> [Anyone have any clues/suggestions as to what is going on?
> The hardware involved is a Microtek 600GS scanner, with the classic
> Adaptec AVA-1502 scsi card, Linux 2.1.107.
> The weird part: scanimage works as non-root, but hangs as root.]
>
> > Date: Wed, 01 Jul 1998 18:06:57 +0200
> > From: Oliver Neukum <neukum@fachschaft.org.chemie.uni-muenchen.de>
>
> > Hi Matto,
> > i fear i have more bad news for you.
> > The scanner's firmware has a bug, it crashes when a lun>0 is referred to.
>
> Not so unexpected....
>
> > I s this done within SANE ? Might it explain hangs as root ?
> > Do you need any more information ?
> > A patch against kernel 2.1.107 is attached.
>
> Isn't there also a kernel configuration option to disable scanning
> all LUNs? That's usually what you want, unless you have a multi-platter
> CDROM that actually uses them... (Either way, I'll put this info somewhere
> on my web page.)
>
> > i can confirm that my inabillity to scan as root is caused by sane touching
> > luns>0.
> > The patch to the kernel, which i sent you yesterday allows scanimage to work.
>
> Can you tell where this happens? The microtek backend shouldn't touch any
> LUN other than 0 --- is this some other backend? (Are the others commented
> out of dll.conf?)
>
> Also, the "BLIST_NOLUN" designation in the patch only affects how the
> mid-level scsi driver scans the scsi bus --- thus your patch should only
> affect what happens a) when machine boots, b) when a scsi driver module
> is loaded. SANE doesn't cause either, unless you have kerneld set up to
> automatically load the scsi modules.
>
> And... I don't see how any part of SANE does anything differently whether
> or not it is used as root. SANE's only interaction with the Linux system
> is to open/read/write/close the sg devices (which is affected by their
> permissions), and open/read/close "/proc/scsi/scsi". Since you can scan
> as non-root, SANE as root would only be touching *other* sg devices
> differently.
>
> Hmm... of course, I'm looking at 2.0.33 kernel code --- maybe the scsi bus
> is rescanned in the 2.1.107 code? Some drastic difference?
>
> -matt m.
>
> --
> Source code, list archive, and docs: http://www.mostang.com/sane/
> To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
Hi Matto,
i can no longer reproduce the root problem unless i add backends to dll.conf.
However those crashes are total; strace information is not written.
How important is this information to you, as your driver at least in patched form is
not to blame ?
Oliver
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com