maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/InstCombine/canonicalize-lack-of-signed...

235 lines
7.5 KiB
LLVM
Raw Normal View History

[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -instcombine -S | FileCheck %s
; https://bugs.llvm.org/show_bug.cgi?id=38149
; Pattern:
; ((%x << MaskedBits) a>> MaskedBits) == %x
; Should be transformed into:
; (add %x, (1 << (KeptBits-1))) u< (1 << KeptBits)
; Where KeptBits = bitwidth(%x) - MaskedBits
; ============================================================================ ;
; Basic positive tests
; ============================================================================ ;
define i1 @p0(i8 %x) {
; CHECK-LABEL: @p0(
[InstCombine] Re-commit: Fold 'check for [no] signed truncation' pattern Summary: [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 This transform is surprisingly frustrating. This does not deal with non-splat shift amounts, or with undef shift amounts. I've outlined what i think the solution should be: ``` // Potential handling of non-splats: for each element: // * if both are undef, replace with constant 0. // Because (1<<0) is OK and is 1, and ((1<<0)>>1) is also OK and is 0. // * if both are not undef, and are different, bailout. // * else, only one is undef, then pick the non-undef one. ``` This is a re-commit, as the original patch, committed in rL337190 was reverted in rL337344 as it broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 Proofs that the fixed folds are ok: https://rise4fun.com/Alive/VYM Differential Revision: https://reviews.llvm.org/D49320 llvm-svn: 337376
2018-07-18 18:55:17 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = add i8 [[X:%.*]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i8 [[TMP1]], 8
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: ret i1 [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
%tmp1 = ashr exact i8 %tmp0, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq i8 %tmp1, %x
ret i1 %tmp2
}
[NFC][InstCombine] i65 tests for 'check for [no] signed truncation' pattern Those initially broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 llvm-svn: 337364
2018-07-18 16:49:51 +08:00
; Big unusual bit width, https://bugs.llvm.org/show_bug.cgi?id=38204
define i1 @pb(i65 %x) {
; CHECK-LABEL: @pb(
[InstCombine] Re-commit: Fold 'check for [no] signed truncation' pattern Summary: [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 This transform is surprisingly frustrating. This does not deal with non-splat shift amounts, or with undef shift amounts. I've outlined what i think the solution should be: ``` // Potential handling of non-splats: for each element: // * if both are undef, replace with constant 0. // Because (1<<0) is OK and is 1, and ((1<<0)>>1) is also OK and is 0. // * if both are not undef, and are different, bailout. // * else, only one is undef, then pick the non-undef one. ``` This is a re-commit, as the original patch, committed in rL337190 was reverted in rL337344 as it broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 Proofs that the fixed folds are ok: https://rise4fun.com/Alive/VYM Differential Revision: https://reviews.llvm.org/D49320 llvm-svn: 337376
2018-07-18 18:55:17 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = add i65 [[X:%.*]], 9223372036854775808
; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt i65 [[TMP1]], -1
[NFC][InstCombine] i65 tests for 'check for [no] signed truncation' pattern Those initially broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 llvm-svn: 337364
2018-07-18 16:49:51 +08:00
; CHECK-NEXT: ret i1 [[TMP2]]
;
%tmp0 = shl i65 %x, 1
%tmp1 = ashr exact i65 %tmp0, 1
%tmp2 = icmp eq i65 %x, %tmp1
ret i1 %tmp2
}
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; ============================================================================ ;
; Vector tests
; ============================================================================ ;
define <2 x i1> @p1_vec_splat(<2 x i8> %x) {
; CHECK-LABEL: @p1_vec_splat(
[InstCombine] Re-commit: Fold 'check for [no] signed truncation' pattern Summary: [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 This transform is surprisingly frustrating. This does not deal with non-splat shift amounts, or with undef shift amounts. I've outlined what i think the solution should be: ``` // Potential handling of non-splats: for each element: // * if both are undef, replace with constant 0. // Because (1<<0) is OK and is 1, and ((1<<0)>>1) is also OK and is 0. // * if both are not undef, and are different, bailout. // * else, only one is undef, then pick the non-undef one. ``` This is a re-commit, as the original patch, committed in rL337190 was reverted in rL337344 as it broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 Proofs that the fixed folds are ok: https://rise4fun.com/Alive/VYM Differential Revision: https://reviews.llvm.org/D49320 llvm-svn: 337376
