#archlinux-ports | Logs for 2018-01-19

[00:28:28] -!- guys has joined #archlinux-ports
[02:13:59] -!- Polichronucci has quit [Ping timeout: 256 seconds]
[02:15:44] -!- Polichronucci has joined #archlinux-ports
[03:32:33] -!- oaken-so1rce has joined #archlinux-ports
[03:35:45] -!- oaken-source has quit [Ping timeout: 248 seconds]
[05:07:24] * buildmaster goes insane.
[05:07:29] -!- rewbycraft has quit [Ping timeout: 248 seconds]
[05:11:32] -!- rewbycraft has joined #archlinux-ports
[05:16:18] * buildmaster resumes sanity.
[05:18:33] -!- hringriin_ has joined #archlinux-ports
[05:21:51] -!- hringriin has quit [Ping timeout: 248 seconds]
[05:21:51] hringriin_ is now known as hringriin
[05:23:39] -!- MrBIOS has joined #archlinux-ports
[05:59:10] -!- bill-auger has quit [Remote host closed the connection]
[06:43:10] -!- MrBIOS has quit [Quit: Going offline, see ya! (www.adiirc.com)]
[07:13:17] -!- bill-auger has joined #archlinux-ports
[07:35:37] bq is now known as a0
[08:22:44] -!- deep42thought has joined #archlinux-ports
[08:23:18] -!- abaumann has joined #archlinux-ports
[08:35:34] -!- Vollzornbrot has joined #archlinux-ports
[08:36:00] <Vollzornbrot> moin
[08:36:04] <deep42thought> Hi
[08:38:27] -!- alyptik has quit [Ping timeout: 240 seconds]
[08:39:25] -!- alyptik has joined #archlinux-ports
[08:40:08] -!- alyptik has quit [Excess Flood]
[08:40:28] <abaumann> hi
[08:41:47] -!- alyptik has joined #archlinux-ports
[08:48:59] <deep42thought> Switzerland and Germany are online :-)
[08:49:11] <deep42thought> Looks like we all get up at approx. the same time ;-)
[08:50:23] <Vollzornbrot> Swittzerland? oO
[08:51:15] <deep42thought> I thought, abaumann was Swiss?
[08:52:29] <Vollzornbrot> oh abaumann hallo, wie ich gehört habe sprichst du "deutsch"? :D
[08:52:35] <Vollzornbrot> good
[09:13:34] <abaumann> yep :-)
[09:14:45] <Vollzornbrot> abaumann: where in switzerland? iam on the bodensee :D
[09:14:53] <abaumann> Zurich
[09:15:19] <Vollzornbrot> oh nice, not so far from me
[09:15:33] <abaumann> you definitely have the bigger lake. :-)
[09:15:54] <Vollzornbrot> yes ^^ but is very dirty....
[09:16:07] <Vollzornbrot> and i mean not the funny dirty :P
[09:16:18] <deep42thought> and I guess, Zürich has the higher mountains?
[09:16:41] <Vollzornbrot> maybe
[09:16:48] <abaumann> aeh. not yet. but you can see them on a clear day.
[09:16:54] <abaumann> it's more hilly.
[09:16:55] <Vollzornbrot> from me too
[09:28:07] -!- FSMaxB has quit [Ping timeout: 248 seconds]
[09:41:21] -!- FSMaxB has joined #archlinux-ports
[11:05:23] oaken-so1rce is now known as oaken-source
[11:32:25] -!- dopsi has quit [Quit: ZNC - http://znc.in]
[11:33:10] -!- dopsi has joined #archlinux-ports
[12:02:48] -!- dopsi has quit [Quit: ZNC - http://znc.in]
[12:04:01] -!- dopsi has joined #archlinux-ports
[12:07:22] -!- dopsi has quit [Client Quit]
[12:08:32] -!- dopsi has joined #archlinux-ports
[12:40:28] a0 is now known as fnargarmarnargil
[12:41:12] fnargarmarnargil is now known as bq
[13:15:30] <buildmaster> linux-lts is broken (says rechenknecht).
[13:26:39] -!- e1z0 has quit [Ping timeout: 248 seconds]
[13:29:10] -!- e1z0 has joined #archlinux-ports
[13:30:27] -!- dopsi has quit [Ping timeout: 240 seconds]
[13:33:35] -!- dopsi has joined #archlinux-ports
[14:44:10] -!- hringriin has quit [Quit: ^Cc12^CbSee ^Cb^Cc04^Cya^Cc!]
[15:01:03] -!- bill-auger has quit [Ping timeout: 256 seconds]
[15:11:01] -!- hringriin has joined #archlinux-ports
[16:07:21] <abaumann> it's more hilly.
[16:13:25] <deep42thought> abaumann: did you have any idea regarding the opcode sniffer?
[16:13:37] <abaumann> yep: I'm on it. :-)
[16:13:43] <deep42thought> nice :-)
[16:14:01] <abaumann> first I thought to fork namcap, as it contains already similar tests.
[16:14:05] <abaumann> but it's Python.
[16:14:47] <deep42thought> if you - by accident - extract a list of linked libraries, too, that could also be of use :-)
[16:15:01] <abaumann> ah :-)
[16:15:12] <abaumann> could be.. could be..
[16:15:21] <deep42thought> just saying
[16:15:24] <abaumann> :-)
[16:27:52] -!- deep42thought has quit [Quit: Leaving.]
[16:53:26] -!- bill-auger has joined #archlinux-ports
[17:09:00] -!- guys has quit [Ping timeout: 256 seconds]
[17:15:56] <abaumann> deep42thought: this looks promising
[17:15:57] <abaumann> https://superuser.com
[17:15:58] <phrik> Title: x86 - How to check if a binary requires SSE4 or AVX on Linux - Super User (at superuser.com)
[18:16:06] -!- deep42thought has joined #archlinux-ports
[18:26:42] <deep42thought> abaumann: yeah, looks promising
[18:31:44] <abaumann> ./check_opcodes -v -a i686 32bad.pkg.xz
[18:31:46] <abaumann> 11d32: f2 0f 10 44 24 20 movsd xmm0,QWORD PTR [esp+0x20]
[18:31:49] <abaumann> ERROR: libQt5Svg.so.5.10.0 conatains MMX, SSE or SSE2 opcodes
[18:31:57] <abaumann> ./check_opcodes -v -a i686 32good
[18:32:08] <abaumann> passes
[18:32:57] <abaumann> I still have some problems to test for i486 compliance, but maybe I can do this later.
[18:34:00] <abaumann> and doing this in shell is not the fastest thing to do :-)
[18:40:29] <deep42thought> about how slow are we talking here?
[18:50:15] <tyzoid> are we going to say i686 packages can use sse2, and move that over to i584/i486? Or are we going to remove SSE2 from i686?
[18:50:30] <tyzoid> i586*
[19:00:46] <deep42thought> dunno
[19:00:53] <deep42thought> I was thinking about the same
[19:01:07] <deep42thought> we could also introduce i786 as "i686+sse2"
[19:05:20] <tyzoid> or maybe we can provide a non-sse2 repo that can be enabled, which would override sse2 packages
[19:18:14] <deep42thought> I'd rather manage a bunch of symlinks
[19:18:29] <deep42thought> and keep i686 and i686-non-sse2 as two separate architectures
[19:18:37] <tyzoid> yeah, makes sense
[19:18:52] <tyzoid> doing dependency resolution across the two would be tricky
[19:19:38] <deep42thought> on the other hand there is the question if we need to include sse2 at all
[19:19:43] <deep42thought> e.g. what we gain by that
[19:20:06] <tyzoid> IIRC it has a performance impact on cryptocode
[19:23:37] <abaumann> I also think this is a nice separation, i486, i686 without MMX, SSE or SSE2, and a i686 with SSE2
[19:24:46] <deep42thought> What instruction was the amd geode thingy in the alix missing again?
[19:24:47] <abaumann> there is one point: packages which need SSE2 or have upstream SSE2 requirements (like vivaldi) can then be clearly identified as being only in the 'i686 with SSE2' repo.
[19:24:54] <abaumann> padded NOPs
[19:25:12] <abaumann> but NOPL and friends were undocumented opcodes by Intel.
[19:25:43] <abaumann> The alix thingy has no NOPL, but has MMX
[19:26:03] <abaumann> but I'm not sure we should split into too many architectures..
[19:26:07] <deep42thought> so it would be neither i686 nor i686 w/o sse
[19:26:33] <abaumann> it reports as an i585, but actuallly MMX got added with the "Pentium MMX"
[19:26:39] <abaumann> before Pentium Pro
[19:26:54] <abaumann> SSE came with the Pentium III
[19:27:01] <deep42thought> but I agree, that we should not have too many architectures
[19:27:41] <abaumann> the i686 without SSE2, SSE and MMX is most likely the most generic one and should be the one we support currently
[19:28:01] <deep42thought> your proposal, abaumann sounds reasonable to me - but on the other hand I have no experience with different processor architectures :-)
[19:28:10] <deep42thought> yes
[19:28:20] <abaumann> me neither. :-)
[19:28:32] <deep42thought> is sse, sse2 and mmx officially in i686?
[19:28:38] <deep42thought> or is there no "official i686"
[19:28:42] <abaumann> no.
[19:28:47] <abaumann> the problem is: what is i686?
[19:29:02] <abaumann> is it the first Pentium Pro without SSE
[19:29:17] <abaumann> Pentium 4 had SSE2
[19:29:45] <abaumann> so, usually i686 refers to generic code which covers all CPUs from Pentium Pro up to Pentium 4, I think.
[19:29:46] <deep42thought> ok, I'd propose the following names: i486, i586 (=i686 w/o sse, sse2, mmx) and i686
[19:29:52] <deep42thought> ah
[19:29:54] <deep42thought> hmpf
[19:30:04] <abaumann> sorry :-)
[19:30:07] <deep42thought> then we'd need i786 or something
[19:30:08] <deep42thought> np
[19:30:25] <abaumann> what is gcc doing, when you sau -march-i686?
[19:30:38] <deep42thought> dunno
[19:30:40] <abaumann> it's not enabling MMX, SSE or SSE2, I think.
[19:30:57] <abaumann> When used with -march, the Pentium Pro instruction set is used, so the code runs on all
[19:31:00] <abaumann> i686 family chips
[19:31:01] <abaumann> i686
[19:31:16] <abaumann> pentium4 or pentium4m is MMX, SSE and SSE2
[19:31:22] <abaumann> it's in the manpage of gcc.
[19:31:36] <deep42thought> ah, ok
[19:31:48] <deep42thought> so we'd call the architecture "pentium4", too?
[19:31:54] <abaumann> this gives a naming problem for i686.
[19:31:59] <deep42thought> huh?
[19:32:19] <abaumann> ./opcode -L is no help here: 8086 186 286 386 486 Pentium Athlon Deschutes Katmai Willamette PentiumM Prescott P6 Barcelona IA64 Penryn Nehalem SandyBridge Haswell Bulldozer
[19:32:44] <abaumann> yeah. But some later Pentium 4s had even SSE2
[19:32:47] <abaumann> SSE3
[19:32:49] <abaumann> I meant.
[19:33:04] <deep42thought> our i686 stays as is (except that we purge all sse/sse2/mmx packages) and we introduce a new "pentium4" arch, which runs on _any_ pentium4
[19:33:14] <deep42thought> hmm?
[19:33:32] <abaumann> yeah.
[19:33:35] <abaumann> this sound sane.
[19:33:37] <abaumann> sounds
[19:33:56] <deep42thought> I should really get the database ready, so we can have multiple architectures ...
[19:34:25] <abaumann> but for now I'm concetrating to do proper checks for "our" i686 without MMX, SSE and SSE2
[19:34:41] <deep42thought> yeah
[19:34:50] <deep42thought> that's more important, I think
[19:35:04] <deep42thought> "pentium4" is a addon, as it only adds speed
[19:35:20] <deep42thought> and "i486" is an addon, as it will only run on my router anyways ;-)
[19:35:24] <abaumann> :-)
[19:35:33] <abaumann> but; the blacklist will vary between architectures.
[19:35:50] <abaumann> for instance vivaldi would be on the blacklist.i686, but not on the blacklist.pentium4?
[19:36:33] <deep42thought> yeah
[19:36:42] <deep42thought> everything will become architecture dependent
[19:36:52] <abaumann> ah.
[19:37:25] <deep42thought> except, the PKGBUILDs which can have the architecture dependence inside one single file
[19:37:43] <deep42thought> but also dependencies will need to be managed per architecture
[19:38:31] <deep42thought> btw: once you have your opcode sniffer ready, feel free to put it into the build scripts and I'll enable the check in "return-assignment".
[19:38:54] -!- guys has joined #archlinux-ports
[19:39:08] <abaumann> so, is it ok that the script takes a package file and returns 0 or 1 and messages?
[19:39:41] <deep42thought> perfect
[19:39:54] -!- isacdaavid has joined #archlinux-ports
[19:39:57] <deep42thought> well, it is perfect if it only emmits messages in case of error :-)
[19:39:57] <abaumann> aha: old forum entry: https://bbs.archlinux.org
[19:39:59] <phrik> Title: repo with packages optimized for pentium-m / Arch Discussion / Arch Linux Forums (at bbs.archlinux.org)
[19:40:31] <abaumann> though pentium4 is a better name.
[19:45:24] <deep42thought> as long as we don't call any architecture "irc-logs", "archisos", "lastsync" or "lastupdate", I'm fine with every name ;-)
[19:49:04] <abaumann> :-)
[19:55:59] -!- guys has quit [Ping timeout: 248 seconds]
[20:00:08] -!- rewbycraft has quit [Ping timeout: 256 seconds]
[20:00:54] -!- guys has joined #archlinux-ports
[20:07:25] -!- rewbycraft has joined #archlinux-ports
[21:19:17] -!- yans has joined #archlinux-ports
[21:20:56] <yans> Hello
[21:20:59] <deep42thought> Hi
[21:21:17] <yans> I have a question, cause i ask on official channel, but someone told me, i should come here.
[21:21:41] <deep42thought> if you're running i686, then you're right here :-)
[21:21:47] <yans> I have old 32 installation on my netbook, i turned it on after 6 months or so, and i cannot perform any upgrades.
[21:21:56] <tyzoid> error messages?
[21:21:57] <yans> I think this is mirror list problem.
[21:22:22] <deep42thought> you read https://www.archlinux.org
[21:22:22] <phrik> Title: Arch Linux - News: The end of i686 support (at www.archlinux.org)
[21:22:23] <deep42thought> ?
[21:22:38] <yans> cause each time, when i try to switch mirrors list i have that output - The requested URL returned error: 404
[21:22:45] <tyzoid> That's probably it.
[21:22:52] <deep42thought> and more precisely followed: https://archlinux32.org
[21:22:53] <phrik> Title: Arch Linux 32 (at archlinux32.org)
[21:22:59] <yans> deep42thought: exacly, someone told me, that's 32 is unsupported.
[21:23:09] <deep42thought> correct
[21:23:10] <tyzoid> 32 is not officially supported, but we maintain it here.
[21:23:14] <deep42thought> but _we_ support 32 bit :-)
[21:23:17] <yans> So i just should switch mirrors list.
[21:23:22] <tyzoid> or rather, deep42thought and abaumann maintain it.
[21:23:26] <yans> If i would like to use latest updates?
[21:23:27] <deep42thought> you should follow the instructions there
[21:23:33] <yans> I check it.
[21:23:59] <yans> Thanks, if i dont't know something...
[21:24:00] <deep42thought> (which includes one step saying to change the mirror)
[21:24:04] <yans> I'll be back...
[21:24:08] <deep42thought> yea, np :-)
[21:26:13] <deep42thought> tyzoid: abaumann and I may contribute the most to the packages, but there are still others (like you), who for example build isos and run the forums ...
[21:30:11] <guys> or just hang out and discuss solving issues.
[21:30:18] <deep42thought> right :-)
[21:31:37] <tyzoid> yeah, that's what I was saying though
[21:31:43] <tyzoid> I don't really "maintain" it, so to speak
[21:31:48] <tyzoid> I just support that work to an extent
[21:31:55] <deep42thought> no packages, but still "archlinux32"
[21:34:06] -!- shentey has joined #archlinux-ports
[21:38:28] <abaumann> yep, agreed, every contribution is welcome. packages are just a part of the whole thing. :-)
[21:47:10] <tyzoid> deep42thought, abaumann: What's your opinion on arch32web requiring javascript?
[21:47:20] <deep42thought> I hate javascript
[21:47:43] <tyzoid> to write? or run?
[21:47:50] <deep42thought> transformed to your question: don't use it, if it's not absolutely necessary
[21:47:58] <deep42thought> to run
[21:48:00] <tyzoid> that's what I figured.
[21:48:04] <guys> I'd say javascript is okay, as long as it is lightweight and serves a meaningfully useful purpose
[21:48:16] <guys> But I cannot think of such a purpose for the current frontpage
[21:48:32] <tyzoid> guys: Well, I was thinking of using it to integrate jsonapis to the frontend.
[21:48:42] <tyzoid> without the need to constantly reload pages.
[21:48:50] <guys> IIRC archweb uses javascript to collapse the dependency lists for packages, for example.
[21:49:01] <guys> reloading pages is what G-d intended.
[21:49:07] <tyzoid> lol
[21:49:10] <abaumann> Well, you can use Javascript for that, and fall back to not showing it or only the current state in the case of no Javascript.
[21:49:31] <abaumann> or make a html refresh=5s :-)
[21:49:35] <tyzoid> I'm not disagreeing, I was just trying to gauge what the opinion was.
[21:49:36] <abaumann> if that still works.
[21:49:38] <guys> abaumann: that's even worse
[21:49:42] <tyzoid> html refresh=5 does work, but it's terrible.
[21:49:50] <tyzoid> If you're typing in a box, it clears it out on refresh
[21:49:56] <abaumann> old netscapes could do server push.
[21:49:56] <tyzoid> and cursor focus breaks
[21:50:17] <abaumann> you sent a constant stream im mime parts (if I recall correctly), and the browser would update them.
[21:50:23] <abaumann> ok. it was full page updates only.
[21:50:32] <guys> reloading should happen when the user desires it to, and not happen when the user desires it not to. refresh violates both of that.
[21:50:34] <abaumann> the refresh was meant for the non-Javascript fallback.
[21:50:47] <tyzoid> yeah, still terrible
[21:50:51] <abaumann> tell this to ad-popups and Google :-)
[21:50:54] <tyzoid> plus, I don't think you can do noscript in head?
[21:51:01] <guys> Oh, so if you disable javascript it is okay to attack the user with malice aforethought?
[21:51:07] <tyzoid> lol
[21:51:10] <abaumann> :-)
[21:51:19] <abaumann> static page then.
[21:51:21] <tyzoid> guys: Clearly. It's the user's fault for trying to keep their system free of propreitary software.
[21:51:46] <guys> I mean, at least google has a well-written webapp in javascript, that updates itself *properly*
[21:52:11] <tyzoid> which google service does that?
[21:52:12] <guys> tyzoid: never mind proprietary, FOSS javascript can be just as annoying
[21:52:17] <guys> tyzoid: the homepage?
[21:52:23] <tyzoid> ok
[21:52:25] <guys> Also everything else they ever write.
[21:52:42] <guys> Look at their page sources someday, the whole thing is a giant javascript blob
[21:52:47] <tyzoid> yup
[21:52:58] <abaumann> but it falls back nicely in lynx. :-)
[21:53:00] <tyzoid> I've been following it throught the years
[21:53:01] <guys> gmail.com does all kinds of nifty things though
[21:53:04] <tyzoid> throughout*
[21:54:03] <abaumann> mmh.. the opcode sniffer uses a giant regex of all assembly opcodes.. really not fast.. I fear, I'll slow down the build progress too much this way..
[21:54:24] <deep42thought> how long does it take on a package?
[21:54:33] <abaumann> 30 seconds on qt5-svg for instance
[21:54:37] <abaumann> 3 shared libraries
[21:54:40] <deep42thought> thats no problem
[21:54:50] <deep42thought> we just run it on the build slaves, then
[21:55:20] <abaumann> I would like to test it on a local pacman cache before putting it into the chain.
[21:55:28] <deep42thought> fair point
[21:55:44] <abaumann> or over the whole mirror :->
[21:55:49] <deep42thought> maybe you can accelerate it by extracting a list of opcodes which you then grep?
[21:55:52] <abaumann> I want to find this KDE black screen..
[21:55:55] <deep42thought> :-)
[21:56:39] <yans> hmmm...
[21:56:51] <abaumann> I'll see into that..
[21:57:43] <yans> Almost fine, but this - https://pastebin.com
[21:57:43] <phrik> Title: [yans@yinsen ~]$ sudo pacman -Syu :: Synchronisiere Paketdatenbanken... core - Pastebin.com (at pastebin.com)
[21:58:13] <deep42thought> och nö
[21:58:28] <yans> nein...
[21:58:32] <yans> Bitte nicht.
[21:58:41] <deep42thought> didn't I move that from testing to extra?
[21:58:43] <deep42thought> ...
[21:59:24] <deep42thought> ah
[21:59:45] <deep42thought> you need to remove ffmpeg0.10, gstreamer0.10-python and thunarx-python
[21:59:52] <deep42thought> I think, they're dropped
[22:00:00] <deep42thought> *have been dropped
[22:00:08] <tyzoid> sounds about right
[22:00:53] <deep42thought> yupp, they have been dropped
[22:00:56] <abaumann> yes, can confirm, they are dropped
[22:01:00] <deep42thought> :-)
[22:01:02] <abaumann> :-)
[22:01:07] <abaumann> there is an echo somewhere
[22:01:19] <deep42thought> lol
[22:01:37] <tyzoid> lol
[22:02:54] <yans> deep42thought: danke schön!
[22:03:01] <yans> Update in progress.
[22:03:04] <tyzoid> deep42thought: do we need to get a phrik32?
[22:03:10] <deep42thought> nah, what for?
[22:03:18] <deep42thought> yans: gern
[22:03:24] <abaumann> German support. :-)
[22:03:34] <deep42thought> rather a phrik-de
[22:03:36] <deep42thought> ;-)
[22:03:39] <yans> German mirror-list server.
[22:03:44] <yans> And works fine, now.
[22:03:53] <deep42thought> I'm fine with english
[22:04:11] <deep42thought> I'm just sometimes surprised how many german speaking people use arch ...
[22:04:23] <tyzoid> !wtf /usr/lib64/libz.so
[22:04:23] <phrik> tyzoid: What the hell is /usr/lib64/libz.so?
[22:04:30] <tyzoid> !wtf libz.so
[22:04:30] <phrik> tyzoid: core/zlib
[22:04:35] <abaumann> huh?
[22:04:46] <tyzoid> As an example
[22:04:58] <tyzoid> currently, phrik just gets all information about the mainline arch
[22:05:03] <abaumann> ah :-)
[22:05:04] <tyzoid> so the helper commands don't really do much for us here
[22:05:16] <deep42thought> hmm
[22:05:19] <abaumann> I was hunting a /usr/lib64/libz.so on my 32-bit machine..
[22:05:23] <tyzoid> wtf is short for 'What's that file?'
[22:05:31] <abaumann> aeh.. :-)
[22:05:31] <deep42thought> maybe we convince upstream to include !ftw ?
[22:05:32] <tyzoid> so it tells you the package that owns something
[22:06:32] <tyzoid> anyway, just a thought
[22:06:51] <deep42thought> I'd rather avoid forking yet another tool ;-)
[22:07:04] <tyzoid> well less fork and more separate tool
[22:07:23] <tyzoid> that said, all the IRC bots I've worked on in python seem to fail over long periods of time.
[22:07:32] <tyzoid> I usually just put a timer on them to kill 'em after 24hrs
[22:07:37] <tyzoid> so...
[22:07:46] <deep42thought> buildmaster: why dont you build sagemath-doc
[22:07:47] <buildmaster> "sagemath-doc" is broken (2x built), but would be built
[22:07:53] <deep42thought> something like that?
[22:08:01] <deep42thought> feel free to enhance it ;-)
[22:08:10] <tyzoid> kind of, but not quite
[22:08:18] <tyzoid> I still hate the idea of an irc bot in bas
[22:08:20] <tyzoid> bash*
[22:08:21] <deep42thought> hmm, that info is inaccurate - sagemath-doc is blocked :-/
[22:08:30] <deep42thought> it's just a start
[22:08:37] <deep42thought> you can write it in any language you like
[22:08:41] <deep42thought> it's basically ii
[22:08:46] <deep42thought> plus a bash script
[22:09:06] <deep42thought> but you can easily replace the bash part by any file system aware software
[22:10:19] <abaumann> ./check_opcodes -a i686 /var/cache/pacman/pkg/vivaldi-1.13.1008.40-1-i686.pkg.tar.xz
[22:10:22] <abaumann> ERROR: libEGL.so conatains MMX, SSE or SSE2 opcodes
[22:10:23] <abaumann> ERROR: libGLESv2.so conatains MMX, SSE or SSE2 opcodes
[22:10:25] <abaumann> ERROR: libwidevinecdmadapter.so conatains MMX, SSE or SSE2 opcodes
[22:10:26] <abaumann> ERROR: libffmpeg.so conatains MMX, SSE or SSE2 opcodes
[22:10:43] <deep42thought> :-)
[22:10:55] <tyzoid> uh
[22:11:04] <tyzoid> can ffmpeg be compiled without mmx/sse/sse2?
[22:11:11] <abaumann> opcode sniffing on the vivaldi webbrowser.
[22:11:19] <abaumann> it's a local version of the library..
[22:11:25] <deep42thought> *phew*
[22:12:02] <tyzoid> I wonder if libssl has SSE
[22:12:11] <abaumann> I'll check in a pull request in bilder with the script with a first version to play around..
[22:14:09] <deep42thought> I can run it on the master mirror - let's see what packages pop up :-)
[22:14:37] <abaumann> tyzoid: try a Configure on openssl src dir.
[22:14:52] <abaumann> you can choose whether to use assembly at all (no-asm)
[22:15:05] <abaumann> and there are several switches around SSE and whatnot options.
[22:15:32] <abaumann> no-sse2
[22:15:58] <abaumann> and you can force it use only 386 opcodes.
[22:16:07] <abaumann> deep42thought: ok. I make the branch.
[22:16:23] <yans> not so easy...
[22:17:35] <yans> https://pastebin.com
[22:17:35] <phrik> Title: Fehler: Konnte den Vorgang nicht durchführen (In Konflikt stehende Dateien) gt - Pastebin.com (at pastebin.com)
[22:17:42] <yans> long list.
[22:17:51] <tyzoid> damn, did a package get renamed?
[22:18:15] <deep42thought> yans: what does 'pacman -Qo /usr/include/gtksourceview-3.0/gtksourceview/gtksource.h' give you?
[22:18:56] <yans> Fehler: Kein Paket besitzt /usr/include/gtksourceview-3.0/gtksourceview/gtksource.h
[22:19:19] <yans> i need to remove or backup, all that files.
[22:19:31] <abaumann> or use --force
[22:19:35] <abaumann> or not?
[22:19:36] <yans> or use --force.
[22:19:43] <yans> can make more damage.
[22:19:49] <abaumann> true.
[22:19:56] <deep42thought> if it's not owned by someone
[22:19:57] <deep42thought> well
[22:20:04] <deep42thought> I'd check each file with pacman -Qo
[22:20:12] <deep42thought> and if all are not owned by any package
[22:20:14] <deep42thought> then use --force
[22:20:30] <yans> But it's friday. What else i can do.
[22:20:44] <yans> deep42thought: at least it's working. Almost.
[22:20:51] <yans> Besser als nix.
[22:29:45] <deep42thought> abaumann: can you move Opcode_list to a different directory - "conf"? it's not executable and should not be in $PATH
[22:30:18] <deep42thought> or is this supposed to be sourced?
[22:30:25] <abaumann> no. it's read.
[22:30:31] <abaumann> and yes. it's config/resource
[22:30:58] <abaumann> anyway. I have to make it more archlinux32-stylish, reading config dir, having a base dir, etc.
[22:31:25] <deep42thought> just source "${0%/*}../conf/default.conf"
[22:31:35] <abaumann> yep.
[22:32:25] <abaumann> ah. and tmpdir is maybe also not so a good idea. maybe I should use the workdir/tmp.xxx
[22:32:36] <deep42thought> depends
[22:32:49] <deep42thought> if you expect large content
[22:33:00] <abaumann> well. I'm unpacking the packages..
[22:33:04] <deep42thought> oh
[22:33:07] <abaumann> :-)
[22:33:09] <deep42thought> then: yes :-)
[22:33:59] <abaumann> where would you put helper scripts like 'opcode'?
[22:34:14] <deep42thought> also in bin/
[22:34:16] <abaumann> ah. I see 'mysql-functions' are also there.
[22:34:25] <deep42thought> but maybe we should put them in a new directory now
[22:34:33] <deep42thought> and common-functions
[22:34:35] <abaumann> a lib or so
[22:34:39] <deep42thought> yeah
[22:34:43] <deep42thought> sounds good
[22:34:49] <deep42thought> bin/ lib/ conf/ work/
[22:34:52] <abaumann> sounds like FHS :-)
[22:35:18] <abaumann> but 'etc' and 'var' are funny names anyway.
[22:36:57] <tyzoid> abaumann: Can your tool be applied to packages without needing their dependencies installed/checked?
[22:37:05] <abaumann> yes.
[22:37:14] <abaumann> it uses bsdtar and unpacks to the tmpdir
[22:37:19] <tyzoid> and how long would it take to run on some large application?
[22:37:29] * tyzoid would like to integrate it into pkgapi
[22:37:49] <abaumann> mmh. I would actually rather write logs as for namcap and use those.
[22:38:03] <abaumann> can take 10 minutes for large packages
[22:38:07] <tyzoid> ah
[22:38:13] <tyzoid> probably a better idea in that case
[22:39:40] <abaumann> readelf and objdump must be installed.. but they should be in base-devel
[22:40:05] <deep42thought> yeah, just run it like namcap on the built packages, save the logs and upload packages, signatures, namcap- and objsniff-logs to the buildmaster which should decide what needs to be done then
[22:41:44] <deep42thought> abaumann: I think, one can skip the untaring of the whole package and just tell tar to run a "check-this-lib" command on each library found within the archive
[22:41:51] <deep42thought> that might save some time and space
[22:42:13] <abaumann> true. but we also have to check binaries for illegal opcodes.
[22:42:23] <deep42thought> at least if "objdump" and "file" can operate on streams
[22:42:29] <abaumann> loadable modules..
[22:42:51] <tyzoid> abaumann: Perhaps some sort of postprocessor should be used, then?
[22:43:11] <tyzoid> i.e. it can go into staging without the check, but must have been checked to get into testing?
[22:43:19] <tyzoid> that way we don't screw with the pipeline too much?
[22:43:27] <deep42thought> no
[22:43:38] <deep42thought> I'd like to run this test on the build slaves
[22:43:46] <deep42thought> this way it's natively parallelized
[22:43:49] <abaumann> I agree.
[22:43:50] <tyzoid> We still can, just as a separate test job
[22:44:00] <abaumann> You don't want to block the main builder
[22:44:28] <deep42thought> what good is it to have potentially broken packages in staging?
[22:44:30] <abaumann> In the OBS style: a builder, a signer, a rpmlinter..
[22:44:33] <deep42thought> that's not, what it's made for
[22:44:46] <abaumann> yeah, I agree.
[22:44:59] <tyzoid> If there's an interface for parallelizing them already, then that makes the most sense.
[22:45:05] <abaumann> the sniffer should block bad packages even to appear in staging and testing.
[22:45:16] <abaumann> that's how the KDE/Qt5 SSE2 thingy happened.
[22:45:31] <deep42thought> wrong opcodes should be considered as serious as failing check()
[22:46:27] <abaumann> If the tests are run and the chroot would actually be a virtual machine with the CPU restricted to the correct CPU features and given the test coverage is high, check() would also fail due to illegal opcodes.
[22:46:44] <abaumann> so that's the place to put it.
[22:46:50] <deep42thought> exactly
[22:47:29] <yans> deep42thought: and what you think about that? After forcing - https://pastebin.com
[22:47:31] <phrik> Title: ( 1/365) Aktualisiere libutil-linux - Pastebin.com (at pastebin.com)
[22:48:00] <deep42thought> /usr/sbin is a symlink now
[22:48:08] <tyzoid> !give yans pastebin
[22:48:08] <deep42thought> (since a few years IIRC)
[22:48:08] <phrik> yans: Pastebin.com is swamped with advertisements and random captchas. Malware found on pastebin.com has resulted in it being blocked for some users. It injects CRLF line-endings. Please, use something else. Use something sane like https://ptpb.pw https://gist.github.com https://bpaste.net https://ix.io
[22:48:14] <tyzoid> !give yans ptpb
[22:48:15] <phrik> yans: “<command> 2>&1 | curl -F c=@- https://ptpb.pw” OR “curl -F c=@path/to/a/file https://ptpb.pw” OR install community/pbpst
[22:49:08] <tyzoid> deep42thought: Yeah, I thought that change happened a whlie ago
[22:49:54] <abaumann> uh. this system has not been updated for a long time.
[22:50:02] <tyzoid> yup
[22:50:04] <deep42thought> https://www.archlinux.org
[22:50:04] <phrik> Title: Arch Linux - News: Binaries move to /usr/bin requiring update intervention (at www.archlinux.org)
[22:50:20] <deep42thought> hmm, not quite the same error ...
[22:51:23] <abaumann> is there something inside /usr/sbin?
[22:51:47] <deep42thought> just move that to /usr/bin and make a symlink from /usr/sbin to /usr/bin
[22:51:54] <yans> *gsmartcontrol - only
[22:53:09] <deep42thought> well, move it out-of-the-way and then try to updade again
[22:54:30] <abaumann> deep42thought: so, I cleaned a little bit the opcode sniffer..
[22:54:45] <deep42thought> :-)
[22:55:33] <deep42thought> abaumann: did you check, that it runs with "set -e", because that's set by conf/default.conf
[22:56:03] <abaumann> ./check-opcodes -v -a i686 /data/arch32/mirror/i686/core/e2fsprogs-1.43.8-1.0-i686.pkg.tar.xz
[22:56:09] <abaumann> this seemed to have worked.
[22:56:38] <deep42thought> ok
[22:56:53] <abaumann> ERROR: VBoxVMM.so contains MMX, SSE or SSE2 opcodes
[22:56:56] <abaumann> mmh. nice.
[22:57:30] <abaumann> One big caveat though: It's not said, that just because there is a library with SSE2 in it it is also used.
[22:57:35] -!- shentey has quit [Quit: Konversation terminated!]
[22:57:40] <deep42thought> yes
[22:57:44] <deep42thought> unfortunately ...
[22:57:47] <abaumann> and there are things lime Embree, which can actually pick the correct libary at runtime.
[22:58:05] <abaumann> it's a little bit like with text relocations: for some packages they are simply bad.
[22:58:29] <abaumann> for others not.
[22:58:55] <abaumann> ERROR: libembree.so.2.17.1 contains MMX, SSE or SSE2 opcodes
[22:58:57] <abaumann> bingo.
[22:59:14] <abaumann> but I have to check: the minimal ISA could be SSE2 there indeed.
[23:00:18] <deep42thought> find /mnt/archlinux32/i686/ -type f -name '*.pkg.tar.xz' -exec bin/check-opcodes -a i686 {} \;
[23:00:23] <deep42thought> let's see, what happens ...
[23:00:27] <abaumann> ui. brave thing to do :-)
[23:00:44] <deep42thought> this computer is idle until monday otherwise
[23:00:55] <deep42thought> ERROR: libjvm.so contains MMX, SSE or SSE2 opcodes
[23:01:03] <deep42thought> hmm, it doesn't print the package :-/
[23:01:19] <abaumann> oh.. maybe use -v.
[23:01:35] <abaumann> I'll do the same with my prelimiary 486 packages.
[23:01:43] <abaumann> not the -v, the test. :-)
[23:02:21] <deep42thought> the exit code indicates errors, right?
[23:02:26] <abaumann> yes.
[23:02:52] <deep42thought> then I'll run: find /mnt/archlinux32/i686/ -type f -name '*.pkg.tar.xz' -not -exec bin/check-opcodes -a i686 {} \; -printf '%p\n'
[23:03:10] <abaumann> rm: cannot remove '/data/arch32/builder/bin/../work/tmp.check-opcodes.fL5A8I/usr/include/netipx/ipx.h': Permission denied
[23:03:21] <abaumann> and i486 checks are a little bit broken.
[23:04:44] <abaumann> I'll fix those errors tomorrow.
[23:05:36] <deep42thought> I'll put my outputs online once they're done, then anyone can have a look and start fixing stuff :-)
[23:05:55] <abaumann> ok.
[23:08:21] <deep42thought> https://buildmaster.archlinux32.org
[23:08:40] <deep42thought> they're refreshed automatically ...
[23:08:53] <abaumann> ah. nice. :-)
[23:15:48] <abaumann> so. tired. Switzerland out..
[23:15:57] <deep42thought> cu
[23:16:01] <abaumann> bye
[23:16:02] -!- abaumann has quit [Quit: leaving]
[23:21:38] -!- yans has quit [Remote host closed the connection]
[23:27:04] -!- deep42thought has quit [Quit: Leaving.]
[23:34:57] -!- guys has quit [Ping timeout: 240 seconds]
[23:36:56] -!- guys has joined #archlinux-ports
[23:42:05] -!- guys has quit [Ping timeout: 240 seconds]
[23:50:20] -!- guys has joined #archlinux-ports