Laurent Pinchart wrote:
> Hi Alan,
>
> On Thursday 07 August 2008, Alan Jenkins wrote:
>
>> Oliver Neukum wrote:
>>
>>> Am Mittwoch 06 August 2008 17:23:43 schrieb Alan Jenkins:
>>>
>>>> [ 290.127788] uvcvideo 1-8:1.0: no reset_resume for driver uvcvideo?
>>>>
>>> [..]
>>>
>>>
>>>> [ 290.128909] [<c023465e>] usb_rebind_intf+0x27/0x4e
>>>> [ 290.128916] [<c02346eb>] do_unbind_rebind+0x4d/0x5a
>>>> [ 290.128924] [<c0234738>] usb_external_resume_device+0x40/0x51
>>>> [ 290.128933] [<c020cd87>] device_resume+0x8a/0x2d2
>>>> [ 290.128941] [<c01334e2>] suspend_devices_and_enter+0x10f/0x138
>>>> [ 290.128954] [<c013363a>] enter_state+0x10b/0x164
>>>> [ 290.128963] [<c0133721>] state_store+0x8e/0xa4
>>>> [ 290.128972] [<c0133693>] state_store+0x0/0xa4
>>>> [ 290.128981] [<c01a3087>] kobj_attr_store+0x18/0x1c
>>>> [ 290.128991] [<c018bedc>] sysfs_write_file+0xaf/0xdc
>>>> [ 290.129000] [<c018be2d>] sysfs_write_file+0x0/0xdc
>>>> [ 290.129007] [<c015ec41>] vfs_write+0x83/0xf6
>>>> [ 290.129016] [<c015efd2>] sys_write+0x3c/0x63
>>>> [ 290.129024] [<c01035c1>] sysenter_do_call+0x12/0x21
>>>> [ 290.129035] [<c02b0000>] wait_for_common+0xe0/0xe3
>>>>
>>> if (reset_resume) {
>>> if (driver->reset_resume) {
>>> status = driver->reset_resume(intf);
>>> if (status)
>>> dev_err(&intf->dev, "%s error %d\n",
>>> "reset_resume", status);
>>> } else {
>>> intf->needs_binding = 1;
>>>
>>> It doesn't have reset_resume, so it is marked for rebinding. That in turn
>>> triggers the warning because the parent is still asleep. That shouldn't
>>> happen if we walk the device tree properly. Can you add a printk
>>> to device_pm_add() to print out the parent? We need to rule out that the
>>> device tree is bad.
>>>
>>> Regards
>>> Oliver
>>>
>> Done. (For the record, I've posted the results at
>> <http://bugzilla.kernel.org/show_bug.cgi?id=11263>).
>>
>
> I've added a reset_resume method to the UVC driver. Could you please test the
> latest SVN revision and tell me if it helps ?
>
Alan Stern said the problem was that USB needed to move to the new PM
ops. He posted a patch for this which fixed my problem (the scary
WARNING with call-trace).
So I'll try to test SVN but it's not very interesting :-).
Ta
Alan
_______________________________________________
Linux-uvc-devel mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/linux-uvc-devel
