llvm-project/clang/lib/CodeGen
Eli Friedman 8f5f4ea466 Change the way clang generates union types a bit so it plays well
with the LLVM C backend.

llvm-svn: 65615
2009-02-27 04:53:40 +00:00
..
ABIInfo.h Pull CodeGenFunction::EmitVAArg into target specific ABIInfo classes. 2009-02-10 21:44:36 +00:00
CGBlocks.cpp Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
CGBuilder.h Disable generation of basic block names in NDEBUG mode. 2008-11-12 00:01:12 +00:00
CGBuiltin.cpp Support IRgen of sqrt -> llvm.sqrt, pow -> llvm.pow. 2009-02-16 22:43:43 +00:00
CGCXX.cpp Pull COdeGenFunction::CreateStaticBlockVarDecl (just for creating the 2009-02-25 19:24:29 +00:00
CGCall.cpp Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
CGCall.h Unbreak CGFunctionInfo::Profile method and reenable caching of ABI 2009-02-05 00:00:23 +00:00
CGDebugInfo.cpp Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
CGDebugInfo.h Add support to emit debug info for objective-c interfaces. 2009-02-26 21:10:26 +00:00
CGDecl.cpp Allow constant initializers to reference their defining decl. 2009-02-25 20:08:33 +00:00
CGExpr.cpp first wave of fixes for @encode sema support. This is part of PR3648. 2009-02-24 22:18:39 +00:00
CGExprAgg.cpp rip out __builtin_overload 2009-02-18 22:14:55 +00:00
CGExprComplex.cpp rip out __builtin_overload 2009-02-18 22:14:55 +00:00
CGExprConstant.cpp CodeGen support for copied BlockDeclRefExprs. 2009-02-25 23:33:13 +00:00
CGExprScalar.cpp CodeGen support for copied BlockDeclRefExprs. 2009-02-25 23:33:13 +00:00
CGObjC.cpp Enable debug info emission for objc methods. 2009-02-25 01:09:46 +00:00
CGObjCGNU.cpp Some refactoring of Ivar offset code gen. 2009-02-10 19:02:04 +00:00
CGObjCMac.cpp Couple of meta-data segments were wrong. This patch fixes them. 2009-02-26 22:30:39 +00:00
CGObjCRuntime.h Some refactoring of Ivar offset code gen. 2009-02-10 19:02:04 +00:00
CGStmt.cpp Make sure to reset the DidCallStackSave variable before emitting a compound statement. Fixes PR3649. 2009-02-22 18:44:21 +00:00
CGValue.h Handle case of none gc'able objects regardless of their 2009-02-21 00:30:43 +00:00
CMakeLists.txt Fix cmake builds. 2009-02-13 15:42:50 +00:00
CodeGenFunction.cpp Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
CodeGenFunction.h Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
CodeGenModule.cpp Minor cleanup: use getDeclAlignInBytes helper. 2009-02-27 04:11:37 +00:00
CodeGenModule.h first wave of fixes for @encode sema support. This is part of PR3648. 2009-02-24 22:18:39 +00:00
CodeGenTypes.cpp Change the way clang generates union types a bit so it plays well 2009-02-27 04:53:40 +00:00
CodeGenTypes.h Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
Makefile Make a major restructuring of the clang tree: introduce a top-level 2008-03-15 23:59:48 +00:00
Mangle.cpp Create a new TypeNodes.def file that enumerates all of the types, 2009-02-26 23:50:07 +00:00
Mangle.h Address Chris's comments regarding C++ name mangling. 2009-02-18 23:53:56 +00:00
ModuleBuilder.cpp Remove some redundant Decl -> Decl castings. 2009-02-17 20:23:54 +00:00
README.txt Add a note about an IRgen optimization opportunity. 2009-02-24 06:34:04 +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!

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

There are some more places where we could avoid generating unreachable code. For
example:
  void f0(int a) { abort(); if (a) printf("hi"); }
still generates a call to printf. This doesn't occur much in real
code, but would still be nice to clean up.

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