[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
gulfemsavrun wrote:
We started seeing the following issue, and bisected to this commit.
```
test.cpp:8:12: error: no matching function for call to 'GetFieldChecked'
8 | return GetFieldChecked(b,
&std::remove_reference::type::has_total);
|^~~
test.cpp:6:5: note: candidate template ignored: deduced conflicting types for
parameter 'Table' ('Bar' vs. 'const Bar')
6 | int GetFieldChecked(const Table&, bool (Table::*)() const);
| ^
1 error generate
```
Here's a small test case:
```
#include
struct Bar {
bool has_total() const;
};
template
int GetFieldChecked(const Table&, bool (Table::*)() const);
int foo(const Bar& b) {
return GetFieldChecked(b,
&std::remove_reference::type::has_total);
}
```
If we just compile it via `clang++ -c test.cpp` with this commit, we ran into
this issue and it is blocking us to roll Clang in our project.
https://github.com/llvm/llvm-project/pull/132401
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: @emaxx-google thanks for the reproducer. I will be off to C++Now soon, so it's unlikely I will have time to take a look at that in the next two weeks, sorry about that. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
emaxx-google wrote: Hello, the minimized reproducer for the determinism issue is there: https://pastebin.com/6aL6rmBe . To build it, unpack it into separate files (via `split-file`), then run `CLANG=path/to/clang make`. The nondeterminism can be checked by looping `make clean` + `make` + `md5sum problem.pcm` - typically a different hash pops up after a few dozens of iterations. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: No, the problem seems to be in serialization/deserialization, there must be some mismatch between `ASTStmtWriter::VisitSubstNonTypeTemplateParmExpr` and `ASTStmtReader::VisitSubstNonTypeTemplateParmExpr`. More precisely, I'm seeing that `ASTStmtWriter::VisitSubstNonTypeTemplateParmExpr` outputs a record of 5 elements, while `ASTStmtReader::VisitSubstNonTypeTemplateParmExpr` gets a record of 4 elements as input, and thus triggers assertion when reading the source location. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Could it be you are hitting an overflow/wrap around perhaps? Some of these nodes store the unsignedOrNone representation in a bitfield, but that's still 15 bits. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: Comparing bcanalyzer --dump outputs for non-deterministic pcms, I see a lot of op values that differ by 1. I wonder if this might be something like the mismatch of Read Write UnsignedOrNone vs unsigned ... https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: In one of these changes we did bump one of these bit fields down to 15 bits, starting from 16. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: Didn't fire so far. Though this is non-deterministic, I might be (un)lucky. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: One thing that looks fishy, but that is even a different node, is this early return on `VisitSubstNonTypeTemplateParmPackExpr` in `ASTReaderStmt`. ```C++ if (ArgPack.getKind() != TemplateArgument::Pack) return; ``` This looks impossible to hit, because `getArgumentPack` calls a constructor which can only return Packs. Can you turn that into an assert and try again? ```C++ assert (ArgPack.getKind() == TemplateArgument::Pack); ``` https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: Using library with assertions, I'm seeing out-of-bounds source location read called from `ASTStmtReader::VisitSubstNonTypeTemplateParmExpr` - [trace](https://gist.github.com/eaeltsin/845fb9cc6f65f47ed03a64aca5aff923) https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: Patched #135450 but still see the same MSan finding on my compilation - uninitialized value while checking location validity in `TranslateSourceLocation` called from `VisitSubstNonTypeTemplateParmExpr`. Sorry but I'm not yet sure how to strip the reproducer from internal code so cannot provide it :| https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Thanks for that stack trace, could be unrelated to this, but that still helped find a bug: https://github.com/llvm/llvm-project/pull/135450 https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: Well, to be honest, I'm not completely sure these are the same problems, these are just sanitizer finding for the same compilation. Here is the [MSan finding](https://gist.github.com/eaeltsin/834192fe5d8bbbc061e1488f113e34e1). Please note my source is somewhat behind the head, so locations might be off a bit https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Do you have a backtrace of that uninitialized read? https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Okay, if the problem is an uninitialized source location somewhere, then that patch doesn't help at all. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: @mizvekov - no, the patch doesn't help, or I did something wrong. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: Memory Sanitizer complains about initialized value [here](https://github.com/llvm/llvm-project/blob/main/clang/include/clang/Serialization/ASTReader.h#L2438) Will try the patch now. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: @eaeltsin speculative fix here, but can you try with this patch? https://github.com/llvm/llvm-project/pull/135434 https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: > We are now seeing non-determinism in `.pcm` files that root-causes to this > commit. > > @mizvekov - might it be something obvious, like pointer-keyed containers or > similar? I don't see anything obvious. The only parts of the patch which touch anything similar to pointer-keyed containers are the type properties cache changes, like the linkage computation. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
