| 00:32.47 | *** join/#uclibc hiyuh (n=hiyuh@ZK138196.ppp.dion.ne.jp) |
| 01:00.01 | *** join/#uclibc ibot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 01:00.02 | *** topic/#uclibc is setting the topic is fun |
| 01:17.58 | *** join/#uclibc ibot_ (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 01:17.58 | *** topic/#uclibc is setting the topic is fun |
| 02:52.42 | *** join/#uclibc CIA-10 (i=cia@cia.navi.cx) |
| 04:13.45 | *** join/#uclibc landley (n=landley@c-24-3-194-136.hsd1.mn.comcast.net) |
| 04:27.35 | *** join/#uclibc uriahheep (n=alexis@bas1-montreal45-1177813830.dsl.bell.ca) |
| 04:27.40 | uriahheep | hey |
| 04:27.55 | uriahheep | uclibc is designed for fpu-less architectures. right? |
| 04:28.36 | Bertl | hmm, why? |
| 04:28.43 | uriahheep | just wondering :P |
| 04:29.10 | uriahheep | i'm wondering if it would be better to use it instead of glibc for a gp2x |
| 04:30.55 | Bertl | that's a dual core arm or so? |
| 04:31.03 | uriahheep | yeah |
| 04:31.10 | uriahheep | well, no |
| 04:31.13 | uriahheep | dual cpu |
| 04:31.17 | Bertl | ah, right |
| 04:31.21 | uriahheep | one has an mmu, one doesn't |
| 04:31.39 | Bertl | well, uClibc will definitely use less resources |
| 04:31.50 | Bertl | glibc got quite bloated over the years |
| 04:31.55 | uriahheep | yeah.. |
| 04:32.34 | uriahheep | what's the compatibility rating on uclibc? nowadays, will pretty much anything compile with uclibc without any patching? |
| 04:33.32 | Bertl | I don't know for sure, but I'd say, you probably will need to do patching for most stuff, unless it was kind of designed for it |
| 04:34.03 | Bertl | but the number of packages which are contained in the buildroot for example, gives a good idea what should work and what not |
| 04:34.34 | uriahheep | hmm, k |
| 04:34.40 | uriahheep | got a link to the buildroot? |
| 04:35.00 | uriahheep | heh never mind |
| 04:35.27 | Bertl | it's linked on the main site :) |
| 04:35.28 | uriahheep | hmm, a ported gcc is in the buildroot? |
| 04:35.41 | Bertl | there are options for several toolchains |
| 04:35.51 | Bertl | some of them work, others dont depends on the arch |
| 04:36.02 | uriahheep | hmm, k |
| 04:36.12 | Bertl | for example, the sparc requires gcc 4.x as I found out |
| 04:36.30 | uriahheep | ah :-/ |
| 04:36.44 | uriahheep | do sparc's have mmu? |
| 04:37.53 | Bertl | yep |
| 04:41.03 | uriahheep | hmm, k |
| 04:41.11 | uriahheep | why are you using uclibc on it then? |
| 04:42.04 | Bertl | I'm trying to get it work there, because I do some testing in a virtual environment (QEMU) for several kernel archs, and it would be fine to have a common boot/test image |
| 04:42.16 | uriahheep | ah, k |
| 04:42.42 | Bertl | and images around 16M and 32M are much nicer than 600M or 2GB :) |
| 04:42.50 | uriahheep | ;) true |
| 04:42.50 | uriahheep | lol |
| 04:58.58 | *** join/#uclibc El-Lotso (n=EnergonC@60.51.120.35) |
| 05:10.34 | *** join/#uclibc garrettkajmowicz (n=garrett@dsl017-059-181.wdc2.dsl.speakeasy.net) |
| 05:11.00 | landley | Bertl: nope, most stuff works with uClibc without patching. |
| 05:11.06 | landley | This has been true since about 0.9.26. |
| 05:11.31 | landley | Hi garrett! |
| 05:11.39 | Bertl | landley: ah, good to hear, didn't succed with strace yesterday though |
| 05:17.26 | garrettkajmowicz | hi, landley. I just got my server/net connection working again. I'm going to go get some sleep and then finish the re-install. |
| 05:18.06 | garrettkajmowicz | No more annoying whine from server machine (528 MB drive removed; was only used for swap, anyways) |
| 05:29.14 | landley | garrett: woot. |
