Re: [PATCH] nptl: Move pthread_equal implementation into libc

[Adhemerval Zanella <adhemerval dot zanella at linaro dot org>]

On 14/08/2019 05:44, Florian Weimer wrote:
> * Zack Weinberg:
> 
>> On Tue, Aug 13, 2019 at 4:18 PM Florian Weimer <fw@deneb.enyo.de> wrote:
>>> My goal is to get rid of the forwarders than do not depend on internal
>>> libpthread state fairly quickly, and then move the rest once the
>>> __libc_single_threaded flag has been around for a while (so that
>>> applications do not have to use pthread_create or something else for
>>> guiding optimization of atomics).
>>
>> FYI, I did some work on this last year, but I never posted the
>> patches, because I discovered that a complete job would be easier if
>> we first merged Adhemerval's cancellation overhaul.
> 
> I think there isn't an inherent dependency, just fewer textual changes
> as the result.
> 
>> That's still waiting for review, right?  We should really get that
>> done _anyway_...
> 
> I can try to review it this cycle.

I plan to push the sigcontext.h [1], this simplifies the code to parse
the program counter from the signal handler on all architectures. It
is a large patch unfortunately, however it is mainly a refactor so it
should be easier to get some review.

I have been working on BZ#12683 sporadically and do hope we could finally
make it upstream for next release. I tested it on more architectures 
recently (ia64, s390, sh4, mips), narrowing down some missing bits and 
in overall I think it is on good shape. The only architectures I haven't 
actually tested are csky, microblaze, nios2, and riscv.

[1] https://sourceware.org/ml/libc-alpha/2019-08/msg00025.html