Changeset d95969a for doc/theses/andrew_beach_MMath/unwinding.tex
- Timestamp:
- Jan 25, 2021, 3:45:42 PM (3 years ago)
- Branches:
- ADT, arm-eh, ast-experimental, enum, forall-pointer-decay, jacob/cs343-translation, master, new-ast-unique-expr, pthread-emulation, qualifiedEnum
- Children:
- c292244
- Parents:
- b6a8b31 (diff), 7158202 (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. - File:
-
- 1 edited
-
doc/theses/andrew_beach_MMath/unwinding.tex (modified) (11 diffs)
Legend:
- Unmodified
- Added
- Removed
-
doc/theses/andrew_beach_MMath/unwinding.tex
rb6a8b31 rd95969a 1 \chapter{ Unwinding in \CFA}1 \chapter{\texorpdfstring{Unwinding in \CFA}{Unwinding in Cforall}} 2 2 3 3 Stack unwinding is the process of removing things from the stack. Within … … 10 10 Even this is fairly simple if nothing needs to happen when the stack unwinds. 11 11 Traditional C can unwind the stack by saving and restoring state (with 12 \codeC{setjmp} \& \codeC{longjmp}). However many languages define actions that12 @setjmp@ \& @longjmp@). However many languages define actions that 13 13 have to be taken when something is removed from the stack, such as running 14 a variable's destructor or a \codeCFA{try} statement's \codeCFA{finally}14 a variable's destructor or a @try@ statement's @finally@ 15 15 clause. Handling this requires walking the stack going through each stack 16 16 frame. … … 29 29 30 30 \CFA uses two primary functions in libunwind to create most of its 31 exceptional control-flow: \codeC{_Unwind_RaiseException}and32 \codeC{_Unwind_ForcedUnwind}.31 exceptional control-flow: @_Unwind_RaiseException@ and 32 @_Unwind_ForcedUnwind@. 33 33 Their operation is divided into two phases: search and clean-up. The search 34 34 phase -- phase 1 -- is used to scan the stack but not unwinding it. The … … 44 44 A personality function performs three tasks, although not all have to be 45 45 present. The tasks performed are decided by the actions provided. 46 \codeC{_Unwind_Action}is a bitmask of possible actions and an argument of46 @_Unwind_Action@ is a bitmask of possible actions and an argument of 47 47 this type is passed into the personality function. 48 48 \begin{itemize} 49 \item \codeC{_UA_SEARCH_PHASE}is passed in search phase and tells the49 \item@_UA_SEARCH_PHASE@ is passed in search phase and tells the 50 50 personality function to check for handlers. If there is a handler in this 51 51 stack frame, as defined by the language, the personality function should 52 return \codeC{_URC_HANDLER_FOUND}. Otherwise it should return53 \codeC{_URC_CONTINUE_UNWIND}.54 \item \codeC{_UA_CLEANUP_PHASE}is passed in during the clean-up phase and52 return @_URC_HANDLER_FOUND@. Otherwise it should return 53 @_URC_CONTINUE_UNWIND@. 54 \item@_UA_CLEANUP_PHASE@ is passed in during the clean-up phase and 55 55 means part or all of the stack frame is removed. The personality function 56 56 should do whatever clean-up the language defines 57 57 (such as running destructors/finalizers) and then generally returns 58 \codeC{_URC_CONTINUE_UNWIND}.59 \item \codeC{_UA_HANDLER_FRAME}means the personality function must install58 @_URC_CONTINUE_UNWIND@. 59 \item@_UA_HANDLER_FRAME@ means the personality function must install 60 60 a handler. It is also passed in during the clean-up phase and is in addition 61 61 to the clean-up action. libunwind provides several helpers for the personality 62 62 function here. Once it is done, the personality function must return 63 \codeC{_URC_INSTALL_CONTEXT}.63 @_URC_INSTALL_CONTEXT@. 64 64 \end{itemize} 65 65 The personality function is given a number of other arguments. Some are for 66 compatability and there is the \codeC{struct _Unwind_Context}pointer which66 compatability and there is the @struct _Unwind_Context@ pointer which 67 67 passed to many helpers to get information about the current stack frame. 68 68 … … 72 72 raise-exception but with some extras. 73 73 The first it passes in an extra action to the personality function on each 74 stack frame, \codeC{_UA_FORCE_UNWIND}, which means a handler cannot be74 stack frame, @_UA_FORCE_UNWIND@, which means a handler cannot be 75 75 installed. 