T2 IRC Log: 2010-09-21

This is the log as captured by an IRC bot in the channel. The statements are those of the individual people and might not neccessarily reflect the policy and legal rules as set forth by the T2 SDE Project.

« prev | next »

--- Log opened Tue Sep 21 00:00:49 2010
01:29 -!- mjungwirth2 [~mjungwirt@213.174.234.68] has joined #t2
01:30 -!- Stealth [stealth@sourcemage/guru/stealth] has quit [Ping timeout: 264 seconds]
01:30 -!- Stealth [stealth@sourcemage/guru/stealth] has joined #t2
01:31 -!- mjungwirth [~mjungwirt@213.174.234.68] has quit [Ping timeout: 264 seconds]
08:24 < ch172> procps-3.2.0 does not compile, installed minimal target, emerged perl, and then ./script/BuildTarget -cfg system fails, anybody sucessfully with building minimal 64bit rarget?
09:32 <@rxr> yes, what exact problem do you get?
09:49 < ch172> Removing old error logs ...
09:50 < ch172> -> Removing var/adm/logs/1-procps.err ...
09:50 < ch172> == 07:44:22 =[1]=> Building base/procps [3.2.8 9.0-trunk].
09:50 < ch172> -> Parallel build using 6 jobs enabled.
09:50 < ch172> -> Reading package config: package/base/procps/procps.conf
09:50 < ch172> -> Preparing build in src.procps.system.20100921.074421.21705.localhost
09:50 < ch172> -> Building. Writing output to $root/var/adm/logs/1-procps.out
09:50 < ch172> !> parser.c:(.text+0x8e2): undefined reference to `display_version'
09:50 < ch172> !> ps/parser.o: In function `parse_all_options':
09:50 < ch172> !> parser.c:(.text+0x1546): undefined reference to `display_version'
09:50 < ch172> !> ps/sortformat.o: In function `do_one_spec':
09:50 < ch172> !> sortformat.c:(.text+0x279): undefined reference to `get_pid_digits'
09:50 < ch172> !> collect2: ld returned 1 exit status
09:50 < ch172> !> Due to previous errors, no 1-procps.log file!
09:50 < ch172> !> (Try enabling xtrace in the config to track an error inside the build system.)
09:50 < ch172> -> $root/var/adm/logs/1-procps.out -> 1-procps.err
09:50 < ch172> == 09/21/10 07:44:27 =[1]=> Aborted building package procps.
09:50 < ch172> -> Unmounting loop mounts ...
09:52 <@rxr> I think I'v not yet seen this
09:52 <@rxr> this is trunk being built on a t2 system?
09:55 < CIA-32> rene * r37637 /trunk/package/base/glibc/ (divdi3.patch.mips glibc.conf subconfig-libc.in):
09:55 < CIA-32> * removed some outdated mips glue from glibc
09:55 < CIA-32> * fixed the build for mips by applying gnu89-inline, likewise
09:55 < CIA-32> * removed some excessive whitespace
09:56 < ch172> yes, it is
10:14 < CIA-32> rene * r37638 /trunk/package/base/linux26/linux26.desc: * updated linux26 (2.6.35.4 -> 2.6.35.5) - SECURITY
10:16 < ch172> It looks like it is a bug, see
10:16 < ch172> http://www.listware.net/201007/gnu-bug-make/104260-make-38191-procps-328.html
10:42 < ch172> Just sent you a patch, on my trunk it now compiles
11:25 < CIA-32> rene * r37639 /trunk/package/database/postgresql/postgresql.desc: * updated postgresql (8.4.4 -> 9.0.0)
11:25 < CIA-32> rene * r37640 /trunk/package/games/ltris/ltris.desc: * updated ltris (1.0.14 -> 1.0.15)
11:25 < CIA-32> rene * r37641 /trunk/package/filesystem/libisoburn/libisoburn.desc: * updated libisoburn (0.6.0.pl00 -> 0.6.2.pl00)
11:25 < CIA-32> rene * r37642 /trunk/package/gnome2/pitivi/pitivi.desc: * updated pitivi (0.13.4.3 -> 0.13.5)
11:27 < CIA-32> rene * r37643 /trunk/package/filesystem/ceph/ceph.desc: * updated ceph (0.21.2 -> 0.21.3)
11:37 < CIA-32> rene * r37644 /trunk/package/base/procps/make-order.patch:
11:37 < CIA-32> Urs Pfister :
11:37 < CIA-32> * fixed procps for make-3.82
12:09 < CIA-32> rene * r37645 /trunk/package/graphic/exact-image/exact-image.desc: * updated exact-image (0.8.1 -> 0.8.2)
13:34 < CIA-32> rene * r37646 /trunk/package/base/embutils/ (embutils.conf embutils.desc): * added mini bunzip2 to embutils
14:33 < rogermason> Hello.
