00:00.01 | JockeHome | that wont work |
00:06.05 | psm | nitinkg: yes |
00:06.36 | psm | JockeHome: it seems to work w/ rewriting the LOAD/RESTOREARGS_1 |
00:07.11 | JockeHome | great, does it run also :) |
00:07.14 | JockeHome | ? |
00:07.58 | JockeHome | 7.72 `.purgem NAME' |
00:07.58 | JockeHome | =================== |
00:07.59 | JockeHome | Undefine the macro NAME, so that later uses of the string will not be |
00:07.59 | JockeHome | expanded. *Note Macro::. |
00:08.03 | psm | cant tell, because I have others problems as well, for ex. a simple libm/s_lib_version.c if within IMA cant compile |
00:08.36 | JockeHome | maybe a .purgem before the macro def will work as well? |
00:08.44 | psm | that would be an option to put it before |
00:09.11 | JockeHome | its cleaner, hope it works |
00:09.43 | JockeHome | whats s_lib_version.c? |
00:10.40 | JockeHome | why does libm prevent you from comp. libc? |
00:11.22 | psm | purgem works the same |
00:11.52 | psm | libm cant be built completely IMA because of this one file (does not have to do w/ libc) |
00:13.15 | psm | I get though a warning "Attempt to purge non-existant macro" |
00:13.29 | psm | but it builds |
00:14.02 | JockeHome | whats the error message for libm? |
00:16.02 | psm | conflicting types for _LIB_VERSION (math.h:252) previous definition of _LIB_VERSION was here (s_lib_version.c) |
00:16.30 | psm | and apparently there is no conflict, this is a "common" IMA failure |
00:18.28 | JockeHome | use #ifdef__USE_MISC in .c file? |
00:18.56 | JockeHome | .c == s_lib_version.c |
00:19.23 | psm | no, but it is compiled w/ _SVID_SOURCE and that iirc implies it |
00:20.30 | psm | yes, it implies __USE_MISC |
00:21.16 | JockeHome | still, math.h has a __USE_MISC around so shold .c as well |
00:22.02 | psm | it does not like the type _LIB_VERSION_TYPE, I have a similar case in libpthread where PT_EI is defined somewhere else, and then used as PT_EI function_name() |
00:23.08 | JockeHome | ds: lets hear it :) |
00:23.17 | psm | added __USE_MISC, no use |
00:23.56 | JockeHome | OK, but leave it in. It should be there |
00:24.35 | ds | JockeHome: (this time...) gcc emits calls to floorf() instead of the actual call to floor() |
00:24.56 | JockeHome | 4.0? |
00:25.20 | ds | 3.4.4 |
00:25.39 | JockeHome | thats bad, I use 3.4.4 too |
00:26.17 | psm | ds: is it floorf or __floorf ? |
00:26.23 | ds | floorf |
00:26.36 | psm | ok, because I removed __floorf recently |
00:26.57 | ds | and it doesn't even do it when you'd expect, i.e., (float) f = floor ((float) g); |
00:27.39 | psm | that is floorf |
00:28.45 | ds | right |
00:28.55 | ds | it seems to be tied in with a memory load |
00:33.46 | JockeHome | gcc is beeing smart and replaces floor with floorf since its arg is a float? |
00:34.27 | psm | JockeHome: the first .purgem emits the warning, can that be suppressed ? |
00:34.29 | ds | yep |
00:34.35 | ds | but only sometimes |
00:36.08 | JockeHome | I guess that depends on the type of 'f'? |
00:36.13 | ds | no |
00:36.42 | JockeHome | psm: don't know, havn't looked |
00:37.06 | JockeHome | ds: :( |
00:40.24 | *** join/#uclibc thraxisp|away (n=thraxisp@ottawa-hs-64-26-147-20.d-ip.magma.ca) |
00:41.09 | JockeHome | psm: Is USE_MISC defined when you compile libm? |
00:41.50 | JockeHome | try removing #ifdef __USE_MISC in math.h |