eaeltsin wrote: We are now seeing non-determinism in `.pcm` files that root-causes to this commit. @mizvekov - might it be something obvious, like pointer-keyed containers or similar? https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
aeubanks wrote: @mizvekov we're seeing pcm non-determinism after this reland, does anything jump out at you? https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
earnol wrote: Could you please look into https://github.com/llvm/llvm-project/issues/133144? It looks like to be caused by your change. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
rupprecht wrote: Still seeing a crash even with the latest fix applied ``` assertion failed at clang/include/clang/AST/Type.h:945 in const ExtQualsTypeCommonBase *clang::QualType::getCommonPtr() const: !isNull() && "Cannot retrieve a NULL type pointer" *** Check failure stack trace: *** @ 0x55e0dffed1f5 DeduceTemplateArgumentsByTypeMatch() @ 0x55e0dffecc3e DeduceTemplateArgumentsByTypeMatch() @ 0x55e0dfff79f1 DeduceTemplateArguments() @ 0x55e0dffe3ce9 DeduceTemplateArguments() @ 0x55e0dffe47c0 clang::Sema::DeduceTemplateArguments() @ 0x55e0e008943b clang::Sema::InstantiateClassTemplateSpecialization() @ 0x55e0e018b8a3 llvm::function_ref<>::callback_fn<>() @ 0x55e0e0b04caf clang::StackExhaustionHandler::runWithSufficientStackSpace() @ 0x55e0e01741a8 clang::Sema::RequireCompleteTypeImpl() @ 0x55e0e01739b5 clang::Sema::RequireCompleteType() @ 0x55e0df8e6914 clang::Sema::RequireCompleteDeclContext() @ 0x55e0dfd44fa9 clang::Sema::LookupParsedName() @ 0x55e0dfb3b113 clang::Sema::BuildQualifiedDeclarationNameExpr() @ 0x55e0e00d83e8 clang::TreeTransform<>::TransformDependentScopeDeclRefExpr() @ 0x55e0e00d2529 clang::TreeTransform<>::TransformCallExpr() @ 0x55e0e00ef218 clang::TreeTransform<>::TransformReturnStmt() @ 0x55e0e00d9018 clang::TreeTransform<>::TransformCompoundStmt() @ 0x55e0e008aa0a clang::Sema::SubstStmt() @ 0x55e0e011f3ce clang::Sema::InstantiateFunctionDefinition() @ 0x55e0e0b04caf clang::StackExhaustionHandler::runWithSufficientStackSpace() @ 0x55e0dffed9d6 clang::Sema::DeduceReturnType() @ 0x55e0dfb2bac5 clang::Sema::DiagnoseUseOfDecl() @ 0x55e0dfeb8361 FinishOverloadedCallExpr() @ 0x55e0dfeb8274 clang::Sema::BuildOverloadedCallExpr() @ 0x55e0dfb322ba clang::Sema::BuildCallExpr() @ 0x55e0dfb492b8 clang::Sema::ActOnCallExpr() @ 0x55e0e00d2704 clang::TreeTransform<>::TransformCallExpr() @ 0x55e0e008d052 clang::TreeTransform<>::TransformExprs() @ 0x55e0e00d258c clang::TreeTransform<>::TransformCallExpr() @ 0x55e0e008d052 clang::TreeTransform<>::TransformExprs() @ 0x55e0e00d258c clang::TreeTransform<>::TransformCallExpr() @ 0x55e0e00ef218 clang::TreeTransform<>::TransformReturnStmt() @ 0x55e0e00d9018 clang::TreeTransform<>::TransformCompoundStmt() @ 0x55e0e008aa0a clang::Sema::SubstStmt() @ 0x55e0e011f3ce clang::Sema::InstantiateFunctionDefinition() @ 0x55e0e01227ba clang::Sema::PerformPendingInstantiations() @ 0x55e0df88aaac clang::Sema::ActOnEndOfTranslationUnitFragment() @ 0x55e0df88b22b clang::Sema::ActOnEndOfTranslationUnit() @ 0x55e0df5a1e5a clang::Parser::ParseTopLevelDecl() @ 0x55e0df59e05e clang::ParseAST() ... ``` In the meantime, I'll try reducing this https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: @nico thanks, that's landed. Let me know if it's all green now, otherwise we proceed with the revert. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: > > @alexfh should be fixed by #133613 > > > > Thank you! The crash is resolved. Do you still need a reduced test case? No worries, the tests included in PR already reproduce it. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
alexfh wrote: > @alexfh should be fixed by #133613 Thank you! The crash is resolved. Do you still need a reduced test case? https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: @alexfh should be fixed by https://github.com/llvm/llvm-project/pull/133613 https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
alexfh wrote: I've found a new crash that's not fixed by https://github.com/llvm/llvm-project/pull/133343. Reducing... https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
llvmbot wrote:
@llvm/pr-subscribers-clang-tools-extra
@llvm/pr-subscribers-lldb
Author: Matheus Izvekov (mizvekov)
Changes
Original PR: #130537
Originally reverted due to revert of dependent commit. Relanding with no
changes.
This changes the MemberPointerType representation to use a NestedNameSpecifier
instead of a Type to represent the base class.
Since the qualifiers are always parsed as nested names, there was an impedance
mismatch when converting these back and forth into types, and this led to
issues in preserving sugar.
The nested names are indeed a better match for these, as the differences which
a QualType can represent cannot be expressed syntatically, and they represent
the use case more exactly, being either dependent or referring to a CXXRecord,
unqualified.
This patch also makes the MemberPointerType able to represent sugar for a
{up/downcast}cast conversion of the base class, although for now the underlying
type is canonical, as preserving the sugar up to that point requires further
work.
As usual, includes a few drive-by fixes in order to make use of the
improvements.
---
Patch is 143.32 KiB, truncated to 20.00 KiB below, full version:
https://github.com/llvm/llvm-project/pull/132401.diff
71 Files Affected:
- (modified) clang-tools-extra/clang-tidy/modernize/UseNullptrCheck.cpp (+1-2)
- (modified) clang-tools-extra/clang-tidy/utils/ExceptionAnalyzer.cpp (+3-1)
- (modified) clang-tools-extra/clangd/unittests/FindTargetTests.cpp (+1-1)
- (modified) clang/docs/ReleaseNotes.rst (+1)
- (modified) clang/include/clang/AST/ASTContext.h (+3-4)
- (modified) clang/include/clang/AST/ASTNodeTraverser.h (+5-2)
- (modified) clang/include/clang/AST/CanonicalType.h (+1-1)
- (modified) clang/include/clang/AST/RecursiveASTVisitor.h (+5-4)
- (modified) clang/include/clang/AST/Type.h (+15-13)
- (modified) clang/include/clang/AST/TypeLoc.h (+19-14)
- (modified) clang/include/clang/AST/TypeProperties.td (+6-3)
- (modified) clang/include/clang/Basic/DiagnosticSemaKinds.td (+2-4)
- (modified) clang/include/clang/Sema/Sema.h (+9-2)
- (modified) clang/lib/AST/ASTContext.cpp (+47-21)
- (modified) clang/lib/AST/ASTImporter.cpp (+9-5)
- (modified) clang/lib/AST/ASTStructuralEquivalence.cpp (+6-2)
- (modified) clang/lib/AST/ItaniumMangle.cpp (+10-1)
- (modified) clang/lib/AST/NestedNameSpecifier.cpp (+1)
- (modified) clang/lib/AST/ODRHash.cpp (+1-1)
- (modified) clang/lib/AST/QualTypeNames.cpp (+4-3)
- (modified) clang/lib/AST/Type.cpp (+30-4)
- (modified) clang/lib/AST/TypePrinter.cpp (+2-2)
- (modified) clang/lib/CodeGen/CGCXXABI.cpp (+1-1)
- (modified) clang/lib/CodeGen/CGPointerAuth.cpp (+2-2)
- (modified) clang/lib/CodeGen/CGVTables.cpp (+2-3)
- (modified) clang/lib/CodeGen/CodeGenModule.cpp (+1-1)
- (modified) clang/lib/CodeGen/ItaniumCXXABI.cpp (+5-5)
- (modified) clang/lib/CodeGen/MicrosoftCXXABI.cpp (+4-3)
- (modified) clang/lib/Sema/SemaAccess.cpp (+18-8)
- (modified) clang/lib/Sema/SemaCast.cpp (+2-2)
- (modified) clang/lib/Sema/SemaExpr.cpp (+4-6)
- (modified) clang/lib/Sema/SemaExprCXX.cpp (+3-1)
- (modified) clang/lib/Sema/SemaOpenMP.cpp (+2-3)
- (modified) clang/lib/Sema/SemaOverload.cpp (+60-27)
- (modified) clang/lib/Sema/SemaTemplate.cpp (+6-2)
- (modified) clang/lib/Sema/SemaTemplateDeduction.cpp (+16-5)
- (modified) clang/lib/Sema/SemaType.cpp (+22-100)
- (modified) clang/lib/Sema/TreeTransform.h (+29-30)
- (modified) clang/lib/Serialization/ASTReader.cpp (+1-1)
- (modified) clang/lib/Serialization/ASTWriter.cpp (+1-1)
- (modified) clang/lib/Serialization/TemplateArgumentHasher.cpp (+3-1)
- (modified) clang/test/AST/ast-dump-template-json-win32-mangler-crash.cpp
(+3-1)
- (modified) clang/test/AST/ast-dump-templates.cpp (+238)
- (modified) clang/test/AST/ast-dump-types-json.cpp (+338-44)
- (modified) clang/test/AST/attr-print-emit.cpp (+1-1)
- (modified) clang/test/Analysis/cxx-uninitialized-object-ptr-ref.cpp (+5-5)
- (modified) clang/test/CXX/class.access/p6.cpp (+2-2)
- (modified) clang/test/CXX/drs/cwg0xx.cpp (+6-6)
- (modified) clang/test/CXX/drs/cwg13xx.cpp (+2-2)
- (modified) clang/test/CXX/drs/cwg26xx.cpp (+3-3)
- (modified) clang/test/CXX/drs/cwg2xx.cpp (+2-2)
- (modified) clang/test/CXX/drs/cwg4xx.cpp (+1-1)