14:37 < rogermason> Having built an x86 generic basic install + xorg and added many packages to it, I now want to build it on (and for) a G5. The G5 in question has 1 Gb RAM and two ~2Ghz processors. Is there any advantage in building a 64 bit system?
14:52 <@rxr> you need to build powerpc64 anyway, the linux kernel only runs on 64bit PowerPC CPUs in 64bit mode
14:52 <@rxr> on ppc64 on t2 by default all binaries are 64bit
14:52 <@rxr> I never had a problem due this, like the 64bitiness
14:52 < rogermason> OK. That makes the decision easy :-).
14:53 <@rxr> on sparc64 however, I endured the pain and let t2 default to 32bit, for the few percent of extra juice on my very aging sparc64 box, ...
14:53 <@rxr> that is sparc64 has a 64bit kernel, too, just the userland is 32bit
14:53 <@rxr> you could make similar changes for ppc64, but the 64bit userland works great on my machines :-)
14:55 < rogermason> My current gentoo install on this G5 is 64 bit with 32 bit userland. I don't like it. All 64 bit will be superior, I think.
14:55 <@rxr> yeah depends
14:55 <@rxr> helps for db and other huge memroy apps
14:55 <@rxr> memory apps
14:56 <@rxr> on RISC, howeve, 64bit is genearlly a few percent slower due to the extra large instructions, and additional spare 0 memory shuffling
14:56 < rogermason> I can live with that.
14:56 <@rxr> on x86-64 that is not that visiblity, because of the variable instruction size, and the vastly increased ammount of general purose registers more than compensating this
14:57 <@rxr> rogermason: yeah me too
14:58 < rogermason> I'm going to try and build trunk on my gentoo G5 as a start. If that fails I'll get the G5 T2 binary. Do you know if it is possible to boot a G5 from a USB stick?
14:59 <@rxr> hm, I think possible but not tested
14:59 <@rxr> the t2 iso2stick script will certainly not work, as it is hardcoded for grub
14:59 <@rxr> I would test with a CD / DVD for a start
15:00 <@rxr> we could work on that when you got t2 on your machine
15:00 <@rxr> we would have to format the stick with an HFS parttion for yaboot, after that it should work
15:00 <@rxr> module potentially hardcoded device numbers in yaboot's config
15:03 < rogermason> Yes, I'm willing to work on the iso2stick thing. I have a version of it that makes the stick with syslinux instead of grub for booting.
15:03 < rogermason> How dis you get T2 on your G5?
15:03 <@rxr> I always ISO installed my ppc's
15:04 < rogermason> from CD?
15:05 <@rxr> sure
15:05 < rogermason> how did you build the ISO? Cross-build on another architecture?
15:08 <@rxr> that is so long ago, honestly I only remember vaguely, ...
15:08 <@rxr> that was still the time of ROCK Linux, ... :-)
15:09 <@rxr> at that time ROCK Linux had some initial PPC support, from the use on IBM machines, ...
15:09 <@rxr> but I think it was too limited to boot on the iBook I purchased
15:09 <@rxr> and cross compiling was still not as advanced as I later implemented
15:09 <@rxr> I think I compiled early ROCK Linux versions on yellow dog for some week until the result booted on my iBook
15:10 <@rxr> if you wanted to know it in that detail
15:10 <@rxr> that was very very long ago
15:10 <@rxr> today you can fully cross compile bootable PPC isos with T2
15:21 < rogermason> hmmm. I wonder if I could get my two T2 boxes to help compile for my G5 using distcc or icecream.
15:23 <@rxr> if the boxes are fast I find this more troubel than worth
15:23 <@rxr> my multi core boxes usually build a minimal iso in just some hours
15:28 < rogermason> OK. Two 2Ghz processors ought to be OK. Thanks.
16:47 <@rxr> for the records, minimal i386 just cross built fine without and error on current trunk
17:07 < CIA-32> rene * r37647 /trunk/target/share/install/ (build_initrd.sh build_stage2.sh init): * finally compress the installer 2nd_stage, again, and even more then ever before
--- Log closed Wed Sep 22 00:00:49 2010