diff options
| author | Bill Wendling <isanbard@gmail.com> | 2011-05-12 19:02:15 +0000 |
|---|---|---|
| committer | Bill Wendling <isanbard@gmail.com> | 2011-05-12 19:02:15 +0000 |
| commit | 0c9643008e5a35cac76cf3419b3308dcad97e53e (patch) | |
| tree | 9f4ed2aec83abb4349e43ed1735c99fce0799931 /lib/CodeGen/CGBuiltin.cpp | |
| parent | 211924b563aa31421836cee7655be729ad02733f (diff) | |
LLVM doesn't always optimize away the four loads from this:
(__m128){ p[0], p[1], p[2], p[3] }
which produces really bad code. This could be done in instcombine, but it's
probably better to do it in the front-end instead.
<rdar://problem/9424836>
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@131237 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/CodeGen/CGBuiltin.cpp')
| -rw-r--r-- | lib/CodeGen/CGBuiltin.cpp | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp index 46546177ae..494dfaeff7 100644 --- a/lib/CodeGen/CGBuiltin.cpp +++ b/lib/CodeGen/CGBuiltin.cpp @@ -2143,6 +2143,8 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID, // If palignr is shifting the pair of vectors more than 32 bytes, emit zero. return llvm::Constant::getNullValue(ConvertType(E->getType())); } + case X86::BI__builtin_ia32_loadups: + case X86::BI__builtin_ia32_loadupd: case X86::BI__builtin_ia32_loaddqu: { const llvm::Type *VecTy = ConvertType(E->getType()); const llvm::Type *IntTy = llvm::IntegerType::get(getLLVMContext(), 128); |