[sane-devel] Brother MFC-8840D fails after switch to Linux kernel 2.6.14.2

[Henning Meier-Geinitz]

Hi,

On Wed, Dec 21, 2005 at 05:43:05PM +0100, Stefan Schwarzer wrote:
> this configuration used to work flawlessly:
> 
> Gentoo Linux with kernel 2.6.12.5
> Brother MFC-8840D (MFC-8440 should be similar), connected via USB,
>   with driver brscan from Brother's Linux page
>   http://solutions.brother.com/linux/sol/printer/linux/sane_drivers.html
> sane-backends 1.0.17

As far as I can see the code from Brother is a complete backend. So I
think sane-backends 1.0.17 is not involved at all.

In this case you may have more luck asking Brother about your trouble
becasue only they know what their binary-only libraries do.

> Scanning stopped working when I updated to Linux kernel 2.6.14.2 .
> This is the vanilla kernel, I also tried the Gentoo-patched

If you haven't changed anything else recently this is a good
indication that something is wrong with that kernel :-)

I haven't tested 2.4.14 until now but I also haven't heard of any
general USB trouble with that kernel.

I'll do some tests this afternoon.

> scanimage > test.pnm   results in the message
> scanimage: open of device brother:bus2;dev1 failed: Error during device I/O

I don't know how this works with the brother backend, but turning on
debugging may help to find the cause of the problem.

> dmesg   emits:
> usb 1-2: usbfs: interface 1 claimed by usbfs while 'scanimage' sets config #1

I think this means that some other application has already claimed
your device, i.e. something else is already using your scanner. 

> Maybe this is related to bug #302207 ("problem with usb_set_configuration()
> in sanei/sanei_usb.c"), listed at
> https://alioth.debian.org/tracker/index.php?func=detail&aid=302207&group_id=30186&atid=410366

But in your case it seems that not a kernel driver (like usblp) claims
the interface but a userspace application using libusb. "usbfs", the
USB file system, is the interface used by libusb to talk to the scanner.

> I don't know whether the problem comes from a bug in Sane, Brother's
> backend or the kernel. Any suggestions?

I think you can exclude sane-backends for now.

Bye,
  Henning