I have a quad-core system that I updated to 8.0 about 2 weeks ago. It
ran fine until yesterday when I used the PATA/ATAPI DVD drive to read a
disk. Then the fun began. I saw two totally different presentations, but
I hope that they both are manifestations of the same problem.
First, the livelock. When attempting to play video off of the disk (xine
or mplayer) while the CPU was busy, xorg would lock up solid. All CPUs
were running a threaded code that run them all at 100%, but they were
running niced to 10.
Couldn't kill xorg, even with a 'kill -9'. Ate 100% of a CPU, though
the
CPU it was running on changed periodically. Could not switch to vty. The
display was frozen and I could do nothing from the console.
I could log in from another system and the system behaved normally.
While I could not free the console, I could and did manage to shut down
the system and reboot. This happened repeatedly and within a few
seconds of starting the video playing.
Now the deadlock. This one is weird in one instance. I had finished with
a disk and unmounted it. No problems. Then I ejected the unmounted
disk. (Yes, df(1) did not show it.) After the disk ejected, the system
deadlocked. Frozen completely. Could not even ping it on the net, so it
was in the kernel. I had no luck doing anything, so I had to kill power.
Then it got worse. The second time this happened, I inserted an erased
DVD while the system was in POST. The system started to boot, but,
shortly after probing acd0, while the boot was "waiting for: usbus7
usbus6 usbus5 usbus4 usbus3 usbus2 usbus1 usbus0", and about the time it
should have probed for cd0 (atapicam), the system locked up and would not
finish the boot. This is prior to mounting root.
After several repeats of this, I ejected the disk from the drive and it
booted just fine. Clearly something with ATAPI is not happy. Should I
not be using atapicam with 8.0?
I won't have access to this system until next Tuesday and the time I
have to tinker will be limited, but I can try setting up to break into
the debugger (if the system is even seeing the keyboard) or do something
else to get an idea of what is going on.
--
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman@es.net Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751