**** BEGIN LOGGING AT Mon Dec 27 02:59:57 2010 Dec 27 03:32:48 mazilo: cool, thanks Dec 27 03:36:57 cshore_: NP. Dec 27 03:37:18 cshore_: If you are still there and don't mind to help a bit. Dec 27 03:43:22 cshore_: When compiling the mod/languages/mod_managed/freeswitch_managed.cpp file, it fails with this error message "freeswitch_managed.h:43: fatal error: glib.h: No such file or directory". I have glib2 compiled and noticed "staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/include/glib-2.0/glib.h". The question I have is should I add "CXXFLAGS=-I$(STAGGING)/usr/include/glib-2.0" to the "src/mod/languages/mod_managed/Makefile" ( Dec 27 03:43:23 through a patch) or can I add it to the "Build/Compile" section of the OpenWRT Makefile for FS? Dec 27 04:06:48 mazilo: It depends on whether glib2 is one of the many libraries that FS doesn't use the system library but rather a copy it downloads and compiles of it it's own. I think what my be missing actually the dependency (+glib2) for the module. Patching the Makefile won't do what you need because it's generated by the configure script Dec 27 04:07:49 mazilo: btw you don't happen to know what compiler is used for your arch do you? Dec 27 04:31:28 what's the reasoning in having ext4 file systems now? are we using ACL's or labels? Dec 27 04:32:53 philipp64: kaloz claims ext3 and ext2 modules are no longer supported upstream, and apparently ext4 is faster without journalling than ext2 Dec 27 04:34:12 or at least deprecated upstream Dec 27 04:35:26 phlipp64: what version of gcc does your platform use? Dec 27 04:40:12 I'm on x86 hardware, so whatever the default is... I was using 4.4.4, but I think it got bumped to 4.5.1 recently? Dec 27 04:41:12 hmmm....I think my config must have kept the old settings then...brcm63xx probably isn't still on 4.3.3 which is what it's using for my compile Dec 27 06:21:53 cshore_: what's the command to normalize a config-default file? I have the impression that target/linux/x86/config-default is missing CONFIG_EXT4_XSATTRS, etc. Dec 27 06:23:40 you mean make kernel_menuconfig? Dec 27 08:12:37 juhosg * r24839 /trunk/target/linux/gemini/config-default: gemini: enable GPIO sysfs interface Dec 27 09:02:09 hi guys Dec 27 09:02:17 when was rc4 released Dec 27 09:14:21 vot: https://forum.openwrt.org/viewtopic.php?id=27466 Dec 27 09:21:20 back.... Dec 27 09:41:38 where are you guys located in the hackcenter @27c3? Dec 27 09:41:40 hi Dec 27 09:41:45 :P Dec 27 09:41:51 :) Dec 27 09:42:05 so ok I will provide the patch for libgphoto 2.4.10.1 Dec 27 09:42:14 how do you think is the bets way to test properly? Dec 27 09:42:26 using an x86 install in a virtual machine? Dec 27 09:42:41 i guess Dec 27 09:43:39 ok Dec 27 09:43:58 look at dev.openwrt.org Dec 27 09:44:12 AFAIK anyway the only needed change is to bump the libgphoto version in the Makefile... Dec 27 09:45:18 alesan: if the package uses custom patches, these might need to be updated, too. Also a short run test is always good ;) Dec 27 09:45:29 of course Dec 27 09:45:34 yes of course Dec 27 09:45:45 there are a couple of patches Dec 27 09:46:14 that I could ask the upstream maintainer to consider actually Dec 27 09:46:32 the best test would be to see if it runs fine on a big endian and on a little endian target, but this assumes you have a big endian one at hand Dec 27 09:47:26 yes interesting Dec 27 09:47:33 I may want to get one actually Dec 27 09:47:41 but I need one with the USB port Dec 27 09:47:50 I only have a wrt54gl at home :( Dec 27 09:50:55 I see the person that updated the package last time is "nico" in the svn Dec 27 09:50:58 is he here? Dec 27 10:04:40 <[florian]> alesan: {Nico} is his nick Dec 27 10:05:29 oh ok Dec 27 10:06:15 by the way, once I have sent you and anothet guy a board to port openwrt on... did you get it? I really lost contact with this... Dec 27 10:06:31 {Nico}, are you around? Dec 27 10:09:07 if only I could remember Dec 27 10:09:19 the name of the other person I had sent the board... Dec 27 10:09:34 [florian], do you remember anything of this? Dec 27 10:09:42 it was possibly before summer... Dec 27 10:12:26 <[florian]> alesan: yes I think it was KanjiMonster but I am not sure Dec 27 10:17:30 nope, I didn't receive one Dec 27 10:30:46 <[florian]> KanjiMonster: -EBADNICK Dec 27 10:34:39 [florian]: sorry ;) Dec 27 10:40:12 nice, the wn2000rpt stock firmware welcomes me with an openwrt banner ;) Dec 27 13:11:09 build #42 of xburst is complete: Failure [failed shell_9] Build details are at http://tksite.gotdns.org:8010/builders/xburst/builds/42 Dec 27 14:34:12 build #42 of mpc52xx is complete: Failure [failed shell_9] Build details are at http://tksite.gotdns.org:8010/builders/mpc52xx/builds/42 Dec 27 15:10:54 build #43 of etrax is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/43 Dec 27 15:38:49 Is there any plan to port mono (http://ftp.novell.com/pub/mono/sources-stable) to OpenWRT? Dec 27 15:39:26 Here is the mono main page http://www.mono-project.com/Main_Page Dec 27 15:56:58 build #42 of octeon is complete: Failure [failed shell_9] Build details are at http://tksite.gotdns.org:8010/builders/octeon/builds/42 Dec 27 16:15:05 any of you at the hackcenter at 27c3? Dec 27 16:16:36 * larsc is Dec 27 16:17:11 larsc: where are you a-sittin'? Dec 27 16:18:22 * nbd is at the wireless table Dec 27 16:19:12 ok let me head over there right quick Dec 27 16:23:45 now that was a fail. nbd where is the wireless table? :P Dec 27 16:24:05 long table next to the entrance of the hackcenter Dec 27 17:50:26 When compiling FS git, some of the library packages are dynamically downloaded and extracted to their respective FS libs. As such, if I need to make a patch for the library source, OpenWRT will not be able to find the package to patch during the prepare stage. Is there a way to instruct OpenWRT to dynamically patch the source once it has been downloaded and extracted? I know there is a 'Build/Prepare' section, but I just don't know Dec 27 17:50:27 how to let OpenWRT to automatically patch the codes. Dec 27 17:51:44 uh Dec 27 17:55:07 xMff: Can I use the "Build/Prep" section to manually patch the source, i.e. patch -p0 < patches/the_patched_file ? Dec 27 17:55:58 xMff: I guess "Build/Prepare" is executed once the main package is extracted. So, this is not possible. Dec 27 17:58:24 oops **** BEGIN LOGGING AT Mon Dec 27 18:49:46 2010 Dec 27 21:35:21 mazilo: you have to patch FS's build system Dec 27 21:35:41 mazilo: FS is evil that way Dec 27 21:44:22 mazilo: are you it's being downloaded and not just extracted those $(FS)/libs has a bunch of source tarballs Dec 27 21:44:32 *extracted though Dec 27 23:34:43 [florian]: http://homedownloads.cisco.com/downloads/gplcode/wrt54gr_v1.01.tgz Dec 27 23:40:26 <[florian]> sn9: woot, thanks! Dec 27 23:41:04 you really couldn't find that? you even got the model right Dec 27 23:44:17 <[florian]> I was very lazy actually Dec 27 23:46:17 <[florian]> by chance do you remember the instructions to build the bootloader image? Dec 27 23:46:50 there is also a 1.18 download, but redboot is apparently unchanged Dec 27 23:47:20 i would have looked for it and been too lazy to e-mail Dec 27 23:47:52 i think you have to have the toolchain at a specific path Dec 27 23:48:02 <[florian]> do you remember if that one had the 768KB limitation? Dec 27 