[PATCH] Add new debug kind LocTrackingOnly.

David Blaikie dblaikie at gmail.com
Mon Jun 23 15:20:02 PDT 2014


On Mon, Jun 23, 2014 at 2:57 PM, Diego Novillo <dnovillo at google.com> wrote:
>
>
>
> On Mon, Jun 23, 2014 at 4:45 PM, David Blaikie <dblaikie at gmail.com> wrote:
>>
>>
>> Well it's not worse - currently -Rpass doesn't enable debug info at
>> all , but I suppose it at least suggests the user enable both -gmlt
>> and -gcolumn-info. But I assume if they enabled -g(mlt or otherwise)
>> they wouldn't be told they should use -gcolumn-info.
>
>
> OK, I've removed the column info setting and modified the documentation to
> match the current semantics.
>
> Let's try to enable column info with -gmlt separately, then. Should I open a
> bug about it? Or do you prefer to handle it directly?

I'm not sure it's something I'm strongly invested in either way -
maybe Alexey (cc'd) has some idea of whether column info would add
anything of significance to backtraces. I would guess currently the
llvm symbolizer doesn't even try to handle column info, so it probably
doesn't add any value to the backtraces at the moment? Would it be
useful/appropriate, Alexey?

Even then, don't mind having a discussion about column info for -Rpass
stuff separately too. I mean it's sort of relevant to discuss whether
column info is good for -g modes because it'll impact whether -g modes
hurt -Rpass diagnostic quality... though if that's really a big deal
we could probably have a mode where column info is added but not used
for actual debug info emission.



More information about the cfe-commits mailing list