[KJ] Re: block/pktcdvd: set_current_state() location

[KJ] Re: block/pktcdvd: set_current_state() location

[Peter Osterlund]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 772 bytes --]

On Fri, 22 Oct 2004, Nishanth Aravamudan wrote:

> Jens & Peter,
>
> I was looking through 2.6.9-bk7 with the latest kjt (Kernel-Janitors
> patchset) applied and noticed a new schedule_timeout() in the code. In
> drivers/block/pktcdvd.c::kcdrwd(), the state is set at the head of the
> for(;;)-loop, while the sleeping doesn't occur until nearly the end of
> the same loop. Is there a reason for the code in-between to run in
> TASK_INTERRUPTIBLE?

Yes, it's needed to avoid race conditions. The code avoids race conditions
in the same way that __wait_event_timeout() does, but the "condition" is
much more complicated, so it wouldn't make sense to use the
__wait_event_timeout() macro directly.

-- 
Peter Osterlund - petero2@telia.com
http://w1.894.telia.com/~u89404340

[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

[KJ] Re: block/pktcdvd: set_current_state() location

[Nishanth Aravamudan]

[-- Attachment #1: Type: text/plain, Size: 881 bytes --]

On Fri, Oct 22, 2004 at 11:55:39PM +0200, Peter Osterlund wrote:
> On Fri, 22 Oct 2004, Nishanth Aravamudan wrote:
> 
> > Jens & Peter,
> >
> > I was looking through 2.6.9-bk7 with the latest kjt (Kernel-Janitors
> > patchset) applied and noticed a new schedule_timeout() in the code. In
> > drivers/block/pktcdvd.c::kcdrwd(), the state is set at the head of the
> > for(;;)-loop, while the sleeping doesn't occur until nearly the end of
> > the same loop. Is there a reason for the code in-between to run in
> > TASK_INTERRUPTIBLE?
> 
> Yes, it's needed to avoid race conditions. The code avoids race conditions
> in the same way that __wait_event_timeout() does, but the "condition" is
> much more complicated, so it wouldn't make sense to use the
> __wait_event_timeout() macro directly.

Great! Thanks for the quick response and information. I just wanted to
make sure.

-Nish

[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors