| 00:09.11 | *** join/#uclibc dieter|work (n=dieter@dslb-088-066-196-071.pools.arcor-ip.net) |
| 00:28.25 | khem | austinf: I have more fixes for nptl |
| 00:28.32 | khem | can you have a quick look |
| 00:28.37 | austinf | sure |
| 00:28.39 | austinf | list? |
| 00:28.46 | khem | let me point you to patch |
| 00:28.49 | austinf | ok |
| 00:28.49 | khem | not sent |
| 00:28.51 | khem | yet |
| 00:29.41 | khem | http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/uclibc/uclibc-nptl/mips-nptl-fix-dynamic-global-tls.patch |
| 00:29.42 | *** join/#uclibc animeloe (n=animeloe@unaffiliated/animeloe) |
| 00:30.06 | khem | austinf: this fixes a problem in uclibc when something build with -fPIC was accessing TLS variables on mips |
| 00:30.50 | khem | e.g. libpthread.so is build with -fPIC and accesses to h_errno are diverted thru h_errno_location from libpthread.so when linked |
| 00:31.05 | khem | and this would segfault if someone said gcc -pthread hello.c |
| 00:31.17 | khem | but it would work if someone just did gcc hello.c |
| 00:31.39 | khem | in short global dynamic accesses were borked on for TLS |
| 00:32.11 | khem | other part of patch hides the __libc_errno __libc_h_errno __libc_resp TLS vars |
| 00:32.24 | khem | makes the access faster to them when inside libc |
| 00:32.52 | khem | because they can use local dynamic sequences which are one relocation less |
| 00:33.26 | austinf | yes |
| 00:33.28 | khem | the second part needs changes in arch specific files for sh, arm and sparc prolly |
| 00:33.34 | austinf | yeah |
| 00:33.38 | khem | I makefile |
| 00:34.31 | austinf | pt-__syscall_rt_sigaction.c flags ? |
| 00:34.43 | khem | yes those |
| 00:34.52 | khem | I know its needed for arm too I guess |
| 00:36.17 | austinf | why was that needed? |
| 00:36.45 | austinf | for hidden __libc_errno access |
| 00:36.45 | austinf | ? |
| 00:36.48 | khem | yep |
| 00:38.02 | austinf | ok, any other syscalls that need that do you think? |
| 00:38.08 | austinf | especially for other archs? |
| 00:45.15 | khem | I will check arm shortly |
| 00:45.29 | khem | but I dont think so |
| 00:46.00 | khem | people outside libc should only see the real variables |
| 00:46.12 | khem | right now we keep __libc ones visible |
| 00:46.17 | khem | which is not right |
| 00:46.21 | austinf | I agree |
| 00:46.28 | khem | plus is suboptimal |
| 00:46.34 | austinf | just need to make sure we don't crash :) |
| 00:46.41 | khem | heh |
| 00:46.52 | *** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe) |
| 00:46.57 | khem | give this a try in your env if you have something for non mips |
| 00:47.05 | khem | sparc ? or x86 ? |
| 00:48.09 | austinf | I can try both, but not today |
| 00:48.18 | khem | sure |
| 00:48.25 | khem | sometime this week ? |
| 00:48.53 | khem | I will actually divide this patch into 2 |
| 00:49.00 | austinf | great |
| 00:49.02 | khem | one which is mips/nptl specific |
| 00:49.08 | khem | and another which is errno stuff |
| 00:49.15 | khem | that way it will be clean |
| 00:49.20 | austinf | that can go to master? |
| 00:49.27 | khem | yeah |
| 00:49.33 | khem | prolly |
| 00:49.36 | khem | I have to check |
| 00:53.08 | austinf | the mips ldso stuff looks good. Curious, did your fixes come from glibc to start with? |
| 01:01.14 | khem | not completely |
| 01:01.22 | khem | I looked at other arches in uclibc |
| 01:01.25 | khem | and debugged the crap |
| 01:01.28 | khem | to see the problem |
| 01:01.33 | khem | :) |
| 01:01.52 | khem | and some part like crt stuff did come from glibc |
| 01:32.11 | austinf | khem, you see landley's emails? |
| 02:05.25 | *** join/#uclibc DuratarskeyK (n=wedge@93-181-200-200.pppoe.yaroslavl.ru) |
| 02:37.45 | *** join/#uclibc dileX (n=sd@p5B2EE338.dip.t-dialin.net) |
| 02:58.54 | *** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net) |
| 03:04.12 | *** join/#uclibc dileX (n=sd@vpn-eu1.unidsl.de) |
| 03:09.10 | *** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe) |
| 03:17.45 | CIA-59 | 03vda.linux 07master * rfeadfe742ae9 10busybox/editors/patch.c: patch: add more long opts; but remove -g for now |
| 03:20.26 | CIA-59 | 03mizvekov 07master * r6f99c91e431a 10busybox/ (editors/diff.c testsuite/diff.tests): diff: fix flag -B, cleanups and a couple more tests V2 |
| 03:25.44 | *** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe) |
| 03:43.25 | *** join/#uclibc sir_lewk (n=jlg95@tux64-10.cs.drexel.edu) |
| 04:06.21 | *** join/#uclibc busybot_ (n=busybot@busybox.osuosl.org) |
| 04:07.23 | *** join/#uclibc dileX_ (n=sd@p5B2EE338.dip.t-dialin.net) |
| 04:11.08 | CIA-59 | 03vda.linux 07master * r8b8d32ea56d4 10busybox/editors/patch.c: patch: accept and ignore -g NUM |
| 04:16.16 | *** join/#uclibc dileX_ (n=sd@vpn-eu2.unidsl.de) |
| 05:19.27 | *** join/#uclibc dileX_ (n=sd@p5B2EE338.dip.t-dialin.net) |
| 05:43.32 | *** join/#uclibc gnomon_ (n=gnomon@CPE0022158a8221-CM000f9f776f96.cpe.net.cable.rogers.com) |
| 06:17.19 | *** join/#uclibc mnt_real_ (n=sinan@bas1-montreal43-1177754708.dsl.bell.ca) |
| 06:42.20 | *** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net) |
| 06:52.12 | *** join/#uclibc ncopa (n=ncopa@245.39.189.109.customer.cdi.no) |
| 08:50.04 | *** join/#uclibc hgb (n=hgb@tussi.moria.no) [NETSPLIT VICTIM] |
| 08:50.04 | *** join/#uclibc Jacmet (i=peko@stolen.plutonium.dk) [NETSPLIT VICTIM] |
| 08:50.04 | *** join/#uclibc neli (i=micha@88.159.211.247) [NETSPLIT VICTIM] |
| 10:05.35 | *** join/#uclibc chezgi (n=abbas@91.98.164.215) |
| 10:07.01 | chezgi | i want to boot my linux from labeled partition without using initrd (i have only one partition). is there any way to do this? |
| 10:25.06 | *** part/#uclibc chezgi (n=abbas@91.98.164.215) |
| 10:25.50 | dougmencken | if your kernel is not very modular, of course |
| 10:29.51 | *** join/#uclibc hgb (n=hgb@tussi.moria.no) [NETSPLIT VICTIM] |
| 10:29.51 | *** join/#uclibc Jacmet (i=peko@stolen.plutonium.dk) [NETSPLIT VICTIM] |
| 10:29.51 | *** join/#uclibc neli (i=micha@88.159.211.247) [NETSPLIT VICTIM] |
| 10:31.49 | *** join/#uclibc hgb (n=hgb@tussi.moria.no) [NETSPLIT VICTIM] |
| 10:31.50 | *** join/#uclibc Jacmet (i=peko@stolen.plutonium.dk) [NETSPLIT VICTIM] |
| 10:31.50 | *** join/#uclibc neli (i=micha@88.159.211.247) [NETSPLIT VICTIM] |
| 10:35.35 | *** join/#uclibc _Lucretia_ (n=munkee@pdpc/supporter/active/lucretia) |
| 10:47.02 | *** join/#uclibc hgb (n=hgb@tussi.moria.no) [NETSPLIT VICTIM] |
| 10:47.02 | *** join/#uclibc Jacmet (i=peko@stolen.plutonium.dk) [NETSPLIT VICTIM] |
| 10:47.02 | *** join/#uclibc neli (i=micha@88.159.211.247) [NETSPLIT VICTIM] |
