Jaroslav, Clemens and Takashi, Thank you for your helpful suggestions. Unfortunately the ALSArc5 OSS-mode sequencer still fails giving invalid input and silent output using an external MIDI keyboard.
The ALSA-mode sequencer does work for input and output,
e.g. "pmidi -p 64:0" plays midi files to external MIDI
keyboards on port 64, and "aconnect 64:0 65:0" enables
notes played on external MIDI keyboards to be played
simultaneously on the soundcard synth. midiloop tests ok.
However, I need the OSS-mode sequencer to run OSS software
like Rosegarden to convert MIDI input to music notation.
I know Noteedit 2.0.16 works with the ALSA-mode sequencer
and I could use it but even though it's very good the
interface is still not as quick to use as Rosegarden for
note/chord editing on big scores.
Config details for my SB AWE64Gold (EISA card) are in attachments
(see also my previous msgs of 6/23/26 Aug 2002).
1) OSS-mode sequencer: Invalid MIDI recordings
Has anyone tested the OSS-mode sequencer in ALSA
by recording from external MIDI instruments? If so,
which OSS-mode software and which keyboard did you use?
Please mail me an example MIDI file you've recorded.
Re: my example of a MIDI file with invalid time values
recorded from the OSS-mode sequencer:
"rawmidi does not care about timing at all"
Ok, I wasn't talking about event delivery speed.
I meant the timing data in the MIDI files you get by
recording an external MIDI instrument were invalid,
i.e. the MIDI tick values that define the timing of
events like note on and note off had silly values.
The tick values should be stored in the midi event
structures which OSS software e.g. Rosegarden reads
from /dev/sequencer but ALSA seems to be corrupting
the tick values. I've tried Rosegarden, Brahms, Muse
and ARTs with the OSS-mode sequencer from ALSA but I
still don't get valid MIDI recordings in any of them.
But if I use the original Linux 2.0.x OSS sound driver
with OSS software like Rosegarden, I get correct time
values and valid MIDI recordings. And if I use the
ALSA-mode sequencer with ALSA software like Noteedit
I get correct time values and valid MIDI recordings.
Please could you test the Rosegarden OSS sequencer on
your hardware by making a recording from a MIDI instrument.
Please download this statically linked bzip2'd version
http://www.demon.net/homepages/liam2/sequencer.static.bz2
Click "Record" in the "File" menu, then "Stop" and
then save the MIDI recording to a file using "Save As"
also in the "File" menu.
This is how Rosegarden reads MIDI events from /dev/sequencer
which works perfectly in OSS/Linux 2.0.x:
typedef struct {
unsigned long int Time;
unsigned char Bytes[3];
} MidiEvent;
unsigned char RawMidiEvent[4];
int devseq = open("/dev/sequencer", O_RDONLY, 0);
MidiEvent ev;
read(devseq, &RawMidiEvent, sizeof(RawMidiEvent));
switch ( RawMidiEvent[0] ) {
case SEQ_WAIT:
ev.Time = ((RawMidiEvent[3]<<16) | (RawMidiEvent[2]<<8) |
RawMidiEvent[1]);
break;
case SEQ_ECHO:
/* process MIDI echo here */
break;
case SEQ_MIDIPUTC:
/* process MIDI putc here */
break;
/* etc */
2) OSS-mode sequencer: Silent external playback
Playing valid MIDI files to external MIDI instruments
using the OSS-mode sequencer gives silence.
Only the ALSA-mode sequencer works.
Clemens wrote:
> Try "cat /dev/midi???" with your device. This will output the
> incoming data as ASCII (i.e. garbage), but the garbage should
> ppear at the same moment you press the keys on your MIDI
> keyboard.
"cat /dev/midi*" (I tried all midi* and sound/midi* devices)
gives hardly any output (a few characters per minute)
and characters appear randomly not at the same time as
external keys are pressed.
Jaroslav wrote:
> Note that throughput for MIDI output for SB16/AWE cards is only 200 bytes
> per second, because of hardware limitations when no loops are used in
> kernel. Try to increase your HZ value in linux/include/asm/params.h to
> 1000 or 2000. You should have better results.
