#archlinux32 | Logs for 2019-06-19

Back
[00:13:20] -!- guys has quit [Ping timeout: 248 seconds]
[00:31:31] <buildmaster> pentium4/parity-ethereum is broken (says rechenknecht).
[00:33:26] -!- traace has quit [Quit: Leaving]
[00:46:30] -!- quequotion has quit [Quit: quequotion]
[00:54:05] -!- lockshaw_ has joined #archlinux32
[00:55:06] <lockshaw_> Are there any pre-existing docker images of archlinux32, or any docs on using the archlinux32/docker repo to create one?
[01:12:37] <elibrokeit> lockshaw_: doesn't docker officially cmllaim that i686 sucks and won't be supported?
[01:15:22] <lockshaw_> elibrokeit: iirc you can run an i686 docker container on a x64 system
[01:37:58] <buildmaster> pentium4/qt5-webkit is broken (says nlopc43).
[01:42:40] -!- samantaz has quit [Ping timeout: 244 seconds]
[01:44:45] -!- samantaz has joined #archlinux32
[02:15:35] -!- eschwartz has joined #archlinux32
[02:35:24] <buildmaster> pentium4/broadcom-wl is broken (says rechenknecht).
[02:42:21] -!- eschwartz has quit [Read error: Connection reset by peer]
[02:42:22] <buildmaster> pentium4/wireguard-arch is broken (says nlopc43).
[02:43:13] -!- eschwartz has joined #archlinux32
[02:55:41] <buildmaster> pentium4/nvidia-390xx is broken (says rechenknecht).
[02:55:52] -!- samantaz has quit [Ping timeout: 245 seconds]
[03:04:46] -!- eschwartz has quit [Read error: Connection reset by peer]
[03:06:24] -!- eschwartz has joined #archlinux32
[03:14:44] <buildmaster> pentium4/geoip2-database is broken (says nlopc43).
[03:29:32] -!- eschwartz has quit [Read error: Connection reset by peer]
[03:30:36] -!- eschwartz has joined #archlinux32
[04:14:59] -!- eschwartz has quit [Read error: Connection reset by peer]
[04:15:48] -!- eschwartz has joined #archlinux32
[04:25:16] <buildmaster> pentium4/ksh is broken (says rechenknecht).
[04:40:09] -!- eschwartz has quit [Read error: Connection reset by peer]
[04:41:00] -!- eschwartz has joined #archlinux32
[05:02:33] -!- eschwartz has quit [Ping timeout: 268 seconds]
[05:04:03] -!- marc2377 has joined #archlinux32
[05:04:21] <marc2377> Hey all.
[05:06:00] <marc2377> The linux package on Archlinux32 refers to upstream git revision: 60592e718b1f9ae072f1a511febd224341a0b65d (https://git.archlinux32.org/archlinux32/packages/src/branch/master/core/linux) but I can't find that revision (nor the previous one) on upstream Arch
[05:06:01] <phrik> Title: archlinux32/packages: Package customizations and pure-i686 packages - Archlinux32 Gitea (at git.archlinux32.org)
[05:06:24] <marc2377> Can anyone instruct me where to find it?
[05:19:02] <bill-auger> marc2377: i dont really understand the question - arch would not have the git sources for linux - you would find that upstream on kernel.org
[05:19:52] <marc2377> Hey bill. The PKGBUILD in the Archlinux32 says, on the first line: 'upstream git revision: 60592e718b1f9ae072f1a511febd224341a0b65d'
[05:20:18] <marc2377> I'm wondering what does that refer to
[05:21:33] <marc2377> From my understanding, the upstream for Archlinux32' kernel is the Arch linux kernel (I can check that by clicking 'Upstream's Source Files' link at https://www.archlinux32.org)
[05:21:36] <phrik> Title: Arch Linux 32 - linux 5.1.6.arch1-1.0 (i686) (at www.archlinux32.org)
[05:24:41] <marc2377> So Archlinux32's upstream is Archlinux, which in turn has upstream kernel.org
[05:24:47] <elibrokeit> bill-auger: the arch packager does in fact use his own personal Linux fork as the PKGBUILD source
[05:24:54] <marc2377> but I can't find revision '60592e718b1f9ae072f1a511febd224341a0b65d' _anywhere_
[05:25:24] <marc2377> Ah, it does?
[05:25:38] <marc2377> he*
[05:26:04] <elibrokeit> marc2377: where are you looking for the revision ?
[05:28:20] <marc2377> I looked at: 1. https://git.archlinux.org 2. https://git.archlinux.org 3. https://git.kernel.org
[05:28:22] <phrik> Title: kernel/git/stable/linux.git - Linux kernel stable tree (at git.kernel.org)
[05:29:20] <marc2377> and also at the Archlinux32 repo I linked previously
[05:30:07] <elibrokeit> That's not even remotely an answer to my question
[05:30:26] <elibrokeit> You gave three links, but only one of them was a git repository
[05:30:40] <elibrokeit> The other two are refs within a git repository
[05:31:02] <elibrokeit> Which implies you don't quite grasp where you're supposed to be looking
[05:31:12] <marc2377> I have all of these checked out
[05:31:21] <marc2377> and I looked by 'git show <hash>' on bash
[05:31:56] <elibrokeit> So why did you link a ref I stead of just packages.git and stable/linux.git :p
[05:32:15] <elibrokeit> At any rate I see it here: https://git.archlinux.org
[05:32:16] <marc2377> Sorry
[05:32:17] <phrik> Title: trunk\linux - svntogit/packages.git - Git clone of the 'packages' repository (at git.archlinux.org)
[05:32:28] <elibrokeit> This is a ref link for the ref you asked about
[05:32:52] <marc2377> Nice, thank you
[05:33:19] <elibrokeit> I'm a very exacting person because I hate confusion, sorry ;)
[05:33:36] <marc2377> Sure, no problem
[05:33:45] <marc2377> you're not wrong
[05:34:19] <elibrokeit> So I guess you probably need to git fetch in packages.git
[05:36:32] <marc2377> Will try
[05:50:07] <marc2377> now, how do I build the Archlinux32 on Arch? (I do have the 32bit chroot, but I don't understand the PKGBUILD)
[05:59:01] <elibrokeit> marc2377: arch32 uses a model of appending their PKGBUILD to the arch one, and cp'ing all other files as replacements
[05:59:27] <elibrokeit> So, cat arch32/PKGBUILD >> arch/PKGBUILD
[06:00:06] <marc2377> Got it
[06:32:00] -!- thePiGrepper has joined #archlinux32
[06:41:05] -!- marc2377 has quit [Ping timeout: 258 seconds]
[06:48:58] -!- davor_ has joined #archlinux32
[06:51:39] -!- davor has quit [Ping timeout: 248 seconds]
[06:51:39] davor_ is now known as davor
[06:53:32] -!- skydrome has quit [Ping timeout: 272 seconds]
[06:59:12] -!- skydrome has joined #archlinux32
[07:16:55] -!- marc2377 has joined #archlinux32
[07:21:17] -!- marc2377 has quit [Ping timeout: 245 seconds]
[07:25:13] -!- tyzoid has quit [Ping timeout: 244 seconds]
[07:26:15] -!- girls has quit [Ping timeout: 244 seconds]
[07:32:53] -!- tyzoid has joined #archlinux32
[07:32:53] <buildmaster> Hi tyzoid!
[07:32:53] <buildmaster> !rq tyzoid
[07:32:55] <phrik> buildmaster: <tyzoid> mpan: I'm bad at trolling because I'm not trolling.
[08:06:44] <buildmaster> pentium4/teamspeak3 is broken (says nlopc43).
[08:30:52] -!- thePiGrepper has quit [Ping timeout: 245 seconds]
[08:39:06] -!- deep42thought has joined #archlinux32
[08:39:07] <buildmaster> Hi deep42thought!
[08:39:07] <buildmaster> !rq deep42thought
[08:39:07] <phrik> buildmaster: <deep42thought> my supervisor when I made my first shot with the laser: He handed me the remote and said: "Hey Erich, want to destroy a 20 Mio€ device?"
[08:41:08] -!- girls has joined #archlinux32
[08:41:09] <buildmaster> Hi girls!
[08:41:09] <buildmaster> !rq girls
[08:41:09] <phrik> buildmaster: <girls> isn't that a typo? shouldn't it be "haskell-base-combat-batteries"?
[08:56:01] -!- deep42thought has quit [Remote host closed the connection]
[09:05:37] -!- deep42thought has joined #archlinux32
[09:05:38] <buildmaster> Hi deep42thought!
[09:05:38] <buildmaster> !rq deep42thought
[09:05:38] <phrik> buildmaster: <deep42thought> documented => the bug becomes a feature
[09:06:12] <deep42thought> an alternative approach to manually appending and copying our modifications is to use "asp32" :-)
[09:06:26] <deep42thought> which should also do the arch=() mangling
[09:20:34] -!- quequotion has joined #archlinux32
[09:21:25] -!- abaumann has joined #archlinux32
[09:21:25] <buildmaster> Hi abaumann!
[09:21:25] <buildmaster> !rq abaumann
[09:21:26] <phrik> buildmaster: <abaumann> still.. I'm close to declare pacman-static the official pacman :->
[09:21:28] <abaumann> hi deep42thought
[09:21:32] <deep42thought> hi abaumann
[09:21:32] abaumann is now known as 17WACOZJF
[09:21:41] <17WACOZJF> "You're now known as 17WACOZJF"
[09:21:43] <17WACOZJF> huh?
[09:21:43] <deep42thought> I know your password now
[09:21:58] <17WACOZJF> that's not remotely my password..
[09:22:17] <17WACOZJF> ..though I have a tendency to mail emails in plain text to people or publish them publicly on github :-)
[09:22:18] -!- 17WACOZJF has quit [Client Quit]
[09:22:40] -!- abaumann has joined #archlinux32
[09:22:40] <buildmaster> Hi abaumann!
[09:22:40] <buildmaster> !rq abaumann
[09:22:41] <phrik> buildmaster: <abaumann> sometimes I get the feeling that the only thing developing fast on Linux is the number of new bugs.
[09:22:48] <abaumann> mmh. something is weird.
[09:24:04] <abaumann> I was wondering: can multiple slaves share packages64 and community64?
[09:24:17] <abaumann> Otherwise I get hell of an I/O load on the new build machine..
[09:24:31] <deep42thought> yes
[09:24:35] <deep42thought> they can
[09:24:39] <deep42thought> we might re-check the locking
[09:24:48] <deep42thought> but it should be not a big issue
[09:24:51] <abaumann> Isn't 'git pull' taking care of this?
[09:24:55] <deep42thought> yes
[09:25:00] <abaumann> let's try :-)
[09:25:10] <deep42thought> the question is: how is the "cannot get lock" error from git handled by the build scripts
[09:25:19] -!- tyzoid has quit [Ping timeout: 276 seconds]
[09:25:24] <abaumann> true.
[09:28:12] <abaumann> "core: ELF 64-bit LSB core file, x86-64, version 1 (SYSV), SVR4-style, from '/usr/lib/systemd/systemd-logind', real uid: 0, effective uid: 0, real gid: 0, effective gid: 0, execfn: '/usr/lib/systemd/systemd-logind', platform: 'x86_64'
[09:28:17] <abaumann> on the buildmaster
[09:28:26] <abaumann> yes. this correlates to my login problems on the Alix
[09:28:55] -!- tyzoid has joined #archlinux32
[09:28:55] <buildmaster> Hi tyzoid!
[09:28:55] <buildmaster> !rq tyzoid
[09:28:57] <phrik> buildmaster: <tyzoid> oblivious user is oblivious
[09:28:58] <abaumann> so, versioning systemd should be 0.242.29 instead of 242.29, never reaching 1.0 :-)
[09:29:02] <abaumann> hi tyzoid
[09:29:07] <abaumann> long no see :-)
[09:29:48] <deep42thought> what does that error mean? Is this a "dumped core"?
[09:29:56] <abaumann> there was :-)
[09:30:13] <abaumann> it means: the systemd people don't test their software.
[09:30:23] <abaumann> and I'm not willing to be their beta testing guinea pig.
[09:30:34] <abaumann> they get money for what they are doing in the end..
[09:31:13] <abaumann> output of 'file /core'
[09:31:23] <abaumann> showing the crashed process.
[09:31:26] <abaumann> I could debug.. :-)
[09:52:43] -!- quequotion has quit [Quit: quequotion]
[10:08:54] -!- marc2377 has joined #archlinux32
[10:09:28] <abaumann> gcc breaks on i686.. nice
[10:10:24] <abaumann> ar: .libs/libstdc++.a: No space left on device
[10:10:24] <abaumann> tee: /logdest/gcc-9.1.0-1.0-i686-build.log: No space left on device
[10:10:32] <abaumann> on 'rechenknecht'
[10:10:36] <deep42thought> oh
[10:10:45] <abaumann> uff. that's an easy one.
[10:10:45] <deep42thought> I thought, it only happened on my test-vm
[10:11:02] <abaumann> I thought, it's a gcc 9 thingy..
[10:13:26] -!- AndrevS has joined #archlinux32
[10:42:36] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.2019-06-19T08:40:07.HyRW2T.stdin"?.
[10:42:51] <deep42thought> oops
[11:03:48] * buildmaster resumes sanity.
[11:31:40] <buildmaster> pentium4/binaryen is broken (says eurobuild1).
[11:42:01] <abaumann> staging-pentium4-build not in the devtools32 package?
[11:42:06] <abaumann> mmh. I build from source
[11:43:41] <abaumann> devtools32-20190201-1
[11:43:44] <abaumann> yep, sounds old.
[11:49:06] <abaumann> oups, my mirror request go over the raspberry pi 1
[11:50:40] <buildmaster> pentium4/grafana is broken (says eurobuild3).
[12:07:15] <abaumann> oh... lunch.. cu later..
[12:07:17] -!- abaumann has quit [Quit: leaving]
[12:30:03] <marc2377> I'm having a hard time creating a new 32-bit chroot on Arch Linux (x86_64). Can it be done with common tools?
[12:36:27] <deep42thought> marc2377: check out devtools32 from our semi-official [releng] repository
[12:36:43] <deep42thought> http://mirror.archlinux32.org
[12:36:43] <phrik> Title: Index of /x86_64/releng/ (at mirror.archlinux32.org)
[12:37:11] <marc2377> cool, thank you
[12:47:32] <marc2377> I should invoke linux32 mkarchroot (...)?
[12:47:46] <deep42thought> yes, if you want to do it manually
[12:48:17] <deep42thought> but have a look at devtools32: /usr/share/devtools/staging-i686-build
[12:48:20] <deep42thought> and friends
[12:48:58] -!- djmoch has quit [Ping timeout: 246 seconds]
[12:49:03] <deep42thought> ah, wrong
[12:49:08] <deep42thought> /usr/bin/staging-i686-build
[12:49:16] <deep42thought> /usr/share/devtools holds the config files
[12:49:58] <marc2377> How can I make it download the packages? It's not working right now because it attempts to download from my system's mirrors (arch64)
[12:50:17] <deep42thought> there is mirrorlist32
[12:50:23] <deep42thought> (also in [releng]
[12:50:24] <deep42thought> )
[12:50:53] <deep42thought> but maybe it is buggy - because all archlinux32 devs use merged mirrors afaik
[12:51:55] <marc2377> Well I got an up-to-date mirrorlist in any case
[12:52:20] <marc2377> but I don't know how to set it up for mkarchroot
[12:52:20] <deep42thought> no, the question is, what mirrors you have on your list
[12:52:41] <deep42thought> e.g. if you use "normal" archlinux mirrors, they (obviously) will not offer i686 packages
[12:52:55] <marc2377> (sorry, meant to say an up-to-date list of arch32 mirrors)
[12:52:59] <deep42thought> however, I use a mirror which does offer x86_64 and i686 packages
[12:53:03] <deep42thought> ah, ok :-)
[12:53:08] <marc2377> oh.
[12:53:34] <marc2377> so I replace my current mirrorlist with that one and it should work?
[12:53:53] <deep42thought> mirrorlist32 is not a replacement
[12:53:59] <deep42thought> it adds a /etc/pacman.d/mirrorlist32
[12:54:05] <deep42thought> which is needed by devtools32
[12:54:13] <deep42thought> (actually, devtools32 should depend on it)
[12:54:43] <deep42thought> but as I said: you might just be hitting a bug here, because we did not actually test whether devtools32 works on a x86_64-only mirrorlist host
[12:55:28] <marc2377> I see. Will try
[12:57:29] <marc2377> Indeed, didn't work
[13:00:52] -!- djmoch has joined #archlinux32
[13:02:34] <marc2377> Hah, now I got it
[13:02:42] <deep42thought> what did you change?
[13:02:59] <marc2377> I had to uncomment the 32bit mirrors haha
[13:03:04] <deep42thought> ok :-)
[13:04:35] <deep42thought> abaumann: just for reference: the 'make oldconfig' inside devops/update-archlinux32-package sees the keyboard from the running console, so if one is asked about new kernel options, one can actually pick a choice
[13:09:48] -!- AndrevS has quit [Remote host closed the connection]
[13:15:18] * buildmaster goes insane.
[13:15:52] <deep42thought> sry
[13:21:54] * buildmaster resumes sanity.
[13:26:04] <marc2377> Are those status reports?
[13:26:24] <marc2377> Sounds accurate ;p
[13:26:36] <deep42thought> yes
[13:26:48] <deep42thought> buildmaster: what's up?
[13:26:48] <buildmaster> up? I'm up for 17 hours, 23 minutes, load average: 1.31, 1.11, 0.90
[13:27:14] <marc2377> lol
[13:27:42] <marc2377> missed opportunity to call it zenmaster
[13:29:05] <marc2377> the 'upstream' (arch main) kernel does not currently depend on 'graphviz', but I noticed the arch32 PKGBUILD includes the dependency - is that correct?
[13:29:12] buildmaster is now known as zenmaster
[13:29:38] <deep42thought> we do? let me check ...
[13:29:54] <marc2377> hahaha
[13:30:14] <marc2377> zenmaster: meaning of life, please
[13:30:42] <zenmaster> marc2377: 42
[13:31:11] <marc2377> [obama_not_bad_face.jpg]
[13:31:14] zenmaster is now known as buildmaster
[13:31:33] <deep42thought> (I'm not sure, how many scripts rely on the correct user name ...)
[13:31:49] <deep42thought> it looks like upstream dropped the graphviz dependency
[13:32:03] <deep42thought> and we still have our patch in place (because we need to drop it for i486)
[13:35:36] <marc2377> right
[13:36:00] <deep42thought> should be fixed in the next build ...
[13:36:33] <marc2377> I'm not entirely sure, but perhaps they dropped it along with python-sphinx, which apparently was causing an error (it's used to build the kernel htmldocs)
[13:37:10] <marc2377> well, gotta go now catch some sleep
[13:37:17] <deep42thought> good n8!
[13:37:17] <marc2377> thanks for the help, see ya another time
[13:37:32] <marc2377> danke
[14:03:33] -!- deep42thought has parted #archlinux32
[14:49:18] <buildmaster> pentium4/js60 is broken (says eurobuild3).
[14:54:29] <elibrokeit> marc2377, deep42thought: I have an x86_64 exclusive machine with a functioning i686 build chroot, which I only use for releasing Pacman-static
[14:58:58] -!- Cthulu201 has quit [Ping timeout: 258 seconds]
[15:18:12] -!- Cthulu201 has joined #archlinux32
[15:18:24] -!- abaumann has joined #archlinux32
[15:18:24] <buildmaster> Hi abaumann!
[15:18:24] <buildmaster> !rq abaumann
[15:18:25] <phrik> buildmaster: <abaumann> bleep. I'm human.
[15:20:48] <abaumann> mmh. eurobuild6-1 was so far only failing packages.. though faster than my other slaves :-)
[15:22:07] <abaumann> ah. manage-slaves. nice. :-)
[15:22:57] <abaumann> ".[sudo] password for slave1: "
[15:22:58] <abaumann> *grmpf*
[15:24:54] <abaumann> I have to update the build descriptions for slaves in the forum.
[15:30:13] <abaumann> "sudo extra-x86_64-build -c -- -- --version
[15:30:21] <abaumann> why that?
[15:31:24] <abaumann> is that for namcap?
[15:33:20] <abaumann> looks. like.
[15:34:49] -!- eschwartz has joined #archlinux32
[15:42:04] -!- isacdaavid has joined #archlinux32
[15:56:51] <abaumann> uh. my eurobuild6-1 buildslave takes all the cores it can find..
[15:57:13] <abaumann> make -j seems to be the default nowadays :-)
[15:57:45] <abaumann> which brings up a good question. Is it better to build one package fast in parallel or build multiple packages on multiple slaves in parallel?
[16:02:34] <abaumann> oh. makepkg.conf :->
[16:03:01] <abaumann> it's _actually_ respected..
[16:04:06] <abaumann> or not..
[16:06:34] <abaumann> ..they are copied during chroot creation only..
[16:10:01] <abaumann> ah well, I'll find out.. :-)
[16:10:02] -!- abaumann has quit [Quit: leaving]
[16:43:16] <buildmaster> i686/qt5-webengine is broken (says eurobuild6-1).
[16:52:50] -!- abaumann has joined #archlinux32
[16:52:51] <buildmaster> Hi abaumann!
[16:52:51] <buildmaster> !rq abaumann
[16:52:51] <phrik> buildmaster: <abaumann> it's not my work. slaves are supposed to do brainless work. ;-)
[16:52:59] <abaumann> eurobuild6-1 works as intended (-j1)
[16:53:09] <abaumann> now I'll add a second and and test the git locking..
[17:13:29] -!- isacdaavid has quit [Quit: Leaving.]
[17:17:32] -!- ofara_ has quit [Ping timeout: 245 seconds]
[17:23:18] -!- ofara_ has joined #archlinux32
[17:25:07] <buildmaster> i686/parity-ethereum is broken (says nlopc43).
[17:34:17] <buildmaster> i686/vocal is broken (says nlopc43).
[17:35:27] -!- ofara_ has quit [Ping timeout: 245 seconds]
[17:38:20] -!- ofara_ has joined #archlinux32
[17:47:39] <buildmaster> i686/ksh is broken (says eurobuild6-1).
[17:48:44] <buildmaster> i686/teamspeak3 is broken (says buildknecht).
[17:54:46] <buildmaster> i686/binaryen is broken (says nlopc43).
[18:02:00] * buildmaster goes insane.
[18:02:16] <abaumann> buildmaster: why?
[18:06:49] * buildmaster resumes sanity.
[18:06:53] <abaumann> aha
[18:06:54] <buildmaster> i686/alacritty is broken (says nlopc43).
[18:07:25] <buildmaster> i686/grafana is broken (says buildknecht).
[18:07:27] <abaumann> auto-repairing capabilities of sanity-check?
[18:17:03] -!- thePiGrepper has joined #archlinux32
[18:19:04] -!- ofara_ has quit [Ping timeout: 258 seconds]
[18:19:37] -!- ofara_ has joined #archlinux32
[18:28:03] -!- Dimtree has quit []
[18:34:17] <abaumann> manage-slaves add works :-)
[18:34:56] <abaumann> I make separate git repos for now in work/repos. git doesn't play well with parallel access and there are too many places to add locks around it in the slaves.
[18:35:09] <abaumann> 2GB of repos per client on a 2TB drive is not a problem for now
[18:42:50] -!- samantaz has joined #archlinux32
[19:21:35] -!- MrBIOS has joined #archlinux32
[19:26:42] -!- MrBIOS has quit [Ping timeout: 245 seconds]
[19:27:01] -!- MrBIOS_ has joined #archlinux32
[19:34:14] -!- thePiGrepper has quit [Ping timeout: 244 seconds]
[19:38:13] <buildmaster> i686/js60 is broken (says eurobuild6-1).
[19:39:59] <slacka123> abaumann: I tested Java on 32-bit Fedora 30 and it works along with ant, even build libreoffice
[19:40:13] <slacka123> https://fedoraproject.org
[19:40:14] <phrik> Title: Changes/Update i686 architectural baseline to include SSE2 - Fedora Project Wiki (at fedoraproject.org)
[19:40:33] <slacka123> should use "-march=i686 -msse2 -mtune=generic -mfpmath=sse -mstackrealign" instead?
[19:42:12] -!- Dimtree has joined #archlinux32
[19:46:22] <abaumann> Yes, Fedora builds for 32-bit Fedora for i586, if I'm not mistaken
[19:46:39] <abaumann> I remember that building glibc must _NOT_ set any funny flags.
[19:48:08] <abaumann> are those flags you are referring to the ones from Fedora?
[19:49:23] <abaumann> at the moment getting the backlog down by adding new build slaves has priority, otherwise we fall behind upstream too much..
[19:52:30] <slacka123> yes those flags are Fedora's flags that work. Their baseline is also Pentium 4
[19:52:50] <abaumann> -mstackrealign sounds promising
[19:53:05] <abaumann> as I'm sure Java plays with stacks.. :-)
[19:53:20] <abaumann> worth a try. I schedule a patch and build..
[19:54:09] <slacka123> Yes, it does - https://bugs.documentfoundation.org
[19:54:12] <phrik> Title: 108619 – (32bitjavacrash) Java Crash on x86 in jfw_plugin_startJavaVirtualMachine w/ recent linux kernels (at bugs.documentfoundation.org)
[19:55:19] <slacka123> Fedora 30 also needs that kernel parameter, "stack_guard_gap=1" to run/build LibreOffice and other java apps
[19:55:34] <abaumann> I read about that too.
[19:56:07] <slacka123> but i686 arch32 also needs it
[19:56:34] <abaumann> the -mstackrealign flag or the kernel flag or both?
[19:58:01] <slacka123> the kernel flag to avoid java runtime errors
[19:58:37] <slacka123> both i686 arch and 32-bit Fedora 32 have working java with that flag
[19:59:03] <abaumann> I just hope, not everything using glibc has to be rebuilt when changing this flag
[20:00:06] <slacka123> me too...thanks. You guys are doing a great job!
[20:00:23] <slacka123> 32-bit Fedora 30
[20:00:35] <slacka123> I don't have a time machine
[20:00:46] <abaumann> sad.. :-)
[20:01:02] <abaumann> I can add a time machine to the todo list ;-)
[20:06:43] <MrBIOS_> abaumann: hey, have you touched your XO-1’s at all?
[20:06:54] <abaumann> Hi MrBIOS
[20:06:59] <abaumann> yes, I touched them.
[20:07:02] <MrBIOS_> As of kernel 5.x, you should be able to build an un-patched XO-1 kernel that will boot
[20:07:07] <abaumann> played with the openfirmware
[20:07:14] <abaumann> but, no Archlinux32 yet.
[20:07:38] <abaumann> The i486 branch is 2800 packages short currently, especially missing things like librsvg, mesa, Xorg, etc.
[20:07:58] <abaumann> I'm confident to get a plain text console though.. :-)
[20:08:09] <MrBIOS_> If you haven’t already, you should unlock them and update them to Q2F20 http://wiki.laptop.org
[20:08:10] <phrik> Title: OLPC Firmware q2f20 - OLPC (at wiki.laptop.org)
[20:08:20] <abaumann> oh. yep.
[20:08:32] <MrBIOS_> keep in mind, Geode is -mostly- i586
[20:08:59] <abaumann> I know. but no cmov
[20:09:12] <abaumann> and some fencing stuff with nops are missing
[20:09:16] <abaumann> CET
[20:09:41] <MrBIOS_> right. See https://www.twam.info
[20:09:42] <phrik> Title: CHOST: i586 vs. i486 on AMD Geode LX (at www.twam.info)
[20:10:26] <MrBIOS_> abaumann: is that just a build capacity limitation, or?
[20:10:38] <abaumann> also a brains/hands limitation :-)
[20:10:50] <MrBIOS_> as usual
[20:11:07] <abaumann> the capacity problem will hopefully be solved with eurobuild6-x slaves, see https://archlinux32.org
[20:11:08] <phrik> Title: Arch Linux 32 - List of Build Slaves (at archlinux32.org)
[20:11:30] <MrBIOS_> but yes, honestly even a CLI-only arch32 would be neat to socialize
[20:12:11] -!- City-busz has quit [Quit: http://quassel-irc.org - Csevegjen kényelmesen. Bárhol.]
[20:12:11] <abaumann> true, that's why we went first to the Alix: see latest post https://bbs.archlinux32.org
[20:12:12] -!- elibrokeit has quit [Quit: A random quit message]
[20:12:14] <phrik> Title: i486 motivation thread / Artwork, Screenshots & Setups / Arch Linux 32 Forums (at bbs.archlinux32.org)
[20:13:22] <abaumann> because deep42thought and I wanted to get the Alix to run some serious text-based stuff
[20:13:51] -!- City-busz has joined #archlinux32
[20:13:52] -!- elibrokeit has joined #archlinux32
[20:14:10] <abaumann> deep42thought: is it a good idea when build slaves on the same machine delete the 'root' part of archbuild? there seems to be locking in place though, I think.
[20:14:36] <abaumann> I end up in some I/O problems, so most likely I get to 4 build slaves with make -j2 on eurobuild6..
[20:14:45] <abaumann> ..at least for now
[20:16:30] <MrBIOS_> do you have need for more build slaves?
[20:18:39] <abaumann> well..if you have some to offer. :-)
[20:18:52] <abaumann> https://bbs.archlinux32.org
[20:18:53] <phrik> Title: How to set up a build slave / Building / Arch Linux 32 Forums (at bbs.archlinux32.org)
[20:19:05] <abaumann> the 18 bullet list at the end.
[20:19:16] <MrBIOS_> are build slaves running as VMs acceptable?
[20:19:23] <abaumann> sure.
[20:19:38] <MrBIOS_> I just installed 10kW of solar at home. I have lots of juice to spare
[20:19:42] <abaumann> i486 build only happen on VMs (because of some setarch Linux processore personality issues)
[20:19:48] <abaumann> wow!
[20:19:52] <abaumann> cool.
[20:20:52] <abaumann> we can easily add slaves now (thanks to deep42thought's manage-slave script) :-)
[20:21:25] <abaumann> though peeking at the buildmaster I can see the day coming when it has to be migrated to new hardware just because there are now more slaves.. but that's not really a problem.. just work..
[20:27:02] <abaumann> I'm actually quite stupid. RAID-1 for the root-partition is fine. For the build data RAID-0 would be much better (and i don't care loosing the data there).
[20:29:09] <abaumann> go-tools outputs tons of debugging output killing my terminal.. yeah.. quality at work..
[21:07:00] <buildmaster> i686/nvidia-390xx is broken (says rechenknecht).
[21:35:55] <abaumann> so, the /data partition is a raid0 now.
[21:36:02] <abaumann> should be faster..let's see..
[21:41:33] <girls> just fyi: I'm online through the alix now - but I still have a little fight to finish with the cable modem
[21:42:26] <girls> abaumann: having different root/ dirs for different build slaves seems wise
[21:42:39] <girls> archbuild itself should lock just fine
[21:43:56] <girls> the questions remaining are: how do we handle failing locks in build-packages and do we somethign to the chroot outside of archbuild (think of the find-umount thing)
[21:44:24] <abaumann> I gave every slave its own work directory and repos
[21:44:29] <girls> in any case, you should run the slaves as different users, otherwise they will lock each other ...
[21:44:33] <abaumann> so at least there we don't have a locking issue
[21:44:35] <girls> ah, perfect
[21:44:52] <girls> by "work" you mean the work dir inside the git repo?
[21:44:58] <girls> or /var/lib/archbuild?
[21:45:03] <abaumann> no. builder/work
[21:45:08] <abaumann> no. builder/work/repos
[21:46:01] <girls> we should maybe make /var/lib/devtools configurable, so each slave can have its own ...
[21:49:32] <abaumann> but the positive thing is: the machine is still damn silent.. the fan of my eeepc is more audible than the build machine..
[21:49:59] <abaumann> yep. and definitely a raid-0 for scratch space.. much faster..
[21:51:29] <abaumann> gpg: skipped "0xC8E8F5A0AF9BA7E7": No secret key
[21:51:32] <abaumann> mmh. bad.
[21:51:52] <abaumann> oh. forgot to import my signing key everywhere :-)
[21:56:38] <abaumann> error: could not rename /var/cache/archbuild32/guile-2.2.4-2.1-i686.pkg.tar.xz.part to /var/cache/archbuild32/guile-2.2.4-2.1-i686.pkg.tar.xz (No such file or directory)
[21:56:50] <abaumann> tons of those errors.. this doesn't sound good.
[21:56:57] <abaumann> so a shared cache seems to be a problem.
[21:57:15] <girls> hmmm
[21:57:26] <abaumann> maybe pacman recovers in the next build..
[21:57:32] <girls> it should
[21:57:42] <abaumann> then it's only a little bit inefficient :-)
[21:57:46] <girls> you could seed the cache from your mirror's pool/ ;-)
[21:57:54] <abaumann> good idea.
[21:58:06] <girls> I'm not sure how practicable it is ...
[21:58:16] <abaumann> would need a continous rsync..
[21:58:39] <abaumann> ==> ERROR: '/var/lib/archbuild/staging-i686/root' does not appear to be an Arch chroot.
[21:58:45] <abaumann> that's somehting I don't like
[21:59:02] <abaumann> I don't thing deleting the 'root' of an 'archbuild' was ever a designed feature.
[21:59:23] <abaumann> it builds a new one, so not a big issue too
[22:00:06] <abaumann> as long as we don't get broken packages I currently don't care
[22:01:04] <girls> yeah
[22:01:19] <girls> an inefficient slave is better than no slave at all
[22:01:32] <abaumann> !grab girls
[22:01:33] <phrik> abaumann: Tada!
[22:01:49] <abaumann> this sounds wrong in so many ways :-)
[22:01:57] <girls> :-/
[22:03:30] <marc2377> hey
[22:04:18] <marc2377> abaumann: about the question you posed earlier about building one package in parallel or many
[22:04:30] <abaumann> hi marc2377
[22:04:32] <abaumann> yes?
[22:04:39] <girls> abaumann: "ip route" on the alix just prints and prints - and never exits
[22:04:43] <girls> is this expected?
[22:04:49] <abaumann> oeh.
[22:04:59] <girls> the routes seem to work fine, though
[22:05:01] <abaumann> let me check on mine.
[22:05:26] <abaumann> maybe you have a loop in your route. ;-)
[22:05:31] <girls> other ip commands also terminate fine
[22:05:50] <abaumann> do we have an strace on i486?
[22:05:53] <girls> btw: openconnect works
[22:06:01] <abaumann> yippie.
[22:06:10] <girls> we do
[22:06:18] <marc2377> I'll hardly saying something you don't already know, but in any case - if your build is bound by IO and the packages are on different storage, building many of them in parallel should be better.
[22:06:51] <marc2377> otherwise, not plenty of difference to expect, but I'd benchmark both cases
[22:06:58] <girls> marc2377: they are not on different storage
[22:07:17] <abaumann> yep. they share a raid-0 with 2 disks as build space now.
[22:07:30] <marc2377> I see
[22:07:30] <abaumann> 4 slaves. I can either do 8 slaves or make -j2 on 4 slaves.
[22:07:58] <girls> rotating disks, not ssds - right?
[22:08:01] <abaumann> OTOH some packages (I don't mention names) just sniff /proc/cpuinfo and take as many CPUs they find, not respecting parallel settings in make..
[22:08:12] <abaumann> yes. good old silent green WDs
[22:08:25] <abaumann> but putting an SSD is an option.
[22:08:40] <abaumann> just not sure, maybe 32GB to 64GB is a better choice (for disk caching)
[22:08:49] <girls> I was just considering the effect of parallel r/ws
[22:09:01] <girls> yeah
[22:09:04] <girls> or building in tmpfs
[22:09:26] <abaumann> tmpfs has to fallback when hitting the configured limit
[22:09:35] <abaumann> at least I think
[22:09:35] <girls> how many lines of strace should I upload? ... it does not end either :-D
[22:10:00] <abaumann> ip route
[22:10:00] <abaumann> default via 192.168.1.1 dev enp0s13 proto dhcp src 192.168.1.25 metric 202
[22:10:00] <abaumann> 192.168.1.0/24 dev enp0s13 proto dhcp scope link src 192.168.1.25 metric
[22:10:04] <abaumann> no problem here
[22:10:17] <abaumann> so, you have a more complex configuration, it seems
[22:10:27] <abaumann> the strace should repeat somewhere.. mmh.
[22:10:31] <marc2377> I'd stick with bulding with j2 on 4 slaves
[22:10:39] <marc2377> tbh I personally like nproc+1
[22:10:53] <girls> why >nproc?
[22:11:13] <abaumann> 1721330 – ip route list infinite loop if 255.255.255.255 route exists
[22:11:28] <abaumann> https://bugzilla.redhat.com
[22:11:31] <phrik> Title: 1721330 – ip route list infinite loop if 255.255.255.255 route exists (at bugzilla.redhat.com)
[22:11:32] <abaumann> this sound interesting.
[22:11:45] <abaumann> I would argue, this is a bug in 'ip' too, yours.
[22:12:06] <girls> I don't have that route
[22:12:18] <girls> or at least it does not show up
[22:12:46] <abaumann> "ip route help", I fail to see a "do not loop endlessly" command line option ;-)
[22:12:47] <girls> unreachable 224.0.0.0/24
[22:12:47] <girls> 224.0.0.0/4 dev wlp0s12
[22:12:50] <girls> what's that???
[22:13:04] <marc2377> in all of my testings, it was faster, and the increased memory is negligible - apparently, having more make tasks than cpu cores seem to account for cpu cycles that are otherwise lost due to IO wait time
[22:13:05] <girls> wlp0s12 has a valid (different) ip - and I'm using it, currently ...
[22:13:15] <abaumann> nice netmask :-)
[22:13:31] <abaumann> multicast?
[22:13:40] <girls> it's an access point
[22:13:48] <abaumann> you have some bonjour/dDNS thingy on the alix?
[22:13:49] <girls> dunno if that influences it
[22:13:56] <abaumann> mmh. shouldn't
[22:13:57] <girls> I try not to
[22:14:17] <abaumann> in the strace you could check the 'write' syscallls and see what gets written to stdout.
[22:14:25] <abaumann> maybe you can narrow it down then.
[22:14:34] <marc2377> this QA over Stack Overflow has more authority on that https://stackoverflow.com
[22:14:34] <phrik> Title: makefile - GNU make: should the number of jobs equal the number of CPU cores in a system? - Stack Overflow (at stackoverflow.com)
[22:14:38] <girls> why? I can see what gets written to stdout
[22:15:08] <abaumann> ah.
[22:15:21] <abaumann> I thought, it's hanging enlessly not printing anything
[22:15:45] <girls> it repeats itself over and over again
[22:15:57] <marc2377> I currently have `MAKEFLAGS="-j$((`nproc`+1))"` on my makepkg.conf
[22:16:03] <girls> aha!
[22:16:07] <girls> now I see: unreachable 255.255.255.255
[22:16:13] <girls> case closed
[22:16:20] <abaumann> :-)
[22:16:38] <girls> (it was there before, I just didn't see it)
[22:17:23] <abaumann> marc2377: this is good for a machine which should not be used for anything else than building a package. I intend to use the machine for other things. :-)
[22:17:57] <marc2377> Oh, right
[22:18:19] <abaumann> /home/slave2/builder/work: Deleting chroot copy 'slave1'...
[22:18:23] <abaumann> yaiks.
[22:18:32] <girls> man, `ssh-keygen -b 4096` takes like forever on the alix ...
[22:18:41] <girls> lol, abaumann
[22:18:49] <marc2377> But do you assign a core count to the build slaves?
[22:19:04] <abaumann> in systemd-nspawn?
[22:19:08] <marc2377> to be honest I don't know how a build slave is set up. for all I know it may be a virtual machine
[22:19:22] <girls> it's literally just different users, isn't it?
[22:19:28] <abaumann> yeah. virtual machines are a little bit too inefficient compared to a build chroot
[22:19:41] <abaumann> cgroups, yep.
[22:19:43] <abaumann> basically
[22:19:53] <girls> arch-nspawn does not assign a core count by default, but I'm sure, one can configure it ...
[22:19:54] <abaumann> I checks whether CPU limitation is possible on systemd-nspawn
[22:20:03] <abaumann> *checked
[22:20:03] <girls> it is
[22:20:08] <marc2377> oh, cool
[22:20:09] <abaumann> a yeah.
[22:20:11] <girls> cpu, ram, cores, ...
[22:20:20] <abaumann> and MAKEFLAGS is also set in the chroot.
[22:20:28] <girls> though I didn't actually tested if/how the limits are enforced
[22:20:32] <abaumann> the only problem is with packages which try to be smart.. :->
[22:20:45] <girls> I think, systemd will kill them :-D
[22:20:53] <marc2377> Seen it happen once
[22:21:18] <girls> it's like watching both your enemies fight each other - no matter who wins, it's fun to watch
[22:21:22] <abaumann> hmum. this find/rm thingy.. it just removes _everything_?
[22:21:37] <girls> abaumann: it does
[22:21:43] <abaumann> cool :-)
[22:21:52] <girls> though it's probably not needed anymore
[22:21:58] <girls> now that the find-umount is in archbuild
[22:22:00] <abaumann> I have a really clena machine afterwards.. ;-)
[22:22:02] <girls> (archbuild32)
[22:22:16] <girls> well, not _every_everything_
[22:22:27] <abaumann> yeah, but the chroots of co-slaves..
[22:22:49] <girls> uhm, oh
[22:22:50] <abaumann> well. for now I let one slave run.. eurobuild6-1
[22:22:54] <abaumann> and sort out the problems later.
[22:23:01] <abaumann> I can remove straws in slave.conf
[22:23:07] -!- thePiGrepper has joined #archlinux32
[22:23:08] <girls> yeah
[22:23:08] <abaumann> make the builds fail fast
[22:23:40] <girls> time for my bed ... cu tomorrow!
[22:23:46] <abaumann> ok. cu. :-)
[22:26:24] -!- abaumann has quit [Quit: leaving]
[22:31:04] <marc2377> cya
[22:50:12] <marc2377> elibrokeit: I managed to create an i686 chroot with 'mkarchroot' by adding the 32-bit mirrors to my mirrorlist
[22:50:32] <marc2377> this also allows me to install packages with 'pacstrap' onto it
[22:51:01] <marc2377> but I'd like to be able to install them by chroot'ing and using 'pacman' instead
[22:51:59] <marc2377> however I currently can't - when I try it, pacman returns: 'error: could not open file: /etc/mtab: No such file or directory \ error: could not determine filesystem mount points'
[22:52:25] <elibrokeit> marc2377: x86_64 devtools does not support this stuff, but you could use mine or deep42thought
[22:52:27] <elibrokeit> 's versions
[22:52:43] <marc2377> right
[23:01:52] <marc2377> Tried with deep42thought's devtools32
[23:03:17] <marc2377> create another chroot, and chroot'ed to it - still I get the same error from pacman
[23:03:25] <marc2377> created*
[23:03:46] <marc2377> tried both traditional 'chroot' and also 'arch-chroot'
[23:04:10] <marc2377> any other hint how I can make it work?
[23:14:49] -!- MrBIOS_ has quit [Quit: MrBIOS_]
[23:16:00] <elibrokeit> you *should* be able to arch-nspawn in and use pacman, though
[23:16:17] <elibrokeit> no different from an x86_64 guest
[23:25:05] -!- MrBIOS has joined #archlinux32
[23:32:53] <marc2377> hmm, after inspecting the output of pacman --debug, I got it to work by disabling 'CheckSpace' on pacman.conf
[23:33:04] <marc2377> and found an old commit for that https://git.seblu.net
[23:33:05] <phrik> Title: pacman-extra.conf · ae7b13646e37ecefb177dffb4e140b7b60451616 · archlinux / devtools · GitLab (at git.seblu.net)
[23:34:22] <marc2377> indeed, arch-nspawn works
[23:35:42] <marc2377> I should default to using it from now on
[23:35:56] <marc2377> it warns 'cp: '/media/Fast/Dev/chroot32-2/root/etc/pacman.conf' and '/media/Fast/Dev/chroot32-2/root/etc/pacman.conf' are the same file' but other than that is ok
[23:44:08] <marc2377> is there a documented list of different kernel config options between arch and arch32?
[23:44:31] <marc2377> I'm about to diff 'em but just wondering