I got my Mustek MFS 12000 SP the day before yesterday. Yesterday I played
around in Win 3.1. The scanner worked, so I tried to get the thing working
in Linux, with the original SCSI card. No luck.
Yesterday I tried again. I had applied the mustek-scsi-patch-0.5.gz to
Linux 3.0.30 (I know I should get pl 32, next thing on my list) and
had gotten one reject, for the Makefile in the drivers/scsi directory.
This Makefile has to be edited by hand, as commented also in the patch (I
figured it out myself before I read it well, always RTFM 2x). My card was
the DTC3181E, a thing that I had written down in the Mustek manual. This
is another lesson from practice, write down the funny codes on the chips
and the jumper settings (if any), before you build in the card. The patch,
and the comment in the patch are not quite up to date for the newer Linux
kernels, but it is easy to find the corresponding bit to change, just look
for the GENERIC_NCR5380_OVERRIDE flag.
Once I had done that, the card was recognised at boot time. The scanner
device ends up at /dev/sga (Redhat 4.0), and scanimage can recognize it.
To my surprise, the scanner works faster now than under Win 3.1! The
machine does not lock up, as claimed for the original 0.2 patch. (and I
verified that for Win 3.1 it does lock up -- totally. This was no surprise
however.) Checking the load with top while doing a scan returns 90% idle
time for the CPU.
Now the only problem I have is that the colours get mixed up on larger
scans. From mailing list I guess I am not alone, and that it affects other
SCSI cards too. I will next try the MUSTEK_FLAG_LD_NONE, and you'll hear
of the results.
--- René van Paassen M.M.vanPaassen@LR.TUDelft.nl Faculty of Mechanical Engineering tel. +31 15 278 5419 Kluijverweg 1, NL-2629 HS Delft fax. +31 15 278 6480 Delft University of Technology
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail firstname.lastname@example.org