llvm-project/clang/lib/CodeGen
Owen Anderson ade90fd1ba Update for LLVM API change.
llvm-svn: 77492
2009-07-29 18:54:39 +00:00
..
ABIInfo.h Update for LLVM API change, and contextify a bunch of related stuff. 2009-07-14 23:10:40 +00:00
CGBlocks.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CGBlocks.h Remove ASTContext::isObjCObjectPointerType(). 2009-07-16 15:41:00 +00:00
CGBuilder.h Disable generation of basic block names in NDEBUG mode. 2008-11-12 00:01:12 +00:00
CGBuiltin.cpp Update for LLVM API change. 2009-07-27 21:00:51 +00:00
CGCXX.cpp Use the existing API for base offset. Use suitable llvm type 2009-07-29 15:54:56 +00:00
CGCXX.h Add CGCXX.h with ctor/dtor type enumerations. No functionality change. 2009-04-15 04:36:55 +00:00
CGCXXTemp.cpp Update for LLVM API change. 2009-07-21 18:06:41 +00:00
CGCall.cpp Allow functions to be marked "implicit return zero" and so mark main(). 2009-07-28 01:00:58 +00:00
CGCall.h Unbreak CGFunctionInfo::Profile method and reenable caching of ABI 2009-02-05 00:00:23 +00:00
CGDebugInfo.cpp Use DICompositeType->replaceAllUsesWith() 2009-07-27 18:42:03 +00:00
CGDebugInfo.h Revert 75648 for now. It is causing test failures. 2009-07-14 21:31:22 +00:00
CGDecl.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CGExpr.cpp Patch to provide cast of objects in member access 2009-07-29 18:40:24 +00:00
CGExprAgg.cpp Update for LLVM API change. 2009-07-24 23:12:58 +00:00
CGExprComplex.cpp Update for LLVM API change. 2009-07-27 21:00:51 +00:00
CGExprConstant.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CGExprScalar.cpp Code refactoring to define getCXXRecordDeclForPointerType 2009-07-29 00:44:13 +00:00
CGObjC.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CGObjCGNU.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CGObjCMac.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CGObjCRuntime.h fix objc codegen to not have its own list of things that eventually get into llvm.used, just 2009-07-17 23:57:13 +00:00
CGRecordLayoutBuilder.cpp No longer need to keep base class offsets in the offset 2009-07-29 18:50:06 +00:00
CGRecordLayoutBuilder.h More CGRecordLayoutBuilder cleanup. 2009-07-28 17:56:36 +00:00
CGStmt.cpp Update for LLVM API change. 2009-07-24 23:12:58 +00:00
CGValue.h Preserve address space information through member accesses, e.g., 2009-07-22 03:08:17 +00:00
CMakeLists.txt Unbreak the CMake build 2009-07-23 15:15:06 +00:00
CodeGenFunction.cpp Update for LLVM API change. 2009-07-24 23:12:58 +00:00
CodeGenFunction.h Some code cleanup. 2009-07-28 18:09:28 +00:00
CodeGenModule.cpp Update for LLVM API change. 2009-07-29 18:54:39 +00:00
CodeGenModule.h Update for LLVM API change, and contextify a bunch of related stuff. 2009-07-14 23:10:40 +00:00
CodeGenTypes.cpp Remove the old RecordOrganizer. 2009-07-28 21:03:34 +00:00
CodeGenTypes.h Move the LLVM field number for bit fields into the BitFieldInfo structure, since it's meaning is completely different than for non-bit fields. 2009-07-23 17:01:21 +00:00
Makefile Build system changes to use TableGen to generate the various 2009-03-16 23:06:59 +00:00
Mangle.cpp Fix a couple recent ABI regressions noticed during code review (fallout from the ObjC type system rewrite). 2009-07-22 17:14:51 +00:00
Mangle.h Add support for generating (very basic) C++ destructors. These aren't called by anything yet. 2009-04-17 01:58:57 +00:00
ModuleBuilder.cpp Update for changes in LLVM. Hopefully this is the last one for a while. 2009-07-01 23:14:14 +00:00
README.txt These IRgen improvements have been done. 2009-07-23 03:03:07 +00:00
TargetABIInfo.cpp No longer need to keep base class offsets in the offset 2009-07-29 18:50:06 +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!

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