[PATCH] D20905: Only attempt to detect AVG if SSE2 is available

[Dimitry Andric via llvm-commits]

dim updated this revision to Diff 59368.
dim added a comment.

Add simplified test case from PR29973.


http://reviews.llvm.org/D20905

Files:
  lib/Target/X86/X86ISelLowering.cpp
  test/CodeGen/X86/pr27973.ll

Index: test/CodeGen/X86/pr27973.ll
===================================================================
--- /dev/null
+++ test/CodeGen/X86/pr27973.ll
@@ -0,0 +1,39 @@
+; REQUIRES: asserts
+; RUN: llc < %s
+
+target datalayout = "e-m:o-p:32:32-f64:32:64-f80:128-n8:16:32-S128"
+target triple = "i386-apple-macosx10.11.0"
+
+; Function Attrs: norecurse nounwind ssp
+define void @fn1() #0 {
+entry:
+  br i1 undef, label %for.body4.preheader, label %for.cond.cleanup
+
+for.body4.preheader:                              ; preds = %entry
+  br label %vector.body
+
+vector.body:                                      ; preds = %vector.body, %for.body4.preheader
+  %0 = zext <16 x i8> zeroinitializer to <16 x i32>
+  %1 = add nuw nsw <16 x i32> %0, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
+  %2 = lshr <16 x i32> %1, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
+  %3 = trunc <16 x i32> %2 to <16 x i8>
+  store <16 x i8> %3, <16 x i8>* undef, align 1, !tbaa !1, !alias.scope !4, !noalias !7
+  br label %vector.body
+
+for.cond.cleanup:                                 ; preds = %entry
+  ret void
+}
+
+attributes #0 = { norecurse nounwind ssp "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="pentium3" "target-features"="+fxsr,+mmx,+sse,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
+
+!llvm.ident = !{!0}
+
+!0 = !{!"clang version 3.9.0 (trunk 271435)"}
+!1 = !{!2, !2, i64 0}
+!2 = !{!"omnipotent char", !3, i64 0}
+!3 = !{!"Simple C/C++ TBAA"}
+!4 = !{!5}
+!5 = distinct !{!5, !6}
+!6 = distinct !{!6, !"LVerDomain"}
+!7 = !{!8}
+!8 = distinct !{!8, !6}
Index: lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- lib/Target/X86/X86ISelLowering.cpp
+++ lib/Target/X86/X86ISelLowering.cpp
@@ -27842,7 +27842,9 @@
   if (InScalarVT.getSizeInBits() <= ScalarVT.getSizeInBits())
     return SDValue();
 
-  if (Subtarget.hasAVX512()) {
+  if (!Subtarget.hasSSE2())
+    return SDValue();
+  else if (Subtarget.hasAVX512()) {
     if (VT.getSizeInBits() > 512)
       return SDValue();
   } else if (Subtarget.hasAVX2()) {


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D20905.59368.patch
Type: text/x-patch
Size: 2425 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160602/f6adbbd5/attachment.bin>