Resume broken in 8.3-PRERELEASE

[Alexey Dokuchaev]

On Fri, Mar 02, 2012 at 04:14:08PM +0100, Hans Petter Selasky wrote:
> On Friday 02 March 2012 09:57:03 Alexey Dokuchaev wrote:
> > On Fri, Mar 02, 2012 at 08:48:13AM +0100, Hans Petter Selasky wrote:
> > > If the reset doesn't work, then try to enable hw.usb.uhub.debug=15 and
> > > see what port change events are coming.
> > 
> > I don't see such oid hw.usb.uhub.debug.  Perhaps it should be hw.usb.debug?

What about this?  I've did hw.usb.debug=15, but didn't see anything new on
the console.

> > > If cfg=255 in usbconfig, then something is wrong.
> > 
> > It seems they are all zeros, per the output I've sent earlier.
> 
> If they are all zeros, then everything is working like it should. If you can 
> dump the device and configuration descriptors, then there is no USB problem.

I can only dump this information *before* suspend, after "resume" I cannot
do almost anything except breaking into debugger (but not being able to
"call doadump").  I had to revert to earlier revision to call usbconfig(8)
after resume.

> I'm thinking it might be some MICROCODE issue that causes the problem. Maybe 
> we shouldn't reset the OHCI and EHCI and UHCI and XHCI before handing over to 
> the MICROCODE suspend handler?
> 
> Have a look in /sys/dev/usb/controller/xhci/ehci/ohci/uhci and search for 
> suspend.

OK, I will try and report back, thanks.

./danfe