[sane-devel] problem with ADF on HP 6300C scanner

[Eugene Reimer]

I've just bought an HP 6300C scanner with ADF, and am trying to use 
scanimage to do multi-page scanning using the document-feeder.  It feeds 
and scans the first page successfully, and then starts to feed the 2nd 
page, when everything stops, with the messages:
    scanimage: sane_read: Error during device I/O
    Scanned page 2. (scanner status = 9)

My software versions:
scanimage (sane-backends) 1.0.17; backend version 1.0.17
Linux 2.6.11-1.1369_FC4

Here is the last part of the output:

[dll] 
sane_read(handle=0x927fdf8,data=0x928c498,maxlen=32768,lenp=0xbfdd586c)
[dll] 
sane_read(handle=0x927fdf8,data=0x928c498,maxlen=32768,lenp=0xbfdd586c)
[dll] sane_cancel(handle=0x927fdf8)
scanimage: read 1339481 bytes in total
Scanned page 1. (scanner status = 5)
Scanning page 2
[dll] sane_start(handle=0x927fdf8)
[dll] sane_get_parameters(handle=0x927fdf8,params=0xbfdd583c)
scanimage: scanning image of size 2550x4199 pixels at 1 bits/pixel
scanimage: acquiring gray frame
[dll] 
sane_read(handle=0x927fdf8,data=0x92944a0,maxlen=32768,lenp=0xbfdd586c)
scanimage: sane_read: Error during device I/O
Scanned page 2. (scanner status = 9)
[dll] sane_close(handle=0x927fdf8)
[dll] sane_exit: exiting
[dll] sane_exit: calling backend `hp's exit function
[dll] sane_exit: finished

from running the following:
export SANE_DEBUG_DLL=128
export PTAL_DEBUG=2
scanimage -v -d hp:libusb:001:006 --batch --source=ADF

I've tried many variations of scanimage commands, including using 
separate single-page scanning commands with both scanimage and xsane, 
but always get the same result, that it fails when starting the 2nd page. 

AHA, the problem is not nearly as bad as I thought, it was just my 
thinking that TWO pages was the right number for testing:-)  When I try 
it with 3 or more pages, then it only fails when starting the last page, 
and occasionally it even handles that one and ends gracefully.  So now 
I'm thinking that using this scanner together with SANE and scanimage 
may well be feasible after all.  I can simply add a blank page to the 
end of every stack I want scanned...

Can anyone tell whether this behaviour indicates my scanner is defective 
in its "document-feeder empty" detection, or is it due to a flaw in the 
sequence of operations the sane-hp-backend is using?