00:07.38 | psm | JockeHome: the solution is ok, but what you showed me is impossible to compile (missing attribute noreturn) |
00:07.41 | psm | andersee: ping |
00:30.14 | CIA-6 | 03gkajmowi * r10424 10uClibc++/ (19 files in 5 dirs): |
00:30.14 | CIA-6 | Changes to iostream code now allow reading for both first and last character of a stream. What a pain. |
00:30.14 | CIA-6 | Added streambuf test |
00:41.21 | CIA-6 | 03gkajmowi * r10425 10uClibc++/ (TODO include/fstream): Make sure all fstream constructors call proper basic_ios constructor with streambuffer as a parameter |
03:28.08 | *** join/#uclibc bjb_ (~bjb@ns1.linuxbutler.ca) |
06:51.41 | *** join/#uclibc simmo (~simmo@adsl-18-27.swiftdsl.com.au) |
08:48.13 | *** join/#uclibc GyrosGeier (~geier@port-212-202-79-27.dynamic.qsc.de) |
11:18.38 | *** join/#uclibc mallok (~mallok20@85.137.30.130) |
12:13.21 | *** join/#uclibc kergoth (~kergoth@covenant.kergoth.com) |
13:33.29 | *** join/#uclibc thomasez (~thomasez@castle.linpro.no) |
13:50.38 | *** join/#uclibc SpanKYx (~UserBah@141.157.191.33) |
14:24.10 | *** join/#uclibc psm (~Peter@host-6.mikroweb.hu) |
14:24.24 | psm | JockeHome: ping |
15:46.30 | JockeHome | psm: pong |
15:46.57 | psm | hi, your solution, w/ some tweaks works |
15:47.14 | JockeHome | good, what tweaks |
15:47.16 | JockeHome | ? |
15:47.49 | psm | added attribute noreturn to old-compat version |
15:48.29 | JockeHome | I see, I missed that, but it compiled for me with gcc 3.4.3 |
15:48.48 | psm | I have also changed extern int main to weak_function, so I can leave -z defs in |
15:48.58 | JockeHome | :) |
15:49.17 | psm | the newly built apps do not use it |
15:49.26 | JockeHome | now you only have to recompile all apps :) |
15:49.45 | JockeHome | use what, the weak main? |
15:50.26 | psm | the new apps get main through uClibc_main and crt1 |
15:50.35 | JockeHome | yeah |
15:51.31 | psm | well, due to the nature of the changes in crt/fini, I want to propose to use __uClibc_main instead of __uClibc_start_main in the new version |
15:51.40 | JockeHome | BTW, you only have to relink your apps |
15:51.53 | psm | this will allow for those using old stuff to have an upgrade path |
15:52.13 | psm | I have done already the patch |
15:52.39 | psm | for archs that were already transited |
15:52.55 | psm | and a FINI_CRT_COMPAT option |
15:53.28 | JockeHome | hmm, not sure thats a good idea, __uclibc_main was used by the even older i/f |
15:53.33 | psm | due to the fact that everything is broken now, it won't matter what name to use |
15:55.07 | JockeHome | send the path to the list, lets see what other think |
15:55.14 | psm | well, but so we could get back to the old name (as the file is called too) ;) but the name does not really matter, only it should be a new one |
15:55.20 | JockeHome | s/path/patch |
15:56.34 | psm | __uC_main ? |
15:57.01 | JockeHome | yes, with the COMPAT option |
15:57.44 | psm | the default is not to enable the compat interface |
15:57.53 | JockeHome | better be :) |
15:58.11 | JockeHome | may not |
15:58.23 | JockeHome | maybe not I mean |
16:01.07 | JockeHome | BTW, I dont think you need the _DL_FINI_IN_LIBC define in atexit.c |
16:01.37 | JockeHome | it is enough, to have it in ldso |
16:18.39 | psm | JockeHome: remember you had to add -e start to ld, so that mips works? |
16:19.01 | psm | -e _start, sorry |
16:19.20 | JockeHome | yes |
16:19.45 | psm | mips is the only arch, that has __start in crt1, all others use _start |
16:20.00 | JockeHome | BTW, you Do need to keep the _DL_FINI_IN_LIBC define in atexit.c |
16:20.38 | JockeHome | yes, I saw that and sent a question to Erik |
16:21.08 | JockeHome | maybe change to __start in ldso for MIPS? |
16:21.52 | psm | Erik removed .size from crt1 (mips/i386), you could remove it from ppc too |
16:22.50 | JockeHome | wounder why, it is polite to say what size a function has. |
16:23.01 | psm | and probably features.h too |
16:23.16 | psm | ask him ;) |
16:23.20 | psm | bbl |
16:24.20 | JockeHome | Oh, it was me that removed size in i386/crt1.S, thats a mistake. |
16:24.50 | JockeHome | features.h ? whats size is missing in there? |
16:26.58 | JockeHome | The _DL_FINI_IN_LIBC define should be renamed to something else. |
16:27.11 | JockeHome | COMPAT ... |
16:30.13 | CIA-6 | 03jocke * r10426 10uClibc/libc/sysdeps/linux/i386/crt1.S: Add back .size for _start |
16:36.52 | CIA-6 | 03jocke * r10427 10uClibc/libc/sysdeps/linux/mips/crt1.S: Add back .size for __start |
16:38.33 | JockeHome | need to feed, back later |
18:26.03 | *** join/#uclibc GyrosGeier (~geier@honey.kvedulv.de) |
19:20.17 | *** join/#uclibc ambroseL (~bjb@67.71.251.210) |
22:18.28 | *** join/#uclibc CIA-10 (~CIA@flapjack.navi.cx) |
23:22.13 | CIA-10 | 03jocke * r10432 10uClibc/libpthread/Makefile: |
23:22.13 | CIA-10 | Filter out any -z defs since this lib will have undefined symbols that |
23:22.13 | CIA-10 | are in gdb. |
23:36.41 | CIA-10 | 03landley * r10433 10busybox/coreutils/watch.c: Patch from Shaun Jackman to save a few bytes. |
23:44.08 | CIA-10 | 03jocke * r10434 10uClibc/ (11 files in 9 dirs): (log message trimmed) |
23:44.09 | CIA-10 | Add Peter Mazinger fini/crt compat patch. Select DL_FINI_CRT_COMPAT to |
23:44.09 | CIA-10 | be able to run apps built with 0.9.27. This also renames |
23:55.27 | CIA-10 | 03landley * r10435 10busybox/coreutils/date.c: (log message trimmed) |
23:55.27 | CIA-10 | Shaun Jackman submitted a patch converting an allocation to use |
23:55.27 | CIA-10 | CONFIG_RESERVE_BUFFER. (Rob Landley removed an #ifdef, per discussion on |