* [PATCH 0/5] ipc/sem.c: multiple fixes
@ 2011-10-15 13:21 Manfred Spraul
2011-10-17 12:43 ` Peter Zijlstra
0 siblings, 1 reply; 2+ messages in thread
From: Manfred Spraul @ 2011-10-15 13:21 UTC (permalink / raw)
To: LKML, Andrew Morton
Cc: Thomas Gleixner, Mike Galbraith, Peter Zijlstra, Manfred Spraul
Hi all,
The wakeup scheme used by semtimedop() does not work as intended on -rt.
The following series:
- fixes one bug that I noticed while thinking about a -rt compatible
implementation
- prevents the code from returning to user space due to spurious wakeups.
(noticed by Peter Zijlstra)
- moves private structures from include/linux/sem.h to ipc/sem.c
- saves some memory if sysvipc is disabled.
- reworks the wakeup scheme and adds a 2nd wakeup scheme that is based
on a completion. For CONFIG_PREEMPT_RT_BASE, this scheme is used.
What do you think?
The rework of the wakeup should also help to convert the code to
using the wake_up_list framework.
Andrew: Could you replace the patches in -mm with this series?
I would prefer to keep the current scheme as the default, it's
~5% faster (6.5 usec vs. 6.8 usec per wakeup).
The patches pass checkpatch.pl and I didn't observe any problems
with my test apps.
--
Manfred
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH 0/5] ipc/sem.c: multiple fixes
2011-10-15 13:21 [PATCH 0/5] ipc/sem.c: multiple fixes Manfred Spraul
@ 2011-10-17 12:43 ` Peter Zijlstra
0 siblings, 0 replies; 2+ messages in thread
From: Peter Zijlstra @ 2011-10-17 12:43 UTC (permalink / raw)
To: Manfred Spraul; +Cc: LKML, Andrew Morton, Thomas Gleixner, Mike Galbraith
On Sat, 2011-10-15 at 15:21 +0200, Manfred Spraul wrote:
> Hi all,
>
> The wakeup scheme used by semtimedop() does not work as intended on -rt.
>
> The following series:
> - fixes one bug that I noticed while thinking about a -rt compatible
> implementation
> - prevents the code from returning to user space due to spurious wakeups.
> (noticed by Peter Zijlstra)
> - moves private structures from include/linux/sem.h to ipc/sem.c
> - saves some memory if sysvipc is disabled.
> - reworks the wakeup scheme and adds a 2nd wakeup scheme that is based
> on a completion. For CONFIG_PREEMPT_RT_BASE, this scheme is used.
>
> What do you think?
> The rework of the wakeup should also help to convert the code to
> using the wake_up_list framework.
>
> Andrew: Could you replace the patches in -mm with this series?
> I would prefer to keep the current scheme as the default, it's
> ~5% faster (6.5 usec vs. 6.8 usec per wakeup).
>
> The patches pass checkpatch.pl and I didn't observe any problems
> with my test apps.
For patches 1-4:
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
I still need to kick my brain in gear and parse patch 5.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2011-10-17 12:44 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-10-15 13:21 [PATCH 0/5] ipc/sem.c: multiple fixes Manfred Spraul
2011-10-17 12:43 ` Peter Zijlstra
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.