On Mon, 21 Jun 2004, Zan Lynx wrote:

> I have attached copies of /proc/driver/uhci/0000:00:13.1
> 
> I can't tell you about the interrupts because other devices on IRQ 11
> are also interrupting.  There isn't much I can do about the SCSI
> controller.  I need that. :-)
> 
> If you think it's important, I could try moving the SCSI controller
> around to see if I can get a different IRQ.

It might be important.  The "before" and "after" copies were
uninteresting; they looked totally normal.  But the "during" copy was
different.  It showed that the USB controller had finished receiving a
data transfer from the mouse, but it apparently had not issued an
interrupt request (because the transfer structures had not been removed
from the active-lists) and it was not trying to issue one.

Another difference was that the "during" copy showed the transfer had the
SPD (Short Packet Detect) flag set.  In principle that should not have
caused this sort of problem.  In fact, the UHCI specification says that
when SPD is set and a short packet occurs (this was a short packet; the
request was for 8 bytes and the mouse only sent 6) then an interrupt
request is issued and the queue header is not updated.  But it was
updated, and an interrupt wasn't requested.

The most likely possibility is that your controller didn't issue an 
interrupt request when it was supposed to.  In other words, a hardware 
failure.  That's my best guess for now.  It doesn't explain why things 
appear to work okay under 2.6.5, though, or why SPD was set.  It's 
possible the two things are related; one of the things that changed 
between 2.6.5 and 2.6.7 was the handling of the SPD flag.

The best way to confirm that the hardware is at fault is to try running 
the same software on another computer with the same type of USB 
controller.  Maybe you can figure out some way to do that.

It also would be nice to know why SPD was set.  That's controlled entirely 
by software.  In this case, the mouse driver or HID driver would have had 
to ask for that flag.  I don't know any reason why they would, however, 
nor do I see any place in the code where they do.

Alan Stern



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-users

Reply via email to