[PATCH] [compiler-rt] Symbolizer refactoring: Make LibbacktraceSymbolizer adopt the SymbolizerTool interface

Alexey Samsonov vonosmas at gmail.com
Mon Mar 2 14:48:27 PST 2015


In http://reviews.llvm.org/D7971#132931, @kubabrecka wrote:

> > CHECK_EQ(addr, last->info.address);
>
> >  Should this be first->info.address?
>
>
> For `llvm-symbolizer`, we always have the same `addr` in all frames coming from one `SymbolizePC(addr)` request.  I am assuming the same holds for libbacktrace symbolizer.  If that's really the case, I'd even vote for having both checks (addr == first->info.address == last->info.address).


Makes sense, let's have both checks.

> Btw. do we have any buildbots that is at least building with `SANITIZER_LIBBACKTRACE=1`?


No, currently it's only used in GCC. Did you build it with it this definition?


http://reviews.llvm.org/D7971

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/






More information about the llvm-commits mailing list