lkml.org 
[lkml]   [2009]   [Aug]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [tip:timers/core] clocksource: Resolve cpu hotplug dead lock with TSC unstable

* Martin Schwidefsky <schwidefsky@de.ibm.com> wrote:

> On Fri, 28 Aug 2009 18:34:00 GMT
> tip-bot for Thomas Gleixner <tglx@linutronix.de> wrote:
>
> > Commit-ID: 7285dd7fd375763bfb8ab1ac9cf3f1206f503c16
> > Gitweb: http://git.kernel.org/tip/7285dd7fd375763bfb8ab1ac9cf3f1206f503c16
> > Author: Thomas Gleixner <tglx@linutronix.de>
> > AuthorDate: Fri, 28 Aug 2009 20:25:24 +0200
> > Committer: Thomas Gleixner <tglx@linutronix.de>
> > CommitDate: Fri, 28 Aug 2009 20:25:24 +0200
> >
> > clocksource: Resolve cpu hotplug dead lock with TSC unstable
> >
> > Martin Schwidefsky analyzed it:
> > To register a clocksource the clocksource_mutex is acquired and if
> > necessary timekeeping_notify is called to install the clocksource as
> > the timekeeper clock. timekeeping_notify uses stop_machine which needs
> > to take cpu_add_remove_lock mutex.
> > Starting a new cpu is done with the cpu_add_remove_lock mutex held.
> > native_cpu_up checks the tsc of the new cpu and if the tsc is no good
> > clocksource_change_rating is called. Which needs the clocksource_mutex
> > and the deadlock is complete.
> >
> > The solution is to replace the TSC via the clocksource watchdog
> > mechanism. Mark the TSC as unstable and schedule the watchdog work so
> > it gets removed in the watchdog thread context.
> >
> > Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> > LKML-Reference: <new-submission>
> > Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
> > Cc: John Stultz <johnstul@us.ibm.com>
>
> Ah, very good. I've been going round in circles to find a solution
> that allows to downgrade the tsc rating when the second cpu is
> enabled. Could not find a solution. Your approach changes
> semantics slightly: the tsc clock will continue with its old
> rating for a while until the watchdog will do the downgrade. If
> that is acceptable then this is a good solution.

Latest timers/core also passed thousands of iterations of -tip
testing so far, so that painful series of locking and stability
troubles has been solved and the bits look good for v2.6.32.

Ingo


\
 
 \ /
  Last update: 2009-08-31 16:41    [W:0.046 / U:11.236 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site