| - vagrantc (QUIT: Ping timeout: 255 seconds) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 00:13 | |
| - ndufresne (QUIT: Quit: Ping timeout (120 seconds)) (~ndufresne@apple.collaboradmins.com) | 01:03 | |
| - mtm (QUIT: Ping timeout: 255 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 01:03 | |
| + ndufresne (~ndufresne@apple.collaboradmins.com) | 01:03 | |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@64.189.201.150) | 01:14 | |
| - mjw (QUIT: Ping timeout: 268 seconds) (~mjw@gnu.wildebeest.org) | 01:23 | |
| digitalrayne | josch, cool trick :) so i guess with that you are skipping over the FAT etc as the firmware is always at a fixed offset? I'll have to give that a try | 02:49 |
|---|---|---|
| digitalrayne | in *BSD that would be dd if=bin/firmware.bin of=/dev/rsd2c bs=512 seek=4 count=64 (replacing sd2 with the disk ID from dmesg), that would also obviate the need for conv=nocreat,notrunc which is nice (only conv=notrunc exists on BSD dd) | 02:51 |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 03:09 | |
| - skipwich (QUIT: Quit: DISCONNECT) (~skipwich@user/skipwich) | 04:01 | |
| + skipwich (~skipwich@user/skipwich) | 04:02 | |
| josch | digitalrayne: yes, according to lukas, the filesystem is created by a firmware located in rom, so it will always look the same -- let me know if it works! :) | 04:37 |
| josch | if your dd tried to re-create the file in the fat (which is not allowed) then that might just explain why it didn't work for you | 04:38 |
| josch | digitalrayne: you could try tracing the syscalls your dd does to find out with what arguments it tries to open the file on the fat | 04:55 |
| + jacobk (~quassel@129.110.242.173) | 05:40 | |
| + chomwitt (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 07:41 | |
| - chomwitt (QUIT: Ping timeout: 272 seconds) (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 08:30 | |
| + chomwitt (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 09:22 | |
| - klardotsh (QUIT: Ping timeout: 264 seconds) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:50 | |
| + mjw (~mjw@gnu.wildebeest.org) | 09:53 | |
| - mjw (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 10:04 | |
| * Guest3473 -> mjw | 10:08 | |
| - jacobk (QUIT: Ping timeout: 268 seconds) (~quassel@129.110.242.173) | 10:26 | |
| - chomwitt (QUIT: Ping timeout: 272 seconds) (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 10:28 | |
| minute | funky https://github.com/ToNi3141/Rasterix | 10:29 |
| josch | the thing can render Warcraft 3??? wow this kind of stuff is so impressive really | 11:12 |
| josch | digitalrayne: and i found another way to flash the lpc firmware without mounting anything: using mtools | 11:39 |
| josch | sudo mcopy -D o -i /dev/sda bin/firmware.bin ::firmware.bin | 11:39 |
| josch | in completely unrelated news, the "Cadence MHDP8501(HDMI/DP) for i.MX8MQ" patch is now at v15 and it looks like more fixes were requested: https://lore.kernel.org/linux-arm-kernel/PAXPR04MB94488A50C2BDD6B859B110DAF42A2@PAXPR04MB9448.eurprd04.prod.outlook.com/T/ | 11:43 |
| - Aard (QUIT: Remote host closed the connection) (~bwachter@edna-edison.lart.info) | 12:23 | |
| + Aard (~bwachter@edna-edison.lart.info) | 12:23 | |
| - murmr____ (QUIT: Ping timeout: 256 seconds) (sid433686@id-433686.helmsley.irccloud.com) | 12:41 | |
| + wakest_ (m-a7d6fe@45.77.48.108) | 12:43 | |
| - wakest (QUIT: Remote host closed the connection) (m-a7d6fe@45.77.48.108) | 12:43 | |
| + murmr____ (sid433686@id-433686.helmsley.irccloud.com) | 12:44 | |
| * wakest_ -> wakest | 12:48 | |
| minute | josch: it's the neverending patch! it's like a perpetual stew | 12:57 |
| AbortRetryFail | What is a DRM bridge driver for? | 12:58 |
| + mark_ (~mjw@gnu.wildebeest.org) | 13:00 | |
| - mtm (QUIT: Ping timeout: 272 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 13:04 | |
| josch | minute: is the qcacld2 out-of-tree driver something you think would make sense to ship in Debian? Or are there efforts to integrate it into mainly before long? | 13:10 |
| minute | josch: it could be shipped as nonfree perhaps, because it has to include bizarre firmware blobs | 13:10 |
| minute | josch: mainline ath10k doesn't work and the maintainer doesn't know why | 13:11 |
| minute | but maybe someone can debug that some day, i am not holding my breath atm though | 13:11 |
| josch | the firmware blobs are those found in firmware-deploy? | 13:12 |
