[rtems commit] posix: Fix pthread_spin_unlock()

[Gedare Bloom]

On Mon, Oct 19, 2020 at 10:59 PM Sebastian Huber
<sebastian.huber at embedded-brains.de> wrote:
>
> On 20/10/2020 00:17, Chris Johns wrote:
> > On 20/10/20 5:06 am, Gedare Bloom wrote:
> >> On Mon, Oct 19, 2020 at 11:11 AM Sebastian Huber
> >> <sebastian.huber at embedded-brains.de>  wrote:
> >>> On 19/10/2020 18:53, Gedare Bloom wrote:
> >>>
> >>>> On Mon, Oct 19, 2020 at 9:48 AM Sebastian Huber
> >>>> <sebastian.huber at embedded-brains.de>   wrote:
> >>>>> On 19/10/2020 17:42, Joel Sherrill wrote:
> >>>>>
> >>>>>> This was reported against 4.11 which means it needs to be committed to
> >>>>>> 4.11, 5, and master.
> >>>>> It depends on the severity of the bug if I create tickets and back ports
> >>>>> for this right now.
> >>>> If we know the bug exists we should at least create tickets on the
> >>>> open, affected branch(es).
> >>>>
> >>>> When we know how to fix it, we should also at least reference that as
> >>>> well. This way someone else can make the fix easier if they need it.
> >>> I thought this is the purpose of the "version" field?
> >>>
> >> https://lists.rtems.org/pipermail/devel/2018-August/050685.html
> >>
> >> The problem is that for example this ticket #4157 is closed. So how
> >> does someone know the bug isn't fixed in 4.11 and 5?
> >>
> >> It is a problem we face and why we end up duplicating all these
> >> tickets in the first place. I don't know (if there is) a right answer.
> > I think cloning tickets is the best solution we have. I cannot see a better path.
>
> Not everything is equally important.
>
> I did run the test suite in an unusual configuration, identified two
> unexpected test failures, debugged the tests, reported two issues with
> the first affected RTEMS version and a target milestone, and fixed the
> issues on the master. This is enough work for bugs which probably affect
> only one person on this planet.
>

I agree with this perspective as a philosophical point. Although, I
suspect there are GSoC students who ran the testsuite on 5 with SMP
enabled and PROFILING turned on. So whether a bug is so esoteric no
one else will encounter it is something of a subjective opinion.

But as a rule/standard, we should at least document known bugs on our
open branch (last released). This means if the bug is known to exist
on 5, there should be a ticket for it. Whether or not you want to fix
it is irrelevant. If creating a ticket is too much work, just say so
and maybe someone else will be bothered enough to handle it.

-Gedare

> --
> Sebastian Huber, embedded brains GmbH
>
> Address : Dornierstr. 4, D-82178 Puchheim, Germany
> Phone   : +49 89 189 47 41-16
> Fax     : +49 89 189 47 41-09
> E-Mail  : sebastian.huber at embedded-brains.de
> PGP     : Public key available on request.
>
> Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.