Hi, I've been trying a few methods with no success of late to allow long term use of alsa. I've tried two paths with two different problems. 1. dmix. This is the default under alsa-lib for this card, and this is using default ICH4.conf settings under alsa-lib. After around 20 hours (I have tried this many, many times), sound totally stops, never to return. I believe this to be a ptr related issue as it doesn't seem to vary in time of occurrence regardless of the amount of sound being played. I have tried this on three different integrated chipsets (all based on intel8x0/ICH5/6, with two differing AC97 chips) with the same result. 2. sans-dmix. (Direct hw). Hardware mixing isn't a requirement in this case, so the lack of dmix is fairly irrelevant functionally. With direct hw, after about the same period of time, but at up to a couple of days, heavy distortion suddenly kicks in. My current thoughts are that this is a result of not handling the wierd period_time and buffer size created by the rate plugins sw->hw params code in snd_pcm_rate_hw_refine_cchange(). (Still not understanding why functions are being called in pcm_rate.c without and reference to it in alsa conf...). These figures are period_size 1115, buffer_size 3345. This is clearly a result of (48000 / 44100) * 1024 = 1114.55, as 44100/16/2/1024 are the requested settings, but I don't understand the actual rate conversion this way with the exception of obvious assumptions. The sans-dmix option doesn't have this problem on other hardware - trying the exact same file system on a machine with an ICH5 and AD1985 pcm chip has no problems, whilst the ICH6 & ALC655 chip combination falls down. The AD1985 seems to support 44100 direct to hw, however, but I did try using the rate plugin to set this to 48000, got the same period_size/buffer_size as above and had no problems. Two final bits of intruiging trivia:- running the same app through OSS (this is an SDL app, so this provides no problems) is problem free, and this is, importantly, only alsa oss emulation, so its following many of the same audio paths. I have tested this path for up to a month of application uptime. Secondly, restarting the program instantly fixes the issue. Some versions for reference: Kernel: 2.6.16.16 Alsa-lib: 1.0.11rc3 SDL: 1.2.8 [ I've also tried with the "Fix hwptr update in rate plugin" patch applied to alsa-lib ] Intel i915 ICH6 | ALC655 [ Broken Dmix & Distorted direct ] Intel 865 ICH5 | AD1985 [ Broken Dmix only. ] I will be leaving a couple of machines with different software configurations to test over the weekend, (latest alsa-lib, different kernel), but I would like any comments anyone has on this configuration or long term alsa use, especially with similar hardware. ------------------------------------------------------------------------- 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-user mailing list Alsa-user@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/alsa-user
