Interesting problem with hotplug and Midisport 2x2 on Mdk 9.0
Hi all,
This is my latest finding. Thanks to the script I got from Jaroslav, I am now able to hotplug my midisport 2x2 whenever I wish to do so. However, there is an interesting problem surrounding Mdk9.0.
If I use the planet ccrma's hotplug ver. 2002_8_26 (by forcing upgrade of the mdk's), I get the fxload and all the goodies making the midisport work right out-of-the-box. However, with one huge flaw:
When I unplug the midisport (after having the alsasound hotplug script installed), my kernel oops and system freezes telling me something about "disabling interrupt handler." (if someone tells me where I can track this information log I will be happy to forward it). So, I am simply not allowed to unplug the device.
In addition my /var/log/messages relatively often spits out the following message:
<date> <host> kernel: usb-uhci.c: interrupt, status 3, frame# <some_number>
I've tracked down the problem to the /sbin/modprobe -r snd-usb-midi (and snd-usb-audio). If I move this thing somewhere else in the alsasound hotplug script (like purging it at load-time and then reloading the driver, rather than on disconnect) then the freeze occurs at that point. When I comment it out, no freezes occur. So, I am pretty sure that is the culprit.
On the other hand, if I downgrade to Mdk's hotplug ver. 2002_4_01, then no oops occur (except one has to put alsasound.usermap and ezusbmidi.usermap to /etc/hotplug/usb.usermap.local, since this older version does not support usb sub-dir placing of usermaps. However in this case my midisport offers me only 1 I/O device (even though it is 2x2), as opposed to the ccrma's which rightly gives me 2. (both report locating 2 I and 2 o devices in /var/log/messages log, but only ccrma's offers both devices via /dev/midi*).
I tried diff-ing the files from the two rpm packages to see if I am able to locate the place where this dramatic change occurs, but found that most of the files involved in the two packages are rather similar, if not the same, major difference being the support of the usb/ subdir for storing module scripts.
So, what gives?
Anyone has a clue? What is so different between the two packages that one causes kernel oops/freeze, while the other one gives me only one i/o midi port?
The files contained within each package are mostly files located in /etc/hotplug dir, some man stuff, and fxload and hotplug from the /sbin/ dir (fxload being present in later hotplug package, while for the older one, I extracted only that file from the ccrma's package and put it in the right place).
In addition to this, I just noticed that the same message is showing up in my /var/log/messages when using the mdk's package, so that one most likely is not the problem, or at least not associated with this issue...
Any help is greatly appreciated! Sinceerely,
Ivica Ico Bukvic, composer, multimedia sculptor,
programmer, webmaster & computer consultant
http://meowing.ccm.uc.edu/~ico
============================
"To be or not to be" - Shakespeare
"To be is to do" - Socrates "To do is to be" - Sartre "Do be do be do" - Sinatra "2b || ! 2b" - ? "I am" - God
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
At Thu, 27 Feb 2003 13:00:10 -0500,
Quoted 22 lines · Ivica Bukvic
Hi all,
This is my latest finding. Thanks to the script I got from Jaroslav, I am now able to hotplug my midisport 2x2 whenever I wish to do so. However, there is an interesting problem surrounding Mdk9.0.
If I use the planet ccrma's hotplug ver. 2002_8_26 (by forcing upgrade of the mdk's), I get the fxload and all the goodies making the midisport work right out-of-the-box. However, with one huge flaw:
When I unplug the midisport (after having the alsasound hotplug script installed), my kernel oops and system freezes telling me something about "disabling interrupt handler." (if someone tells me where I can track this information log I will be happy to forward it). So, I am simply not allowed to unplug the device.
In addition my
/var/log/messagesrelatively often spits out the following message:
<date> <host> kernel:usb-uhci.c: interrupt, status 3, frame# <some_number>
Quoted 6 lines
I've tracked down the problem to the
/sbin/modprobe-r snd-usb-midi (and snd-usb-audio). If I move this thing somewhere else in the alsasound hotplug script (like purging it at load-time and then reloading the driver, rather than on disconnect) then the freeze occurs at that point. When I comment it out, no freezes occur. So, I am pretty sure that is the culprit.
hmm, the hot unplugging of ALSA usb devices was improved in the recent version - basically i didn't see problems with the devices i have. but i've not tested with midi devices, so this might be a problem.
just make sure that you unload snd-usb-audio (snd-usb-midi doesn't exist any longer) before unplugging the device.
Takashi
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Actually, now that I have reinstalled the src.rpm package of hotplug 8_26_2002 (from sourceforge), I do get both the 2 midi ports and audio (crashes still occur). I've now discovered two more interesting issues:
1) Upon booting and installing the midisport 2x2 for the first time (regardless whether that is at boot time or afterwards), I get 2 midi ports and everything is ok. If I then disconnect and reconnect it, everything is still ok, but now I only have 1 midi port. Seems like only the first time is a charm. This is no biggie, since I've overcome the major problems, but still may point to some kind of a bug in (I presume) Alsa.
