--- Log opened Tue Jan 18 00:00:21 2005
--- Day changed Tue Jan 18 2005
00:00 < mnemoc> that's our only 'known issue' on wiki :p
00:00 < nzg> can you include a linux-src on the next isos?
00:00 < mnemoc> linux-src was removed and will neve come back :'(
00:01 < mnemoc> because it is considered harmful
00:01 < nzg> why that?
00:01 < nzg> at least drop a tarball in /usr/src
00:03 < mnemoc> rene considered it harmfull because you need to rebuild many others packages when you rebuild your kernel :\
00:04 < rxr> nzg: you still have the patches and can tar xvfj ; patch ... when you want it ...
00:04 < nzg> but there is no tarball i can see
00:05 < nzg> except for rock-src
00:05 < rxr> well - ok - I meant as downlaoded by the t2 system or from kernel.org
00:06 < rxr> the various kernel sources just kept the CD a bit full ...
00:07 < nzg> just if you need to rebuild the kernel for the network driver....
00:07 < mnemoc> i miss those bits
00:07 < rxr> and there is this rebuild inconsistency problem where we introduced postlinux.conf - so that you get all 3rd party packages for any kernel (not just some default one) - and all of them automatically rebuild with your kernel
00:07 < rxr> all network modules should be enabled by default ...
00:08 < rxr> mnemoc: if you miss it that badly - then readd them ...
00:08 < nzg> at least my network doesn't work... maybe its some problem with the asus hardware...
00:08 < rxr> hm
00:10 < mnemoc> i only miss them due to the lack of tuned configuration
00:11 < rxr> http://www.heise.de/newsticker/meldung/55232
00:11 < rxr> mnemoc: chile - that is t2 based I hope ?
00:11 -!- praenti [~praenti@82.139.198.99] has quit [Read error: 104 (Connection reset by peer)]
00:11 < mnemoc> both are rocks 2.0.x
00:12 < mnemoc> mm
00:12 < mnemoc> mom
00:12 < mnemoc> i though you were talking about mirrors ;)
00:12 < rxr> nope ...
00:12 < rxr> about the news
00:12 < mnemoc> no, fedora
00:12 < rxr> some chilean linux stufff ...
00:12 -!- praenti [~praenti@82.139.198.99] has joined #t2
00:13 < mnemoc> rms was on chile today
00:13 < rxr> Xen 2.0.3
00:20 * mnemoc moving home
00:20 < mnemoc> bbl
00:22 < rxr> cu
00:30 < nzg> rxr: sorry... works without rebuilding the kernel *g*
00:37 < rxr> what was the problem ?
00:38 < rxr> -> Compiler Cache Hits while pkg build: 99.97% (3656 hits, 1 misses)
00:38 < rxr> ^- X.org
00:39 < jsaw> leaving... to tired to accomplish anything... cu
00:41 < nzg> cryptic module name... its sk98lin for marvel youkon chipset
00:42 < nzg> found it while digging through make menuconfig
00:44 < rxr> rockplug did not match it ?
00:45 < nzg> no
00:45 < nzg> maybe because it matched the firewire first
00:45 < rxr> well - has it one PCI ID?
00:46 < rxr> I guess you have one for the IEE1394 and one for the NIC, no?
00:46 < rxr> what does lspci say ?
00:47 < _Ragnar_> okay, so I now got some packages in TOOLCHAIN/pkgs ... is it enough to slap those on a 64bit machine and then reboot?:)
00:48 < rxr> you did a full minimal build or so ?
00:48 < _Ragnar_> full generic
00:48 < rxr> at what stage-package are you ?
00:48 < _Ragnar_> (cross compile)
00:49 < _Ragnar_> last was 1-sysfiles
00:49 < nzg> 02:00.0 Ethernet Controller Marvell blabla
00:50 < nzg> 02:01.1 Fire Wire IEEE 1394 blabla
00:52 < rxr> what does this yield for you:
00:52 < rxr> rockscan pci
00:52 < rxr> query if too much ...
00:53 < rxr> _Ragnar_: no - let 1-* finish
00:53 < _Ragnar_> it finished
00:53 < rxr> well - than you could chroot into it and run stuff in it
00:54 < rxr> momemn
00:54 < rxr> ah - yep - now it is finished anyway, isn't it
00:54 < _Ragnar_> no I can't, since I'm not on a 64bit machine
00:55 < rxr> ouhm - why do you ask then ;-) ?
00:55 < _Ragnar_> I'm not exactly clear as how to go on from here
00:56 < _Ragnar_> just unpack TOOLCHAIN/pkgs/* on the target system, then reboot into it?
00:56 < valentin> hh
00:57 < valentin> sorry
00:58 < valentin> my terminal was kind of stucked
00:59 < rxr> _Ragnar_: yes - either unpack the tarballs - or cp build/whatever/*
01:00 < _Ragnar_> ^_^ fine :) thanks, will do after work
01:06 < _Ragnar_> .oO(why would libcrypto.so.0.9.7 require executable stack?)
01:25 < _Ragnar_> .oO(and why would it need to be disabled on a non-stack-protected system?)
01:26 < rxr> _Ragnar_: where do you see that ?
01:29 < _Ragnar_> on a 2.4.5 kernel after trying a full update (including openssl and glibc)
01:36 < _Ragnar_> this is weird error day;)
01:37 < rxr> hm?
01:40 < _Ragnar_> well all googling I did about this points to some grsec or similar patch being running, which I do know isn't installed here
01:50 < mnemoc> re!
01:50 < rxr> hi!
01:50 < _Ragnar_> hi
01:51 < _Ragnar_> *grrr* ... how are the chances of being able to build a minimal router with uclibc?
01:51 < mnemoc> rxr: rsync finished, thanks :)
01:51 < rxr> nice
01:51 < rxr> have fun
01:51 < rxr> I hope it works well
01:52 < rxr> thanks to the traffic shaping I use for some weeks now I have not even noticed your rsync run ....
01:52 < mnemoc> :)
01:54 < mnemoc> rxr: old glibc segfaulted too
01:59 < mnemoc> _Ragnar_: as far as you don't need to enter stage 2 everything will go great ;)
02:02 < _Ragnar_> ;p
02:03 < _Ragnar_> meaning I'd have to do a cross-build for that, too?
02:04 < mnemoc> or found why the heck fl_wrapper segfaults :)
02:08 < rxr> you found out ?
02:08 < mnemoc> me?
02:08 < rxr> Davyd Madeley wrote a Sneak Peek at Gnome 2.10, scheduled for release on the March 9, 2005.
02:09 < rxr> 02:04 < mnemoc> or found why ...
02:10 < _Ragnar_> hmmm ... make[1]: Leaving directory `/export/t2-trunk/src.uclibc.1106011179.23982.fad8824d/uClibc-0.9.27/libc'
02:10 < _Ragnar_> No 1-uclibc.log file.
02:10 < mnemoc> that was to _Ragnar_, he has two options, cross-build or find why fl_Wrapper segfaults
02:10 < mnemoc> _Ragnar_: and the error is?
02:10 < _Ragnar_> doesn't even get to stage2
02:10 < _Ragnar_> none
02:12 < mnemoc> what's the error?
02:12 < rxr> http://www.gnome.org/~davyd/gnome-2-10/
02:12 < _Ragnar_> No 1-uclibc.log file. <-- that's the only error I get
02:12 < rxr> _Ragnar_: try addint the [F] NOPARALLEL
02:12 < rxr> flag to the uclibc.desc
02:13 < rxr> _Ragnar_: what target are you building that it includes the uclibc?
02:14 < _Ragnar_> generic, with [*] Use uClibc instead of fat glibc, for athlon 32
02:14 < rxr> oh - interesting build that is ..
02:16 < _Ragnar_> I've deselected basically all of x11/kde/gnome/... usw
02:17 < mnemoc> _Ragnar_: to see what is failing you need to check at least 10 lines above that you pasted
02:17 < _Ragnar_> currently trying with 1 build job instead of 2
02:19 < rxr> post when it built then ...
02:22 < mnemoc> rxr: did you saw my message about glibc?
02:22 < _Ragnar_> okay worked with 1 build job
02:22 < rxr> which one ?
