Skip to content

release/20.x: [InstCombine] Do not fold logical is_finite test (#136851) #137606

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.

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

Merged
merged 1 commit into from
Apr 29, 2025

Conversation

llvmbot
Copy link
Member

@llvmbot llvmbot commented Apr 28, 2025

Backport 8abc917

Requested by: @dtcxzyw

@llvmbot
Copy link
Member Author

llvmbot commented Apr 28, 2025

@nikic What do you think about merging this PR to the release branch?

@github-project-automation github-project-automation bot moved this to Needs Triage in LLVM Release Status Apr 28, 2025
@llvmbot llvmbot added llvm:instcombine Covers the InstCombine, InstSimplify and AggressiveInstCombine passes llvm:transforms labels Apr 28, 2025
@llvmbot
Copy link
Member Author

llvmbot commented Apr 28, 2025

@llvm/pr-subscribers-llvm-transforms

Author: None (llvmbot)

Changes

Backport 8abc917

Requested by: @dtcxzyw


Full diff: https://github.com/llvm/llvm-project/pull/137606.diff

2 Files Affected:

  • (modified) llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp (+3-1)
  • (modified) llvm/test/Transforms/InstCombine/and-fcmp.ll (+28)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp b/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
index ca8a20b4b7312..ebb84d177a832 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineAndOrXor.cpp
@@ -1475,7 +1475,9 @@ Value *InstCombinerImpl::foldLogicOfFCmps(FCmpInst *LHS, FCmpInst *RHS,
     }
   }
 
-  if (IsAnd && stripSignOnlyFPOps(LHS0) == stripSignOnlyFPOps(RHS0)) {
+  // This transform is not valid for a logical select.
+  if (!IsLogicalSelect && IsAnd &&
+      stripSignOnlyFPOps(LHS0) == stripSignOnlyFPOps(RHS0)) {
     // and (fcmp ord x, 0), (fcmp u* x, inf) -> fcmp o* x, inf
     // and (fcmp ord x, 0), (fcmp u* fabs(x), inf) -> fcmp o* x, inf
     if (Value *Left = matchIsFiniteTest(Builder, LHS, RHS))
diff --git a/llvm/test/Transforms/InstCombine/and-fcmp.ll b/llvm/test/Transforms/InstCombine/and-fcmp.ll
index c7bbc8ab56f9a..ec1b6ad2ea168 100644
--- a/llvm/test/Transforms/InstCombine/and-fcmp.ll
+++ b/llvm/test/Transforms/InstCombine/and-fcmp.ll
@@ -4990,6 +4990,34 @@ define i1 @clang_builtin_isnormal_inf_check_copysign(half %x, half %y) {
   ret i1 %and
 }
 
+define i1 @clang_builtin_isnormal_inf_check_copysign_logical_select(half %x, half %y) {
+; CHECK-LABEL: @clang_builtin_isnormal_inf_check_copysign_logical_select(
+; CHECK-NEXT:    [[COPYSIGN_X:%.*]] = call half @llvm.copysign.f16(half [[X:%.*]], half [[Y:%.*]])
+; CHECK-NEXT:    [[ORD:%.*]] = fcmp ord half [[X]], 0xH0000
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp ueq half [[COPYSIGN_X]], 0xH7C00
+; CHECK-NEXT:    [[AND:%.*]] = select i1 [[ORD]], i1 [[CMP]], i1 false
+; CHECK-NEXT:    ret i1 [[AND]]
+;
+  %copysign.x = call half @llvm.copysign.f16(half %x, half %y)
+  %ord = fcmp ord half %x, 0.0
+  %cmp = fcmp uge half %copysign.x, 0xH7C00
+  %and = select i1 %ord, i1 %cmp, i1 false
+  ret i1 %and
+}
+
+define i1 @clang_builtin_isnormal_inf_check_fabs_nnan_logical_select(half %x) {
+; CHECK-LABEL: @clang_builtin_isnormal_inf_check_fabs_nnan_logical_select(
+; CHECK-NEXT:    [[COPYSIGN_X:%.*]] = call half @llvm.fabs.f16(half [[X:%.*]])
+; CHECK-NEXT:    [[AND:%.*]] = fcmp oeq half [[COPYSIGN_X]], 0xH7C00
+; CHECK-NEXT:    ret i1 [[AND]]
+;
+  %copysign.x = call nnan half @llvm.fabs.f16(half %x)
+  %ord = fcmp ord half %x, 0.0
+  %cmp = fcmp uge half %copysign.x, 0xH7C00
+  %and = select i1 %ord, i1 %cmp, i1 false
+  ret i1 %and
+}
+
 define i1 @isnormal_logical_select_0(half %x) {
 ; CHECK-LABEL: @isnormal_logical_select_0(
 ; CHECK-NEXT:    [[FABS_X:%.*]] = call half @llvm.fabs.f16(half [[X:%.*]])

@github-project-automation github-project-automation bot moved this from Needs Triage to Needs Merge in LLVM Release Status Apr 28, 2025
This patch disables the fold for logical is_finite test (i.e., `and
(fcmp ord x, 0), (fcmp u* x, inf) -> fcmp o* x, inf`).
It is still possible to allow this fold for several logical cases (e.g.,
`stripSignOnlyFPOps(RHS0)` does not strip any operations). Since this
patch has no real-world impact, I decided to disable this fold for all
logical cases.

Alive2: https://alive2.llvm.org/ce/z/aH4LC7
Closes llvm#136650.

(cherry picked from commit 8abc917)
@tstellar tstellar merged commit 57a31e1 into llvm:release/20.x Apr 29, 2025
7 of 10 checks passed
@github-project-automation github-project-automation bot moved this from Needs Merge to Done in LLVM Release Status Apr 29, 2025
Copy link

@dtcxzyw (or anyone else). If you would like to add a note about this fix in the release notes (completely optional). Please reply to this comment with a one or two sentence description of the fix. When you are done, please add the release:note label to this PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
llvm:instcombine Covers the InstCombine, InstSimplify and AggressiveInstCombine passes llvm:transforms
Projects
Development

Successfully merging this pull request may close these issues.

4 participants