**** BEGIN LOGGING AT Mon Nov 04 02:59:58 2013 Nov 04 08:46:05 <_trine> luka, are you available? Nov 04 09:27:04 <_trine> This is a pastebin of my dockstar booting in serial. This is a new check out and is still at all the default values http://pastebin.com/p0qP0gAL Nov 04 10:18:31 nbd: took a while, but just had a disconnect again on the Buffalo WZR-HP-G300NH running AA + aa-mac80211, r38462. now when trying to reconnect I had 8 times this message on the client: [856385.231361] wlan0: 4c:e6:76:3f:09:4d denied authentication (status 14) Nov 04 10:19:11 ping jow_laptop related to a firewall logging spam - https://dev.openwrt.org/ticket/14405 Nov 04 10:20:49 nbd: the usual in dmesg, http://pastebin.com/hfiWNMz1 Nov 04 10:24:01 and here the stuff /sys/kernel/debug/ieee80211/phy0/ath9k Nov 04 10:24:03 http://pastebin.com/d7FQSPKj Nov 04 10:30:25 hm, denied authentication due to MIC failure Nov 04 10:31:04 do you have TKIP enabled or something? Nov 04 10:31:24 please show me /var/run/hostapd-phy0.conf Nov 04 10:38:40 nbd: http://pastebin.com/SjJZfAF6 Nov 04 10:38:43 TKIP is indeed enabled Nov 04 10:39:08 ah, you're using wpa1/2 mixed Nov 04 10:39:12 try setting encryption to psk2 Nov 04 10:39:42 it should make it more stable Nov 04 10:39:54 I'd like to keep it for compatibility (AP @ work ) Nov 04 10:40:01 but I can force it in wpa_supplicant I guess Nov 04 10:41:43 or set it to mixed+ccmp Nov 04 10:41:50 that way it'll still do wpa1/2 Nov 04 10:42:25 but keeping tkip around should be unnecessary Nov 04 10:42:37 chipsets that don't do aes/ccmp are very rare Nov 04 10:45:49 build #359 of kirkwood is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/kirkwood/builds/359 Nov 04 10:57:56 dape: ponging that back to cyrusff Nov 04 10:58:16 alrighty, i will ping him later, thanks Nov 04 10:59:04 dape: thats due to the new ifupdate event which fires all the time, we should probably add some filtering so that fw3 only reacts on update if something relevant changed Nov 04 11:01:26 got it, my ipv6 /128 and /64 pd have infinite lifetime anyway so as long as the pppoe-wan is up.. Nov 04 11:02:39 nbd: looks like an apple device is having trouble connecting now, this keeps appearing: Nov 4 12:02:20 OpenWrt daemon.info hostapd: wlan0: STA e4:ce:8f:20:6d:1e IEEE 802.11: authenticated Nov 04 11:03:36 my mba with 10.9 connects fine though Nov 04 11:05:23 might be coincidence, I'll check after lunch Nov 04 12:16:14 dape: yeah, it's probably caused by your ISP sending many RAs Nov 04 12:16:50 i filter these down to once per minute Nov 04 12:16:52 cyrusff, thanks for responding, actually the ISP has the tdhcp in place and now RAs Nov 04 12:17:04 could be rogue RAs though Nov 04 12:17:11 dape: there must be RAs Nov 04 12:17:18 dhcpv6 doesn't setup any routing Nov 04 12:17:30 can something be done for this? Nov 04 12:17:52 well yes, we can optimize the firewall logic Nov 04 12:17:57 awesome! Nov 04 12:18:09 the thing is, its probably mostly irrelevant and only the logmessage itself is pretty annoying Nov 04 12:18:17 (so i should install radvd dump and check for ras) Nov 04 12:18:26 yes, it is.. Nov 04 12:23:35 cyrusff, i only see this - https://pastee.org/r9t3g Nov 04 12:26:09 dape: looks like RAs Nov 04 12:26:16 how often do they arrive? Nov 04 12:26:28 seems like every 10 or so seconds Nov 04 12:26:47 ok Nov 04 12:27:12 unnecessarily high given the big AdvDefaultLifetime but yeah, comcast is worse with every 3 seconds Nov 04 12:29:14 correction: in between 40 and 50 seconds Nov 04 12:29:55 cyrusff: wouldn't it be enough to issue ifupdate only when something actually changed? Nov 04 12:30:01 yes Nov 04 12:30:24 or maybe keep the current sematics and add more env vars that indicate the kind of change Nov 04 12:30:41 e.g. CHANGE_LINK=0, CHANGE_ADDR=1 Nov 04 12:30:46 or something like that Nov 04 12:31:00 the problem is there is always a change as the address and routing timers are always reset Nov 04 12:31:07 but we could ignore timer changes i think Nov 04 12:31:11 that would be CHANGE_VALIDITY=1 Nov 04 12:31:39 but yeah, filter timer changes makes probably sense Nov 04 12:31:55 the common use case for hotplug is dealing with configuration changes Nov 04 12:32:01 mainly link down, up or ip changed Nov 04 12:32:28 yeah ifupdate was intenden mostly for ip changed / prefix changed / some other config data changed Nov 04 12:32:36 not so much for every RA that only sets timers Nov 04 12:32:58 i have to think about how we can use the AVLs version things to detect if there was an actual change or so Nov 04 13:31:08 Hauke: no, probably removed it because none of the devices I used for testing needed it (and one required some other stuff) Nov 04 13:32:07 KanjiMonster: _trine: is using a device with the root fs on a hard disk using ext2 Nov 04 13:33:10 the only devices I tested on were harddisk-less routers Nov 04 13:34:08 he used some nas device Nov 04 13:34:30 okay? Nov 04 13:34:58 this: http://wiki.openwrt.org/toh/seagate/dockstar Nov 04 13:35:44 for all kirkwood things you should mainly ask luka Nov 04 13:35:52 I do not have any of these devices I just saw that kernel 3.3 had ext4 support and 3.10 does not have Nov 04 13:36:08 luka: ping Nov 04 14:06:34 Hauke: kernel 3.10 has introduced a whole lot more new features than kernel 3.3. However, the new kernel 3.10 fails to enable the CONFIG_MACH_DOCKSTAR and among other Seagate FreeAgent devices as shown here (https://forum.openwrt.org/viewtopic.php?pid=216702#p216702) or here (http://pastebin.com/D32aStqF - see line #147 and #152 for CONFIG_MACH_DOCKSTAR). Nov 04 14:09:48 Hauke: That explains why some Seagate Dockstar and/or PogoPlug users complained about the new kernel 3.10 does not boot. Nov 04 14:10:05 mazilo: yes that explains that all Nov 04 14:10:14 could you send a patch? Nov 04 14:11:33 Hauke: There is no patch needed. Just enable the new kernel 3.10 with features found on the old kernel 3.3. Nov 04 14:14:06 Hauke: OTOH, both kernel 3.3 and 3.10 also enable some networking stuffs, i.e. NAT, etc., which a Seagate DockStar, GoFLEX HOME/NET, etc., device don't use. They just waste RAM. Perhaps, these should only be enabled by OpenWRT menuconfig for the specific platforms. Nov 04 14:15:19 mazilo: I haven't looked into how these kernels are beeing build, but often we use one kernel for all devices with the same SoC, so deactivating nat would also affect routers Nov 04 14:16:46 mazilo: there is no CONFIG_MACH_DOCKSTAR anymore in 3.10 Nov 04 14:18:57 at least there wasn't in my tree Nov 04 14:20:20 nbd: I'm getting a hung dhcpc child process because of a hung write, netifd detaches stdin/stdout/stderr and never returns the write call for some reason. Can the code cause a deadlock or infinite loop causing this ? Nov 04 14:20:33 mazilo: even if you should use an kernel with appended dtb, not with the legacy board code Nov 04 14:23:08 mazilo: and I plan to add support for a kirkwood based router soonish, so disabling NAT would be a no-go Nov 04 14:23:38 also I never saw a kirkwood device with less than 128 MiB RAM, so the memory usage is sort of a non issue ;) Nov 04 14:32:56 KanjiMonster: I just pulled target/linux/kirkwood/config-3.10 from OpenWRT SVN trunk and CONFIG_MACH_DOCKSTAR (line #142) is still there, except it isn't set. Nov 04 14:33:45 mazilo: yeah, saw that the old legacy board code is still present. But you should use the devicetree setup variant, not the old board code Nov 04 14:34:11 the board code will lilkely be removed soon Nov 04 14:34:49 it actually is already removed Nov 04 14:35:44 KanjiMonster: Which device tree variant should I enable for a Seagate Dockstar and/or GOFLEX NET/HOME? Nov 04 14:37:14 merbanan: details on openwrt version, platform, device, etc. please Nov 04 14:37:34 mazilo: there's nothing to enable, just use kirkwood-goflexnet.dtb Nov 04 14:38:10 KanjiMonster: Where is kirkwood-goflexnet.dtb? Nov 04 14:39:01 $(LINUX_DIR)/arch/arm/boot/dts/ Nov 04 14:39:08 well, thre is the .dts Nov 04 14:39:19 which will be compiled into the dtb Nov 04 14:39:28 nbd: is there a place I can send a bug report ? Nov 04 14:40:12 nbd: it is kind of hard to replicate the issue, so I suspect a race Nov 04 14:40:21 KanjiMonster: Yes, but how do I enable that? Nov 04 14:41:21 merbanan: before you make a detailed bug report, i'd like to know what openwrt version you're running Nov 04 14:43:28 nbd: we are running git checkout 2674941b06c1ec67f1aff1bff9212e1372106641 Nov 04 14:44:39 what tree does that hash refer to? Nov 04 14:44:40 the box that hangs has 2 dhcpc clients while one of them works Nov 04 14:45:27 ah, seems to be a netifd commit Nov 04 14:45:36 but i was actually asking for the openwrt revision Nov 04 14:45:40 git://nbd.name/luci2/netifd.git Nov 04 14:45:53 our own custom Nov 04 14:46:20 please update to latest versions of at least libubox and netifd (probably need to update ubus as well) Nov 04 14:46:33 the bug you're describing could be something that i already fixed Nov 04 14:46:43 might be an issue in ustream (part of libubox) Nov 04 14:46:51 ok, thanks will try Nov 04 14:46:51 or might be one of the netifd issues that i fixed Nov 04 14:49:01 mazilo: you just need to make the same calls as ib62x0, just with goflexnet instead of ib62x0 in kirkwood/image/Makefile Nov 04 14:50:21 KanjiMonster: How? I just couldn't find any option on 'make menuconfig' nor 'make kernel_menuconfig' to do that. Nov 04 14:50:57 mazilo: there is neither a menuconfig nor kernel menuconfig option for it Nov 04 14:51:41 Hauke: pong Nov 04 14:51:42 all kirkwood-*.dts files in arch/arm/boot/dts/ are always built regardless of any menuconfig options Nov 04 14:52:02 KanjiMonster: Then, the problem still persists that uImage compiled for a Marvel Kirkwood won't boot on a Seagate Dockstar. :( Nov 04 14:52:37 mazilo: have you tested an uImage with an dtb appended kernel? Nov 04 14:52:47 luka: some people have problems with kernel 3.10 on the Kirkwood target Nov 04 14:52:53 KanjiMonster: The question is how do I do that? Nov 04 14:53:12 mazilo: the way I just told you Nov 04 14:53:31 KanjiMonster: Unfortunately, I don't see how to do that. :( Nov 04 14:53:41 you need to *edit* the kirkwood/image/Makefile Nov 04 14:53:47 Hauke: i did not port all boards from 3.10 Nov 04 14:54:15 Hauke: mostly because i couldnt test the ubi(fs) stuff on new boards Nov 04 14:54:36 KanjiMonster: Is this 'KERNELNAME:="zImage dtbs"' the one I need to change? Nov 04 14:54:43 luka: for _trine it booted, but ext4 support was missing in the kernel Nov 04 14:55:41 Hauke: i dont know if it was _trine but somebody said they have a board which has only 512K and they need to boot from hard disk Nov 04 14:55:42 mazilo: is this from kirkwood/image/Makefile? No? Then you don't need to change that. Nov 04 14:55:45 Hauke: I believe _trine also enabled the CONFIG_MACH_DOCKSTAR. Nov 04 14:56:24 KanjiMonster: Yes. Then, it probably won't work for a Seagate Dockstar. Nov 04 14:56:33 Hauke: so they need would need ext4 build in, others can simply use a module Nov 04 14:57:01 Hauke: if their rootfs is on nand of course Nov 04 14:57:04 mazilo: No it isn't FFS you are looking at kirkkwod/Makefile, not kirkwood/*image*/Makefile Nov 04 14:58:21 KanjiMonster: So, which one? Nov 04 14:59:13 mazilo: as I wrote the last billion times, you need to edit kirkwood/*image*/Makefile Nov 04 14:59:44 KanjiMonster: OK, I am looking at it, but what do I need to change? Nov 04 14:59:59 I already told you Nov 04 15:00:08 look in your backlog Nov 04 15:00:30 KanjiMonster: You probably did tell me, but I don't understand what you said. :( Nov 04 15:02:41 KanjiMonster: Is this (http://pastebin.com/eMZae4YV on line #36 and #37)? Nov 04 15:03:48 [15:49:01] mazilo: you just need to make the same calls as ib62x0, just with goflexnet instead of ib62x0 in kirkwood/image/Makefile Nov 04 15:10:38 KanjiMonster: Is there a way to automate this procedure for everyone? Nov 04 15:12:15 mazilo: as I said, do it like for ib62x0 or pogo_e02 (and that means add the appropriate lines, not replace it), and then you should get a patch out of it Nov 04 15:12:21 then you can submit it Nov 04 15:12:50 then it's automated or everyone ;D Nov 04 15:12:57 KanjiMonster: I will let you guys do that. Nov 04 15:13:32 well then I can't help you Nov 04 15:14:33 KanjiMonster: You sure did already help me. But, I am just lost somewhere, unfortunately, :( Nov 04 17:00:24 nbd: I was reading a thread about qmi here http://patchwork.openwrt.org/patch/3402/ - what is the status of qmi support in owrt trunk? Nov 04 17:21:09 <_trine> Hauke, the dockstar boots now Nov 04 17:21:41 <_trine> I do get this message at the endo of the boot process though Nov 04 17:21:46 <_trine> root@OpenWrt:/# [ 54.114714] [sched_delayed] sched: RT throttling activated Nov 04 18:38:42 juhosg r38642 trunk/target/linux/ x86/generic/config-3.10 generic/config-3.8 generic/config-3.12 generic/config-3.10 * kernel: refresh kernel configs with kconfig.pl Nov 04 18:38:44 juhosg r38643 trunk/target/linux/ (16 files in 2 dirs) * kernel/3.1[02]: move MTD root device setup code to mtdcore Nov 04 18:38:45 juhosg r38644 trunk/target/linux/ar71xx/image/Makefile * ar71xx: image: dd root.* images in fs specific macros Nov 04 18:38:46 juhosg r38645 trunk/target/linux/ar71xx/config-3.10 * ar71xx: remove yaffs from the generic rootfstype list Nov 04 18:38:47 juhosg r38646 trunk/target/linux/ ar71xx/nand/target.mk ar71xx/Makefile ar71xx/nand ar71xx/nand/config-default * ar71xx: Add new subtarget nand for WNDR4300 Nov 04 18:38:50 juhosg r38647 trunk/target/linux/ ar71xx/nand/profiles ar71xx/nand/profiles/netgear.mk ar71xx/generic/profiles/netgear.mk ar71xx/image/Makefile * ar71xx: move WNDR4300 profile to the nand subtarget Nov 04 20:19:35 build #411 of uml is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/411 Nov 04 20:22:57 build #399 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/399 Nov 04 21:01:44 I may have managed to update ar71xx to linux 3.12. However patch line numbers are wildly off. Is there an automated way to fix them? Nov 04 21:03:43 make target/linux/refresh Nov 04 21:07:15 Done. Nov 04 22:12:24 is sysupgrade possible when booted from a ramdisk? Nov 04 22:42:35 jow r38648 trunk/package/network/ services/dnsmasq/files/dnsmasq.init services/dnsmasq/Makefile * dnsmasq: fix backward compatibility with existing configs Nov 04 23:33:02 build #432 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/432 **** BEGIN LOGGING AT Tue Nov 05 00:58:43 2013 **** ENDING LOGGING AT Tue Nov 05 02:59:58 2013