
                                                 malloc_calls, malloc_storage,
+                                                aalloc_calls, aalloc_storage,
                                                 calloc_calls, calloc_storage,
                                                 memalign_calls, memalign_storage,
+                                                amemalign_calls, amemalign_storage,
                                                 cmemalign_calls, cmemalign_storage,
+                                                resize_calls, resize_storage,
                                                 realloc_calls, realloc_storage,
                                                 free_calls, free_storage,
 …
-static inline void checkAlign( size_t alignment ) {
-        if ( alignment < libAlign() || ! libPow2( alignment ) ) {
-                abort( "Alignment %zu for memory allocation is less than %d and/or not a power of 2.", alignment, libAlign() );
-        } // if
-} // checkAlign
-static inline bool setHeapExpand( size_t value ) {
-  if ( heapExpand < pageSize ) return true;
-        heapExpand = value;
-        return false;
-} // setHeapExpand
 // thunk problem
 size_t Bsearchl( unsigned int key, const unsigned int * vals, size_t dim ) {
 …
 static inline bool setMmapStart( size_t value ) {               // true => mmapped, false => sbrk
   if ( value < pageSize || bucketSizes[NoBucketSizes - 1] < value ) return true;
+  if ( value < pageSize || bucketSizes[NoBucketSizes - 1] < value ) return false;
         mmapStart = value;                                                                      // set global
 …
         assert( maxBucketsUsed < NoBucketSizes );                       // subscript failure ?
         assert( mmapStart <= bucketSizes[maxBucketsUsed] ); // search failure ?
         return false;
+        return true;
 } // setMmapStart
+// <-------+----------------------------------------------------> bsize (bucket size)
+// |header |addr
+//==================================================================================
+//                   align/offset |
+// <-----------------<------------+-----------------------------> bsize (bucket size)
+//                   |fake-header | addr
+#define headerAddr( addr ) ((HeapManager.Storage.Header *)( (char *)addr - sizeof(HeapManager.Storage) ))
+#define realHeader( header ) ((HeapManager.Storage.Header *)((char *)header - header->kind.fake.offset))
+// <-------<<--------------------- dsize ---------------------->> bsize (bucket size)
+// |header |addr
+//==================================================================================
+//                   align/offset |
+// <------------------------------<<---------- dsize --------->>> bsize (bucket size)
+//                   |fake-header |addr
+#define dataStorage( bsize, addr, header ) (bsize - ( (char *)addr - (char *)header ))
+static inline void checkAlign( size_t alignment ) {
+        if ( alignment < libAlign() || ! libPow2( alignment ) ) {
+                abort( "Alignment %zu for memory allocation is less than %d and/or not a power of 2.", alignment, libAlign() );
+        } // if
+} // checkAlign
 …
 static inline void fakeHeader( HeapManager.Storage.Header *& header, size_t & alignment ) {
         if ( unlikely( (header->kind.fake.alignment & 1) == 1 ) ) { // fake header ?
-                size_t offset = header->kind.fake.offset;
                 alignment = header->kind.fake.alignment & -2;   // remove flag from value
                 #ifdef __CFA_DEBUG__
                 checkAlign( alignment );                                                // check alignment
                 #endif // __CFA_DEBUG__
                 header = (HeapManager.Storage.Header *)((char *)header - offset);
+                header = realHeader( header );                                  // backup from fake to real header
         } // if
 } // fakeHeader
-// <-------+----------------------------------------------------> bsize (bucket size)
-// |header |addr
-//==================================================================================
-//                                | alignment
-// <-----------------<------------+-----------------------------> bsize (bucket size)
-//                   |fake-header | addr
-#define headerAddr( addr ) ((HeapManager.Storage.Header *)( (char *)addr - sizeof(HeapManager.Storage) ))
-// <-------<<--------------------- dsize ---------------------->> bsize (bucket size)
-// |header |addr
-//==================================================================================
-//                                | alignment
-// <------------------------------<<---------- dsize --------->>> bsize (bucket size)
-//                   |fake-header |addr
-#define dataStorage( bsize, addr, header ) (bsize - ( (char *)addr - (char *)header ))
 …
         #ifdef __CFA_DEBUG__
         checkHeader( addr < heapBegin || header < (HeapManager.Storage.Header *)heapBegin, name, addr ); // bad low address ?
+        checkHeader( addr < heapBegin, name, addr );            // bad low address ?
         #endif // __CFA_DEBUG__
 …
         // along with the block and is a multiple of the alignment size.
   if ( unlikely( size > ~0ul - sizeof(HeapManager.Storage) ) ) return 0p;
+  if ( unlikely( size > ULONG_MAX - sizeof(HeapManager.Storage) ) ) return 0p;
         size_t tsize = size + sizeof(HeapManager.Storage);
         if ( likely( tsize < mmapStart ) ) {                            // small size => sbrk
 …
                 block->header.kind.real.home = freeElem;                // pointer back to free list of apropriate size
         } else {                                                                                        // large size => mmap
   if ( unlikely( size > ~0ul - pageSize ) ) return 0p;
+  if ( unlikely( size > ULONG_MAX - pageSize ) ) return 0p;
                 tsize = libCeiling( tsize, pageSize );                  // must be multiple of page size
                 #ifdef __STATISTICS__
 …
         } // if
+        block->header.size = size;                                                      // store allocation size
         void * addr = &(block->data);                                           // adjust off header to user bytes
 …
         #endif // FASTLOOKUP
         if ( setMmapStart( default_mmap_start() ) ) {
+        if ( ! setMmapStart( default_mmap_start() ) ) {
                 abort( "HeapManager : internal error, mmap start initialization failure." );
         } // if
 …
         char * end = (char *)sbrk( 0 );
+        sbrk( (char *)libCeiling( (long unsigned int)end, libAlign() ) - end ); // move start of heap to multiple of alignment
+        heapBegin = heapEnd = sbrk( 0 );                                        // get new start point
+        heapBegin = heapEnd = sbrk( (char *)libCeiling( (long unsigned int)end, libAlign() ) - end ); // move start of heap to multiple of alignment
 } // HeapManager
 …
         //assert( heapManager.heapBegin != 0 );
         //heapManager{};
         if ( heapManager.heapBegin == 0p ) heapManager{};
+        if ( heapManager.heapBegin == 0p ) heapManager{};       // sanity check
 } // memory_startup
 …
         //assert( heapManager.heapBegin != 0 );
         if ( unlikely( heapManager.heapBegin == 0p ) ) heapManager{}; // called before memory_startup ?
+#if __SIZEOF_POINTER__ == 8
+        verify( size < ((typeof(size_t))1 << 48) );
+#endif // __SIZEOF_POINTER__ == 8
         void * addr = doMalloc( size );
         if ( unlikely( addr == 0p ) ) errno = ENOMEM;           // POSIX
 …
 static inline void * callocNoStats( size_t noOfElems, size_t elemSize ) {
         size_t size = noOfElems * elemSize;
+static inline void * callocNoStats( size_t dim, size_t elemSize ) {
+        size_t size = dim * elemSize;
         char * addr = (char *)mallocNoStats( size );
   if ( unlikely( addr == 0p ) ) return 0p;
 …
 static inline void * cmemalignNoStats( size_t alignment, size_t noOfElems, size_t elemSize ) {
         size_t size = noOfElems * elemSize;
+static inline void * cmemalignNoStats( size_t alignment, size_t dim, size_t elemSize ) {
+        size_t size = dim * elemSize;
         char * addr = (char *)memalignNoStats( alignment, size );
   if ( unlikely( addr == 0p ) ) return 0p;
 …
         #endif // __CFA_DEBUG__
                 memset( addr, '\0', dataStorage( bsize, addr, header ) ); // set to zeros
+        header->kind.real.blockSize |= 2;                               // mark as zero filled
+        header->kind.real.blockSize |= 2;                                       // mark as zero filled
         return addr;
 } // cmemalignNoStats
 …
 extern "C" {
+        // The malloc() function allocates size bytes and returns a pointer to the allocated memory. The memory is not
+        // initialized. If size is 0, then malloc() returns either 0p, or a unique pointer value that can later be
+        // successfully passed to free().
+        // Allocates size bytes and returns a pointer to the allocated memory.  The contents are undefined. If size is 0,
+        // then malloc() returns a unique pointer value that can later be successfully passed to free().
         void * malloc( size_t size ) {
                 #ifdef __STATISTICS__
 …
         } // malloc
+        // The calloc() function allocates memory for an array of nmemb elements of size bytes each and returns a pointer to
+        // the allocated memory. The memory is set to zero. If nmemb or size is 0, then calloc() returns either 0p, or a
+        // unique pointer value that can later be successfully passed to free().
+        void * calloc( size_t noOfElems, size_t elemSize ) {
+        // Same as malloc() except size bytes is an array of dim elements each of elemSize bytes.
+        void * aalloc( size_t dim, size_t elemSize ) {
+                #ifdef __STATISTICS__
+                __atomic_add_fetch( &aalloc_calls, 1, __ATOMIC_SEQ_CST );
+                __atomic_add_fetch( &aalloc_storage, dim * elemSize, __ATOMIC_SEQ_CST );
+                #endif // __STATISTICS__
+                return mallocNoStats( dim * elemSize );
+        } // aalloc
+        // Same as aalloc() with memory set to zero.
+        void * calloc( size_t dim, size_t elemSize ) {
                 #ifdef __STATISTICS__
                 __atomic_add_fetch( &calloc_calls, 1, __ATOMIC_SEQ_CST );
                 __atomic_add_fetch( &calloc_storage, noOfElems * elemSize, __ATOMIC_SEQ_CST );
                 #endif // __STATISTICS__
                 return callocNoStats( noOfElems, elemSize );
+                __atomic_add_fetch( &calloc_storage, dim * elemSize, __ATOMIC_SEQ_CST );
+                #endif // __STATISTICS__
+                return callocNoStats( dim, elemSize );
         } // calloc
+        // The realloc() function changes the size of the memory block pointed to by ptr to size bytes. The contents will be
+        // unchanged in the range from the start of the region up to the minimum of the old and new sizes. If the new size
+        // is larger than the old size, the added memory will not be initialized.  If ptr is 0p, then the call is
+        // equivalent to malloc(size), for all values of size; if size is equal to zero, and ptr is not 0p, then the call
+        // is equivalent to free(ptr). Unless ptr is 0p, it must have been returned by an earlier call to malloc(),
+        // calloc() or realloc(). If the area pointed to was moved, a free(ptr) is done.
+        // Change the size of the memory block pointed to by oaddr to size bytes. The contents are undefined.  If oaddr is
+        // 0p, then the call is equivalent to malloc(size), for all values of size; if size is equal to zero, and oaddr is
+        // not 0p, then the call is equivalent to free(oaddr). Unless oaddr is 0p, it must have been returned by an earlier
+        // call to malloc(), alloc(), calloc() or realloc(). If the area pointed to was moved, a free(oaddr) is done.
+        void * resize( void * oaddr, size_t size ) {
+                #ifdef __STATISTICS__
+                __atomic_add_fetch( &resize_calls, 1, __ATOMIC_SEQ_CST );
+                __atomic_add_fetch( &resize_storage, size, __ATOMIC_SEQ_CST );
+                #endif // __STATISTICS__
+                // If size is equal to 0, either NULL or a pointer suitable to be passed to free() is returned.
+          if ( unlikely( size == 0 ) ) { free( oaddr ); return mallocNoStats( size ); } // special cases
+          if ( unlikely( oaddr == 0p ) ) return mallocNoStats( size );
+                HeapManager.Storage.Header * header;
+                HeapManager.FreeHeader * freeElem;
+                size_t bsize, oalign = 0;
+                headers( "resize", oaddr, header, freeElem, bsize, oalign );
+                size_t odsize = dataStorage( bsize, oaddr, header ); // data storage available in bucket
+                // same size, DO NOT preserve STICKY PROPERTIES.
+          if ( oalign == 0 && size <= odsize && odsize <= size * 2 ) { // allow 50% wasted storage for smaller size
+                        header->kind.real.blockSize &= -2;                      // no alignment and turn off 0 fill
+                        return oaddr;
+                } // if
+                // change size, DO NOT preserve STICKY PROPERTIES.
+                free( oaddr );
+                void * naddr = mallocNoStats( size );                   // create new area
+                return naddr;
+        } // resize
+        // Same as resize() but the contents are unchanged in the range from the start of the region up to the minimum of
+        // the old and new sizes.
         void * realloc( void * oaddr, size_t size ) {
                 #ifdef __STATISTICS__
                 __atomic_add_fetch( &realloc_calls, 1, __ATOMIC_SEQ_CST );
+                __atomic_add_fetch( &realloc_storage, size, __ATOMIC_SEQ_CST );
                 #endif // __STATISTICS__
 …
                         // Do not know size of original allocation => cannot do 0 fill for any additional space because do not know
                         // where to start filling, i.e., do not overwrite existing values in space.
-                        //
-                        // This case does not result in a new profiler entry because the previous one still exists and it must match with
-                        // the free for this memory.  Hence, this realloc does not appear in the profiler output.
                         return oaddr;
                 } // if
-                #ifdef __STATISTICS__
-                __atomic_add_fetch( &realloc_storage, size, __ATOMIC_SEQ_CST );
-                #endif // __STATISTICS__
                 // change size and copy old content to new storage
 …
         } // realloc
+        // The obsolete function memalign() allocates size bytes and returns a pointer to the allocated memory. The memory
+        // address will be a multiple of alignment, which must be a power of two.
+        // Same as malloc() except the memory address is a multiple of alignment, which must be a power of two. (obsolete)
         void * memalign( size_t alignment, size_t size ) {
                 #ifdef __STATISTICS__
 …
         // The cmemalign() function is the same as calloc() with memory alignment.
         void * cmemalign( size_t alignment, size_t noOfElems, size_t elemSize ) {
+        // Same as aalloc() with memory alignment.
+        void * amemalign( size_t alignment, size_t dim, size_t elemSize ) {
                 #ifdef __STATISTICS__
                 __atomic_add_fetch( &cmemalign_calls, 1, __ATOMIC_SEQ_CST );
+                __atomic_add_fetch( &cmemalign_storage, noOfElems * elemSize, __ATOMIC_SEQ_CST );
+                #endif // __STATISTICS__
+                return cmemalignNoStats( alignment, noOfElems, elemSize );
+                __atomic_add_fetch( &cmemalign_storage, dim * elemSize, __ATOMIC_SEQ_CST );
+                #endif // __STATISTICS__
+                return memalignNoStats( alignment, dim * elemSize );
+        } // amemalign
+        // Same as calloc() with memory alignment.
+        void * cmemalign( size_t alignment, size_t dim, size_t elemSize ) {
+                #ifdef __STATISTICS__
+                __atomic_add_fetch( &cmemalign_calls, 1, __ATOMIC_SEQ_CST );
+                __atomic_add_fetch( &cmemalign_storage, dim * elemSize, __ATOMIC_SEQ_CST );
+                #endif // __STATISTICS__
+                return cmemalignNoStats( alignment, dim, elemSize );
         } // cmemalign
         // The function aligned_alloc() is the same as memalign(), except for the added restriction that size should be a
         // multiple of alignment.
+        // Same as memalign(), but ISO/IEC 2011 C11 Section 7.22.2 states: the value of size shall be an integral multiple
+    // of alignment. This requirement is universally ignored.
         void * aligned_alloc( size_t alignment, size_t size ) {
                 return memalign( alignment, size );
 …
         // The function posix_memalign() allocates size bytes and places the address of the allocated memory in *memptr. The
         // address of the allocated memory will be a multiple of alignment, which must be a power of two and a multiple of
         // sizeof(void *). If size is 0, then posix_memalign() returns either 0p, or a unique pointer value that can later
         // be successfully passed to free(3).
+        // Allocates size bytes and places the address of the allocated memory in *memptr. The address of the allocated
+        // memory shall be a multiple of alignment, which must be a power of two and a multiple of sizeof(void *). If size
+        // is 0, then posix_memalign() returns either 0p, or a unique pointer value that can later be successfully passed to
+        // free(3).
         int posix_memalign( void ** memptr, size_t alignment, size_t size ) {
           if ( alignment < sizeof(void *) || ! libPow2( alignment ) ) return EINVAL; // check alignment
 …
         } // posix_memalign
         // The obsolete function valloc() allocates size bytes and returns a pointer to the allocated memory. The memory
         // address will be a multiple of the page size.  It is equivalent to memalign(sysconf(_SC_PAGESIZE),size).
+        // Allocates size bytes and returns a pointer to the allocated memory. The memory address shall be a multiple of the
+        // page size.  It is equivalent to memalign(sysconf(_SC_PAGESIZE),size).
         void * valloc( size_t size ) {
                 return memalign( pageSize, size );
 …
+        // The free() function frees the memory space pointed to by ptr, which must have been returned by a previous call to
+        // malloc(), calloc() or realloc().  Otherwise, or if free(ptr) has already been called before, undefined behavior
+        // occurs. If ptr is 0p, no operation is performed.
+        // Same as valloc but rounds size to multiple of page size.
+        void * pvalloc( size_t size ) {
+                return memalign( pageSize, libCeiling( size, pageSize ) );
+        } // pvalloc
+        // Frees the memory space pointed to by ptr, which must have been returned by a previous call to malloc(), calloc()
+        // or realloc().  Otherwise, or if free(ptr) has already been called before, undefined behaviour occurs. If ptr is
+        // 0p, no operation is performed.
         void free( void * addr ) {
                 #ifdef __STATISTICS__
 …
         // The malloc_alignment() function returns the alignment of the allocation.
+        // Returns the alignment of an allocation.
         size_t malloc_alignment( void * addr ) {
           if ( unlikely( addr == 0p ) ) return libAlign();      // minimum alignment
 …
                         return header->kind.fake.alignment & -2;        // remove flag from value
                 } else {
                         return libAlign ();                                                     // minimum alignment
+                        return libAlign();                                                      // minimum alignment
                 } // if
         } // malloc_alignment
+        // The malloc_zero_fill() function returns true if the allocation is zero filled, i.e., initially allocated by calloc().
+        // Set the alignment for an the allocation and return previous alignment or 0 if no alignment.
+        size_t $malloc_alignment_set( void * addr, size_t alignment ) {
+          if ( unlikely( addr == 0p ) ) return libAlign();      // minimum alignment
+                size_t ret;
+                HeapManager.Storage.Header * header = headerAddr( addr );
+                if ( (header->kind.fake.alignment & 1) == 1 ) { // fake header ?
+                        ret = header->kind.fake.alignment & -2;         // remove flag from old value
+                        header->kind.fake.alignment = alignment | 1; // add flag to new value
+                } else {
+                        ret = 0;                                                                        // => no alignment to change
+                } // if
+                return ret;
+        } // $malloc_alignment_set
+        // Returns true if the allocation is zero filled, e.g., allocated by calloc().
         bool malloc_zero_fill( void * addr ) {
           if ( unlikely( addr == 0p ) ) return false;           // null allocation is not zero fill
                 HeapManager.Storage.Header * header = headerAddr( addr );
                 if ( (header->kind.fake.alignment & 1) == 1 ) { // fake header ?
                         header = (HeapManager.Storage.Header *)((char *)header - header->kind.fake.offset);
                 } // if
                 return (header->kind.real.blockSize & 2) != 0;  // zero filled (calloc/cmemalign) ?
+                        header = realHeader( header );                          // backup from fake to real header
+                } // if
+                return (header->kind.real.blockSize & 2) != 0;  // zero filled ?
         } // malloc_zero_fill
+        // The malloc_usable_size() function returns the number of usable bytes in the block pointed to by ptr, a pointer to
+        // a block of memory allocated by malloc(3) or a related function.
+        // Set allocation is zero filled and return previous zero filled.
+        bool $malloc_zero_fill_set( void * addr ) {
+          if ( unlikely( addr == 0p ) ) return false;           // null allocation is not zero fill
+                HeapManager.Storage.Header * header = headerAddr( addr );
+                if ( (header->kind.fake.alignment & 1) == 1 ) { // fake header ?
+                        header = realHeader( header );                          // backup from fake to real header
+                } // if
+                bool ret = (header->kind.real.blockSize & 2) != 0; // zero filled ?
+                header->kind.real.blockSize |= 2;                               // mark as zero filled
+                return ret;
+        } // $malloc_zero_fill_set
+        // Returns original total allocation size (not bucket size) => array size is dimension * sizeif(T).
+        size_t malloc_size( void * addr ) {
+          if ( unlikely( addr == 0p ) ) return false;           // null allocation is not zero fill
+                HeapManager.Storage.Header * header = headerAddr( addr );
+                if ( (header->kind.fake.alignment & 1) == 1 ) { // fake header ?
+                        header = realHeader( header );                          // backup from fake to real header
+                } // if
+                return header->size;
+        } // malloc_size
+        // Set allocation size and return previous size.
+        size_t $malloc_size_set( void * addr, size_t size ) {
+          if ( unlikely( addr == 0p ) ) return false;           // null allocation is not zero fill
+                HeapManager.Storage.Header * header = headerAddr( addr );
+                if ( (header->kind.fake.alignment & 1) == 1 ) { // fake header ?
+                        header = realHeader( header );                          // backup from fake to real header
+                } // if
+                size_t ret = header->size;
+                header->size = size;
+                return ret;
+        } // $malloc_size_set
+        // Returns the number of usable bytes in the block pointed to by ptr, a pointer to a block of memory allocated by
+        // malloc or a related function.
         size_t malloc_usable_size( void * addr ) {
           if ( unlikely( addr == 0p ) ) return 0;                       // null allocation has 0 size
 …
+        // The malloc_stats() function prints (on default standard error) statistics about memory allocated by malloc(3) and
+        // related functions.
+        // Prints (on default standard error) statistics about memory allocated by malloc and related functions.
         void malloc_stats( void ) {
                 #ifdef __STATISTICS__
 …
         } // malloc_stats
         // The malloc_stats_fd() function changes the file descripter where malloc_stats() writes the statistics.
+        // Changes the file descripter where malloc_stats() writes statistics.
         int malloc_stats_fd( int fd __attribute__(( unused )) ) {
                 #ifdef __STATISTICS__
 …
+        // The mallopt() function adjusts parameters that control the behavior of the memory-allocation functions (see
+        // malloc(3)). The param argument specifies the parameter to be modified, and value specifies the new value for that
+        // parameter.
+        // Adjusts parameters that control the behaviour of the memory-allocation functions (see malloc). The param argument
+        // specifies the parameter to be modified, and value specifies the new value for that parameter.
         int mallopt( int option, int value ) {
                 choose( option ) {
                   case M_TOP_PAD:
                         if ( setHeapExpand( value ) ) return 1;
+                        heapExpand = ceiling( value, pageSize ); return 1;
                   case M_MMAP_THRESHOLD:
                         if ( setMmapStart( value ) ) return 1;
+                        break;
                 } // switch
                 return 0;                                                                               // error, unsupported
         } // mallopt
+        // The malloc_trim() function attempts to release free memory at the top of the heap (by calling sbrk(2) with a
+        // suitable argument).
+        // Attempt to release free memory at the top of the heap (by calling sbrk with a suitable argument).
         int malloc_trim( size_t ) {
                 return 0;                                                                               // => impossible to release memory
 …
         // The malloc_info() function exports an XML string that describes the current state of the memory-allocation
         // implementation in the caller.  The string is printed on the file stream stream.  The exported string includes
         // information about all arenas (see malloc(3)).
+        // Exports an XML string that describes the current state of the memory-allocation implementation in the caller.
+        // The string is printed on the file stream stream.  The exported string includes information about all arenas (see
+        // malloc).
         int malloc_info( int options, FILE * stream ) {
                 if ( options != 0 ) { errno = EINVAL; return -1; }
 …
         // The malloc_get_state() function records the current state of all malloc(3) internal bookkeeping variables (but
         // not the actual contents of the heap or the state of malloc_hook(3) functions pointers).  The state is recorded in
         // a system-dependent opaque data structure dynamically allocated via malloc(3), and a pointer to that data
         // structure is returned as the function result.  (It is the caller's responsibility to free(3) this memory.)
+        // Records the current state of all malloc internal bookkeeping variables (but not the actual contents of the heap
+        // or the state of malloc_hook functions pointers).  The state is recorded in a system-dependent opaque data
+        // structure dynamically allocated via malloc, and a pointer to that data structure is returned as the function
+        // result.  (The caller must free this memory.)
         void * malloc_get_state( void ) {
                 return 0p;                                                                              // unsupported
 …
         // The malloc_set_state() function restores the state of all malloc(3) internal bookkeeping variables to the values
         // recorded in the opaque data structure pointed to by state.
+        // Restores the state of all malloc internal bookkeeping variables to the values recorded in the opaque data
+        // structure pointed to by state.
         int malloc_set_state( void * ptr ) {
                 return 0;                                                                               // unsupported
 …
 // Must have CFA linkage to overload with C linkage realloc.
 void * realloc( void * oaddr, size_t nalign, size_t size ) {
+void * resize( void * oaddr, size_t nalign, size_t size ) {
         #ifdef __STATISTICS__
+        __atomic_add_fetch( &realloc_calls, 1, __ATOMIC_SEQ_CST );
+        __atomic_add_fetch( &resize_calls, 1, __ATOMIC_SEQ_CST );
+        __atomic_add_fetch( &resize_storage, size, __ATOMIC_SEQ_CST );
         #endif // __STATISTICS__
         // If size is equal to 0, either NULL or a pointer suitable to be passed to free() is returned.
+  if ( unlikely( size == 0 ) ) { free( oaddr ); return mallocNoStats( size ); } // special cases
+  if ( unlikely( oaddr == 0p ) ) return mallocNoStats( size );
+  if ( unlikely( size == 0 ) ) { free( oaddr ); return memalignNoStats( nalign, size ); } // special cases
+  if ( unlikely( oaddr == 0p ) ) return memalignNoStats( nalign, size );
         if ( unlikely( nalign == 0 ) ) nalign = libAlign();     // reset alignment to minimum
 …
         HeapManager.FreeHeader * freeElem;
         size_t bsize, oalign = 0;
         headers( "realloc", oaddr, header, freeElem, bsize, oalign );
+        headers( "resize", oaddr, header, freeElem, bsize, oalign );
         size_t odsize = dataStorage( bsize, oaddr, header ); // data storage available in bucket
+  if ( oalign != 0 && (uintptr_t)oaddr % nalign == 0 ) { // has alignment and just happens to work out
+                headerAddr( oaddr )->kind.fake.alignment = nalign | 1; // update alignment (could be the same)
+                return realloc( oaddr, size );
+        } // if
+        #ifdef __STATISTICS__
+        __atomic_add_fetch( &realloc_storage, size, __ATOMIC_SEQ_CST );
+        #endif // __STATISTICS__
+        // change size and copy old content to new storage
+        if ( oalign <= nalign && (uintptr_t)oaddr % nalign == 0 ) { // <= alignment and new alignment happens to match
+                if ( oalign >= libAlign() ) {                                   // fake header ?
+                        headerAddr( oaddr )->kind.fake.alignment = nalign | 1; // update alignment (could be the same)
+                } // if
+                if ( size <= odsize && odsize <= size * 2 ) {   // allow 50% wasted storage for smaller size
+                        header->kind.real.blockSize &= -2;                      // turn off 0 fill
+                        return oaddr;
+                } // if
+        } // if
+        // change size
         void * naddr;
 …
         } // if
+        free( oaddr );
+        return naddr;
+} // resize
+void * realloc( void * oaddr, size_t nalign, size_t size ) {
+        if ( unlikely( nalign == 0 ) ) nalign = libAlign();     // reset alignment to minimum
+        #ifdef __CFA_DEBUG__
+        else
+                checkAlign( nalign );                                                   // check alignment
+        #endif // __CFA_DEBUG__
+        HeapManager.Storage.Header * header;
+        HeapManager.FreeHeader * freeElem;
+        size_t bsize, oalign = 0;
+        headers( "realloc", oaddr, header, freeElem, bsize, oalign );
+        size_t odsize = dataStorage( bsize, oaddr, header ); // data storage available in bucket
+        if ( oalign <= nalign && (uintptr_t)oaddr % nalign == 0 ) { // <= alignment and new alignment happens to match
+                if ( oalign >= libAlign() ) {                                   // fake header ?
+                        headerAddr( oaddr )->kind.fake.alignment = nalign | 1; // update alignment (could be the same)
+                } // if
+                return realloc( oaddr, size );
+        } // if
+        // change size and copy old content to new storage
+        #ifdef __STATISTICS__
+        __atomic_add_fetch( &realloc_calls, 1, __ATOMIC_SEQ_CST );
+        __atomic_add_fetch( &realloc_storage, size, __ATOMIC_SEQ_CST );
+        #endif // __STATISTICS__
+        // If size is equal to 0, either NULL or a pointer suitable to be passed to free() is returned.
+  if ( unlikely( size == 0 ) ) { free( oaddr ); return memalignNoStats( nalign, size ); } // special cases
+  if ( unlikely( oaddr == 0p ) ) return memalignNoStats( nalign, size );
+        void * naddr;
+        if ( unlikely( header->kind.real.blockSize & 2 ) ) { // previous request zero fill
+                naddr = cmemalignNoStats( nalign, 1, size );    // create new aligned area
+        } else {
+                naddr = memalignNoStats( nalign, size );                // create new aligned area
+        } // if
         headers( "realloc", naddr, header, freeElem, bsize, oalign );
         size_t ndsize = dataStorage( bsize, naddr, header ); // data storage avilable in bucket
+        size_t ndsize = dataStorage( bsize, naddr, header ); // data storage available in bucket
         // To preserve prior fill, the entire bucket must be copied versus the size.
         memcpy( naddr, oaddr, MIN( odsize, ndsize ) );          // copy bytes

libcfa/src/interpose.cfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Wed Mar 29 16:10:31 2017
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Mon Feb 17 10:18:53 2020
 // Update Count     : 166
+// Last Modified On : Fri Mar 13 17:35:37 2020
+// Update Count     : 178
 //
 #include <stdarg.h>                                                                             // va_start, va_end
+#include <stdio.h>
 #include <string.h>                                                                             // strlen
 #include <unistd.h>                                                                             // _exit, getpid
 …
 void abort( const char fmt[], ... ) __attribute__(( format(printf, 1, 2), __nothrow__, __leaf__, __noreturn__ ));
 void abort( bool signalAbort, const char fmt[], ... ) __attribute__(( format(printf, 2, 3), __nothrow__, __leaf__, __noreturn__ ));
+void __abort( bool signalAbort, const char fmt[], va_list args ) __attribute__(( __nothrow__, __leaf__, __noreturn__ ));
 extern "C" {
 …
                 va_list argp;
                 va_start( argp, fmt );
                 abort( false, fmt, argp );
+                __abort( false, fmt, argp );
                 va_end( argp );
+        }
 …
+}
+void abort( bool signalAbort, const char fmt[], ... ) {
+// Cannot forward va_list.
+void __abort( bool signalAbort, const char fmt[], va_list args ) {
         void * kernel_data = kernel_abort();                            // must be done here to lock down kernel
         int len;
 …
         assert( fmt );
-        va_list args;
-        va_start( args, fmt );
         len = vsnprintf( abort_text, abort_text_size, fmt, args );
-        va_end( args );
         __cfaabi_bits_write( STDERR_FILENO, abort_text, len );
         if ( fmt[strlen( fmt ) - 1] != '\n' ) {                         // add optional newline if missing at the end of the format text
                 __cfaabi_dbg_write( "\n", 1 );
+                __cfaabi_bits_write( STDERR_FILENO, "\n", 1 );
         } // if
         kernel_abort_msg( kernel_data, abort_text, abort_text_size );
 …
         va_list args;
         va_start( args, fmt );
+        abort( false, fmt, args );
+        __abort( false, fmt, args );
+    // CONTROL NEVER REACHES HERE!
         va_end( args );
+}
+void abort( bool signalAbort, const char fmt[], ... ) {
+    va_list args;
+    va_start( args, fmt );
+    __abort( signalAbort, fmt, args );
+    // CONTROL NEVER REACHES HERE!
+    va_end( args );
+}

libcfa/src/iostream.cfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Wed May 27 17:56:53 2015
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Thu Feb 20 15:53:23 2020
 // Update Count     : 829
+// Last Modified On : Sat May  2 18:30:25 2020
+// Update Count     : 1017
 //
 …
 #include <complex.h>                                                                    // creal, cimag
 } // extern "C"
+#include <bitmanip.hfa>                                                                 // fms
 …
+\
                 if ( f.base == 'b' || f.base == 'B' ) {                 /* bespoke binary format */ \
+                        int bits;                                                                                                       \
+                        if ( f.val == (T){0} ) bits = 1;                        /* force at least one bit to print */ \
+                        else bits = sizeof(long long int) * 8 - __builtin_clzll( f.val ); /* position of most significant bit */ \
+                        bits = bits > sizeof(f.val) * 8 ? sizeof(f.val) * 8 : bits; \
+                        int spaces = f.wd - bits;                                       /* can be negative */ \
+                        if ( ! f.flags.nobsdp ) { spaces -= 2; }        /* base prefix takes space */ \
+                        /* printf( "%d %d\n", bits, spaces ); */ \
+                        int bits = high1( f.val );                                      /* position of most significant bit */ \
+                        if ( bits == 0 ) bits = 1;                                      /* 0 value => force one bit to print */ \
+                        int spaces; \
                         if ( ! f.flags.left ) {                                         /* right justified ? */ \
                                 /* Note, base prefix then zero padding or spacing then prefix. */ \
+                                if ( f.flags.pad0 || f.flags.pc ) { \
+                                if ( f.flags.pc ) { \
+                                        spaces = f.wd - f.pc; \
+                                        if ( ! f.flags.nobsdp ) { spaces -= 2; } /* base prefix takes space */ \
+                                        if ( spaces > 0 ) fmt( os, "%*s", spaces, " " ); /* space pad */ \
                                         if ( ! f.flags.nobsdp ) { fmt( os, "0%c", f.base ); } \
                                         if ( f.flags.pc ) spaces = f.pc - bits; \
+                                        spaces = f.pc - bits; \
                                         if ( spaces > 0 ) fmt( os, "%0*d", spaces, 0 ); /* zero pad */ \
                                 } else { \
+                                        if ( spaces > 0 ) fmt( os, "%*s", spaces, " " ); /* space pad */ \
+                                        if ( ! f.flags.nobsdp ) { fmt( os, "0%c", f.base ); } \
+                                        spaces = f.wd - bits; \
+                                        if ( ! f.flags.nobsdp ) { spaces -= 2; } /* base prefix takes space */ \
+                                        if ( f.flags.pad0 ) { \
+                                                if ( ! f.flags.nobsdp ) { fmt( os, "0%c", f.base ); } \
+                                                if ( spaces > 0 ) fmt( os, "%0*d", spaces, 0 ); /* zero pad */ \
+                                        } else { \
+                                                if ( spaces > 0 ) fmt( os, "%*s", spaces, " " ); /* space pad */ \
+                                                if ( ! f.flags.nobsdp ) { fmt( os, "0%c", f.base ); } \
+                                        } /* if */ \
                                 } /* if */ \
+                        } else if ( ! f.flags.nobsdp ) { \
+                                fmt( os, "0%c", f.base ); \
+                        } else { \
+                                if ( ! f.flags.nobsdp ) fmt( os, "0%c", f.base ); \
+                                if ( f.flags.pc ) { \
+                                        spaces = f.pc - bits; \
+                                        if ( spaces > 0 ) fmt( os, "%0*d", spaces, 0 ); /* zero pad */ \
+                                        spaces = f.wd - f.pc; \
+                                } else { /* pad0 flag ignored with left flag */ \
+                                        spaces = f.wd - bits; \
+                                } /* if */ \
+                                if ( ! f.flags.nobsdp ) { spaces -= 2; } /* base prefix takes space */ \
                         } /* if */ \
                         int shift = (bits - 1) / 4 * 4; /* floor( bits - 1, 4 ) */ \
+                        int shift = floor( bits - 1, 4 ); \
                         typeof( f.val ) temp = f.val; \
                         fmt( os, "%s", shortbin[(temp >> shift) & 0xf] ); \
 …
 #define IntegralFMTImpl128( T, SIGNED, CODE, IFMTNP, IFMTP ) \
 forall( dtype ostype | ostream( ostype ) ) \
+static void base10_128( ostype & os, _Ostream_Manip(T) fmt ) { \
+        if ( fmt.val > UINT64_MAX ) { \
+                fmt.val /= P10_UINT64; \
+                base10_128( os, fmt ); /* recursive */ \
+                _Ostream_Manip(unsigned long long int) fmt2 @= { (uint64_t)(fmt.val % P10_UINT64), 0, 19, 'u', { .all : 0 } }; \
+                fmt2.flags.nobsdp = true; \
+                printf( "fmt2 %c %lld %d\n", fmt2.base, fmt2.val, fmt2.all );   \
+static void base10_128( ostype & os, _Ostream_Manip(T) f ) { \
+        if ( f.val > UINT64_MAX ) { \
+                unsigned long long int lsig = f.val % P10_UINT64; \
+                f.val /= P10_UINT64; /* msig */ \
+                base10_128( os, f ); /* recursion */ \
+                _Ostream_Manip(unsigned long long int) fmt @= { lsig, 0, 19, 'u', { .all : 0 } }; \
+                fmt.flags.nobsdp = true; \
+                /* printf( "fmt1 %c %lld %d\n", fmt.base, fmt.val, fmt.all ); */ \
                 sepOff( os ); \
                 (ostype &)(os | fmt2); \
+                (ostype &)(os | fmt); \
         } else { \
+                printf( "fmt %c %lld %d\n", fmt.base, fmt.val, fmt.all ); \
+                /* printf( "fmt2 %c %lld %d\n", f.base, (unsigned long long int)f.val, f.all ); */ \
+                _Ostream_Manip(SIGNED long long int) fmt @= { (SIGNED long long int)f.val, f.wd, f.pc, f.base, { .all : f.all } }; \
                 (ostype &)(os | fmt); \
         } /* if */ \
 } /* base10_128 */                                                 \
+} /* base10_128 */ \
 forall( dtype ostype | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
                 if ( $sepPrt( os ) ) fmt( os, "%s", $sepGetCur( os ) ); \
+\
                 if ( f.base == 'b' | f.base == 'o' | f.base == 'x' | f.base == 'X' ) { \
+                if ( f.base == 'b' | f.base == 'B' | f.base == 'o' | f.base == 'x' | f.base == 'X' ) { \
                         unsigned long long int msig = (unsigned long long int)(f.val >> 64); \
                         unsigned long long int lsig = (unsigned long long int)(f.val); \
 …
                         } else { \
                                 fmt2.flags.pad0 = fmt2.flags.nobsdp = true;     \
+                                if ( f.base == 'b' ) { \
+                                        if ( f.wd > 64 ) fmt.wd = f.wd - 64; \
+                                        fmt2.wd = 64; \
+                                if ( f.base == 'b' | f.base == 'B' ) { \
+                                        if ( fmt.flags.pc && fmt.pc > 64 ) fmt.pc -= 64; else { fmt.flags.pc = false; fmt.pc = 0; } \
+                                        if ( fmt.flags.left ) { \
+                                                fmt.flags.left = false; \
+                                                fmt.wd = 0; \
+                                                /* printf( "L %llo %llo %llo %d %d '%c' %x\n", msig, lsig, fmt.val, fmt.wd, fmt.pc, fmt.base, fmt.all ); */ \
+                                                fmt2.flags.left = true; \
+                                                int msigd = high1( msig ); \
+                                                fmt2.wd = f.wd - (fmt.pc > msigd ? fmt.pc : msigd); \
+                                                if ( ! fmt.flags.nobsdp ) fmt2.wd -= 2; /* compensate for 0b base specifier */ \
+                                                if ( (int)fmt2.wd < 64 ) fmt2.wd = 64; /* cast deals with negative value */ \
+                                                fmt2.flags.pc = true; fmt2.pc = 64; \
+                                        } else { \
+                                                if ( fmt.wd > 64 ) fmt.wd -= 64; \
+                                                else fmt.wd = 1; \
+                                                /* printf( "R %llo %llo %llo %d %d '%c' %x\n", msig, lsig, fmt.val, fmt.wd, fmt.pc, fmt.base, fmt.all ); */ \
+                                                fmt2.wd = 64; \
+                                        } /* if */ \
+                                        /* printf( "C %llo %d %d '%c' %x\n", fmt2.val, fmt2.wd, fmt2.pc, fmt2.base, fmt2.all ); */ \
                                         (ostype &)(os | fmt | "" | fmt2); \
                                 } else if ( f.base == 'o' ) { \
+                                        if ( fmt.flags.pc && fmt.pc > 22 ) fmt.pc -= 22; else { fmt.flags.pc = false; fmt.pc = 0; } \
                                         fmt.val = (unsigned long long int)fmt.val >> 2; \
+                                        if ( f.wd > 21 ) fmt.wd = f.wd - 21; \
+                                        fmt2.wd = 1; \
+                                        fmt2.val = ((msig & 0x3) << 1) + 1; \
+                                        (ostype &)(os | fmt | "" | fmt2); \
+                                        sepOff( os ); \
+                                        fmt2.wd = 21; \
+                                        fmt2.val = lsig & 0x7fffffffffffffff; \
+                                        fmt2.val = ((msig & 0x3) << 1) + ((lsig & 0x8000000000000000U) != 0); \
+                                        if ( fmt.flags.left ) { \
+                                                fmt.flags.left = false; \
+                                                fmt.wd = 0; \
+                                                /* printf( "L %llo %llo %llo %d %d '%c' %x %llo %d %d '%c' %x\n", msig, lsig, fmt.val, fmt.wd, fmt.pc, fmt.base, fmt.all, fmt2.val, fmt2.wd, fmt2.pc, fmt2.base, fmt2.all ); */ \
+                                                (ostype &)(os | fmt | "" | fmt2); \
+                                                sepOff( os ); \
+                                                fmt2.flags.left = true; \
+                                                int msigd = ceiling( high1( fmt.val ), 3 ); \
+                                                fmt2.wd = f.wd - (fmt.pc > msigd ? fmt.pc : msigd); \
+                                                if ( ! fmt.flags.nobsdp ) fmt2.wd -= 1; /* compensate for 0 base specifier */ \
+                                                if ( (int)fmt2.wd < 21 ) fmt2.wd = 21; /* cast deals with negative value */ \
+                                                fmt2.flags.pc = true; fmt2.pc = 21; \
+                                        } else { \
+                                                if ( fmt.wd > 22 ) fmt.wd -= 22; \
+                                                else fmt.wd = 1; \
+                                                /* printf( "R %llo %llo %llo %d %d '%c' %x %llo %d %d '%c' %x\n", msig, lsig, fmt.val, fmt.wd, fmt.pc, fmt.base, fmt.all, fmt2.val, fmt2.wd, fmt2.pc, fmt2.base, fmt2.all ); */ \
+                                                (ostype &)(os | fmt | "" | fmt2); \
+                                                sepOff( os ); \
+                                                fmt2.wd = 21; \
+                                        } /* if */ \
+                                        fmt2.val = lsig & 0x7fffffffffffffffU; \
+                                        /* printf( "\nC %llo %d %d '%c' %x\n", fmt2.val, fmt2.wd, fmt2.pc, fmt2.base, fmt2.all ); */ \
                                         (ostype &)(os | fmt2); \
+                                } else { \
+                                        if ( f.flags.left ) { \
+                                                if ( f.wd > 16 ) fmt2.wd = f.wd - 16;   \
+                                                fmt.wd = 16;                                                    \
+                                } else { /* f.base == 'x'  | f.base == 'X' */ \
+                                        if ( fmt.flags.pc && fmt.pc > 16 ) fmt.pc -= 16; else { fmt.flags.pc = false; fmt.pc = 0; } \
+                                        if ( fmt.flags.left ) { \
+                                                fmt.flags.left = false; \
+                                                fmt.wd = 0; \
+                                                /* printf( "L %llo %llo %llo %d %d '%c' %x\n", msig, lsig, fmt.val, fmt.wd, fmt.pc, fmt.base, fmt.all ); */ \
+                                                fmt2.flags.left = true; \
+                                                int msigd = high1( msig ); \
+                                                fmt2.wd = f.wd - (fmt.pc > msigd ? fmt.pc : msigd); \
+                                                if ( ! fmt.flags.nobsdp ) fmt2.wd -= 2; /* compensate for 0x base specifier */ \
+                                                if ( (int)fmt2.wd < 16 ) fmt2.wd = 16; /* cast deals with negative value */ \
+                                                fmt2.flags.pc = true; fmt2.pc = 16; \
                                         } else { \
+                                                if ( f.wd > 16 ) fmt.wd = f.wd - 16;    \
+                                                fmt2.wd = 16;                                                   \
+                                                if ( fmt.wd > 16 ) fmt.wd -= 16; \
+                                                else fmt.wd = 1; \
+                                                /* printf( "R %llo %llo %llo %d %d '%c' %x\n", msig, lsig, fmt.val, fmt.wd, fmt.pc, fmt.base, fmt.all ); */ \
+                                                fmt2.wd = 16; \
                                         } /* if */ \
+                                        /* printf( "C %llo %d %d '%c' %x\n", fmt2.val, fmt2.wd, fmt2.pc, fmt2.base, fmt2.all ); */ \
                                         (ostype &)(os | fmt | "" | fmt2); \
                                 } /* if */ \
                         } /* if */ \
                 } else { \
+                        if ( CODE == 'd' ) { \
+                                if ( f.val < 0 )  { fmt( os, "-" ); sepOff( os ); f.val = -f.val; f.flags.sign = false; } \
+                        } /* if */ \
                         base10_128( os, f ); \
                 } /* if */ \

