Re: [linux-audio-user] 2.6.8-rc1-mm1 [and alsa xrun debugging]

Re: [linux-audio-user] 2.6.8-rc1-mm1 [and alsa xrun debugging]

[Lee Revell]

On Fri, 2004-07-16 at 10:25, Florian Schmidt wrote:
> after reading lee's email i had to try this kernel.. so i went over to www.de.kernel.org, grabbed 2.6.8-rc1 and patched it up with this patch: 
> 
> http://www.ussg.iu.edu/hypermail/linux/kernel/0407.1/1453.html
> 
> On first sight it looks very good. many of the sporadic xruns i experienced 
> with jack in RT mode are gone. even a "find /" parallel to a "make bzImage" 
> seems not to provoke any xruns [i use ext3]. But: i use fluxbox and it supports 
> desktop wheeling [switching desktop via scrollwheel - very handy] and doing this 
> excessively and rapidly provokes xruns easily. Also starting mozilla just provoked 
> an xrun..

Try the included patch.  Andrew Morton suggested this, and it works
great for me, but it is not in -mm1.

Lee

--- drivers/char/tty_io.c_orig  2004-07-16 16:10:11.000000000 -0400
+++ drivers/char/tty_io.c       2004-07-16 16:10:31.000000000 -0400
@@ -679,17 +679,13 @@
                return -ERESTARTSYS;
        }
        if ( test_bit(TTY_NO_WRITE_SPLIT, &tty->flags) ) {
-               lock_kernel();
                written = write(tty, file, buf, count);
-               unlock_kernel();
        } else {
                for (;;) {
                        unsigned long size = max((unsigned long)PAGE_SIZE*2, 16384UL);
                        if (size > count)
                                size = count;
-                       lock_kernel();
                        ret = write(tty, file, buf, size);
-                       unlock_kernel();
                        if (ret <= 0)
                                break;
                        written += ret;

Re: [linux-audio-user] 2.6.8-rc1-mm1 [and alsa xrun debugging]

[Lee Revell]

On Fri, 2004-07-16 at 12:51, Florian Schmidt wrote:
> On Fri, 16 Jul 2004 16:54:28 +0100
> David M <eseol@tiscali.co.uk> wrote:
> 
> > I think you have to echo 2 > /proc/asound/card0/etc 
> > 
> > On debian they are in /var/log/messages,var/log/syslog , and a few other 
> > places.
> 
> oh ok, that works.. i'll also enable frame pointers in the kernel. but for now, here's some of the xruns i can provoke by rapid desktop wheeling [though kernel still w/o rame pointers]:
> 
> 
> Jul 16 18:47:31 mango kernel:  [<c0107761>] do_IRQ+0x91/0x130
> Jul 16 18:47:31 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20
> Jul 16 18:47:32 mango kernel: ALSA sound/core/pcm_lib.c:169: XRUN: pcmC0D0c
> Jul 16 18:47:32 mango kernel:  [<f08c0357>] snd_pcm_period_elapsed+0x2e7/0x460 [snd_pcm]
> Jul 16 18:47:32 mango kernel:  [<f0bc232e>] _nv000897rm+0x4e/0x70 [nvidia]
> Jul 16 18:47:32 mango kernel:  [<f08edb8e>] snd_cs46xx_interrupt+0x1be/0x1f0 [snd_cs46xx]
> Jul 16 18:47:32 mango kernel:  [<f0d28962>] nv_kern_isr+0x26/0x59 [nvidia]
> Jul 16 18:47:32 mango kernel:  [<c01073fa>] handle_IRQ_event+0x3a/0x70
> Jul 16 18:47:32 mango kernel:  [<c0107761>] do_IRQ+0x91/0x130
> Jul 16 18:47:32 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20

