microsoft
/
clang-1
зеркало из https://github.com/microsoft/clang-1.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
clang-1/lib/CodeGen
История
Eric Christopher acae011241 Carry the debug information from single exit unified return block 
along with the new insert point.

Fixes PR10829

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@139416 91177308-0d34-0410-b5e6-96231b3b80d8
 		2011-09-09 21:53:04 +00:00
..
ABIInfo.h clang side to match the LLVM IR type system rewrite patch. 2011-07-09 17:41:47 +00:00
BackendUtil.cpp [driver] Add -mglobal-merge/-mno-global-merge machine options to enable/disable merging of 2011-08-26 00:26:29 +00:00
CGBlocks.cpp Treat the weak export of block runtime symbols as a deployment-target 2011-09-09 20:41:01 +00:00
CGBlocks.h de-constify llvm::Type, patch by David Blaikie! 2011-07-18 04:24:23 +00:00
CGBuilder.h Make CGBuilderTy a typedef again; its functionality has been rolled back 2010-07-06 18:43:48 +00:00
CGBuiltin.cpp Switch clang over to using fence/atomicrmw/cmpxchg instead of the intrinsics (which will go away). LLVM CodeGen does almost exactly the same thing with these and the old intrinsics, so I'm reasonably confident this will not break anything. 2011-09-07 01:41:24 +00:00
CGCXX.cpp Disable the optimization that skips emission of complete, non-virtual 2011-07-26 23:18:30 +00:00
CGCXXABI.cpp remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports 2011-07-23 10:55:15 +00:00
CGCXXABI.h clean up forward declarations of raw_ostream to use the new LLVM.h 2011-07-23 10:35:09 +00:00
CGCall.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGCall.h now that we have a centralized place to do so, add some using declarations for 2011-07-20 06:58:45 +00:00
CGClass.cpp Declare and define implicit move constructor and assignment operator. 2011-08-30 19:58:05 +00:00
CGCleanup.cpp Simplify EH control flow by observing that EH scopes form a simple 2011-08-11 02:22:43 +00:00
CGCleanup.h Simplify EH control flow by observing that EH scopes form a simple 2011-08-11 02:22:43 +00:00
CGDebugInfo.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGDebugInfo.h Use new DIBuilder::finalize() at the end to wrap up debug info encoding for a translation unit. 2011-08-15 23:01:55 +00:00
CGDecl.cpp Track whether an AggValueSlot is potentially aliased, and do not 2011-08-25 23:04:34 +00:00
CGDeclCXX.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGException.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGException.h now that we have a centralized place to do so, add some using declarations for 2011-07-20 06:58:45 +00:00
CGExpr.cpp Give conversions of block pointers to ObjC pointers a different cast kind 2011-09-09 05:25:32 +00:00
CGExprAgg.cpp Give conversions of block pointers to ObjC pointers a different cast kind 2011-09-09 05:25:32 +00:00
CGExprCXX.cpp Rearrange code so that we pass the right pointer to delete[] when an exception is thrown constructing the array elements in an array new expression. Fixes PR10870. 2011-09-06 18:53:03 +00:00
CGExprComplex.cpp Give conversions of block pointers to ObjC pointers a different cast kind 2011-09-09 05:25:32 +00:00
CGExprConstant.cpp Give conversions of block pointers to ObjC pointers a different cast kind 2011-09-09 05:25:32 +00:00
CGExprScalar.cpp Give conversions of block pointers to ObjC pointers a different cast kind 2011-09-09 05:25:32 +00:00
CGObjC.cpp Give conversions of block pointers to ObjC pointers a different cast kind 2011-09-09 05:25:32 +00:00
CGObjCGNU.cpp switch clang to use the new-new way of creating llvm::StructType's. 2011-08-12 17:43:31 +00:00
CGObjCMac.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGObjCRuntime.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGObjCRuntime.h Support for catching objc pointer objects in c++ catch-statement 2011-06-23 19:00:08 +00:00
CGRTTI.cpp CodeGen: rename CodeGenModule::Runtime to ObjCRuntime 2011-07-27 20:29:46 +00:00
CGRecordLayout.h clean up forward declarations of raw_ostream to use the new LLVM.h 2011-07-23 10:35:09 +00:00
CGRecordLayoutBuilder.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGStmt.cpp Track whether an AggValueSlot is potentially aliased, and do not 2011-08-25 23:04:34 +00:00
CGTemporaries.cpp Generalize Cleanup::Emit's "isForEH" parameter into a set 2011-07-12 20:27:29 +00:00
CGVTT.cpp remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports 2011-07-23 10:55:15 +00:00
CGVTables.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CGVTables.h now that we have a centralized place to do so, add some using declarations for 2011-07-20 06:58:45 +00:00
CGValue.h What say we document some of these AggValueSlot flags a bit 2011-08-26 08:02:37 +00:00
CMakeLists.txt Wire up the -ftest-coverage and -fprofile-arcs flags to .gcno file emission (at 2011-04-21 23:44:07 +00:00
CodeGenAction.cpp remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports 2011-07-23 10:55:15 +00:00
CodeGenFunction.cpp Carry the debug information from single exit unified return block 2011-09-09 21:53:04 +00:00
CodeGenFunction.h Track whether an AggValueSlot is potentially aliased, and do not 2011-08-25 23:04:34 +00:00
CodeGenModule.cpp Treat the weak export of block runtime symbols as a deployment-target 2011-09-09 20:41:01 +00:00
CodeGenModule.h Do a lookup for the blocks runtime globals to see if they were declared, 2011-09-09 06:44:21 +00:00
CodeGenTBAA.cpp remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports 2011-07-23 10:55:15 +00:00
CodeGenTBAA.h now that we have a centralized place to do so, add some using declarations for 2011-07-20 06:58:45 +00:00
CodeGenTypes.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
CodeGenTypes.h now that we have a centralized place to do so, add some using declarations for 2011-07-20 06:58:45 +00:00
ItaniumCXXABI.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
Makefile BUILD_ARCHIVE is the default for libraries, no need to set it. 2010-07-18 00:14:47 +00:00
MicrosoftCXXABI.cpp remove unneeded llvm:: namespace qualifiers on some core types now that LLVM.h imports 2011-07-23 10:55:15 +00:00
ModuleBuilder.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
README.txt These IRgen improvements have been done. 2009-07-23 03:03:07 +00:00
TargetInfo.cpp Extend the ASTContext constructor to delay the initialization of 2011-09-02 00:18:52 +00:00
TargetInfo.h The size of struct UnwindException varies by platform with no 2011-08-30 01:42:09 +00:00

README.txt 

IRgen optimization opportunities.

//===---------------------------------------------------------------------===//

The common pattern of
--
short x; // or char, etc
(x == 10)
--
generates an zext/sext of x which can easily be avoided.

//===---------------------------------------------------------------------===//

Bitfields accesses can be shifted to simplify masking and sign
extension. For example, if the bitfield width is 8 and it is
appropriately aligned then is is a lot shorter to just load the char
directly.

//===---------------------------------------------------------------------===//

It may be worth avoiding creation of alloca's for formal arguments
for the common situation where the argument is never written to or has
its address taken. The idea would be to begin generating code by using
the argument directly and if its address is taken or it is stored to
then generate the alloca and patch up the existing code.

In theory, the same optimization could be a win for block local
variables as long as the declaration dominates all statements in the
block.

NOTE: The main case we care about this for is for -O0 -g compile time
performance, and in that scenario we will need to emit the alloca
anyway currently to emit proper debug info. So this is blocked by
being able to emit debug information which refers to an LLVM
temporary, not an alloca.

//===---------------------------------------------------------------------===//

We should try and avoid generating basic blocks which only contain
jumps. At -O0, this penalizes us all the way from IRgen (malloc &
instruction overhead), all the way down through code generation and
assembly time.

On 176.gcc:expr.ll, it looks like over 12% of basic blocks are just
direct branches!

//===---------------------------------------------------------------------===//