| 00:30.12 | ds-work | is uclibc known to work when compiled with Linux-2.6? |
| 00:32.12 | andersee | ds-work: yes |
| 00:32.23 | andersee | ds-work: I know mjn3 has tested it |
| 00:32.34 | mjn3 | andersee: uh.. no |
| 00:32.37 | andersee | ds-work: though there are some warnings |
| 00:32.39 | mjn3 | i only tested building it |
| 00:32.47 | mjn3 | i've never tried running it |
| 00:32.52 | andersee | mjn3: ahh |
| 00:33.18 | ds-work | there's lots of warnings, but it builds fine |
| 00:33.22 | DavidM | we have tried uClibc with 2.6 |
| 00:33.30 | DavidM | It worked fine, without a problem |
| 00:33.50 | ds-work | I'm wondering if I should switch the Debian package to use linux-kernel-headers instead of extracting kernel headers from the linux source |
| 00:34.15 | ds-work | it would have the benefit of tracking known-good userspace versions of the Linux headers |
| 00:35.35 | dsmith | Wasn't there some talk about removing the depenancy on kernel headers? |
| 00:36.39 | mjn3 | yes. unfortunately, there's only be time for talking about it... not doing it |
| 00:51.15 | dsmith | Do you guys (andersee, mjn3, and anyone else) get psin to work on uclibc/buildroot? |
| 00:51.31 | dsmith | Woah! s/psin/paid/ |
| 00:52.03 | mjn3 | dsmith: the softfloat stuff i checked in two weeks ago was for a contract |
| 00:52.56 | dsmith | mjn3: Yeah, I talked to you about that about a month ago. So they paid up? |
| 00:53.08 | mjn3 | yeah |
| 00:53.10 | dsmith | Or was that a different gig? |
| 00:53.18 | mjn3 | no, that was it |
| 00:54.36 | dsmith | mjn3: Are you contracting by yourself, or work for a firm, or? |
| 00:55.10 | dsmith | (if you don't mind me asking) |
| 00:57.21 | mjn3 | well, i'm essentially subcontracting under erik's codepoet |
| 00:59.05 | dsmith | <- smacks head |
| 00:59.21 | dsmith | codepoet, I forgot about that |
| 00:59.46 | ds-work | how do I compile gcc so that it uses /lib/ld-uClibc.so instead of /lib/ld-linux.so? |
| 01:01.29 | DavidM | ds-work: I use a 3 line perl script to. Hold on let me find it |
| 01:02.25 | mjn3 | ds-work: you could look at the various search-and-replace lines in buildroot's gcc-3.3.mk |
| 01:02.29 | DavidM | perl -i -p -e "s,#.*define.*ELF_DYNAMIC_INTERPRETER.*\".*\" ,#define ELF_DYNAMIC_INTERPRETER \"/lib/ld-uClibc.so.0\",;" `grep -lr "#.*define.*ELF_DYNAMIC_INTERPRETER.*\".*\"" ${BINUTILS_PATH}`) |
| 01:02.57 | DavidM | ds-work: Actually I grabbed that from a very early version of buildroot |
| 01:05.07 | andersee | ds-work: yep |
| 01:05.23 | andersee | ds-work: the buildroot stuff is the reference implementation |
| 01:07.54 | andersee | still updating the debian packaging |
| 01:09.05 | dsmith | andersee: Debian Packaging? Oh. For native use? (not cross?) |
| 01:40.14 | mjn3 | andersee: i'm experimenting with bumping __GLIBC_MINOR__ to 3 |
| 01:41.52 | andersee | mjn3: find anything that cares one way or the other thus far? |
| 01:43.15 | mjn3 | i haven't really looked. but it affects the expected ABI for the locale-specific functions. for instance, __wctype_l changed to wctype_l (which is what we support) |
| 02:01.00 | mjn3 | andersee: i'm not going to worry about it right now. to do what i was trying to do requires some of the planned locale internal changes |
| 02:01.50 | mjn3 | but after this release, i'll bump it in my version so that i can start testing for problems |
| 02:05.22 | andersee | mjn3: k |
| 02:05.57 | andersee | mjn3: after this release, we can bump that, make the stdio ABI changes, etc |
| 02:07.20 | mjn3 | ok. i didn't realize until i started looking into that stdlibc++ related post that the wchar-specific specializations weren't getting built. started to track it down and realized that fixing it is too big a change to try to make before this release |
| 02:26.06 | andersee | mjn3: sounds like a plan |
| 02:26.21 | andersee | mjn3: we definately want to tread lightly at the moment |
| 02:26.47 | andersee | mjn3: since most everything seems to be working |
| 02:27.06 | mjn3 | yeah. i just posted a reply about libstc++ |
| 02:27.28 | andersee | <for a sufficiently lax definition of working, or a carefully chosen but still rather large test set> |
| 02:27.35 | andersee | :-) |
| 02:27.38 | mjn3 | heh |
| 03:16.49 | *** join/#uclibc bug1 (~bug1@c211-28-81-139.sunsh3.vic.optusnet.com.au) |
| 05:36.02 | *** join/#uclibc yashi (~yashi@dns1.atmark-techno.com) |
| 07:40.53 | *** part/#uclibc kyo (~kyo@216.155.40.190) |
| 07:41.24 | *** join/#uclibc tahorg (~loic@perax.net1.nerim.net) |
| 08:06.55 | *** join/#uclibc mr_claus (~mr_claus@p5080364C.dip.t-dialin.net) |
| 08:28.57 | *** join/#uclibc ChanServ (ChanServ@services.) |
| 08:28.57 | *** join/#uclibc mr_claus (~mr_claus@p5080364C.dip.t-dialin.net) [NETSPLIT VICTIM] |
| 08:28.57 | *** join/#uclibc yashi (~yashi@dns1.atmark-techno.com) [NETSPLIT VICTIM] |
| 08:28.57 | *** join/#uclibc ds-work (ds@condor.ucsf.edu) [NETSPLIT VICTIM] |
| 08:28.57 | *** join/#uclibc kergoth`bbl ([a13twLYzb@zelda.tuxnami.org) [NETSPLIT VICTIM] |
| 08:28.57 | *** join/#uclibc waldi (~waldi@80.190.101.130) [NETSPLIT VICTIM] |
| 08:28.57 | *** join/#uclibc TheMasterMind1 (~aman@h-68-166-70-220.MCLNVA23.dynamic.covad.net) [NETSPLIT VICTIM] |
| 08:28.57 | *** join/#uclibc sjhill-afk (~sjhill@65.117.135.105) [NETSPLIT VICTIM] |
| 08:28.57 | *** mode/#uclibc [+o ChanServ] by tolkien.freenode.net |
| 08:29.31 | *** join/#uclibc tahorg (~loic@perax.net1.nerim.net) [NETSPLIT VICTIM] |
| 09:39.37 | *** join/#uclibc andersee (~andersee@codepoet.org) |
| 11:28.25 | *** join/#uclibc mr_claus (~mr_claus@217.9.121.56) |
| 11:28.30 | mr_claus | hi |
| 11:30.27 | andersee | mr_claus: hi |
| 11:32.42 | mr_claus | You can enable UCLIBC_HAS_SYS_SIGLIST in your uClibc .config. |
| 11:32.42 | mr_claus | Or just change all instances of "sys_siglist[sig]" to instead |
| 11:32.42 | mr_claus | use "strsignal(sig)" |
| 11:32.57 | mr_claus | andersee: you wrote that in the mailing list |
| 11:33.18 | mr_claus | andersee: would you prefer use a patch which change the instances? |
| 11:33.37 | mr_claus | andersee: or would you prefer to enable the parameter in .config |
| 11:34.07 | andersee | mr_claus: for buildroot, I prefer keeping UCLIBC_HAS_SYS_SIGLIST disabled |
| 11:34.22 | mr_claus | ok, so i can make a patch and send it to you? |
| 11:34.24 | andersee | mr_claus: why, what is using it? |
| 11:34.47 | mr_claus | fstab.c: In function `handler': |
| 11:34.47 | mr_claus | fstab.c:345: error: `sys_siglist' undeclared (first use in this function) |
| 11:34.47 | mr_claus | fstab.c:345: error: (Each undeclared identifier is reported only once |
| 11:34.47 | mr_claus | fstab.c:345: error: for each function it appears in.) |
| 11:34.47 | mr_claus | make[2]: *** [fstab.o] Error 1 |
| 11:34.55 | mr_claus | i try to compile util-linux |
| 11:35.02 | mr_claus | with buildroot |
| 11:35.19 | mr_claus | i just added to the targets |
| 11:35.41 | mr_claus | but i can change all the instances and make a diff of that changes |
| 11:36.42 | mr_claus | it's only one line |
| 11:40.11 | mr_claus | the patch |
| 11:40.13 | mr_claus | --- util-linux-2.11z.orig/mount/fstab.c Tue Nov 26 10:53:41 2002 |
| 11:40.13 | mr_claus | +++ util-linux-2.11z/mount/fstab.c Thu Nov 13 12:37:36 2003 |
| 11:40.13 | mr_claus | @@ -342,7 +342,7 @@ |
| 11:40.13 | mr_claus | <PROTECTED> |
| 11:40.13 | mr_claus | <PROTECTED> |
| 11:40.14 | mr_claus | <PROTECTED> |
| 11:40.17 | mr_claus | - die (EX_USER, "%s", sys_siglist[sig]); |
| 11:40.18 | mr_claus | + die (EX_USER, "%s", strsignal(sig)); |
| 11:40.20 | mr_claus | <PROTECTED> |
| 11:40.24 | mr_claus | <PROTECTED> |
| 11:40.53 | mr_claus | or should i mail it to you, with the changed util-linux.mk? |
| 11:45.19 | mr_claus | hmm, another problem |
| 11:45.21 | mr_claus | Your libc thinks that swapon has 1 arg only. |
| 11:45.37 | mr_claus | swapon.c: In function `main_swapon': |
| 11:45.37 | mr_claus | swapon.c:321: warning: implicit declaration of function `access' |
| 11:45.37 | mr_claus | swapon.c:321: error: `R_OK' undeclared (first use in this function) |
| 11:45.37 | mr_claus | swapon.c:321: error: (Each undeclared identifier is reported only once |
| 11:45.37 | mr_claus | swapon.c:321: error: for each function it appears in.) |
| 11:45.37 | mr_claus | make[2]: *** [swapon.o] Error 1 |
| 11:46.09 | mr_claus | i can remember that i compiled util-linux with uclibc before, but with another version of uclibc |
| 11:52.07 | andersee | mr_claus: re swapon -- umm no |
| 11:52.29 | andersee | mr_claus: uClibc quite clearly uses two args for swapon |
| 11:53.28 | mr_claus | andersee: i think it's only a info message, between that line and the error it tried to compile swapon.c |
| 11:54.07 | mr_claus | andersee: i think these things a two different problems |
| 11:54.26 | mr_claus | andersee: perhaps the argument thing isn't a real problem but the R_OK |
| 12:01.57 | mr_claus | hmm, a #include <fcntl.h> in swapon.c is enough |
| 12:02.07 | mr_claus | that solves the problem |
| 12:02.28 | mr_claus | andersee: should i enhance the patch and send these things to you for including it into the buildroot? |
| 12:02.59 | andersee | mr_claus: send patches to the uClibc list |
| 12:03.03 | mr_claus | ok |
| 12:38.18 | *** join/#uclibc TheMasterMind1 (~aman@h-66-167-145-129.MCLNVA23.dynamic.covad.net) |
| 13:01.39 | *** topic/#uclibc by andersee -> Discussion of uClibc, busybox, tinylogin, and related stuff | November 13 -- uClibc 0.9.23 released | Sept 12 -- Busybox 1.0.0-pre3 released | Sept 30 -- uClibc dev systems for x86, arm, powerpc, and mips updated | http://codepoet.org/uClibc-2003-10-22.png |
| 13:06.39 | mr_claus | andersee: 0.9.23? 0.9.22 is only a few days old |
| 13:06.48 | andersee | mr_claus: yup |
| 13:07.06 | andersee | mr_claus: there were several problems with the last release |
| 13:07.22 | andersee | mr_claus: Some things like soft-float were broken |
| 13:07.45 | mr_claus | ok, i will upgrade again |
| 13:09.11 | mr_claus | i try to compile shadow and configure tell me "cannot check setpgrp if cross compiling", how can i solved problems like that? |
| 13:12.18 | andersee | You need to either not cross compile, or you need to fix the configure script so it can deal with cross compiling |
| 13:12.49 | mr_claus | hmm, it seems that the script do the rest well but only with that function it doesn't want cross compile |
| 13:13.00 | mr_claus | configure:8920: checking for setpgrp |
| 13:13.00 | mr_claus | configure:8957: /mnt/daten/projects/lfs-project/buildroot-0.9.22/build_i386/stag |
| 13:13.00 | mr_claus | ing_dir/bin/i386-linux-gcc -o conftest -O2 -Wall -s conftest.c >&5 |
| 13:13.00 | mr_claus | configure:8960: $? = 0 |
| 13:13.01 | mr_claus | configure:8963: test -s conftest |
| 13:13.01 | mr_claus | configure:8966: $? = 0 |
| 13:13.03 | mr_claus | configure:8976: result: yes |
| 13:13.05 | mr_claus | configure:8979: checking whether setpgrp takes no argument |
| 13:13.07 | mr_claus | configure:8985: error: cannot check setpgrp if cross compiling |
| 13:14.54 | andersee | mr_claus: http://www.uclibc.org/FAQ.html |
| 13:15.05 | andersee | <PROTECTED> |
| 13:15.13 | andersee | <PROTECTED> |
| 13:15.13 | andersee | scripts. |
| 13:17.38 | andersee | mr_claus: if you look in buildroot/sources/openssh.patch |
| 13:18.01 | andersee | mr_claus: you can see the approach I use to work around this |
| 13:18.46 | andersee | mr_claus: i.e. providing a correct-for-uClibc default when tests cannot be run due to cross compiling |
| 14:04.56 | mr_claus | andersee: thx for your informations, i will try to fix that |
| 14:44.15 | *** join/#uclibc dsmith (~dsmith@mail.actron.com) |
| 15:17.46 | *** join/#uclibc andersee (~andersee@codepoet.org) |
| 15:57.12 | mr_claus | andersee: you changed uclibc.mk in buildroot to use more seds, these sed commands use parameter -ie, sed don't know -i |
| 15:57.48 | kergoth | time to upgrade your sed? :) |
| 15:58.03 | kergoth | <PROTECTED> |
| 15:58.03 | kergoth | <PROTECTED> |
| 15:58.27 | mr_claus | come on, i use debian woody, sed should be fine |
| 15:58.32 | andersee | mr_claus: You need a newer version of sed. |
| 15:58.43 | mr_claus | grrrrrr |
| 15:59.24 | andersee | debian woody has sed-3.02 |
| 15:59.47 | andersee | The -i option was added in ~sed-4.03 |
| 16:00.23 | mr_claus | hmm, so it isn't possible with woody, one backport more |
| 16:00.53 | andersee | mr_claus: lots of people were unhappy with my using perl |
| 16:01.26 | andersee | mr_claus: switching to sed was fairly easy using the -i option |
| 16:01.36 | mr_claus | andersee: i think it's better to use sed too |
| 16:02.01 | mr_claus | andersee: but then i need a testing plattform for buildroot, bad luck for me |
| 16:02.03 | andersee | it would have been messier to use sed-3.02, since I would have to copy files around |
| 16:03.28 | mr_claus | andersee: perhaps it's a way to compile sed in front staticly for using in buildroot |
| 16:03.58 | mr_claus | andersee: i will backport sed-4.03 to woody |
| 16:07.51 | mr_claus | andersee: you told me you're using debian testing, right? |
| 16:08.24 | andersee | mr_claus: I recommend sed 4.07 from testing |
| 16:08.31 | mr_claus | i have done |
| 16:08.32 | andersee | mr_claus: works nicely |
| 16:08.39 | mr_claus | andersee: i backported at this moment |
| 16:09.24 | mr_claus | andersee: if you're using testing, do you have problems with it? or do you think it's ready for a production environment? |
| 16:09.47 | andersee | mr_claus: it works for me |
| 16:10.21 | mr_claus | andersee: so you dont have problems but i should not come to you if i will get some problems :)) |
| 16:10.33 | andersee | heh |
| 16:10.39 | andersee | yeah |
| 16:11.09 | mr_claus | a default response, "it's good for me, but if you want to use it, it's your own risk" |
| 16:11.17 | mr_claus | :)) |
| 16:11.45 | mr_claus | i think it's working for a lot of people so i will try |
| 16:13.12 | mr_claus | nice, much better with sed-4.0.7 |
| 16:29.49 | *** join/#uclibc dsmith (~dsmith@mail.actron.com) |
| 16:30.34 | dsmith | Lost power. Did I miss anything? |
| 16:37.27 | mr_claus | nope |
| 16:49.59 | dsmith | Yay. just got the uclibc update. |
| 16:50.29 | dsmith | Arrg. Same problem. |
| 16:50.31 | mr_claus | yep, buildroot runs fine after updating sed |
| 16:50.34 | dsmith | /usr/local/src/buildroot/build_mipsel/staging_dir/lib/gcc-lib/mipsel-linux/3.3.2/../../../../mipsel-linux/bin/ld: cannot open crt1 .o: No such file or directory |
| 16:51.25 | andersee | dsmith: when did you last update? |
| 16:51.33 | dsmith | About 10 min ago. |
| 16:51.58 | andersee | dsmith: and you rebuilt the toolchain? |
| 16:52.17 | andersee | dsmith: there was a problem I fixed last night with binutils that could cause that |
| 16:53.08 | dsmith | What make *clean will clar the toolchain? |
| 17:02.26 | mr_claus | andersee: in bzip you use sed to replace "ln" into "ln -sf" and you use that syntax |
| 17:02.28 | mr_claus | sed -ie "s,ln \\$\(,ln -sf \\$\(,g" |
