llvm-project/clang/test/OpenMP/declare_variant_messages.c

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

178 lines
22 KiB
C
Raw Normal View History

// RUN: %clang_cc1 -triple=x86_64-pc-win32 -verify -fopenmp -x c -std=c99 -fms-extensions -Wno-pragma-pack %s
// RUN: %clang_cc1 -triple=x86_64-pc-win32 -verify -fopenmp-simd -x c -std=c99 -fms-extensions -Wno-pragma-pack %s
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare // expected-error {{expected an OpenMP directive}}
int foo(void);
#pragma omp declare variant // expected-error {{expected '(' after 'declare variant'}}
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant( // expected-error {{expected expression}} expected-error {{expected ')'}} expected-note {{to match this '('}}
#pragma omp declare variant(foo // expected-error {{expected ')'}} expected-error {{expected 'match' clause on 'omp declare variant' directive}} expected-note {{to match this '('}}
#pragma omp declare variant(x) // expected-error {{use of undeclared identifier 'x'}} expected-error {{expected 'match' clause on}}
#pragma omp declare variant(foo) // expected-error {{expected 'match' clause on 'omp declare variant' directive}}
#pragma omp declare variant(foo) // expected-error {{expected 'match' clause on 'omp declare variant' directive}}
#pragma omp declare variant(foo) xxx // expected-error {{expected 'match' clause on 'omp declare variant' directive}}
#pragma omp declare variant(foo) match // expected-error {{expected '(' after 'match'}}
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(foo) match( // expected-error {{expected ')'}} expected-warning {{expected identifier or string literal describing a context set; set skipped}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match() // expected-warning {{expected identifier or string literal describing a context set; set skipped}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx=) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx=yyy) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx=yyy}) // expected-error {{expected ')'}} expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(xxx={) // expected-error {{expected ')'}} expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx={vvv, vvv}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx={vvv} xxx) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(xxx={vvv}) xxx // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp declare variant(foo) match(implementation={xxx}) // expected-warning {{'xxx' is not a valid context selector for the context set 'implementation'; selector ignored}} expected-note {{context selector options are: 'vendor' 'extension' 'unified_address' 'unified_shared_memory' 'reverse_offload' 'dynamic_allocators' 'atomic_default_mem_order'}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(implementation={vendor}) // expected-warning {{the context selector 'vendor' in context set 'implementation' requires a context property defined in parentheses; selector ignored}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(implementation={vendor(}) // expected-error {{expected ')'}} expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{context property options are: 'amd' 'arm' 'bsc' 'cray' 'fujitsu' 'gnu' 'ibm' 'intel' 'llvm' 'pgi' 'ti' 'unknown'}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(implementation={vendor()}) // expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{context property options are: 'amd' 'arm' 'bsc' 'cray' 'fujitsu' 'gnu' 'ibm' 'intel' 'llvm' 'pgi' 'ti' 'unknown'}}
#pragma omp declare variant(foo) match(implementation={vendor(score ibm)}) // expected-error {{expected '(' after 'score'}} expected-warning {{expected '':'' after the score expression; '':'' assumed}}
#pragma omp declare variant(foo) match(implementation={vendor(score( ibm)}) // expected-error {{use of undeclared identifier 'ibm'}} expected-error {{expected ')'}} expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{context property options are: 'amd' 'arm' 'bsc' 'cray' 'fujitsu' 'gnu' 'ibm' 'intel' 'llvm' 'pgi' 'ti' 'unknown'}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(implementation={vendor(score(2 ibm)}) // expected-error {{expected ')'}} expected-error {{expected ')'}} expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{to match this '('}} expected-note {{context property options are: 'amd' 'arm' 'bsc' 'cray' 'fujitsu' 'gnu' 'ibm' 'intel' 'llvm' 'pgi' 'ti' 'unknown'}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(implementation={vendor(score(foo()) ibm)}) // expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{score expressions in the OpenMP context selector need to be constant; foo() is not and will be ignored}}
#pragma omp declare variant(foo) match(implementation={vendor(score(5): ibm), vendor(llvm)}) // expected-warning {{the context selector 'vendor' was used already in the same 'omp declare variant' directive; selector ignored}} expected-note {{the previous context selector 'vendor' used here}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(implementation={vendor(score(5): ibm), kind(cpu)}) // expected-warning {{the context selector 'kind' is not valid for the context set 'implementation'; selector ignored}} expected-note {{the context selector 'kind' can be nested in the context set 'device'; try 'match(device={kind(property)})'}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(device={xxx}) // expected-warning {{'xxx' is not a valid context selector for the context set 'device'; selector ignored}} expected-note {{context selector options are: 'kind' 'arch' 'isa'}} expected-note {{the ignored selector spans until here}}
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(foo) match(device={kind}) // expected-warning {{the context selector 'kind' in context set 'device' requires a context property defined in parentheses; selector ignored}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(device={kind(}) // expected-error {{expected ')'}} expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{context property options are: 'host' 'nohost' 'cpu' 'gpu' 'fpga' 'any'}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(device={kind()}) // expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{context property options are: 'host' 'nohost' 'cpu' 'gpu' 'fpga' 'any'}}