| 06:14.04 | *** join/#uclibc garrettkajmowicz (n=garrett@dsl017-059-181.wdc2.dsl.speakeasy.net) [NETSPLIT VICTIM] |
| 06:14.04 | *** join/#uclibc landley (n=landley@c-24-3-194-136.hsd1.mn.comcast.net) [NETSPLIT VICTIM] |
| 06:14.04 | *** join/#uclibc hiyuh (n=hiyuh@ZK138196.ppp.dion.ne.jp) [NETSPLIT VICTIM] |
| 06:14.05 | *** join/#uclibc common (n=common@port-212-202-23-8.dynamic.qsc.de) [NETSPLIT VICTIM] |
| 06:14.05 | *** join/#uclibc Xires (n=Xires@68-119-244-202.dhcp.wrbg.mo.charter.com) |
| 06:48.25 | *** join/#uclibc Newsome (n=sorenson@adsl-75-0-136-70.dsl.chcgil.sbcglobal.net) |
| 07:54.42 | *** join/#uclibc CIA-10 (i=cia@cia.navi.cx) |
| 08:13.28 | *** join/#uclibc blindvt_ (n=bf@M794P015.adsl.highway.telekom.at) |
| 08:18.30 | *** join/#uclibc CIA-10 (i=cia@cia.navi.cx) |
| 08:56.42 | *** join/#uclibc landley_ (n=landley@c-24-3-194-136.hsd1.mn.comcast.net) |
| 08:57.56 | *** part/#uclibc landley_ (n=landley@c-24-3-194-136.hsd1.mn.comcast.net) |
| 10:16.33 | *** join/#uclibc wrobbie (n=rob@cm30.sigma248.maxonline.com.sg) |
| 10:33.25 | *** join/#uclibc CIA-10 (i=cia@cia.navi.cx) |
| 11:12.22 | Kaloz | anyone crosscompiled xorg 7.1 (or 7.2rc) for busybox? |
| 11:12.33 | Kaloz | erm, i'm too sleepy |
| 11:12.39 | Kaloz | s/busybox/uclibc/ |
| 12:36.24 | *** join/#uclibc Sgt-Donan (n=Loutre@feu30-1-82-242-59-35.fbx.proxad.net) |
| 14:20.44 | *** join/#uclibc wrobbie (n=rob@cm30.sigma248.maxonline.com.sg) |
| 15:57.47 | *** join/#uclibc blindvt__ (n=bf@M943P002.adsl.highway.telekom.at) |
| 16:32.52 | *** join/#uclibc garrettkajmowicz (n=garrett@dsl017-059-181.wdc2.dsl.speakeasy.net) |
| 17:27.13 | *** join/#uclibc landley (n=landley@c-71-199-114-26.hsd1.pa.comcast.net) |
| 17:28.09 | CIA-10 | 03vda * r16981 10busybox/include/usage.h: fix/shorten some usage strings |
| 17:29.01 | landley | Denis is kind of energetic, isn't he? |
| 17:29.07 | landley | (He's vda.) |
| 17:29.31 | solar | <PROTECTED> |
| 17:29.56 | solar | vda was the .ru guy that did the orig e2fs stuff for bb before vapier commited his? |
| 17:30.03 | CIA-10 | 03vda * r16982 10busybox/ (debianutils/start_stop_daemon.c libbb/read.c): |
| 17:30.03 | CIA-10 | start_stop_daemon: fix bug where any program name was "matching" |
| 17:30.03 | CIA-10 | processes for which readlink(/proc/N/exe) fails |
| 17:38.23 | landley | solar: dunno about the e2fs stuff, I'd have to check. |
| 17:38.46 | landley | vda is a .ru guy. |
| 17:40.29 | landley | And his judgement's generally pretty good. And he listens when people explain stuff, without being led by the nose either. |
| 17:40.49 | *** join/#uclibc alesan (n=alesan@adsl-ull-129-6.47-151.net24.it) |
| 17:43.29 | landley | So I sit down at the coffee shop to get FWL phase 2 finished today, and what am I doing? |
| 17:43.39 | landley | Reading a history of alcohol in the james bond films. |
| 17:43.44 | garrettkajmowicz | LOL |
| 17:43.52 | landley | http://www.atomicmartinis.com/007/brands.htm |
| 17:43.54 | landley | Welcome to the internet. |
| 19:08.33 | CIA-10 | 03vda * r16983 10busybox/archival/tar.c: |
| 19:08.33 | CIA-10 | tar: report error if child dies while writing out the end of tarball |
| 19:08.33 | CIA-10 | (e.g. out of disk space). |
| 19:12.12 | landley | What's gcrt1.o? |
| 19:12.15 | landley | Dunno that one. |
| 19:15.38 | landley | Nevermind, profiling. zapped. |
| 19:20.08 | SpanKY | solar: no, that was vodz |
| 19:20.26 | SpanKY | you can tell ... vda's english isnt completely broken ;p |
| 19:21.14 | CIA-10 | 03vda * r16984 10busybox/coreutils/od_bloaty.c: od: remove bug-for-bug compat (because the bug was fixed in coreuitls) |
| 19:31.37 | landley | SpanKY: vda actually speaks english. |
| 19:31.43 | landley | Vodz used a translation program. |
| 19:32.03 | landley | The difference is, no matter how much trouble we may have communicating with vda at any given time, he's _learning_. |
| 19:32.16 | landley | Vodz's translator never improved, even after 3 years. |
| 19:43.19 | CIA-10 | 03vda * r16985 10busybox/debianutils/start_stop_daemon.c: start_stop_daemon: make code a bit more readable |
| 19:48.13 | Bertl | SpanKY: so you are working on the sparc port? |
| 19:50.29 | SpanKY | Bertl: idly |
| 19:50.54 | Bertl | well, I already figured a few broken things |
| 19:51.22 | Bertl | but I'm still unable to get a basic 'working' setup |
| 19:51.44 | SpanKY | sec |
| 19:51.51 | Bertl | np |
| 19:53.10 | SpanKY | i give up on buildroot ... installing a sparc-gentoo-linux-uclibc cross-compiler |
| 19:53.51 | landley | SpanKY I've never made buildroot do much for me. I find nested makefiles totally opaque. |
| 19:54.05 | landley | However, http://landley.net/code/firmware is coming along nicely... |
| 19:55.00 | Bertl | yeah, well, the nice part of buildroot (if it worked) would be that you can build for all the archs in an automated way |
| 19:55.27 | SpanKY | i've integrated it into Gentoo so i can automate everything with emerge |
| 19:56.16 | landley | Integrated uClibc, you mean? |
| 19:56.44 | landley | Bertl: you invoke my scripts with "./cross-compiler armv4l" |
| 19:57.03 | landley | Once it has a config file for each architecture you're interested in, automating it to iterate through them is fairly trivial. |
| 19:57.47 | landley | SpanKY: does gentoo embedded cross compile, or build natively? |
| 20:01.13 | Bertl | landley: what exactly do your scripts do? |
| 20:02.17 | landley | download.sh downloads the source code tarballs (and verifies the sha1 sums if it has one for that package), and sets up version-independent symlinks so the rest of the scripts dont' have to care about versions. |
| 20:03.18 | Bertl | okay |
| 20:03.50 | landley | cross-compiler.sh builds a relocatable cross compile toolchain out of binutils, gcc-core (C only, no c++ yet), an updated version of the old uClibc wrapper (which I use to make the toolchain relocatable), kernel headers (from make headers_install in the current linux kernel), and uClibc. |
| 20:04.18 | landley | It also builds a hello world program and does a test run of it under qemu to verify that the toolchain works (this assumes you have qemu already installed, still pondering how I want to handle that). |
| 20:04.32 | Bertl | sounds good |
| 20:04.49 | landley | Then it tars it up. By "relocatable" I mean you can extract that tarball into any directory, add its' bin to your $PATH, and use it. It should just work. |
| 20:05.03 | Bertl | okay |
| 20:05.09 | landley | (And actually $path is optional, you can call it via ../blah too. :) |
| 20:05.39 | landley | mini-native.sh is what I'm working on now. It builds a minimal native environment (including native compiler) that I can boot into and run under qemu, and build the rest of the system natively. |
| 20:05.53 | Bertl | for which archs have you tested the cross build? |
| 20:06.30 | landley | I'm trying to get armv4l to work all the way first. |
| 20:06.35 | landley | Depth first, not breadth first. |
| 20:06.41 | landley | I've poked at x86_64, though. |
| 20:06.45 | Bertl | ah, so early stage ... |
| 20:07.00 | landley | And I've got the guts of armv5l and armv4be too. |
| 20:07.13 | landley | I'll be doing ppc and mips when I get around to it, plus x86. |
| 20:07.30 | landley | There are design documents linked from the website that go into a lot more detail. |
| 20:08.06 | Bertl | okay, will look into your scripts, as I have kernel toolchains for almost all linux-kernel archs, I'll see, maybe I can integrate your uClibc part and fix up the brokenness for some of them |
| 20:08.21 | landley | All the magic is in the wrapper script. |
| 20:08.37 | Bertl | what's the license of those scripts? |
| 20:08.38 | landley | I spent weeks trying to patch gcc into submission before giving up and deciding that its path logic was too stupid to live. |
| 20:08.45 | landley | The wrapper overrides _everything_. |
| 20:08.48 | landley | license is gplv2. |
| 20:08.56 | Bertl | okay, great |
| 20:12.08 | landley | In some ways my approach is hideously wasteful of disk space compared to multilib. |
| 20:12.35 | landley | But if you have an x86-gcc and an x86_64-gcc in your path, and all their binaries and libraries and such live in separate directories, the result is likely to _work_. :) |
| 20:18.08 | CIA-10 | 03vda * r16986 10busybox/networking/Config.in: |
| 20:18.08 | CIA-10 | httpd: we don't have FEATURE_HTTPD_WITHOUT_INETD anymore, |
| 20:18.08 | CIA-10 | remove it from Config.in |
| 20:49.13 | CIA-10 | 03vapier * r16987 10uClibc/extra/Configs/Config.in.arch: for now, dont allow mmu users to select file format |
| 20:51.51 | SpanKY | blah, this sparc box is slow ... takes like 20 seconds to build uclibc |
| 20:56.05 | CIA-10 | 03vapier * r16988 10uClibc/libc/stdlib/malloc-standard/malloc.c: rename local "brk" var so it doesnt namespace collide with the brk() function |
| 20:56.24 | SpanKY | ah nice, sparc gives me segv |
| 21:02.57 | Bertl | yep, that was what I got on the cross compiling too :) |
| 21:03.11 | Bertl | (after I realized that gcc has to be at least 4.0.1 :) |
| 21:05.02 | SpanKY | i'm using 4.1.1 here |
| 21:05.13 | SpanKY | waiting for gdb to emerge |
| 21:05.49 | Bertl | btw, is uCLibc supposed to be built against sanitized kernel headers or should it work 'out of the box'? |
| 21:06.23 | Bertl | what I mean is, would a 'make headers_install' in a recent kernel tree give me what uClibc expects? |
| 21:09.40 | SpanKY | both |
| 21:09.56 | SpanKY | http://www.uclibc.org/~vapier/ <- those are against sanitized headers |
| 21:10.07 | SpanKY | atm, i think only cris has screwed kernel headers |
| 21:10.15 | SpanKY | well, bfin too, but i cheat so it doesnt matter |
| 21:20.50 | landley | Bertl: linux-2.6.19.1 is the first one that produces usable headers. |
| 21:21.02 | landley | Using that, it builds for me just fine. And I can build busybox with the resulting toolchain. |
| 21:21.21 | landley | Haven't tried bfin or cris. |
| 21:21.31 | landley | Tried x86, x86_64, and armv4l. |
| 21:25.10 | Bertl | k, tx |
| 21:26.27 | SpanKY | broken pthread call, great |
| 21:30.45 | landley | SpanKY: ? |
| 21:55.49 | *** join/#uclibc CIA-10 (i=cia@cia.navi.cx) |
| 23:20.37 | Bertl | any idea why I would get this: |
| 23:20.38 | Bertl | socketcalls.c:179: error: syntax error before "__libc_recvfrom" |
| 23:20.39 | Bertl | socketcalls.c:180: warning: type defaults to `int' in declaration of `_syscall6' |
| 23:21.43 | Bertl | when building uClibc 0.9.28 for i386, missing include? |
| 23:22.21 | landley | I'd try the current -svn. It's close to release. |
| 23:22.38 | landley | (If I have to _drive_ to utah... :) |
| 23:33.21 | Bertl | looks good, at least it compiles :) |
| 23:35.07 | landley | Current svn should work ok on i386 unless they broke it today. |
| 23:35.16 | landley | armv4l has a weird problem I'm chasing. |
| 23:35.22 | landley | armv5 (EABI) works, as do most of the other platforms. |