| 17:02.46 | mr_claus | sed use the \( as a starting ( for a \1 in the replace |
| 17:13.05 | mr_claus | should be --> sed -ie "s,ln \$(,ln -sf \$(,g" |
| 17:14.40 | mr_claus | should be --> sed -ie "s,ln \$$(,ln -sf \$$(,g" |
| 17:33.29 | *** join/#uclibc DavidM (~david@h24-207-7-221.dlt.dccnet.com) |
| 17:43.17 | *** join/#uclibc dsmith (~dsmith@mail.actron.com) |
| 17:43.46 | dsmith | Power! |
| 17:54.01 | *** join/#uclibc DavidM (~david@h24-207-7-221.dlt.dccnet.com) [NETSPLIT VICTIM] |
| 17:54.01 | *** join/#uclibc kergoth ([a13twLYzb@zelda.tuxnami.org) [NETSPLIT VICTIM] |
| 18:21.43 | *** join/#uclibc DavidM (~david@h24-207-7-221.dlt.dccnet.com) [NETSPLIT VICTIM] |
| 18:21.43 | *** join/#uclibc kergoth ([a13twLYzb@zelda.tuxnami.org) [NETSPLIT VICTIM] |
| 18:23.26 | dsmith | WhoooHoooo! |
| 18:23.44 | mjn3 | dsmith: you seem quite excitable this morning |
| 18:31.01 | DavidM | dsmith: you sure do! Is there a story? |
| 19:04.12 | dsmith | Everything built properly. First time with builtroot. |
| 19:04.19 | dsmith | Thank You. |
| 19:04.29 | dsmith | Very much.! |
| 19:09.09 | mjn3 | glad to hear it |
| 19:11.33 | DavidM | building uClibc 0.9.23: crt0.S:127:27: ./gmon-start.S: No such file or directory |
| 19:11.50 | DavidM | has support for profiling been removed? |
| 19:13.57 | mjn3 | not that i know of. but i haven't tried it lately |
| 19:14.18 | mjn3 | it may have been a casualty of the recent initfini changes |
| 19:14.57 | DavidM | the file has been removed. I'm going to go check the web cvs logs |
| 19:14.59 | mjn3 | also, even when i used it there were some issues |
| 19:15.41 | *** join/#uclibc Russ (user@ip68-2-183-161.ph.ph.cox.net) |
| 19:15.47 | *** part/#uclibc Russ (user@ip68-2-183-161.ph.ph.cox.net) |
| 19:16.01 | DavidM | mjn3: actually I never used it. Some customers were asking about it, so I enabled it, but I never tested it |
| 19:16.55 | mjn3 | i was using it about a month ago (i think). but i had to do something weird like compile as PIC but link staticly or something |
| 19:18.32 | DavidM | did this used to be a generated file? I can't seem to find it in CVS. |
| 19:19.43 | DavidM | It is I found it in the Makefile |
| 19:21.34 | DavidM | http://www.uclibc.org/cgi-bin/cvsweb/uClibc/libc/sysdeps/linux/arm/Makefile?r1=1.49&r2=1.50&hideattic=0 |
| 19:21.47 | DavidM | That would be where it got removed |
| 19:29.50 | *** join/#uclibc Russ (user@ip68-2-183-161.ph.ph.cox.net) |
| 19:43.29 | Russ | anyone compile stuff for uclinux? |
| 20:04.26 | mjn3 | not me. want to donate some hardware so i'd have reason to? :-) |
| 20:05.13 | DavidM | I take it to get soft floating point to work I have to get gcc to use soft floating point for its objects (crtbegin.o ...)? |
| 20:05.28 | mjn3 | yep |
| 20:05.42 | mjn3 | buildroot handles that |
| 20:05.50 | Russ | I built a kernel with my standard arm toolchain |
| 20:06.04 | Russ | but the kernel chokes on any userspcae applications I build with it |
| 20:07.26 | Russ | what is different about the uclinux toolchains? |
| 20:07.28 | mjn3 | DavidM: or at least it should. well, except for uClibc with has it's own config setting |
| 20:07.56 | DavidM | mjn3: How come there is no option in uClibc to use soft or hard floating point? |
| 20:07.59 | mjn3 | Russ: no idea... other than the flat file format uclinux expects in apps |
| 20:08.17 | Russ | what is the flat file format |
| 20:08.25 | Russ | (that is the problem I'm having) |
| 20:08.27 | DavidM | mjn3: I haven't looked that the kconfig files yet but I take it, that uClibc figures it out from the architeture |
| 20:10.19 | mjn3 | DavidM: if you configure uClibc with HAS_FLOATS but not HAS_FPU then it will build in soft-float mode |
| 20:11.06 | mjn3 | Russ: you need to do some browsing at http://ucdot.org/ |
