On Thu, 18 Mar 2004, Alan Stern wrote:
> > Hotplug would be the only interesting one running at the same time.
>
> Of course! Just put the clues together, Watson!
>
> 1. Problem is caused by control messages, but strace shows iscan
> only uses bulk messages.
>
> 2. You observe an excessive amount of hotplug activity.
I call it excessive because I wonder if it's correct to feed a hotplug
event just because $USER_SPACE_PROGRAM has decided to open a
/proc/bus/usb/MMM/NNN device.
> 3. No such problems exist under 2.4.
Well, 2.4 is not 100% stable, I've had it lock up with vuescan (the one
that works for 2.6 most of the time) three hours ago as well.
> Conclusion: the hotplug programs are submitting the control requests and
> confusing the scanner.
Might be.
> What version of the hotplug package do you have installed?
2002_08_26-22 (as shipped with SuSE Linux 8.2)
To answer further questions, I changed /etc/hotplug/usb.usermap adding
this line:
epson_scanner 0x0003 0x04b8 0x0110 0x0000 0x0000 0x00 0x00 0x00 0x00 0x00 0x00
0x00000000
And /etc/hotplug/usb/epson_scanner is:
------------------------------------------------------------------------
#! /bin/sh
if [ "$ACTION" = add ] && [ -f "$DEVICE" ] ; then
chown root.trusted "$DEVICE"
chmod 0660 "$DEVICE"
fi
------------------------------------------------------------------------
(Yes I know that [(1) supports the -a option -- it's just an old habit
of using && and separate [(1) commands instead.)
This works with 2.6 but not 2.4 (it never gets executed under 2.4, but I
run chmod 0666 /proc/bus/usb/003/002 manually on 2.4 before trying to
scan).
> I'll have to look more deeply to see why it's starting up so many times
> and why it's using timeouts that are way too short. In the meantime, you
> could simply try disabling hotplug entirely. See if that doesn't fix the
> problem.
I'll try replacing /sbin/hotplug by /bin/true and also updating the
hotplug package.
> > > Attached are versions of the patch for both 2.6 and 2.4. It will be
> > > interesting to compare the results.
> >
> > I wonder if it will really be interesting.
> >
> > Linux 2.4 produces no more proc_control: logs after starting iscan, but
> > I have successfully done a preview and then a full scan in 400 dpi of an
> > ISO-A4 sheet - this should amount to c. 46 MB data through USB 1.1.
> >
> > I'm gzipping the logs as they are 50 kByte each and gain >= 90%.
>
> Clearly there's a difference between when the two kernels generate hotplug
> events. I'll look into that too.
For sure, 2.4 has severely lacking coldplug support, as mentioned above.
Might of course be that a newer hotplug package fixes _that_, I've never
bothered much. sudo chmod 0666 /proc/bus/usb/003/002
happens to be in my shell's history. :)
> In any case, the control requests being submitted by the hotplug program
> are completely unnecessary. They are asking the scanner to transmit data
> which the kernel already knows and makes available through sysfs.
Seems as though scanners were not necessarily multi-tasking devices ;)
--
Matthias Andree
Encrypt your mail: my GnuPG key ID is 0x052E7D95
-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
