Bug fix + better support of 18i20 Gen3 device

[sadko4u]

This set of patches enhances support of Focusrite Scarlett 18i20 Gen3 and other Gen3 devices.

The list of changes:

• Fixed bug that moved clients (like alsamixer) into infinite loop when changing any control in the UI.
• Added proper INST button controlling and synchronization with audio device.
• Added proper PAD button controlling and synchronization with audio device.
• Added software 48V phantom power switching support.
• Implemented proper ALT button synchronization with audio device.
• Implemented TALKBACK button support and synchronization with device.
• Added myself to the list of authors.

So there are no buttons left on the 18i20 device (except power button) that can not be enabled by software. Also software now properly handles all manual button presses and triggers client mixer software to update.

[sadko4u]

We're moving forward. Now talkback function can be assigned to each output of the internal mixer.
The thing I still don't like is that the driver doesn't obtain the settings of the internal matrix mixer from the device. This also should be fixed.

[mchodzikiewicz]

Hi, I am willing to test these patches soon, is this branch worth trying or should I wait for for more commits?

[sadko4u]

Hi, I am willing to test these patches soon, is this branch worth trying or should I wait for for more commits?

This branch is still in progress but I always check what I commit and try to not commit half-working functions.
Currently you can get almost full control over device except persistent routing settings. This still requires reverse-engineering.
So, in fact, you may try this branch and report your experience. But you should take into consideration that some functions are still not working.

[sadko4u]

Latest changes: now mixer state is persistent, the driver handles software configuration area of device and configures the mixer when the device is connected to USB port.

[skobkin]

@mchodzikiewicz I've tried to apply this PR as a patch (as of today 16 of October) to 5.9.0, but looks like there are too many differences already 😞

patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 753.
Hunk #2 FAILED at 2660.
2 out of 2 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 succeeded at 228 with fuzz 1 (offset -31 lines).
Hunk #2 succeeded at 238 (offset -34 lines).
Hunk #3 FAILED at 759.
Hunk #4 succeeded at 1258 (offset -419 lines).
Hunk #5 succeeded at 1317 (offset -419 lines).
Hunk #6 FAILED at 1760.
Hunk #7 FAILED at 2517.
Hunk #8 succeeded at 1956 (offset -713 lines).
Hunk #9 FAILED at 2716.
Hunk #10 succeeded at 2099 (offset -735 lines).
4 out of 10 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 4.
1 out of 1 hunk FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 succeeded at 230 (offset -32 lines).
Hunk #2 succeeded at 238 with fuzz 1 (offset -35 lines).
Hunk #3 FAILED at 2324.
Hunk #4 succeeded at 1215 with fuzz 1 (offset -1124 lines).
Hunk #5 FAILED at 2411.
Hunk #6 succeeded at 1998 (offset -739 lines).
Hunk #7 FAILED at 2765.
3 out of 7 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 succeeded at 229 (offset -32 lines).
Hunk #2 succeeded at 241 (offset -35 lines).
Hunk #3 FAILED at 763.
Hunk #4 succeeded at 1201 (offset -421 lines).
Hunk #5 succeeded at 1567 with fuzz 2 (offset -111 lines).
Hunk #6 FAILED at 1730.
Hunk #7 FAILED at 1763.
Hunk #8 FAILED at 1789.
Hunk #9 FAILED at 1813.
Hunk #10 FAILED at 2022.
Hunk #11 FAILED at 2636.
Hunk #12 succeeded at 2014 (offset -775 lines).
Hunk #13 FAILED at 2858.
Hunk #14 succeeded at 2179 (offset -801 lines).
8 out of 14 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 169.
Hunk #2 FAILED at 242.
Hunk #3 FAILED at 269.
Hunk #4 succeeded at 242 (offset -35 lines).
Hunk #5 FAILED at 481.
Hunk #6 FAILED at 533.
Hunk #7 FAILED at 600.
Hunk #8 FAILED at 681.
Hunk #9 FAILED at 819.
Hunk #10 FAILED at 888.
Hunk #11 succeeded at 1208 (offset -421 lines).
Hunk #12 succeeded at 1223 (offset -421 lines).
Hunk #13 succeeded at 1237 (offset -421 lines).
Hunk #14 succeeded at 1407 with fuzz 1 (offset -443 lines).
Hunk #15 succeeded at 1565 (offset -443 lines).
Hunk #16 succeeded at 1590 (offset -452 lines).
Hunk #17 FAILED at 2665.
Hunk #18 succeeded at 2004 (offset -676 lines).
Hunk #19 FAILED at 2693.
Hunk #20 FAILED at 2705.
Hunk #21 succeeded at 2069 (offset -725 lines).
Hunk #22 succeeded at 2096 (offset -725 lines).
12 out of 22 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 908.
Hunk #2 succeeded at 1415 (offset -466 lines).
Hunk #3 succeeded at 1426 (offset -466 lines).
1 out of 3 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 246.
Hunk #2 FAILED at 274.
Hunk #3 succeeded at 379 with fuzz 2 (offset -323 lines).
Hunk #4 FAILED at 843.
Hunk #5 FAILED at 947.
Hunk #6 FAILED at 2435.
Hunk #7 FAILED at 2446.
Hunk #8 FAILED at 2496.
Hunk #9 FAILED at 2535.
Hunk #10 FAILED at 2543.
Hunk #11 FAILED at 2695.
Hunk #12 FAILED at 2718.
Hunk #13 FAILED at 2787.
Hunk #14 succeeded at 2163 (offset -765 lines).
12 out of 14 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 246.
1 out of 1 hunk FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 4.
Hunk #2 succeeded at 244 (offset -41 lines).
Hunk #3 FAILED at 797.
Hunk #4 FAILED at 848.
Hunk #5 FAILED at 951.
Hunk #6 FAILED at 2509.
Hunk #7 FAILED at 2572.
Hunk #8 FAILED at 2588.
Hunk #9 FAILED at 2604.
Hunk #10 FAILED at 2613.
Hunk #11 FAILED at 2782.
Hunk #12 succeeded at 2050 with fuzz 1 (offset -823 lines).
10 out of 12 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 FAILED at 3253.
1 out of 1 hunk FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/pcm.c
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 succeeded at 122 (offset -24 lines).
Hunk #2 succeeded at 257 (offset -41 lines).
Hunk #3 succeeded at 501 with fuzz 1 (offset -322 lines).
Hunk #4 succeeded at 574 (offset -332 lines).
Hunk #5 succeeded at 601 with fuzz 2 (offset -393 lines).
Hunk #6 FAILED at 1058.
Hunk #7 succeeded at 1361 (offset -447 lines).
Hunk #8 FAILED at 2194.
Hunk #9 succeeded at 1770 (offset -472 lines).
Hunk #10 FAILED at 2896.
Hunk #11 succeeded at 2158 (offset -900 lines).
Hunk #12 succeeded at 2394 (offset -903 lines).
Hunk #13 succeeded at 2474 (offset -926 lines).
3 out of 13 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej
patching file sound/usb/mixer_scarlett_gen2.c
Hunk #1 succeeded at 147 (offset -24 lines).
Hunk #2 succeeded at 1392 (offset -453 lines).
Hunk #3 FAILED at 2231.
Hunk #4 FAILED at 2248.
Hunk #5 succeeded at 1830 (offset -480 lines).
Hunk #6 succeeded at 2004 (offset -804 lines).
Hunk #7 FAILED at 2941.
Hunk #8 succeeded at 2287 (offset -916 lines).
Hunk #9 succeeded at 2299 (offset -916 lines).
3 out of 9 hunks FAILED -- saving rejects to file sound/usb/mixer_scarlett_gen2.c.rej

So looks like we should either use old kernel or wait until it finished and updated to the new versions.

[sadko4u]

@skobkin
Have you applied the original branch to which the PR is submitted to the kernel?

[sadko4u]

@skobkin
Added patch to the vanilla 5.9.0 kernel:
https://github.com/sadko4u/focusrite-scarlett-backports/blob/master/vanilla-linux-5.9.0-scarlett-gen3.patch

[skobkin]

@sadko4u

Have you applied the original branch to which the PR is submitted to the kernel?

No. Maybe that's the source of the problem.

Added patch to the vanilla 5.9.0 kernel

Should I apply it AFTER applying changes from the original repo or directly to the 5.9.0 sources?

BTW, Thanks!

[sadko4u]

No. Maybe that's the source of the problem.

I am sure that's the source of problem.

Should I apply it AFTER applying changes from the original repo or directly to the 5.9.0 sources?

