[ConstantFolding] Fix nvvm_round folding on PPC

[LewisCrawford]

Fix a failing test for constant-folding the nvvm_round intrinsic. The original implementation added in #141233 used a native libm call to the "round" function, but on PPC this produces +0.0 if the input is -0.0, which caused a test failure.

This patch updates it to use APFloat functions instead of native libm calls to ensure cross-platform consistency.

[llvmbot]

@llvm/pr-subscribers-llvm-analysis

Author: Lewis Crawford (LewisCrawford)

Changes

Fix a failing test for constant-folding the nvvm_round intrinsic. The original implementation added in #141233 used a native libm call to the "round" function, but on PPC this produces +0.0 if the input is -0.0, which caused a test failure.

This patch updates it to use APFloat functions instead of native libm calls to ensure cross-platform consistency.

---

Full diff: https://github.com/llvm/llvm-project/pull/149837.diff

1 Files Affected:

• (modified) llvm/lib/Analysis/ConstantFolding.cpp (+9-5)

diff --git a/llvm/lib/Analysis/ConstantFolding.cpp b/llvm/lib/Analysis/ConstantFolding.cpp
index f5a88b6e0368e..eb7369fcc7513 100644
--- a/llvm/lib/Analysis/ConstantFolding.cpp
+++ b/llvm/lib/Analysis/ConstantFolding.cpp
@@ -2677,11 +2677,15 @@ static Constant *ConstantFoldScalarCall1(StringRef Name,
 
       case Intrinsic::nvvm_round_ftz_f:
       case Intrinsic::nvvm_round_f:
-      case Intrinsic::nvvm_round_d:
-        return ConstantFoldFP(
-            round, APF, Ty,
-            nvvm::GetNVVMDenromMode(
-                nvvm::UnaryMathIntrinsicShouldFTZ(IntrinsicID)));
+      case Intrinsic::nvvm_round_d: {
+        // Use APFloat implementation instead of native libm call, as some
+        // implementations (e.g. on PPC) do not preserve the sign of negative 0.
+        APFloat Res = nvvm::UnaryMathIntrinsicShouldFTZ(IntrinsicID)
+                          ? FTZPreserveSign(APF)
+                          : APF;
+        Res.roundToIntegral(APFloat::rmNearestTiesToAway);
+        return ConstantFP::get(Ty->getContext(), U);
+      }
 
       case Intrinsic::nvvm_saturate_ftz_f:
       case Intrinsic::nvvm_saturate_d:

[LewisCrawford]

This should fix the test failure here: https://lab.llvm.org/buildbot/#/builders/64/builds/4929

• /home/llvm/llvm-external-buildbots/workers/aix-ppc64/clang-ppc64-aix/build/bin/FileCheck /home/llvm/llvm-external-buildbots/workers/aix-ppc64/clang-ppc64-aix/llvm-project/llvm/test/Transforms/InstSimplify/const-fold-nvvm-unary-arithmetic.ll
  /home/llvm/llvm-external-buildbots/workers/aix-ppc64/clang-ppc64-aix/llvm-project/llvm/test/Transforms/InstSimplify/const-fold-nvvm-unary-arithmetic.ll:421:15: error: CHECK-NEXT: expected string not found in input
  ; CHECK-NEXT: ret double -0.000000e+00
  ^
  :196:44: note: scanning from here
  define double @test_round_d_neg_subnorm() {
  ^
  :197:2: note: possible intended match here
  ret double 0.000000e+00
  ^
  /home/llvm/llvm-external-buildbots/workers/aix-ppc64/clang-ppc64-aix/llvm-project/llvm/test/Transforms/InstSimplify/const-fold-nvvm-unary-arithmetic.ll:429:15: error: CHECK-NEXT: expected string not found in input
  ; CHECK-NEXT: ret float -0.000000e+00
  ^
  :200:43: note: scanning from here
  define float @test_round_f_neg_subnorm() {
  ^
  :201:2: note: possible intended match here
  ret float 0.000000e+00
  ^
  Input file:
  Check file: /home/llvm/llvm-external-buildbots/workers/aix-ppc64/clang-ppc64-aix/llvm-project/llvm/test/Transforms/InstSimplify/const-fold-nvvm-unary-arithmetic.ll
  -dump-input=help explains the following input dump.
  Input was:
  <<<<<<
  .
  .
  .
  191:
  192: define float @test_round_ftz_f_neg_1_5() {
  193: ret float -2.000000e+00
  194: }
  195:
  196: define double @test_round_d_neg_subnorm() {
  next:421'0 X error: no match found
  197: ret double 0.000000e+00
  next:421'0 ~~~~~~~~~~~~~~~~~~~~~~~~~
  next:421'1 ? possible intended match
  198: }
  next:421'0 ~~
  199:
  next:421'0 ~
  200: define float @test_round_f_neg_subnorm() {
  next:421'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  next:429'0 X error: no match found
  201: ret float 0.000000e+00
  next:429'0 ~~~~~~~~~~~~~~~~~~~~~~~~
  next:429'1 ? possible intended match
  202: }
  next:429'0 ~~
  203:
  next:429'0 ~
  204: define float @test_round_ftz_f_neg_subnorm() {
  next:429'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  205: ret float -0.000000e+00
  206: }
  .
  .
  .

[jholewinski]

LGTM. Thanks for jumping on this!