lkml.org 
[lkml]   [2008]   [Jun]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH 5/5] futex: fix miss ordered wakeups
From
Date

On Thu, 2008-06-12 at 10:56 +0200, Thomas Gleixner wrote:
> On Wed, 11 Jun 2008, Daniel Walker wrote:
> > Adds an additional function call to the sched_setscheduler to update the
> > waiter position of a task if it happens to be waiting on a futex. This
> > ensures that the kernel level waiter ordering is correctly maintained
> > based on the changed priority of the task.
> >
> > I fixed the locking issue noticed by Thomas Gleixner.
> >
> > This doesn't address userspace at all, only the kernel level wakeups and
> > kernel level ordering.
> >
> > The additional locking added to the futex_wait function has no visible speed
> > impact, and only effects waiters which actual enter the kernel.
>
> The additional locking is just broken and you did not even bother to
> test your changes with lockdep.

I ran it with lockdep enabled , I didn't get any warnings..

> Aside of this, these patches still add 100 lines of code to achieve
> nothing - as dicussed when you previously submitted your changes.
>
> Please stop wasting everyone's time with that.

It achieves correct ordering of the futex waiters inside the kernel,
that is in fact _something_ ..

Daniel




\
 
 \ /
  Last update: 2008-06-12 15:33    [W:0.059 / U:1.520 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site