On Mon, 2002-10-28 at 13:04, Jon Wikne wrote:
> In terms of source code, I can not. If I could, I would probably
> have suggested a fix and posted a patch.
In the discussion during your absence, the conclusion was:
a) The "funky result: -75" messages are causes when the *application*
issues reads which are not *exactly* the same size as the USB
transaction size.
This is a Sane/VueScan bug.
b) When applications break this API restriction by trying to read *more*
than the size of a the next USB transaction that the scanner wants to
write, scanner.c does not handle the resulting EOVERFLOW (75) error and
stays stuck in the read_scanner loop.
This is the reason you see the oopses: you are trying to rmmod a module
that is still executing code.
This is a kernel scanner.c bug.
> >>EIP; fa9404f8 <[scanner]read_scanner+e8/240> <=====
> Trace; c0139b26 <sys_read+96/110>
> Trace; c010a766 <do_IRQ+c6/f0>
> Trace; c0108c3b <system_call+33/38>
> Code; fa9404f8 <[scanner]read_scanner+e8/240>
> 00000000 <_EIP>:
> Code; fa9404f8 <[scanner]read_scanner+e8/240> <=====
> 0: 8b 11 mov (%ecx),%edx <=====
> Code; fa9404fa <[scanner]read_scanner+ea/240>
> 2: c1 e0 0f shl $0xf,%eax
> Code; fa9404fd <[scanner]read_scanner+ed/240>
> 5: c1 e2 08 shl $0x8,%edx
> Code; fa940500 <[scanner]read_scanner+f0/240>
> 8: 09 c2 or %eax,%edx
> Code; fa940502 <[scanner]read_scanner+f2/240>
> a: 89 d0 mov %edx,%eax
> Code; fa940504 <[scanner]read_scanner+f4/240>
> c: 0d 00 00 00 04 or $0x4000000,%eax
> Code; fa940509 <[scanner]read_scanner+f9/240>
> 11: 83 79 04 00 cmpl $0x0,0x4(%ecx)
--
,_
/_) /| /
/ i e t e r / |/ a g e l
http://www.nagel.co.za
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
