Last week we were generating code with duplicate induction variables in this
authorBob Wilson <bob.wilson@apple.com>
Mon, 15 Feb 2010 21:56:40 +0000 (21:56 +0000)
committerBob Wilson <bob.wilson@apple.com>
Mon, 15 Feb 2010 21:56:40 +0000 (21:56 +0000)
test, but the problem seems to have gone away today.  Add a check to make sure
it doesn't come back.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@96277 91177308-0d34-0410-b5e6-96231b3b80d8

test/CodeGen/Thumb2/2010-02-11-phi-cycle.ll

index 0f23ee75683c8ebcdc2f1c3bfd3edee6cbcc8ae2..363f5719d17cfab039b019ed3ae3fdcabef7d6b0 100644 (file)
@@ -58,6 +58,10 @@ bb1:                                              ; preds = %bb
   br label %bb2
 
 bb2:                                              ; preds = %bb1, %bb
+; also check for duplicate induction variables (radar 7645034)
+; CHECK: subs r{{.*}}, #1
+; CHECK-NOT: subs r{{.*}}, #1
+; CHECK: pop
   %u.0 = phi i64 [ %ins, %bb1 ], [ %u.17, %bb ]   ; <i64> [#uses=2]
   %indvar.next = add i32 %indvar, 1               ; <i32> [#uses=2]
   %exitcond = icmp eq i32 %indvar.next, %tmp      ; <i1> [#uses=1]