Skip to content

[InstCombine] Fix FMF propagation in foldSelectIntoOp #114356

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Oct 31, 2024
Merged

[InstCombine] Fix FMF propagation in foldSelectIntoOp #114356

merged 1 commit into from
Oct 31, 2024

Conversation

@dtcxzyw
Copy link
Member

@dtcxzyw dtcxzyw commented Oct 31, 2024

Closes #113423.

@dtcxzyw dtcxzyw requested review from arsenm and goldsteinn October 31, 2024 05:02
@dtcxzyw dtcxzyw requested a review from nikic as a code owner October 31, 2024 05:02
@dtcxzyw dtcxzyw mentioned this pull request Oct 31, 2024
Open
@llvmbot
Copy link
Member

llvmbot commented Oct 31, 2024

@llvm/pr-subscribers-llvm-transforms

Author: Yingwei Zheng (dtcxzyw)

Changes

Closes #113423.

Full diff: https://github.com/llvm/llvm-project/pull/114356.diff

3 Files Affected:

  • (modified) llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp (+2-3)
  • (modified) llvm/test/Transforms/InstCombine/fold-select-fmul-if-zero.ll (+9-9)
  • (modified) llvm/test/Transforms/InstCombine/select-binop-foldable-floating-point.ll (+8-8)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
index c5f39a4c381ed1..9e193e7faa8ac3 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
@@ -529,9 +529,6 @@ Instruction *InstCombinerImpl::foldSelectIntoOp(SelectInst &SI, Value *TrueVal,
     if (!OpToFold)
       return nullptr;
 
-    // TODO: We probably ought to revisit cases where the select and FP
-    // instructions have different flags and add tests to ensure the
-    // behaviour is correct.
     FastMathFlags FMF;
     if (isa<FPMathOperator>(&SI))
       FMF = SI.getFastMathFlags();
@@ -564,6 +561,8 @@ Instruction *InstCombinerImpl::foldSelectIntoOp(SelectInst &SI, Value *TrueVal,
     BinaryOperator *BO =
         BinaryOperator::Create(TVI->getOpcode(), FalseVal, NewSel);
     BO->copyIRFlags(TVI);
+    if (isa<FPMathOperator>(&SI))
+      BO->andIRFlags(NewSel);
     return BO;
   };
 
