This patch modifies the behavior of the EHCI driver in an unlink path
that seems to be causing various issues on some systems.  Those problems
have included issues with disconnection, driver unbinding, and similar
cases where urb unlinking would just not work right.

Basically it just avoids turning off processing of the async (control/bulk)
schedule until it's no longer expecting an "async advance" IRQ, which comes
from the processing passing the ring head.  Whether the driver attempts to
do such things is dependent on various system timings, including both
hardware (problems are seen most consistently on VIA EHCI) and software.

EXPERIMENTAL ... please report results to linux-usb-devel.

--- g26.orig/drivers/usb/host/ehci-q.c	2006-01-05 17:35:38.000000000 -0800
+++ g26/drivers/usb/host/ehci-q.c	2006-01-05 21:10:11.000000000 -0800
@@ -1015,12 +1015,14 @@ static void start_unlink_async (struct e
 	/* stop async schedule right now? */
 	if (unlikely (qh == ehci->async)) {
 		/* can't get here without STS_ASS set */
-		if (ehci_to_hcd(ehci)->state != HC_STATE_HALT) {
+		if (ehci_to_hcd(ehci)->state != HC_STATE_HALT
+				&& !ehci->reclaim) {
+			/* ... and CMD_IAAD clear */
 			writel (cmd & ~CMD_ASE, &ehci->regs->command);
 			wmb ();
 			// handshake later, if we need to
+			timer_action_done (ehci, TIMER_ASYNC_OFF);
 		}
-		timer_action_done (ehci, TIMER_ASYNC_OFF);
 		return;
 	}