I recompiled different 2.4.19 kernels after setting
HZ=2000 in asm-i386/param.h (I had to fix the SHIFT_HZ
definition which breaks when HZ>=1536 in linux/timex.h)
but recordings are still silent. Setting HZ=2000 also
has the horrible side-effect of converting my speedy
1800Mhz CPU into a crawling 66MHz CPU (BYTE benchmark
fell from 32 to 0.08).
Is there a way to get external MIDI playback working on
the OSS-mode sequencer without increasing HZ? Why does
the ALSA-mode sequencer work perfectly without requiring
a higher HZ?
3) Will MIDI RESET be implemented in ALSA 1.0.0 ?
Jaroslav wrote:
> I've removed these unimplemented functions. Actually, the current
> implementation sends 0xfe byte (active sensing) at close().
> Sure, we may modify this code to satisfy some non-standard devices.
I'm not using non-standard devices. I've got standard
MIDI instruments like Yamaha, Roland etc. I want
to send MIDI reset from ALSA to my MIDI instruments.
It's extremely useful when you need to quickly restore
the default settings in a MIDI instrument. I don't
want to close() the device every time I need to do a
reset. Having a reliable MIDI RESET function like the
seq_reset() in OSS/Linux 2.0.x which explicitly switches
off notes as well as sending 0xfe and which doesn't
close() the device would be useful. Would you like
seq_reset() from OSS/Linux 2.0.x ported to ALSA?
Takashi wrote:
> > > > 4) BTW, is anyone working on MIDI RESET for ALSA 1.0.0?
> > > > alsa-kernel/core/rawmidi.c says MIDI reset is in the
> > > > TODO stage. I have instruments that sometimes need
> > > > MIDI reset. OSS/Linux 2.0.x had a handy seq_reset()
> > > > in drivers/sound/sequencer.c which mostly worked well.
> > >
> > > I've removed these unimplemented functions. Actually, the current
> > > implementation sends 0xfe byte (active sensing) at close(). Sure, we
> > > may modify this code to satisfy some non-standard devices.
> >
> > I don't understand. Can't an application that wants to reset a device
> > just send 0xff?
>
> yes. it should work (mostly).
>
> > And why send 0xfe at close? So the receiving device
> > will detect a connection loss 300msec later?
> > Shouldn't this all be left to the application?
>
> some (mainly old) devices required the active sensing for resetting.
> and please note that this feature can be enabled/disabled via API,
> too.
Some of my Yamaha devices need explicit note off for reliable resetting
like seq_reset() does in the old Linux 2.0.x OSS driver.
Do you mean an API exists to en(dis)able sending 0xfe at close?
Where is this API?
Thanks,
Will
-------------------------------
recording.mid
Description: MIDI audio
The MPU registration which you do get in the original Linux 2.0.x OSS driver output from /dev/sndstat under "Card config:" is missing in ALSA. The /dev/sndstat outputs below (full outputs in attachments) are different but maybe that's simply because ALSA uses /dev/sndstat differently? The MPU401 port is in /proc/ioports (see attachment). ALSA/Linux 2.4.x: [...] Card config: Sound Blaster 16 at 0x220, irq 5, dma 1&5 >>>[ MPU is not mentioned here ] [...] Midi devices: 0: MPU-401 (UART) [...] OSS/Linux 2.0.x: [...] Card config: Sound Blaster at 0x220 irq 5 drq 1,5 >>>SB MPU-401 at 0x300 irq 5 OPL-2/OPL-3 FM at 0x388 drq 0 [...] Midi devices: 0: Sound Blaster 16 [...]
