mstorsjo added a comment. In D70840#1792498 <https://reviews.llvm.org/D70840#1792498>, @labath wrote:
> In D70840#1791292 <https://reviews.llvm.org/D70840#1791292>, @mstorsjo wrote: > > > And irrespectively if the ArchSpec vs Architecture design, can you (either > > of you) comment on the updated form of the patch? > > > The code still seems somewhat schizophrenic to me. :/ The line tables are > fixed up super late, but DW_AT_low_pc is adjusted very early. The line table > adjustment happens even after sorting, which means the fixup could alter the > sort order. It probably wouldn't matter in practice, as everything would just > get decremented by one, but it still seems like a bad design. And adjusting > the low_pc so early will complicate the move to the llvm dwarf parser. > > I think I'd most prefer some middle ground where the fixup happens after the > lowest extraction layers are finished, but before the data hits the "generic" > code. It's possible that no such place exists right now, but it might be > possible to create something with a bit of refactoring... I tried to revisit this a bit now. Thanks to D72920 <https://reviews.llvm.org/D72920>, some of the more problematic cases went away, and I tried to trace all callers of the relevant methods and moving the fixups into them. Now the DWARFDebugInfoEntry class is no longer touched at all. I also tried to move fixups to before sorting. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D70840/new/ https://reviews.llvm.org/D70840 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
