[ValueTracking] Use KnownOnes to provide a better bound on known zeros for ctlz/cttz...

This patch uses KnownOnes of the input of ctlz/cttz to bound the value that can be returned from these intrinsics. This makes these intrinsics more similar to the handling for ctpop which already uses known bits to produce a similar bound.

Differential Revision: https://reviews.llvm.org/D32521

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302444 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Analysis/ValueTracking.cpp b/lib/Analysis/ValueTracking.cpp
index 7a0cb28ea4602dfeac922ea18ae7dbcc2dc9b34f..39a92066a487c527294e8938bc01d84d4acbce44 100644 (file)
--- a/lib/Analysis/ValueTracking.cpp
+++ b/lib/Analysis/ValueTracking.cpp
@@ -1392,12 +1392,25 @@ static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
         Known.Zero |= Known2.Zero.byteSwap();
         Known.One |= Known2.One.byteSwap();
         break;
-      case Intrinsic::ctlz:
+      case Intrinsic::ctlz: {
+        computeKnownBits(I->getOperand(0), Known2, Depth + 1, Q);
+        // If we have a known 1, its position is our upper bound.
+        unsigned PossibleLZ = Known2.One.countLeadingZeros();
+        // If this call is undefined for 0, the result will be less than 2^n.
+        if (II->getArgOperand(1) == ConstantInt::getTrue(II->getContext()))
+          PossibleLZ = std::min(PossibleLZ, BitWidth - 1);
+        unsigned LowBits = Log2_32(PossibleLZ)+1;
+        Known.Zero.setBitsFrom(LowBits);
+        break;
+      }
       case Intrinsic::cttz: {
-        unsigned LowBits = Log2_32(BitWidth)+1;
+        computeKnownBits(I->getOperand(0), Known2, Depth + 1, Q);
+        // If we have a known 1, its position is our upper bound.
+        unsigned PossibleTZ = Known2.One.countTrailingZeros();
         // If this call is undefined for 0, the result will be less than 2^n.
         if (II->getArgOperand(1) == ConstantInt::getTrue(II->getContext()))
-          LowBits -= 1;
+          PossibleTZ = std::min(PossibleTZ, BitWidth - 1);
+        unsigned LowBits = Log2_32(PossibleTZ)+1;
         Known.Zero.setBitsFrom(LowBits);
         break;
       }

diff --git a/test/Transforms/InstCombine/intrinsics.ll b/test/Transforms/InstCombine/intrinsics.ll
index 66ab7f48aeffc86715a48f335547cfed246dabca..5654b265da5865fc64a781ec00757ef15491d7fc 100644 (file)
--- a/test/Transforms/InstCombine/intrinsics.ll
+++ b/test/Transforms/InstCombine/intrinsics.ll
@@ -1,3 +1,4 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt -instcombine -S < %s | FileCheck %s
 
 %overflow.result = type {i8, i1}
@@ -283,14 +284,24 @@ define i32 @cttz(i32 %a) {
 
 define i1 @cttz_knownbits(i32 %arg) {
 ; CHECK-LABEL: @cttz_knownbits(
+; CHECK-NEXT:    ret i1 false
+;
+  %or = or i32 %arg, 4
+  %cnt = call i32 @llvm.cttz.i32(i32 %or, i1 true) nounwind readnone
+  %res = icmp eq i32 %cnt, 4
+  ret i1 %res
+}
+
+define i1 @cttz_knownbits2(i32 %arg) {
+; CHECK-LABEL: @cttz_knownbits2(
 ; CHECK-NEXT:    [[OR:%.*]] = or i32 [[ARG:%.*]], 4
 ; CHECK-NEXT:    [[CNT:%.*]] = call i32 @llvm.cttz.i32(i32 [[OR]], i1 true)
-; CHECK-NEXT:    [[RES:%.*]] = icmp eq i32 [[CNT]], 4
+; CHECK-NEXT:    [[RES:%.*]] = icmp eq i32 [[CNT]], 2
 ; CHECK-NEXT:    ret i1 [[RES]]
 ;
   %or = or i32 %arg, 4
   %cnt = call i32 @llvm.cttz.i32(i32 %or, i1 true) nounwind readnone
-  %res = icmp eq i32 %cnt, 4
+  %res = icmp eq i32 %cnt, 2
   ret i1 %res
 }
 
@@ -306,14 +317,24 @@ define i8 @ctlz(i8 %a) {
 
 define i1 @ctlz_knownbits(i8 %arg) {
 ; CHECK-LABEL: @ctlz_knownbits(
+; CHECK-NEXT:    ret i1 false
+;
+  %or = or i8 %arg, 32
+  %cnt = call i8 @llvm.ctlz.i8(i8 %or, i1 true) nounwind readnone
+  %res = icmp eq i8 %cnt, 4
+  ret i1 %res
+}
+
+define i1 @ctlz_knownbits2(i8 %arg) {
+; CHECK-LABEL: @ctlz_knownbits2(
 ; CHECK-NEXT:    [[OR:%.*]] = or i8 [[ARG:%.*]], 32
 ; CHECK-NEXT:    [[CNT:%.*]] = call i8 @llvm.ctlz.i8(i8 [[OR]], i1 true)
-; CHECK-NEXT:    [[RES:%.*]] = icmp eq i8 [[CNT]], 4
+; CHECK-NEXT:    [[RES:%.*]] = icmp eq i8 [[CNT]], 2
 ; CHECK-NEXT:    ret i1 [[RES]]
 ;
   %or = or i8 %arg, 32
   %cnt = call i8 @llvm.ctlz.i8(i8 %or, i1 true) nounwind readnone
-  %res = icmp eq i8 %cnt, 4
+  %res = icmp eq i8 %cnt, 2
   ret i1 %res
 }