LineageOS, Current status : 16.0 Test Builds

[SteffenWi 139]

Hello, welcome 'back' @tdm and thank you to y'all for investing time here!

[EskeRahn 5,461]

16 minutes ago, SteffenWi said:

Hello, welcome 'back' @tdm and thank you to y'all for investing time here!

Indeed. And for anyone that would like to say a more substantial thanks, I will refer to this:

 

[Wheeljack 212]

'I have some bluetooth issues with the handsfree system of both an Audi A1 and a VW Golf 7. Not entirely sure if this is related to this thread: https://community.fxtec.com/topic/2899-bluetooth-issues/

On both test11 and test12 my pro1 can connect properly to both cars. I can start and receive a call with the car's controls. The audio however remains with the pro1 earpiece and microphone. When I look at the pro1 display, audio is set to bluetooth. Setting it to phone and back doesn't make a difference. I can set it to speakerphone though.

I tried setting BT audio codec to SBC only.

I also have a Sena 10uPad for my bike helmet and a Sennheiser BT headset. Both work fine as handsfree devices with the pro1.

[tdm 2,322]

7 hours ago, acrux said:

I tried hard to find the codes (mostly as trial&error procedure) for still missing Fn+ keys, but failed. It seems that the symbols missing are currently tied to the Sym key acting as Altgr key for keys corresponding to pc105 keyboard german qwertz layout not for Pro1 german qwertz layout?

If somebody guides me to the usable keycode table then I'll continue.

It was very easy in SailfishOS to get the keyboard layout you like, but that is Linux... 😉

The original output.txt was missing some corrections. I provided the corrected one in my original post and also here.

output1.txt 596 B · 2 downloads

I am not very familiar with international keyboards.  It is only through this project that I even learned the basics about how non-US keyboards worked.  But I'm learning.  🙂

 

Unfortunately, keyboards are hard.  There are multiple layers of translation involved from the physical key press to what you see on the screen.  The kernel keymap is the lowest level, which makes it difficult to choose the correct values that generate the symbol you wish to see.

 

I've downloaded your attached file and each line has three symbols on the left and a comment on the right.  I presume these three symbols represent what you see from the keypress for (1) no modifiers, (2) shift, and (3) yellow-arrow.  Is that correct?  If so, then the first two columns seem fine for all keys and there are eleven keys which need the yellow-arrow symbol adjusted.  Was this done with my modified qwertz boot image, or with the standard test11/test12 boot image?

 

I will try to share my thought process for finding the codes for the keymap, in hopes that it will help:

 

1. Select a key and symbol that you wish to map.  Let's choose Q since it is the first key in your text file that needs adjusted.  This should be key 46 (you can use drivers/input/keyboard/qx1000.c in my kernel source to look this up, see qwertz_keyz[] around line 304..329).  The yellow symbol is @.

 

2. Note the normal unshifted key value in Linux kernel file "include/uapi/linux/input-event-codes.h".  In this case, KEY_Q has value 16.  In hex, this is 0x0010.

 

3. Look up the desired yellow symbol on a standard German or other US-INTL keyboard and see how it is generated.  This is the part that is difficult for me, as I do not have a QWERTZ Pro1 and I have never used a US-INTL keyboard -- I have only images to work with for both.  It seems @ is generated by shift+2.

 

4. Look up the key code for the base key in Linux kernel file "include/uapi/linux/input-event-codes.h".  in this case, KEY_2 has value 3.

 

5. Note any modifiers needed.  In this case, a normal shift.  My keymap file uses 0x8000 for this.

 

6. Combining the above, we get 0x8000 + 3, or 0x8003 for our desired yellow-shift value.

 

7. Our keymap line should be "46:0010:8003".

 

Does that make sense?  Can you test and see if it works?

 

Note that some symbols may be generated with AltGr.  I am not sure what physical keycode AltGr generates.  I am guessing KEY_RIGHTALT?  We will need to investigate that.

 

PS: Both Sailfish and Android are Linux, technically.  I think it would be more accurate to say that Sailfish was easy because it is X11 based and that is easier to work with because it's a known system with much more documentation.

