Reapply "[CGP] Check for existing inttotpr before creating new one"

Original commit: r351582

llvm-svn: 351618

2 files changed, 81 insertions, 4 deletions

diff --git a/llvm/lib/CodeGen/CodeGenPrepare.cpp b/llvm/lib/CodeGen/CodeGenPrepare.cpp
index c35f8666fa3..e33178c961d 100644
--- a/llvm/lib/CodeGen/CodeGenPrepare.cpp
+++ b/llvm/lib/CodeGen/CodeGenPrepare.cpp
@@ -4664,13 +4664,26 @@ bool CodeGenPrepare::optimizeMemoryInst(Instruction *MemoryInst, Value *Addr,
     // will look through it and provide only the integer value. In that case,
     // use it here.
     if (!DL->isNonIntegralPointerType(Addr->getType())) {
+      const auto getResultPtr = [MemoryInst, Addr,
+                                 &Builder](Value *Reg) -> Value * {
+        BasicBlock *BB = MemoryInst->getParent();
+        for (User *U : Reg->users())
+          if (auto *I2P = dyn_cast<IntToPtrInst>(U))
+            if (I2P->getType() == Addr->getType() && I2P->getParent() == BB) {
+              if (isa<Instruction>(Reg) &&
+                  cast<Instruction>(Reg)->getParent() == BB)
+                I2P->moveAfter(cast<Instruction>(Reg));
+              else
+                I2P->moveBefore(BB->getFirstNonPHI());
+              return I2P;
+            }
+        return Builder.CreateIntToPtr(Reg, Addr->getType(), "sunkaddr");
+      };
       if (!ResultPtr && AddrMode.BaseReg) {
-        ResultPtr = Builder.CreateIntToPtr(AddrMode.BaseReg, Addr->getType(),
-                                           "sunkaddr");
+        ResultPtr = getResultPtr(AddrMode.BaseReg);
         AddrMode.BaseReg = nullptr;
       } else if (!ResultPtr && AddrMode.Scale == 1) {
-        ResultPtr = Builder.CreateIntToPtr(AddrMode.ScaledReg, Addr->getType(),
-                                           "sunkaddr");
+        ResultPtr = getResultPtr(AddrMode.ScaledReg);
         AddrMode.Scale = 0;
       }
     }
diff --git a/llvm/test/Transforms/CodeGenPrepare/X86/sink-addrmode-cse-inttoptrs.ll b/llvm/test/Transforms/CodeGenPrepare/X86/sink-addrmode-cse-inttoptrs.ll
new file mode 100644
index 00000000000..ad33a2da5de
--- /dev/null
+++ b/llvm/test/Transforms/CodeGenPrepare/X86/sink-addrmode-cse-inttoptrs.ll
@@ -0,0 +1,64 @@
+; RUN: opt -mtriple=x86_64-- -codegenprepare                        %s -S -o - | FileCheck %s --check-prefixes=CGP,COMMON
+; RUN: opt -mtriple=x86_64-- -codegenprepare -load-store-vectorizer %s -S -o - | FileCheck %s --check-prefixes=LSV,COMMON
+
+; Make sure CodeGenPrepare doesn't emit multiple inttoptr instructions
+; of the same integer value while sinking address computations, but
+; rather CSEs them on the fly: excessive inttoptr's confuse SCEV
+; into thinking that related pointers have nothing to do with each other.
+;
+; Triggering this problem involves having just right addressing modes,
+; and verifying that the motivating pass (LoadStoreVectorizer) is able
+; to benefit from it - just right LSV-policies. Hence the atypical combination
+; of the target and datalayout / address spaces in this test.
+
+target datalayout = "p1:32:32:32"
+
+define void @test1(i32 %tmp, i32 %off) {
+; COMMON-LABEL: @test1
+; CGP:     = inttoptr
+; CGP-NOT: = inttoptr
+; LSV:     = load <2 x float>
+; LSV:     = load <2 x float>
+entry:
+  %tmp1 = inttoptr i32 %tmp to float addrspace(1)*
+  %arrayidx.i.7 = getelementptr inbounds float, float addrspace(1)* %tmp1, i32 %off
+  %add20.i.7 = add i32 %off, 1
+  %arrayidx22.i.7 = getelementptr inbounds float, float addrspace(1)* %tmp1, i32 %add20.i.7
+  br label %for.body
+
+for.body:
+  %tmp8 = phi float [ undef, %entry ], [ %tmp62, %for.body ]
+  %tmp28 = load float, float addrspace(1)* %arrayidx.i.7
+  %tmp29 = load float, float addrspace(1)* %arrayidx22.i.7
+  %arrayidx.i321.7 = getelementptr inbounds float, float addrspace(1)* %tmp1, i32 0
+  %tmp43 = load float, float addrspace(1)* %arrayidx.i321.7
+  %arrayidx22.i327.7 = getelementptr inbounds float, float addrspace(1)* %tmp1, i32 1
+  %tmp44 = load float, float addrspace(1)* %arrayidx22.i327.7
+  %tmp62 = tail call fast float @foo(float %tmp8, float %tmp44, float %tmp43, float %tmp29, float %tmp28)
+  br label %for.body
+}
+
+define void @test2(i64 %a, i64 %b, i64 %c) {
+; COMMON-LABEL: @test2
+; CGP:    loop:
+; CGP-NEXT: %mul =
+; CGP-NEXT: = inttoptr i64 %mul
+; CGP-NOT:  = inttoptr
+; LSV:      store <2 x i64>
+entry:
+  %mul.neg.i630 = add nsw i64 %a, -16
+  br label %loop
+
+loop:
+  %mul = mul nsw i64 %b, -16
+  %sub.i631 = add nsw i64 %mul.neg.i630, %mul
+  %tmp = inttoptr i64 %sub.i631 to i8*
+  %tmp1 = inttoptr i64 %sub.i631 to i64*
+  store i64 %c, i64* %tmp1, align 16
+  %arrayidx172 = getelementptr inbounds i8, i8* %tmp, i64 8
+  %tmp2 = bitcast i8* %arrayidx172 to i64*
+  store i64 42, i64* %tmp2, align 8
+  br label %loop
+}
+
+declare float @foo(float, float, float, float, float)