chromebook_cherry: status: tomato (acer cp513-2h) #61
Comments
hexdump0815
changed the title
|
Hi, thanks a lot for your efforts! What works and what not:
|
|
thanks a lot for testing - that is actually way better than i expected :) for audio: what does "cat /proc/asound/cards" show? |
|
|
the chromeos ucm config would be a good start i guess - i have collected some notes about getting audio working here: https://github.com/hexdump0815/imagebuilder/blob/main/info/generic/alsa-ucm-pulseaudio-hacking.txt ... there is some limitation in the length of the ucm dir name, so it should go to /usr/share/alsa/ucm2/m8195_r1019_568 (see also the output of the /proc/alsa/cards) ... i think it is a good idea to put a "Syntax 2" on top of m8195_r1019_5682s.tomato.conf and to comment out all devices except the speakers at the beginning to check if the proper ucm file gets loaded it is a good idea to run "strace alsaucm -c myconfig reload" (will have to add it to the notes mentioned above) in general when trying to bring up audio never use headphones directly until you have it working as you might get high volume level in the process which is not good for the ears regarding video accel: i guess the chance is low to get this working with a regular linux without a lot of extra patches i guess (but i might be wrong) good luck and best wishes - hexdump |
|
this might be a good read about getting audio to work: #54 |
|
Thanks for the detail info, will try. |
|
Wow! That worked, audio is working now ! hahaha What works and what not:
Many many thanks! |
|
@zhxt - that sounds very good! could you please create a pull request with your working ucm files or upload/post them somewhere, so that i can include them? |
|
@hexdump0815 PR #64 created :D |
|
@zhxt - thanks a lot - pull request is already merged :) regarding the default to headphone what i have learned over time is that the last entry in the ucm file is used by default as it seems - so maybe you can get this fixed by just reordering the entries to have the speaker last ... same applies for the mics this is the ugly hack for kukui chromebooks to initially set the devices: https://github.com/hexdump0815/imagebuilder/blob/main/systems/chromebook_kukui/rc-local-additions.txt#L29-L73 best wishes - hexdump |
|
@hexdump0815 thanks, reorder the entries did the trick :-D for initialize: maybe add a and btw, do you think is it possible to get GPU work ? |
|
@zhxt - i think in newer ucm versions (syntax 3 and 4 i think) they restructured things completely with proper init sections (so no more enadev etc. required) and maybe there are more keywords ... chromeos has some extra keywords in their ucm files which are specific to their cras (chrome-audio-server i think) and not really part of reglar ucm i think gpu support will come with mainline support via the panfrost project, but all that is in very early stages and will still need quite a few months until it might start to get useable, so most probably some patience is required for this |
|
@hexdump0815 tried, and yeah, seems cras uses a separate program called for GPU, I was thinking, chromeos came with GPU support, and debian uses the same kernel, the point is cros is 32bit, but debian is 64bit, so if the user space binary blobs can be get from somewhere, will it work on debian ? I found that there is a overlay-cherry64 directory, maybe there will be a 64bit cros in the future :) |
|
the mali blobs are not really useable for linux as they are for the chromeos windows system or low level drm (not sure) and thus have a completely different api than what xorg or wayland are using ... but initial mali valhal support is there in panfrost, so its just a matter of time that it will be useable |
reorder the devices so that speaker and internal mic are the last ones and thus used as default devices - with this done headphones and external mic can be enabled again as well see also: #61 (comment) and the comments before
|
i have reordered the devices in the ucm files in the repo now as well: b659e96 |
yeah, need to wait for now. |
|
kukui is in very good shape - with my latest images nearly everything is working quite well except of hw video decoding, external monitors and on some devices the cameras if no usb video class devices are used ... otherwise gpu is working with panfrost, wifi, bt, suspend/resume, sound all works quite well with little bugs here and there maybe :) - see for example #52 |
|
@zhxt - i read your conversation on panfrost irc - good luck for trying mainline - please let me know here if you get anywhere near a working kernel with this ... |
Sounds great!
