#archlinux32 | Logs for 2023-04-17

Back
[12:01:52] Foxboron is now known as FXB
[12:05:59] FXB is now known as Foxboron
[12:37:34] -!- zxrom has quit [Quit: Leaving]
[13:55:25] -!- drathir_tor has quit [Remote host closed the connection]
[13:55:51] -!- drathir_tor has joined #archlinux32
[15:22:12] -!- ConSiGno has joined #archlinux32
[15:24:48] <ConSiGno> I checked the logs, actually :D
[15:25:05] <KitsuWhooa> awesome
[15:25:32] <KitsuWhooa> What are you using to test sdl2 anyway?
[15:25:48] <ConSiGno> chocolate-doom mostly
[15:26:08] <ConSiGno> I know it's an SSE instruction on sdl2 that's causing the SIGILL
[15:26:21] <KitsuWhooa> do you remember which one?
[15:27:12] <KitsuWhooa> Also, is chocolate-doom in the repos?
[15:27:25] <ConSiGno> it's not in the repo, I built from source
[15:27:29] <KitsuWhooa> cross compiled it?
[15:27:35] <ConSiGno> the instruction was movss
[15:27:42] <ConSiGno> I built it on the box itself
[15:27:50] <ConSiGno> doesn't take long
[15:28:07] <KitsuWhooa> I remember building a version of doom on my P3 laptop and it took forever
[15:28:20] <ConSiGno> it takes about 10 minutes at the most
[15:29:07] <KitsuWhooa> Anyway, yeah, this looks to be SSE (not SSE2), so I can't really help there
[15:30:06] <ConSiGno> https://pkgsrc.se this might be of help?
[15:30:08] <phrik> Title: pkgsrc.se | The NetBSD package collection (at pkgsrc.se)
[15:31:07] <KitsuWhooa> oh abaumann: if you haven't seen it yet, this is all broken https://www.archlinux32.org
[15:31:42] <KitsuWhooa> ConSiGno: I doubt it. The package uses cmake
[15:32:19] <KitsuWhooa> that said
[15:32:36] <KitsuWhooa> cmake -DSSEMATH=OFF -DMMX=OFF -D3DNOW=OFF -DSSE=OFF -DSSE2=OFF -DSSE3=OFF
[15:34:12] <KitsuWhooa> looks like they renamed them https://github.com
[15:34:51] <KitsuWhooa> 2 years ago https://github.com
[15:35:15] <KitsuWhooa> well this should be easy to fix :p
[15:35:23] <ConSiGno> yay! \o/
[15:37:15] <KitsuWhooa> want to try building it yourself?
[15:37:20] <KitsuWhooa> I have a patch
[15:38:29] <KitsuWhooa> Otherwise I can send you a package to test
[15:41:30] <KitsuWhooa> actually, idk if a build on i686 will work on i486
[15:54:56] <KitsuWhooa> looks like this is recent too. The SDL that's on my laptop works fine with zdoom
[15:55:11] <KitsuWhooa> so I'll update first, make sure it's broken for me (due to SSE2) then try my patched version, and if it works, submit the patch
[15:55:34] <ConSiGno> hmm... okay :D
[15:55:49] <ConSiGno> I have no idea how to build from source on the arch package manager
[15:56:15] <KitsuWhooa> typically you clone the repo, cd to the package and run makepkg, but on arch32 it's complicated
[15:57:36] <KitsuWhooa> https://tasossah.com if you feel like giving it a try
[15:57:42] <KitsuWhooa> 7d004080f1b7020380460ace749a2812fc11d16bf1c3b046d30dbd1006425319
[16:02:07] <KitsuWhooa> Not sure. zdoom runs even with updated SDL2
[16:02:17] <KitsuWhooa> it is extremely slow though
[16:06:42] <KitsuWhooa> this is what happens when there's no video acceleration I guess
[16:13:26] <KitsuWhooa> Anyway, I don't have an i486 at the moment, and I can't reproduce it on i686 (even though it should be broken there too), so I guess please test the package
[16:36:08] <ConSiGno> i'm rebuilding chocolate-doom now, will let you know :)
[16:36:17] <KitsuWhooa> you don't need to rebuild it
[16:36:24] <KitsuWhooa> at least, I don't think you do
[16:36:43] <KitsuWhooa> it'd be weird to have asm in headers
[16:40:30] <ConSiGno> i still got an illegal instruction so i'm recompiling choco just to make sure it's not choco this time
[16:41:02] <KitsuWhooa> ah...
[16:41:31] <ConSiGno> remember it stops at any sigill, so if there's another sigill that's showing up in choco, I will need to report that one to fraggle
[16:41:47] <KitsuWhooa> if you have gdb available, you can quickly check it
[16:41:58] <ConSiGno> yeah oops
[16:41:59] <KitsuWhooa> (idk if it works on i486)
[16:42:00] <ConSiGno> let me do that haha
[16:42:10] -!- abaumann has joined #archlinux32
[16:42:10] <buildmaster> Hi abaumann!
[16:42:10] <buildmaster> !rq abaumann
[16:42:11] <phrik> buildmaster: <abaumann> if analog tv would have been engineered nowadays with all this software nonsense, it would explode every second weekend..
[16:42:23] <KitsuWhooa> ehem dvb-t
[16:42:28] <KitsuWhooa> also hi :p
[16:42:36] <girls> Hi abaumann!
[16:42:51] <KitsuWhooa> :p
[16:42:57] <abaumann> hi there.
[16:43:00] <abaumann> https://git.archlinux32.org
[16:43:02] <phrik> Title: PKGBUILD « sdl2 « extra - packages - Archlinux32 package modifications (at git.archlinux32.org)
[16:43:03] <girls> and hi all others, I first have to check from the logs who is online/active ;)
[16:43:06] <abaumann> this actually looks correct.
[16:43:17] <KitsuWhooa> abaumann: it's not
[16:43:33] <KitsuWhooa> https://tasossah.com
[16:43:52] <abaumann> oh.
[16:43:55] <abaumann> sorry. I misread.
[16:44:02] <abaumann> Yes. this is easy to fix.
[16:44:15] <KitsuWhooa> feel free to apply the patch if you can test it
[16:44:21] <KitsuWhooa> cmake definitely shows the correct flags enabled/disabled
[16:44:32] <KitsuWhooa> Also, I'd suggest getting work done on storing successful build logs
[16:44:47] <KitsuWhooa> it'd be easy to add a step that greps for things like "unrecognised flag" or whatever the cmake warning is
[16:44:57] <KitsuWhooa> and would let us know when projects change flags around
[16:45:07] <abaumann> https://buildmaster.archlinux32.org
[16:45:08] <ConSiGno> ENDBR32 is showing up in gdb on the package you gave me, kitsu
[16:45:08] <phrik> Title: Index of /build-logs/success (at buildmaster.archlinux32.org)
[16:45:18] <KitsuWhooa> oh ffs
[16:45:25] <KitsuWhooa> yeah sorry I forgot I can't build for i486
[16:45:29] <abaumann> well, blame cmake to run if you specify illegal flags.. :->
[16:45:59] <KitsuWhooa> abaumann: yeah, those logs aren't helpful :p
[16:46:01] <KitsuWhooa> that's just namcap
[16:46:12] <abaumann> uh.
[16:46:20] <KitsuWhooa> https://buildmaster.archlinux32.org
[16:46:28] <girls> should we conserver the log of the build() step, even if that succeeds?
[16:46:46] <KitsuWhooa> in this case, it's whenever cmake first runs
[16:46:50] <KitsuWhooa> but I don't see why not honestly
[16:46:54] <KitsuWhooa> it shouldn't take up much space
[16:47:02] <KitsuWhooa> cmake runs in prepare() I think
[16:47:04] <girls> I'll look into it
[16:47:57] <KitsuWhooa> ConSiGno: anyway, sorry, I can't build for i486. I can send you the PKGBUILD I used if you want
[16:48:27] <KitsuWhooa> should require you to just download it and run `makepkg -s` in that directory
[16:48:47] <KitsuWhooa> girls: appreciated
[16:48:51] <KitsuWhooa> also I'm wrong
[16:48:55] <KitsuWhooa> https://github.com
[16:48:57] <KitsuWhooa> it's in build
[16:48:59] <KitsuWhooa> so yeah
[16:49:53] <KitsuWhooa> Also, I converted the arch32 logo into an svg that should be ready to go to put on the website, because every time I zoom in the existing one gets blurry and it annoys me
[16:49:55] <abaumann> asp32 export extra/sdl2, then ( CARCH=pentium4; . PKGBUILD; declare -f build)
[16:49:56] <ConSiGno> I am forcing an upgrade (downgrade in pacman's case) now
[16:50:01] <KitsuWhooa> feel free to use it https://tasossah.com
[16:50:07] <abaumann> respectively CARCH=i686 and CARCH-i486
[16:50:27] <ConSiGno> abaumann, I'm using a Pentium MMX 200MHz for this distro, I am happy to report any issues I find :)
[16:50:29] <girls> KitsuWhooa: thx!
[16:50:33] <abaumann> you can also build then with 'makepkg' or the devtools 'staging-i486-build'
[16:50:52] <KitsuWhooa> abaumann: idk if that's directed at me, but idk how that will bypass the compiler emitting endbr32
[16:51:15] <KitsuWhooa> I really should install an i486 VM...
[16:51:15] <ConSiGno> is that a compiler bug?
[16:51:16] <abaumann> the compiler should have cet disabled
[16:51:28] <abaumann> you need patches in binutils and gcc
[16:51:32] <KitsuWhooa> ConSiGno: yes
[16:51:38] <KitsuWhooa> yeah, I don't have those patches as I built it in i686
[16:52:32] <abaumann> that's why you build with the devtools, there is a makepkg-i486.conf which sets flags correctly..
[16:52:43] <KitsuWhooa> ah I see
[16:52:48] <KitsuWhooa> I really need to look into it
[16:53:03] <abaumann> ..https://git.archlinux32.org/packages/tree/core/binutils/PKGBUILD and https://git.archlinux32.org
[16:53:04] <phrik> Title: PKGBUILD « binutils « core - packages - Archlinux32 package modifications (at git.archlinux32.org)
[16:53:33] <abaumann> but it can easily happen, that the toolchain is rebuild for i486 without the seds having had any effect.
[16:53:40] <ConSiGno> ah noooo I'm getting movss instructions in here now
[16:53:40] <abaumann> and then you end up with endbr32.
[16:54:02] <abaumann> why setting -march=i486 enabled endbr32 in gcc is beyond me, this should be considered a compiler bug
[16:54:16] <abaumann> ConSiGno: welcome :-)
[16:54:21] <KitsuWhooa> it looks like it built with -march=i686
[16:54:27] <abaumann> quite a nightmare, I know.
[16:54:30] <KitsuWhooa> either way, https://tasossah.com the flags are set correctly
[16:54:31] <phrik> Title: sdl2 flags (at tasossah.com)
[16:54:49] <KitsuWhooa> wait what
[16:54:51] <KitsuWhooa> > -- SDL_SSE (Wanted: OFF): ON
[16:55:04] <KitsuWhooa> I'm confused
[16:56:07] <ConSiGno> I will be putting in a 6x86 in this box soon but I will still run into SSE issues - at least it won't be as bad
[16:56:30] <ConSiGno> I still need to build a kernel I think, I'm using the stock one that came on the disc
[16:56:46] <KitsuWhooa> you're most likely not using the one that came with the disc
[16:57:30] <KitsuWhooa> ...cmake caches the build flags
[16:57:33] <girls> KitsuWhooa: either I broke the website, or it will have your svg - when it next pulls its updates from git (in the next few minutes, probably)
[16:57:38] <KitsuWhooa> hah
[16:57:43] <KitsuWhooa> thanks!
[16:57:47] <KitsuWhooa> I tested it with the dev tools and it was fine
[16:57:54] <girls> perfect
[16:58:04] <girls> then I just need to not have made any typo ;)
[16:58:06] <KitsuWhooa> I explicitly messed with the viewport to get it to work out of the box
[16:58:11] <KitsuWhooa> otherwise it overflowed
[16:58:53] <KitsuWhooa> Anyway, after cleaning the build and re-running makepkg, it built without SSE but with march=i686
[17:02:05] <ConSiGno> I honestly think the other issue I have with my FX5200 is, sadly, lack of power from the motherboard. I think nouveau is requesting the card to go into the mode it's supposed to, but it can't get enough power to run anything besides basic VGA framebuffer stuff, and even that is glitchy. if only my mach64 card were supported I could throw my top of the line RageXL in here :P
[17:02:20] <KitsuWhooa> the i486 iso doesn't seem to have working networking
[17:02:38] <abaumann> not automatic networking. no
[17:02:38] <ConSiGno> you have to use dhclient I think?
[17:02:44] <abaumann> dhclient should be on it.
[17:02:47] <abaumann> not all the wifi stuff
[17:02:57] <ConSiGno> yeah I'll never put this on Wifi :O
[17:03:00] <KitsuWhooa> I shouldn't need dhclient for ipv6, right?
[17:03:13] <KitsuWhooa> I just want it to get a v6 address
[17:03:54] <abaumann> oeh. I really never tested the ISO with IPv6, really. *blush*
[17:04:12] <girls> i486 is not compatible with ipv6 </troll>
[17:04:16] <KitsuWhooa> :p
[17:04:19] <KitsuWhooa> Oh I see
[17:04:19] <abaumann> :-)
[17:04:20] <KitsuWhooa> I'm dumb
[17:04:23] <KitsuWhooa> the interface wasn't up
[17:04:53] <girls> say the one running i486 arch32 on their router ... which ran a 4to6 tunnel (to ukraine ^^) until they got native ipv6 :D
[17:04:56] <abaumann> like me today: why isn't the old Apple ][ not booting after a PSU change.. oh wait.. I might have to reinsert the boot ROMS..
[17:05:00] <abaumann> ..old technology :-)
[17:05:07] <ConSiGno> I had ipv6 working on netbsd, but netbsd's package management was making me pull my hair out
[17:05:49] <KitsuWhooa> https://tasossah.com yeah it works
[17:06:24] <abaumann> \o/
[17:07:41] <ConSiGno> https://archlinux.org this makes me so sad. I hate that these were all pulled out
[17:07:58] <KitsuWhooa> can you not build it manually?
[17:08:00] <KitsuWhooa> well
[17:08:05] <KitsuWhooa> new kernel dropped dri support for these old cards
[17:08:09] <KitsuWhooa> but also, I don't think mesa even works
[17:08:16] <abaumann> https://aur.archlinux.org
[17:08:17] <ConSiGno> it does
[17:08:20] <abaumann> ach64-dri
[17:08:25] <ConSiGno> it's just slow as a glacier
[17:08:28] <KitsuWhooa> it doesn't on my laptop :p
[17:08:35] <abaumann> there is an orphaned one in the AUR, but when I tried it, it didn't compile for me
[17:08:37] <ConSiGno> interesting...
[17:08:46] <KitsuWhooa> I definitely had proper hw acceleration working
[17:08:48] <abaumann> its DRI1 anyway, so I think that disappeared from Linux also?
[17:08:53] <KitsuWhooa> yes
[17:08:57] <KitsuWhooa> they nuked dri1 from the kernel
[17:09:03] <abaumann> sad, ok.
[17:09:08] <KitsuWhooa> Eventually I want to try to build the savage one using dkms
[17:09:13] <ConSiGno> yeah it was a "security issues"
[17:09:24] <ConSiGno> and no one converted the drivers to dri2
[17:09:31] <ConSiGno> I think rage128 was the only one that got converted
[17:09:38] <ConSiGno> but I don't have a Rage128 so, what good that does :)
[17:09:46] <abaumann> ok, sdl2 seems ok, at least it runs on my alix..
[17:09:52] <KitsuWhooa> abaumann: with my patch?
[17:09:57] <abaumann> yep.
[17:10:01] <KitsuWhooa> ConSiGno: I'd convert them if I knew how
[17:10:03] <KitsuWhooa> abaumann: cool
[17:10:04] <abaumann> but the alix has mmx and sse.
[17:10:07] <KitsuWhooa> oh
[17:10:11] <KitsuWhooa> yeah it works on i686
[17:10:14] <KitsuWhooa> I tested that already
[17:10:16] <KitsuWhooa> :p
[17:10:22] <KitsuWhooa> well, sse
[17:10:27] <abaumann> ah, only mmx
[17:10:29] <abaumann> flags : fpu de pse tsc msr cx8 sep pge cmov clflush mmx mmxext 3dnowext 3dnow cpuid 3dnowprefetch vmmcall
[17:10:41] <KitsuWhooa> then it works :p
[17:10:43] <ConSiGno> (if there was a 586-mmx build of arch32 I would be all over that lol)
[17:10:52] <abaumann> my real i486 is a little bit - aeh - slow for testing -
[17:10:56] <abaumann> and for booting Archlinux32
[17:11:14] <abaumann> more sub-architectures :-)
[17:11:43] <abaumann> usually, it's quite enough to compile the specific package and/or glibc just to use the optimized code
[17:12:44] <ConSiGno> I would love to see if I could get this FX5200 working, the card works but nouveau hates it, and I can't really debug that since the screen goes bonkers
[17:12:57] <ConSiGno> I know it works as it worked in Win98 >_>
[17:12:58] <abaumann> lemme see. last time I had a hard time runnig Arch32 on my unisys 486: the network card is a NE1000 ISA, the kernel was supporting it, but it OOPSed somewhere when adding a network interface.
[17:13:26] <abaumann> mmh. nope. nuked the sd card..
[17:13:36] <abaumann> I'll test also in a 486 VM..
[17:13:41] <ConSiGno> yeah I can't imagine running on a stock 486 now, I remember doing that in high school. I would set stuff to compile, go to school, and it would be halfway done when I came back.
[17:13:44] <KitsuWhooa> https://tasossah.com
[17:13:56] <abaumann> LOL
[17:14:11] <KitsuWhooa> old, but I had working hw acceleration with mesa
[17:14:19] <abaumann> oh. nice.
[17:14:23] <KitsuWhooa> I could play openttd and doom and other things, albeit low res
[17:14:26] <ConSiGno> I used Gentoo and I had a dedicated HDD just for swap
[17:14:38] <KitsuWhooa> now my existing zdoom build (from back when it worked) runs at 0.5FPS
[17:14:39] <abaumann> openttd doesn't make much use of hw acceleration, does it?
[17:14:43] <KitsuWhooa> it renders a new frame every 2 seconds
[17:14:49] <KitsuWhooa> abaumann: just updating the screen was faster
[17:15:17] <abaumann> I (buildmaster.archlinux32.org) could not complete a mysql query:
[17:15:22] <KitsuWhooa> it went from "usable" to completely unsable
[17:15:24] <abaumann> ok, SDL2 is stuck again..
[17:15:26] <ConSiGno> @kitsuWhooa honestly SDL1, as old as it is, seems to be a LOT faster due to still having some old X11 stuff that it uses that SDL2 seems to have dropped in favour of grabbing a GLX context and running with it
[17:15:41] <ConSiGno> if you don't have GLX, you get the slowest thing imaginable
[17:15:45] <KitsuWhooa> I tried a game that uses sdl1 and that was also unusable
[17:15:50] <KitsuWhooa> (was that OTTD?)
[17:15:56] <KitsuWhooa> just an old build I had lying around of something
[17:16:04] <abaumann> openttd has some startup flags for bbp and color..
[17:16:12] <abaumann> ..greatly changing speed. IIRC
[17:16:25] <KitsuWhooa> sure, but I'm just saying, the exact same build I last played on that laptop with the last used config worked previously
[17:16:27] <KitsuWhooa> and now it's unusable
[17:16:40] <ConSiGno> weird... I got 35FPS (max fps) on xdoom with my rageXL but that was on NetBSD, which, like I said, is great until you run into things like the memory manager being a pile of crap
[17:16:42] <KitsuWhooa> I tried zdoom at what, 320x240?
[17:16:56] <KitsuWhooa> kernel dropped DRI, so...
[17:17:18] <ConSiGno> NetBSD still has it since it's not linux, and someone hacked that in because they're insane
[17:17:47] <ConSiGno> netBSD doesn't have the security flaw that caused DRI to be removed
[17:18:03] <KitsuWhooa> I don't think it was the security flaw that caused them to be dropped
[17:18:15] <KitsuWhooa> it's what caused them to be disabled by default
[17:18:24] <ConSiGno> Well, that's what I was told from the smiling guy on phoronix dot com!!! that always did sound liek an excuse more than a legit thing to be concerned about.
[17:18:40] <ConSiGno> that guy revels in old hardware being blammed from the kernel, I have no idea why
[17:18:47] <KitsuWhooa> It's more old code on the chopping block because "it feels good to remove old stuff"
[17:19:13] <ConSiGno> I hate that so much augh
[17:19:13] <abaumann> phoronix is the "we celebrate deleting old things" platform
[17:19:14] <KitsuWhooa> I doubt there's much maintenance overhead for stuff like this
[17:19:34] <abaumann> it still claims linux 6.2 doesn't work on i486, yet it builds and runs just fine.
[17:19:37] <KitsuWhooa> I remember the comments on the phoronix article about my fix for the floppy driver
[17:19:40] <ConSiGno> @abaumann exactly, it's wild, but they somehow have so much sway
[17:19:48] <KitsuWhooa> why would anyone still use floppies in 2022?!
[17:20:01] <abaumann> ah, yes. util-linux disabled floppy support per default, you just have to ./configure it, to get it.
[17:20:07] <abaumann> this is then "floppy support was removed"
[17:20:25] <ConSiGno> I use a floppy to boot into arch32 so uh, heh :P
[17:20:32] <abaumann> :-)
[17:20:38] <KitsuWhooa> Meanwhile, I still want to get floppies to work with modern AMD IOMMUs...
[17:21:59] <ConSiGno> My BIOS has an issue with booting in that it doesn't allow things to get "ready" before giving up and thinking there is no OS installed
[17:22:06] <KitsuWhooa> Question regarding i486: It fails with keyring is not writable. Shouldn't it be disabled?
[17:22:10] <ConSiGno> so I use a boot floppy
[17:22:30] <ConSiGno> yeah I disabled that eventually, I had to fall into a lot of holes before I got the kernel installed
[17:22:45] <ConSiGno> mostly becuase I didn't know arch at all, and on top of that, running an esoteric port of it
[17:22:52] <ConSiGno> so I had double the rakes to step on
[17:23:05] <KitsuWhooa> I used to boot my Pentium 3 router using a CD, because it didn't support USB boot (OS was in a USB)
[17:23:31] <KitsuWhooa> until the optical drive died and it was easier to just put an internal SSD with an adapter
[17:23:45] <KitsuWhooa> I tried using floppies but I couldn't fit a kernel + kexec
[17:23:54] <abaumann> ah.
[17:23:55] <KitsuWhooa> Anyway
[17:24:03] <KitsuWhooa> > Shouldn't it be disabled?
[17:24:04] <abaumann> https://git.archlinux32.org
[17:24:05] <phrik> Title: floppy - archi486 - Archlinux32 i486 tools (at git.archlinux32.org)
[17:24:06] <KitsuWhooa> I meant, shouldn't it be disabled by default?
[17:24:08] <ConSiGno> now I'm to the point I reached in NetBSD where things just aren't compiled correctly due to some psycho (me) coming in and finding the hard way how many things are misconfigured and downloading an irc client hoping for answers :V
[17:24:19] <abaumann> the USTAR idea from NetBSD "stolen" to boot a Linux kernel.
[17:24:25] <abaumann> But it needs much more work
[17:24:30] <KitsuWhooa> cool
[17:24:47] <abaumann> I wanted to code something in unreal mode :-)
[17:25:05] <KitsuWhooa> I see the arch32 logo is cut off in git.archlinux32.org :
[17:25:07] <KitsuWhooa> * :p
[17:25:20] <KitsuWhooa> (and it's not even my svg)
[17:25:21] * girls whistles inocently
[17:25:22] <abaumann> uh. right. :-)
[17:25:31] * abaumann also whistles
[17:25:45] <abaumann> could have been anybody of us
[17:26:18] <abaumann> this is some css thingi in cgitc or so.
[17:26:21] -!- deep42thought has joined #archlinux32
[17:26:22] <buildmaster> Hi deep42thought!
[17:26:22] <buildmaster> !rq deep42thought
[17:26:23] <phrik> buildmaster: * deep42thought considers locking away abaumann's keyboard until noon
[17:26:24] <ConSiGno> also while I'm here, I know the maintainer of wine in AUR and he's never gonna care about me asking to get those scripts running, so if anyone could get wine-stable working on i486/i586 and throw me a file, I was gonna compile the whole thing myself but it's not 2002 anymore so there's a lot more to compile and sift through :|
[17:26:32] * deep42thought can also whistle
[17:26:52] <ConSiGno> I hate it when a maintainer is impossible to contact because he has prior angry interactions
[17:27:04] <ConSiGno> maybe I need an alt lmao
[17:27:06] <KitsuWhooa> ConSiGno: you can look at what is in the packages git for arch32
[17:27:09] <KitsuWhooa> I know wine works in i686
[17:27:19] -!- deep42thought has parted #archlinux32
[17:27:38] <ConSiGno> yeah there's a i686 port but no 486 port - but it does work in 486 as I compiled it for NetBSD (with an ancient compiler unfortunately)
[17:27:46] <ConSiGno> I had chips challenge running at the most
[17:27:48] <KitsuWhooa> Change the width to width: 230px !important; in #archnavbarlogo I guess
[17:28:04] <KitsuWhooa> that fixes it :p
[17:30:00] <girls> now at least https://archlinux32.org should have it
[17:30:01] <phrik> Title: Arch Linux 32 (at archlinux32.org)
[17:30:09] <ConSiGno> also I had a chat with a pcem dev, she put in the work to get i486/Pentium/6x86 support working so it's _possible_ to check stuff without having to pull out a dusty box
[17:30:10] <KitsuWhooa> yup
[17:30:12] <KitsuWhooa> looks good :p
[17:30:23] <girls> note: do not git pull as root - it creates root-owned files in .git, which the regular cronjob cannot edit/replace
[17:30:32] <KitsuWhooa> ConSiGno: does it actually detect instructions properly?
[17:30:41] <ConSiGno> yep
[17:30:45] <KitsuWhooa> ConSiGno: I tried using qemu-i386 but it straight up doesn't care about SSE2 on pentium3
[17:30:57] <ConSiGno> yeah that's why I switching to pcem
[17:31:01] <KitsuWhooa> cool
[17:31:02] <KitsuWhooa> I didn't know that
[17:31:11] <girls> bbs, bugs, git are another construction site ;)
[17:33:20] <abaumann> (damn: the second power supply now failing, also my AMD Duron is constantly rebooting)
[17:33:40] <KitsuWhooa> time for a recap? :p
[17:33:51] <abaumann> I already rebuilt the power-supply of the Apple ][ clone, so I'm skilled now (and not electocuted myself, apparently)
[17:33:54] <ConSiGno> oh yeah I got an AT Switching PSU for my box
[17:34:10] <ConSiGno> it's so quiet, the entire PC is quiet now
[17:34:14] <abaumann> recap? well, those things look so alien to me: huge caps and small transformers *shudder*
[17:34:39] <abaumann> the 400V caps still held power..
[17:34:44] <abaumann> ..quite a scary monster.
[17:34:57] <KitsuWhooa> Yeah, you usually press the power button to discharge them while unplugged, and then short them out with a resistor
[17:34:59] <KitsuWhooa> to be 100% sure
[17:35:00] <abaumann> and the humming and hissing sound I didn't like. And I thing a X-cap also blew.
[17:35:09] <abaumann> I used a light bulp. :-)
[17:35:12] <KitsuWhooa> rifa? :p
[17:35:19] <ConSiGno> oh the ol light bulb trick
[17:35:20] <abaumann> yeah, by the smell :-)
[17:35:26] <KitsuWhooa> gotta love the rifas
[17:35:33] <ConSiGno> gotta get the tubes in the 486 to warm up huh
[17:35:34] <abaumann> he.. I'm a _software_ engineer.
[17:35:40] <ConSiGno> Rifa caps are the worst
[17:35:45] <abaumann> screwdrivers usually jump away, when they see me.. ;-)
[17:35:52] <ConSiGno> I had a technics turntable pop at an event, rifa cat
[17:35:54] <ConSiGno> cap*
[17:36:01] <ConSiGno> they smell AWFUL.
[17:36:05] <abaumann> indeed
[17:36:45] <abaumann> flags : fpu vme de pse tsc msr cx8 pge mmx syscall 3dnow k6_mtrr cpuid 3dnowprefetch vmmcall
[17:36:48] <abaumann> model name : AMD-K6(tm) 3D processor
[17:36:52] <abaumann> yea, let's see.
[17:36:54] <ConSiGno> I'm tempted to recap my mobo and go a little above spec to see if I can get the FX5200 to be stable, it's a PCI card so you have to put them on the _LAST_ slot to get the most power, not the first!
[17:37:36] <KitsuWhooa> When I was trying to get my old nvidia cards working with nouveau, I had loads of power problems and it was AGP
[17:37:37] <abaumann> urgh. this sounds like driving things at the limit. :-)
[17:37:48] <ConSiGno> yeah it's usually caps, it's always caps
[17:37:50] <KitsuWhooa> A Radeon from a similar era worked fine, but the screen flickers depending on load...
[17:37:57] <ConSiGno> remember this is when cap plague was starting
[17:38:03] <KitsuWhooa> yeah but those just bulge up
[17:38:25] <abaumann> electrolitic are quite harmless in that regard.
[17:38:30] <abaumann> they usually just bulp up.
[17:38:35] <KitsuWhooa> ...not when they leak all over the place
[17:38:40] <KitsuWhooa> and corrode everything
[17:38:43] <abaumann> but they can run outside of specs. causing all kind of strange effect.
[17:38:46] <abaumann> *effects
[17:38:50] <ConSiGno> I wish I were better at soldering so I could replace all the caps on this board and the card
[17:39:00] <abaumann> I used sand-paper and Deoxit on old 74x ICs today.
[17:39:04] <abaumann> works quite well. :-)
[17:39:04] <KitsuWhooa> you definitely need experience to recap motherboards
[17:39:11] <KitsuWhooa> there are so many layers, it's difficult
[17:39:30] <KitsuWhooa> that said, I think my pinecil will probably make it easier
[17:39:33] <ConSiGno> I do know the VX97 mobos have under-spec caps
[17:39:37] <KitsuWhooa> (amazing soldering iron btw)
[17:39:37] <abaumann> yes. all modern (sau mid 90is) PCs are already too small to solder (at least for me)
[17:39:44] <abaumann> 286 is my leage..
[17:39:56] <KitsuWhooa> I think the newest board I replaced caps in was an AM2 board
[17:40:25] <abaumann> The PSU had 2 layers, so has the motherboard on my apple ][ clone, this makes soldering really easy.
[17:40:30] <abaumann> that would be 8086 or so?
[17:40:36] <abaumann> in PC-time..
[17:41:03] <KitsuWhooa> Where do I suggest changes for the message that's printed when you boot an i486 iso? :p
[17:41:11] <KitsuWhooa> pacstrap -c doesn't work
[17:41:30] <KitsuWhooa> and SigLevel = Never is required regardless of RAM as it's read only
[17:41:40] <abaumann> aeh, here, or a bug report or so.. why? :-)
[17:41:48] <KitsuWhooa> should be pacstrap -C
[17:41:57] <KitsuWhooa> and ideally SigLevel = Never is the default
[17:42:01] <abaumann> is it. oh.
[17:42:11] <abaumann> mmh. so memory usage increased.
[17:42:15] <KitsuWhooa> no
[17:42:23] <KitsuWhooa> it doesn't work because the keyring is read only
[17:42:34] <abaumann> ah. ok.
[17:42:48] <abaumann> Mmh. maybe having a writeable overlay for the ISO would be an idea then..
[17:42:51] <KitsuWhooa> https://tasossah.com
[17:42:57] <KitsuWhooa> I think that'd increase ram usage
[17:43:10] <abaumann> ..otherwise too many workarounds are necessary (you can put the keyring also to /mnt and confiure /etc/pacman.conf accordingly)
[17:43:31] <abaumann> do you have swap space?
[17:43:39] <KitsuWhooa> This is a VM with tons of RAM
[17:43:43] <abaumann> ah,.
[17:43:44] <KitsuWhooa> I'm talking about a case of someone installing on an actual machine
[17:43:58] <abaumann> right. we established that already. sorry. :-)
[17:44:00] <KitsuWhooa> the base "should just work" in a VM
[17:44:06] <KitsuWhooa> but as it is, it doesn't
[17:44:12] <abaumann> i486 is alpha
[17:44:16] <KitsuWhooa> of course
[17:44:19] <KitsuWhooa> that's why I'm giving feedback
[17:44:20] <KitsuWhooa> :p
[17:44:21] <abaumann> I'm also pretty sure, tons of stuff is missing on the ISO.
[17:44:29] <abaumann> and, I'm deviating heavily from upstream.
[17:45:04] <abaumann> but, hey, thanks all for testing i486 :-)
[17:45:24] <abaumann> there are not many "crazy" people around anymore trying that.. ;-)
[17:45:43] <KitsuWhooa> Yeah, I wouldn't really do it either :p
[17:46:03] <KitsuWhooa> Pentium 3 is already a chore
[17:46:17] <girls> abaumann: I also have to disappoint you, my next network setup will not route through the ALIX
[17:46:33] <abaumann> oh no! ;-)
[17:46:43] <abaumann> you go all MIPS now with WRT54? ;-)
[17:46:54] <KitsuWhooa> lmao
[17:46:57] <girls> I can put an ALIX in front of my collocation space - so only *your* connections will suffer from i486 issues :D
[17:47:14] <abaumann> that's fair :-)
[17:47:23] <abaumann> I'm using OpenBSD on the Alix, so.. :->
[17:50:48] <KitsuWhooa> Hmmm
[17:50:59] <KitsuWhooa> After chrooting I'm getting "package architecture is not valid" when trying to install anything
[17:51:13] <abaumann> pacman guess stuff failing?
[17:51:19] <KitsuWhooa> Oh right
[17:51:22] <KitsuWhooa> "Architecture = auto"
[17:51:28] <KitsuWhooa> that seems broken
[17:51:46] <KitsuWhooa> yeah, setting that to i486 fixes it
[17:51:50] <abaumann> not really broken, just very hard to get right for all Intel/AMD/Cyrix/whatever CPUs ;-)
[17:52:03] <abaumann> it constantly miscalculates
[17:52:22] <KitsuWhooa> guess it doesn't know what to do with a ryzen =p
[17:52:25] <girls> actually, gcc accepted our patch, so i686 vs pentium4 should be fine now on non-amd and non-intel cpus :)
[17:52:42] <abaumann> oh, that's nice :-)
[17:53:03] <girls> but I'm reluctant to remove our hack and switch to the gcc-native version
[17:53:34] <abaumann> setting the Architecure by hand is not a biggie..
[17:53:39] <girls> yeah
[17:54:07] <girls> btw: the build-slaves (with recent git master) will now upload logs on successful build, now only the build-master needs to learn, how to save them
[17:54:08] <KitsuWhooa> fair
[17:54:13] <KitsuWhooa> cool!
[17:54:25] <abaumann> yes, cool.
[17:54:39] <abaumann> We have plenty of space on the buildmaster now, so no problem.
[17:54:57] <girls> do you use youtube as storage backend now?
[17:54:59] <KitsuWhooa> I don't expect them to take up much space when compressed though
[17:55:01] <KitsuWhooa> pfft
[17:55:23] <abaumann> actually, no, much easier, I got a third disk :-)
[17:55:32] -!- GNUtoo has quit [Ping timeout: 255 seconds]
[17:55:36] <ConSiGno> I'm hoping to be of much help while I can :)
[17:55:36] <girls> youtube would be cheaper
[17:56:07] <abaumann> probably :-)
[17:58:43] <abaumann> that's nice behaviour, using and crypto routing (ssh, gpg check in pacman) just reboots the machine (because the PSU cannot deliver enough power)
[17:58:53] <abaumann> *routines
[18:00:10] <abaumann> Indeed, setting SigLevel=Never works..
[18:02:30] <KitsuWhooa> `reboot` is broken, too
[18:02:52] <abaumann> ah.. naeh.. who wants to do that anyway ;-)
[18:02:53] <abaumann> really?
[18:03:06] <KitsuWhooa> times out activating org.freedesktop.login1
[18:03:07] <abaumann> ah, maybe systemd wants to write stuff during reboot?
[18:03:12] <KitsuWhooa> Oh, it works
[18:03:13] <abaumann> *sigh*
[18:03:14] <KitsuWhooa> just takes ages
[18:03:17] <abaumann> ah.
[18:03:53] <abaumann> yeah, I tried once systemd on a i486 with Arch32, every other init system is beating the crap out of it. in terms of performance
[18:04:19] <abaumann> I think, systemd creates a shutdown ramdisk during reboot/shutdown..
[18:04:27] <abaumann> ..it might compress it :->
[18:04:29] <KitsuWhooa> it was definitely trying to do something and failing
[18:04:34] <KitsuWhooa> and maybe it fell back into something else
[18:04:41] <KitsuWhooa> perhaps some service isn't running
[18:04:48] <abaumann> quite likely
[18:04:59] <KitsuWhooa> dbus activation failing sounds like whatever it wanted to start wouldn't start
[18:05:00] <abaumann> systemd on the i486 might not be the best idea either..
[18:05:14] <abaumann> I blame it all to the read-only nature of the i486 ISO.
[18:05:20] <KitsuWhooa> maybe
[18:05:24] <abaumann> Pretty sure, if there is an overlay, it would work.
[18:05:38] <abaumann> But I have to read on ISO9660 and overlay file systems first..
[18:06:01] <abaumann> that's some - aeh - 20 years ago - I might have done something like that.. :-)
[18:06:26] <girls> do you mind, if the success logs don't contain the architecture in the name?
[18:06:30] <KitsuWhooa> ah yeah, systemd-logind fails to start
[18:06:38] <girls> otherwise I would need to add another db query to return-assignment
[18:07:10] <girls> I could heuristically just take the architecture from any of the built packages
[18:07:24] <KitsuWhooa> would there be any other way to identify them?
[18:07:36] <abaumann> the build machine, at least for i486?
[18:07:46] <girls> time stamp :D
[18:08:12] <KitsuWhooa> abaumann: systemd-logind fails to start through systemd, but running the binary manually works...
[18:08:17] <girls> I'll take the least-any'ish architecture from binary packages, that should be a good-enough pointer
[18:08:26] <abaumann> KitsuWhooa: interesting..
[18:09:26] <KitsuWhooa> wonder if it's the sandboxing that breaks it
[18:09:57] <KitsuWhooa> but yeah, if I start it manually and then "reboot" it works instantly
[18:09:58] <abaumann> some systemd parts are alergic to seccomp things IIRC
[18:10:49] <KitsuWhooa> https://bugzilla.redhat.com
[18:10:52] <phrik> Title: 1737217 – systemd-logind is unable to start terminals after latest updates. (at bugzilla.redhat.com)
[18:11:05] <KitsuWhooa> idk if related, but it's the same error code
[18:11:55] <abaumann> selinux should not be enabled per default..
[18:14:02] <KitsuWhooa> CONFIG_SECURITY_SELINUX=y
[18:14:35] <abaumann> urgh
[18:15:16] <abaumann> but the filesystem not labelled and we don't have policy rules or so?
[18:15:19] <KitsuWhooa> idk how to check if it's enabled
[18:15:22] <KitsuWhooa> there's no getenforce
[18:15:25] <abaumann> so it should just sit there and do nothing.
[18:15:57] <KitsuWhooa> or sestatus
[18:16:21] <abaumann> I don't think upstream either enabled selinux per default or bundles the selinux tools
[18:16:25] <KitsuWhooa> yeah
[18:16:28] <KitsuWhooa> I think it's unrelated
[18:19:39] <abaumann> CONFIG_SECURITY_SELINUX=y
[18:19:39] <abaumann> CONFIG_SECURITY_SELINUX_BOOTPARAM=y
[18:19:47] <abaumann> this is upstream's config
[18:19:51] <KitsuWhooa> yeah, fair
[18:19:56] <abaumann> so, it can be disabled on startup
[18:20:05] <KitsuWhooa> which I don't know how to do on the i486 iso
[18:20:07] <KitsuWhooa> :p
[18:20:22] <abaumann> press TAB when the syslinux thingy comes up
[18:20:34] <abaumann> then you have an 'arch32' target which you can modifyg
[18:20:50] <abaumann> would also need some -- aeh, how do you call that? -- documentation :-)
[18:22:39] <KitsuWhooa> yeah, okay, selinux=0 didn't do anything
[18:26:41] <abaumann> so, I pushed to official SDL2 to stable with all correct patching.
[18:26:52] <KitsuWhooa> thanks!
[18:26:54] <abaumann> at least on my oldest running machine with just MMX it seems to work fine.
[18:26:55] <abaumann> np
[18:26:56] <KitsuWhooa> ConSiGno: ^
[18:28:34] <abaumann> thanks for finding the fix :-)
[18:28:39] <KitsuWhooa> no worries
[18:29:00] <KitsuWhooa> meanwhile, of course my vm won't boot. I named the file "gru.cfg"
[18:29:09] <abaumann> lol
[18:29:15] <girls> b-key stuck?
[18:29:26] <KitsuWhooa> doubt it
[18:30:47] <KitsuWhooa> Yeah, systemd-logind works fine on an actual installation
[18:30:58] <KitsuWhooa> so it's either ro, or some package was bugged and has been fixed since
[18:31:07] <abaumann> sounds plausible
[18:31:29] <KitsuWhooa> (it's not worth spending any more time on it, especially since I need strace for this)
[18:40:13] -!- T`aZ has quit [Ping timeout: 252 seconds]
[18:41:56] -!- T`aZ has joined #archlinux32
[18:45:53] <ConSiGno> woo thank you!
[18:50:36] <abaumann> np
[18:51:09] <abaumann> so, me off for today 25/98 ICs removed, checked, sanded, deoxided, reseatet. Errors persisting :-)
[18:51:16] <abaumann> bye
[18:51:23] <girls> bye
[18:51:43] -!- abaumann has quit [Quit: leaving]
[18:51:44] <KitsuWhooa> take care
[18:51:57] <KitsuWhooa> I suspect bad ram
[18:53:59] -!- GNUtoo has joined #archlinux32
[18:55:09] <ConSiGno> I still can't see packages on the site for i486, huh
[18:55:26] <KitsuWhooa> yeah it's still broken
[20:48:32] -!- T`aZ has quit [Ping timeout: 248 seconds]
[20:50:40] -!- T`aZ has joined #archlinux32
[21:02:19] <ConSiGno> next on the agenda: building my own kernel D:
[21:09:29] <ConSiGno> hnng no xscreensaver/wine ports for i486, tempted to compile my own