This patch I've specially done for 5.9.0 kernel. I've made it by manually applying all necessary changes and taking the diff with original source, so it may be applied directly.

[skobkin]

@sadko4u

This patch I've specially done for 5.9.0 kernel. I've made it by manually applying all necessary changes and taking the diff with original source, so it may be applied directly.

Thanks! I'll try to test it soon!

[skobkin]

@sadko4u Hi :)

Did quick test of the patch.

First it didn't work at all and printed this message in the dmesg:

[    4.992620] usb 5-1: Focusrite Scarlett Gen 2/3 Mixer Driver disabled; use options snd_usb_audio vid=0x1235 pid=0x8215 device_setup=1 to enable and report any issues to g@b4.vu

I did that and added new config file in the /etc/modprobe.d:

# /etc/modprobe.d/alsa-focusrite-gen3.conf
options snd_usb_audio vid=0x1235 pid=0x8215 device_setup=1

Good things:

• A lot of parameters are exposed to the ALSA configuration. I've even saw phantom power control which is not available via Focusrite Control in Windows.
• All controls I checked (phantom power, air, 1-2 input mode switching) were causing corresponding LED's to turn on or off on the interface.

Bad things:

I think that something related to interface configuration is broken.
First thing I noticed is that when I tried to turn on phantom power my mic didn't work. Even integrated VU-meter on the interface didn't show typical peaking level when enabling phantom power via QasHctl.
Then I found out that output isn't working too. For some reason sound isn't working on Linux now even after rolling back to the kernel without the patch.
But the interface itself is fine because it's still working under Windows. But on Linux it looks misconfigured somehow. It doesn't show anything on VU-meter even when I connect my guitar or speak to the mic with phantom power enabled (manually by the press of the button). Audio input and output in the system (currently I'm using jack with pulseaudio sink) isn't working too. Turning the interface off and on using hardware power switch isn't helping too.

Quite interesting result.

If you have a suggestion about how I can fix it (on non-patched kernel), I'll be grateful.

Thanks for the work anyway and good luck with finishing and merging upstream.

[sadko4u]

First thing I noticed is that when I tried to turn on phantom power my mic didn't work.

This still has not been checked by me. I believe there's something missing in the configuration of the device. Anyway, I'll check the input when end up with routing settings.

It doesn't show anything on VU-meter even when I connect my guitar or speak to the mic with phantom power enabled

Which one? The hardware VU-meter which is on the device or the software one?

If you have a suggestion about how I can fix it (on non-patched kernel), I'll be grateful.

Currentlly have no idea. Seems really that I've something missed when configuring the device.

[skobkin]

@sadko4u

Which one? The hardware VU-meter which is on the device or the software one?

The hardware one. But software indicators also shows nothing (input and output both looks broken). Maybe routing is broken, or something like that. I'm not too familiar with low-level details.

Seems really that I've something missed when configuring the device.

If you want, I can provide a video of how it behaves and any logs (in the Telegram probably?) later.

[flowernert]

@sadko4u I'm trying to build vanilla 5.9 + your patch on my Linux Mint 20, keep you posted

If you already applied patch to your kernel once, then you just need to replace corresponding driver file - sound/usb/mixer_scarlett_gen2.c. Disconnect the device from USB port and rebuild driver:

Well actually I took this occasion to upgrade my system to its next release (Linux Mint 20, which is Ubuntu 20.04 based)
Then I tried to build the current release of Ubuntu 5.8 kernel. I had 2 conflict when applying the patch but I've been able to fix them manually and build without error, I'm using it at the moment. I can generate and provide you the patch if you want to add it to this repo for Ubuntu/Mint users.

[sadko4u]

I can generate and provide you the patch if you want to add it to this repo for Ubuntu/Mint users.

Yes, you can do a pull-request to my repository and I'll add the corresponding patch for the kernel.

[kahles]

@sadko4u Now everything seems to work - thank you!
The 8i6 has a "retain 48v"-switch and it's working now.
Further I applied the commit to my ubuntu-hwe-kernel and it also works fine there.
One strange thing: Is it normal, that the mappings of the Analogue-Inputs to PCM-channels ("PCM Out xx Source") appear in alsamixer in playback-section?

