SLPVectorizer: compare entire intrinsic for SLP compatibility.

Some Intrinsics are overloaded to the extent that return type equality (all
that's been checked up to now) does not guarantee that the arguments are the
same. In these cases SLP vectorizer should not recurse into the operands, which
can be achieved by comparing them as "Function *" rather than simply the ID.

llvm-svn: 205424

3 files changed, 23 insertions, 2 deletions

diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
index f6b5b122742..ee322279df3 100644
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -955,11 +955,11 @@ void BoUpSLP::buildTree_rec(ArrayRef<Value *> VL, unsigned Depth) {
         return;
       }
 
-      Intrinsic::ID ID = II->getIntrinsicID();
+      Function *Int = II->getCalledFunction();
 
       for (unsigned i = 1, e = VL.size(); i != e; ++i) {
         IntrinsicInst *II2 = dyn_cast<IntrinsicInst>(VL[i]);
-        if (!II2 || II2->getIntrinsicID() != ID) {
+        if (!II2 || II2->getCalledFunction() != Int) {
           newTreeEntry(VL, false);
           DEBUG(dbgs() << "SLP: mismatched calls:" << *II << "!=" << *VL[i]
                        << "\n");
diff --git a/llvm/test/Transforms/SLPVectorizer/ARM64/lit.local.cfg b/llvm/test/Transforms/SLPVectorizer/ARM64/lit.local.cfg
new file mode 100644
index 00000000000..84ac9811f01
--- /dev/null
+++ b/llvm/test/Transforms/SLPVectorizer/ARM64/lit.local.cfg
@@ -0,0 +1,3 @@
+targets = set(config.root.targets_to_build.split())
+if not 'ARM64' in targets:
+    config.unsupported = True
diff --git a/llvm/test/Transforms/SLPVectorizer/ARM64/mismatched-intrinsics.ll b/llvm/test/Transforms/SLPVectorizer/ARM64/mismatched-intrinsics.ll
new file mode 100644
index 00000000000..3d6da124fc4
--- /dev/null
+++ b/llvm/test/Transforms/SLPVectorizer/ARM64/mismatched-intrinsics.ll
@@ -0,0 +1,18 @@
+; RUN: opt -S -slp-vectorizer %s | FileCheck %s
+target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
+target triple = "arm64-apple-ios5.0.0"
+
+define i64 @mismatched_intrinsics(<4 x i32> %in1, <2 x i32> %in2) nounwind {
+; CHECK-LABEL: @mismatched_intrinsics
+; CHECK: call i64 @llvm.arm64.neon.saddlv.i64.v4i32
+; CHECK: call i64 @llvm.arm64.neon.saddlv.i64.v2i32
+
+  %vaddlvq_s32.i = tail call i64 @llvm.arm64.neon.saddlv.i64.v4i32(<4 x i32> %in1) #2
+  %vaddlv_s32.i = tail call i64 @llvm.arm64.neon.saddlv.i64.v2i32(<2 x i32> %in2) #2
+  %tst = icmp sgt i64 %vaddlvq_s32.i, %vaddlv_s32.i
+  %equal = sext i1 %tst to i64
+  ret i64 %equal
+}
+
+declare i64 @llvm.arm64.neon.saddlv.i64.v4i32(<4 x i32> %in1)
+declare i64 @llvm.arm64.neon.saddlv.i64.v2i32(<2 x i32> %in1)