 

[tdm 2,322]

56 minutes ago, Wheeljack said:

'I have some bluetooth issues with the handsfree system of both an Audi A1 and a VW Golf 7. Not entirely sure if this is related to this thread: https://community.fxtec.com/topic/2899-bluetooth-issues/

On both test11 and test12 my pro1 can connect properly to both cars. I can start and receive a call with the car's controls. The audio however remains with the pro1 earpiece and microphone. When I look at the pro1 display, audio is set to bluetooth. Setting it to phone and back doesn't make a difference. I can set it to speakerphone though.

I tried setting BT audio codec to SBC only.

I also have a Sena 10uPad for my bike helmet and a Sennheiser BT headset. Both work fine as handsfree devices with the pro1.

 

Audio issues in cars can be difficult, both because it is more challenging to get logs (most folks don't carry their laptops with them in the car) and because each car manufacturer seems to have their own unique BT implementation.

 

First question, how does stock behave?

 

Second question, are you able to get a logcat of this happening?

 

EDIT: Also, please file an issue on github.  Issues get lost easily on the forum.

 

Edited June 3, 2020 by tdm

[VaZso 1,998]

6 hours ago, tdm said:

I am not very familiar with international keyboards.  It is only through this project that I even learned the basics about how non-US keyboards worked.  But I'm learning.  🙂

 

6 hours ago, tdm said:

Note that some symbols may be generated with AltGr.  I am not sure what physical keycode AltGr generates.  I am guessing KEY_RIGHTALT?  We will need to investigate that.

Basically international layouts were based on US keyboard map but some of the keys, mostly non-alphanumeric ones were changed (as we have more alphanumeric letters than in English).
They kept CTRL + ALT, in Hungarian layout, their name are the same, in German layout, they were renamed (but still the same).
The "Alt Gr" key is an additional modifier, which is technically the right Alt button on US keyboard.

For example, in Hungarian layout, "[" and "]" can be reached by pressing AltGr+F,G and "{" and "}" by pressing AltGr+B,N and also Z / Y are interchanged and 0 moved to the right of 9 - basically this is the reason I don't use this layout for programming.

[EskeRahn 5,461]

15 hours ago, tdm said:

3. Look up the desired yellow symbol on a standard German or other US-INTL keyboard and see how it is generated.  This is the part that is difficult for me, as I do not have a QWERTZ Pro1 and I have never used a US-INTL keyboard -- I have only images to work with for both.  It seems @ is generated by shift+2.

Please note that the QWERTZ and QWERTY version of the Prpo1 are COMPLETELY identical internally, the ONLY difference is what is printed on the keys.

So you can test qwertZ on a qwertY Pro1 (or vice versa). (Obviously testing like this is slightly confusing as the qwertY letter print is shifted compared to the qwertZ)

[acrux 56]

17 hours ago, tdm said:

I've downloaded your attached file and each line has three symbols on the left and a comment on the right.  I presume these three symbols represent what you see from the keypress for (1) no modifiers, (2) shift, and (3) yellow-arrow.  Is that correct?  If so, then the first two columns seem fine for all keys and there are eleven keys which need the yellow-arrow symbol adjusted.  Was this done with my modified qwertz boot image, or with the standard test11/test12 boot image?

Yes, correct - (1) no modifiers, (2) shift, and (3) yellow-arrow.

This is the result with your modified qwertz boot image. Thanks!

Please note that " lLL     should be lL' " can be easily fixed with the line "47:0026:802b" which by some reason did not get into that "modified qwertz boot image".

Edited June 4, 2020 by acrux

[acrux 56]

18 hours ago, tdm said:

7. Our keymap line should be "46:0010:8003".

Does that make sense?  Can you test and see if it works?

Unfortunately not. 😞 This generates qQ" eg yellow-arrow + q generates ".