- (modified) clang/test/CXX/drs/cwg7xx.cpp (+1-2)
- (modified) clang/test/CXX/temp/temp.arg/temp.arg.nontype/p1.cpp (+1-1)
- (modified) clang/test/CXX/temp/temp.arg/temp.arg.nontype/p5.cpp (+3-3)
- (modified) clang/test/Index/print-type.cpp (+1-1)
- (modified) clang/test/SemaCXX/addr-of-overloaded-function.cpp (+13-13)
- (modified) clang/test/SemaCXX/builtin-ptrtomember-ambig.cpp (+2-2)
- (modified) clang/test/SemaCXX/calling-conv-compat.cpp (+21-21)
- (modified) clang/test/SemaCXX/err_init_conversion_failed.cpp (+1-1)
- (modified) clang/test/SemaCXX/member-pointer.cpp (+13-4)
- (modified) clang/test/SemaOpenACC/combined-construct-if-ast.cpp (+2-2)
- (modified) clang/test/SemaOpenACC/combined-con
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
nico wrote: > Thanks, this will be fixed here: #133113 That also fixes an `Assertion `!isNull() && "Cannot retrieve a NULL type pointer"' failed.` we started seeing after this here landed (repro: https://issues.chromium.org/issues/406497227#comment3) It'd be good if we could either land that soon or revert this here until the follow-up is ready, to keep HEAD green. (#133113 looks kind of involved, so maybe reverting first is safer? You'll need: ``` % git revert 960615954e4cb3150ae4a479fa7f9d0d17035eea % git revert 5999be0f4770e9dd0f88ee9051a37119c8f5a1e4 % git revert 45270853192db53022ccadf4767999af4fe4e332 % git revert 14f7bd63b95d0f61a6f47119ac66398ca230559a ``` to also revert the existing 3 follow-ups, else you'll get conflicts.) https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Thanks, this will be fixed here: https://github.com/llvm/llvm-project/pull/133113 https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Thanks, that would be appreciated. https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mstorsjo wrote: > Thanks, for the report, will be fixed by #132551 Thanks for the quick fix, I can confirm that Qt builds fine for me again! https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mizvekov wrote: Thanks, for the report, will be fixed by https://github.com/llvm/llvm-project/pull/132551 https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
mstorsjo wrote:
This change broke building Qt (tested with 6.8), ending up with errors like
this:
```
qtbase/src/corelib/kernel/qcoreapplication.cpp:2946:78: error: 'this' cannot be
used in a static member function declaration
2946 | : slotObject(std::move(slotObject)), context(context ?
context : this)
|
^
qtbase/src/corelib/kernel/qobjectdefs_impl.h:405:23: note: in instantiation of
template class 'QtPrivate::CallableHelper' requested here
405 | struct Callable : CallableHelper::Type
| ^
qtbase/src/corelib/kernel/qobjectdefs.h:437:38: note: in instantiation of
template class 'QtPrivate::Callable' requested here
437 | typename QtPrivate::Callable::ReturnType *ret)
| ^
qtbase/src/corelib/kernel/qcoreapplication.cpp:2980:13: note: while
substituting deduced template arguments into function template 'invokeMethod'
[with Func = void (PermissionReceiver::*)(const QPermission &)]
2980 | QMetaObject::invokeMethod(receiver,
| ^
```
This issue also persists on the latest git main as of right now.
I've reduced the issue down to the following small reproducer:
```c++
template struct CallableHelper {
static auto Resolve() -> Func;
};
struct QIODevice {
void d_func() { d_ptr; }
int d_ptr;
};
struct Callable : CallableHelper {};
```
Reproducible with e.g. `clang -target x86_64-linux-gnu -c repro.cpp`
(presumably the same way for any target).
https://github.com/llvm/llvm-project/pull/132401
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
https://github.com/mizvekov closed https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-tools-extra] [lldb] Reland: [clang] preserve class type sugar when taking pointer to member (PR #132401)
https://github.com/Michael137 commented: LLDB changes LGTM https://github.com/llvm/llvm-project/pull/132401 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
