[PATCH] D68465: [DebugInfo] Trim call-clobbered location list entries when tuning for GDB

Wed Oct 9 05:49:53 PDT 2019

dstenb added a comment.

In D68465#1698682 <https://reviews.llvm.org/D68465#1698682>, @dblaikie wrote:

> In D68465#1697407 <https://reviews.llvm.org/D68465#1697407>, @dstenb wrote:
>
> > In D68465#1695482 <https://reviews.llvm.org/D68465#1695482>, @dblaikie wrote:
> >
> > > Thanks for bringing this up!
> > >
> > > A few thoughts from me:
> > >
> > > 1. Yeah, I tend to agree with the DWARF Committee folks & the fact that LLDB can do the right thing without this change sort of points to this being a "fix it in GDB" situation. Have you tried asking the GDB folks about it/submitting patches there rather than here?
> >
> >
> > No, we have not done that yet.
>
>
> I think it'd be worthwhile having at least a statement from GDB that they feel this should be the responsibility of the producer. Though even if that's the answer they provide - I think some amount of pushback (especially given the existence proof of LLDB's behavior, by the sounds of it/if I'm understanding you correctly) might be worthwhile.

Yes, that's fair. I sent a mail to the GDB mailing list now: https://sourceware.org/ml/gdb/2019-10/msg00002.html. As mentioned in the mail, it appears that registers actually are described in that way for some targets, e.g. RS/6000 and S/390. Sorry for not noticing that before. (Although, I did not test any such targets in practice, so I might have misunderstood that code.)

Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D68465/new/

https://reviews.llvm.org/D68465