maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/polly/test/ScopInfo/memcpy-raw-source.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

62 lines
3.4 KiB
LLVM
Raw Normal View History

[FIX] Resolve test failure in polly/test/ScopInfo/memcpy-raw-source.ll scoped-noalias -> scoped-noalias-aa reference: https://reviews.llvm.org/D84542 Reviewed By: aeubanks Differential Revision: https://reviews.llvm.org/D84720
2020-07-29 00:12:31 +08:00
; RUN: opt %loadPolly -basic-aa -scoped-noalias-aa -tbaa -polly-scops -analyze < %s
[ScopInfo] Fix: use raw source pointer. When adding an llvm.memcpy instruction to AliasSetTracker, it uses the raw source and target pointers which preserve bitcasts. MemAccInst::getPointerOperand() also returns the raw target pointers, but Scop::buildAliasGroups() did not for the source pointer. This lead to mismatches between AliasSetTracker and ScopInfo on which pointer to use. Fixed by also using raw pointers in Scop::buildAliasGroups(). llvm-svn: 285071
2016-10-25 21:37:43 +08:00
;
; Ensure that ScopInfo's alias analysis llvm.memcpy for,
; like the AliasSetTracker, preserves bitcasts.
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
@tonemasks = external global [17 x [6 x [56 x float]]], align 16
; Function Attrs: argmemonly nounwind
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture writeonly, i8* nocapture readonly, i64, i32, i1) #0
; Function Attrs: nounwind uwtable
define void @setup_tone_curves() #1 {
entry:
%workc = alloca [17 x [8 x [56 x float]]], align 16
br label %for.cond7.preheader
for.cond7.preheader: ; preds = %for.cond7.preheader, %entry
%indvars.iv45 = phi i64 [ %indvars.iv.next46, %for.cond7.preheader ], [ 0, %entry ]
%indvars.iv.next46 = add nuw nsw i64 %indvars.iv45, 1
%exitcond48 = icmp ne i64 %indvars.iv.next46, 56
br i1 %exitcond48, label %for.cond7.preheader, label %for.body36
for.body36: ; preds = %for.body36, %for.cond7.preheader
%indvars.iv49 = phi i64 [ %indvars.iv.next50, %for.body36 ], [ 0, %for.cond7.preheader ]
%arraydecay42 = getelementptr inbounds [17 x [8 x [56 x float]]], [17 x [8 x [56 x float]]]* %workc, i64 0, i64 0, i64 0, i64 0
%0 = bitcast float* %arraydecay42 to i8*
%arraydecay47 = getelementptr inbounds [17 x [6 x [56 x float]]], [17 x [6 x [56 x float]]]* @tonemasks, i64 0, i64 0, i64 %indvars.iv49, i64 0
%1 = bitcast float* %arraydecay47 to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* %1, i64 224, i32 16, i1 false)
%indvars.iv.next50 = add nuw nsw i64 %indvars.iv49, 1
br i1 false, label %for.body36, label %for.end50
for.end50: ; preds = %for.body36
%arrayidx38 = getelementptr inbounds [17 x [8 x [56 x float]]], [17 x [8 x [56 x float]]]* %workc, i64 0, i64 0
%2 = bitcast [8 x [56 x float]]* %arrayidx38 to i8*
%arraydecay58 = getelementptr inbounds [17 x [6 x [56 x float]]], [17 x [6 x [56 x float]]]* @tonemasks, i64 0, i64 0, i64 0, i64 0
%3 = bitcast float* %arraydecay58 to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %2, i8* %3, i64 224, i32 16, i1 false)
br label %for.body74
for.body74: ; preds = %for.body74, %for.end50
%indvars.iv53 = phi i64 [ %indvars.iv.next54, %for.body74 ], [ 0, %for.end50 ]
%arrayidx99 = getelementptr inbounds [17 x [8 x [56 x float]]], [17 x [8 x [56 x float]]]* %workc, i64 0, i64 0, i64 0, i64 %indvars.iv53
%4 = load float, float* %arrayidx99, align 4
store float undef, float* %arrayidx99, align 4
%indvars.iv.next54 = add nuw nsw i64 %indvars.iv53, 1
%exitcond57 = icmp ne i64 %indvars.iv.next54, 56
br i1 %exitcond57, label %for.body74, label %for.inc104
for.inc104: ; preds = %for.body74
[tests] Make sure tests do not end in 'unreachable' There is no point in optimizing unreachable code, hence our test cases should always return. This commit is part of a series that makes Polly more robust on the presence of unreachables. llvm-svn: 297147
2017-03-07 23:17:23 +08:00
ret void
[ScopInfo] Fix: use raw source pointer. When adding an llvm.memcpy instruction to AliasSetTracker, it uses the raw source and target pointers which preserve bitcasts. MemAccInst::getPointerOperand() also returns the raw target pointers, but Scop::buildAliasGroups() did not for the source pointer. This lead to mismatches between AliasSetTracker and ScopInfo on which pointer to use. Fixed by also using raw pointers in Scop::buildAliasGroups(). llvm-svn: 285071
2016-10-25 21:37:43 +08:00
}
attributes #0 = { argmemonly nounwind }
Revert "[NFC] remove explicit default value for strboolattr attribute in tests" This reverts commit bda6e5bee04c75b1f1332b4fd1ac4e8ef6c3c247. See https://lab.llvm.org/buildbot/#/builders/109/builds/15424 for instance
2021-05-25 01:43:40 +08:00
attributes #1 = { nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
[ScopInfo] Fix: use raw source pointer. When adding an llvm.memcpy instruction to AliasSetTracker, it uses the raw source and target pointers which preserve bitcasts. MemAccInst::getPointerOperand() also returns the raw target pointers, but Scop::buildAliasGroups() did not for the source pointer. This lead to mismatches between AliasSetTracker and ScopInfo on which pointer to use. Fixed by also using raw pointers in Scop::buildAliasGroups(). llvm-svn: 285071
2016-10-25 21:37:43 +08:00
!llvm.ident = !{!0}
!0 = !{!"clang version 4.0.0 (trunk 285057) (llvm/trunk 285063)"}