15:43 <bamse> steev: unfortunately i haven't paid much attention to the audio situation lately...

15:43 <steev> well.. it works on one... and not the other and i'm just so fucking lost

15:43 <steev> userland is identical

15:45 <steev> bamse: fwiw... since it does work on one - with the 5.12 (maybe it got into 5.13 and was backported -either way, with that patch and the patch from shawn to fix the gpio in the dts, it should work now with the normal userland stuff already installed)

15:46 <bamse> i need to respin the backlight patches, will see if i have audio while doing that...

15:51 <steev> pls :)

15:51 <steev> well, the respin regardless :D

16:00 <steev> am about to test 5.13.3 but i doubt anything has changed there

16:17 <steev> bamse: it's also weird because on the c630 that doesn't have working audio, the ipa device doesn't seem to probe automatically... but on the one that does work, it does. i tried restoring windows and then re-installing just on the off chance the hardware got into a weird state, but with the 5.11 kernel from the installer, audio works perfectly

16:17 <steev> and srini still doesn't seem to have moved to oftc :(

16:19 <bamse> steev: if ipa doesn't probe automatically then your dt isn't updated...

16:20 <steev> hm

16:20 <steev> maybe it's an issue with dtbloader?

16:21 <bamse> steev: it's just a platform driver that need to be enabled in the dts for it to probe...it just used to not work after that

16:21 <bamse> steev: but with that patch i think it would probe a few versions back

16:23 <steev> https://usercontent.irccloud-cdn.com/file/I9AGFQ50/Screenshot%20from%202021-07-19%2011-22-54.png

16:23 <steev> i'm assuming the grep matching means its in the dtb that is loaded?

16:25 <steev> the one where everything works does *not* use dtbloader, grub loads the dtb file

16:26 <bamse> what's the content in soc@ipa../status ?

16:26 <steev> disabled

16:27 <bamse> and on your other machine?

16:27 <steev> good question, lets check

16:32 <steev> okay

16:34 <steev> though, 5.12 it says okay, but i don't get the device, only on 5.13+ do i get the device

16:36 <bamse> perhaps there's some dependency that i'm forgetting about...

16:36 <steev> should just be rmnet

16:39 <steev> in fact, when ipa doesn't work, it at least complains that it can't find the firmware

16:39 <bamse> rmnet depends on ipa, not the other way around

16:40 <steev> oh, interesting

16:44 <bamse> so iirc having just ipa working should give you rmnet_ipa0

16:44 <bamse> and with that you can link rmnet_data* interfaces ontop, which requires the rmnet driver

16:47 <steev> okay, we're on to something!

16:47 <steev> forcing loading the dtb from grub via "devicetree" entry, does give me the ipa module loading

16:48 <bamse> and audio?

16:49 <steev> not sure just yet - https://paste.ubuntu.com/p/32M2qqGdxj/

16:49 <steev> well, there's not the audio errors but there's also no audio output

16:49 <steev> does just give dummy output :/

16:51 <steev> hm

16:52 <steev> i do have to actually modprobe -r ipa and then modprobe ipa after booting for it to load the firmware; maybe i should shove it into the initrd

16:53 <steev> and i still dunno which firmware i'm supposed to be using here - on one, where it's "ipa_fws.mdt.orig" that is a file size of 6.7K, on the other it's ipa_fws.mdt.elf which is 33K

17:04 <steev> oh, interesting, i see

17:05 <steev> the 6.7k one is one that pulls in all the b0X ones, and the 33K is one that has all of them combined

17:05 <steev> because i copied only the 6.7K mdt into the initrd, and it complained it couldn't load ipa_fws.b02

17:13 <steev> yeah, okay so use the 33K size one, and throw it into the initrd, and then it'll actually come up, otherwise the module probes before the firmware is available and doesnt seem to retry at all

17:13 <steev> oh, and audio worked this boot

17:14 <steev> shawnguo: okay so... how the heck do i go about debugging why loading the dtb via dtbloader isn't working properly, but when forcing grub to via 'devicetree /usr/lib/linux-image-5.13.3/sdm850-lenovo-yoga-c630.dtb' works

17:26 <robclark> steev: did the \dtb\$sha.dtb not get updated for new kernel?

17:29 <steev> no idea

17:29 <steev> i would assume it did, but who knows

17:32 <steev> i have 2 kernels, but 3 dtbs in there?

17:35 <steev> okay, 2 of them match the 5.11 kernel, and the last one... no idea wtf it matches because both of them say they differ

17:36 <steev> robclark: i don't suppose you have some non-edk cli tool to get the chid

17:37 <robclark> hmm, I think fwupd has something?

17:37 <robclark> 3 dtb's means three separate devices supported, not three separate kernels, fwiw

17:38 <robclark> I have 12c0e5b0-8886-5444-b42b-93692fa736df.dtb and 30b031c0-9de7-5d31-a61c-dee772871b7d.dtb .. one is c630 and one should be flex5g

17:38 <steev> ah yeah mine matches the 30b...

17:39 <steev> `fwupdtool hwids` is the command to spit them all out

17:39 <steev> okay so, for some reason, dpkg -i linux-image-5.13.3blah.deb isn't updating the dtb in there

17:40 <steev> lets test manually copying

17:40 <robclark> steev: https://github.com/aarch64-laptops/edk2/blob/dtbloader-app/EmbeddedPkg/Application/ConfigTableLoader/install-dtbs.py#L44

17:41 <robclark> I guess you can hack in a print to that script to show the actual chid

17:41 <steev> yeah, that did it

17:41 <steev> okay, so... is install-dtbs.py supposed to be run as a post kernel install hook?

17:43 <steev> hm, audio doesn't error out, but definitely doesn't come up every time in 5.13.3

17:44 <robclark> I guess it could run as post install hook.. I'd kinda envisioned it as being run to generate a "dtbloader fw update" (but then never had time to write a fwupd pluging to install those fw updates)

17:45 <steev> well, now that i know i need to update the dtb manually... :D

17:45 <robclark> you kinda want to only move forward, ie. if downgrading the kernel keep newer dtb..

17:47 <steev> yeah, i'm fine with that :D

17:47 <steev> i just wanted to get off the 5.11 kernel but thought the dtb got updated automatically

18:51 <derzahl> man, so seeing i put everything back from /dev/sdf the way it was and windows wifi now works. i cant figure out wth is causing grub not to automatically boot. so when grub-mkconfig runs, it thinks the /dev/sda is hd0, but when grub actually boots it thinks its hd5 and drops me to a shell. how i force grub to use hd5? its gotta be more than just changing referencers in grub.cfg, right?

18:55 <steev> i have no idea, i've never seen that

18:55 <steev> robclark: did we ever come to a conclusion about those enc frame timeouts?

18:56 <steev> ah, from the looks of it... i never pasted when i ran into it again

18:57 <robclark> I don't think so, but also haven't had a chance to try and repro or debug it.. haven't noticed anything like that on sc7180 but not sure if that is down to some sdm845 vs sc7180 thing or some difference between our userspaces..

18:58 <steev> i do happen to be using x11 (again), sdaly

19:00 <steev> oh happy day, i do still have the dumpstack hack in

19:00 <steev> robclark: https://paste.ubuntu.com/p/rYKjfhhrDw

19:01 <robclark> hmm, it's called from a timer.. ok, so that is a bit less useful

19:02 <steev> it does spam the ever loving shit out of... everything, when it happens

19:25 <derzahl> does the grub version built into debian-cdimage not support specifying devicetree in the boot stanza? i seem to recall something like that

19:31 <derzahl> so many crazy things happening. after doing a fresh install of https://github.com/aarch64-laptops/debian-cdimage rel4(latest), it setup that dtbloader thing that i havent used in forever. but then after recompiling steevs 5.11rc7, neither kernel boots(hangs like wrong or no device tree is loaded). tried removing dtbloader and just speicfying 'devicetree correct.dtb' from the kerne package and that gives the same results

19:32 <steev> you can specify devicetree in grub with it, that's grub, not specific to the cdimage

19:32 <steev> it has been in grub for a long time, what is missing is the ability to automatically add it to the config; ubuntu patches that, but upstream hasn't moved on it and/or ubuntu hasn't passed the patch up, i don't remember which

19:32 <derzahl> well im lost then. why do things keep changing on me? lol

19:33 <steev> well

19:33 <steev> as i said, or maybe you already took off - 5.11.0-custom is *not* the same as my 5.11-rc7

19:33 <steev> 5.11.0-custom is based on aarch64-laptops/linux's 5.11

19:33 <derzahl> nah,i saw that that

19:33 <derzahl> makes sense

19:34 <steev> while the dtb *should* work, no idea

19:34 <derzahl> thatswhy i compiled your 5.11rc7, so id have a working kernel with headers

19:34 <steev> 5.12 works pgood

19:34 <derzahl> and i know dtbs change per kernel, so i always try to use the one that comeswith the kernel

19:35 <derzahl> well. 5.12 works well except....doesnt let me type my password to unlock encrypted root

19:35 <derzahl> while 5.11.rc7 does

19:35 <derzahl> go figure

20:01 <steev> oh right, i forgot you're doing that weird zfs shit

20:39 <derzahl> well trying

20:40 <derzahl> always soo close

20:53 <derzahl> gonna try a fresh install. does anyone else have that /boot/Persisted_capsules.bin file? not sure if it came from winsows or linux

20:53 <steev> its from windows

20:54 <steev> afaiui, it's so that things will update next time you reboot while on mains, since it won't upgrade the bios or ec firmware if mains aren't applied

20:54 <steev> which is kinda... moot since there hasn't been an update to either since 2019

20:55 <derzahl> ah

20:55 <derzahl> thanksfor the background

20:56 <derzahl> its like 1/3rd of the efi partition

20:56 <steev> that's just my understanding. doesn't mean i'm correct

20:58 <derzahl> well at least I shouldbeableto safely back it up somewhere

20:58 <steev> i guess, really, it's for capsule updates, but that's why they wouldd exist?

21:02 <derzahl> not sure what a 'capsule' even is myself

21:03 <derzahl> so if i use dtbloader, do you know if it have some mechanism to update dtbs when with new kernels? it seems kinda static, and i know Ive has issues in the past using old dtbs with newer kernels or vice versa

21:11 <steev> https://uefi.org/sites/default/files/resources/UEFI%20Fall%202018%20Intel%20UEFI%20Capsules.pdf

21:11 <steev> nope, you have to copy them in yourself, i just learned that myself earlier :D

21:12 <steev> that's kinda the issue tho... because of a gpio change, 5.12+ dtb won't work with 5.11

21:12 <steev> well, they will but audio won't work

21:12 <steev> typically tho, newer dtb with older kernel is fine, but not vice versa

21:29 <derzahl> can you just specify a static dtb at grub install though? or is that broken

21:29 <derzahl> the option is there

21:30 <derzahl> grub-install --dtb=FILE embed a specific DTB

21:31 <derzahl> i had added on to grub.d/10_linux to check for a variable in /etc/default/grub and add the devicetree line. forgot to back it up tho:(

21:33 <derzahl> one more question and then im done bothering you for now. is there one already, or would it be relatively easy for me to make from your 5.11rc7 sources, a later version of the 5.11 kernel?

21:34 <steev> disclaimer: i suck at git, but you might be able to pull it down, add stable as a remote, then `git merge --rebase stable/linux-5.11.y` - i'm JUST guessing though

21:34 <steev> and then fix up anything that's broken along the way

21:35 <steev> sometimes it's just a matter of applying the patch manually and adding the files and running git am --continue or maybe it would be git merge --continue, not sure

21:35 <derzahl> oh, you know what. i just remembered that zfs 2.1.0 has any easy 'make deb-pkg' or similar option

21:36 <derzahl> so i can use 5.12.y:)

21:36 <derzahl> the zfs-dkms version in sid still doesnt support 5.12

21:36 <derzahl> but ill try that if i need to

21:36 <steev> probably

21:37 <steev> doesn't zfs have a way to make a patch that you apply to the kernel?

21:37 <derzahl> ok, one more thing i PROMISE:) wheres they latest most stripped down config for 5.12.y.

21:37 <steev> uhhh

21:37 <derzahl> i keep running into problems using my arch build config that works fine on arch, on debian

21:38 <steev> dunno how stripped down it is, but this is the one i use https://paste.ubuntu.com/p/pbXMbZGxvJ/

21:38 <derzahl> i dont think they maintain one anymore

21:38 <derzahl> since linus got upset

21:39 <derzahl> awesome. ty

21:39 <steev> i used to talk to ryao about it a bit back when i was still a gentoo dev, trying to get him to fix arm 32bit but gave up at how long it was taking

21:39 <steev> i dunno why but i absolutely positively have to use that config on the ubuntu (origins) c630, but as of 5.13, i just use shawn's distro_defconfig

21:47 <derzahl> im pleasently surprised how well zfs runs on arm64

21:47 <derzahl> not sure if itd be worth it on arm32

21:48 <derzahl> especially if theres not enough ram

21:48 <derzahl> but it runs great on my pi4

21:49 <derzahl> which is why im so deadset in getting it as my root on the c630

21:49 <derzahl> the debian-cdimage config?

21:49 <derzahl> its pretty big

21:49 <derzahl> lot of stuff can be removed

21:51 <derzahl> even the one on arch i has stripped down had tons of bloat still thati never got around to removing. since id always go too far and then it wouldnt boot for some reason

21:57 <derzahl> whewww 5.11.17 slready

22:05 <derzahl> any exciting news on more powerful arm laptops coming out?

22:07 <derzahl> oh, looks like alot of progress with the m1

23:02 <derzahl> damn. nobooty on 5.12.17

23:04 <derzahl> hangs for a bit and then black screens after xhci-hcd

23:05 <obeliskblack> Prebuilt images seem to brick the usb key after first boot? are these live cd isos or supposed to be flashed on eMMC built on the device

23:06 <obeliskblack> sd835 miix 630 accelerated graphics still a no go on those?

23:06 <derzahl> this is a newly compiled kernel on a working c630 ssd

23:32 <derzahl> well 5.12.14 that i previously compiled but didnt work, works now

23:32 <derzahl> yay

23:50 <derzahl> which is funny, cuz i was just telling steev that my prior custom config wasnt working on debian

23:51 <derzahl> oh well egrep '=m|=y' config-5.12.14-custom | wc -l

23:51 <derzahl> 3086

23:51 <derzahl> about 1000 less enabled options