labath added a comment. Hm... looking at @jankratochvil's patches, I realized that I completely omitted HashedNameToDIE from this patch (it used `emplace_back` to construct DIERefs, so it snuck past me). Thinking about it, I guess the best way to handle that would be to set the cu_idx to the invalid value for DIERefs created there, as the offset is sufficient to identify the die in those cases (which I guess is also the reason why all tests passed). That shouldn't impact the performance in any way as one way or the other one has to do a binary search to convert the offset into a DWARFUnit*. I'll try to update that on monday...
CHANGES SINCE LAST ACTION https://reviews.llvm.org/D61482/new/ https://reviews.llvm.org/D61482 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
