maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/polly/test/ScopDetectionDiagnostics/ReportVariantBasePtr-01.ll

103 lines
5.2 KiB
LLVM
Raw Normal View History

tests: Drop -polly-detect-unprofitable and -polly-no-early-exit These flags are now always passed to all tests and need to be disabled if not needed. Disabling these flags, rather than passing them to almost all tests, significantly simplfies our RUN: lines. llvm-svn: 249422
2015-10-06 23:36:44 +08:00
; RUN: opt %loadPolly -pass-remarks-missed="polly-detect" -polly-detect-track-failures -polly-detect -analyze < %s 2>&1| FileCheck %s
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
; struct b {
; double **b;
; };
;
; void a(struct b *A) {
; for (int i=0; i<32; i++)
Allow invariant loads in the SCoP description This patch allows invariant loads to be used in the SCoP description, e.g., as loop bounds, conditions or in memory access functions. First we collect "required invariant loads" during SCoP detection that would otherwise make an expression we care about non-affine. To this end a new level of abstraction was introduced before SCEVValidator::isAffineExpr() namely ScopDetection::isAffine() and ScopDetection::onlyValidRequiredInvariantLoads(). Here we can decide if we want a load inside the region to be optimistically assumed invariant or not. If we do, it will be marked as required and in the SCoP generation we bail if it is actually not invariant. If we don't it will be a non-affine expression as before. At the moment we optimistically assume all "hoistable" (namely non-loop-carried) loads to be invariant. This causes us to expand some SCoPs and dismiss them later but it also allows us to detect a lot we would dismiss directly if we would ask e.g., AliasAnalysis::canBasicBlockModify(). We also allow potential aliases between optimistically assumed invariant loads and other pointers as our runtime alias checks are sound in case the loads are actually invariant. Together with the invariant checks this combination allows to handle a lot more than LICM can. The code generation of the invariant loads had to be extended as we can now have dependences between parameters and invariant (hoisted) loads as well as the other way around, e.g., test/Isl/CodeGen/invariant_load_parameters_cyclic_dependence.ll First, it is important to note that we cannot have real cycles but only dependences from a hoisted load to a parameter and from another parameter to that hoisted load (and so on). To handle such cases we materialize llvm::Values for parameters that are referred by a hoisted load on demand and then materialize the remaining parameters. Second, there are new kinds of dependences between hoisted loads caused by the constraints on their execution. If a hoisted load is conditionally executed it might depend on the value of another hoisted load. To deal with such situations we sort them already in the ScopInfo such that they can be generated in the order they are listed in the Scop::InvariantAccesses list (see compareInvariantAccesses). The dependences between hoisted loads caused by indirect accesses are handled the same way as before. llvm-svn: 249607
2015-10-08 04:17:36 +08:00
; A[i].b[i] = 0;
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
; }
[ScopDetection] Require LoadInst base pointers to be hoisted. Only when load-hoisted we can be sure the base pointer is invariant during the SCoP's execution. Most of the time it would be added to the required hoists for the alias checks anyway, except with -polly-ignore-aliasing, -polly-use-runtime-alias-checks=0 or if AliasAnalysis is already sure it doesn't alias with anything (for instance if there is no other pointer to alias with). Two more parts in Polly assume that this load-hoisting took place: - setNewAccessRelation() which contains an assert which tests this. - BlockGenerator which would use to the base ptr from the original code if not load-hoisted (if the access expression is regenerated) Differential Revision: https://reviews.llvm.org/D30694 llvm-svn: 297195
2017-03-08 04:28:43 +08:00
; The loads are currently just adds %7 to the list of required invariant loads
; and only -polly-scops checks whether it is actionally possible the be load
; hoisted. The SCoP is still rejected by -polly-detect because it may alias
; with %A and is not considered to be eligble for runtime alias checking.
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
; CHECK: remark: ReportVariantBasePtr01.c:6:8: The following errors keep this region from being a Scop.
[ScopDetection] Require LoadInst base pointers to be hoisted. Only when load-hoisted we can be sure the base pointer is invariant during the SCoP's execution. Most of the time it would be added to the required hoists for the alias checks anyway, except with -polly-ignore-aliasing, -polly-use-runtime-alias-checks=0 or if AliasAnalysis is already sure it doesn't alias with anything (for instance if there is no other pointer to alias with). Two more parts in Polly assume that this load-hoisting took place: - setNewAccessRelation() which contains an assert which tests this. - BlockGenerator which would use to the base ptr from the original code if not load-hoisted (if the access expression is regenerated) Differential Revision: https://reviews.llvm.org/D30694 llvm-svn: 297195
2017-03-08 04:28:43 +08:00
; CHECK: remark: ReportVariantBasePtr01.c:7:5: Accesses to the arrays "A", " <unknown> " may access the same memory.
