[12:03] <jbailey> Eh?  I have 2 RT installations that I help manage now, and I used it at my last job.
[12:03] <jbailey> It's a nice little package.
[12:03] <jbailey> Have you seen it version >= 3?
[12:11] <lamont> ncftp, not rt
[12:18] <jbailey> ah, I've never hacked on the ncftp code. =)
[01:15] <jbailey> -march=i686 -mcpu=i686
[01:15] <jbailey> Err.
[01:15] <jbailey> -march=i486 -mcpu=i686
[01:15] <jbailey> Should that be -march=i386 -mtune=i686 ?
[01:39] <jbailey> I *definetly* need better ia32 hardware.
[01:44] <jbailey> lamont-away: It's building fine for me here...
[01:45] <jbailey> I'm using gcc-3.4 from Debian, current lkh and amd64-libs{-dev} from Ubuntu
[01:45] <jbailey> libgcc1 4.0 from Ubuntu
[05:31] <fabbione> morning
[10:40] <svenl> about biarch ppc64 compilers and 4.0 : 
[10:40] <svenl> 10:37 < alanm> svenl, gcc-4.0 cvs might be ok, but it hasn't been tested anywhere near as much as gcc-3.4.  eg. the 4.0
[10:40] <svenl>                release is known to miscompile the linux kernel
[10:40] <svenl> doko: do you have any insight on this ? 
[10:41] <doko> we have gcc-4.0 CVS 20050509
[10:42] <svenl> doko: ok.
[10:42] <svenl> doko: question is if you can compile 64bit kernels with it.
[10:43] <svenl> doko: btw my debian/experimental gcc-4.0 build failed with : 
[10:43] <svenl> fatal error: file gnat1drv.ali is incorrectly formatted
[10:43] <svenl> make sure you are using consistent versions of gcc-3.3/gnatbind
[10:44] <svenl> doko: are you sure the gnat build-dependency is as strict as is needed ? 
[10:44] <doko> if you disable add, no
[10:47] <svenl> doko: disable add ? 
[10:47] <doko> ada
[10:47] <svenl> doko: i just did a plain dpkg-buildpackage -B
[10:47] <doko> WITHOUT_LANG=ada dpkg-buildpackage ...
[10:47] <svenl> so i didn't touch the package.
[10:47] <svenl> mmm.
[10:48] <svenl> doko: so you want me to build and upload gcc-4.0 for powerpc without ada ? 
[11:52] <doko> svenl: ehh? no!
[11:54] <svenl> doko: so ? 
[11:54] <svenl> doko: i just did a plain dpkg-buildpackage -B, and the gnat stuff failed to build.
[11:54] <doko> misunderstanding, I thought you had problems wuth the biarch setup ...
[11:54] <doko> is gnat-3.3 or gnat-3.4 installed?
[11:55] <svenl> doko: so i was wondering if this was because i have gnat-3.3 or something.
[11:55] <svenl> ii  gnat-3.3       3.3.6-4        The GNU Ada compiler
[11:55] <doko> very strange ... you can check with gnat-3.4 maybe
[11:55] <svenl> but the build-depends are ok, so if gnat-3.4 is needed, the build-depends need tightening.
[11:55] <doko> save the build logs
[11:55] <svenl> yep, will do a build this evening.
[11:56] <svenl> doko: you want the gnat-3.3 build log ? 
[11:57] <doko> yes, could you email it?
[11:57] <svenl> doko : http://people.debian.org/~luther/gcc-4.0-gnat-3.3.log
[11:59] <doko> strange, it did build before, and on the ubuntu buildd's as well
[12:05] <svenl> with gnat-3.3 ? 
[12:06] <doko> yes
[12:06] <svenl> doko: on powerpc ? 
[12:06] <doko> http://people.ubuntu.com/~lamont/buildLogs/g/gcc-4.0/4.0.0-7ubuntu1/gcc-4.0_4.0.0-7ubuntu1_20050509-2138-powerpc-successful.bz2
[12:06] <svenl> strange.
[12:07] <svenl> I will let aba build it, and we will see.
[04:22] <fabbione> hey ladies
[04:23] <fabbione> doko: did you read what i wrote on #ubuntu-devel ?
[04:23] <doko> fabbione, that's not an under 18 channel ;-P
[04:25] <fabbione> ahhaa
[06:08] <jbailey> lamont, doko: I can't reproduce the configure time failure here for glibc on i386 targetting amd64.
[06:08] <jbailey> I did notice, however, that my thing moving the -O0 hack to a generic place didn't work. =(
[06:11] <lamont> jbailey: ew...
[06:11] <lamont> hrm...
[07:58] <svenl> jbailey: hi.
[07:58] <svenl> jbailey: glibc build failed because the 64bit build dir was not populated.
[08:05] <jbailey> svenl: Eh?
[08:05] <jbailey> svenl: Here, lemme produce a .deb for you.
[08:05] <jbailey> With both passes, it'll take about 40 minutes.
[08:05] <jbailey> ...
[08:05] <jbailey> to build.
[08:05] <jbailey> Lemme assemble it first.
[08:05] <svenl> jbailey: i rebuild glibc-2.3.5 (from your web page), with the gcc from doko, and it failed when it tried to build the ppc64 stuff, because that dir was empty.
[08:06] <svenl> jbailey: that's not the problem, i am investigating the build on ppc32 part of it.
[08:08] <svenl> jbailey: let me relocate on the powerbook, so i can tell you more about this error.
[08:09] <jbailey> svenl: 'kay.  I can't talk long today.  I munged my wrist last night, so I'm avoiding typing today.
[08:10] <svenl> jbailey: but you can read :)
[08:10] <jbailey> Right. =)
[08:10] <svenl> jbailey: or we can try a voip debugging session :)
[08:10] <jbailey> True.  If I were at home, I'd just call you, it's 5 a minute for me to europe.
[08:11] <jbailey> I should setup voip sometime.
[08:12] <svenl> root@tael:~/ubuntu/glibc-2.3.5/build-tree/powerpc-ppc64# ls
[08:12] <svenl> config.log  configparms
[08:12] <svenl> you see there is only this.
[08:12] <svenl> /usr/bin/make -C build-tree/powerpc-ppc64 -j 1 \
[08:12] <svenl>   install_root=/home/sven/ubuntu/glibc-2.3.5/debian/tmp-ppc64 install
[08:12] <svenl> make[1] : entrant dans le rpertoire  /home/sven/ubuntu/glibc-2.3.5/build-tree/powerpc-ppc64 
[08:13] <svenl> make[1] : *** Pas de rgle pour fabriquer la cible  install . Arrt.
[08:13] <jbailey> What's the failure at the bottom of the log file?
[08:13] <svenl> and there it fsails.
[08:13] <jbailey> No, it fails before that.
[08:13] <svenl> well, i guess there is no Makedile in that dir.
[08:13] <jbailey> It just doesn't kill the build.
[08:13] <jbailey> We pipe stuff through tee, so we lose return codes.
[08:13] <jbailey> It can fail *way* before that point.
[08:13] <svenl> touch /home/sven/ubuntu/glibc-2.3.5/stamp-dir/install_libc
[08:13] <svenl> Installing ppc64
[08:13] <svenl> rm -rf /home/sven/ubuntu/glibc-2.3.5/debian/tmp-ppc64
[08:13] <svenl> /usr/bin/make -C build-tree/powerpc-ppc64 -j 1 \
[08:13] <svenl>   install_root=/home/sven/ubuntu/glibc-2.3.5/debian/tmp-ppc64 install
[08:13] <svenl> make[1] : entrant dans le rpertoire  /home/sven/ubuntu/glibc-2.3.5/build-tree/powerpc-ppc64 
[08:13] <svenl> make[1] : *** Pas de rgle pour fabriquer la cible  install . Arrt.
[08:13] <svenl> make[1] : quittant le rpertoire  /home/sven/ubuntu/glibc-2.3.5/build-tree/powerpc-ppc64 
[08:13] <jbailey> The the top directory of the tree, there should be a log-powerpc64-*
[08:13] <svenl> make: *** [/home/sven/ubuntu/glibc-2.3.5/stamp-dir/install_ppc64]  Erreur 2
[08:14] <svenl> ah, ok.
[08:14] <jbailey> I need to see where configure failed.
[08:14] <svenl> i have the fukll log normally, let me check.
[08:14] <jbailey> Right.  The full log doesn't make sense in our case, because there's many places it can fail.
[08:14] <jbailey> And just to piss lamont off, it's 40 megs.
[08:14] <svenl> configure: error: forced unwind support is required
[08:14] <svenl> make[1] : entrant dans le rpertoire  /home/sven/ubuntu/glibc-2.3.5/build-tree/powerpc-ppc64 
[08:14] <jbailey> Because he *loves* getting those in his email. =)
[08:14] <svenl> make[1] : *** Pas de cibles spcifies et aucun makefile n'a t trouv. Arrt.
[08:15] <svenl> make[1] : quittant le rpertoire  /home/sven/ubuntu/glibc-2.3.5/build-tree/powerpc-ppc64 
[08:15] <jbailey> That'd be the one. =(
[08:15] <svenl> ok.
[08:15] <svenl> checking for forced unwind support... no
[08:15] <svenl> configure: error: forced unwind support is required
[08:15] <jbailey> I need a typing break, back in a couple of minutes (sorry)
[08:16] <svenl> no problem
[08:18] <jbailey> Bah, found ice, can't find anything to wrap it in.
[08:18] <svenl> how did you damage your hand ?
[08:18] <jbailey> Can you take a look at that config.log file, please?
[08:19] <jbailey> svenl: Playing pool last night.
[08:19] <jbailey> svenl: I have hypermobile joints, so if I'm not careful I can do incredibly bad things to them.
[08:19] <svenl> yep.
[08:19] <jbailey> I was a little too enthusiastic on a shot.
[08:19] <jbailey> (On the upside, I did sink two balls off of it)
[08:19] <svenl> :/
[08:19] <svenl> :)
[08:20] <svenl> so what you want to see in config.log ?
[08:20] <jbailey> It's not horrible, just throbby.  The nurse I spoke to says that I've likely just pulled something, given that there's no swelling and whatnot, and that I need to ice it and rest it when it gets uncomfortable.
[08:20] <jbailey> Look for the word "unwind" and find a failure near there.
[08:20] <svenl> configure:5402: checking for libunwind-support in compiler
[08:20] <svenl> configure:5419: result: no
[08:21] <jbailey> Next one, force unwind is later.
[08:21] <svenl> jbailey: you could plug out that liquid cooling system of the pmac G5 and put it around your hand :)
[08:21] <jbailey> svenl: The pmac is 5000km from here.  I haven't finished travelling home from UDU yet. =)
[08:21] <svenl> /usr/bin/ld: cannot find -lc
[08:21] <svenl> Ah.
[08:22] <jbailey> Eh?  It shouldn't need a working C library to build the biarch compiler.
[08:22] <jbailey> Oh...
[08:22] <jbailey> wait
[08:22] <svenl> configure:27: checking for forced unwind support
[08:22] <svenl> configure:51: gcc-3.4 -m64 -o conftest -g -O2 -isystem /home/sven/ubuntu/glibc-2.3.5/debian/include  conftest.c  >&5
[08:22] <jbailey> DAMMIT
[08:22] <svenl> /usr/bin/ld: skipping incompatible /usr/lib/gcc/powerpc-linux/3.4.4/../../../libc.so when searching for -lc
[08:22] <jbailey> I forgot about that.
[08:22] <svenl> /usr/bin/ld: skipping incompatible /usr/lib/gcc/powerpc-linux/3.4.4/../../../libc.a when searching for -lc
[08:22] <svenl> /usr/bin/ld: skipping incompatible /usr/bin/../lib/libc.so when searching for -lc
[08:22] <svenl> /usr/bin/ld: skipping incompatible /usr/bin/../lib/libc.a when searching for -lc
[08:22] <svenl> /usr/bin/ld: skipping incompatible /usr/lib/libc.so when searching for -lc
[08:22] <svenl> /usr/bin/ld: skipping incompatible /usr/lib/libc.a when searching for -lc
[08:22] <doko> jbailey: it does for libgcc
[08:22] <svenl> /usr/bin/ld: cannot find -lc
[08:22] <svenl> collect2: ld returned 1 exit status
[08:22] <svenl> configure:57: $? = 1
[08:22] <svenl> configure: failed program was:
[08:22] <jbailey> doko: No, it's a bug in the nptl configury.
[08:23] <jbailey> doko: Linuxthreads doesn't need it.
[08:23] <doko> ahh, better :)
[08:23] <svenl> mmm.
[08:23] <jbailey> svenl: Lemme send you my backup copy of glibc with lt. =)
[08:23] <svenl> doko: btw, i also trried building gcc 3.4.3 cvs from scratch to build a biarch toolchain by hand.
[08:23] <svenl> huh ?
[08:24] <svenl> doko: but it failed too with a libiberty error.
[08:24] <jbailey> svenl: I have a copy of glicb that you need to install for this build to work.
[08:24] <svenl> jbailey: :)
[08:24] <svenl> jbailey: a glibc package ?
[08:24] <jbailey> svenl: Yes.
[08:24] <jbailey> svenl: A transitional one.
[08:24] <jbailey> I need to remember to get this bug fixed.
[08:24] <svenl> mmmm.
[08:25] <svenl> ok.
[08:38] <svenl> jbailey: how will you send me the copy and when ?
[08:38] <jbailey> http://testhaus.cns.utoronto.ca/~jbailey/ppc64-lt/
[08:38] <jbailey> lamont: It just occured to me this is probably our amd64 failure too. =(  Same bootstrap case, really.
[08:39] <lamont> ??
[08:40] <svenl> jbailey: you know that you just violated the GPL :)
[08:40] <jbailey> svenl: I have not, if you ask me for the source I will provide it.
[08:40] <svenl> (meaning the above does not contain the source packages :)
[08:40] <svenl> jbailey: you didn give me a written offer for it though :)
[08:41] <jbailey> Just as soon as a self addressed stamped envelope with an international postage voucher shows up with a blank cdrw. ;)
[08:41] <svenl> ok, i am downloading the stuff.
[08:42] <lamont> I had a very confused person recently who didn't realize that you could make a derivitive work based on a GPL'ed work and not return your patches upstream.
[08:42] <lamont> (specifically, source must be made available to anyone who can get the binaries, which is not necessarily everyone)
[08:43] <jbailey> lamont: Right.  I've wondered a few times when we'll start to see embedded linux devices with a spare chip, not hooked up to anything, that has the entire kernel source in it.
[08:43] <svenl> BTW, why does sudo need to do a gethostbyname ?
[08:44] <svenl> jbailey: it needs the prefered form of modification, so this wouldn work.
[08:44] <lamont> svenl: hostname is one of the components of the sudoers file
[08:44] <jbailey> svenl: It's still in source form.  If you have the tools to modify the embedded binary, you have the tools to get the source code.
[08:44] <lamont> so you can have one file that goes everywhere, and only grants bob access to reboot _his_ machines
[08:45] <lamont> jbailey: socketed?
[08:45] <svenl> lamont: ah.
[08:45] <svenl> would add about 15$ per board though as of today.
[08:45] <jbailey> lamont: Could even be soldered as long as you could reasonably dump it with the some tools you'd use to program the roms.
[08:45] <lamont> mind you, an sudod would be nice to jave, so you didn't have to push said file everywhere...
[08:46] <jbailey> lamont: The trick is that every bios update would have to also update the source image.
[08:46] <lamont> true
[08:47] <jbailey> svenl: Right, cost prohibative for now, but given that it can be quite slow to write to, it might be cheaper than the support overhead of making source available for 3 years, or shipping it on demand if they managed to get quantities up.
[08:47] <svenl> jbailey: yep.
[08:47] <svenl> jbailey: especially as flash sizes grow.
[08:48] <svenl> jbailey: launched the build with your transitional glibc.
[08:48] <svenl> will take a while.
[08:49] <jbailey> lamont: Do you still have that chroot lying arond that you used for the i386/amd64 biarch build?
[08:49] <lamont> uh, used?
[08:50] <doko> ehh, yes, jbailey wanted to fix it ...
[11:36] <jbailey> doko: You there?
[11:42] <doko> yes
[11:42] <jbailey> We currently have -march=i486 -mcpu=i686
[11:42] <jbailey> (In various configs)
[11:43] <jbailey> I get a gcc-3.4 warning saying that one of those is deprecated.
[11:43] <jbailey> -mcpu
[11:43] <jbailey> Should that just change to -mtune ?
[11:44] <doko> yes, I mailed you about it
[11:44] <jbailey> You did?
[11:45] <jbailey> I see the thing where you point out the message...
[11:46] <jbailey> But a'ight, I'll change it to -mtune
[11:46] <doko> yes, I meant that :)
[11:47] <jbailey> Thanks.  Starting another build and running off for lunch.