23:49:15 it was the very same redboot that was on the ar525w, iirc Dec 27 23:50:31 <[florian]> woah that is ugly code Dec 27 23:50:51 yep Dec 27 23:51:10 well, it's a bootloader after all Dec 27 23:52:35 <[florian]> I do not see the 768KB limitation here Dec 27 23:58:06 build #53 of ubicom32 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/53 Dec 27 23:58:38 umm, i may not be reading this right, but according to the way im reading this, certain options are never going to work :/ Dec 27 23:59:53 er Dec 28 00:00:02 inside thr dnsmasq init script Dec 28 00:00:05 *the Dec 28 00:00:41 we have some append_* functions Dec 28 00:01:05 inside those functions its doing append args "blah" Dec 28 00:01:11 with no definition of what "args" is Dec 28 00:02:02 unless my understanding of sh functions is a bit off .. Dec 28 00:03:13 the only reason i looked at it was because even though i've defined nowildcards and notinterface, it still inststs n listening where i dont want it to :/ Dec 28 00:04:46 <[florian]> sn9: if you happen to remember where it was, I take it ;) Dec 28 00:07:48 oh i see Dec 28 00:07:54 build #56 of at91 is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/56 Dec 28 00:08:05 its global unless told otherwise :/ Dec 28 00:23:13 The src/mod/languages/mod/perl/Makefile of FS git has an entry "PERL = `which perl`" that will look for the host perl and causes error in the compilation. I modified the line to become 'PERL="$(STAGING_DIR)/usr/bin/perl""' and that fixes the compilation problem. I wonder if there is a better way to do this. Dec 28 00:24:07 mazilo: probably need to modify the appropriate configure stuff instead of the Makefile directly Dec 28 00:25:26 cshore_: I would agree with you on this; however, the mod_perl package doesn't use autoconf nor automake. That's why I had to manually create the ugly patch. Dec 28 00:27:27 florian * r24840 /trunk/target/linux/brcm63xx/patches-2.6.35/240-spi.patch: Dec 28 00:27:28 [brcm63xx] spi driver: remove bcm_rset usage Dec 28 00:27:28 Since bcm636x platform embeds two spi master device, Dec 28 00:27:28 the attached patch removes static bcm_rset usage, replaced by Dec 28 00:27:28 "bs->regs" field for all I/O operation. Dec 28 00:27:28 [florian]: ecos/ecos-2.0/tools/bin/BenTest_install/include/pkgconf/r8610.h Dec 28 00:27:28 Signed-off-by: Miguel Gaio Dec 28 00:30:03 cshore_:: Here is the original Makfile (http://pastebin.com/p2AeRJVn) and here is the modification (http://pastebin.com/GGmV8QA3). Dec 28 01:40:45 ping [florian] Dec 28 01:41:39 mazilo: if it doesn't use autotools, or equivalent then yeah what you did is what is necessary. Dec 28 02:06:33 cshore_: Thanks. Apparently, it is more involved than that. Looks like I have to disable the mod_perl. :( Dec 28 02:10:44 mazilo: it might not like using perl that's not in the final destination? Dec 28 02:24:38 cshore_: Honestly, I am not sure. But, it looks like using '$(PERL) -MExtUtils::Embed -e ccopts' gives compilation error where PERL="$(STAGING_DIR)/usr/bin/perl" Dec 28 02:26:25 os ExtUtils::Embed built with perl? Dec 28 02:26:28 is Dec 28 02:28:12 cshore: I see $(STAGING_DIR)/usr/libl/perl5/5.10/ExtUtils/Embed.pm file on mine. Dec 28 02:28:51 what's the error? Dec 28 02:29:34 cshore_: If I manually execute 'staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/bin/perl -MExtUtils::Embed -e ccopts', I get this error: Dec 28 02:29:38 Can't locate ExtUtils/Embed.pm in @INC (@INC contains: /usr/local/lib/perl5/5.10.0/x86_64-linux /usr/local/lib/perl5/5.10.0 /usr/local/lib/perl5/site_perl/5.10.0/x86_64-linux /usr/local/lib/perl5/site_perl/5.10.0 .). Dec 28 02:29:40 BEGIN failed--compilation aborted. Dec 28 02:30:17 ok, that's an enviroment problem....it's not using the STAGING_DIR version of perl Dec 28 02:30:51 it's searching the the worth paths Dec 28 02:30:55 *wrong Dec 28 02:31:06 cshore_: If I do it this way 'staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/bin/perl -I/opt/tmp/openwrt-svn-trunk/staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/lib/perl5/* -MExtUtils::Embed -e ccopts', it is ok and the outut is: Dec 28 02:31:31 "/opt/tmp/openwrt-svn-trunk/staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/bin/perl: No match" Dec 28 02:32:40 mazilo: what happens when you compile that way? Dec 28 02:32:58 cshore_: I meant if I do it this way "staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/bin/perl -I/opt/tmp/openwrt-svn-trunk/staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/lib/perl5/5.10 -MExtUtils::Embed -e ccopts", the output is as follows: Dec 28 02:33:00 -Os -pipe -march=armv5te -mtune=marvell-f -funit-at-a-time -fhonour-copts -msoft-float -DUSE_CROSS_COMPILE -I/opt/tmp/openwrt-svn-trunk/staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/include -I/opt/tmp/openwrt-svn-trunk/staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/include -I/opt/tmp/openwrt-svn-trunk/staging_dir/toolchain-arm_v5te_gcc-4.4.1+cs_uClibc-0.9.31_eabi/usr/include -I/opt/tmp/openwrt-svn-trunk/staging_dir/toolchain- Dec 28 02:33:01 arm_v5te_gcc-4.4.1+cs_uClibc-0.9.31_eabi/include -I/CORE Dec 28 02:33:42 so far that looks ok Dec 28 02:34:13 or did your post get cut off? Dec 28 02:34:21 cshore_: Let me try compile it using that simple hack. Dec 28 02:35:53 cshore_: It seems to take it, but another problem crops up where it says it could not find EXTERN.h file which is in "staging_dir/target-arm_v5te_uClibc-0.9.31_eabi/usr/lib/perl5/5.10/CORE/EXTERN.h" Dec 28 02:36:28 mazilo: that's another -I Dec 28 02:36:47 cshore_: So, looks like another patch to change '#include "EXTERN.h"' to '#include "CORE/EXTERN.h"' Dec 28 02:37:03 cshore_: Let me try. Dec 28 02:37:07 or add -I to include the path Dec 28 02:37:36 cshore_: Yup. Dec 28 02:44:34 cshore_: That quick and ugly hack seems to work and the compilation produces the mod_perl now. I will show you the ugly patch in a minute and who knows you will have a better solution. ;) Dec 28 02:46:24 cshore_: Here is the src/mod/languages/mod_perl/Makefile file patch (http://pastebin.com/LyiSP7uB). It looks ugly. Dec 28 02:50:27 well the 'right' way to get perl to include paths is probably some kind of environment or configuration file which would be applied to the OpenWRT Perl Dec 28 02:50:55 then you would only need the which perl replacement Dec 28 02:51:46 cshore_: Correct. But, my knowledge is currently limited and am hoping to learn more from anyone here who can chime in. Dec 28 02:53:37 cshore_: Under the 'Build/Compile' section, I believe the environment for perl can be declared there. Dec 28 02:53:41 mazilo: I think you would need a host-perl in tools which uses the staging dir paths in configure Dec 28 02:54:04 mazilo: AFAICT from my desktop system Perl doesn't use an env Dec 28 02:54:18 cshore_: IC. Dec 28 02:55:03 mazilo: but that may only be true for the default env....it might be possible to use an env to override Dec 28 02:56:20 chsore_: Did you mean the host env to point to the perl in staging dir? Dec 28 02:58:06 mazilo: include paths for modules (e.g. extutils embed), can be set with PERLLIB=/path/to/module/base Dec 28 02:58:44 cshore_: Thanks. I will try this tomorrow and will let you know. Dec 28 02:59:14 so PERLLIB points to the dir containing ExtUtils Dec 28 02:59:30 chsore_: OK. **** ENDING LOGGING AT Tue Dec 28 02:59:56 2010