| 20:12.00 | Russ | grrr...why can't it just support elf!?! |
| 20:12.31 | DavidM | I don't have an option for HAS_FPU. It seems to be automatically set by ARM720T. At least I know where it is being set |
| 20:12.46 | mjn3 | Russ: uh... because you don't have an MMU perhaps? |
| 20:14.59 | mjn3 | Russ: while some of the targets still use some of the older coff tools, all the recent ones use elf. but you need to do a conversion afterwards to generate the flat format files |
| 20:17.19 | DavidM | Russ: you should also be clear on what not having an MMU means, and what uClinux does. I've tried to stay away from this, since I have nightmares(memories of an old project) about terrible bugs that are hard to find |
| 20:21.45 | Russ | I'm aware of mmuless and mmufull stuff |
| 20:22.02 | Russ | it just confuses me if you can have a tool that converts between two formats, why the kernel can't load both |
| 20:25.59 | DavidM | I don't know why it wasn't implemented in uClinux, but at a guess I'd say because most uClinux machines are slow, and this would keep the runtime overhead down |
| 20:26.15 | DavidM | or maybe no one has gotten around to implementing it |
| 20:26.20 | Russ | usually bin formats are a kernel compile time option |
| 20:27.06 | Russ | how does one use elf2flt? |
| 20:28.07 | DavidM | never used it. I'd look at the man page for objcopy though |
| 20:28.27 | DavidM | I think this is what you are looking for, unless elf2flt is a simplified tool |
| 20:28.27 | Russ | it aparently doesn't convert elf's |
| 20:28.36 | Russ | elf2flt busybox |
| 20:28.41 | Russ | 'File is not an object file' |
| 20:28.57 | DavidM | gotta go, it's lunch time |
| 20:30.25 | mjn3 | Russ: i'd suggest looking at the uClinux distribution and see how they build apps |
| 20:31.00 | Russ | it uses a patched compiler |
| 20:31.09 | Russ | I was hoping not to have to use their toolchain though |
| 22:05.05 | *** join/#uclibc A_Young (~knoppix@adsl-18-27.swiftdsl.com.au) |
| 22:05.12 | A_Young | hi anyone up? |
| 22:05.16 | mjn3 | yes |
| 22:05.55 | A_Young | is there a working uclibc UML at the moment? |
| 22:06.05 | A_Young | eg buildroot |
| 22:06.38 | mjn3 | well, buildroot should be working right now. i haven't tried UML in a while though |
| 22:06.48 | A_Young | ok |
| 22:06.59 | A_Young | i should get the cvs one? |
| 22:07.44 | A_Young | oh i c 0.9.23 is out |
| 22:07.48 | mjn3 | erik rebuilt the rootfs images last night/this morning before this release i believe |
| 22:07.59 | A_Young | cool :) i will try that then :D |
| 22:09.39 | A_Young | is the woody build available somewhere? or an apt cache? |
| 22:10.48 | mjn3 | erik has a uclibc-based debian system running, but it hasn't been released yet... unless he did it early this morning and i haven't noticed |
| 22:11.13 | mjn3 | i'm sure it is something he would have announced though |
| 22:11.38 | A_Young | k .. you think he will be releasing it? |
| 22:11.51 | A_Young | otherwise i build my own i guess :) |
| 22:12.00 | A_Young | or something different |
| 22:12.01 | mjn3 | at some point, yes. don't know when though |
| 22:12.13 | A_Young | cool |
| 22:12.38 | mjn3 | be aware that it took him maybe a week to build. lots of hacking involved to get around circular dependencies i think |
| 22:12.56 | A_Young | i see |
| 22:13.24 | A_Young | i may try slackware in it |
| 22:13.35 | A_Young | or something |
| 22:14.05 | A_Young | woody is a bit old for most of the stuff i want |
| 22:14.24 | A_Young | but attractive as a base since it's well tested |
| 22:14.34 | mjn3 | i'm not sure which version he built. my guess would be testing (sarge?) |
| 22:15.04 | A_Young | oh well that would be really cool |
| 22:16.14 | A_Young | then we can have ucknoppix |
