[llvm] c868335 - [SCEV] ScalarEvolution::createSCEV(): clarify no-wrap flag propagation for shift by bitwidth-1

[Roman Lebedev via llvm-commits]

Author: Roman Lebedev
Date: 2020-06-06T13:02:07+03:00
New Revision: c868335e243262876b0db3328f79153bd2d27d00

URL: https://github.com/llvm/llvm-project/commit/c868335e243262876b0db3328f79153bd2d27d00
DIFF: https://github.com/llvm/llvm-project/commit/c868335e243262876b0db3328f79153bd2d27d00.diff

LOG: [SCEV] ScalarEvolution::createSCEV(): clarify no-wrap flag propagation for shift by bitwidth-1

Summary:
There was this comment here previously:
```
-        // It is currently not resolved how to interpret NSW for left
-        // shift by BitWidth - 1, so we avoid applying flags in that
-        // case. Remove this check (or this comment) once the situation
-        // is resolved. See
-        // http://lists.llvm.org/pipermail/llvm-dev/2015-April/084195.html
-        // and http://reviews.llvm.org/D8890 .
```
But langref was fixed in rL286785, and the behavior is pretty obvious:
http://volta.cs.utah.edu:8080/z/MM4WZP
^ nuw can always be propagated. nsw can be propagated if
either nuw is specified, or the shift is by *less* than bitwidth-1.

This mimics similar D81189 Reassociate change, alive2 is happy about that one.

I'm not sure `NUW` isn't being printed, but that seems unrelated.

Reviewers: mkazantsev, reames, sanjoy, nlopes, craig.topper, efriedma

Reviewed By: efriedma

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D81243

Added: 
    

Modified: 
    llvm/lib/Analysis/ScalarEvolution.cpp
    llvm/test/Analysis/ScalarEvolution/flags-from-poison.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Analysis/ScalarEvolution.cpp b/llvm/lib/Analysis/ScalarEvolution.cpp
index 27530f103011..4d7def2e4d0e 100644
--- a/llvm/lib/Analysis/ScalarEvolution.cpp
+++ b/llvm/lib/Analysis/ScalarEvolution.cpp
@@ -6453,15 +6453,19 @@ const SCEV *ScalarEvolution::createSCEV(Value *V) {
         if (SA->getValue().uge(BitWidth))
           break;
 
-        // It is currently not resolved how to interpret NSW for left
-        // shift by BitWidth - 1, so we avoid applying flags in that
-        // case. Remove this check (or this comment) once the situation
-        // is resolved. See
-        // http://lists.llvm.org/pipermail/llvm-dev/2015-April/084195.html
-        // and http://reviews.llvm.org/D8890 .
+        // We can safely preserve the nuw flag in all cases. It's also safe to
+        // turn a nuw nsw shl into a nuw nsw mul. However, nsw in isolation
+        // requires special handling. It can be preserved as long as we're not
+        // left shifting by bitwidth - 1.
         auto Flags = SCEV::FlagAnyWrap;
-        if (BO->Op && SA->getValue().ult(BitWidth - 1))
-          Flags = getNoWrapFlagsFromUB(BO->Op);
+        if (BO->Op) {
+          auto MulFlags = getNoWrapFlagsFromUB(BO->Op);
+          if ((MulFlags & SCEV::FlagNSW) &&
+              ((MulFlags & SCEV::FlagNUW) || SA->getValue().ult(BitWidth - 1)))
+            Flags = (SCEV::NoWrapFlags)(Flags | SCEV::FlagNSW);
+          if (MulFlags & SCEV::FlagNUW)
+            Flags = (SCEV::NoWrapFlags)(Flags | SCEV::FlagNUW);
+        }
 
         Constant *X = ConstantInt::get(
             getContext(), APInt::getOneBitSet(BitWidth, SA->getZExtValue()));

diff  --git a/llvm/test/Analysis/ScalarEvolution/flags-from-poison.ll b/llvm/test/Analysis/ScalarEvolution/flags-from-poison.ll
index 2b60192b1fb2..122d7dc5f171 100644
--- a/llvm/test/Analysis/ScalarEvolution/flags-from-poison.ll
+++ b/llvm/test/Analysis/ScalarEvolution/flags-from-poison.ll
@@ -559,11 +559,11 @@ loop:
   %i = phi i32 [ %nexti, %loop ], [ %start, %entry ]
 
 ; CHECK: %index32 =
-; CHECK: --> {(-2147483648 * %start),+,-2147483648}<%loop>
+; CHECK: --> {(-2147483648 * %start),+,-2147483648}<nsw><%loop>
   %index32 = shl nuw nsw i32 %i, 31
 
 ; CHECK: %index64 =
-; CHECK: --> (sext i32 {(-2147483648 * %start),+,-2147483648}<%loop>
+; CHECK: --> {(sext i32 (-2147483648 * %start) to i64),+,-2147483648}<nsw><%loop>
   %index64 = sext i32 %index32 to i64
 
   %ptr = getelementptr inbounds float, float* %input, i64 %index64