| 11:04.24 | *** join/#uclibc hgb (n=hgb@tussi.moria.no) [NETSPLIT VICTIM] |
| 11:19.05 | *** join/#uclibc Jacmet (i=peko@stolen.plutonium.dk) |
| 11:22.43 | *** join/#uclibc hiyuh (n=hiyuh@KD124214245222.ppp-bb.dion.ne.jp) |
| 11:40.20 | _Lucretia_ | hi, just tried to build for gcc-4.4.x, and it's failing with the following line: cp: cannot stat `/home/laguest/opt/arm-linux/lib/gcc-lib/arm-linux-uclibcgnueabi/4.4.2/include/syslimits.h': No such file or directory |
| 12:03.46 | CIA-59 | 03raj.khem 070_9_30 * r2911103dd4a0 10uClibc/libc/sysdeps/linux/mips/crt1.S: Make use of macros from sys/asm.h in crt1.S |
| 12:03.47 | CIA-59 | 03rep.dot.nop 070_9_30 * r83a09cd5c9ed 10uClibc/libc/inet/rpc/rpc_prot.c: rpc: fix typo in version mismatch msg |
| 12:10.17 | CIA-59 | 03rep.dot.nop 07refs/tags/0_9_30_2 * raa319dcdbb0a 10uClibc/Rules.mak: bump version to 0.9.30.2 |
| 12:13.27 | *** join/#uclibc gustavoz (n=gustavoz@host5.201-253-139.telecom.net.ar) |
| 12:31.30 | *** join/#uclibc PortaLu (n=Lucretia@5ac95b14.bb.sky.com) |
| 12:41.00 | CIA-59 | 03rep.dot.nop 070_9_30 * rd75ad2e129b3 10uClibc/ (extra/locale/programs/locale.c libc/misc/wchar/wchar.c): fix make {,install_}{,host}utils |
| 12:41.01 | CIA-59 | 03rep.dot.nop 070_9_30 * r47e88c04e012 10uClibc/Makefile.in: {,host}utils depend on headers |
| 12:42.55 | dileX | cool uclibc locale (wchar) fixes |
| 12:58.41 | CIA-59 | 03rep.dot.nop 07master * r8a2d6677c3d4 10uClibc/libc/sysdeps/linux/common/sysctl.c: silence warning about implicit decl of sysctl |
| 13:54.07 | *** join/#uclibc sjhill (n=sjhill@home.bethel-hill.org) |
| 13:59.01 | dileX | I am trying to build a mipsel toolchain for a TI-ar7 router platform on an i386 debian/sid host. |
| 13:59.01 | dileX | Unfortunately, building gcc-4.4.3 for the target fails with any combination of gmp (4.2.4/4.3.1/4.3.2/5.0.0) and mpfr (2.4.1/2.4.2). |
| 13:59.01 | dileX | Target toolchain has uclibc-0.9.30.2 and binutils-2.20.51.0.5. |
| 13:59.01 | dileX | It always breaks... here http://paste.pocoo.org/show/168363/. |
| 13:59.02 | dileX | Any idea? |
| 14:01.26 | *** join/#uclibc Christos_N (n=Christos@ppp-94-66-26-41.home.otenet.gr) |
| 14:13.10 | dileX | mpfr-2.4.2/config.log: http://pastebin.ca/1761554 and gmp-5.0.0/config.log: http://pastebin.ca/1761555 |
| 14:20.13 | *** join/#uclibc Christos_N_ (n=Christos@ppp-94-66-26-41.home.otenet.gr) |
| 14:21.39 | blindvt | lunch time |
| 14:34.04 | *** join/#uclibc KaiForce (n=chatzill@adsl-70-228-89-235.dsl.akrnoh.ameritech.net) |
| 14:35.00 | *** join/#uclibc gnomon (n=gnomon@CPE0022158a8221-CM000f9f776f96.cpe.net.cable.rogers.com) |
| 15:18.21 | *** join/#uclibc likewise (n=likewise@215-65-ftth.onsneteindhoven.nl) |
| 15:20.49 | *** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa) |
| 15:22.27 | *** join/#uclibc austinf (n=austinf@75-92-1-28.war.clearwire-wmx.net) |
| 15:27.20 | dileX | just for the records: I am trying to build a new target toolchain now with following patch |
| 15:27.23 | dileX | Due to a change in GCC 4.4 internals for MIPS, the MPFR build fails on MIPS processors with GCC 4.4 and later. The longlong.h patch (written by Aurélien Jarno) solves this problem. |
| 15:27.23 | dileX | [Changeset: 6638] |
| 16:03.58 | *** join/#uclibc wberrier (n=wberrier@65-100-208-11.slkc.qwest.net) |
| 16:15.36 | *** join/#uclibc wberrier (n=wberrier@65-100-208-11.slkc.qwest.net) |
