CSE: allow PerformTrivialCoalescing to check copies across basic block

boundaries.

Given the following case:
BB0
  %vreg1<def> = SUBrr %vreg0, %vreg7
  %vreg2<def> = COPY %vreg7
BB1
  %vreg10<def> = SUBrr %vreg0, %vreg2
We should be able to CSE between SUBrr in BB0 and SUBrr in BB1.

rdar://12462006

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168717 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/MachineCSE.cpp b/lib/CodeGen/MachineCSE.cpp
index dbc41de..56cea4e 100644 (file)
--- a/lib/CodeGen/MachineCSE.cpp
+++ b/lib/CodeGen/MachineCSE.cpp
@@ -126,8 +126,6 @@ bool MachineCSE::PerformTrivialCoalescing(MachineInstr *MI,
       // deleted.
       continue;
     MachineInstr *DefMI = MRI->getVRegDef(Reg);
-    if (DefMI->getParent() != MBB)
-      continue;
     if (!DefMI->isCopy())
       continue;
     unsigned SrcReg = DefMI->getOperand(1).getReg();

diff --git a/test/CodeGen/ARM/machine-cse-cmp.ll b/test/CodeGen/ARM/machine-cse-cmp.ll
index 3ac7d77..03abd76 100644 (file)
--- a/test/CodeGen/ARM/machine-cse-cmp.ll
+++ b/test/CodeGen/ARM/machine-cse-cmp.ll
@@ -45,3 +45,35 @@ for.cond1.preheader:                              ; preds = %entry
 }
 
 declare void @llvm.memset.p0i8.i32(i8* nocapture, i8, i32, i32, i1) nounwind
+
+; rdar://12462006
+define i8* @f3(i8* %base, i32* nocapture %offset, i32 %size) nounwind {
+entry:
+; CHECK: f3:
+; CHECK-NOT: sub
+; CHECK: cmp
+; CHECK: blt
+%0 = load i32* %offset, align 4
+%cmp = icmp slt i32 %0, %size
+%s = sub nsw i32 %0, %size
+%size2 = sub nsw i32 %size, 0
+br i1 %cmp, label %return, label %if.end
+
+if.end:
+; We are checking cse between %sub here and %s in entry block.
+%sub = sub nsw i32 %0, %size2
+%s2 = sub nsw i32 %s, %size
+%s3 = sub nsw i32 %sub, %s2
+; CHECK: sub [[R1:r[0-9]+]], [[R2:r[0-9]+]], r2
+; CHECK: sub [[R3:r[0-9]+]], [[R1]], r2
+; CHECK: sub [[R4:r[0-9]+]], [[R1]], [[R3]]
+; CHECK-NOT: sub
+; CHECK: str
+store i32 %s3, i32* %offset, align 4
+%add.ptr = getelementptr inbounds i8* %base, i32 %sub
+br label %return
+
+return:
+%retval.0 = phi i8* [ %add.ptr, %if.end ], [ null, %entry ]
+ret i8* %retval.0
+}