libcfa/src/startup.cfa

-              rb7d6a36
+              r6a490b2
 //
+#include <time.h>                                                                               // tzset
+#include <time.h>                // tzset
+#include <locale.h>        // setlocale
 #include "startup.hfa"
 …
     void __cfaabi_appready_startup( void ) {
                 tzset();                                                                                // initialize time global variables
+                setlocale(LC_NUMERIC, "");
                 #ifdef __CFA_DEBUG__
                 extern void heapAppStart();
 …
 struct __spinlock_t;
 extern "C" {
         void __cfaabi_dbg_record(struct __spinlock_t & this, const char prev_name[]) __attribute__(( weak )) {}
+        void __cfaabi_dbg_record_lock(struct __spinlock_t & this, const char prev_name[]) __attribute__(( weak )) {}
+}

libcfa/src/stdhdr/malloc.h

-              rb7d6a36
+              r6a490b2
 // Created On       : Thu Jul 20 15:58:16 2017
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Sat Aug 11 09:06:31 2018
 // Update Count     : 10
+// Last Modified On : Thu Apr 16 22:44:06 2020
+// Update Count     : 13
 //
 …
 extern "C" {
+void * aalloc( size_t noOfElems, size_t elemSize );
+void * amemalign( size_t alignment, size_t noOfElems, size_t elemSize );
+void * cmemalign( size_t alignment, size_t noOfElems, size_t elemSize );
 size_t malloc_alignment( void * );
 bool malloc_zero_fill( void * );
+size_t malloc_size( void * );
 int malloc_stats_fd( int fd );
-void * cmemalign( size_t alignment, size_t noOfElems, size_t elemSize );
 } // extern "C"

libcfa/src/stdlib.cfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Thu Jan 28 17:10:29 2016
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Tue Feb  4 08:27:08 2020
 // Update Count     : 486
+// Last Modified On : Thu Apr 16 22:43:33 2020
+// Update Count     : 498
 //
 …
 #define _XOPEN_SOURCE 600                                                               // posix_memalign, *rand48
 #include <string.h>                                                                             // memcpy, memset
-#include <malloc.h>                                                                             // malloc_usable_size
 //#include <math.h>                                                                             // fabsf, fabs, fabsl
 #include <complex.h>                                                                    // _Complex_I
 …
         } // alloc_set
+        T * alloc_set( T ptr[], size_t dim, T fill ) {          // realloc array with fill
+                size_t olen = malloc_usable_size( ptr );                // current allocation
+                void * nptr = (void *)realloc( (void *)ptr, dim * sizeof(T) ); // C realloc
+                size_t nlen = malloc_usable_size( nptr );               // new allocation
+                if ( nlen > olen ) {                                                    // larger ?
+                        for ( i; malloc_size( ptr ) / sizeof(T) ~ dim ) {
+                                memcpy( &ptr[i], &fill, sizeof(T) );    // initialize with fill value
+                        } // for
+                } // if
+                return (T *)nptr;
+        } // alloc_align_set
         T * alloc_align_set( T ptr[], size_t align, char fill ) { // aligned realloc with fill
                 size_t olen = malloc_usable_size( ptr );                // current allocation
 …
                 if ( nlen > olen ) {                                                    // larger ?
                         memset( (char *)nptr + olen, (int)fill, nlen - olen ); // initialize added storage
+                } // if
+                return (T *)nptr;
+        } // alloc_align_set
+        T * alloc_align_set( T ptr[], size_t align, size_t dim, T fill ) { // aligned realloc with fill
+                size_t olen = malloc_usable_size( ptr );                // current allocation
+                void * nptr = (void *)realloc( (void *)ptr, align, sizeof(T) ); // CFA realloc
+                // char * nptr = alloc_align( ptr, align );
+                size_t nlen = malloc_usable_size( nptr );               // new allocation
+                if ( nlen > olen ) {                                                    // larger ?
+                        for ( i; dim ) { memcpy( &ptr[i], &fill, sizeof(T) ); } // initialize with fill value
                 } // if
                 return (T *)nptr;

libcfa/src/stdlib.hfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Thu Jan 28 17:12:35 2016
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Tue Feb  4 08:27:01 2020
 // Update Count     : 401
+// Last Modified On : Thu Apr 16 22:44:05 2020
+// Update Count     : 432
 //
 …
 #include <stdlib.h>                                                                             // *alloc, strto*, ato*
+// Reduce includes by explicitly defining these routines.
 extern "C" {
         void * memalign( size_t align, size_t size );           // malloc.h
+        size_t malloc_usable_size( void * ptr );                        // malloc.h
+        size_t malloc_size( void * addr );                                      // CFA heap
+        void * cmemalign( size_t alignment, size_t noOfElems, size_t elemSize ); // CFA heap
         void * memset( void * dest, int fill, size_t size ); // string.h
         void * memcpy( void * dest, const void * src, size_t size ); // string.h
     void * cmemalign( size_t alignment, size_t noOfElems, size_t elemSize ); // CFA heap
+        void * resize( void * oaddr, size_t size );                     // CFA heap
 } // extern "C"
+void * resize( void * oaddr, size_t nalign, size_t size ); // CFA heap
 void * realloc( void * oaddr, size_t nalign, size_t size ); // CFA heap
 …
 static inline forall( dtype T | sized(T) ) {
         // C dynamic allocation
+        // Cforall safe equivalents, i.e., implicit size specification
         T * malloc( void ) {
 …
                 return posix_memalign( (void **)ptr, align, sizeof(T) ); // C posix_memalign
         } // posix_memalign
+        // Cforall dynamic allocation
+} // distribution
+static inline forall( dtype T | sized(T) ) {
+        // Cforall safe general allocation, fill, resize, array
         T * alloc( void ) {
 …
         } // alloc
+        T * alloc( T ptr[], size_t dim ) {                                      // realloc
+                return (T *)(void *)realloc( (void *)ptr, dim * sizeof(T) ); // C realloc
+        forall( dtype S | sized(S) )
+        T * alloc( S ptr[], size_t dim = 1 ) {                          // singleton/array resize
+                size_t len = malloc_usable_size( ptr );                 // current bucket size
+                if ( sizeof(T) * dim > len ) {                                  // not enough space ?
+                        T * temp = alloc( dim );                                        // new storage
+                        free( ptr );                                                            // free old storage
+                        return temp;
+                } else {
+                        return (T *)ptr;
+                } // if
+        } // alloc
+        T * alloc( T ptr[], size_t dim, bool copy = true ) {
+                if ( copy ) {                                                                   // realloc
+                        return (T *)(void *)realloc( (void *)ptr, dim * sizeof(T) ); // C realloc
+                } else {
+                        struct __Unknown {};
+                        return alloc( (__Unknown *)ptr, dim );          // reuse, cheat making T/S different types
+                } // if
         } // alloc
 …
 forall( dtype T | sized(T) ) {
         T * alloc_set( T ptr[], size_t dim, char fill );        // realloc array with fill
+        T * alloc_set( T ptr[], size_t dim, T fill );           // realloc array with fill
 } // distribution
 …
         T * alloc_align( T ptr[], size_t align ) {                      // aligned realloc array
                 return (T *)(void *)realloc( (void *)ptr, align, sizeof(T) ); // CFA realloc
+        } // alloc_align
+        forall( dtype S | sized(S) )
+        T * alloc_align( S ptr[], size_t align ) {                      // aligned reuse array
+                return (T *)(void *)resize( (void *)ptr, align, sizeof(T) ); // CFA realloc
         } // alloc_align
 …
 forall( dtype T | sized(T) ) {
+        T * alloc_align_set( T ptr[], size_t align, char fill ); // aligned realloc with fill
+        T * alloc_align_set( T ptr[], size_t align, T fill ); // aligned realloc with fill
         T * alloc_align_set( T ptr[], size_t align, size_t dim, char fill ); // aligned realloc array with fill
+} // distribution
+static inline forall( dtype T | sized(T) ) {
+        // data, non-array types
+        T * alloc_align_set( T ptr[], size_t align, size_t dim, T fill ); // aligned realloc array with fill
+} // distribution
+static inline forall( dtype T | sized(T) ) {
+        // Cforall safe initialization/copy, i.e., implicit size specification, non-array types
         T * memset( T * dest, char fill ) {
                 return (T *)memset( dest, fill, sizeof(T) );
 …
 static inline forall( dtype T | sized(T) ) {
         // data, array types
+        // Cforall safe initialization/copy, i.e., implicit size specification, array types
         T * amemset( T dest[], char fill, size_t dim ) {
                 return (T *)(void *)memset( dest, fill, dim * sizeof(T) ); // C memset
 …
 } // distribution
 // allocation/deallocation and constructor/destructor, non-array types
+// Cforall allocation/deallocation and constructor/destructor, non-array types
 forall( dtype T | sized(T), ttype Params | { void ?{}( T &, Params ); } ) T * new( Params p );
 forall( dtype T | sized(T) | { void ^?{}( T & ); } ) void delete( T * ptr );
 forall( dtype T, ttype Params | sized(T) | { void ^?{}( T & ); void delete( Params ); } ) void delete( T * ptr, Params rest );
 // allocation/deallocation and constructor/destructor, array types
+// Cforall allocation/deallocation and constructor/destructor, array types
 forall( dtype T | sized(T), ttype Params | { void ?{}( T &, Params ); } ) T * anew( size_t dim, Params p );
 forall( dtype T | sized(T) | { void ^?{}( T & ); } ) void adelete( size_t dim, T arr[] );

src/AST/Convert.cpp

-              rb7d6a36
+              r6a490b2
+        }
+        const ast::Stmt * visit(const ast::SuspendStmt * node ) override final {
+                if ( inCache( node ) ) return nullptr;
+                auto stmt = new SuspendStmt();
+                stmt->then   = get<CompoundStmt>().accept1( node->then   );
+                switch(node->type) {
+                        case ast::SuspendStmt::None     : stmt->type = SuspendStmt::None     ; break;
+                        case ast::SuspendStmt::Coroutine: stmt->type = SuspendStmt::Coroutine; break;
+                        case ast::SuspendStmt::Generator: stmt->type = SuspendStmt::Generator; break;
+                }
+                return stmtPostamble( stmt, node );
+        }
         const ast::Stmt * visit( const ast::WaitForStmt * node ) override final {
                 if ( inCache( node ) ) return nullptr;
 …
+        }
+        virtual void visit( const SuspendStmt * old ) override final {
+                if ( inCache( old ) ) return;
+                ast::SuspendStmt::Type type;
+                switch (old->type) {
+                        case SuspendStmt::Coroutine: type = ast::SuspendStmt::Coroutine; break;
+                        case SuspendStmt::Generator: type = ast::SuspendStmt::Generator; break;
+                        case SuspendStmt::None     : type = ast::SuspendStmt::None     ; break;
+                        default: abort();
+                }
+                this->node = new ast::SuspendStmt(
+                        old->location,
+                        GET_ACCEPT_1(then  , CompoundStmt),
+                        type,
+                        GET_LABELS_V(old->labels)
+                );
+                cache.emplace( old, this->node );
+        }
         virtual void visit( const WaitForStmt * old ) override final {
                 if ( inCache( old ) ) return;

src/AST/Decl.hpp

-              rb7d6a36
+              r6a490b2
         bool is_coroutine() { return kind == Coroutine; }
+        bool is_monitor() { return kind == Monitor; }
+        bool is_thread() { return kind == Thread; }
+        bool is_generator() { return kind == Generator; }
+        bool is_monitor  () { return kind == Monitor  ; }
+        bool is_thread   () { return kind == Thread   ; }
         const Decl * accept( Visitor & v ) const override { return v.visit( this ); }

src/AST/Fwd.hpp

rb7d6a36	r6a490b2
53	53	class CatchStmt;
54	54	class FinallyStmt;
	55	class SuspendStmt;
55	56	class WaitForStmt;
56	57	class WithStmt;

src/AST/Pass.hpp

rb7d6a36	r6a490b2
111	111	const ast::Stmt * visit( const ast::CatchStmt * ) override final;
112	112	const ast::Stmt * visit( const ast::FinallyStmt * ) override final;
	113	const ast::Stmt * visit( const ast::SuspendStmt * ) override final;
113	114	const ast::Stmt * visit( const ast::WaitForStmt * ) override final;
114	115	const ast::Decl * visit( const ast::WithStmt * ) override final;

src/AST/Pass.impl.hpp

-              rb7d6a36
+              r6a490b2
 //--------------------------------------------------------------------------
+// FinallyStmt
+template< typename pass_t >
+const ast::Stmt * ast::Pass< pass_t >::visit( const ast::SuspendStmt * node ) {
+        VISIT_START( node );
+        VISIT(
+                maybe_accept( node, &SuspendStmt::then   );
+        )
+        VISIT_END( Stmt, node );
+}
+//--------------------------------------------------------------------------
 // WaitForStmt
 template< typename pass_t >

src/AST/Print.cpp

-              rb7d6a36
+              r6a490b2
                 safe_print( node->body );
                 --indent;
+                return node;
+        }
+        virtual const ast::Stmt * visit( const ast::SuspendStmt * node ) override final {
+                os << "Suspend Statement";
+                switch (node->type) {
+                        case ast::SuspendStmt::None     : os << " with implicit target"; break;
+                        case ast::SuspendStmt::Generator: os << " for generator"; break;
+                        case ast::SuspendStmt::Coroutine: os << " for coroutine"; break;
+                }
+                os << endl;
+                ++indent;
+                if(node->then) {
+                        os << indent << " with post statement :" << endl;
+                        safe_print( node->then );
+                }
+                ++indent;
                 return node;

src/AST/Stmt.hpp

-              rb7d6a36
+              r6a490b2
 };
+/// Suspend statement
+class SuspendStmt final : public Stmt {
+public:
+        ptr<CompoundStmt> then;
+        enum Type { None, Coroutine, Generator } type = None;
+        SuspendStmt( const CodeLocation & loc, const CompoundStmt * then, Type type, std::vector<Label> && labels = {} )
+        : Stmt(loc, std::move(labels)), then(then), type(type) {}
+        const Stmt * accept( Visitor & v ) const override { return v.visit( this ); }
+private:
+        SuspendStmt * clone() const override { return new SuspendStmt{ *this }; }
+        MUTATE_FRIEND
+};
 /// Wait for concurrency statement `when (...) waitfor (... , ...) ... timeout(...) ... else ...`
 class WaitForStmt final : public Stmt {

src/AST/Visitor.hpp

rb7d6a36	r6a490b2
47	47	virtual const ast::Stmt * visit( const ast::CatchStmt * ) = 0;
48	48	virtual const ast::Stmt * visit( const ast::FinallyStmt * ) = 0;
	49	virtual const ast::Stmt * visit( const ast::SuspendStmt * ) = 0;
49	50	virtual const ast::Stmt * visit( const ast::WaitForStmt * ) = 0;
50	51	virtual const ast::Decl * visit( const ast::WithStmt * ) = 0;

