[17:53] *** CorePorkRation has quit IRC (Quit: CorePorkRation)
[00:34] <Juan> I've noticed this regression and I wonder if there's anything else that I could report http://article.gmane.org/gmane.linux.drivers.video-input-infrastructure/85583/
[01:23] <pinchartl> Juan: you've reported it to the right mailing list
[01:23] <pinchartl> if nobody responds, just ping the list
[01:23] <pinchartl> and CC the maintainer
[01:24] <pinchartl> in this case that would be Mauro
[01:24] <pinchartl> who is currently mostly offline due to holidays
[01:24] <Juan> I see
[01:24] <pinchartl> until the end of the month
[01:24] <pinchartl> he may read his e-mails
[01:25] <Juan> As far as any information to gather, can you think of anything else I could do?
[01:26] <pinchartl> it would be good if you could explain how the device stopped working
[01:27] <pinchartl> e.g. do you get video but no sound, or the sound is noisy, or the video is all black, ...
[01:27] <pinchartl> it could also be useful to capture the output of
[01:27] <pinchartl> lsusb -v -d eb1a:2861
[01:28] <Juan> It's just sound not working
[01:28] <pinchartl> with the old and new kernel
[01:28] <pinchartl> and compare the two
[01:28] <pinchartl> if there are differences, report them
[01:28] <Juan> Will do that, on a newer kernel though
[01:28] <pinchartl> otherwise you can just report they're the same
[01:28] <pinchartl> I have no experience with the em28xx driver
[01:29] <Juan> Still, you've been very helpful
[01:29] <pinchartl> can you easily modify the kernel, recompile it and test it ?
[01:30] <Juan> I'm open to testing, sure
[01:33] <pinchartl> I can't really help you more than that
[01:34] <pinchartl> Mauro probably can, when he will be back from vacations
[01:34] <Juan> It's cool
[01:35] <Juan> The device itself is really nice for the price
[01:35] <Juan> Which was like 15 usd
[01:36] <pinchartl> time for me to go to bed
[01:36] <Juan> Good night
[01:37] <pinchartl> it's 3:38am here
[01:37] <pinchartl> thank you
[18:06] <RageYL> hi
[18:11] <gnac> where would be the best place to report an issue in the saa7164 firmware for a haupage 2250 card?
[19:20] <RageYL> i do not see any difference (in term of speed or anything else) between having one buffer or multiple. what are the pros/cons from having multiple buffers ?
[19:31] *** smartin__ changes his alias to smartin
[19:34] <pinchartl> gnac: here or on the linux-media mailing list
[19:34] <pinchartl> RageYL: more buffers consume more memory and can introduce delays
[19:34] <pinchartl> but if you don't have enough buffers peak cpu loads can result in lost frames
[19:35] <RageYL> ok thanks =)
[19:35] <pinchartl> in the general case you want at least two buffers
[19:36] <pinchartl> once the first buffer is full it gets passed to userspace, and the device captures the second frame in the second buffer
[19:36] <pinchartl> userspace then processes the first frame in parallel, and queues the buffer back when done
[19:36] <pinchartl> on average the processing time must be lower than the frame time, otherwise you'll lose frames
[19:37] <pinchartl> but the processing time can vary depending on many factors
[19:37] <pinchartl> and can sometimes be higher than the frame time
[19:37] <pinchartl> in that case, if you have two buffers only, the driver won't have any buffer to capture to after the second frame has been captured, as the first buffer is still being processed by userspace
[19:38] <pinchartl> raising the number of buffers can help there
[19:39] <RageYL> in my case i have choose a format which say: Interval: Discrete 0.033s (30.000 fps), so with your explanation if i have the good amount of buffer, i can have 33 picture per seconds
[19:40] <pinchartl> 30 pictures per second
[19:41] <RageYL> yep sorry
[21:45] <gnac> So my tuners stop working and I see tons of errors logged, specifically saa7164_api_i2c_read(), s5h1411_readreg and saa7164_cmd_send() in that order to the tune of 100's of megs a day.
[21:46] <gnac> I've tried a couple of different firmwares from the stevehoth site, but they all seem to exhibit the same behavior after a while (<24 hours).
[21:46] <gnac> rebooting clears the error for a while.
[02:16] *** debris'_ changes his alias to debris'
[02:29] *** _jmleo changes his alias to jmleo
[02:49] *** _jmleo changes his alias to jmleo
[04:57] *** hverkuil_ changes his alias to hverkuil