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)
<steev>
you could always restore (again) and start over :P
<steev>
are you trying to point at the dtb on your encrypted root that isn't decrypted yet?
<derzahl>
yeh, thats what im tryingto avoid:)
<derzahl>
no, dtb is just on the normal efi partition
<derzahl>
and im just using debian, not even the distro that shall not be names
<derzahl>
*named
<derzahl>
but like i was saying, i dont *think* its a dtb problem. even when i boot dtbloader, which uses the old dtb that 5.12 cant boot with. i can still boot 5.12 if i specify the proper dtb with devicetree
<steev>
why not copy in the correct dtb?
<derzahl>
for dtbloader?
<derzahl>
i could try that
<derzahl>
not sure how it works
<derzahl>
theres a /boot/dtb dir
<derzahl>
that contains 3 dtbs with uuids for file names
<derzahl>
oh, i seem to recall the system booting into linux after the install
<derzahl>
maybe i remembered wrong
<shawnguo>
if you ever installed Linux before, that could be the case, since bootorder editing is needed for the first time when you install Linux
<derzahl>
yeh, im talking about since the first time i installed it
<derzahl>
but i was just grasping at straws. pretty sure now i remembered it wrong
iivanov has joined #aarch64-laptops
<derzahl>
whelp. finally got the c630 booting from encrypted zfs root!
<derzahl>
and i was just about ready to give up
<derzahl>
and hot damn, its got sound too!
<derzahl>
weird. its also showing two wireless adapters
FizzBuzz has quit []
derzahl has quit [Ping timeout: 480 seconds]
pg12 has joined #aarch64-laptops
pg12_ has quit [Ping timeout: 480 seconds]
FizzBuzz has joined #aarch64-laptops
<steev>
shawnguo: how do you feel about making CONFIG_HZ_1000 the default instead of CONFIG_HZ_250?
<robclark>
fwiw, CONFIG_HZ_1000 is what fedora and CrOS use.. afaiu you really only want to use lower for server use cases where interactivity doesn't matter as much
<steev>
robclark: yeah, i was kinda surprised that arm64 doesn't set it by default... but then again... arm64 servers are quite common these days
<steev>
but i figure since shawn has the distro_defconfig (and i've added it there in my fork), it's a decent place to put it
FizzBuzz has quit []
<bamse>
look at that...if i just turn on the right bus the adreno driver doesn't crash on sc8180x
<robclark>
well, fortunately I refactored things recently when adding a660/a635.. which makes it a nice small patch to add new devices of an already known sub-generation
<bamse>
anything particular i need to do to build mesa btw?
<bamse>
flags etc
<robclark>
defaults should be ok if you are compiling natively but if you give me a sec I'll pastebin what I use (which is a bit more trimmed down than defaults)
<bamse>
presumably i can save some time by not building !freedreno?
<robclark>
and then `ninja -C release && sudo ninja -C release install` should work
<robclark>
(I normally don't install to /usr but I guess that is probably what you want to do at this point)
<robclark>
oh, you might want to uncomment glvnd if arch uses it
derzahl has joined #aarch64-laptops
<bamse>
robclark: patching up the mesa-git arch package, so that i don't trash my installation...and it seems to be building now
<robclark>
cool
<HdkR>
I've still been fighting with my install. Haven't had GPU acceleration for a while now
<HdkR>
Trying to figure out why DRI2 can't authenticate is always a nightmare
<robclark>
is that even still needed with dri3? I thought dri3 was using dma-buf fd's instead of flink names?
<HdkR>
This is a level of the stack I'm don't know enough about to know
derzahl has quit [Ping timeout: 480 seconds]
<HdkR>
Updating the OS, maybe I got a bad set of packages
FizzBuzz has joined #aarch64-laptops
<HdkR>
yay. Updated the OS and it fixed the issue
<steev>
that makes one of us :(
<steev>
jk, i don't have 3d issues
<bamse>
robclark: glxinfo says "freedreno and FD680...
<robclark>
ship it!
<robclark>
does it actually work?
<bamse>
now that i finally rebuilt the package again with proper prefix :)
<bamse>
glxgears are spinning!
<robclark>
\o/
<bamse>
webgl aquarium gives me 30-35 fps
<robclark>
that is kinda low, but there could be other reasons for that..
<bamse>
but compiling mesa only got me to ~40C on the cpus...so i'm missing something on my side...
<bamse>
like bus bandwidth
<robclark>
heh, yeah, that could certainly do it
<bamse>
running aquarium for a few minutes now...cpus are 33-38C
<bamse>
anyway, i will polish up the driver patch that i came up with and send that out
<bamse>
what's needed to get the mesa change landed?
<robclark>
I guess if aquarium works, then it is at least not completely broken..
<robclark>
I guess wouldn't hurt to do a bit more sanity checking.. but I think a680 should be closely related to a640 which is pretty well tested
<bamse>
yeah, but i guess landing it will increase the number of people testing it dramatically?
<bamse>
either way, i'll continue to throw my kernel pieces out there...so that you and others can reproduce this
<robclark>
well, I guess there aren't a lot of people yet who have setup to test it without kernel patches ;-)
<bamse>
yeah, and i presume anyone that has my kernel patches would prefer to have mesa+your patch, even though the stability etc is somewhat unknown
<bamse>
but how does mesa releases work? when is the next one?
<robclark>
It doesn't have to be *perfect*.. a660/a635 are passing all of deqp-gles2/gles3, but only 95% or so of deqp-gles31.. but if things like desktop + chrome or firefox, etc are working, that is good enough
<robclark>
we are already past the branchpoint for 21.2.. so I guess this won't be in distros until 21.3 in a few months
<bamse>
okay, but i'll continue testing this as we land the other kernel patches and the displayport patches and we can do some more testing in the coming weeks
<bamse>
and then merge it, well in time for 21.3
<robclark>
*maybe* I can talk folks into a backport since it is a small patch, but the release managers probably won't go for it.. at least the series that the patch depends on got in before the 21.2 brnchpoint
<robclark>
hopefully we can get kernel side in for this upcoming merge window
<bamse>
i'm hoping we can reach a point where shawn can update the debian image with something useful before "a few months"
<bamse>
but i presume we have some means for dealing with such things
<robclark>
if debian package builds from git (or at least regenerates the .py generated tables) it should be easy enough to add that patch to deb pkg once 21.2 is released
<robclark>
(I think it is on ~rc2 atm)
<robclark>
(tbh, I don't follow the release branches too closely, we've been using ToT in CrOS)
<bamse>
anyway, i have more dpu and dp patches that i need to convince you to pick, before this is a real problem
<robclark>
I think we are pretty interested in getting the eDP bits landed ;-)
<bamse>
i know
<bamse>
cleaning out my tree right now, that's why i noticed that i forgot to turn the busses on
iivanov has quit [Read error: Connection reset by peer]
iivanov has joined #aarch64-laptops
iivanov has quit [Remote host closed the connection]
derzahl has joined #aarch64-laptops
derzahl has quit [Remote host closed the connection]
derzahl has joined #aarch64-laptops
derzahl has quit [Remote host closed the connection]
derzahl has joined #aarch64-laptops
derzahl has quit []
derzahl has joined #aarch64-laptops
derzahl has quit [Remote host closed the connection]