[LoopIdiom] check for safety while expanding
Loop Idiom recognition was generating memset in a case that would result generating a division operation to an unsafe location. Differential Revision: https://reviews.llvm.org/D32674 llvm-svn: 302238
This commit is contained in:
parent
e608f6a632
commit
1c42d135e1
|
@ -783,6 +783,11 @@ bool LoopIdiomRecognize::processLoopStridedStore(
|
||||||
if (NegStride)
|
if (NegStride)
|
||||||
Start = getStartForNegStride(Start, BECount, IntPtr, StoreSize, SE);
|
Start = getStartForNegStride(Start, BECount, IntPtr, StoreSize, SE);
|
||||||
|
|
||||||
|
// TODO: ideally we should still be able to generate memset if SCEV expander
|
||||||
|
// is taught to generate the dependencies at the latest point.
|
||||||
|
if (!isSafeToExpand(Start, *SE))
|
||||||
|
return false;
|
||||||
|
|
||||||
// Okay, we have a strided store "p[i]" of a splattable value. We can turn
|
// Okay, we have a strided store "p[i]" of a splattable value. We can turn
|
||||||
// this into a memset in the loop preheader now if we want. However, this
|
// this into a memset in the loop preheader now if we want. However, this
|
||||||
// would be unsafe to do if there is anything else in the loop that may read
|
// would be unsafe to do if there is anything else in the loop that may read
|
||||||
|
@ -814,6 +819,11 @@ bool LoopIdiomRecognize::processLoopStridedStore(
|
||||||
SCEV::FlagNUW);
|
SCEV::FlagNUW);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// TODO: ideally we should still be able to generate memset if SCEV expander
|
||||||
|
// is taught to generate the dependencies at the latest point.
|
||||||
|
if (!isSafeToExpand(NumBytesS, *SE))
|
||||||
|
return false;
|
||||||
|
|
||||||
Value *NumBytes =
|
Value *NumBytes =
|
||||||
Expander.expandCodeFor(NumBytesS, IntPtr, Preheader->getTerminator());
|
Expander.expandCodeFor(NumBytesS, IntPtr, Preheader->getTerminator());
|
||||||
|
|
||||||
|
|
|
@ -0,0 +1,55 @@
|
||||||
|
; RUN: opt -S < %s -loop-idiom | FileCheck %s
|
||||||
|
; CHECK-NOT: memset
|
||||||
|
; check that memset is not generated (for stores) because that will result
|
||||||
|
; in udiv hoisted out of the loop by the SCEV Expander
|
||||||
|
; TODO: ideally we should be able to generate memset
|
||||||
|
; if SCEV expander is taught to generate the dependencies
|
||||||
|
; at the right point.
|
||||||
|
|
||||||
|
@a = global i32 0, align 4
|
||||||
|
@b = global i32 0, align 4
|
||||||
|
@c = external local_unnamed_addr global [1 x i8], align 1
|
||||||
|
|
||||||
|
define void @e() local_unnamed_addr {
|
||||||
|
entry:
|
||||||
|
%d0 = load i32, i32* @a, align 4
|
||||||
|
%d1 = load i32, i32* @b, align 4
|
||||||
|
br label %for.cond1thread-pre-split
|
||||||
|
|
||||||
|
for.cond1thread-pre-split: ; preds = %for.body5, %entry
|
||||||
|
%div = udiv i32 %d0, %d1
|
||||||
|
br label %for.body5
|
||||||
|
|
||||||
|
for.body5: ; preds = %for.body5, %for.cond1thread-pre-split
|
||||||
|
%indvars.iv = phi i64 [ 0, %for.cond1thread-pre-split ], [ %indvars.iv.next, %for.body5 ]
|
||||||
|
%divx = sext i32 %div to i64
|
||||||
|
%0 = add nsw i64 %divx, %indvars.iv
|
||||||
|
%arrayidx = getelementptr inbounds [1 x i8], [1 x i8]* @c, i64 0, i64 %0
|
||||||
|
store i8 0, i8* %arrayidx, align 1
|
||||||
|
%indvars.iv.next = add nsw i64 %indvars.iv, 1
|
||||||
|
%1 = trunc i64 %indvars.iv.next to i32
|
||||||
|
%tobool4 = icmp eq i32 %1, 0
|
||||||
|
br i1 %tobool4, label %for.cond1thread-pre-split, label %for.body5
|
||||||
|
}
|
||||||
|
|
||||||
|
; The loop's trip count is depending on an unsafe operation
|
||||||
|
; udiv. SCEV expander hoists it out of the loop, so loop-idiom
|
||||||
|
; should check that the memset is not generated in this case.
|
||||||
|
define void @f(i32 %a, i32 %b, i8* nocapture %x) local_unnamed_addr {
|
||||||
|
entry:
|
||||||
|
br label %for.body
|
||||||
|
|
||||||
|
for.body: ; preds = %for.body6, %entry
|
||||||
|
%div = udiv i32 %a, %b
|
||||||
|
%conv = zext i32 %div to i64
|
||||||
|
br label %for.body6
|
||||||
|
|
||||||
|
for.body6: ; preds = %for.body6, %for.body
|
||||||
|
%i.09 = phi i64 [ %inc, %for.body6 ], [ 0, %for.body ]
|
||||||
|
%arrayidx = getelementptr inbounds i8, i8* %x, i64 %i.09
|
||||||
|
store i8 0, i8* %arrayidx, align 1
|
||||||
|
%inc = add nuw nsw i64 %i.09, 1
|
||||||
|
%cmp3 = icmp slt i64 %inc, %conv
|
||||||
|
br i1 %cmp3, label %for.body6, label %for.body
|
||||||
|
}
|
||||||
|
|
Loading…
Reference in New Issue