Re: False positive (well not really) on RT backward clock check

[john stultz <johnstul@us.ibm.com>]

On Fri, 2005-10-21 at 17:52 +0200, Thomas Gleixner wrote:
> On Fri, 2005-10-21 at 11:46 -0400, Steven Rostedt wrote:
> > 
> > I think you're right about that.  Here it is.
> > 
> > Sorry for the late reply.  The original was from my custom kernel and I
> > wanted to get an output from Ingo's.  This is rc5-rt1.
> > 
> > -- Steve
> > 
> > isapnp: Scanning for PnP cards...
> > Time: tsc clocksource has been installed.
> > WARNING: non-monotonic time!
> > Ktimers: Switched to high resolution mode CPU 0
> > softirq-timer/1/14[CPU#1]: BUG in ktime_get at kernel/ktimers.c:101
> >  [<c0104433>] dump_stack+0x23/0x30 (20)
> >  [<c0121427>] __WARN_ON+0x67/0x80 (44)
> >  [<c013ad82>] ktime_get+0xd2/0x100 (48)
> >  [<c013c2b6>] ktimer_run_queues+0x56/0x130 (60)
> >  [<c012abbe>] run_timer_softirq+0x12e/0x450 (56)
> >  [<c01268b0>] ksoftirqd+0x120/0x1a0 (40)
> >  [<c01376eb>] kthread+0xbb/0xc0 (48)
> >  [<c01014a5>] kernel_thread_helper+0x5/0x10 (538427420)
> > ---------------------------
> > | preempt count: 00000001 ]
> > | 1-level deep critical section nesting:
> > ----------------------------------------
> > .. [<c0143edc>] .... add_preempt_count+0x1c/0x20
> > .....[<c01213db>] ..   ( <= __WARN_ON+0x1b/0x80)
> > 
> > Ktimers: Switched to high resolution mode CPU 1
> 
> This is at the moment where the clock source is switched over. I check
> what might be the reason.
> 
> John, any idea ?

This is using the B7 tod code, correct? If so, B8 corrected a problem
specifically with changing clocksources (most easily seen while changing
to the TSC based clocksource), so I would not be surprised if this is
the same bug.

thanks
-john