02:24 < mnemoc> rxr: 2.3.4-2004-08-01 segfault just as xmas-edition
02:24 < rxr> _Ragnar_: adding the NOPARALLEL flag would have had the same effect withotu loosing parallel builds for the next packages ...
02:25 < rxr> mnemoc: hm
02:25 < rxr> IIRC I never saw those with that older version
02:25 < _Ragnar_> rxr: yeah I've noticed, I've added the flag and then set the parallel builds to 2 again ;) thx for the reminder
02:26 < mnemoc> rxr: me neither, so it is a regretion of another change
02:27 < rxr> maybe the other flist hacking ...
02:27 < rxr> flist wrapper I meant
02:27 < mnemoc> i know
02:28 < mnemoc> but i can't imagine what
02:28 < CIA-10> rene * r5636 /trunk/package/base/uclibc/uclibc.desc: * fixed uclibc to not build in parallel
02:35 < rxr> Ironically, Microsoft Research provided funding for the Xen group when it was founded two years ago, but has since back out, Xen officials said.
02:38 -!- ideal [~idealm@211.100.227.107] has joined #t2
02:44 < rxr> hi ideal
03:00 -!- rxr_ [~rene@p213.54.195.130.tisdip.tiscali.de] has joined #t2
03:00 -!- Topic for #t2: T2 | 2.1.0-beta2 RELEASED | The next generation of System Development Enviroments (SDE) | http://www.t2-project.org/ | CIA, don't ever leave us again!
03:00 -!- Topic set by mnemoc [] [Wed Jan 12 03:04:43 2005]
03:00 [Users #t2]
03:00 [ _Ragnar_] [ daja77] [ jsaw ] [ nzg] [ praenti] [ rxr_ ]
03:00 [ CIA-10 ] [ ideal ] [ mnemoc] [ ops] [ rxr ] [ valentin]
03:00 -!- Irssi: #t2: Total of 12 nicks [0 ops, 0 halfops, 0 voices, 12 normal]
03:00 -!- Channel #t2 created Sun Aug 8 21:15:33 2004
03:00 -!- [freenode-info] help freenode weed out clonebots, please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup
03:00 -!- Irssi: Join to #t2 was synced in 11 secs
03:03 -!- rxr [~rene@p213.54.195.112.tisdip.tiscali.de] has quit [Read error: 54 (Connection reset by peer)]
03:03 -!- sparc-kly [~mubex@65-23-204-29.prtc.net] has joined #t2
03:04 < rxr_> hi sparc-kly
03:04 -!- You're now known as rxr
03:04 < sparc-kly> hi :P
03:04 < rxr> how are you ?
03:06 < sparc-kly> im fine . and you .
03:06 < sparc-kly> today i preorder mac mini
03:06 < rxr> oh - quite well ,too
03:06 < rxr> oh!
03:07 < sparc-kly> for my work . i need one for my house LOL
03:08 < rxr> a mac mini for your work ?
03:10 < sparc-kly> yes
03:10 < rxr> what is your job that you need a mac mini ?
03:12 < sparc-kly> i replace my desktop p3 1ghz . for G4 is good idea
03:12 < sparc-kly> im lab admin
03:12 < sparc-kly> computer lab LOL
03:32 < rxr> ic ;-)
03:32 < rxr> what was your actual plan with t2 ? ;-)
03:33 -!- KGB355 [~cookbot@200.75.27.34] has joined #t2
03:33 < mnemoc> :D
03:34 < mnemoc> rot13 hello T2
03:34 < KGB355> uryyb G2
03:34 < mnemoc> :D
03:34 -!- KGB355 [~cookbot@200.75.27.34] has quit [Remote closed the connection]
03:35 -!- KGB387 [~cookbot@200.75.27.34] has joined #t2
03:35 -!- KGB387 [~cookbot@200.75.27.34] has quit [Remote closed the connection]
03:36 -!- KGB397 [~cookbot@200.75.27.34] has joined #t2
03:37 -!- KGB397 [~cookbot@200.75.27.34] has quit [Remote closed the connection]
03:37 < rxr> ?
03:37 < mnemoc> he is joining #kgb now
03:37 < rxr> seems mnemoc is playing with his shiny new bot ,-)
03:37 < mnemoc> :p
03:38 < mnemoc> to avoid noice
03:38 < rxr> test noise ?
03:38 < mnemoc> yep
03:39 < rxr> I'm looking forward ;-)
03:42 < rxr> oh - so late ...
03:44 < mnemoc> late? < 4am is not late ;)
03:44 < rxr> 3:44 ...
03:44 < mnemoc> <4am ;)
03:46 < rxr> wow
03:46 < rxr> I start a just4fun test build on my sparc ...
03:46 < rxr> and then i'm in bed ~ 4 ....
03:48 < ideal> Hi, rxr. Sorry, just now, busy in working:)
04:04 < rxr> mnemoc: how is your kdb don
04:04 < rxr> do'n even ...
04:05 < _Ragnar_> hmmm ncurses fails with unknown refs to __ctype_b
04:07 < rxr> xo x86-64 cross ?
04:07 < rxr> are you ad :HEAD ?
04:07 < _Ragnar_> no, x86-32 with uclibc
04:08 < rxr> ouh
04:16 < mnemoc> re
04:16 < mnemoc> rxr: i'm re-learning perl with it ;)
04:21 < _Ragnar_> hmm
04:21 < rxr> mnemoc: can you imagine some negative unix globbing pattern for not *.cache ?
04:22 < mnemoc> if we are sure we have only one .cache file per package, no problem
04:22 < mnemoc> some on .conf and .desc
04:23 < rxr> I mean as in "ls *.{^cache}"
04:23 < rxr> just some pattern that works
04:24 < _Ragnar_> btw: !> uclibc: Unknown package category: NOPARALLEL
04:24 < _Ragnar_> I think you meant [F], not [C]
04:25 < mnemoc> *G*
04:25 < rxr> argh
04:25 < rxr> this NOPARALLEL does not like me ...
04:25 < _Ragnar_> *g*
04:26 < mnemoc> rxr: why not ls package/*/*.cache ?
04:26 < rxr> I do _not_ wan the cache file
04:26 < rxr> everything else but not it ...
04:26 < rxr> and it does not need to work over multiple packages
04:26 < rxr> cd $confdir ; ls the-globbing is enough
04:28 < mnemoc> package/*/$pkg/*.cache :)
04:28 < mnemoc> it's the same
04:28 < mnemoc> reducing .cache is another thing, and we agree.
04:29 < mnemoc> whata re you trying to implement?
04:36 < rxr> ?
04:36 < rxr> mnemoc: I want to ls any other file - but _not_ the cache !!!!
04:36 < mnemoc> aha
04:37 < rxr> your pattern only gives the cache ;-)
04:37 < mnemoc> sure
04:37 < rxr> mnemoc: hey - I might be sleepy - but I would not ask for the pattern to list a file *g* ;-)))
04:37 < mnemoc> find ! -name '*.cache'
04:37 < mnemoc> we are both sleepy ;)
04:37 < rxr> *lol*
04:37 < rxr> ok
04:37 < rxr> so far I have
04:37 < rxr> ls | grep -v .cache
04:38 < rxr> is there really no globbing to supress some files ?
04:38 < rxr> I mean - so far I never searched - but - that looks a bit poor ...
04:38 < mnemoc> you have packages with subdirs
04:39 < rxr> oh !
04:39 < rxr> thanks for that reminder!
04:40 < rxr> even a stronger argument to rework the current checksumming
04:40 < rxr> it does not take that into account
04:40 < rxr> that is even more vital for ROCK Linux, since they have many more of those
04:40 < rxr> due to the package forking ..
04:41 < rxr> well - anyway - that the rock stuff is not cooked up is - well - some well known fact ... :-)
04:41 < rxr> mnemoc: I wanted to alter the pacakge $confdir checksumming for so long
04:41 < rxr> and since I feel a bit too awaken to go to bed and watch my U30 building t2 I thought it might be the time to do it ...
04:42 < mnemoc> ew
04:42 < mnemoc> re*
04:42 < rxr> e.g. I do not want to checksum unimporant feels - so typographic corrections and white space changes do not touch the checksum
04:42 < rxr> also not the copyright ...
04:42 < rxr> and comments ...
04:42 < rxr> the rock people want so ugly, error prone, manual revision counter
04:42 < rxr> the extra version field
04:43 < rxr> and that only because clifford things he checksums the .cache file - what he does not ...
04:43 < mnemoc> typographic corrections? how can you ignore those?
04:43 < rxr> but anyway - all the others just nodded on the video and after 20 minutes of strange talking acked it ..
04:43 < rxr> mnemoc: only in tags like I T, ...
04:44 < mnemoc> on compare.sh i remove some stuff like that
04:45 < mnemoc> spacing can be ignored output-ing via | tr '\t' ' ' | tr -s ' '
04:47 < rxr> nah - all in one sed call ....
04:47 < rxr> I already have that part:
04:47 < rxr> sed -e '/^[ ]*#.*/d' -e '/^\[COPY\]/d' -e '/^\[[T,I,U,A,M,L,S,C]\]/d' -e '/^[\t ]*$/d' -e 's/[\t ]*//g'
04:49 < mnemoc> CV-URL ;)
04:49 < rxr> find ! -path './.*' ! -name '*.cache'
04:49 < rxr> cat `find ! -path './.*' ! -name '*.cache'` | ...
04:49 < rxr> what do you think ?
04:50 < rxr> cat `find ! -path '*/.*' ! -name '*.cache'` | ...
04:51 < rxr> that would get us a quite sanitized checksum IMHO
04:52 < mnemoc> does that find do what we can imagine it do?
04:53 < _Ragnar_> okay back
04:53 < rxr> yes
04:53 < rxr> tested
04:53 < CIA-10> mnemoc * r5637 /trunk/scripts/Download: * changed scripts/Download to use svn:// and svn+http:// instead of dirty svn://svn://
04:53 < rxr> thus my 2nd corrected version where I had to fix it for packages with subdirs like uclibc ...
04:54 < CIA-10> mnemoc * r5638 /trunk/package/games/pingus/pingus.desc: * changed pingus [D] to use new svn:// schema
04:54 < mnemoc> dietlibc has subdirs too, don't blame me :)
04:55 < rxr> no - I did not
04:55 < rxr> I just said I tested it ... w/ it ...
04:55 < rxr> no pun intended ...
04:56 < mnemoc> forgiven
04:58 < mnemoc> CIA-10!
04:59 < CIA-10> mnemoc * r5639 /trunk/package/gnome2/gtksourceview-sharp/ (. gtksourceview-sharp.conf gtksourceview-sharp.desc): * added gtksourceview-sharp (r38552)
04:59 < mnemoc> thanks
04:59 < mnemoc> rxr: can you add that to your reference and see if monodevelop builds?
04:59 < rxr> ouhm
04:59 < rxr> tomorrow
05:00 < mnemoc> sure
05:00 < rxr> I switched my athlon off - to save energy (and thsu money) - in favour to blasting the dust out of the U30 ....
05:00 < mnemoc> :)
05:07 < mnemoc> i have read every diff after the last good 5-glibc i knew, without finding some to blame.
05:08 < rxr> ouhm
05:08 < rxr> I just notice md5sum is written by Ullrich Drepper
05:08 < rxr> seems we have to switch to some other ;-)
05:09 < mnemoc> we have to learn to live with his software
05:10 < rxr> rsync daemon stopped ...
05:10 < rxr> does the minimal work for you ?
05:11 < mnemoc> it finished when i was already at home
05:11 < mnemoc> tomorrow i'll test it
05:11 < rxr> ah - ok
05:11 < mnemoc> but i'm worried about that glibc thing
05:11 < rxr> the most intruisive change in this area was the fl_wrapper cleanup
05:13 < mnemoc> only .c.sh to sync .c
05:13 < mnemoc> .c output is exactly the same than before
05:13 < rxr> and the ... thing
05:13 < mnemoc> that's older
05:13 < mnemoc> uhm
05:13 < mnemoc> i'll test without ...
05:15 < rxr> 0-binutils takes 25 minutes on the U30 ...
05:15 < mnemoc> oh
05:15 < mnemoc> test without ... started
05:16 < rxr> me in bed - cu
05:17 < rxr> n8 all
05:17 < mnemoc> LFS does mesure using bintuils time?
05:17 < mnemoc> n8 rxr
05:43 < mnemoc> rxr: rebuilding glibc without ... on fl_wrapper didn't help. i started a fresh bootdisk without it now
05:43 < mnemoc> gn8 *
05:47 < _Ragnar_> *grr* *kicks ncurses*
07:29 -!- madtux [~mike@196.40.44.28] has joined #t2
07:30 < madtux> 'lo
07:36 -!- madtux [~mike@196.40.44.28] has quit ["leaving"]
07:52 -!- ideal [~idealm@211.100.227.107] has quit [Remote closed the connection]
07:54 -!- mipe [~mipe@dsl10040.japo.fi] has joined #t2
09:57 -!- mnemoc [~amery@200.75.27.34] has quit [Read error: 113 (No route to host)]
10:00 -!- mnemoc [~amery@200.75.27.2] has joined #t2
10:01 < nzg> moin
11:05 < mipe> anyone awake?
11:32 -!- ideal [~idealm@211.100.227.107] has joined #t2
12:36 < rxr> moin
12:37 < CIA-10> rene * r5640 /trunk/package/base/uclibc/uclibc.desc: * fixed uclibc NOPARALLEL _FLAG_
12:40 < mipe> rxr:ltrace failed to build,but it solved by upgrading it, and libelf needs to be built before ltrace
12:41 < mipe> btw. how are build priorities handled. is it build in list order at packages file?
12:41 < rxr> mipe: so your x86-64 build is still on track?
12:41 < rxr> mipe: yes - linear build order
12:41 < mipe> rxr:well it failed to build at parted havent yet checked the error at it
12:42 < rxr> we plan to get rid ot if for 2.(1+x)
12:42 < rxr> are you sure libelf needs ltrace ?
12:42 < mipe> yeah,it searches for gelf.h file which libelf provides?
12:44 < mipe> well atleast filelist from libelf build says that it has it
12:45 < mipe> damm:parted error is:/usr/lib/gcc/x86_64-unknown-linux-gnu/3.4.3/../../../../lib/libuuid.a: could not
12:45 < mipe> read symbols: Bad value
12:45 < mipe> ah -fPIC is needed
12:45 < rxr> yes on the e2fsprogs package
12:45 < rxr> I add those
12:46 < mipe> i really need understand conf files and stuff so i dont need to bother you with silly things :)
12:47 < CIA-10> rene * r5641 /trunk/package/filesystem/e2fsprogs/e2fsprogs.conf: * fixed e2fsprogs for x86-64
12:51 < mipe> yeah,parted now builds
12:53 < rxr> to what ltrace version did you update to ?
12:54 < mipe> 0.3.36
12:57 < rxr> mnemoc: btw. there is the xfce -final release ;-)
13:03 < rxr> mipe: the .orig tarball?
13:04 < rxr> or did you patch the .2 patch in, too ?
13:05 < rxr> ok - the -2 only contains debian junk ...
13:08 < rxr> done
13:08 < CIA-10> rene * r5642 /trunk/package/base/ltrace/ltrace.desc:
13:08 < CIA-10> * updated powerpc (0.3.31 -> 0.3.36) - should fix x86-64
13:08 < CIA-10> * marked it powerpc since it just "accidently" built here ;-)
13:09 < mipe> rxr:orig tarball,didnt use patch
13:10 < rxr> erhm - what commit message did I inect there ..
13:10 < rxr> urgs - updated powerpc ... *loL*
13:10 < rxr> I should better wake up ..
13:10 < mipe> :)
13:10 < mipe> just woke up?
13:14 < CIA-10> rene * r5643 /trunk/package/base/ (libelf/libelf.desc ltrace/ltrace.cache):
13:14 < CIA-10> * fixed the libelf [P] to build before ltrace, updated the ltrace.cache
13:14 < CIA-10> to list the new libelf dependency
13:14 < rxr> yeah - sort of ...
13:19 < rxr> valentin: valentin !
13:19 < rxr> yeah
13:19 < rxr> I can not belive this
13:19 < rxr> yeah
13:19 < rxr> grat
13:20 < rxr> T2:HEAD builds a non crashing dynamic linker out of glibc on sparc64 !!!
13:20 < rxr> # chroot . ./lib/libc-2.3.4.so
13:20 < rxr> GNU C Library stable release version 2.3.4, by Roland McGrath et al.
13:20 < rxr> Compiled by GNU CC version 3.4.3.
13:20 < rxr> Compiled on a Linux 2.6.8.1-dist system on 2005-01-18.
13:21 < rxr> # file lib/libc-2.3.4.so
13:21 < rxr> lib/libc-2.3.4.so: ELF 64-bit MSB shared object, SPARC V9, version 1 (SYSV), stripped
13:21 < rxr> the only build error so far is 1-dietlibc
13:21 < rxr> which is no wonder when it uses uname to get the current system type and then wants to build a 64 bit sparc binary with a 32bit in-system toolchain ..
13:22 < rxr> I'll fix this one and for all by removing this 2 dietlibc building from the cross dietlibc ..
13:22 < rxr> after showering and breakfast that is ..
13:28 * rxr happy and off under the shower - cu
13:31 < mipe> nice,5-glibc segfaults
13:32 < mipe> and linux26 fails because of nvidia build
13:43 < rxr> the 5-glibc is know
13:43 < rxr> ignore it for know
13:43 < rxr> now even
13:44 < rxr> it fails on any arch right now - not x86-64 specific
13:44 < rxr> hm - what to do for nvidia - package the x86-64 binary, too
13:44 < rxr> or just disable it for x86-64 ?
13:47 < mipe> error for linux24:
13:47 < mipe> ld -r -o nv-linux.o nv.o nv-vm.o os-agp.o os-interface.o os-registry.o
13:47 < mipe> ld -r -o nvidia.o nv-linux.o nv-kernel.o
13:47 < mipe> ld: Relocatable linking with relocations from format elf32-i386 (nv-kernel.o) to
13:47 < mipe> format elf64-x86-64 (nvidia.o) is not supported
13:47 < mipe> when its building nvidia
13:47 < rxr> yes I imagined
13:47 < mipe> {standard input}:539: Error: suffix or operands invalid for `mov'
13:48 < mipe> for linux26 when building nvidia
13:48 < rxr> we only have the 32bit nvidia binary right now ..
13:48 < mipe> yeah,disable it for now?
13:48 < rxr> sure
13:48 * rxr doing
13:51 < mipe> well theres diffrent source package for x86_64 for nvidia. have to test it sometime. or is there a easy way to put diffrent [D] tag for x86_64?
13:51 < rxr> no - not yet
13:51 < rxr> we need to think if we want this ...
13:51 < rxr> cou can just add a 2nd and case $arch_machine in ... in the package .conf ...
13:55 < CIA-10> rene * r5644 /trunk/package/x11/nvidia/ (nvidia.conf postlinux.conf):
13:55 < CIA-10> * fixed x11/nvidia to only install stuff on x86 - that is also no
13:55 < CIA-10> x86-64 (for now)
13:55 < rxr> ^- untested - please test
13:59 < mnemoc> moin
14:00 < mnemoc> rxr: yes, i saw xfce4.2.0 but i need to build it before commiting ;)
14:00 < mnemoc> rxr: same segfault on a fresh build without '...' on fl_wrapper.... i'll try without FLWRAPPER_LIBC
14:04 < mnemoc> rxr: if FLWRAPPER_LIBC did this i can replace it with a #define
14:24 < jsaw> re
14:25 < jsaw> did anybody read clifford's comment on the dietlibc mailing list?
14:25 < mnemoc> paste paste paste
14:25 < daja77> ?
14:25 < jsaw> there was some flaming going on
14:26 < daja77> clifford flaming fefe?
14:26 < daja77> url?
14:26 < jsaw> a guy flamed at Felix in german language..
14:26 < jsaw> and cliff commented it
14:26 < jsaw> ok, you can't resist sending this to the list instead of sending it to the
14:26 < jsaw> list owner (fefe). but you should know that sending german off-topic flames
14:26 < jsaw> to an english mailing list might have a negative impact on your karma..
14:26 < jsaw> do you always use that kind of style for your bugreports? if so, please
14:26 < jsaw> don't even try to participate in any of my projects.
14:27 < daja77> hehe
14:27 < daja77> sounds like a valid claim
14:29 < mnemoc> :)
14:32 < jsaw> <- doesn't like world police
14:32 < jsaw> mnemoc: is ncurses building over there?
14:32 < mnemoc> it builds, but kills bash
14:33 < jsaw> I have !> : undefined reference to `__ctype_b'
14:33 < rxr> uclibc?
14:33 < jsaw> yep. uboot target.
14:33 < jsaw> should I start from scratch?
14:34 < rxr> nope
14:34 < rxr> that seems to be a general problem with uclibc
14:34 < jsaw> ?
14:34 < rxr> _Ragnar_ posted it here, too
14:34 < rxr> the same build error
14:34 < mnemoc> i never got that
14:34 < rxr> without uboot
14:34 < jsaw> oh seems like I missed something
14:34 < rxr> generic + uclibc
14:34 < jsaw> but I have uboot
14:35 < rxr> which is also uclibc ...
14:35 < mnemoc> yes, i understand that
14:35 < jsaw> any idea, rxr?
14:35 < mnemoc> but ncurses builds on stage-1 and my tests didn't fail there
14:35 -!- sparc-kly [~mubex@65-23-204-29.prtc.net] has quit [Read error: 54 (Connection reset by peer)]
14:36 < mnemoc> i had some problems with #include_next, maybe that's the reason of that undefined
14:37 < rxr> jsaw: no - no idea, I have not looked into uclibc that recently ...
14:37 < jsaw> it's a undefined symbol while linking...
14:37 < jsaw> :(
14:37 < mnemoc> :\
14:37 < mnemoc> sorry
14:38 < jsaw> oh, defined in locale.h -eehhh?-
14:40 < jsaw> #ifdef __UCLIBC_HAS_XLOCALE__
14:40 < jsaw> const __ctype_mask_t *__ctype_b;
14:40 < jsaw> ...
14:41 < mnemoc> uhm
14:44 < mnemoc> jsaw,rxr: FLWRAPPER_LIBC was responsable of 5-glibc segfault, do you see any problem if i change that to a #define ?
14:45 < rxr> mnemoc: can you paste the revision
14:45 < rxr> I need to reread that diff ...
14:45 < mnemoc> FLWRAPPER_LIBC?
14:45 < rxr> yes
14:47 < mnemoc> svn diff -r5406:5407
14:47 < rxr> thanks
14:50 < jsaw> mnemoc: how did it segfault?
14:50 < rxr> jsaw: I have not usable backtrace
14:50 < jsaw> maybe getenv not available...?
14:50 < rxr> the first program during the glibc rebuild in stage5 that is run with the fresh glibc segfault
14:51 < mnemoc> == 01/18/05 10:50:23 =[5]=> Finished building package glibc.
14:51 < jsaw> mnemoc: yes, change it to a define!!!
14:51 < mnemoc> ok
14:51 < jsaw> most probably getenv related
14:53 < mnemoc> yep
14:53 < rxr> I somehow can not believe that ...
14:54 < jsaw> you do dlopen to get symbols from the libc... which means, libc has to be loaded already -- isn't it?
14:54 < jsaw> for getenv...I mean
14:54 < mnemoc> :D
14:55 < jsaw> mnemoc: commited?
14:55 < mnemoc> yes
14:55 < CIA-10> mnemoc * r5645 /trunk/misc/tools-source/ (fl_wrapper.c fl_wrapper.c.sh): * changed FLWRAPPER_LIBC to be a #define instead of an env var (5-glibc was segfaulting)
14:58 * mnemoc testing new Build-{Tools,Pkg} and commit the rest
14:59 < rxr> I know why the prev. code segfaulted
14:59 < rxr> you passed the absolute path into it, right ?
14:59 < mnemoc> arg
14:59 < mnemoc> no, absolute path was never passed
15:00 < mnemoc> mom, typo
15:02 < CIA-10> mnemoc * r5646 /trunk/misc/tools-source/ (fl_wrapper.c fl_wrapper.c.sh): * fixed just introduced typo on fl_wrapper.c
15:07 < jsaw> gotta go, cu later
15:09 < rxr> cu
15:11 < mnemoc> cu jsaw
15:11 < CIA-10> mnemoc * r5647 /trunk/scripts/ (Build-Pkg Build-Tools): * adapted Build-Tools and Build-Pkg to FLWRAPPER_LIBC as #define
15:12 < mnemoc> rxr: can you see if that -D is properly given? my brain is not working. at least it builds ;)
15:15 < rxr> mom
15:28 -!- ideal [~idealm@211.100.227.107] has quit ["Leaving"]
15:34 < mnemoc> fsck! segfault
15:34 < mnemoc> rxr: what did you wanted to say about absolute path?
15:37 < mnemoc> the difference i see between #define and my previous test was that i ignored previous (static) value of libc_handle, i always opened on my succesful test
15:44 < rxr> hm
15:47 < rxr> can you add a printf before the dlopen and otuput what FLWRAPPER_LIBC
15:47 < rxr> is set to ?
15:47 < rxr> (if printf is available at that time ...=
15:47 < rxr> )
15:48 < rxr> cu a bit later
15:48 < mnemoc> we have lot of fprintf(stderr,..) there
15:49 < rxr> maybe even an error happens and the printf crashes ...
15:49 < rxr> no idea if it is valid to call ad that preload time ..
15:49 < rxr> try to obtain the LIBC opened ...
15:49 < mnemoc> strace
15:49 < rxr> I somehow have the feeling it is the wrong one ..
15:49 < rxr> yeah - or strace
15:52 < mnemoc> open("/lib/librt.so.1", O_RDONLY) = 3 <0.000035>
15:52 < mnemoc> uhm
15:53 < rxr> and more ?
15:53 < mnemoc> go, i'll paste both (with and without PRELOAD) somewhere
15:53 < rxr> ok
16:03 < mnemoc> http://t2.geeks.cl/~amery/ <-- wopl (without preload), wpl (with preload) ... PIDs hacked to be diffeable
16:22 < mipe> 1384 builds total, 256 completed fine, 15 with errors.
16:23 < mnemoc> arch?
16:23 < mipe> x86_64
16:23 < mnemoc> what 'failure tolerance' did you set?
16:24 < mipe> and where do i set it?
16:24 < mipe> :)
16:24 < mnemoc> Config
16:24 < mipe> didnt notice anything like that
16:25 < mipe> in there
16:25 < mnemoc> export ROCKCFG_ABORT_ON_ERROR_AFTER='4'
16:26 < mnemoc> you have 15 errors on low stages, that's bad
16:26 < mipe> on stage 5
16:27 < mipe> i mean i'm building on stage 5. and its set to 4
16:27 < mnemoc> yes
16:27 < mnemoc> that's ok
16:27 < mnemoc> the value is ok, the amount of errors no :)
16:28 < mipe> :)
16:45 < mnemoc> http://svn.exactcode.de/big.png <-- looking good :)
16:48 < mipe> ash,cdparanoia,lprng,gnupth there packages only work if parallel is set to 1,so add NOPARALLEL flag?
16:48 < mipe> there=these
16:48 < mnemoc> yes
16:48 < mnemoc> please
16:50 < mipe> i mean someone with svn access :)
16:50 < mnemoc> mail rene to get access ;)
16:51 < mipe> or i could make a patch and send it to ml,after i'm tested others too. is svn diff -r HEAD enough for the patch?
16:51 < mnemoc> yes
16:51 < mnemoc> and a comment :)
16:51 < mipe> :)
16:52 < mnemoc> Name :
16:52 < mnemoc> * blah
16:52 < mnemoc> * blah
16:52 < mnemoc>
16:53 < mnemoc> or $name <$mail>:\n\t* blah\n\t* blah\n\n if your prefer
17:47 -!- sparc-kly [~mubex@65-23-193-95.prtc.net] has joined #t2
17:56 < rxr> re
17:57 < mnemoc> rxr: it was readlink
17:57 < mnemoc> i can't understand why, but that seems to produce the segfault
17:58 < mnemoc> make -C resolv subdir_install
17:58 < mnemoc> make[1]: *** [resolv/subdir_install] Segmentation fault
17:58 < rxr> aha
17:58 < mnemoc> FSCK!!!
18:02 < mnemoc> - FLWRAPPER_LIBC=$( readlink -f `echo $xroot/lib/libc.*` )
18:02 < mnemoc> + #FLWRAPPER_LIBC=$( readlink -f `echo $xroot/lib/libc.*` )
18:02 < mnemoc> + FLWRAPPER_LIBC=`echo $xroot/lib/libc.*`
18:02 < mnemoc> this moved the segfault from rpcgen to make -C resolv subdir_install
18:02 < mnemoc> i can't understand this
18:03 < rxr> and you are really sure that when you patch the conditional away completely it does isntall the glibc just fine
18:03 < rxr> ?
18:04 < mnemoc> i'll re-test that
18:04 < mnemoc> glibc takes ages to segfault :\
18:05 < mnemoc> i'll re-revert glibc update again (on local copy)
18:05 < rxr> yep
18:07 < mnemoc> 5-glibc started without defininf FLWRAPPER_LIBC and with downdated glibc
18:08 < rxr> why downdated
18:08 < rxr> the same is more interesting ...
18:08 < mnemoc> ok
18:08 < mnemoc> un-reverting downdate
18:09 < mnemoc> == 14:09:25 =[5]=> Building base/glibc [2.3.4-2004-12-24 2.1.0-beta3].
18:11 < mipe> huoh,gcc cant produce 32bit code
18:13 < _Ragnar_> *grrr*
18:14 < _Ragnar_> how am I supposed to compile anything without even an editor or swapon or anything?:P
18:14 < rxr> nvi should be there
18:14 < rxr> there is no swapon ?
18:14 < mnemoc> you can edit with sed :D
18:15 < _Ragnar_> ;p
18:15 < CIA-10> mnemoc * r5648 /trunk/scripts/Build-Pkg: * improved Build-Pkg to abort immediately if /usr/bin/time crashed
18:15 < _Ragnar_> nvi complains about not being able to open terminal: linux
18:15 < rxr> what the heck ?
18:15 < rxr> ^- mnemoc
18:16 < mnemoc> a sanity check
18:16 < mnemoc> complains?
18:16 < mnemoc> objections?
18:16 < mnemoc> comments?
18:16 < _Ragnar_> oh, found swapon
18:17 < mnemoc> :D
18:17 < _Ragnar_> scripts/config.func: line 194: /dev/fd/62: No such file or directory
18:17 < mnemoc> mount /dev
18:17 < _Ragnar_> it's mounted
18:18 < mnemoc> ln -s /proc/self/fd /dev/
18:18 < rxr> ack
18:18 < rxr> devfsd does do this on startup ...
18:19 < mnemoc> isn't this an evil restriction?
18:19 < _Ragnar_> and thats not running, since no init
18:19 < mnemoc> we should use /proc/self/fd directly, imho
18:20 < mnemoc> rxr: i think ncurses is not ready for stage-1 :)
18:20 < mnemoc> or i should improve uclibc
18:22 < rxr> disable ncurses for stage 1 and nvi, too
18:23 < mnemoc> i did :p
18:23 < mnemoc> but i guess it will fail on next stage too ;) if i reach 'next stage' some day
18:24 < _Ragnar_> scripts/basic/fixdep.c: In function 'use_config':
18:25 < _Ragnar_> scripts/basic/fixdep.c:193: error: 'PATH_MAX' undeclared ...
18:25 < _Ragnar_> (for linux-kernel)
18:25 < _Ragnar_> oops
18:25 < _Ragnar_> linux-header
18:25 < mnemoc> uhm? uclibc?
18:25 < _Ragnar_> no
18:25 < _Ragnar_> that's the 64-cross-compile
18:26 < mnemoc> i got the same on my p4 cross-base
18:26 < mnemoc> that's why local cpp was considered not sane by gcc
18:27 < mnemoc> rxr: ^--- read that? ;)
18:28 < _Ragnar_> (it is defined in /usr/include/linux/limits.h)
18:30 < mnemoc> #include_next on /usr/include/limits.h should include it
18:31 < mnemoc> i was hunting too ;)
18:31 < _Ragnar_> but why isn't it included?
18:32 < _Ragnar_> ah;) Build-Pkg at least built less
18:32 < mnemoc> :)
18:33 < _Ragnar_> so gcc/cpp is okay
18:34 < _Ragnar_> hmm nvi seems to wanna read terminfo from /export/t2-trunk-build/a64*/usr/share/terminfo/l/linux"
18:34 < _Ragnar_> which of course isn't present on the target system
18:35 < CIA-10> mnemoc * r5649 /trunk/package/base/uclibc/config-650.in: * fixed uclibc config to not build ncurses or nvi on stage 1 if uclibc is enabled
18:37 < _Ragnar_> well I have to attack the 64bit cross compile later, I need to head to work :/
18:37 < mnemoc> payed work is bad for your soul :)
18:38 < daja77> good for the purse
18:38 < mnemoc> :)
18:38 < _Ragnar_> *g*
18:38 < mnemoc> and for the stomach
18:42 < mnemoc> rxr: don't forget to re-build broken packages on reference with last updates (of course not re-schedule glibc ;-)
18:44 < mnemoc> rxr: why libc.so.6 doesn't have a path on fl_wrapper?
18:45 * mnemoc moving home to eat something
19:09 < rxr> ?
19:12 -!- mipe [~mipe@dsl10040.japo.fi] has quit ["su"]
19:20 < mnemoc> ? ?
19:21 < mnemoc> == 14:09:25 =[5]=> Building base/glibc [2.3.4-2004-12-24 2.1.0-beta3].
19:21 < mnemoc> == 01/18/05 14:53:49 =[5]=> Finished building package glibc.
19:22 < mnemoc> without -DFLWRAPPER_LIB=...
19:22 < rxr> that is without all of the dynamic WRAPPER code, right ?
19:26 < CIA-10> rene * r5650 /trunk/package/base/findutils/4.2.10-enable-debug.patch:
19:26 < CIA-10> * fixed the findutils/4.2.10-enable-debug.patch to not depend on an
19:26 < CIA-10> installed aclocal-1.9 and friends
19:32 < mnemoc> rxr: FLWRAPPER lines on Build-Tools were removed
19:32 < _Ragnar_> btw: (related to trying to build the uclibc 32 bit system, for which I selected to compile executables statically)
19:32 < _Ragnar_> i686-unknown-linux-uclibc-gcc -fPIC -O3 -DUSE_MMAP -o example example.o -L. libz.so.1.2.1
19:32 < mnemoc> tha'ts the only diff with HEAD
19:32 < _Ragnar_> /export/t2-trunk/build/a32-2.1.0-beta3-x86-athlon-xp-32-cross-generic-expert/TOOLCHAIN/tools.cross/crosscc/../lib/gcc/i686-unknown-linux-uclibc/3.4.3/../../../../i686-unknown-linux-uclibc/bin/ld: attempted static link of dynamic object `libz.so.1.2.1'
19:33 < mnemoc> uhm?
19:33 < mnemoc> building what?
19:33 < _Ragnar_> 1-zlib
19:34 < rxr> seems zlib needs some fixes for static builds ...
19:35 -!- sparc-kly [~mubex@65-23-193-95.prtc.net] has quit [Read error: 104 (Connection reset by peer)]
19:43 < _Ragnar_> var_insert confopt " " "--shared" <-- that should be omitted for static builds
19:43 < mnemoc> *G*
19:44 < mnemoc> [ "$ROCKCFG_STATIC" == 0 ] &&
19:46 < _Ragnar_> *g*
19:48 < _Ragnar_> uhm what's it doing now ...
19:49 < _Ragnar_> i686-unknown-linux-uclibc-gcc -O3 -DUSE_MMAP -c -o minigzip.o minigzip.c
19:49 < _Ragnar_> libz.a adler32.o compress.o crc32.o gzio.o uncompr.o deflate.o trees.o zutil.o inflate.o infback.o inftrees.o inffast.o
19:49 < _Ragnar_> make: libz.a: Command not found
19:49 < mnemoc> o_O
19:49 < mnemoc> a missing \ on Makefile?
19:50 < _Ragnar_> seems like it's lacking ar
19:50 < mnemoc> rxr: i'll try with a hardcoded -DFLWRAPPER_LIBC=\"libc.so.6\"
19:52 < mnemoc> == 15:51:25 =[5]=> Building base/glibc [2.3.4-2004-12-24 2.1.0-beta3].
19:52 < mnemoc> .oO( this debugging is weird )o
19:53 < mnemoc> _Ragnar_: $(AR) ?
19:55 < _Ragnar_> yeah
19:55 < _Ragnar_> in the Makefile: AR=ar rc
19:56 < _Ragnar_> so why's that empty later on?
19:58 < mnemoc> does 1-zlib.err shows AR=ar on make command line?
19:59 < _Ragnar_> ah, there's an AR= there
20:00 < mnemoc> :D
20:00 < mnemoc> brb
20:07 < _Ragnar_> okay now it works ;p
20:09 < _Ragnar_> should I make a patch?
20:14 < rxr> sure
20:20 < _Ragnar_> http://66.146.166.62/~loki/zlib.patch
20:28 < mnemoc> re
20:29 < mnemoc> _Ragnar_: use diff -u to make your patches
20:30 < _Ragnar_> kk
20:31 * _Ragnar_ moves on to curl
20:33 < mnemoc> and the patch? ;)
20:35 < _Ragnar_> bllll;p
20:36 < _Ragnar_> refresh the url
20:36 < mnemoc> thanks :)
20:39 < mnemoc> _Ragnar_: why hook_add preconf 3 "export AR='ar rc'" instead of var_append makeopt ' ' 'AR="ar rc"' ?
20:39 < rxr> == 01/18/05 20:38:58 =[1]=> Finished building package silo.
20:40 < _Ragnar_> *shrugs*
20:41 < rxr> ?
20:41 < _Ragnar_> -hook_add preconf 3 "unset AR" <- modifying that
20:42 < mnemoc> did you tryed just commenting it?
20:42 < mnemoc> tried*
20:42 < mnemoc> rxr: == 01/18/05 16:38:48 =[5]=> Finished building package glibc.
20:42 < _Ragnar_> adding to the very generous commenting that's already present in the file?
20:42 < _Ragnar_> ;)
20:43 < rxr> mnemoc: which permutation is that ?
20:43 < mnemoc> i meaned # :)
20:43 < mnemoc> - -DFLWRAPPER_LIBC=\"${FLWRAPPER_LIBC##*/}\" \
20:43 < mnemoc> + -DFLWRAPPER_LIBC=\"libc.so.6\" \
20:43 < mnemoc> that worked
20:43 < rxr> ic
20:44 < mnemoc> i don't know why ${FLWRAPPER_LIBC##*/} != libc.so.6
20:44 < rxr> i'm in a 64bit UltraSPARC system right now ...
20:45 < rxr> GNU bash, version 3.00.16(2)-release (sparc64-unknown-linux-gnu)
20:45 < mnemoc> what about jsaw's 17?
20:45 < rxr> but one of my glibc hacks is not effective ... :-(
20:45 < rxr> mnemoc: he removed the patchlevel patching again ...
20:45 < mnemoc> :\
20:46 < rxr> it's no official bash patch and will cause confusion
20:46 < mnemoc> 3.00.16-jsaw ;)
20:47 < rxr> yep
20:47 < rxr> hm - for some strane reason my glibc lib64 patch is not effective ...
20:50 < _Ragnar_> http://66.146.166.62/~loki/zlib.patch & http://66.146.166.62/~loki/curl.patch
20:50 < _Ragnar_> (last try)
20:52 < mnemoc> _Ragnar_: did you tested _last_ zlib's?
20:52 < mnemoc> curl.patch is blank
20:53 < _Ragnar_> reload curls
20:54 < _Ragnar_> define 'tested'
20:54 < mnemoc> used to build and builds
20:55 < mnemoc> built*
20:55 < _Ragnar_> builds for my system with uclibc and static execs
20:55 < _Ragnar_> did not build without the patch
20:55 < mnemoc> good
20:56 < _Ragnar_> otherwise it would be kinda pointless, wouldn't it?
