Radar 9012638.

llvm-svn: 126127

but which is responsible for us doing really bad things to 256.bzip2.

llvm-svn: 126126

llvm-svn: 126125

should be that if the phi is used by a side-effect free instruction with
no uses then the phi and the instruction now get zapped (checked by the
unittest).

llvm-svn: 126124

llvm-svn: 126116

when checking an initialization.

llvm-svn: 126115

logic from CXXMemberCallExpr and by making it check for
CXXOperatorCallExpr in order to defer.  This is not really an awesome solution,
but I don't have a better idea.

llvm-svn: 126114

for ?: on record types.

llvm-svn: 126113

"dllimport" function must not be GlobalVariable, but Function. It is enough to check with GlobalValue.
test/CodeGen/X86/dll-linkage.ll is updated to check llc -O0.

llvm-svn: 126110

llvm-svn: 126108

CMake: Define LLVM_INCLUDE_TESTS in cmake/modules/HandleLLVMOptions.cmake. It affects testing stuff on clang-standalone build.

llvm-svn: 126107

llvm-svn: 126106

CMake: Eliminate duplicate -D__STDC_LIMIT_MACROS and -D__STDC_CONSTANT_MACROS. llvm/cmake/modules/HandleLLVMOptions.cmake defines them.

llvm-svn: 126105

for consistency with other tests (and to remove "scalar" from the name).

llvm-svn: 126104

llvm-svn: 126103

llvm-svn: 126102

llvm-svn: 126101

on Core 2 and Nehalem, so the code we generate is better than GCC's here.

llvm-svn: 126100

independent vectors.

llvm-svn: 126099

appropriate attribute. Add a bit more testing that finds a pretty bad
regression (since ~forever) in this warning. Fix it with a nice 2 line
change. =]

llvm-svn: 126098

of a constant had a minor typo introduced when copying it from the book, which
caused it to favor negative approximations over positive approximations in many
cases. Positive approximations require fewer operations beyond the multiplication.

In the case of division by 3, we still generate code that is a single instruction
larger than GCC's code.

llvm-svn: 126097

-Wformat-security on -Wformat, not vice-versa.

Fixes PR8486. Patch by Oleg Slezberg.

llvm-svn: 126096

Requires CMake 2.8.3 or newer.

llvm-svn: 126094

llvm-svn: 126093

Requires CMake 2.8.3 or newer.

llvm-svn: 126092

PR9245.

llvm-svn: 126091

CMake complained about this while generating VS project files but was
okay with it while generating makefiles on Linux.

llvm-svn: 126090

used by the linker and not by nm or ar.

llvm-svn: 126089

test for that. With this change, test/CodeGen/X86/codegen-dce.ll no longer finds
any instructions to DCE, so delete the test.

Also renamed J and JP to I and IP in RecursivelyDeleteDeadPHINode.

llvm-svn: 126088

llvm-svn: 126087

libxul links in 7m0.403s.

llvm-svn: 126085

llvm-svn: 126084

Visual Studio.

llvm-svn: 126083

llvm-svn: 126082

We usually catch this kind of optimization through InstSimplify's distributive
magic, but or doesn't distribute over xor in general.

"A | ~(A | B) -> A | ~B" hits 24 times on gcc.c.

llvm-svn: 126081

The DAGCombiner folds the zext into complex load instructions. This patch
prevents this optimization on vectors since none of the supported targets
knows how to perform load+vector_zext in one instruction.

llvm-svn: 126080

llvm-svn: 126078

by Andrew Clinton!

llvm-svn: 126077

one Value set. This is faster because we only need to use the set when there
isn't already an entry in the map. No functionality change!

llvm-svn: 126076

llvm-svn: 126073