Re: [BUG on PREEMPT_RT, 2.6.23.1-rt5] in rt-mutex code and signals

[Ingo Molnar <mingo@elte.hu>]

* Daniel Walker <dwalker@mvista.com> wrote:

> On Sat, 2007-11-17 at 18:46 +0100, Ingo Molnar wrote:
> > * Daniel Walker <dwalker@mvista.com> wrote:
> > 
> > > > Actually, IMO, compat_semaphores behave like semaphores should 
> > > > behave, and thus the same as they behave on a non-RT kernel, and at 
> > > > the locations where the semaphores are now misused as mutexes on RT, 
> > > > we should replace them by differently-named-mutex-type-semaphores, 
> > > > or better: real-RT-mutexes..
> > > 
> > > The vast majority of semaphore are actually binary semaphores in the 
> > > Linux kernel .. So it's easier to mass convert semaphores to mutexes, 
> > > then address the ones that don't conform.. Usually they are converted 
> > > to the complete API in mainline..
> > 
> > right now there are 3992 mutex_lock() critical sections in the kernel 
> > and only 351 down() based critical sections are left.
> > 
> > fixing the top 20:
> > 
> >       4 &vuart_bus_priv.probe_mutex
> >       5 &connections_lock
> >       5 &irq_ptr->setting_up_sema
> >       5 &kbd->sem
> >       5 &pnp_res_mutex
> >       5 &port->port_lock
> >       5 &tq_init_sem
> >       6 &adb_handler_sem
> >       6 &dev->parent->sem
> >       6 &driver_lock
> >       6 &ha->vport_sem
> >       7 &big_buffer_sem
> >       8 &dir_f->sem
> >       9 &c->alloc_sem
> >      11 &dev->sem
> >      11 &usbvision->lock
> >      12 &c->erase_free_sem
> >      15 &u132->scheduler_lock
> >      16 &zfcp_data.config_sema
> >      17 &f->sem
> > 
> > would remove 164 of them, so it would convert half of the remaining 
> > semaphore use in the kernel. So the job is almost finished - would 
> > anyone like to go for the final grand feat: complete removal of 
> > semaphores from the kernel? :-)
> 
> Sure, you want to split the list?

split the list with you? Feel free to take any of those :-) dev->sem is 
nontrivial and probably not possible right now - and some of the others 
might be problematic too. But there might be fixable ones in the list. 
This shouldnt become like the BKL conversion - never truly finished.

	Ingo