If you look at the current keymap (/sys/bus/i2c/drivers/aw9523b/6-0058/keymap), then you can find a line "60:0003:8003" - it is for key 2 and 8003 generates " as it should do. Actually that kind of problem restricted the keys I was able to adjust to these I posted in issue35 😉

However @ can be generated with the help of Sym key. Sym + q generates @. 😉 And all other these still missing yellow keys which should be mapped to "yellow-arrow"...

Created a file symkey.txt which describes the current Sym key mappings.

symkey.txt

Edited June 4, 2020 by acrux

[Wheeljack 212]

20 hours ago, tdm said:

Second question, are you able to get a logcat of this happening?

Okay, I did a logcat with the pro1 connecting to my car (BT name is "Audi A1").

From what I can tell the call starts at 06-04 13:52:06.897

I did cut several hours of stuff from this morning - still very verbose. Hope this helps for now. I'll see if I can set up a proper strategy for doing a full backup and restore, so I can test with stock.

logcat_BT.txt

[tdm 2,322]

Okay I see what's going on with QWERTZ now.  All of the incorrect keys (except possibly apostrophe) are accessed via AltGr.  So I'll add that as a modifier in my driver and redo the layout for these keys.  I should have a new boot image later today.

 

[tdm 2,322]

Alright qwertz folks, here's another test boot image.  Please try and let me know how the keyboard works.  Most keys should be correct, but I may have made a mistake or two...

 

http://files.nwwn.com/android/pro1/boot-qwertz-test2.img

 

Note that I have added a flag for AltGr = 0x1000.  So we have:

0x8000 = shift (left shift)

0x4000 = ctrl (left ctrl)

0x2000 = alt (left alt)

0x1000 = altgr (right alt)

 

Note there is only space for one more flag.  If more flags are needed, I will probably expand the codes from 16 bits to 32 bits.

 

[acrux 56]

48 minutes ago, tdm said:

Alright qwertz folks, here's another test boot image.  Please try and let me know how the keyboard works.  Most keys should be correct, but I may have made a mistake or two...

Thanks a lot! Seems to be mostly correct.

I guess the yellow-key pluss l should give ' (apostrophe), at moment it gives ' (acute accent) if my eyes see them correctly

Also yellow-key pluss ü gives ~ but for it to be really usefull should give dead_tilde...

Thanks again for a great progress!

Edited June 4, 2020 by acrux

[tdm 2,322]

7 hours ago, Wheeljack said:

Okay, I did a logcat with the pro1 connecting to my car (BT name is "Audi A1").

From what I can tell the call starts at 06-04 13:52:06.897

I did cut several hours of stuff from this morning - still very verbose. Hope this helps for now. I'll see if I can set up a proper strategy for doing a full backup and restore, so I can test with stock.

logcat_BT.txt 385.44 kB · 1 download

 

This looks interesting:

06-04 13:52:08.223   731  1420 D voice   : voice_start_usecase: enter usecase:voicemmode1-call
06-04 13:52:08.224   731  1420 D audio_hw_primary: select_devices for use case (voicemmode1-call)
06-04 13:52:08.224   731  1420 D audio_hw_extn: audio_extn_should_use_handset_anc: AANC enabled in the property
06-04 13:52:08.224   731  1420 D audio_hw_extn: audio_extn_should_use_handset_anc: AANC enabled in the property
06-04 13:52:08.224   731  1420 D audio_hw_primary: select_devices: out_snd_device(15: voice-handset) in_snd_device(72: handset-mic)

 

I wonder if you can provide a log from a call with your BT headset... perhaps comparing them would make the issue stand out.

 

Edited June 4, 2020 by tdm

[Wheeljack 212]

12 hours ago, tdm said:

I wonder if you can provide a log from a call with your BT headset... perhaps comparing them would make the issue stand out.

Sure, no problem.

Is this the part you were looking for?

06-04 22:54:39.475   736  1768 D voice   : voice_stop_usecase: exit: status(0)
06-04 22:54:39.475   736  1768 D voice_extn: update_calls: cur_state=1 new_state=1 vsid=11dc5000
06-04 22:54:39.475   736  1768 D audio_hw_extn: audio_extn_set_anc_parameters: anc_enabled:0
06-04 22:54:39.475   736  1768 D audio_hw_spkr_prot: audio_extn_fbsp_set_parameters: Speaker protection disabled
06-04 22:54:39.475   794  7273 I bt_a2dp_hw: adev_set_parameters: state 5
06-04 22:54:39.475   794  7273 I bt_a2dp_hw: out_set_parameters: state 5 kvpairs call_state=1;vsid=297816064

...

06-04 22:54:39.482   736  1300 D audio_hw_primary: adev_set_parameters: enter: call_state=1;vsid=297816064
06-04 22:54:39.483   736  1300 D voice_extn: update_call_states is_call_active:0 in_call:1, mode:2
06-04 22:54:39.483   736  1300 D voice_extn: update_calls: enter:

 

logcat_headset.txt

logcat_sena.txt

Edited June 5, 2020 by Wheeljack
logcat for Sena also added

[Derecho 7]

19 hours ago, tdm said:

Alright qwertz folks, here's another test boot image.  Please try and let me know how the keyboard works.  Most keys should be correct, but I may have made a mistake or two...

 

http://files.nwwn.com/android/pro1/boot-qwertz-test2.img

 

Note that I have added a flag for AltGr = 0x1000.  So we have:

0x8000 = shift (left shift)

0x4000 = ctrl (left ctrl)

0x2000 = alt (left alt)

0x1000 = altgr (right alt)

 

Note there is only space for one more flag.  If more flags are needed, I will probably expand the codes from 16 bits to 32 bits.

 

Ooh, am I reading this correctly? Is it now possible to make custom keymappings using the slanted arrow key to get special characters that require AltGr and thus far needed the Sym key to be pressed? This would be a lifechanger for me! As I described in a post a month back, on test build 2 I'd use the slanted arrow keys for all the special Polish characters I need, but updating to build 11 (iirc) meant I was stuck to using Sym for those keys and it was impossible to remap them manually in the way I wanted. So if this is now possible thanks to this additional modifier, that's great! It really was the biggest drawback with the newer builds for me, enough of a bother to actually consider running the outdated build 2 or going stock again. Over the past month or more I've worked from home due to the pandemic thing anyway, so I've been mostly avoiding typing on the phone altogether, but in the few situations I had to it was very frustrating as Sym is not a convenient key to use in combination with many of the other keys on the keyboard. As the slanted arrows are represented on both the left and the right side, they're much more suitable for this.

I do wonder about one combination though, the L key for example; This has a question mark printed on it and currently the slanted arrow is used to type it. I'd remap the slanted arrow to produce altgr with L in order to get Ł. How would I go about typing a question mark? Ideally I'd do this by pressing Sym+L, but as I understand that still sends a AltGr+L code so that would also result in Ł instead of a question mark. I guess for now I could have L be the one exception, fortunately it is on the right side of the keyboard so using it with Sym is doable. It'd be nice to make it consistent with the rest of the keys though, in that I'd be using the slanted arrow keys for it. If there's something I can do to help out on this matter please let me know.

And one other issue/bug with the build 11 that is quite noticable is the headphone volume. At first I thought the slider for in-call volume didn't work at all, but after more testing I found out that it does. It's just that the minimum possible volume, is already too loud for most of my earphones. I'm now limited to using one specific set of earphones and try and position them as far out of my ear as possible, along with having the volume for in-call set to the lowest position. I've done some researching online and I've come to understand that it'd be possible to change the starting volume for these things in a file called mixer_paths.xml , but I'm not sure which of those values need changing and to what level. I believe this is also an issue on stock, it's been mentioned on the forum by several people already with no seeming fix. Is this something that we could see fixed in LOS?

My apologies for welcoming you back by burying you with more requests/issues, I do really appreciate all the work you've done for us into making LOS a thing for this phone and it's amazing how it's generally better than the stock rom. If I can help out someway into polishing the LOS rom into something perfect, do tell and I'll give it a try.