2018-07-18 18:55:17 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = add <2 x i8> [[X:%.*]], <i8 4, i8 4>
; CHECK-NEXT: [[TMP2:%.*]] = icmp ult <2 x i8> [[TMP1]], <i8 8, i8 8>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: ret <2 x i1> [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl <2 x i8> %x, <i8 5, i8 5>
%tmp1 = ashr exact <2 x i8> %tmp0, <i8 5, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq <2 x i8> %tmp1, %x
ret <2 x i1> %tmp2
}
define <2 x i1> @p2_vec_nonsplat(<2 x i8> %x) {
; CHECK-LABEL: @p2_vec_nonsplat(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl <2 x i8> [[X:%.*]], <i8 5, i8 6>
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact <2 x i8> [[TMP0]], <i8 5, i8 6>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i8> [[TMP1]], [[X]]
; CHECK-NEXT: ret <2 x i1> [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl <2 x i8> %x, <i8 5, i8 6>
%tmp1 = ashr exact <2 x i8> %tmp0, <i8 5, i8 6>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq <2 x i8> %tmp1, %x
ret <2 x i1> %tmp2
}
define <3 x i1> @p3_vec_undef0(<3 x i8> %x) {
; CHECK-LABEL: @p3_vec_undef0(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl <3 x i8> [[X:%.*]], <i8 5, i8 undef, i8 5>
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact <3 x i8> [[TMP0]], <i8 5, i8 5, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <3 x i8> [[TMP1]], [[X]]
; CHECK-NEXT: ret <3 x i1> [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl <3 x i8> %x, <i8 5, i8 undef, i8 5>
%tmp1 = ashr exact <3 x i8> %tmp0, <i8 5, i8 5, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq <3 x i8> %tmp1, %x
ret <3 x i1> %tmp2
}
define <3 x i1> @p4_vec_undef1(<3 x i8> %x) {
; CHECK-LABEL: @p4_vec_undef1(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl <3 x i8> [[X:%.*]], <i8 5, i8 5, i8 5>
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact <3 x i8> [[TMP0]], <i8 5, i8 undef, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <3 x i8> [[TMP1]], [[X]]
; CHECK-NEXT: ret <3 x i1> [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl <3 x i8> %x, <i8 5, i8 5, i8 5>
%tmp1 = ashr exact <3 x i8> %tmp0, <i8 5, i8 undef, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq <3 x i8> %tmp1, %x
ret <3 x i1> %tmp2
}
define <3 x i1> @p5_vec_undef2(<3 x i8> %x) {
; CHECK-LABEL: @p5_vec_undef2(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl <3 x i8> [[X:%.*]], <i8 5, i8 undef, i8 5>
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact <3 x i8> [[TMP0]], <i8 5, i8 undef, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <3 x i8> [[TMP1]], [[X]]
; CHECK-NEXT: ret <3 x i1> [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl <3 x i8> %x, <i8 5, i8 undef, i8 5>
%tmp1 = ashr exact <3 x i8> %tmp0, <i8 5, i8 undef, i8 5>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq <3 x i8> %tmp1, %x
ret <3 x i1> %tmp2
}
; ============================================================================ ;
; Commutativity tests.
; ============================================================================ ;
declare i8 @gen8()
define i1 @c0() {
; CHECK-LABEL: @c0(
; CHECK-NEXT: [[X:%.*]] = call i8 @gen8()
[InstCombine] Re-commit: Fold 'check for [no] signed truncation' pattern Summary: [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 This transform is surprisingly frustrating. This does not deal with non-splat shift amounts, or with undef shift amounts. I've outlined what i think the solution should be: ``` // Potential handling of non-splats: for each element: // * if both are undef, replace with constant 0. // Because (1<<0) is OK and is 1, and ((1<<0)>>1) is also OK and is 0. // * if both are not undef, and are different, bailout. // * else, only one is undef, then pick the non-undef one. ``` This is a re-commit, as the original patch, committed in rL337190 was reverted in rL337344 as it broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 Proofs that the fixed folds are ok: https://rise4fun.com/Alive/VYM Differential Revision: https://reviews.llvm.org/D49320 llvm-svn: 337376
2018-07-18 18:55:17 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = add i8 [[X]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i8 [[TMP1]], 8
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: ret i1 [[TMP2]]
;
%x = call i8 @gen8()
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
%tmp1 = ashr exact i8 %tmp0, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq i8 %x, %tmp1 ; swapped order
ret i1 %tmp2
}
; ============================================================================ ;
; One-use tests.
; ============================================================================ ;
declare void @use8(i8)
define i1 @n_oneuse0(i8 %x) {
; CHECK-LABEL: @n_oneuse0(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl i8 [[X:%.*]], 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: call void @use8(i8 [[TMP0]])
[InstCombine] Re-commit: Fold 'check for [no] signed truncation' pattern Summary: [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 This transform is surprisingly frustrating. This does not deal with non-splat shift amounts, or with undef shift amounts. I've outlined what i think the solution should be: ``` // Potential handling of non-splats: for each element: // * if both are undef, replace with constant 0. // Because (1<<0) is OK and is 1, and ((1<<0)>>1) is also OK and is 0. // * if both are not undef, and are different, bailout. // * else, only one is undef, then pick the non-undef one. ``` This is a re-commit, as the original patch, committed in rL337190 was reverted in rL337344 as it broke chromium build: https://bugs.llvm.org/show_bug.cgi?id=38204 and https://crbug.com/864832 Proofs that the fixed folds are ok: https://rise4fun.com/Alive/VYM Differential Revision: https://reviews.llvm.org/D49320 llvm-svn: 337376
2018-07-18 18:55:17 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = add i8 [[X]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i8 [[TMP1]], 8
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: ret i1 [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
call void @use8(i8 %tmp0)
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp1 = ashr exact i8 %tmp0, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq i8 %tmp1, %x
ret i1 %tmp2
}
define i1 @n_oneuse1(i8 %x) {
; CHECK-LABEL: @n_oneuse1(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl i8 [[X:%.*]], 5
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact i8 [[TMP0]], 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: call void @use8(i8 [[TMP1]])
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i8 [[TMP1]], [[X]]
; CHECK-NEXT: ret i1 [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
%tmp1 = ashr exact i8 %tmp0, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
call void @use8(i8 %tmp1)
%tmp2 = icmp eq i8 %tmp1, %x
ret i1 %tmp2
}
define i1 @n_oneuse2(i8 %x) {
; CHECK-LABEL: @n_oneuse2(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl i8 [[X:%.*]], 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: call void @use8(i8 [[TMP0]])
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact i8 [[TMP0]], 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: call void @use8(i8 [[TMP1]])
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i8 [[TMP1]], [[X]]
; CHECK-NEXT: ret i1 [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
call void @use8(i8 %tmp0)
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp1 = ashr exact i8 %tmp0, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
call void @use8(i8 %tmp1)
%tmp2 = icmp eq i8 %tmp1, %x
ret i1 %tmp2
}
; ============================================================================ ;
; Negative tests
; ============================================================================ ;
define i1 @n0(i8 %x) {
; CHECK-LABEL: @n0(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl i8 [[X:%.*]], 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact i8 [[TMP0]], 3
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i8 [[TMP1]], [[X]]
; CHECK-NEXT: ret i1 [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
%tmp1 = ashr exact i8 %tmp0, 3 ; not 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq i8 %tmp1, %x
ret i1 %tmp2
}
define i1 @n1(i8 %x) {
; CHECK-LABEL: @n1(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i8 [[X:%.*]], 8
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: ret i1 [[TMP1]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
%tmp1 = lshr exact i8 %tmp0, 5 ; not ashr
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq i8 %tmp1, %x
ret i1 %tmp2
}
define i1 @n2(i8 %x, i8 %y) {
; CHECK-LABEL: @n2(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl i8 [[X:%.*]], 5
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact i8 [[TMP0]], 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i8 [[TMP1]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl i8 %x, 5
%tmp1 = ashr exact i8 %tmp0, 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq i8 %tmp1, %y ; not %x
ret i1 %tmp2
}
define <2 x i1> @n3_vec_nonsplat(<2 x i8> %x) {
; CHECK-LABEL: @n3_vec_nonsplat(
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = shl <2 x i8> [[X:%.*]], <i8 5, i8 5>
; CHECK-NEXT: [[TMP1:%.*]] = ashr exact <2 x i8> [[TMP0]], <i8 5, i8 3>
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i8> [[TMP1]], [[X]]
; CHECK-NEXT: ret <2 x i1> [[TMP2]]
;
Revert test changes part of "Revert "[InstCombine] Fold 'check for [no] signed truncation' pattern"" We want the test to remain good anyway. I think the fix is incoming. This reverts part of commit rL337344. llvm-svn: 337359
2018-07-18 16:15:13 +08:00
%tmp0 = shl <2 x i8> %x, <i8 5, i8 5>
%tmp1 = ashr exact <2 x i8> %tmp0, <i8 5, i8 3> ; 3 instead of 5
[NFC][InstCombine] Tests for 'check for [no] signed truncation' pattern [[ https://bugs.llvm.org/show_bug.cgi?id=38149 | PR38149 ]] As discussed in https://reviews.llvm.org/D49179#1158957 and later, the IR for 'check for [no] signed truncation' pattern can be improved: https://rise4fun.com/Alive/gBf ^ that pattern will be produced by Implicit Integer Truncation sanitizer, https://reviews.llvm.org/D48958 https://bugs.llvm.org/show_bug.cgi?id=21530 in signed case, therefore it is probably a good idea to improve it. The DAGCombine will reverse this transform, see https://reviews.llvm.org/D49266 llvm-svn: 337042
2018-07-14 04:33:34 +08:00
%tmp2 = icmp eq <2 x i8> %tmp1, %x
ret <2 x i1> %tmp2
}