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)
flowriser has quit [Ping timeout: 480 seconds]
<qzed> bamse: is it possible that this uses some QSPI flash chip to store those things?
<qzed> although that might just be UEFI flash...
<qzed> just looks like there might be two SPI flash chips, one marked as UEFI SPI ROM and another as QCOM store...
flowriser has joined #aarch64-laptops
<steev> i'd bet the qcom store one is it
<qzed> seems likely
<qzed> unfortunately I don't have any idea how to describe the quad-spi it's connected to in the DT
<qzed> and while I've been able to dump some stuff from various partitions that show up in the efi shell, the things we're looking for aren't there
<qzed> but there are a bunch of efi drivers and some interesting config files
<qzed> also the boot logo... now I'm wondering whether I can just replace that with a tux xD
<qzed> also a bunch of files with GUIDs as names...
<bamse> qzed: regarding your github update on the gpu firmware, you can use firmware-name in the zap-shader node to point it to the windows .mbn file directly, you don't need to split it
<bamse> qzed: but perhaps you knew that already
<bamse> qzed: it's definitely a possibility that bootloader and some other things live in a qspi flash
<qzed> or something else?
<bamse> exactly like that
<bamse> wasn't sure if you did something else when i read the github update
<qzed> yeah, i really need to write something more detailed on how to get everything working at some point...
<qzed> while the whole uefi stuff is quite interesting, it doesn't look like there's anything related to those partitions in there
hexdump01 has joined #aarch64-laptops
<qzed> interestingly, there seem to be some lzma compressed images for various stuff in there... also references to a ram partition, so things may get extracted from those to ram, I guess
<qzed> anyway... I've tried creating empty files and that gets me a step further
<qzed> it's now complaining about not finding board data
<qzed> meaning the remote proc isn't crashing immediately
hexdump0815 has quit [Ping timeout: 480 seconds]
<qzed> but that's something I'll try to figure out tomorrow
<bamse> i probably created my boardfile using the ath10k swiss army knife tool and the bdwlan files from windows...
<bamse> have a good night
<qzed> yeah, I remember playing around with that a long time ago...
<qzed> thanks! you as well
tomf_ has joined #aarch64-laptops
jenneron has joined #aarch64-laptops
<jenneron> macc24: i added support for minnie and speedy. i need someone with jerry with depthcharge on SPI (someone who didn't switch to u-boot). i can also add support for other veyron devices, it's quite trivial, in this case i also need someone to test it
<steev> bamse: i know it's not yet rc1 but using the defconfig on torvalds tree, i'm getting bam-dma-engine 17184000.dma-controller: Runtime PM usage count underflow!
<steev> and by getting, i mean it's spamming the shit outta my log
jhovold has joined #aarch64-laptops
jhovold has quit [Ping timeout: 480 seconds]
<macc24> javierm: ok
<macc24> oops, * jenneron
jenneron_ has joined #aarch64-laptops
jenneron has quit [Ping timeout: 480 seconds]
<qzed> bamse: so the problem with the board files is that the chip is reporting a board_id of 0xff
<qzed> specifically qmi chip_id 0x30224 chip_family 0x4001 board_id 0xff soc_id 0x40060000
<qzed> and qmi fw_version 0x32080009 fw_build_timestamp 2020-11-16 14:44 fw_build_id QC_IMAGE_VERSION_STRING=WLAN.HL.3.2.0.c8-00009-QCAHLSWSC8180XMTPLZ-1
<qzed> so there seems to be an option for matching board files over the chip_id
<qzed> I've been poking a little bit at the windows driver to see if I can figure out which file it's loading
<qzed> but before I go further down that rabbit hole... how does that look like on the flex?
<qzed> or is there any other way to figure out which board file I need for that chip id?
<steev> qzed: it's ff on the flex5g too -
<qzed> there are also a ton of entries with ff in https://github.com/kvalo/ath10k-firmware/tree/master/WCN3990/hw1.0
<qzed> so I guess there has to be some way to figure out what the right file is... or is it just trial and error?
<steev> yeah
<steev> i'm not sure
<steev> that's what the flex5g reports for wifi
<qzed> thanks! seems to be the exact same thing
<qzed> you don't happen to know which of those bNN files it's using? xD
<steev> one momento
<steev> https://paste.debian.net/1243041/ this is what we use on the flex5g to grab the firmware from the windows partition
<steev> should just be the one wlanmdsp.mbn ?
<qzed> hmm, I thought that was used for something else?
<qzed> or is that somehow integrated into the board-2.bin?
