[LV] Deny irregular types in interleavedAccessCanBeWidened

Summary:
Avoid that loop vectorizer creates loads/stores of vectors
with "irregular" types when interleaving. An example of
an irregular type is x86_fp80 that is 80 bits, but that
may have an allocation size that is 96 bits. So an array
of x86_fp80 is not bitcast compatible with a vector
of the same type.

Not sure if interleavedAccessCanBeWidened is the best
place for this check, but it solves the problem seen
in the added test case. And it is the same kind of check
that already exists in memoryInstructionCanBeWidened.

Reviewers: fhahn, Ayal, craig.topper

Reviewed By: fhahn

Subscribers: hiraditya, rkruppe, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D63386

llvm-svn: 363547

2 files changed, 36 insertions, 0 deletions

diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 79528e5927e..d0d72d8de90 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -4412,6 +4412,13 @@ bool LoopVectorizationCostModel::interleavedAccessCanBeWidened(Instruction *I,
   auto *Group = getInterleavedAccessGroup(I);
   assert(Group && "Must have a group.");
 
+  // If the instruction's allocated size doesn't equal it's type size, it
+  // requires padding and will be scalarized.
+  auto &DL = I->getModule()->getDataLayout();
+  auto *ScalarTy = getMemInstValueType(I);
+  if (hasIrregularType(ScalarTy, DL, VF))
+    return false;
+
   // Check if masking is required.
   // A Group may need masking for one of two reasons: it resides in a block that
   // needs predication, or it was decided to use masking to deal with gaps.
diff --git a/llvm/test/Transforms/LoopVectorize/X86/x86_fp80-interleaved-access.ll b/llvm/test/Transforms/LoopVectorize/X86/x86_fp80-interleaved-access.ll
new file mode 100644
index 00000000000..80c7eecdeed
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/X86/x86_fp80-interleaved-access.ll
@@ -0,0 +1,29 @@
+; RUN: opt < %s -enable-interleaved-mem-accesses=true -force-vector-width=4 -loop-vectorize -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-macosx10.7.0"
+
+; Verify that we do not get any loads of vectors with x86_fp80 elements.
+;
+; CHECK-NOT: load {{.*}} x x86_fp80
+define x86_fp80 @foo(x86_fp80* %a) {
+entry:
+  br label %for.body
+
+for.cond.cleanup:
+  ret x86_fp80 %3
+
+for.body:
+  %i.09 = phi i16 [ 0, %entry ], [ %add3, %for.body ]
+  %res.08 = phi x86_fp80 [ undef, %entry ], [ %3, %for.body ]
+  %arrayidx = getelementptr inbounds x86_fp80, x86_fp80* %a, i16 %i.09
+  %0 = load x86_fp80, x86_fp80* %arrayidx, align 1
+  %add = or i16 %i.09, 1
+  %arrayidx2 = getelementptr inbounds x86_fp80, x86_fp80* %a, i16 %add
+  %1 = load x86_fp80, x86_fp80* %arrayidx2, align 1
+  %2 = fadd fast x86_fp80 %0, %res.08
+  %3 = fadd fast x86_fp80 %2, %1
+  %add3 = add nuw nsw i16 %i.09, 2
+  %cmp = icmp ult i16 %add3, 400
+  br i1 %cmp, label %for.body, label %for.cond.cleanup
+}