diff options
| author | Evan Cheng <evan.cheng@apple.com> | 2008-09-04 22:59:58 +0000 |
|---|---|---|
| committer | Evan Cheng <evan.cheng@apple.com> | 2008-09-04 22:59:58 +0000 |
| commit | 4a03775777785ef758cff8c0699a6bf571a1f2b9 (patch) | |
| tree | a3e8d527e079840bf5627f2e68f73f901728eb1e /test/CodeGen/X86/combine-lds.ll | |
| parent | deafefabae915cf20884fd829d2542f9626786e8 (diff) | |
| download | llvm-4a03775777785ef758cff8c0699a6bf571a1f2b9.tar.gz llvm-4a03775777785ef758cff8c0699a6bf571a1f2b9.tar.bz2 llvm-4a03775777785ef758cff8c0699a6bf571a1f2b9.tar.xz | |
For whatever the reason, x86 CallingConv::Fast (i.e. fastcc) was not passing scalar arguments in registers. This patch defines a new fastcc CC which is slightly different from the FastCall CC. In addition to passing integer arguments in ECX and EDX, it also specify doubles are passed in 8-byte slots which are 8-byte aligned (instead of 4-byte aligned). This avoids a potential performance hazard where doubles span cacheline boundaries.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@55807 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/combine-lds.ll')
| -rw-r--r-- | test/CodeGen/X86/combine-lds.ll | 5 |
1 files changed, 2 insertions, 3 deletions
diff --git a/test/CodeGen/X86/combine-lds.ll b/test/CodeGen/X86/combine-lds.ll index 5e0ad99667..a78a042d7e 100644 --- a/test/CodeGen/X86/combine-lds.ll +++ b/test/CodeGen/X86/combine-lds.ll @@ -1,7 +1,6 @@ -; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep movsd -; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep mov | count 1 +; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep fldl | count 1 -define fastcc double @doload64(i64 %x) nounwind { +define double @doload64(i64 %x) nounwind { %tmp717 = bitcast i64 %x to double ret double %tmp717 } |