| minute | josch: i need to put all of these in /usr/firmware like that, or patch the driver to fix up the paths somehow https://source.mnt.re/reform/qcacld2/-/tree/boundary-CNSS.LEA.NRT_3.1/firmware-deploy?ref_type=heads | 13:12 |
| josch | what is their license? | 13:12 |
| minute | josch: yes | 13:12 |
| minute | josch: i don't know :3 i need to check where i found them | 13:12 |
| minute | ohh probably this https://github.com/boundarydevices/qca-firmware | 13:13 |
| josch | the rest of the driver seems to be bsd 2, 3 and 4-clause and isc licensed | 13:13 |
| minute | josch: i think the driver is dual licensed | 13:13 |
| minute | not sure what this means exactly https://source.mnt.re/reform/qcacld2/-/blob/boundary-CNSS.LEA.NRT_3.1/Makefile?ref_type=heads#L10 | 13:13 |
| minute | josch: ah yeah you're right | 13:14 |
| josch | then that would be two packages, one for main and one for non-free-firmware | 13:15 |
| minute | ok! | 13:15 |
| minute | in any case i managed to build the driver on the target like it is in the repo now, yesterday, and it works | 13:15 |
| josch | the repo on source.mntre.com is the one from https://github.com/TechNexion/qcacld-2.0 with your own modifications? | 13:15 |
| minute | hmm no not from technexion | 13:15 |
| josch | what was your source? | 13:16 |
| minute | https://github.com/boundarydevices/qcacld-2.0/tree/boundary-CNSS.LEA.NRT_3.1 | 13:16 |
| josch | nice, the master branch only has a README in it :D | 13:16 |
| minute | yeah because you can choose your branch :D | 13:16 |
| minute | it is like choose your own adventure | 13:17 |
| minute | btw we only need the driver for wifi. the bluetooth portion works with mainline (it's connected via an internal UART) | 13:17 |
| josch | thanks, now i can upload some more stuff to NEW while we wait for tuigreet and reform-setup-wizard to get accepted :) | 13:17 |
| minute | josch: great, thanks a ton! | 13:18 |
| josch | thank you for the great hardware :) | 13:18 |
| - mark_ (QUIT: Remote host closed the connection) (~mjw@gnu.wildebeest.org) | 13:20 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 13:21 | |
| josch | huh and qcom_cfg.ini and cfg.dat also are somehow needed and need to go into /usr/lib/firmware? | 13:26 |
| - mark_ (QUIT: Ping timeout: 268 seconds) (~mjw@gnu.wildebeest.org) | 13:41 | |
| josch | minute: there are 1.4k lines of diff between https://github.com/boundarydevices/qcacld-2.0/tree/boundary-CNSS.LEA.NRT_3.1 and https://source.mnt.re/reform/qcacld2/-/tree/boundary-CNSS.LEA.NRT_3.1?ref_type=heads are those your copyright? https://mister-muffin.de/p/cQjl.diff | 13:52 |
| minute | josch: qcom_cfg.ini and cfg.dat need to go to /usr/lib/firmware/wlan | 13:54 |
| minute | josch: yeah, those changes are just fixes to make it work with current kernels | 14:04 |
| josch | cool, just making sure that this is from you (as i have to document that) and to make sure that i am not accidentally packaging a slightly incorrect branch :) | 14:07 |
| + mark_ (~mjw@gnu.wildebeest.org) | 14:39 | |
| josch | minute: i chose the ISC license for your debian packaging so that your additions remain trivially compatible with the upstream source, okay? | 14:40 |
| minute | josch: great | 14:54 |
| josch | minute: if you would like to review the finished product, here it is: https://salsa.debian.org/debian/ezurio-qcacld-2.0-dkms | 14:57 |
| minute | josch: amazing, thank you! | 15:08 |
| josch | minute: would you like to review and comment? If yes, I wait with my upload until then, if not I upload some time tonight (afk now) | 15:09 |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 15:09 | |
| - cobra (QUIT: Ping timeout: 245 seconds) (~cobra@user/Cobra) | 16:05 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 16:45 | |
| + chomwitt (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 17:18 | |
| - mark_ (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 17:27 | |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 17:58 | |
| minute | josch: testing your REFORM_DEBIAN_PACKAGES_BRANCH feature in CI now for the first time | 18:00 |
| minute | josch: i won't need to review just yet, to be able to really do that i need to fiddle that dkms package into the system image somehow | 18:01 |
| minute | aw > refusing to build image using packages from reform-debian-packages branch | 18:02 |
| minute | then i can try it locally maybe | 18:02 |
| josch | minute: to set variables you must do it on a branch other than "main" | 18:08 |
| josch | minute: the reason for that is to avoid poisoning what https://mnt.re/system-image points to | 18:11 |
| josch | same with reform-debian-packages | 18:12 |
| josch | i wanted to avoid that pipeline runs with custom variable values would result in https://mnt.re/system-image as well as https://mntre.com/reform-debian-repo containing custom build results | 18:12 |
| josch | unfortunately, as we learned recently, this attempt of mine was futile | 18:13 |
| josch | as gitlab will happily link to build jobs from any branch with the same top commit as the requested branch | 18:13 |
| josch | minute: i listed possible solutions here: https://mntre.com/reform-irc-logs/2024-03-25.log.html#t17:37:00 | 18:15 |
| josch | minute: i also added information to variables.sh so that the script that updates https://mntre.com/reform-debian-repo can make sure that it's not populating the repo with stuff that does not come from the main branch | 18:16 |
| josch | but see the irc log for details | 18:17 |
| minute | josch: ahh my mistake. i didn't want it to do on main | 18:37 |
| josch | minute: you might want to run it in your fork so that you do not accidentally change the contents behind https://mnt.re/system-image | 18:38 |
| josch | after having slept over this problem for two nights, my personally preferred solution is to just forbid running the pipelines with custom variables set if $CI_PROJECT_NAMESPACE is equal to "reform" | 18:41 |
| josch | minute: do you have any uncommitted changes to the qcacld-2.0 packaging? | 18:47 |
| minute | josch: ok, sounds good, and we can just use a fork for CI, makes sense | 18:47 |
| josch | when installing the package it doesn't build the module | 18:47 |
| josch | and it looks like it never calls make in the first place which sounds like an issue you had yesterday | 18:47 |
| minute | josch: oh hm. let me check | 18:47 |
| minute | ok what i did yesterday is that i went into the dkms build dir after installing the package and ran make, which succeeded | 18:48 |
| minute | and then i added the MAKE[0] line to qcacld2-dkms.dkms | 18:48 |
| minute | but didn't run that again on the target | 18:48 |
| josch | ah okay | 18:48 |
| josch | then i'll investigate this later tonight | 18:49 |
| minute | oh thank you | 18:49 |
| minute | i'm currently fighting with the imx8mp/pocket kernel config | 18:49 |
| minute | i found out that we need IMX2_WDT baked into the kernel. if the module is on disk and is not loaded for some reason or there's too much of a delay, the system just resets, which makes debugging more than uncomfortable | 18:49 |
| - chomwitt (QUIT: Ping timeout: 268 seconds) (~chomwitt@2a02:587:7a17:f700:1ac0:4dff:fedb:a3f1) | 18:50 | |
| josch | seems to become a pattern that using features as modules appears to be something that driver developers don't test much | 19:08 |
| + mark_ (~mjw@gnu.wildebeest.org) | 20:04 | |
| - mjw (QUIT: Killed (mercury.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:15 | |
| * mark_ -> mjw | 20:15 | |
| + Guest5307 (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:15 | |
| - hairu (QUIT: Remote host closed the connection) (m-uotkmd@user/hairu) | 20:24 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 20:42 | |
| + hairu (m-uotkmd@user/hairu) | 20:57 | |
| hramrach | it means something already enabled the watchdog before loading the kernel, it's not the faualt of the driver | 20:57 |
| hramrach | of course, if you want to paper over lockup during kernel boot that's the thing to do so it makes sens in some setups | 20:58 |
| - violet (QUIT: Remote host closed the connection) (~vi@user/violet) | 21:07 | |
| + violet (~vi@user/violet) | 21:07 | |
| minute | josch: yeah | 21:08 |
| minute | hramrach: possibly something in uboot | 21:08 |
| + reform1926 (~root@102.44.63.188.dynamic.wline.res.cust.swisscom.ch) | 21:16 | |
| reform1926 | hello | 21:17 |
| vkoskiv | Hello! | 21:17 |
| reform1926 | this is great! | 21:17 |
| reform1926 | sorry im new here | 21:18 |
| vkoskiv | Welcome | 21:18 |
| vkoskiv | Did you just set up a reform? | 21:18 |
| reform1926 | yes | 21:18 |
| vkoskiv | Very cool! What kind of setup did you get? | 21:19 |
| reform1926 | im trying to get familiar with it | 21:19 |
| reform1926 | mnt reform | 21:19 |
| reform1926 | at the sametime trying to learn system | 21:19 |
| reform1926 | i have few questions... | 21:20 |
| vkoskiv | Yep, lots to read. Do ask if you run into any issues/confusion, many people here can help | 21:20 |
| reform1926 | would you be able to help? | 21:20 |
| vkoskiv | Sure | 21:20 |
| reform1926 | ok | 21:20 |
| vkoskiv | I've had mine for a while now | 21:20 |
| reform1926 | first question..---i have installed a DE keyboard but the system dont recogniyed and type with the EN-settings | 21:22 |
| vkoskiv | Are you using sway or some other graphical environment? With sway, I configure it in ~/.config/sway/config | 21:23 |
| vkoskiv | The system image assumes US layout, it doesn't detect it automatically I think. | 21:23 |
| josch | or, even better, in ~/.config/sway/config.d/input | 21:23 |
| vkoskiv | Yeah, that is the recommended way. I just prefer to keep my configs to single file per thing :D | 21:24 |
| vkoskiv | Wouldn't it be possible to do some kind of automatic detection for that, since we do control the firmware? | 21:24 |
| vkoskiv | Not sure if there is any standard way of doing that. | 21:24 |
| josch | vkoskiv: the firmware is only neo2, US and "normal" | 21:24 |
| reform1926 | how can i change the setting for DE instead of EN, to tzpe the right keys... for example, where the kez z is the y and where the y is the z... is sometime overwhelming | 21:24 |
| vkoskiv | josch: I meant the layout of the keeb | 21:25 |
| josch | reform1926: put this into your sway config: https://paste.debian.net/1312218/ | 21:26 |
| josch | vkoskiv: the firmware does not know what is printed on the keys | 21:26 |
| vkoskiv | Good point. Extra complexity would be needed to tell it that info when flashing | 21:27 |
| reform1926 | @vkoskiv- im using root and sway (both) | 21:28 |
| reform1926 | the 2nd question.. my screen is constently flickering, with time it get enozing | 21:32 |
| reform1926 | sorry /// enoying | 21:32 |
| josch | reform1926: do you have an ls1028a reform? | 21:33 |
| reform1926 | dont the reason why is constently doing this | 21:33 |
| reform1926 | @josch- yes | 21:34 |
| josch | reform1926: there is a hardware fix for this issue. You should get into contact with support@mntre.com | 21:35 |
| reform1926 | ok thanks | 21:35 |
| josch | reform1926: details here: https://community.mnt.re/t/research-into-ls1028a-wifi-to-edp-interference/1762 | 21:35 |
| josch | on my ls1028a i solved the problem by wrapping tape around the flex cable | 21:36 |
| josch | but there is a more professional solution :) | 21:36 |
| reform1926 | im sure... :) | 21:37 |
| reform1926 | thank you all for the time and answers.. | 21:39 |
| reform1926 | will leave now and try to cahnge the settings | 21:40 |
| reform1926 | until the next time | 21:40 |
| reform1926 | exit | 21:40 |
| josch | it should be /quit | 21:41 |
| reform1926 | hahahaha | 21:41 |
| josch | irc commands are prefixed with a slash | 21:41 |
| reform1926 | thanks | 21:41 |
| - reform1926 (QUIT: Quit: Leaving) (~root@102.44.63.188.dynamic.wline.res.cust.swisscom.ch) | 21:42 | |
| josch | minute: the magic runes were, that one has to set WLAN_ROOT, MODNAME and CONFIG_* options for the make invocation to build the module. There is a Kconfig which expects these to be set. The Android.mk also hints that the final module name is not wlan.ko but qcacld_wlan.ko. Unfortunately, when installing and building it, I'm getting build failures like: | 21:45 |
| josch | CORE/HDD/src/wlan_hdd_assoc.c:2161:13: error: ‘struct cfg80211_roam_info’ has no member named ‘bss’ | 21:45 |
| josch | CORE/HDD/src/wlan_hdd_hostapd.c:1576:9: error: too few arguments to function ‘cfg80211_ch_switch_notify’ | 21:45 |
| josch | sounds like i'm on the wrong kernel version? | 21:45 |
| josch | this is 6.7.9 | 21:46 |
| digitalrayne | that reminds me i have to work out how to adjust my v2 keyboard for qwerty-jp now that i have the keycaps | 21:55 |
| josch | minute: seems there was a kernel version check around the code which errors out now: https://salsa.debian.org/debian/ezurio-qcacld-2.0-dkms/-/blob/master/debian/patches/fixup.patch?ref_type=heads#L91 | 21:56 |
| josch | full failure log here: https://salsa.debian.org/debian/ezurio-qcacld-2.0-dkms/-/jobs/5506371 | 21:57 |
| josch | the bss member for cfg80211_roam_info seems to have been introduced in git efbabc1165004 which first appeared in v6.0 which is exactly what the patched out version check suggests... hrm... | 21:58 |
| josch | same with cfg80211_ch_switch_notify | 21:59 |
| - robin (QUIT: Remote host closed the connection) (~robin@user/terpri) | 22:23 | |
| + robin (~robin@user/terpri) | 22:23 | |
| + jacobk (~quassel@64.189.201.150) | 23:13 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!