src/Common/PassVisitor.h

-              rb7d6a36
+              r6a490b2
         virtual void visit( FinallyStmt * finallyStmt ) override final;
         virtual void visit( const FinallyStmt * finallyStmt ) override final;
+        virtual void visit( SuspendStmt * suspendStmt ) override final;
+        virtual void visit( const SuspendStmt * suspendStmt ) override final;
         virtual void visit( WaitForStmt * waitforStmt ) override final;
         virtual void visit( const WaitForStmt * waitforStmt ) override final;
 …
         virtual Statement * mutate( CatchStmt * catchStmt ) override final;
         virtual Statement * mutate( FinallyStmt * finallyStmt ) override final;
+        virtual Statement * mutate( SuspendStmt * suspendStmt ) override final;
         virtual Statement * mutate( WaitForStmt * waitforStmt ) override final;
         virtual Declaration * mutate( WithStmt * withStmt ) override final;

src/Common/PassVisitor.impl.h

-              rb7d6a36
+              r6a490b2
 //--------------------------------------------------------------------------
+// SuspendStmt
+template< typename pass_type >
+void PassVisitor< pass_type >::visit( SuspendStmt * node ) {
+        VISIT_START( node );
+        maybeAccept_impl( node->then  , *this );
+        VISIT_END( node );
+}
+template< typename pass_type >
+void PassVisitor< pass_type >::visit( const SuspendStmt * node ) {
+        VISIT_START( node );
+        maybeAccept_impl( node->then  , *this );
+        VISIT_END( node );
+}
+template< typename pass_type >
+Statement * PassVisitor< pass_type >::mutate( SuspendStmt * node ) {
+        MUTATE_START( node );
+        maybeMutate_impl( node->then  , *this );
+        MUTATE_END( Statement, node );
+}
+//--------------------------------------------------------------------------
 // WaitForStmt
 template< typename pass_type >

src/CompilationState.cc

rb7d6a36	r6a490b2
27	27	nopreludep = false,
28	28	genproto = false,
	29	deterministic_output = false,
29	30	nomainp = false,
30	31	parsep = false,

src/CompilationState.h

rb7d6a36	r6a490b2
28	28	nopreludep,
29	29	genproto,
	30	deterministic_output,
30	31	nomainp,
31	32	parsep,

src/Concurrency/Keywords.cc

-              rb7d6a36
+              r6a490b2
 #include "Concurrency/Keywords.h"
+#include <cassert>                 // for assert
+#include <string>                  // for string, operator==
+#include "Common/PassVisitor.h"    // for PassVisitor
+#include "Common/SemanticError.h"  // for SemanticError
+#include "Common/utility.h"        // for deleteAll, map_range
+#include "CodeGen/OperatorTable.h" // for isConstructor
+#include "InitTweak/InitTweak.h"   // for getPointerBase
+#include "SynTree/LinkageSpec.h"   // for Cforall
+#include "SynTree/Constant.h"      // for Constant
+#include "SynTree/Declaration.h"   // for StructDecl, FunctionDecl, ObjectDecl
+#include "SynTree/Expression.h"    // for VariableExpr, ConstantExpr, Untype...
+#include "SynTree/Initializer.h"   // for SingleInit, ListInit, Initializer ...
+#include "SynTree/Label.h"         // for Label
+#include "SynTree/Statement.h"     // for CompoundStmt, DeclStmt, ExprStmt
+#include "SynTree/Type.h"          // for StructInstType, Type, PointerType
+#include "SynTree/Visitor.h"       // for Visitor, acceptAll
+#include <cassert>                        // for assert
+#include <string>                         // for string, operator==
+#include "Common/PassVisitor.h"           // for PassVisitor
+#include "Common/SemanticError.h"         // for SemanticError
+#include "Common/utility.h"               // for deleteAll, map_range
+#include "CodeGen/OperatorTable.h"        // for isConstructor
+#include "ControlStruct/LabelGenerator.h" // for LebelGenerator
+#include "InitTweak/InitTweak.h"          // for getPointerBase
+#include "SynTree/LinkageSpec.h"          // for Cforall
+#include "SynTree/Constant.h"             // for Constant
+#include "SynTree/Declaration.h"          // for StructDecl, FunctionDecl, ObjectDecl
+#include "SynTree/Expression.h"           // for VariableExpr, ConstantExpr, Untype...
+#include "SynTree/Initializer.h"          // for SingleInit, ListInit, Initializer ...
+#include "SynTree/Label.h"                // for Label
+#include "SynTree/Statement.h"            // for CompoundStmt, DeclStmt, ExprStmt
+#include "SynTree/Type.h"                 // for StructInstType, Type, PointerType
+#include "SynTree/Visitor.h"              // for Visitor, acceptAll
 class Attribute;
 …
         //      int data;                                  int data;
         //      a_struct_t more_data;                      a_struct_t more_data;
         //                                =>             thread_desc __thrd_d;
+        //                                =>             $thread __thrd_d;
         // };                                        };
         //                                           static inline thread_desc * get_thread( MyThread * this ) { return &this->__thrd_d; }
+        //                                           static inline $thread * get_thread( MyThread * this ) { return &this->__thrd_d; }
         //
         class ThreadKeyword final : public ConcurrentSueKeyword {
 …
                 ThreadKeyword() : ConcurrentSueKeyword(
                         "thread_desc",
+                        "$thread",
                         "__thrd",
                         "get_thread",
 …
         //      int data;                                  int data;
         //      a_struct_t more_data;                      a_struct_t more_data;
         //                                =>             coroutine_desc __cor_d;
+        //                                =>             $coroutine __cor_d;
         // };                                        };
         //                                           static inline coroutine_desc * get_coroutine( MyCoroutine * this ) { return &this->__cor_d; }
+        //                                           static inline $coroutine * get_coroutine( MyCoroutine * this ) { return &this->__cor_d; }
         //
         class CoroutineKeyword final : public ConcurrentSueKeyword {
 …
                 CoroutineKeyword() : ConcurrentSueKeyword(
                         "coroutine_desc",
+                        "$coroutine",
                         "__cor",
                         "get_coroutine",
 …
         };
         //-----------------------------------------------------------------------------
         //Handles monitor type declarations :
 …
         //      int data;                                  int data;
         //      a_struct_t more_data;                      a_struct_t more_data;
         //                                =>             monitor_desc __mon_d;
+        //                                =>             $monitor __mon_d;
         // };                                        };
         //                                           static inline monitor_desc * get_coroutine( MyMonitor * this ) { return &this->__cor_d; }
+        //                                           static inline $monitor * get_coroutine( MyMonitor * this ) { return &this->__cor_d; }
         //
         class MonitorKeyword final : public ConcurrentSueKeyword {
 …
                 MonitorKeyword() : ConcurrentSueKeyword(
                         "monitor_desc",
+                        "$monitor",
                         "__mon",
                         "get_monitor",
 …
         //-----------------------------------------------------------------------------
+        //Handles generator type declarations :
+        // generator MyGenerator {                   struct MyGenerator {
+        //      int data;                                  int data;
+        //      a_struct_t more_data;                      a_struct_t more_data;
+        //                                =>             int __gen_next;
+        // };                                        };
+        //
+        class GeneratorKeyword final : public ConcurrentSueKeyword {
+          public:
+                GeneratorKeyword() : ConcurrentSueKeyword(
+                        "$generator",
+                        "__generator_state",
+                        "get_generator",
+                        "Unable to find builtin type $generator\n",
+                        true,
+                        AggregateDecl::Generator
+                )
+                {}
+                virtual ~GeneratorKeyword() {}
+                virtual bool is_target( StructDecl * decl ) override final { return decl->is_generator(); }
+                static void implement( std::list< Declaration * > & translationUnit ) {
+                        PassVisitor< GeneratorKeyword > impl;
+                        mutateAll( translationUnit, impl );
+                }
+        };
+        //-----------------------------------------------------------------------------
+        class SuspendKeyword final : public WithStmtsToAdd, public WithGuards {
+        public:
+                SuspendKeyword() = default;
+                virtual ~SuspendKeyword() = default;
+                void  premutate( FunctionDecl * );
+                DeclarationWithType * postmutate( FunctionDecl * );
+                Statement * postmutate( SuspendStmt * );
+                static void implement( std::list< Declaration * > & translationUnit ) {
+                        PassVisitor< SuspendKeyword > impl;
+                        mutateAll( translationUnit, impl );
+                }
+        private:
+                DeclarationWithType * is_main( FunctionDecl * );
+                bool is_real_suspend( FunctionDecl * );
+                Statement * make_generator_suspend( SuspendStmt * );
+                Statement * make_coroutine_suspend( SuspendStmt * );
+                struct LabelPair {
+                        Label obj;
+                        int   idx;
+                };
+                LabelPair make_label() {
+                        labels.push_back( gen.newLabel("generator") );
+                        return { labels.back(), int(labels.size()) };
+                }
+                DeclarationWithType * in_generator = nullptr;
+                FunctionDecl * decl_suspend = nullptr;
+                std::vector<Label> labels;
+                ControlStruct::LabelGenerator & gen = *ControlStruct::LabelGenerator::getGenerator();
+        };
+        //-----------------------------------------------------------------------------
         //Handles mutex routines definitions :
         // void foo( A * mutex a, B * mutex b,  int i ) {                  void foo( A * a, B * b,  int i ) {
         //                                                                       monitor_desc * __monitors[] = { get_monitor(a), get_monitor(b) };
+        //                                                                       $monitor * __monitors[] = { get_monitor(a), get_monitor(b) };
         //                                                                       monitor_guard_t __guard = { __monitors, 2 };
         //    /*Some code*/                                       =>           /*Some code*/
 …
         //Handles mutex routines definitions :
         // void foo( A * mutex a, B * mutex b,  int i ) {                  void foo( A * a, B * b,  int i ) {
         //                                                                       monitor_desc * __monitors[] = { get_monitor(a), get_monitor(b) };
+        //                                                                       $monitor * __monitors[] = { get_monitor(a), get_monitor(b) };
         //                                                                       monitor_guard_t __guard = { __monitors, 2 };
         //    /*Some code*/                                       =>           /*Some code*/
 …
                 CoroutineKeyword        ::implement( translationUnit );
                 MonitorKeyword  ::implement( translationUnit );
+                GeneratorKeyword  ::implement( translationUnit );
+                SuspendKeyword    ::implement( translationUnit );
+        }
 …
         Expression * ConcurrentSueKeyword::postmutate( KeywordCastExpr * cast ) {
                 if ( cast_target == cast->target ) {
                         // convert (thread &)t to (thread_desc &)*get_thread(t), etc.
+                        // convert (thread &)t to ($thread &)*get_thread(t), etc.
                         if( !type_decl ) SemanticError( cast, context_error );
                         if( !dtor_decl ) SemanticError( cast, context_error );
 …
                         get_type,
                         nullptr,
                         noAttributes,
+                        { new Attribute("const") },
                         Type::Inline
                 );
 …
                 declsToAddAfter.push_back( get_decl );
+                // get_decl->fixUniqueId();
+        }
+        }
+        //=============================================================================================
+        // Suspend keyword implementation
+        //=============================================================================================
+        DeclarationWithType * SuspendKeyword::is_main( FunctionDecl * func) {
+                if(func->name != "main") return nullptr;
+                if(func->type->parameters.size() != 1) return nullptr;
+                auto param = func->type->parameters.front();
+                auto type  = dynamic_cast<ReferenceType * >(param->get_type());
+                if(!type) return nullptr;
+                auto obj   = dynamic_cast<StructInstType *>(type->base);
+                if(!obj) return nullptr;
+                if(!obj->baseStruct->is_generator()) return nullptr;
+                return param;
+        }
+        bool SuspendKeyword::is_real_suspend( FunctionDecl * func ) {
+                if(isMangled(func->linkage)) return false; // the real suspend isn't mangled
+                if(func->name != "__cfactx_suspend") return false; // the real suspend has a specific name
+                if(func->type->parameters.size() != 0) return false; // Too many parameters
+                if(func->type->returnVals.size() != 0) return false; // Too many return values
+                return true;
+        }
+        void SuspendKeyword::premutate( FunctionDecl * func ) {
+                GuardValue(in_generator);
+                in_generator = nullptr;
+                // Is this the real suspend?
+                if(is_real_suspend(func)) {
+                        decl_suspend = decl_suspend ? decl_suspend : func;
+                        return;
+                }
+                // Is this the main of a generator?
+                auto param = is_main( func );
+                if(!param) return;
+                if(func->type->returnVals.size() != 0) SemanticError(func->location, "Generator main must return void");
+                in_generator = param;
+                GuardValue(labels);
+                labels.clear();
+        }
+        DeclarationWithType * SuspendKeyword::postmutate( FunctionDecl * func ) {
+                if( !func->statements ) return func; // Not the actual definition, don't do anything
+                if( !in_generator     ) return func; // Not in a generator, don't do anything
+                if( labels.empty()    ) return func; // Generator has no states, nothing to do, could throw a warning
+                // This is a generator main, we need to add the following code to the top
+                // static void * __generator_labels[] = {&&s0, &&s1, ...};
+                // goto * __generator_labels[gen.__generator_state];
+                const auto & loc = func->location;
+                const auto first_label = gen.newLabel("generator");
+                // for each label add to declaration
+                std::list<Initializer*> inits = { new SingleInit( new LabelAddressExpr( first_label ) ) };
+                for(const auto & label : labels) {
+                        inits.push_back(
+                                new SingleInit(
+                                        new LabelAddressExpr( label )
+                                )
+                        );
+                }
+                auto init = new ListInit(std::move(inits), noDesignators, true);
+                labels.clear();
+                // create decl
+                auto decl = new ObjectDecl(
+                        "__generator_labels",
+                        Type::StorageClasses( Type::Static ),
+                        LinkageSpec::AutoGen,
+                        nullptr,
+                        new ArrayType(
+                                Type::Qualifiers(),
+                                new PointerType(
+                                        Type::Qualifiers(),
+                                        new VoidType( Type::Qualifiers() )
+                                ),
+                                nullptr,
+                                false, false
+                        ),
+                        init
+                );
+                // create the goto
+                assert(in_generator);
+                auto go_decl = new ObjectDecl(
+                        "__generator_label",
+                        noStorageClasses,
+                        LinkageSpec::AutoGen,
+                        nullptr,
+                        new PointerType(
+                                Type::Qualifiers(),
+                                new VoidType( Type::Qualifiers() )
+                        ),
+                        new SingleInit(
+                                new UntypedExpr(
+                                        new NameExpr("?[?]"),
+                                        {
+                                                new NameExpr("__generator_labels"),
+                                                new UntypedMemberExpr(
+                                                        new NameExpr("__generator_state"),
+                                                        new VariableExpr( in_generator )
+                                                )
+                                        }
+                                )
+                        )
+                );
+                go_decl->location = loc;
+                auto go = new BranchStmt(
+                        new VariableExpr( go_decl ),
+                        BranchStmt::Goto
+                );
+                go->location = loc;
+                go->computedTarget->location = loc;
+                auto noop = new NullStmt({ first_label });
+                noop->location = loc;
+                // wrap everything in a nice compound
+                auto body = new CompoundStmt({
+                        new DeclStmt( decl ),
+                        new DeclStmt( go_decl ),
+                        go,
+                        noop,
+                        func->statements
+                });
+                body->location   = loc;
+                func->statements = body;
+                return func;
+        }
+        Statement * SuspendKeyword::postmutate( SuspendStmt * stmt ) {
+                SuspendStmt::Type type = stmt->type;
+                if(type == SuspendStmt::None) {
+                        // This suspend has a implicit target, find it
+                        type = in_generator ? SuspendStmt::Generator : SuspendStmt::Coroutine;
+                }
+                // Check that the target makes sense
+                if(!in_generator && type == SuspendStmt::Generator) SemanticError( stmt->location, "'suspend generator' must be used inside main of generator type.");
+                // Act appropriately
+                switch(type) {
+                        case SuspendStmt::Generator: return make_generator_suspend(stmt);
+                        case SuspendStmt::Coroutine: return make_coroutine_suspend(stmt);
+                        default: abort();
+                }
+        }
+        Statement * SuspendKeyword::make_generator_suspend( SuspendStmt * stmt ) {
+                assert(in_generator);
+                // Target code is :
+                //   gen.__generator_state = X;
+                //   { THEN }
+                //   return;
+                //   __gen_X:;
+                // Save the location and delete the old statement, we only need the location from this point on
+                auto loc = stmt->location;
+                // Build the label and get its index
+                auto label = make_label();
+                // Create the context saving statement
+                auto save = new ExprStmt( new UntypedExpr(
+                        new NameExpr( "?=?" ),
+                        {
+                                new UntypedMemberExpr(
+                                        new NameExpr("__generator_state"),
+                                        new VariableExpr( in_generator )
+                                ),
+                                new ConstantExpr(
+                                        Constant::from_int( label.idx )
+                                )
+                        }
+                ));
+                assert(save->expr);
+                save->location = loc;
+                stmtsToAddBefore.push_back( save );
+                // if we have a then add it here
+                auto then = stmt->then;
+                stmt->then = nullptr;
+                delete stmt;
+                if(then) stmtsToAddBefore.push_back( then );
+                // Create the return statement
+                auto ret = new ReturnStmt( nullptr );
+                ret->location = loc;
+                stmtsToAddBefore.push_back( ret );
+                // Create the null statement with the created label
+                auto noop = new NullStmt({ label.obj });
+                noop->location = loc;
+                // Return the null statement to take the place of the previous statement
+                return noop;
+        }
+        Statement * SuspendKeyword::make_coroutine_suspend( SuspendStmt * stmt ) {
+                if(stmt->then) SemanticError( stmt->location, "Compound statement following coroutines is not implemented.");
+                // Save the location and delete the old statement, we only need the location from this point on
+                auto loc = stmt->location;
+                delete stmt;
+                // Create the call expression
+                if(!decl_suspend) SemanticError( loc, "suspend keyword applied to coroutines requires coroutines to be in scope, add #include <coroutine.hfa>\n");
+                auto expr = new UntypedExpr( VariableExpr::functionPointer( decl_suspend ) );
+                expr->location = loc;
+                // Change this statement into a regular expr
+                assert(expr);
+                auto nstmt = new ExprStmt( expr );
+                nstmt->location = loc;
+                return nstmt;
+        }
         //=============================================================================================
 …
         void MutexKeyword::postvisit(StructDecl* decl) {
                 if( decl->name == "monitor_desc" && decl->body ) {
+                if( decl->name == "$monitor" && decl->body ) {
                         assert( !monitor_decl );
                         monitor_decl = decl;
 …
                 );
                 //monitor_desc * __monitors[] = { get_monitor(a), get_monitor(b) };
+                //$monitor * __monitors[] = { get_monitor(a), get_monitor(b) };
                 body->push_front( new DeclStmt( monitors) );
+        }
 …
                 );
                 //monitor_desc * __monitors[] = { get_monitor(a), get_monitor(b) };
+                //$monitor * __monitors[] = { get_monitor(a), get_monitor(b) };
                 body->push_front( new DeclStmt( monitors) );
+        }
 …
         //=============================================================================================
         void ThreadStarter::previsit( StructDecl * decl ) {
                 if( decl->name == "thread_desc" && decl->body ) {
+                if( decl->name == "$thread" && decl->body ) {
                         assert( !thread_decl );
                         thread_decl = decl;

src/Concurrency/Waitfor.cc

rb7d6a36	r6a490b2
244	244	decl_mask = decl;
245	245	}
246		else if( decl->name == "~~monitor_desc~~" ) {
	246	else if( decl->name == "$monitor" ) {
247	247	assert( !decl_monitor );
248	248	decl_monitor = decl;