diff --git a/llvm/test/Transforms/InstCombine/fold-select-fmul-if-zero.ll b/llvm/test/Transforms/InstCombine/fold-select-fmul-if-zero.ll
index 1c28b151825c12..caf38c676e20d7 100644
--- a/llvm/test/Transforms/InstCombine/fold-select-fmul-if-zero.ll
+++ b/llvm/test/Transforms/InstCombine/fold-select-fmul-if-zero.ll
@@ -468,7 +468,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_nsz_fmul(float %x, float %y) {
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_nsz_fmul(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nsz float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -482,7 +482,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_nsz_ninf_fmul(float %x, float %y) {
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_nsz_ninf_fmul(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul ninf nsz float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -496,7 +496,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_nsz_nnan_fmul(float %x, float %y) {
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_nsz_nnan_fmul(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan nsz float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -510,7 +510,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_nnan_ninf_fmul(float %x, float %y) {
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_nnan_ninf_fmul(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan ninf float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -559,7 +559,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_nsz(float %x, float %
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_nsz(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan ninf nsz float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -572,7 +572,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_nsz_commuted(float %x
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_nsz_commuted(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan ninf nsz float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -586,7 +586,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_select_known_never_ne
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_select_known_never_negzero(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan ninf float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -599,7 +599,7 @@ define float @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_select_known_never_ne
 ; CHECK-LABEL: @fmul_by_var_if_0_oeq_zero_f32_fmul_nnan_ninf_select_known_never_negzero_negsub(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[Y:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan ninf float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
@@ -706,7 +706,7 @@ define float @fmul_by_self_if_0_oeq_zero_f32_fmul_nnan_ninf_nsz(float %x) {
 ; CHECK-LABEL: @fmul_by_self_if_0_oeq_zero_f32_fmul_nnan_ninf_nsz(
 ; CHECK-NEXT:    [[X_IS_ZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
 ; CHECK-NEXT:    [[SCALED_X:%.*]] = select nnan i1 [[X_IS_ZERO]], float [[X]], float 1.000000e+00
-; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan ninf nsz float [[X]], [[SCALED_X]]
+; CHECK-NEXT:    [[SCALED_IF_DENORMAL:%.*]] = fmul nnan float [[X]], [[SCALED_X]]
 ; CHECK-NEXT:    ret float [[SCALED_IF_DENORMAL]]
 ;
   %x.is.zero = fcmp oeq float %x, 0.0
diff --git a/llvm/test/Transforms/InstCombine/select-binop-foldable-floating-point.ll b/llvm/test/Transforms/InstCombine/select-binop-foldable-floating-point.ll
index e5ad312bb85c17..df02eb7bb83ac6 100644
--- a/llvm/test/Transforms/InstCombine/select-binop-foldable-floating-point.ll
+++ b/llvm/test/Transforms/InstCombine/select-binop-foldable-floating-point.ll
@@ -48,7 +48,7 @@ define float @select_nnan_fadd_swapped(i1 %cond, float %A, float %B) {
 define float @select_nnan_fadd_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fadd_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float [[B:%.*]], float -0.000000e+00
-; CHECK-NEXT:    [[D:%.*]] = fadd fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fadd nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fadd fast float %A, %B
@@ -59,7 +59,7 @@ define float @select_nnan_fadd_fast_math(i1 %cond, float %A, float %B) {
 define float @select_nnan_fadd_swapped_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fadd_swapped_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float -0.000000e+00, float [[B:%.*]]
-; CHECK-NEXT:    [[D:%.*]] = fadd fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fadd nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fadd fast float %A, %B
@@ -125,7 +125,7 @@ define float @select_nnan_fmul_swapped(i1 %cond, float %A, float %B) {
 define float @select_nnan_fmul_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fmul_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float [[B:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[D:%.*]] = fmul fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fmul nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fmul fast float %A, %B
@@ -136,7 +136,7 @@ define float @select_nnan_fmul_fast_math(i1 %cond, float %A, float %B) {
 define float @select_nnan_fmul_swapped_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fmul_swapped_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float 1.000000e+00, float [[B:%.*]]
-; CHECK-NEXT:    [[D:%.*]] = fmul fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fmul nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fmul fast float %A, %B
@@ -169,7 +169,7 @@ define float @select_nnan_fsub_swapped(i1 %cond, float %A, float %B) {
 define float @select_nnan_fsub_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fsub_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float [[B:%.*]], float 0.000000e+00
-; CHECK-NEXT:    [[D:%.*]] = fsub fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fsub nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fsub fast float %A, %B
@@ -180,7 +180,7 @@ define float @select_nnan_fsub_fast_math(i1 %cond, float %A, float %B) {
 define float @select_nnan_fsub_swapped_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fsub_swapped_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float 0.000000e+00, float [[B:%.*]]
-; CHECK-NEXT:    [[D:%.*]] = fsub fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fsub nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fsub fast float %A, %B
@@ -247,7 +247,7 @@ define float @select_nnan_fdiv_swapped(i1 %cond, float %A, float %B) {
 define float @select_nnan_fdiv_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fdiv_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float [[B:%.*]], float 1.000000e+00
-; CHECK-NEXT:    [[D:%.*]] = fdiv fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fdiv nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fdiv fast float %A, %B
@@ -258,7 +258,7 @@ define float @select_nnan_fdiv_fast_math(i1 %cond, float %A, float %B) {
 define float @select_nnan_fdiv_swapped_fast_math(i1 %cond, float %A, float %B) {
 ; CHECK-LABEL: @select_nnan_fdiv_swapped_fast_math(
 ; CHECK-NEXT:    [[C:%.*]] = select nnan i1 [[COND:%.*]], float 1.000000e+00, float [[B:%.*]]
-; CHECK-NEXT:    [[D:%.*]] = fdiv fast float [[A:%.*]], [[C]]
+; CHECK-NEXT:    [[D:%.*]] = fdiv nnan float [[A:%.*]], [[C]]
 ; CHECK-NEXT:    ret float [[D]]
 ;
   %C = fdiv fast float %A, %B

@dtcxzyw dtcxzyw mentioned this pull request Oct 31, 2024
Closed
@goldsteinn
Copy link
Contributor

goldsteinn commented Oct 31, 2024

Ill defer to matt but LGTM.

