**** BEGIN LOGGING AT Fri Aug 05 02:59:56 2011 Aug 05 05:29:12 build #67 of at91 is complete: Failure [failed shell_5 compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/67 Aug 05 05:53:37 build #65 of ubicom32 is complete: Failure [failed compile_3] Build details are at http://buildbot.openwrt.org:8010/builders/ubicom32/builds/65 Aug 05 09:34:24 who's Peter Naulls? Aug 05 09:54:14 http://www.riscos.info/index.php/Peter_Naulls ?? Aug 05 09:54:27 there is also apparently an Peter Naulls appreciation society Aug 05 10:00:40 i wanted to know his nick ins this channel Aug 05 10:02:51 mirko: peter@chocky... -> Chocky Aug 05 10:05:12 KanjiMonster: thanks! Aug 05 10:05:43 Chocky: ping Aug 05 10:32:55 juhosg * r27907 /trunk/target/linux/ar71xx/image/Makefile: ar71xx: fix image generation Aug 05 10:35:02 Hi, what's the correct procedure to follow to push a package update to the openwrt trunk? Aug 05 11:20:58 hello Aug 05 11:21:01 * Chocky hides Aug 05 11:21:34 internet past always catches up with you, and I've been on the internet a loooong time Aug 05 11:24:53 mirko: ping Aug 05 11:30:21 Chocky: btw, looks like the confirmation email finally arrived ;) Aug 05 11:30:33 I did it again, and it came in seconds Aug 05 11:31:41 Chocky: i was reading your mail regarding the libc-stuff Aug 05 11:31:49 \o/ Aug 05 11:32:01 Chocky: i'd like to get it cleaned and working as well Aug 05 11:32:14 do you need anything? Aug 05 11:32:30 since i'm working currently on a project using the eglibc i was changing quite some things Aug 05 11:32:50 Chocky: it sounded like you're a bit upset with the current setup / how it is handled within openwrt Aug 05 11:33:25 no; I'm just frustrated and the endless glibc bashing, and the apathy towards this whole matter Aug 05 11:33:31 s/and/at/ Aug 05 11:33:55 I've had this stuff for months now Aug 05 11:38:04 Chocky: i see Aug 05 11:38:18 ultimately, I have to make a working product. that's my focus. Aug 05 11:38:43 Chocky: same here - that's why I invested lot's of time fixing that eglibc stuff Aug 05 11:38:56 lots too, I bet :o Aug 05 11:39:17 so, you are sitting on a load of patches too? Aug 05 11:40:29 bbl. Aug 05 11:41:27 Chocky: i committed all my stuff Aug 05 11:41:33 focused on on eglibc 2.12 Aug 05 12:03:15 Chocky: if you have pending patches - let's try to get them upstream.. Aug 05 12:06:10 Chocky: btw. what about the imagebuilder issue, can i close the ticket now? Aug 05 12:54:52 Chocky: hmm some if/else statements for glibx -> Xglibc migrations are wishfull instaid of patches that simply remove whole lines and somtimes even more :) Aug 05 12:55:03 s/glibx/glibc Aug 05 12:55:18 s/Xglibc/Xlibc Aug 05 12:55:28 X as uc or eg ;-) Aug 05 12:55:59 then it will not include big chunks on small targets and still keep support on "defaults" Aug 05 12:58:02 also this will not apply for all patches just some.. Aug 05 13:35:54 hi, I have trouble to compile mac80211 package for mips architecture, can someone look pls ? http://pastebin.com/YL28Jtit Aug 05 13:39:48 voyo: using an external kernel tree? Aug 05 13:42:18 nbd: yes, 2.6.31 . tried default compat-wireless 2010-03-24 then switched to try to 2011-06-22 after disabling all patches. all with no luck. Aug 05 13:42:33 why are you doing that? Aug 05 13:42:41 doing what ? Aug 05 13:42:59 using 2.6.31 as an external kernel tree? Aug 05 13:43:11 did you backport ar71xx patches manually to that tree? Aug 05 13:43:59 have some important to me changes in that kernel , some proprietary kernel things, can't easily and fast port this to any newer kernel Aug 05 13:44:09 what kind of proprietary things? Aug 05 13:44:47 improvements to gre tunnels, bonding stuff, routing. Aug 05 13:45:07 nothing related to wifi Aug 05 13:45:20 well, your tree seems to be messed up, as it uses the wrong cflags Aug 05 13:47:02 hm, its my first approach to build for mips target, for sure it was not changed regarding mips.. Aug 05 13:47:13 where should I look ? Aug 05 13:47:14 for mips target? Aug 05 13:47:18 not all mips are created equal Aug 05 13:47:36 yes. I played with arm and *-x86 Aug 05 13:47:48 what kind of mips is it? Aug 05 13:48:08 1sec, dont remember, will find url Aug 05 13:48:56 ah, so you're using the ar71xx target even though your device is not ar71xx based Aug 05 13:49:04 in that case you need to change the target Aug 05 13:49:15 since your kernel tree implies that it's something mips3 based Aug 05 13:49:18 which is mips32r1 Aug 05 13:49:23 whereas ar71xx is mips32r2 Aug 05 13:49:31 florian * r27908 /trunk/target/linux/generic/config-3.0: [kernel] add missing config symbol Aug 05 13:49:35 florian * r27909 /trunk/target/linux/generic/patches-3.0/101-perf_arm_uclibc.patch: [kernel] add 101-perf_arm_uclibc.patch to 3.0 patches Aug 05 13:49:38 florian * r27910 /trunk/target/linux/realview/Makefile: [realview] update to kernel 3.0 Aug 05 13:49:41 http://ubnt.com/routerstation I choosed for that Target System (Atheros AR71xx/AR7240/AR913x) -> Target Profile (Ubiquiti RouterStation Pro) ---> Aug 05 13:49:42 florian * r27911 /trunk/target/linux/generic/config-3.0: [kernel] add more missing config symbols Aug 05 13:49:45 florian * r27912 /trunk/target/linux/malta/ (4 files in 3 dirs): [malta] update to kernel 3.0 Aug 05 13:50:07 ah, ok Aug 05 13:50:19 seems ok for me , right ? Aug 05 13:50:20 if it really is a routerstation that you're building for, then your kernel tree selects the wrong cflags Aug 05 13:50:56 btw. what kind of gre / routing changes are those? Aug 05 13:50:58 anything interesting? Aug 05 13:51:14 ok, I'll check that in kenrel_menuconfig Aug 05 13:51:28 maybe something is not/wrongly populated to kernel's .config Aug 05 13:51:47 Im using it for company Im workin for Aug 05 13:51:56 what code is the .31 tree based on? Aug 05 13:52:02 is it a backport of the openwrt ar71xx support? Aug 05 13:52:10 or is it from the atheros sdk? Aug 05 13:52:34 link bonders, and loadbalancer stuff.. Aug 05 13:52:44 ah, ok Aug 05 13:52:48 erm, not sure what you mean ? Aug 05 13:52:59 well, a normal 2.6.31 kernel does not have support for that SoC Aug 05 13:53:32 so if you want to use it on ar71xx, you either need the openwrt patches for ar71xx support, or the atheros code Aug 05 13:53:39 aaaaaa.... Aug 05 13:53:57 damn, so thats probably my case.. Aug 05 13:54:09 so if you don't have that, then you should probably forward port your company's changes Aug 05 13:54:14 I just blindly tried to compile my orginal old kernel with that Aug 05 13:54:52 I'd rather and easier probably port those ar71xx patches to 2.6.31 then... Aug 05 13:56:11 btw. next time you ask for help on such issues, you should probably mention upfront that you're doing something completely non-standard and unsupported ;) Aug 05 13:56:32 do you know maybe which kernel start supporting ar71xx ? Aug 05 13:56:53 no kernel fully supports it without the openwrt patches Aug 05 13:57:02 there's some initial work in 2.6.39 Aug 05 13:57:03 ha, right, I tought Im known already on this channel that Im doing completly non-standard and unsupported things ;) Aug 05 13:57:19 i mean non-standard and unsupported relative to what openwrt usually does ;) Aug 05 13:57:32 yes yes , exactly ;) Aug 05 13:58:10 the 'backfire' branch of openwrt has 2.6.32 patches for ar71xx Aug 05 13:58:20 Im using owrt only as "base" (more like linux distribution), with my own things inside, own configuration etc. Aug 05 13:58:29 but if you forward port your stuff to 2.6.39 from openwrt trunk, you get much better performance Aug 05 13:59:22 i recently approximately doubled routing performance for ar71xx there Aug 05 14:00:22 hm. thats probably long term solution, not for now. (my stuff in newer kernel). and honestly - its quite dangerous to change kernel in commercial product (without proper and long testing) ;) Aug 05 14:00:49 well, what you're doing sounds to me like you're making something new anyway Aug 05 14:01:02 considering that your kernel has no support for the platform that you're trying to build for ;) Aug 05 14:02:11 [florian]: (regarding malta update) - doesn't it need the i8259 compile fix that didn't make it for 3.0? Aug 05 14:02:28 <[florian]> it did not trigger the build failure Aug 05 14:02:29 this board I have its for testing, Im not yet sure if I will use it, depends on issues, performance, etc etc. currently Im using ARM and x86 boxes. Aug 05 14:04:37 nbd so I think I have all those patches which I need to apply to my kernel in target/linux/ar71xx/patches-*/ , should be enough, right ? Aug 05 14:05:23 well, mostly Aug 05 14:05:35 to run properly, openwrt will also need some of the generic patches Aug 05 14:05:38 which are relevant for rootfs mounting Aug 05 14:06:04 and maybe other things as well Aug 05 14:06:15 bbl Aug 05 14:06:35 ok, Im not using those patches ;) Aug 05 14:06:37 ty Aug 05 14:06:58 well, you will need at least some of them if you want to boot a normal openwrt with that kernel Aug 05 14:08:09 you mean like something from target/linux/generic-2.6/patches-*/ ? Aug 05 14:08:19 yes Aug 05 14:08:51 ok, Im not using any of those patches rightnow for my 2.6.31, so I think I dont need it also for ar71xx Aug 05 14:09:55 but thank for the tip, I'll look there in case of problems with patching for ar71xx Aug 05 14:22:55 build #67 of s3c24xx is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/67 Aug 05 14:48:56 willienl: I can't make any sense of that. I appreciate your input, but *slow it down*. Take more care in what you type and the intended meaning. It will save us both time. Aug 05 14:49:26 mirko: ping Aug 05 14:50:24 nbd: manually providing a list of packages on cli is certainly a backwards step. Do you disagree with the ability to override the subarch in the profile? That seems like an easy thing to do. Aug 05 15:17:55 * Chocky applies for sponsorship to go to UDS in Florida end of October Aug 05 15:33:23 acinonyx * r27913 /packages/net/quagga/patches/120-quagga_manet.patch: [packages] quagga: Fix installing of OLSR host routes Aug 05 15:33:27 acinonyx * r27914 /packages/net/olsrd/patches/910-quagga_workaround_remove.patch: [packages] olsrd: Remove 'destination == gateway' routes workaround - should be fixed with new Quagga patch Aug 05 16:00:27 I'll ask again ;-) can anybody apply my getline() darwin fixes.. https://lists.openwrt.org/pipermail/openwrt-devel/2011-July/011758.html and https://lists.openwrt.org/pipermail/openwrt-devel/2011-July/011757.html thanks! Aug 05 16:13:48 dirtyfreebooter: is the second patch really needed? Aug 05 16:23:02 larsc: both patches are need, or build doesn't work Aug 05 16:23:37 mkimage defines its own getline.h and getline.c Aug 05 16:23:46 separate from openwrt's getline.h Aug 05 16:25:05 ah Aug 05 16:25:10 i guess its probably better to make the patch use openwrt's getline.h instead, along with adding in the __DARWIN_C_LEVEL Aug 05 16:25:43 if you rather that, I can try it and re-work the patches Aug 05 16:27:43 nevermind, i thought mkimage might be an openwrt package, but i doubt the upstream author would want to remove that, since it still useful outside of the openwrt build env Aug 05 16:28:50 dirtyfreebooter: why did you move the ifdef beneath the includes in patch 1? Aug 05 16:29:15 because __DARWIN_C_LEVEL is not defined by the compiler Aug 05 16:29:31 i see Aug 05 16:29:33 you have at least include stdio.h or whatever it includes to have it defined Aug 05 16:30:54 i tried looking at the compiler built-in defines, for 10.7 and nothing helped, __DARWIN_C_LEVEL wraps getinline() in stdio.h and it seemed most appropriate Aug 05 16:31:07 getinline/getline Aug 05 16:40:08 nbd * r27915 /trunk/toolchain/kernel-headers/Makefile: kernel-headers: fix compile error caused by wrong host include path when the toolchain is already built Aug 05 16:53:51 lars * r27916 /trunk/tools/include/getline.h: (log message trimmed) Aug 05 16:53:51 [tools] host/include/getline.h - Mac OS X 10.7 (Lion) fix Aug 05 16:53:51 Apple has finally updated their version of Libc, and it now includes getline(). \o/ Aug 05 16:53:51 The following patch fixes OpenWRT getline.h to not redefine it for 10.7. The one Aug 05 16:53:51 caveat is, the includes (at least stdio.h) are needed before the #ifdef so that Aug 05 16:53:51 __DARWIN_C_LEVEL could be properly defined (it is not a compiler built-in define Aug 05 16:53:51 like __linux__ or __APPLE__). Aug 05 16:54:49 lars * r27917 /trunk/tools/mkimage/patches/020-darwin-10_7-getline-fix.patch: Aug 05 16:54:49 [tools] mkimage/getline.h - Mac OS X 10.7 (Lion) fix Aug 05 16:54:49 This mends mkimage for getline() in Mac OS X 10.7. Aug 05 16:54:49 Signed-off-by: Joe Roback Aug 05 16:55:23 silly Apple Aug 05 17:00:29 yea silly Apple, but at least they are updating things.. maybe not as fast as linux, but certainly faster than systems like AIX, HP-UX, and other UNIX dinosaurs Aug 05 17:01:19 hah Aug 05 17:01:22 but "no one uses those" Aug 05 17:02:00 I doubt many people using "big unix" have much interest in OpenWrt on them. much easier to get a Linux system etc Aug 05 17:03:55 yea, agreed, ppl that use big unix, have no interest in updating anything, their attitude is, it works, no changes, uptime is more valuable then having the latest and greatest.. Aug 05 17:04:39 than, too Aug 05 17:08:23 build #65 of brcm47xx is complete: Exception [exception interrupted] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/65 Aug 05 17:08:24 build #64 of atheros is complete: Exception [exception interrupted] Build details are at http://buildbot.openwrt.org:8010/builders/atheros/builds/64 Aug 05 17:08:26 build #63 of cobalt is complete: Exception [exception interrupted] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/63 Aug 05 17:08:28 build #62 of orion is complete: Exception [exception interrupted] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/62 Aug 05 18:25:02 does gpioctl not work on ar71xx? Aug 05 18:25:26 I can't seem to find /dev/gpio on a ubiquiti unifi AP Aug 05 18:26:54 build #61 of ramips is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/61 Aug 05 18:49:16 build #61 of pxcab is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/pxcab/builds/61 Aug 05 18:49:17 build #61 of ps3 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/ps3/builds/61 Aug 05 20:06:04 build #57 of ppc44x is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/57 Aug 05 21:13:58 build #57 of uml is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/57 Aug 05 23:17:25 i get this error (genext2fs: not enough memory for filesystem) when building trunk. any hints? Aug 05 23:18:13 please supply complete log in pastebin Aug 05 23:20:47 here it is http://pastebin.com/9NwS8rRg Aug 05 23:22:15 "complete log" Aug 05 23:22:25 including the command you type Aug 05 23:22:26 +d Aug 05 23:22:35 make target/linux/install V=99 Aug 05 23:25:11 seem the host build of genext2fs has a problem? Aug 05 23:25:25 "complete log" Aug 05 23:25:35 heilterminalcons: looks like a 50M rootfs image howmutch packages did you include ? check "du -hs /srv/owrt/trunk/bin/x86/packages/" for the total packages size Aug 05 23:25:39 crap Aug 05 23:25:44 heilterminalcons: looks like a 50M rootfs image howmutch packages did you include ? Aug 05 23:25:59 much Aug 05 23:26:07 but you meant "many" Aug 05 23:26:48 i'll say his image is to small.. Aug 05 23:27:11 and indeed howmany* Aug 05 23:27:50 if selected glibc and no stripping + debug on "all" selected packages 70M can go up +500M :) Aug 05 23:28:12 but we dunno wat heilterminalcons exactly have done so thats hard to say at this moment Aug 05 23:28:27 let's stop guessing. Aug 05 23:28:42 * WillieNL continue's smoking. Aug 05 23:28:48 yuck Aug 05 23:28:54 the smoking, and the apostrophe Aug 05 23:28:58 even changing the image size to 1GB doesnt change Aug 06 02:40:42 hey! Aug 06 02:40:58 do you think it's a good idea update from LINUX_VERSION:=2.6.32.27 Aug 06 02:41:07 to LINUX_VERSION:=2.6.32.47 in all backfire targets? Aug 06 02:42:36 http://pastebin.com/xZuzw4e3 Aug 06 02:43:05 s/all/all (applicable)/ **** ENDING LOGGING AT Sat Aug 06 02:59:57 2011