Sound Driver:3.8.1a-980706 (ALSA v0.9.0rc5 emulation code) Kernel: Linux alsatest 2.4.19 #7 Thu Nov 14 17:52:19 GMT 2002 i686 Config options: 0 Installed drivers: Type 10: ALSA emulation Card config: Sound Blaster 16 at 0x220, irq 5, dma 1&5 Audio devices: 0: DSP v4.16 (DUPLEX) Synth devices: 0: OPL3 FM Midi devices: 0: MPU-401 (UART) Timers: 7: system timer Mixers: 0: mixer00
Sound Driver:3.5.4-960630 (Fri Dec 15 14:42:06 GMT 2000 will, Linux osstest 2.0.39 #6 Thu Dec 14 19:01:13 GMT 2000 i686) Kernel: Linux osstest 2.0.39 #8 Fri Dec 15 08:21:13 GMT 2000 i686 Config options: 0 Installed drivers: Type 1: OPL-2/OPL-3 FM Type 2: Sound Blaster Type 7: SB MPU-401 Card config: Sound Blaster at 0x220 irq 5 drq 1,5 SB MPU-401 at 0x300 irq 5 OPL-2/OPL-3 FM at 0x388 drq 0 Audio devices: 0: Sound Blaster 16 (4.13) Synth devices: 0: Yamaha OPL-3 Midi devices: 0: Sound Blaster 16 Timers: 0: System clock Mixers: 0: Sound Blaster
alias usb-controller usb-uhci # For ALSA driver and (EISA) SoundBlaster 32/64 AWE only: alias sound snd-sbawe alias char-major-14 soundcore alias char-major-116 sound alias /dev/dsp sound alias /dev/music sound alias /dev/sound sound alias /dev/sequencer sound alias /dev/sndstat sound options snd-sbawe snd_isapnp=1 snd_awe_port=0x620 # if using devfsd you must use add-above to get the OSS modules loaded: add above snd-sbawe snd-pcm-oss snd-mixer-oss snd-seq-oss snd-opl3-synth alias snd-card-1 off alias snd-card-2 off alias snd-card-3 off alias snd-card-4 off alias snd-card-5 off alias snd-card-6 off alias snd-card-7 off
Module Size Used by Tainted: P snd-seq-midi 3328 0 (autoclean) (unused) snd-emu8000-synth 8584 0 (autoclean) (unused) snd-pcm-oss 35684 0 (autoclean) (unused) snd-mixer-oss 9212 1 (autoclean) [snd-pcm-oss] snd-emux-synth 24060 0 (autoclean) [snd-emu8000-synth] snd-seq-virmidi 2584 0 (autoclean) [snd-emux-synth] snd-util-mem 1200 0 (autoclean) [snd-emu8000-synth snd-emux-synth] snd-seq-oss 22336 0 (autoclean) (unused) snd-seq-midi-event 2984 0 (autoclean) [snd-seq-midi snd-seq-virmidi snd-seq-oss] snd-opl3-synth 8836 0 (autoclean) (unused) snd-seq-instr 4304 0 (autoclean) [snd-opl3-synth] snd-seq-midi-emul 4320 0 (autoclean) [snd-emux-synth snd-opl3-synth] snd-seq 33936 2 (autoclean) [snd-seq-midi snd-emux-synth snd-seq-virmidi snd-seq-oss snd-seq-midi-event snd-opl3-synth snd-seq-instr snd-seq-midi-emul] snd-ainstr-fm 1396 0 (autoclean) [snd-opl3-synth] snd-sbawe 16128 1 (autoclean) [snd-emu8000-synth] snd-sb16-csp 15080 0 (autoclean) [snd-sbawe] isa-pnp 27908 0 (autoclean) [snd-sbawe] snd-opl3-lib 5092 0 (autoclean) [snd-opl3-synth snd-sbawe] snd-hwdep 3424 0 (autoclean) [snd-sb16-csp snd-opl3-lib] snd-sb16-dsp 5224 0 (autoclean) [snd-sbawe] snd-pcm 48128 0 (autoclean) [snd-emu8000-synth snd-pcm-oss snd-sb16-dsp] snd-timer 9576 0 (autoclean) [snd-seq snd-opl3-lib snd-pcm] snd-sb-common 5960 0 (autoclean) [snd-sbawe snd-sb16-csp snd-sb16-dsp] snd-mpu401-uart 2544 0 (autoclean) [snd-sbawe snd-sb16-dsp] snd-rawmidi 11968 0 (autoclean) [snd-seq-midi snd-seq-virmidi snd-mpu401-uart] snd-seq-device 3824 0 (autoclean) [snd-seq-midi snd-emu8000-synth snd-emux-synth snd-seq-oss snd-opl3-synth snd-seq snd-sbawe snd-opl3-lib snd-rawmidi] snd 24204 0 (autoclean) [snd-seq-midi snd-emu8000-synth snd-pcm-oss snd-mixer-oss snd-emux-synth snd-seq-virmidi snd-util-mem snd-seq-oss snd-seq-midi-event snd-opl3-synth snd-seq-instr snd-seq snd-sbawe snd-sb16-csp snd-opl3-lib snd-hwdep snd-sb16-dsp snd-pcm snd-timer snd-sb-common snd-mpu401-uart snd-rawmidi snd-seq-device]
# cat /proc/asound/cards
0 [card0 ]: SB AWE - Sound Blaster 16
Sound Blaster 16 at 0x220, irq 5, dma 1&5
# cat /proc/asound/devices
0: [0- 0]: ctl
4: [0- 0]: hardware dependent
8: [0- 0]: raw midi
16: [0- 0]: digital audio playback
24: [0- 0]: digital audio capture
1: : sequencer
9: [0- 1]: raw midi
10: [0- 2]: raw midi
17: [0- 1]: digital audio playback
33: : timer
# cat /proc/asound/oss/devices
10: [0-10]: hardware dependent
2: [0- 2]: raw midi
1: : sequencer
8: : sequencer
0: [0- 0]: mixer
3: [0- 3]: digital audio
13: [0-13]: raw midi
12: [0-12]: digital audio
# cat /proc/bus/isapnp/devices
0100 CTL009eCTL0044
0102200330038800000000000000000000010100000105050000000000000000000000000000000000
0101 CTL009eCTL7002
0000000000000000000000000000000000000200000404040000000000000000000000000000000000
0102 CTL009eCTL0023
0106200a200e2000000000000000000000000200000404040000000000000000000000000000000000
# cat /proc/interrupts
CPU0
[...]
5: 1 XT-PIC SoundBlaster
[...]
NMI: 0
LOC: 4
ERR: 5
MIS: 0
# cat /proc/ioports
0000-001f : dma1
[...]
0080-008f : dma page reg
[...]
00c0-00df : dma2
[...]
0213-0213 : isapnp read
0220-022f : SoundBlaster
[...]
0330-0331 : MPU401 UART
[...]
0388-0389 : OPL2/3 (left)
038a-038b : OPL2/3 (right)
[...]
0620-0623 : Emu8000-1
0a20-0a23 : Emu8000-2
0a79-0a79 : isapnp write
0cf8-0cff : PCI conf1
0e20-0e23 : Emu8000-3
[...]
% cat /etc/devfsd.conf # Enable full compatibility mode for old device names. You may comment these # out if you don't use the old device names. Make sure you know what you're # doing! REGISTER .* MKOLDCOMPAT UNREGISTER .* RMOLDCOMPAT # You may comment these out if you don't use the original "new" names REGISTER .* MKNEWCOMPAT UNREGISTER .* RMNEWCOMPAT # Enable module autoloading. You may comment this out if you don't use # autoloading LOOKUP .* MODLOAD # # Uncomment this if you want the old /dev/cdrom symlink # (e.g. those specifying CD-ROM type, mouse port, modem port etc) # REGISTER ^cdroms/cdrom0$ CFUNCTION GLOBAL mksymlink $devname cdrom UNREGISTER ^cdroms/cdrom0$ CFUNCTION GLOBAL unlink cdrom REGISTER ^sound/audio$ CFUNCTION GLOBAL mksymlink $devname audio UNREGISTER ^sound/audio$ CFUNCTION GLOBAL unlink audio REGISTER ^sound/midi00$ CFUNCTION GLOBAL mksymlink $devname midi UNREGISTER ^sound/midi00$ CFUNCTION GLOBAL unlink midi REGISTER ^sound/sequencer2$ CFUNCTION GLOBAL mksymlink $devname music UNREGISTER ^sound/sequencer2$ CFUNCTION GLOBAL unlink music REGISTER ^/proc/asound/oss/sndstat$ CFUNCTION GLOBAL mksymlink $devname sndstat UNREGISTER ^/proc/asound/oss/sndstat$ CFUNCTION GLOBAL unlink sndstat # Set the correct owner/group permissions on the sound driver device files: LOOKUP snd MODLOAD ACTION snd REGISTER sound/.* PERMISSIONS root.audio 660 REGISTER snd/.* PERMISSIONS root.audio 660
