Support and General Use > Recording
Recording Enhancements Pack
Mmmm:
I've been doing some testing with the PCM buffer display and with svn there is a slight 'twitch' every 5 seconds or so to 8000 and then fairly quickly back to 0....
With the REP installed this happens every 0.5 seconds or so....
I have been experimenting with yields in various places and I now have it so that there isn't even the suggestion of a twitch at all, and that's with REP installed! So better than SVN it seems!
Only testing will tell if this fixes the problem, so any testers out there please give the build in my sig a go (H120/140)... I'll do an H3xx build later (got to rush off to dinner now!) :)
EDIT: H3xx build HERE
jhMikeS:
Well, peturs lastest bringing of something to my attention has shown that some atomicity issues in the scheduler must be fixed and fast. The cliplight may very well increase the likelyhood of a thread to get orphaned or not scheduled properly due to posts from IRQ handlers -- increased backlight activity and all that. The problem will be worse on targets with fading in combination with cliplight. It caused thread orphaning not just in recording but playback and other places. It's not likely, but a large number of copies of rockbox running will make it a situation someone is sure to encounter regularly. A change I made earlier made it less likely but hardly impossible...time for a thorough scheduler scan and bugfix.
The report petur pointed me to is a dead ringer for such an occurrence. The timing changes from this patch are probably relevant to the race condition.
Mmmm:
I've updated the patch in post#1 to latest SVN and put in the yields that improve the buffer usage. So hopefully that and JSMikes updates mean the problems that have been seen recently are no more....
jhMikeS:
I should have the proper code guarding in the scheduler/queues in SVN today or tomorrow. I think it'll be a red line on the delta table but it's a decent solution with more detailed panicf as well. Really, I want to take my core lock patch and commit it sans locking but I fear it still might bork the PP5020 targets since all the IRQ fiddling seem to be what keeps them running. Without the locking mechanism in place it will basically be a rework to keep IRQs completely away from the scheduler which is really the _right_ thing imo.
DeepBlueEditor:
I have asked this in another forum but didn't get a good answer. Here's the question.
I am looking to record some short films and am looking at using my H320 to record field audio. Is there a "pretty" and more useful recording interface? Something with much larger text and metering that anyone has come up with? I was pointed to this branch of the forums by a few others who thought you guys would know best.
The recording in 44.2KHz audio isn't the best answer for video (it's 48KHz) but it will do with conversion so that's all cool and since it's hard coded, it'll do. As I am getting older, the tiny little meters built in to the current build RockBox are just too tiny to be usable for a real audio operator under stressful situations on a film shoot.
Anyone have an easy to install option? I would do the programming for it myself if I had the time but I simply don't. IF anyone has ideas or even better, screen shots of various record screens, that would help a lot. Otherwise I'll breakdown and buy an H4 or the new H2 for these projects and keep the H320 as a media player. I love it for that.
Thanks folks,
Sean McHenry
Sean@DeepBlueEdit.com
http://www.DeepBlueEdit.com
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version