felipepiovezan wrote: > > we should probably fix the underlying issue instead: #77696 > > We still have binaries that are floating around for now that contain this > issue and this was causing crashes. So it would be nice to fix this in LLDB > for now and back this out after we have a stable and trustable .debug_names > section?
That's ok by me, but I still don't understand how the problem being solved here is related to IDX_parent. As I said in the email you quoted, we don't add links to parents that are forward declarations. When the parent chain of a DIE is built, if we see an `AT_declaration` (just like what is being done in this PR), we pretend the parent is not indexed. In other others, the queries involving IDX_parent will _never_ reach the ProcessEntry function. Note that only `DebugNamesDWARFIndex::GetFullyQualifiedType` makes use of IDX_parent, but there are tons of other queries that call `ProcessEntry`. Maybe the slow path you are reaching is from one of those other queries? https://github.com/llvm/llvm-project/pull/91808 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
