Do not throw away alignment when generating the DAG for
memset; we may need it to decide between MOVAPS and MOVUPS
later.  Adjust a test that was looking for wrong code.
PR 3866 / 8675131.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@119605 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index 5477e7b..08ef505 100644
--- a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -3487,7 +3487,7 @@
     SDValue Store = DAG.getStore(Chain, dl, Value,
                                  getMemBasePlusOffset(Dst, DstOff, DAG),
                                  DstPtrInfo.getWithOffset(DstOff),
-                                 isVol, false, 0);
+                                 isVol, false, Align);
     OutChains.push_back(Store);
     DstOff += VTSize;
   }
diff --git a/test/CodeGen/X86/memset64-on-x86-32.ll b/test/CodeGen/X86/memset64-on-x86-32.ll
index c0cd271..3f069b4 100644
--- a/test/CodeGen/X86/memset64-on-x86-32.ll
+++ b/test/CodeGen/X86/memset64-on-x86-32.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=i386-apple-darwin   -mcpu=nehalem | grep movaps | count 5
+; RUN: llc < %s -mtriple=i386-apple-darwin   -mcpu=nehalem | grep movups | count 5
 ; RUN: llc < %s -mtriple=i386-apple-darwin   -mcpu=core2   | grep movl   | count 20
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=core2   | grep movq   | count 10
 
diff --git a/test/CodeGen/X86/misaligned-memset.ll b/test/CodeGen/X86/misaligned-memset.ll
new file mode 100644
index 0000000..21f8bf2
--- /dev/null
+++ b/test/CodeGen/X86/misaligned-memset.ll
@@ -0,0 +1,15 @@
+; RUN: llc -mtriple=x86_64-apple-darwin -mcpu=nehalem < %s | FileCheck %s
+
+@a = common global [3 x i64] zeroinitializer, align 16
+
+define i32 @main() nounwind ssp {
+; CHECK: movups
+entry:
+  %retval = alloca i32, align 4
+  store i32 0, i32* %retval
+  call void @llvm.memset.p0i8.i64(i8* bitcast (i64* getelementptr inbounds ([3 x i64]* @a, i32 0, i64 1) to i8*), i8 0, i64 16, i32 1, i1 false)
+  %0 = load i32* %retval
+  ret i32 %0
+}
+
+declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) nounwind