[sane-devel] Scsi problem with Solaris 2.6

Henning Meier-Geinitz henning@meier-geinitz.de
Thu, 26 Jun 2003 10:09:33 +0200


Hi,

On Wed, Jun 25, 2003 at 03:16:35PM +0200, Henrik Schmidt wrote:
> I'm trying to use a Microtek ScanMaker III with Solaris 2.6 and the scg
> SCSI driver. I'm using sane-backends 1.0.12 with frontend 1.0.10.

I'm not sure how old this scanner is but as you can see in some recent
threads on this list some scanners don't like "advanced" SCSI features
like reconnect/disconnect or synchronous negotiation. Maybe it helps
to turn them off in the SCSI driver or adapter BIOS.

> The scanner is detected but as soon as I start a scan or prescan either
> with xscanimage or xsane the application locks up. I get errors in
> /var/adm/messages. After 2 minutes the applications return with a timeout
> and i/o error.

> Jun 25 13:08:25 cortex unix:    data transfer overrun: current esp state:

Hm, maybe the SCSI buffer is too big? Try setting the environment
variable  SANE_SG_BUFFER_SIZE to e.g. 32768.

> I checked the scanner on a Windows computer and it did work. I checked it
> on another computer with sane and it failed too.

Also with Solaris? Same SCSI controller?

At least for Solaris/x86 we have a positive report about a SCSI
scanner working (with SANE 1.0.9).

> I had it running some a ago. When I moved the computer to a different
> location I updated sane and it failed to work afterwards. Going back to
> older sane backend versions (up to 1.0.8) didn't help either.

Then it doesn't look like a SANE problem, does it?

> I tried to contact both the microtek backend coder as well as the contact
> in README.solaris but got no reply.

Matthew Marjanovic posted some days ago on this list. Maybe you can
try it again?

Bye,
  Henning