[llvm] r194876 - LoopVectorizer: Use abi alignment for accesses with no alignment
Arnold Schwaighofer
aschwaighofer at apple.com
Fri Nov 15 15:09:33 PST 2013
Author: arnolds
Date: Fri Nov 15 17:09:33 2013
New Revision: 194876
URL: http://llvm.org/viewvc/llvm-project?rev=194876&view=rev
Log:
LoopVectorizer: Use abi alignment for accesses with no alignment
When we vectorize a scalar access with no alignment specified, we have to set
the target's abi alignment of the scalar access on the vectorized access.
Using the same alignment of zero would be wrong because most targets will have a
bigger abi alignment for vector types.
This probably fixes PR17878.
Added:
llvm/trunk/test/Transforms/LoopVectorize/align.ll
Modified:
llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp
Modified: llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp?rev=194876&r1=194875&r2=194876&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp (original)
+++ llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp Fri Nov 15 17:09:33 2013
@@ -1205,6 +1205,10 @@ void InnerLoopVectorizer::vectorizeMemor
Type *DataTy = VectorType::get(ScalarDataTy, VF);
Value *Ptr = LI ? LI->getPointerOperand() : SI->getPointerOperand();
unsigned Alignment = LI ? LI->getAlignment() : SI->getAlignment();
+ // An alignment of 0 means target abi alignment. We need to use the scalar's
+ // target abi alignment in such a case.
+ if (!Alignment)
+ Alignment = DL->getABITypeAlignment(ScalarDataTy);
unsigned AddressSpace = Ptr->getType()->getPointerAddressSpace();
unsigned ScalarAllocatedSize = DL->getTypeAllocSize(ScalarDataTy);
unsigned VectorElementSize = DL->getTypeStoreSize(DataTy)/VF;
Added: llvm/trunk/test/Transforms/LoopVectorize/align.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoopVectorize/align.ll?rev=194876&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/LoopVectorize/align.ll (added)
+++ llvm/trunk/test/Transforms/LoopVectorize/align.ll Fri Nov 15 17:09:33 2013
@@ -0,0 +1,33 @@
+; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -S | FileCheck %s
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
+target triple = "x86_64-apple-macosx10.8.0"
+
+; Make sure we output the abi alignment if no alignment is specified.
+
+;CHECK-LABEL: @align
+;CHECK: load <4 x i32>* {{.*}} align 4
+;CHECK: load <4 x i32>* {{.*}} align 4
+;CHECK: store <4 x i32> {{.*}} align 4
+
+define void @align(i32* %a, i32* %b, i32* %c) nounwind uwtable ssp {
+ br label %1
+
+; <label>:1 ; preds = %1, %0
+ %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ]
+ %2 = getelementptr inbounds i32* %b, i64 %indvars.iv
+ %3 = load i32* %2
+ %4 = getelementptr inbounds i32* %c, i64 %indvars.iv
+ %5 = load i32* %4
+ %6 = add nsw i32 %5, %3
+ %7 = getelementptr inbounds i32* %a, i64 %indvars.iv
+ store i32 %6, i32* %7
+ %indvars.iv.next = add i64 %indvars.iv, 1
+ %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+ %exitcond = icmp eq i32 %lftr.wideiv, 128
+ br i1 %exitcond, label %8, label %1
+
+; <label>:8 ; preds = %1
+ ret void
+}
+
More information about the llvm-commits
mailing list