[sane-devel] 44-10 vs epson 1250

Till Kamppeter till.kamppeter@gmx.net
Fri, 18 Oct 2002 18:14:24 +0200


Thank you for the patch, I have forwarded it to the kernel maintainers 
here at MandrakeSoft.

    Till


Henning Meier-Geinitz wrote:
> Hi,
> 
> On Fri, Oct 18, 2002 at 05:25:44PM +0200, Till Kamppeter wrote:
> 
>>When I have only an Epson Perfection 1260 connected ("plustek" driver) 
>>the "snapscan" backend takes around 15 seconds, with "snapscan" 
>>commented out xsnae needs less than 10 seconds to start, with "snapscan" 
>>around 25. "mustek-usb" is on the second place with something like two 
>>seconds.
> 
> 
> The USB detection funtion in sanei_usb isn't very effective.
> For every sanei_usb_attach_matching_devices it checks all the
> /dev/usb/scanner devices, opens them, sends the vendor/product id
> ioctls and closes them. That's done for every scanner in the backend.
> For a scanimage -L, about 1300 files are opened.
> 
> However, the real culprit seems to be the USB scanner driver:
> 
> Times for scanimage -L on my system with to Mustek USB scanners
> attached:
> 
> * without scanner driver (with libusb):            0.437s
> * with the scanner driver from Linux 2.4.20pre10: 19.385s
> 
> Now I'm using a patched kernel scanner driver that disables the
> warning messages sent to syslog, fixes the "unable-to-rmmod" bug and
> adds support for devices without bulk-out-endpoints:
> 
> * 0.437s
> 
> That's not a joke. I don't know which of the changes makes the
> difference, but it works (and the annoying messages in syslog are
> gone). The patch is from Sergey Vlasov and was already submitted to
> the maintainer of the USB scanner module and the linux-usb-devel
> mailinglist. They don't seem to be very interested until now, however.
> 
> Bye,
>   Henning
> 
> 
> 
>