@arsenm arsenm added llvm:instcombine Covers the InstCombine, InstSimplify and AggressiveInstCombine passes floating-point Floating-point math labels Oct 31, 2024
arsenm
arsenm approved these changes Oct 31, 2024
View reviewed changes
llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
Comment on lines 563 to +565
BO->copyIRFlags(TVI);
if (isa<FPMathOperator>(&SI))
BO->andIRFlags(NewSel);
Copy link
Contributor

@arsenm arsenm Oct 31, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This API is so clumsy.

We could preserve more value flags if the same operands appear in both instructions, but that's probably not worth the complexity

@dtcxzyw dtcxzyw merged commit cf1963a into llvm:main Oct 31, 2024
10 of 12 checks passed
@dtcxzyw dtcxzyw deleted the fix-113423 branch October 31, 2024 15:26
@dyung
Copy link
Collaborator

dyung commented Oct 31, 2024

Hi @dtcxzyw I suspect your change is causing the test failure in predicated-reduction.ll that is failing on https://lab.llvm.org/buildbot/#/builders/190/builds/8601. Can you take a look>

@dyung
Copy link
Collaborator

dyung commented Oct 31, 2024

This change also seems to be causing a test failure in Transforms/LoopVectorize/ARM/mve-selectandorcost.ll:
https://lab.llvm.org/buildbot/#/builders/174/builds/7674

@gulfemsavrun
Copy link
Contributor

gulfemsavrun commented Oct 31, 2024

We also started seeing a test failure:

FAIL: LLVM :: Transforms/PhaseOrdering/AArch64/predicated-reduction.ll (46966 of 56618)
******************** TEST 'LLVM :: Transforms/PhaseOrdering/AArch64/predicated-reduction.ll' FAILED ********************
Exit Code: 1

Command Output (stderr):
--
RUN: at line 2: /b/s/w/ir/x/w/llvm_build/bin/opt -passes="default<O3>" -S < /b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll  | /b/s/w/ir/x/w/llvm_build/bin/FileCheck /b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
+ /b/s/w/ir/x/w/llvm_build/bin/opt '-passes=default<O3>' -S
+ /b/s/w/ir/x/w/llvm_build/bin/FileCheck /b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
/b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll:26:15: error: CHECK-NEXT: expected string not found in input
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x double> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP18:%.*]], %[[VECTOR_BODY]] ]
              ^
<stdin>:26:67: note: scanning from here
 %index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
                                                                  ^
<stdin>:26:67: note: with "VECTOR_PH" equal to "vector\\.ph"
 %index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
                                                                  ^
<stdin>:26:67: note: with "VECTOR_BODY" equal to "vector\\.body"
 %index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
                                                                  ^
<stdin>:39:9: note: possible intended match here
 %8 = fcmp fast ogt <2 x double> %6, zeroinitializer
        ^
/b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll:349:12: error: undefined variable: LOOP0
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
           ^
/b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll:349:35: error: undefined variable: LOOP0
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
                                  ^
<stdin>:172:2: note: possible intended match here
!0 = distinct !{!0, !1, !2}
 ^

Input file: <stdin>
Check file: /b/s/w/ir/x/w/llvm-llvm-project/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll

-dump-input=help explains the following input dump.

