I've had similar behavior on EHCI/UHCI equiped i915 based boards. It's
like sometimes interrupts were never caught or fired in GENERIC. I
switched to GENERIC.MPACPI and that seemed to improve matters, though I
haven't been testing it much recently though. In my USB debug kernel,
there was zero activity from the USB side of things when I would plug in a
device at the highest debug level, but, plug something into another random
port, and then all the sudden everything shows up. It's weird.


On Tue, 29 Mar 2005, George Michaelson wrote:

>
> I have an IBM X31 with CF slot, and have found from time to time my cards
> get rejected as 'bogus CIS' with a long timeout. They insert, they detect
> to some level, then nothing happens. no wd<x> is detected. When you eject
> you get the count in seconds since insertion as a debug message.
>
> If I do an insert-remove on a USB device at the same time, I can often wake
> these devices up. Its like some other timer check on the pending requests
> walks through the queue, and sweeps it up.
>
> I noticed this with two USB disks, one an iPod Mini, which would go into
> its bulk update state, but not detect a drive reliably (sometimes scsictl
> can wake it up) unless another device is plug/unplugged when its there.
>
> Odd. Anybody else seen this?
>
> -George
>

-- 
The nice thing about Windows is - It does not just crash, it displays a
dialog box and lets you press 'OK' first.
						-- Arno Schaefer