[sane-devel] VIA chipset and HP4100C USB scanner

Bill Moseley moseley@hank.org
Thu, 1 Jul 2004 17:59:48 -0700


On Tue, Jun 29, 2004 at 07:34:08PM -0400, m. allan noah wrote:
> when you run scanimage with no -d argument, sane will load all the 
> backends that are listed in the dll.conf file.

Ok.

  bumby:~# cat /etc/sane.d/dll.conf
  hp

> as to the timeouts, that could have any number of causes, but seems really 
> common with 2.6.x can you switch back to a 2.4 kernel just to test?

Yes, using 2.4 seems more reliable.

    bumby:~# uname -a
    Linux bumby 2.4.21-xfs-athlon #1 Mon Jul 28 21:36:07 PDT 2003 i686 GNU/Linux

    bumby:~# fgrep usb /proc/filesystems 
    nodev   usbdevfs
    nodev   usbfs

Under 2.4.21 it loads scanner.o.

    Jul  1 16:50:13 bumby kernel: hub.c: new USB device 00:11.2-2, assigned address 3
    Jul  1 16:50:13 bumby kernel: usb.c: USB device 3 (vend/prod 0x3f0/0x101) is not claimed by any active driver.
    Jul  1 16:50:16 bumby kernel: usb.c: registered new driver usbscanner
    Jul  1 16:50:16 bumby kernel: scanner.c: USB scanner device (0x03f0/0x0101) now attached to scanner0
    Jul  1 16:50:16 bumby kernel: scanner.c: 0.4.12:USB Scanner Driver
    Jul  1 16:50:16 bumby usb.agent[1011]:      scanner: loaded sucessfully
    Jul  1 16:50:16 bumby usb.agent[1011]:      libusbscanner: loaded sucessfully

And, yes this works:

    bumby:~# scanimage --mode Color > out.pnm

So under 2.4.21 I'm using the scanner module, but under 2.6.6 I'm using
libusb, correct?  So does that mean the problem is with libusb?  Sorry,
I don't really understand how the parts fit together.

In 2.4 there's direct access via scanner.o, but in 2.6 libusb provides
access via usbdev, right?

By the way, once in a while I'm still seeing this under 2.4 kernel:

   Jul  1 16:58:44 bumby kernel: usb_control/bulk_msg: timeout

which I guess doesn't tell you much.

So, in short, it's not my scanner it's just a problem with my
motherboard's use of VIA chipset.  Any suggestions on what software to
watch for updates that might address this problem sometime in the
future?

Thanks,

-- 
Bill Moseley
moseley@hank.org