Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

--- Takashi Iwai <tiwai@suse.de> wrote:
> At Sat, 16 Apr 2005 02:31:06 +0200 (CEST),
> karsten wiese wrote:
> > 
> > --- Clemens Ladisch <clemens@ladisch.de> wrote:
> > > Takashi Iwai wrote:
> > > > karsten wiese wrote:
> > > > > (-: You apply my patch now?
> > > >
> > > > No, not enough :)  This is exactly why
> > > snd_card_free_in_thread() was
> > > > introduced.
> > > >
> > > > As I asked, please make sure that the files are all
> > > released before
> > > > snd_card_free().  Otherwise the disconnect callback
> > > hangs at this
> > > > point because of this check.
> > > 
> > > Most programs close the device file when an error
> occurs,
> > > so you have
> > > to test it with something like this to keep the file
> > > open:
> > > 
> > > #include <stdio.h>
> > > int main()
> > > {
> > >   fopen("/dev/snd/hwC1D0", "rb");
> > >   getchar();
> > > }
> > > 
> > Tested with it, see attached shell log (copied+pasted).
> > Any more tests needed?
> 
> Did you check whether the release fops callback
> (e.g. snd_hwdep_release) is called?

Yes: All snd_*_release() functions call 
snd_card_file_remove() and snd_card_file_remove() definitly
was called as usX2Y_usb_disconnect() finished correctly,
when the client closed it's fd(s).
Only snd_card_file_remove() calls
wake_up(&card->shutdown_sleep).
Only other places where snd_card_file_remove() is called
are the snd_*_open() functions error paths, but those where
not involved, as lsof stabily showed that /dev/snd/hwC1D0
was held open by a.out (the client).

Karsten


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Tue, 19 Apr 2005 23:22:06 +0200 (CEST),
karsten wiese wrote:
> 
> 
> --- Takashi Iwai <tiwai@suse.de> wrote:
> > At Sat, 16 Apr 2005 02:31:06 +0200 (CEST),
> > karsten wiese wrote:
> > > 
> > > --- Clemens Ladisch <clemens@ladisch.de> wrote:
> > > > Takashi Iwai wrote:
> > > > > karsten wiese wrote:
> > > > > > (-: You apply my patch now?
> > > > >
> > > > > No, not enough :)  This is exactly why
> > > > snd_card_free_in_thread() was
> > > > > introduced.
> > > > >
> > > > > As I asked, please make sure that the files are all
> > > > released before
> > > > > snd_card_free().  Otherwise the disconnect callback
> > > > hangs at this
> > > > > point because of this check.
> > > > 
> > > > Most programs close the device file when an error
> > occurs,
> > > > so you have
> > > > to test it with something like this to keep the file
> > > > open:
> > > > 
> > > > #include <stdio.h>
> > > > int main()
> > > > {
> > > >   fopen("/dev/snd/hwC1D0", "rb");
> > > >   getchar();
> > > > }
> > > > 
> > > Tested with it, see attached shell log (copied+pasted).
> > > Any more tests needed?
> > 
> > Did you check whether the release fops callback
> > (e.g. snd_hwdep_release) is called?
> 
> Yes: All snd_*_release() functions call 
> snd_card_file_remove() and snd_card_file_remove() definitly
> was called as usX2Y_usb_disconnect() finished correctly,
> when the client closed it's fd(s).
> Only snd_card_file_remove() calls
> wake_up(&card->shutdown_sleep).
> Only other places where snd_card_file_remove() is called
> are the snd_*_open() functions error paths, but those where
> not involved, as lsof stabily showed that /dev/snd/hwC1D0
> was held open by a.out (the client).

OK, thanks for the confirmation!

Now I applied your patch (also modified usbaudio.c) together with
the kernel version check in *.patch files.


Takashi


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

--- Takashi Iwai <tiwai@suse.de> wrote:
> At Sat, 16 Apr 2005 02:31:06 +0200 (CEST),
> karsten wiese wrote:
> > 
> > [1  <text/plain; iso-8859-1 (8bit)>]
> > 
> > --- Clemens Ladisch <clemens@ladisch.de> wrote:
> > > Takashi Iwai wrote:
> > > > karsten wiese wrote:
> > > > > (-: You apply my patch now?
> > > >
> > > > No, not enough :)  This is exactly why
> > > snd_card_free_in_thread() was
> > > > introduced.
> > > >
> > > > As I asked, please make sure that the files are all
> > > released before
> > > > snd_card_free().  Otherwise the disconnect callback
> > > hangs at this
> > > > point because of this check.
> > > 
> > > Most programs close the device file when an error
> occurs,
> > > so you have
> > > to test it with something like this to keep the file
> > > open:
> > > 
> > > #include <stdio.h>
> > > int main()
> > > {
> > >   fopen("/dev/snd/hwC1D0", "rb");
> > >   getchar();
> > > }
> > > 
> > Tested with it, see attached shell log (copied+pasted).
> > Any more tests needed?
> > I bet a kasten bier that it oopses reliably on first
> try,
> > if I revert to snd_card_free_in_thread() nothing else
> > changed.
> > And I'll inform you , if any oops should show when i
> shut
> > down (pc is still in the state like @ the end of the
> log).
> > 
> > Regards,
> > Karsten
> > 
> > P.S.: how about adding a kernelversion depending macro
> for
> > backward compatibility,... when was
> > snd_card_free_in_thread() introduced?
> 
> Agreed.  The problem is that how to know which version
> has still this
> problem..?  At least, 2.4 kernel must have a problem
> without
> snd_card_free_in_thread().

>From memory I've been hitting the keyboard dead bug since
quite a long time ago. So as a first estimate I suggest 
"kernelversion >= 2.6.0" as condition for use of
snd_card_free().

Karsten


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

[-- Attachment #1: Type: text/plain, Size: 1335 bytes --]


--- Clemens Ladisch <clemens@ladisch.de> wrote:
> Takashi Iwai wrote:
> > karsten wiese wrote:
> > > (-: You apply my patch now?
> >
> > No, not enough :)  This is exactly why
> snd_card_free_in_thread() was
> > introduced.
> >
> > As I asked, please make sure that the files are all
> released before
> > snd_card_free().  Otherwise the disconnect callback
> hangs at this
> > point because of this check.
> 
> Most programs close the device file when an error occurs,
> so you have
> to test it with something like this to keep the file
> open:
> 
> #include <stdio.h>
> int main()
> {
>   fopen("/dev/snd/hwC1D0", "rb");
>   getchar();
> }
> 
Tested with it, see attached shell log (copied+pasted).
Any more tests needed?
I bet a kasten bier that it oopses reliably on first try,
if I revert to snd_card_free_in_thread() nothing else
changed.
And I'll inform you , if any oops should show when i shut
down (pc is still in the state like @ the end of the log).

Regards,
Karsten

P.S.: how about adding a kernelversion depending macro for
backward compatibility,... when was
snd_card_free_in_thread() introduced?


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: test-snd_card_free-open_hw1 --]
[-- Type: text/x-java; name=test-snd_card_free-open_hw1, Size: 9137 bytes --]

[ka@p4 debug]$ /home/ka/kernel/linux-2.6.12-rc2/sound/usb/usx2y/a.out

[root@p4 ~]# lsof /dev/snd/*
COMMAND  PID USER   FD   TYPE DEVICE SIZE  NODE NAME
kmix    3797   ka   11u   CHR  116,0       4515 /dev/snd/controlC0
a.out   7124   ka    4r   CHR 116,36      24757 /dev/snd/hwC1D0

[root@p4 ~]# dmesg | tail -n5
[4297447.497000] CLASS: registering class device: ID = 'pcmC1D2c'
[4297447.497000] kobject pcmC1D2c: registering. parent: sound, set: class_obj
[4297447.497000] fill_kobj_path: path = '/class/sound/pcmC1D2c'
[4297447.497000] class_hotplug - name = pcmC1D2c
[4297447.497000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'

#switch off us428

[root@p4 ~]# dmesg | tail -n29
[4297447.497000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298506.550000] hub 3-0:1.0: state 5 ports 2 chg 0000 evt 0004
[4298506.550000] uhci_hcd 0000:00:10.1: port 2 portsc 008a,00
[4298506.550000] hub 3-0:1.0: port 2, status 0100, change 0003, 12 Mb/s
[4298506.550000] usb 3-2: USB disconnect, address 6
[4298506.550000] usb 3-2: usb_disable_device nuking all URBs
[4298506.550000] usb 3-2: unregistering interface 3-2:1.0
[4298506.550000] usX2Y_usb_disconnect(f1946c00,e8483c00)
[4298506.550000] CLASS: Unregistering class device. ID = 'mixer1'
[4298506.550000] fill_kobj_path: path = '/class/sound/mixer1'
[4298506.550000] class_hotplug - name = mixer1
[4298506.569000] kobject mixer1 @ f74b1f90: cleaning up
[4298506.569000] device class 'mixer1': release.
[4298506.569000] CLASS: Unregistering class device. ID = 'adsp1'
[4298506.569000] fill_kobj_path: path = '/class/sound/adsp1'
[4298506.569000] class_hotplug - name = adsp1
[4298506.588000] kobject adsp1 @ e79a5790: cleaning up
[4298506.588000] device class 'adsp1': release.
[4298506.588000] CLASS: Unregistering class device. ID = 'dsp1'
[4298506.588000] fill_kobj_path: path = '/class/sound/dsp1'
[4298506.588000] class_hotplug - name = dsp1
[4298506.608000] kobject dsp1 @ f7841110: cleaning up
[4298506.608000] device class 'dsp1': release.
[4298506.608000] CLASS: Unregistering class device. ID = 'audio1'
[4298506.608000] fill_kobj_path: path = '/class/sound/audio1'
[4298506.608000] class_hotplug - name = audio1
[4298506.629000] kobject audio1 @ f6c9da90: cleaning up
[4298506.629000] device class 'audio1': release.
[4298507.572000] uhci_hcd 0000:00:10.1: suspend_hc

[root@p4 ~]# lsof /dev/snd/*
COMMAND  PID USER   FD   TYPE DEVICE SIZE  NODE NAME
kmix    3797   ka   11u   CHR  116,0       4515 /dev/snd/controlC0
a.out   7124   ka    4r   CHR 116,36      24757 /dev/snd/hwC1D0


#way 1 , see below what happens when us428 is switched on while "/dev/snd/hwC1D0" is still open

