Re: PATCH (as112) Re: USB APM suspend

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

Alan Stern wrote:
> On Sun, 21 Sep 2003, David Brownell wrote:
> 
>>>Why was this routine called twice?  (Don't be fooled by the timestamps; I 
>>>think the "suspend D4 --> D3" message was created during the suspend but 
>>>not read by syslogd until after the resume.)
>>
>>That's happened for as long as I remember (2.4 also).
>>Still seems buglike to me, maybe 2.6 will finally squish it...
> 
> 
> Well, the code path is easy enough to find.  If you look at suspend() in
> arch/i386/kernel/apm.c, you'll see calls to pm_send_all() and
> device_suspend().  They both end up filtering down to the USB HC drivers.  
> The bad one is pm_send_all(); it comes too soon.

Rather, "it comes at all".  Call device_{suspend,resume} should
suffice.  It shouldn't pm_send_all() -- either of the two calls.
(The 2.4 bug is necessarily a different issue.)

Does it work if you remove those calls?


> By the way, David, apparently core/hcd-pci.c wants the HC drivers to set 
> the hcd state to USB_STATE_SUSPENDED, but a simple grep shows that neither 
> the EHCI nor the OHCI driver does so.  That certainly looks like an 
> oversight, though I'm not sure in which source file.

And what's odd is that it was working before, too!  I'll have a look,
next I get a chance.  It's good to know that APM is almost behaving again.


> Meanwhile, here's a simple patch to improve logging during suspend and
> resume.  Greg, if David approves please apply it.

Reads OK to me -- go for it!

- Dave