2) The other rather important observation is that the crash (kernel oops) occurs, when I omit the /sbin/modprobe -r snd-usb-midi line from the Jaroslav's script as you have suggested (this has happened several times and every time I comment it out, the thing crashes with the same interrupt handler async oops, while the system remains rock solid regardless of numbers of reconnects when the thing is not commented out).
It makes me wonder whether that call to remove a module that does not exist gives the kernel enough time to clean up something usb-related on its side before invoking the removal of the module, therefore avoiding the oops (the second line /sbin/modprobe -r snd-usb-audio). It would be perhaps a good idea to do a small sleep command there and see whether that would work as well as the above-mentioned removal of the non-existent module (but I don't know how to put one there since I am not that good with shell scripts -- any ideas are welcome though).
Still, the issue of reconnect giving you only 1 midi port remains. :-)
Note: mandrake-expert developers, if you need the script and config e-mailed in order to have Mandrake 9.0 auto-detect alsasound devices and dynamically load them, please contact me and I'll e-mail them to ya.
Ico
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Quoted 9 lines · Takashi Iwai
hmm, the hot unplugging of ALSA usb devices was improved in the recent version - basically i didn't see problems with the devices i have. but i've not tested with midi devices, so this might be a problem.
just make sure that you unload snd-usb-audio (snd-usb-midi doesn't exist any longer) before unplugging the device.
I see it too with the quattro and another user has reported it with the audigy. It would be nicer if we didn't have to remember to unload the usb-audio driver before hotplugging. Is it possible to make that happen.
Signature · 16 lines
Patrick Shirkey - Boost Hardware Ltd. Http://www.boosthardware.com Http://www.djcj.org - The Linux Audio Users guide ======================================== Being on stage with the band in front of crowds shouting, "Get off! No! We want normal music!", I think that was more like acting than anything I've ever done. Goldie, 8 Nov, 2002 The Scotsman ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf
At Fri, 28 Feb 2003 14:51:08 +0900,
Quoted 14 lines · Patrick Shirkey
Quoted 9 lines · Takashi Iwai
hmm, the hot unplugging of ALSA usb devices was improved in the recent version - basically i didn't see problems with the devices i have. but i've not tested with midi devices, so this might be a problem.
just make sure that you unload snd-usb-audio (snd-usb-midi doesn't exist any longer) before unplugging the device.
Quoted 3 lines
I see it too with the quattro and another user has reported it with the audigy. It would be nicer if we didn't have to remember to unload the usb-audio driver before hotplugging. Is it possible to make that happen.
then there is still a bug in this regard.
are you using MIDI on quattro, too?
Takashi
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
Quoted 7 lines · Takashi Iwai
Quoted 2 lines
I see it too with the quattro and another user has reported it with the audigy. It would be nicer if we didn't have to remember to unload the usb-audio driver before hotplugging. Is it possible to make that happen.
then there is still a bug in this regard.
are you using MIDI on quattro, too?
It is automatically used. Clemens would have to explain as I cannot remember the exact details.
Signature · 16 lines
Patrick Shirkey - Boost Hardware Ltd. Http://www.boosthardware.com Http://www.djcj.org - The Linux Audio Users guide ======================================== Being on stage with the band in front of crowds shouting, "Get off! No! We want normal music!", I think that was more like acting than anything I've ever done. Goldie, 8 Nov, 2002 The Scotsman ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf
Quoted 7 lines · Patrick Shirkey
Quoted 5 lines · Takashi Iwai
Quoted 2 lines
I see it too with the quattro and another user has reported it with the audigy. It would be nicer if we didn't have to remember to unload the usb-audio driver before hotplugging. Is it possible to make that happen.
Quoted 6 lines
Quoted 3 lines
then there is still a bug in this regard.
are you using MIDI on quattro, too?It is automatically used. Clemens would have to explain as I cannot remember the exact details.
I another bug I was just reminded of is that if I remove the alsa modules while the usb device is still switched on I get a complete hang where I have to restart with alt+sysreq+b
Signature · 16 lines
Patrick Shirkey - Boost Hardware Ltd. Http://www.boosthardware.com Http://www.djcj.org - The Linux Audio Users guide ======================================== Being on stage with the band in front of crowds shouting, "Get off! No! We want normal music!", I think that was more like acting than anything I've ever done. Goldie, 8 Nov, 2002 The Scotsman ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf
Quoted 7 lines · Patrick Shirkey
Quoted 6 lines · Patrick Shirkey
Quoted 5 lines · Takashi Iwai
Quoted 4 lines
I see it too with the quattro and another user has reported it with the audigy. It would be nicer if we didn't have to remember to unload the usb-audio driver before hotplugging. Is it possible to make that happen.
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines
then there is still a bug in this regard.
are you using MIDI on quattro, too?It is automatically used. Clemens would have to explain as I cannot remember the exact details.
MIDI input always has a URB submitted.
Quoted 3 lines
I another bug I was just reminded of is that if I remove the alsa modules while the usb device is still switched on I get a complete hang where I have to restart with alt+sysreq+b
All these crashes look like the bug which has been fixed on 13th January, so it should work with 0.9.0rc7. Nothing like this happens with my devices (UA-1A and SC-8820), and I regularly plug/unplug them and load/unload the drivers during debugging.
HTH
Clemens
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf