Bug#516752: [etchnhalf] ide cd-rom drive causes kernel problem
With apologies for the somewhat sporadic testing, I have the following
to report:
On Sun, Apr 29, 2012 at 9:10 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:
> Jonathan McCune wrote:
>
>> That machine now sits gathering dust. If you tell me it's a
>> worthwhile effort, I will dig it back out and give it a try.
Actually I could not find the exact same machine, but I did find
another Dell Optiplex gx620 with an IDE CD ROM drive and it exhibits
similar problems with the "etchnhalf" kernel. I did not go back and
try with regular Etch.
> Yes, it would be interesting to try.
Here we are. :)
> If I understood correctly, the bug you found was in the piix driver.
> Squeeze and wheezy kernels use the newer ata_piix driver, which we
> would like to ensure works well.
I tested the amd64 version of squeeze, installing a standard desktop
system. Tons of information here:
http://www.ece.cmu.edu/~jmmccune/idecd/debian6.0/
The following did appear in the output of dmesg after inserting a CD:
[10894.233350] UDF-fs: No VRS found
[10894.233355] UDF-fs: No partition found (1)
[10894.250339] ISO 9660 Extensions: Microsoft Joliet Level 3
[10894.275514] ISO 9660 Extensions: RRIP_1991A
[10921.426157] ata1: drained 1024 bytes to clear DRQ.
[10921.426183] ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
[10921.426190] sr 0:0:0:0: [sr0] CDB: Volume set (in), Read cd: be 00
ff ff ff ff 00 00 01 10 00 00
[10921.426207] ata1.00: cmd a0/00:00:00:00:00/00:00:00:00:00/a0 tag 0
[10921.426209] res 58/00:02:00:00:08/00:00:00:00:00/a0 Emask
0x2 (HSM violation)
[10921.426212] ata1.00: status: { DRDY DRQ }
[10921.484026] ata1: soft resetting link
[10921.662296] ata1.00: configured for UDMA/33
[10921.663893] ata1: EH complete
Other than that, I was able to copy the full contents of a CD in a
reasonable period of time, and automount, etc worked as expected. I
did not try to burn a CD.
Now, regarding the old etchnhalf install, I grabbed an ISO from
archive.debian.org:
cd732225b84c5a9ef1051dc86b3acbe8916cc62e
/export/iso/debian-40r9-etchnhalf-i386-netinst.iso
I burned that to disk on a newer system, and used that disk to install
Etchnhalf on the gx620 (clobbering the squeeze install; apologies; I
can redo if needed).
Strangely, the installation works. Because Etch is "retired", I had
to use expert mode and manually specify a package mirror, and even
then the normal package selection did not behave very well. But I did
end up with a bootable system and apt-get did the rest. :)
Plenty of CD-related errors as the system boots, even with an empty
drive. Attempting to 'sudo mount /media/cdrom' takes forever, and
tons of kernel stack traces roll by on the primary text console. It
does actually "work" eventually, though. I can run 'ls' on the
contents of /media/cdrom. I didn't pursue it much further, since
repairing Etch isn't really the goal.
Similar logs available:
http://www.ece.cmu.edu/~jmmccune/idecd/debian4.0r9/
For both releases, I really only ran the following meaningful commands:
apt-get install aptitude openssh-server
# just a list of packages I often find useful
aptitude install build-essential kernel-package libncurses5-dev
fakeroot wget bzip2 libncurses5-dev autoconf automake libtool
libgtk2.0-dev libssl-dev less minicom vim subversion openssh-server
emacs nfs-common nfs-client mercurial rsync traceroute bridge-utils
gawk python-dev git-core git-svn gitk libpci-dev bcc bin86 uuid-dev
iasl ntpdate acpidump jsvc secure-delete lshw hwinfo
bash getinfo.sh gx620
That 'getinfo.sh' script is visible here: http://www.ece.cmu.edu/~jmmccune/idecd
To me, the main question is whether the dmesg contents on the
_squeeze_ system are any indication of trouble. I did not try wheezy.
Regards,
-Jon
Reply to: