Re: 2.6.14-rc3-rt10 build problem (now rt12)

[Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU>]

On Sun, 2005-10-09 at 06:33 +0200, Ingo Molnar wrote:
> * Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU> wrote:
> 
> > This appears to be triggered from Freqtweak (a Jack application):
> > Oct  8 18:48:00 cmn3 kernel: freqtweak:4705 userspace BUG: scheduling in
> > user-atomic context!
> > Oct  8 18:48:00 cmn3 kernel:  [<c037c05b>] schedule+0xeb/0x100 (8)
> > Oct  8 18:48:00 cmn3 kernel:  [<c037d745>] rwsem_down_read_failed
> > +0xa5/0x1c0 (28)
> > Oct  8 18:48:00 cmn3 kernel:  [<c01467ee>] .text.lock.futex+0xa9/0x2db
> > (52)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0130c46>] try_to_del_timer_sync
> > +0x46/0x50 (32)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0130c71>] del_timer_sync+0x21/0x30 (16)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0121330>] default_wake_function
> > +0x0/0x20 (32)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0151b5c>] audit_syscall_exit+0x4c/0x400
> > (12)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0146589>] do_futex+0x69/0xf0 (24)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0146674>] sys_futex+0x64/0x120 (24)
> > Oct  8 18:48:00 cmn3 kernel:  [<c0103471>] syscall_call+0x7/0xb (60)
> 
> could you enable CONFIG_DEBUG_PREEMPT and send me the same assert (which 
> will hopefully include a critical section list too).

I will try to do that tomorrow. 

I forgot to mention, all this is with the SMP kernel running on a dual
core Athlon, PREEMPT_DESKTOP and HZ=1000. 

The symptoms would seem to indicate some basic problem with timing and
what is reported to userspace (ie: wrong warnings about delayed
interrupts in Jack - I hear no audible problems and the time difference
reported by Jack keeps growing no matter what I do, seems to only be
realted to the time the kernel has been up, key repeat on X speeds up,
apparently the screensaver kicks in randomly, etc). 

-- Fernando