00:42.41 | *** join/#uclibc andersen (n=andersee@codepoet.org) |
00:43.00 | psm | -D_SVID_SOURCE is defined and that pulls in through math.h/features.h __USE_MISC |
00:43.29 | *** join/#uclibc ashes (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
00:45.54 | CIA-8 | 03psm * r11968 10uClibc/ (Rules.mak extra/Configs/Config.in): Add configure option to allow IMA compiling |
00:46.01 | JockeHome | does all files include math.h? |
00:46.19 | psm | probably |
00:46.44 | psm | I get the error many times, not only once |
00:46.44 | JockeHome | still, try removing #ifdef __USE_MISC in math.h |
00:47.25 | psm | the same |
00:48.13 | psm | I get the error probably for each and every source or object file |
00:51.22 | JockeHome | strange |
00:52.21 | psm | I found compiling like this some real conflicts (in fcntl.h for ex.) , but this one seems to be really ok |
00:52.50 | JockeHome | is the error message the same for all files |
00:52.52 | JockeHome | ? |
00:52.59 | psm | exactly |
00:53.07 | CIA-8 | 03vapier * r11969 10uClibc/ (Makefile Makefile.in): cleanup the release target so that it doesnt kill my svn checkout anymore :/ |
00:53.21 | JockeHome | what line in s_lib_version.c? |
00:53.50 | psm | the one active (for you the last with _IEEE_ |
00:55.24 | JockeHome | redef _LIB_VERSION_TYPE to an int to what happens |
00:55.44 | psm | I use _POSIX_MODE, so mine is the line w/ _POSIX_ |
00:56.00 | psm | in *.c or math.h ? |
00:56.09 | JockeHome | math.h |
00:59.37 | psm | #define _LIB_VERSION_TYPE int allows to build |
01:00.14 | psm | done between enum and extern |
01:00.59 | JockeHome | smells like an gcc bug |
01:02.36 | JockeHome | maybe you need some enum compiler option? |
01:04.28 | JockeHome | what if you do typedef int _LIB_VERSION_TYPE? |
01:05.00 | psm | you are right, gcc-4.0.0 through gcc-4.0.2-20050929 builds it, 3.4.4 fails |
01:06.45 | psm | I have to disable the enum before, and then _POSIX_/_SVID_/_IEEE_ undeclared |
01:09.12 | ashes | # ip addr add ::1/64 dev lo |
01:09.12 | ashes | Error: an inet prefix is expected rather than "::1/64". |
01:09.18 | ashes | what am i doing wrong? |
01:10.24 | JockeHome | just rename the existing _LIB_VERSION_TYPE |
01:11.54 | psm | where rename it ? |
01:12.13 | JockeHome | in math.h |
01:12.22 | psm | glibc has exactly the same in s_lib_version.c and math.h |
01:12.32 | JockeHome | yup |
01:14.56 | psm | then it builds |
01:15.50 | JockeHome | gcc 3.4.4 seems to be buggy, just ask ds :) |
01:16.21 | psm | thx, until now it worked, but some have trouble rebuilding gcc w/ itself |
01:17.35 | JockeHome | did libc get any smaller with IMA? |
01:18.22 | psm | libc is far away of being doable, until I do have all other libs, I dont touch it (I managed only about 1/4 sources to get into IMA) |
01:19.20 | JockeHome | switch compiler :) |
01:19.35 | JockeHome | need to sleep now |
01:19.42 | psm | I will probably limit IMA to gcc-4 |
01:19.46 | psm | sleep well |
01:19.48 | psm | thx |
01:19.50 | JockeHome | bye |
01:21.41 | psm | bye |
01:26.02 | CIA-8 | 03psm * r11970 10uClibc/Rules.mak: Limit IMA compiling to gcc-4, gcc-3.4.4 can't cope w/ some sources |
01:36.32 | CIA-8 | 03psm * r11971 10uClibc/Rules.mak: Add -o /dev/null to AS_NEEDED check linker script check, else a.out polution all over the place |
01:49.22 | solar | ahh i see. |