<steev> i think the bd files are board-2.bin
<qzed> yeah, that's what I'm trying to do, but the problem is that that script goes by board-id, which is 0xff
<qzed> so I have a bunch of bdwlan.bNN files but no bff file, so it seems it's matching via chip id, which that script doesn't support
<steev> i don't show any bff file here either (on the c630)
<steev> let me plug the flex in
<steev> i accidentally left it on overnight and the battery is dead
<qzed> I guess ff is just some placeholder that indicates that matching should be done by chip id instead of board id
<steev> that's over my knowledge level
<qzed> at least looking at the latest board-2.bin from https://github.com/kvalo/ath10k-firmware/tree/master/WCN3990/hw1.0 has a ton of entries with ff and some chip-id
<steev> actually... i don't think the script worked for me on the flex5g to extract the firmware, because i'd bumped to windows 11 and it couldn't mount the drive properly anymore
<steev> so i copied them all manually onto a usb drive, or used wsl
<steev> on my flex, in the extracted firmware folder that firmware script creates, i just show a bdwlanu.b5f file
<qzed> i copied everything manually from the windows partition a while ago, and I now also have an official "recovery image" which should also have all that, if you know where to look for it
<qzed> ah, I'll go have a try with that then
<steev> it'd be in the filerepository/driverstore - qcwlanXXXXXX
<qzed> yeah
<qzed> have it already
<steev> 8180 iirc
<steev> ah, right
<steev> it's a symlink on my flex5g to "unsupported reparse point"
<steev> i haven't tried recently to run it and see if there were upgrades to ntfs3g
<qzed> phoronix mentioned some fixes for 5.19... but no clue what they're for
<steev> the ath10k didn't get much
<steev> i *think* that thread i linked earlier will be in 5.19, but i don't recall
<qzed> so... that didn't work
<qzed> but I'm not sure if it's the board file
<qzed> remoteproc crashes
<qzed> fatal error received: err_qdi.c:1024:EF:wlan_process:0x1:WLAN RT:0x107b:cmnos_thread.c:3998:Asserted in halphy_reset.c:halphy_reset_UpdateResetParms:776
<qzed> this line seems a bit weird:
<qzed> wcn3990 hw1.0 target 0x00000008 chip_id 0x00000000 sub 0000:0000
<steev> i don't think i've ever seen that line, so i won't be of much help
<qzed> no worries, thanks for your help anyways!
<steev> i'm also more of a cosplay kernel dev than an actual kernel dev
<qzed> to be honest... working with all that qualcomm stuff I feel pretty out of my depth too
<steev> there are definitely a lot of moving parts
<qzed> hmm, so another board file kind of works... and with that I mean it doesn't crash
<qzed> but spams the log with `chan info: invalid frequency 0 (idx 41 out of bounds)`
<qzed> so I guess I'll go with trial and error
jenneron_ has quit [Remote host closed the connection]
jenneron_ has joined #aarch64-laptops
<steev> i see that on the flex5g as well
<qzed> oh, neat
<steev> not really spammed, i think it only does it 3-4 times, and then occasionally a few times
<qzed> yeah, initially it was almost every second, but it's quieted down a bit now
<qzed> also I can see some networks
<steev> i think i might have grabbed the board-2.bin from chromeos, because mine is ~715K in size
<qzed> just not my home network... but connection isn't great in my room
<qzed> ah, guess I'll need to have a look at that
<qzed> hmm, so something doesn't seem to work quite right... if I switch my network to channel 5 it gets recognized
<qzed> but not on channel 12
<qzed> okay so for the record, bdwlanu.b58 worked for me
<steev> wifi may not be entirely stable either
<steev> at one point, after a while, the device would end up rebooting, or the wireless device would just stop working
<qzed> huh, interesting... can't be easy now, can it xD
<qzed> I guess I'll run into that one as well at some point
<steev> i still experience the device just not working, and the patch thread i linked above allows you to unbind and then rebind it to get it working again, at least that happens on the flex5g, i can unbind and rebind on the c630 to get the device back but i end up having to reboot to get working wifi if i unbind/rebind on there
<steev> just stop working*
<qzed> neat, I'll apply that to my branch then
<steev> speak of the devil... it just stopped working here
<qzed> seems like the pro x is pretty similar in that regard, at least going by the hardware IDs
<qzed> thanks again for all your help!
<steev> no problem, glad it's working for you - i think i saw something in net-next regarding channels?
<qzed> neat, I'll look through that later
<steev> https://kernel.googlesource.com/pub/scm/linux/kernel/git/kvalo/ath/+log/refs/heads/master may have some things as well, kalle is the atheros maintainer
<qzed> ah nice, thanks!
<steev> and krzk wrote the patch i linked above, so they may have some atheros insights as well?