Input was:
<<<<<<
             .
             .
             .
            21:  %broadcast.splatinsert17 = insertelement <2 x double> poison, double %Z, i64 0 
            22:  %broadcast.splat18 = shufflevector <2 x double> %broadcast.splatinsert17, <2 x double> poison, <2 x i32> zeroinitializer 
            23:  br label %vector.body 
            24:  
            25: vector.body: ; preds = %vector.body, %vector.ph 
            26:  %index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ] 
next:26'0                                                                         X error: no match found
next:26'1                                                                           with "VECTOR_PH" equal to "vector\\.ph"
next:26'2                                                                           with "VECTOR_BODY" equal to "vector\\.body"
            27:  %vec.phi = phi double [ 0.000000e+00, %vector.ph ], [ %17, %vector.body ] 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            28:  %vec.phi15 = phi double [ 0.000000e+00, %vector.ph ], [ %19, %vector.body ] 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            29:  %0 = getelementptr inbounds float, ptr %samples, i64 %index 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            30:  %1 = getelementptr inbounds i8, ptr %0, i64 8 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            31:  %wide.load = load <2 x float>, ptr %0, align 4 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
             .
             .
             .
            34:  %3 = fpext <2 x float> %wide.load16 to <2 x double> 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            35:  %4 = fmul fast <2 x double> %broadcast.splat, %2 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            36:  %5 = fmul fast <2 x double> %broadcast.splat, %3 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            37:  %6 = fsub fast <2 x double> %4, %broadcast.splat18 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            38:  %7 = fsub fast <2 x double> %5, %broadcast.splat18 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            39:  %8 = fcmp fast ogt <2 x double> %6, zeroinitializer 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
next:26'3               ?                                             possible intended match
            40:  %9 = fcmp fast ogt <2 x double> %7, zeroinitializer 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            41:  %10 = fmul fast <2 x double> %6, %6 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            42:  %11 = fmul fast <2 x double> %7, %7 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            43:  %12 = tail call fast <2 x double> @llvm.maxnum.v2f64(<2 x double> %6, <2 x double> <double -0.000000e+00, double -0.000000e+00>) 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            44:  %13 = tail call fast <2 x double> @llvm.maxnum.v2f64(<2 x double> %7, <2 x double> <double -0.000000e+00, double -0.000000e+00>) 
next:26'0       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
             .
             .
             .
           146:  
           147: for.end10: ; preds = %for.body, %for.cond1.for.inc8_crit_edge.us, %entry 
           148:  %v0.0.lcssa = phi double [ 0.000000e+00, %entry ], [ %v0.2.us, %for.cond1.for.inc8_crit_edge.us ], [ 0.000000e+00, %for.body ] 
           149:  %v1.0.lcssa = phi double [ 0.000000e+00, %entry ], [ %v1.2.us, %for.cond1.for.inc8_crit_edge.us ], [ 0.000000e+00, %for.body ] 
           150:  %add11 = fadd fast double %v1.0.lcssa, %v0.0.lcssa 
           151:  ret double %add11 
check:349'0                       X error: match failed for invalid pattern
check:349'1                         undefined variable: LOOP0
check:349'2                         undefined variable: LOOP0
           152: } 
check:349'0     ~~
           153:  
check:349'0     ~
           154: declare void @resample(i32 noundef, ptr noundef) local_unnamed_addr 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           155:  
check:349'0     ~
           156: ; Function Attrs: mustprogress nocallback nofree nosync nounwind speculatable willreturn memory(none) 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
             .
             .
             .
           167:  
check:349'0     ~
           168: attributes #0 = { nofree norecurse nosync nounwind memory(argmem: read) } 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           169: attributes #1 = { mustprogress nocallback nofree nosync nounwind speculatable willreturn memory(none) } 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           170: attributes #2 = { nocallback nofree nosync nounwind speculatable willreturn memory(none) } 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           171:  
check:349'0     ~
           172: !0 = distinct !{!0, !1, !2} 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
check:349'3      ?                           possible intended match
           173: !1 = !{!"llvm.loop.isvectorized", i32 1} 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           174: !2 = !{!"llvm.loop.unroll.runtime.disable"} 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           175: !3 = distinct !{!3, !2, !1} 
check:349'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>>>>

--

********************

https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-base-linux-x64/b8732549597609293617/overview

gulfemsavrun added a commit that referenced this pull request Oct 31, 2024
@gulfemsavrun
Revert "[InstCombine] Fix FMF propagation in foldSelectIntoOp (#114356
4b61314 
)"

This reverts commit cf1963a.
@gulfemsavrun gulfemsavrun mentioned this pull request Oct 31, 2024
Merged
gulfemsavrun added a commit that referenced this pull request Oct 31, 2024
@gulfemsavrun
Revert "[InstCombine] Fix FMF propagation in foldSelectIntoOp" (#11…
d183dc7 
…4458)

Reverts #114356 because it caused test failures.
https://lab.llvm.org/buildbot/#/builders/190/builds/8601

https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-base-linux-x64/b8732549597609293617/overview
dtcxzyw added a commit to dtcxzyw/llvm-project that referenced this pull request Nov 1, 2024
@dtcxzyw
[InstCombine] Fix FMF propagation in foldSelectIntoOp (llvm#114356)
163698f 
Closes llvm#113423.
@dtcxzyw dtcxzyw mentioned this pull request Nov 1, 2024
Merged
dtcxzyw added a commit that referenced this pull request Nov 1, 2024
@dtcxzyw
[Reland][InstCombine] Fix FMF propagation in foldSelectIntoOp (#114499
96b14f2 
)

Relands #114356. Compared to the last version, this patch only merges
poison-generating/nsz flags from the select to fix LV regression in
`llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll`.
smallp-o-p pushed a commit to smallp-o-p/llvm-project that referenced this pull request Nov 3, 2024
@dtcxzyw
[InstCombine] Fix FMF propagation in foldSelectIntoOp (llvm#114356)
146da23 
Closes llvm#113423.
smallp-o-p pushed a commit to smallp-o-p/llvm-project that referenced this pull request Nov 3, 2024
@gulfemsavrun
Revert "[InstCombine] Fix FMF propagation in foldSelectIntoOp" (llv…
4e96184 
…m#114458)

Reverts llvm#114356 because it caused test failures.
https://lab.llvm.org/buildbot/#/builders/190/builds/8601

https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-base-linux-x64/b8732549597609293617/overview
smallp-o-p pushed a commit to smallp-o-p/llvm-project that referenced this pull request Nov 3, 2024
@dtcxzyw
[Reland][InstCombine] Fix FMF propagation in foldSelectIntoOp (llvm…
a84ca28 
…#114499)

Relands llvm#114356. Compared to the last version, this patch only merges
poison-generating/nsz flags from the select to fix LV regression in
`llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll`.
NoumanAmir657 pushed a commit to NoumanAmir657/llvm-project that referenced this pull request Nov 4, 2024
@dtcxzyw @NoumanAmir657
[InstCombine] Fix FMF propagation in foldSelectIntoOp (llvm#114356)
67a4d01 
Closes llvm#113423.
NoumanAmir657 pushed a commit to NoumanAmir657/llvm-project that referenced this pull request Nov 4, 2024
@gulfemsavrun @NoumanAmir657
Revert "[InstCombine] Fix FMF propagation in foldSelectIntoOp" (llv…
2ae30bd 
…m#114458)

Reverts llvm#114356 because it caused test failures.
https://lab.llvm.org/buildbot/#/builders/190/builds/8601

https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-base-linux-x64/b8732549597609293617/overview
NoumanAmir657 pushed a commit to NoumanAmir657/llvm-project that referenced this pull request Nov 4, 2024
@dtcxzyw @NoumanAmir657
[Reland][InstCombine] Fix FMF propagation in foldSelectIntoOp (llvm…
b0eaa62 
…#114499)

Relands llvm#114356. Compared to the last version, this patch only merges
poison-generating/nsz flags from the select to fix LV regression in
`llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll`.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@arsenm arsenm arsenm approved these changes

@goldsteinn goldsteinn Awaiting requested review from goldsteinn

@nikic nikic Awaiting requested review from nikic nikic is a code owner

Assignees

No one assigned

Labels

floating-point Floating-point math llvm:instcombine Covers the InstCombine, InstSimplify and AggressiveInstCombine passes llvm:transforms

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

Wrong folding of select into floating number ops

6 participants

@dtcxzyw @llvmbot @goldsteinn @dyung @gulfemsavrun @arsenm