#pragma omp declare variant(foo) match(device={kind(score cpu)}) // expected-error {{expected '(' after 'score'}} expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{the context selector 'kind' in the context set 'device' cannot have a score ('<invalid>'); score ignored}}
#pragma omp declare variant(foo) match(device = {kind(score(ibm) }) // expected-error {{use of undeclared identifier 'ibm'}} expected-error {{expected ')'}} expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{the context selector 'kind' in the context set 'device' cannot have a score ('<recovery-expr>()'); score ignored}} expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{context property options are: 'host' 'nohost' 'cpu' 'gpu' 'fpga' 'any'}} expected-note {{to match this '('}}
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(foo) match(device={kind(score(2 gpu)}) // expected-error {{expected ')'}} expected-error {{expected ')'}} expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{the context selector 'kind' in the context set 'device' cannot have a score ('2'); score ignored}} expected-warning {{expected identifier or string literal describing a context property; property skipped}} expected-note {{to match this '('}} expected-note {{context property options are: 'host' 'nohost' 'cpu' 'gpu' 'fpga' 'any'}} expected-note {{to match this '('}}
#pragma omp declare variant(foo) match(device={kind(score(foo()) ibm)}) // expected-warning {{expected '':'' after the score expression; '':'' assumed}} expected-warning {{the context selector 'kind' in the context set 'device' cannot have a score ('foo()'); score ignored}} expected-warning {{'ibm' is not a valid context property for the context selector 'kind' and the context set 'device'; property ignored}} expected-note {{try 'match(implementation={vendor(ibm)})'}} expected-note {{the ignored property spans until here}}
#pragma omp declare variant(foo) match(device={kind(score(5): host), kind(llvm)}) // expected-warning {{the context selector 'kind' in the context set 'device' cannot have a score ('5'); score ignored}} expected-warning {{the context selector 'kind' was used already in the same 'omp declare variant' directive; selector ignored}} expected-note {{the previous context selector 'kind' used here}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(device={kind(score(5): nohost), vendor(llvm)}) // expected-warning {{the context selector 'kind' in the context set 'device' cannot have a score ('5'); score ignored}} expected-warning {{the context selector 'vendor' is not valid for the context set 'device'; selector ignored}} expected-note {{the context selector 'vendor' can be nested in the context set 'implementation'; try 'match(implementation={vendor(property)})'}} expected-note {{the ignored selector spans until here}}
#pragma omp declare variant(foo) match(implementation={extension("aaa")}) // expected-warning {{'aaa' is not a valid context property for the context selector 'extension' and the context set 'implementation'; property ignored}} expected-note {{context property options are: 'match_all' 'match_any' 'match_none'}} expected-note {{the ignored property spans until here}}
int bar(void);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(foo) match(implementation = {vendor(score(foo) :llvm)}) // expected-warning {{score expressions in the OpenMP context selector need to be constant; foo is not and will be ignored}}
#pragma omp declare variant(foo) match(implementation = {vendor(score(foo()) :llvm)}) // expected-warning {{score expressions in the OpenMP context selector need to be constant; foo() is not and will be ignored}}
#pragma omp declare variant(foo) match(implementation = {vendor(score(<expr>) :llvm)}) // expected-error {{expected expression}} expected-error {{use of undeclared identifier 'expr'}} expected-error {{expected expression}}
#pragma omp declare variant(foo) match(user = {condition(foo)}) // expected-error {{the user condition in the OpenMP context selector needs to be constant; foo is not}}
#pragma omp declare variant(foo) match(user = {condition(foo())}) // expected-error {{the user condition in the OpenMP context selector needs to be constant; foo() is not}}
#pragma omp declare variant(foo) match(user = {condition(<expr>)}) // expected-error {{expected expression}} expected-error {{use of undeclared identifier 'expr'}} expected-error {{expected expression}} expected-note {{the ignored selector spans until here}}
int score_and_cond_non_const();
#pragma omp declare variant(foo) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int a; // expected-error {{'#pragma omp declare variant' can only be applied to functions}}
#pragma omp declare variant(foo) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
#pragma omp threadprivate(a) // expected-error {{'#pragma omp declare variant' can only be applied to functions}}
int var;
#pragma omp threadprivate(var)
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(foo) match(xxx={}) // expected-error {{function declaration is expected after 'declare variant' directive}}
#pragma omp declare // expected-error {{expected an OpenMP directive}}
#pragma omp declare variant(foo) match(xxx={}) // expected-error {{function declaration is expected after 'declare variant' directive}}
#pragma omp declare variant(foo) match(xxx={}) // expected-error {{function declaration is expected after 'declare variant' directive}}
#pragma options align=packed
int main();
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
