**** BEGIN LOGGING AT Sun Apr 15 02:59:59 2012 Apr 15 03:14:02 build #121 of ep93xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/121 Apr 15 06:52:18 build #117 of adm5120 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/117 Apr 15 09:09:57 build #141 of uml is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/141 Apr 15 09:38:22 build #118 of gemini is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/118 Apr 15 09:49:18 build #164 of brcm47xx is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/164 Apr 15 10:21:40 hauke * r31288 /trunk/target/linux/at91/config-default: (log message trimmed) Apr 15 10:21:40 at91: Missing kernel config symbols for AT91 target platform Apr 15 10:21:40 I looked into buildbot snapshots and noticed that some platforms have Apr 15 10:21:40 not successfully compiled for a long time. One example is at91, which Apr 15 10:21:40 seems to be missing device-related kernel config symbols and the Apr 15 10:21:41 snapshot build crashes already there. Apr 15 10:21:42 Please find attached a patch that makes the at91 kernel config and Apr 15 10:28:23 build #142 of x86 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/142 Apr 15 11:19:43 build #135 of avr32 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/135 Apr 15 12:26:16 build #154 of orion is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/154 Apr 15 12:57:06 build #160 of atheros is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/atheros/builds/160 Apr 15 14:05:20 build #155 of lantiq is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq/builds/155 Apr 15 15:52:43 jow * r31289 /trunk/package/base-files/ (Makefile files/etc/banner image-config.in): [package] base-files: reformat banner to better work with varying version string lengths, do not set image options if menu is disabled (#11280) Apr 15 15:57:28 jow * r31290 /branches/backfire/package/base-files/ (Makefile files/etc/banner image-config.in): [backfire] backport r31289, remove default repo (#11280) Apr 15 16:39:08 stintel: ping Apr 15 16:39:12 nbd: pong Apr 15 16:39:22 you had some issues with default route with recent netifd versions, right? Apr 15 16:39:27 yeah Apr 15 16:39:48 ok, please try putting in commit id bc214ca470015b12a4f905f57ba6f2352c60f5a5 into package/netifd/Makefile Apr 15 16:39:55 and see if that fixes the issue Apr 15 16:40:17 i've been having some default route issues on my router as well, but unfortunately they only show up after the 24h pppoe disconnect, not if i try to trigger them manually Apr 15 16:40:27 so debugging this is a bit slow for me at the moment ;) Apr 15 16:41:47 can't we somehow change pppd to retain its allocated ppp device? The changing ifindexes every 24h are a pita Apr 15 16:42:02 also for other services which resolve the ifindex onle once on startup Apr 15 16:42:13 i need pppd to exit between attempts Apr 15 16:42:19 ah, right Apr 15 16:42:21 so that netifd can pass the return code to the proto handler Apr 15 16:42:52 why are the changing ifindexes a pita? Apr 15 16:42:54 yes yes. So we need to fix the rest Apr 15 16:43:29 because there are various packages lacking a hotplug handler to reinit themselves whenever $WAN (pppoe-xyz) changes Apr 15 16:43:40 hm Apr 15 16:43:40 for example snmp stuff Apr 15 16:44:32 i'd say turning those subtle bugs into not-so-subtle bugs is a good thing Apr 15 16:44:42 because stuff can never rely on ifindex (or even ifname) to remain the same Apr 15 16:44:46 configuration can change Apr 15 16:44:59 oh they're not that subtle. It usually goes like $foo stops working after 24 hours, my wan is pppoe ... Apr 15 16:45:12 the rest is clear then Apr 15 16:45:24 yes, we know that ifindexes are not stable etc. etc. Apr 15 16:45:35 but a lot of software out there is simply not well written :) Apr 15 16:45:46 true that Apr 15 16:46:20 my idea for fixing this in a crude but generic way would be to use the yet-to-be-written config reload stuff that i was planning Apr 15 16:46:31 yeah Apr 15 16:46:48 when there's something that depends on ifindexes to stay the same, we simply grab the ifindexes at startup and push them into the active config state Apr 15 16:47:00 and trigger a reload on every hotplug event Apr 15 16:50:40 until recently even radvd suffered from the changing ifindex problem Apr 15 16:50:56 and I believe the dhcpv6 stuff still has issues with it Apr 15 16:51:40 the generic solution so far was to add a hotplug handler which did a killall -HUP on the associated service whenever an ifup occured Apr 15 16:51:50 dirty but