Debian-Stable:Modules are not loaded randomly & hdparm:set-dma error
Dear Debian-Users!
I have started this as a bug report (Bug#426432) but I think that it actually
deserves a broader discussion. To begin by stating the problem: I am using
Debian/stable with a 2.6.18-kernel (self-made) and the offical
2.6.18-4-686-kernel-image (i.e. you may assume that all relevant packages are
up-to-date versions of this branch). During booting I get randomly error
messages concerning the loading of modules, i.e. sometimes they are loaded
without any error messages, sometimes not. In my opinion, this corresponds in
some way to Bug#333522 (udev: modules randomly aren't loaded at startup). In
addition to that, hdparm complains about setting dma on my disk (i.e. I get
error messages of the kind "hda: status error: status=0x58 { DriveReady
SeekComplete DataRequest }").
To discuss things in more detail, let me start with the latter. Evidently,
this might have a lot of reasons which are related to a faulty hardware. But
up to now, I have checked my disk with smartctl, and I have changed the IDE
cable as well as the kernel -with no effect. Though this leaves the
possibility of a damaged IDE controller, I think that it has to do with the
booting process. After all, I am able to set the same parameters (after
booting) without any problems. It is also interesting to note that these
problems just right occurred after I had to repair some disk problems (bad
blocks, etc.) (successfully), which affected my old kernel (and modules of it
respectively).
The second problem (random problems concerning the loading of modules) has
been from the very beginning of my Etch-installation. But they have gotten
enhanced since the already-mentioned disk repair, i.e. they show up more
often. Affected modules are ide_cd, uhci_hcd, usbcore, usb_storage, 8139too,
lp (if I would try to related it to my disk repair, then it is probably more
interesting where those modules actually reside on the disk). Worthwile to
mention is that those errors do not occur if booting the offical
kernel-image. The problem is not solved by installing or compiling a new
kernel.
So much for this. Hopefully things might get clearer by looking at the
relevant logs. For this reason, I have provided this link [1], which is
unfortunately limited to 100 downloads & 7 days (you may distribute the file
as needed). Boot and boot.0 show the results of booting the SAME kernel (one
time with errors and one time with no errors). kern.log and syslog contain
the relevant entries from these processes.
In this spirit
With thanks for your efforts
Yours Gernot
---
[1] http://download.yousendit.com/6ABD24B71044D3EC
http://download.yousendit.com/C81C745658EEC5C7 (mirror)
Reply to: