**** BEGIN LOGGING AT Wed Jun 20 02:59:59 2012 Jun 20 06:51:36 nbd: I'm "losing" my WGT with netifd, the main router reports after some time: DHCPNAK(br-lan) 192.168.1.39 00:0f:b5:0b:a7:c0 lease not found Jun 20 06:52:17 nbd: it's the snapshot build from yesterday, which is not the latest revision though Jun 20 07:48:49 blogic * r32466 /trunk/rules.mk: add and export HOSTCC_NOCACHE Jun 20 07:48:56 luka12345: ^^^^ Jun 20 07:52:56 blogic: thx Jun 20 12:09:47 build #3 of adm5120 is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/3 Jun 20 12:09:52 build #3 of octeon is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/3 Jun 20 12:10:09 build #3 of mpc52xx is complete: Failure [failed compile_2] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/3 Jun 20 13:10:12 build #6 of s3c24xx is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/6 Jun 20 13:42:35 build #3 of ar71xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ar71xx/builds/3 Jun 20 13:45:06 nbd * r32467 /trunk/package/ (4 files in 3 dirs): toolchain: move eglibc-files from base-files to toolchain, it is packaged as part of the libc package Jun 20 15:03:01 build #3 of au1000 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/au1000/builds/3 Jun 20 15:46:46 build #3 of iop32x is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/3 Jun 20 15:48:21 if I've got some feeds in my own github repo, what's the nicest/best way of submitting them ot OpenWRT packages? copy the latest and greatest into my own packages checkout and submit it as just a single diff? Jun 20 15:48:44 that way makes the patch just one single patch, and you don't have to get any development work that was done in my own packaging repo? Jun 20 15:49:05 presumably no-one cares about that? Jun 20 15:49:29 karlp: send 1 patch per change/addition/deletion/... Jun 20 15:49:45 properly annotated and with a commit messagre that describes that specific patch Jun 20 15:49:51 then mail them to the -devel ML Jun 20 15:50:18 build #6 of orion is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/6 Jun 20 15:50:58 so yes, ignore completely any string of commits that led to the final version in my own repository, and just make a single patch against owrt Jun 20 15:53:26 yes Jun 20 15:53:45 although i am sure its all interesting stuff... we dont care what it was like last week :-) Jun 20 15:53:54 karlp: got a link ? Jun 20 15:54:07 we have been considering to allow people to setup staging trees that we can oull from Jun 20 15:54:42 https://github.com/karlp/owrt_pub_feeds Jun 20 15:54:56 some new, some old. Jun 20 15:55:14 I understand not wanting my history in owrt, Jun 20 15:55:26 just not sure how to get owrt back itno my history for a future request. Jun 20 15:55:40 I guess I just manually copy over the latest into my owrt tree each time to generate the diffs Jun 20 15:55:45 workable enough. Jun 20 15:56:37 rebase ? Jun 20 15:56:50 i mean ... git rebase is your friend Jun 20 15:56:53 yeah, in theory. Jun 20 15:57:08 ;) Jun 20 16:26:52 nbd: happens with latest trunk as well, this is my config: http://pastebin.com/7HBRZtG6 Jun 20 16:30:03 loswillios: any specific firewall rules? Jun 20 16:32:46 I am fleshing out more stuff in the rb-751g support, so far, I've got it up and going with a squashfs, and overlays working Jun 20 16:33:02 now I'm hunting in startup code to do some fixups in mac addresses Jun 20 16:33:30 struggling to understand the creation of /etc/config/network on a fresh boot Jun 20 16:34:01 I need to insert a couple more bits into the default setup to change a mac address Jun 20 16:34:07 so that wifi and eth0 are not on the same mac Jun 20 16:35:12 which script actually kicks off the creation of /etc/config/network if one doesn't exist on the first startup ? Jun 20 16:38:42 /rom/etc/uci-defaults/network Jun 20 16:39:01 thanks Jun 20 16:39:44 if I can get this one done, then I think I have everything covered Jun 20 16:39:58 so that the rb-751g will work like any other device Jun 20 16:40:05 you want to fix the macs in the network config? Jun 20 16:40:13 in terms of squash+jffs and a functional sysupgrade etc Jun 20 16:40:26 ya, as this one comes up, it sets the same mac address for wifi, and ethernet Jun 20 16:40:40 that should be solved in the board setup imo Jun 20 16:40:47 as it comes from microtik, it actually uses 6 of them Jun 20 16:40:53 one for each port, and one for the wifi Jun 20 16:41:22 ok, so you think it's better to go do it in the drivers themselves Jun 20 16:41:34 board starts up with a kmac paramater in kernel parms, fromt he boot loader Jun 20 16:41:38 and the drivers parse that Jun 20 16:41:52 I'm just trying to get the ethernet and wifi onto different macs Jun 20 16:42:17 as it comes up right now, wan, lan, and wifi, all have the same mac on them Jun 20 16:42:59 all matching the kernel command line kmac parameter Jun 20 16:43:43 for the wndr3700 I solved it like this: http://luci.subsignal.org/~jow/ar71xx-wndr3700-locally-administered-eth0mac.patch Jun 20 16:44:50 what I am looking at for this one, looking at the bottom of the box itself Jun 20 16:44:59 it has a range of 6 consecutive macs assigned Jun 20 16:45:11 the first one of that range is the kmac parameter it gets at boot Jun 20 16:45:27 so, what I was thinking, best way would be to take that, and increment one to set the eth0 mac Jun 20 16:45:33 yes Jun 20 16:45:50 and then call ar71xx_init_mac(...) for each calculated one Jun 20 16:45:51 but i was thinking of doign it in startup scripts, rather than the drivers Jun 20 16:45:56 easy enough to do in drivers too Jun 20 16:45:59 its not in the drivers Jun 20 16:46:04 it is in the board setup code Jun 20 16:46:13 yah, that's what i meant Jun 20 16:46:21 i have the file up Jun 20 16:46:33 now, i have another question if you have a sec Jun 20 16:46:41 this one works like this, for flash layout Jun 20 16:46:50 yaffs partion with file called kernel in it Jun 20 16:46:55 then following that, rootfs Jun 20 16:47:06 the upgrade procedure, put a new kernel elf file into the yaffs Jun 20 16:47:13 then, mtd write the squash into rootfs Jun 20 16:47:17 that works, got it running Jun 20 16:47:35 so the question becomes, is there already a 'standard' way for packaging these for sysupgrade ? Jun 20 16:47:47 you can use the combined image format Jun 20 16:47:48 what I did to test it all out here, is just put kernel and rootfs into a tgz Jun 20 16:47:57 scripts/combined-image.sh Jun 20 16:48:09 then I added a script to use that in the sysupgrade stuff Jun 20 16:48:12 and see /lib/upgrade/platform.sh on the router to see how it is handled Jun 20 16:48:13 and, got it all working Jun 20 16:48:30 I did, but, it's not handled int he code in trunk for this one Jun 20 16:48:39 the code in trunk was set up for yaffs on both file systems Jun 20 16:48:50 yeah, so add a new case Jun 20 16:48:50 and didn't support sysupgrade at all Jun 20 16:49:02 what I did was get it all going same as the rest of them Jun 20 16:49:12 yeah Jun 20 