2007-10-11 23:45:13

by Martin

[permalink] [raw]
Subject: CMD 64x regression from 2.6.21 to 2.6.22 and 2.6.23?

Hi guys,

it has been 12 years since I was last actively working on linux 68k. Anyway,
this time I have a problem report on i386:

Background:
I upgraded one machine from 2.6.21.5 to 2.6.23 and experienced boot failure
due to the IDE disks being inaccessible. I worked out that the problem was
introduced with 2.6.22-rc1. The machine has an Asus CUBX motherboard. The
PIIX3 is switched off in the BIOS. All disks are hanging off the onboard
CMD648 IDE controller which is connected to IRQ 15.

Error description:
The kernel initializes normally, goes through init, into runlevel S. The
following is printed on the screen (there is no log saved on disk for obvious
reasons):

<normal kernel messages>
INIT: version 2.86 booting
proc on /proc type proc (rw)
sysfs on /sys type sysfs (rw)
Starting udevd: /sbin/udevd --daemon
Triggering udev events: /bin/udevtrigger
irg 15: nobody cared (try booting with the "irqpoll" option)
<call stack information>
Disabling IRQ #15
<endless loop printing /dev/hda & /dev/hdc lost interrupt, time out etc.>

When booting with the "irqpoll" option, the last line of output is "Triggering
udev events: /bin/udevtrigger" after which the machine appears to hang (no
disk activity, no reaction).

Attachments:
kernel configuration (of the 2.6.22-rc1 not working)
system configuration (taken under 2.6.21)
dmesg output (taken under 2.6.21)

Please let me know if you need further information or if you'd like me to try
something. Also, please CC me on your reply. Thanks,

Martin


Attachments:
(No filename) (1.48 kB)
kernel_config (33.11 kB)
system_config (13.54 kB)
dmesg.out (11.96 kB)
Download all attachments