On Saturday 30 August 2003 20:35, Alan Stern wrote:
> On Sat, 30 Aug 2003, Duncan Sands wrote:
> > Long ago I posted a bug fix for a VIA add on uhci card (plugging in a new
> > device disconnects existing devices)  The workaround is to suspend the
> > hc before doing the PORT_RESET for the new device, and wake it up
> > afterwards.
> >
> > However this is problematic in 2.5/6: if you suspend the hc, it
> > immediately wakes up (need for wakeup detected in uhci_irq).  So I need a
> > way to forceably suspend the hc, so that it only wakes up when I say so. 
> > This certainly means complicating the existing code in uhci-hcd.  Any
> > comments and suggestions for the best way to do this before I start
> > hacking?
>
> This turns out to be a very good time to ask that question, as I am just
> about to start fixing up the suspend/resume code in uhci-hcd.  I'll put in
> a hook for forceably suspending and sent it to you when it's ready.  Is
> that okay?

Sounds great, thanks Alan.  I guess it needs to deal with the case that
the forced suspend comes while the hc is already suspended; then the
corresponding wake up shouldn't really wake it up.  Also, multiple forced
suspends should nest properly.  Sounds like a counter is needed :)
However, for my workaround this isn't really needed: even if someone
plugs in two devices at exactly the same time, the connects are
dealt with serially.

Thanks again,

Duncan.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to