Changeset 365c8dcb for doc/theses/thierry_delisle_PhD

Timestamp:

Apr 14, 2022, 3:00:28 PM (4 years ago)

Author:

JiadaL <j82liang@…>

Branches:

ADT, ast-experimental, enum, master, pthread-emulation, qualifiedEnum, stuck-waitfor-destruct

Children:

bfd5512

Parents:

30d91e4 (diff), 4ec9513 (diff)
Note: this is a merge changeset, the changes displayed below correspond to the merge itself.
Use the (diff) links above to see all the changes relative to each parent.

Message:

Merge branch 'master' into enum

Location:

doc/theses/thierry_delisle_PhD/thesis

Files:

• Makefile (modified) (1 diff)
• fig/cache-noshare.fig (added)
• fig/cache-share.fig (added)
• local.bib (modified) (1 diff)
• text/core.tex (modified) (2 diffs)
• text/io.tex (modified) (1 diff)
• text/practice.tex (modified) (1 diff)

doc/theses/thierry_delisle_PhD/thesis/Makefile

@@ -30 +30 @@
         base \
         base_avg \
+        cache-share \
+        cache-noshare \
         empty \
         emptybit \

doc/theses/thierry_delisle_PhD/thesis/local.bib

@@ -685 +685 @@
   note = "[Online; accessed 9-February-2021]"
 }
+
+@misc{wiki:rcu,
+  author = "{Wikipedia contributors}",
+  title = "Read-copy-update --- {W}ikipedia{,} The Free Encyclopedia",
+  year = "2022",
+  url = "https://en.wikipedia.org/wiki/Linear_congruential_generator",
+  note = "[Online; accessed 12-April-2022]"
+}
+
+@misc{wiki:rwlock,
+  author = "{Wikipedia contributors}",
+  title = "Readers-writer lock --- {W}ikipedia{,} The Free Encyclopedia",
+  year = "2021",
+  url = "https://en.wikipedia.org/wiki/Readers%E2%80%93writer_lock",
+  note = "[Online; accessed 12-April-2022]"
+}

doc/theses/thierry_delisle_PhD/thesis/text/core.tex

@@ -32 +32 @@
         \item Faster than other schedulers that have equal or better fairness.
 \end{itemize}
+
+\subsection{Fairness Goals}
+For this work fairness will be considered as having two strongly related requirements: true starvation freedom and ``fast'' load balancing.
+
+\paragraph{True starvation freedom} is more easily defined: As long as at least one \proc continues to dequeue \ats, all read \ats should be able to run eventually.
+In any running system, \procs can stop dequeing \ats if they start running a \at that will simply never park.
+Traditional workstealing schedulers do not have starvation freedom in these cases.
+Now this requirement begs the question, what about preemption?
+Generally speaking preemption happens on the timescale of several milliseconds, which brings us to the next requirement: ``fast'' load balancing.
+
+\paragraph{Fast load balancing} means that load balancing should happen faster than preemption would normally allow.
+For interactive applications that need to run at 60, 90, 120 frames per second, \ats having to wait for several millseconds to run are effectively starved.
+Therefore load-balancing should be done at a faster pace, one that can detect starvation at the microsecond scale.
+With that said, this is a much fuzzier requirement since it depends on the number of \procs, the number of \ats and the general load of the system.
 
 \subsection{Fairness vs Scheduler Locality} \label{fairnessvlocal}
@@ -223 +237 @@
 Therefore this unprotected read of the timestamp and average satisfy the limited correctness that is required.
 
+\begin{figure}
+        \centering
+        \input{cache-share.pstex_t}
+        \caption[CPU design with wide L3 sharing]{CPU design with wide L3 sharing \smallskip\newline A very simple CPU with 4 \glspl{hthrd}. L1 and L2 are private to each \gls{hthrd} but the L3 is shared across to entire core.}
+        \label{fig:cache-share}
+\end{figure}
+
+\begin{figure}
+        \centering
+        \input{cache-noshare.pstex_t}
+        \caption[CPU design with a narrower L3 sharing]{CPU design with a narrower L3 sharing \smallskip\newline A different CPU design, still with 4 \glspl{hthrd}. L1 and L2 are still private to each \gls{hthrd} but the L3 is shared some of the CPU but there is still two distinct L3 instances.}
+        \label{fig:cache-noshare}
+\end{figure}
+
+With redundant tiemstamps this scheduling algorithm achieves both the fairness and performance requirements, on some machines.
+The problem is that the cost of polling and helping is not necessarily consistent across each \gls{hthrd}.
+For example, on machines where the motherboard holds multiple CPU, cache misses can be satisfied from a cache that belongs to the CPU that missed, the \emph{local} CPU, or by a different CPU, a \emph{remote} one.
+Cache misses that are satisfied by a remote CPU will have higher latency than if it is satisfied by the local CPU.
+However, this is not specific to systems with multiple CPUs.
+Depending on the cache structure, cache-misses can have different latency for the same CPU.
+The AMD EPYC 7662 CPUs that is described in Chapter~\ref{microbench} is an example of that.
+Figure~\ref{fig:cache-share} and Figure~\ref{fig:cache-noshare} show two different cache topologies with highlight this difference.
+In Figure~\ref{fig:cache-share}, all cache instances are either private to a \gls{hthrd} or shared to the entire system, this means latency due to cache-misses are likely fairly consistent.
+By comparison, in Figure~\ref{fig:cache-noshare} misses in the L2 cache can be satisfied by a hit in either instance of the L3.
+However, the memory access latency to the remote L3 instance will be notably higher than the memory access latency to the local L3.
+The impact of these different design on this algorithm is that scheduling will scale very well on architectures similar to Figure~\ref{fig:cache-share}, both will have notably worst scalling with many narrower L3 instances.
+This is simply because as the number of L3 instances grow, so two does the chances that the random helping will cause significant latency.
+The solution is to have the scheduler be aware of the cache topology.
+
 \subsection{Per CPU Sharding}
+Building a scheduler that is aware of cache topology poses two main challenges: discovering cache topology and matching \procs to cache instance.
+Sadly, there is no standard portable way to discover cache topology in C.
+Therefore, while this is a significant portability challenge, it is outside the scope of this thesis to design a cross-platform cache discovery mechanisms.
+The rest of this work assumes discovering the cache topology based on Linux's \texttt{/sys/devices/system/cpu} directory.
+This leaves the challenge of matching \procs to cache instance, or more precisely identifying which subqueues of the ready queue are local to which cache instance.
+Once this matching is available, the helping algorithm can be changed to add bias so that \procs more often help subqueues local to the same cache instance
+\footnote{Note that like other biases mentioned in this section, the actual bias value does not appear to need precise tuinng.}.
+
+The obvious approach to mapping cache instances to subqueues is to statically tie subqueues to CPUs.
+Instead of having each subqueue local to a specific \proc, the system is initialized with subqueues for each \glspl{hthrd} up front.
+Then \procs dequeue and enqueue by first asking which CPU id they are local to, in order to identify which subqueues are the local ones.
+\Glspl{proc} can get the CPU id from \texttt{sched\_getcpu} or \texttt{librseq}.
+
+This approach solves the performance problems on systems with topologies similar to Figure~\ref{fig:cache-noshare}.
+However, it actually causes some subtle fairness problems in some systems, specifically systems with few \procs and many \glspl{hthrd}.
+In these cases, the large number of subqueues and the bias agains subqueues tied to different cache instances make it so it is very unlikely any single subqueue is picked.
+To make things worst, the small number of \procs mean that few helping attempts will be made.
+This combination of few attempts and low chances make it so a \at stranded on a subqueue that is not actively dequeued from may wait very long before it gets randomly helped.
+On a system with 2 \procs, 256 \glspl{hthrd} with narrow cache sharing, and a 100:1 bias, it can actually take multiple seconds for a \at to get dequeued from a remote queue.
+Therefore, a more dynamic matching of subqueues to cache instance is needed.
 
 \subsection{Topological Work Stealing}
-
-
+The approach that is used in the \CFA scheduler is to have per-\proc subqueue, but have an excplicit data-structure track which cache instance each subqueue is tied to.
+This is requires some finess because reading this data structure must lead to fewer cache misses than not having the data structure in the first place.
+A key element however is that, like the timestamps for helping, reading the cache instance mapping only needs to give the correct result \emph{often enough}.
+Therefore the algorithm can be built as follows: Before enqueuing or dequeing a \at, each \proc queries the CPU id and the corresponding cache instance.
+Since subqueues are tied to \procs, each \proc can then update the cache instance mapped to the local subqueue(s).
+To avoid unnecessary cache line invalidation, the map is only written to if the mapping changes.
+

doc/theses/thierry_delisle_PhD/thesis/text/io.tex

@@ -406 +406 @@
 Finally, the last important part of the \io subsystem is it's interface. There are multiple approaches that can be offered to programmers, each with advantages and disadvantages. The new \io subsystem can replace the C runtime's API or extend it. And in the later case the interface can go from very similar to vastly different. The following sections discuss some useful options using @read@ as an example. The standard Linux interface for C is :
 