| 16:36.11 | *** join/#uclibc mnt_real (n=sinan@bas1-montreal43-1177754708.dsl.bell.ca) |
| 17:00.43 | *** join/#uclibc chezgi (n=abbas@91.98.164.215) |
| 17:07.16 | dileX | OK, mpfr-2.4.2-p3 and gmp-5.0.0 was successful. unfortunately, I have the next breakage: http://paste.pocoo.org/show/168422/ |
| 17:07.19 | dileX | might someone look at this? |
| 17:11.48 | *** join/#uclibc austinf (n=austinf@74.92.231.110) |
| 17:14.39 | dileX | /home/sd/pbuilder/freetz/freetz-trunk/toolchain/build/gcc-4.4.3-uClibc-0.9.30.3/mipsel-linux-uclibc/usr/include/bits/types.h:133:3: error: #error your machine is neither 32 bit or 64 bit ... it must be magical |
| 17:14.47 | dileX | what about this? |
| 17:16.17 | chezgi | where is the problem when "udevadmin monitor" doesn't show anything. |
| 17:51.52 | *** join/#uclibc neli (n=neli@88.159.211.247) |
| 18:03.18 | *** join/#uclibc austinf (n=austinf@74.92.231.110) |
| 18:23.40 | dileX | gcc-3.4.6/gcc/config/mips/linux.h: http://pastebin.ca/1761825 |
| 18:23.47 | dileX | gcc-4.4.3/gcc/config/mips/linux.h: http://pastebin.ca/1761826 |
| 18:24.03 | dileX | builtin_define ("_MIPS_SZPTR=32"); |
| 18:24.21 | dileX | so there is no _MIPS_SZPTR defined for gcc-4.4.3 |
| 18:26.44 | dougmencken | it was about that uClibc's 'bits/wordsize.h' for MIPS depends on _MIPS_SZPTR, which is now not defined |
| 18:27.43 | dileX | uclibc/libc/sysdeps/linux/mips/bits/wordsize.h: #define __WORDSIZE _MIPS_SZPTR |
| 18:28.04 | dileX | can someone fix that, please? |
| 18:28.32 | austinf | khem, do you know what it should be instead? |
| 18:28.45 | dougmencken | 32 or 64 :) |
| 18:28.59 | austinf | right, but what define does gcc provide now? |
| 18:31.26 | dougmencken | gcc/config/mips/mips.h |
| 18:32.03 | dougmencken | #define ABI_32 0 #define ABI_N32 1 #define ABI_64 2 #define ABI_EABI 3 #define ABI_O64 4 |
| 18:35.00 | dougmencken | btw, why don't use absolutely portable #define WORDSIZE ((sizeof(int))*8) ? |
| 18:42.23 | austinf | sounds good to me, send patch to mailing list please. |
| 18:43.59 | khem | _MIPS_SZPTR is gcc internal define |
| 18:44.08 | khem | builtin |
| 18:44.20 | dougmencken | now it is not |
| 18:45.45 | khem | dougmencken: it is still in there |
| 18:46.02 | khem | mips-oe-linux-uclibc-gcc -x c /dev/null -E -dD|grep SZPTR |
| 18:46.08 | khem | #define _MIPS_SZPTR 32 |
| 18:46.24 | dougmencken | which is your gcc version? |
| 18:46.35 | khem | trunk latest |
| 18:49.47 | dougmencken | <PROTECTED> |
| 18:50.02 | khem | dougmencken: #define WORDSIZE ((sizeof(int))*8) is wrong too |
| 18:50.17 | khem | int is always 4 byte long |
| 18:50.20 | dougmencken | well, then long |
| 18:50.26 | khem | #define WORDSIZE ((sizeof(long))*8) |
| 18:50.28 | khem | is ok |
| 18:52.02 | dileX | khem: wanna push that fix? |
| 18:52.34 | khem | yes so what its internal define in gcc |
| 18:53.01 | khem | dileX: no I dont think there is something to fix |
| 18:53.53 | *** join/#uclibc PortaLu (n=Lucretia@5ac95b14.bb.sky.com) |
| 18:54.52 | dougmencken | khem, it is: empty __WORDSIZE results in " #error your machine is neither 32 bit or 64 bit ... it must be magical" |
| 18:55.09 | khem | why does it end up empty |
| 18:56.12 | dougmencken | wait, I'm checkouting the fresh trunk; in newly released gcc 4.4.3 doesn't have _MIPS_SZPTR |
| 18:56.41 | dileX | dougmencken: yeah, I had that |
| 18:56.43 | dileX | [18:13:57] <dileX> /home/sd/pbuilder/freetz/freetz-trunk/toolchain/build/gcc-4.4.3-uClibc-0.9.30.3/mipsel-linux-uclibc/usr/include/bits/types.h:133:3: error: #error your machine is neither 32 bit or 64 bit ... it must be magical |
| 18:57.04 | khem | dileX: what were you trying to build |
| 18:57.14 | khem | and can you paste the commandline somewhere |
| 18:57.33 | dileX | I am rebuilding from scratch right now |
| 18:58.23 | dileX | thats the source of freetz project - build-system is based on buildroot (but modified) |
| 18:59.08 | dileX | http://trac.freetz.org/wiki/WikiStart.en |
| 19:00.16 | khem | dileX: this would happen if you compile something with -undef |
| 19:07.42 | dougmencken | so why don't try "# ifdef _MIPS_SZPTR .... #else #define WORDSIZE ((sizeof(long))*8)"? |
| 19:08.03 | dougmencken | also, http://pastebin.com/m2043b073 |
| 19:09.00 | khem | that wont work either |
| 19:09.04 | khem | it needs a static define |
| 19:09.09 | khem | and mips has 2 abi's |
| 19:09.31 | khem | the current approarch is nice that you can use same wordsize.h for all abis |
| 19:09.48 | khem | and let gcc decide upon machine length |
| 19:09.52 | khem | of word |
| 19:10.20 | khem | as I said problem is because you must be passing -undef on the commandline to gcc |
| 19:10.56 | dougmencken | khem, the problem is because gcc 4.4.3 doesn't have *any* define for _MIPS_SZPTR in its headers |
| 19:12.11 | dileX | khem: http://paste.pocoo.org/show/168493/ |
| 19:12.13 | khem | hmm hard to believe |
| 19:12.35 | blindvt | dougmencken, that sounds dubious |
| 19:12.43 | blindvt | s40:/scratch/src/gcc-4.5# grep _MIPS_SZPTR gcc/config/mips/mips.h |
| 19:12.43 | blindvt | <PROTECTED> |
| 19:13.06 | dougmencken | 4.4.3 |
| 19:13.13 | blindvt | so no, it's 99.999% a configury issue |
| 19:13.33 | blindvt | dougmencken, nonsense. http://gcc.gnu.org/viewcvs?view=revision&revision=127113 |
| 19:13.40 | blindvt | dougmencken, been like this since 2007 |
| 19:14.23 | khem | dileX: /home/sd/pbuilder/freetz/freetz-trunk/toolchain/target/bin/mipsel-linux-uclibc-gcc |
| 19:14.39 | khem | -x c /dev/null -E -dD|grep SZPTR |
| 19:14.50 | khem | what does that say ? |
| 19:15.12 | khem | blindvt: hi |
| 19:15.19 | dileX | /home/sd/pbuilder/freetz/freetz-trunk/toolchain/target/bin/mipsel-linux-uclibc-gcc -x c /dev/null -E -dD | grep SZPTR |
| 19:15.19 | dileX | #define _MIPS_SZPTR 32 |
| 19:15.29 | dileX | hmm |
| 19:15.30 | khem | well there you go then |
| 19:15.32 | dileX | funny |
| 19:15.48 | khem | I asked for failing commandline |
| 19:15.51 | blindvt | khem, hi |
| 19:15.58 | khem | and I also said you must be using -undef |
| 19:16.01 | khem | did you check ? |
| 19:16.20 | khem | blindvt: I have another patch for mips nptl |
| 19:16.31 | khem | which lets global dynamic model work now |
| 19:17.04 | blindvt | khem, sure, please feel free to push it right to the branch |
| 19:17.41 | khem | if you want to have a look I committed it into OE here http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/uclibc/uclibc-nptl/mips-nptl-fix-dynamic-global-tls.patch?id=5cc04010e70378484efe549b30b411f840e825f5 |
| 19:18.05 | khem | blindvt: ok |
| 19:18.14 | khem | I have tested it |
