From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266221AbUAQXmj (ORCPT ); Sat, 17 Jan 2004 18:42:39 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266227AbUAQXmj (ORCPT ); Sat, 17 Jan 2004 18:42:39 -0500 Received: from b0504.opsion.fr ([62.39.122.72]:11017 "HELO b0504.idoo.com") by vger.kernel.org with SMTP id S266221AbUAQXmc convert rfc822-to-8bit (ORCPT ); Sat, 17 Jan 2004 18:42:32 -0500 Subject: Re : Alsa create high problems... From: Eddahbi Karim To: linux-kernel@vger.kernel.org Content-Type: text/plain; charset=ISO-8859-1 Message-Id: <1074382859.29525.20.camel@gamux> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Sun, 18 Jan 2004 00:41:00 +0100 Content-Transfer-Encoding: 8BIT if-filter0: N Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org I'm posting the answer to a mail that I received because it will maybe help a developper to fix the problem. Le sam 10/01/2004 à 11:31, Jaroslav Kysela a écrit : > On Sat, 10 Jan 2004, Eddahbi Karim wrote: > > > Hiya, > > > > I've pretty bad problems with the ALSA driver and that drives me mad. > > The driver seems to, sometimes, broke. > > > > I'm using the kernel 2.6.1-mm1. > > This bug doesn't happen on my 2.6.0 with hasn't PREEMPT enabled. > > > > For example, in XMMS, if a program requires a lot of CPU in 1 second, > > the song will stop and freeze in a loop (The song will repeat the last > > second like a while(1)). > > It's with via82xx or emu10k1 driver? Can you send me contents of > /proc/asound/card0/pcm0p/sub0/* files when the crash occurs? Ok here it is : access: MMAP_INTERLEAVED format: S16_LE subformat: STD channels: 1 rate: 22050 (22050/1) period_size: 1225 buffer_size: 11025 tick_time: 1000 card: 0 device: 0 subdevice: 0 stream: PLAYBACK id: VIA 82C686A/B rev50 name: VIA 82C686A/B rev50 subname: subdevice #0 class: 0 subclass: 0 subdevices_count: 1 subdevices_avail: 0 64 state: RUNNING trigger_time: 1073850885.877994000 tstamp : 1073850900.726626000 delay : 10083 avail : 942 avail_max : 1433 ----- hw_ptr : 10077 appl_ptr : 20160 tstamp_mode: NONE period_step: 1 sleep_min: 0 avail_min: 1225 xfer_align: 1225 start_threshold: 1 stop_threshold: 11025 silence_threshold: 0 silence_size: 0 boundary: 1445068800 By the way, If I run cat /proc/asound/card0/pcm0p/sub0/* two or three times, it unlock XMMS and continue to play With madplay, it just unlock 1 second. But I've found something interesting with madplay. When the music lockup these two values are different : avail : 0 avail_max : 9301 And If continue until avail_max reaches 0, madplay will play another second of the music. I've run madplay on a terminal and "script" on another one. The log created by script is attached to the mail. In the log, each time that avail and avail_max are equal after I typed cat, madplay will play a second of the music. If the value are different, madplay won't continue. By the way, some informations : It doesn't appear after a high load create by the system. The system seems to lag after a certain amount of time. I just power on my computer, let the screensaver plays and I saw some lags after 30 minutes. So I launched XMMS, try to