💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2021-06-03.txt captured on 2024-05-26 at 16:15:59.
⬅️ Previous capture (2021-12-17)
-=-=-=-=-=-=-
[2021-06-03T00:02:29Z] <illiliti> then that's 100% problem with efibootmgr/uefi
[2021-06-03T00:04:04Z] <sad_plan> you know what? my config also seems to have been reset. no wonder it wont boot, I dont have the neccessary drivers compiled either. gimme a few minutes and Ill have the new kernel back as its supose to be.
[2021-06-03T00:04:43Z] <sad_plan> i figured so myself, untill I noticed mlre stuff was off. like intel microcode, which I know for sure I dont have enabled.
[2021-06-03T00:04:53Z] <illiliti> keep digging XD
[2021-06-03T00:07:23Z] <kqz> i've had issues with PARTUUID and efistub in the past as well, for now I just have "root=/dev/nvme0n1p3" in my kernel cmdline
[2021-06-03T00:11:22Z] <sad_plan> the kernel config seems to be the issue here. waiting for kernel to build now, and we shall see.
[2021-06-03T00:14:53Z] <riteo> well, gtg for now, bye!
[2021-06-03T00:25:44Z] <sad_plan> and it boots!
[2021-06-03T00:35:34Z] <dante_boi> sad_plan
[2021-06-03T00:35:41Z] <dante_boi> Solved the problem?
[2021-06-03T01:41:54Z] <dante_boi> Anyone uses kakoune or micro?
[2021-06-03T03:36:20Z] <illiliti> fuck, now i know why libinput is slow as shit
[2021-06-03T03:36:28Z] <illiliti> it uses libevdev library which uses ioctl() to fetch input bits
[2021-06-03T03:36:37Z] <illiliti> https://github.com/freedesktop/libevdev/blob/master/libevdev/libevdev.c#L454
[2021-06-03T03:37:12Z] <illiliti> ioctl() is ridiculously slow
[2021-06-03T03:40:22Z] <noocsharp> is there a better way to do what it's using ioctls to do or are you stuck?
[2021-06-03T03:40:43Z] <schillingklaus> is freedesktop the work of the systemd mafia?
[2021-06-03T03:41:00Z] <illiliti> the better way is parsing bits from sysfs
[2021-06-03T03:41:19Z] <noocsharp> lmao sounds fun
[2021-06-03T03:41:44Z] <illiliti> /sys/class/input/*/device/capabilities/*
[2021-06-03T03:42:57Z] <illiliti> schillingklaus: freedesktop == redhat == systemd, so yeah :)
[2021-06-03T03:43:25Z] <illiliti> noocsharp: this is equivalent to ioctl()
[2021-06-03T03:45:01Z] <noocsharp> i see, is there a good reason to use the ioctl over sysfs?
[2021-06-03T03:45:55Z] <illiliti> no reason
[2021-06-03T03:46:03Z] <illiliti> just bad design
[2021-06-03T03:46:08Z] <noocsharp> that's pretty fucked
[2021-06-03T03:47:49Z] <noocsharp> it seems like a pretty easy fix right? literally just substitute sysfs reads for the ioctls?
[2021-06-03T03:48:00Z] <illiliti> yeah
[2021-06-03T03:50:15Z] <noocsharp> also out of curiosity, what are you doing where you are able to notice the latency from ioctls?
[2021-06-03T03:51:42Z] <illiliti> https://github.com/illiliti/libudev-zero/commit/fca883ed54688481e761e92ec0955af284191732
[2021-06-03T03:55:38Z] <noocsharp> wait is that 2.964 seconds?!?!
[2021-06-03T03:58:55Z] <schillingklaus> ratpoison does not follow the mess called freedesktop
[2021-06-03T04:00:27Z] <illiliti> it was until i dropped ioctl in favor of sysfs. if libinput will drop ioctl, the whole xorg/wayland startup would be much faster
[2021-06-03T04:01:36Z] <illiliti> for now, it calls open/ioctl/close for every input device which is very slow
[2021-06-03T04:01:53Z] <illiliti> read() from sysfs is much faster
[2021-06-03T04:04:11Z] <noocsharp> i think this would be of interest in #oasis
[2021-06-03T04:05:13Z] <schillingklaus> because oasis is addicted to wayland?
[2021-06-03T04:07:40Z] <noocsharp> well yes, oasis uses wayland, and mcf is not averse to patching programs
[2021-06-03T04:21:31Z] <noocsharp> xorg is freedesktop btw schillingklaus
[2021-06-03T04:24:17Z] <acheam> lol
[2021-06-03T04:32:42Z] <illiliti> ah, there is actually the only one reason to keep ioctl. this is freebsd compatility
[2021-06-03T04:32:52Z] <illiliti> but seriously ...
[2021-06-03T04:33:58Z] <illiliti> i would rather to add fast/slow path to code.
[2021-06-03T04:34:36Z] <illiliti> e.g: if exist("/sys"); then do read("/sys/..."); else ioctl()
[2021-06-03T05:21:03Z] <noocsharp> or even #ifdef __linux__
[2021-06-03T05:22:26Z] <noocsharp> or just checking for /sys at compile time
[2021-06-03T05:32:26Z] <testuser[m]_> hi
[2021-06-03T05:36:48Z] <testuser[m]_> dilyn accordimg to gentoo ff is brokem with gcc 11
[2021-06-03T05:36:59Z] <testuser[m]_> forcing clang isnt a big deal since its needed amyway
[2021-06-03T10:41:26Z] <sad_plan> dante_boi: yeah, it was appearantly that my kernel config got reset at some point, so it was all wrong. so no wonder it didnt boot :')
[2021-06-03T14:16:21Z] <zr> node-irc says goodbye?
[2021-06-03T14:16:33Z] <zr> oh, probably matrix bridge having an oopsie
[2021-06-03T14:17:12Z] <testuser[m]_> yea
[2021-06-03T15:15:51Z] <dante_boi> testuser[m]_
[2021-06-03T15:16:22Z] <dante_boi> The steam script in your dotfiles need something of the void chroot?
[2021-06-03T15:16:44Z] <testuser[m]_> you can point STEAMROOT to any chroot containing the relevant libs for steam
[2021-06-03T15:18:02Z] <testuser[m]_> hi dilyn
[2021-06-03T15:18:16Z] <dilyn> o/
[2021-06-03T15:19:21Z] <dante_boi> All other stuff is your kiss dir right?
[2021-06-03T15:19:37Z] <testuser[m]_> wdym
[2021-06-03T15:21:08Z] <testuser[m]_> it takes pseudo-fs stuff like from /dev from kiss host, libs from chroot dir, and mounts the dir of your choice on $HOME
[2021-06-03T15:21:40Z] <soliwilos> Is the change to using gmake in nnn's community build file intentional?
[2021-06-03T15:22:06Z] <dilyn> well that wasn't supposed to happen. thx soliwilos
[2021-06-03T15:22:22Z] <dilyn> testuser: I did figure it was a gcc 11 problem, was just about to start looking! thx
[2021-06-03T15:22:35Z] <soliwilos> No problem.
[2021-06-03T15:22:40Z] <testuser[m]_> nice
[2021-06-03T16:11:18Z] <zr> dilyn: /18
[2021-06-03T16:11:20Z] <zr> oops
[2021-06-03T16:11:22Z] <zr> sorry
[2021-06-03T16:24:55Z] <dante_boi> testuser[m]_:
[2021-06-03T16:25:10Z] <dante_boi> Your script had bug?
[2021-06-03T16:25:18Z] <testuser[m]_> not that im aware of
[2021-06-03T16:25:31Z] <dante_boi> Bwrap can't locate execvp
[2021-06-03T16:25:53Z] <dante_boi> And lone 16 i don't know why it not recognize --bind
[2021-06-03T16:25:53Z] <testuser[m]_> that means it cant find the binary you're passing it
[2021-06-03T16:26:33Z] <testuser[m]_> show the exact error
[2021-06-03T16:28:45Z] <dante_boi> Do you have matrix? Is better i talk in pv chat
[2021-06-03T16:29:04Z] <dilyn> ` 3:11.52 cargo:warning=/root/.cache/kiss/proc/13804/build/firefox/obj-x86_64-pc-linux-musl/dist/system_wrappers/type_traits:3:15: fatal error: 'type_traits' file not found` :|
[2021-06-03T16:30:46Z] <dante_boi> sad
[2021-06-03T16:31:41Z] <dilyn> dumb, dumb dumb
[2021-06-03T16:31:55Z] <dilyn> some std problem, I assume. ~love it~
[2021-06-03T16:32:05Z] <dilyn> restarting petition to drop firefox
[2021-06-03T16:32:15Z] <dante_boi> Lol
[2021-06-03T16:32:16Z] <testuser[m]_> https://github.com/git-bruh/kiss-repo/commit/a856822ba350c266229f5efa4959202ff89eea05 this works for me
[2021-06-03T16:32:32Z] <testuser[m]_> but on glibc untested on musl
[2021-06-03T16:32:38Z] <testuser[m]_> dante_boi: testuser:kde.org
[2021-06-03T16:36:12Z] <dilyn> yeah I have roughly those patches
[2021-06-03T16:36:19Z] <dilyn> hmhmhmhmhm
[2021-06-03T17:16:55Z] <omanom> i think i'll take a stab at rewriting kiss in python
[2021-06-03T17:17:19Z] <omanom> name it `piss`
[2021-06-03T17:24:52Z] <testuser[m]_> appropriate
[2021-06-03T18:54:25Z] <mmatongo> whats perntium
[2021-06-03T18:55:20Z] <omanom> https://libera.irclog.whitequark.org/kisslinux/2021-06-02#30008243;perntium
[2021-06-03T18:55:36Z] <mmatongo> oh ghad
[2021-06-03T18:55:40Z] <omanom> just a mistype XD
[2021-06-03T18:56:31Z] <mmatongo> hehehe
[2021-06-03T20:27:02Z] <acheam> anyone know of a replacement to the default web server directory browser with tarball generation?
[2021-06-03T21:08:27Z] <noocsharp> illiliti: when you said removing ioctls took you from 3 seconds to 1 second, what were you measuring?
[2021-06-03T21:15:00Z] <illiliti> libinput list-devices
[2021-06-03T21:30:20Z] <noocsharp> do you just have a lot of devices or are you running on a slow computer? it only takes 0.14 seconds for me
[2021-06-03T21:30:59Z] <noocsharp> although i'm running eudev
[2021-06-03T21:47:03Z] <kqz> takes 0.8-0.9 seconds for me with eudev, with 13 devices
[2021-06-03T21:48:10Z] <illiliti> [e]udev is different, it caches the results. that's why it faster. libudev-zero can't cache results because there is no way to do it by design
[2021-06-03T21:49:27Z] <omanom> strongswan is such a PITA compared to wireguard