Re: [Libunwind-devel] patch: dwarf_find_save improvements

[Doug Moore]

I can't repro a problem on my machine.  I don't really grasp what  
you're saying about sending a pull request, but okay:

$ git pull https://github.com/libunwind/libunwind
From https://github.com/libunwind/libunwind
 * branch            HEAD       -> FETCH_HEAD
Already up-to-date.

Did a travis-CI run kick off for me somewhere, whatever that means?

I'm at a bit of a loss here.

Doug Moore


Quoting Dave Watson <address@hidden>:

> Hi Doug,
>
>
> On 04/26/17 05:19 PM, Doug Moore wrote:
> > <div class="PlainText">This patch addresses two issues.&nbsp;  
> > First, when caching is disabled, a failed call to  
> > create_state_record_for from uncached_dwarf_find_save_locs results  
> > in a failure to call put_unwind_info to release resources.&nbsp;  
> > Second, in the case of
> >  a cache miss, there is an extra, unnecessary copying of the  
> > register data. This change addresses both issues.<br>
>
> This one looks like it breaks one of the tests, Gtest-resume-sig
>
> https://travis-ci.org/djwatson/libunwind/jobs/226511961
>
> It seems to be related to the 'rs = &sr.rs_current;' on cache miss,
> but I haven't tracked it down exactly.
>
> If you send pull requests to https://github.com/libunwind/libunwind
> the travis-CI run should kick off for you.  It doesn't seem to repro
> on my local dev machine.
>
> Thanks
>
> !DSPAM:10223,59023d99294695992465132!