[tdm 2,322]

2 hours ago, Derecho said:

Ooh, am I reading this correctly? Is it now possible to make custom keymappings using the slanted arrow key to get special characters that require AltGr and thus far needed the Sym key to be pressed? This would be a lifechanger for me!

 

Yes, with the qwertz test2 boot image you may specify AltGr as a modifier.  It seems this is very close to the final desired mappings, so I will incorporate it into test13.

 

2 hours ago, Derecho said:

I do wonder about one combination though, the L key for example; This has a question mark printed on it and currently the slanted arrow is used to type it. I'd remap the slanted arrow to produce altgr with L in order to get Ł. How would I go about typing a question mark? Ideally I'd do this by pressing Sym+L, but as I understand that still sends a AltGr+L code so that would also result in Ł instead of a question mark.

 

Because the '?' symbol is generated with shift, you would probably need to choose another key to map for use with the yellow-arrow.  Perhaps the "O" key, for example:

44:0018:8035

 

And, now that you mention it, the SYM key does also generate KEY_RIGHTALT, aka. AltGr.  So I am wondering if that is still desirable...?

 

2 hours ago, Derecho said:

And one other issue/bug with the build 11 that is quite noticable is the headphone volume. At first I thought the slider for in-call volume didn't work at all, but after more testing I found out that it does. It's just that the minimum possible volume, is already too loud for most of my earphones.

 

Thanks for the report.  Please file an issue on github.  I should be able to fix it.

 

2 hours ago, Derecho said:

My apologies for welcoming you back by burying you with more requests/issues, I do really appreciate all the work you've done for us into making LOS a thing for this phone and it's amazing how it's generally better than the stock rom. If I can help out someway into polishing the LOS rom into something perfect, do tell and I'll give it a try.

 

Thank you.  🙂  I'm not sure how you can help... that depends on your experience.  If you have never built Android, that would be a good start... build your own ROM.

 

[Craig 1,435]

I'm looking forward to being able to re-assign at least the right slant arrow.  For people using international layouts (including us international), it could be useful as another right-alt key in addition to Sym.  For people using plain english on qwerty, it could be useful as a slash/quesitonmark key.

Edited June 5, 2020 by Craig

[Derecho 7]

On 6/5/2020 at 7:12 PM, tdm said:

 

Yes, with the qwertz test2 boot image you may specify AltGr as a modifier.  It seems this is very close to the final desired mappings, so I will incorporate it into test13.

 

 

Because the '?' symbol is generated with shift, you would probably need to choose another key to map for use with the yellow-arrow.  Perhaps the "O" key, for example:

44:0018:8035

 

And, now that you mention it, the SYM key does also generate KEY_RIGHTALT, aka. AltGr.  So I am wondering if that is still desirable...?

 

 

Thanks for the report.  Please file an issue on github.  I should be able to fix it.

 

 

Thank you.  🙂  I'm not sure how you can help... that depends on your experience.  If you have never built Android, that would be a good start... build your own ROM.

 

The boot image would be the recovery one, yes? I didn't think the keyboard driver could would be in there, I'm actually still running the recovery image from build 2 I think yet I do have the updated keyboard behaviour. Maybe I upgraded and forgot?

Hm I guess I could put the question mark on O indeed, then I could have L behave consistently as the other keys would, with using slanted arrows for the AltGr key combinations. Thanks.

Well, ideally the Sym key would be configurable too. Personally, what'd fix things for me the best, is simply swapping current behaviour of the slanted arrows and sym key. Then I'd use Sym for all the yellow characters, and the slanted arrows for AltGr with stock keyboard files and no custom remapping. But I had the impression that the slanted arrows and sym keys operated differently in the driver, one being on a keyboard matrix and one being on some GPIO lines, so I'm not sure how trivial it is to have it all this configurable.

Issue created (#57).

I have built Android before, LineageOS specifically even, but this was way back on the Photon Q. My Android and Java (and build tools) experience is very minimal, but I am quite familiar with other embedded sytems. 