02:37.09 | *** join/#uclibc thraxisp|away (n=thraxisp@ottawa-hs-64-26-147-20.d-ip.magma.ca) |
02:38.30 | *** join/#uclibc andersee (n=andersen@codepoet.org) |
02:58.47 | psm | andersee: ping |
03:00.03 | andersee | psm: evening |
03:00.24 | ashes | if i dont normally take vitamins, and im sick, is it fairly safe to go to the overdose threshhold of vitamin intake just while im sick? |
03:01.37 | psm | when do you think the new Makefiles should go active? should the links be created on startup or you want it all in the svn repo? |
03:02.08 | ashes | i may have taken two multi-vitamins by mistake. and im popping vitamin C and two odourless garlics every hour. and 2-3 vitamin A's |
03:02.37 | ashes | 2-3 vitamin A's per day |
03:02.49 | andersee | psm: according to: http://subversion.tigris.org/faq.html#symlinks |
03:03.03 | andersee | psm: svn can store symlinks w/o any problems |
03:03.17 | andersee | psm: so lets just go with that |
03:04.14 | psm | do you want them all (in every subdir) or only the minimal set needed to allow building , having a script that creates the rest? |
03:04.58 | andersee | hmm |
03:08.07 | andersee | psm: lets just run the script |
03:08.33 | andersee | psm: I think we will have some unhappy people if we archive symlinks |
03:08.53 | andersee | like the twisted souls that build under windows with cygwin |
03:09.34 | psm | we could optionally copy instead of link |
03:09.34 | andersee | we will need to however teach svn to ignore the symlinks |
03:10.15 | andersee | the makefiles are trivial enough a copy wouldn't be painful |
03:10.57 | andersee | psm: lets do that. Just copy them into place and check them in. |
03:11.07 | andersee | no mucking about with symlinks |
03:13.43 | psm | I would leave the "master copies" and the script, so that if these get updated (TOPDIR for ex. being removed), we can update all of them at once |
03:13.59 | andersee | seems reasonable |
03:14.16 | andersee | just need to change create_links.sh to copy instead of link |
03:14.21 | andersee | and perhaps rename it |
03:14.47 | psm | create_makefiles.sh ;) |
03:16.01 | psm | when should the change happen, I have tested on x86 only, and another one on vax (no shared support) |
03:16.27 | andersee | psm: go for it |
03:17.10 | andersee | psm: might as well get it done. If something is broken, its best to find out sooner rather than later |
03:17.47 | psm | what should happen w/ the Makefiles for archs that are not supported (old crt0.S), remove them ? |
03:20.02 | andersee | well, arches such as i960 and such have been sortof bitrotting for a while now |
03:20.38 | andersee | perhaps we need to mv then to libc/sysdeps/linux/broken/ |
03:20.43 | andersee | :) |
03:21.25 | andersee | suppose if anyone cares that will encourage them to fix things |
03:21.44 | andersee | and if nobody cares, we can nuke them after a while |
03:48.06 | ashes | from inetutils-cvs: undefined reference to `rpl_realloc' |
06:11.30 | *** join/#uclibc __cow (n=cow@M850P029.adsl.highway.telekom.at) |
06:22.19 | vapier | ashes: it's because you dont have gnu malloc turned on |
06:29.56 | ashes | inetutils-cvs died for another reason after |
06:30.12 | ashes | er |
06:30.20 | ashes | ping6 doesn't compile |
06:31.23 | vapier | ok ... ? |
06:31.36 | vapier | look at the configure output ... find the lines talking about malloc |