| 19:18.39 | dileX | khem: I have done s/#define __WORDSIZE _MIPS_SZPTR/#define __WORDSIZE 32. so I must re-build again? |
| 19:18.41 | blindvt | khem, will need adjustment for 0_9_30 since there's no tls yet in there |
| 19:18.58 | dileX | (/me thinks yes) |
| 19:19.33 | blindvt | dileX, you really shouldn't be doing this. It's wrong and just papers over the real thing, if anything |
| 19:19.33 | khem | blindvt: ok |
| 19:19.40 | khem | blindvt: we should plan .31 now |
| 19:19.47 | khem | with all supported nptl |
| 19:20.01 | dileX | blindvt: yupp, it was only for testing purposes |
| 19:20.02 | khem | atleast it will give people change to test it whatever we have |
| 19:20.18 | khem | s/change/chance |
| 19:22.23 | blindvt | khem, sounds ok. Do !mips also need guards in dl-tls.h? If it's needed for 0.9.30.3 too, then please install an appropriate thing on the 0_9_30 branch too |
| 19:23.14 | austinf | khem, dont forget to patch the arm makefile as well for pt-__syscall-rt-sigaction |
| 19:23.14 | blindvt | khem, did austinf sync from upstream yet and cleanup the stuff we talked about previously? |
| 19:23.26 | austinf | I haven't had a chance to sync with upstream |
| 19:23.31 | blindvt | austinf, heya |
| 19:23.50 | khem | austinf: I will |
| 19:24.27 | blindvt | i've NPTL for powerpc lying around here for some time now that i will try to commit soonish |
| 19:24.50 | austinf | blindvt, I don't recall if I talked to you since I had to revert that nptl_headers work |
| 19:25.12 | blindvt | it survives the testsuite so should be quote useable i think |
| 19:25.14 | austinf | it broke 'make headers', since it requires the cross compiler to generate |
| 19:25.23 | khem | blindvt: powerpc nptl cool |
| 19:25.37 | khem | blindvt: do you have target to test it on |
| 19:25.43 | austinf | blindvt, did you do cancellation for ppc too? |
| 19:26.18 | blindvt | austinf, not yet, i think (at least i don't remember offhand). If some stuff needs to be generated, then it has to be in pregen and _NOT_ in headers (as you certainly know). What was the problem? |
| 19:26.52 | austinf | it is in pregen now, but we had talked about moving it to headers before. I was just letting you know the end result |
| 19:27.54 | blindvt | khem, no, i don't have a target. I used qemu (don't ask me which machine exactly, it could have been a \"mpc8544ds\", whatever that is) |
| 19:28.14 | khem | blindvt: cool |
| 19:28.44 | khem | blindvt: what kernel version boots on it and the .config for kernel |
| 19:28.47 | blindvt | austinf, i admit that i don't remember offhand, i think the cancellation test worked, yes |
| 19:29.00 | khem | if you have handy I would be interested to have it |
| 19:29.16 | blindvt | khem, let me have a look.. mompls |
| 19:29.28 | khem | blindvt: cool thx |
| 19:29.37 | dileX | khem: shall I rebuild? |
| 19:29.56 | khem | dileX: I dont know what the problem is |
| 19:30.18 | khem | I could build gcc well on mips uclibc |
| 19:30.25 | khem | so it must be some of your config |
| 19:34.01 | dileX | uclibc-config? or configure line? |
| 19:34.14 | blindvt | khem, wow, no idea where my PPC.img is.. the kernel .config was based off the debian sid kernel .config, but adjusted for plain 2.6.30.something (which was current back then) |
| 19:34.28 | khem | blindvt: ok |
