Number of subdevices on midi.

Number of subdevices on midi.

[Peter Enderborg]

I have tryed to get some answer from Jaroslav about subdevices on midi.
Im trying to write
a driver for midisport 8x8/s and that have 9 midi ports. On my hack to
figure out
the protocol I use virmidi to handle the interface to alsa. But virmidi
only have four
subdevices per device. I need to see the driver for a midi interface
that using all the
16 subdevices. Since I don't get answer from Jaroslav i guess he don't
have the time
or my question is stupid. But can someone help me out? Where should I
look?

Re: Number of subdevices on midi.

[Michael Ashton]

On Wednesday 03 April 2002 12:23 pm, you wrote:
> subdevices per device. I need to see the driver for a midi interface
> that using all the
> 16 subdevices. Since I don't get answer from Jaroslav i guess he don't
> have the time
> or my question is stupid. But can someone help me out? Where should I
> look?

I have done a little work on a driver I'm writing for a PCI-MIDI interface 
system I designed (card is built and seems to work, breakout box isn't built 
yet). I first used the MTP-AV driver as a model, but had quite a time 
figuring out how exactly it hooked up with the sequencer, and eventually 
decided to write a kernel sequencer client that talks to the card directly. 
Much easier, and I can add some interesting features using user packets, and 
also I think it will be faster. This does have the disadvantage of leaving 
rawmidi apps in the cold, though, so I'd be interested in whatever you come 
up with.

cheers -- Michael Ashton <data@ieee.org>

Re: Number of subdevices on midi.

[Jaroslav Kysela]

On Wed, 3 Apr 2002, Peter Enderborg wrote:

> I have tryed to get some answer from Jaroslav about subdevices on midi.
> Im trying to write a driver for midisport 8x8/s and that have 9 midi
> ports. On my hack to figure out the protocol I use virmidi to handle the
> interface to alsa. But virmidi only have four subdevices per device. I

You can set up to eight virmidi devices (snd_midi_devs parameter controls
this count). On rawmidi side, there is 16 rawmidi substreams (subdevices)
per one virmidi device. These subdevices are used for event merging 
(write) and coping (read). On sequencer side, there is one client and 
one port connected to virmidi device.

> need to see the driver for a midi interface that using all the 16
> subdevices. Since I don't get answer from Jaroslav i guess he don't have
> the time or my question is stupid. But can someone help me out? Where
> should I look?

There are two ways:

1) we can add a new mode to virmidi code which suppress copying and 
   merging and connects more sequencer ports per a sequencer client
   to rawmidi subdevices to handle such devices
2) without any change - simply use more than one virmidi card and connect 
   your code to more clients:

	modprobe snd-virmidi snd_index=1,2 snd_enable=1,1 snd_midi_devs=8,8

	(use index values as you wish)

	In this case, you should have available this routing table:

	sequencer (client:port)		rawmidi (card:device:subdevice)
	72:0				1:0:any
	73:0				1:1:any
	....				....
	79:0				1:7:any
	80:0				2:0:any
	....				...
	87:0				2:7:any


						Jaroslav

-----
Jaroslav Kysela <perex@suse.cz>
Linux Kernel Sound Maintainer
ALSA Project  http://www.alsa-project.org
SuSE Linux    http://www.suse.com

Re: Number of subdevices on midi.

[Peter Enderborg]

Jaroslav Kysela wrote:

> On Wed, 3 Apr 2002, Peter Enderborg wrote:
>
> > I have tryed to get some answer from Jaroslav about subdevices on midi.
> > Im trying to write a driver for midisport 8x8/s and that have 9 midi
> > ports. On my hack to figure out the protocol I use virmidi to handle the
> > interface to alsa. But virmidi only have four subdevices per device. I
>
> You can set up to eight virmidi devices (snd_midi_devs parameter controls
> this count). On rawmidi side, there is 16 rawmidi substreams (subdevices)
> per one virmidi device. These subdevices are used for event merging
> (write) and coping (read). On sequencer side, there is one client and
> one port connected to virmidi device.
>
> > need to see the driver for a midi interface that using all the 16
> > subdevices. Since I don't get answer from Jaroslav i guess he don't have
> > the time or my question is stupid. But can someone help me out? Where
> > should I look?
>
> There are two ways:
>
> 1) we can add a new mode to virmidi code which suppress copying and
>    merging and connects more sequencer ports per a sequencer client
>    to rawmidi subdevices to handle such devices
> 2) without any change - simply use more than one virmidi card and connect
>    your code to more clients:
>

That is what I do now. But it is not pretty. My plan is to do a real driver
not a virmidi user.


>
>         modprobe snd-virmidi snd_index=1,2 snd_enable=1,1 snd_midi_devs=8,8
>
>         (use index values as you wish)
>

Have you done that? This is what I get.

[root@pescadero pme]# /sbin/modprobe snd-virmidi snd_enable=1,1
snd_midi_devs=8,8
/lib/modules/2.4.16/misc/snd-virmidi.o: init_module: No such device
Hint: insmod errors can be caused by incorrect module parameters, including
invalid IO or IRQ parameters
/lib/modules/2.4.16/misc/snd-virmidi.o: insmod
/lib/modules/2.4.16/misc/snd-virmidi.o failed
/lib/modules/2.4.16/misc/snd-virmidi.o: insmod snd-virmidi failed


[root@pescadero pme]# /sbin/modprobe snd-virmidi snd_enable=1,1
snd_midi_devs=8
/lib/modules/2.4.16/misc/snd-virmidi.o: init_module: No such device
Hint: insmod errors can be caused by incorrect module parameters, including
invalid IO or IRQ parameters
/lib/modules/2.4.16/misc/snd-virmidi.o: insmod
/lib/modules/2.4.16/misc/snd-virmidi.o failed
/lib/modules/2.4.16/misc/snd-virmidi.o: insmod snd-virmidi failed


[root@pescadero pme]# /sbin/modprobe snd-virmidi snd_enable=1,1
snd_midi_devs=4,4
[root@pescadero pme]#

I also get the same with 5.

>
>         In this case, you should have available this routing table:
>
>         sequencer (client:port)         rawmidi (card:device:subdevice)
>         72:0                            1:0:any
>         73:0                            1:1:any
>         ....                            ....
>         79:0                            1:7:any
>         80:0                            2:0:any
>         ....                            ...
>         87:0                            2:7:any
>
>                                                 Jaroslav
>
> -----
> Jaroslav Kysela <perex@suse.cz>
> Linux Kernel Sound Maintainer
> ALSA Project  http://www.alsa-project.org
> SuSE Linux    http://www.suse.com
>
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/alsa-devel

--
foo!

Re: Number of subdevices on midi.

[Jaroslav Kysela]

On Thu, 4 Apr 2002, Peter Enderborg wrote:

> Jaroslav Kysela wrote:
> 
> > On Wed, 3 Apr 2002, Peter Enderborg wrote:
> >
> > > I have tryed to get some answer from Jaroslav about subdevices on midi.
> > > Im trying to write a driver for midisport 8x8/s and that have 9 midi
> > > ports. On my hack to figure out the protocol I use virmidi to handle the
> > > interface to alsa. But virmidi only have four subdevices per device. I
> >
> > You can set up to eight virmidi devices (snd_midi_devs parameter controls
> > this count). On rawmidi side, there is 16 rawmidi substreams (subdevices)
> > per one virmidi device. These subdevices are used for event merging
> > (write) and coping (read). On sequencer side, there is one client and
> > one port connected to virmidi device.
> >
> > > need to see the driver for a midi interface that using all the 16
> > > subdevices. Since I don't get answer from Jaroslav i guess he don't have
> > > the time or my question is stupid. But can someone help me out? Where
> > > should I look?
> >
> > There are two ways:
> >
> > 1) we can add a new mode to virmidi code which suppress copying and
> >    merging and connects more sequencer ports per a sequencer client
> >    to rawmidi subdevices to handle such devices
> > 2) without any change - simply use more than one virmidi card and connect
> >    your code to more clients:
> >
> 
> That is what I do now. But it is not pretty. My plan is to do a real driver
> not a virmidi user.
> 
> 
> >
> >         modprobe snd-virmidi snd_index=1,2 snd_enable=1,1 snd_midi_devs=8,8
> >
> >         (use index values as you wish)
> >
> 
> Have you done that? This is what I get.
> 
> [root@pescadero pme]# /sbin/modprobe snd-virmidi snd_enable=1,1
> snd_midi_devs=8,8
> /lib/modules/2.4.16/misc/snd-virmidi.o: init_module: No such device
> Hint: insmod errors can be caused by incorrect module parameters, including
> invalid IO or IRQ parameters
> /lib/modules/2.4.16/misc/snd-virmidi.o: insmod
> /lib/modules/2.4.16/misc/snd-virmidi.o failed
> /lib/modules/2.4.16/misc/snd-virmidi.o: insmod snd-virmidi failed

