On 06/12/10 08:42, Andy Walls wrote: > On Fri, 2010-06-11 at 22:03 -0400, Dale Pontius wrote: >> On 06/10/10 23:14, Andy Walls wrote: > >> Thanks for all of the thought, but last night/today/just-now I think I >> shot any theories full of holes. >> >> The weather has been warming today, but last night I shut the computer >> down. Regular shutdown, not full power-off, so 5VSB was still on, which >> is at least significant to the red-screen. This morning, in the cold, I >> powered on - no audio. A little later I rebooted, still no audio. >> >> So much for it being largely about temperature. >> >> It has warmed quite a bit today, and this evening I had a few minutes, >> so I removed all of the modules, then modprobed cx18 and cx18-alsa. I >> have sound, again. >> >> It's just flakey. Maybe when I get a break and take this system down, a >> bit of dust-off and reseating cards will work wonders. > > Well, some things to try: > > 1. remove all instances of the cx18-alsa.ko module from > under /lib/modules. It shouldn't matter, but most users never need it, > If hald or pulseaudio has it open, you may be unable to switch away from > 48 ksps and then back.
My kernel is getting a little long in the tooth, so it wouldn't hurt me at some point to build a new one. 2.6.33 seemed a bit disruptive to lirc, but I think that's worked around by now. Do I even need cx18-alsa? I sort of built it because it was there and I have a cx18, even though I never needed it before. What is cx18-alsa for, anyway? > > 2. When you have no audio, try using v4l2-ctl to switch from the tuner > audio to line in audio and back. You could also stop then capture, > switch to 44.1 ksps or 32 ksps and then restart the capture. I'll give that a shot. I've no doubt I'll get the opportunity. I'm pleased enough that the unload/reload worked, because nothing else has so far. I don't know if it "worked" or if it was just luck, like booting occasionally is. > > > >> I had been focusing on initialization, partly because when we're >> analyzing, we analyze the major modes of operation practically to death, >> especially the performance-critical ones. There are things you can do >> to simulate end-of-life device shifts, etc, and we do those. For the >> power circuitry you analyze power-up to death, too. But initialization >> is one of those typically non-critical things that you logically verify. > > Whenever I see > > Verification method: inspection > > in a design review, my internal alarm bells go off. When I said "logically verify" I meant logical simulation, as opposed to full multi-corner analog transistor-level simulation. I don't trust "inspection" either. (There are those other phrases, "correct by design" and "correct by construction.") Dale _______________________________________________ ivtv-users mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-users
