AaronBallman wrote: > Hi @AaronBallman > > I see that if I build clang with ASAN with this patch and run the testcase > `clang/test/C/C99/n590.c` it crashes and I see this > > ``` > AddressSanitizer:DEADLYSIGNAL > ================================================================= > ==2063954==ERROR: AddressSanitizer: SEGV on unknown address 0xb5c8001f7e52 > (pc 0x7fe2bd0e8baf bp 0x7fe2bd718370 sp 0x7fe2bd7181a0 T0) > ==2063954==The signal is caused by a WRITE memory access. > #0 0x7fe2bd0e8baf in raise (/lib64/libpthread.so.0+0x12baf) (BuildId: > 1962602ac5dc3011b6d697b38b05ddc244197114) > #1 0x5610e162f93e in SignalHandler(int, siginfo_t*, void*) > /repo/llvm/build-all-bbisdk-asan/../lib/Support/Unix/Signals.inc > #2 0x7fe2bd0e8d0f (/lib64/libpthread.so.0+0x12d0f) (BuildId: > 1962602ac5dc3011b6d697b38b05ddc244197114) > #3 0x5610e888912a in AnalyzeImplicitConversions(clang::Sema&, > clang::Expr*, clang::SourceLocation, bool) > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:12638 > #4 0x5610e888a70b in CheckCommaOperand > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:11653:3 > #5 0x5610e888a70b in AnalyzeImplicitConversions > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:12510:7 > #6 0x5610e888a70b in AnalyzeImplicitConversions(clang::Sema&, > clang::Expr*, clang::SourceLocation, bool) > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:12642:5 > [...] > #730 0x5610e888a70b in CheckCommaOperand > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:11653:3 > #731 0x5610e888a70b in AnalyzeImplicitConversions > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:12510:7 > #732 0x5610e888a70b in AnalyzeImplicitConversions(clang::Sema&, > clang::Expr*, clang::SourceLocation, bool) > /repo/llvm/build-all-bbisdk-asan/../../clang/lib/Sema/SemaChecking.cpp:12642:5 > > AddressSanitizer can not provide additional info. > SUMMARY: AddressSanitizer: SEGV (/lib64/libpthread.so.0+0x12baf) (BuildId: > 1962602ac5dc3011b6d697b38b05ddc244197114) in raise > ==2063954==ABORTING > ```
I'm trying to reproduce the issue on Windows with MSVC + ASAN and I'm not getting any failures. The stack trace looks valid, but also implies that `E = E->IgnoreParenImpCasts();` is somehow resulting in an invalid pointer being passed to `AnalyzeImplicitConversions()` which is a surprise; I would expect `E->` to be an issue if there was an invalid pointer involved with the changes in this PR. I may need a bit of help on this one if I can't reproduce locally. https://github.com/llvm/llvm-project/pull/138752 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
