Re: [tip:core/rcu] rcu: Fix holdoff for accelerated GPs for lastnon-dynticked CPU

[Ingo Molnar]

* Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx> wrote:

> On Wed, Mar 17, 2010 at 06:53:11PM +0100, Ingo Molnar wrote:
> > 
> > * Ingo Molnar <mingo@xxxxxxx> wrote:
> > 
> > > incidentally i have a RCU badness crash log from another box:
> > > 
> > > [    4.573332] initcall ipmi_init_msghandler_mod+0x0/0xd returned 0 after 0 usecs                 
> > > [    4.573332] calling  init_ipmi_si+0x0/0x435 @ 1
> > > [    4.573332] IPMI System Interface driver.
> > > [    4.573340]  [<ffffffff81022cdc>] no_context+0x1fc/0x20b
> > > [    4.576672]  [<ffffffff81002a16>] ? ftrace_call+0x5/0x2b
> > > [    4.580021]  [<ffffffff81022e77>] __bad_area_nosemaphore+0x18c/0x1af
> > > [    4.583338]  [<ffffffff81002a16>] ? ftrace_call+0x5/0x2b
> > > [    4.586672]  [<ffffffff81022ead>] bad_area_nosemaphore+0x13/0x15
> > > [    4.590005]  [<ffffffff81023159>] do_page_fault+0x138/0x284
> > > [    4.593354]  [<ffffffff81a632bf>] page_fault+0x1f/0x30
> > 
> > I spent a few hours yesterday and today trying to track this down, but failed 
> > to bisect it. The crash is sporadic on many configs - which confused my 
> > bisection attempts. Sometimes it crashes only after 10 reboots. Was able to 
> > crash v2.6.30 as well, so it's an older bug. v2.6.29 was not testable on that 
> > box due to network driver problems so i stopped there.
> 
> OK, I give up...
> 
> What did you change about your testing?  ;-)

Nothing really - and it has not reoccured on that box since i let it move on 
to another random config. This might be a 1:100000 kind of blue moon bug. 
(literally)

My suspicion is that it's an ftrace bug, all the crashes happened during 
function-tracer self-tests.

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/