**** BEGIN LOGGING AT Sat Jul 16 02:59:57 2011 Jul 16 05:50:38 hi Jul 16 05:50:59 I have ported my board to Openwrt Jul 16 05:53:55 1) it seems complete and stable, do you think it can be accepted in the "git"? Jul 16 05:54:27 2) what about the software packages? Jul 16 06:08:12 what about them? Jul 16 06:11:24 what exactly is your question? Jul 16 06:47:39 could all the packages be available to users via a simple webinf install Jul 16 06:48:01 or the available packages are hosted on openwrt serves only for specific architectures? Jul 16 07:33:48 applications are usually specific to a n-tuple of gcc version/c library type and version (e.g. uClibc, eglibc, or glibc)/and linux version. Jul 16 07:35:47 oh, and processor type, of course. Jul 16 07:36:25 you can always enable CONFIG_ALL=y and build *all* of the packages. it takes about 2 days. Jul 16 07:36:46 and then you can use ipkg to install the ones you want afterwards. Jul 16 07:37:31 if you look at target/linux/x86/geos/target.mk you'll see that the default image contains a reasonable number of packages as the default. Jul 16 07:38:07 this being an x86 machine, the assumption is that you have a fair amount of flash (usually 512MB or more)... so enough to put quite a few packages. Jul 16 07:38:30 what's your platform like? processor type, speed, DRAM and Flash capacities? Jul 16 08:44:13 florian * r27622 /trunk/target/linux/brcm63xx/ (7 files in 2 dirs): [brcm63xx] add ehci support to DWV-S0 board (#9059) Jul 16 09:25:36 gcc 4.4.5 without linaro patches is not comming back ? Jul 16 09:25:57 https://dev.openwrt.org/changeset/27333/trunk/toolchain/gcc Jul 16 09:26:43 was using 4.5.2 without patches for my glibc 2.6.1 builds :S Jul 16 09:27:56 like i said before there are a lot changes in 2 weeks on just gcc.. Jul 16 09:33:07 at 27155 i was still able producing a glibc based build.. for damned collectd with failing rrdtool Jul 16 09:33:27 now nothing is working any more.. only untested stuff that is left. Jul 16 09:34:16 +15 changes at the toolchain in 2 weeks and bye bye glibc support.. or a new bug has been created. *THANKS* Jul 16 09:34:34 why not remove whole x86 maby i just go then . Jul 16 09:35:03 got any details about your bug? Jul 16 09:35:32 rofl did nbody test his own changes ? Jul 16 09:35:46 since nobody test glibc 2.6.1 or 2.7.x Jul 16 09:36:13 i did report a working situation.. and 2 weeks ago half the toolchain has been removed. Jul 16 09:36:54 in the mean time spending/whasting time on uclibc and nodebug options asking for how to continue can get threaded like some troll bye a few trolls :s Jul 16 09:37:16 can repaste my stuff from yesterday since nobody is going to even look at glibc based openwrt builds. Jul 16 09:37:25 all i want is damned rrdtool working Jul 16 09:37:45 tried eglibc? Jul 16 09:37:53 no Jul 16 09:37:55 that's probably a better choice than glibc anyway Jul 16 09:37:58 glibc or uclibc.. Jul 16 09:38:11 somehow debugging on uclibc is not posible. Jul 16 09:38:18 because it's compatible to glibc, but the people that maintain it actually care about cross compiling, and support for non-x86 stuff Jul 16 09:38:21 and on glibc based builds it just worked. Jul 16 09:38:24 unlike the glibc maintainer Jul 16 09:38:48 i tried an eglibc x86 build recently and it worked Jul 16 09:39:03 i run on plain i686.. so gues wat i think about linaro patches on a gcc package .. Jul 16 09:39:13 :S Jul 16 09:39:24 i was using the linaro toolchain as well Jul 16 09:39:25 can try whasting an other weekend.. Jul 16 09:39:39 i don't expect the linaro stuff to cause issues Jul 16 09:39:39 else must make some decisions monday Jul 16 09:39:51 glibc is probably the least maintained libc in openwrt Jul 16 09:39:58 eglibc and uclibc are used more Jul 16 09:40:05 nbd: i dunno linaro first google results tell me i will never need it with 4G flash storage Jul 16 09:40:22 on x86 Jul 16 09:40:23 that makes no sense Jul 16 09:41:05 anyway yesterday i had gdb+rrdtool in a setup using uclibc(defaults) Jul 16 09:41:17 the main reason why we kick out older toolchains is that we want to reduce the number of toolchains that we have to maintain Jul 16 09:41:23 so we have to test fewer combinations Jul 16 09:41:42 somehow gdb keeps wining about pthreads and someone told me here thats imposible on uclib so i asked how to continue there.. somehow thats a hard question Jul 16 09:41:58 nbd: and remove working situations .. Jul 16 09:42:39 gcc 4.6.1 will work with uclibc but glibc is broken afther 27155 Jul 16 09:42:43 i don't think thread debugging is impossible on uclibc Jul 16 09:42:52 so try eglibc then Jul 16 09:42:56 and stop whining Jul 16 09:43:14 i enabled pthread_debug indeed. had the modules installed but my pthread libs segfault. Jul 16 09:43:33 and the people yesterday keep saying not posible pthread on uclib. Jul 16 09:44:14 can give it a try but i expect nothing at all and wat if problems.. Jul 16 09:44:37 i'm a glibc user and already hate uclib due you guys. imagine eglibc and issues.. Jul 16 09:45:08 nbd: and eglibs is the fix for rrdtool segfaulting with no info ? Jul 16 09:45:23 they way you tell/push it now Jul 16 09:46:53 it might not work out of the box, you might need to enable a config option and copy the libthread_db library or something Jul 16 09:46:56 but i don't think it's impossible Jul 16 09:47:06 ehh wait Jul 16 09:47:09 i know that issue Jul 16 09:47:17 there is no thread_db we have pthread Jul 16 09:47:41 ? Jul 16 09:47:47 and then wat ? since thats where i stopped yesterday more or less.. Jul 16 09:48:03 have phtreads installed and gdb wining about thread_db .. Jul 16 09:48:03 libthread_db is a debugging support library used by gdb to figure out thread state information Jul 16 09:48:20 only gdb needs it, so by default it is not installed Jul 16 09:48:39 not even availible afther 2 days messing around Jul 16 09:49:04 thats my whole problem.. somehow yesterday someone refused to read that part of my line and keep saying not understand like a dummy Jul 16 09:49:24 while i told him gdb no working Jul 16 09:50:00 i can not even think about cacti or mrtg since that damn rrdtool is broken. Jul 16 09:50:20 without tools dunno why that happend right . Jul 16 09:51:55 nbd: don't get me wrong but when your working on a package and the damn toolchain get broken by someone else.. Jul 16 09:52:07 like gcc sources with a new md5 hasj Jul 16 09:52:28 don't get me wrong, but i'm not interested in whining Jul 16 09:52:46 no just waiting till other people fix without needing help right ? Jul 16 09:53:05 if people not ignore and try to help then it will me less a mobnologe. Jul 16 09:53:22 thats the whole problem when i'm here this damn channel get silent Jul 16 09:53:28 if you start testing the suggestions i gave you, or give me some usable details, then i'll continue to look into the issues Jul 16 09:53:34 if not, then i won't care Jul 16 09:53:40 thats how to thank someone for spending his privater time on your project right ? Jul 16 09:53:43 simple as that Jul 16 09:53:56 the damn eglib is already compiling yes Jul 16 09:54:00 ok Jul 16 09:54:18 but i don't like compile the whole thing time over time without any results Jul 16 09:54:36 i'm just sitting on a train home after a >24 hour trip involving planes busses and trains Jul 16 09:54:39 specialy when i'm not able making a backup .config working anymore Jul 16 09:54:41 so i have low tolerance for whining at the moment ;) Jul 16 09:55:01 nbd: i have nothing better todo then sit here and fix the damn problem.. Jul 16 09:55:05 else i have nothing todo Jul 16 09:55:09 get that ? Jul 16 09:55:32 since a week or so Jul 16 09:55:34 ping me again when you have something interesteding for me Jul 16 09:56:06 gcc 4.6.1 with eglibc 2.13.. will wait and see. Jul 16 09:56:32 needed cc 4.4 for rrdtool but nevermind