#
…

2026-02-16

00:21n3ph has quit [Ping timeout: 480 seconds]
00:30alice has joined #asahi-alt
00:30alice is now known as Guest2841
00:46mixi has quit [Ping timeout: 480 seconds]
00:57mixi has joined #asahi-alt
01:20chadmed has quit [Quit: Konversation terminated!]
01:20chadmed has joined #asahi-alt
01:37hightower3 has joined #asahi-alt
01:44vimproved has quit [Read error: No route to host]
01:44vimproved has joined #asahi-alt
02:06hightower4 has joined #asahi-alt
02:06ink has joined #asahi-alt
02:07inkfinishing up a gentoo install on my 2021 m1 pro mbp and i have a couple of questions if anyone can maybe help
02:09inkmain thing is i've noticed that main profile hevc just does not want to decode - mpv keeps giving 'Illegal instruction'
02:12JamesCalligeros[m]youve compiled something wrong then
02:13hightower3 has quit [Ping timeout: 480 seconds]
02:14inkwould it be better then to start again on a fresh install, or how hard would it be to track down the issue
02:21hightower3 has joined #asahi-alt
02:29hightower4 has quit [Ping timeout: 480 seconds]
02:30JamesCalligeros[m]launch mpv under a debugger and do a backtrace to see what is trying to use an illegal instruction. i doubt it's mpv itself.
02:42Guest2841 has quit [Ping timeout: 480 seconds]
02:49alice has joined #asahi-alt
02:50alice is now known as Guest2851
02:51inklibavcodec.so
03:04Guest2851 has quit [Ping timeout: 480 seconds]
03:14tobhe has joined #asahi-alt
03:21tobhe_ has quit [Ping timeout: 480 seconds]
03:22KongrooParadox has joined #asahi-alt
03:25[KongrooParadox] has quit [Ping timeout: 480 seconds]
03:30inkSpecifically it's from ff_hevc_put_hevc_qpel_h32_8_neon_i8mm () in libavcodec.so
03:31inkWhich is from ffmpeg
03:34mrkcee has joined #asahi-alt
03:39chadmed has quit [Quit: Konversation terminated!]
03:40chadmed has joined #asahi-alt
03:41inki've tried recompiling ffmpeg before this though
03:41inkit was one of the first things i did
03:47[KongrooParadox] has joined #asahi-alt
03:48JamesCalligeros[m]what USE flags are set for ffmpeg, what cflags are you using to build it, and what CPU_FLAGS_ARM do you have set for the package
03:50KongrooParadox has quit [Ping timeout: 480 seconds]
03:51alice has joined #asahi-alt
03:51alice is now known as Guest2857
03:53inkI have most use flags set for ffmpeg - pretty much everything minus a few (doc, frei0r, jack, jpeg2k, librtmp, libsoxr, modplug, nvenc, speex, srt, and verify-sig)
03:53mrkcee has quit [Quit: Konversation terminated!]
03:54inkCFLAGS I didn't change from what I have in make.conf, which is the set that was recommended in the old guide
03:55ink-march=armv8.5-a+fp16+simd+crypto+i8mm -mtune=native -O2 -pipe
03:56inkDidn't set any CPU_FLAGS_ARM
03:56JamesCalligeros[m]no guide i have written has recommended -mtune
03:56JamesCalligeros[m]do emerge -pv ffmpeg and show me what set of CPU_FLAGS_ARM it has enabled
04:09inkThe guide was the Gentoo on an M1 Mac made by Jared
04:11inkAs far as CPU_FLAGS_ARM I'm not fully sure how to see specifically them - there is no categorization in the USE output when I do emerge -pv ffmpeg
04:20inkWhen I check against the output of cpuid2cpuflags on my machine ffmpeg (according to the wiki) has none of them enabled
04:20inkffmpeg-chromium does though
04:21inkffmpeg just doesn't have any cpu_flags_arm listed
04:21inkvlc has neon
04:22JamesCalligeros[m]try setting -march to native instead of that jumble of isa extensions on top of 8.5
04:22JamesCalligeros[m]and then rebuild ffmpeg
04:23Guest2857 has quit [Ping timeout: 480 seconds]
04:25inkso -march=native -mtune=native -02 -pipe
04:25inkor wait no mtune you said
04:26inkthis is inherited from my common flags should i change those to just -march=native -02 -pipe
04:28inkthe fem-overlay ffmpeg does actually have arm flags
04:29inkbut that seems irrelevant
04:29JamesCalligeros[m]yeah just get rid of mtune entirely
04:31inkafter this should i do a full recompile from scratch
04:32JamesCalligeros[m]nah dont waste your time. stuff will recompile as updates happen
04:33inkthank you
04:33inkalso quick other question
04:34inksteam_aarch64 currently opens the app in low res mode
04:35inki think maybe bc i'm using sway and that forces the display to scale=2?
04:35inki tried switching to scale 1.5 but the icons are so tiny i'm not sure i can deal with it just for a steam fix
04:36inkalso holy moly that worked
04:36inkthank you
04:36inki assume this will help with other stuff as well
04:36inki might just run a full emerge overnight anyways
04:36inkmy world list isn't that big yet
04:37inkshould i set my cpu_flags_arm to the output of cpuid2cpuflags?
04:37inkis that a useful thing to do
04:38inkedsp neon neon-fp16 thumb vfp vfpv3 vfpv4 vfp-d32 aes sha1 sha2 crc32 asimd asimddp asimdfhm asimdhp v4 v5 v6 v7 v8 thumb2
04:38inkthis is what it spits out
04:45JamesCalligeros[m]yeah it will help with some packages
04:45JamesCalligeros[m]theres not a lot we can really do about steam, theres a bunch of stuff in between it and your compositor...
04:46KongrooParadox has joined #asahi-alt
04:46inki was just curious if there was something that could be done to turn the scaling off for steam
04:46inklike crossover has the enable retina option
04:47inkand you can do that with regedit in wine as well
04:48[KongrooParadox] has quit [Ping timeout: 480 seconds]
04:51inkhuh hevc is still broken in legcord-bin for me
04:52JamesCalligeros[m]are you sure legcord-bin doesnt vendor its own ffmpeg
04:52inki'm sure that it probably does
04:53inkmight just uninstall and try compiling from source
04:53alice has joined #asahi-alt
04:53inkor use webcord, but then push to talk doesn't work globally
04:53JamesCalligeros[m]yeah if it vendors a binary ffmpeg built with e.g. sve2 then its not going to work lol
04:53inkand probably has the same issues anyways
04:53alice is now known as Guest2860
04:54JamesCalligeros[m]i just use discord in the browser if i need to use discord
04:54inki wonder if firefox has a push to talk global extension like chrome does
04:57inkok well the problematic videos that weren't playing in firefox before are playing in firefox now
04:58inkso yeah discord browser i guess is the way to go
05:10inkthank you very much though for being patient with me
05:12leioink: why are you using -02?
05:14inkthat was also on the list of recommendations
05:14leiothat cpuid2cpuflags is total bullshit with thumb2 and co in there, but should theoretically not hurt if they are enabled, as arm32 only (which apple silicon does not have)
05:14leioNo, it was not, also how did you end up with Jared guide, it's some unofficial guide that has various issues
05:14leioit might have had -O2, but not -02.
05:14inkthen i'm blind
05:15inki swear i did copy paste though
05:15leioand all your binaries are zero optimized by compiler
05:15leioif it is -02
05:16leiowhich I guess is not, because then it would gcc: error: unrecognized command-line option ‘-02’
05:16inkit wasn't
05:16inkit was in fact -O2
05:17leiook, so no need to recompile all for that reason
05:17leioI really should look into what's up with cpuid2cpuflags there :(
05:17JamesCalligeros[m]i wish we had just insisted that this guide be taken down
05:17JamesCalligeros[m]im not going to say i told you so though ;)
05:18leioI don't recall having a strong opinion of that, but how did ink end up with that instead of the real guide
05:18inkit was the first one i found
05:18inki didn't actually use it that much beyond the make.conf
05:19leioand what did you search for, I'm getting correct guide as first for searches; maybe we need to SEO this thing
05:19inki searched gentoo apple silicon i think
05:19inkor maybe just gentoo m1
05:20inkgentoo m1 would make sense since m1 is in the title of the jared guide
05:20leiook, I need to use google to get that for that search, others are more sensible
05:20leio(in case of gentoo apple silicon)
05:21inki mean i kinda just jumped into this blindly
05:21inkto be fair
05:22inkbut i've generally had a good time!
05:22leioI don't quite get why ffmpeg went bad though with those original CFLAGS
05:24inkso which cpu flags are sensible?
05:24leioah, probably the main armv8.5-a part
05:24leioas, well, M1 is a armv8.4
05:25leiopastebin the lscpu output?
05:26inkthe flags part
05:26leioI think what it spit out is probably fine, it's just including stuff that is wrong but isn't applicable to arm64, so won't hurt
05:26inkor the whole thing
05:26leioyes, Flags
05:26inkfor lscpu
05:26inki mean i could just copy it here
05:27inkunless that's a problem
05:27leioif it's just a single line picked out like that, yes
05:27inkfp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 asimddp sha512 asimdfhm dit uscat ilrcpc flagm
05:27inkssbs sb paca pacg dcpodp flagm2 frint
05:29leioso I guess that means we filter out some to start with and are left with "neon aes sha1 sha2 crc32 asimd asimddp asimdfhm asimdhp"
05:30leioI wonder if I could nerdsnipe chaos_princess or someone to look at cpuid2cpuflags instead of talking to the tools maintainer/author or filing bugs :P
05:30JamesCalligeros[m]whats the point when the whole CPU_FLAGS_ARM situation is fucked anyway
05:30leioit's not that critical as those extras are just masked on arm64 anyways afair
05:30JamesCalligeros[m]no point fixing cpuid2cpuflags until we know whats going to happen long term with arm64
05:31JamesCalligeros[m]are we getting our own USE_EXPAND?
05:31leioI think we know what's going to happen, just... need to do the work :(
05:31leioI don't understand why the tool invents those thumb and co things on arm64 in the first place
05:32leio(yes, separate USE_EXPAND)
05:32inkwhy is it not already that wa
05:34JamesCalligeros[m]historical reasons
05:35JamesCalligeros[m]like all non-x86 platforms, arm64 has been considered an afterthought for tinkerers playing with grown up toys until relatively recently
05:35JamesCalligeros[m]now that there are multiple desktop-class arm64 platforms, problems like this are becoming more difficult to ignore
05:36inkwait neon isn't even on my list of flags
05:36inkeither
05:40leiothat's a whole other mess from this. But it was there for you.
05:41leiobut it is masked on arm64, except some things that unmask it
05:41leiosome ebuilds that react to CPU_FLAGS_ARM=neon are for 32-bit neon and will fail to compile with 64-bit neon
05:42leioso it is masked to not break, on arm64 packages should just unconditionally enable 64-bit neon on arm64 without flags, and many do; some don't and needs sorting.
05:42leio(when I say many, it's not many in count; not many packages overall have hand-written assembly to be using CPU_FLAGS_*)
05:43inkwhat are v4 through v8
05:44inki assume for armv4 through armv8, and if m1 is 8.4 why would those tags not be relevant
05:44leioarm32 things, profiles/desc/cpu_flags_arm.desc document these, I don't recall what tools output those for user convenience
05:44leioand that's the problem with sharing CPU_FLAGS_ARM in a nutshell
05:45leiosharing would be kind of needed if we did 32-bit multilib on arm64, but we do not
05:47inkwhat would i type in to recompile everything
05:47leiomeh, why do we have vfp flags unmasked on arm64
05:47leio-e flag to your normal @world line added
05:48inkwith update in it still?
05:48inkor literally just emerge -e @world
05:48leiosure, shouldn't matter, -e makes it think nothing is installed for dep resolution
05:49inkso the update actually doesn't matter then if it just recompiles everything from scratch anyways?
05:49leioI said on top of your normal world line to get stuff like "--ask" (-a) or such ;)
05:49inki should probably run it through genlop first i want to see how long it thinks it will take
05:50leiopersonally I would just gradually migrate and rebuild specific things if something looks broken like ffmpeg did; the wrong armv8.5-a wasn't that great
05:51ink1 day 7 hours 24 minutes
05:51leioand on downtimes chip away at the package list that was compiled with armv8.5-a via some grepping of /var/db/pkg/*/*/CFLAGS
05:51leiobut that's me, to others I'd have to say -e :D
05:52inki mean i have a back up laptop
05:52inki could just let it rip
05:52leiobut if your -e gets interrupted at some point, might be handy to not rebuild all those things yet again.
05:52leioif you just started, most things might be safe from stage3 still
05:52inkthe most i've seen it chain is like
05:52ink30 packages
05:52inkbefore an install chunk
05:53inkbut i guess in that case i caught it and stopped it before it continued
05:53inkdefault behavior is just to keep going right?
05:53JamesCalligeros[m]Honestly who cares just let stuff update
05:54JamesCalligeros[m]Its not like those flags dont work, theyre just not optimal
05:54inki probably will
05:54JamesCalligeros[m]They were from when gcc did not do proper feature detection for apple socs. That was fixed in like gcc 13
05:55leiobut it was armv8.5-a, isn't M1 a armv8.4-a?
05:55leioif that doesn't hurt, then why did ffmpeg break
05:56inki'm also having random gl errors on firefox, but they're not actually breaking anything
05:57leiobut I mostly agree, just would at least at some point clean it up by recompiling what didn't naturally get away from those CFLAGS
05:58chadmed has quit [Quit: Konversation terminated!]
05:58chadmed has joined #asahi-alt
05:59Guest2860 has quit [Ping timeout: 480 seconds]
06:04inkok a total of 5 packages wanted to be updated
06:05inkfrom the cpu_flags_arm definition
06:05inkactually 4 one was from when i cleaned out some package.use files earlier
06:07alice has joined #asahi-alt
06:08alice is now known as Guest2867
06:09leiodo you have the list still?
06:09leioas I would have expected around 0 rebuilds from that, so might be a good list to review (as a gentoo dev)
06:09[KongrooParadox] has joined #asahi-alt
06:12KongrooParadox has quit [Ping timeout: 480 seconds]
06:12inkxz-utils (crc32 flag added), nettle (aes, sha1, sha2 added), libgcrypt (aes, sha1, sha2 added), libaom (crc2 added)
06:13inkalso realized for push to talk i can also just bind a toggle mic mute and that does literally the same thing
06:19leioah, ok, just additions as you hadn't ran with the cpuid2cpuflags output yet or so
06:23inkthere really are not that many packages that actually use the flags
06:36jannauthe problem was '+i8mm' in the CPU flags. that extension is not present on M1*
06:38jannauM1 is armv8.4-a but armv8.5-a has afaik no mandatory features which matter to compilers
07:24Guest2867 has quit [Ping timeout: 480 seconds]
07:34alice has joined #asahi-alt
07:34alice is now known as Guest2873
07:42Guest2873 has quit [Ping timeout: 480 seconds]
07:47nst_ has quit [Quit: WeeChat 4.7.2]
07:48nst has joined #asahi-alt
07:50leioI shouldn't have been checking the +i8mm stupidly against my own M2 flags
08:00chadmedoh its only optional in v8.2-a
08:01chadmed8.1* even
08:01chadmedand mandatory only for 8.6
08:10ink has quit [Ping timeout: 480 seconds]
08:35alice has joined #asahi-alt
08:36alice is now known as Guest2881
09:28KongrooParadox has joined #asahi-alt
09:31[KongrooParadox] has quit [Ping timeout: 480 seconds]
09:50pionaiki has joined #asahi-alt
10:19hightower4 has joined #asahi-alt
10:26hightower3 has quit [Ping timeout: 480 seconds]
10:26hightower3 has joined #asahi-alt
10:27hightower4 has quit [Ping timeout: 480 seconds]
11:02Atid[m] has joined #asahi-alt
11:26n3ph has joined #asahi-alt
11:37yuka has quit [Remote host closed the connection]
11:38yuka has joined #asahi-alt
11:43Guest2881 has quit [Ping timeout: 480 seconds]
12:17[KongrooParadox] has joined #asahi-alt
12:19KongrooParadox has quit [Ping timeout: 480 seconds]
12:24alice has joined #asahi-alt
12:24alice is now known as Guest2895
12:27pionaiki has quit [Quit: Leaving.]
13:03pionaiki has joined #asahi-alt
13:17craftyguy has quit [Ping timeout: 480 seconds]
13:18sam_ has quit [Ping timeout: 480 seconds]
13:29pionaiki has quit [Remote host closed the connection]
13:35craftyguy has joined #asahi-alt
13:47Guest2895 has quit [Remote host closed the connection]
13:53sam_ has joined #asahi-alt
13:54kraem is now known as Guest2898
13:54kraem has joined #asahi-alt
13:58Guest2898 has quit [Ping timeout: 480 seconds]
14:00alice has joined #asahi-alt
14:00alice has quit [Read error: Connection reset by peer]
15:20sergeyt[m] has joined #asahi-alt
15:22kraem is now known as Guest2903
15:23kraem has joined #asahi-alt
15:26Guest2903 has quit [Ping timeout: 480 seconds]
15:47chadmed has quit [Quit: Konversation terminated!]
15:47chadmed has joined #asahi-alt
15:55zumi has quit [Remote host closed the connection]
18:26Dcow[m] has joined #asahi-alt
18:29fumnanya has joined #asahi-alt
18:49fumnanya has quit [Quit: fumnanya]
18:49fumnanya has joined #asahi-alt
18:55Super has joined #asahi-alt
18:55Super has quit []
19:08ink has joined #asahi-alt
19:21ink has quit [Quit: Leaving]
19:34n3ph has quit [Ping timeout: 480 seconds]
20:17ink has joined #asahi-alt
20:37inkso i'm probably going to start my gentoo install over for a variety of reasons (primarily that i had no idea what i was doing and was treating the first install mostly as a learning experience, but now i have a very messy install as a result, and it's fairly fresh so i'm not losing much by restarting)
20:37inkbut one thing i want to do is encrypt my root partition
20:38inkwhich i didn't do the first time around
20:38inkso (1) is this advisable, and (2) if so what settings in luksFormat are well suited for an m1 pro
20:39chaos_princess1. i do so, seems to be working fine. 2. default ones
20:39inkthanks!
20:39inkdo you use a keyfile or a passcode
20:41inkis there anything special i need to do with dracut/grub to make things work, or will using dps standards just make things work
20:42chaos_princessiirc it was just standard dracut stuff
20:42chaos_princess+plymouth for graphical password prompt
20:44inkbut basically just `cryptsetup luksFormat /dev/nvme0n1p#` -> `cryptsetup luksOpen /dev/nvme0n1p# root` and follow the guide as usual?
20:44jannautty text prompt works as well although it might be pushed of screen by kernel log messages. input still works
20:45inkquiet should fix the log messages right?
20:46jannaudoesn't the gentoo handbook cover luks?
20:46chaos_princessi think i also changed the partition type to whatever one causes systemd to mount it as rootfs automatically
20:46inkLinux root (ARM-64)
20:46chaos_princesssounds about right
20:46inkthe handbook does cover it but i was wondering if anything other than default was advisable
21:01n3ph has joined #asahi-alt
21:23inkwhat makeopts do y'all use
21:25hightower4 has joined #asahi-alt
21:25inkm1 or m2?
21:26JamesCalligeros[m]Both but mostly m2 pros and maxes
21:27JamesCalligeros[m]It doesnt matter, they all have either 8 or 10 cpus
21:27inki'm worried about j8 on my m1 pro because it does have 8 cores but not all are performance
21:27inkor i guess not worried per say but wondering if that's still worth it
21:27JamesCalligeros[m]The ecores are fine to build on
21:28inkalrighty
21:32hightower3 has quit [Ping timeout: 480 seconds]
21:36inkso the old jared guide which i originally followed (woops) mounts the efi partition under /boot inside the chroot, but should it actually be /efi?
21:38chaos_princessjust mount it at /boot, it is fine
21:49fumnanya has quit [Quit: fumnanya]
21:49fumnanya has joined #asahi-alt
21:54JamesCalligeros[m]Making the esp /boot makes things a bit easier and is safe
21:54JamesCalligeros[m]Just remember to prune old kernel versions
22:00n3ph has quit [Ping timeout: 480 seconds]
22:09n3ph has joined #asahi-alt
22:13inkand for global use flags, i assume wayland is a good idea
22:13JamesCalligeros[m]i have -X set globally and just turn it back on for whatever packages portage complains about
22:14inki was actually wondering about that
22:14inkbut that makes sense
22:14inkare there any other particularly important ones for asahi
22:15chaos_princessthe only specific one is VIDEO_CARDS=asahi
22:36inkwayland is set by the desktop profile
22:36inki guess that makes sense
22:36inkand X is set in package.use.force in a desktop profile too
22:37chaos_princessi wouldn't go with -X imo
22:38jannaume neither except for maybe a few packages there that results in installing xorg-xserver. and even that is mostly harmless
22:39chaos_princessi think some packages use X to mean any gui at all
22:40inkpipewire?
22:40inki assume that's a good idea
22:40inki believe pulseaudio is set in desktop already
22:41chaos_princesspipewire is _the_ sound server in current year
22:42inkyeah, but the desktop profile doesn't set it
22:42inkjust pulseaudio
22:42inkhence i assume putting it in global use is good
22:42chaos_princessthat is fine too, it will just cause things to talk to pipewire using the pulseaudio protocol
22:43JamesCalligeros[m]Set pipewire and screencast. Some packages still use screencast to mean “build with pipewire support”
22:44JamesCalligeros[m]Notably firefox iirc
22:44inki like the policykit use.force comment
22:48inki think my final list is "pipewire screencast x265" for now
22:48inkbut i'll probably add some system-*'s
22:49chaos_princessi think you'd want system-llvm at least
22:49chaos_princesssince something will surely want rust, and something else will surely want clang
22:50chaos_princessm-chips are fast, but building llvm twice is still annoying
22:51inkis there a dedicated list somewhere of available system-* flags
22:51JamesCalligeros[m]yeah especially if youre not evil and naughty and force-disable most of the llvm targets
22:51inkthey're not in use.desc
22:52chaos_princessthey are mostly per-package
22:52chaos_princessjust do emerge -a and see what is there
22:52inki'll probably at least set system-ffmpeg
22:53inkconsidering vendored ffmpeg breaks a lot of video playback in legcord-bin for me
22:54inksetting EMERGE_DEFAULT_OPTS to jobs=1 still kills the log torrent right
22:55inkor rather --jobs 1
22:55chaos_princessyou can also do -q
22:56chaos_princesswhat machine are you on?
22:56chaos_princessand how much ram
22:56ink2021 m1 pro mackbook pro
22:56ink16GB ram
22:56chaos_princess--jobs 500 :P
22:57chaos_princessm1 pro mbp here, 32gb, i have -j (which is unlimited)
22:58inkis it not worth it then to mount tmpfs to /var/tmp
22:58inki guess if you have 32gb of ram
22:58inkit's less of an issue
22:58inki assume mounting tmpfs makes compilation at least a little faster
22:59inkbut then it's possible to run out of memory
22:59chaos_princesspage cache exists, it is really good on linux
23:00inkso it's not necessary?
23:00chaos_princessno, this mattered more when /var/tmp was on spinning rust, not a nvme ssd
23:00inkunderstandable
23:01chaos_princessand i think we do the same cheating at fsync that macos does
23:02inkso i guess EMERGE_DEFAULT_OPTS="-q"
23:02jannauyes, see commit 09075fd17a4f ("apple-nvme: defer cache flushes by a specified amount")
23:03JamesCalligeros[m]does... does q actually stop portage from logging each packages' build stdout to disk?
23:03chaos_princessneither does -j
23:04inkbut it will stop the log spam right
23:04chaos_princesswell, j still dumps stdout to terminal on failure, so i assume q does to
23:05inkshould i set video cards now
23:05JamesCalligeros[m]Only if theres a single failure. If you set keep-going and more than one package fails it wont dump both
23:05inki assume doing it now causes no harm and i already have make.conf open
23:06inkVIDEO_CARDS="asahi"
23:06JamesCalligeros[m]Yeah just set video_cards. It only really affects mesa and virglrenderer but you definitely absolutely want it set
23:07inkok another thing
23:07ink--changeduse or --newuse
23:08chaos_princessshouldn't matter much
23:08inkit seems like it really only changes very edge cases
23:08inkbut i like the look of -avuDN
23:08chaos_princessmy default incantation is `-auDNj --with-bdeps=y`
23:11JamesCalligeros[m]Same but i add --keep-going so that i can go do other stuff and have a higher chance that it will have actually gotten through more than three packages when i come back
23:13inkoh wait
23:13inkcpu flags
23:14inki can just check the logs from yesterday
23:14chaos_princessJamesCalligeros[m]: what even is up with your system that it fails often enough to matter
23:15inki have no idea
23:15inkthat was a reply
23:15inki am not used to irc
23:15inkCPU_FLAGS_ARM64="aes sha1 sha2 crc32 asimd asimddp asimdfhm asimdhp"
23:15inki think this is what it was
23:16JamesCalligeros[m]lmao it doesnt matter often. there was a recent bug in qtwebengine that made it try to build SVE2 and SME stuff, and then the occasional fex and llvm-mingw things we deal with
23:16JamesCalligeros[m]youve seen my make.conf it's not that controversial :P
23:16n3ph has quit [Ping timeout: 480 seconds]
23:17chaos_princessi think i mostly got llvm-mingw in a good shape
23:17JamesCalligeros[m]yeah i havent had it explode since -r3 got merged into ::gentoo
23:18fumnanya has quit [Ping timeout: 480 seconds]
23:19inkhow inadvisable is it to be on ~arm64
23:19chaos_princessit is inadvisable to be on arm64 on this platform :P
23:19inkok well
23:20inkthat's fair
23:21inki set ACCEPT_KEYWORDS="~arm64" on my original install
23:21inkand nothing seemed to break too bad that wasn't for a different reason entirely
23:23JamesCalligeros[m]keep in mind that "unstable" doesnt necessarily mean "broken and crashes a lot"
23:24chaos_princesssometimes it means "everyone just runs ~arm64 and forgets to stable things"
23:26inki feel like if i don't set ~arm64 globally my package.accept_keywords will be terribly bloated
23:27inkoh dear god this is gonna be a monster emerge
23:27inkit's probably better to do this post first reboot
23:28fumnanya has joined #asahi-alt
23:28inkthough i guess it doesn't matter that much
23:36fumnanya has quit [Ping timeout: 480 seconds]
23:40inkis system-service useful for a systemd install
23:41chaos_princessthat sounds package-specific
23:56n3ph has joined #asahi-alt
…