-@ssize_t read(int fd, void *buf, size_t count);@.
+@ssize_t read(int fd, void *buf, size_t count);@
 
 \subsection{Replacement}
-Replacing the C \glsxtrshort{api}
+Replacing the C \glsxtrshort{api} is the more intrusive and draconian approach.
+The goal is to convince the compiler and linker to replace any calls to @read@ to direct them to the \CFA implementation instead of glibc's.
+This has the advantage of potentially working transparently and supporting existing binaries without needing recompilation.
+It also offers a, presumably, well known and familiar API that C programmers can simply continue to work with.
+However, this approach also entails a plethora of subtle technical challenges which generally boils down to making a perfect replacement.
+If the \CFA interface replaces only \emph{some} of the calls to glibc, then this can easily lead to esoteric concurrency bugs.
+Since the gcc ecosystems does not offer a scheme for such perfect replacement, this approach was rejected as being laudable but infeasible.
 
 \subsection{Synchronous Extension}
+An other interface option is to simply offer an interface that is different in name only. For example:
+
+@ssize_t cfa_read(int fd, void *buf, size_t count);@
+
+\noindent This is much more feasible but still familiar to C programmers.
+It comes with the caveat that any code attempting to use it must be recompiled, which can be a big problem considering the amount of existing legacy C binaries.
+However, it has the advantage of implementation simplicity.
 
 \subsection{Asynchronous Extension}
+It is important to mention that there is a certain irony to using only synchronous, therefore blocking, interfaces for a feature often referred to as ``non-blocking'' \io.
+A fairly traditional way of doing this is using futures\cit{wikipedia futures}.
+As simple way of doing so is as follows:
+
+@future(ssize_t) read(int fd, void *buf, size_t count);@
+
+\noindent Note that this approach is not necessarily the most idiomatic usage of futures.
+The definition of read above ``returns'' the read content through an output parameter which cannot be synchronized on.
+A more classical asynchronous API could look more like:
+
+@future([ssize_t, void *]) read(int fd, size_t count);@
+
+\noindent However, this interface immediately introduces memory lifetime challenges since the call must effectively allocate a buffer to be returned.
+Because of the performance implications of this, the first approach is considered preferable as it is more familiar to C programmers.
 
 \subsection{Interface directly to \lstinline{io_uring}}
+Finally, an other interface that can be relevant is to simply expose directly the underlying \texttt{io\_uring} interface. For example:
+
+@array(SQE, want) cfa_io_allocate(int want);@
+
+@void cfa_io_submit( const array(SQE, have) & );@
+
+\noindent This offers more flexibility to users wanting to fully use all of the \texttt{io\_uring} features.
+However, it is not the most user-friendly option.
+It obviously imposes a strong dependency between user code and \texttt{io\_uring} but at the same time restricting users to usages that are compatible with how \CFA internally uses \texttt{io\_uring}.
+
+

doc/theses/thierry_delisle_PhD/thesis/text/practice.tex

@@ -2 +2 @@
 The scheduling algorithm discribed in Chapter~\ref{core} addresses scheduling in a stable state.
 However, it does not address problems that occur when the system changes state.
-Indeed the \CFA runtime, supports expanding and shrinking the number of KTHREAD\_place \todo{add kthrd to glossary}, both manually and, to some extent automatically.
+Indeed the \CFA runtime, supports expanding and shrinking the number of \procs, both manually and, to some extent, automatically.
 This entails that the scheduling algorithm must support these transitions.
 
