maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/combine-srem.ll

164 lines
4.8 KiB
LLVM
Raw Normal View History

[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
[X86] Add common CHECK prefix for tests without SSE/AVX codegen llvm-svn: 321810
2018-01-05 02:23:46 +08:00
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX2
[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
[X86] Add srem/udiv/urem by one combine tests llvm-svn: 321826
2018-01-05 06:08:36 +08:00
; fold (srem x, 1) -> 0
define i32 @combine_srem_by_one(i32 %x) {
; CHECK-LABEL: combine_srem_by_one:
; CHECK: # %bb.0:
; CHECK-NEXT: xorl %eax, %eax
; CHECK-NEXT: retq
%1 = srem i32 %x, 1
ret i32 %1
}
define <4 x i32> @combine_vec_srem_by_one(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_by_one:
; SSE: # %bb.0:
; SSE-NEXT: xorps %xmm0, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_srem_by_one:
; AVX: # %bb.0:
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = srem <4 x i32> %x, <i32 1, i32 1, i32 1, i32 1>
ret <4 x i32> %1
}
[X86] Show missed combine for X/X for SDIV/UDIV and X%X for SREM/UREM llvm-svn: 321807
2018-01-05 02:20:46 +08:00
; TODO fold (srem x, x) -> 0
define i32 @combine_srem_dupe(i32 %x) {
[X86] Add common CHECK prefix for tests without SSE/AVX codegen llvm-svn: 321810
2018-01-05 02:23:46 +08:00
; CHECK-LABEL: combine_srem_dupe:
; CHECK: # %bb.0:
; CHECK-NEXT: movl %edi, %eax
; CHECK-NEXT: cltd
; CHECK-NEXT: idivl %edi
; CHECK-NEXT: movl %edx, %eax
; CHECK-NEXT: retq
[X86] Show missed combine for X/X for SDIV/UDIV and X%X for SREM/UREM llvm-svn: 321807
2018-01-05 02:20:46 +08:00
%1 = srem i32 %x, %x
ret i32 %1
}
define <4 x i32> @combine_vec_srem_dupe(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_dupe:
; SSE: # %bb.0:
; SSE-NEXT: pextrd $1, %xmm0, %ecx
; SSE-NEXT: movl %ecx, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %ecx
; SSE-NEXT: movl %edx, %ecx
; SSE-NEXT: movd %xmm0, %esi
; SSE-NEXT: movl %esi, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %esi
; SSE-NEXT: movd %edx, %xmm1
; SSE-NEXT: pinsrd $1, %ecx, %xmm1
; SSE-NEXT: pextrd $2, %xmm0, %ecx
; SSE-NEXT: movl %ecx, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %ecx
; SSE-NEXT: pinsrd $2, %edx, %xmm1
; SSE-NEXT: pextrd $3, %xmm0, %ecx
; SSE-NEXT: movl %ecx, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %ecx
; SSE-NEXT: pinsrd $3, %edx, %xmm1
; SSE-NEXT: movdqa %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_srem_dupe:
; AVX: # %bb.0:
; AVX-NEXT: vpextrd $1, %xmm0, %ecx
; AVX-NEXT: movl %ecx, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %ecx
; AVX-NEXT: movl %edx, %ecx
; AVX-NEXT: vmovd %xmm0, %esi
; AVX-NEXT: movl %esi, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %esi
; AVX-NEXT: vmovd %edx, %xmm1
; AVX-NEXT: vpinsrd $1, %ecx, %xmm1, %xmm1
; AVX-NEXT: vpextrd $2, %xmm0, %ecx
; AVX-NEXT: movl %ecx, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %ecx
; AVX-NEXT: vpinsrd $2, %edx, %xmm1, %xmm1
; AVX-NEXT: vpextrd $3, %xmm0, %ecx
; AVX-NEXT: movl %ecx, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %ecx
; AVX-NEXT: vpinsrd $3, %edx, %xmm1, %xmm0
; AVX-NEXT: retq
%1 = srem <4 x i32> %x, %x
ret <4 x i32> %1
}
[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
; fold (srem x, y) -> (urem x, y) iff x and y are positive
define <4 x i32> @combine_vec_srem_by_pos0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_by_pos0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[DAGCombiner] Enable srem(x.y) -> urem(x,y) combine for vectors SelectionDAG::SignBitIsZero (via SelectionDAG::computeKnownBits) has supported vectors since rL280927 llvm-svn: 285123
2016-10-26 05:20:18 +08:00
; SSE-NEXT: andps {{.*}}(%rip), %xmm0
[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
; SSE-NEXT: retq
;
[X86][SSE] Add AVX1 tests to sdiv/udiv srem/urem combine tests As requested on D27714 llvm-svn: 289652
2016-12-14 22:39:51 +08:00
; AVX1-LABEL: combine_vec_srem_by_pos0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX1: # %bb.0:
[X86][SSE] Add AVX1 tests to sdiv/udiv srem/urem combine tests As requested on D27714 llvm-svn: 289652
2016-12-14 22:39:51 +08:00
; AVX1-NEXT: vandps {{.*}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: combine_vec_srem_by_pos0:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX2: # %bb.0:
Regenerate broadcast constant comment llvm-svn: 321808
2018-01-05 02:21:33 +08:00
; AVX2-NEXT: vbroadcastss {{.*#+}} xmm1 = [3,3,3,3]
[X86][SSE] Add AVX1 tests to sdiv/udiv srem/urem combine tests As requested on D27714 llvm-svn: 289652
2016-12-14 22:39:51 +08:00
; AVX2-NEXT: vandps %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
%1 = and <4 x i32> %x, <i32 255, i32 255, i32 255, i32 255>
%2 = srem <4 x i32> %1, <i32 4, i32 4, i32 4, i32 4>
ret <4 x i32> %2
}
define <4 x i32> @combine_vec_srem_by_pos1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_by_pos1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; SSE: # %bb.0:
[DAGCombiner] Try to use SelectionDAG::isKnownToBeAPowerOfTwo instead of just APInt::isPowerOf2 Generalize sdiv/udiv/srem/urem combines using APInt::isPowerOf2, which only works for const/splat-const values, to call SelectionDAG::isKnownToBeAPowerOfTwo instead which recognises many more cases. Added a DAGCombiner::BuildLogBase2 helper since PowerOf2 combines often involve taking the log2 of such a value. Differential Revision: https://reviews.llvm.org/D27714 llvm-svn: 289654
2016-12-14 23:08:13 +08:00
; SSE-NEXT: andps {{.*}}(%rip), %xmm0
[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_srem_by_pos1:
[CodeGen] Unify MBB reference format in both MIR and debug output As part of the unification of the debug format and the MIR format, print MBB references as '%bb.5'. The MIR printer prints the IR name of a MBB only for block definitions. * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g' * find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g' * find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g' * grep -nr 'BB#' and fix Differential Revision: https://reviews.llvm.org/D40422 llvm-svn: 319665
2017-12-05 01:18:51 +08:00
; AVX: # %bb.0:
[DAGCombiner] Try to use SelectionDAG::isKnownToBeAPowerOfTwo instead of just APInt::isPowerOf2 Generalize sdiv/udiv/srem/urem combines using APInt::isPowerOf2, which only works for const/splat-const values, to call SelectionDAG::isKnownToBeAPowerOfTwo instead which recognises many more cases. Added a DAGCombiner::BuildLogBase2 helper since PowerOf2 combines often involve taking the log2 of such a value. Differential Revision: https://reviews.llvm.org/D27714 llvm-svn: 289654
2016-12-14 23:08:13 +08:00
; AVX-NEXT: vandps {{.*}}(%rip), %xmm0, %xmm0
[X86][SSE] Added vector srem combine tests llvm-svn: 285121
2016-10-26 05:14:11 +08:00
; AVX-NEXT: retq
%1 = and <4 x i32> %x, <i32 255, i32 255, i32 255, i32 255>
%2 = srem <4 x i32> %1, <i32 1, i32 4, i32 8, i32 16>
ret <4 x i32> %2
}
[DAGCombine] visitREM - Don't assume that one divrem isn't driving another Under some circumstances the divrems won't have been combined together before getting to this code. So replace the assertion with a if() guard to not expand to X-((X/C)*C) to give the other combine chance to happen. Reduced from OSS-Fuzz #6883 https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=6883 llvm-svn: 327424
2018-03-14 01:17:15 +08:00
; OSS-Fuzz #6883
; https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=6883
define i32 @ossfuzz6883() {
; CHECK-LABEL: ossfuzz6883:
; CHECK: # %bb.0:
; CHECK-NEXT: movl (%rax), %ecx
; CHECK-NEXT: movl %ecx, %eax
; CHECK-NEXT: cltd
; CHECK-NEXT: idivl %ecx
; CHECK-NEXT: movl %edx, %esi
; CHECK-NEXT: movl $1, %edi
; CHECK-NEXT: cltd
; CHECK-NEXT: idivl %edi
; CHECK-NEXT: movl %edx, %edi
; CHECK-NEXT: xorl %edx, %edx
; CHECK-NEXT: movl %ecx, %eax
; CHECK-NEXT: divl %edi
; CHECK-NEXT: andl %esi, %eax
; CHECK-NEXT: retq
%B17 = or i32 0, 2147483647
%L6 = load i32, i32* undef
%B11 = sdiv i32 %L6, %L6
%B13 = udiv i32 %B17, %B17
%B14 = srem i32 %B11, %B13
%B16 = srem i32 %L6, %L6
%B10 = udiv i32 %L6, %B14
%B6 = and i32 %B16, %B10
ret i32 %B6
}