play and change to another music and the bug appears. The screensaver is rss-glx : * x11-misc/rss-glx Latest version available: 0.7.4-r1 Latest version installed: 0.7.4-r1 Size of downloaded files: 4,771 kB Homepage: http://rss-glx.sourceforge.net/ Description: Really Slick Screensavers using OpenGL for XScreenSaver My graphic card is a Geforce FX 5200, I'm using nvidia binary drivers. Dmesg displays that about nvidia drivers : Debug: sleeping function called from invalid context at mm/slab.c:1868 in_atomic():1, irqs_disabled():0 Call Trace: [] __might_sleep+0xab/0xd0 [] __kmalloc+0x89/0x90 [] os_alloc_mem+0x7c/0x90 [nvidia] [] _nv001308rm+0x10/0x28 [nvidia] [] _nv001518rm+0x7c9/0xb34 [nvidia] [] _nv002463rm+0x78/0x194 [nvidia] [] _nv001338rm+0x1d/0x24 [nvidia] [] _nv000858rm+0x300/0xe14 [nvidia] [] ppp_start_xmit+0x105/0x290 [] qdisc_restart+0x18/0x140 [] dev_queue_xmit+0x21c/0x2b0 [] _nv002962rm+0x2c5/0x3b8 [nvidia] [] _nv000899rm+0x4c9/0xf70 [nvidia] [] _nv000899rm+0x4dc/0xf70 [nvidia] [] dst_output+0x15/0x30 [] dst_output+0x0/0x30 [] ip_queue_xmit+0x49e/0x5b0 [] dst_output+0x0/0x30 [] qdisc_restart+0x18/0x140 [] dev_queue_xmit+0x21c/0x2b0 [] ip_finish_output2+0x11f/0x1e0 [] tcp_v4_send_check+0x4a/0xf0 [] tcp_transmit_skb+0x3c4/0x5e0 [] ip_finish_output+0x234/0x240 [] tcp_send_ack+0x82/0xd0 [] _nv001556rm+0x5b/0x6c [nvidia] [] _nv001556rm+0x5b/0x6c [nvidia] [] _nv003622rm+0x15/0x1c [nvidia] [] _nv001826rm+0xb0/0xbc [nvidia] [] _nv001556rm+0x5b/0x6c [nvidia] [] _nv003622rm+0x15/0x1c [nvidia] [] _nv001826rm+0xb0/0xbc [nvidia] [] _nv001556rm+0x5b/0x6c [nvidia] [] _nv003622rm+0x15/0x1c [nvidia] [] _nv003795rm+0xad9/0xaec [nvidia] [] _nv001532rm+0x1f/0x28 [nvidia] [] _nv004240rm+0x180/0x18c [nvidia] [] _nv001534rm+0x20/0x28 [nvidia] [] _nv001223rm+0x12/0x18 [nvidia] [] _nv004046rm+0x3a3/0x3b0 [nvidia] [] _nv001223rm+0x12/0x18 [nvidia] [] _nv001476rm+0x452/0x45c [nvidia] [] wake_up_state+0x1a/0x20 [] send_group_sig_info+0x2f/0x50 [] send_sigio_to_task+0x10b/0x130 [] recalc_task_prio+0x90/0x1a0 [] __switch_to+0x145/0x1c0 [] recalc_task_prio+0x90/0x1a0 [] __switch_to+0x145/0x1c0 [] schedule+0x35b/0x5c0 [] avc_has_perm+0x6c/0x79 [] convert_fxsr_from_user+0x96/0x150 [] inode_has_perm+0x5f/0x90 [] selinux_file_ioctl+0x121/0x420 [] rm_ioctl+0x19/0x20 [nvidia] [] nv_kern_ioctl+0x7c/0x490 [nvidia] [] sys_ioctl+0x115/0x2b0 [] syscall_call+0x7/0xb Here is the md5sum of the log : db8069da9e6a8c239db8a7b74706926a typescript.gz Thanks for the help, I'm going down to my 2.6 which has the PREEMPT option activated :). UPDATE : It still happens on 2.6.1-mm4 while true; do cat /proc/asound/card0/pcm0p/sub0/*; done freeze if I play with XMMS. That's a weird and annoying problem which doesn't appears on my 2.6.0. Here is the link for typescript.gz because I don't think that I can attach a file on the LKML : http://thetemplar.free.fr/typescript.gz -- -- Eddahbi Karim Phone : (33) (0)6 61 30 57 77 France _____________________________________________________________________ Envie de discuter en "live" avec vos amis ? Télécharger MSN Messenger http://www.ifrance.com/_reloc/m la 1ère messagerie instantanée de France