-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Am 18.12.2011 20:37, schrieb Laurent Pinchart:
Hi Alexey,
Thanks for the patch, and sorry for the late reply.
On Monday 21 November 2011 15:51:47 Alexey Fisher wrote:
Hi Laurent,
here is my version of bug_report file for debugfs. It include
usb.id, dmi.system.product (for upsidedown report). And stream
specific information. I think it is good size for initial patch.
I near future i would like to add control call count and control
error rate to stats. And probably firs urb error number to report
file. I think it will be enough to provide needed info for
troubleshooting.
I got a couple of comments regarding both the purpose of the patch
and its implementation. I'd like to discuss the purpose first.
My understanding is that this new debugfs entry is meant to help
diagnosing and debugging issues. This is a nice thing, but most of
the debugging information (such as the USB VID:PID and the DMI
information) can already be obtained using lsusb and dmidecode.
Duplicating in debugfs information that is already available
through other means isn't something I'm very fond of. Sure,
dmidecode needs to be run as root, but if the user can't figure
that out I don't think getting him to moutn debugfs will be a great
success either :-)
Ok, i agree.
The i suggest to add two updates for FAQ.
1)
sudo echo 0x /sys/module/uvcvideo/parameters/trace
and
sudo echo 0 /sys/module/uvcvideo/parameters/trace
should be probably corrected to:
sudo sh -c echo 0x /sys/module/uvcvideo/parameters/trace
sudo sh -c echo 0 /sys/module/uvcvideo/parameters/trace
at least on ubuntu it is imposable to execute the previous version.
2) add in case of upside down webcam.
sudo dmidecode -t system -t baseboard
after some more time in this list i got some reasonable amount of
skepticism. so i can agree to remove many of my previous suggestions :D
Here i my current thoughts:
- - Looks like bandwidth keeps to be a major problem. Last month there
was cases with: lowspeed cam; high speed on low speed bus(?); advanced
cases with more then one cam using jpeg compression (usually buggy
cams)...
i also noticed that people (including me) wont to know video format
currently used by cam. So probably packet size, and frame format
should be included in to debugfs report.
- - other problem i noticed is: looks like uvcvideo collides some time
with snd-usb-audio. For example if audio use lots of ctrl messages and
get time out because of video stream, it resets usb device and kill
both audio and video.
Other example from one of last mails:
=
uvcvideo: uvc_set_video_ctrl: PROBE; bmHint: 1; bFormatIndex: 1;
bFrameIndex: 8; dwFrameInterval: 33; wKeyFrameRate: 0;
wCompQuality: 0; wCompWindowSize: 0; wDelay: 0; dwMaxVideoFrameSize 0;
dwMaxPayloadTransferSize: 0
5:3:1: cannot get freq at ep 0x82
uvcvideo: Failed to set UVC probe control : -32 (exp. 26).
=
if i interpret it correctly: video and audio send ctrl at same time
and fail. -32 mean wrong pipe. am i correct? stand alone each works fine.
for this case i have no idea for debugfs report. probably dmesg is the
best way to do it.
- --
Regards,
Alexey
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk7ucGMACgkQw8E0jNwoJm/SGgCcCeorwiDQCaWQ+x4gV1S3wfS2
XjsAoKwFbcuHaggi3dL2wdlAcwrQ647m
=3sIb
-END PGP SIGNATURE-
___
Linux-uvc-devel mailing list
Linux-uvc-devel@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/linux-uvc-devel