20:57 < mnemoc> yep :)
21:00 < _Ragnar_> e2fsprogs:
21:00 < _Ragnar_> if [ $stagelevel -gt 1 -o "$crossnative" == "cross" ] ; then var_append confopt " " "--enable-elf-shlibs"
21:00 < _Ragnar_> fi
21:01 < _Ragnar_> obviously not gonna work on a static system either
21:17 < _Ragnar_> http://66.146.166.62/~loki/e2fsprogs.patch
21:18 < mnemoc> rxr: http://www.geeks.cl/~amery/flwrapper_debug_uclibc.txt <--- this happens with uclibc but without readlink, fopen in same minute resolves to nil
21:18 < _Ragnar_> .oO(but if no one bothers I'll just stop)
21:18 < mnemoc> ?
21:19 < mnemoc> what do you mean?
21:20 < _Ragnar_> nevermind, the crossbuild-environment finished
21:21 < mnemoc> :)
21:22 < mnemoc> can you try to rebuild 1-zlib without var_append makeopt ' ' 'AR="ar rc"' ?, just removing the unset
21:22 < rxr> http://developer.apple.com/documentation/Cocoa/Conceptual/GeometryandRange/index.html#//apple_ref/doc/uid/10000046i
21:23 * mnemoc has never seem obj-c code
21:23 < mnemoc> seen*
21:23 < _Ragnar_> I tried that before, it failed
21:23 < mnemoc> ok
21:23 < mnemoc> do you have a name and an email address for the changelog?
21:25 < _Ragnar_> Roland Schulz
21:26 < mnemoc> thanks :)
21:26 < mnemoc> curl and zlib will conflict when you update due to indentation changes, but the rest will be identical
21:27 < CIA-10> mnemoc * r5651 /trunk/package/ (3 files in 3 dirs):
21:27 < CIA-10> Roland Schulz :
21:27 < CIA-10> * fixed zlib, e2fsprogs and curl to build when static is choosen on Config
21:28 < _Ragnar_> you're welcome
21:29 < _Ragnar_> probably going to be a few more for the whole build on the target
21:29 < mnemoc> will you build a native uclibc build?
21:30 < _Ragnar_> yeah
21:31 < _Ragnar_> well the main reason is to get a static system, but that seems to be about impossible with glibc
21:31 < mnemoc> zlib on non-static --> outch!
21:31 < mnemoc> ===[zlib_pm:28 (last $?=0)> make libz.a
21:31 < mnemoc> i586-unknown-linux-uclibc-ar libz.a adler32.o compress.o crc32.o gzio.o uncompr.o deflate.o trees.o zutil.o inflate.o infback
21:31 < mnemoc> .o inftrees.o inffast.o
21:31 < mnemoc> i586-unknown-linux-uclibc-ar: illegal option -- z
21:31 < _Ragnar_> oops
21:31 < mnemoc> i'll unset it on non-static, and set it as you did on static
21:32 < _Ragnar_> okay
21:34 < mnemoc> == 01/18/05 17:34:38 =[1]=> Finished building package zlib.
21:34 -!- praenti [~praenti@82.139.198.99] has quit [Read error: 104 (Connection reset by peer)]
21:34 < _Ragnar_> why does it do the zlib_pm stuff anyway? on reading the zlib Makefile, it already installs the .a, doesn't it?
21:35 -!- praenti [~praenti@82.139.198.99] has joined #t2
21:36 < mnemoc> it may be legacy, .conf life is much longer than releases ;)
21:36 < CIA-10> mnemoc * r5652 /trunk/package/archiver/zlib/zlib.conf: * fixed zlib to unset AR on dynamic and set it on static
21:37 < _Ragnar_> ;)
21:39 < mnemoc> rxr: == 01/18/05 17:39:56 =[5]=> Finished building package glibc.
21:41 < mnemoc> rxr: look here:
21:41 < mnemoc> fl_wrapper.so debug [31869]: Symbol 'open' in libc (0x804af90) has been resolved to 0x400421c0.
21:41 < mnemoc> fl_wrapper.so debug [31869]: Symbol 'fopen' in libc (0x804af90) has been resolved to (nil).
21:42 < rxr> this is with glibc ?
21:42 < mnemoc> no, uclibc
21:42 < mnemoc> entering stage-2
21:44 < _Ragnar_> oops
21:44 < _Ragnar_> in 00-dirtree on the target system:
21:44 < _Ragnar_> ===[add_flist:348 (last $?=0)> fl_wrparse -D -r /t2-trunk/build/a32-2.1.0-beta3-x86-athlon-xp-32-cross-generic-expert/TOOLCHAIN/tools.cross/
21:44 < _Ragnar_> Segmentation fault (core dumped)
21:44 < mnemoc> uhm
21:45 < mnemoc> fl_wrparse code is at misc/tools-source if you want to debug :)
21:46 < _Ragnar_> *lol*
21:46 < mnemoc> you have a beatiful core file ;)
21:47 < rxr> [ 943 active packages ] <- I need to put the desktop target on a diet soon
21:48 < mnemoc> :)
21:49 < _Ragnar_> =)
21:50 < _Ragnar_> oh, on the target system I have to set ROCKCFG_CROSSBUILD='0' ?
21:50 < mnemoc> yep
21:50 < mnemoc> it's a native build
21:51 < _Ragnar_> okay
21:51 < _Ragnar_> (didn't change anything tho
21:51 < mnemoc> but nobody has ever tested building t2/rock over a uclibc system :)
21:51 < _Ragnar_> *hrhr*
21:51 < _Ragnar_> I've noticed
21:55 < mnemoc> what a WEAK symbol is?
21:56 < rxr> weak defines how it is bound on link time
21:56 < rxr> IIRC WEAK may be redifned or overwritten or so ..
21:59 < _Ragnar_> how do I compile the fl_wrparse with symbols?
21:59 < _Ragnar_> ROCKCFG_DEBUG doesn't do the trick
21:59 < mnemoc> Build-Tools
22:00 < mnemoc> there is a BUILDCC line to build fl_wrparse
22:00 < mnemoc> -g it ;)
22:04 < mnemoc> rxr: and global?
22:04 < rxr> mnemoc: ?
22:05 < _Ragnar_> uhm
22:05 < mnemoc> if weak means it can be overwritten, what global means?
22:05 < mnemoc> 819: 0001bb78 36 FUNC GLOBAL DEFAULT 7 fopen
22:05 < rxr> i do not have the details in my mind
22:05 < _Ragnar_> FLWRAPPER_LIBC=$( readlink -f `echo $xroot/lib/libc.*` ) <-- so 'take the first libc.* you find and hope it's the right one?'
22:05 < mnemoc> 1192: 000351c0 71 FUNC WEAK DEFAULT 7 open
22:05 < mnemoc> _Ragnar_: you may have only one
22:06 < mnemoc> rxr: open, the weak, works... but fopen, the global fails
22:07 < rxr> hm - what are they in glibc ?
22:09 < mnemoc> 1948: 000a5d50 124 FUNC WEAK DEFAULT 11 open@@GLIBC_2.0
22:09 < mnemoc> 1238: 00058f34 105 FUNC GLOBAL DEFAULT 11 fopen@GLIBC_2.0
22:10 < _Ragnar_> ah I see ... uclibc has libc.so.0, glibc has libc.so.6 (and I need glibc in this case for running gdb)
22:11 < mnemoc> why glibc?
22:12 < mnemoc> rxr: more spam on t2@ec.de
22:12 < _Ragnar_> for running gdb
22:12 < mnemoc> why not uclibc?
22:14 < _Ragnar_> cuz I can't get to compile anything with uclibc yet, since fl_wrparse isn't working?
22:14 < rxr> mnemoc: ok - in the next free minute I migrate to spamassasin and tune the rules again ...
