[Lldb-commits] [PATCH] D112058: [lldb/DWARF] Ignore debug info pointing to the low addresses

Greg Clayton via Phabricator via lldb-commits lldb-commits at lists.llvm.org
Wed Oct 20 10:48:53 PDT 2021


clayborg added a comment.

In D112058#3074618 <https://reviews.llvm.org/D112058#3074618>, @labath wrote:

> In D112058#3073451 <https://reviews.llvm.org/D112058#3073451>, @clayborg wrote:
>
>> Looks good. Do we need a follow up patch to avoid creating functions that should have been stripped?
>
> I'm not entirely sure what you mean by that. Are you referring to the fact that `SymbolFileDWARF::ResolveFunction` creates an `lldb_private::Function`, which it does not add the to result SymbolContextList (instead of not creating the object in the first place?
>
> If yes, then that sounds like a good idea. I don't know if this has more user-facing implications, but I was definitely puzzled by the fact that the stripped function still showed up in "image dump symfile". I'll see what can be done about that.

Yeah, that is what I was talking about. I think if we modify SymbolFileDWARF::ParseFunction() that would be the easiest to quickly get the ranges for the DW_TAG_subprogram and avoid calling DWARFASTParser::ParseFunctionFromDWARF() if the low PC is less than m_first_code_address we can avoid bloat in the ASTs and keep our memory footprint down in the LLDB process.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D112058/new/

https://reviews.llvm.org/D112058



More information about the lldb-commits mailing list