Oops. I've fixed this problem in CVS. Change SNDRV_RAWMIDI_DEVICES in 
alsa-kernel/include/rawmidi.h to 8.

						Jaroslav

-----
Jaroslav Kysela <perex@suse.cz>
Linux Kernel Sound Maintainer
ALSA Project  http://www.alsa-project.org
SuSE Linux    http://www.suse.com

Re: Number of subdevices on midi.

[Takashi Iwai]

At Sun, 7 Apr 2002 22:22:42 +0200 (CEST),
Jaroslav wrote:
> 
> On Thu, 4 Apr 2002, Peter Enderborg wrote:
> 
> > Jaroslav Kysela wrote:
> > 
> > > On Wed, 3 Apr 2002, Peter Enderborg wrote:
> > >
> > > > I have tryed to get some answer from Jaroslav about subdevices on midi.
> > > > Im trying to write a driver for midisport 8x8/s and that have 9 midi
> > > > ports. On my hack to figure out the protocol I use virmidi to handle the
> > > > interface to alsa. But virmidi only have four subdevices per device. I
> > >
> > > You can set up to eight virmidi devices (snd_midi_devs parameter controls
> > > this count). On rawmidi side, there is 16 rawmidi substreams (subdevices)
> > > per one virmidi device. These subdevices are used for event merging
> > > (write) and coping (read). On sequencer side, there is one client and
> > > one port connected to virmidi device.
> > >
> > > > need to see the driver for a midi interface that using all the 16
> > > > subdevices. Since I don't get answer from Jaroslav i guess he don't have
> > > > the time or my question is stupid. But can someone help me out? Where
> > > > should I look?
> > >
> > > There are two ways:
> > >
> > > 1) we can add a new mode to virmidi code which suppress copying and
> > >    merging and connects more sequencer ports per a sequencer client
> > >    to rawmidi subdevices to handle such devices
> > > 2) without any change - simply use more than one virmidi card and connect
> > >    your code to more clients:
> > >
> > 
> > That is what I do now. But it is not pretty. My plan is to do a real driver
> > not a virmidi user.
> > 
> > 
> > >
> > >         modprobe snd-virmidi snd_index=1,2 snd_enable=1,1 snd_midi_devs=8,8
> > >
> > >         (use index values as you wish)
> > >
> > 
> > Have you done that? This is what I get.
> > 
> > [root@pescadero pme]# /sbin/modprobe snd-virmidi snd_enable=1,1
> > snd_midi_devs=8,8
> > /lib/modules/2.4.16/misc/snd-virmidi.o: init_module: No such device
> > Hint: insmod errors can be caused by incorrect module parameters, including
> > invalid IO or IRQ parameters
> > /lib/modules/2.4.16/misc/snd-virmidi.o: insmod
> > /lib/modules/2.4.16/misc/snd-virmidi.o failed
> > /lib/modules/2.4.16/misc/snd-virmidi.o: insmod snd-virmidi failed
> 
> Oops. I've fixed this problem in CVS. Change SNDRV_RAWMIDI_DEVICES in 
> alsa-kernel/include/rawmidi.h to 8.

in alsa-kernel/include/minors.h,

#define SNDRV_MINOR_RAWMIDI		8	/* 8 - 11 */
#define SNDRV_MINOR_RAWMIDIS		4
#define SNDRV_MINOR_PCM_PLAYBACK	16	/* 16 - 23 */

so we need to change this too..
it seems no problem to assign 8 rawmidi devices - 8 devices are free,
although the comment says 8 - 11..