Are you using the binary-only NVIDIA module?  If your kernel is tainted
(by having a binary-only module loaded), then you cannot get valid data.
> Jul 16 18:49:04 mango kernel: ALSA sound/core/pcm_lib.c:169: XRUN: pcmC0D0c
> Jul 16 18:49:04 mango kernel:  [<f08c0357>] snd_pcm_period_elapsed+0x2e7/0x460 [snd_pcm]
> Jul 16 18:49:04 mango kernel:  [<f08edb8e>] snd_cs46xx_interrupt+0x1be/0x1f0 [snd_cs46xx]
> Jul 16 18:49:04 mango kernel:  [<c0105737>] handle_signal+0xb7/0x120
> Jul 16 18:49:04 mango kernel:  [<c01073fa>] handle_IRQ_event+0x3a/0x70
> Jul 16 18:49:04 mango kernel:  [<c0107761>] do_IRQ+0x91/0x130
> Jul 16 18:49:04 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20
> Jul 16 18:49:04 mango kernel:  [<c011a2f0>] __do_softirq+0x30/0x80
> Jul 16 18:49:04 mango kernel:  [<c011a366>] do_softirq+0x26/0x30
> Jul 16 18:49:04 mango kernel:  [<c01077cd>] do_IRQ+0xfd/0x130
> Jul 16 18:49:04 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20
> Jul 16 18:49:04 mango kernel: ALSA sound/core/pcm_lib.c:169: XRUN: pcmC0D0p
> Jul 16 18:49:04 mango kernel:  [<f08c0357>] snd_pcm_period_elapsed+0x2e7/0x460 [snd_pcm]
> Jul 16 18:49:04 mango kernel:  [<c011e1c6>] update_process_times+0x46/0x60
> Jul 16 18:49:04 mango kernel:  [<f08edb8e>] snd_cs46xx_interrupt+0x1be/0x1f0 [snd_cs46xx]
> Jul 16 18:49:04 mango kernel:  [<c011e4bf>] do_timer+0xdf/0xf0
> Jul 16 18:49:04 mango kernel:  [<c010b378>] timer_interrupt+0xb8/0x140
> Jul 16 18:49:04 mango kernel:  [<c01073fa>] handle_IRQ_event+0x3a/0x70
> Jul 16 18:49:04 mango kernel:  [<c0107761>] do_IRQ+0x91/0x130
> Jul 16 18:49:04 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20
> Jul 16 18:49:04 mango kernel:  [<c011a2f0>] __do_softirq+0x30/0x80
> Jul 16 18:49:04 mango kernel:  [<c011a366>] do_softirq+0x26/0x30
> Jul 16 18:49:04 mango kernel:  [<c01077cd>] do_IRQ+0xfd/0x130
> Jul 16 18:49:04 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20

Takashi Iwai mentioned there were some patches available to daemonize
softirq processing, and that these might help.  Takashi, can you provide
some details on this?

> Jul 16 18:49:05 mango kernel: ALSA sound/core/pcm_lib.c:169: XRUN: pcmC0D0c
> Jul 16 18:49:05 mango kernel:  [<f08c0357>] snd_pcm_period_elapsed+0x2e7/0x460 [snd_pcm]
> Jul 16 18:49:05 mango kernel:  [<c028d579>] unix_stream_sendmsg+0x269/0x3e0
> Jul 16 18:49:05 mango kernel:  [<f08edb8e>] snd_cs46xx_interrupt+0x1be/0x1f0 [snd_cs46xx]
> Jul 16 18:49:05 mango kernel:  [<c01073fa>] handle_IRQ_event+0x3a/0x70
> Jul 16 18:49:05 mango kernel:  [<c0107761>] do_IRQ+0x91/0x130
> Jul 16 18:49:05 mango kernel:  [<c0105be4>] common_interrupt+0x18/0x20
> Jul 16 18:49:05 mango kernel:  [<c0112cab>] finish_task_switch+0x2b/0x90
> Jul 16 18:49:05 mango kernel:  [<c0290e4f>] schedule+0x2cf/0x560
> Jul 16 18:49:05 mango kernel:  [<c02915e3>] schedule_timeout+0x63/0xc0
> Jul 16 18:49:05 mango kernel:  [<c011e5b0>] process_timeout+0x0/0x10
> Jul 16 18:49:05 mango kernel:  [<c015faaf>] do_select+0x18f/0x2d0
> Jul 16 18:49:05 mango kernel:  [<c015f770>] __pollwait+0x0/0xd0
> Jul 16 18:49:05 mango kernel:  [<c015fedf>] sys_select+0x2bf/0x4c0
> Jul 16 18:49:05 mango kernel:  [<c014c30b>] vfs_read+0x10b/0x140
> Jul 16 18:49:05 mango kernel:  [<c0105a77>] syscall_call+0x7/0xb

This one I am seeing too.  I do not have a good explanation for it yet.

Lee