search

takaswie / linux-firewire-dkms

Currently this repository is maintained for Linux firewire subsystem and unit drivers.
http://ieee1394.docs.kernel.org/
39 stars 8 forks source link

Add support for MOTU 896 HD #29

Closed danielmellado closed 3 years ago

danielmellado commented 4 years ago

Coming from https://github.com/takaswie/snd-firewire-improve/issues/28

danielmellado commented 4 years ago

@takaswie this is the output from the script how ;)

$ ./motu896hd.py 
Detect clock status:
  0x0b14: 00000000
    rate configuration: 44100
    src configuration: internal
Detect I/O configuration:
  0x0c04: 03000006
    opt out iface configuration: None
    opt in iface configuration: None

In terms of the physical knobs, the do control monitor level/volume + mix 1/4 and meters, if pressed.

takaswie commented 4 years ago

@danielmellado Great ;)

In terms of the physical knobs, the do control monitor level/volume + mix 1/4 and meters, if pressed.

Hm. Although I expected that users can configure digital I/O port (optical interface) and clock configurations (source and frequency), the configuration is just available via GUI software according to manual. (In MOTU 828mkII or later, users can configure them by physical knobs.)

It means that I should update the script so that we can configure these features and check current status. Would you give me a bit time for the additional work?

Thanks for your patience.

danielmellado commented 4 years ago

Hi @takaswie sure, I have also checked via ffado-mixer, and although the optical I/O can be set to ADAT/Disabled, it doesn't seem to honor it, as well as sample rate. I have been able to change samplerate using jack and can provide another output just so we make sure it works there ;)

$ ./motu896hd.py 
Detect clock status:
  0x0b14: 00000018
    rate configuration: 96000
    src configuration: internal
Detect I/O configuration:
  0x0b14: 03000006
    opt out iface configuration: None
    opt in iface configuration: None

Regarding the time, for sure! You have been extremely responsive, so thanks a lot! ;)

danielmellado commented 4 years ago

Ok, so in the end I was able to change them using ffado-mixer and it does seem to work ;), @takaswie that might save you some work ;)

[dmellado@fedora  alsa]$ ./motu896hd.py 
Detect clock status:
  0x0b14: 00000000
    rate configuration: 44100
    src configuration: internal
Detect I/O configuration:
  0x0b14: 03000506
    opt out iface configuration: ADAT
    opt in iface configuration: ADAT
danielmellado commented 4 years ago

btw, forgot to mention that I'm also logged in as dmellado in freenode's #alsa channel, should you want to ping me. I do also run a bouncer so would appear usually online.

takaswie commented 4 years ago

Hi @danielmellado ,

I'm sorry to be late for reply, but I was a bit busy for new releases of libhinawa[1]/libhinoko[2] and their rust bindings[3][4].

Ok, so in the end I was able to change them using ffado-mixer and it does seem to work ;), @takaswie that might save you some work ;)

It might be against your expectation that ALSA firewire stack is not a port of libffado from user land to kernel land. I have been doing investigation and reverse-engineering again apart from FFADO implementation. Thus it is not preferable that the check procedure is done by FFADO tools, somehow...

The most of MOTU devices have on-board 'cuemix' feature and I can get current the status of device via LCD by controlling knobs after operating by software. However, this procedure can not be applied to your case, mmm.

I attempt to get 896HD in used market for reverse-enginnering, but at present there's no guarantee, sorry.

[1] https://github.com/alsa-project/libhinawa [2] https://github.com/takaswie/libhinoko [3] https://github.com/alsa-project/hinawa-rs/ [4] hinoko-rs is in my private repository.

takaswie commented 4 years ago

@danielmellado ,

I attempt to get 896HD in used market for reverse-enginnering, but at present there's no guarantee, sorry.

If I fortunately get it, I'll prepare for kernel patch to add support for the device. Before posting, I'm going to ask you for test.

takaswie commented 4 years ago

If I fortunately get it, I'll prepare for kernel patch to add support for the device. Before posting, I'm going to ask you for test.

The misfortune comes to me. The bought MOTU 896HD is in disorder and can not detected on IEEE 1394 bus, sigh...

danielmellado commented 4 years ago

Hi @takaswie, just read all this, wow! Thanks a lot for trying that out! May I help you with some contribution for the MOTU thing? In any case if it's used and not working that's a pain... Should we be leaving somehow close I would ship you mine and lend it to you for the testing... Is there anything I can do from my side?

Basically my idea on the test was, as I can't modify the outputs physically as we'd do with a 828mkII or something, just using the hardware buttons, to do that using the ffado mixer, but only as a way to test that with alsa, so just somehow mimicking the 'hardware buttons'. Another option would do that I try to hook it up into a amc and try changing the settings with the cuemix (let's see, latest software seems to be "MOTU Universal Audio Installer for Mac OS X (1.6.59644), from 2014".

So, summarizing, if we use ffado-mixer as a way to mimic the settings on the hardware, them later on they seem totally detected from your script, both ADAT and sample rate changes. Is this something we could use from a starting point?

Thanks a lot for your help and efforts!

danielmellado commented 4 years ago

What we might do is to have a shared tmate session maybe, so you can get access to the hardware, what do you think?

takaswie commented 4 years ago

Hi,

I'm sorry to be late reply, but was a bit busy for this PR[1].

Is there anything I can do from my side?

Basically my idea on the test was, as I can't modify the outputs physically as we'd do with a 828mkII or something, just using the hardware buttons, to do that using the ffado mixer, but only as a way to test that with alsa, so just somehow mimicking the 'hardware buttons'.

Another option would do that I try to hook it up into a amc and try changing the settings with the cuemix (let's see, latest software seems to be "MOTU Universal Audio Installer for Mac OS X (1.6.59644), from 2014".

So, summarizing, if we use ffado-mixer as a way to mimic the settings on the hardware, them later on they seem totally detected from your script, both ADAT and sample rate changes. Is this something we could use from a starting point?

Thanks for your dedication and suggestions. Here, I explain my way to work for ALSA firewire stack.

I have IEEE 1394 protocol hardware analyzer[2]. My work is based on analysis of actual packet on IEEE 1394 bus, sniffed by the analyzer. I manage to rely on the analysis result to exclude FFADO implementation from ALSA firewire stack. This might be against your expectation. But I'm walking on the winding road apart from FFADO since my work is not porting to ALSA land. (In this point, it's similar to Wayland project against Xorg, maybe).

So it's our last resort to utilize FFADO stuffs for the work.

What we might do is to have a shared tmate session maybe, so you can get access to the hardware, what do you think?

I open 896HD and ensure that the hardware is quite similar to MOTU Traveler, expectedly[3]. The hardware consists of four ICs to process isochronous packets as Traveler does:

I have a conjecture that 896HD can be controlled by the same way for Traveler. It's mostly reliable that operations in my scripts are appropriate to 896HD.

However, 896HD has some features which Traveler doesn't have. Furthermore, hardware notification is not clear yet[4].

Anyway, I need time to make enough strategy for 896HD with our cases.

[1] motu: add service program for devices in MOTU FireWire series https://github.com/alsa-project/snd-firewire-ctl-services/pull/13 [2] DAP Design FireSpy 801 https://www.daptechnology.com/products/bus-analyzer/single-bus/single-bus-gen2/firespy810410bt/ [3] You can see my code comment in source code of Linux kernel. https://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git/tree/sound/firewire/motu/motu-protocol-v2.c#n173 [4] FFADO doesn't handle the hardware notification.

takaswie commented 3 years ago

Hi @danielmellado ,

Recently I worked to add support for MOTU 896 in development period for v5.14 kernel[1] and realized that both 896 and 896HD have similar features except for packet streaming function. The experience take me to make inference about specification of 896HD.

My 896HD is still broken. If you are still interested in ALSA support for 896HD, would I ask you to test two softwares?

  1. I pushed topic/motu-896hd remote branch, which includes the recent backport drivers and patches for 896HD. When installing it by DKMS or anything, snd-firewire-motu driver is available to handle 896HD.

  2. I also pushed topic/motu-896hd remote branch into snd-firewire-ctl-services project[2]. When installing, snd-firewire-motu-ctl-service program is available via cargo run, which can dispatch requests from ALSA control applications to control 896HD. You can configure 896HD by standard ALSA control applications such as amixer or alsamixer.

The points we should check are:

I note that CueMix feature is not implemented yet, thus it's preferrable to use phone output, to which usually the first two channels in PCM frame are assigned.

[1] [PATCH 0/2] ALSA: firewire-motu: add support for MOTU 828 and 896 https://lore.kernel.org/alsa-devel/20210616082847.124688-1-o-takashi@sakamocchi.jp/ [2] https://github.com/alsa-project/snd-firewire-ctl-services

danielmellado commented 3 years ago

Hi @takaswie, sure! thanks for your efforts on this, I'll take a look as soon as I can allocate some time ;)

takaswie commented 3 years ago

@danielmellado Thanks to grant my request. As another request, would I ask you to make merge request to add the image of configuration ROM for your MOTU 896HD?

The image is available by below command when detecting the node as fw1:

$ cat /sys/bus/firewire/devices/fw1/config_rom > am-config-rom/audio_and_music/motu/motu-896.img

The content of image is used by Linux FireWire stack to detect operable instances of any node on IEEE 1394 bus. I use the image to create better list of sytemd hardware database:

For MOTU 896HD, I've already added corresponding entry to the database (see line 981-985). However, metadata in the image is still useful to me.

Regards

takaswie commented 3 years ago

Hi @danielmellado ,

24 days past since I contact to you but no reactions I have.

If you already lose interests in the topic, I'm sorry to say that I think it better to close the issue. Would I ask your situation?

takaswie commented 3 years ago

Further 7 days past, but no reaction. I'll close the issue and abandon the topic remote branches.

Thanks

takaswie commented 3 years ago

Today I got a new 896HD in used market. Fortunately communication function is alive.

I prepared and posted a path to upstream: https://lore.kernel.org/alsa-devel/20210823085741.33864-1-o-takashi@sakamocchi.jp/

And it was applied to main branch: https://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git/commit/?id=23c671be97b9e49846d03ceb0bce3731f4b869ac

The support will be included in Linux kernel v5.15, scheduled to be released a few months later.

danielmellado commented 3 years ago

Hi Takashi, sorry I didn't have the chance to test this, covid around family... Thanks for all your help! You made my day! I'll take a look at it!

El lun., 23 ago. 2021 11:37, 坂本 貴史 @.***> escribió:

Today I got a new 896HD in used market. Fortunately communication function is alive.

I prepared and posted a path to upstream:

@.***/

And it was applied to main branch:

https://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git/commit/?id=23c671be97b9e49846d03ceb0bce3731f4b869ac

The support will be included in Linux kernel v5.15, scheduled to be released a few months later.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/takaswie/snd-firewire-improve/issues/29#issuecomment-903605117, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKAVQ2RZU7R47D6TMNNK73T6IJGHANCNFSM4PWOFM4Q . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email .

takaswie commented 3 years ago

Basic control functions are also available via snd-firewire-motu-ctl-service program as well.

[PATCH 0/2] motu-protocols/motu-runtime: add support for MOTU 896HD https://github.com/alsa-project/snd-firewire-ctl-services/pull/61