sivachandra added a comment. About performance regression, I am not sure if it really does cause any; For those cases where looking up a function with mangled name actually works, the new lookup should not trigger. For those where it does not work, additional lookup could now make it work.
About cases where in looking up with full (demangled) name helps, the two test cases from which the expectedFailure decorators were removed are examples. Essentially, the issue is that the mangled name used by the inferior compiler is different from the mangled name as generated by LLDB (via Clang). However, since both the manglings resolve to the same demangled name, looking up a function using the demangled name works. Another example is std::vector<int>::size(); GCC does not generate complete DWARF for the this method, and hence LLDB (via Clang) generates a different mangled name as compared to what is present in the SymTab. http://reviews.llvm.org/D12613 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
