[llvm-commits] [PATCH] llvm-symbolizer: symbolize global variables

[Dmitry Vyukov]

  > I've got zero knowledge of lldb interface, so that part of the change should be reviewed by someone else (getNameFromSymbolTable, etc).


  This is not related to lldb, just elf.


  > I'm a bit worried about llvm-symbolizer interface becoming more complicated. Why do we need this CODE/DATA distinction anyway? Is it not possible to figure out if we are looking at code or data section just by the module offset?


  I think explicit is better than implicit. We always know what we want to symbolize code or data. Output for code and data differs, so client should guess what the symbolizer guessed. Moreover, potentially there can TLS addresses that overlap with code/data.

http://llvm-reviews.chandlerc.com/D277