| 00:12.08 | CIA-49 | 03vda.linux 07master * re66a09ba9c16 10busybox/networking/dnsd.c: dnsd: add more -v messages |
| 00:35.19 | *** join/#uclibc trem (~trem@mol92-1-81-57-136-23.fbx.proxad.net) |
| 00:45.33 | *** join/#uclibc malloc (~malloc@2001:470:1f11:80b:216:3eff:fe55:f8b0) |
| 00:48.15 | *** join/#uclibc _Lucretia_ (~munkee@pdpc/supporter/active/lucretia) |
| 01:45.54 | CIA-49 | 03vda.linux 07master * r343dfd7abe44 10busybox/ (include/usage.h networking/dnsd.c): dnsd: add -s option. This allows (clumsy) operation with read dns servers |
| 02:40.15 | *** join/#uclibc dougmencken (~Douglas@93.123.156.139) |
| 02:46.44 | *** join/#uclibc _Lucretia__ (~munkee@5ac2a28d.bb.sky.com) |
| 02:53.45 | *** join/#uclibc neutronscott (~irc@157.132.121.18) |
| 02:54.14 | *** part/#uclibc neutronscott (~irc@157.132.121.18) |
| 05:25.14 | *** join/#uclibc animeloe[net] (~animeloe@unaffiliated/animeloe) |
| 05:42.28 | *** join/#uclibc animeloe[net] (~animeloe@unaffiliated/animeloe) |
| 10:45.01 | *** join/#uclibc trem (~trem@mol92-1-81-57-136-23.fbx.proxad.net) |
| 13:19.00 | *** join/#uclibc tsukasa (~tsukasa@unaffiliated/tsukasa) |
| 15:12.46 | *** join/#uclibc animeloe[net] (~animeloe@unaffiliated/animeloe) |
| 16:14.33 | *** join/#uclibc austinf (~austinf@75-92-1-28.war.clearwire-wmx.net) |
| 16:18.18 | dougmencken | austinf, hi! |
| 16:18.29 | dougmencken | I have a question for you |
| 16:19.15 | austinf | hello |
| 16:19.55 | dougmencken | so the question: I'm building uclibc with LINUXTHREADS_OLD=y and it then says "library contains unsupported TLS" |
| 16:20.21 | dougmencken | i.e. ld-uclibc.so blames libc.so |
| 16:20.32 | dougmencken | what is workaround? |
| 16:22.12 | austinf | can you tell which symbol it is complaining about? |
| 16:23.06 | austinf | and also, is tls enabled in your gcc? |
| 16:23.07 | dougmencken | "/bin/sh: '/lib/libc.so.0' library contains unsupported TLS" |
| 16:23.51 | dougmencken | nothing special about TLS in gcc -v |
| 16:24.38 | dougmencken | also http://pastebin.com/m3fb21bb4 <-- the previously build uclibc and the last one; how to get symbol? |
| 16:25.36 | austinf | objdump -t |
| 16:25.39 | austinf | is one way |
| 16:26.01 | austinf | my libc.a doesn't contain any .tdata or .tbss though |
| 16:26.12 | austinf | does yours? |
| 16:26.42 | dougmencken | objdump -t /lib/libc.so.0? wait a moment |
| 16:26.57 | austinf | that might not work if it has been stripped |
| 16:27.14 | dougmencken | for old one, it prints "SYMBOL TABLE:no symbols"; yep, it was stripped |
| 16:27.23 | austinf | try the .a |
| 16:27.50 | dougmencken | can't find .a |
| 16:28.07 | dougmencken | /usr/lib, okay |
| 16:28.29 | dougmencken | millions of lines |
| 16:29.13 | dougmencken | pastebin them? |
| 16:34.37 | austinf | you can just search it for tdata and tbss |
| 16:35.07 | austinf | and also, why is it in /? |
| 16:35.15 | austinf | if you are just building uclibc |
| 16:35.36 | austinf | are you building something linked against uclibc? |
| 16:35.52 | dougmencken | it's chrooted env |
| 16:36.05 | dougmencken | I'm building inside it |
| 16:36.16 | dougmencken | anything is linked against uclibc |
| 16:36.26 | austinf | and what is failing? |
| 16:36.56 | dougmencken | uclibc; as I understood, ld-uclibc.so doesn't like libc.so |
| 16:39.25 | austinf | hmm, I don't think that is enough info |
| 16:39.43 | austinf | can you explain what you are trying to build/do? |
| 16:39.49 | dougmencken | before I used (and forced to still use) 778-g784e157, and there was no TLS in it, the version I tried (which has TLS in it) 852-gf7eba78 |
| 16:39.57 | dougmencken | I'm trying to re-build uclibc |
| 16:40.15 | dougmencken | with FTW support (because e2fsprogs require it) |
| 16:41.16 | dougmencken | objdump -t --> http://pastebin.com/m4038c080 |
| 16:41.58 | austinf | so, no tls symbols in there |
| 16:42.03 | austinf | hmm |
| 16:42.46 | austinf | can you run make V=1 and pastebin that? |
| 16:43.22 | dougmencken | rebuild the same version with the same config and with make V=1? |
| 16:43.36 | dougmencken | and what to pastebin then? make V=1 > somelog? |
| 16:45.34 | austinf | you don't have to rebuild fully, I just want to the see the failing command |
| 16:47.27 | dougmencken | it rebuilds now |
| 16:50.23 | *** join/#uclibc tsukasa (~tsukasa@unaffiliated/tsukasa) |
| 16:56.03 | dougmencken | oops, I forgot about 2>somelog too, so warnings will be omitted from somelog |
| 16:56.49 | dougmencken | usually, it tooks 30-40 minutes to build uclibc |
| 17:01.14 | austinf | I don't need the whole build log, just the last command that failed most likely |
| 17:02.28 | dougmencken | libc/stdio/fflush.c:101:2: warning: #warning unnecessary check if no threads << something about threads |
| 17:02.52 | austinf | nope, not relevent |
| 17:03.01 | austinf | and that's just a warning |
| 17:06.52 | dougmencken | btw (while uclibc still builds), e2fsprogs has a very trivial error (they missed 'struct' keyword: http://manulix.wikidot.com/local--files/history/closefs.patch), but it refuses to build; I wonder how *nobody* noticed that before? so I think warnings are important |
| 17:09.15 | dougmencken | last lines of somelog --> http://pastebin.com/m2150191 |
| 17:09.27 | dougmencken | (yep, build finished) |
| 17:11.05 | dougmencken | paste from tty output --> http://pastebin.com/m1f0595e1 |
| 17:17.18 | dougmencken | # readelf -d lib/libc.so.0 | grep -i tls |
| 17:17.19 | dougmencken | <PROTECTED> |
| 17:19.00 | dougmencken | I can upload just-built-uclibc, I think readelf and objdump will work |
| 17:19.30 | dougmencken | ls -l |
| 17:19.41 | dougmencken | oh sorry |
| 17:25.24 | dougmencken | austinf, ^^ |
| 17:27.41 | austinf | hmm, so the build did finish? where is the error you were hitting? |
| 17:29.39 | dougmencken | the error "/bin/sh: '/lib/libc.so.0' library contains unsupported TLS" (and then the same another one, and then "/bin/sh: can't load library 'libc.so.0'")? |
| 17:30.07 | dougmencken | it's when I use new uclibc on chrooted env and trying to chroot |
| 17:30.57 | dougmencken | it's a message from ld-uclibc.so, from source file ./ldso/ldso/dl-elf.c |
| 17:32.55 | dougmencken | if (ppnt->p_type == PT_TLS) { ... |
| 17:33.38 | dougmencken | and it appeared only recently, I have successfully build and ran uclibc-git-nptl before |
| 17:36.33 | austinf | got it |
| 17:36.37 | *** join/#uclibc tsukasa (~tsukasa@unaffiliated/tsukasa) |
| 17:39.08 | dougmencken | austinf, so what to do? |
| 17:40.22 | austinf | I don't have time to setup a repro environment right now unfortunately |
| 17:40.42 | austinf | this is the nptl branch? |
| 17:40.48 | dougmencken | yes |
| 17:40.59 | dougmencken | 852-gf7eba78 |
| 17:41.40 | austinf | are you sure it isn't becaue you are mixing old and new binaries? |
| 17:42.25 | dougmencken | nope, I'm using bootstrap script to quickly copy-and-extract the specified packages |
| 17:43.18 | dougmencken | also, config --> http://manulix.wikidot.com/files:uclibc-config |
| 17:46.51 | dougmencken | part of the script is rm -rf "${CHROOT_DIR}" && mkdir "${CHROOT_DIR}", so there's no chance to mix old uclibc and new uclibc |
| 17:49.41 | austinf | so it is definitely a regression then |
| 17:49.59 | austinf | ok |
| 17:51.23 | austinf | hmm, I don't see that git id on the nptl branch |
| 17:51.42 | austinf | top of nptl branch is f7eba78a7464c for me |
| 17:51.53 | dougmencken | btw, non-working-on-my-platform ("make: *** No rule to make target `nptl_arch_headers', needed by `pregen'. Stop.") UCLIBC_HAS_THREADS_NATIVE=y sets UCLIBC_HAS_TLS=y |
| 17:52.06 | austinf | ah |
| 17:52.07 | austinf | ppc |
| 17:52.12 | dougmencken | git pull uClibc-git... already up-to-date (0_9_30_rc3-852-gf7eba78) |
| 17:52.17 | austinf | yes, blindvt has patches for that |
| 17:52.45 | *** join/#uclibc blindvt` (~b@chello213047219050.surfer.at) |
| 17:52.52 | dougmencken | blindvt`, hi |
| 17:53.00 | blindvt | hi |
| 17:53.21 | dougmencken | can you help me with the problem mentioned above? |
| 17:53.28 | blindvt | i didn't follow any discussion |
| 17:53.40 | austinf | 0_9_30_rc3-852-gf7eba78? Is that your local branch name? |
| 17:53.55 | dougmencken | git describe tells me so |
| 17:54.15 | austinf | heh |
| 17:54.43 | dougmencken | blindvt, have you tried to build the latest git-nptl? |
| 17:54.53 | dougmencken | and to run it then |
| 17:54.57 | blindvt | my local branch name? no. my local branch name is 0.9.30 |
| 17:55.22 | austinf | blindvt, sorry I was trying to figure out what code dougmencken is running |
| 17:55.42 | blindvt | dougmencken, what do you want to do and what do you do ? |
| 17:56.11 | austinf | (09:31:33 AM) dougmencken: the error "/bin/sh: '/lib/libc.so.0' library contains unsupported TLS" (and then the same another one, and then "/bin/sh: can't load library 'libc.so.0'")? |
| 17:56.11 | austinf | (09:32:01 AM) dougmencken: it's when I use new uclibc on chrooted env and trying to chroot |
| 17:56.11 | austinf | (09:32:51 AM) dougmencken: it's a message from ld-uclibc.so, from source file ./ldso/ldso/dl-elf.c |
| 17:56.12 | austinf | (09:34:50 AM) dougmencken: if (ppnt->p_type == PT_TLS) { ... |
| 17:56.12 | austinf | (09:35:32 AM) dougmencken: and it appeared only recently, I have successfully build and ran uclibc-git-nptl before |
| 17:56.29 | dougmencken | blindvt, I want to re-build the bleeding-edge uclibc with FTW support |
| 17:56.58 | blindvt | dougmencken, then you shouldn't be on nptl but either on 0_9_30 or master (preferably the former) |
| 17:57.21 | dougmencken | blindvt, but I built nptl branch before and it was okay |
| 17:57.28 | austinf | dougmencken, git-describe on nptl branch is kind of useless, since no relevant tags have been put in there |
| 17:57.51 | dougmencken | so what to depend on in versioning? |
| 17:58.07 | dougmencken | how to call package, for example? |
| 17:58.16 | blindvt | austinf, don't know offhand, turn off forcing shareable segs if you have relocs in there or handle tls properly, i'd guess |
| 17:58.25 | dougmencken | now it is uClibc-0.9.30-rc3-852-gf7eba78 |
| 17:58.33 | austinf | ...well since it is an in development branch, use the git hash directly |
| 17:58.45 | austinf | that string git-desribe spits out is useless |
| 17:58.50 | austinf | for this branch |
| 17:59.37 | austinf | "git rev-list --max-count=1 HEAD" is what use |
| 18:01.36 | dougmencken | a huge hex line, no connection with base version |
| 18:02.07 | dougmencken | even in .config, it has 0.9.30-git |
| 18:04.25 | dougmencken | but on TLS, what happened with uclibc-nptl? why it complains on itself now? |
| 18:04.48 | dougmencken | "Yup, the user was an idiot and tried to sneak in a library with" |
| 18:07.02 | dougmencken | also, if I will use master, and then at some time you'll switch to ntpl in master, then what? |
| 18:19.10 | austinf | re: huge hex line. It uniquely identifies what code you are running, so that is very helpful for developers trying to help you |
| 18:20.18 | austinf | re: tls, sounds like there is potentially a regression. you might try blindvt's suggestions, or look through the commits to that code since you knew it was last working. |
| 18:22.52 | dougmencken | imho, additional 20 hex digits in "f7eba78a7464c7b31326abf92dab254608835028" vs "852-gf7eba78" (also excluding commit number) doesn't make any sense; git log -p what? |
| 18:23.36 | dougmencken | I don't want to revert to anything older than my previous 778-g784e157 |
| 18:24.01 | austinf | I don't know what commits those correspond to |
| 18:24.16 | austinf | so I can't look at the affected code and see if anything might have been changed |
| 18:24.38 | austinf | also, you can typically just use the first 10 digits of the git commit id |
| 18:25.10 | dougmencken | git name-rev 852-gf7eba78 --> 852-gf7eba78 nptl |
| 18:25.38 | CIA-49 | 03rep.dot.nop 07master * rf16d7c4bdf0d 10busybox/scripts/bloat-o-meter: bloat-o-meter: fix quadric behavior in alias resolution |
| 18:26.22 | austinf | that still isn't a commit id |
| 18:27.15 | dougmencken | okay, I will name this package uClibc-0.9.30-rc3-852-f7eba78a7464c7b31326abf92dab254608835028 |
| 18:28.43 | austinf | :) |
| 18:28.46 | dougmencken | (if I can successfully use it, even uClibc-0.9.30-rc3-852-f7eba78a7464c7b31326abf92dab254608835028-austinf-blindvt) |
| 18:29.07 | austinf | anyways |
| 18:29.37 | austinf | what was the date/time of the last known good for you |
| 18:29.41 | austinf | we can look it up that way |
| 18:30.21 | dougmencken | 778-g784e157, i.e. 784e157xxxxxxxxxxxxxxxxxxxxxxxxxx |
| 18:30.52 | dougmencken | I will ask on #git how to get info based on commit number |
| 18:30.56 | austinf | oh ok, so not that long ago |
| 18:31.08 | austinf | was the info I was asking for |
| 18:31.10 | austinf | 784e157 |
| 18:32.29 | austinf | there are a few relevant commits to ldso in that time period |
| 18:32.42 | austinf | I've got to take off now, but I'll look at it later |
| 18:32.53 | austinf | bye |
| 18:33.36 | dougmencken | good luck |
| 18:34.10 | *** join/#uclibc animeloe[net] (~animeloe@unaffiliated/animeloe) |
| 18:41.28 | *** join/#uclibc animeloe[net] (~animeloe@unaffiliated/animeloe) |
| 18:42.14 | dougmencken | git rev-parse f7eba78 --> f7eba78a7464c7b31326abf92dab254608835028 |
| 18:42.16 | CIA-49 | 03rep.dot.nop 07master * reb255752c146 10busybox/scripts/bloat-o-meter: bloat-o-meter: convert to print() calls |
| 18:42.26 | dougmencken | so there's no need in huge hex line |
| 18:44.09 | animeloe[net] | hehe |
| 18:54.08 | dougmencken | animeloe[net], what? they looked me as on an idiot and said "dougmencken: it's after the g", "dougmencken: it's enough", "dougmencken: it is, if it's a unique prefix" |
| 18:59.01 | dougmencken | blindvt, what is that bloat-o-meter? |
| 18:59.37 | dougmencken | /bin/sh: bloat-o-meter: not found |
| 19:04.33 | solar | a busybox/scripts/bloat-o-meter |
| 19:05.15 | solar | it also calls #!/usr/bin/python |
| 19:06.02 | dougmencken | ah, my system has only perl now |
| 19:06.50 | dougmencken | btw, why "#!/bin/sh" but "#!/usr/bin/perl" << *usr*; why to hardcode |
| 19:07.26 | dougmencken | (paths; isn't it so easy to `which perl`?) |
| 19:07.38 | *** join/#uclibc PortaLu (~LucretiaL@5ac2a28d.bb.sky.com) |
| 19:07.44 | solar | perl and most scripting interps live in /usr/bin/ FHS |
| 19:08.34 | dougmencken | solar, you can --prefix="/"; and by deafult, prefis is "/usr/local" |
| 19:08.42 | dougmencken | prefix* |
| 19:08.58 | dougmencken | ibot, who is steve ballmer |
| 19:08.59 | ibot | i guess steve ballmer is http://www.youtube.com/watch?v=tGvHNNOLnCk, or http://www.youtube.com/watch?v=wvsboPUjrGc&feature=related |
| 19:09.07 | dougmencken | I like him |
| 19:11.24 | dougmencken | (not steve ballmer, but ibot :) |
| 19:11.46 | dougmencken | ibot, who is dougmencken |
| 19:11.47 | ibot | you are, like, Douglas Mencken, the legenday creator of powerpc-uclibc 100% free gnu/linux distro located at manulix.wikidot.com |
| 19:12.03 | dougmencken | I love him |
| 19:12.31 | dougmencken | ibot, s/legenday/legendary/ |
| 19:14.39 | solar | dougmencken: cuz most scripts have it hard coded to those loccations. |
| 19:15.00 | solar | you can add some logic to a script to replace all the shebangs before being installed |
| 19:15.25 | *** part/#uclibc plaes (~plaes@ns.store20.com) |
| 19:19.10 | *** join/#uclibc tsukasa (~tsukasa@unaffiliated/tsukasa) |
| 19:21.14 | *** join/#uclibc malloc (~malloc@2001:470:1f11:80b:216:3eff:fe32:7c19) |
| 19:41.33 | *** join/#uclibc Dj-Death (~djdeath@potipota.net) |
| 19:50.25 | *** join/#uclibc Dj-Death (~djdeath@potipota.net) |
| 20:37.01 | *** join/#uclibc tchan (~tchan@lunar-linux/developer/tchan) |
| 21:09.30 | *** join/#uclibc austinf (~austinf@75-92-1-28.war.clearwire-wmx.net) |
| 21:11.01 | Mirell | More people need to use #!/usr/bin/env |
| 21:11.34 | CIA-49 | 03vapier 07master * r6dba356ee09f 10busybox-website/ (FAQ.html lists.html): drop "subversion" from a few more places |
| 21:21.12 | *** join/#uclibc trem (~trem@AToulouse-256-1-4-224.w90-38.abo.wanadoo.fr) |
| 22:28.51 | CIA-49 | 03vda.linux 07master * r3205c24a822c 10busybox/coreutils/md5_sha1_sum.c: shaNNNsum: accept and ignore -b and -t |
| 22:32.41 | trem | nite all, sweet dreams |
| 22:44.35 | dougmencken | good morining |
| 23:25.35 | *** join/#uclibc animeloe[net] (~animeloe@unaffiliated/animeloe) |