ScalarEvolution: do not set nuw when creating exprs of form <expr> + <all-ones>.

The nuw constraint will not be satisfied unless <expr> == 0.

This bug has been around since r102234 (in 2010!), but was uncovered by
r251052, which introduced more aggressive optimization of nuw scev expressions.

Differential Revision: http://reviews.llvm.org/D14850

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@253627 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Analysis/ScalarEvolution.cpp b/lib/Analysis/ScalarEvolution.cpp
index 5f927121fc29ae191e4734463c3e951330198267..837461a84bbbb11a8cc76c5e2c5fb335cd09f47a 100644 (file)
--- a/lib/Analysis/ScalarEvolution.cpp
+++ b/lib/Analysis/ScalarEvolution.cpp
@@ -7045,16 +7045,14 @@ bool ScalarEvolution::SimplifyICmpOperands(ICmpInst::Predicate &Pred,
       Pred = ICmpInst::ICMP_ULT;
       Changed = true;
     } else if (!getUnsignedRange(LHS).getUnsignedMin().isMinValue()) {
-      LHS = getAddExpr(getConstant(RHS->getType(), (uint64_t)-1, true), LHS,
-                       SCEV::FlagNUW);
+      LHS = getAddExpr(getConstant(RHS->getType(), (uint64_t)-1, true), LHS);
       Pred = ICmpInst::ICMP_ULT;
       Changed = true;
     }
     break;
   case ICmpInst::ICMP_UGE:
     if (!getUnsignedRange(RHS).getUnsignedMin().isMinValue()) {
-      RHS = getAddExpr(getConstant(RHS->getType(), (uint64_t)-1, true), RHS,
-                       SCEV::FlagNUW);
+      RHS = getAddExpr(getConstant(RHS->getType(), (uint64_t)-1, true), RHS);
       Pred = ICmpInst::ICMP_UGT;
       Changed = true;
     } else if (!getUnsignedRange(LHS).getUnsignedMax().isMaxValue()) {

diff --git a/test/Transforms/IndVarSimplify/zext-nuw.ll b/test/Transforms/IndVarSimplify/zext-nuw.ll
new file mode 100644 (file)
index 0000000..13138de
--- /dev/null
+++ b/test/Transforms/IndVarSimplify/zext-nuw.ll
@@ -0,0 +1,49 @@
+; RUN: opt -indvars -S %s | FileCheck %s
+
+%struct.A = type { i8 }
+
+@c = global %struct.A* null
+@d = global i32 4
+
+define void @_Z3fn1v() {
+  %x2 = load i32, i32* @d
+  %x3 = icmp slt i32 %x2, 1
+  %x4 = select i1 %x3, i32 1, i32 %x2
+  %x5 = load %struct.A*, %struct.A** @c
+  %j.sroa.0.0..sroa_idx = getelementptr %struct.A, %struct.A* %x5, i64 0, i32 0
+  %j.sroa.0.0.copyload = load i8, i8* %j.sroa.0.0..sroa_idx
+  br label %.preheader4.lr.ph
+
+.preheader4.lr.ph:                                ; preds = %0
+  ; CHECK-NOT: add i64 {{.*}}, 4294967296
+  br label %.preheader4
+
+.preheader4:                                      ; preds = %x22, %.preheader4.lr.ph
+  %k.09 = phi i8* [ undef, %.preheader4.lr.ph ], [ %x25, %x22 ]
+  %x8 = icmp ult i32 0, 4
+  br i1 %x8, label %.preheader.lr.ph, label %x22
+
+.preheader.lr.ph:                                 ; preds = %.preheader4
+  br label %.preheader
+
+.preheader:                                       ; preds = %x17, %.preheader.lr.ph
+  %k.17 = phi i8* [ %k.09, %.preheader.lr.ph ], [ %x19, %x17 ]
+  %v.06 = phi i32 [ 0, %.preheader.lr.ph ], [ %x20, %x17 ]
+  br label %x17
+
+x17:                                              ; preds = %.preheader
+  %x18 = sext i8 %j.sroa.0.0.copyload to i64
+  %x19 = getelementptr i8, i8* %k.17, i64 %x18
+  %x20 = add i32 %v.06, 1
+  %x21 = icmp ult i32 %x20, %x4
+  br i1 %x21, label %.preheader, label %._crit_edge.8
+
+._crit_edge.8:                                    ; preds = %x17
+  %split = phi i8* [ %x19, %x17 ]
+  br label %x22
+
+x22:                                              ; preds = %._crit_edge.8, %.preheader4
+  %k.1.lcssa = phi i8* [ %split, %._crit_edge.8 ], [ %k.09, %.preheader4 ]
+  %x25 = getelementptr i8, i8* %k.1.lcssa
+  br label %.preheader4
+}