-\section{Resizing}
+More precise \CFA supports adding \procs using the RAII object @processor@.
+These objects can be created at any time and can be destroyed at any time.
+They are normally create as automatic stack variables, but this is not a requirement.
+
+The consequence is that the scheduler and \io subsystems must support \procs comming in and out of existence.
+
+\section{Manual Resizing}
+The consequence of dynamically changing the number of \procs is that all internal arrays that are sized based on the number of \procs neede to be \texttt{realloc}ed.
+This also means that any references into these arrays, pointers or indexes, may need to be fixed when shrinking\footnote{Indexes may still need fixing because there is no guarantee the \proc causing the shrink had the highest index. Therefore indexes need to be reassigned to preserve contiguous indexes.}.
+
+There are no performance requirements, within reason, for resizing since this is usually considered as part of setup and teardown.
+However, this operation has strict correctness requirements since shrinking and idle sleep can easily lead to deadlocks.
+It should also avoid as much as possible any effect on performance when the number of \procs remain constant.
+This later requirement prehibits simple solutions, like simply adding a global lock to these arrays.
+
+\subsection{Read-Copy-Update}
+One solution is to use the Read-Copy-Update\cite{wiki:rcu} pattern.
+In this pattern, resizing is done by creating a copy of the internal data strucures, updating the copy with the desired changes, and then attempt an Idiana Jones Switch to replace the original witht the copy.
+This approach potentially has the advantage that it may not need any synchronization to do the switch.
+The switch definitely implies a race where \procs could still use the previous, original, data structure after the copy was switched in.
+The important question then becomes whether or not this race can be recovered from.
+If the changes that arrived late can be transferred from the original to the copy then this solution works.
+
+For linked-lists, dequeing is somewhat of a problem.
+Dequeing from the original will not necessarily update the copy which could lead to multiple \procs dequeing the same \at.
+Fixing this requires making the array contain pointers to subqueues rather than the subqueues themselves.
+
+Another challenge is that the original must be kept until all \procs have witnessed the change.
+This is a straight forward memory reclamation challenge but it does mean that every operation will need \emph{some} form of synchronization.
+If each of these operation does need synchronization then it is possible a simpler solution achieves the same performance.
+Because in addition to the classic challenge of memory reclamation, transferring the original data to the copy before reclaiming it poses additional challenges.
+Especially merging subqueues while having a minimal impact on fairness and locality.
+
+\subsection{Read-Writer Lock}
+A simpler approach would be to use a \newterm{Readers-Writer Lock}\cite{wiki:rwlock} where the resizing requires acquiring the lock as a writer while simply enqueing/dequeing \ats requires acquiring the lock as a reader.
+Using a Readers-Writer lock solves the problem of dynamically resizing and leaves the challenge of finding or building a lock with sufficient good read-side performance.
+Since this is not a very complex challenge and an ad-hoc solution is perfectly acceptable, building a Readers-Writer lock was the path taken.
+
+To maximize reader scalability, the readers should not contend with eachother when attempting to acquire and release the critical sections.
+This effectively requires that each reader have its own piece of memory to mark as locked and unlocked.
+Reades then acquire the lock wait for writers to finish the critical section and then acquire their local spinlocks.
+Writers acquire the global lock, so writers have mutual exclusion among themselves, and then acquires each of the local reader locks.
+Acquiring all the local locks guarantees mutual exclusion between the readers and the writer, while the wait on the read side prevents readers from continously starving the writer.
+\todo{reference listings}
+
+\begin{lstlisting}
+void read_lock() {
+        // Step 1 : make sure no writers in
+        while write_lock { Pause(); }
+
+        // May need fence here
+
+        // Step 2 : acquire our local lock
+        while atomic_xchg( tls.lock ) {
+                Pause();
+        }
+}
+
+void read_unlock() {
+        tls.lock = false;
+}
+\end{lstlisting}
+
+\begin{lstlisting}
+void write_lock()  {
+        // Step 1 : lock global lock
+        while atomic_xchg( write_lock ) {
+                Pause();
+        }
+
+        // Step 2 : lock per-proc locks
+        for t in all_tls {
+                while atomic_xchg( t.lock ) {
+                        Pause();
+                }
+        }
+}
+
+void write_unlock() {
+        // Step 1 : release local locks
+        for t in all_tls {
+                t.lock = false;
+        }
+
+        // Step 2 : release global lock
+        write_lock = false;
+}
+\end{lstlisting}
 
 \section{Idle-Sleep}
+
+\subsection{Tracking Sleepers}
+
+\subsection{Event FDs}
+
+\subsection{Epoll}
+
+\subsection{\texttt{io\_uring}}
+
+\subsection{Reducing Latency}