[PATCH] D144571: [DAGCombine] Fix an ICE in combineMinNumMaxNum(...)

[Matt Arsenault via Phabricator via llvm-commits]

arsenm accepted this revision.
arsenm added a comment.
This revision is now accepted and ready to land.

LGTM with test nits



================
Comment at: llvm/test/CodeGen/X86/2023-02-22-combineMinNumMaxNum.ll:5
+
+define float @test_combinemaxnum() local_unnamed_addr #0 {
+L.entry:
----------------
Don't need local_unnamed_addr


================
Comment at: llvm/test/CodeGen/X86/2023-02-22-combineMinNumMaxNum.ll:7
+L.entry:
+  %0 = tail call float @llvm.maxnum.f32(float 0.000000e+00, float 0.000000e+00) 
+  %1 = fsub fast float poison, poison
----------------
I suspect you can simplify this too


================
Comment at: llvm/test/CodeGen/X86/2023-02-22-combineMinNumMaxNum.ll:13
+  %2 = fneg fast float %1
+  %3 = fcmp fast ule float %0, %2
+  %4 = fneg fast float %0
----------------
Use named values in test


================
Comment at: llvm/test/CodeGen/X86/2023-02-22-combineMinNumMaxNum.ll:21
+
+attributes #0 = { "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" }
----------------
Do we really need the attributes (IIRC, yes for this case which really should be fixed)


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144571/new/

https://reviews.llvm.org/D144571