Hi Kristian.
I think the problem here is that ide-cd tries to access /dev/hdd which is already used by ide-scsi. That looks like a forgotten check in ide-cd.
Could you first try 2.4.19-rc1 ? I couldn't reproduce a hang with it. A second "modprobe ide-cd" only stalls linux for 2 seconds and I'm getting:
hdd: set_drive_speed_status: status=0xd0 { Busy }
which means that modprobe is stucking in D state:
root 1288 0.0 2.2 5740 5740 pts/1 SL 09:59 0:00 cdrecord blank=fa
root 1290 14.9 0.2 1712 756 pts/1 D 09:59 0:10 modprobe ide-cd i
After finished burning (hdc is accessible again) it comes back.
Kristian Amlie <k_amlie@yahoo.com> wrote:
> Hi!
>
> [2.]
> I have a configuration with two CD-ROM drives; they are:
> 1. HL-DT-ST RW/DVD GCC-4120B (Goldstar)
> 2. Pioneer CD-ROM ATAPI Model DR-A04S 0105
> according to "/proc/ide/hdX/model".
>
> I use drive number one to burn CDs using the scsi interface with
> scsi_mod ide-scsi, sr_mod and sg. The program I use is cdrecord.
> My "modules.conf" file is set up to load "ide-cd ignore=hdc" before the
> sg interface is opened. This way, the ide interface takes care of my
> second CD-ROM (hdd) and the scsi interface takes care of the first
> (hdc).
> This configuration works fine. However, a problem arises because the
> ide-cd module is often autocleaned while a burning is in progress. If I
> then try to access the second CD-ROM (hdd), the ide-cd module hangs on
> initialization (at least that's what lsmod tells me).
> The burning process also stops, and the programs that tried to access
> the second drive are caught in their uninterruptable sleeps.
> Shortly after this (about a minute or two) I get a message that loading
> the module failed.
> And that is almost immediately followed by a complete kernel crash. I
> am afraid I cannot provide any screen output, as it starts printing hex
> values at unlimited speed, and I am unable to read it.
> The kernel doesn't react to any keypresses (like ScrollLock to stop the
> output), not even the Magic SysRq key, which I have activated.
If this still persists:
It could be helpful for the developers to have the oops-output. Try to attach a serial console. First enable CONFIG_SERIAL_CONSOLE in your kernel-config. If done, append the following options to lilo. For example my lilo.conf:
image=/boot/vmlinuz-2.4.19-rc1-kp1
label=udmacda
append="console=ttyS0,57600n8,vt102 console=tty0"
read-only
root=/dev/hda1
Then you need a second computer to capture that output (with mingetty or anything similar). Maybe you could redirect the klog output via syslog to another machine. I hope you have the possibilities.
*Kristian
> #!/bin/bash
>
> modprobe ide-cd ignore=hdc
> modprobe sg
> cdrecord dev=<insert scsi device> <SomeBigIsoFile> &
> # Give it some time to start the burning.
> sleep 120
> modprobe -r ide-cd
> modprobe ide-cd
> echo "Shouldn't get here! The problem obviously doesn't reveal itself
> under this environment!"
:... [snd.science] ...:
:: _o)
:: http://www.korseby.net /\\
:: http://gsmp.sf.net _\_V
:.........................:
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Jul 15 2002 - 22:00:25 EST