GlobalISel: Fix MMO creation with non-power-of-2 mem size

It should probably just be mandatory for getTgtMemIntrinsic to return
the alignment.

llvm-svn: 352817

diff --git a/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp b/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp
index 49a2262..09bbbe2 100644 (file)
--- a/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp
+++ b/llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp
@@ -1213,12 +1213,13 @@ bool IRTranslator::translateCall(const User &U, MachineIRBuilder &MIRBuilder) {
   TargetLowering::IntrinsicInfo Info;
   // TODO: Add a GlobalISel version of getTgtMemIntrinsic.
   if (TLI.getTgtMemIntrinsic(Info, CI, *MF, ID)) {
-    uint64_t Size = Info.memVT.getStoreSize();
-    if (Info.align == 0)
-      Info.align = Size;
+    unsigned Align = Info.align;
+    if (Align == 0)
+      Align = DL->getABITypeAlignment(Info.memVT.getTypeForEVT(F->getContext()));
 
+    uint64_t Size = Info.memVT.getStoreSize();
     MIB.addMemOperand(MF->getMachineMemOperand(MachinePointerInfo(Info.ptrVal),
-                                               Info.flags, Size, Info.align));
+                                               Info.flags, Size, Align));
   }
 
   return true;

diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll b/llvm/test/CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll
index 4ee1fd3..dd89f9a 100644 (file)
--- a/llvm/test/CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll
+++ b/llvm/test/CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll
@@ -2323,3 +2323,12 @@ define float @test_sqrt_f32(float %x) {
   %y = call float @llvm.sqrt.f32(float %x)
   ret float %y
 }
+
+; CHECK-LABEL: name: test_llvm.aarch64.neon.ld3.v4i32.p0i32
+; CHECK: %1:_(s384) = G_INTRINSIC_W_SIDE_EFFECTS intrinsic(@llvm.aarch64.neon.ld3), %0(p0) :: (load 48 from %ir.ptr, align 64)
+define void @test_llvm.aarch64.neon.ld3.v4i32.p0i32(i32* %ptr) {
+  %arst = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0i32(i32* %ptr)
+  ret void
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0i32(i32*) #3