[391] in linux-scsi channel archive
NCR53c810 lockups
daemon@ATHENA.MIT.EDU (Geoffrey Bennett)
Tue Jul 18 02:47:07 1995
From: Geoffrey Bennett <geoffrey@tafe.sa.edu.au>
To: linux-scsi@vger.rutgers.edu
Date: Wed, 21 Jun 1995 22:14:30 +0930 (CST)
'lo all...
I'm having a bit of trouble at the moment with lockups on a computer
here with an NCR53c810 controller.
If I run disk intensive stuff (e.g. cat /dev/sda > /dev/null and do
something else) then I get messages like:
Jun 21 22:04:42 Kara kernel: scsi0 : DANGER : abort_connected() called
Jun 21 22:04:42 Kara kernel: scsi0 : DMA FIFO not empty
Jun 21 22:04:42 Kara kernel: scsi0 : DMA FIFO not empty
Jun 21 22:04:42 Kara kernel: scsi0 : unexpected phase unknown at dsp = 0x1d1270
Jun 21 22:04:42 Kara kernel: 001d1270 : 0x0e000001 0x001d0a99
Jun 21 22:04:42 Kara kernel: 001d1278 : 0x48000000 0x00000000
appearing. These messages didn't seem to be a problem before, but
if I do enough to trigger *LOTS* of them then the scsi system will die
eventually. I was running top in another window when I last triggered
the lockup and it was showing:
7 root 1 0 32 168 340 D 0.0 1.1 0:00 update (bdflush)
46 root 1 0 81 284 332 D 0.0 1.8 0:04 /usr/sbin/syslogd
The machine is running headless, so I don't know whether there were
any parting messages from the kernel before the scsi stuff hung up.
Other stuff: it's a 486DX4/100; I've got a Pentium/66 running the same
controller, and it has never given me these danger messages, so I'll
try swapping the controllers to see where the problem lies, but I just
thought I'd ask to see if anyone has seen this problem before, and
whether they have an explanation or cure for it.
Thanks,
--
___
/ __
\___|eoffrey D. Bennett!-) geoffrey@tafe.sa.edu.au