Takashi

Re: Number of subdevices on midi.

[Jaroslav Kysela]

On Mon, 8 Apr 2002, Takashi Iwai wrote:

> At Sun, 7 Apr 2002 22:22:42 +0200 (CEST),
> Jaroslav wrote:
> > 
> > On Thu, 4 Apr 2002, Peter Enderborg wrote:
> > 
> > > Jaroslav Kysela wrote:
> > > 
> > > > On Wed, 3 Apr 2002, Peter Enderborg wrote:
> > > >
> > > > > I have tryed to get some answer from Jaroslav about subdevices on midi.
> > > > > Im trying to write a driver for midisport 8x8/s and that have 9 midi
> > > > > ports. On my hack to figure out the protocol I use virmidi to handle the
> > > > > interface to alsa. But virmidi only have four subdevices per device. I
> > > >
> > > > You can set up to eight virmidi devices (snd_midi_devs parameter controls
> > > > this count). On rawmidi side, there is 16 rawmidi substreams (subdevices)
> > > > per one virmidi device. These subdevices are used for event merging
> > > > (write) and coping (read). On sequencer side, there is one client and
> > > > one port connected to virmidi device.
> > > >
> > > > > need to see the driver for a midi interface that using all the 16
> > > > > subdevices. Since I don't get answer from Jaroslav i guess he don't have
> > > > > the time or my question is stupid. But can someone help me out? Where
> > > > > should I look?
> > > >
> > > > There are two ways:
> > > >
> > > > 1) we can add a new mode to virmidi code which suppress copying and
> > > >    merging and connects more sequencer ports per a sequencer client
> > > >    to rawmidi subdevices to handle such devices
> > > > 2) without any change - simply use more than one virmidi card and connect
> > > >    your code to more clients:
> > > >
> > > 
> > > That is what I do now. But it is not pretty. My plan is to do a real driver
> > > not a virmidi user.
> > > 
> > > 
> > > >
> > > >         modprobe snd-virmidi snd_index=1,2 snd_enable=1,1 snd_midi_devs=8,8
> > > >
> > > >         (use index values as you wish)
> > > >
> > > 
> > > Have you done that? This is what I get.
> > > 
> > > [root@pescadero pme]# /sbin/modprobe snd-virmidi snd_enable=1,1
> > > snd_midi_devs=8,8
> > > /lib/modules/2.4.16/misc/snd-virmidi.o: init_module: No such device
> > > Hint: insmod errors can be caused by incorrect module parameters, including
> > > invalid IO or IRQ parameters
> > > /lib/modules/2.4.16/misc/snd-virmidi.o: insmod
> > > /lib/modules/2.4.16/misc/snd-virmidi.o failed
> > > /lib/modules/2.4.16/misc/snd-virmidi.o: insmod snd-virmidi failed
> > 
> > Oops. I've fixed this problem in CVS. Change SNDRV_RAWMIDI_DEVICES in 
> > alsa-kernel/include/rawmidi.h to 8.
> 
> in alsa-kernel/include/minors.h,
> 
> #define SNDRV_MINOR_RAWMIDI		8	/* 8 - 11 */
> #define SNDRV_MINOR_RAWMIDIS		4
> #define SNDRV_MINOR_PCM_PLAYBACK	16	/* 16 - 23 */
> 
> so we need to change this too..
> it seems no problem to assign 8 rawmidi devices - 8 devices are free,
> although the comment says 8 - 11..

It's right, I don't remember, if these device are reserved, but we can 
allocate next major when this problem arises. Fixed and thanks.

						Jaroslav

-----
Jaroslav Kysela <perex@suse.cz>
Linux Kernel Sound Maintainer
ALSA Project  http://www.alsa-project.org
SuSE Linux    http://www.suse.com