irclog2html for #uclibc on 20100117

00:06.31CIA-4303vda.linux 07master * r5b9a91074917 10/networking/ntpd.c: ntpd: disable frequency estimation code
00:08.29*** join/#uclibc kolbyjack (n=jkolb@113.229.189.72.cfl.res.rr.com)
01:52.06CIA-4303vda.linux 07master * r9b20adca4b2b 10/networking/ntpd.c: ntpd: add anti-clock-hopping code
02:01.54CIA-4303vda.linux 07master * r57f46c1afd28 10/networking/ntpd.c: ntpd: remove some code which is at best unneeded and at worst wrong
03:34.31*** join/#uclibc Salec (n=Salec@c-174-54-255-117.hsd1.pa.comcast.net)
03:35.04Salechey guys, just finished making a buildroot image and I cannot seem to use gcc on the target machine, when I tried compiling a simple hello world app I got can't load library 'libmpfr.so'
03:35.07Salecany ideas?
04:39.22*** join/#uclibc Salec_ (n=Salec@c-174-54-255-117.hsd1.pa.comcast.net)
04:57.49*** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan)
05:49.02*** join/#uclibc swrh (n=mind@189-55-80-116-nd.cpe.vivax.com.br)
05:53.35swrhhi. is it possible to compile a kernel and "attach" the initramfs to it so it is automatically executed without needing to tell grub/lilo/loadlin/syslinux/whatever that i have an initramfs/initrd?
06:10.58*** part/#uclibc swrh (n=mind@189-55-80-116-nd.cpe.vivax.com.br)
07:47.31*** join/#uclibc dileX (n=sd@p5B2EEDBE.dip.t-dialin.net)
07:47.37dileXhi
07:48.18dileXI have several problems while creating a new target toolchain for a ti-ar7 router-platform (mipsel arch)
07:49.11dileXfirst, there exists no uClibc-locale-20081111-32-el.tgz file for downloading
07:49.21dileX(see http://old.nabble.com/uclibc-locale-td23865388.html)
07:50.39dileXwas this file ever published?
07:51.45dileXthe next problem I had to solve was "...mipsel-linux-uclibc/bin/ld: this linker was not configured to use sysroots"
07:52.02dileXthis worked fine with adding "--with-sysroot"
07:52.13dileXpatch see http://pastebin.ca/1754439
07:53.19dileXthe first problem with missing uclibc-locale I could override via http://pastebin.ca/1754441
07:53.49dileXUCLIBC*PREGENERATED_LOCALE_DATA is not set
07:55.55dileXnow, I have a problems with uclibc/utils, especially with iconv.c
07:56.16dileXbroken build-log: http://pastebin.ca/1754444
07:56.51dileXiconv.c:(.text+0x18e): undefined reference to `__iconv_codesets'
07:56.51dileXiconv.c:(.text+0x1b4): undefined reference to `__iconv_codesets'
07:56.56dileXany idea? help?
07:57.27dileX(note: target-toolchain is gcc-4.4.2-uClibc-0.9.30.2 for mipsel)
08:01.55dileXtoolchain/make/target/uclibc/Config.mod.0.9.30.2: http://pastebin.ca/1754448 (for the sake of completeness)
08:04.02dileXI tried binutils 2.19.1 and 2.20.51 (snapshot) as "target-binutils" (both break)
08:06.00dileXuClibc-fix-locale-version-string.patch: http://pastebin.ca/1754453 (same breakage with uclibc/utils/iconv.c)
08:24.42*** join/#uclibc _Lucretia_ (n=munkee@pdpc/supporter/active/lucretia)
08:25.06*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
10:41.43*** join/#uclibc trem (n=trem@mol92-1-81-57-136-23.fbx.proxad.net)
10:55.08*** join/#uclibc ymorin_ (n=ymorin@ARennes-252-1-3-165.w83-195.abo.wanadoo.fr)
11:56.42*** join/#uclibc gatty (n=agatward@2a01:348:11b:beef:25e6:4c76:fbcd:84a5)
12:31.14*** join/#uclibc gatty^ (n=agatward@2a01:348:11b:beef:44d0:1a85:c3c5:eb1)
12:47.10*** join/#uclibc ymorin_ (n=ymorin@ARennes-252-1-3-165.w83-195.abo.wanadoo.fr)
13:33.31*** join/#uclibc adb (n=adb@144.85.45.157)
15:10.26*** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa)
15:13.22*** join/#uclibc _Lucretia__ (n=munkee@5acb92f5.bb.sky.com)
16:00.34*** join/#uclibc ymorin_ (n=ymorin@ARennes-252-1-3-165.w83-195.abo.wanadoo.fr)
16:04.13*** join/#uclibc swrh (n=mind@189-55-80-116-nd.cpe.vivax.com.br)
16:25.09*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
16:56.39*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
16:56.55*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
16:58.07*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
18:09.23CIA-4303vda.linux 07master * r96a6bdcb7764 10/testsuite/bunzip2.tests: bunzip2.tests: replace $? with \$?
18:16.03*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
18:19.49*** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net)
18:20.12dougmenckenhi! how to get super strip (sstrip) without buildroot?
18:21.44solarhttp://dev.gentoo.org/~solar/misc/sstrip-mjn3.c
18:22.36dougmenckenonly one source file? all I need is to cuild that one file?
18:22.44dougmenckens/cuild/build/
18:22.46solaryep
18:22.53dougmenckenwow, thanks
18:23.43dougmenckenis it really much better than binutils' strip?
18:24.28solarit's multiplatform and strips off a lot more then standard binutils.
18:24.53dougmenckencool
18:25.01dileXhi
18:25.04solarif you know you wont ever be debugging the binary via gdb. then it's suited.
18:25.19dougmenckenI have no gdb on my system ;)
18:25.50solardon't use it blindly on a desktop system
18:25.50dileXI have a build problem with uclibc-0.9.30.2 and utilc/iconv.c: http://pastebin.ca/1754444
18:26.21dileXbuilding a toolchain with gcc-4.4.2 for mipsel
18:27.09dougmenckendileX, use libiconv
18:27.27dougmencken(as me does :)
18:27.59dileXor dont built utils
18:28.04dougmenckenit works well and doesn't require any "locales" thing
18:28.04dileXbuild*
18:28.41dileXhow to use libiconv? via configure option/flag or how?
18:29.19dougmenckendileX, just download it from http://www.gnu.org/software/libiconv/
18:29.37dougmenckenhttp://ftp.gnu.org/pub/gnu/libiconv/libiconv-1.13.1.tar.gz
18:30.26dileXhow to include in uclibc?
18:30.38dougmenckendileX, why?
18:32.21dougmenckenany reason to include character set conversion lib and program in the core C library?
18:34.01dileXI didnt get the point on how to use/build-against uclibc and libiconv
18:35.01dougmenckendileX, if your system has libiconv.so, autoconf or ./configure will find it
18:35.37dileXhmm, I am cross-compiling on an i386 host for mipsel
18:36.27dougmenckenif your target system has ....
18:38.59dileXno. and means re-writing existing make-/build-system.
18:39.18dileXfrom freetz project
18:39.26dougmenckendileX, the main gotcha is that the libiconv tooks about 1-1.5 MiB; you don't need to rebuild toolchain
18:41.23dileXyes, I understand
18:41.56dileXchanging to utils-dir and re-invoking make, iconv is built
18:42.07dileXafter breakage
18:49.26solardileX: are you planing on running gtk/kde based apps on the device?
18:49.45solarif not. You 99% probably don't need any iconv/local support at all.
18:50.10dileXno. toolchain is for a ti-ar7 based router-platform.
18:50.28solara router. I would leave iconv out of uclibc all together then
18:52.17dileXyeah, I agree
18:52.50dileXcould it be my linker is causing troubles? next breakage: http://paste.debian.net/56953/
18:55.29dougmenckendileX, have you tried to ./configure --without-nsl ?
18:55.33dougmenckennls
19:00.13dileXbetter would be to know UCLIBC_* for dot-config
19:01.42dileXseems to be ENABLE_NLS
19:03.36dougmenckendileX, do you really need "native language support" for your router?
19:06.30dougmenckenbtw, before sstrip: "-rwxr-xr-x    1 root     0           50771 Jan 15 07:58 iconv"
19:06.38dougmenckenafter: "-rwxr-xr-x    1 root     0           20752 Jan 17 19:06 iconv"
19:06.41dougmenckenvery cool
19:06.55dileXhttp://trac.freetz.org/browser/trunk/toolchain/make/target/uclibc/uclibc.mk#L123
19:07.15dileXcant switch off build of utils completely
19:08.50dougmenckendileX, yes, just edit your .config file and correct the following lines:
19:08.52dougmencken# UCLIBC_HAS_LOCALE is not set
19:09.05dougmenckenthen make oldconfig
19:09.33dileXUCLIBC_HAS_LOCALE=y
19:09.40dileXOK, I will change that
19:21.44dougmenckensolar, so maybe you know how to automate sstrip for any binary in just-built package? for example, GCC
19:22.02dougmenckens/any binary/any ELF/
19:23.03dileXdougmencken: thanks in advance. I build the toolchain now.
19:24.29dougmenckendileX, any time :)
19:32.54*** join/#uclibc Salec (n=Salec@c-174-54-255-117.hsd1.pa.comcast.net)
19:33.48Salechey guys, for a system that isn't very limited on size (256mb memory and 8gb of disk space)  is uclibc still faster than glibc?
19:34.29dougmenckenSalec, it is; due to "no s%#$1 legacy"
19:34.39Saleclol
19:35.25Saleccool :)  next question, in buildroot's config there is an option for copying the toolchain to the target, does thi build a separate version of the toolchain that can execute on the target?  or does it just copy it?
19:35.50dougmenckenif the main goal is to support legacy proprietary programs, then the lib is just a crap
19:36.17Salecyea I don't care about legacy or even localization
19:36.59dougmenckenSalec, wait a minute, there definitely EXISTS that option
19:37.22Saleci've seen it in 2 places, actually it was a bit confusing as to what it actually did though
19:37.44Salecon Build Options there's a "development files on the target filesystem"
19:38.01dougmenckenyes, IT is
19:38.08Salecthere's also a "native toolchain in the target filesystem" option for Package Selection for the target
19:38.25Salecdoes that mean buildroot will actually build gcc twice?
19:38.27dougmenckenSalec, dev files are headers
19:38.31Salecohhh
19:38.37dougmenckenyou need them both
19:40.58Saleck
19:41.08Salecthese scripts are pretty awesome
19:42.19Saleci keep finding a bunch of oddities though
19:43.39dougmenckenbtw, I found a one-liner to find all ELFs: "find / -print0 | xargs -0 file | grep ELF"
19:45.27dougmenckenalso, after sstrip 'file' prints "corrupted section header size" << is it okay?
19:45.29solarheh
19:45.49dougmenckenfor non-sstripped, it prints "not stripped"
19:46.31dougmenckensolar, any known patch for 'file'?
19:46.55solarin what terms?
19:47.14dougmenckenwell, "corrupted" sounds like "not working"
19:47.21dougmenckenbut it works
19:47.24solarwell that is expected.
19:47.36solarcvs -d :pserver:anonymous@anoncvs.gentoo.org:/var/cvsroot co gentoo-projects/pax-utils
19:47.40dougmenckenwhat is expected?
19:47.58solarthat file will say a section header is corrupted after it's been sstriped
19:48.20dougmenckensolar, oh, but why?
19:48.20solaryou should only sstrip ET_EXEC files at first imo vs all ELF files.
19:48.27solarwhy does it say that?
19:48.50solarcuz it's checking for a a header that does not exist. sstrip pulled it off.
19:48.56dougmenckenyep and how to make it say "stripped" instead of ""corrupted section header"?
19:49.14solaryou don't.
19:49.29solarand it's handy this way cuz you know what files have been sstriped
19:49.45solarother elf formats show up the same way. This is a limitation of file.
19:50.15dougmenckenah okay; so this "corrupted section header" just means "you've done stripping well", right?
19:50.17solarI think a upx packed elf will say the same
19:51.39dougmenckenalso how to sstrip all of them? "find / -print0 | xargs -0 file | grep ELF | sstrip" just prints sstrip usage
19:51.40*** join/#uclibc gatty (n=agatward@2a01:348:11b:beef:f5f0:160d:981e:a35b)
19:51.57dougmenckeneven with "-"
19:52.43dougmenckenoh, I'm stupid, sorry; 'file's output isn't paths to files
19:54.49solaryeah but don't do that
19:55.03dougmenckensolar, I want to automate sstripping
19:55.06solarthat would try and sstrip ET_REL files..
19:55.18solarI hear ya. thats why I pointed ya at pax-utils
19:57.13solarwhich has scanelf.
19:57.15solarscanelf -R . -ym -E ET_EXEC  -BF%
19:57.32solarwould give you a list of executables it's safer to sstrip
19:58.11solarin general you would never want to run sstrip on an elf till you are ready to make the final firmware image
19:58.54solarcuz if it's been sstriped you wont be able to link things you might still be building and linking to
20:01.37dougmenckensolar, i.e. sstripped lib isn't recognizable by ld-uClibc.so*?
20:02.39dougmenckenbtw, cpio is a part of busybox (and a part of pax-utils)
20:02.50solarwrong paxutils
20:02.56dougmenckengnu ones
20:03.18dougmenckenhttp://www.gnu.org/software/paxutils/ << GNU paxutils is a suite of archive utilities: it will provide cpio, tar and POSIX pax archivers.
20:03.32solarit's a set of tools that vapier and myself wrote that deal with elf files.
20:03.33solarhttp://sources.gentoo.org/viewcvs.py/gentoo-projects/pax-utils/
20:03.34dougmenckennot this one?
20:03.51solarhttp://www.gentoo.org/proj/en/hardened/pax-utils.xml
20:04.05dougmenckenhmm.. it looks like gentoo-specific thing
20:04.13solarnot at all
20:04.45dougmenckenlet me test (stripping anything in /lib and /usr/lib)
20:04.52*** join/#uclibc adb (n=adb@144.85.45.157)
20:04.56solarnot real /lib I hope.
20:05.02dougmencken"Text file busy"
20:05.05dougmenckenlol
20:05.54dougmencken"missing ELF signature" << not so lol :
20:06.27dougmenckenbut... it works well after sstripping ;)
20:07.17dougmenckenso I can say any lib may be safely sstripped
20:07.36solarnot really
20:07.44dougmenckenany exception?
20:07.48dougmenckenhow to test it?
20:07.58solarthere are times you should not strip an elf based on it's .note section flags
20:08.35solarI know
20:09.26dougmenckenbut I know that is a "section" in a binary container
20:15.01*** part/#uclibc dazman (n=dazman@unaffiliated/dazman)
20:40.45dileXdougmencken: where can I find sstrip?
20:44.34dougmenckenhttp://dev.gentoo.org/~solar/misc/sstrip-mjn3.c
20:46.34dougmenckenthen just build it via "gcc -O3 sstrip-mjn3.c -o sstrip"
20:54.33*** join/#uclibc swrh (n=mind@189-55-85-3-nd.cpe.vivax.com.br)
21:02.10dougmenckensolar, you were right; no my gcc can't compile anything :(
21:31.53*** join/#uclibc trem_ (n=trem@AToulouse-256-1-32-159.w90-38.abo.wanadoo.fr)
21:33.07CIA-4303vda.linux 07master * r33f9dc08e55b 10/networking/ftpd.c: ftpd: code shrink
21:50.23*** join/#uclibc swrh (n=mind@189-55-85-3-nd.cpe.vivax.com.br)
22:28.48*** join/#uclibc kolbyjack (n=jkolb@113.229.189.72.cfl.res.rr.com)
23:02.36tremnite all, sweet dreams
23:16.08CIA-4303jacmet 07master * r6a772086b7c4 10/package/mtd/mtd-utils/mtd.mk: package/mtd: mtd-utils: fix fec.c build (bcopy reference)
23:48.38*** join/#uclibc swrh (n=mind@189-55-85-3-nd.cpe.vivax.com.br)

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