PR18060 - When we RAUW values with ExtractElement instructions in some cases

we generate PHI nodes with multiple entries from the same basic block but
with different values. Enabling CSE on ExtractElement instructions make sure
that all of the RAUWed instructions are the same.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@195773 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Transforms/Vectorize/SLPVectorizer.cpp b/lib/Transforms/Vectorize/SLPVectorizer.cpp
index f2a099ac691bfb800f5eb041e766b9a7d2d1fcb8..78caab3b50a2e009898c060854bad8b21030031f 100644 (file)
--- a/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -1591,6 +1591,8 @@ Value *BoUpSLP::vectorizeTree() {
     if (PHINode *PN = dyn_cast<PHINode>(Vec)) {
       Builder.SetInsertPoint(PN->getParent()->getFirstInsertionPt());
       Value *Ex = Builder.CreateExtractElement(Vec, Lane);
+      if (Instruction *Ins = dyn_cast<Instruction>(Ex))
+        GatherSeq.insert(Ins);
       User->replaceUsesOfWith(Scalar, Ex);
     } else if (isa<Instruction>(Vec)){
       if (PHINode *PH = dyn_cast<PHINode>(User)) {
@@ -1598,17 +1600,23 @@ Value *BoUpSLP::vectorizeTree() {
           if (PH->getIncomingValue(i) == Scalar) {
             Builder.SetInsertPoint(PH->getIncomingBlock(i)->getTerminator());
             Value *Ex = Builder.CreateExtractElement(Vec, Lane);
+            if (Instruction *Ins = dyn_cast<Instruction>(Ex))
+              GatherSeq.insert(Ins);
             PH->setOperand(i, Ex);
           }
         }
       } else {
         Builder.SetInsertPoint(cast<Instruction>(User));
         Value *Ex = Builder.CreateExtractElement(Vec, Lane);
+        if (Instruction *Ins = dyn_cast<Instruction>(Ex))
+          GatherSeq.insert(Ins);
         User->replaceUsesOfWith(Scalar, Ex);
      }
     } else {
       Builder.SetInsertPoint(F->getEntryBlock().begin());
       Value *Ex = Builder.CreateExtractElement(Vec, Lane);
+      if (Instruction *Ins = dyn_cast<Instruction>(Ex))
+        GatherSeq.insert(Ins);
       User->replaceUsesOfWith(Scalar, Ex);
     }
 

diff --git a/test/Transforms/SLPVectorizer/X86/pr18060.ll b/test/Transforms/SLPVectorizer/X86/pr18060.ll
new file mode 100644 (file)
index 0000000..de9a4c3
--- /dev/null
+++ b/test/Transforms/SLPVectorizer/X86/pr18060.ll
@@ -0,0 +1,47 @@
+; RUN: opt < %s -basicaa -slp-vectorizer -dce -S -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7
+
+target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32-S128"
+target triple = "i386-pc-linux"
+
+; Function Attrs: nounwind
+define i32 @_Z16adjustFixupValueyj(i64 %Value, i32 %Kind) {
+entry:
+  %extract.t = trunc i64 %Value to i32
+  %extract = lshr i64 %Value, 12
+  %extract.t6 = trunc i64 %extract to i32
+  switch i32 %Kind, label %sw.default [
+    i32 0, label %return
+    i32 1, label %return
+    i32 129, label %sw.bb1
+    i32 130, label %sw.bb2
+  ]
+
+sw.default:                                       ; preds = %entry
+  call void @_Z25llvm_unreachable_internalv()
+  unreachable
+
+sw.bb1:                                           ; preds = %entry
+  %shr = lshr i64 %Value, 16
+  %extract.t5 = trunc i64 %shr to i32
+  %extract7 = lshr i64 %Value, 28
+  %extract.t8 = trunc i64 %extract7 to i32
+  br label %sw.bb2
+
+sw.bb2:                                           ; preds = %sw.bb1, %entry
+  %Value.addr.0.off0 = phi i32 [ %extract.t, %entry ], [ %extract.t5, %sw.bb1 ]
+  %Value.addr.0.off12 = phi i32 [ %extract.t6, %entry ], [ %extract.t8, %sw.bb1 ]
+  %conv6 = and i32 %Value.addr.0.off0, 4095
+  %conv4 = shl i32 %Value.addr.0.off12, 16
+  %shl = and i32 %conv4, 983040
+  %or = or i32 %shl, %conv6
+  %or11 = or i32 %or, 8388608
+  br label %return
+
+return:                                           ; preds = %sw.bb2, %entry, %entry
+  %retval.0 = phi i32 [ %or11, %sw.bb2 ], [ %extract.t, %entry ], [ %extract.t, %entry ]
+  ret i32 %retval.0
+}
+
+; Function Attrs: noreturn
+declare void @_Z25llvm_unreachable_internalv()
+