#archlinux-ports | Logs for 2026-04-02

Back
[02:46:51] -!- Daanct12 has joined #archlinux-ports
[02:52:20] -!- Daanct12 has quit [Quit: WeeChat 4.8.2]
[02:55:21] -!- Daanct12 has joined #archlinux-ports
[03:10:11] -!- hcmb has quit [Ping timeout: 272 seconds]
[03:10:12] -!- hcmb_ has joined #archlinux-ports
[03:10:12] hcmb_ is now known as hcmb
[04:08:40] -!- drathir_tor has quit [Remote host closed the connection]
[05:41:48] -!- drathir_tor has joined #archlinux-ports
[06:17:29] -!- SpieringsAE has joined #archlinux-ports
[07:22:44] <solskogen|M> ==> Build started: composable-kernel 7.2.1-1 (Wed Apr 01 17:53:02 2026)
[07:22:44] <solskogen|M> ==> Build finished: composable-kernel 7.2.1-1 (Thu Apr 02 07:55:30 2026) [SUCCESS]
[08:21:52] <solskogen|M> no, that was pytorch :-) when ever rocm and/or cuda is released a lot of stuff needs to be recompiled for some reason.
[08:26:09] -!- Daanct12 has quit [Ping timeout: 245 seconds]
[08:28:17] -!- Daanct12 has joined #archlinux-ports
[09:54:57] -!- drathir_tor has quit [*.net *.split]
[10:02:58] <yuvadm> solskogen|M: how are you orchestrating the builds?
[10:04:00] <solskogen|M> utilizing these scripts: https://github.com
[10:04:02] <phrik> Title: GitHub - solskogen/archlinux-aarch64-builder · GitHub (at github.com)
[10:05:44] <solskogen|M> running them on a machine on AWS kindly provided by @drzee
[10:06:12] <yuvadm> that's what i was looking for, thanks!
[10:06:58] <yuvadm> i guess this has some upstream equivalent for proper x86_64 arch?
[10:06:59] <solskogen|M> 99% of it is coded by AI, so judge the code with that in mind. But it works!
[10:07:08] <yuvadm> 99% slop :)
[10:07:46] <solskogen|M> Well, no. Not really. But it is using makechrootpkg behind the scenes, which is the same as upstream.
[10:08:43] <solskogen|M> But what is basically does it parse the .db files and check if we're out of date, and it then generates a json that provides the dependencies etc.
[10:09:11] <solskogen|M> the dependency checking isn't always working, but in most if the cases it just works.
[10:11:03] <solskogen|M> Sometimes some packages goes from pkgrel 1 to pkgrel 3 (because pkgrel 2 only went into staging) - and that's a bit hard to fix.
[10:11:45] <solskogen|M> The scripts aren't perfect, but they do seem to get the job done.
[10:13:06] <solskogen|M> I'm a huge fan of the motto: If it's stupid and it works, it aint stupid.
[10:14:09] <solskogen|M> I'm not commited to those scripts at all, so if anyone have a better solution I'm all for it.
[10:18:57] -!- kitlith has quit [Ping timeout: 265 seconds]
[10:19:56] -!- kitlith has joined #archlinux-ports
[10:20:54] <yuvadm> they seem to work which is great, from first glance i think there's some better ways to to do dependency resolution with ALPM and maybe networkx for graph stuff
[10:21:04] <yuvadm> will play around and maybe open a PR
[10:32:03] <solskogen|M> Go ahead - and have a lot of fun!
[10:43:56] -!- Daanct12 has quit [Quit: WeeChat 4.8.2]
[11:21:34] -!- yjun123 has quit [Quit: Konversation terminated!]
[11:29:42] -!- yjun123 has joined #archlinux-ports
[12:24:32] -!- qwertviop has joined #archlinux-ports
[12:38:42] -!- qwertviop has quit [Quit: qwertviop]
[12:39:24] -!- qwertviop has joined #archlinux-ports
[12:46:47] -!- yjun123 has quit [Remote host closed the connection]
[13:12:24] <solskogen|M> Oh, I lied. rocm stuff is done.
[13:13:56] <solskogen|M> gromit: and no aarch64 hacks needed at all! Everything just compiled without any modifications
[13:17:52] <gromit> solskogen|M: but is it also usable?
[13:18:52] <solskogen|M> I have no idea. How can I check? :-)
[13:37:54] -!- qwertviop has quit [Ping timeout: 248 seconds]
[13:38:08] -!- qwertviop has joined #archlinux-ports
[13:48:05] -!- drathir_tor has joined #archlinux-ports
[13:57:03] -!- solsTiCe has joined #archlinux-ports
[13:59:15] -!- solsTiCe has quit [Client Quit]
[14:07:23] -!- solsTiCe has joined #archlinux-ports
[14:23:07] <bertptrs> do you have an ARM device with an AMD GPU?
[14:23:36] <SpieringsAE> the latest kernel seems to have broken booting on my x elite machine, will have to investigate a bit
[14:24:34] <SpieringsAE> sadly it also doesn't seem to connect to a network, so I can't ssh in
[14:27:11] -!- SpieringsAE has quit [Quit: SpieringsAE]
[14:28:35] -!- qwertviop has quit [Ping timeout: 252 seconds]
[14:29:35] -!- qwertviop has joined #archlinux-ports
[14:48:48] -!- qwertviop has quit [Ping timeout: 256 seconds]
[15:37:09] -!- qwertviop has joined #archlinux-ports
[15:39:12] -!- drathir_tor has quit [Remote host closed the connection]
[15:40:51] -!- drathir_tor has joined #archlinux-ports
[15:46:34] -!- drathir87 has joined #archlinux-ports
[15:49:13] -!- drathir_tor has quit [Remote host closed the connection]
[15:49:15] drathir87 is now known as drathir_tor
[16:15:50] <solskogen|M> spierings_ae: damn :/ Probably my fault while as I putting stuff as modules.
[16:17:03] <solskogen|M> but it's a bit hard for me to investigate since I don't have that particular device. Perhaps bschnei has a clue?
[16:40:58] <bschnei> check CONFIG_REGMAP_SPMI
[16:41:54] <bschnei> also CONFIG_I2C_QCOM_GENI
[16:43:25] <bschnei> CONFIG_PINCTRL_MSM was also unset. solskogen: did you intentionally set y to m or n or was that a byproduct of changing other settings?
[17:04:49] <solskogen|M> Everything that I broke as not intentionally at all. A byproduct of my lack of understanding the consequences.
[17:17:27] <bschnei> I'm going to try to reapply SpieringsAE's changes on top of yours
[17:18:00] <solskogen|M> Sorry :/
[17:25:56] <bschnei> all good. I don't _think_ it's that bad... (famous last words lol)
[18:10:09] -!- linkmauve has parted #archlinux-ports
[18:21:29] -!- linkmauve has joined #archlinux-ports
[18:22:03] -!- linkmauve has parted #archlinux-ports
[18:28:28] -!- cjc7373 has joined #archlinux-ports
[18:28:34] -!- coherence42 has quit [Ping timeout: 248 seconds]
[18:51:10] <solskogen|M> say when, bschnei - and I'll rebuild it asap.
[19:28:38] <bschnei> I accidentally closed menuconfig w/o saving. :sigh: it's one of those days
[20:04:11] <tpkessler|M> bertptrs: I think not even AMD has one as most ARM devices don't implement the full PCIe specs.
[20:20:22] <bschnei> solskogen: I'll probably need to revisit this later tonight. Obviously if people need to get their device back on track ASAP, I haven't squashed the commits so you can just go to HEAD^1 on the aarch64 branch and build/release that.
[20:26:36] <yuvadm> bschnei: i also ahve some updates for odroid-c4 that i overlooked in my original patch, a few modules that need to be baked into kernel for better support (to avoid adding them as modules in initrmafs)
[20:27:43] <bschnei> yuvadm: may want to hold off on an MR for the time being. we gotta fix devices we borked :)
[20:28:03] <yuvadm> i think it's just three configs for meson devices, but no problem
[20:30:25] <bschnei> ok, that's small enough. go for it, i just might have to sit on it for a sec. we've got amlogic, marvell, rockchip and a bit of broadcom (early pi5 stuff) so i'm trying to figure out if i can keep the expanded support from solksogen without also breaking everyone else's stuff
[20:32:28] <bschnei> one area i know relatively litte about is PHY subsystem. kernel config allows nearly everything to be built as modules, but it seems there could be cases where that needs to be built in based on the description. if anyone knows that area better and can let me know if we can default to everything as module or if i need to try to keep all 'y' options that are already 'y' that would be helpful
[20:36:28] <yuvadm> i dont know this very well but some drivers are better off just being in the kernel as they're needed at boot time
[20:36:45] <yuvadm> for each device probably a good idea to test this prior to updating the config
[20:37:10] <yuvadm> e.g. minimal drivers in kernel for boot, everything else built as modules
[20:37:54] <yuvadm> the alternative for that is that each device has to updated MODULES=() in mkinitcpio manually, not that elegant
[21:49:44] -!- marmis has quit [Quit: Bye!]
[21:54:39] <solskogen|M> spierings_ae: Can you try the kernel in core-testing?
[21:57:03] -!- marmis has joined #archlinux-ports
[22:10:18] -!- titus_livius has joined #archlinux-ports
[22:40:50] -!- hcmb has quit [Ping timeout: 252 seconds]
[22:50:50] -!- hcmb has joined #archlinux-ports