**** BEGIN LOGGING AT Sun Feb 12 02:59:56 2012 Feb 12 07:18:02 kerneis * r30464 /packages/net/babel/ (Makefile files/babeld.init): Feb 12 07:18:02 [packages] babeld: fix keep_unfeasible parameter Feb 12 07:18:02 Thanks to Juliusz Chroboczek. Feb 12 07:35:31 build #112 of avr32 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/112 Feb 12 09:01:08 hauke * r30465 /packages/net/freeradius2/ (4 files in 2 dirs): [packages] freeradius2: update to version 2.1.12 Feb 12 09:20:24 kaloz * r30466 /trunk/toolchain/gcc/common.mk: [toolchain/gcc]: update to latest Linaro releases Feb 12 09:44:24 kaloz * r30467 /trunk/toolchain/gcc/ (Config.in Config.version): [tolchain/gcc]: use Linaro GCC 4.6 by default Feb 12 09:45:35 kaloz * r30468 /trunk/toolchain/uClibc/ (Config.in Config.version): [toolchain/uClibc]: use uClibc 0.9.33 by default Feb 12 09:48:00 kaloz * r30469 /trunk/toolchain/binutils/Config.in: [toolchain/binutils]: use binutils 2.22 by default Feb 12 10:18:45 kaloz * r30470 /trunk/toolchain/gcc/ (8 files in 3 dirs): [toolchain/gcc]: upgrade vanilla gcc to 4.6.2 Feb 12 10:26:54 kaloz * r30471 /trunk/toolchain/binutils/Config.in: [toolchain/binutils]: enable 2.22 for unsupported archs if BROKEN is enabled Feb 12 11:52:22 kaloz * r30472 /trunk/target/linux/cns3xxx/config-default: [cns3xxx]: enable another l2 cache errate on newer kernels Feb 12 12:46:50 who was it who had a Huawei router? my BT homehub V3B has a firmware verison V100R001C01B031SP09_L_B - and a quick search on this type of pattenr leads to Huawei Feb 12 13:22:16 juhosg * r30473 /trunk/target/linux/ramips/files/arch/mips/ralink/rt305x/ (21 files): ramips: rt305x: remove per-board physmap_flash_data instances Feb 12 13:22:17 juhosg * r30474 /trunk/target/linux/ramips/files/arch/mips/ralink/rt288x/ (6 files): ramips: rt288x: remove per-board physmap_flash_data instances Feb 12 13:22:19 juhosg * r30475 /trunk/target/linux/ramips/files/drivers/spi/ramips_spi.c: ramips: include 'linux/module.h' in the ramips_spi driver Feb 12 13:22:20 juhosg * r30476 /trunk/target/linux/ramips/ (14 files in 4 dirs): ramips: add initial support for 3.2 Feb 12 13:57:45 WillieNL: ping Feb 12 16:10:26 nbd: pong Feb 12 16:11:41 WillieNL: got patches for me for glibc and stuff? Feb 12 16:11:45 philipp64|laptop: pong Feb 12 16:13:20 nbd * r30477 /trunk/package/broadcom-diag/src/diag.c: broadcom-diag: fix WRT150N/WRT160N detection (patch from #10836) Feb 12 16:18:10 nbd: raw pastebin -> http://pastebin.com/raw.php?i=UPe6tjiB Feb 12 16:18:32 nbd: must i post them in the ticked from chocky ? Feb 12 16:19:27 my build using 2.14 is now up and running for about 20 hours including collectd and rrdtool without modificating those :) Feb 12 16:19:49 those segfaults i had months ago where a problem in that ancient libc at the time :) Feb 12 16:20:35 i can post the package sizes if you like.. image size is about 9~13M Feb 12 16:27:34 WillieNL: no need to post them Feb 12 16:28:17 nbd: also this is not fixing anything for uc or eg libc yet :( Feb 12 16:28:39 but like we talked before i have a diffrent target purpose for using this wonderfull project Feb 12 16:32:45 i'm also happy that i can help somehow ;-) always asking for help is not the right way :p Feb 12 16:42:07 nbd * r30478 /trunk/toolchain/glibc/ (9 files in 5 dirs): glibc: remove all previous (ancient) versions, add 2.14 (partially based on a patch and testing by WillieNL) Feb 12 16:42:14 nbd * r30479 /trunk/tools/genext2fs/Makefile: tools/genext2fs: do not include the host getline.h, it breaks the tool in subtle ways, and the source already has a getline replacement Feb 12 16:43:47 nbd: for the comming few days i'll spend some time i setting up my home httpd box i can share my x86 build output, if you like for other users for usage in vbox/vmware for example ? Feb 12 16:48:43 nbd: thanks for r30479 ! Feb 12 17:08:16 hauke * r30480 /trunk/target/linux/brcm47xx/patches-3.2/ (14 files): brcm47xx: update usb driver for brcm47xx. Feb 12 17:15:34 hauke * r30481 /trunk/target/linux/brcm47xx/patches-3.2/ (8 files): brcm47xx: refresh patches Feb 12 18:34:53 juhosg * r30482 /trunk/target/linux/ramips/image/Makefile: ramips: rt305x: pass MTD partition map to the ALL0256N via cmdline Feb 12 18:34:53 juhosg * r30483 /trunk/target/linux/ramips/image/Makefile: ramips: use GENERIC_{4,8}M image template for WR5123GN Feb 12 19:10:26 nbd * r30484 /trunk/package/grub/patches/030-add-stage2-ldscripts.patch: (log message trimmed) Feb 12 19:10:26 package/grub: fix stage2 link with recent binutils Feb 12 19:10:26 We want the `_start's symbols to be at the very beginning of the `stage2' Feb 12 19:10:26 binary's `.text' section. However, nothing enforces this requirement. With Feb 12 19:10:26 recent binutils (ie. post 2.19.1), this results in a corrupt binary, as ld(1) Feb 12 19:10:26 puts another symbol (eg. `journal_init') at the beginning of the section. Feb 12 19:10:26 Fix this by providing a stripped down version of the default linker script Feb 12 19:13:41 nbd: i use binutils 2.20 or newer even 2.22 in my last build Feb 12 19:14:07 yeah, in my own test i just ran into the error described in that ticket Feb 12 19:14:33 i had that before here in a random mess of errors inducating a toolchain problem :) Feb 12 19:15:06 also wich version of gcc do you use ? 4.5 or 4.6 Feb 12 19:15:10 i used 4.5 Feb 12 19:15:26 right now 4.6, because unfortunately it was made the default Feb 12 19:15:30 prematurely, in my opinion Feb 12 19:15:54 will test 4.6 when have my router up and running also a local svn/git host for putting my own stuff in a ordered way safe Feb 12 19:16:21 don't try 4.6 on ar71xx (in case you have something like that Feb 12 19:16:22 nbd: agree but if nobody test 4.6 we dunno where it need some extra work Feb 12 19:16:34 people did test 4.6 on some platforms already Feb 12 19:16:37 and it was found to be broken Feb 12 19:16:52 atm i'm focusing on x86 and in the future have some broadcom stuff Feb 12 19:17:25 sounds like what i tryed to tell a while ago by saying it wrong when blame the toolchain while asking why dropping older gcc for newer ;-) Feb 12 19:17:54 well, 4.5 is well-tested Feb 12 19:17:57 the linaro version of it Feb 12 19:18:02 also i can't speak for atheros/mips and other platforms.. Feb 12 19:18:21 remeber a short while mips32el/mips64el with gentoo but thats it Feb 12 19:20:09 either way, with gcc 4.6, binutils 2.22 and glibc 2.14 the system seems to be working now Feb 12 19:20:13 (tried in qemu) Feb 12 19:20:35 ok thats quick or you had multiple builds running ? Feb 12 19:20:45 nice to hear that it works with gcc 4.6 Feb 12 19:20:58 i had started a build already before i got your patch Feb 12 19:21:16 also i wonder why/where it did go wrong with eglibc and uclibc on x86 wich resulted in library linkage issues i think Feb 12 19:21:25 what kind of issues? Feb 12 19:22:29 hmm i had some problems with grub during the builds also when the builds did go well the produced librarys and apps segfaulted in my case multiple times with rrdtool/collectd Feb 12 19:22:50 now they run smooth. Feb 12 19:23:15 same about grub i was expecting some trouble theire but surpriced about no probs with grub. Feb 12 19:24:58 also when select create a ramdisk with iso image the system stopped before joining the rootf with the kernel on a isofs image :( gzip was looking for the combined archive before it was produced.. i have not looked in the lists if it is known already. Feb 12 19:25:19 since vbox images where ok for me i have ignored this. Feb 12 19:29:02 http://pastebin.com/jGukDbtb Feb 12 19:30:02 WillieNL on what hardware you use x86 openwrt Feb 12 19:30:47 crow: aopen i855gmem-lfs with 1.7ghz centrino dothan cpu Feb 12 19:31:25 with the current results my 4gb cf is overkill.. Feb 12 19:32:45 i see, well i wanna compile openwrt x86 for my alix 3d3 board. i think i have 8gb cf on it Feb 12 19:33:01 updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/index.html Feb 12 19:33:04 hmm my image glibc based was around 9~13MB Feb 12 19:33:13 so 16MB compact flash can do :) Feb 12 19:34:18 i never tried x86 of openwrt, need to check wiki if there is something to be done and not to compile not booting image. Feb 12 19:35:49 hmm before i give/send the wrong idea out.. x86 works as it is and theire are some precompiled images uclibc based on the public servers Feb 12 19:36:57 i'm doing it on my own way for personal reasons end result that i have in mind is a diy system from scratch and not willing to reinvent a wheel while you can help someone else :) Feb 12 19:37:42 some are ok with a embedded dlink/linksys router others like to mess with a playtoy :) Feb 12 19:39:32 since this machine can do more then route 4 nics + wlan and have the options to run a 128M ramdisk for rootfs in 512mb ram why not use the pata/sata storage for samba/logs/statistics Feb 12 19:39:59 it it ever reach a bottle neck it will be funky :) Feb 12 19:40:13 :) Feb 12 19:40:30 seems like you are having fun Feb 12 19:41:08 well i also do like to compile my own image (even for wndr-3700v2), but would like to use openwrt on alix to. Feb 12 19:43:13 also on embedded devices with flash storage they are easyer to brake.. Feb 12 19:43:34 or you must be able to boot from usb storage then they are easy to debug Feb 12 19:43:57 but when have a couple of identical machines i had a idea like wtf why not. Feb 12 19:44:37 also i decide when i update it.. also i have to fix the problems i encounter without waiting endless for someone else that possible have enough work for his own todo Feb 12 19:45:06 since i can live with it as a diy hobby :) Feb 12 19:46:18 anyway bbl Feb 12 19:57:08 nbd * r30485 /trunk/package/swconfig/src/swlib.c: swconfig: fix off-by-one error, causing segfaults on gcc 4.6+ (fixes #9765) Feb 12 20:55:55 nbd * r30486 /trunk/toolchain/gcc/patches/4.6-linaro/850-use_shared_libgcc.patch: gcc 4.6: port over the missing patch 850-use_shared_libgcc.patch to prevent libgcc crap from leaking into every single binary Feb 12 23:00:00 nbd, so far, the earliest revision that fails is revert-06 (that is, reverting the ath5k patches 12, 11, 10, 9, 8, 7 and 6) Feb 12 23:01:43 revert-06, -07, -08, -12 and no-reversions (relative to r30427) have all failed, i am still testing -01 through -05, and -09 through -11. Feb 12 23:03:03 ok, based on that, my guess is patch 04 is the one that breaks Feb 12 23:03:21 0005 i think Feb 12 23:03:44 ath5k: Use usleep_range where possible Feb 12 23:03:56 04 is a calibration rework Feb 12 23:04:02 and the error looked a lot like a calibration issue Feb 12 23:04:08 it's still possible that my revert-05 will fail too Feb 12 23:04:34 it's been running about 24 hours, and i had one that only failed after 30 hours Feb 12 23:04:43 although others failed after about 4 Feb 12 23:09:23 build #112 of ixp4xx is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/112 Feb 12 23:09:39 hm, i have an idea Feb 12 23:09:50 i am not seeing the same problem with ath5k on wgt (mipsel) or x86-net45xx Feb 12 23:10:04 the gain calibration that is failing is specific to some wifi chipsets Feb 12 23:10:07 hauke * r30487 /trunk/target/linux/brcm47xx/ (config-3.0 config-3.2): brcm47xx: activate config bug to show warnings Feb 12 23:10:09 i think mostly the older ones Feb 12 23:10:34 i've seen it on meraki mini too Feb 12 23:10:40 one thing i noticed is that it stops tx during calibration to prevent the gain calibration from messing with traffic Feb 12 23:10:53 and if my interpretation of the calibration type is correct, this might be a very bad idea Feb 12 23:11:06 because that type of calibration may actually need some tx activity Feb 12 23:11:26 hauke * r30488 /trunk/target/linux/brcm47xx/ (config-3.0 config-3.2): brcm47xx: deactivate CONFIG_NO_HZ as it seams not to be needed. Feb 12 23:11:51 O_o Feb 12 23:12:26 i have one device ready to test any idea you might have (and more, as they fail) Feb 12 23:13:14 i have lots more ports available in my switch, although i'm starting to scrape the bottom for ready-made patch cables ;-) Feb 12 23:13:20 http://nbd.name/440-ath5k_calibrate_no_queue_stop.patch Feb 12 23:13:24 copy that to package/mac80211/patches Feb 12 23:13:30 without any of the other reverts Feb 12 23:13:35 right Feb 12 23:17:18 nice something else i can test :) Feb 12 23:18:51 building r30427 + your patch now Feb 12 23:19:05 aironet pi21ag + cb21ag + a 2.4ghz only ath5k card :) Feb 12 23:20:04 did you see this error before? Feb 12 23:21:48 i do not have problems building madwifi and ath5k drivers here.. loading shows some errors/warnings in my vbox runs Feb 12 23:21:49 mac80211: Unknown symbol cfg80211_probe_status (err 0) Feb 12 23:22:54 but theire is no ath5k card pressent and i have done something wrong in my kernel/driver setup for wifi.. i'll dive in that tomorrow and feed this build to that machine with one of the ath5k cards in it Feb 12 23:24:06 cisco chipsets will be older (super ag) compared to the single band extended range 54g card. Feb 12 23:24:33 so yes wil test Feb 12 23:29:02 revert-09 just failed too, after 5 hours Feb 12 23:29:10 (as anticipated) Feb 12 23:33:47 * russell-- reflashes it with your patch as well Feb 12 23:46:24 I just compiled Openwrt r25712 applying the neccessary patches from https://dev.openwrt.org/ticket/8938 and want to get it running on http://wiki.openwrt.org/toh/linksys/wag160n (v1) which should work imho ... Feb 12 23:52:47 can't upload it through webinterface though Feb 13 00:33:36 yuck, linksys routers **** ENDING LOGGING AT Mon Feb 13 02:59:58 2012