Re: [patch] voluntary-preempt-2.6.9-rc1-bk4-R3

[Ingo Molnar]

* K.R. Foley <kr@xxxxxxxxxx> wrote:

> I am glad that it's reproducible for you as well. How did you trigger
> it? Because it seems to only crash under heavy load for me. The system
> has been up since I rebooted last night after the crash and I haven't
> seen any problems. Same thing goes for up until last night when I
> booted the new patch. Even building the new patch didn't seem to be
> enough to trigger it.

i triggered it via two IO-intense scripts running on a 256 MB RAM
testbox:

  cd /tmp; while true; do dd if=/dev/zero of=bigfile bs=1000000 
  count=500 >/dev/null 2>/dev/null; sync; date; sleep 60; done &

  while true; do du /usr >/dev/null 2>/dev/null; date; done &

usually it triggers within half an hour or so of runtime. With -R4 it
didnt crash yet after ~1 hour of running so i'm quite optimistic that 
this particular bug has been fixed.

i think the key to reproduce was to use KP=0,SP=0,HP=0 - this
concentrates all preemption activity to the places the VP patch adds -
amongst them the one in the do_exit() path that is necessary to trigger
the race.

	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/