#control-c to quit a.out
[ka@p4 debug]$

[root@p4 ~]# dmesg | tail -n78
[4298507.572000] uhci_hcd 0000:00:10.1: suspend_hc
[4298653.844000] CLASS: Unregistering class device. ID = 'controlC1'
[4298653.844000] fill_kobj_path: path = '/class/sound/controlC1'
[4298653.844000] class_hotplug - name = controlC1
[4298653.844000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.865000] kobject controlC1 @ f6e2c310: cleaning up
[4298653.865000] device class 'controlC1': release.
[4298653.865000] CLASS: Unregistering class device. ID = 'pcmC1D2p'
[4298653.865000] fill_kobj_path: path = '/class/sound/pcmC1D2p'
[4298653.865000] class_hotplug - name = pcmC1D2p
[4298653.865000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.886000] kobject pcmC1D2p @ e743c410: cleaning up
[4298653.886000] device class 'pcmC1D2p': release.
[4298653.886000] CLASS: Unregistering class device. ID = 'pcmC1D2c'
[4298653.886000] fill_kobj_path: path = '/class/sound/pcmC1D2c'
[4298653.886000] class_hotplug - name = pcmC1D2c
[4298653.886000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.906000] kobject pcmC1D2c @ e743c390: cleaning up
[4298653.906000] device class 'pcmC1D2c': release.
[4298653.906000] CLASS: Unregistering class device. ID = 'hwC1D1'
[4298653.906000] fill_kobj_path: path = '/class/sound/hwC1D1'
[4298653.906000] class_hotplug - name = hwC1D1
[4298653.906000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.927000] kobject hwC1D1 @ e79a5610: cleaning up
[4298653.927000] device class 'hwC1D1': release.
[4298653.927000] CLASS: Unregistering class device. ID = 'pcmC1D1c'
[4298653.927000] fill_kobj_path: path = '/class/sound/pcmC1D1c'
[4298653.927000] class_hotplug - name = pcmC1D1c
[4298653.927000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.944000] kobject pcmC1D1c @ e79a5810: cleaning up
[4298653.944000] device class 'pcmC1D1c': release.
[4298653.944000] CLASS: Unregistering class device. ID = 'pcmC1D0p'
[4298653.944000] fill_kobj_path: path = '/class/sound/pcmC1D0p'
[4298653.944000] class_hotplug - name = pcmC1D0p
[4298653.944000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.950000] kobject pcmC1D0p @ f0e06710: cleaning up
[4298653.950000] device class 'pcmC1D0p': release.
[4298653.950000] CLASS: Unregistering class device. ID = 'pcmC1D0c'
[4298653.950000] fill_kobj_path: path = '/class/sound/pcmC1D0c'
[4298653.950000] class_hotplug - name = pcmC1D0c
[4298653.950000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298653.968000] kobject pcmC1D0c @ f0e06990: cleaning up
[4298653.968000] device class 'pcmC1D0c': release.
[4298653.968000] CLASS: Unregistering class device. ID = 'midi1'
[4298653.968000] fill_kobj_path: path = '/class/sound/midi1'
[4298653.968000] class_hotplug - name = midi1
[4298653.994000] kobject midi1 @ e8be1790: cleaning up
[4298653.994000] device class 'midi1': release.
[4298653.994000] CLASS: Unregistering class device. ID = 'dmmidi1'
[4298653.994000] fill_kobj_path: path = '/class/sound/dmmidi1'
[4298653.994000] class_hotplug - name = dmmidi1
[4298654.013000] kobject dmmidi1 @ e8be1690: cleaning up
[4298654.013000] device class 'dmmidi1': release.
[4298654.013000] CLASS: Unregistering class device. ID = 'midiC1D0'
[4298654.013000] fill_kobj_path: path = '/class/sound/midiC1D0'
[4298654.013000] class_hotplug - name = midiC1D0
[4298654.013000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298654.031000] kobject midiC1D0 @ e8be1310: cleaning up
[4298654.031000] device class 'midiC1D0': release.
[4298654.031000] CLASS: Unregistering class device. ID = 'hwC1D0'
[4298654.031000] fill_kobj_path: path = '/class/sound/hwC1D0'
[4298654.031000] class_hotplug - name = hwC1D0
[4298654.031000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298654.049000] kobject hwC1D0 @ f6e2c090: cleaning up
[4298654.049000] device class 'hwC1D0': release.
[4298654.049000] usX2Y_usb_disconnect() leaving
[4298654.049000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2/3-2:1.0'
[4298654.049000] usb 3-2:1.0: hotplug
[4298654.067000] kobject 3-2:1.0 @ f5423178: cleaning up
[4298654.067000] usb 3-2: unregistering device
[4298654.067000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
[4298654.067000] usb 3-2: hotplug
[4298654.099000] kobject 3-2 @ f1946c88: cleaning up
[4298654.206000] hub 3-0:1.0: debounce: port 2: total 100ms stable 100ms status 0x100
[4298654.206000] hub 1-0:1.0: state 5 ports 8 chg 0000 evt 0010
[4298654.206000] ehci_hcd 0000:00:10.4: GetStatus port 4 status 001002 POWER sig=se0  CSC
[4298654.206000] hub 1-0:1.0: port 4, status 0100, change 0001, 12 Mb/s
[4298654.311000] hub 1-0:1.0: debounce: port 4: total 100ms stable 100ms status 0x100


-----------------------------------------

#way 2, us428 is switched on again while a.out still holds  "/dev/snd/hwC1D0" open

[root@p4 ~]# dmesg | tail -n10
[4298822.089000] fill_kobj_path: path = '/class/sound/dsp1'
[4298822.089000] class_hotplug - name = dsp1
[4298822.109000] kobject dsp1 @ f6883b10: cleaning up
[4298822.109000] device class 'dsp1': release.
[4298822.109000] CLASS: Unregistering class device. ID = 'audio1'
[4298822.109000] fill_kobj_path: path = '/class/sound/audio1'
[4298822.109000] class_hotplug - name = audio1
[4298822.129000] kobject audio1 @ f6883890: cleaning up
[4298822.129000] device class 'audio1': release.
[4298822.972000] uhci_hcd 0000:00:10.1: suspend_hc

# looks the same ,  except for memory pointers and timer marks , compared to when we took way 1
# also note: us428 USB light stay's OFF !!

#control-c to quit a.out
[ka@p4 debug]$

# us428 USB light is ON , yesz! also us428control has been started by hotplug:
[root@p4 ~]# lsof /dev/snd/*
COMMAND    PID USER   FD   TYPE DEVICE SIZE  NODE NAME
kmix      3797   ka   11u   CHR  116,0       4515 /dev/snd/controlC0
us428cont 9183 root  mem    CHR 116,36      37064 /dev/snd/hwC1D0
us428cont 9183 root    3u   CHR 116,36      37064 /dev/snd/hwC1D0
us428cont 9183 root    4u   CHR  116,1      12293 /dev/snd/seq

[ka@p4 usx2y]$ arecord -Dhw:1 -fcd | aplay -Dhw:1
Recording WAVE 'stdout' : Signed 16 bit Little Endian, Rate 44100 Hz, Stereo
Playing WAVE 'stdin' : Signed 16 bit Little Endian, Rate 44100 Hz, Stereo

#...... sounds just fine ....no oops, everything ok  :-)

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Sat, 16 Apr 2005 02:31:06 +0200 (CEST),
karsten wiese wrote:
> 
> [1  <text/plain; iso-8859-1 (8bit)>]
> 
> --- Clemens Ladisch <clemens@ladisch.de> wrote:
> > Takashi Iwai wrote:
> > > karsten wiese wrote:
> > > > (-: You apply my patch now?
> > >
> > > No, not enough :)  This is exactly why
> > snd_card_free_in_thread() was
> > > introduced.
> > >
> > > As I asked, please make sure that the files are all
> > released before
> > > snd_card_free().  Otherwise the disconnect callback
> > hangs at this
> > > point because of this check.
> > 
> > Most programs close the device file when an error occurs,
> > so you have
> > to test it with something like this to keep the file
> > open:
> > 
> > #include <stdio.h>
> > int main()
> > {
> >   fopen("/dev/snd/hwC1D0", "rb");
> >   getchar();
> > }
> > 
> Tested with it, see attached shell log (copied+pasted).
> Any more tests needed?
> I bet a kasten bier that it oopses reliably on first try,
> if I revert to snd_card_free_in_thread() nothing else
> changed.
> And I'll inform you , if any oops should show when i shut
> down (pc is still in the state like @ the end of the log).
> 
> Regards,
> Karsten
> 
> P.S.: how about adding a kernelversion depending macro for
> backward compatibility,... when was
> snd_card_free_in_thread() introduced?

Agreed.  The problem is that how to know which version has still this
problem..?  At least, 2.4 kernel must have a problem without
snd_card_free_in_thread().


Takashi


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Clemens Ladisch]

Takashi Iwai wrote:
> karsten wiese wrote:
> > P.S.: how about adding a kernelversion depending macro for
> > backward compatibility,... when was
> > snd_card_free_in_thread() introduced?
>
> Agreed.  The problem is that how to know which version has still this
> problem..?  At least, 2.4 kernel must have a problem without
> snd_card_free_in_thread().

I think all 2.6 kernels have reference counted kernel objects.

The actual crash is caused by the class_simple_device_remove() call in
snd_unregister_device() (which is a dummy in kernels before 2.6).
Perhaps this call should be put into the snd_disconnect_device()
function (which doesn't yet exist).


Regards,
Clemens



-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Sat, 16 Apr 2005 02:31:06 +0200 (CEST),
karsten wiese wrote:
> 
> --- Clemens Ladisch <clemens@ladisch.de> wrote:
> > Takashi Iwai wrote:
> > > karsten wiese wrote:
> > > > (-: You apply my patch now?
> > >
> > > No, not enough :)  This is exactly why
> > snd_card_free_in_thread() was
> > > introduced.
> > >
> > > As I asked, please make sure that the files are all
> > released before
> > > snd_card_free().  Otherwise the disconnect callback
> > hangs at this
> > > point because of this check.
> > 
> > Most programs close the device file when an error occurs,
> > so you have
> > to test it with something like this to keep the file
> > open:
> > 
> > #include <stdio.h>
> > int main()
> > {
> >   fopen("/dev/snd/hwC1D0", "rb");
> >   getchar();
> > }
> > 
> Tested with it, see attached shell log (copied+pasted).
> Any more tests needed?

Did you check whether the release fops callback
(e.g. snd_hwdep_release) is called?


Takashi


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

--- Takashi Iwai <tiwai@suse.de> wrote:
> At Fri, 15 Apr 2005 16:28:54 +0200 (CEST),
> karsten wiese wrote:
> > 
> > 
> > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> > > karsten wiese wrote:
> > > > 
> > > > 
> > > > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > > > On Thu, 2005-04-14 at 20:59 +0200, karsten wiese
> > > wrote:
> > > > > > Hi Clemens & Takashi
> > > > > > 
> > > > > > Please commit the patch.
> > > > > 
> > > > > Is the ALSA in kernel 2.6.11 affected?  If so,
> this
> > > > 
> > > > Think so, yes. The keyboard dead effect + described
> > > oops
> > > > has been around for quiet a long time.
> > > > (There is a lkml mail from yesterday about an
> > > usb-cellphone
> > > > setup,
> > > > where the same oops + dead keyboard happens also:
> > > >
> > >
> >
>
http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > > > )
> > > > There is also an alsa-bug with snd-usb-audio in the
> > > > bugbase.
> > > > Clemens seams not to be convinced about this
> aproach
> > > yet.
> > > > maybe 'cause the comment of snd_card_free_in_thread
> > > > explicitly states that its the one to be called for
> > > > hotplug:
> > > > /**
> > > >  *  snd_card_free_in_thread - call snd_card_free()
> in
> > > > thread
> > > >  *  @card: soundcard structure
> > > >  *
> > > >  *  This function schedules the call of
> snd_card_free()
> > > > function in a
> > > >  *  work queue.  When all devices are released
> > > (non-busy),
> > > > the work
> > > >  *  is woken up and calls snd_card_free().
> > > >  *
> > > >  *  When a card can be disconnected at any time by
> > > hotplug
> > > > service,
> > > >  *  this function should be used in disconnect (or
> > > detach)
> > > > callback
> > > >  *  instead of calling snd_card_free() directly.
> > > >  *  
> > > >  *  Returns - zero otherwise a negative error code
> if
> > > the
> > > > start of thread failed.
> > > >  */
> > > > 
> > > > Also I remember , that snd_card_free didn't work ok
> > > when i
> > > > used it in the first implementations of
> usx2y-module
> > > some 2
> > > > years ago.
> > > > but that was on kernel 2.4.something.......
> > > > maybe above comment is from kernel 2.4 also and the
> > > hotplug
> > > > system has changed ? anybody?
> > > 
> > > Might be.  To be sure, check whether the release fops
> is
> > > called
> > > properly before you calling snd_card_free().
> > 
> > You mean snd_card_disconnect()?
> > That's called with the same argument and before
> > snd_card_free().
> 
> No, I mean snd_card_free(), which you changed from
> snd_card_free_in_thread().  Someone has to release the
> files
> (i.e. call release fops eventually) before clean up
> everything via
> snd_card_free().
> 
Erm, you mean the close (for
snd_pcm_ops_t,snd_rawmidi_ops_t) and release (for
snd_hwdep_ops_t) callbacks then? They should all be called
before snd_card_free()?
in other words:
<snip>
int snd_card_free_in_thread(snd_card_t * card)
{
	if (card->files == NULL) {
		snd_card_free(card);
		return 0;
	}
</snip>
the "(card->files == NULL)" MUST be true before
snd_card_free() can be called safely?