06:32.55 | vapier | http://rafb.net/paste/results/QCcikq40.html |
06:34.59 | CIA-8 | 03vapier * r11972 10uClibc/docs/uclibc.org/FAQ.html: add an entry about gnu malloc compat option |
06:35.52 | vapier | andersen: ping ... how come the FAQ has a ton of <br>'s at the end of it ? |
06:39.20 | CIA-8 | 03vapier * r11973 10uClibc/docs/uclibc.org/header.html: add a charset def and update the doctype |
07:01.35 | CIA-8 | 03vapier * r11974 10buildroot/package/ (6 files in 3 dirs): rename dropbear_sshd to just dropbear |
07:16.39 | CIA-8 | 03vapier * r11975 10buildroot/package/zlib/zlib.mk: use := instead of = for variables and make sure zlib builds with target cflags |
07:50.22 | CIA-8 | 03vapier * r11976 10uClibc/docs/uclibc.org/FAQ.html: close <a> tags like they should be for valid html |
07:55.31 | CIA-8 | 03vapier * r11977 10uClibc/docs/uclibc.org/products.html: use & instead of &, tweak style, and add link for Embedded Gentoo |
07:57.52 | CIA-8 | 03vapier * r11978 10busybox/docs/busybox.net/header.html: update doctype and add a charset encoding |
08:03.55 | CIA-8 | 03vapier * r11979 10busybox/docs/busybox.net/FAQ.html: fix <p> and <a> tags |
08:14.35 | CIA-8 | 03vapier * r11980 10busybox/docs/busybox.net/shame.html: add ZyXEL to the shame list |
09:45.23 | CIA-8 | 03vapier * r11981 10buildroot/package/gnuconfig/ (CVS/Entries ChangeLog config.guess): sync with config cvs |
10:08.21 | *** join/#uclibc Sgt-Donan (n=Loutre@mto30-1-82-242-53-8.fbx.proxad.net) |
10:12.00 | CIA-8 | 03psm * r11982 10uClibc/extra/scripts/ (create_links.sh create_makefiles.sh): Renamed script to create_makefiles.sh, copy instead of link (to allow cygwin users to build), creates all Makefiles now, should be only used if the master copies are updated |
10:16.20 | CIA-8 | 03psm * r11983 10uClibc/ (70 files in 70 dirs): Replace all Makefiles for new build infrastucture |
10:18.44 | CIA-8 | 03psm * r11984 10uClibc/libc/misc/internals/ (Makefile.in __uClibc_main.c static.c): Hopefully correct undefined __libc_stack_end issue |
11:32.46 | *** join/#uclibc ico2 (n=ico2@modem-1638.chimpanzee.dialup.pol.co.uk) |
11:32.48 | ico2 | hi |
11:32.55 | ico2 | having some trouble with pivot_root |
11:33.09 | ico2 | i create the ramdisk etc |
11:33.25 | ico2 | i use unionfs to link everything onto it |
11:33.30 | ico2 | then i cd into it |
11:33.38 | ico2 | and pivot_root . old_root |
11:33.48 | ico2 | (the old_root dir exists) |
11:34.01 | ico2 | and i get an Invalid argument error |
11:34.04 | ico2 | why is this? |
11:41.52 | *** join/#uclibc woglinde (i=woglinde@e178076105.adsl.alicedsl.de) |
11:54.46 | woglinde | jo zecke |
11:55.05 | woglinde | ups |
12:00.41 | *** join/#uclibc ashes_ (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
13:00.16 | *** join/#uclibc thraxisp (n=thraxisp@ottawa-hs-64-26-147-20.d-ip.magma.ca) |
13:46.28 | CIA-8 | 03jocke * r11985 10uClibc/ldso/ldso/ (dl-elf.c ldso.c powerpc/elfinterp.c): |
13:46.29 | CIA-8 | Hopefully fix the bug Oleg reported in http://uclibc.org/lists/uclibc/2005-October/012809.html |
13:46.29 | CIA-8 | This will only fix powerpc. Should be easy to fix the other arches. |
13:49.39 | *** join/#uclibc ico2 (n=ico2@modem-1400.elephant.dialup.pol.co.uk) |
14:34.29 | *** join/#uclibc ico2 (n=ico2@modem-2229.chameleon.dialup.pol.co.uk) |
16:11.45 | *** join/#uclibc __cow (n=cow@M801P017.adsl.highway.telekom.at) |
