On Tue, Oct 05, 2010 at 11:05:11PM +0200, Tobias Ulmer wrote: > On Tue, Oct 05, 2010 at 05:48:08PM +0000, Jacob Meuser wrote: > > On Tue, Oct 05, 2010 at 04:27:59PM +0200, Landry Breuil wrote: > > > On Tue, Oct 05, 2010 at 02:15:29PM +0000, Jacob Meuser wrote: > > > > On Mon, Oct 04, 2010 at 05:11:47PM +0000, Jacob Meuser wrote: > > > > > On Mon, Oct 04, 2010 at 05:57:50PM +0200, Tobias Ulmer wrote: > > > > > > I'm against adding any cargo cult stuff like this and would rather > > > > > > see > > > > > > someone who knows aucat and the audio stuff figure out why it skips > > > > > > and > > > > > > clicks like mad. I doubt it's mpds fault. > > > > > > > > > > > > But that's probably not going to happen... Oh well. Commit it if you > > > > > > really have to. > > > > > > > > > > this is the first I heard of mpd skipping with auich. skipping > > > > > was reported with azalia, but since I haven't heard anything about > > > > > that since the interrupt issue was dealt with, I figured that's > > > > > what was causing it. > > > > > > > > > > anyway, I have no idea how to use mpd, and don't really care enough > > > > > to figure it out on my own. there are many mpd users here, afaik. > > > > > > > > so, I pkg_add'd mpd and mpc ... > > > > > > > > $ mpd > > > > log: problem opening log file "/var/log/mpd/mpd.log" (config line 37) > > > > for writing > > > > Abort trap (core dumped) > > > > $ > > > > > > oh come on. > > > > > > pkg_info -M mpd will show you that by default it's supposed to be > > > launched by root, and /etc/mpd.conf has shitload of comments. > > > > no, it tells you how to launch it from rc.local. it does not say mpd > > will crash and burn if you start it as a regular user without changing > > the config file. > > Hey Jacob, cool down a bit ;) > > The core dump comes from the fact that it uses glib error message > routines that are intended for debugging. Thanks to David, that has > been fixed upstream recently. > > Another stupid thing you will likely run into is that it doesn't create > some files on its own. "touch" them... > > Now, mpd is a daemon and the default config reflects that. You can of > course change it, but that's not how upstream distributes it. I don't > like the Debian way of doing things, and I'm probably not alone here... > > Anyway, I don't know what Davids exact problem is. I can only describe > mine. > > I have a cmpci(4) which usually works great: > cmpci0 at pci0 dev 16 function 0 "C-Media Electronics CMI8738/C3DX > Audio" rev 0x10: apic 2 int 19 (irq 11) > > Under heavy IO + crypto softraid, it sometimes skips. That's ok by me, I > don't expect a 2x500mhz machine to run cvs up and backups and nfs etc ^^^^^^^^ > while providing excellent audio.
(fwiw, I use a 266 MHz PII equipped old laptop as my music player in my workshop. it has an ISA soundcard. it's *very* difficult to make it skip.) > What is new since a couple of months (guessing 2ish), it goes into an ^^^^^^^^^^^^^^^^^^^^^^^^ > endless clicking noise intermixed with music and doesn't recover from ^^^^^^^^^^^^^^^ > that even if the machine is idle again. > > It's not that easy to reproduce, I have to do quite a few things in > parallel to get it there, and often it doesn't happen at all. > > mpc stop && sleep 2 && mpc play fixes it for a short time under load. > There is a noticable delay between the end of clicking and the end of > music (~.3 sec or so) > > Afair play.errors increases during the "clicking time". > > What makes me think it's not mpd is that it also happens with mplayer. > The difference is that mplayer consumes 1% cpu compared to 10% mpd. It > just happens less often. > > I can't really pin it down, and I didn't want to write a bug report > which basically consists of wild guessing. I'll try without aucat for a > while to check if that has any influence. sounds like the mp "delayed interrupt notification" issue that wreaks havoc on audio. thanks for explaining the problem. > Davids problem sounds different to mine, could well be they are unrelated. > > Tobias > > > > > > And yes, mpd's error handling is totally awful. MAINTAINER is aware of > > > it. > > > > > > Landry > > > > and so why should I care, really? if it doesn't work as a regular user > > try running it as root? talk about "oh come on" ... > > > > -- > > jake...@sdf.lonestar.org > > SDF Public Access UNIX System - http://sdf.lonestar.org > > -- jake...@sdf.lonestar.org SDF Public Access UNIX System - http://sdf.lonestar.org