Karsten


	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

--- Takashi Iwai <tiwai@suse.de> wrote:
> At Fri, 15 Apr 2005 16:28:54 +0200 (CEST),
> karsten wiese wrote:
> > 
> > 
> > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> > > karsten wiese wrote:
> > > > 
> > > > 
> > > > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > > > On Thu, 2005-04-14 at 20:59 +0200, karsten wiese
> > > wrote:
> > > > > > Hi Clemens & Takashi
> > > > > > 
> > > > > > Please commit the patch.
> > > > > 
> > > > > Is the ALSA in kernel 2.6.11 affected?  If so,
> this
> > > > 
> > > > Think so, yes. The keyboard dead effect + described
> > > oops
> > > > has been around for quiet a long time.
> > > > (There is a lkml mail from yesterday about an
> > > usb-cellphone
> > > > setup,
> > > > where the same oops + dead keyboard happens also:
> > > >
> > >
> >
>
http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > > > )
> > > > There is also an alsa-bug with snd-usb-audio in the
> > > > bugbase.
> > > > Clemens seams not to be convinced about this
> aproach
> > > yet.
> > > > maybe 'cause the comment of snd_card_free_in_thread
> > > > explicitly states that its the one to be called for
> > > > hotplug:
> > > > /**
> > > >  *  snd_card_free_in_thread - call snd_card_free()
> in
> > > > thread
> > > >  *  @card: soundcard structure
> > > >  *
> > > >  *  This function schedules the call of
> snd_card_free()
> > > > function in a
> > > >  *  work queue.  When all devices are released
> > > (non-busy),
> > > > the work
> > > >  *  is woken up and calls snd_card_free().
> > > >  *
> > > >  *  When a card can be disconnected at any time by
> > > hotplug
> > > > service,
> > > >  *  this function should be used in disconnect (or
> > > detach)
> > > > callback
> > > >  *  instead of calling snd_card_free() directly.
> > > >  *  
> > > >  *  Returns - zero otherwise a negative error code
> if
> > > the
> > > > start of thread failed.
> > > >  */
> > > > 
> > > > Also I remember , that snd_card_free didn't work ok
> > > when i
> > > > used it in the first implementations of
> usx2y-module
> > > some 2
> > > > years ago.
> > > > but that was on kernel 2.4.something.......
> > > > maybe above comment is from kernel 2.4 also and the
> > > hotplug
> > > > system has changed ? anybody?
> > > 
> > > Might be.  To be sure, check whether the release fops
> is
> > > called
> > > properly before you calling snd_card_free().
> > 
> > You mean snd_card_disconnect()?
> > That's called with the same argument and before
> > snd_card_free().
> 
> No, I mean snd_card_free(), which you changed from
> snd_card_free_in_thread().  Someone has to release the
> files
> (i.e. call release fops eventually) before clean up
> everything via
> snd_card_free().
> 
Erm, you mean the close (for
snd_pcm_ops_t,snd_rawmidi_ops_t) and release (for
snd_hwdep_ops_t) callbacks then? They should all be called
before snd_card_free()?
in other words:
<snip>
int snd_card_free_in_thread(snd_card_t * card)
{
	if (card->files == NULL) {
		snd_card_free(card);
		return 0;
	}
</snip>
the "(card->files == NULL)" MUST be true before
snd_card_free() can be called safely?

Karsten


	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Fri, 15 Apr 2005 17:14:40 +0200 (CEST),
karsten wiese wrote:
> 
> 
> --- Takashi Iwai <tiwai@suse.de> wrote:
> > At Fri, 15 Apr 2005 16:28:54 +0200 (CEST),
> > karsten wiese wrote:
> > > 
> > > 
> > > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > > At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> > > > karsten wiese wrote:
> > > > > 
> > > > > 
> > > > > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > > > > On Thu, 2005-04-14 at 20:59 +0200, karsten wiese
> > > > wrote:
> > > > > > > Hi Clemens & Takashi
> > > > > > > 
> > > > > > > Please commit the patch.
> > > > > > 
> > > > > > Is the ALSA in kernel 2.6.11 affected?  If so,
> > this
> > > > > 
> > > > > Think so, yes. The keyboard dead effect + described
> > > > oops
> > > > > has been around for quiet a long time.
> > > > > (There is a lkml mail from yesterday about an
> > > > usb-cellphone
> > > > > setup,
> > > > > where the same oops + dead keyboard happens also:
> > > > >
> > > >
> > >
> >
> http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > > > > )
> > > > > There is also an alsa-bug with snd-usb-audio in the
> > > > > bugbase.
> > > > > Clemens seams not to be convinced about this
> > aproach
> > > > yet.
> > > > > maybe 'cause the comment of snd_card_free_in_thread
> > > > > explicitly states that its the one to be called for
> > > > > hotplug:
> > > > > /**
> > > > >  *  snd_card_free_in_thread - call snd_card_free()
> > in
> > > > > thread
> > > > >  *  @card: soundcard structure
> > > > >  *
> > > > >  *  This function schedules the call of
> > snd_card_free()
> > > > > function in a
> > > > >  *  work queue.  When all devices are released
> > > > (non-busy),
> > > > > the work
> > > > >  *  is woken up and calls snd_card_free().
> > > > >  *
> > > > >  *  When a card can be disconnected at any time by
> > > > hotplug
> > > > > service,
> > > > >  *  this function should be used in disconnect (or
> > > > detach)
> > > > > callback
> > > > >  *  instead of calling snd_card_free() directly.
> > > > >  *  
> > > > >  *  Returns - zero otherwise a negative error code
> > if
> > > > the
> > > > > start of thread failed.
> > > > >  */
> > > > > 
> > > > > Also I remember , that snd_card_free didn't work ok
> > > > when i
> > > > > used it in the first implementations of
> > usx2y-module
> > > > some 2
> > > > > years ago.
> > > > > but that was on kernel 2.4.something.......
> > > > > maybe above comment is from kernel 2.4 also and the
> > > > hotplug
> > > > > system has changed ? anybody?
> > > > 
> > > > Might be.  To be sure, check whether the release fops
> > is
> > > > called
> > > > properly before you calling snd_card_free().
> > > 
> > > You mean snd_card_disconnect()?
> > > That's called with the same argument and before
> > > snd_card_free().
> > 
> > No, I mean snd_card_free(), which you changed from
> > snd_card_free_in_thread().  Someone has to release the
> > files
> > (i.e. call release fops eventually) before clean up
> > everything via
> > snd_card_free().
> > 
> Erm, you mean the close (for
> snd_pcm_ops_t,snd_rawmidi_ops_t) and release (for
> snd_hwdep_ops_t) callbacks then?

I meant the "file operations".  Anyway, the callback of PCM or rawmidi
will be called from the release callback of fops, though...

>  They should all be called
> before snd_card_free()?

Yes.

> in other words:
> <snip>
> int snd_card_free_in_thread(snd_card_t * card)
> {
> 	if (card->files == NULL) {
> 		snd_card_free(card);
> 		return 0;
> 	}
> </snip>
> the "(card->files == NULL)" MUST be true before
> snd_card_free() can be called safely?

Yes, exactly.
Otherwise, snd_card_free() removes the device entry, and the remainig
files will have no way to release themselves after that.


Takashi


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

 --- Takashi Iwai <tiwai@suse.de> schrieb: 
> At Fri, 15 Apr 2005 17:14:40 +0200 (CEST),
> karsten wiese wrote:
> > 
> > 
> > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > At Fri, 15 Apr 2005 16:28:54 +0200 (CEST),
> > > karsten wiese wrote:
> > > > 
> > > > 
> > > > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > > > At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> > > > > karsten wiese wrote:
> > > > > > 
> > > > > > 
> > > > > > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > > > > > On Thu, 2005-04-14 at 20:59 +0200, karsten
> wiese
> > > > > wrote:
> > > > > > > > Hi Clemens & Takashi
> > > > > > > > 
> > > > > > > > Please commit the patch.
> > > > > > > 
> > > > > > > Is the ALSA in kernel 2.6.11 affected?  If
> so,
> > > this
> > > > > > 
> > > > > > Think so, yes. The keyboard dead effect +
> described
> > > > > oops
> > > > > > has been around for quiet a long time.
> > > > > > (There is a lkml mail from yesterday about an
> > > > > usb-cellphone
> > > > > > setup,
> > > > > > where the same oops + dead keyboard happens
> also:
> > > > > >
> > > > >
> > > >
> > >
> >
>
http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > > > > > )
> > > > > > There is also an alsa-bug with snd-usb-audio in
> the
> > > > > > bugbase.
> > > > > > Clemens seams not to be convinced about this
> > > aproach
> > > > > yet.
> > > > > > maybe 'cause the comment of
> snd_card_free_in_thread
> > > > > > explicitly states that its the one to be called
> for
> > > > > > hotplug:
> > > > > > /**
> > > > > >  *  snd_card_free_in_thread - call
> snd_card_free()
> > > in
> > > > > > thread
> > > > > >  *  @card: soundcard structure
> > > > > >  *
> > > > > >  *  This function schedules the call of
> > > snd_card_free()
> > > > > > function in a
> > > > > >  *  work queue.  When all devices are released
> > > > > (non-busy),
> > > > > > the work
> > > > > >  *  is woken up and calls snd_card_free().
> > > > > >  *
> > > > > >  *  When a card can be disconnected at any time
> by
> > > > > hotplug
> > > > > > service,
> > > > > >  *  this function should be used in disconnect
> (or
> > > > > detach)
> > > > > > callback
> > > > > >  *  instead of calling snd_card_free()
> directly.
> > > > > >  *  
> > > > > >  *  Returns - zero otherwise a negative error
> code
> > > if
> > > > > the
> > > > > > start of thread failed.
> > > > > >  */
> > > > > > 
> > > > > > Also I remember , that snd_card_free didn't
> work ok
> > > > > when i
> > > > > > used it in the first implementations of
> > > usx2y-module
> > > > > some 2
> > > > > > years ago.
> > > > > > but that was on kernel 2.4.something.......
> > > > > > maybe above comment is from kernel 2.4 also and
> the
> > > > > hotplug
> > > > > > system has changed ? anybody?
> > > > > 
> > > > > Might be.  To be sure, check whether the release
> fops
> > > is
> > > > > called
> > > > > properly before you calling snd_card_free().
> > > > 
> > > > You mean snd_card_disconnect()?
> > > > That's called with the same argument and before
> > > > snd_card_free().
> > > 
> > > No, I mean snd_card_free(), which you changed from
> > > snd_card_free_in_thread().  Someone has to release
> the
> > > files
> > > (i.e. call release fops eventually) before clean up
> > > everything via
> > > snd_card_free().
> > > 
> > Erm, you mean the close (for
> > snd_pcm_ops_t,snd_rawmidi_ops_t) and release (for
> > snd_hwdep_ops_t) callbacks then?
> 
> I meant the "file operations".  Anyway, the callback of
> PCM or rawmidi
> will be called from the release callback of fops,
> though...
> 
> >  They should all be called
> > before snd_card_free()?
> 
> Yes.
> 
> > in other words:
> > <snip>
> > int snd_card_free_in_thread(snd_card_t * card)
> > {
> > 	if (card->files == NULL) {
> > 		snd_card_free(card);
> > 		return 0;
> > 	}
> > </snip>
> > the "(card->files == NULL)" MUST be true before
> > snd_card_free() can be called safely?
> 
> Yes, exactly.
> Otherwise, snd_card_free() removes the device entry, and
> the remainig
> files will have no way to release themselves after that.