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
%struct.b = type { double** }
Adjust debug metadata to LLVM changes in 252219 llvm-svn: 252273
2015-11-06 14:27:39 +08:00
define void @a(%struct.b* nocapture readonly %A) #0 !dbg !4 {
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
entry:
br label %entry.split
entry.split: ; preds = %entry
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
tail call void @llvm.dbg.value(metadata %struct.b* %A, i64 0, metadata !16, metadata !DIExpression()), !dbg !23
tail call void @llvm.dbg.value(metadata i32 0, i64 0, metadata !17, metadata !DIExpression()), !dbg !25
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
br label %for.body, !dbg !27
for.body: ; preds = %for.body, %entry.split
%indvar4 = phi i64 [ %indvar.next, %for.body ], [ 0, %entry.split ]
Allow invariant loads in the SCoP description This patch allows invariant loads to be used in the SCoP description, e.g., as loop bounds, conditions or in memory access functions. First we collect "required invariant loads" during SCoP detection that would otherwise make an expression we care about non-affine. To this end a new level of abstraction was introduced before SCEVValidator::isAffineExpr() namely ScopDetection::isAffine() and ScopDetection::onlyValidRequiredInvariantLoads(). Here we can decide if we want a load inside the region to be optimistically assumed invariant or not. If we do, it will be marked as required and in the SCoP generation we bail if it is actually not invariant. If we don't it will be a non-affine expression as before. At the moment we optimistically assume all "hoistable" (namely non-loop-carried) loads to be invariant. This causes us to expand some SCoPs and dismiss them later but it also allows us to detect a lot we would dismiss directly if we would ask e.g., AliasAnalysis::canBasicBlockModify(). We also allow potential aliases between optimistically assumed invariant loads and other pointers as our runtime alias checks are sound in case the loads are actually invariant. Together with the invariant checks this combination allows to handle a lot more than LICM can. The code generation of the invariant loads had to be extended as we can now have dependences between parameters and invariant (hoisted) loads as well as the other way around, e.g., test/Isl/CodeGen/invariant_load_parameters_cyclic_dependence.ll First, it is important to note that we cannot have real cycles but only dependences from a hoisted load to a parameter and from another parameter to that hoisted load (and so on). To handle such cases we materialize llvm::Values for parameters that are referred by a hoisted load on demand and then materialize the remaining parameters. Second, there are new kinds of dependences between hoisted loads caused by the constraints on their execution. If a hoisted load is conditionally executed it might depend on the value of another hoisted load. To deal with such situations we sort them already in the ScopInfo such that they can be generated in the order they are listed in the Scop::InvariantAccesses list (see compareInvariantAccesses). The dependences between hoisted loads caused by indirect accesses are handled the same way as before. llvm-svn: 249607
2015-10-08 04:17:36 +08:00
%b = getelementptr inbounds %struct.b, %struct.b* %A, i64 %indvar4, i32 0, !dbg !26
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
%0 = mul i64 %indvar4, 4, !dbg !26
%1 = add i64 %0, 3, !dbg !26
%2 = add i64 %0, 2, !dbg !26
%3 = add i64 %0, 1, !dbg !26
Update Polly tests to handle explicitly typed load changes in LLVM. llvm-svn: 230796
2015-02-28 05:22:50 +08:00
%4 = load double**, double*** %b, align 8, !dbg !26, !tbaa !28
Update Polly tests to handle explicitly typed gep changes in LLVM llvm-svn: 230784
2015-02-28 03:20:19 +08:00
%arrayidx = getelementptr double*, double** %4, i64 %0, !dbg !26
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
store double* null, double** %arrayidx, align 8, !dbg !26, !tbaa !33
Update Polly tests to handle explicitly typed load changes in LLVM. llvm-svn: 230796
2015-02-28 05:22:50 +08:00
%5 = load double**, double*** %b, align 8, !dbg !26, !tbaa !28
Update Polly tests to handle explicitly typed gep changes in LLVM llvm-svn: 230784
2015-02-28 03:20:19 +08:00
%arrayidx.1 = getelementptr double*, double** %5, i64 %3, !dbg !26
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
store double* null, double** %arrayidx.1, align 8, !dbg !26, !tbaa !33
Update Polly tests to handle explicitly typed load changes in LLVM. llvm-svn: 230796
2015-02-28 05:22:50 +08:00
%6 = load double**, double*** %b, align 8, !dbg !26, !tbaa !28
Update Polly tests to handle explicitly typed gep changes in LLVM llvm-svn: 230784
2015-02-28 03:20:19 +08:00
%arrayidx.2 = getelementptr double*, double** %6, i64 %2, !dbg !26
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
store double* null, double** %arrayidx.2, align 8, !dbg !26, !tbaa !33
Update Polly tests to handle explicitly typed load changes in LLVM. llvm-svn: 230796
