Alan Stern replied:
>>
>> >If it works initially, before you unplug the device, then interrupts
are
>> >enabled.
>>
>> But...
>> I put in a printk() at the start of the interrupt handler: uhci_irq()
>> and it _never_ gets called!... ever!
>
>Well, that's bad. Didn't you say before that when the device was
>coldplugged it would work okay? If there were no interrupts the device
>couldn't have been working.
Unfortunately...
Yes, the device _does_ work when cold plugged, even though it _never_
got to the uhci_irq() printk statement!
Thats why I emphasized the fact.
>> But usb_disconnect() does.
>> How can you get to usb_disconnect without having gone through uhci_irq?
>> (wishing I had a print_stack_trace() function.)
>
>There is such a function (although it might not help you here); it's
>called dump_stack().
Great, thanks. It'll come in handy.
>usb_disconnect() is called when the hub driver notices that the connect
>status for one of the ports has changed. No interrupts are required, but
>the connection status bits must be valid, obviously.
>
>On the other hand, usb_disconnect() won't get called unless a
>usb_new_device() was called first, and that won't happen without a valid
>interrupt.
Ok, I'll add the debug statements and report back (tomorrow).
>> For what its worth:
>> Looking at /proc/interrupts on 2.6.6 (transcribed and edited):
>>
>> 10: uhci_hcd, [EMAIL PROTECTED]:1:0:0
>> 11: uhci_hcd, uhci_hcd, eth0, yenta, Intel 82801DB-ICH4
>>
>> Whereas on RH9 it says:
>>
>> 10: usb_uhci, [EMAIL PROTECTED]:1:0:0
>> 11: ehci_hcd, usb_uhci, usb_uhci, ohci1394, eth0,
>> PCI device 1524:1410 (Ene Technology Inc)
>
>I don't know much about this, but there are kernel boot options for ACPI
>that will control interrupt allocation.
>
>Also, you could simply try removing the drivers for all the devices that
>share an interrupt with UHCI. Not that I think it would make any
>difference, but you never know...
I'm not too sure how to do that for the video card and the ICH4
ie. what to specify on "modprobe -r".
And what about the fact that the uhci _is in_ the ICH4?
The rest, eth0 and yenta, I can live without during the test.
-------------------------------------------------------
This SF.Net email is sponsored by Sleepycat Software
Learn developer strategies Cisco, Motorola, Ericsson & Lucent use to
deliver higher performing products faster, at low TCO.
http://www.sleepycat.com/telcomwpreg.php?From=osdnemail3
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