Just found that snd_card_free() also waits for all files
being released:
<snip>
int snd_card_free(snd_card_t * card)
{
	struct snd_shutdown_f_ops *s_f_ops;

	if (card == NULL)
		return -EINVAL;
	write_lock(&snd_card_rwlock);
	snd_cards[card->number] = NULL;
	write_unlock(&snd_card_rwlock);

#ifdef CONFIG_PM
	wake_up(&card->power_sleep);
#ifdef CONFIG_SND_GENERIC_PM
	if (card->pm_dev) {
		snd_generic_device_unregister(card->pm_dev);
		card->pm_dev = NULL;
	}
#endif
#endif

	/* wait, until all devices are ready for the free
operation */
	wait_event(card->shutdown_sleep, card->files == NULL);
</snip>

(-: You apply my patch now?
It really works here better than before.
If you still doubt my patch is correct:
Please line out a testcase, where a bug might happen, that
would not happen without the patch.
So far I tested jackd and "arecord | aplay" ok with
unplugging the usb cable, switching of usx2y and "rmmod
uhci_hcd". several times. no oops. without the patch the
oops came every 3rd test or so.

Karsten


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Fri, 15 Apr 2005 17:45:44 +0200 (CEST),
karsten wiese wrote:
> 
>  --- Takashi Iwai <tiwai@suse.de> schrieb: 
> > At Fri, 15 Apr 2005 17:14:40 +0200 (CEST),
> > karsten wiese wrote:
> > > 
> > > 
> > > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > > At Fri, 15 Apr 2005 16:28:54 +0200 (CEST),
> > > > karsten wiese wrote:
> > > > > 
> > > > > 
> > > > > --- Takashi Iwai <tiwai@suse.de> wrote:
> > > > > > At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> > > > > > karsten wiese wrote:
> > > > > > > 
> > > > > > > 
> > > > > > > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > > > > > > On Thu, 2005-04-14 at 20:59 +0200, karsten
> > wiese
> > > > > > wrote:
> > > > > > > > > Hi Clemens & Takashi
> > > > > > > > > 
> > > > > > > > > Please commit the patch.
> > > > > > > > 
> > > > > > > > Is the ALSA in kernel 2.6.11 affected?  If
> > so,
> > > > this
> > > > > > > 
> > > > > > > Think so, yes. The keyboard dead effect +
> > described
> > > > > > oops
> > > > > > > has been around for quiet a long time.
> > > > > > > (There is a lkml mail from yesterday about an
> > > > > > usb-cellphone
> > > > > > > setup,
> > > > > > > where the same oops + dead keyboard happens
> > also:
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
> http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > > > > > > )
> > > > > > > There is also an alsa-bug with snd-usb-audio in
> > the
> > > > > > > bugbase.
> > > > > > > Clemens seams not to be convinced about this
> > > > aproach
> > > > > > yet.
> > > > > > > maybe 'cause the comment of
> > snd_card_free_in_thread
> > > > > > > explicitly states that its the one to be called
> > for
> > > > > > > hotplug:
> > > > > > > /**
> > > > > > >  *  snd_card_free_in_thread - call
> > snd_card_free()
> > > > in
> > > > > > > thread
> > > > > > >  *  @card: soundcard structure
> > > > > > >  *
> > > > > > >  *  This function schedules the call of
> > > > snd_card_free()
> > > > > > > function in a
> > > > > > >  *  work queue.  When all devices are released
> > > > > > (non-busy),
> > > > > > > the work
> > > > > > >  *  is woken up and calls snd_card_free().
> > > > > > >  *
> > > > > > >  *  When a card can be disconnected at any time
> > by
> > > > > > hotplug
> > > > > > > service,
> > > > > > >  *  this function should be used in disconnect
> > (or
> > > > > > detach)
> > > > > > > callback
> > > > > > >  *  instead of calling snd_card_free()
> > directly.
> > > > > > >  *  
> > > > > > >  *  Returns - zero otherwise a negative error
> > code
> > > > if
> > > > > > the
> > > > > > > start of thread failed.
> > > > > > >  */
> > > > > > > 
> > > > > > > Also I remember , that snd_card_free didn't
> > work ok
> > > > > > when i
> > > > > > > used it in the first implementations of
> > > > usx2y-module
> > > > > > some 2
> > > > > > > years ago.
> > > > > > > but that was on kernel 2.4.something.......
> > > > > > > maybe above comment is from kernel 2.4 also and
> > the
> > > > > > hotplug
> > > > > > > system has changed ? anybody?
> > > > > > 
> > > > > > Might be.  To be sure, check whether the release
> > fops
> > > > is
> > > > > > called
> > > > > > properly before you calling snd_card_free().
> > > > > 
> > > > > You mean snd_card_disconnect()?
> > > > > That's called with the same argument and before
> > > > > snd_card_free().
> > > > 
> > > > No, I mean snd_card_free(), which you changed from
> > > > snd_card_free_in_thread().  Someone has to release
> > the
> > > > files
> > > > (i.e. call release fops eventually) before clean up
> > > > everything via
> > > > snd_card_free().
> > > > 
> > > Erm, you mean the close (for
> > > snd_pcm_ops_t,snd_rawmidi_ops_t) and release (for
> > > snd_hwdep_ops_t) callbacks then?
> > 
> > I meant the "file operations".  Anyway, the callback of
> > PCM or rawmidi
> > will be called from the release callback of fops,
> > though...
> > 
> > >  They should all be called
> > > before snd_card_free()?
> > 
> > Yes.
> > 
> > > in other words:
> > > <snip>
> > > int snd_card_free_in_thread(snd_card_t * card)
> > > {
> > > 	if (card->files == NULL) {
> > > 		snd_card_free(card);
> > > 		return 0;
> > > 	}
> > > </snip>
> > > the "(card->files == NULL)" MUST be true before
> > > snd_card_free() can be called safely?
> > 
> > Yes, exactly.
> > Otherwise, snd_card_free() removes the device entry, and
> > the remainig
> > files will have no way to release themselves after that.
> 
> Just found that snd_card_free() also waits for all files
> being released:
> <snip>
> int snd_card_free(snd_card_t * card)
> {
> 	struct snd_shutdown_f_ops *s_f_ops;
> 
> 	if (card == NULL)
> 		return -EINVAL;
> 	write_lock(&snd_card_rwlock);
> 	snd_cards[card->number] = NULL;
> 	write_unlock(&snd_card_rwlock);
> 
> #ifdef CONFIG_PM
> 	wake_up(&card->power_sleep);
> #ifdef CONFIG_SND_GENERIC_PM
> 	if (card->pm_dev) {
> 		snd_generic_device_unregister(card->pm_dev);
> 		card->pm_dev = NULL;
> 	}
> #endif
> #endif
> 
> 	/* wait, until all devices are ready for the free
> operation */
> 	wait_event(card->shutdown_sleep, card->files == NULL);
> </snip>
> 
> (-: You apply my patch now?

No, not enough :)  This is exactly why snd_card_free_in_thread() was
introduced.

As I asked, please make sure that the files are all released before
snd_card_free().  Otherwise the disconnect callback hangs at this
point because of this check.

(IOW, the fact that calling snd_card_free doesn't hang implies that
 the files are released.  But it's better to check this explicitly.)


Takashi


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Clemens Ladisch]

Takashi Iwai wrote:
> karsten wiese wrote:
> > (-: You apply my patch now?
>
> No, not enough :)  This is exactly why snd_card_free_in_thread() was
> introduced.
>
> As I asked, please make sure that the files are all released before
> snd_card_free().  Otherwise the disconnect callback hangs at this
> point because of this check.

Most programs close the device file when an error occurs, so you have
to test it with something like this to keep the file open:

#include <stdio.h>
int main()
{
  fopen("/dev/snd/hwC1D0", "rb");
  getchar();
}


Regards,
Clemens



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Clemens Ladisch]

karsten wiese wrote:
> Erm, you mean the close (for snd_pcm_ops_t,snd_rawmidi_ops_t) and
> release (for snd_hwdep_ops_t) callbacks then? They should all be
> called before snd_card_free()?
> in other words: ...
> the "(card->files == NULL)" MUST be true before
> snd_card_free() can be called safely?

Yes.  The files cannot be removed while they are still opened by an
application.


To make snd_card_free_in_thread() work, We could increase the
reference count of card->dev as long as snd_card_free_in_thread() is
still being executed, but I don't know what happens to the rest of the
kernel if the device cleanup is delayed indefinitely.

Another possibility would be to just not assign the USB device to the
sound card and its files, i.e., just remove the snd_card_set_dev()
call from the USB drivers.  This shouldn't affect anything else
because card->dev isn't used except for the device files.


Regards,
Clemens



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

--- Takashi Iwai <tiwai@suse.de> wrote:
> At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> karsten wiese wrote:
> > 
> > 
> > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > On Thu, 2005-04-14 at 20:59 +0200, karsten wiese
> wrote:
> > > > Hi Clemens & Takashi
> > > > 
> > > > Please commit the patch.
> > > 
> > > Is the ALSA in kernel 2.6.11 affected?  If so, this
> > 
> > Think so, yes. The keyboard dead effect + described
> oops
> > has been around for quiet a long time.
> > (There is a lkml mail from yesterday about an
> usb-cellphone
> > setup,
> > where the same oops + dead keyboard happens also:
> >
>
http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > )
> > There is also an alsa-bug with snd-usb-audio in the
> > bugbase.
> > Clemens seams not to be convinced about this aproach
> yet.
> > maybe 'cause the comment of snd_card_free_in_thread
> > explicitly states that its the one to be called for
> > hotplug:
> > /**
> >  *  snd_card_free_in_thread - call snd_card_free() in
> > thread
> >  *  @card: soundcard structure
> >  *
> >  *  This function schedules the call of snd_card_free()
> > function in a
> >  *  work queue.  When all devices are released
> (non-busy),
> > the work
> >  *  is woken up and calls snd_card_free().
> >  *
> >  *  When a card can be disconnected at any time by
> hotplug
> > service,
> >  *  this function should be used in disconnect (or
> detach)
> > callback
> >  *  instead of calling snd_card_free() directly.
> >  *  
> >  *  Returns - zero otherwise a negative error code if
> the
> > start of thread failed.
> >  */
> > 
> > Also I remember , that snd_card_free didn't work ok
> when i
> > used it in the first implementations of usx2y-module
> some 2
> > years ago.
> > but that was on kernel 2.4.something.......
> > maybe above comment is from kernel 2.4 also and the
> hotplug
> > system has changed ? anybody?
> 
> Might be.  To be sure, check whether the release fops is
> called
> properly before you calling snd_card_free().

You mean snd_card_disconnect()?
That's called with the same argument and before
snd_card_free().