src/ControlStruct/ExceptTranslate.cc

-              rb7d6a36
+              r6a490b2
 // Author           : Andrew Beach
 // Created On       : Wed Jun 14 16:49:00 2017
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Fri Dec 13 23:40:15 2019
 // Update Count     : 12
+// Last Modified By : Andrew Beach
+// Last Modified On : Fri Mar 27 11:58:00 2020
+// Update Count     : 13
 //
 …
                         ThrowStmt *throwStmt ) {
                 // __throw_terminate( `throwStmt->get_name()` ); }
                 return create_given_throw( "__cfaabi_ehm__throw_terminate", throwStmt );
+                return create_given_throw( "__cfaehm_throw_terminate", throwStmt );
+        }
 …
                         ) ) );
                 result->push_back( new ExprStmt(
                         new UntypedExpr( new NameExpr( "__cfaabi_ehm__rethrow_terminate" ) )
+                        new UntypedExpr( new NameExpr( "__cfaehm_rethrow_terminate" ) )
                         ) );
                 delete throwStmt;
 …
                         ThrowStmt *throwStmt ) {
                 // __throw_resume( `throwStmt->get_name` );
                 return create_given_throw( "__cfaabi_ehm__throw_resume", throwStmt );
+                return create_given_throw( "__cfaehm_throw_resume", throwStmt );
+        }
 …
                         local_except->get_attributes().push_back( new Attribute(
                                 "cleanup",
                                 { new NameExpr( "__cfaabi_ehm__cleanup_terminate" ) }
+                                { new NameExpr( "__cfaehm_cleanup_terminate" ) }
                                 ) );
 …
                         FunctionDecl * terminate_catch,
                         FunctionDecl * terminate_match ) {
                 // { __cfaabi_ehm__try_terminate(`try`, `catch`, `match`); }
+                // { __cfaehm_try_terminate(`try`, `catch`, `match`); }
                 UntypedExpr * caller = new UntypedExpr( new NameExpr(
                         "__cfaabi_ehm__try_terminate" ) );
+                        "__cfaehm_try_terminate" ) );
                 std::list<Expression *>& args = caller->get_args();
                 args.push_back( nameOf( try_wrapper ) );
 …
                 // struct __try_resume_node __resume_node
                 //      __attribute__((cleanup( __cfaabi_ehm__try_resume_cleanup )));
+                //      __attribute__((cleanup( __cfaehm_try_resume_cleanup )));
                 // ** unwinding of the stack here could cause problems **
                 // ** however I don't think that can happen currently **
                 // __cfaabi_ehm__try_resume_setup( &__resume_node, resume_handler );
+                // __cfaehm_try_resume_setup( &__resume_node, resume_handler );
                 std::list< Attribute * > attributes;
 …
                         std::list< Expression * > attr_params;
                         attr_params.push_back( new NameExpr(
                                 "__cfaabi_ehm__try_resume_cleanup" ) );
+                                "__cfaehm_try_resume_cleanup" ) );
                         attributes.push_back( new Attribute( "cleanup", attr_params ) );
+                }
 …
                 UntypedExpr *setup = new UntypedExpr( new NameExpr(
                         "__cfaabi_ehm__try_resume_setup" ) );
+                        "__cfaehm_try_resume_setup" ) );
                 setup->get_args().push_back( new AddressExpr( nameOf( obj ) ) );
                 setup->get_args().push_back( nameOf( resume_handler ) );
 …
         ObjectDecl * ExceptionMutatorCore::create_finally_hook(
                         FunctionDecl * finally_wrapper ) {
                 // struct __cfaabi_ehm__cleanup_hook __finally_hook
+                // struct __cfaehm_cleanup_hook __finally_hook
                 //      __attribute__((cleanup( finally_wrapper )));
 …
                         // Skip children?
                         return;
                 } else if ( structDecl->get_name() == "__cfaabi_ehm__base_exception_t" ) {
+                } else if ( structDecl->get_name() == "__cfaehm_base_exception_t" ) {
                         assert( nullptr == except_decl );
                         except_decl = structDecl;
                         init_func_types();
                 } else if ( structDecl->get_name() == "__cfaabi_ehm__try_resume_node" ) {
+                } else if ( structDecl->get_name() == "__cfaehm_try_resume_node" ) {
                         assert( nullptr == node_decl );
                         node_decl = structDecl;
                 } else if ( structDecl->get_name() == "__cfaabi_ehm__cleanup_hook" ) {
+                } else if ( structDecl->get_name() == "__cfaehm_cleanup_hook" ) {
                         assert( nullptr == hook_decl );
                         hook_decl = structDecl;

src/Parser/ParseNode.h

rb7d6a36	r6a490b2
428	428	Statement * build_asm( bool voltile, Expression * instruction, ExpressionNode * output = nullptr, ExpressionNode * input = nullptr, ExpressionNode * clobber = nullptr, LabelNode * gotolabels = nullptr );
429	429	Statement * build_directive( std::string * directive );
	430	SuspendStmt * build_suspend( StatementNode *, SuspendStmt::Type = SuspendStmt::None);
430	431	WaitForStmt * build_waitfor( ExpressionNode * target, StatementNode * stmt, ExpressionNode * when );
431	432	WaitForStmt * build_waitfor( ExpressionNode * target, StatementNode * stmt, ExpressionNode * when, WaitForStmt * existing );

src/Parser/StatementNode.cc

-              rb7d6a36
+              r6a490b2
 } // build_finally
+SuspendStmt * build_suspend( StatementNode * then, SuspendStmt::Type type ) {
+        auto node = new SuspendStmt();
+        node->type = type;
+        std::list< Statement * > stmts;
+        buildMoveList< Statement, StatementNode >( then, stmts );
+        if(!stmts.empty()) {
+                assert( stmts.size() == 1 );
+                node->then = dynamic_cast< CompoundStmt * >( stmts.front() );
+        }
+        return node;
+}
 WaitForStmt * build_waitfor( ExpressionNode * targetExpr, StatementNode * stmt, ExpressionNode * when ) {
         auto node = new WaitForStmt();

src/Parser/TypeData.cc

rb7d6a36	r6a490b2
769	769	case AggregateDecl::Struct:
770	770	case AggregateDecl::Coroutine:
	771	case AggregateDecl::Generator:
771	772	case AggregateDecl::Monitor:
772	773	case AggregateDecl::Thread:

src/Parser/lex.ll

-              rb7d6a36
+              r6a490b2
 #define FLOATXX(v) KEYWORD_RETURN(v);
 #else
 #define FLOATXX(v) IDENTIFIER_RETURN();
+#define FLOATXX(v) IDENTIFIER_RETURN();
 #endif // HAVE_KEYWORDS_FLOATXX
 …
 _Static_assert  { KEYWORD_RETURN(STATICASSERT); }               // C11
 struct                  { KEYWORD_RETURN(STRUCT); }
+        /* suspend                      { KEYWORD_RETURN(SUSPEND); }                    // CFA */
+suspend                 { KEYWORD_RETURN(SUSPEND); }                    // CFA
 switch                  { KEYWORD_RETURN(SWITCH); }
 thread                  { KEYWORD_RETURN(THREAD); }                             // C11

src/Parser/parser.yy

-              rb7d6a36
+              r6a490b2
 // Created On       : Sat Sep  1 20:22:55 2001
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Sun Feb 16 08:22:14 2020
 // Update Count     : 4461
+// Last Modified On : Mon Apr 27 12:25:42 2020
+// Update Count     : 4483
 //
 …
 %token OTYPE FTYPE DTYPE TTYPE TRAIT                                    // CFA
 %token SIZEOF OFFSETOF
+// %token SUSPEND RESUME                                                                        // CFA
+// %token RESUME                                                                        // CFA
+%token SUSPEND                                                                  // CFA
 %token ATTRIBUTE EXTENSION                                                              // GCC
 %token IF ELSE SWITCH CASE DEFAULT DO WHILE FOR BREAK CONTINUE GOTO RETURN
 …
         conditional_expression
         | unary_expression assignment_operator assignment_expression
+                { $$ = new ExpressionNode( build_binary_val( $2, $1, $3 ) ); }
+                {
+                        if ( $2 == OperKinds::AtAssn ) {
+                                SemanticError( yylloc, "C @= assignment is currently unimplemented." ); $$ = nullptr;
+                        } else {
+                                $$ = new ExpressionNode( build_binary_val( $2, $1, $3 ) );
+                        } // if
+                }
         | unary_expression '=' '{' initializer_list_opt comma_opt '}'
                 { SemanticError( yylloc, "Initializer assignment is currently unimplemented." ); $$ = nullptr; }
 …
 tuple_expression_list:
+        assignment_expression_opt
+        | tuple_expression_list ',' assignment_expression_opt
+        assignment_expression
+        | '@'                                                                                           // CFA
+                { SemanticError( yylloc, "Eliding tuple element with '@' is currently unimplemented." ); $$ = nullptr; }
+        | tuple_expression_list ',' assignment_expression
                 { $$ = (ExpressionNode *)($1->set_last( $3 )); }
+        | tuple_expression_list ',' '@'
+                { SemanticError( yylloc, "Eliding tuple element with '@' is currently unimplemented." ); $$ = nullptr; }
+        ;
 …
         | RETURN '{' initializer_list_opt comma_opt '}' ';'
                 { SemanticError( yylloc, "Initializer return is currently unimplemented." ); $$ = nullptr; }
+        // | SUSPEND ';'
+        //      { SemanticError( yylloc, "Suspend expression is currently unimplemented." ); $$ = nullptr; }
+        // | SUSPEND compound_statement ';'
+        //      { SemanticError( yylloc, "Suspend expression is currently unimplemented." ); $$ = nullptr; }
+        | SUSPEND ';'
+                { $$ = new StatementNode( build_suspend( nullptr ) ); }
+        | SUSPEND compound_statement
+                { $$ = new StatementNode( build_suspend( $2 ) ); }
+        | SUSPEND COROUTINE ';'
+                { $$ = new StatementNode( build_suspend( nullptr, SuspendStmt::Coroutine ) ); }
+        | SUSPEND COROUTINE compound_statement
+                { $$ = new StatementNode( build_suspend( $3, SuspendStmt::Coroutine ) ); }
+        | SUSPEND GENERATOR ';'
+                { $$ = new StatementNode( build_suspend( nullptr, SuspendStmt::Generator ) ); }
+        | SUSPEND GENERATOR compound_statement
+                { $$ = new StatementNode( build_suspend( $3, SuspendStmt::Generator ) ); }
         | THROW assignment_expression_opt ';'                           // handles rethrow
                 { $$ = new StatementNode( build_throw( $2 ) ); }
 …
                 // type_specifier can resolve to just TYPEDEFname (e.g., typedef int T; int f( T );). Therefore this must be
                 // flattened to allow lookahead to the '(' without having to reduce identifier_or_type_name.
         cfa_abstract_tuple identifier_or_type_name '(' push cfa_parameter_ellipsis_list_opt pop ')'
+        cfa_abstract_tuple identifier_or_type_name '(' push cfa_parameter_ellipsis_list_opt pop ')' attribute_list_opt
                 // To obtain LR(1 ), this rule must be factored out from function return type (see cfa_abstract_declarator).
                 { $$ = DeclarationNode::newFunction( $2, $1, $5, 0 ); }
         | cfa_function_return identifier_or_type_name '(' push cfa_parameter_ellipsis_list_opt pop ')'
                 { $$ = DeclarationNode::newFunction( $2, $1, $5, 0 ); }
+                { $$ = DeclarationNode::newFunction( $2, $1, $5, 0 )->addQualifiers( $8 ); }
+        | cfa_function_return identifier_or_type_name '(' push cfa_parameter_ellipsis_list_opt pop ')' attribute_list_opt
+                { $$ = DeclarationNode::newFunction( $2, $1, $5, 0 )->addQualifiers( $8 ); }
+        ;
 …
 aggregate_control:                                                                              // CFA
         GENERATOR
+                { yyy = true; $$ = AggregateDecl::Coroutine; }
+                { yyy = true; $$ = AggregateDecl::Generator; }
+        | MONITOR GENERATOR
+                { SemanticError( yylloc, "monitor generator is currently unimplemented." ); $$ = AggregateDecl::NoAggregate; }
         | COROUTINE
                 { yyy = true; $$ = AggregateDecl::Coroutine; }
         | MONITOR
                 { yyy = true; $$ = AggregateDecl::Monitor; }
+        | MONITOR COROUTINE
+                { SemanticError( yylloc, "monitor coroutine is currently unimplemented." ); $$ = AggregateDecl::NoAggregate; }
         | THREAD
                 { yyy = true; $$ = AggregateDecl::Thread; }
+        | MONITOR THREAD
+                { SemanticError( yylloc, "monitor thread is currently unimplemented." ); $$ = AggregateDecl::NoAggregate; }
+        ;

src/ResolvExpr/AlternativeFinder.cc

-              rb7d6a36
+              r6a490b2
                 try {
                         // Attempt 1 : turn (thread&)X into (thread_desc&)X.__thrd
+                        // Attempt 1 : turn (thread&)X into ($thread&)X.__thrd
                         // Clone is purely for memory management
                         std::unique_ptr<Expression> tech1 { new UntypedMemberExpr(new NameExpr(castExpr->concrete_target.field), castExpr->arg->clone()) };
 …
                 } catch(SemanticErrorException & ) {}
                 // Fallback : turn (thread&)X into (thread_desc&)get_thread(X)
+                // Fallback : turn (thread&)X into ($thread&)get_thread(X)
                 std::unique_ptr<Expression> fallback { UntypedExpr::createDeref( new UntypedExpr(new NameExpr(castExpr->concrete_target.getter), { castExpr->arg->clone() })) };
                 // don't prune here, since it's guaranteed all alternatives will have the same type

src/ResolvExpr/Resolver.cc

-              rb7d6a36
+              r6a490b2
 // Author           : Aaron B. Moss
 // Created On       : Sun May 17 12:17:01 2015
 // Last Modified By : Aaron B. Moss
 // Last Modified On : Wed May 29 11:00:00 2019
 // Update Count     : 241
+// Last Modified By : Andrew Beach
+// Last Modified On : Fri Mar 27 11:58:00 2020
+// Update Count     : 242
 //
 …
                 // TODO: Replace *exception type with &exception type.
                 if ( throwStmt->get_expr() ) {
                         const StructDecl * exception_decl = indexer.lookupStruct( "__cfaabi_ehm__base_exception_t" );
+                        const StructDecl * exception_decl = indexer.lookupStruct( "__cfaehm_base_exception_t" );
                         assert( exception_decl );
                         Type * exceptType = new PointerType( noQualifiers, new StructInstType( noQualifiers, const_cast<StructDecl *>(exception_decl) ) );
 …
                 if ( throwStmt->expr ) {
                         const ast::StructDecl * exceptionDecl =
                                 symtab.lookupStruct( "__cfaabi_ehm__base_exception_t" );
+                                symtab.lookupStruct( "__cfaehm_base_exception_t" );
                         assert( exceptionDecl );
                         ast::ptr< ast::Type > exceptType =

src/ResolvExpr/TypeEnvironment.cc

-              rb7d6a36
+              r6a490b2
 #include <utility>                     // for pair, move
+#include "CompilationState.h"          // for deterministic_output
 #include "Common/utility.h"            // for maybeClone
 #include "SynTree/Type.h"              // for Type, FunctionType, Type::Fora...
 …
         void EqvClass::print( std::ostream &os, Indenter indent ) const {
+                os << "( ";
+                std::copy( vars.begin(), vars.end(), std::ostream_iterator< std::string >( os, " " ) );
+                os << ")";
+                if( !deterministic_output ) {
+                        os << "( ";
+                        std::copy( vars.begin(), vars.end(), std::ostream_iterator< std::string >( os, " " ) );
+                        os << ")";
+                }
                 if ( type ) {
                         os << " -> ";
 …
                 // check safely bindable
                 if ( r.type && occursIn( r.type, s.vars.begin(), s.vars.end(), *this ) ) return false;
                 // merge classes in
                 r.vars.insert( s.vars.begin(), s.vars.end() );

src/SynTree/Declaration.h

-              rb7d6a36
+              r6a490b2
         bool is_coroutine() { return kind == Coroutine; }
+        bool is_monitor() { return kind == Monitor; }
+        bool is_thread() { return kind == Thread; }
+        bool is_generator() { return kind == Generator; }
+        bool is_monitor  () { return kind == Monitor  ; }
+        bool is_thread   () { return kind == Thread   ; }
         virtual StructDecl * clone() const override { return new StructDecl( *this ); }

src/SynTree/LinkageSpec.cc

-              rb7d6a36
+              r6a490b2
 // Author           : Rodolfo G. Esteves
 // Created On       : Sat May 16 13:22:09 2015
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Mon Dec 16 15:02:29 2019
 // Update Count     : 28
+// Last Modified By : Andrew Beach
+// Last Modified On : Mon Mar  2 16:13:00 2020
+// Update Count     : 29
 //
 …
 #include "LinkageSpec.h"
+#include "Common/CodeLocation.h"
 #include "Common/SemanticError.h"

src/SynTree/LinkageSpec.h

-              rb7d6a36
+              r6a490b2
 // Author           : Rodolfo G. Esteves
 // Created On       : Sat May 16 13:24:28 2015
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Mon Dec 16 15:03:43 2019
 // Update Count     : 20
+// Last Modified By : Andrew Beach
+// Last Modified On : Mon Mar  2 16:13:00 2020
+// Update Count     : 21
 //
 …
 #include <string>
+#include "Common/CodeLocation.h"
+struct CodeLocation;
 namespace LinkageSpec {

src/SynTree/Mutator.h

rb7d6a36	r6a490b2
51	51	virtual Statement * mutate( CatchStmt * catchStmt ) = 0;
52	52	virtual Statement * mutate( FinallyStmt * catchStmt ) = 0;
	53	virtual Statement * mutate( SuspendStmt * suspendStmt ) = 0;
53	54	virtual Statement * mutate( WaitForStmt * waitforStmt ) = 0;
54	55	virtual Declaration * mutate( WithStmt * withStmt ) = 0;

src/SynTree/Statement.cc

-              rb7d6a36
+              r6a490b2
+}
+SuspendStmt::SuspendStmt( const SuspendStmt & other )
+        : Statement( other )
+        , then( maybeClone(other.then) )
+{}
+SuspendStmt::~SuspendStmt() {
+        delete then;
+}
+void SuspendStmt::print( std::ostream & os, Indenter indent ) const {
+        os << "Suspend Statement";
+        switch (type) {
+                case None     : os << " with implicit target"; break;
+                case Generator: os << " for generator"       ; break;
+                case Coroutine: os << " for coroutine"       ; break;
+        }
+        os << endl;
+        indent += 1;
+        if(then) {
+                os << indent << " with post statement :" << endl;
+                then->print( os, indent + 1);
+        }
+}
 WaitForStmt::WaitForStmt() : Statement() {
         timeout.time      = nullptr;

src/SynTree/Statement.h

-              rb7d6a36
+              r6a490b2
 };
+class SuspendStmt : public Statement {
+  public:
+        CompoundStmt * then = nullptr;
+        enum Type { None, Coroutine, Generator } type = None;
+        SuspendStmt() = default;
+        SuspendStmt( const SuspendStmt & );
+        virtual ~SuspendStmt();
+        virtual SuspendStmt * clone() const override { return new SuspendStmt( *this ); }
+        virtual void accept( Visitor & v ) override { v.visit( this ); }
+        virtual void accept( Visitor & v ) const override { v.visit( this ); }
+        virtual Statement * acceptMutator( Mutator & m )  override { return m.mutate( this ); }
+        virtual void print( std::ostream & os, Indenter indent = {} ) const override;
+};
 class WaitForStmt : public Statement {
   public:

src/SynTree/SynTree.h

rb7d6a36	r6a490b2
54	54	class CatchStmt;
55	55	class FinallyStmt;
	56	class SuspendStmt;
56	57	class WaitForStmt;
57	58	class WithStmt;

src/SynTree/Visitor.h

rb7d6a36	r6a490b2
78	78	virtual void visit( FinallyStmt * node ) { visit( const_cast<const FinallyStmt *>(node) ); }
79	79	virtual void visit( const FinallyStmt * finallyStmt ) = 0;
	80	virtual void visit( SuspendStmt * node ) { visit( const_cast<const SuspendStmt *>(node) ); }
	81	virtual void visit( const SuspendStmt * suspendStmt ) = 0;
80	82	virtual void visit( WaitForStmt * node ) { visit( const_cast<const WaitForStmt *>(node) ); }
81	83	virtual void visit( const WaitForStmt * waitforStmt ) = 0;

src/main.cc

-              rb7d6a36
+              r6a490b2
 static const char optstring[] = ":c:ghlLmNnpP:S:twW:D:";
+static const char optstring[] = ":c:ghlLmNnpdP:S:twW:D:";
 enum { PreludeDir = 128 };
 …
         { "no-prelude", no_argument, nullptr, 'n' },
         { "prototypes", no_argument, nullptr, 'p' },
+        { "deterministic-out", no_argument, nullptr, 'd' },
         { "print", required_argument, nullptr, 'P' },
         { "prelude-dir", required_argument, nullptr, PreludeDir },
 …
         "do not read prelude",                                // -n
         "generate prototypes for prelude functions",            // -p
+        "don't print output that isn't deterministic",        // -d
         "print",                                              // -P
         "<directory> prelude directory for debug/nodebug",      // no flag
 …
                         genproto = true;
                         break;
+                  case 'd':                                     // don't print non-deterministic output
+                    deterministic_output = true;
+                        break;
                   case 'P':                                                                             // print options
                         for ( int i = 0;; i += 1 ) {

