Re: [PATCH]: expose SIGRTMIN and SIGRTMAX

[Charles Cui <charles.cui1984%gmail.com@localhost>]

Thanks Martin letting me know the problems in the patch.
I will read the documents recommended by you carefully.
Let me know if you have specific requirements for those signals in kernel
space.
For me, my principle is to fix bugs in posix test suite and do necessary
work in kernel.
Those two signals are just guards (the range for reliable or non-reliable
signals),
So, before reading your docs, I will guess exposing to userland is enough.
Also thanks Ngie letting me know the github version of the benchmark.

2016-05-15 0:24 GMT-07:00 Ngie Cooper (yaneurabeya) <yaneurabeya%gmail.com@localhost>:

>
> > On May 15, 2016, at 00:20, Martin Husemann <martin%duskware.de@localhost> wrote:
> >
> > On Sat, May 14, 2016 at 10:48:23PM -0700, Charles Cui wrote:
> >> Hi Guys,
> >>
> >>   I have attached a patch which is used to expose SIGRTMIN and SIGRTMAX
> to
> >> user land. This will remove a lot of errors related with those macros in
> >> posix test benchmark (http://posixtest.sourceforge.net/).  Comments are
> >> welcomed. Besides this mail list, I believe it would be useful to also
> send
> >> to netbsd kernel mail list, I am wondering which list is the good place
> to
> >> review this patch?
> >
> > Please do not cross-post to several mailing lists. In this case actually
> > tech-kern@ might be a good place to ask, but it is not clear, so we can
> > just continue discussion here.
> >
> > I don't think exposing the bounds to userland is enough to fix the issue,
> > we need to check that we have all parts implemented for
> >
> >       2.4.2 Realtime Signal Generation and Delivery
> >
> > and we don't, see "man 3 sigevent" and search for SIGEV_THREAD.
> >
> > So this is a non-trivial fix (Christos would know better how hard it
> actually
> > is to implement), and a good step for your project would be to examine
> the
> > current state and write a PR describing the bits we are missing.
> >
> > Then either, as you go along with other stuff, somebody (Christos?),
> > will fill the missing parts, or in the end this part will (for now) stay
> > incomplete.
>
>         Quick sidenote: the project noted above from sourceforge.net is
> dated and a bit flawed. Please use the version from ltp:
> https://github.com/linux-test-project/ltp/tree/master/testcases/open_posix_testsuite
> — if you have any questions or concerns, please feel free to contact me
> directly (yaneurabeya@ on github, etc). I’ll do my best to push through
> compatibility improvements where it makes sense.
> Thanks,
> -Ngie