* Matthew Wilcox <matthew@wil.cx> wrote:

quoted text
> > > Current XFS CVS has a fix from myself and Christoph that gets rid 
> > > of the l_flushsema and replaces it with a staggered wakeup of each 
> > > task that's waiting as the previously woken task clears the 
> > > critical section.
> > 
> > the solution is to reduce semaphore usage by converting them to 
> > mutexes. Is anyone working on removing legacy semaphore use from 
> > XFS?
> 
> This race is completely irrelevant to converting semaphores to 
> mutexes. [...]

i was not talking about the race. I was just reacting on your comments 
about thundering herds and staggered wakeups - which is a performance 
detail. Semaphores should not regress AIM7 by 50% but otherwise they are 
legacy code and their use should be reduced monotonically, so i was 
asking why anyone still cares about tuning semaphore details in XFS 
instead of just working on removing semaphore use from them.

quoted text
> [...] It can only occur for semaphores which /can't/ be converted to 
> mutexes.

exactly what usecase is that? Perhaps it could be converted to an atomic 
counter + the wait_event() APIs.

	Ingo
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/