effective Apr 15 16:52:43 you guys know of anyone working on Mikrotik RB2011 support ? Apr 15 16:53:46 thepeople * r31291 /trunk/package/kernel/modules/fs.mk: disable xfs on avr32 Apr 15 16:58:10 build #147 of rb532 is complete: Exception [exception compile_10 compile_13] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/147 Apr 15 16:58:10 build #146 of ppc40x is complete: Exception [exception shell_16 compile_13] Build details are at http://buildbot.openwrt.org:8010/builders/ppc40x/builds/146 Apr 15 17:01:14 nbd: looks like I still don't get a default route Apr 15 17:02:55 I don't need to change anything in /etc/config/network do I ? Apr 15 17:04:02 another issue recently, my country code is ignored Apr 15 17:05:30 but only on my rspro with 2 wifi nics, hmmm Apr 15 17:05:37 show me your /etc/config/network Apr 15 17:06:27 oh dear Apr 15 17:06:48 my bad, sorry for the noise :( Apr 15 17:06:54 wonder how that got there Apr 15 17:07:01 option defaultroute '0' Apr 15 17:07:25 what is weird then, is that without netifd I do get a default route Apr 15 17:09:33 with netifd that option became generic Apr 15 17:09:44 probably the non-netifd version ignored it Apr 15 17:10:22 ah. sigh. checking my backups, the option defaultroute 0 is in there since january 2011 Apr 15 17:10:49 ok after changing that to 1 and rebooting the router all is well Apr 15 17:11:12 :) Apr 15 17:12:19 for wan/ipv4/dhcp that is. Apr 15 17:12:36 seems that with 6in4 I also get no default route Apr 15 17:12:54 and I added the option defaultroute 1 there to be sure Apr 15 17:13:20 no ::/0 and no 2001::/3 ? Apr 15 17:13:59 jow_laptop: http://pastebin.com/wYnHHmAc Apr 15 17:14:32 there is no 6in4 at all as it seems Apr 15 17:14:43 shouldn't there be a sit0 or 6in4-whatever ? Apr 15 17:14:50 ah, you're right Apr 15 17:15:03 so nothing happens. maybe 6in4 isn't supported yet in netifd Apr 15 17:15:07 it is Apr 15 17:15:24 but maybe it failed to derive the local v4 addr or something Apr 15 17:15:24 should there be a /lib/netifd/proto/6in4.sh ? Apr 15 17:15:34 can you post a redacted version of your current 6in4 config? Apr 15 17:16:20 stintel: yes, there should - and it is still provided by the external 6in4 package, its not included in netifd Apr 15 17:16:32 jow_laptop: http://pastebin.com/30yE88wb Apr 15 17:16:41 jow_laptop: that might be the problem then, don't see the 6in4.sh Apr 15 17:17:43 also the 6in4 package must be "compiled" with CONFIG_USE_NETIFD=y Apr 15 17:17:50 ... in .config Apr 15 17:17:58 ah. make clean first Apr 15 17:17:59 otherwise the legacy scripts are packaged, not the ones for netifd Apr 15 17:20:43 i have 6in4 running on my router Apr 15 17:20:45 ok did some make clean first Apr 15 17:20:47 works just fine Apr 15 17:23:57 when I get everything working on my router with netifd I'm gonna try looking into openvswitch some more Apr 15 17:24:04 but meh where does one buy more time Apr 15 17:24:30 :) Apr 15 17:26:42 don't think I want to run openvswitch on a router with flash storage though Apr 15 17:27:35 the flash might not like the ovsdb-server Apr 15 17:28:06 should be able to run it somewhere else though Apr 15 17:28:15 and have ovs connect to it over tcp Apr 15 17:32:18 smells like breakage though when that remove server is off Apr 15 17:32:22 remote* Apr 15 17:37:45 mirko * r31292 /feeds/xorg/lib/qt4/ (5 files in 3 dirs): Apr 15 17:37:46 [feeds/xorg/qt4] upgrade Qt version 4.7.0 -> 4.8.1 Apr 15 17:37:46 - remove libc checks since by now every libc available in OpenWrt has full nptl support Apr 15 17:37:46 - add '-fpermissive' to CFLAGS to compile with GCC 4.7.0 Apr 15 17:37:46 - package qt4-xmlpatterns Apr 15 17:37:46 - remove patches which went upstream meanwhile Apr 15 17:45:35 the whole regulatory stuff really doesn't like > 1 cards :( Apr 15 17:46:03 the whole regulatory stuff is a mess Apr 15 17:46:34 its design is quite flawed Apr 15 17:47:22 I just ran sh -x /sbin/wifi , and it runs "iw reg set BE" several times, but iw reg get keeps giving country:00 Apr 15 17:47:54 try enabling CONFIG_ATH_USER_REGD in your openwrt .config Apr 15 17:48:22 nbd: is enabled. on my 2nd rspro with only 1 card it gives country:BE Apr 15 17:48:29 weird Apr 15 17:48:41 try setting the country code only for one card (the first one) Apr 15 17:48:51 tried it too. will try for the other one Apr 15 17:48:54 and give it another reboot just to be sure Apr 15 17:55:14 that seems to have worked. Apr 15 17:55:19 nbd: In #10764 https://dev.openwrt.org/ticket/10764 did I rework the provided patch. Now it