tests/.expect/alloc-ERROR.txt

-              rb7d6a36
+              r6a490b2
 alloc.cfa:310:1 error: No reasonable alternatives for expression Applying untyped:
+alloc.cfa:362:1 error: No reasonable alternatives for expression Applying untyped:
   Name: ?=?
 ...to:
   Name: p
+  Name: ip
   Applying untyped:
     Name: realloc
 …
 alloc.cfa:311:1 error: No reasonable alternatives for expression Applying untyped:
+alloc.cfa:363:1 error: No reasonable alternatives for expression Applying untyped:
   Name: ?=?
 ...to:
+  Name: p
+  Applying untyped:
+    Name: alloc
+  ...to:
+    Name: stp
+    Applying untyped:
+      Name: ?*?
+    ...to:
+      Name: dim
+      Sizeof Expression on: Applying untyped:
+          Name: *?
+        ...to:
+          Name: stp
+alloc.cfa:312:1 error: No reasonable alternatives for expression Applying untyped:
+  Name: ?=?
+...to:
+  Name: p
+  Name: ip
   Applying untyped:
     Name: memset
 …
 alloc.cfa:313:1 error: No reasonable alternatives for expression Applying untyped:
+alloc.cfa:364:1 error: No reasonable alternatives for expression Applying untyped:
   Name: ?=?
 ...to:
   Name: p
+  Name: ip
   Applying untyped:
     Name: memcpy

tests/.expect/alloc.txt

-              rb7d6a36
+              r6a490b2
 xefefefef 0xefefefef 0xefefefef 0xefefefef 0xefefefef 0xefefefef 0xefefefef 0xefefefef 0xefefefef 0xefefefef 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101
 CFA resize array alloc
+CFA realloc array alloc
 xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef
 CFA resize array alloc
+CFA realloc array alloc
 xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101
 CFA resize array alloc
+CFA realloc array alloc
 xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef
 CFA resize array alloc
+CFA realloc array alloc, fill
 xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede
 CFA resize array alloc
+CFA realloc array alloc, fill
 xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef
 CFA resize array alloc, fill
+CFA realloc array alloc, fill
 xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0xdeadbeef 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0x1010101 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede 0xdededede

tests/Makefile.am

rb7d6a36	r6a490b2
41	41	-quiet @CFA_FLAGS@ \
42	42	-DIN_DIR="${abs_srcdir}/.in/"
	43
	44	AM_CFAFLAGS = -XCFA --deterministic-out
43	45
44	46	# get the desired cfa to test

tests/Makefile.in

rb7d6a36	r6a490b2
408	408	-DIN_DIR="${abs_srcdir}/.in/"
409	409
	410	AM_CFAFLAGS = -XCFA --deterministic-out
410	411
411	412	# get the desired cfa to test

tests/alloc.cfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Wed Feb  3 07:56:22 2016
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Sun Feb 16 09:21:13 2020
 // Update Count     : 405
+// Last Modified On : Mon Apr  6 21:08:23 2020
+// Update Count     : 428
 //
 …
         size_t dim = 10;
         char fill = '\xde';
         int * p, * p1;
+        int * ip, * ip1;
         // allocation, non-array types
         p = (int *)malloc( sizeof(*p) );                                        // C malloc, type unsafe
         *p = 0xdeadbeef;
         printf( "C   malloc %#x\n", *p );
         free( p );
         p = malloc();                                       // CFA malloc, type safe
         *p = 0xdeadbeef;
         printf( "CFA malloc %#x\n", *p );
         free( p );
         p = alloc();                                        // CFA alloc, type safe
         *p = 0xdeadbeef;
         printf( "CFA alloc %#x\n", *p );
         free( p );
         p = alloc_set( fill );                                                          // CFA alloc, fill
         printf( "CFA alloc, fill %08x\n", *p );
         free( p );
         p = alloc_set( 3 );                                                                     // CFA alloc, fill
         printf( "CFA alloc, fill %d\n", *p );
         free( p );
+        ip = (int *)malloc( sizeof(*ip) );                                      // C malloc, type unsafe
+        *ip = 0xdeadbeef;
+        printf( "C   malloc %#x\n", *ip );
+        free( ip );
+        ip = malloc();                                                                          // CFA malloc, type safe
+        *ip = 0xdeadbeef;
+        printf( "CFA malloc %#x\n", *ip );
+        free( ip );
+        ip = alloc();                                                                           // CFA alloc, type safe
+        *ip = 0xdeadbeef;
+        printf( "CFA alloc %#x\n", *ip );
+        free( ip );
+        ip = alloc_set( fill );                                                         // CFA alloc, fill
+        printf( "CFA alloc, fill %08x\n", *ip );
+        free( ip );
+        ip = alloc_set( 3 );                                                            // CFA alloc, fill
+        printf( "CFA alloc, fill %d\n", *ip );
+        free( ip );
 …
         printf( "\n" );
         p = (int *)calloc( dim, sizeof( *p ) );                         // C array calloc, type unsafe
+        ip = (int *)calloc( dim, sizeof( *ip ) );                       // C array calloc, type unsafe
         printf( "C   array calloc, fill 0\n" );
         for ( i; dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
         free( p );
         p = calloc( dim );                                  // CFA array calloc, type safe
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        free( ip );
+        ip = calloc( dim );                                                                     // CFA array calloc, type safe
         printf( "CFA array calloc, fill 0\n" );
         for ( i; dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
         free( p );
         p = alloc( dim );                                   // CFA array alloc, type safe
         for ( i; dim ) { p[i] = 0xdeadbeef; }
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        free( ip );
+        ip = alloc( dim );                                                                      // CFA array alloc, type safe
+        for ( i; dim ) { ip[i] = 0xdeadbeef; }
         printf( "CFA array alloc, no fill\n" );
         for ( i; dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
         free( p );
         p = alloc_set( 2 * dim, fill );                                         // CFA array alloc, fill
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        free( ip );
+        ip = alloc_set( 2 * dim, fill );                                        // CFA array alloc, fill
         printf( "CFA array alloc, fill %#hhx\n", fill );
         for ( i; 2 * dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
         free( p );
         p = alloc_set( 2 * dim, 0xdeadbeef );                           // CFA array alloc, fill
+        for ( i; 2 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        free( ip );
+        ip = alloc_set( 2 * dim, 0xdeadbeef );                          // CFA array alloc, fill
         printf( "CFA array alloc, fill %#hhx\n", 0xdeadbeef );
         for ( i; 2 * dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
         // do not free
         p1 = alloc_set( 2 * dim, p );                                           // CFA array alloc, fill
+        for ( i; 2 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip1 = alloc_set( 2 * dim, ip );                                         // CFA array alloc, fill
         printf( "CFA array alloc, fill from array\n" );
+        for ( i; 2 * dim ) { printf( "%#x %#x, ", p[i], p1[i] ); }
+        free( p1 );
+        printf( "\n" );
+        for ( i; 2 * dim ) { printf( "%#x %#x, ", ip[i], ip1[i] ); }
+        free( ip1 );
+        printf( "\n" );
+        // realloc, non-array types
+        printf( "\n" );
+        ip = (int *)realloc( ip, dim * sizeof(*ip) );           // C realloc
+        printf( "C realloc\n" );
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = realloc( ip, 2 * dim * sizeof(*ip) );                      // CFA realloc
+        for ( i; dim ~ 2 * dim ) { ip[i] = 0x1010101; }
+        printf( "CFA realloc\n" );
+        for ( i; 2 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        // realloc, array types
+        printf( "\n" );
+        ip = alloc( ip, dim );                                                          // CFA realloc array alloc
+        for ( i; dim ) { ip[i] = 0xdeadbeef; }
+        printf( "CFA realloc array alloc\n" );
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc( ip, 2 * dim );                                                      // CFA realloc array alloc
+        for ( i; dim ~ 2 * dim ) { ip[i] = 0x1010101; }         // fill upper part
+        printf( "CFA realloc array alloc\n" );
+        for ( i; 2 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc( ip, dim );                                                          // CFA realloc array alloc
+        printf( "CFA realloc array alloc\n" );
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc_set( ip, 3 * dim, fill );                            // CFA realloc array alloc, fill
+        printf( "CFA realloc array alloc, fill\n" );
+        for ( i; 3 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc_set( ip, dim, fill );                                        // CFA realloc array alloc, fill
+        printf( "CFA realloc array alloc, fill\n" );
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc_set( ip, 3 * dim, fill );                            // CFA realloc array alloc, fill
+        printf( "CFA realloc array alloc, fill\n" );
+        for ( i; 3 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+#if 0 // FIX ME
+        ip = alloc_set( ip, 5 * dim, 5 );                                       // CFA realloc array alloc, 5
+        printf( "CFA realloc array alloc, 5\n" );
+        for ( i; 5 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc_set( ip, dim, 5 );                                           // CFA realloc array alloc, 5
+        printf( "CFA realloc array alloc, 5\n" );
+        for ( i; dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+        // do not free
+        ip = alloc_set( ip, 5 * dim, 5 );                                       // CFA realloc array alloc, 5
+        printf( "CFA realloc array alloc, 5\n" );
+        for ( i; 5 * dim ) { printf( "%#x ", ip[i] ); }
+        printf( "\n" );
+#endif // 0
+        free( ip );
         // resize, non-array types
+        printf( "\n" );
         p = (int *)realloc( p, dim * sizeof(*p) );                      // C realloc
         printf( "C realloc\n" );
+        for ( i; dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
         // do not free
         p = realloc( p, 2 * dim * sizeof(*p) );             // CFA realloc
+        for ( i; dim ~ 2 * dim ) { p[i] = 0x1010101; }
         printf( "CFA realloc\n" );
+        for ( i; 2 * dim ) { printf( "%#x ", p[i] ); }
         printf( "\n" );
+        // do not free
+        struct S {
+                int a[5];
+        };
+    ip = alloc();
+        *ip = 5;
+    double * dp = alloc( ip );
+        *dp = 5.5;
+    S * sp = alloc( dp );
+        *sp = (S){ {0, 1, 2, 3, 4} };
+    ip = alloc( sp );
+        *ip = 3;
+    free( ip );
         // resize, array types
+        printf( "\n" );
+        p = alloc( p, dim );                                // CFA resize array alloc
+        for ( i; dim ) { p[i] = 0xdeadbeef; }
+        printf( "CFA resize array alloc\n" );
+        for ( i; dim ) { printf( "%#x ", p[i] ); }
+        printf( "\n" );
+        // do not free
+        p = alloc( p, 2 * dim );                            // CFA resize array alloc
+        for ( i; dim ~ 2 * dim ) { p[i] = 0x1010101; }          // fill upper part
+        printf( "CFA resize array alloc\n" );
+        for ( i; 2 * dim ) { printf( "%#x ", p[i] ); }
+        printf( "\n" );
+        // do not free
+        p = alloc( p, dim );                                // CFA resize array alloc
+        printf( "CFA resize array alloc\n" );
+        for ( i; dim ) { printf( "%#x ", p[i] ); }
+        printf( "\n" );
+        // do not free
+        p = alloc_set( p, 3 * dim, fill );                                      // CFA resize array alloc, fill
+        printf( "CFA resize array alloc\n" );
+        for ( i; 3 * dim ) { printf( "%#x ", p[i] ); }
+        printf( "\n" );
+        // do not free
+        p = alloc_set( p, dim, fill );                                          // CFA resize array alloc, fill
+        printf( "CFA resize array alloc\n" );
+        for ( i; dim ) { printf( "%#x ", p[i] ); }
+        printf( "\n" );
+        // do not free
+        p = alloc_set( p, 3 * dim, fill );                                      // CFA resize array alloc, fill
+        printf( "CFA resize array alloc, fill\n" );
+        for ( i; 3 * dim ) { printf( "%#x ", p[i] );; }
+        printf( "\n" );
+        free( p );
+    ip = alloc( 5 );
+        for ( i; 5 ) { ip[i] = 5; }
+    dp = alloc( ip, 5 );
+        for ( i; 5 ) { dp[i] = 5.5; }
+    sp = alloc( dp, 5 );
+        for ( i; 5 ) { sp[i] = (S){ {0, 1, 2, 3, 4} }; }
+    ip = alloc( sp, 3 );
+        for ( i; 3 ) { ip[i] = 3; }
+    ip = alloc( ip, 7 );
+        for ( i; 7 ) { ip[i] = 7; }
+    ip = alloc( ip, 7, false );
+        for ( i; 7 ) { ip[i] = 7; }
+    free( ip );
 …
         free( stp );
         stp = &(*memalign( Alignment )){ 42, 42.5 };          // CFA memalign
+        stp = &(*memalign( Alignment )){ 42, 42.5 };            // CFA memalign
         assert( (uintptr_t)stp % Alignment == 0 );
         printf( "CFA memalign %d %g\n", stp->x, stp->y );
 …
         free( fp - 1 );
         p = foo( bar( baz( malloc(), 0 ), 0 ), 0 );
         *p = 0xdeadbeef;
         printf( "CFA deep malloc %#x\n", *p );
         free( p );
+        ip = foo( bar( baz( malloc(), 0 ), 0 ), 0 );
+        *ip = 0xdeadbeef;
+        printf( "CFA deep malloc %#x\n", *ip );
+        free( ip );
 #ifdef ERR1
         stp = malloc();
         printf( "\nSHOULD FAIL\n" );
+        p = realloc( stp, dim * sizeof( *stp ) );
+        p = alloc( stp, dim * sizeof( *stp ) );
+        p = memset( stp, 10 );
+        p = memcpy( &st1, &st );
+        ip = realloc( stp, dim * sizeof( *stp ) );
+        ip = memset( stp, 10 );
+        ip = memcpy( &st1, &st );
 #endif
 } // main

tests/concurrent/.expect/monitor.txt

rb7d6a36	r6a490b2
1		4000000
	1	3000000

tests/concurrent/coroutineYield.cfa

rb7d6a36	r6a490b2
33	33	sout \| "Coroutine 2";
34	34	#endif
35		suspend();
	35	suspend;
36	36	}
37	37	}

tests/concurrent/monitor.cfa

rb7d6a36	r6a490b2
29	29
30	30	void main( MyThread & this ) {
31		for(int i = 0; i < ~~1_00~~0_000; i++) {
	31	for(int i = 0; i < 750_000; i++) {
32	32	increment( global );
33	33	}

tests/concurrent/multi-monitor.cfa

-              rb7d6a36
+              r6a490b2
 void increment( monitor_t & mutex p1, monitor_t & mutex p2, int & value ) {
+        assert(active_thread() == get_monitor(p1)->owner);
+        assert(active_thread() == get_monitor(p2)->owner);
         value += 1;
+        assert(active_thread() == get_monitor(p1)->owner);
+        assert(active_thread() == get_monitor(p2)->owner);
+}

tests/concurrent/signal/block.cfa

-              rb7d6a36
+              r6a490b2
 monitor global_data_t {
         thread_desc * last_thread;
         thread_desc * last_signaller;
+        $thread * last_thread;
+        $thread * last_signaller;
 };
 …
         if( !is_empty( cond ) ) {
                 thread_desc * next = front( cond );
+                $thread * next = front( cond );
                 if( ! signal_block( cond ) ) {

tests/concurrent/suspend_then.cfa

-              rb7d6a36
+              r6a490b2
 #include <fstream.hfa>
 #include <kernel.hfa>
-#include <monitor.hfa>
 #include <thread.hfa>
 #include <time.hfa>
 …
 #include "long_tests.hfa"
-#ifndef PREEMPTION_RATE
-#define PREEMPTION_RATE 10`ms
-#endif
 Duration default_preemption() {
         return PREEMPTION_RATE;
+        return 0;
+}
 …
 #if !defined(TEST_FOREVER)
         static inline void print(const char * const text ) {
                 write( STDERR_FILENO, text, strlen(text) );
+                write( STDOUT_FILENO, text, strlen(text) );
+        }
 #else
 …
 #endif
 coroutine Coroutine {};
+generator Coroutine { int i; };
 volatile bool done = false;
 …
 void main(Coroutine& this) {
+        suspend();
+        for(int i = 0; TEST(i < N); i++) {
+        this.i = 0;
+        suspend;
+        for(;TEST(this.i < N); this.i++) {
                 print("C - Suspending");
                 void publish() {
                         print("C - Publishing");
+                print("C - Suspending\n");
+                suspend{
+                        print("C - Publishing\n");
                         assert(!the_cor);
                         store( this );
+                }
-                suspend_then(publish);
                 assert(!the_cor);
                 print("Coroutine 2");
+                print("C - Back\n");
                 KICK_WATCHDOG;
                 yield();
+        }
         done = true;
         suspend();
+        suspend;
+}
 …
                 if(!mine) continue;
                 print("T - took");
+                print("T - took\n");
                 resume(*mine);
-                print("T - back");
+        }
+}

tests/coroutine/.expect/fmtLines.txt

-              rb7d6a36
+              r6a490b2
 {                                                         // f  or n  ewli
 ne c  hara  cter  s                                     su
 spen  d();                                      if   ( fm
 t.ch   !=   '\n'   ) b  reak
 ;               /  / ig  nore   new  line
                                   } //   for                              sout
  | f  mt.c  h;                                                  //
 prin  t ch  arac  ter                   }
 // f  or                        sou  t |   "  "
 ;                                                               //   prin  t bl
 ock   sepa  rato  r             }   //
 for             sou  t |   nl;
                                   // p  rint   gro  up s
 epar  ator      } /  / fo  r} /
 / ma  invo  id p  rt(   Form
 at &   fmt  , ch  ar c  h )
 {      fmt  .ch   = ch  ;
  res  ume(   fmt   );}   //
 prti  nt m  ain(  ) {     Form
 at f  mt;         char   ch;    for
  ( ;  ; )   {           s  in |   ch;
                                                                                 //   rea  d on
 e ch  arac  ter     if   ( e
 of(   sin   ) )   brea  k;
                                         //   eof   ?            p  rt(
 fmt,   ch   );  }   //   for}
  //   main  // L  ocal   Var
 iabl  es:   ////   tab  -wid
 th:   4 //  // c  ompi  le-c
 omma  nd:   "cfa   fmt  Line
 s.cf  a" /  ///   End:   //
+spen  d;                                        i  f (   fmt.
+ch !  = '\  n' )   bre  ak;
+        //   igno  re n  ewli  ne
+                }   // f  or                            so  ut |
+ fmt  .ch;                                                      /  / pr
+int   char  acte  r                       } //
+ for                    s  out   | "    ";
+                                                        /  / pr  int   bloc
+k se  para  tor         } /  / fo
+r               s  out   | nl  ;
+                //   pri  nt g  roup   sep
+arat  or        }   //   for}   //
+main  void   prt  ( Fo  rmat
+ & f  mt,   char   ch   ) {
+   f  mt.c  h =   ch;      r
+esum  e( f  mt )  ;} /  / pr
+tint   mai  n()   {     Fo  rmat
+ fmt  ; ch  ar c  h;    f  or (
+ ;;   ) {               sin   | c  h;
+                                                                  // r  ead   one
+char  acte  r       if (   eof
+( si  n )   ) br  eak;
+                        /  / eo  f ?            prt  ( fm
+t, c  h );      } /  / fo  r} /
+/ ma  in//   Loc  al V  aria
+bles  : //  // t  ab-w  idth
+: 4   ////   com  pile  -com
+mand  : "c  fa f  mtLi  nes.
+cfa"   ///  / En  d: /  /

tests/coroutine/.in/fmtLines.txt

rb7d6a36	r6a490b2
35	35	for ( fmt.b = 0; fmt.b < 4; fmt.b += 1 ) { // blocks of 4 characters
36	36	for ( ;; ) { // for newline characters
37		suspend();
	37	suspend;
38	38	if ( fmt.ch != '\n' ) break; // ignore newline
39	39	} // for

