**** BEGIN LOGGING AT Sun Jul 21 02:59:58 2013 Jul 21 03:43:47 build #279 of ixp4xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/279 Jul 21 08:08:48 nbd r37486 trunk/package/kernel/linux/modules/netdevices.mk Jul 21 08:08:48 kernel: remove an exotic ethernet driver from the tulip package which does not build on some platforms Jul 21 08:17:13 build #270 of ar71xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ar71xx/builds/270 Jul 21 08:20:30 build #250 of iop32x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/250 Jul 21 08:21:47 nbd r37487 trunk/package/kernel/linux/modules/usb.mk * kernel: add missing config symbol for the chipidea usb driver Jul 21 08:21:51 nbd r37488 trunk/package/kernel/linux/modules/crypto.mk * kernel: fix kmod-crypto-hw-ppc4xx dependencies Jul 21 10:00:35 build #244 of gemini is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/244 Jul 21 10:27:47 build #66 of mpc85xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/mpc85xx/builds/66 Jul 21 10:55:30 build #251 of mcs814x is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mcs814x/builds/251 Jul 21 11:10:28 jogo r37489 trunk/package/kernel/linux/modules/netdevices.mk * kernel: fix config breakage from removing the de4x5 tulip driver Jul 21 11:12:07 KanjiMonster: th Jul 21 11:12:08 thx Jul 21 11:12:17 yw Jul 21 11:13:02 I wonder from where bluetooth's rfkill dependency comes from Jul 21 11:13:39 the rfkill mess needs to be reworked Jul 21 11:13:53 there's a rather arbitrary list of platforms in 001-depends.mk related to it Jul 21 11:14:05 no idea what that's based on or what it's for Jul 21 11:14:29 and gemini is part of one list, but not the other below it Jul 21 11:14:37 no idea what that is all about Jul 21 11:15:07 i think we should add a global config option for rfkill support Jul 21 11:16:38 I rather mean that I don't find any references to rfkill in drivers/bluetooth Jul 21 11:17:21 ah, I missed the net/bluetooth part Jul 21 11:17:23 no wonder ;) Jul 21 11:18:01 :) Jul 21 11:18:22 oh, another hamlet-dependency Jul 21 11:18:38 depends on RFKILL || !RFKILL Jul 21 11:19:30 heh Jul 21 11:19:43 this is becoming quite common Jul 21 11:19:55 well, mostly with IPV6 Jul 21 11:20:04 looking at the results and not reading it literally, it does make some sense Jul 21 11:20:51 e.g. if RFKILL=m, then this stuff needs to be =m as well Jul 21 11:20:55 yes Jul 21 11:23:57 what is the reasoning of depending on rfkill for only a select number of platforms? Jul 21 11:24:28 build #254 of adm5120 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/254 Jul 21 11:27:26 build #222 of octeon is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/222 Jul 21 11:27:50 build #253 of ep93xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/253 Jul 21 11:30:19 nbd: ultimately "[kernel] extract the rfkill module to an extra package, only build rfkill module if it is not integrated into the kernel." - so I guess we can just now just always depend on it (or rather do it a +PACKAGE_kmod-rfkill:kmod-rfkill dependency) Jul 21 11:30:30 -just Jul 21 11:30:46 no need for awkward target-dependent dependencies anymore Jul 21 11:31:33 let's have a separate option for it Jul 21 11:31:51 i'd like to avoid having the rfkill dependency in all targets in the snapshot builds Jul 21 11:32:19 i think that was the reason for the quirky target list in the first place Jul 21 11:32:39 we could then have a target feature flag for enabling the option by default Jul 21 11:32:39 "@(TARGET_ar71xx||TARGET_brcm47xx||TARGET_s3c24xx||TARGET_x86||TARGET_gemini||TARGET_cns3xxx||TARGET_ixp4xx)" <- these targets already have it Jul 21 11:33:28 build #237 of mpc52xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/237 Jul 21 11:34:34 KanjiMonster: yeah, and i think that's a mistake Jul 21 11:34:39 one that we should correct Jul 21 11:37:02 i'll hack up a patch for it Jul 21 11:43:25 KanjiMonster: http://nbd.name/rfkill.patch Jul 21 11:45:06 nbd: looks good, but maybe first add the rfkill_support stuff (and add it to the appropriate targets), then add use_rfkill to rfkill Jul 21 11:45:06 lars r37490 trunk/target/ (11 files in 7 dirs) * kernel: Remove gpio-pwm patches Jul 21 11:45:32 KanjiMonster: i think the target list was completely bogus to begin with Jul 21 11:46:47 nbd: from what I understood the target list was at once those targets that could build it as a module, for when we kmods for stuff built into the kernel broke everything Jul 21 11:47:38 that doesn't make any sense to me Jul 21 11:48:21 at least from what i can see, any target can build rfkill as module Jul 21 11:48:28 at least in current kernels Jul 21 11:49:18 i checked 3.3, seems to be the same Jul 21 11:54:02 i think i'll just commit this, and if anybody wants rfkill to be enabled in the shapshot builds for a particular platform, that can be done later Jul 21 11:54:03 nbd: remember the time where we did not have the "build an empty ipk if the file/.ko is not available/built into the kernel"? ;) Jul 21 11:54:16 yes, but i see no platform specific dependencies for the rfkill stuff Jul 21 11:54:29 nor do i see any targets building it into the kernel Jul 21 11:55:02 maybe that was different at some point (if so, probably a long time ago) Jul 21 11:55:10 but as it stands right now, the list seems 100% bogus Jul 21 11:56:17 so i'm going to commit this now, and if anybody wants rfkill to be enabled in snapshot builds for a particular platform, he can add a feature flag Jul 21 11:57:19 nbd r37491 trunk/ (5 files in 4 dirs) Jul 21 11:57:19 kernel: add a new global config symbol for enabling rfkill support (can be enabled by default via target feature flag) Jul 21 11:59:23 nbd: going from the original commit, it was once a exclusion list (where rfkill was not built) that was inverted some time (to switch to the targets where it was not excluded), which is likely why the list looks like that Jul 21 12:00:11 also at that time some targets indeed had rfkill built into the kenrel Jul 21 12:00:20 https://dev.openwrt.org/browser/trunk/target/linux/brcm47xx/config-2.6.25?rev=14780 Jul 21 12:23:16 build #2 of imx23 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/imx23/builds/2 Jul 21 12:29:12 hauke r37492 trunk/target/linux/generic/config-3.10 * kernel: add missing config option CONFIG_SND_MXS_SOC Jul 21 13:55:44 nbd r37493 trunk/target/ linux/generic/patches-3.10/621-sched_act_connmark.patch linux/generic/patches-3.9/621-sched_act_connmark.patch * kernel: fix crash in act_connmark in 3.9 and 3.10 (fixes #13916, #13876) Jul 21 14:00:50 nbd r37494 packages/net/dsl-qos-queue * dsl-qos-queue: remove from /packages Jul 21 15:51:24 nbd: i'll be receiving this tomorrow and i'll report on differences from the tp-link variant: http://www.rakuten.com/pr/product.aspx?sku=245965249 Jul 21 16:01:32 updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/index.html Jul 21 18:25:36 build #316 of cobalt is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/316 Jul 21 19:37:20 what is missing here: http://pastebin.com/qepzTwYM ? Jul 21 20:09:25 nbd r37495 packages/net/siit/src/siit.c * siit: add MODULE_LICENSE, get rid of some obsolete #ifdef crap Jul 21 21:12:10 build #346 of brcm63xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/brcm63xx/builds/346 Jul 21 21:33:09 i've run into a strange issue with block-mount in ubox, specifically +198 in libblkid-tiny.c Jul 21 21:33:28 pr is getting corrupted after pr->err = idinfos[i]->probefunc(pr, mag); returns Jul 21 21:33:50 so it segfaults Jul 21 21:34:17 i couldnt figure out where it is getting corrupt. right before probefunc (squashfs) return 0, the address of pr is intact Jul 21 21:34:48 but after it returns to probe_block, pr is at a different address. so when pr->err is attempted to be set, it segfaults Jul 21 21:35:11 does it sound like a compiler issue? uclibc? Jul 21 21:36:26 is the return address correct when it's about to return? Jul 21 21:36:42 yes Jul 21 21:36:58 is the pr pointer re-loaded from the stack after the return? Jul 21 21:36:59 right before probefunc returns, it is the same as before probefunc was entered Jul 21 21:37:09 or is it kept in a register Jul 21 21:37:21 :| Jul 21 21:37:23 dunno? Jul 21 21:37:29 :X Jul 21 21:37:52 if the former, it could be a bug in probefunc, if the latter, it's a calling convention or compiler issue Jul 21 21:38:28 well, nobody else seems to have the issue. i had ssp enabled in uclibc Jul 21 21:38:29 simple stack overflow Jul 21 21:38:37 in probe_squashfs Jul 21 21:38:51 nbd: guess or confirmed? Jul 21 21:39:02 i can see it just by looking at the code Jul 21 21:39:06 ah, ok Jul 21 21:39:13 uint8_t md5[4]; Jul 21 21:39:17 (i didn't look at the code, too lazy ;) ) Jul 21 21:39:20 4 byte is not enough to hold an md5 hash ;) Jul 21 21:39:22 lol Jul 21 21:39:25 blogic: ping Jul 21 21:39:26 ^ Jul 21 21:44:01 m4t: please try making the md5 array bigger and see if that fixes your issue Jul 21 21:44:16 blogic will probably take care of this tomorrow Jul 21 21:45:25 hrm nope Jul 21 21:45:59 nope? Jul 21 21:46:43 oh wait Jul 21 21:46:46 not big enough Jul 21 21:46:49 :) Jul 21 21:48:25 yeah i think that fixed it Jul 21 21:49:30 cool Jul 21 21:49:33 yeah Jul 21 21:51:48 thanks Jul 21 21:58:40 while i have your attention, what do you think about setsid() in procd_early Jul 21 21:59:36 why? Jul 21 21:59:47 because htop thinks its a kernel process Jul 21 21:59:49 ;) Jul 21 22:00:13 ah, ok Jul 21 22:00:26 you can tell that to blogic, he maintains the procd init functionality Jul 21 22:00:31 and ubox Jul 21 22:00:48 will do Jul 22 00:09:12 where has blogic disappeared to? Jul 22 00:28:56 oh, interesting Jul 22 00:29:19 looks like remote syslogging is working on a today-rev Jul 22 00:29:39 at least on a ubiquiti airrouter Jul 22 00:36:39 build #328 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/328 Jul 22 00:38:25 trying again with a buffalo wzr600dhp (also didn't work on alix2 as of r37374) Jul 22 00:39:11 it occured to me today to look at tcpdump traffic for more clues, and was surprised to see it work ;-) Jul 22 00:41:14 build #295 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/295 Jul 22 00:44:35 build #295 of ppc44x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/295 Jul 22 00:55:06 curiouser and curiouser Jul 22 00:55:14 wzr600dhp still not working Jul 22 00:56:53 tcpdump shows no udp syslog traffic (despite adding log_proto udp in /etc/config/system) on boot, but it does on restart, same rev (r37493) on ubnt-airrouter works. Jul 22 00:57:14 * russell-- is baffled Jul 22 01:11:08 * russell-- has been pinging blogic for a week, begining to think blogic is yella (http://yella.urbanup.com/1078315) Jul 22 01:39:02 build #309 of uml is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/309 Jul 22 02:48:36 build #342 of at91 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/342 Jul 22 02:50:58 build #279 of sibyte is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/279 **** ENDING LOGGING AT Mon Jul 22 02:59:58 2013