[PATCH] D22377: [SCEV] trip count calculation for loops with unknown stride
Pankaj Chawla via llvm-commits
llvm-commits at lists.llvm.org
Thu Jul 14 17:08:44 PDT 2016
pankajchawla updated this revision to Diff 64066.
pankajchawla added a comment.
Added check for loopHasNoAbnormalExits() and a comment about bailing out for do-while loops.
https://reviews.llvm.org/D22377
Files:
llvm/lib/Analysis/ScalarEvolution.cpp
llvm/test/Analysis/ScalarEvolution/trip-count-assumed-positive-stride.ll
Index: llvm/lib/Analysis/ScalarEvolution.cpp
===================================================================
--- llvm/lib/Analysis/ScalarEvolution.cpp
+++ llvm/lib/Analysis/ScalarEvolution.cpp
@@ -8659,8 +8659,12 @@
const SCEV *Stride = IV->getStepRecurrence(*this);
+ bool PositiveStride = isKnownPositive(Stride);
+
// Avoid negative or zero stride values
- if (!isKnownPositive(Stride))
+ // We can assume stride to be positive if NoWrap is true and loop does not
+ // have an abnormal exit.
+ if (!PositiveStride && (!NoWrap || !loopHasNoAbnormalExits(L)))
return getCouldNotCompute();
// Avoid proven overflow cases: this will ensure that the backedge taken count
@@ -8674,8 +8678,15 @@
: ICmpInst::ICMP_ULT;
const SCEV *Start = IV->getStart();
const SCEV *End = RHS;
- if (!isLoopEntryGuardedByCond(L, Cond, getMinusSCEV(Start, Stride), RHS))
+ if (!isLoopEntryGuardedByCond(L, Cond, getMinusSCEV(Start, Stride), RHS)) {
+ // If loop entry guard is missing, we could be dealing with a single trip
+ // do-while loop. Stride cannot be assumed to be positive for such loops so
+ // we bail out.
+ if (!PositiveStride)
+ return getCouldNotCompute();
+
End = IsSigned ? getSMaxExpr(RHS, Start) : getUMaxExpr(RHS, Start);
+ }
const SCEV *BECount = computeBECount(getMinusSCEV(End, Start), Stride, false);
Index: llvm/test/Analysis/ScalarEvolution/trip-count-assumed-positive-stride.ll
===================================================================
--- llvm/test/Analysis/ScalarEvolution/trip-count-assumed-positive-stride.ll
+++ llvm/test/Analysis/ScalarEvolution/trip-count-assumed-positive-stride.ll
@@ -0,0 +1,55 @@
+; RUN: opt < %s -analyze -scalar-evolution | FileCheck %s
+
+; ScalarEvolution should be able to compute trip count by assuming stride to be
+; positive by checking the nsw flag on the IV and the presence of a loop entry
+; guard.
+
+; CHECK: Determining loop execution counts for: @foo1
+; CHECK: backedge-taken count is ((-1 + %n) /u %s)
+
+target datalayout = "e-m:e-p:32:32-f64:32:64-f80:32-n8:16:32-S128"
+
+; Function Attrs: norecurse nounwind
+define void @foo1(i32* nocapture %A, i32 %n, i32 %s) #0 {
+entry:
+ %cmp4 = icmp sgt i32 %n, 0
+ br i1 %cmp4, label %for.body, label %for.end
+
+for.body: ; preds = %entry, %for.body
+ %i.05 = phi i32 [ %add, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i.05
+ %0 = load i32, i32* %arrayidx, align 4
+ %inc = add nsw i32 %0, 1
+ store i32 %inc, i32* %arrayidx, align 4
+ %add = add nsw i32 %i.05, %s
+ %cmp = icmp slt i32 %add, %n
+ br i1 %cmp, label %for.body, label %for.end
+
+for.end: ; preds = %for.body, %entry
+ ret void
+}
+
+
+; Loop should have unpredictable trip count without the loop entry guard.
+; CHECK: Determining loop execution counts for: @foo2
+; CHECK: Unpredictable backedge-taken count
+
+; Function Attrs: norecurse nounwind
+define void @foo2(i32* nocapture %A, i32 %n, i32 %s) #0 {
+entry:
+ br label %for.body
+
+for.body: ; preds = %entry, %for.body
+ %i.05 = phi i32 [ %add, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i.05
+ %0 = load i32, i32* %arrayidx, align 4
+ %inc = add nsw i32 %0, 1
+ store i32 %inc, i32* %arrayidx, align 4
+ %add = add nsw i32 %i.05, %s
+ %cmp = icmp slt i32 %add, %n
+ br i1 %cmp, label %for.body, label %for.end
+
+for.end: ; preds = %for.body, %entry
+ ret void
+}
+
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D22377.64066.patch
Type: text/x-patch
Size: 3676 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160715/510c200f/attachment.bin>
More information about the llvm-commits
mailing list