Re: ov511 verbose startup.

[Mark McClelland <mark@alpha.dyndns.org>]

Greg KH wrote:

>On Thu, Jan 31, 2002 at 02:34:57AM +0100, Dave Jones wrote:
>
>>The changes to ov511 in 2.5.3 seem to generate excessive
>>amounts of blurb on boot up for me..
>>
>>ov511.c: USB OV511+ camera found
>>ov511.c: model: Creative Labs WebCam 3
>>ov511.c: Sensor is an OV7620
>>ov511.c: Device registered on minor 0
>>usbdevfs: USBDEVFS_CONTROL failed dev 2 rqt 128 rq 6 len 137 ret -75
>>usb_control/bulk_msg: timeout
>>usbdevfs: USBDEVFS_CONTROL failed dev 2 rqt 128 rq 6 len 18 ret -110
>>usb_control/bulk_msg: timeout
>>usbdevfs: USBDEVFS_CONTROL failed dev 2 rqt 128 rq 6 len 18 ret -110
>>...
>>repeat last two lines another dozen or so times...
>>
>
>What userspace program are you using that is talking to the usb device
>through usbfs?  Or is this usbutils trying to determine what driver to
>load?
>
I have been getting reports of -75 (babble?) and -110 errors with both 
control and iso transfers. The problematic kernels seem to be 2.4.17+, 
with uhci HCD. IIRC, nearly all of the reports mentioned a Via chipset. 
There are sporadic reports of corrupted iso packets (blocks of zeros 
inserted randomly) with uhci under 2.4.17 as well.

This is the first case of a usbfs-related ov511 error that I have seen. 
Very strange.

Greg (if you know): usbfs is not allowed to access claimed interfaces, 
correct? (ie. ones that are implicitly claimed because of a successful 
return from probe()). Are interfaces treated as claimed while probe() is 
active, so that user-space "probes" cannot interfere with driver probes()?

I use usb-uhci, uhci, and usb-ohci regularly with ov511, and I have 
never seen any of these problems. There are clearly a number of factors 
involved here.

-- 
Mark McClelland
mmcclell@bigfoot.com