At Tue, 13 Mar 2007 03:01:17 +0100, I wrote: > > At Tue, 13 Mar 2007 00:46:52 +0100, > I wrote: > > > > At Sat, 10 Mar 2007 14:17:13 +0000, > > John Rigg wrote: > > > > > > On Fri, Mar 09, 2007 at 05:38:34PM +0100, Takashi Iwai wrote: > > > > At Thu, 8 Mar 2007 21:51:28 +0000, > > > > John Rigg wrote: > > > > > > > > > > On Thu, Mar 08, 2007 at 07:58:23PM +0100, Takashi Iwai wrote: > > > > > > At Thu, 08 Mar 2007 18:16:02 +0100, > > > > > > Simon Lewis wrote: > > > > > > > Many JACK users have successfully used the following patch: > > > > > > > > > > > > > > http://www.sound-man.co.uk/linuxaudio/pcm_multi-patch > > > > > > > > > > > > Well, the patch is no real fix. It's likely a workaround for JACK, > > > > > > though. Someone has to take time to track down this bug more deeply. > > > > > > > > > > True, the patch is just a workaround for JACK. However, I would > > > > > argue that most users of pcm_multi are probably JACK users. > > > > > > > > No, multi plugin is used in many surround PCM definitions as default. > > > > It's pretty hidden, but multi streams are no rare case. > > > > > > I hadn't noticed that. Thanks for pointing it out (and thank you for > > > looking at this problem!) > > > > > > > > Surely adding code that breaks something for the majority of users (no > > > > > matter how correct that code is in isolation) is not a good thing. > > > > > > > > Sure. However, this code addition was for bugfixes of major other use > > > > cases like above, IIRC. The regression should be avoided and should > > > > be fixed. But the point is, we (at least, I) haven't been enough > > > > informed, unfortunately (or simply burried in a big TODO list :) > > > > > > > > Now, let's back to the original problem: Could someone give a pointer > > > > describing for this problem, or just explain a bit details here? > > > > I vaguely remember but not precisely at all now... > > > > I'm willing to dig down after knowing how I can reproduce the bug. > > > > > > To summarise, using multiple sound cards with pcm_multi and jackd > > > no longer works in duplex mode ever since extra linking code was > > > added to pcm_multi.c in alsa-lib-1.0.9rc1. > > > > > > Trying to start jackd in duplex fails with a poll timeout message. > > > It still works in playback-only or capture-only modes. > > > The lack of duplex operation makes overdubbing in a recording studio, > > > for example, impossible. > > > > > > A configuration which produces the problem with two ice1712 cards > > > set up as a 16 channel multi device is shown here: > > > http://www.sound-man.co.uk/linuxaudio/ice1712multi.html > > > > > > With the .asoundrc described in the above link the following > > > command fails with a poll timeout message: > > > > > > jackd -d alsa -P multi_playback -C multi_capture > > > > > > The following still work: > > > jackd -d alsa -P multi_playback > > > jackd -d alsa -C multi_capture > > > > OK, played a bit around this bug. A temporary fix is attached below. > > It's applied to HG tree now, too. Please give it a try. > > > > The real fix will come later... > > More complete fix is below. Apply after the previous patch. > (Or more easily use the HG version.) BTW, while debugging this, I found some potential problems in the current alsa support in JACK: - The arguments of snd_pcm_link() is master, slave. And, the master PCM should be used as a trigger. In the case of JACK, it should call like snd_pcm_link(playback_handle, capture_handle). - Mangles of poll revents are sometimes needed (e.g. dmix) by calling snd_pcm_poll_descriptor_revents() after poll. JACK might work without this, though, because it checks only revents !=0 or not. Once if it checks the direction (IN/OUT), there could be a problem. Takashi ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Alsa-devel mailing list Alsa-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/alsa-devel