| 22:16.45 | A_Young | And Erik has built Debian stable (woody) for x86 with uClibc and it runs great |
| 22:16.52 | A_Young | that's from uclibc.org main page |
| 22:16.58 | mjn3 | ah. ok |
| 22:17.16 | A_Young | anyway, i will have a play :) |
| 22:24.42 | Russ | hmm...new uclibc doesn't let me select mmuless |
| 22:24.48 | Russ | (0.9.23) |
| 22:24.53 | mjn3 | what arch? |
| 22:24.57 | Russ | arm |
| 22:25.45 | mjn3 | looking |
| 22:26.07 | Russ | its in the menuconfig, but it just has three dashes |
| 22:26.17 | Russ | instead of a [ ] |
| 22:27.26 | Russ | doesn't have the compiler optimization in the menuconfig for the arm940T yet either, but, eh |
| 22:27.58 | DavidM | Russ: the mmu information is in extra/Configs/Config.arm |
| 22:28.43 | Russ | oh, so if I add in a config CONFIG_ARM_S3C2500, everything will work |
| 22:28.50 | DavidM | Which is actually wrong too, since there shouldn't be any assumptions made on the MMU depending on the ARM CPU |
| 22:28.59 | Russ | I was going to say |
| 22:29.21 | Russ | most of the cores listed are used in mmu and mmuless processors |
| 22:29.43 | DavidM | Russ: I know, I think a patch should be made for MMU to be optional in all of those processors |
| 22:30.34 | DavidM | Russ: It seems like you have a current need for this. Maybe you should be the one volunteering to make the patch :) you could even test it while you are at it |
| 22:30.56 | Russ | I just hacked it :( |
| 22:31.55 | mjn3 | erik changed Config.arm over a week ago. i don't know the details |
| 22:32.44 | Russ | why is it ARCH_HAS_NO_MMU/UCLIBC_HAS_MMU? |
| 22:32.53 | Russ | why not UCLIBC_HAS_NO_MMU? |
| 22:33.24 | Russ | (btw, removing the select mmu lines from the CONFIG_ARM### stanzas fixes it) |
| 22:34.58 | DavidM | I think the FPU information is also an incorrect assumption. I believe the cores can optionally have floating point. Don't quote me as I've never used an ARM with floating point |
| 22:35.33 | Russ | just assign stuff on the actual cpu choices then |
| 22:35.39 | Russ | (ie xscale, sa1100) |
| 22:36.11 | Russ | arch/armnommu really needs a resync with arch/arm... |
| 22:36.24 | DavidM | Russ: both xscale and sa1100 could have MMU or not and FPU or not |
| 22:36.49 | Russ | umm.... |
| 22:36.52 | Russ | I don't think so |
| 22:37.00 | Russ | sa1100 refers to an actual cpu, not a core |
| 22:37.02 | Russ | same with xscale |
| 22:37.42 | Russ | grr....internal compiler error |
| 22:38.01 | DavidM | strong arms, and xscales are just intel cores, which don't have the same numbering convention as ARM cores. The big difference is that ARM licenses it's cores so the licensee's have all sorts of different configurations |
| 22:38.20 | Russ | but it doesn't say strongarm |
| 22:38.24 | Russ | it says sa1100 |
| 22:39.34 | Russ | (iirc |
| 22:39.57 | DavidM | ok you've got me. I've never used a strong arm, and I don't know the numbering convetion |
| 22:40.12 | Russ | sa1100 is a strongarm core with a DRAM controller |
| 22:40.18 | DavidM | hmmm, install_target is gone, but there are still references to it |
| 22:40.20 | Russ | sa1110 is a strongarm core with an SDRAM controller |
| 22:42.08 | *** join/#uclibc TheMasterMind1 (~aman@h-69-3-84-103.MCLNVA23.dynamic.covad.net) |
| 22:45.14 | Russ | the compiler is b0rking on elf_resolve |
| 22:45.19 | Russ | er |
| 22:45.26 | Russ | on _dl_load_elf_shared_library |
| 22:45.41 | Russ | I don't think I need that function anyway... |
| 22:46.01 | Russ | do I? |
| 22:48.28 | mjn3 | are you trying NOMMU? |
| 22:48.33 | Russ | ya |
| 22:48.48 | mjn3 | i honestly don't know if erik tested that |
| 22:49.10 | Russ | does uclibc compile shared under uclinux? |
| 22:50.06 | mjn3 | i've never used uclinux |
| 22:50.38 | *** join/#uclibc andersee (~andersee@codepoet.org) |