tests/coroutine/cntparens.cfa

-              rb7d6a36
+              r6a490b2
 //
+//
 // Cforall Version 1.0.0 Copyright (C) 2017 University of Waterloo
 //
 // The contents of this file are covered under the licence agreement in the
 // file "LICENCE" distributed with Cforall.
 //
+//
 // cntparens.cfa -- match left/right parenthesis
 //
+//
 // Author           : Peter A. Buhr
 // Created On       : Sat Apr 20 11:04:45 2019
 …
 // Last Modified On : Sat Apr 20 11:06:21 2019
 // Update Count     : 1
 //
+//
 #include <fstream.hfa>
 …
 void main( CntParens & cpns ) with( cpns ) {
         for ( ; ch == '('; cnt += 1 ) {                                         // left parenthesis
                 suspend();
+                suspend;
+        }
         for ( ; ch == ')' && cnt > 1; cnt -= 1 ) {                      // right parenthesis
                 suspend();
+                suspend;
+        }
         status = ch == ')' ? Match : Error;
 } // main
 void ?{}( CntParens & cpns ) with( cpns ) { status = Cont; cnt = 0; }

tests/coroutine/devicedriver.cfa

-              rb7d6a36
+              r6a490b2
 //
+//
 // Cforall Version 1.0.0 Copyright (C) 2017 University of Waterloo
 //
 // The contents of this file are covered under the licence agreement in the
 // file "LICENCE" distributed with Cforall.
 //
 // devicedriver.cfa --
 //
+//
+// devicedriver.cfa --
+//
 // Author           : Peter A. Buhr
 // Created On       : Sat Mar 16 15:30:34 2019
 …
 // Last Modified On : Sat Apr 20 09:07:19 2019
 // Update Count     : 90
 //
+//
 #include <fstream.hfa>
 …
 void checkCRC( Driver & d, unsigned int sum ) with( d ) {
         suspend();
+        suspend;
         unsigned short int crc = byte << 8;                                     // sign extension over written
         suspend();
+        suspend;
         // prevent sign extension for signed char
         status = (crc | (unsigned char)byte) == sum ? MSG : ECRC;
 …
                 status = CONT;
                 unsigned int lnth = 0, sum = 0;
                 while ( byte != STX ) suspend();
+                while ( byte != STX ) suspend;
           emsg: for () {
                         suspend();
+                        suspend;
                         choose ( byte ) {                                                       // process byte
                           case STX:
                                 status = ESTX; suspend(); continue msg;
+                                status = ESTX; suspend; continue msg;
                           case ETX:
                                 break emsg;
                           case ESC:
                                 suspend();
+                                suspend;
                         } // choose
                         if ( lnth >= MaxMsg ) {                                         // buffer full ?
                                 status = ELNTH; suspend(); continue msg;
+                                status = ELNTH; suspend; continue msg;
                         } // if
                         msg[lnth++] = byte;
 …
                 msg[lnth] = '\0';                                                               // terminate string
                 checkCRC( d, sum );                                                             // refactor CRC check
                 suspend();
+                suspend;
         } // for
 } // main

tests/coroutine/fibonacci.cfa

-              rb7d6a36
+              r6a490b2
         int fn1, fn2;                                                                           // retained between resumes
         fn = 0;  fn1 = fn;                                                                      // 1st case
         suspend();                                                                                      // restart last resume
+        suspend;                                                                                        // restart last resume
         fn = 1;  fn2 = fn1;  fn1 = fn;                                          // 2nd case
         suspend();                                                                                      // restart last resume
+        suspend;                                                                                        // restart last resume
         for () {
                 fn = fn1 + fn2;  fn2 = fn1;  fn1 = fn;                  // general case
                 suspend();                                                                              // restart last resume
+                suspend;                                                                                // restart last resume
         } // for
+}

tests/coroutine/fibonacci_1.cfa

-              rb7d6a36
+              r6a490b2
 // Last Modified On : Thu Mar 21 08:10:45 2019
 // Update Count     : 25
 //
+//
 #include <fstream.hfa>
 …
         [fn1, fn] = [0, 1];                                                                     // precompute first two states
         for () {
                 suspend();                                                                              // restart last resume
+                suspend;                                                                                // restart last resume
                 [fn1, fn] = [fn, fn1 + fn];                                             // general case
         } // for

tests/coroutine/fmtLines.cfa

rb7d6a36	r6a490b2
27	27	for ( b = 0; b < 4; b += 1 ) { // blocks of 4 characters
28	28	for () { // for newline characters
29		suspend();
	29	suspend;
30	30	if ( ch != '\n' ) break; // ignore newline
31	31	} // for

tests/coroutine/raii.cfa

rb7d6a36	r6a490b2
39	39	Raii raii = { "Coroutine" };
40	40	sout \| "Before Suspend";
41		suspend();
	41	suspend;
42	42	sout \| "After Suspend";
43	43	}

tests/coroutine/runningTotal.cfa

rb7d6a36	r6a490b2
25	25	void update( RunTotal & rntl, int input ) with( rntl ) { // helper
26	26	total += input; // remember between activations
27		suspend(); // inactivate on stack
	27	suspend; // inactivate on stack
28	28	}
29	29

tests/coroutine/suspend_then.cfa

-              rb7d6a36
+              r6a490b2
 #include <fstream.hfa>
-#include <coroutine.hfa>
+void then() {
+        sout | "Then!";
+}
+coroutine Fibonacci { int fn; };                                                // used for communication
+generator Fibonacci {
+        int fn;                                                                         // used for communication
+        int fn1, fn2;                                                           // retained between resumes
+};
 void main( Fibonacci & fib ) with( fib ) {                              // called on first resume
-        int fn1, fn2;                                                           // retained between resumes
         fn = 0;  fn1 = fn;                                                      // 1st case
         suspend_then(then);                                                     // restart last resume
+        suspend { sout | "Then!"; }                                             // restart last resume
         fn = 1;  fn2 = fn1;  fn1 = fn;                                  // 2nd case
         suspend_then(then);                                                     // restart last resume
+        suspend { sout | "Then!"; }                                             // restart last resume
         for () {
                 fn = fn1 + fn2;  fn2 = fn1;  fn1 = fn;                  // general case
                 suspend_then(then);                                             // restart last resume
+                suspend { sout | "Then!"; }                                     // restart last resume
         } // for
+}

tests/errors/.expect/completeType.txt

-              rb7d6a36
+              r6a490b2
     void
+  )
   Environment:( _83_4_DT ) -> instance of struct A with body 0 (no widening)
+  Environment: -> instance of struct A with body 0 (no widening)
 …
     void
+  )
   Environment:( _83_4_DT ) -> instance of struct B with body 1 (no widening)
+  Environment: -> instance of struct B with body 1 (no widening)
 …
           void
+        )
         Environment:( _102_0_T ) -> instance of type T (not function type) (no widening)
+        Environment: -> instance of type T (not function type) (no widening)
       Could not satisfy assertion:
 ?=?: pointer to function
         ... with parameters
           reference to instance of type _102_0_T (not function type)
           instance of type _102_0_T (not function type)
+          reference to instance of type _104_0_T (not function type)
+          instance of type _104_0_T (not function type)
         ... returning
           _retval__operator_assign: instance of type _102_0_T (not function type)
+          _retval__operator_assign: instance of type _104_0_T (not function type)
           ... with attributes:
             Attribute with name: unused

tests/manipulatorsOutput1.cfa

-              rb7d6a36
+              r6a490b2
 // Created On       : Sat Jun  8 18:04:11 2019
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Mon Jun 10 12:37:28 2019
 // Update Count     : 8
+// Last Modified On : Fri May  1 11:51:44 2020
+// Update Count     : 9
 //
 …
         signed char sc = -12;
         printf( "%hhd %2hhd %5.2hhd %-5.2hhd %hho %#hho %hhx %#hhx %#8hhx %#8.10hhx %#8.3hhX %+-8.3hhd %08hhd\n", sc, sc, sc, sc, sc, sc, sc, sc, sc, sc, sc, sc, sc );
+        sout | sc | wd(2,sc) | wd(5,2,sc) | left(wd(5,2,sc)) | nobase(oct(sc)) | oct(sc) | nobase(hex(sc)) | hex(sc) | wd(8,hex(sc)) | wd(8,10,hex(sc)) | upcase(wd(8,3,hex(sc))) | left(sign(upcase(wd(8,3,sc)))) | pad0(wd(8,sc));
+        sout | sc | wd(2,sc) | wd(5,2,sc) | left(wd(5,2,sc)) | nobase(oct(sc)) | oct(sc) | nonl;
+        sout | nobase(hex(sc)) | hex(sc) | wd(8,hex(sc)) | wd(8,10,hex(sc)) | upcase(wd(8,3,hex(sc))) | nonl;
+        sout | left(sign(upcase(wd(8,3,sc)))) | pad0(wd(8,sc));
         sout | "unsigned char";
         unsigned char usc = 12;
         printf( "%hhu %2hhu %5.2hhu %-5.2hhu %hho %#hho %hhx %#hhx %#8hhx %#8.10hhx %#8.3hhX %-8.3hhu %08hhu\n", usc, usc, usc, usc, usc, usc, usc, usc, usc, usc, usc, usc, usc );
+        sout | usc | wd(2,usc) | wd(5,2,usc) | left(wd(5,2,usc)) | nobase(oct(usc)) | oct(usc) | nobase(hex(usc)) | hex(usc) | wd(8,hex(usc)) | wd(8,10,hex(usc)) | upcase(wd(8,3,hex(usc))) | left(upcase(wd(8,3,usc))) | pad0(wd(8,usc));
+        sout | usc | wd(2,usc) | wd(5,2,usc) | left(wd(5,2,usc)) | nobase(oct(usc)) | oct(usc) | nonl;
+        sout | nobase(hex(usc)) | hex(usc) | wd(8,hex(usc)) | wd(8,10,hex(usc)) | upcase(wd(8,3,hex(usc))) | nonl;
+        sout | left(upcase(wd(8,3,usc))) | pad0(wd(8,usc));
         sout | "signed short int";
         signed short int si = -12;
         printf( "%hd %2hd %5.2hd %-5.2hd %ho %#ho %hx %#hx %#8hx %#8.10hx %#8.3hX %+-8.3hd %08hd\n", si, si, si, si, si, si, si, si, si, si, si, si, si );
+        sout | si | wd(2,si) | wd(5,2,si) | left(wd(5,2,si)) | nobase(oct(si)) | oct(si) | nobase(hex(si)) | hex(si) | wd(8,hex(si)) | wd(8,10,hex(si)) | upcase(wd(8,3,hex(si))) | left(sign(upcase(wd(8,3,si)))) | pad0(wd(8,si));
+        sout | si | wd(2,si) | wd(5,2,si) | left(wd(5,2,si)) | nobase(oct(si)) | oct(si) | nonl;
+        sout | nobase(hex(si)) | hex(si) | wd(8,hex(si)) | wd(8,10,hex(si)) | upcase(wd(8,3,hex(si))) | nonl;
+        sout | left(sign(upcase(wd(8,3,si)))) | pad0(wd(8,si));
         sout | "unsigned short int";
         unsigned short int usi = 12;
         printf( "%hu %2hu %5.2hu %-5.2hu %ho %#ho %hx %#hx %#8hx %#8.10hx %#8.3hX %-8.3hu %08hu\n", usi, usi, usi, usi, usi, usi, usi, usi, usi, usi, usi, usi, usi );
+        sout | usi | wd(2,usi) | wd(5,2,usi) | left(wd(5,2,usi)) | nobase(oct(usi)) | oct(usi) | nobase(hex(usi)) | hex(usi) | wd(8,hex(usi)) | wd(8,10,hex(usi)) | upcase(wd(8,3,hex(usi))) | left(upcase(wd(8,3,usi))) | pad0(wd(8,usi));
+        sout | usi | wd(2,usi) | wd(5,2,usi) | left(wd(5,2,usi)) | nobase(oct(usi)) | oct(usi) | nonl;
+        sout | nobase(hex(usi)) | hex(usi) | wd(8,hex(usi)) | wd(8,10,hex(usi)) | upcase(wd(8,3,hex(usi))) | nonl;
+        sout | left(upcase(wd(8,3,usi))) | pad0(wd(8,usi));
         sout | "signed int";
         signed int i = -12;
         printf( "%d %2d %5.2d %-5.2d %o %#o %x %#x %#8x %#8.10x %#8.3X %+-8.3d %08d\n", i, i, i, i, i, i, i, i, i, i, i, i, i );
+        sout | i | wd(2,i) | wd(5,2,i) | left(wd(5,2,i)) | nobase(oct(i)) | oct(i) | nobase(hex(i)) | hex(i) | wd(8,hex(i)) | wd(8,10,hex(i)) | upcase(wd(8,3,hex(i))) | left(sign(upcase(wd(8,3,i)))) | pad0(wd(8,i));
+        sout | i | wd(2,i) | wd(5,2,i) | left(wd(5,2,i)) | nobase(oct(i)) | oct(i) | nonl;
+        sout | nobase(hex(i)) | hex(i) | wd(8,hex(i)) | wd(8,10,hex(i)) | upcase(wd(8,3,hex(i))) | nonl;
+        sout | left(sign(upcase(wd(8,3,i)))) | pad0(wd(8,i));
         sout | "unsigned int";
         unsigned int ui = 12;
         printf( "%u %2u %5.2u %-5.2u %o %#o %x %#x %#8x %#8.10x %#8.3X %-8.3u %08u\n", ui, ui, ui, ui, ui, ui, ui, ui, ui, ui, ui, ui, ui );
+        sout | ui | wd(2,ui) | wd(5,2,ui) | left(wd(5,2,ui)) | nobase(oct(ui)) | oct(ui) | nobase(hex(ui)) | hex(ui) | wd(8,hex(ui)) | wd(8,10,hex(ui)) | upcase(wd(8,3,hex(ui))) | left(upcase(wd(8,3,ui))) | pad0(wd(8,ui));
+        sout | ui | wd(2,ui) | wd(5,2,ui) | left(wd(5,2,ui)) | nobase(oct(ui)) | oct(ui) | nonl;
+        sout | nobase(hex(ui)) | hex(ui) | wd(8,hex(ui)) | wd(8,10,hex(ui)) | upcase(wd(8,3,hex(ui))) | nonl;
+        sout | left(upcase(wd(8,3,ui))) | pad0(wd(8,ui));
         sout | "signed long long int";
         signed long long int lli = -12;
         printf( "%lld %2lld %5.2lld %-5.2lld %llo %#llo %llx %#llx %#8llx %#8.10llx %#8.3llX %+-8.3lld %08lld\n", lli, lli, lli, lli, lli, lli, lli, lli, lli, lli, lli, lli, lli );
+        sout | lli | wd(2,lli) | wd(5,2,lli) | left(wd(5,2,lli)) | nobase(oct(lli)) | oct(lli) | nobase(hex(lli)) | hex(lli) | wd(8,hex(lli)) | wd(8,10,hex(lli)) | upcase(wd(8,3,hex(lli))) | left(sign(upcase(wd(8,3,lli)))) | pad0(wd(8,lli));
+        sout | lli | wd(2,lli) | wd(5,2,lli) | left(wd(5,2,lli)) | nobase(oct(lli)) | oct(lli) | nonl;
+        sout | nobase(hex(lli)) | hex(lli) | wd(8,hex(lli)) | wd(8,10,hex(lli)) | upcase(wd(8,3,hex(lli))) | nonl;
+        sout | left(sign(upcase(wd(8,3,lli)))) | pad0(wd(8,lli));
         sout | "unsigned long long int";
         unsigned long long int ulli = 12;
         printf( "%llu %2llu %5.2llu %-5.2llu %llo %#llo %llx %#llx %#8llx %#8.10llx %#8.3llX %-8.3llu %08llu\n", ulli, ulli, ulli, ulli, ulli, ulli, ulli, ulli, ulli, ulli, ulli, ulli, ulli );
+        sout | ulli | wd(2,ulli) | wd(5,2,ulli) | left(wd(5,2,ulli)) | nobase(oct(ulli)) | oct(ulli) | nobase(hex(ulli)) | hex(ulli) | wd(8,hex(ulli)) | wd(8,10,hex(ulli)) | upcase(wd(8,3,hex(ulli))) | left(upcase(wd(8,3,ulli))) | pad0(wd(8,ulli));
+        sout | ulli | wd(2,ulli) | wd(5,2,ulli) | left(wd(5,2,ulli)) | nobase(oct(ulli)) | oct(ulli) | nonl;
+        sout | nobase(hex(ulli)) | hex(ulli) | wd(8,hex(ulli)) | wd(8,10,hex(ulli)) | upcase(wd(8,3,hex(ulli))) | nonl;
+        sout | left(upcase(wd(8,3,ulli))) | pad0(wd(8,ulli));
         sout | nl | "binary integral";
+        sout | bin(0) | bin(13) | upcase(bin(13)) | nobase(bin(13)) | left(wd(8,bin(13))) | wd(8,bin(13)) | pad0(left(wd(8,bin(13)))) | pad0(wd(8,bin(13))) | pad0(wd(8,10,bin(13))) | pad0(wd(8,6,bin(13)));
+        sout | bin(0) | bin(13) | upcase(bin(13)) | nobase(bin(13)) | left(wd(8,bin(13))) | wd(8,bin(13)) | nonl;
+        sout | pad0(left(wd(8,bin(13)))) | pad0(wd(8,bin(13))) | pad0(wd(8,10,bin(13))) | pad0(wd(8,6,bin(13)));
 …
         printf( "%g  %8g %#8g %g %8g %8.0g %#8.0g %8.2g %#8.2g %-8.2g %-8.2g %-#8.2g %-+8.2g %-+#8.2g %08.2g %8.2E %8.2a %#8.2A %#8.2e\n",
 .0,3.0F,3.0F, f,  f,    f,     f,    f,     f,  3.0F,      f,      f,      f,       f,     f,    f,    f,     f,     f );
+        sout | 0.0 | wd(8, 3.0F) | nodp(wd(8, 3.0F)) | f | wd(8, f) | ws(8,0, f) | nodp(ws(8,0, f)) | ws(8,2, f) | nodp(ws(8,2, f)) | left(ws(8,2, 3.0F)) | left(ws(8,2, f)) | left(nodp(ws(8,2, f))) | left(sign(ws(8,2, f))) | left(sign(nodp(ws(8,2, f)))) | pad0(ws(8,2, f)) | upcase(wd(8,2, sci(f))) | wd(8,2, hex(f)) | upcase(wd(8,2, hex(f))) | nodp(wd(8,2, sci(f)));
+        sout | 0.0 | wd(8, 3.0F) | nodp(wd(8, 3.0F)) | f | wd(8, f) | ws(8,0, f) | nodp(ws(8,0, f)) | ws(8,2, f) | nodp(ws(8,2, f)) | nonl;
+        sout | left(ws(8,2, 3.0F)) | left(ws(8,2, f)) | left(nodp(ws(8,2, f))) | left(sign(ws(8,2, f))) | left(sign(nodp(ws(8,2, f)))) | nonl;
+        sout | pad0(ws(8,2, f)) | upcase(wd(8,2, sci(f))) | wd(8,2, hex(f)) | upcase(wd(8,2, hex(f))) | nodp(wd(8,2, sci(f)));
         sout | "double";
 …
         printf( "%g  %#8f %g %8f %#8.0f %8.0f %8.2f %-8.2f %-+#8.2f %08.2F %8.2E %8.2a %8.2A %8.2e\n",
 .0,  3.0, d,  d,     d,    d,    d,     d,       d,     d,    d,    d,    d,    d );
+        sout | 0.0 | wd(8, 3.0) | d | wd(8, d) | nodp(wd(8,0, d)) | wd(8,0, d) | wd(8,2, d) | left(wd(8,2, d)) | left(sign(wd(8,2, d))) | pad0(upcase(wd(8,2, d))) | upcase(wd(8,2, sci(d))) | wd(8,2, hex(d)) | upcase(wd(8,2, hex(d))) | wd(8,2, sci(d));
+        sout | 0.0 | wd(8, 3.0) | d | wd(8, d) | nodp(wd(8,0, d)) | wd(8,0, d) | wd(8,2, d) | nonl;
+        sout | left(wd(8,2, d)) | left(sign(wd(8,2, d))) | pad0(upcase(wd(8,2, d))) | upcase(wd(8,2, sci(d))) | wd(8,2, hex(d)) | upcase(wd(8,2, hex(d))) | wd(8,2, sci(d));
         sout | "long double";
 …
         printf( "%Lg  %#8Lf %Lg %8Lf %#8.0Lf %8.0Lf %8.2Lf %-8.2Lf %-+#8.2Lf %08.2LF %8.2LE %8.2La %8.2LA %8.2Le\n",
 .0L,  3.0L, ld,  ld,     ld,    ld,    ld,     ld,       ld,     ld,    ld,    ld,    ld,    ld );
+        sout | 0.0L | wd(8, 3.0L) | ld | wd(8, ld) | nodp(wd(8,0, ld)) | wd(8,0, ld) | wd(8,2, ld) | left(wd(8,2, ld)) | left(sign(wd(8,2, ld))) | pad0(upcase(wd(8,2, ld))) | upcase(wd(8,2, sci(ld))) | wd(8,2, hex(ld)) | upcase(wd(8,2, hex(ld))) | wd(8,2, sci(ld));
+        sout | 0.0L | wd(8, 3.0L) | ld | wd(8, ld) | nodp(wd(8,0, ld)) | wd(8,0, ld) | wd(8,2, ld) | nonl;
+        sout | left(wd(8,2, ld)) | left(sign(wd(8,2, ld))) | pad0(upcase(wd(8,2, ld))) | upcase(wd(8,2, sci(ld))) | wd(8,2, hex(ld)) | upcase(wd(8,2, hex(ld))) | wd(8,2, sci(ld));
 …
         char c = 'a';
         printf( "%c %2c %5c %-5c %hho %#hho %hhx %#hhx %#8hhx %#8hhX %-8c %8c\n", c, c, c, c, c, c, c, c, c, c, c, c );
+        sout | c | ' ' | wd(2,c) | wd(5,c) | left(wd(5,c)) | nobase(oct(c)) | oct(c) | nobase(hex(c)) | hex(c) | wd(8,hex(c)) | upcase(wd(8,hex(c))) | left(wd(8,c)) | wd(8,c);
+        sout | c | ' ' | wd(2,c) | wd(5,c) | left(wd(5,c)) | nobase(oct(c)) | oct(c) | nonl;
+        sout | nobase(hex(c)) | hex(c) | wd(8,hex(c)) | upcase(wd(8,hex(c))) | left(wd(8,c)) | wd(8,c);
         sout | nl | "string";