Karsten


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Fri, 15 Apr 2005 16:28:54 +0200 (CEST),
karsten wiese wrote:
> 
> 
> --- Takashi Iwai <tiwai@suse.de> wrote:
> > At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
> > karsten wiese wrote:
> > > 
> > > 
> > > --- Lee Revell <rlrevell@joe-job.com> wrote:
> > > > On Thu, 2005-04-14 at 20:59 +0200, karsten wiese
> > wrote:
> > > > > Hi Clemens & Takashi
> > > > > 
> > > > > Please commit the patch.
> > > > 
> > > > Is the ALSA in kernel 2.6.11 affected?  If so, this
> > > 
> > > Think so, yes. The keyboard dead effect + described
> > oops
> > > has been around for quiet a long time.
> > > (There is a lkml mail from yesterday about an
> > usb-cellphone
> > > setup,
> > > where the same oops + dead keyboard happens also:
> > >
> >
> http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> > > )
> > > There is also an alsa-bug with snd-usb-audio in the
> > > bugbase.
> > > Clemens seams not to be convinced about this aproach
> > yet.
> > > maybe 'cause the comment of snd_card_free_in_thread
> > > explicitly states that its the one to be called for
> > > hotplug:
> > > /**
> > >  *  snd_card_free_in_thread - call snd_card_free() in
> > > thread
> > >  *  @card: soundcard structure
> > >  *
> > >  *  This function schedules the call of snd_card_free()
> > > function in a
> > >  *  work queue.  When all devices are released
> > (non-busy),
> > > the work
> > >  *  is woken up and calls snd_card_free().
> > >  *
> > >  *  When a card can be disconnected at any time by
> > hotplug
> > > service,
> > >  *  this function should be used in disconnect (or
> > detach)
> > > callback
> > >  *  instead of calling snd_card_free() directly.
> > >  *  
> > >  *  Returns - zero otherwise a negative error code if
> > the
> > > start of thread failed.
> > >  */
> > > 
> > > Also I remember , that snd_card_free didn't work ok
> > when i
> > > used it in the first implementations of usx2y-module
> > some 2
> > > years ago.
> > > but that was on kernel 2.4.something.......
> > > maybe above comment is from kernel 2.4 also and the
> > hotplug
> > > system has changed ? anybody?
> > 
> > Might be.  To be sure, check whether the release fops is
> > called
> > properly before you calling snd_card_free().
> 
> You mean snd_card_disconnect()?
> That's called with the same argument and before
> snd_card_free().

No, I mean snd_card_free(), which you changed from
snd_card_free_in_thread().  Someone has to release the files
(i.e. call release fops eventually) before clean up everything via
snd_card_free().


Takashi


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

--- Lee Revell <rlrevell@joe-job.com> wrote:
> On Thu, 2005-04-14 at 20:59 +0200, karsten wiese wrote:
> > Hi Clemens & Takashi
> > 
> > Please commit the patch.
> 
> Is the ALSA in kernel 2.6.11 affected?  If so, this

Think so, yes. The keyboard dead effect + described oops
has been around for quiet a long time.
(There is a lkml mail from yesterday about an usb-cellphone
setup,
where the same oops + dead keyboard happens also:
http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
)
There is also an alsa-bug with snd-usb-audio in the
bugbase.
Clemens seams not to be convinced about this aproach yet.
maybe 'cause the comment of snd_card_free_in_thread
explicitly states that its the one to be called for
hotplug:
/**
 *  snd_card_free_in_thread - call snd_card_free() in
thread
 *  @card: soundcard structure
 *
 *  This function schedules the call of snd_card_free()
function in a
 *  work queue.  When all devices are released (non-busy),
the work
 *  is woken up and calls snd_card_free().
 *
 *  When a card can be disconnected at any time by hotplug
service,
 *  this function should be used in disconnect (or detach)
callback
 *  instead of calling snd_card_free() directly.
 *  
 *  Returns - zero otherwise a negative error code if the
start of thread failed.
 */

Also I remember , that snd_card_free didn't work ok when i
used it in the first implementations of usx2y-module some 2
years ago.
but that was on kernel 2.4.something.......
maybe above comment is from kernel 2.4 also and the hotplug
system has changed ? anybody?

rgds,
Karsten 


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Takashi Iwai]

At Fri, 15 Apr 2005 15:59:25 +0200 (CEST),
karsten wiese wrote:
> 
> 
> --- Lee Revell <rlrevell@joe-job.com> wrote:
> > On Thu, 2005-04-14 at 20:59 +0200, karsten wiese wrote:
> > > Hi Clemens & Takashi
> > > 
> > > Please commit the patch.
> > 
> > Is the ALSA in kernel 2.6.11 affected?  If so, this
> 
> Think so, yes. The keyboard dead effect + described oops
> has been around for quiet a long time.
> (There is a lkml mail from yesterday about an usb-cellphone
> setup,
> where the same oops + dead keyboard happens also:
> http://marc.theaimsgroup.com/?i=<20050414230621.49663f75.vsu%20()%20altlinux%20!%20ru>
> )
> There is also an alsa-bug with snd-usb-audio in the
> bugbase.
> Clemens seams not to be convinced about this aproach yet.
> maybe 'cause the comment of snd_card_free_in_thread
> explicitly states that its the one to be called for
> hotplug:
> /**
>  *  snd_card_free_in_thread - call snd_card_free() in
> thread
>  *  @card: soundcard structure
>  *
>  *  This function schedules the call of snd_card_free()
> function in a
>  *  work queue.  When all devices are released (non-busy),
> the work
>  *  is woken up and calls snd_card_free().
>  *
>  *  When a card can be disconnected at any time by hotplug
> service,
>  *  this function should be used in disconnect (or detach)
> callback
>  *  instead of calling snd_card_free() directly.
>  *  
>  *  Returns - zero otherwise a negative error code if the
> start of thread failed.
>  */
> 
> Also I remember , that snd_card_free didn't work ok when i
> used it in the first implementations of usx2y-module some 2
> years ago.
> but that was on kernel 2.4.something.......
> maybe above comment is from kernel 2.4 also and the hotplug
> system has changed ? anybody?

Might be.  To be sure, check whether the release fops is called
properly before you calling snd_card_free().


Takashi


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

[-- Attachment #1: Type: text/plain, Size: 593 bytes --]

--- Clemens Ladisch <clemens@ladisch.de> wrote:
> karsten wiese wrote:
> > Please commit the patch.
> >
> > - snd_card_free_in_thread((snd_card_t*)ptr);
> > + snd_card_free((snd_card_t*)ptr);
> 
> What happens when you unplug the device while it's still
> playing?

See attached excerpts of /var/log/messages.
they are a bit looong, but critical points are marked with
a comment starting with //.





	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: usx2y-hot-unplug_snd_card_free_in_thread --]
[-- Type: text/x-java; name=usx2y-hot-unplug_snd_card_free_in_thread, Size: 17740 bytes --]

with snd_card_free_in_thread, it oopses:
this is triggered by "rmmod uhci_hcd" while jackd was runnnig.
the effect of unplugging the cable is nearly identicall.
comments begin with //

Apr 13 13:22:21 p4 kernel: [4299685.996000] uhci_hcd 0000:00:10.1: remove, state 1
Apr 13 13:22:21 p4 kernel: [4299685.996000] uhci_hcd 0000:00:10.1: roothub graceful disconnect
// usx2y is connnected to usb3:
Apr 13 13:22:21 p4 kernel: [4299685.996000] usb usb3: USB disconnect, address 1
Apr 13 13:22:22 p4 kernel: [4299685.996000] usb 3-2: USB disconnect, address 2
Apr 13 13:22:22 p4 kernel: [4299685.996000] usb 3-2: usb_disable_device nuking all URBs
Apr 13 13:22:22 p4 kernel: [4299685.996000] uhci_hcd 0000:00:10.1: shutdown urb f75cfd80 pipe 40020280 ep4in-intr
Apr 13 13:22:22 p4 kernel: [4299685.996000] uhci_hcd 0000:00:10.1: shutdown urb ed73f480 pipe 40020280 ep4in-intr
// usx2y notices disconnect.....
Apr 13 13:22:22 p4 kernel: [4299685.996000] urb status -108
Apr 13 13:22:22 p4 kernel: [4299685.996000] usb_submit_urb() returned -19
Apr 13 13:22:22 p4 kernel: [4299685.996000] 
// and stops streaming
Apr 13 13:22:22 p4 kernel: [4299685.996000] snd_usX2Y_pcm_trigger(STOP)
Apr 13 13:22:22 p4 kernel: [4299685.996000] snd_usX2Y_pcm_trigger(STOP)
Apr 13 13:22:22 p4 kernel: [4299685.996000] 0 status=-115 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.996000] 1 status=0 start_frame=810
Apr 13 13:22:22 p4 kernel: [4299685.996000] 0 status=0 start_frame=808
Apr 13 13:22:22 p4 kernel: [4299685.996000] 1 status=-115 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.996000] 0 status=0 start_frame=808
Apr 13 13:22:22 p4 kernel: [4299685.996000] 1 status=-115 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.996000] 0 edd64d80 state=4
Apr 13 13:22:22 p4 kernel: [4299685.996000] 1 edd64f80 state=4
Apr 13 13:22:22 p4 kernel: [4299685.996000] 3 edd64e40 state=4
Apr 13 13:22:22 p4 kernel: [4299685.996000] snd_usX2Y_pcm_prepare(e8a8fd00)
Apr 13 13:22:22 p4 kernel: [4299685.996000] usX2Y_subs_prepare(edd64f80) ep=8 urb0=c5f89b00 urb1=c5f89b80
Apr 13 13:22:22 p4 kernel: [4299685.996000] starting capture pipe for self
Apr 13 13:22:22 p4 kernel: [4299685.997000] hcd_frame=1834 ep=8in status=-2 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.997000] hcd_frame=1834 ep=10out status=0 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.997000] hcd_frame=1834 ep=10in status=0 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.997000] cannot usb_submit_urb() for urb 0, err = -19
Apr 13 13:22:22 p4 kernel: [4299685.997000] 0 status=0 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.997000] 1 status=0 start_frame=810
Apr 13 13:22:22 p4 kernel: [4299685.997000] 0 status=0 start_frame=-1
Apr 13 13:22:22 p4 kernel: [4299685.997000] 1 status=-2 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.997000] 0 status=0 start_frame=808
Apr 13 13:22:22 p4 kernel: [4299685.997000] 1 status=0 start_frame=809
Apr 13 13:22:22 p4 kernel: [4299685.997000] 0 edd64d80 state=0
Apr 13 13:22:22 p4 kernel: [4299685.997000] 1 edd64f80 state=3
Apr 13 13:22:22 p4 kernel: [4299685.997000] 3 edd64e40 state=0
// snd_card_free_in_thread starts working ok
Apr 13 13:22:22 p4 kernel: [4299685.997000] usb 3-2: unregistering interface 3-2:1.0
Apr 13 13:22:22 p4 kernel: [4299685.997000] CLASS: Unregistering class device. ID = 'mixer1'
Apr 13 13:22:22 p4 kernel: [4299685.997000] fill_kobj_path: path = '/class/sound/mixer1'
Apr 13 13:22:22 p4 kernel: [4299685.997000] class_hotplug - name = mixer1
Apr 13 13:22:22 p4 kernel: [4299686.113000] kobject mixer1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.113000] device class 'mixer1': release.
Apr 13 13:22:22 p4 kernel: [4299686.113000] CLASS: Unregistering class device. ID = 'adsp1'
Apr 13 13:22:22 p4 kernel: [4299686.113000] fill_kobj_path: path = '/class/sound/adsp1'
Apr 13 13:22:22 p4 kernel: [4299686.113000] class_hotplug - name = adsp1
Apr 13 13:22:22 p4 kernel: [4299686.123000] kobject adsp1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.123000] device class 'adsp1': release.
Apr 13 13:22:22 p4 kernel: [4299686.123000] CLASS: Unregistering class device. ID = 'dsp1'
Apr 13 13:22:22 p4 kernel: [4299686.123000] fill_kobj_path: path = '/class/sound/dsp1'
Apr 13 13:22:22 p4 kernel: [4299686.123000] class_hotplug - name = dsp1
Apr 13 13:22:22 p4 kernel: [4299686.133000] kobject dsp1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.133000] device class 'dsp1': release.
Apr 13 13:22:22 p4 kernel: [4299686.133000] CLASS: Unregistering class device. ID = 'audio1'
Apr 13 13:22:22 p4 kernel: [4299686.133000] fill_kobj_path: path = '/class/sound/audio1'
Apr 13 13:22:22 p4 kernel: [4299686.133000] class_hotplug - name = audio1
Apr 13 13:22:22 p4 kernel: [4299686.143000] kobject audio1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.143000] device class 'audio1': release.
Apr 13 13:22:22 p4 kernel: [4299686.143000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2/3-2:1.0'
Apr 13 13:22:22 p4 kernel: [4299686.143000] usb 3-2:1.0: hotplug
Apr 13 13:22:22 p4 kernel: [4299686.161000] kobject 3-2:1.0: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.161000] usb 3-2: unregistering device
Apr 13 13:22:22 p4 kernel: [4299686.161000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'