| 22:50.44 | Russ | afternoon andersee |
| 22:50.52 | andersee | Russ: hey |
| 22:50.59 | andersee | Russ: how goes it? |
| 22:51.02 | Russ | does uclibc compile shared under uclinux |
| 22:51.10 | Russ | good, except I have a internal compiler error |
| 22:54.27 | andersee | Russ: Shared Libraries are not currently supported by uClibc on MMU-less systems. SnapGear has implemented shared library support for MMU-less systems, however, so if you need MMU-less shared library support they may be able to help. |
| 22:54.39 | andersee | http://www.snapgear.com/ |
| 22:58.05 | Russ | you have any plans of implementing that? |
| 22:58.13 | Russ | seems a bit of a waste |
| 22:59.32 | andersee | Russ: alledgedly their stuff is open source and available to the interested student wanting to sift through uClinux-dist and extract the needed stuff |
| 22:59.57 | andersee | Russ: but it has never been submitted upstream |
| 22:59.57 | Russ | their uClinux-dist stuff does it? |
| 23:00.00 | Russ | arg...I really hate the uclinux community |
| 23:00.12 | andersee | at least for m68k/coldfire |
| 23:00.31 | andersee | dunno about arm |
| 23:00.40 | andersee | Russ: which arch are you looking into? |
| 23:02.04 | DavidM | andersee: When I logged on this monring you were logging off. That isn't very much sleep |
| 23:02.24 | andersee | DavidM: yep |
| 23:02.29 | DavidM | andersee: If I were you I'd be dead tired |
| 23:03.19 | andersee | Tomorrow I take off for a week of business travel |
| 23:04.09 | DavidM | andersee: hmmm, working hard to prepare for a week of working hard. You have long weekend coming up though |
| 23:05.23 | DavidM | Russ and I were talking about extra/Configs/Config.arm, and we don't think that the MMU and FPU information should be set there. Those cores could be either, and should be optional |
| 23:06.05 | mjn3 | andersee: oh, jeff posted a patch for various mips processors |
| 23:07.07 | DavidM | if that is alright, I'll make a patch for it |
| 23:12.26 | andersee | DavidM: ok |
| 23:12.46 | andersee | DavidM: I was debating changeing that right before the release, then decided against it |
| 23:13.01 | andersee | DavidM: no problem, feel free to change it |
| 23:13.12 | andersee | mjn3: ok |
| 23:13.14 | andersee | mjn3: cool |
| 23:13.18 | DavidM | andersee:k, patch coming |
| 23:13.41 | andersee | DavidM: I'm taking off for a few hours. I've got some errands to run etc |
| 23:14.36 | DavidM | andersee: I'll just e-mail the list |
| 23:14.50 | andersee | DavidM: k |
| 23:22.09 | DavidM | e-mail with patch sent |
| 23:41.10 | *** join/#uclibc bug1 (~bug1@c211-28-81-139.sunsh3.vic.optusnet.com.au) |
| 23:45.04 | bug1 | if the old fork() and pipe() method is so slow, how does the shell do "bzcat <filename> | tar -x", it must use another method |
| 23:48.25 | mjn3 | it shouldn't be that slow. do you have a test app? |
| 23:49.35 | bug1 | i was only testing it by doing "time tar -zxf <kernel>" |
| 23:50.04 | bug1 | when i changed get_header_tar_gz back to the fork method it took 15 minuts, the ucrrent method took 3 minutes |
| 23:50.26 | bug1 | it looks like ash just does fork and pipe internally |
| 23:51.33 | bug1 | it must be the buffering rather than the pipe itself iguess |
| 23:52.28 | bug1 | i should experiment with mixingf streams and pipes together |
| 23:52.30 | DavidM | andersee: when you get back, this was the other thing I was curious about http://www.uclibc.org/cgi-bin/cvsweb/uClibc/libc/sysdeps/linux/arm/Makefile?r1=1.49&r2=1.50&hideattic=0 This removes the build rule for gmon-start.S |
| 23:54.03 | mjn3 | bug1: try timing the steps seperately to see if it might be something in the implementation |
| 23:55.19 | mjn3 | i.e. time zcat and tar -xf |