16:49:16 with squashfs that does the rootfs_data trick Jun 20 16:49:24 and it's working nicely, like this router with this setup Jun 20 16:49:39 and, boot loader with give a 'reset defaults' parameter if you hold in button while starting Jun 20 16:49:59 so, i'm parsing for that in pre-init, and mtd erase rootfs_data if you start with the button in Jun 20 16:50:10 to get a nice clean 'reset to defaults' Jun 20 16:50:39 with all of this in place, 64 meg ram, 64 meg flash, ar71xx platform Jun 20 16:50:47 with gigabit ports, it's actually quite a nice little router now Jun 20 16:51:38 cool Jun 20 16:51:49 I think it is best if you mail juhosg to discuss the mac setup Jun 20 16:51:54 buy em locally from retail for a hundred bucs Jun 20 16:52:35 probly be chinese knock-offs available for half that Jun 20 16:53:08 the other nice thing, if you hold the button long enough during a power up, it'll netboot by default Jun 20 16:53:17 so, very easy to recover from a bad ooops Jun 20 16:53:56 but lack of a serial port, makes it a difficult one to deal with initial startup stuff Jun 20 17:09:53 acinonyx * r32468 /packages/net/lispmob/ (. Makefile patches/ patches/010-disable-modprobe.patch): [packages] lispmob: New package Jun 20 17:40:13 jow_laptop: none Jun 20 17:40:42 jow_laptop: AFAICS the device gets lost after the first dhcp renew Jun 20 17:40:57 asked because of https://dev.openwrt.org/ticket/4108 Jun 20 17:41:57 ah, hm. Jun 20 17:43:15 well no, this started happening with a netifd image Jun 20 17:45:23 whats the lease time? Jun 20 17:47:10 Jun 20 11:50:56 OpenWrt daemon.info dnsmasq-dhcp[10067]: DHCPACK(br-lan) 192.168.1.39 00:0f:b5:0b:b2:10 wgt Jun 20 17:47:16 Jun 20 17:50:57 OpenWrt daemon.info dnsmasq-dhcp[10067]: DHCPNAK(br-lan) 192.168.1.39 00:0f:b5:0b:a7:c0 lease not found Jun 20 17:49:11 ah, now I see it Jun 20 17:49:15 huh, different mac address there? Jun 20 17:49:30 the bridge has a different mac-address than eth0 Jun 20 17:50:10 yup, the device has a different IP now Jun 20 17:50:35 is that intentional (different MAC)? Jun 20 17:50:40 it seems another dhcp server NACKs your refresh attempt Jun 20 17:50:48 :) Jun 20 17:52:07 I guess the first dhcp request comes from 00:0f:b5:0b:b2:10 (eth0), then the wifi bridge kicks in or something, and the next renew is from 00:0f:b5:0b:a7:c0 (br-lan) Jun 20 17:55:38 and 00:0F:B5:0B:A7:C0 is the MAC from the wifi card Jun 20 17:56:15 nbd: can you reproduce it with http://pastebin.com/7HBRZtG6 ? Jun 20 17:57:12 it's a dumb switch with wifi in a bridge Jun 20 17:57:56 thats the linux bridge code Jun 20 17:58:09 the bridge assumes the lowest mac of all its member ifaces Jun 20 17:58:24 if the wifi mac < lan mac it will swap when wifi is brought up Jun 20 17:59:12 ah so actually the kernel needs patching. changing the mac address of an already up interface is never a good idea Jun 20 17:59:18 imho Jun 20 17:59:26 i did add some code to netifd to fix up the bridge mac Jun 20 17:59:30 maybe that part isn't working Jun 20 17:59:33 i'll look into it Jun 20 17:59:38 thanks Jun 20 18:10:48 seeing a crash&burn when building trunk after an update yesterday... target is x86/alix: http://fpaste.org/BLHy/ Jun 20 18:11:14 complains about cpuid.h when building eglibc Jun 20 18:11:22 is this a known issue? Jun 20 18:12:14 check config.log Jun 20 18:13:12 probably just some missing flag Jun 20 18:13:20 like -std=gnu99 or something Jun 20 18:28:27 ok... not sure why it stopped working. nothing has changed at my end. Jun 20 18:29:19 configure:3005: i486-openwrt-linux-gnu-gcc -std -c -march=geode -Os -mmmx -m3dnow -fno-align-jumps -fno-align-functions -fno-align-labels -fno-align-loops -pipe -fomit-frame-pointer -fhonour-copts -Wno-error=unused-but-set-variable conftest.c >&5 Jun 20 18:29:21 cc1: error: unrecognized command line option '-std' Jun 20 18:29:32 hmmm..... where is that coming from? Jun 20 18:30:37 you'd think it would bail rather than continuing. Jun 20 18:35:01 it bails Jun 20 18:35:10 configure: WARNING: cpuid.h: present but cannot be compiled Jun 20 18:35:46 no, I mean sooner. if the args to the compiler itself aren't understood, that's a separate issue than a test stub failing to compile. Jun 20 18:36:43 you should be able to compile /dev/null with no problems. if not, then the args being generated themselves are bogus. Jun 20 18:44:58 jow_laptop: I can't tell where the -std is creeping in from. Jun 20 18:45:12 me neither Jun 20 18:49:16 I searched all recent commit mails and nothing comes up either. Jun 20 18:49:24 so it must be in the package itself. Jun 20 19:00:18 hmm... staging_dir/host/share/autoconf/autoconf/c.m4 Jun 20 19:02:19 do we build our own autoconf, and if so when did it last change? Jun 20 19:04:38 mirko * r32469 /trunk/target/linux/xburst/Makefile: [target/xburst] verified to work - remove broken flag Jun 20 19:41:48 jow thanks for the tips earlier, I got this working now, the way I want / need for mac addresses, just have 'clean up the patches' left Jun 20 21:57:48 florian * r32470 /trunk/target/linux/mcs814x/ (3 files in 3 dirs): Jun 20 21:57:48 [mcs814x] do not use MULTI_IRQ_HANDLER it is bogus on our platform Jun 20 21:57:48 This caused stalls in the Ethernet DMA block, so until properly Jun 20 21:57:48 written and sorted out, fallback to the assembly version instead. Jun 20 21:57:50 florian * r32471 /trunk/target/linux/mcs814x/files-3.3/drivers/net/ethernet/mcs8140/nuport_mac.c: [mcs814x] improve ethernet driver debugging Jun 20 21:57:54 florian * r32472 /trunk/target/linux/mcs814x/files-3.3/drivers/net/ethernet/mcs8140/nuport_mac.c: [mcs814x] remove uselss cache invalidations Jun 20 21:57:57 florian * r32473 /trunk/target/linux/mcs814x/base-files/etc/uci-defaults/leds: [mcs814x] use generic uci defaults functions Jun 20 21:58:06 florian * r32474 /trunk/target/linux/mcs814x/ (Makefile base-files/etc/uci-defaults/leds): [mcs814x] fix USB led triggers Jun 20 22:07:10 luka * r32475 /packages/net/dhcp/ (4 files in 3 dirs): [packages] dhcp: remove package, it's replaced by dhcp4 Jun 20 22:07:11 luka * r32476 /packages/net/ (21 files in 6 dirs): [packages] dhcp4: update to 4.2.4 and rename to dhcp Jun 20 22:12:20 luka * r32477 /packages/net/dhcp4/: [packages] dhcp4: remove empty folders Jun 20 22:40:18 luka * r32478 /packages/net/dhcp-forwarder/ (Makefile files/dhcp-fwd.init): [packages] dhcp-forwarder: update to 0.10 Jun 20 23:06:05 swalker * r32479 /packages/net/bind/Makefile: [packages] bind: update to 9.9.1-P1 (CVE-2012-1667) Jun 20 23:07:30 swalker * r32480 /packages/utils/bluelog/Makefile: [packages] bluelog: update to 1.0.4, remove unused files from bluelog-live, thanks Tom Nardi Jun 20 23:54:11 build #3 of etrax is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/etrax/builds/3 Jun 21 00:13:10 build #3 of ep93xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/3 Jun 21 01:03:21 build #5 of avr32 is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/5 **** ENDING LOGGING AT Thu Jun 21 02:59:58 2012