| 19:34.58 | khem | blindvt: did it boot zImage or uImage ? |
| 19:35.30 | blindvt | khem, iirc zImage, but i'm not 100% sure |
| 19:35.45 | blindvt | khem, let's wait for find to finish :) |
| 19:40.49 | *** join/#uclibc tchan1 (n=tchan@c-69-243-144-70.hsd1.il.comcast.net) |
| 19:40.52 | khem | :) |
| 19:46.36 | blindvt | khem, what's that installfix.patch? |
| 19:47.06 | khem | blindvt: helps to get parallel build |
| 19:47.10 | khem | going |
| 19:47.23 | khem | blindvt: I have not proposed it upstream yet |
| 19:49.00 | dileX | khem: the initial problem occured from this broken build-log http://paste.pocoo.org/show/168422/ |
| 19:49.15 | dileX | initial discussion* |
| 19:54.47 | khem | dileX: its gcc configure problem I think |
| 19:54.55 | khem | that needs to be fixed |
| 19:55.08 | khem | I have fixed it in Openembedded have to look there |
| 19:55.55 | *** join/#uclibc sjhill (n=sjhill@home.bethel-hill.org) |
| 19:56.04 | dileX | khem: you have a link? |
| 19:56.20 | khem | google is your friend |
| 19:56.20 | blindvt | khem, uclibc.org/~aldot/nptl-ppc |
| 19:56.46 | blindvt | khem, that was my configury |
| 19:58.02 | blindvt | khem, for 2.6.31.6 which was current back then. Beware, the kernel config is _huge_ but good enough for testing ;) |
| 19:58.34 | *** join/#uclibc Dj-Death (n=djdeath@potipota.net) |
| 19:58.44 | khem | blindvt: ok thx |
| 19:58.53 | khem | I will try it out |
| 19:59.59 | khem | MPC52xx |
| 20:00.02 | khem | is what you used |
| 20:01.54 | blindvt | khem, history tells me that the last thing i ran was qemu-system-ppc -kernel vmlinuxz -initrd initrd.xz -hda ../PPC.img -net nic,vlan=0 -net tap,vlan=0,ifname=tap0,script=/etc/qemu-brup -append "root=/dev/hdc4" |
| 20:02.12 | *** join/#uclibc neli (n=neli@88.159.211.247) [NETSPLIT VICTIM] |
| 20:02.13 | blindvt | khem, HTH |
| 20:02.26 | khem | vmlinuxz |
| 20:02.37 | dileX | khem: hmm, I am browsing through http://git.openembedded.org/cgit.cgi/openembedded/tree/, but where shall this gcc-configure problem be fixed? |
| 20:03.09 | blindvt | khem, i was also playing with xz (formerly known as lzma2), just for the kicks :P |
| 20:04.40 | blindvt | dileX, recipes/gcc |
| 20:04.50 | dileX | yeah, I found it |
| 20:05.02 | dileX | but I dont see gcc-4.4.3 |
| 20:05.43 | khem | so what |
| 20:06.02 | khem | you can use 4.4.2 ones |
| 20:07.37 | dileX | http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/gcc/gcc-4.4.2 |
| 20:07.51 | blindvt | khem, ack for your installfix.patch |
| 20:08.16 | khem | blindvt: thx ok |
| 20:09.35 | dileX | khem: which fix do you mean in special? |
| 20:11.01 | dileX | think of zecke-no-host-includes.patch |
| 20:11.45 | khem | gcc-flags-for-build.patch |
| 20:12.51 | dileX | will see if they apply to gcc-4.4.3 |
| 20:13.16 | khem | they should in worst case you might need to regenerate configure scripts |
| 20:13.26 | khem | buts thats not a rocket science |
| 20:14.14 | blindvt | khem, i was thinking about pushing these patches to stable, but unfortunately cannot test any of them in a sensible way.. What do you think? uclibc.org/~aldot/uClibc/for-0.9.30.3 |
| 20:15.53 | blindvt | khem, your o_cloexec patch is a bit dubious in that it seems to have forgotten to add the kernel-features snippet? |
| 20:15.54 | khem | blindvt: people who were using arm/oabi complained |
| 20:16.30 | khem | with the new syscall common stuff |
| 20:16.40 | khem | I remember running into a roadblock |
| 20:16.47 | khem | which was discussed on ml |
| 20:16.58 | khem | blindvt: all those patches would need extensive testing |
| 20:17.16 | dougmencken | I can test on powerpc |
| 20:17.17 | khem | people try weird architectures |
| 20:17.18 | blindvt | khem, i think i've fixed the arm-oabi vs. unified syscall issue, no? |
| 20:17.23 | khem | like microblaze |
| 20:17.47 | khem | blindvt: hmm I forgot .. which patch was it |
| 20:17.55 | khem | in any case |
| 20:18.15 | khem | we need to build and test atleast few arches if this needs to go in |
| 20:19.30 | blindvt | khem, i mean, i remember the discussion and there was indeed a short time when it failed, but at least since 374e9300e8ba8e9b1a01bd35de981b32e95db3c8 it's supposed to be fine for both OABI as well as sane setups :) |
| 20:21.20 | khem | oh that was EABI |
| 20:21.25 | khem | I was talking about oabi |
| 20:21.31 | khem | let me surf the ml |
| 20:21.45 | blindvt | khem, these syscall unification doesn't need to go in 0_9_30, no. In contrast, i'd much prefer to leave such cleanups to master and don't fool around with the stable branch, you're right |
| 20:22.12 | khem | http://www.mail-archive.com/uclibc@uclibc.org/msg04343.html |
| 20:22.15 | khem | this thread |
| 20:22.22 | khem | I dont know if that issue it moot or not |
| 20:22.45 | khem | blindvt: only pull bugfixes |
| 20:22.56 | khem | into release branches IMO |
| 20:23.37 | blindvt | khem, first we switched to pure eabi (which broke oabi) then reverted back to make oabi happy and added that eabi.S which didn't work right as in the error in abovementioned commit but that was due fixed, so both abis should be gold again now |
| 20:25.32 | khem | blindvt: ah I remember now :) rusted |
| 20:46.25 | dileX | khem: gcc-link-libgcc-with-libm.patch needs a refresh |
| 20:47.21 | dileX | http://paste.pocoo.org/show/168557/ |
| 21:04.21 | dileX | gcc-link-libgcc-with-libm-v2.patch: http://pastebin.ca/1762027 |
| 21:09.31 | *** join/#uclibc neli (n=neli@88.159.211.247) [NETSPLIT VICTIM] |
| 21:11.24 | *** join/#uclibc neli (n=neli@88.159.211.247) [NETSPLIT VICTIM] |
| 21:22.32 | *** join/#uclibc trem (n=trem@mol92-1-81-57-136-23.fbx.proxad.net) |
| 21:26.31 | *** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa) |
| 21:42.46 | dileX | khem: so I applied the gcc-4.4.2 patches from openembedded. but got still the same build-crash: http://paste.pocoo.org/show/168592/ |
| 21:43.06 | dileX | khem: gcc-line included in paste |
| 21:44.03 | khem | switch to using OpenEmbedded then it works there |
| 21:47.41 | dileX | more readable gcc-line: http://paste.pocoo.org/show/168594/ |
| 21:56.30 | *** join/#uclibc y_morin (n=ymorin@ARennes-252-1-56-112.w83-195.abo.wanadoo.fr) |
| 23:02.13 | *** join/#uclibc kojack (n=jkolb@113.229.189.72.cfl.res.rr.com) |
| 23:03.45 | *** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe) |
| 23:13.50 | CIA-59 | 03mizvekov 07master * r31416d589ebb 10busybox/coreutils/ (Config.in chown.c): chown: support long options |
| 23:13.52 | CIA-59 | 03vda.linux 07master * r9882b34fe22a 10busybox/coreutils/chown.c: chown: -2 bytes |
| 23:14.29 | *** join/#uclibc likewise (n=likewise@215-65-ftth.onsneteindhoven.nl) |