Am Dienstag, den 26.08.2014, 17:36 -0400 schrieb Rich Felker:
> Fixing it should be trivial via the design I mentioned earlier: don't
> use a waiter flag like this, but instead offset the initial value of
> ref by +1 and a_dec it just before waiting. As in other places, of
> course, a wake to an invalid address is possible either way; this is
> "fixable" if necessary via FUTEX_WAKE_OP (having the kernel do the
> atomic dec after acquiring the futex bin locks).

generally it would be nice to have such a lock functionality that
takes care of the waiters inside the int itself, this could perhaps be
used in other places

> > Ok, I'll try to switch to one-step-at-a-time-mode.
> 
> OK.

ok, so in a follow up I will send two patches that are the essence of
that. It doesn't do the booking from above, let's keep that for later.

Jens Gustedt (2):
  Let the signaler do all work concerning consistency of the list
  avoid taking _c_lock if we know it isn't necessary

 src/thread/pthread_cond_timedwait.c |   62 +++++++++++++++++++++++++++--------
 1 file changed, 48 insertions(+), 14 deletions(-)


> > > Or, if you don't have time to spend on side projects like test
> > > cases, maybe someone else could test it?
> > 
> > Not at the moment. I would prefer to have a C thread implementation by
> > the end of the week that we could agree upon.
> 
> Yes, that would be very nice!

ok, so I will be turning back to that, and as said that will be a
version of C threads that use the POSIX lock structures transparently.

Jens


-- 
:: INRIA Nancy Grand Est ::: AlGorille ::: ICube/ICPS :::
:: ::::::::::::::: office Strasbourg : +33 368854536   ::
:: :::::::::::::::::::::: gsm France : +33 651400183   ::
:: ::::::::::::::: gsm international : +49 15737185122 ::
:: http://icube-icps.unistra.fr/index.php/Jens_Gustedt ::