Re: [linux-usb-devel] Re: serious 2.6 bug in USB subsystem?

[David Brownell <david-b@pacbell.net>]

David Mosberger wrote:

> Here is patch #3.  It also Works For Me.  I was wondering whether it

I've had several "Works For Me" patches too, but then if the
silicion got kicked a bit differently it'd not behave... :(


> it is really safe to mess with the OHCI control registers the way
> ed_deschedule() does at a time the OHCI is running.  To test this

It must be, or we'd not have had a driver working for several
years now!

The quick stop/restart cycles haven't seemed to be a problem
with any OHCI silicion in the way they are with, for example,
VIA EHCI.


> theory, I delayed the ed_deschedule() handling to finish_unlinks(), as
> shown in the patch below.  I don't know whether this is really safe as
> far as the host's lists are concerned, but it does avoid the crashes.

My suspicions have been focussing on finish_unlinks().

That's really the only place the HCD does anything
that could corrupt the ED queues, which is what looks
to be happening.

Your change doesn't actually _unlink_ in the same way;
interesting change, I'll have to think about it.  It
certainly changes timings.


> What's the argument as to why it's safe to update the OHCI control
> registers in ed_deschedule() at the time start_ed_unlink() is running?

It's always safe to update those registers, except
that some silicon doesn't support that while the
controller is suspended.

- Dave