applies against SVN trunk. What else needs to be done to get it in SVN trunk? Apr 15 18:11:20 maybe [florian] or KanjiMonster can review it, i don't work on the brcm63xx target Apr 15 18:12:03 <[florian]> stappers: you are missing the patch for 3.1, 3.2 and 3.3 Apr 15 18:14:53 <[florian]> otherwise, the patch looks good Apr 15 18:17:43 build #165 of s3c24xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/165 Apr 15 18:19:35 nbd: heh, I reverted https://dev.openwrt.org/changeset/28387 and now works fine again when specifying country for both interfaces Apr 15 18:19:50 well I didn't revert it, I just commented the "iw reg set .." line Apr 15 18:19:52 odd Apr 15 18:20:25 yeah weird, looks like in this case it tries to set it too early ? Apr 15 18:28:56 would it make sence to have a "common" section or something in /etc/config/wireless, with country there, and do the iw reg stuff before starting to configure the interface(s) Apr 15 18:29:15 sense? Apr 15 18:29:17 sigh Apr 15 18:34:48 gonna add a second card in my 2nd rspro, so I can test there without interrupting my internet connection Apr 15 18:38:19 hope I can reproduce it with 1 ath5k and 1 ath9k card though Apr 15 18:44:20 #10764 missing 3.1, 3.2 and 3.3 otherwise, the patch looks good Apr 15 18:44:59 where is defined which kernel to use? ( /me didn't find it in `make menuconfig` ) Apr 15 18:46:02 <[florian]> in target/linux/brcm63xx/Makefile Apr 15 18:48:37 of course it works fine with 1 ath5k and 1 ath9k card as well Apr 15 18:48:47 le sigh Apr 15 19:01:01 [florian]: Acknowledge Apr 15 19:03:05 stintel: the only way to be sure is to nuke the entire regulatory handling from orbit :) Apr 15 19:10:23 aye Apr 15 19:11:20 maybe it would help to loop over the iw reg get | grep -q "^country $country:" thing a few times, Apr 15 19:11:26 but that's ugly Apr 15 19:11:36 <[florian]> stappers: please don't put irc conversation on trac, it is not useful Apr 15 19:12:50 [florian]: ack. Where else to store such notes? Apr 15 19:13:02 <[florian]> you don't have to, just keep them for you Apr 15 19:13:13 <[florian]> people hacking on openwrt usually know about this detail Apr 15 19:13:48 * stappers admits that he is is new to the OpenWRT world Apr 15 19:24:26 florian * r31293 /trunk/target/linux/generic/patches-2.6.32/033-x86_unused_variables.patch: [generic/2.6.32] silence an annoying x86 warning Apr 15 19:24:29 florian * r31294 /trunk/target/linux/rdc/patches-2.6.32/015-r6040_fix_multicast.patch: [rdc] refresh 2.6.32 patches Apr 15 19:24:33 florian * r31295 /trunk/target/linux/rb532/patches-2.6.32/001-cmdline_hack.patch: [rb532] refresh 2.6.32 patches Apr 15 19:24:35 florian * r31296 /trunk/target/linux/rb532/patches-2.6.32/003-unused_variables.patch: [rb532] mark set but unused variables as such Apr 15 19:31:14 [florian]: is refreshing kernel patches manual craftsperson work, a program or both? Apr 15 19:32:34 <[florian]> you can do it automatically using make target/linux/refresh Apr 15 19:41:50 * stappers makes a note about "make target/linux/refresh" and leaves keyboard Apr 15 19:50:47 mirko * r31297 /packages/libs/boost/Makefile: (log message trimmed) Apr 15 19:50:47 [packages/boost] level up, fix several issues Apr 15 19:50:47 - level up to version 1.49.0 Apr 15 19:50:47 - package boost-chrono Apr 15 19:50:47 - package boost-locale Apr 15 19:50:48 - package boost-random Apr 15 19:50:48 - package boost-timer Apr 15 20:16:27 mhei * r31298 /packages/lang/php5/Makefile: [packages] php5: build/enable shmop module Apr 15 20:26:43 mhei * r31299 /packages/lang/php5/Makefile: Apr 15 20:26:43 [packages] php5: fix missing dependency for http module (fixes #11257) Apr 15 20:26:43 Thanks ryzhov_al for spotting this out. Apr 15 21:12:12 hm, why did the three patches I just sent not show up as superseding my old versions, in patchwork? What'd I do wrong? Apr 15 21:12:31 OK, the third one had a slightly different subject but at least the first two (2/4 and 3/4) should have been OK, shouldn't they? Apr 15 21:15:48 build #123 of etrax is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/etrax/builds/123 Apr 15 21:34:46 jow_laptop: why are the unversioned shared libs symlinks getting added (back?) in r31286 & r31287? Apr 15 21:45:55 updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/index.html Apr 16 01:08:20 dwmw2_gone: patchwork isn't that smart... Apr 16 01:08:44 swalker: uhm no idea, I supposed they are needed somehow, why else would someone make the effort to change it? **** ENDING LOGGING AT Mon Apr 16 02:59:58 2012