--- Log opened Mon Apr 09 00:00:04 2007
--- Day changed Mon Apr 09 2007
00:00 < sepp> huuhuu
00:13 < sepp> :)
00:54 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has joined #t2
01:07 < sepp> debian 4.0 with only 21 cds !
01:08 < sepp> that is worse than my amiga 500 with only 1 floppy drive :p
01:10 < Stelz> ahaha
01:37 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has quit [Remote closed the connection]
01:39 -!- mqueiros [n=mqueiros@c-217-70-65-169.bragatel.pt] has quit [Connection timed out]
01:39 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has joined #t2
01:51 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has quit [Remote closed the connection]
01:58 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has joined #t2
01:58 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has quit [Remote closed the connection]
02:08 < AStorm> sepp: ok, figured all of it out :>
02:08 < AStorm> Just wanted to rebuild stage1 uclibc w/o static
02:08 < AStorm> modified my target to switch off STATIC for it in parse-config
02:11 < AStorm> The dynamic loader of uclibc should be toggled by some other option, much like dietlibc
02:11 < AStorm> not just CONFIG_STATIC
02:14 < sepp> :)
02:14 < AStorm> Well, and perl has to be fixed for it too :>
02:15 < AStorm> (by turning off DynaLoader)
02:15 < AStorm> more patches will follow :D
02:17 < AStorm> Hmm...
02:17 < AStorm> embutils failed
02:18 < AStorm> with 'env.c:20: error: 'environ' undeclared
02:18 < AStorm> Someone forgot to link libcompat in
02:39 < AStorm> Hmm, where should I look for C Library Options menu?
02:41 < AStorm> ok, got it
02:41 < AStorm> just subconfig-libc.in
02:44 < sepp> good night :)
02:44 < AStorm> nighty night :-)
02:58 -!- mqueiros [n=mqueiros@c-217-70-65-169.bragatel.pt] has joined #t2
02:59 -!- mqueiros [n=mqueiros@c-217-70-65-169.bragatel.pt] has quit [Client Quit]
03:10 < AStorm> Hmm, uclibc doesn't have md5.h :>
03:12 < AStorm> Ah, nope, that should be from PAM
03:14 < AStorm> Someone wasn't very thoughtful when he included that in dietlibc
03:15 < AStorm> Yes, there's no md5.h in uclibc it seems
03:15 < AStorm> or it's optional
03:24 < AStorm> I'll add some predefined feature sets to uclibc :>
03:24 < AStorm> (like vim
03:24 < AStorm> and some options too, like IPv6 or RPC support
03:25 < AStorm> or Unicode support (will toggle some things)
03:46 < AStorm> (actually, Unicode and wide character support)
04:45 -!- emte_ [n=emte@d64-180-45-14.bchsia.telus.net] has joined #t2
04:47 -!- emte [n=emte@d64-180-45-14.bchsia.telus.net] has quit [Read error: 110 (Connection timed out)]
06:16 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has joined #t2
06:18 -!- idealm [n=ideal@c58-107-18-60.belrs2.nsw.optusnet.com.au] has quit [Client Quit]
09:24 < sepp> moin :)
09:40 -!- fke_ [n=fke@p3e9e3778.dip0.t-ipconnect.de] has quit []
11:28 < rxr> re
11:28 < rxr> moin sepp 
11:30 < sepp> hi rxr :)
12:25 < CIA-18> rene * r23473 /trunk/package/graphic/imagemagick/imagemagick.desc: * updated imagemagick (6.3.3-5 -> 6.3.3-6)
12:42 < CIA-18> rene * r23474 /trunk/package/contrib/vidix/vidix.desc: * updated vidix (0.9.9.3 -> 1.0.0)
12:44 < CIA-18> rene * r23475 /trunk/package/network/ipsec-tools/ipsec-tools.desc: * updated ipsec-tools (0.6.6 -> 0.6.7)
12:47 < CIA-18> rene * r23476 /trunk/package/network/whois/whois.desc: * updated whois (4.7.20 -> 4.7.21)
12:56 -!- sepp_ [n=sepp@p213.54.60.20.tisdip.tiscali.de] has joined #t2
13:20 -!- sepp [n=sepp@p85.212.42.70.tisdip.tiscali.de] has quit [Read error: 110 (Connection timed out)]
13:47 < AStorm> moin guys
13:47 < AStorm> Ok, now on to patching embutils (they should be banned: busybox is much better)
13:49 < rxr> one prefers that, another prefers the other :-)
13:49 < rxr> free of cohice
13:49 < rxr> choice
13:49 < rxr> moin moin AStorm 
13:49 < AStorm> stupid embutils doesn't have extern char**environ :>
13:50 < AStorm> Instead, it depends on GNU behaviour w/o #define _GNU_SOURCES
13:50 < AStorm> *_GNU_SOURCe
13:51 < AStorm> rxr: indeed, expect _a lot_ of patches :-)
13:56 < rxr> please send them as-you-go
13:56 < rxr> this a) makes working and commenting them more pleassuant
13:57 < rxr> and b) we have the patches even if you decide to do a three week holiday before submitting them :-)
13:58 < AStorm> rxr: well, no mail client :>
13:58 < AStorm> First, I have to finish building :P
13:59 < AStorm> Still sitting on T2-Resc :P
14:03 < AStorm> Hmm, where should be PAGE_SIZE defined? :>
14:04 < AStorm> linux/sys/user.h?
14:04 < AStorm> (or sys/user.h)
14:05 < AStorm> Ah, yes
14:05 < AStorm> __USE_XOPEN
14:05 < AStorm> and limits.h
14:06 < AStorm> so, _XOPEN_SOURCE 600
14:06 < AStorm> otherwise, it's _SC_PAGE_SIZE
14:08 < AStorm> embutils programmers can't read manpages ;-)
14:09 < AStorm> (and dietlibc is lax)
14:11 < AStorm> dd.c:129: warning: incompatible implicit declaration of built-in function 'memcpy'
14:11 < AStorm> Hmm...
14:12 < AStorm> But I won't fix this warning. Later :>
14:16 < AStorm> Hah, uclibc doesn't define PAGE_SIZE
14:16 < AStorm> one should use sysconfig interface instead
14:21 < rxr> yes, static PAGE_SIZE is gone in latest kernel headers
14:24 < AStorm> another one: MNT_DETACH
14:24 < AStorm> Let's grep a bit...
14:28 < AStorm> hmm, doesn't exist
14:29 < AStorm> It's defined in linux/fs.h
14:29 < AStorm> it _was_ defined there
14:29 < AStorm> Not in the 2.6.19.1 headers
14:32 < AStorm> Now, any idea on how to work around it?
14:34 < rxr> hm
14:35 < AStorm> Hmm, should be in <sys/mount.h>
14:36 < AStorm> Aha, uclibc doesn't define it :P
14:36 < AStorm> Probably doesn't support it.
14:37 < AStorm> Ok, it does support it - umount2 is just a syscall there
14:37 < AStorm> But the enum doesn't have that entry
14:37 < AStorm> uclibc patch upcoming :>
14:50 < AStorm> Hmm, splitting debug from uclibc takes ages... but works fine
14:53 < AStorm> even more build errors, ehh
14:53 < AStorm> allinone.c:198: undefi
14:54 < AStorm> ned reference to `__syscall_syslog'
14:57 < AStorm> Yes, they are stupid: there is a "syslog" function using this syscall
14:57 < AStorm> and they don't use it
15:02 < rxr> guess they did so because dietlibc had or has no such wrapper function
15:02 < AStorm> No, it's for dmesg
15:02 < AStorm> something else
15:03 < AStorm> I'll have to find the syscall number
15:03 < AStorm> and wrap it :>
15:04 < AStorm> yep, __SYSCALL(__NR_syslog, sys_ni_syscall) probably
15:08 -!- Capey [n=admin@a83-245-240-112.elisa-laajakaista.fi] has joined #T2
15:10 < AStorm> someone forgot to include <asm/unistd.h> :>
15:10 < AStorm> dietlibc does include it, some other libcs don't
15:18 < AStorm> yes, klogctl call
15:21 < AStorm> in <sys/klog.h>
15:21 < AStorm> this is the standard way to access klog
15:24 < AStorm> Ok, that was the last of fixes :D
15:25 < AStorm> That's why embutils sucks :D
15:28 < AStorm> uh, cpio failed too :>
15:28 < AStorm> multiple definition of `program_invocation_{,short}_name'
15:29 < AStorm> Ok, easy to fix.
15:31 < rxr> i Capey 
15:31 < rxr> hi even
15:31 < rxr> AStorm: better get a MUA running sooner than later :-)
15:32 < rxr> so - lunch - cu
15:32 < AStorm> When the T2 is built, it will be the first thing I do
15:32 < AStorm> (exim at least)
15:36 < AStorm> ok, fixed
15:40 < AStorm> linux is dling :-)
15:50 < Capey> hi rxr
15:51 < Capey> rxr: i got wrt2 successfully build on my laptop but for somereason it fails on my desktop pc
15:52 < Capey> and sadly i forgot the laptop to my parents but i'm currently getting diff from trunk tree
15:53 < Capey> but they should be identical
15:58 < AStorm> rxr: eh, donate me faster broadband :P
15:58 < AStorm> this 256k DSL can't even aspire to that name
16:05 < AStorm> rxr: when is your book scheduled (for buying and download?)
16:58 < AStorm> uhh, kernel is building like 100 years :>
16:58 < AStorm> and w/o -j3 somehow
17:08 < rxr> re
17:10 < rxr> Capey: hm
17:10 < rxr> AStorm: german version will go to a publisher in june
17:10 < rxr> in the "regular stores" a bit later then, Q3
17:10 < rxr> international, EN version will receive updates soon
17:11 < rxr> and will be available via print-on-demand (whatever, we're still looking) around june, likewise
17:11 < AStorm> German version is ok with me too (know a bit)
17:12 < rxr> we still have to write^3
17:17 < AStorm> rxr: woohoo, kernel build failed
17:17 < AStorm> !> /t2-trunk/build/default-7.0-trunk-multipen-x86-64-cross/sbin/mkinitrd: line 56: which: comman ..
17:17 < AStorm> !> cpio not found!
17:17 < AStorm> It's weird, as cpio has been installed properly
17:18 < AStorm> Ahh.
17:18 < AStorm> I don't have cpio in my stage
17:18 < AStorm> Add a paranoia check against that, please.
17:31 < Capey> rxr: there's around 1.9Mb of imported patches from openwrt. i'll tar those and upload somewhere and add new download tag to needed packages, rather than adding them directly to svn.
17:32 < Capey> at later stage i'll review those patches and remove those that are not directly needed
18:52 < rxr> did you use their 2.4 or 2.6 patches?
18:52 < rxr> I scrolled thru the site some days ago and noticed they have booth
18:52 < rxr> with still 2.4 in their stable or so ?
18:53 < rxr> what is your experience ?
18:58 -!- sepp_ is now known as sepp
19:52 < Capey> rxr: i used they 2.6 patches. 2.4 is for current stable.
19:53 < Capey> rxr: but most of the patches i imported where actually from broadcom
19:53 < Capey> rxr: just rediffed against 2.6 IIRC
19:59 < rxr> of course 2.6 would be soo much nicer
19:59 < rxr> any idea if their 2.6 rediff does work?
20:06 < AStorm> rxr: uhoh
20:06 < AStorm> linux26 is broken, requires which
20:06 < AStorm> senseless
20:06 < AStorm> should use type -p
20:09 < AStorm> actually, it's the mkinitrd script
20:11 < rxr> you think type -p is more universal than which?
20:11 < rxr> you do not want to have which in your system?
20:11 < AStorm> You're using /bin/bash anyway
20:11 < AStorm> Not in the stage.
20:11 < rxr> using bash anyway is a good piont
20:11 < rxr> point
20:12 < rxr> change the line and include it in your monster patch
20:12 < AStorm> In my monster patches, you mean :-)
20:13 < AStorm> Uh, why is the kernel built w/o -j?
20:15 < AStorm> Where should I change it? (kills perfomance needlessly on my C2D - the build takes 30 minutes or more)
20:15 -!- Capey [n=admin@a83-245-240-112.elisa-laajakaista.fi] has quit [Read error: 113 (No route to host)]
20:25 < AStorm> Where is the code for parallel make?
20:25 < AStorm> I can't find any place that does add -j
20:27 < AStorm> Ahh, you have your own parallel build simulator
20:28 < AStorm> I wonder why isn't it used with kernel build
20:29 < sepp> package/base/make/parse-config
20:29 < AStorm> Ir looks like the -j is added...
20:30 < AStorm> but then stripped somewhere (kcc?)
20:31 < AStorm> the logs show -j
20:31 < AStorm> but ps doesn't show multiple gccs running
20:31 < AStorm> Something is wrong :>
20:31 < Stelz> everything's wrong until you'll make it right :)
20:31 < AStorm> Uh..
20:32 < AStorm> the thing is that it works with "normal" packages
20:32 < AStorm> But not the kernel
20:32 < sepp> in the biginning of the log is what flags are removed and inserted into the gcc wrapper
20:32 < rxr> -j is for make
20:32 < rxr> I also saw sub-optimal load on kernel builds
20:33 < rxr> I think we had or even have special code in linux24.conf (or so) even adding +1 to the PARALLEL_MAX value
20:33 < AStorm> Usually, when I do the kernel myself, -j works all right
20:33 < rxr> does "ps ax" or so list the make -j ... ?
20:33 < AStorm> Let's see
20:33 < AStorm>  make -j3 ARCH=x86_64 CROSS_COMPILE=x86_64-t2-linux-uclibc- KCC=x86_64-t2-linux-uclibc-kcc vmlinux modules bzImage
20:34 < AStorm> But then...
20:34 < AStorm> make -f scripts/Makefile.build obj=arch/x86_64/ia32
20:35 < rxr> no idea what the kernel makefiles do ...
20:35 < rxr> for me also T2 controlled kernel builds did work on multi cpu boxes
20:36 < rxr> maybe a new kbuild regression?
20:36 < AStorm> Hmm.
20:36 < AStorm> Well, 2.6.20 worked fine here when made not by T2...
20:36 < AStorm> but not in cross
20:36 < AStorm> maybe it's some cross build issue
20:36 < rxr> maybe due to our MAKE environment variable ...
20:36 < rxr> maybe "some environment variable invluencing kbuild" issue ...
20:37 < AStorm> :/
20:37 < AStorm> Could be.
20:37 < rxr> please hunt :-)
20:37 < AStorm> Will do, once it builds.
20:44 < rxr> :-)
20:44 < rxr> cu all
21:08 < Stelz> cu rxr
21:28 < AStorm> DAMN
21:28 < AStorm> Yet another build error
21:29 < AStorm> /sbin/depmod: No s
21:29 < AStorm> uch file or directory
21:29 < AStorm> ?!?
21:29 < AStorm> Drat, my stage doesn't have module-init-tools either
21:29 < AStorm> I'll really have to fix the paranoia check
21:44 < AStorm> Another 45 minutes down the drain
21:47 < sepp> :/
22:04 < LMJ> moin moin
22:09 < AStorm> moin
22:20 < AStorm> Hey guys, what about rebuilding all packages depending on x when x is updated?
22:20 < AStorm> Good for static systems...
22:22 < AStorm> or when the new version is binary-incompatible
22:45 < AStorm> yet another 45 minutes down the drain
22:46 < AStorm> I installed module-init-tools to /usr/local and not /
22:46 < AStorm> :/
23:03 < sepp> :\
23:06 < AStorm> Who sane calls a program with an absolute path :>
23:06 < AStorm> (e.g. /sbin/depmod)
23:18 -!- emte_ [n=emte@d64-180-45-14.bchsia.telus.net] has quit [Read error: 110 (Connection timed out)]
23:19 -!- emte_ [n=emte@d64-180-45-14.bchsia.telus.net] has joined #t2
23:44 < AStorm> finally, linux26 built :-)
23:47 < sepp> hurra
23:47 < AStorm> paranoia check is bound for fixing, once I get the system up
23:48 < AStorm> now, the big things: stage1 binutils, gcc, libcs
23:48 < AStorm> Oh, binutils just bombed on me :P
23:51 < AStorm> http://pastebin.ca/431781
23:51 < AStorm> Any idea? :>
23:52 < AStorm> It should be static, not -fPIC :/
23:53 < AStorm> with -z relro
23:53 < AStorm> (that's forced by libc afaik)
23:54 < AStorm> Does anyone there test uclibc builds? :P
23:58 < AStorm> Probably it's because of CONFIG_STATIC
--- Log closed Tue Apr 10 00:00:10 2007