// since its snd_card_free_IN_THREAD, usX2Y_usb_disconnect() has finished and hotplug thinks,
// it may continue cleaning up:
Apr 13 13:22:22 p4 kernel: [4299686.161000] usb 3-2: hotplug
Apr 13 13:22:22 p4 kernel: [4299686.174000] usb usb3: usb_disable_device nuking all URBs
Apr 13 13:22:22 p4 kernel: [4299686.174000] uhci_hcd 0000:00:10.1: shutdown urb e8336b00 pipe 40408180 ep1in-intr
Apr 13 13:22:22 p4 kernel: [4299686.174000] usb usb3: unregistering interface 3-0:1.0
Apr 13 13:22:22 p4 kernel: [4299686.174000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-0:1.0'
Apr 13 13:22:22 p4 kernel: [4299686.174000] usb 3-0:1.0: hotplug
Apr 13 13:22:22 p4 kernel: [4299686.191000] kobject 3-0:1.0: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.191000] usb usb3: unregistering device
Apr 13 13:22:22 p4 kernel: [4299686.191000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3'
Apr 13 13:22:22 p4 kernel: [4299686.191000] usb usb3: hotplug
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_usbpcm_hw_free(e8a8fb80)
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_urbs_release() 10
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_urbs_release() 8
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_urbs_release() 10
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_usbpcm_hw_free(e8a8fd00)
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_urbs_release() 8
Apr 13 13:22:22 p4 kernel: [4299686.193000] snd_usX2Y_urbs_release() 10
Apr 13 13:22:22 p4 kernel: [4299686.193000] CLASS: Unregistering class device. ID = 'controlC1'
Apr 13 13:22:22 p4 kernel: [4299686.193000] fill_kobj_path: path = '/class/sound/controlC1'
Apr 13 13:22:22 p4 kernel: [4299686.193000] class_hotplug - name = controlC1
Apr 13 13:22:22 p4 kernel: [4299686.193000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.198000] kobject controlC1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.198000] device class 'controlC1': release.
Apr 13 13:22:22 p4 kernel: [4299686.198000] CLASS: Unregistering class device. ID = 'pcmC1D2p'
Apr 13 13:22:22 p4 kernel: [4299686.198000] fill_kobj_path: path = '/class/sound/pcmC1D2p'
Apr 13 13:22:22 p4 kernel: [4299686.198000] class_hotplug - name = pcmC1D2p
Apr 13 13:22:22 p4 kernel: [4299686.198000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.199000] kobject pcmC1D2p: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.199000] device class 'pcmC1D2p': release.
Apr 13 13:22:22 p4 kernel: [4299686.199000] CLASS: Unregistering class device. ID = 'pcmC1D2c'
Apr 13 13:22:22 p4 kernel: [4299686.199000] fill_kobj_path: path = '/class/sound/pcmC1D2c'
Apr 13 13:22:22 p4 kernel: [4299686.199000] class_hotplug - name = pcmC1D2c
Apr 13 13:22:22 p4 kernel: [4299686.199000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.203000] kobject pcmC1D2c: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.203000] device class 'pcmC1D2c': release.
Apr 13 13:22:22 p4 kernel: [4299686.203000] CLASS: Unregistering class device. ID = 'hwC1D1'
Apr 13 13:22:22 p4 kernel: [4299686.203000] fill_kobj_path: path = '/class/sound/hwC1D1'
Apr 13 13:22:22 p4 kernel: [4299686.203000] class_hotplug - name = hwC1D1
Apr 13 13:22:22 p4 kernel: [4299686.203000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.207000] kobject hwC1D1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.207000] device class 'hwC1D1': release.
Apr 13 13:22:22 p4 kernel: [4299686.207000] CLASS: Unregistering class device. ID = 'pcmC1D1c'
Apr 13 13:22:22 p4 kernel: [4299686.207000] fill_kobj_path: path = '/class/sound/pcmC1D1c'
Apr 13 13:22:22 p4 kernel: [4299686.207000] class_hotplug - name = pcmC1D1c
Apr 13 13:22:22 p4 kernel: [4299686.207000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.212000] kobject pcmC1D1c: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.212000] device class 'pcmC1D1c': release.
Apr 13 13:22:22 p4 kernel: [4299686.212000] CLASS: Unregistering class device. ID = 'pcmC1D0p'
Apr 13 13:22:22 p4 kernel: [4299686.212000] fill_kobj_path: path = '/class/sound/pcmC1D0p'
Apr 13 13:22:22 p4 kernel: [4299686.212000] class_hotplug - name = pcmC1D0p
Apr 13 13:22:22 p4 kernel: [4299686.212000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.216000] kobject pcmC1D0p: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.216000] device class 'pcmC1D0p': release.
Apr 13 13:22:22 p4 kernel: [4299686.216000] CLASS: Unregistering class device. ID = 'pcmC1D0c'
Apr 13 13:22:22 p4 kernel: [4299686.216000] fill_kobj_path: path = '/class/sound/pcmC1D0c'
Apr 13 13:22:22 p4 kernel: [4299686.216000] class_hotplug - name = pcmC1D0c
Apr 13 13:22:22 p4 kernel: [4299686.216000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:22 p4 kernel: [4299686.218000] kobject pcmC1D0c: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.218000] device class 'pcmC1D0c': release.
Apr 13 13:22:22 p4 kernel: [4299686.218000] CLASS: Unregistering class device. ID = 'midi1'
Apr 13 13:22:22 p4 kernel: [4299686.218000] fill_kobj_path: path = '/class/sound/midi1'
Apr 13 13:22:22 p4 kernel: [4299686.218000] class_hotplug - name = midi1
Apr 13 13:22:22 p4 kernel: [4299686.263000] uhci_hcd 0000:00:10.1: USB bus 3 deregistered
Apr 13 13:22:22 p4 kernel: [4299686.263000] fill_kobj_path: path = '/class/usb_host/usb3'
Apr 13 13:22:22 p4 kernel: [4299686.263000] class_hotplug - name = usb3
Apr 13 13:22:22 p4 kernel: [4299686.263000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1'
Apr 13 13:22:22 p4 kernel: [4299686.319000] kobject midi1: cleaning up
Apr 13 13:22:22 p4 kernel: [4299686.319000] device class 'midi1': release.
Apr 13 13:22:22 p4 kernel: [4299686.319000] CLASS: Unregistering class device. ID = 'dmmidi1'
Apr 13 13:22:22 p4 kernel: [4299686.319000] fill_kobj_path: path = '/class/sound/dmmidi1'
Apr 13 13:22:23 p4 kernel: [4299686.319000] class_hotplug - name = dmmidi1
Apr 13 13:22:23 p4 kernel: [4299686.319000] uhci_hcd 0000:00:10.2: remove, state 1
Apr 13 13:22:23 p4 kernel: [4299686.319000] uhci_hcd 0000:00:10.2: roothub graceful disconnect
Apr 13 13:22:23 p4 kernel: [4299686.319000] usb usb4: USB disconnect, address 1
Apr 13 13:22:23 p4 kernel: [4299686.319000] usb usb4: usb_disable_device nuking all URBs
Apr 13 13:22:23 p4 kernel: [4299686.319000] uhci_hcd 0000:00:10.2: shutdown urb e8336780 pipe 40408180 ep1in-intr
Apr 13 13:22:23 p4 kernel: [4299686.319000] usb usb4: unregistering interface 4-0:1.0
Apr 13 13:22:23 p4 kernel: [4299686.319000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.2/usb4/4-0:1.0'
Apr 13 13:22:23 p4 kernel: [4299686.319000] usb 4-0:1.0: hotplug
Apr 13 13:22:23 p4 kernel: [4299686.327000] kobject dmmidi1: cleaning up
Apr 13 13:22:23 p4 kernel: [4299686.327000] device class 'dmmidi1': release.
Apr 13 13:22:23 p4 kernel: [4299686.327000] CLASS: Unregistering class device. ID = 'midiC1D0'
Apr 13 13:22:23 p4 kernel: [4299686.327000] fill_kobj_path: path = '/class/sound/midiC1D0'
Apr 13 13:22:23 p4 kernel: [4299686.327000] class_hotplug - name = midiC1D0
Apr 13 13:22:23 p4 kernel: [4299686.327000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 13 13:22:23 p4 kernel: [4299686.335000] kobject 4-0:1.0: cleaning up
Apr 13 13:22:23 p4 kernel: [4299686.335000] usb usb4: unregistering device
Apr 13 13:22:23 p4 kernel: [4299686.335000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.2/usb4'
Apr 13 13:22:23 p4 kernel: [4299686.335000] usb usb4: hotplug
Apr 13 13:22:23 p4 kernel: [4299686.353000] kobject midiC1D0: cleaning up
Apr 13 13:22:23 p4 kernel: [4299686.353000] device class 'midiC1D0': release.
Apr 13 13:22:23 p4 kernel: [4299686.353000] CLASS: Unregistering class device. ID = 'hwC1D0'
Apr 13 13:22:23 p4 kernel: [4299686.353000] kobject 3-2: cleaning up
Apr 13 13:22:23 p4 kernel: [4299686.353000] kobject usb3: cleaning up
Apr 13 13:22:23 p4 kernel: [4299686.353000] kobject usb3: cleaning up
Apr 13 13:22:23 p4 kernel: [4299686.353000] device class 'usb3': release.

