Pointer types were treated as zero-size by MergeICmps

Summary:
The visitICmp analysis function would record compares of pointer types, as size 0. This causes the resulting memcmp() call to have the wrong total size.
Found with "self-build" of clang/LLVM on Windows.

Reviewers: christylee, trentxintong, courbet

Reviewed By: courbet

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D53536

llvm-svn: 345413

diff --git a/llvm/lib/Transforms/Scalar/MergeICmps.cpp b/llvm/lib/Transforms/Scalar/MergeICmps.cpp
index f68662b488cbb053839698715f54f0f6c09e3277..3633485d5d5c87d1fd34626b5cf4533c6b4d5671 100644 (file)
--- a/llvm/lib/Transforms/Scalar/MergeICmps.cpp
+++ b/llvm/lib/Transforms/Scalar/MergeICmps.cpp
@@ -283,8 +283,9 @@ BCECmpBlock visitICmp(const ICmpInst *const CmpI,
     if (!Lhs.Base()) return {};
     auto Rhs = visitICmpLoadOperand(CmpI->getOperand(1));
     if (!Rhs.Base()) return {};
+    const auto &DL = CmpI->getModule()->getDataLayout();
     return BCECmpBlock(std::move(Lhs), std::move(Rhs),
-                       CmpI->getOperand(0)->getType()->getScalarSizeInBits());
+                       DL.getTypeSizeInBits(CmpI->getOperand(0)->getType()));
   }
   return {};
 }

diff --git a/llvm/test/Transforms/MergeICmps/X86/int64-and-ptr.ll b/llvm/test/Transforms/MergeICmps/X86/int64-and-ptr.ll
new file mode 100644 (file)
index 0000000..78924ae
--- /dev/null
+++ b/llvm/test/Transforms/MergeICmps/X86/int64-and-ptr.ll
@@ -0,0 +1,39 @@
+; RUN: opt < %s -mtriple=x86_64-unknown-unknown -mergeicmps -S | FileCheck %s --check-prefix=X86
+
+; 8-byte int and 8-byte pointer should merge into a 16-byte memcpy.
+; X86: memcmp(i8* {{.*}}, i8* {{.*}}, i64 16)
+
+%struct.outer = type { i64, %struct.inner* }
+%struct.inner = type { i32, i32, i32 }
+
+; Function Attrs: nounwind uwtable
+define dso_local i1 @"?foo@@YAHAEAUouter@@0@Z"(%struct.outer* align 8 dereferenceable(16) %o1, %struct.outer* align 8 dereferenceable(116) %o2) local_unnamed_addr #0 {
+entry:
+  %p1 = getelementptr inbounds %struct.outer, %struct.outer* %o1, i64 0, i32 0
+  %0 = load i64, i64* %p1, align 8
+  %p11 = getelementptr inbounds %struct.outer, %struct.outer* %o2, i64 0, i32 0
+  %1 = load i64, i64* %p11, align 8
+  %cmp = icmp eq i64 %0, %1
+  br i1 %cmp, label %if.then, label %if.end5
+
+if.then:                                          ; preds = %entry
+  %p2 = getelementptr inbounds %struct.outer, %struct.outer* %o1, i64 0, i32 1
+  %2 = load %struct.inner*, %struct.inner** %p2, align 8
+  %p22 = getelementptr inbounds %struct.outer, %struct.outer* %o2, i64 0, i32 1
+  %3 = load %struct.inner*, %struct.inner** %p22, align 8
+  %cmp3 = icmp eq %struct.inner* %2, %3
+  br label %if.end5
+
+if.end5:                                          ; preds = %if.then, %entry
+  %rez.0 = phi i1 [ %cmp3, %if.then ], [ false, %entry ]
+  ret i1 %rez.0
+}
+; Function Attrs: argmemonly nounwind
+declare void @llvm.lifetime.start.p0i8(i64, i8* nocapture) #1
+
+; Function Attrs: argmemonly nounwind
+declare void @llvm.lifetime.end.p0i8(i64, i8* nocapture) #1
+
+attributes #0 = { nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #1 = { argmemonly nounwind }
+attributes #2 = { nounwind }