2015-02-28 05:22:50 +08:00
%7 = load double**, double*** %b, align 8, !dbg !26, !tbaa !28
Update Polly tests to handle explicitly typed gep changes in LLVM llvm-svn: 230784
2015-02-28 03:20:19 +08:00
%arrayidx.3 = getelementptr double*, double** %7, i64 %1, !dbg !26
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
store double* null, double** %arrayidx.3, align 8, !dbg !26, !tbaa !33
%indvar.next = add i64 %indvar4, 1, !dbg !27
%exitcond = icmp eq i64 %indvar.next, 8, !dbg !27
br i1 %exitcond, label %for.end, label %for.body, !dbg !27
for.end: ; preds = %for.body
ret void, !dbg !34
}
Upgrade testcases after LLVM r234181 Until r234181 we were silently upgrading old `@llvm.dbg` intrinsics. Fix testcases in polly that were relying on that. llvm-svn: 234192
2015-04-07 02:25:51 +08:00
declare void @llvm.dbg.value(metadata, i64, metadata, metadata) #1
Add diagnostic remark for ReportVariantBasePtr llvm-svn: 211777
2014-06-26 21:33:35 +08:00
attributes #0 = { nounwind uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #1 = { nounwind readnone }
!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!20, !21}
!llvm.ident = !{!22}
Update debug metadata after LLVM commits r266445+r266446 llvm-svn: 266473
2016-04-16 04:51:27 +08:00
!0 = distinct !DICompileUnit(language: DW_LANG_C99, producer: "clang version 3.5.0 ", isOptimized: true, emissionKind: 1, file: !1, enums: !2, retainedTypes: !2, globals: !2, imports: !2)
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!1 = !DIFile(filename: "ReportVariantBasePtr01.c", directory: "test/ScopDetectionDiagnostics")
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!2 = !{}
Update debug metadata after LLVM commits r266445+r266446 llvm-svn: 266473
2016-04-16 04:51:27 +08:00
!4 = distinct !DISubprogram(name: "a", line: 5, isLocal: false, isDefinition: true, virtualIndex: 6, flags: DIFlagPrototyped, isOptimized: true, unit: !0, scopeLine: 5, file: !1, scope: !5, type: !6, variables: !15)
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!5 = !DIFile(filename: "ReportVariantBasePtr01.c", directory: "test/ScopDetectionDiagnostics")
!6 = !DISubroutineType(types: !7)
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!7 = !{null, !8}
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!8 = !DIDerivedType(tag: DW_TAG_pointer_type, size: 64, align: 64, baseType: !9)
!9 = !DICompositeType(tag: DW_TAG_structure_type, name: "b", line: 1, size: 64, align: 64, file: !1, elements: !10)
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!10 = !{!11}
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!11 = !DIDerivedType(tag: DW_TAG_member, name: "b", line: 2, size: 64, align: 64, file: !1, scope: !9, baseType: !12)
!12 = !DIDerivedType(tag: DW_TAG_pointer_type, size: 64, align: 64, baseType: !13)
!13 = !DIDerivedType(tag: DW_TAG_pointer_type, size: 64, align: 64, baseType: !14)
!14 = !DIBasicType(tag: DW_TAG_base_type, name: "double", size: 64, align: 64, encoding: DW_ATE_float)
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!15 = !{!16, !17}
Fix polly tests after LLVM IR change in r243774 llvm-svn: 243801
2015-08-01 07:58:50 +08:00
!16 = !DILocalVariable(name: "A", line: 5, arg: 1, scope: !4, file: !5, type: !8)
!17 = !DILocalVariable(name: "i", line: 6, scope: !18, file: !5, type: !19)
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!18 = distinct !DILexicalBlock(line: 6, column: 3, file: !1, scope: !4)
!19 = !DIBasicType(tag: DW_TAG_base_type, name: "int", size: 32, align: 32, encoding: DW_ATE_signed)
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!20 = !{i32 2, !"Dwarf Version", i32 4}
Update Polly tests for the great metadata schema change llvm-svn: 231089
2015-03-04 02:17:26 +08:00
!21 = !{i32 2, !"Debug Info Version", i32 3}
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!22 = !{!"clang version 3.5.0 "}
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!23 = !DILocation(line: 5, column: 18, scope: !4)
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!24 = !{i32 0}
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!25 = !DILocation(line: 6, column: 12, scope: !18)
!26 = !DILocation(line: 7, column: 5, scope: !18)
!27 = !DILocation(line: 6, column: 8, scope: !18)
Run upgrade script from PR21532 to match LLVM changes Update tests for LLVM assembly format change in r224257 using the script attached to PR21532. I'm hoping this unsticks the bot [1]. [1]: http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/25432 llvm-svn: 224269
2014-12-16 04:28:50 +08:00
!28 = !{!29, !30, i64 0}
!29 = !{!"b", !30, i64 0}
!30 = !{!"any pointer", !31, i64 0}
!31 = !{!"omnipotent char", !32, i64 0}
!32 = !{!"Simple C/C++ TBAA"}
!33 = !{!30, !30, i64 0}
Update polly for LLVM rename of debug info metadata with DI* prefix Ran the same rename-md-di-prefix.sh script attached to PR23080 as in LLVM r236120 and CFE r236121. llvm-svn: 236127
2015-04-30 01:02:14 +08:00
!34 = !DILocation(line: 8, column: 1, scope: !4)