// usb3 is gone, but snd_card_free_in_thread() hasn't finished yet:
Apr 13 13:22:23 p4 kernel: [4299686.353000] fill_kobj_path: path = '/class/sound/hwC1D0'
Apr 13 13:22:23 p4 kernel: [4299686.353000] class_hotplug - name = hwC1D0
Apr 13 13:22:23 p4 kernel: [4299686.353000] Unable to handle kernel NULL pointer dereference at virtual address 00000000
Apr 13 13:22:23 p4 kernel: [4299686.353000]  printing eip:
Apr 13 13:22:23 p4 kernel: [4299686.353000] c01bd569
Apr 13 13:22:23 p4 kernel: [4299686.353000] *pde = 00000000
Apr 13 13:22:23 p4 kernel: [4299686.353000] Oops: 0000 [#1]
Apr 13 13:22:23 p4 kernel: [4299686.353000] PREEMPT 
Apr 13 13:22:23 p4 kernel: [4299686.353000] Modules linked in: uhci_hcd snd_usb_usx2y snd_seq snd_usb_lib snd_hwdep nfs nvidia agpgart realtime commoncap nfsd exportfs lockd md5 ipv6 autofs4 sunrpc video button ac ehci_hcd i2c_viapro i2c_core snd_via82xx snd_ac97_codec snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd_page_alloc snd_mpu401_uart snd_rawmidi snd_seq_device snd floppy ext3 jbd
Apr 13 13:22:23 p4 kernel: [4299686.353000] CPU:    0
Apr 13 13:22:23 p4 kernel: [4299686.353000] EIP:    0060:[<c01bd569>]    Tainted: P      VLI
Apr 13 13:22:23 p4 kernel: [4299686.353000] EFLAGS: 00010246   (2.6.12-rc2) 
Apr 13 13:22:23 p4 kernel: [4299686.353000] EIP is at get_kobj_path_length+0x19/0x30
Apr 13 13:22:23 p4 kernel: [4299686.353000] eax: 00000000   ebx: 00000000   ecx: ffffffff   edx: e12c7c88
Apr 13 13:22:23 p4 kernel: [4299686.353000] esi: 00000001   edi: 00000000   ebp: f7f09d44   esp: f7f09d38
Apr 13 13:22:23 p4 kernel: [4299686.353000] ds: 007b   es: 007b   ss: 0068
Apr 13 13:22:23 p4 kernel: [4299686.353000] Process events/0 (pid: 3, threadinfo=f7f08000 task=c18da020)
Apr 13 13:22:23 p4 kernel: [4299686.353000] Stack: c0345d40 e12c7c64 e8336418 f7f09d68 c01bd60a e12c7c88 e8336400 e8336418 
Apr 13 13:22:23 p4 kernel: [4299686.353000]        f7f09d68 c0345d40 e12c7c64 e8336418 f7f09dac c0229c19 e12c7c88 000000d0 
Apr 13 13:22:23 p4 kernel: [4299686.353000]        e8336e54 ffffffff fffffffd ffffffff c0345d40 e8336400 00000000 e8336e08 
Apr 13 13:22:23 p4 kernel: [4299686.353000] Call Trace:
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c010402f>] show_stack+0x7f/0xa0
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c01041ca>] show_registers+0x15a/0x1d0
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c01043d0>] die+0xf0/0x180
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c0115914>] do_page_fault+0x464/0x66a
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c0103c4b>] error_code+0x4f/0x54
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c01bd60a>] kobject_get_path+0x1a/0x70
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c0229c19>] class_hotplug+0xa9/0x250
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c01be40a>] kobject_hotplug+0x2ba/0x2e0
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c01bd9ca>] kobject_del+0x1a/0x30
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c022a1a1>] class_device_del+0xc1/0xd0
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c022a1d5>] class_device_unregister+0x25/0x40
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<f885059f>] snd_unregister_device+0x7f/0xd0 [snd]
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<f8a39ac2>] snd_hwdep_dev_unregister+0x52/0xb0 [snd_hwdep]
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<f8855bb6>] snd_device_free+0xa6/0xc0 [snd]
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<f8855de4>] snd_device_free_all+0x54/0x70 [snd]
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<f8850d8d>] snd_card_free+0x9d/0x2b0 [snd]
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<f8850ff7>] snd_card_free_thread+0x57/0xd0 [snd]
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c012c3ee>] worker_thread+0x1be/0x2a0
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c0130a4c>] kthread+0x9c/0xd0
Apr 13 13:22:23 p4 kernel: [4299686.353000]  [<c0101365>] kernel_thread_helper+0x5/0x10
Apr 13 13:22:23 p4 kernel: [4299686.353000] Code: 89 34 24 e8 7a 29 fe ff eb dd 90 8d b4 26 00 00 00 00 55 89 e5 57 8b 55 08 56 be 01 00 00 00 53 31 db 8b 3a b9 ff ff ff ff 89 d8 <f2> ae f7 d1 49 8b 52 24 8d 74 31 01 85 d2 75 e7 5b 89 f0 5e 5f 

// keyboard (usb or ps/2, doesn't matter) stops working

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #3: usx2y-hot-unplug_snd_card_free --]
[-- Type: text/x-objcsrc; name=usx2y-hot-unplug_snd_card_free, Size: 13412 bytes --]

with snd_card_free, it's ok:
this is triggered by unplugging the usb cable while jackd was runnnig.
comments begin with //


// jackd starts streamin
Apr 14 20:12:10 p4 kernel: [4299805.544000] snd_usX2Y_pcm_trigger(START)
Apr 14 20:12:10 p4 kernel: [4299805.544000] snd_usX2Y_pcm_trigger(START)
Apr 14 20:12:10 p4 kernel: [4299805.546000] cap start 31

// after 10seconds the usb cable is unplugged
Apr 14 20:12:20 p4 kernel: [4299815.118000] ep=10 stalled with status=-84
Apr 14 20:12:20 p4 kernel: [4299815.118000] snd_usX2Y_pcm_trigger(STOP)
Apr 14 20:12:20 p4 kernel: [4299815.118000] snd_usX2Y_pcm_trigger(STOP)
Apr 14 20:12:20 p4 kernel: [4299815.118000] 0 status=0 start_frame=489
Apr 14 20:12:20 p4 kernel: [4299815.118000] 1 status=-115 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.118000] 0 status=-115 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.118000] 1 status=0 start_frame=489
Apr 14 20:12:20 p4 kernel: [4299815.118000] 0 status=-115 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.118000] 1 status=0 start_frame=489
Apr 14 20:12:20 p4 kernel: [4299815.118000] 0 e4a19ac0 state=4
Apr 14 20:12:20 p4 kernel: [4299815.118000] 1 c8ebe8c0 state=4
Apr 14 20:12:20 p4 kernel: [4299815.118000] 3 e4a19c00 state=4
Apr 14 20:12:20 p4 kernel: [4299815.118000] snd_usX2Y_pcm_prepare(ed36b1c0)
Apr 14 20:12:20 p4 kernel: [4299815.118000] usX2Y_subs_prepare(c8ebe8c0) ep=8 urb0=e4d0a200 urb1=e4d0a180
Apr 14 20:12:20 p4 kernel: [4299815.118000] starting capture pipe for self
Apr 14 20:12:20 p4 kernel: [4299815.119000] hcd_frame=491 ep=8in status=-2 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.119000] hcd_frame=491 ep=10out status=0 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.119000] hcd_frame=491 ep=10in status=-84 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.119000] 501
Apr 14 20:12:20 p4 kernel: [4299815.119000] 501
Apr 14 20:12:20 p4 kernel: [4299815.119000] 502
Apr 14 20:12:20 p4 kernel: [4299815.119000] 502
Apr 14 20:12:20 p4 kernel: [4299815.130000] ep=8 stalled with status=-84
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 status=0 start_frame=489
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 status=0 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 status=0 start_frame=501
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 status=-115 start_frame=502
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 status=-84 start_frame=501
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 status=-115 start_frame=502
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 e4a19ac0 state=0
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 c8ebe8c0 state=4
Apr 14 20:12:20 p4 kernel: [4299815.130000] 3 e4a19c00 state=4
Apr 14 20:12:20 p4 kernel: [4299815.130000] hcd_frame=502 ep=10in status=-84 start_frame=501
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 status=0 start_frame=489
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 status=0 start_frame=490
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 status=0 start_frame=501
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 status=-115 start_frame=502
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 status=-84 start_frame=501
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 status=-115 start_frame=502
Apr 14 20:12:20 p4 kernel: [4299815.130000] 0 e4a19ac0 state=0
Apr 14 20:12:20 p4 kernel: [4299815.130000] 1 c8ebe8c0 state=0
Apr 14 20:12:20 p4 kernel: [4299815.130000] 3 e4a19c00 state=0
Apr 14 20:12:20 p4 kernel: [4299815.131000] hcd_frame=503 ep=8in status=-84 start_frame=502
Apr 14 20:12:20 p4 kernel: [4299815.131000] hcd_frame=503 ep=10in status=-84 start_frame=502
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_usbpcm_hw_free(ed36b340)
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_urbs_release() 10
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_urbs_release() 8
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_urbs_release() 10
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_usbpcm_hw_free(ed36b1c0)
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_urbs_release() 8
Apr 14 20:12:20 p4 kernel: [4299815.133000] snd_usX2Y_urbs_release() 10
Apr 14 20:12:20 p4 kernel: [4299815.828000] hub 3-0:1.0: state 5 ports 2 chg 0000 evt 0004
Apr 14 20:12:20 p4 kernel: [4299815.828000] uhci_hcd 0000:00:10.1: port 2 portsc 008a,00
Apr 14 20:12:20 p4 kernel: [4299815.828000] hub 3-0:1.0: port 2, status 0100, change 0003, 12 Mb/s
Apr 14 20:12:20 p4 kernel: [4299815.828000] usb 3-2: USB disconnect, address 2
Apr 14 20:12:20 p4 kernel: [4299815.828000] usb 3-2: usb_disable_device nuking all URBs
Apr 14 20:12:20 p4 kernel: [4299815.828000] usb 3-2: unregistering interface 3-2:1.0