22:19 < _Ragnar_> #0 0x40089300 in ?? ()
22:19 < _Ragnar_> #1 0x4000a3ff in open (f=0xbfffc168 "/proc/6547/stat", a=0) at misc/tools-source/fl_wrapper.c:112
22:20 < mnemoc> where did you got gdb? ;)
22:20 < mnemoc> on misc/tools-source/fl_wrapper.c there is a DEBUG=0 line, change that
22:20 < mnemoc> i'm fighting with something very similar entering stage-2
22:21 < _Ragnar_> moving libc.so.6 around ;)
22:21 < mnemoc> outch
22:23 < _Ragnar_> fl_wrapper.so debug [9619]: begin of handle_file_access_before("open", "/proc/9619/stat", xxx)
22:23 < _Ragnar_> fl_wrapper.so debug [9619]: end of handle_file_access_before("open", "/proc/9619/stat", xxx)
22:23 < _Ragnar_> fl_wrapper.so debug [9619]: Symbol 'open' in libc (0x804afc0) has been resolved to 0x4008a300.
22:24 < _Ragnar_> fl_wrapper.so debug [9619]: going to run original open() at 0x4008a300 (wrapper is at 0x4000a650).
22:24 < _Ragnar_> Segmentation fault (core dumped)
22:24 < mnemoc> aaaarg!
22:25 < mnemoc> can you try to cross-build gdb?
22:25 < _Ragnar_> .oO(?)
22:25 < mnemoc> (add 1 to it's .desc's [P]
22:26 < _Ragnar_> I can try
22:26 < mnemoc> to get a uclibc gdb
22:28 < _Ragnar_> okay trying
22:30 * _Ragnar_ also adds debug symbols and recompiles uclibc
22:30 < mnemoc> :)
22:35 < _Ragnar_> the uclibc patch for gdb fails
22:35 < mnemoc> too hard to adapt?
22:49 < _Ragnar_> mnemoc: lot of failed hunks ... I'll try in a few
22:51 < CIA-10> mnemoc * r5653 /trunk/scripts/Build-Tools: * finally fixed 5-glibc/fl_wrapper segfault (i hope)
22:52 < mnemoc> _Ragnar_: don't patch!
22:52 < mnemoc> _Ragnar_: that should be automatized now
22:53 < mnemoc> all except this:
22:53 < mnemoc> +linux-uclibc*)
22:53 < mnemoc> 130 + lt_cv_deplibs_check_method=pass_all
22:53 < mnemoc> 131 + lt_cv_file_magic_test_file=`echo /lib/libuClibc-*.so`
22:53 < mnemoc> 132 + ;;
22:53 < mnemoc> 133 +
22:53 < mnemoc> look on bintuils patching
22:54 < mnemoc> i'll move uclibc_fix_addcases and uclibc_fix_addcases2 to parse-config so you can use it on patches/gbd.conf as i do on bintuils
22:56 < mnemoc> _Ragnar_: alive?
22:58 < mnemoc> now i can't see the difference between uclibc_fix_configsub and uclibc_fix_addcases
22:59 < _Ragnar_> huh?
23:00 < mnemoc> that was a brain dump, did you understood something?
23:00 -!- praenti [~praenti@82.139.198.99] has quit [Read error: 104 (Connection reset by peer)]
23:00 < _Ragnar_> no not really
23:00 -!- praenti [~praenti@82.139.198.99] has joined #t2
23:00 < mnemoc> mom
23:01 < mnemoc> gdb patch does lot of:
23:01 < mnemoc> - alpha*-*-linux-gnu* | alpha*-*-elf*)
23:01 < mnemoc> 9 + alpha*-*-linux-gnu* | alpha*-*-linux-uclibc* | alpha*-*-elf*)
23:01 < mnemoc> that is automaticed
23:02 < mnemoc> the only real thing to patch is:
23:02 < mnemoc> +linux-uclibc*)
23:02 < mnemoc> + lt_cv_deplibs_check_method=pass_all
23:02 < mnemoc> + lt_cv_file_magic_test_file=`echo /lib/libuClibc-*.so`
23:02 < mnemoc> + ;;
23:02 < mnemoc> +
23:02 < _Ragnar_> should I delete the rest of the patch?
23:03 < mnemoc> yes, _but_ replace it with hacks like those done on patches/bintuils.conf
23:04 < mnemoc> .patch patching is very sensitive to updates
23:05 < mnemoc> i would love to unify uclibc_fix_configsub, uclibc_fix_addcases and uclibc_fix_addcases2 on one single sed function, but my brain is not clear enough
23:05 < mnemoc> i don't even see the differences between uclibc_fix_configsub and uclibc_fix_addcases now :\
23:06 < _Ragnar_> *g*]
23:06 < mnemoc> move those functions from patches/binutils.conf to parse-config (bellow uclibc_fix_configsub and friends)
23:07 < _Ragnar_> who?
23:07 < mnemoc> you :)
23:08 < mnemoc> i can't test it here, you can
23:08 < mnemoc> and you want, don't you? ;)
23:08 < mnemoc> i can do it, but it will be blind patching
23:09 < mnemoc> ok, i'll do it
23:09 < _Ragnar_> *lol*
23:10 < rxr> == 01/18/05 23:09:08 =[5]=> Finished building package glibc.
23:10 < _Ragnar_> I'm not familiar enough with the build process as of now and this seems to be lots of black magic ;)
23:10 < rxr> ^- mnemoc it was not the whitespace change
23:10 < rxr> readline -> echo alone did it ...
23:10 < mnemoc> rxr: only readlink?
23:10 < rxr> think so
23:11 < mnemoc> _Ragnar_: isn't black magic nice? ;)
23:11 < _Ragnar_> *lol*
23:11 < rxr> e.g. I build with my modification - and I had no whitespace change
23:12 < mnemoc> it builds :)
23:12 < mnemoc> and that's great
23:13 < mnemoc> $F regresion is dead finally
23:14 < rxr> row, row, row your boat
23:14 < rxr> ...
23:16 < _Ragnar_> okay building
23:17 < mnemoc> :D
23:17 < mnemoc> building what?
23:18 < _Ragnar_> gdb
23:18 < _Ragnar_> make: *** [configure-readline] Error 1
23:18 < _Ragnar_> make: *** Waiting for unfinished jobs....
23:19 < mnemoc> how did you patched it?
23:19 < rxr> _Ragnar_: you may try NOPARALLEL ...
23:19 < CIA-10> mnemoc * r5654 /trunk/package/base/uclibc/ (parse-config-9 patches/binutils.conf): * moved uclibc binutils hacks to parse-config
23:19 < rxr> or jobs=1 ...
23:20 < _Ragnar_> mnemoc: as you told me ;)
23:20 < mnemoc> o_O
23:20 < mnemoc> publish an svn diff please ;)
23:22 < _Ragnar_> sorry, I need to go for lunch, see http://66.146.166.62/~loki/gdb-10-uclibc_arch_target.patch / gdb.conf / 1-gdb.err
23:22 < _Ragnar_> be back in an hour or so
23:24 < mnemoc> _Ragnar_: i'll commit mine, when you get back see how it conflicts yours and if it builds please
23:25 < CIA-10> mnemoc * r5655 /trunk/package/base/uclibc/ (parse-config-9 patches/gdb-10-uclibc_arch_target.patch): * fixed uclibc gdb patching to the automatized way (NOT TESTED!)
23:29 < rxr> mnemoc: if we are too crazzy we could optimize the t2 package creation on linux
23:29 < mnemoc> rxr: how?
23:29 < rxr> write a fuse plugin that transparnly mounts over a subtree and stores files writes into a tarball ...
23:30 < mnemoc> what about files wrote outside $builddir by mistake?
23:30 < mnemoc> who will alert?
23:30 < rxr> ok ;-)
23:30 < rxr> was just some crazzy flash anyway ... *g*
23:30 < mnemoc> i mean, i like the idea of using fuse
23:30 < rxr> (with build as user we could do this - no / access ;-)
23:30 < mnemoc> i told it to jsaw before
23:31 < mnemoc> but we could use fl_wrapper just to segfault if it tries to write outside $builddir
23:31 < mnemoc> i.e. wrappinf writes only
23:31 < mnemoc> wrapping*
23:34 < rxr> let's not overcomplicate it ... ,-)
23:35 < mnemoc> :p
23:35 < mnemoc> fuse + luser builds ;)
23:35 < mnemoc> for 2.2
23:48 < rxr> == 01/18/05 23:44:54 =[5]=> Finished building package glibc.
23:48 < rxr> ^- t2:HEAD
23:48 < mnemoc> :D
--- Log closed Wed Jan 19 00:00:47 2005