**** BEGIN LOGGING AT Wed Jul 20 02:59:57 2011 Jul 20 05:10:30 wtf is cerowrt? Jul 20 06:56:00 luka12345|wiik: yes but owsip has a bug right now and wont work Jul 20 07:35:05 build #51 of ppc44x is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/51 Jul 20 08:52:57 nbd * r27699 /trunk/package/comgt/ (3 files in 2 dirs): comgt: clean up patch, rename gcom binary to comgt, but add a compatibility symlink Jul 20 10:57:09 I want to add a countdown to the ledtrig-timer.c, to stop blinking ..., any idea? Jul 20 11:14:32 build #51 of uml is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/51 Jul 20 11:14:56 build #52 of x86 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/52 Jul 20 11:58:36 <^Willie^> hmm i belive that i must have seen almosth every problem around the toolchain by now :S Jul 20 12:02:54 <^Willie^> hmm is there some easyway rebuilding only terminfo for ncurses ? Jul 20 12:03:07 <^Willie^> package/terminfo is not ok Jul 20 12:04:06 <^Willie^> at least every solution for working around problems result in new problems like right now Jul 20 12:04:28 <^Willie^> with cross builds i do not have any dep issues with ncurses.. so wonder why that do happend native :s Jul 20 12:04:47 <^Willie^> cross builds failed at 86_64 as hostscc where we expect i486 :S Jul 20 12:09:05 <^Willie^> anyway i giveup Jul 20 12:09:59 nbd * r27700 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ag71xx_main.c: ar71xx: fix an unused variable warning Jul 20 12:10:04 nbd * r27701 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ag71xx_main.c: ar71xx: increase the delay after the ethernet MAC reset Jul 20 12:10:07 nbd * r27702 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ag71xx_main.c: ar71xx: make sure that rx and interrupts are disabled before issuing the hardware reset Jul 20 12:10:11 nbd * r27703 /trunk/target/linux/ar71xx/files/ (arch/mips/ar71xx/devices.c drivers/net/ag71xx/ag71xx_main.c): ar71xx: reset the phy in the ethernet init on ar724x Jul 20 12:10:39 <^Willie^> eglibc and rrdtool -> libpthread segfault.. uclibc builds just rrdtool segfault. glibc builds don't work anymore since all gcc releases before 4.5 without patches.. Jul 20 12:11:04 <^Willie^> and no threading is not even acceptible for a x86 i686 setup.. Jul 20 12:12:26 <^Willie^> also for the wiki pages.. try adding a page for x86 users that they better can use a external toolchain and stop crosscompiling i486 on i686 machines for no reason. Jul 20 12:13:55 <^Willie^> also there is no i386 support for threading in glibc anymore sonce 2006 or so.. Jul 20 12:14:15 <^Willie^> sonce/since Jul 20 12:15:27 <^Willie^> and why is collectd configured to be compiled threaded and rrdtool half yes half no :S that will segfault indeed. Jul 20 12:39:22 ^Willie^: you can try to compile only specific package like this: 'make package/name_of_package/compile V=99' Jul 20 12:39:48 or combine that with clean like this: 'make package/name_of_package/{clean,compile} V=99' Jul 20 12:43:11 can anybody explain what are used for *pc files in staging_dir/target-xxx/usr/lib/pkgconfig/ Jul 20 12:44:08 luka12345|wiik: gnu autofail gets cflags and ldflags for libraries from there Jul 20 12:45:26 autofail? Jul 20 12:45:52 the retarded automake / autoconf / configure / libtool family Jul 20 12:46:36 <^Willie^> luka12345|wiik: best is that i go for a walk right now.. i do need to cool down a bit.. Jul 20 12:48:03 jow_laptop: i guess than that every package with InstallDev should place *pc files in staging_dir/target-xxx/usr/lib/pkgconfig/, or not? Jul 20 12:48:30 <^Willie^> luka12345|wiik: somehow package/terminfo is not working but ncurses include it and later complain about libtic must rebuild terminfo with tic -x from there no clue where to look and failed with some patches in the wrong directory so i can start from scratch again anyway.. Jul 20 12:49:16 luka12345|wiik: in theory yes, in practise of course those files are often tainted with host crap so you have to sanitize them first Jul 20 12:49:47 otherwise configure scripts in unrelated packages will then pick up some /usr/include and one starts to wonder where it comes from Jul 20 12:50:51 <^Willie^> jow_laptop: now i see you.. have you seen any bug posts about glibc 2.7 that include nptl/sysdeps/x86_64 instaid of i386 ? Jul 20 12:51:32 dunno, I do not follow glibc Jul 20 12:51:59 if I want glibc I use debian Jul 20 12:52:00 <^Willie^> result in this bug http://gcc.gnu.org/ml/gcc-bugs/2005-08/msg00500.html Jul 20 12:52:14 <^Willie^> jow_laptop: hmm is debwrt a option or totally diffrent project ? Jul 20 12:52:37 debwrt is just a debian with an openwrt kernel Jul 20 12:52:41 <^Willie^> and yes i do build on a amd64 but in vbox as i686 host so dunno why it include that part Jul 20 12:53:02 <^Willie^> jow_laptop: hmm all i want is luci web panel on don't care wat distro as long it is glibc based Jul 20 12:53:22 luci will not work on anything besides openwrt userland Jul 20 12:53:51 <^Willie^> have seen luci dev tree can make a testrun, but did not expect lua stuff to work on the gentoo box afther noticing some missing info Jul 20 12:54:19 it works but it will not do anything useful Jul 20 12:54:19 jow_laptop: is this good enough .pc file or it still needs some love: http://pastebin.com/6Gh2mkWT Jul 20 12:54:36 <^Willie^> also will try avoiding cross compilations for my case since that is the last thing i want/need here for i686 targets :) Jul 20 12:55:51 luka12345|wiik: no Jul 20 12:56:08 luka12345|wiik: it will result in -I/usr/include, -L/usr/lib Jul 20 12:56:15 exactly what I meant Jul 20 12:56:27 from my build machine? Jul 20 12:56:31 yes Jul 20 12:56:53 just sed out the -I and -L flags Jul 20 12:57:03 jow_laptop: thx, i'll fix it Jul 20 12:57:36 something like $(SED) 's/-[IL][^[:space:]]* // Jul 20 12:57:55 something like $(SED) 's/-[IL][^[:space:]]* //' $(PKG_BUILD_DIR)/foo/bar.pc Jul 20 12:58:11 prior to installing it Jul 20 12:58:30 this one if from lua: http://pastebin.com/j8ae7AYt Jul 20 12:59:09 it should look like something like that, i have grep'ed packages directory and there are sed examples Jul 20 13:00:30 jow_laptop: thx Jul 20 13:00:38 luka12345|wiik: ok I have to revise it a bit Jul 20 13:01:00 as long as flags are given in the for ${prefix}/foo or ${exec_prefix}/foo its fine Jul 20 13:01:09 because those prefix variables are overriden when calling pkg-config Jul 20 13:01:37 its important that there is nothing literally in there, like -L/usr/lib instead of -L${prefix}/lib Jul 20 13:02:10 so -L${prefix}/lib is ok? Jul 20 13:03:36 yes, you can test it: Jul 20 13:03:37 PATH=./staging_dir/host/bin:$PATH STAGING_PREFIX=/test PKG_CONFIG_PATH=$(echo staging_dir/target-*/usr/lib/pkgconfig) ./staging_dir/host/bin/pkg-config --cflags lua Jul 20 13:03:49 should result in e.g. /test/include Jul 20 13:04:44 <^Willie^> jow_laptop: for glibc if anybody ever wanna test it losts of fixes are already here -> https://dev.openwrt.org/ticket/9483 Jul 20 13:05:42 jow_laptop: thx, worked like a charm Jul 20 13:06:09 i'll send patch tonight for libosip & siproxd Jul 20 13:11:15 <^Willie^> jow_laptop: one question about building the trunk when using a external toolchain. when is that tested for the last time ? Jul 20 13:12:01 I tested it three months ago for ar7 Jul 20 13:12:08 <^Willie^> hmm ok Jul 20 13:12:25 <^Willie^> will give it an other try tonight from the debian host here.. Jul 20 13:13:18 from observing your irc messages here it looks like you erratically hop back and forth between different trees, host systems, configs, setups patches and goals Jul 20 13:13:53 <^Willie^> yes i started glibc moved uclibc then eglibc.. now back glibc due the pthread segfaults i have seen on the eglibc builds. Jul 20 13:13:59 did you ever simply rebuilt collectd and rrdtool with -ggdb3 and used cross-gdb to pinpoint the segfault? Jul 20 13:14:07 <^Willie^> and moved from i486 cross build to native i686 yes Jul 20 13:14:13 oh well Jul 20 13:14:25 <^Willie^> hope nothing is hard patched.. Jul 20 13:14:27 lets continue this if you settled on one system Jul 20 13:14:37 <^Willie^> yes thats wat i want.. Jul 20 13:15:12 <^Willie^> and have run gdb a few times on collectd indeed libm was missing fixed that and run into the libpthread segfaults :( Jul 20 13:15:45 then the next step is to debug uclibc Jul 20 13:15:58 its a bit more complex but doable Jul 20 13:16:24 oh btw, openwrt for x86 is crosscompiled because it is intended to run on soekris hardware Jul 20 13:16:30 <^Willie^> hmm ok but before i do that can uclibc give me the same threading performance as generic glibc based debian builds ? Jul 20 13:16:34 which cannot host toolchains Jul 20 13:16:43 and also only has i486 level x86 compatibility Jul 20 13:16:57 <^Willie^> hmm Jul 20 13:19:48 <^Willie^> before the changes from 3 weeks ago i was able building a working glibc based build but the md5 hash of gcc 4.5 with linaro patches was updated resulting in not able to download it afther a distclean/dirclean and since then not even able building a working image due more issues then i ever have seen in that week also part due my own actions by hopping from version to version.. Jul 20 13:20:11 <^Willie^> it is pretty frustrating. Jul 20 13:20:56 yes Jul 20 13:21:56 <^Willie^> i basicly started with the toolchain then collectd but the toolchain did not rebuild anymore since then :) Jul 20 13:22:23 <^Willie^> sounds weird but then i do hop back to 0 and wanna start from scratch avoiding jumping over hidden issues that popup next build that can be weeks later Jul 20 13:22:50 hint: if some issue persists over a distclean but nobody can reproduce it its usually due to a bad .config Jul 20 13:23:20 <^Willie^> hmm i did go a bit further.. rm -rf trunk ; svn co Jul 20 13:23:44 <^Willie^> created patches from wat i had done on luci-collectd and rrdtool update. Jul 20 13:23:58 <^Willie^> so do have those for speeding up next time Jul 20 13:24:53 <^Willie^> right now i must first get the toolchain building without "new" errors :) Jul 20 13:25:10 <^Willie^> anyway bbl must go to the supermarked before they are closed again. Jul 20 13:53:09 build #54 of rb532 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/54 Jul 20 14:45:22 nbd * r27704 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ag71xx_ar7240.c: ar71xx: configure address aging on ar7240 Jul 20 14:45:26 nbd * r27705 /trunk/target/linux/ar71xx/files/drivers/net/ag71xx/ag71xx_ar7240.c: ar71xx: reinitialize global switch settings after reset on ar7240 Jul 20 14:52:16 nbd * r27706 /branches/backfire/target/linux/ar71xx/files/ (3 files in 2 dirs): ar71xx: merge ar7240 fixes from trunk Jul 20 14:52:36 <_lore_> ping nbd Jul 20 14:54:28 _lore_: pong Jul 20 14:54:40 <_lore_> hi Jul 20 14:54:43 <_lore_> how are you? Jul 20 14:55:12 good, how are you? Jul 20 14:55:28 <_lore_> fine tnx Jul 20 14:55:57 <_lore_> I have a question... Jul 20 14:56:33 <_lore_> I am wondering how ath9k queues work.. Jul 20 14:56:41 <_lore_> I mean Jul 20 14:57:03 jow * r27707 /packages/libs/cyrus-sasl/Makefile: [packages] cyrus-sasl: autoreconf sasldb subdir as well Jul 20 14:57:46 <_lore_> is there a reserved bandwidth also for low priority queue (BE- BK) if I have traffic in VO queue? Jul 20 14:58:21 each queue has a limit of how many buffers are allowed to be pending Jul 20 14:58:32 the rest is handled by hardware queueing on the chip Jul 20 14:58:41 <_lore_> ok Jul 20 15:00:57 <_lore_> I wan to to be sure that I a particular frame is the last trasmitted.. Jul 20 15:01:03 <_lore_> *want Jul 20 15:02:44 <_lore_> this frame is used to syncronize the channel access and it has to be trasmitted at the end .. Jul 20 15:04:39 jow * r27708 /packages/net/freeradius2/Makefile: [packages] freeradius2: conditionally depend on libopenldap Jul 20 15:06:04 <_lore_> if I have 100 frame in VO queue and just one in VI...does the HW tx all VO frames and at the end the VI one? or is there a reseved bandwidth also for VI traffic? Jul 20 15:09:42 depends on the queue settings Jul 20 15:09:51 not sure exactly Jul 20 15:10:04 <_lore_> ok Jul 20 15:10:21 i don't think you can count on the VI frame being sent last Jul 20 15:11:03 hi Jul 20 15:11:03 <_lore_> ok Jul 20 15:11:37 in my board's default kernel config, do I *need* to disable for example I2C and USB to have OpenWRT properly take care of the kmod-usb and kmid-i2c packages? Jul 20 15:12:53 <_lore_> to be sure VI traffic is sent last I have to wait for the VO queue to be empty Jul 20 15:13:06 <_lore_> and then tx this management frame Jul 20 15:14:02 <_lore_> so I think I can stop the mac80211 local_pending_queue and the schedule a tasklet to check ath9k queue are empty Jul 20 15:14:16 <_lore_> and then tx the token frame Jul 20 15:14:28 <_lore_> is this resonable? Jul 20 15:16:42 not sure, depends on the structure of your code i guess Jul 20 15:16:59 somehow you probably need to limit the number of frames that you put into your queue anyway Jul 20 15:17:17 <_lore_> already done Jul 20 15:17:52 so when you've hit the limit, you can just throw your mgmt frame at the end of the queue Jul 20 15:17:58 maybe set the last frame for bursting Jul 20 15:18:06 or something like that Jul 20 15:18:27 when you've not reached the quota, you could watch for a TXEOL interrupt or something like that Jul 20 15:18:43 and use that as a trigger to send the mgmt frame Jul 20 15:18:53 loswillios * r27709 /packages/sound/mpdas/ (Makefile patches/001-flags.patch): [packages] mpdas: remove /usr/local PREFIX Jul 20 15:18:55 loswillios * r27710 /packages/sound/mpdas/Makefile: [packages] mpdas: update description Jul 20 15:19:00 though you have to watch out for race conditions there Jul 20 15:20:41 <_lore_> TXEOL means the hw queue is empty, tight? Jul 20 15:20:45 <_lore_> right? Jul 20 15:20:46 i think so Jul 20 15:21:54 bbl Jul 20 15:22:06 <_lore_> ok Jul 20 15:22:12 <_lore_> tnx Jul 20 15:34:49 jow * r27711 /trunk/package/firewall/ (Makefile files/lib/core_interface.sh): [package] firewall: fix serious bug in state var handling (#9746) Jul 20 15:39:18 isn't the topic here a bit out of date? :) Jul 20 15:55:56 loswillios: I believe you can fix 27709 by just doing MAKE_VARS += PREFIX="/usr" Jul 20 15:56:20 loswillios: or rather MAKE_VARS += "CONFIG=/etc" Jul 20 16:14:36 build #48 of avr32 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/48 Jul 20 16:39:18 nbd: ping Jul 20 17:23:02 OutBackDingo: pong Jul 20 17:24:27 nbd: any smart way to decipher the switch on this Ubiquiti AirRouter, wireless is up now, though we only show a lan interface, where it is a single wan port, and 4 port switch Jul 20 17:25:17 does the airrouter have a gigabit or 10/100 switch? Jul 20 17:26:28 if it has a 10/100 switch, then it's most likely the one built into the cpu Jul 20 17:30:53 nbd: its 10/100 Jul 20 17:32:35 should be trivial to extend the mach-ubnt.c file for the new router and adding a new image for it Jul 20 17:38:44 nbd: ok ill dig deeper into it Jul 20 18:16:36 ping jow_laptop Jul 20 19:47:39 RealOpty: pong Jul 20 20:00:24 jow_laptop, can u test out https://dev.openwrt.org/ticket/9718 Jul 20 20:00:41 florian seems to be having issues i cant duplicate Jul 20 20:05:27 looks ok to me Jul 20 20:25:32 jow * r27712 /packages/ipv6/radvd/ (Makefile patches/ patches/100-RS-RA-warning-to-debug.patch): [packages] radvd: reduce log spam introduced by v1.8, it logs every RA it emits as warning. change it to a debug level message Jul 20 20:26:50 jow * r27713 /branches/packages_10.03.1/ipv6/radvd/ (Makefile patches/): [packages_10.03.1] merge r27712 Jul 20 21:24:56 <^Willie^> jow_laptop: afther a little seeking at a few mailing lists.. my binutils was to new for glibc.. Jul 20 21:26:06 <^Willie^> atleast the toolchain builds ok using glibc, and directly wrote some notes. Jul 20 21:54:30 jow * r27714 /packages/libs/libgphoto2/Makefile: [packages] libgphoto2: disable iconv support if built against the stub, this allows drivers with optional conversion to work which would otherwise fail Jul 20 21:55:52 jow * r27715 /packages/libs/libgphoto2/Makefile: [packages] libgphoto2: bump PKG_REVISION to make the fix identifiable Jul 20 21:56:32 jow * r27716 /branches/packages_10.03.1/libs/libgphoto2/Makefile: [packages_10.03.1] merge r27714, r27715 Jul 20 22:10:22 jow * r27717 /packages/net/udpxy/Makefile: [packages] udpxy: update to version Chipmunk-20 **** ENDING LOGGING AT Thu Jul 21 02:59:56 2011