// usX2Y_usb_disconnect() is called. I added this debugstatement only now, not in the other trace.
Apr 14 20:12:20 p4 kernel: [4299815.828000] usX2Y_usb_disconnect(e25ae400, ef741000)
Apr 14 20:12:20 p4 kernel: [4299815.828000] CLASS: Unregistering class device. ID = 'mixer1'
Apr 14 20:12:20 p4 kernel: [4299815.828000] fill_kobj_path: path = '/class/sound/mixer1'
Apr 14 20:12:20 p4 kernel: [4299815.828000] class_hotplug - name = mixer1
Apr 14 20:12:20 p4 kernel: [4299815.828000] kobject mixer1 @ e2c9e310: cleaning up
Apr 14 20:12:20 p4 kernel: [4299815.828000] device class 'mixer1': release.
Apr 14 20:12:20 p4 kernel: [4299815.828000] CLASS: Unregistering class device. ID = 'adsp1'
Apr 14 20:12:20 p4 kernel: [4299815.828000] fill_kobj_path: path = '/class/sound/adsp1'
Apr 14 20:12:20 p4 kernel: [4299815.828000] class_hotplug - name = adsp1
Apr 14 20:12:20 p4 kernel: [4299815.840000] kobject adsp1 @ e4ebef10: cleaning up
Apr 14 20:12:20 p4 kernel: [4299815.840000] device class 'adsp1': release.
Apr 14 20:12:20 p4 kernel: [4299815.840000] CLASS: Unregistering class device. ID = 'dsp1'
Apr 14 20:12:20 p4 kernel: [4299815.840000] fill_kobj_path: path = '/class/sound/dsp1'
Apr 14 20:12:20 p4 kernel: [4299815.840000] class_hotplug - name = dsp1
Apr 14 20:12:20 p4 kernel: [4299815.852000] kobject dsp1 @ e2dcdb10: cleaning up
Apr 14 20:12:20 p4 kernel: [4299815.852000] device class 'dsp1': release.
Apr 14 20:12:21 p4 kernel: [4299815.852000] CLASS: Unregistering class device. ID = 'audio1'
Apr 14 20:12:21 p4 kernel: [4299815.852000] fill_kobj_path: path = '/class/sound/audio1'
Apr 14 20:12:21 p4 kernel: [4299815.852000] class_hotplug - name = audio1
Apr 14 20:12:21 p4 kernel: [4299815.863000] kobject audio1 @ e2dcda90: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.863000] device class 'audio1': release.
Apr 14 20:12:21 p4 kernel: [4299815.863000] CLASS: Unregistering class device. ID = 'controlC1'
Apr 14 20:12:21 p4 kernel: [4299815.863000] fill_kobj_path: path = '/class/sound/controlC1'
Apr 14 20:12:21 p4 kernel: [4299815.863000] class_hotplug - name = controlC1
Apr 14 20:12:21 p4 kernel: [4299815.863000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.876000] kobject controlC1 @ e2c9e810: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.876000] device class 'controlC1': release.
Apr 14 20:12:21 p4 kernel: [4299815.876000] CLASS: Unregistering class device. ID = 'pcmC1D2p'
Apr 14 20:12:21 p4 kernel: [4299815.876000] fill_kobj_path: path = '/class/sound/pcmC1D2p'
Apr 14 20:12:21 p4 kernel: [4299815.876000] class_hotplug - name = pcmC1D2p
Apr 14 20:12:21 p4 kernel: [4299815.876000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.887000] kobject pcmC1D2p @ ec9f1d90: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.887000] device class 'pcmC1D2p': release.
Apr 14 20:12:21 p4 kernel: [4299815.887000] CLASS: Unregistering class device. ID = 'pcmC1D2c'
Apr 14 20:12:21 p4 kernel: [4299815.887000] fill_kobj_path: path = '/class/sound/pcmC1D2c'
Apr 14 20:12:21 p4 kernel: [4299815.887000] class_hotplug - name = pcmC1D2c
Apr 14 20:12:21 p4 kernel: [4299815.887000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.899000] kobject pcmC1D2c @ e4d0ac90: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.899000] device class 'pcmC1D2c': release.
Apr 14 20:12:21 p4 kernel: [4299815.900000] CLASS: Unregistering class device. ID = 'hwC1D1'
Apr 14 20:12:21 p4 kernel: [4299815.900000] fill_kobj_path: path = '/class/sound/hwC1D1'
Apr 14 20:12:21 p4 kernel: [4299815.900000] class_hotplug - name = hwC1D1
Apr 14 20:12:21 p4 kernel: [4299815.900000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.911000] kobject hwC1D1 @ e4ebed90: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.911000] device class 'hwC1D1': release.
Apr 14 20:12:21 p4 kernel: [4299815.911000] CLASS: Unregistering class device. ID = 'pcmC1D1c'
Apr 14 20:12:21 p4 kernel: [4299815.911000] fill_kobj_path: path = '/class/sound/pcmC1D1c'
Apr 14 20:12:21 p4 kernel: [4299815.911000] class_hotplug - name = pcmC1D1c
Apr 14 20:12:21 p4 kernel: [4299815.911000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.923000] kobject pcmC1D1c @ e4ebef90: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.923000] device class 'pcmC1D1c': release.
Apr 14 20:12:21 p4 kernel: [4299815.923000] CLASS: Unregistering class device. ID = 'pcmC1D0p'
Apr 14 20:12:21 p4 kernel: [4299815.923000] fill_kobj_path: path = '/class/sound/pcmC1D0p'
Apr 14 20:12:21 p4 kernel: [4299815.923000] class_hotplug - name = pcmC1D0p
Apr 14 20:12:21 p4 kernel: [4299815.923000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.936000] kobject pcmC1D0p @ e2dcdc10: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.936000] device class 'pcmC1D0p': release.
Apr 14 20:12:21 p4 kernel: [4299815.936000] CLASS: Unregistering class device. ID = 'pcmC1D0c'
Apr 14 20:12:21 p4 kernel: [4299815.936000] fill_kobj_path: path = '/class/sound/pcmC1D0c'
Apr 14 20:12:21 p4 kernel: [4299815.936000] class_hotplug - name = pcmC1D0c
Apr 14 20:12:21 p4 kernel: [4299815.936000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.959000] kobject pcmC1D0c @ e2dcdb90: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.959000] device class 'pcmC1D0c': release.
Apr 14 20:12:21 p4 kernel: [4299815.959000] CLASS: Unregistering class device. ID = 'midi1'
Apr 14 20:12:21 p4 kernel: [4299815.959000] fill_kobj_path: path = '/class/sound/midi1'
Apr 14 20:12:21 p4 kernel: [4299815.959000] class_hotplug - name = midi1
Apr 14 20:12:21 p4 kernel: [4299815.959000] kobject midi1 @ f79e0210: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.959000] device class 'midi1': release.
Apr 14 20:12:21 p4 kernel: [4299815.959000] CLASS: Unregistering class device. ID = 'dmmidi1'
Apr 14 20:12:21 p4 kernel: [4299815.959000] fill_kobj_path: path = '/class/sound/dmmidi1'
Apr 14 20:12:21 p4 kernel: [4299815.959000] class_hotplug - name = dmmidi1
Apr 14 20:12:21 p4 kernel: [4299815.969000] kobject dmmidi1 @ f79e0390: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.969000] device class 'dmmidi1': release.
Apr 14 20:12:21 p4 kernel: [4299815.969000] CLASS: Unregistering class device. ID = 'midiC1D0'
Apr 14 20:12:21 p4 kernel: [4299815.969000] fill_kobj_path: path = '/class/sound/midiC1D0'
Apr 14 20:12:21 p4 kernel: [4299815.969000] class_hotplug - name = midiC1D0
Apr 14 20:12:21 p4 kernel: [4299815.969000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.983000] kobject midiC1D0 @ f79e0190: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.983000] device class 'midiC1D0': release.
Apr 14 20:12:21 p4 kernel: [4299815.983000] CLASS: Unregistering class device. ID = 'hwC1D0'
Apr 14 20:12:21 p4 kernel: [4299815.983000] fill_kobj_path: path = '/class/sound/hwC1D0'
Apr 14 20:12:21 p4 kernel: [4299815.983000] class_hotplug - name = hwC1D0
Apr 14 20:12:21 p4 kernel: [4299815.983000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299815.992000] kobject hwC1D0 @ e2c9e890: cleaning up
Apr 14 20:12:21 p4 kernel: [4299815.992000] device class 'hwC1D0': release.
// snd_card_free() has finished.
// now hotplug continues cleaning up usb3:
Apr 14 20:12:21 p4 kernel: [4299815.992000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2/3-2:1.0'
Apr 14 20:12:21 p4 kernel: [4299815.992000] usb 3-2:1.0: hotplug
Apr 14 20:12:21 p4 kernel: [4299816.001000] kobject 3-2:1.0 @ f7940078: cleaning up
Apr 14 20:12:21 p4 kernel: [4299816.001000] usb 3-2: unregistering device
Apr 14 20:12:21 p4 kernel: [4299816.002000] fill_kobj_path: path = '/devices/pci0000:00/0000:00:10.1/usb3/3-2'
Apr 14 20:12:21 p4 kernel: [4299816.002000] usb 3-2: hotplug
Apr 14 20:12:21 p4 kernel: [4299816.033000] kobject 3-2 @ e25ae488: cleaning up
Apr 14 20:12:21 p4 kernel: [4299816.137000] hub 3-0:1.0: debounce: port 2: total 100ms stable 100ms status 0x100
Apr 14 20:12:21 p4 kernel: [4299816.137000] hub 1-0:1.0: state 5 ports 8 chg 0000 evt 0010
Apr 14 20:12:21 p4 kernel: [4299816.137000] ehci_hcd 0000:00:10.4: GetStatus port 4 status 001002 POWER sig=se0  CSC
Apr 14 20:12:21 p4 kernel: [4299816.137000] hub 1-0:1.0: port 4, status 0100, change 0001, 12 Mb/s
Apr 14 20:12:21 p4 kernel: [4299816.241000] hub 1-0:1.0: debounce: port 4: total 100ms stable 100ms status 0x100
Apr 14 20:12:21 p4 kernel: [4299816.713000] uhci_hcd 0000:00:10.1: suspend_hc

// everything still ok. no oops, keyboard workin.

[PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[karsten wiese]

[-- Attachment #1: Type: text/plain, Size: 1016 bytes --]

Hi Clemens & Takashi

Please commit the patch.
Btw, snd-usb-audio could use the same treatment.
Whats the use of:
	chip->num_interfaces--;
	if (chip->num_interfaces <= 0) {
in snd_usb_audio_disconnect() ?
is snd_usb_audio_disconnect() called for each usb
interface?

best regards,
Karsten
-----------------------------------------------------------
Header:
snd-usb-usx2y: prevent oops & dead keyboard on usb
unplugging while the device is being used

Description:
Without this patch, some usb kobjects, which are parents to
the usx2y's kobjects can be freed before the usx2y's.
This led to an oops in get_kobj_path_length() and a dead
keyboard, when the usx2y's kobjects were freed.
The patch ensures the correct sequence.
Tested ok on kernel 2.6.12-rc2.

Signed-of-by: Karsten Wiese <annabellesgarden@yahoo.de>


	

	
		
___________________________________________________________ 
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: usbusx2y.c_1.10_patch --]
[-- Type: text/x-diff; name="usbusx2y.c_1.10_patch", Size: 1572 bytes --]

Index: usbusx2y.c
===================================================================
RCS file: /cvsroot/alsa/alsa-kernel/usb/usx2y/usbusx2y.c,v
retrieving revision 1.9
diff -c -r1.9 usbusx2y.c
--- usbusx2y.c	28 Jan 2005 10:21:48 -0000	1.9
+++ usbusx2y.c	14 Apr 2005 18:27:29 -0000
@@ -1,6 +1,11 @@
 /*
  * usbusy2y.c - ALSA USB US-428 Driver
  *
+2005-04-14 Karsten Wiese
+	Version 0.8.7.2:
+	Call snd_card_free() instead of snd_card_free_in_thread() to prevent oops with dead keyboard symptom.
+	Tested ok with kernel 2.6.12-rc2.
+
 2004-12-14 Karsten Wiese
 	Version 0.8.7.1:
 	snd_pcm_open for rawusb pcm-devices now returns -EBUSY if called without rawusb's hwdep device being open.
@@ -143,7 +148,7 @@
 
 
 MODULE_AUTHOR("Karsten Wiese <annabellesgarden@yahoo.de>");
-MODULE_DESCRIPTION("TASCAM "NAME_ALLCAPS" Version 0.8.7.1");
+MODULE_DESCRIPTION("TASCAM "NAME_ALLCAPS" Version 0.8.7.2");
 MODULE_LICENSE("GPL");
 MODULE_SUPPORTED_DEVICE("{{TASCAM(0x1604), "NAME_ALLCAPS"(0x8001)(0x8005)(0x8007) }}");
 
@@ -430,8 +435,6 @@
 	if (ptr) {
 		usX2Ydev_t* usX2Y = usX2Y((snd_card_t*)ptr);
 		struct list_head* p;
-		if (usX2Y->chip_status == USX2Y_STAT_CHIP_HUP)	// on 2.6.1 kernel snd_usbmidi_disconnect()
-			return;					// calls us back. better leave :-) .
 		usX2Y->chip.shutdown = 1;
 		usX2Y->chip_status = USX2Y_STAT_CHIP_HUP;
 		usX2Y_unlinkSeq(&usX2Y->AS04);
@@ -443,7 +446,7 @@
 		}
 		if (usX2Y->us428ctls_sharedmem) 
 			wake_up(&usX2Y->us428ctls_wait_queue_head);
-		snd_card_free_in_thread((snd_card_t*)ptr);
+		snd_card_free((snd_card_t*)ptr);
 	}
 }
 

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Lee Revell]

On Thu, 2005-04-14 at 20:59 +0200, karsten wiese wrote:
> Hi Clemens & Takashi
> 
> Please commit the patch.

Is the ALSA in kernel 2.6.11 affected?  If so, this should be sent to
stable@kernel.org.

Lee



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: [PATCH]usx2y: prevent oops & dead keyboard on hot usb unplugging

[Clemens Ladisch]

karsten wiese wrote:
> Please commit the patch.
>
> - snd_card_free_in_thread((snd_card_t*)ptr);
> + snd_card_free((snd_card_t*)ptr);

What happens when you unplug the device while it's still playing?

> Whats the use of:
> 	chip->num_interfaces--;
> 	if (chip->num_interfaces <= 0) {
> in snd_usb_audio_disconnect() ?
> is snd_usb_audio_disconnect() called for each usb
> interface?

Yes, the kernel interface treats each interface as a separate device.


Regards,
Clemens



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click