1

I'm using a 16-to-18-years-old laptop, a Clevo Zeus M3CW, on which I recently installed antiX Linux 19.3. The kernel version is 4.9.235 (compiled for 486).

Now, whenever I boot, the kernel complains about I/O errors from device /dev/sr0 - my laptop's built-in CD-ROM. This is what lsblk says about the device:

# lsblk -o +VENDOR,MODEL | head -1 ; lsblk -o +VENDOR,MODEL | grep sr0
NAME   MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT VENDOR   MODEL
sr0     11:0    1  1024M  0 rom             TEAC     DW-224E-A

The drive is empty - no CD in there.

Later while the system runs, I get innumerable dmesg entries such as the following:

[   53.250051] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.250056] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.250066] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.250071] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.250076] sr 1:0:0:0: [sr0] tag#0 unaligned transfer

and also this:

[   53.312725] REISERFS warning (device sr0): sh-2006 read_super_block: bread failed (dev sr0, block 16, size 512)
[   53.312746] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.312754] REISERFS warning (device sr0): sh-2006 read_super_block: bread failed (dev sr0, block 128, size 512)
[   53.359602] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.359617] EXT4-fs (sr0): unable to read superblock
[   53.405593] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.405606] EXT4-fs (sr0): unable to read superblock
[   53.464594] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[   53.464608] EXT2-fs (sr0): error: unable to read superblock
[   53.533449] sr 1:0:0:0: [sr0] tag#0 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[   53.533461] sr 1:0:0:0: [sr0] tag#0 Sense Key : Medium Error [current] 
[   53.533478] sr 1:0:0:0: [sr0] tag#0 Add. Sense: Unable to recover table-of-contents
[   53.533490] sr 1:0:0:0: [sr0] tag#0 CDB: Read(10) 28 00 00 00 00 00 00 00 02 00
[   53.533516] SQUASHFS error: squashfs_read_data failed to read block 0x0
[   53.533528] squashfs: SQUASHFS error: unable to read squashfs_super_block

and these:

[ 2852.951421] sr 1:0:0:0: [sr0] tag#0 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[ 2852.951431] sr 1:0:0:0: [sr0] tag#0 Sense Key : Medium Error [current] 
[ 2852.951444] sr 1:0:0:0: [sr0] tag#0 Add. Sense: Unable to recover table-of-contents
[ 2852.951453] sr 1:0:0:0: [sr0] tag#0 CDB: Read(10) 28 00 00 07 ff fc 00 00 02 00
[ 2852.951458] blk_update_request: 1680 callbacks suppressed
[ 2852.951462] blk_update_request: I/O error, dev sr0, sector 2097136
[ 2852.951509] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[ 2852.951514] blk_update_request: I/O error, dev sr0, sector 2097136
[ 2852.951521] buffer_io_error: 1670 callbacks suppressed
[ 2852.951525] Buffer I/O error on dev sr0, logical block 2097136, async page read
[ 2852.951540] sr 1:0:0:0: [sr0] tag#0 unaligned transfer
[ 2852.951544] blk_update_request: I/O error, dev sr0, sector 2097137
[ 2852.951550] Buffer I/O error on dev sr0, logical block 2097137, async page read
[ 2852.951563] sr 1:0:0:0: [sr0] tag#0 unaligned transfer

I should also note that if I insert a CD, it automounts and reads fine.

My questions:

  1. What makes the kernel try to read from /dev/sr0 all the time? Especially on boot, when it's not in /etc/fstab nor mentioned in the grub configuration.
  2. How can I make the kernel give up on this device? From the get-go or after failing a few times?
einpoklum
  • 8,772
  • 19
  • 65
  • 129
  • Saw your posts about that Clevo. You either have a Pentium M 1.6, Pentium M 725, Pentium M 725A, Pentium M 730 or a Pentium M LV 778. The oldest one is from March 12, 2003 while the oldest is from July 2005, which means your Clevo is not a 19-22 years old laptop, but a 18-16 year old's one. And good news, these are all able to run i686 PAE distros, which means you can install a regular Debian on it. The issue with single core CPUs is that putting more than 2GB of RAM is often useless and sometimes the BIOS won't even recognize them. Pentium M are as powerful as Athlon 64 at the same clocks. – X.LINK May 08 '21 at 17:09
  • This means that XFCE will do it (just make sure to disable compositing to use even less RAM), IceWM or the bare xfwm from XFCE if you want to use even less RAM. Do note that LXDE and LXQT uses the same amount of RAM than XFCE. I saw you put Antix on this Clevo, but there's one big catch: You can't upgrade (dist-upgrade) Antix as you would with Debian (e.g Debian 10 to Debian 11). Which means it'll only have 3 years of support/updates instead of 5 years.. and more. – X.LINK May 08 '21 at 17:12
  • @X.LINK: 1. It's a Pentium M 1.6GHz 2. I don't use XFCE... antiX has IceWM by default 3. It happens even during boot. – einpoklum May 08 '21 at 17:13
  • Yup, that's just in case you'd want to switch to Debian considering Antix's issue (see above, I've edited it meanwhile). – X.LINK May 08 '21 at 17:14
  • @X.LINK: You mean Devuan... the thing is, antiX is supposedly friendlier to weaker hardware. And it's Debian-based, so not that much of a difference. Plus, I want to help the antiX people fix these issues. – einpoklum May 08 '21 at 17:15
  • Nope, pure Debian. antiX is friendlier on old hardawre because it uses the RAM friendly IceWM by default, which you can install on Debian too. As I said, the big difference is that you can't `dist-upgrade` antiX (like going from Windows 7 to Windows 10), while Debian can since antiX guys made too much internal changes to allow that. This means 3 years of support/updates only on antiX instead of the 5 years and more you would get on Debian (when `dist-upgrade`) Also, Debian 11 is coming out soon this summer and some software may need PAE (e.g Chrome dropped support for non SSE3 CPUs). – X.LINK May 08 '21 at 17:20
  • 2
    @X.LINK: I can't consider Debian until they stop baking in systemd. But you do make me wonder what modifications antiX has for catering to weaker hardware, other than defaulting to IceWM. – einpoklum May 08 '21 at 17:42
  • Well, Devuan it is then. AntiX has the same modifications than MX Linux since both are made by the same team. You can see the impossibility to upgrade there: https://mxlinux.org/migration/ with "From any version of MX-18 to MX-19" and "From MX-16 or MX-16.1 (or earlier) to MX-17". Since these modifications are shared, it means that they seems to not be related on making them friendlier on weaker hardware. I may not remember it well, but Devuan should have both XFCE and IceWM on the same ISO. – X.LINK May 08 '21 at 17:54
  • What makes you think that the kernel is the problem? – ctrl-alt-delor May 08 '21 at 17:56
  • @ctrl-alt-delor: The fact that I also see this early in the boot sequence, before the init process takes over. But will rephrase the title. – einpoklum May 08 '21 at 18:16
  • How can `lsblk` give a size of 1024M to an empty drive? The driver must think it has a disk in it, and some part of the system is trying to probe for different filesystem types. – meuh May 08 '21 at 18:21
  • @meuh: Good point. But why would the driver think that? Is this a known issue? – einpoklum May 08 '21 at 18:46
  • 1
    FYI, you can install debian [without systemd](https://wiki.debian.org/systemd#Installing_without_systemd). Just don't install stuff like gnome that requires it - Debian's IceWM package doesn't depend on systemd. In short: `apt-get install sysvinit-core` or `apt-get install openrc` after the base system is installed. I still have one sysvinit system on my network (it fails to boot at all with systemd), running `sid`. I gave up on the others because it was too much effort constantly fighting systemd. Fortunately, Debian's systemd packaging disables most of systemd's Borg-like tendencies. – cas May 09 '21 at 03:35
  • 1
    See also http://linuxmafia.com/kb/Debian/openrc-conversion.html. Finally, I've found it useful to run `apt-mark hold sysv-rc sysvinit-core sysvinit-utils` to prevent any upgrades from silently removing sysvinit due to temporary packaging conflicts (sid is volatile). BTW, my non-systemd box is my mythv machine, running xfce, lightdm, and other stuff. – cas May 09 '21 at 03:35
  • 1
    @cas: The dependence on systemd is very deep, like you said, in too much of the distribution. For my desktop machine, I don't want to be stuck to only that. So, Devuan. But for this older, weaker machine - I would consider Debian if I thought antix didn't offer anything beyond defaulting to IceWM. – einpoklum May 09 '21 at 07:18
  • @meuh It sometimes does that. There's even some motherboards that will show you a floppy drive or even some USB drive (not even recovery partitions) when you don't at all and will try to look at it... but nothing in it of course. – X.LINK May 28 '21 at 14:03

0 Answers0