that Jul 16 09:56:41 cc=gcc Jul 16 09:57:13 linaro gcc probably would have been a better choice, but ok Jul 16 09:57:27 generally, the fewer things you change the more likely it is to work Jul 16 09:57:49 and linaro gcc has been tested on the biggest number of targets at the moment Jul 16 09:57:54 heh only on precompiled backfire and glibc based trunk builds i have seen rrdtool working. Jul 16 09:58:39 since pthreading librarys also segfault yesterday and rrdtool depend on it as option Jul 16 09:59:03 but bbl in +- 15 min i guess. Jul 16 10:26:24 nbd: will retry with gcc linaro.. -> http://pastebin.com/TXh4UL7b Jul 16 10:27:53 looks like this http://www.eglibc.org/archives/patches/msg00966.html Jul 16 10:28:49 oh, that option has a wrong dependency in the code Jul 16 10:29:07 hmm easy fix by switch eglibc a version up or down ? Jul 16 10:29:37 no, but i'll quickly go through the options to remove all bogus dependencies Jul 16 10:29:53 maybe i can fix it before the train arrives in berlin Jul 16 10:30:10 berlin sounds like 500km away from here :) Jul 16 10:31:04 hmm berlin is a bit further driving them hamburg from here .. Jul 16 10:31:14 well, yesterday i was in denpasar ;) Jul 16 10:31:27 ehh denpasar.. never heard of it :p Jul 16 10:31:45 indonesia :D Jul 16 10:32:09 yep Jul 16 10:32:41 not that i have seen mutch from the world anyway.. just europe.. Jul 16 10:32:55 nbd * r27623 /trunk/toolchain/eglibc/config/Config.in: eglibc: remove obsolete dependencies to fix missing config options for eglibc 2.13 Jul 16 10:33:12 svn up :) Jul 16 10:33:18 and make oldconfig Jul 16 10:33:22 or make menuconfig Jul 16 10:33:27 that's important after this change Jul 16 10:33:28 hmm nah i'll redownload a new tree.. Jul 16 10:33:31 or that Jul 16 10:33:56 and redo from scratch.. in case one of my vim edits are still there. Jul 16 10:34:30 in case you're wondering what state your .config is in, ./scripts/diffconfig.sh should give you a nice overview Jul 16 10:34:46 it shows you everything that's different compared to the defaults Jul 16 10:35:54 hmm ahh also when i change the kernel config settings or in the toolchain enable a feature ? Jul 16 10:36:02 since those changes are not handles via menuconfig anyway Jul 16 10:36:09 handles/handleD Jul 16 10:36:13 it only shows menuconfig based stuff Jul 16 10:36:28 ahh ok wise choice redownload in a seperate directory :) Jul 16 10:36:32 but kernel config changes show up in svn diff Jul 16 10:36:42 and as for the toolchain, depends on where you made those Jul 16 10:36:56 if you touched files under version control, then svn diff will show that stuff as well Jul 16 10:38:50 ahh and poof there is a complete eglibc menu :) Jul 16 10:39:28 maybe it won't show all options for 2.13 yet, but at least it will show those that were present in 2.12 already Jul 16 10:39:42 the rest will be fixed later Jul 16 10:40:07 hmm i never used eglibc but if it works better/easyer then uclibc i'll like it Jul 16 10:40:24 it aims to be binary-compatible with glibc Jul 16 10:40:28 or atleast on gentoo uclibc builds never workedout for me.. Jul 16 10:40:40 gentoo is different Jul 16 10:40:49 we have lots of uclibc related fixes that are not in gentoo Jul 16 10:40:50 jup i know it is a lot diffrent Jul 16 10:41:10 also way how the crew fix issues :) Jul 16 10:41:22 does gentoo even have uclibc 0.9.32 yet? Jul 16 10:41:34 hmm can look Jul 16 10:41:35 because before 0.9.32, threading was way too crappy Jul 16 10:41:45 didn't even support tls Jul 16 10:42:40 i will do some more testing with uclibc threading soon, because i will resume my work on porting android stuff to openwrt Jul 16 10:42:53 and that makes extensive use of threading Jul 16 10:43:17 hmm sort of :) Jul 16 10:43:18 https://bugs.gentoo.org/show_bug.cgi?id=352597 Jul 16 10:43:58 nbd: hmm i did think about rolling out premade images with a kind of flavour.. example nasbox/routerbox Jul 16 10:45:29 also think about running apache on openwrt and use luci als management not like production lamp but for a hobby home httpd with php not shure if that will run on uclibc i do not have to think about that. Jul 16 10:45:45 why apache? Jul 16 10:45:50 chereoke or a other httpd posible will do easyer then apache Jul 16 10:46:05 hmm i know apache without having to think about stuff that was +5 years ago Jul 16 10:46:06 nginx or lighttpd is probably much better than apache Jul 16 10:46:11 haven't tried cherokee Jul 16 10:46:33 my server still runs apache and it annoys me Jul 16 10:46:38 too unstable Jul 16 10:46:45 sometimes i simply have to restart the server for no reason Jul 16 10:46:53 hmm it is oversized and the mpm workers in apache always annoyed me :( Jul 16 10:47:17 and i'm not even using crappy modules for php Jul 16 10:47:30 nbd: hmm nah never had that i moved away from iis for those service reboot issues while edit php code and keep receive cached code Jul 16 10:47:32 all php stuff is isolated in extremely minimal fastcgi chroots Jul 16 10:47:54 i have a script to autogenerate a chroot whenever i add a new site Jul 16 10:48:32 bbl, train arriving soon Jul 16 10:48:41 i go make some breakfast here.. Jul 16 10:48:47 ehh lunch due the time Jul 16 10:49:34 rebuilding with eglibc and linaro gcc 4.5.x will ping when it finish/fail Jul 16 12:02:41 nbd: thanks the toolchain continue's afther having gcc and eglibc build :) now i can focus on my issue ;-) Jul 16 12:14:45 nbd: finish no errors -> make[1]: Leaving directory `/mnt/backup1/willie/openwrt/r27623/trunk' Jul 16 12:14:51 thanks ! Jul 16 17:48:15 nbd, speaking of gcc, i havent been able to download it cause all the mirrors are broken and the file names are wrong Jul 16 17:49:56 RealOpty: which target/gcc version? also trunk or backfire? Jul 16 17:50:34 trunk - x86 - 4.6.something i think, its a package Jul 16 17:51:08 ah, you mean the gcc package, not the toolchain cross compiler? Jul 16 17:51:19 i was gonna make a wiki page on how to make a build environment inside of a openwrt virtualbox image lol Jul 16 17:51:23 yes package :) Jul 16 17:52:14 gcc as a package is x86 only ;) Jul 16 17:52:56 i know? Jul 16 17:53:14 was intending to use it in virtualbox on x86 target Jul 16 17:57:16 I meant if you said at first that you are talking about the gcc package, then you the target would be obvious ;) Jul 16 17:58:17 hm, I see the problem Jul 16 17:58:50 o Jul 16 18:02:38 the gcc package is broken since 5 month ago -> https://dev.openwrt.org/changeset/25515/ Jul 16 18:03:15 (see the Config.version change) Jul 16 18:03:44 Kaloz: ^ Jul 16 18:05:38 ahhhh i was wondering why it just said lanaro Jul 16 18:07:29 but the default mirrors on the package are broken too, only miror2.openwrt works Jul 16 18:07:49 someone please commit 1206, which fixes a bug. Jul 16 18:07:57 s/works/has the file Jul 16 18:08:32 RealOpty: interestingly, googling for the md5sum suggest its actually for make 3.81 ;) Jul 16 18:08:47 (the one in the gcc package makefile) Jul 16 18:17:13 RealOpty: you could try hardcoding the PKG_VERSION Jul 16 18:18:05 KanjiMonster, i have, just creates more issues lol Jul 16 18:54:27 blogic: ping Jul 16 19:04:34 hi Jul 16 19:06:01 blogic: i'm having trouble with button and gpio you have found Jul 16 19:06:23 i'm using this line: '{ .desc = "reset", .type = EV_KEY, .code = BTN_0, .threshold = 3, .gpio = 14, .active_low = 1, },' for gpio Jul 16 19:06:54 but in dmesg i get this error 'input: input_set_capability: unknown type 168 (code 33)' Jul 16 19:08:29 ok Jul 16 19:08:30 you have used in uboot: 'CONFIG_BUTTON_PIN 14'; isn't that gpio pin? Jul 16 19:09:00 what is CONFIG_BUTTON_PORT set to ? Jul 16 19:09:19 'CONFIG_BUTTON_PORT0 1' Jul 16 19:09:46 in that case it is 14 Jul 16 19:09:59 when do you get "nput: input_set_capability: unknown" Jul 16 19:10:01 during boot ? Jul 16 19:10:05 yes Jul 16 19:10:12 hmm Jul 16 19:10:26 kernel files responsible for this are drivers/input/input.c and include/linux/input.h Jul 16 19:10:30 ok Jul 16 19:10:35 try to figure out why :) Jul 16 19:10:45 i am :) Jul 16 19:12:18 what is ebu used for? Jul 16 19:12:43 you are not using a ebu Jul 16 19:12:53 well the board does not use it Jul 16 19:12:59 its a 16bit prallel bus Jul 16 19:13:08 like a intel bus Jul 16 19:13:41 im not using ebu, i thought maybe that is the error... Jul 16 19:14:13 nope Jul 16 19:14:38 i have selected im 'make menuconfig' kmod-input-gpio-buttons Jul 16 19:14:54 should i use something else? Jul 16 19:15:37 never used hthe gpio stuff, sorry Jul 16 19:16:41 ok, but the buttons defined in mach-arv45xx.c are working right? i used that as a reference Jul 16 19:17:04 never tested it Jul 16 19:17:12 it was contributed by someone Jul 16 19:17:16 look at the ar71xx Jul 16 19:17:24 i think it was copied from there Jul 16 19:17:53 ok, thx Jul 16 19:19:16 i'll send patches tonight (with or without gpio) so it would be great if you could commit them before your vacation Jul 16 19:20:37 sure Jul 16 19:39:10 blogic: if/when need to create a patch for one of the utils packages wat will be the best way creating that patch ? Jul 16 19:40:03 regulair methods result in build scripts look like being in a other path then where i'm adding changes. Jul 16 19:41:44 Simply diff file.org file.bak > file.patch works for me but the generated output is useless . Jul 16 19:42:53 patch result: can't find file to patch at input line 1 Jul 16 19:44:54 while svn diff not is working in the build_dir or i do something wrong.. if someone may can point me at where i go wrong :( Jul 16 19:45:47 nbd: ping*. Jul 16 19:46:04 WillieNL: here is described patching process: http://wiki.openwrt.org/doc/devel/patches Jul 16 19:46:45 that is how you manage patches in build_dir Jul 16 19:47:58 luka12345|wiik: ahh ok thanks. Jul 16 19:48:55 np Jul 16 19:50:07 installing quilt.. make coffee and read the whole article before doing anything.. mosthly easyer then read while do and wonder why. Jul 16 19:50:18 bbl Jul 16 19:55:24 i mainly use quilt (pop|push|diff|refresh) for patching; you should also try zsh shell it will make your life easier Jul 16 19:56:46 WillieNL: use quilt Jul 16 19:56:58 make package/foo/{clean,compile,install} V=99 QUILT=! Jul 16 19:56:59 make package/foo/{clean,compile,install} V=99 QUILT=1 Jul 16 19:57:12 cd build_dir/target...../foo/ Jul 16 19:57:25 quilt new/add ... Jul 16 19:57:26 cd - Jul 16 19:57:32 make package/foo/update Jul 16 19:57:42 the new patch is then copied to the package/foo/patches folder Jul 16 20:00:26 hmm jup first remove the old failed patch attempt.. and coffee before i'm forgetting that.. moment Jul 16 20:01:30 also ehm rrdtool 1.3.8 depends on cairo and i try to include svg support but no sure if i do have all cairo deps right jet :) Jul 16 20:01:52 but brb Jul 16 20:16:15 maddes * r27624 /packages/net/sslh/ (Makefile files/sslh.config): [package] sslh move to submenu Routing and Redirection, mark config files Jul 16 20:20:19 blogic and luka12345|wiik both thanks for the info. Jul 16 20:45:33 blogic: got the gpio button to work :) Jul 16 20:45:39 nice Jul 16 20:45:51 my wifi is still not working Jul 16 20:45:51 i used ar71xx for reference like you said Jul 16 20:45:56 but i think i know why now Jul 16 20:46:06 i have a w303V typA here Jul 16 20:46:11 and all works but wifi Jul 16 20:46:23 what madwifi or ath5k? Jul 16 20:47:11 i googled, its a new target :) Jul 16 20:48:04 rt2800 Jul 16 20:57:19 maddes * r27625 /packages/net/sslh/Makefile: [package] sslh: add maintainer information Jul 16 23:09:08 suppose libmxl2 not to be depending on libpthread-stubs instaid of libpthread ? Jul 16 23:09:30 libxml2* Jul 16 23:11:16 (join #freifunk Jul 17 01:56:55 * luka12345|wiik has done enough patching for today... **** ENDING LOGGING AT Sun Jul 17 02:59:57 2011