yeah, I haven't tried that kernel yet. I will let you know if I get any progress :D |
hexdump0815
changed the title
hexdump0815
changed the title
|
Hey, finally got some time to try. 😆 |
|
@zhxt - from irc i remember some trouble with the /dev/dri device numbering (https://oftc.irclog.whitequark.org/panfrost/2022-10-24 and the next days) - maybe try something like https://github.com/hexdump0815/imagebuilder/blob/main/files/extra-files/etc/X11/xorg.conf.d.samples/13-panfrost-swapped-dri-nodes.conf best wishes and good luck - hexdump |
|
@hexdump0815 yeah, I saw that too, but I didn't find any log related Xorg, when I start lightdm manually from terminal, it just jump to the black screen with a flicking cursor. |
|
@hexdump0815 wow, tried the patch, that did the trick. thanks. |
|
@zhxt - are you using the latest version of mesa? |
|
|
@zhxt - i have built an experimental image with a v6.1 kernel which has working panfrost gpu support - it has not yet been tested on mt8195/tomato (only on mt8192 so far) and it would be interesting to hear if it is working - in case you want to give it a try: https://github.com/hexdump0815/imagebuilder/releases/tag/230109-01 |
|
following the untimely demise of my CP513-1H was I was 90% of the way setting it up, i got the CP513-2H |
|
And I am back |
|
@bruceloco - first would be trying to figure out if it boots at all - for that it might be good to add something like "dmesg > /root/dmesg.txt; shutdown -h now" to rc.local and then you can see if the system will shutdown and if the file exists afterwards ... if that is working then it is a good idea to make sure (with some well working kernel) that you end up with a working network connection when fully booted (best lower level like usb ethernet as it is easier than when using wifi) - if that works you may login via ssh even if the screen, kdb etc. do not work and then you can move on debugging from there ... alternatives would be to get or rather build nowadays a suzyqable yourself which can give you a serial console on chromebooks (see https://github.com/hexdump0815/linux-mainline-on-arm-chromebooks) ... in some situations also something like Sonicadvance1/linux#27 (comment) might be helpful ... |
|
@hexdump
The first cherry image you provided booted.Will try the first approach on dumping the dmesg to a file and I was halfway through a homebrew suzyqable.Does the double serial work as a suzyqable?
|
|
@bruceloco - i just remembered, that on tomato maybe reading around this one here might be useful: https://oftc.irclog.whitequark.org/aarch64-laptops/2023-10-06#32545749 - this mentions a patch required for the display to work: https://lore.kernel.org/all/20230816190427.2137768-1-nfraprado@collabora.com/#r ... not sure if this is your problem, but worth looking into maybe ... in general that user on irc seems to have tomato running quite well with mainline or the collabora chromeos-close-to-mainline tree from https://gitlab.collabora.com/google/chromeos-kernel/-/tree/for-kernelci ... maybe some of this information is useful for you to move forward ... best wishes and good luck |
|
@hexdump0815 thanks, I will take a look at it |
|
i think colabora is doing the mainlining of the mediatek chromebook kernels for google, so they use a kernel which is always updated to the latest mainline dev kernel version and has some of the chromeos special patches in it on top - they use this kernel to upstream the mt81xx support to mainline ... it might be a good option to try if pure mainline is still not reliable enough on cherry tomatoes ... |
|
i just noticed that o forgot to answer the question about the dual usb-serial vs suzyqable from above: suzyqable is special for chromebooks and gives a full serial console while the dual us-bserial will only give serial output if usb is at least working |
|
Thanks for that tip, that guy is one of the contributors to this git: https://git.kernel.org/pub/scm/linux/kernel/git/chunkuang.hu/linux.git/commit/?h=mediatek-drm-fixes |
|
@bruceloco - that are very good news - please keep me/us up to date here in case of any further progress |
|
@hexdump0815 When I sort this out I will provide you with a cherry tomato config that is working on the DRM fixes branch of the 6.7 RC1 from the above git repo. |
|
you mentioned that you saw the penguins - right? that means the console initially came on i guess - at some point it is taken over by the drm system - maybe this is where it breaks? does it still crash, i.e. no dmesg if you write it to the filesystem, sync, shutdown via rc.local? |
|
@hexdump0815 that was building with your defconfig |
|
ignore that, i was being an idiot, there are some dependencies that I needed to follow for options enabling |
|
@hexdump0815 got a cherry config |
|
@hexdump0815 I fixed the wifi bad assert, Chun-Kuang Hu did not pull the code from nl80211.c where they lock the wiphy mutex. |
|
@hexdump0815 adding this fix otherwise wifi produce error messages A fix for audio as its loading the wrong firwware file Or if you don't want to fiddle the code just hardlink or symlink Attaching the current .config |
|
@bruceloco - thanks a lot for this information - i hope to have time to look a bit closer at all this next week as maybe it helps me getting corsola (mt8186) forward a bit with mainline as well |
|
@hexdump0815 it's just contributing, i would have not gone far without you so I have to pay it forward :) The advantage is that I do my builds on my chromebook in a chroot and ssh into the terminal so i can have an actual terminal and a browser for searches. so its easy to cat the output of modules and things that are loaded, compare the dmesg output, etc |
|
i just want to say that i'm very happy with such kinds of contribution as it helps a lot to bring the idea to run mainline linux on chromebooks forward quite a bit if more and more people start researching how to get and keep it working on all the different hardware - so all your effort is very welcome |
|
@hexdump0815 https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/tree/ |
|
and for extra fun - we have a non changeable mode 2256x1504 (3:2) I no longer feel blind |
this issue is just a placeholder to potentially give feedback if my untested experimental images from https://github.com/hexdump0815/imagebuilder/releases/tag/220710-01 are working at all and in case they are working to note down what works and what not ...
The text was updated successfully, but these errors were encountered: