[llvm] r185955 - Don't crash in SE dealing with ashr x, -1

[Hal Finkel]

Author: hfinkel
Date: Tue Jul  9 13:16:16 2013
New Revision: 185955

URL: http://llvm.org/viewvc/llvm-project?rev=185955&view=rev
Log:
Don't crash in SE dealing with ashr x, -1

ScalarEvolution::getSignedRange uses ComputeNumSignBits from ValueTracking on
ashr instructions. ComputeNumSignBits can return zero, but this case was not
handled correctly by the code in getSignedRange which was calling:
  APInt::getSignedMinValue(BitWidth).ashr(NS - 1)
with NS = 0, resulting in an assertion failure in APInt::ashr.

Now, we just return the conservative result (as with NS == 1).

Another bug found by llvm-stress.

Added:
    llvm/trunk/test/CodeGen/PowerPC/ashr-neg1.ll
Modified:
    llvm/trunk/lib/Analysis/ScalarEvolution.cpp

Modified: llvm/trunk/lib/Analysis/ScalarEvolution.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/ScalarEvolution.cpp?rev=185955&r1=185954&r2=185955&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/ScalarEvolution.cpp (original)
+++ llvm/trunk/lib/Analysis/ScalarEvolution.cpp Tue Jul  9 13:16:16 2013
@@ -3589,7 +3589,7 @@ ScalarEvolution::getSignedRange(const SC
     if (!U->getValue()->getType()->isIntegerTy() && !TD)
       return setSignedRange(U, ConservativeResult);
     unsigned NS = ComputeNumSignBits(U->getValue(), TD);
-    if (NS == 1)
+    if (NS <= 1)
       return setSignedRange(U, ConservativeResult);
     return setSignedRange(U, ConservativeResult.intersectWith(
       ConstantRange(APInt::getSignedMinValue(BitWidth).ashr(NS - 1),

Added: llvm/trunk/test/CodeGen/PowerPC/ashr-neg1.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/ashr-neg1.ll?rev=185955&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/ashr-neg1.ll (added)
+++ llvm/trunk/test/CodeGen/PowerPC/ashr-neg1.ll Tue Jul  9 13:16:16 2013
@@ -0,0 +1,18 @@
+; RUN: llc -mtriple=powerpc64-unknown-linux-gnu -mcpu=pwr7 < %s
+target triple = "powerpc64-unknown-linux-gnu"
+
+define void @autogen_SD30723(i32) {
+BB:
+  br label %CF80
+
+CF80:                                             ; preds = %CF80, %BB
+  %B = ashr i32 %0, -1
+  br i1 undef, label %CF80, label %CF84
+
+CF84:                                             ; preds = %CF84, %CF80
+  %Cmp62 = icmp sge i32 undef, %B
+  br i1 %Cmp62, label %CF84, label %CF85
+
+CF85:                                             ; preds = %CF85, %CF84
+  br label %CF85
+}