DragonFlyBSD

:New submission from Peter Avalos <peter@theshell.com>:
:
:I woke up to the console spewing this message infinitely. I couldn't seem to
:break out of it, and had to power cycle the machine. Kernel is: DragonFly
:v2.9.1.747.gf7b29d-DEVELOPMENT #42: Sun Feb 20 19:53:37 HST 2011
:
:status: unread
:title: lwkt_setcpu_remote: td->td_flags 00800621 console flood

Hmm. Too bad we couldn't get a core. I suspect the IPIQ fifo has
filled up and caused a livelock against another cpu trying to
migrate a thread at the same time.

I think I will have to change the cpu migration mechanics, the current
algorithm is clearly too fragile.

-Matt
Matthew Dillon
<dillon@backplane.com>

:New submission from Peter Avalos <peter@theshell.com>:
:
:I woke up to the console spewing this message infinitely. I couldn't seem to
:break out of it, and had to power cycle the machine. Kernel is: DragonFly
:v2.9.1.747.gf7b29d-DEVELOPMENT #42: Sun Feb 20 19:53:37 HST 2011

I'm taking a shot in the dark but I've made a change to the setcpu code
in commit 95858b9186e25a62d68c58e909b9fe7f50ee81d7 which *MIGHT* fix the
problem. It's going to reproduce fairly rarely but it's the only thing
I see that can cause the problem.

-Matt
Matthew Dillon
<dillon@backplane.com>

On Sun, Mar 06, 2011 at 10:26:49AM -0800, Matthew Dillon wrote:
> :New submission from Peter Avalos <peter@theshell.com>:
> :
> :I woke up to the console spewing this message infinitely. I couldn't seem to
> :break out of it, and had to power cycle the machine. Kernel is: DragonFly
> :v2.9.1.747.gf7b29d-DEVELOPMENT #42: Sun Feb 20 19:53:37 HST 2011
>
> I'm taking a shot in the dark but I've made a change to the setcpu code
> in commit 95858b9186e25a62d68c58e909b9fe7f50ee81d7 which *MIGHT* fix the
> problem. It's going to reproduce fairly rarely but it's the only thing
> I see that can cause the problem.
>

Yeh, I've only seen it happen once. I'll compile up a new kernel, and I
guess we'll just see if we ever see it again.

--Peter