AW: RFC: POSIX timers and threads in a realtime context

[Joseph Myers]

On Tue, 6 Oct 2015, Zack Weinberg wrote:

> On Tue, Oct 6, 2015 at 11:46 AM, Joseph Myers <joseph@codesourcery.com> wrote:
> > On Tue, 6 Oct 2015, Zack Weinberg wrote:
> >
> >> It seems to me that Christof's second proposal (add a SIGEV_PTHREAD_ID
> >> that takes a pthread_t) is *much* simpler than doing all this work,
> >> and also far more likely to get standardized.
> >
> > However, my understanding is that there are several other interfaces where
> > tids are needed as well.  (The two proposals need not be mutually
> > exclusive.)
> 
> Do you have a list?  That would really help me, at least, understand
> the tradeoffs.

Several are mentioned in bug 6399.  (Note that some cases may be ones 
where POSIX specifies that a pid not a tid is taken - fixing such POSIX 
conformance bugs would require new symbol versions for the affected 
functions, and possibly new kernel interfaces to make it possible to pass 
a pid there.)

-- 
Joseph S. Myers
joseph@codesourcery.com