[sadko4u]

@sadko4u Now everything seems to work - thank you!
The 8i6 has a "retain 48v"-switch and it's working now.
Further I applied the commit to my ubuntu-hwe-kernel and it also works fine there.
One strange thing: Is it normal, that the mappings of the Analogue-Inputs to PCM-channels ("PCM Out xx Source") appear in alsamixer in playback-section?

This is the result of that driver is not following the ALSA parameter naming feature. Probably I'll fix it when I will polish the driver. Currently that's true.

The development is still in progress since there are still a couple of things that is required to be implemented. I'll continue my job at the end of week.

[sadko4u]

Hi all!
I've just committed experimental change to the debug driver:
sadko4u/focusrite-scarlett-backports@2a964b7

This change adds mute controls to Analogue, S/PDIF and ADAT outputs.
Since the original driver uses MUX to perform mutes of all outputs, I won't add this change to the pull request until the mixer driver will read the MUX settings from the software configuration area of the device configuration space.

Anyway, you can test mutes now but remember: if you'll restart the driver, the routing of muted channels may be broken.

[kahles]

Hi @sadko4u,

sorry for the delay, but I finally tested your debug driver:
Mute of analogue and SPDIF-outputs now works on my 8i6 - and the routing was "broken" afterwards.

Is anyone here using PulseAudio? I don't know if this is a driver-related issue:
After booting and logging in (to Cinnamon), I first have to kill the pulseaudio daemon and wait for it to respawn in order to have sound. When I turn the Saffire off and on again or when waking up from hibernate, there are no problems ...

I'm trying to get logs from pulseaudio, but this seems a bit complicated, since it automatically restarts when I kill it and it doesn't want to read my config that should enable debug output ...

[sadko4u]

Mute of analogue and SPDIF-outputs now works on my 8i6 - and the routing was "broken" afterwards.

Okay, I'll continue the development ASAP. There were couple of things I was required to do. Now I'm more free and can resume the work.

[andmarti1424]

k different to the major Gen3 devices.

Hello folks. Does this driver work for 2i2 gen3?
Thanks!

[sadko4u]

@andmarti1424
The support of 2i2 in the mixer has been added but not tested yet.

[andmarti1424]

@sadko4u Thank you for your reply. I have a question about the use of this driver though. Would this enable the routing, so each output (headphone output as well) can be set up, to mix between playback and input signals?? Thanks!!

[sadko4u]

@sadko4u Thank you for your reply. I have a question about the use of this driver though. Would this enable the routing, so each output (headphone output as well) can be set up, to mix between playback and input signals?? Thanks!!

2i2 doesn't support routing. It has no internal mixer/router at all. The only things that can be controlled are several buttons like pad/air and phantom power. And I believe that not all of them. For this purpose I need a more proper protocol dump than I have currently.

[andmarti1424]

@sadko4u Actually I do not have an 2i2 yet. Was planning on getting one and that was why I was asking that. I though the Linksrite control app allowed to control the mix between playback and input, in the monitor output. This is not the case then? If thats the case.. what a bummer. thank you again!

[sadko4u]

@andmarti1424

The best support you'll get if you buy 18i20 device :). Since I'm the owner of 18i20 and can guarantee that most of functions fork for it.

[andmarti1424]

@andmarti1424

The best support you'll get if you buy 18i20 device :). Since I'm the owner of 18i20 and can guarantee that most of functions fork for it.

@sadko4u what about 4i4 3rd Gen..? Is the mix and routing supported with this patch? Thanks!

[sadko4u]

@andmarti1424

what about 4i4 3rd Gen..? Is the mix and routing supported with this patch?

Yes, 4i4 should be supported but I didn't test it. So be ready to take dumps from device :).

[kahles]

Now I managed to get logs from pulseaudio - I tried to play a test sound, and diffed the logs ...

1. I think line 3 could be a problem:
   I: [pulseaudio] sink.c: Cannot update sample spec, SINK_IS_RUNNING, will keep using s32le and 48000 Hz
   After line 56 it hangs a few seconds.
   Full logs: pulse.no-sound.log

2. I kill pulseaudio and at first try, I also hear nothing, because the device seems to be suspended by ALSA (line 120)
   pulse.after_restart.log

3. Afterwards I have sound:
   pulse.working.log

Now I can do almost everything and still have sound. Except when I now turn off the device, kill the pulseaudio daemon and turn it on again, I seem to have the same problem as at 1. But now it can change the sample_spec and uses a input channel 3 (line 68). After line 119 it hangs again a few seconds ...
pulse.after-power-cycle.log

When I now do pulseaudio -k, I'm back at step 2.

[sadko4u]

Hi folks! I've just updated the device driver. The routing, probably the hardest part of the driver, is now working nice for 18i20 device.

• Added 'Mute' switches for all non-PCM outputs.
• Added 'Mute' switches for all mixer lines.
• Added gain halos control for 18i8 gen3 device.
• Driver code refactoring and prettifying.
• Signal routing is now fully persistent (at least for 18i20).

[sadko4u]

@kahles I can not say anything special about your problem now. I'm not a pulseaudio fan. But I got 18i20 properly working for ALSA and JACK.

Also @dlkw had some issues with patched sound/usb/pcm.c file. Try to use the file from original kernel without changes. Maybe that helps.

[sadko4u]

Added Scarlett 2i2 gen3 device support:

• Line/Inst switches support
• Air switches support
• 48V phantom power switch support
• Direct monitor switching support
• Retain 48V support
• Gain halos custom colors support

[andmarti1424]

Added Scarlett 2i2 gen3 device support:

* Line/Inst switches support

* Air switches support

* 48V phantom power switch support

* Direct monitor switching support

* Retain 48V support

* Gain halos custom colors support

Is the mixer working for 4i4 gen3? Thanks!

[sadko4u]

Is the mixer working for 4i4 gen3? Thanks!

I have no device to test. If you're ready to test, then you need to apply kernel patches from this repository and try it:
https://github.com/sadko4u/focusrite-scarlett-backports

[kahles]

Hello again,

@kahles I can not say anything special about your problem now. I'm not a pulseaudio fan. But I got 18i20 properly working for ALSA and JACK.

I'm neither a fan of pulseaudio, but I feared the trouble of replacing it ;)

Now I discovered another problem when trying to record something using the newest version of the driver (e9428f6):
I wanted route Input 1 to PCM 1, but Analog In 1 wasn't selectable in alsamixer.

After writing the first lines here I wanted to look up what is available, and suddenly all options are present o.O

[Pauldb8]

Hi,
I can confirm I have downloaded a generic 5.9.0 kernel's sources, patched it following https://github.com/sadko4u/focusrite-scarlett-backports using the vanilla-linux-5.9.0-scarlettt-gen3.patch.

Since I use Ubuntu 20.10 I helped myself with https://phoenixnap.com/kb/build-linux-kernel (up until 'sudo make install', which also updates bootloader) to download, patch and build the kernel.

When I restarted I was out of video driver. I had to uninstall nvidia proprietary drivers, cleaned them following https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#removing-cuda-tk-and-driver (Ubuntu section). Then restart, reinstall them, and restart I had proper video driver working again.

Regarding the Scarlett 18i20 driver I was hoping it would work... And it does ! I can now see a lot of options when doing 'alsamixer' in the terminal:

• PCM out 01 Source - Analogue 01 x 12
• PCM out 12 Source - S/PDIF, ADAT x 8
• Mix A In 24 - volume, Mix A Talkback
  Mix B In 24 - volume, Mix B Talkback
  ...
  Mix L In 24 - volume, Mix I Talkback
• Mix In 1... 24 Source
• ADAT Out 1...8 Source/Mute
• Analog In 01...8 Pad / Air
• Analog Out 01...8 Mute / Source
• Dim / Mute

Every Dim, Air, Instrument button works, hw and sf.
Really amazing nice work.
Now I must say there are so many things I am a bit lost, is there a GUI to better visualize this or something ?
Also, still, work using jackd and I use 18 input / 20 output. The patchbay there makes more sense than alsa :/ .

Do you have a date in mind when it will be available in mainstream kernel ?

Thank you for the patch it works great.

[sadko4u]

@Pauldb8
There are still issues with 18i8 device. @dlkw sent dumps to me but I still have no time to look at them, sorry.
Also, I found that Talkback feature is not working properly for my device, so I believe I need to check it again. It seems that I've missed something because I can not hear anyhing sent to talkback channel.

Anyway, I believe it will be ready when I end up with 18i8 device and talkback feature.
By the other side, things could get easier if someone could donate the 18i8 device to me :))).