tests/pybin/settings.py

-              rb7d6a36
+              r6a490b2
 class Architecture:
         KnownArchitectures = {
                 'x64'           : 'x64',
                 'x86-64'        : 'x64',
                 'x86_64'        : 'x64',
                 'x86'           : 'x86',
                 'aarch64'       : 'arm',
                 'i386'          : 'x86',
                 'i486'          : 'x86',
                 'i686'          : 'x86',
                 'Intel 80386'   : 'x86',
                 'arm'           : 'arm',
                 'ARM'           : 'arm',
+                'x64'         : 'x64',
+                'x86-64'      : 'x64',
+                'x86_64'      : 'x64',
+                'x86'         : 'x86',
+                'aarch64'     : 'arm',
+                'i386'        : 'x86',
+                'i486'        : 'x86',
+                'i686'        : 'x86',
+                'Intel 80386' : 'x86',
+                'arm'         : 'arm',
+                'ARM'         : 'arm',
+        }
 …
                         print("updated to %s" % self.target)
+        def match(self, arch):
+        def filter(self, tests):
+                return [test for test in tests if not test.arch or self.target == test.arch]
                 return True if not arch else self.target == arch
         @classmethod
         def make_canonical(_, arch):
+        @staticmethod
+        def make_canonical(arch):
                 return Architecture.KnownArchitectures[arch]
 …
                 self.total  = Timeouts.check(tg)
         @classmethod
         def check(_, value):
+        @staticmethod
+        def check(value):
                 if value < 1:
                         print("Timeouts must be at least 1 second", file=sys.stderr)
 …
 def init( options ):
+        global all_arch
+        global all_debug
+        global all_install
         global arch
         global archive
+        global continue_
         global debug
-        global distcc
         global dry_run
         global generating
 …
         global output_width
         global timeout
+        global timeout2gdb
+        arch         = Architecture(options.arch)
+        all_arch     = [Architecture(o) for o in list(dict.fromkeys(options.arch   ))] if options.arch else [Architecture(None)]
+        all_debug    = [Debug(o)        for o in list(dict.fromkeys(options.debug  ))]
+        all_install  = [Install(o)      for o in list(dict.fromkeys(options.install))]
         archive      = os.path.abspath(os.path.join(original_path, options.archive_errors)) if options.archive_errors else None
         debug        = Debug(options.debug)
+        continue_    = options.continue_
         dry_run      = options.dry_run # must be called before tools.config_hash()
-        distcc       = "DISTCC_CFA_PATH=~/.cfadistcc/%s/cfa" % tools.config_hash()
         generating   = options.regenerate_expected
-        install      = Install(options.install)
         make         = ['make']
         output_width = 24
         timeout      = Timeouts(options.timeout, options.global_timeout)
+        timeout2gdb  = options.timeout_with_gdb
         # if we distribute, distcc errors will fail tests, use log file for distcc
 …
 def validate():
+        """Validate the current configuration and update globals"""
+        global distcc
+        distcc       = "DISTCC_CFA_PATH=~/.cfadistcc/%s/cfa" % tools.config_hash()
         errf = os.path.join(BUILDDIR, ".validate.err")
         make_ret, out = tools.make( ".validate", error_file = errf, output_file=subprocess.DEVNULL, error=subprocess.DEVNULL )

tests/pybin/test_run.py

-              rb7d6a36
+              r6a490b2
                 return os.path.normpath( os.path.join(settings.BUILDDIR, self.path, self.name) )
         @classmethod
         def valid_name(_, name):
+        @staticmethod
+        def valid_name(name):
                 return not name.endswith( ('.c', '.cc', '.cpp', '.cfa') )
         @classmethod
         def from_target(_, target):
+        @staticmethod
+        def new_target(target, arch):
                 test = Test()
                 test.name = os.path.basename(target)
                 test.path = os.path.relpath (os.path.dirname(target), settings.SRCDIR)
                 test.arch = settings.arch.target if settings.arch.cross_compile else ''
+                test.arch = arch.target if arch else ''
                 return test
 …
                 return text
         @classmethod
         def fmtDur( cls, duration ):
+        @staticmethod
+        def fmtDur( duration ):
                 if duration :
                         hours, rem = divmod(duration, 3600)

tests/pybin/tools.py

-              rb7d6a36
+              r6a490b2
                                         return proc.returncode, out.decode("utf-8") if out else None
                                 except subprocess.TimeoutExpired:
+                                        proc.send_signal(signal.SIGABRT)
+                                        proc.communicate()
+                                        return 124, str(None)
+                                        if settings.timeout2gdb:
+                                                print("Process {} timeout".format(proc.pid))
+                                                proc.communicate()
+                                                return 124, str(None)
+                                        else:
+                                                proc.send_signal(signal.SIGABRT)
+                                                proc.communicate()
+                                                return 124, str(None)
         except Exception as ex:
 …
         raise argparse.ArgumentTypeError(msg)
+# Convert a function that converts a string to one that converts comma separated string.
+def comma_separated(elements):
+    return lambda string: [elements(part) for part in string.split(',')]
 def fancy_print(text):
         column = which('column')

tests/test.py

-              rb7d6a36
+              r6a490b2
 import argparse
+import itertools
 import re
 import sys
 …
                         test.path = match.group(1)
                         test.arch = match.group(3)[1:] if match.group(3) else None
+                        if settings.arch.match(test.arch):
+                                expected.append(test)
+                        expected.append(test)
         path_walk( match_test )
 …
+                ]
+        # sort the test alphabetically for convenience
+        test_list.sort(key=lambda t: ('~' if t.arch else '') + t.target() + (t.arch if t.arch else ''))
         return test_list
 …
                 for testname in options.tests :
                         testname = canonical_path( testname )
+                        # first check if this is a valid name to regenerate
                         if Test.valid_name(testname):
+                                # this is a valid name, let's check if it already exists
                                 found = [test for test in all_tests if canonical_path( test.target() ) == testname]
+                                tests.append( found[0] if len(found) == 1 else Test.from_target(testname) )
+                                if not found:
+                                        # it's a new name, create it according to the name and specified architecture
+                                        if options.arch:
+                                                # user specified one or multiple architectures, assume the tests will have architecture specific results
+                                                tests.extend( [Test.new_target(testname, arch) for arch in settings.all_arch] )
+                                        else:
+                                                # user didn't specify an architecture, just create a cross platform test
+                                                tests.append( Test.new_target( testname, None ) )
+                                elif len(found) == 1 and not found[0].arch:
+                                        # we found a single test, the user better be wanting to create a cross platform test
+                                        if options.arch:
+                                                print('ERROR: "%s", test has no specified architecture but --arch was specified, ignoring it' % testname, file=sys.stderr)
+                                        else:
+                                                tests.append( found[0] )
+                                else:
+                                        # this test is already cross platform, just add a test for each platform the user asked
+                                        tests.extend( [Test.new_target(testname, arch) for arch in settings.all_arch] )
+                                        # print a warning if it users didn't ask for a specific architecture
+                                        if not options.arch:
+                                                print('WARNING: "%s", test has architecture specific expected files but --arch was not specified, regenerating only for current host' % testname, file=sys.stderr)
                         else :
                                 print('ERROR: "%s", tests are not allowed to end with a C/C++/CFA extension, ignoring it' % testname, file=sys.stderr)
 …
                         if test :
                                 tests.append( test[0] )
+                                tests.extend( test )
                         else :
                                 print('ERROR: No expected file for test %s, ignoring it' % testname, file=sys.stderr)
 …
         # create a parser with the arguments for the tests script
         parser = argparse.ArgumentParser(description='Script which runs cforall tests')
+        parser.add_argument('--debug', help='Run all tests in debug or release', type=yes_no, default='yes')
+        parser.add_argument('--install', help='Run all tests based on installed binaries or tree binaries', type=yes_no, default='no')
+        parser.add_argument('--arch', help='Test for specific architecture', type=str, default='')
+        parser.add_argument('--debug', help='Run all tests in debug or release', type=comma_separated(yes_no), default='yes')
+        parser.add_argument('--install', help='Run all tests based on installed binaries or tree binaries', type=comma_separated(yes_no), default='no')
+        parser.add_argument('--arch', help='Test for specific architecture', type=comma_separated(str), default=None)
+        parser.add_argument('--continue', help='When multiple specifications are passed (debug/install/arch), sets whether or not to continue if the last specification failed', type=yes_no, default='yes', dest='continue_')
         parser.add_argument('--timeout', help='Maximum duration in seconds after a single test is considered to have timed out', type=int, default=60)
         parser.add_argument('--global-timeout', help='Maximum cumulative duration in seconds after the ALL tests are considered to have timed out', type=int, default=7200)
+        parser.add_argument('--timeout-with-gdb', help='Instead of killing the command when it times out, orphan it and print process id to allow gdb to attach', type=yes_no, default="no")
         parser.add_argument('--dry-run', help='Don\'t run the tests, only output the commands', action='store_true')
         parser.add_argument('--list', help='List all test available', action='store_true')
 …
                 else:
+                        with open (out_file, "r") as myfile:
+                                error = myfile.read()
+                        if os.stat(out_file).st_size < 1048576:
+                                with open (out_file, "r") as myfile:
+                                        error = myfile.read()
+                        else:
+                                error = "Output log can't be read, file is bigger than 1MB, see {} for actual error\n".format(out_file)
                         ret, info = core_info(exe_file)
 …
                 return False, ""
         except Exception as ex:
                 print("Unexpected error in worker thread: %s" % ex, file=sys.stderr)
+                print("Unexpected error in worker thread running {}: {}".format(t.target(), ex), file=sys.stderr)
                 sys.stderr.flush()
                 return False, ""
 …
         make('clean', output_file=subprocess.DEVNULL, error=subprocess.DEVNULL)
         return 1 if failed else 0
+        return failed
 …
         settings.init( options )
-        # fetch the liest of all valid tests
-        all_tests = list_tests( options.include, options.exclude )
-        # if user wants all tests than no other treatement of the test list is required
-        if options.all or options.list or options.list_comp or options.include :
-                tests = all_tests
-        #otherwise we need to validate that the test list that was entered is valid
-        else :
-                tests = valid_tests( options )
-        # make sure we have at least some test to run
-        if not tests :
-                print('ERROR: No valid test to run', file=sys.stderr)
-                sys.exit(1)
-        # sort the test alphabetically for convenience
-        tests.sort(key=lambda t: (t.arch if t.arch else '') + t.target())
         # users may want to simply list the tests
         if options.list_comp :
+                print("-h --help --debug --dry-run --list --arch --all --regenerate-expected --archive-errors --install --timeout --global-timeout -j --jobs ", end='')
+                # fetch the liest of all valid tests
+                tests = list_tests( None, None )
+                # print the possible options
+                print("-h --help --debug --dry-run --list --arch --all --regenerate-expected --archive-errors --install --timeout --global-timeout --timeout-with-gdb -j --jobs -I --include -E --exclude --continue ", end='')
                 print(" ".join(map(lambda t: "%s" % (t.target()), tests)))
         elif options.list :
+                print("Listing for %s:%s"% (settings.arch.string, settings.debug.string))
+                # fetch the liest of all valid tests
+                tests = list_tests( options.include, options.exclude )
+                # print the available tests
                 fancy_print("\n".join(map(lambda t: t.toString(), tests)))
         else :
+                # check the build configuration works
+                # fetch the liest of all valid tests
+                all_tests = list_tests( options.include, options.exclude )
+                # if user wants all tests than no other treatement of the test list is required
+                if options.all or options.include :
+                        tests = all_tests
+                #otherwise we need to validate that the test list that was entered is valid
+                else :
+                        tests = valid_tests( options )
+                # make sure we have at least some test to run
+                if not tests :
+                        print('ERROR: No valid test to run', file=sys.stderr)
+                        sys.exit(1)
+                # prep invariants
                 settings.prep_output(tests)
+                settings.validate()
+                options.jobs, forceJobs = job_count( options, tests )
+                settings.update_make_cmd(forceJobs, options.jobs)
+                print('%s %i tests on %i cores (%s:%s)' % (
+                        'Regenerating' if settings.generating else 'Running',
+                        len(tests),
+                        options.jobs,
+                        settings.arch.string,
+                        settings.debug.string
+                ))
+                # otherwise run all tests and make sure to return the correct error code
+                sys.exit( run_tests(tests, options.jobs) )
+                failed = 0
+                # for each build configurations, run the test
+                for arch, debug, install in itertools.product(settings.all_arch, settings.all_debug, settings.all_install):
+                        settings.arch    = arch
+                        settings.debug   = debug
+                        settings.install = install
+                        # filter out the tests for a different architecture
+                        # tests are the same across debug/install
+                        local_tests = settings.arch.filter( tests )
+                        options.jobs, forceJobs = job_count( options, local_tests )
+                        settings.update_make_cmd(forceJobs, options.jobs)
+                        # check the build configuration works
+                        settings.validate()
+                        # print configuration
+                        print('%s %i tests on %i cores (%s:%s)' % (
+                                'Regenerating' if settings.generating else 'Running',
+                                len(local_tests),
+                                options.jobs,
+                                settings.arch.string,
+                                settings.debug.string
+                        ))
+                        # otherwise run all tests and make sure to return the correct error code
+                        failed = run_tests(local_tests, options.jobs)
+                        if failed:
+                                result = 1
+                                if not settings.continue_:
+                                        break
+                sys.exit( failed )

tests/vector.cfa

-              rb7d6a36
+              r6a490b2
 //
+#include <vector.hfa>
 #include <fstream.hfa>
-#include <vector.hfa>
 #undef assert
 …
 int main() {
         vector( int ) iv;
+        assert( ((uintptr_t)&iv.storage.storage ) == (((uintptr_t)&iv)) );
+        assert( ((uintptr_t)&iv.storage.capacity) == (((uintptr_t)&iv) + sizeof(void *)) );
+        assert( ((uintptr_t)&iv.size            ) == (((uintptr_t)&iv) + sizeof(void *) + sizeof(size_t)) );
         assert( empty( &iv ) );

tools/build/push2dist.sh

rb7d6a36	r6a490b2
19	19	# echo "Copying to machines : ${hosts} (hash=${hash})"
20	20
21		files="../../../driver/cfa ../../../driver/cfa-cpp ../../../driver/cc1 ../../../driver/as $(find . -name '.c' \| tr '\n' ' ')"
	21	files="../../../driver/cfa ../../../driver/cfa-cpp ../../../driver/cc1 ../../../driver/as defines.hfa $(find . -name '.c' \| tr '\n' ' ')"
22	22	# echo "Files ${files}"
23	23

tools/cfa.nanorc

-              rb7d6a36
+              r6a490b2
 # Declarations
 color brightgreen "\<(struct|union|typedef|trait|coroutine|monitor|thread)\>"
 color brightgreen "\<(with)\>"
+color brightgreen "\<(struct|union|typedef|trait|coroutine|generator)\>"
+color brightgreen "\<(monitor|thread|with)\>"
 # Control Flow Structures
 color brightyellow "\<(if|else|while|do|for|switch|choose|case|default)\>"
 color brightyellow "\<(disable|enable|waitfor|when|timeout)\>"
+color brightyellow "\<(disable|enable|waitfor|when|timeout|suspend)\>"
 color brightyellow "\<(try|catch(Resume)?|finally)\>"
 …
 # Escaped Keywords, now Identifiers.
 color white "`\w+`"
+color white "``\w+"
 # Operator Names
 …
 ## Update/Redistribute
 # GCC builtins
 color cyan "__attribute__[[:space:]]*\(\([^()]*(\([^()]*\)[^()]*)*\)\)"
+color cyan "__attribute__[[:space:]]*\(\(([^)]|[^)]\))*\)\)"
 ##color cyan "__(aligned|asm|builtin|hidden|inline|packed|restrict|section|typeof|weak)__"

tools/vscode/uwaterloo.cforall-0.1.0/package.json

-              rb7d6a36
+              r6a490b2
         "name": "cforall",
         "version": "0.1.0",
         "displayName": "Cforall Language Support",
+        "displayName": "C∀ (C-for-all) Language Support",
         "description": "Cforall - colorizer, grammar and snippets.",
         "publisher": "uwaterloo",
 …
                 "vscode": "^1.5.0"
         },
         "icon": "images/icon.svg",
+        "icon": "images/icon.png",
         "categories": [
                 "Languages",
+                "Programming Languages",
                 "Linters",
                 "Other"
         ],
+        "activationEvents": [
+                "onLanguage:cforall"
+        ],
+        "main": "./client/main.js",
         "contributes": {
                 "languages": [
 …
                                 "aliases": [
                                         "C∀",
+                                        "CForAll",
                                         "Cforall",
-                                        "CForAll",
                                         "cforall"
                                 ],
                                 "extensions": [
+                                        ".cf"
+                                        ".cfa",
+                                        ".hfa",
+                                        ".ifa"
                                 ],
                                 "configuration": "./cforall.configuration.json"
 …
+                        {
                                 "language": "cforall",
                                 "scopeName": "source.cf",
                                 "path": "./syntaxes/cfa.tmLanguage"
+                                "scopeName": "source.cfa",
+                                "path": "./syntaxes/cfa.tmLanguage.json"
+                        }
+                ]
+                ],
+                "configuration": {
+                        "type": "object",
+                        "title": "Example configuration",
+                        "properties": {
+                                "cforall.maxNumberOfProblems": {
+                                        "scope": "resource",
+                                        "type": "number",
+                                        "default": 100,
+                                        "description": "Controls the maximum number of problems produced by the server."
+                                },
+                                "cforall.trace.server": {
+                                        "scope": "window",
+                                        "type": "string",
+                                        "enum": [
+                                                "off",
+                                                "messages",
+                                                "verbose"
+                                        ],
+                                        "default": "off",
+                                        "description": "Traces the communication between VS Code and the language server."
+                                }
+                        }
+                }
+        },
+        "dependencies": {
+                "vscode-languageclient": "^4.1.4"
+        },
+        "devDependencies": {
+                "vscode-languageclient": "^4.1.4"
+        }
+}

Context Navigation

Legend:

Download in other formats: