#archlinux32 | Logs for 2019-08-21

Back
[00:41:08] -!- davor has quit [Ping timeout: 248 seconds]
[00:43:26] -!- infides has quit [Ping timeout: 258 seconds]
[00:44:21] -!- davor has joined #archlinux32
[00:59:53] -!- ofara_ has quit [Ping timeout: 268 seconds]
[01:01:42] -!- ofara_ has joined #archlinux32
[02:43:44] <buildmaster> any/firefox-noscript is broken (says nlopc46-i486bs0): https://archlinux32.org
[03:07:48] -!- djmoch has quit [Quit: ZNC 1.7.3 - https://znc.in]
[03:09:48] -!- djmoch has joined #archlinux32
[05:29:27] <buildmaster> any/npm is broken (says eurobuild6-3): https://archlinux32.org
[07:48:39] <buildmaster> i686/gqrx is broken (says eurobuild6-1): https://archlinux32.org
[07:52:14] <buildmaster> pentium4/gqrx is broken (says eurobuild6-4): https://archlinux32.org
[09:00:09] -!- deep42thought has joined #archlinux32
[09:00:10] <buildmaster> Hi deep42thought!
[09:00:10] <buildmaster> !rq deep42thought
[09:00:11] <phrik> buildmaster: <deep42thought> ah, maybe, borg is borged, too
[09:03:14] -!- infides has joined #archlinux32
[09:24:31] -!- ofara_ has quit [Ping timeout: 244 seconds]
[09:53:58] -!- thePiGrepper has quit [Ping timeout: 272 seconds]
[10:12:52] -!- thePiGrepper has joined #archlinux32
[10:13:09] -!- AndrevS has joined #archlinux32
[10:20:53] -!- infides has quit [Ping timeout: 245 seconds]
[12:06:39] -!- AndrevS has quit [Ping timeout: 250 seconds]
[12:32:35] -!- thePiGrepper has quit [Ping timeout: 244 seconds]
[12:40:59] -!- ofara_ has joined #archlinux32
[14:56:00] -!- deep42thought has quit [Remote host closed the connection]
[14:56:21] -!- deep42thought has joined #archlinux32
[14:56:21] <buildmaster> Hi deep42thought!
[14:56:21] <buildmaster> !rq deep42thought
[14:56:22] <phrik> buildmaster: <deep42thought> doesn't one usually try to avoid rust on metals?
[15:07:23] -!- thePiGrepper has joined #archlinux32
[16:07:25] -!- deep42thought has quit [Quit: Leaving.]
[16:52:49] <thePiGrepper> eschwartz[m]: hi again, regarding the issue I had yesterday with compiling for a different -march and -m. what did you do to get the following error? 'cc1: error: CPU you selected does not support x86-64 instruction set'. Because, what I tried to do was basically modify -march to 'pentium4' in makepkg.conf to build the linux package for pentium4 in my 64bit system. I didnt add the -m32 option or
[16:52:55] <thePiGrepper> anything, just modified -march to 'pentium4' and it worked just fine.
[16:53:53] <T`aZ> building the kernel probably doesnt take into accoount the vars set in makepkg
[16:58:34] <elibrokeit> thePiGrepper: the Linux kconfig build system explicitly ignores cflags
[16:59:04] <elibrokeit> You must set your cflags via kconfig, bad flags can wreck your boot
[17:02:19] <elibrokeit> The linux developers specifically don't want leaky bashrc flags to corrupt your kernel builds
[17:07:06] <elibrokeit> thePiGrepper: the kernel kconfig actually has its own setting to determine whether you want a 32-bit or 64-bit kernel :)
[17:51:48] <thePiGrepper> elibrokeit: indeed. I already knew that, I dont know what I was thinking. sorry
[17:51:58] <thePiGrepper> how did you get that error then?
[17:53:06] -!- abaumann has joined #archlinux32
[17:53:06] <buildmaster> Hi abaumann!
[17:53:07] <buildmaster> !rq abaumann
[17:53:08] <phrik> buildmaster: <abaumann> who would have thought digital books burn too. ;-)
[17:55:11] <elibrokeit> $ gcc -march=pentium4 /tmp/hello.c -o /tmp/hello
[17:55:11] <elibrokeit> cc1: error: CPU you selected does not support x86-64 instruction set
[17:56:20] <abaumann> Don't you have to explicitely install gcc from multilib when you want 32-bit support on 64-bit?
[17:56:47] <elibrokeit> Not for a very long time
[17:56:55] <abaumann> thought so :-)
[17:56:59] <elibrokeit> lib32-gcc-libs is what you need, it is in core
[17:59:10] <abaumann> gcc -o hello -march=pentium4 -m32 hello.c
[17:59:14] <abaumann> hello: ELF 32-bit LSB pie executable, Intel 80386, version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux.so.2, BuildID[sha1]=bcfe4dfd3e06886fea22418e78ee128f7e41d6a1, for GNU/Linux 3.2.0, not stripped
[17:59:18] <abaumann> works fine.
[18:00:22] <abaumann> basically, gcc has to support -m32 as long as there is a 32-bit library on 64-bit which needs compiling..
[18:16:17] <thePiGrepper> I think I understand now, however, the msg itself still seems to me kinda weird 'cc1: error: CPU you selected does not support x86-64 instruction set'. I thought the -m32/64 only changed the size of some types of variables, the options responsible for the instruction sets are the -march option as kinda of sets of intructions sets and the specific instruction set options such as -msse2 or -mrtm,
[18:16:23] <thePiGrepper> to name some.
[18:17:59] <thePiGrepper> maybe Im totally wrong about how do the -m options work. if so, can anyone explain how does this work, because I still dont quite understand after reading the gcc man.
[18:19:33] <eschwartz> thePiGrepper: well, you're wrong :)
[18:19:50] <abaumann> better than _totally_ wrong ;-)
[18:19:53] <eschwartz> -m32 generates 32-bit code, not "just the size of variables"
[18:20:21] <eschwartz> -march is useful for selecting presets for the hardware CPU you have, for example, does it support sse2
[18:22:02] <eschwartz> But consider, you may have many CPUs, which should all support -m32 code, not all support -m64 code though
[18:23:04] <eschwartz> So pentium4 does not imply -m32, just like haswell does not imply -m64
[18:24:12] <thePiGrepper> then what the gcc manual says is quite incomplete https://termbin.com :-(
[18:26:14] <thePiGrepper> eschwartz: indeed there are some instructions extensions such as sseX which some processor can have,in theory, independently if they are 32bit or 64bit. that's indeed true
[18:26:50] -!- infides has joined #archlinux32
[18:27:19] <eschwartz> thePiGrepper: "The -mx32 option sets "int", "long", and pointer types to 32 bits, and generates code for the x86-64 architecture."
[18:27:36] <eschwartz> Clearly pointer types, while important, aren't "everything"
[18:28:23] <eschwartz> "The -m32 option sets "int", "long", and pointer types to 32 bits, and generates code that runs on any i386 system."
[18:28:29] <eschwartz> Note the and.
[18:28:29] <thePiGrepper> sure, but that's x32, clearly that text focuses on size of types, be that int/long or pointersm that's why there's that mixed case which is x32
[18:28:54] <eschwartz> If the bit after the "and" was automatically implied by the first half, then they would not bother with the first half?
[18:29:36] <thePiGrepper> when you say that m32 implies more than "just the size of variables", that's not really clear when you read the manual, or is it?
[18:30:02] <eschwartz> thePiGrepper: The x32 ABI is a fantastic (in the fantasy sense) architecture which uses 32-bit types, but generates x86_64 executables!
[18:30:42] <thePiGrepper> Im not saying it doesnt do that, it clearly does after testing it, however the manual isnt that clear , that's what Im saying
[18:30:50] <buildmaster> i686/akonadi-contacts are broken (says eurobuild3): https://archlinux32.org
[18:30:55] <thePiGrepper> eschwartz: hehe, yeah x32 is 'great'
[18:30:58] <eschwartz> And yes, the manual does seem to distinguish between the two things that -m32 does. 1) define the pointer sizes, 2) choose between i386 and x86_64 binary types
[18:31:34] <eschwartz> It even highlights how -m32 and -m64 are accompanied by -mx32 which is half of one and half of the other
[18:31:39] <thePiGrepper> and the combination between what the manual says (or doesnt say) and the error you got is what's confusing to me
[18:33:15] <abaumann> I could imagine having 64-bit registers for computation and 32-bit for pointers is handy for something small, like a Linux kernel. :-)
[18:34:39] <thePiGrepper> I actually wanted to test that, but never got to it. the only build 'alive' I think it's a debian one right? (talking about x32 systems)
[18:38:50] * abaumann shrugs
[18:38:51] <abaumann> dunno
[18:40:00] <abaumann> yeah, looks like they have a x32 chroot and a x32Port documentation page
[18:40:21] <buildmaster> i486/akonadi-notes are broken (says eurobuild6-7-i486): https://archlinux32.org
[18:41:29] <abaumann> yah. there is definitely a need for this type of architecture, but wasn't there rumours support of x32 would get dropped soonish in the Linux kernel?
[18:42:25] <abaumann> Gentoo guys are surely also playing with x32.. :-)
[18:46:04] <buildmaster> any/npm is broken (says buildknecht): https://archlinux32.org
[18:48:40] <buildmaster> pentium4/akonadi-contacts are broken (says eurobuild3): https://archlinux32.org
[18:52:27] <thePiGrepper> there have been those kind of talks for years, but when that happens, those x32 users come out of who knows where saying that there're a lot of ppl and companies still using it. so who knows.
[18:53:57] <thePiGrepper> abaumann: are they?(Gentoo ppl) I didnt know that
[18:54:28] <thePiGrepper> finding an irc channel for x32 users would be nice
[19:01:58] <abaumann> https://github.com
[19:02:00] <phrik> Title: GitHub - 1camper/gnux32: gnux32 (glibc + x32 abi) overlay for Gentoo Linux (at github.com)
[19:02:17] <abaumann> https://www.phoronix.com
[19:02:20] <phrik> Title: Gentoo Does An x32 Stage 3 Release Candidate - Phoronix (at www.phoronix.com)
[19:02:30] <abaumann> not sure, if they continued, because that was 2012..
[19:03:48] -!- infides has quit [Ping timeout: 245 seconds]
[19:04:58] <abaumann> Required version of freetype: 2.2.1
[19:04:59] <abaumann> Detected freetype headers: 2.10.1
[19:05:04] <abaumann> is done with a strcmp :-)
[19:05:15] <abaumann> so, the first version with 2 digits fails?
[19:05:20] <abaumann> openjdk7 people, tsts
[19:39:59] -!- deep42thought has joined #archlinux32
[19:39:59] <buildmaster> Hi deep42thought!
[19:39:59] <buildmaster> !rq deep42thought
[19:40:00] <phrik> buildmaster: <deep42thought> my cluster of 1k 4/86's will be happy to iterate Maxwell-Vlasov-equations :-D
[19:40:11] <deep42thought> abaumann: you know the story, why there is no windows 9?
[19:40:37] <abaumann> hi deep42thought
[19:40:47] <abaumann> because of some internal API f*up?
[19:41:08] <abaumann> :-)
[19:41:29] <deep42thought> no, because some third party software tests for "windows 9x" literally to detect "old" versions :-)
[19:41:49] <abaumann> ah.. I remember now :-)
[19:42:04] <abaumann> nothing M$ could do about it..
[19:43:15] <abaumann> I like software which unzips files as part of configure *grmpf*
[19:43:30] <abaumann> should I write now a diff file for a compressed ZIP file?
[19:44:07] <abaumann> but to be mean: I don't expect anything other from Java people :->
[19:46:48] <deep42thought> why not patch after configure?
[20:25:07] <abaumann> because configure compiles the C program and runs it inside configure
[20:26:13] <abaumann> just realized. I don't need Java 7 for libreoffice rebuilds
[20:26:37] <eschwartz> deep42thought: I heard that Windows 7 ate 9.
[20:26:44] <deep42thought> !grab eschwartz
[20:26:45] <phrik> deep42thought: 🎉
[20:55:55] -!- abaumann has quit [Quit: leaving]
[21:00:51] <buildmaster> pentium4/kmymoney is broken (says eurobuild6-3): https://archlinux32.org
[21:08:03] * deep42thought just overwrote a µsd card of some raspi with the backup of the wrong machine :-(