17:42.22 | *** join/#uclibc ico2 (n=ico2@modem-2835.cougar.dialup.pol.co.uk) |
18:16.04 | *** join/#uclibc ambrose (n=bjb@ns1.linuxbutler.ca) |
18:39.32 | *** join/#uclibc nickv111 (n=nick@70-33-44-221.clspco.adelphia.net) |
18:40.01 | nickv111 | I'm trying to build a small uclibc/busybox based rootfs. How would I go about linking busybox against uClibc? |
18:40.57 | nickv111 | I have uClibc installed on my rootfs and I need busybox to not link against my host's glibc, but rather the new rootfs's uClibc |
18:46.51 | CIA-8 | 03sjhill 07uClibc-nptl * r11986 10/libc/sysdeps/linux/mips/sysdep.h: |
18:46.51 | CIA-8 | Only include 'regdef.h' when using assembly code. Your character has |
18:46.51 | CIA-8 | gained +1 idiot experience points. |
19:07.31 | ashes_ | nickv111: use CC="gcc --dynamic-linker /lib/ld-uClibc.so.0" |
19:08.06 | nickv111 | ashes: Oh, thanks |
19:08.21 | nickv111 | ashes: Unfortunately, uClibc is refusing to build |
19:08.29 | nickv111 | Probably because I'm using the svn version |
19:08.35 | nickv111 | I'm going to try using the sane versions |
19:09.07 | CIA-8 | 03sjhill 07uClibc-nptl * r11987 10/test/Makefile: (log message trimmed) |
19:09.07 | CIA-8 | Do not add entries: pthread, dlopen, tls to ALL_SUBDIRS unless thos |
19:09.07 | CIA-8 | features are actually enabled. Add 'tls' as a new test suite in the |
19:09.54 | CIA-8 | 03sjhill 07uClibc-nptl * r11988 10/ldso/libdl/Makefile: Add '-DSHARED' flag when building the shared 'libdl'. |
19:11.50 | CIA-8 | 03sjhill 07uClibc-nptl * r11989 10/ldso/include/ldsodefs.h: |
19:11.50 | CIA-8 | Fix up dependencies for when 'extern' is used. This file is a mess |
19:11.50 | CIA-8 | and I look forward to eliminating it soon. |
19:14.29 | nickv111 | Yeah, using a sane version of uClibc seems to work fine |
19:14.50 | CIA-8 | 03sjhill 07uClibc-nptl * r11990 10/ldso/ (include/ldso.h ldso/dl-debug.c ldso/dl-tls.c): (log message trimmed) |
19:14.50 | CIA-8 | Removed old 'assert' function and created new '_dl_assert' macro and |
19:14.51 | CIA-8 | update usage of it. Also removed '_dl_fatal_printf' macro and usage |
19:48.51 | CIA-8 | 03sjhill 07uClibc-nptl * r11991 10/Makerules: Add new file to branch. |
20:02.05 | *** join/#uclibc ico2 (n=ico2@modem-2678.elephant.dialup.pol.co.uk) |
20:02.23 | nickv111 | ashes: Hmm... Your suggestion doesn't seem to help, as I don't have a --dynamic-linker option in my gcc. |
20:14.21 | ashes | take away one - |
20:15.31 | ashes | -Wl,--dynamic-linker,/lib/ld-uClibc.so.0 |
20:15.44 | ashes | or |
20:15.47 | ashes | CC="gcc -Wl,--dynamic-linker,/lib/ld-uClibc.so.0 /lib/libc.so.0" |
20:16.31 | ashes | --dynamic-linker is an ld option |
20:16.36 | *** part/#uclibc ico2 (n=ico2@modem-2678.elephant.dialup.pol.co.uk) |
20:16.41 | nickv111 | Oh... |
20:17.31 | nickv111 | Wow. I think it's working :) |
20:17.35 | nickv111 | Thank you |
20:17.42 | ashes | :) |
20:17.44 | nickv111 | Nobody else would help me |
20:18.39 | nickv111 | Actually, I don't think it did work :-\ |
20:18.50 | nickv111 | Doing "ldd busybox" shows that it's still linked to glibc |
20:21.19 | nickv111 | I don't think it's using my $CC variable |
20:25.35 | nickv111 | No, it's definitely not using my $CC |
20:25.39 | CIA-8 | 03sjhill 07uClibc-nptl * r11992 10/libc/pwd_grp/pwd_grp_internal.c: Add from trunk. |
20:46.24 | nickv111 | Got it. |