76 76 … … 83 83 stack frames have been removed. By the standard API this is marked by setting 84 84 the stack pointer inside the context passed to the stop function. However both 85 GCC and Clang add an extra action for this case \codeC{_UA_END_OF_STACK}.85 GCC and Clang add an extra action for this case @_UA_END_OF_STACK@. 86 86 87 87 Each time function the stop function is called it can do one or two things. 88 When it is not the end of the stack it can return \codeC{_URC_NO_REASON}to88 When it is not the end of the stack it can return @_URC_NO_REASON@ to 89 89 continue unwinding. 90 90 % Is there a reason that NO_REASON is used instead of CONTINUE_UNWIND? … … 93 93 are provided to do it. 94 94 95 \section{\ CFA Implementation}95 \section{\texorpdfstring{\CFA Implementation}{Cforall Implementation}} 96 96 97 97 To use libunwind, \CFA provides several wrappers, its own storage, … … 113 113 114 114 The stop function is very simple. It checks the end of stack flag to see if 115 it is finished unwinding. If so, it calls \codeC{exit}to end the process,115 it is finished unwinding. If so, it calls @exit@ to end the process, 116 116 otherwise it returns with no-reason to continue unwinding. 117 117 % Yeah, this is going to have to change. … … 128 128 location of the instruction pointer and stack layout, which varies with 129 129 compiler and optimization levels. So for frames where there are only 130 destructors, GCC's attribute cleanup with the \texttt{-fexception}flag is130 destructors, GCC's attribute cleanup with the @-fexception@ flag is 131 131 sufficient to handle unwinding. 132 132 133 133 The only functions that require more than that are those that contain 134 \codeCFA{try} statements. A \codeCFA{try} statement has a \codeCFA{try} 135 clause, some number of \codeCFA{catch} clauses and \codeCFA{catchResume}136 clauses and may have a \codeCFA{finally} clause. Of these only \codeCFA{try}137 statements with \codeCFA{catch}clauses need to be transformed and only they138 and the \codeCFA{try}clause are involved.134 @try@ statements. A @try@ statement has a @try@ 135 clause, some number of @catch@ clauses and @catchResume@ 136 clauses and may have a @finally@ clause. Of these only @try@ 137 statements with @catch@ clauses need to be transformed and only they 138 and the @try@ clause are involved. 139 139 140 The \codeCFA{try}statement is converted into a series of closures which can140 The @try@ statement is converted into a series of closures which can 141 141 access other parts of the function according to scoping rules but can be 142 passed around. The \codeCFA{try}clause is converted into the try functions,143 almost entirely unchanged. The \codeCFA{catch}clauses are converted into two142 passed around. The @try@ clause is converted into the try functions, 143 almost entirely unchanged. The @catch@ clauses are converted into two 144 144 functions; the match function and the catch function. 145 145 … … 153 153 runs the handler's body. 154 154 155 These three functions are passed to \codeC{try_terminate}. This is an155 These three functions are passed to @try_terminate@. This is an 156 156 % Maybe I shouldn't quote that, it isn't its actual name. 157 157 internal hand-written function that has its own personality function and … … 167 167 handler was found in this frame. If it was then the personality function 168 168 installs the handler, which is setting the instruction pointer in 169 \codeC{try_terminate}to an otherwise unused section that calls the catch169 @try_terminate@ to an otherwise unused section that calls the catch 170 170 function, passing it the current exception and handler index. 171 \codeC{try_terminate}returns as soon as the catch function returns.171 @try_terminate@ returns as soon as the catch function returns. 172 172 173 173 At this point control has returned to normal control flow.
Note: See TracChangeset
for help on using the changeset viewer.
