ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630)
alpernebbi has joined #aarch64-laptops
user has joined #aarch64-laptops
user is now known as Guest6301
Guest6301 has quit []
obeliskblack has joined #aarch64-laptops
<obeliskblack>
Any good tutorials on setting up stable linux distro for Snapdragon 835 like Miix 630?
<obeliskblack>
Flashed prebuilt images to usb but got no keyboard/mouse or touch input on boot.
<steev>
can you use an external keyboard/mouse ?
<steev>
looks like jhugo added the first commit, not sure what the status is of it now
<obeliskblack>
steev will need to buy a usb-c hub/dock to test it out
<obeliskblack>
are images like Live CD with installer gui or supposed to flashed dd if=img of=to the internal UFS directly? https://github.com/aarch64-laptops/build
<obeliskblack>
flashed usb seems to corrupt itself after booting from it once
<steev>
they should be flashed to a usb, not the ufs
<robclark>
so, I remember the c630 had a fw bug where it would helpfully try to "recover" a backup GPT (which could exist in the part of usb drive you didn't overwrite while flashing).. it wouldn't surprise me that the 835 devices had the same fw bug
<robclark>
I don't remember the magic cmds to delete the backup GPT off the top of my head.. I suppose dd'ing /dev/zero to the usb drive is a brute force way, but IIRC there is some fdisk/cfdisk/etc way to do it
<robclark>
anyways, the symptom of that is usb drive being corrupted after booting from it once
FizzBuzz has joined #aarch64-laptops
hexdump01 has joined #aarch64-laptops
hexdump01 has quit []
hexdump0815 has joined #aarch64-laptops
<hexdump0815>
robclark: that magic command was sgdisk -Z /dev/xyz - much faster than a full dd of /dev/zero
<hexdump0815>
obeliskblack: that one worked very well for me on my novago as well - maybe you can give that one a try ... i also tried to build a 5.11 kernel based on the openplus 5 tree (which i think should be the most advanced msm8998/835 tree at the moment)
<hexdump0815>
robclark: while you are around a question i wanted to aks for quite some time already and maybe you know the answer: why none of the arm systems support loading of color calibration profiles or setting gamma in xorg for instance via xcalib?
<hexdump0815>
it just returns without an error, but does not change anything on screen - at which level this is being done actually - xorg, drm, gpu? on intel systems it just works, but i did not see it working on any arm system yet
<robclark>
hmm, I can't say I remember what modesetting/glamor does with that.. newer snapdragons (stuff that uses dpu) does support kms CTM.. or at least sc7180 does (we use it) and I think it should work for 845
<hexdump0815>
something like this has to be in place at least in android and chromeos, as they allow to change the color temperature of the screen depending on time
<robclark>
right
<hexdump0815>
so its implemented separately for each soc/gpu then?
<robclark>
assuming xorg is just passing it thru to the kernel, then it needs to be supported by the display
<hexdump0815>
i think i read that it requires/uses the xvideo xorg extension, but do not know anything more about it
<hexdump0815>
would be nice to be able to warm up the color temp of the screen on more systems - more healthy for the eyes i think
<robclark>
it's possible that xorg uses some older driver specific mechanism.. tbh it has been ages since I looked at xorg
<robclark>
a compositing window manager should be able to do it in gl
<robclark>
which is more or less fine, since xorg isn't really using overlay compositing like CrOS/android
<hexdump0815>
ok - that might explain why it works on android/cros
<robclark>
for android, it probably uses whatever crazy mechanism the vendor hwc uses.. for CrOS it uses the standard kms CTM properties.. we (or rather qcom.. we have most of the display register docs but the color processing is something they omitted) implemented that for sc7180 exactly for that use-case.. on the dpu side there are multiple generations of the color processing block, so maybe what we did doesn't support
<robclark>
*everything*
<robclark>
for $random_thing, I'd recommend just doing it in GL.. once you are compositing on the GPU applying a color transform matrix is basically free
<obeliskblack>
hexdump0815 will try novago image, do you just need to replace the bootrom EFI (device specific) using the same flashtool procedure as aarch64-laptops/build?
<hexdump0815>
obeliskblack: just flash that image completely to a usb drive and it should be bootable - i think you'll have to adjust the dtb used in the grub.cfg to your model
<hexdump0815>
obeliskblack: if you had that old image booting then this one should just boot as well