irclog2html for #uclibc on 20070111

00:00.06landleyCombination Linux expo and science fiction convention.
00:00.56landleyYear 1: Terry Pratchett's and Rob Malda of slashdot lost a quake deathmatch to Pete Abrams of sluggy freelance and Illiad of user friendly.
00:01.02landleyStill hard to top that...
00:01.15landley(there was a word "team" in there that got lost...)
00:03.09solarSpanKY: http://geekz.co.uk/schneierfacts/
00:09.33landleysolar: yes, we were pondering printing some of those in the program book.
00:11.54*** join/#uclibc mtr_ (n=Michael@kobz-590cb888.pool.einsundeins.de)
01:30.36*** join/#uclibc hiyuh (n=hiyuh@ZK138196.ppp.dion.ne.jp)
02:25.33*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-26-197.dsl.pltn13.sbcglobal.net)
02:48.17*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-26-197.dsl.pltn13.sbcglobal.net)
03:10.20*** join/#uclibc KhemHome_ (n=KhemHome@adsl-71-146-26-197.dsl.pltn13.sbcglobal.net)
03:11.34*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-26-197.dsl.pltn13.sbcglobal.net)
04:54.50*** join/#uclibc hiyuh (n=hiyuh@ZK138196.ppp.dion.ne.jp) [NETSPLIT VICTIM]
04:54.50*** join/#uclibc SpanKY (n=UserBah@pool-141-154-238-235.bos.east.verizon.net) [NETSPLIT VICTIM]
05:05.36*** join/#uclibc hiyuh (n=hiyuh@ZK138196.ppp.dion.ne.jp) [NETSPLIT VICTIM]
05:05.37*** join/#uclibc SpanKY (n=UserBah@pool-141-154-238-235.bos.east.verizon.net) [NETSPLIT VICTIM]
06:17.08CIA-1003andersen * r17238 10buildroot/package/xorg/ (Config.in xorg-config.patch xorg.mk): more updates to ensure subsequent packages build
06:17.25CIA-1003andersen * r17239 10buildroot/package/ (6 files in 2 dirs): Add pango
06:46.43*** join/#uclibc PhB (n=PhB@dD5E060DA.access.telenet.be)
07:20.08*** join/#uclibc blindvt (n=bf@M829P012.adsl.highway.telekom.at)
07:24.42*** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan)
07:25.03CIA-1003andersen * r17240 10buildroot/package/expat/expat.mk: fix dependancies and install paths
07:25.30CIA-1003andersen * r17241 10buildroot/package/jpeg/jpeg.mk: be explicit about not doing X
07:26.03CIA-1003andersen * r17242 10buildroot/package/libpng/libpng.mk: be explicit about not depending on X, dep should go the other way
07:27.39CIA-1003andersen * r17243 10buildroot/package/ (Config.in tiff/ tiff/Config.in tiff/tiff.mk): add tiff package
07:36.01CIA-1003andersen * r17244 10buildroot/package/cairo/cairo.mk: fix build deps
07:36.46CIA-1003andersen * r17245 10buildroot/package/pango/pango.mk: use a slightly more sensible dep, strip the correct library names
07:44.15CIA-1003andersen * r17246 10buildroot/package/ (Config.in atk/ atk/Config.in atk/atk.mk): add atk package
07:52.07*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-42-240.dsl.pltn13.sbcglobal.net)
08:42.38CIA-1003andersen * r17247 10buildroot/package/ (19 files in 2 dirs): add Gtk+ version 2
08:45.29CIA-1003andersen * r17248 10buildroot/package/cairo/cairo.mk: it seems pdf and ps support are required for gtk2
10:38.12CIA-1003andersen * r17249 10buildroot/package/xorg/ (mcookie.c xorg.mk): a copy of mcookie will come in handy...
11:32.17whirmhi all
12:38.04*** join/#uclibc sjhill (n=sjhill@real.realitydiluted.com)
13:39.58whirmI need UCLIBC_HAS_LOCALE enabled in uclib to be able to build dbus, what's the "right way" to check for it/enable it  if dbus is selected BR2_PACKAGE_DBUS is enabled?
13:40.34whirmshould I grep for it in the dbus makefile and add the optiond,rebuild uclibc, etc...?
13:40.47whirmor there's a cleaner way?
13:43.17blindvt`dbus has to cope with locale support turned off. If it doesn't then you have to fix it (turning on locale support just because dbus is broken is not a reasonable solution)
13:45.08whirmpufff, that can't be easy to do :-/
13:45.48blindvt`very exciting. compilation took about 9.5 hours, now it's starting the final linking.. *holds thumbs*
13:47.22whirmblindvt: compiling X?
13:49.13blindvt`nope, that'd take about 20 minutes
13:49.35whirmthen ooffice?
13:49.37whirmXDDD
14:11.31*** join/#uclibc hiyuh (n=hiyuh@KD125054017176.ppp-bb.dion.ne.jp)
15:33.03*** join/#uclibc blindvt_ (n=bf@80.123.57.90)
15:53.43whirmdbus needs iconv.h
15:54.12whirmand I only can use iconv.h from uclibc if I enable locales at compile time :-(
15:54.15*** join/#uclibc ambroseL (n=bjb@router.fidus.ca)
16:07.02solarwork around it. make dbus no longer need it :)
16:07.32*** join/#uclibc rob____ (i=rw@217.22.94.106)
16:07.40rob____hello
16:13.22whirmsolar: I just saw that dbus doesn't need iconv.h  but libxml2 does, buildroot includes /usr/include/libxml2/libxml/encoding.h (yup, from the host system)
16:13.46whirmand this header is the one that includes iconv.h
16:17.04sjhillwhirm: i am reaching a point where i can submit the buildroot changes for dbus
16:17.10sjhilli had to upgrade and merge
16:17.32sjhilludev is also being changed
16:18.11whirmsjhill: great!
16:18.22whirmthen Ill drop the attempt
16:18.37whirmwhen do you think the changes will be up?
16:19.57sjhilltonight
16:20.32whirmok, then Ill continue coding other stuff. ;-)
16:23.56solarwhirm: I got both libxml2 and dbus installed in a uClibc devel env which has no locales/iconv/nls support
16:25.18whirmsolar: yes, but buildroot is including my local libxml2 headers, i supose because I have not libxml2  in buildroot env....
16:26.18solarperhaps.. either way just letting you know what you want can be accomplished without the bloat
16:26.21rob____how do i go about compiling busybox against uclibc? I've compiled and installed uclibc and have configured busybox 1.3.1 how i'd like it but when i run make it /
16:26.39rob____attempts to link against glibc
16:27.24solarthe suggested normal method is to use buildroot for such things
16:27.51whirmsolar: Ill wait for sjhill commit and continue my work from there, now I'm reinventing the wheel, and I'm doing it worse than the first :-)
16:28.16rob____buildroot uses busybox 1.2.2.1
16:30.12solarbuildroot is a series of Makefiles that the end user can edit as he/she sees fit
16:37.02*** join/#uclibc landley (n=landley@c-71-199-114-26.hsd1.pa.comcast.net)
16:44.32*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-3-195.dsl.pltn13.sbcglobal.net)
16:45.05rob____i'd prefer not to use the buildroot scripts as im not exactly sure how they work
16:46.40landleyrob: What are you trying to accomplish?
16:46.52rob____following this guide: http://gentoo-wiki.com/SECURITY_System_Encryption_DM-Crypt_with_LUKS#Creating_initramfs_image
16:47.10rob____i dont have this option in menuconfig for busybox:   (/usr/i386-linux-uclibc/bin/i386-uclibc-) Cross Compiler prefix
16:47.23rob____which seems to be the only thing related to uclibc and busybox in this article
16:50.25CIA-1003vda * r17250 10busybox/ (5 files in 4 dirs):
16:50.25CIA-10ipv6-ization efforts continue. Few bugs are found,
16:50.25CIA-10unknown number likely introduced...
16:51.22CIA-1003vda * r17251 10busybox/ (4 files in 2 dirs): fix verbose output; remove commented-out includes.
16:51.25rob____i cant find any documentation what-so-ever on compiling busybox against uclibc on a glibc system
16:51.53rob____which is wierd since most people would be in the same situation as me - wanting to compile busybox on a glibc system
16:52.41rob____ive tried modifying the Makefile as follows: HOSTCFLAGS      = -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -I/usr/i386-linux-uclibc/usr/include /usr/i386-linux-uclibc/usr/lib/libc.a -nodefaultlibs -static -static-libgcc
16:52.48rob____but this didnt seem to help
16:52.50BertlAFAIK, the uclibc developers suggest to use the buildroot stuff
16:53.14Bertlwhich provides the entire uclibc toolchains and libraries
16:53.43Bertl(if you succeed to create the buildroot or find one to download, that is :)
16:53.46rob____ive built buildroot but by default it seems to use an old version of busybox
16:54.16Bertlyou can select the versions in the config, IIRC
16:54.44Bertland you could download a specific version yourself and compile it _inside_
16:55.24rob____ah ok :)
16:55.27rob____i think i see now
16:56.52rob____the problem i have is
16:57.14rob____i want to configure busybox using menuconfig to set some specific options
16:57.21rob____but when you make buildroot it seems to automatically do this
16:58.21whirmrob____: use make busybox-menuconfig
16:59.41rob____excellent, thanks
17:01.00whirmrob____: but thake in account that the config file is in the source tree, so if you change busybox version you will have to copy it to the new dir
17:02.29rob____thanks for the advice!
17:07.19Mirellinfo rob____
17:07.23MirellAck.
17:07.39MirellStupid wrong window and tab completion and...
17:11.58landleyrob: I've got a build system of my own that creates a cross compiler you can use to build uClibc stuff on x86.
17:12.07landleyhttp://landley.net/hg/firmware
17:20.04CIA-1003vda * r17252 10busybox/ (48 files in 19 dirs): Trailing whitespace removal over entire tree
17:20.09solarrob____: if using gentoo the suggested method is to use crossdev to build a proper cross compiler then either cross compile busybox or simp;y just grab the current uclibc-stageball and build in a chroot
17:20.28solardepending on the arch
17:20.36rob____im not using gentoo, its an LFS system im developing
17:21.26rob____i think if i use the buildroot makefiles then I can make menuconfig;make busybox-menuconfig; make and just disregard everything apart from the busybox stuff i need
17:21.54rob____i basically just need a statically compiled busybox executable
17:22.30rob____and since it wont compile against glibc i was faffing about trying to get it to compile with uclibc, but buildroot seems to take care of that
17:33.47*** join/#uclibc Khe1 (i=Khem@nat/montavista/x-cbade8dac8fd2090)
17:34.34landleySorry, I'm distracted in other windows...
17:35.20*** part/#uclibc Khem (i=Khem@nat/montavista/x-8e912f75f9adbffc)
17:35.28landleyYou can download a cross compiler tarball (prebuilt) from the download directory of http://landley.net/code/firmware/downloads
17:35.58landleyGrab cross-compiler-i686.tar.bz2, extract it, and add its' "bin" directory to your path.  The use i686-gcc as your cross compiler.  It'll build against uClibc.
17:36.03*** join/#uclibc Gerrath (n=Shane_@unaffiliated/gerrath)
17:36.19landleyhaven't tried it on other Linux distros yet, I dunno if it depends against specific libc versions or some such from the ubuntu I built it on.
17:36.29landley(That's a function of the host toolchain, really.)
17:36.44landleyIf it does, you'll need to build it from source yourself.  But try the binary first.
17:37.55landleyWith busybox you can go "make CONFIG_STATIC=y CROSS_COMPILE=i686-"
17:38.16landleyWith 1.2.2, anyway.  Dunno about 1.3.0, they might have renamed CROSS_COMPILE to just CROSS when the new build system went in...
17:40.04rob____i thought cross-compiling was just for building different architectures?
17:45.20landleyrob: cross compiling means building something that won't run on the host.
17:45.30landleyIf you dynamically link and don't have uClibc installed on the host, it won't run.
17:45.34landleyTherefore, we're cross compiling.
17:45.53landleyBuilding i486 code on an i686 is also sort of cross compiling.
17:46.12landleyIt'll run on the host, but if you're not careful you'll leak i686 bits into what you're building, and then it won't run on a real 486.
17:46.52landleyBasically, we're using a specially prepared compiler that does something specific, rather than the compiler your system comes with that does any old thing.
17:47.14landleyThis compiler outputs i686 code linked against uClibc.
17:47.37landleyUsing the headers and libraries and stuff contained in its directory, _NOT_ the stuff contained in /usr and /lib and such.
17:48.42rob____ok, that what i originally thought but my lodger told me I was wrong!
17:49.49rob____im going to come back to this tomorrow as i have to go out
17:49.57landleyok.
17:50.00rob____but the stuff on your site looks interesting so i'll have a read tomorrow :)
17:51.03landleyLet me know what you think. :)
17:51.32rob____will do, what im doing is a continuation of a project i was doing at my old job
17:51.40landleyJoin the club. :)
17:51.57rob____building a linux distribution build system based on the LFS project and netbsd pkgsrc
17:52.14rob____mainly as a personal project but i've been thinking about documenting it recently
17:52.37rob____ive just been trying to add encrypted filesystem stuff because its nerdy
17:55.26MirellSpeaking of encrypted filesystems, when I first got the MacBook, I went, "Oh, FileSafe! Encrypted home directory! Spiffy!" Then realized it was totally worthless.
17:56.01rob____since someones just going to modify your binaries and do a man in the middle attack :)
17:56.02MirellI used to have a Gentoo system though with encrypted swap and encrypted root partitions though.
17:56.07Mirelldm-crypt is marvelous for that.
17:56.19rob____yeah, im using dm-crypt with LUKS
17:56.34rob____im a bit skeptical of luks though, the bloke who writes it seems to be a bit mad
17:56.41MirellThen my hard drive decided to slightly crash.
17:56.48MirellWhich is why I no longer encrypted my stuff.
17:56.58rob____did you disable write caching?
17:57.35rob____it shouldnt matter if your machine crashes
17:58.13landleyrob: where do you get working dm-crypt source?  The stuff I found was ancient and out of date.
17:58.29landleyI want to add dm-crypt support to toybox, but the calls you make don't seem particularly documented...
17:58.34rob____dm-crypt is in the kernel
17:58.45landleyrob: I'm talking about the userspace utilities to use it.
17:58.59landleyWhat command do you configure it via?
17:59.07landleydmsetup?
17:59.25rob____http://luks.endorphin.org/dm-crypt
17:59.29rob____cryptsetup
17:59.51landleyHmmm...  I think I tried that one and couldn't get it to build on ubuntu, but I'll try again.  Thanks.
18:00.07rob____it compiled ok here
18:00.34rob____yep, just checked
18:01.10rob____toybox is your cut down version of busybox right?
18:01.22landleyrob: sort of.
18:01.33landleyIt's me starting over from scratch with a project that doesn't have Bruce Disease.
18:01.46landleyFirmware LInux has taken up most of my time for the past couple months, but I'm coming back to toybox now.
18:02.18rob____im going to take a look at your firmware linux stuff tomorrow i think
18:02.34rob____my original intention was to build my system to run off compact flash
18:02.37MirellI need to install Gentoo in Parallels now...
18:02.50rob____fed up of dying harddisks
18:03.08MirellI initially was going to dual-boot the MacBook, but just read the MadWifi bug report, and they *still* don't have access to the new HAL for the wireless
18:12.41sjhillwhirm: you will be getting dbus, dbus-glib and hald
18:40.34Bertllandley: just read your forensics analysis ... are you still planning to rewrite the init?
18:40.58Bertl(or did that already happen? :)
18:43.05landleybertl: I'm not touching busybox anymore.
18:43.08landleyIt has Bruce Disease.
18:43.15landleyBut I plan to put an improved init into toybox.
18:43.28landleyHaven't decided if it should be a system v style init, or more like upstart.
18:43.36landleyPossibly I should just implement both...
18:45.42Bertlwas not talking about busybox stuff :)
18:45.52landleyBut I have about 100 other applets to add to toybox.
18:46.11landleyAt the moment, I'm working on mke2fs/gene2fs because I've found a need for it in firmware linux.
18:46.31BertlI can probably test that in 'unusual' setups, if you are interested in feedback ...
18:46.40landley(There's been a new release of the old genext2fs from sourceforge in the last few days, for the first time in a loooong while.)
18:46.53landleyBut of course it happened _after_ I started writing my own, and I've got up a head of steam, so...
18:47.48landleyI'm also halfway through adapting my bunzip2 code and mdev to busybox.  (Both of which I wrote, so I can use 'em without worrying about Bruce showing up years later and going "I claim this project in the name of my ego, everything you've done has been in the name of Bruce Almighty.".)
18:47.52landleyWhich is a movie I still haven't seen yet...
18:49.46Bertlwell, I do not care that much about Bruce as you seem to do ... but as I said, if you are interested in feedback (i.e. actually want to fix issues/bugs) in toybox, it shouldn't be a problem for me to testdrive it ...
18:55.30*** join/#uclibc Khem (i=Khem@nat/montavista/x-5d00b02714fd81c3)
19:14.00landleyBertl: I'm always interested in feedback. :)
19:16.12Bertlgood to hear ...
19:41.24*** join/#uclibc sjhill_ (n=sjhill@real.realitydiluted.com)
19:48.13Bertllandley: web page says I should bug you regarding tinygcc (as I'm interested :)
19:50.45landleybertl: ah yes.
19:50.51MirellThat project is pretty much dead methought.
19:50.55landleyI need to update the web page: I wound up doing a wrapper script instead.
19:51.13landleyBasically, chopping unnecessary crap out of the gcc front-end turned out to be a heroic underdaking.
19:51.29Bertllike uClibc used to have?
19:51.33landleyI gave up after 10,000 lines removed with no end in sight.  I don't want to maintain a patch that big with no chance of going upstream.
19:51.43landleyBertl: actually based on the old uClibc wrapper script.
19:51.46Bertl(before it was decided that it cannot be maintained)
19:51.53landleyIt's in firmware linux, sources/toys/
19:51.57landleyI'm maintaining it.
19:52.05Bertlah, good! appreciate that!
19:52.07landleyBasically the wrapper overrides all the glibc path searching logic.
19:52.23landleyStarts with --nostdlib --nostdinc and works its way back up from there with "ok, now look HERE".
19:52.40landleyCreating a command line at runtime with the correct paths based on where the wrapper executable lives.
19:52.57Bertltrying to build your firmware toolchain for sparc as we speak
19:53.01landleyThat's how firmware linux is generating a relocatable toolchain.
19:53.10landleyBertl: I don't think I ever made a sparc config file for it.
19:53.17landley(The config files are in packages/configs I think.)
19:53.20Bertlnp, I made one :)
19:53.31landleyI really need to get the miniconfig stuff into the kernel and uClibc.
19:53.43landleyThe big evil part of those config files is that the uClibc and kernel .config files are here documents.
19:53.57landleyWhich makes sense as miniconfig (they're a couple dozen lines, tops.  Generally smaller than that.)
19:54.04landleyBut makes very little sense as full-sized config files.
19:54.05Bertlsparc is more or less my 'hard test case' as it constantly fails with various issues
19:54.10landleyThere's an implicit todo item in there. :)
19:54.20landleyBertl: If you get a working sparc config, please send it my way and I'll include it.
19:54.48Bertlwill do so, once the thing boots with qemu :)
19:54.48landleyIf you need to disable the qemu application emulation sanity test, look at x86_64 to see how to do that.
19:54.54landleyBertl: woot.
19:55.08Mirellqemu is the current killer in the whole thing now.
19:55.18landleyI'm working on adding a gene2fs to toybox as we speak, so I can have the script package up a hard drive image for qemu to boot system emulation from.
19:55.25landleyMirell: yup.
19:55.32landleyExtremely cool project but darn brittle at the moment.
19:55.47Bertllandley: my problem up to now was/is the uclibc, I have working kernel toolchains for all linux archs (well, except fvr atm)
19:56.03landleyAfter they get 0.8.3 out, they're merging Paul Brook's qops rewrite of the code generator, and _that_ should finally make the dependencies on gcc 3.3 go away.
19:56.12Bertl(cross compiling toolchains, that is)
19:56.29landleyBertl: in theory I've beaten uClibc into submission.  In practice, 0.9.29 is a bit untested on some platforms still.
19:56.51landley(Although the armv4 segfault fix went in yesterday and I should upgrade to last night's snapshot to see if I can drop the fix patch.)
19:57.21landleyThat's in download.sh.
19:57.41landleyThat one has the sha1sum check disabled (blank) so if you just point it at a new URL, it should download and use that instead.
19:57.54landleyIt automatically makes version independent symlinks for the build to use, so you don't have to adjust anything else.
19:58.22landleyDid I mention there's a mailing list?  (Very low traffic at present...)
20:06.43Bertlurl to subscribe?
20:07.10Bertlnevermind, just found it :)
20:09.53Bertllandley: is there an IRC channel too? :)
20:10.08landleyBertl: afraid not.
20:10.17landleyIf they start complainign here I'm happy to start one, though. :)
20:10.29Bertlhow about making one on oftc or freenode right now?
20:10.36landley(I hang out here when I'm online.  Here, #cross-lfs, #gentoo-embedded, occasionally a few others.)
20:10.46landleyI don't have a server set up to do a bot to stay logged in.
20:10.54landleyAnd I'm not sure of the procedure for registering channels.
20:11.21BertlI can do that for you if you like
20:11.30landleyBoth #fwl and #firmware seem free at the moment.
20:11.45Bertl(at least for oftc I know the procedures :)
20:11.57landleyprobably #firmware is the better choice.
20:12.09landleyI don't hang out on oftc, though.
20:12.23landleyIsn't that the one Debian started because freenode wasn't idealogically pure enough?
20:12.32landleyI vaguely remember reading about that on lwn...
20:12.59Bertlat some point it was freenode vs oftc
20:13.35Bertlbut shouldn't be too hard on freenode either ...
20:15.29Bertllandley: okay, please join #firmware, so you 'own' it atm
20:19.59*** join/#uclibc Gerrath (n=Shane_@unaffiliated/gerrath)
20:20.42landleyBertl: done
20:22.55*** join/#uclibc sjhill (n=sjhill@real.realitydiluted.com)
21:52.38*** part/#uclibc PhB (n=PhB@dD5E060DA.access.telenet.be)
21:56.55*** join/#uclibc psyco (n=psyco@APuteaux-153-1-15-189.w82-124.abo.wanadoo.fr)
22:39.17CIA-1003jocke * r17253 10uClibc/ (7 files in 5 dirs):
22:39.17CIA-10Peter Mazinger pointed out that my last commit was faulty.
22:39.17CIA-10This should fix it.
22:39.28CIA-1003vda * r17254 10busybox/networking/ftpgetput.c: fix my own breakage
22:47.01*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-3-195.dsl.pltn13.sbcglobal.net)
22:52.26CIA-1003jocke * r17255 10uClibc/TODO: I fixed that item long ago.
23:17.04*** join/#uclibc blindvt__ (n=bf@62.47.131.130)
23:26.14CIA-1003vda * r17256 10busybox/ (include/usage.h networking/fakeidentd.c networking/inetd.c): (log message trimmed)
23:26.15CIA-10fakeinetd: attempted ipv6-ization (and it's done)
23:26.15CIA-10but entire applet can be orders of magnitude smaller
23:46.44*** join/#uclibc KhemHome (n=KhemHome@adsl-71-146-3-195.dsl.pltn13.sbcglobal.net)

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with blootbot logs, split per channel, etc.