Re: [PATCH, newlib] Allow locking routine to be retargeted

[Thomas Preudhomme <thomas dot preudhomme at foss dot arm dot com>]

Hi Freddie,

Have you had time to work on the last version of the patch?

Best regards,

Thomas

On 16/01/17 15:59, Thomas Preudhomme wrote:
> Please find attached the updated patch.
>
> I've noticed that there is also some target and OS specific locks:
> _arc4random_mutex and many Linux/rtems/tirtos specific locks.
>
> I haven't included those.
>
> Best regards,
>
> Thomas
>
> On 13/01/17 18:16, Thomas Preudhomme wrote:
> > On 13/01/17 18:05, Freddie Chopin wrote:
> > > On Fri, 2017-01-13 at 13:24 +0000, Thomas Preudhomme wrote:
> > > > And here's the patch. Tested with full and partial retargeting as
> > > > well as no
> > > > retargeting.
> > >
> > > I confirm - if I provide just some of the functions/objects, the
> > > linking will fail with "multiple definition" error. Great!
> > >
> > > Any chance for adding all locks used by newlib internally? These 4 are
> > > missing from what I've seen:
> > > newlib/libc/posix/telldir.c:__LOCK_INIT(static, dd_hash_lock);
> > > newlib/libc/time/tzlock.c:__LOCK_INIT(static, __tz_lock_object);
> > > newlib/libc/stdlib/quick_exit.c:__LOCK_INIT(static, atexit_mutex);
> > > newlib/libc/stdlib/envlock.c:__LOCK_INIT_RECURSIVE(static, __env_lock_object);
> > >
> > > I know these are not "popular", but nevertheless required. For example
> > > time-zone locking is pulled by any function dealing with "local" time,
> > > like a very useful mktime(). 4 bytes here or there is not such a big
> > > deal (;
> >
> > Oh yes absolutely. I forgot that I added the lock only based on testing rather
> > than analysis of the code. Thanks for pointing that out.
> >
> > Best regards,
> >
> > Thomas