[Wheeljack 212]

1 hour ago, Derecho said:

I'm actually still running the recovery image from build 2 I think yet I do have the updated keyboard behaviour. Maybe I upgraded and forgot?

Yeah... I thought so, too. But tdm cleared that up. Flashing Lineage also updates the bootloader.

On 6/3/2020 at 12:23 AM, tdm said:

Please note that the separate boot image is provided only for booting into Lineage recovery when switching from stock to Lineage.  The main Lineage zip file has its own (identical) copy of the boot image that is flashed along with the rest of the system.

 

[tdm 2,322]

test13 is up.  Changes:

* Updated QWERTZ keymap.

* Addressed some SELinux denials.

 

If you have a QWERTY device, there is pretty much no reason to upgrade from test12.

 

If you have a QWERTZ device, please test and provide feedback.  Please remember to remove any custom keymap before testing.

 

[tdm 2,322]

4 hours ago, Derecho said:

The boot image would be the recovery one, yes? I didn't think the keyboard driver could would be in there, I'm actually still running the recovery image from build 2 I think yet I do have the updated keyboard behaviour. Maybe I upgraded and forgot?

 

The Pro1 has a combined boot/recovery partition.  This is why my Lineage install page instructs you to download and flash the boot image and then boot into Lineage recovery.

 

4 hours ago, Derecho said:

I have built Android before, LineageOS specifically even, but this was way back on the Photon Q. My Android and Java (and build tools) experience is very minimal, but I am quite familiar with other embedded sytems. 

 

Great, so you should be able to build your own ROM.  Once you can do that, let me know and we can talk about stuff to fix.  🙂

 

[EskeRahn 5,461]

10 hours ago, tdm said:

test13 is up.  Changes:

* Updated QWERTZ keymap.

Thanks.
It took me a while to guess how to find the qwertY/qwertZ setting. I browsed through the Settings-tree unable to guess where it is.

Gave up and tried searching for "qwerty" or "qwertz" that did not find it either.

BUT if I do a search for "keyboard" or "layout" it do pop up on the search-list. 🙂

Would be nice if it had some natural entry in the settings-tree, preferably under

• Settings ► Language & Input ► Physical keyboard

Or perhaps in a special top level setting group for all the customizations settings that are device specific. e.g. "Pro¹" or "Pro¹ specific"

Both the qwertY and qwertZ seem to work just fine for German and Danish. (haven't looked for a lot of details, just checked the obvious)

Choosing Danish, it works really well for qwertZ. I'm a little confused on the functionality triggered by FN+Æ (that is the L key on shifted qwertY). I get a special cursor and has "Sticky FN" . By the Print I expected a Backslash. The similar goes for Fn+Z (that is the ] key on shifted qwertY), where the print indicates vertical bar), and a couple of others.

Please note I'm not actually using Lineage on the qwertZ device so not currently a real issues for me, just reporting it.

 

Really interesting with the custom keymap too. I just have to find out have to do that, but guess I have to re-read some of the posts from the last week or so. 😉

ADD: found the path in this post some months old

On 2/25/2020 at 9:32 PM, tdm said:

So you can create a file "/persist/data/keyboard/keymap" that has the keymap that you like and it will be loaded at boot.

[EskeRahn 5,461]

11 hours ago, tdm said:

If you have a QWERTZ device, please test and provide feedback.  Please remember to remove any custom keymap before testing.

Would it be possible to make access to this folder available without root, or to fetch it from a different user-available location?

C:\util\ADB>adb shell ls -R /persist/data/keyboard/
ls: /persist/data/keyboard/: Permission denied

Would deselect of "Custom  layout" under keyboards not work just as well as removing any added file, for the test purpose?

IDEA:
A bit similar to switching languages with Ctrl-Space it COULD be handy with some master key-combo that turned off the usage of any custom layout, if people have messed it up completely. I would suggest a combo hard to do (so unlikely to conflict with anything sensible), say something like Ctrl+Sh+Fn+Alt+Sym+ESC

[Wheeljack 212]

The L key hasn´t changed since the qwertz-test2.img has it?

Still getting the áccént ´instead of '