Messages in this thread | | | From | Dmitry Torokhov <> | Subject | Re: mmotm 2010-05-19 BUG weirdness... | Date | Thu, 20 May 2010 09:45:24 -0700 |
| |
On Thursday 20 May 2010 05:55:20 am Andrew Morton wrote: > On Thu, 20 May 2010 11:47:50 -0400 Valdis.Kletnieks@vt.edu wrote: > > On Wed, 19 May 2010 16:13:09 PDT, akpm@linux-foundation.org said: > > > The mm-of-the-moment snapshot 2010-05-19-16-12 has been uploaded to > > > > > > http://userweb.kernel.org/~akpm/mmotm/ > > > > So I'm looking closer at the BUG I just posted > > I can't see that BUG report on lkml or in inbox. > > > - I had deleted two further > > BUGs because they were obviously follow-ons to the original. But then... > > > > Note the following 2 lines: > > > > [ 35.357018] note: keymap[2481] exited with preempt_count 1 > > [ 35.360503] BUG: scheduling while atomic: keymap/2481/0x10000002 > > > > The kernel reports the instigating process exited - and then reports it > > as the offender for a "scheduling while atomic". Insufficient attempted > > cleanup after the first BUG? Do we care because this is a sign of a > > scheduler bug that could trip on a non-BUG as well, or is it "all bets > > are off" because of the first BUG? > > Yes, the oops code will end up calling do_exit() to get rid of this > process and to try to keep the machine limping along. So if you hit an > oops with (say) a spinlock held, the task will end up calling do_exit() > with a non-zero preempt_count. > > So the only problem I'm seeing here is .... Dmitry's ;)
Hmm, any chance you could stick a printk in input_set_keycode and print the id/name of the input device?
-- Dmitry
| |