RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-30 07:22:55 +01:00
Code Issues Projects Releases Wiki Activity
ec249f2467
llvm-mirror/test/CodeGen/X86/memset-sse-stack-realignment.ll

78 lines
1.9 KiB
LLVM
Raw Normal View History

X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; Make sure that we realign the stack. Mingw32 uses 4 byte stack alignment, we
; need 16 bytes for SSE and 32 bytes for AVX.
; RUN: llc < %s -mtriple=i386-pc-mingw32 -mcpu=pentium2 | FileCheck %s -check-prefix=NOSSE
; RUN: llc < %s -mtriple=i386-pc-mingw32 -mcpu=pentium3 | FileCheck %s -check-prefix=SSE1
; RUN: llc < %s -mtriple=i386-pc-mingw32 -mcpu=yonah | FileCheck %s -check-prefix=SSE2
; RUN: llc < %s -mtriple=i386-pc-mingw32 -mcpu=corei7-avx | FileCheck %s -check-prefix=AVX1
; RUN: llc < %s -mtriple=i386-pc-mingw32 -mcpu=core-avx2 | FileCheck %s -check-prefix=AVX2
define void @test1(i32 %t) nounwind {
%tmp1210 = alloca i8, i32 32, align 4
call void @llvm.memset.p0i8.i64(i8* %tmp1210, i8 0, i64 32, i32 4, i1 false)
%x = alloca i8, i32 %t
call void @dummy(i8* %x)
ret void
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; NOSSE-LABEL: test1:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; NOSSE-NOT: and
; NOSSE: movl $0
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; SSE1-LABEL: test1:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; SSE1: andl $-16
; SSE1: movl %esp, %esi
; SSE1: movaps
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; SSE2-LABEL: test1:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; SSE2: andl $-16
; SSE2: movl %esp, %esi
; SSE2: movaps
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; AVX1-LABEL: test1:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; AVX1: andl $-32
; AVX1: movl %esp, %esi
; AVX1: vmovaps %ymm
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; AVX2-LABEL: test1:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; AVX2: andl $-32
; AVX2: movl %esp, %esi
; AVX2: vmovaps %ymm
}
define void @test2(i32 %t) nounwind {
%tmp1210 = alloca i8, i32 16, align 4
call void @llvm.memset.p0i8.i64(i8* %tmp1210, i8 0, i64 16, i32 4, i1 false)
%x = alloca i8, i32 %t
call void @dummy(i8* %x)
ret void
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; NOSSE-LABEL: test2:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; NOSSE-NOT: and
; NOSSE: movl $0
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; SSE1-LABEL: test2:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; SSE1: andl $-16
; SSE1: movl %esp, %esi
; SSE1: movaps
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; SSE2-LABEL: test2:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; SSE2: andl $-16
; SSE2: movl %esp, %esi
; SSE2: movaps
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; AVX1-LABEL: test2:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; AVX1: andl $-16
; AVX1: movl %esp, %esi
; AVX1: vmovaps %xmm
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
; AVX2-LABEL: test2:
X86: Enable SSE memory intrinsics even when stack alignment is less than 16 bytes. The stack realignment code was fixed to work when there is stack realignment and a dynamic alloca is present so this shouldn't cause correctness issues anymore. Note that this also enables generation of AVX instructions for memset under the assumptions: - Unaligned loads/stores are always fast on CPUs supporting AVX - AVX is not slower than SSE We may need some tweaked heuristics if one of those assumptions turns out not to be true. Effectively reverts r58317. Part of PR2962. llvm-svn: 167967
2012-11-14 21:08:40 +01:00
; AVX2: andl $-16
; AVX2: movl %esp, %esi
; AVX2: vmovaps %xmm
}
declare void @dummy(i8*)
declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) nounwind
Reference in New Issue Copy Permalink