[OpenMP] `omp begin/end declare variant` - part 2, sema ("+CG") This is the second part loosely extracted from D71179 and cleaned up. This patch provides semantic analysis support for `omp begin/end declare variant`, mostly as defined in OpenMP technical report 8 (TR8) [0]. The sema handling makes code generation obsolete as we generate "the right" calls that can just be handled as usual. This handling also applies to the existing, albeit problematic, `omp declare variant support`. As a consequence a lot of unneeded code generation and complexity is removed. A major purpose of this patch is to provide proper `math.h`/`cmath` support for OpenMP target offloading. See PR42061, PR42798, PR42799. The current code was developed with this feature in mind, see [1]. The logic is as follows: If we have seen a `#pragma omp begin declare variant match(<SELECTOR>)` but not the corresponding `end declare variant`, and we find a function definition we will: 1) Create a function declaration for the definition we were about to generate. 2) Create a function definition but with a mangled name (according to `<SELECTOR>`). 3) Annotate the declaration with the `OMPDeclareVariantAttr`, the same one used already for `omp declare variant`, using and the mangled function definition as specialization for the context defined by `<SELECTOR>`. When a call is created we inspect it. If the target has an `OMPDeclareVariantAttr` attribute we try to specialize the call. To this end, all variants are checked, the best applicable one is picked and a new call to the specialization is created. The new call is used instead of the original one to the base function. To keep the AST printing and tooling possible we utilize the PseudoObjectExpr. The original call is the syntactic expression, the specialized call is the semantic expression. [0] https://www.openmp.org/wp-content/uploads/openmp-TR8.pdf [1] https://reviews.llvm.org/D61399#change-496lQkg0mhRN Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim, aaron.ballman Subscribers: bollu, guansong, openmp-commits, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D75779
2020-02-26 06:04:06 +08:00
#pragma omp declare variant(foo) match(implementation={vendor(llvm)}) // expected-error {{function declaration is expected after 'declare variant' directive}}
#pragma omp declare variant(foo) match(implementation={vendor(llvm)}) // expected-error {{function declaration is expected after 'declare variant' directive}}
#pragma init_seg(compiler)
int main();
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(foo) match(xxx={}) // expected-error {{single declaration is expected after 'declare variant' directive}} expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int b, c;
int no_proto();
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(no_proto) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int no_proto_too();
int proto1(int);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(proto1) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int diff_proto(); // expected-note {{previous declaration is here}}
int diff_proto(double); // expected-error {{conflicting types for 'diff_proto'}}
#pragma omp declare variant(no_proto) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int diff_proto1(double);
int after_use_variant(void);
int after_use();
int bar() {
return after_use();
}
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(after_use_variant) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-warning {{'#pragma omp declare variant' cannot be applied for function after first usage; the original function might be used}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int after_use(void);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(after_use_variant) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int defined(void) { return 0; }
int defined1(void) { return 0; }
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(after_use_variant) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-warning {{'#pragma omp declare variant' cannot be applied to the function that was defined already; the original function might be used}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
int defined1(void);
int diff_cc_variant(void);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(diff_cc_variant) match(xxx={}) // expected-error {{variant in '#pragma omp declare variant' with type 'int (void)' is incompatible with type 'int (void) __attribute__((vectorcall))'}} expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
__vectorcall int diff_cc(void);
int diff_ret_variant(void);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(diff_ret_variant) match(xxx={}) // expected-error {{variant in '#pragma omp declare variant' with type 'int (void)' is incompatible with type 'void (void)'}} expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
void diff_ret(void);
void marked(void);
void not_marked(void);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(not_marked) match(implementation={vendor(unknown)}, device={kind(cpu)}) // expected-note {{marked as 'declare variant' here}}
void marked_variant(void);
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant(marked_variant) match(xxx={}) // expected-warning {{'xxx' is not a valid context set in a `declare variant`; set ignored}} expected-warning {{variant function in '#pragma omp declare variant' is itself marked as '#pragma omp declare variant'}} expected-note {{context set options are: 'construct' 'device' 'implementation' 'user'}} expected-note {{the ignored set spans until here}}
void marked(void);
#pragma omp declare variant(foo) match(device = {isa("foo")})
int unknown_isa_trait(void);
#pragma omp declare variant(foo) match(device = {isa(foo)})
int unknown_isa_trait2(void);
#pragma omp declare variant(foo) match(device = {kind(fpga), isa(bar)})
int ignored_isa_trait(void);
void caller() {
unknown_isa_trait(); // expected-warning {{isa trait 'foo' is not known to the current target; verify the spelling or consider restricting the context selector with the 'arch' selector further}}
unknown_isa_trait2(); // expected-warning {{isa trait 'foo' is not known to the current target; verify the spelling or consider restricting the context selector with the 'arch' selector further}}
ignored_isa_trait();
}
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
// Unknown arch
#pragma omp begin declare variant match(device={isa(sse2020)}) // expected-warning {{isa trait 'sse2020' is not known to the current target; verify the spelling or consider restricting the context selector with the 'arch' selector further}}
#pragma omp end declare variant
// Unknown arch guarded by arch.
#pragma omp begin declare variant match(device={isa(sse2020), arch(ppc)})
#pragma omp end declare variant
[OpenMP][Part 2] Use reusable OpenMP context/traits handling This patch implements an almost complete handling of OpenMP contexts/traits such that we can reuse most of the logic in Flang through the OMPContext.{h,cpp} in llvm/Frontend/OpenMP. All but construct SIMD specifiers, e.g., inbranch, and the device ISA selector are define in `llvm/lib/Frontend/OpenMP/OMPKinds.def`. From these definitions we generate the enum classes `TraitSet`, `TraitSelector`, and `TraitProperty` as well as conversion and helper functions in `llvm/lib/Frontend/OpenMP/OMPContext.{h,cpp}`. The above enum classes are used in the parser, sema, and the AST attribute. The latter is not a collection of multiple primitive variant arguments that contain encodings via numbers and strings but instead a tree that mirrors the `match` clause (see `struct OpenMPTraitInfo`). The changes to the parser make it more forgiving when wrong syntax is read and they also resulted in more specialized diagnostics. The tests are updated and the core issues are detected as before. Here and elsewhere this patch tries to be generic, thus we do not distinguish what selector set, selector, or property is parsed except if they do behave exceptionally, as for example `user={condition(EXPR)}` does. The sema logic changed in two ways: First, the OMPDeclareVariantAttr representation changed, as mentioned above, and the sema was adjusted to work with the new `OpenMPTraitInfo`. Second, the matching and scoring logic moved into `OMPContext.{h,cpp}`. It is implemented on a flat representation of the `match` clause that is not tied to clang. `OpenMPTraitInfo` provides a method to generate this flat structure (see `struct VariantMatchInfo`) by computing integer score values and boolean user conditions from the `clang::Expr` we keep for them. The OpenMP context is now an explicit object (see `struct OMPContext`). This is in anticipation of construct traits that need to be tracked. The OpenMP context, as well as the `VariantMatchInfo`, are basically made up of a set of active or respectively required traits, e.g., 'host', and an ordered container of constructs which allows duplication. Matching and scoring is kept as generic as possible to allow easy extension in the future. --- Test changes: The messages checked in `OpenMP/declare_variant_messages.{c,cpp}` have been auto generated to match the new warnings and notes of the parser. The "subset" checks were reversed causing the wrong version to be picked. The tests have been adjusted to correct this. We do not print scores if the user did not provide one. We print spaces to make lists in the `match` clause more legible. Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim Subscribers: merge_guards_bot, rampitec, mgorny, hiraditya, aheejin, fedor.sergeev, simoncook, bollu, guansong, dexonsmith, jfb, s.egerton, llvm-commits, cfe-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D71830
2019-12-20 10:42:12 +08:00
#pragma omp declare variant // expected-error {{function declaration is expected after 'declare variant' directive}}
#pragma omp declare variant // expected-error {{function declaration is expected after 'declare variant' directive}}
// FIXME: If the scores are equivalent we should detect that and allow it.
#pragma omp begin declare variant match(implementation = {vendor(score(2) \
: llvm)})
#pragma omp declare variant(foo) match(implementation = {vendor(score(2) \
: llvm)}) // expected-error@-1 {{nested OpenMP context selector contains duplicated trait 'llvm' in selector 'vendor' and set 'implementation' with different score}}
int conflicting_nested_score(void);
#pragma omp end declare variant
// FIXME: We should build the conjuction of different conditions, see also the score fixme above.
#pragma omp begin declare variant match(user = {condition(1)})
#pragma omp declare variant(foo) match(user = {condition(1)}) // expected-error {{nested user conditions in OpenMP context selector not supported (yet)}}
int conflicting_nested_condition(void);
#pragma omp end declare variant