**** BEGIN LOGGING AT Tue Nov 19 03:00:00 2013 Nov 19 03:54:13 Today, I flashed my FON2100 with the snapshot and it boots to get a DHCP address, but it can't be pinged nor telnet to. Does anyone know what is going on? Nov 19 03:54:39 It sounded like netifd is broken. Nov 19 04:00:27 mazilo: i flashed my FON2201 with trunk a couple days ago and it was fine, except this: https://lists.openwrt.org/pipermail/openwrt-devel/2013-November/022349.html Nov 19 04:02:05 DonkeyHotei: Can you please tell me what SVN revision you have flashed to your FON2100? This way, I can build mine using the same revision. Nov 19 04:03:42 DonkeyHotei: Do you know if it is possible to turn a FON2100 into an SPI programmer through its GPIO port using flashrom software? Nov 19 04:03:44 r38705 Nov 19 04:03:52 DonkeyHotei: Thank.s Nov 19 04:04:20 probably. Nov 19 04:04:47 DonkeyHotei: Thanks Nov 19 08:49:47 interesting, doku.php just barfed on me Nov 19 08:50:00 barfed how? Nov 19 08:50:14 directs me to www.ioerror.us Nov 19 08:50:26 a bad behaviour Nov 19 08:50:42 Your technical support key is: d863-d607-b40c-8ddc Nov 19 08:50:43 sou probably triggered the antispam heuristics Nov 19 08:51:00 most often it is related to cookie issues Nov 19 08:51:34 try clearing the cookie and log in again if the issue persists Nov 19 08:54:45 okay Nov 19 09:02:05 better Nov 19 09:02:08 thanks Nov 19 09:08:47 * russell-- encountered an ethernet run that suddenly started dropping 75% of packets, after a reboot it worked better for a little bit then started dropped again Nov 19 09:09:15 had been working fine for years, is a little longish, not sure the length Nov 19 09:10:13 the upstream end reports no errors on the interface, but would disable the interface in the bridge (although maybe that was me unplugging/rebooting) Nov 19 09:10:18 will look closer tomorrow Nov 19 09:11:09 i brought the end point home to test, didn't misbehave here Nov 19 09:11:35 end point is a wzr600dhp, upstream end is an alix2 Nov 19 12:09:28 blogic: i found a quick fix by just pointing the ubus socket to /tmp/run/ubus.socket. how about adding a /run tmpfs and just link /tmp/run to it . This it would be possible to nomal partitions for /tmp and /var and just let /var/run link to /run Nov 19 12:10:50 it would be possible to mount normal partitions for /tmp and /var and just let /var/run link to /run - the /run tmpfs aaproach is used in a lot of distributions and it would also allow to save ram when extroot is used Nov 19 13:02:30 nbd r38865 trunk/target/linux/ mpc85xx/patches-3.10/200-gianfar_napi_poll_revert.patch mpc85xx/patches-3.10/200-fix_gianfar_napi_poll.patch * mpc85xx: Fix NAPI poll mechanism in GIANFAR ethernet driver Nov 19 13:31:10 is there any way to directly write values to a register on using the ag71xx driver? like the ethreg tool from the atheros development kit Nov 19 13:31:12 see http://gpl.back2roots.org/source/fritzbox/7270_5.05/GPL-release_kernel/linux/drivers/net/ag7240/tools/ethreg.c Nov 19 14:08:19 build #410 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/410 Nov 19 14:09:32 tripolar: i iwll push a fix in a sec Nov 19 14:09:39 build #410 of ppc44x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/410 Nov 19 14:11:16 build #411 of x86 is complete: Failure [failed compile_1] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/411 Nov 19 14:58:26 build #380 of sibyte is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/380 Nov 19 15:34:44 blogic nice :) Nov 19 15:34:52 allready pushed? **** BEGIN LOGGING AT Tue Nov 19 15:44:20 2013 Nov 19 15:46:33 blogic: was https://lists.openwrt.org/pipermail/openwrt-devel/2013-November/022504.html OK? Nov 19 16:10:12 tripolar: is it in the timeline ? Nov 19 16:10:30 or shall i open my browser for you and opent he page and then proxy the content of the www to you via irc ? :) Nov 19 16:12:21 15:09:32 < blogic> tripolar: i iwll push a fix in a sec Nov 19 16:13:08 tripolar: why do oyu copy and paste that line ? Nov 19 16:13:16 it is unrelated to what i said Nov 19 16:13:55 tripolar: that fix i have only fixes part of your problem Nov 19 16:14:05 blogic: okay Nov 19 16:14:11 the rest of your problem will be fixed when you do your /tmp magic in the mount step Nov 19 16:14:24 you seem to be doing it post mount and post preinit Nov 19 16:14:29 this causes your problem Nov 19 16:14:38 ... hwich is that you delete the ubus socket at runtime Nov 19 16:14:43 and there is no way for us to detect this Nov 19 16:14:49 and we dont even want to detect it Nov 19 16:15:10 you need to make sure that the tmpfs mount never happens rathern than delete it after boot Nov 19 16:15:14 does that make sense ? Nov 19 16:15:24 or is any of that unclear ? Nov 19 16:17:51 blogic: but do i a make shure that the tmpfs never mounts? i never deleted the tmpfs - i just removed the symlink /var and created a dir /var and rebooted with the hd spzified as "/" Nov 19 16:17:59 but/but how Nov 19 16:18:58 blogic: setting the socket in usbus to /tmp/run/ubus.socket - fixes he problem and the device comes up Nov 19 16:19:22 yes but thta is a ugly hack that fixes your problem by breaking the code structure Nov 19 16:19:29 that is not how problems should be solved Nov 19 16:19:52 so i assume you have a /var/run folder ? Nov 19 16:19:58 and all the other folders oyu deleted ? Nov 19 16:19:59 blogic: yes Nov 19 16:21:09 hmmm Nov 19 16:30:41 blogic: i rebind / and copy all to the hd - then i remove /var symlink and create a dir and create only the dir /var/lock in it - normally i also use another partition for /var Nov 19 16:30:58 thats all Nov 19 16:31:47 blogic: also crearting a link from /var/run to /tmp/run and then boot from hd doesn't work Nov 19 16:32:06 this was all working before without the latest updates Nov 19 16:33:41 blogic r38866 trunk/package/libs/libubox/Makefile * libubox: update to latest git head Nov 19 16:35:35 blogic r38867 trunk/target/linux/ramips/mt7620n/config-3.10 * ralink: add missing CONFIG_ symbol to mt7620n Nov 19 16:35:43 blogic r38868 trunk/package/system/procd/Makefile * procd: update to latest git head Nov 19 16:36:14 blogic: how would i have to change it to make it work? how about using a seperate /run tmpfs that just holdes the run files - /tmp/run is just a symlink - so it would be possible to get rid of /tmp and /var at boot time Nov 19 16:36:36 make a bug report that is usefull or fix the bug by hunting it Nov 19 16:36:42 however camping me wont help Nov 19 16:37:00 try if that fix from just now is already enough for your problem please Nov 19 16:37:13 if not tell me and i will flash an image with extroot and try it out Nov 19 16:37:58 blogic: okay thans - i will flash it in a few hours and report back Nov 19 16:38:02 thanks Nov 19 16:38:44 the respawn timeou was swizzled Nov 19 16:38:59 so if oyur ubusd died for what ever reason it will take several years to repawn Nov 19 16:39:08 because i swapped timeout 1&2 Nov 19 16:39:41 thanks for the info Nov 19 16:40:28 DonkeyHotei: that patch looks really broken Nov 19 16:41:07 - if (!ret) Nov 19 16:41:07 - ath9k_eep_loaded = 1; Nov 19 16:41:07 + if (ret) Nov 19 16:41:07 + ath9k_eep_load = 1; Nov 19 16:42:14 like the pci_fixup_device() bit ... where was it before ? do you leave dead code ln the kernel ? Nov 19 16:42:32 DonkeyHotei: basically resend the patch with a description of what, why and so on .. Nov 19 16:42:51 * blogic wonders if he should even try to boot it Nov 19 16:44:40 tripolar: you are doing this on ar71xx ? Nov 19 16:45:18 yes Nov 19 17:00:13 build #390 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/390 Nov 19 17:16:19 blogic, do you you know the reason behind octeon using 32 bit userland, on top of 64 bit kernel ? Ie, why it's not all just done as 64 ? Nov 19 17:16:36 starting to dig into the issue of iptables being borked Nov 19 17:17:16 and it does look like it's 32/64 issues on the data going into kernel Nov 19 17:25:57 build #447 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/447 Nov 19 17:29:31 build #368 of kirkwood is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/kirkwood/builds/368 Nov 19 17:32:16 build #329 of iop32x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/329 Nov 19 18:25:01 groz: hey Nov 19 18:25:16 1) the 64bit support in uclibc on mips fails Nov 19 18:25:30 and i failed to get eglibc running with 64bit Nov 19 18:25:47 2) because 32 bit binaries are much smaller and there is no real performance gain Nov 19 18:25:55 so its common to run a 32bit userland on 64 bit Nov 19 18:26:02 and yes, its the dtaa going into the kernel Nov 19 18:26:33 i once did a bit of digging and dumped the struct at the point where it enters the 64 bit compat layer code and directly after and its not the same Nov 19 18:27:53 ok, well, I'm fussing along here Nov 19 18:28:03 so far, no progress of interest Nov 19 18:28:36 altho I did confirm, setting 64 bit options does fail in the uclibc build Nov 19 18:28:37 hehe Nov 19 18:29:05 but there is a ton of storage on this, so, size isn't a big deal at this point Nov 19 18:29:25 i may fuss with options to try get full 64 bit, if for no other reason, a point of reference Nov 19 18:29:42 I got an 8 core sitting in the closet that can do various builds Nov 19 18:29:48 it's got nothing else to do today Nov 19 18:30:10 good Nov 19 18:30:28 once this works i will order the 5 port version of the board as my new core router int he office Nov 19 18:30:51 I'm thinking along similar lines Nov 19 18:31:05 but with the 5 port, gotta figure out how to deal with the poe Nov 19 18:31:12 so one doesn't accidentally cook things attached Nov 19 18:32:29 well, you get those boxes that extract the poe and put the eth on a output rj45 and the power on a barrel jack Nov 19 18:33:23 that's one route, but, what I want to make sure of, is i am able to turn it off, so I can plug anything into those ports Nov 19 18:33:42 then turn on 24v style to plug in ubiquity access points for one or two Nov 19 18:34:10 yes Nov 19 18:34:18 the ports allow none, 24V and 48V Nov 19 18:34:22 software setable Nov 19 18:34:32 http://www.amazon.com/CP-Technologies-POS-1002-LevelOne-Splitter/dp/B0046SSX56/ref=sr_1_2?ie=UTF8&qid=1384886033&sr=8-2&keywords=poe+splitter Nov 19 18:34:36 yes, so it makes it a very flexible thing, if one can get it all going Nov 19 18:34:40 i plan to use those to poe power !poe devices Nov 19 18:34:58 the switch is supported already Nov 19 18:34:59 I want to use two ports for uplinks, one to the cable company, one to the telco Nov 19 18:35:15 then the other 3 ports, for different segments Nov 19 18:35:30 one is office, one is 'home', and one is a run out to the observatory Nov 19 18:35:53 right now, I'm using a D-510 board with two on board, and another card Nov 19 18:37:19 ok Nov 19 18:37:38 my thoughts on this one tho, size doesn't matter, ram footprint, who cares Nov 19 18:37:50 it's got a usb stick for storage, tons of space there, and, its got a ton of ram Nov 19 18:37:51 i mainly want the router to be able to run various vpns and the pppoe server, iperf, radius, .... Nov 19 18:38:04 groz: yes Nov 19 18:38:06 ditto, I have numerous vpn connections on my router Nov 19 18:38:15 two different data centers Nov 19 18:38:29 then I have a vpn to my dads place, where I have an offsite backup machine parked Nov 19 18:39:49 hehe Nov 19 18:40:09 I may have to think about a new location for that soon tho, he's getting on in years, 80 now, and I think he is going to take us up on the offer to move in with us soon Nov 19 18:40:23 winter in his part of the world is brutal on somebody his age Nov 19 18:40:32 it's not nearly that cold here in the south Nov 19 18:41:18 hehe, when he and mom were snowbirds, before she passed away, they ran into a problem in the usa, couldn't get the canadian online coverage of the olympics Nov 19 18:41:38 so, I set up a router at thier place, to tunnel thru my place, and present a canadian ip to the broadcasters Nov 19 18:41:46 allowed them to watch the local broadcasts online Nov 19 18:42:24 they thought that was way cool, and had lotsa folks coming by to watch various events over that two weeks Nov 19 18:42:41 they were living in a snowbird park in california for the winters back then Nov 19 18:46:38 at least you have lots of bandwidth, in my case, I have 3Mbit upload at home and 5 at work Nov 19 18:46:47 in other words.. nothing Nov 19 18:46:50 nbd: http://patchwork.openwrt.org/patch/4437/ shouldn't this patch be set to accepted? Nov 19 18:47:15 cable company keeps saying, bandwidth upgrades 'next month' Nov 19 18:47:23 but, no upgrade available yet Nov 19 18:47:40 still stuck at 5mb up Nov 19 18:48:01 supposed to have 250/15 available real soon now tho Nov 19 18:48:02 lol Nov 19 18:48:44 telco sent us letters last week, apparently they have been putting in more fiber, and offering higher speeds in the area now Nov 19 18:48:52 is 3Mbit enough for that kind of traffic you had? Nov 19 18:49:15 probably could do it, but it's more about uplink that download Nov 19 18:49:28 we have 5 up here, which worked out just fine Nov 19 18:50:20 yup, at home I have 35/3, at work, 5/5 Nov 19 18:50:49 I really don't understand why they cap upload so bad Nov 19 18:50:52 home, work, all the same place for me Nov 19 18:51:44 I dream to have FTTH again :( Nov 19 19:02:09 blogic: i tried the latest version and how i device i unreachable even when /var is symlink Nov 19 19:02:14 how=now Nov 19 19:03:21 blogic: would a patch be accepeted where the socket would be created in /run and /run is a tmpfs with /tmp/run linking to /run? Nov 19 19:03:47 no Nov 19 19:04:11 can you not just be patient for -2 days till its fixed ? Nov 19 19:04:33 yes i can Nov 19 19:04:47 that would be awesome Nov 19 19:05:04 i'll look into it tomorrow Nov 19 19:05:54 okay thanks, just thought when this special case is not supported i could try to make a fix . i just wanted to avoid investing hours for a patch that will never be accepted Nov 19 19:17:13 do you use extroot a lot ? Nov 19 19:27:30 tripolar r38869 packages/net/nfs-kernel-server/Makefile * [packages] nfs-kernel-server: update to 1.2.9 Nov 19 20:00:48 [Tue 2013-11-19 08:42:15 AM PST] like the pci_fixup_device() bit ... where was it before ? do you leave dead code ln the kernel ? <---- it was simply not there, since it's called indirectly by pcibios_init() Nov 19 20:02:09 i inverted the meaning of the ath9k_eep_load flag because i needed to check for it before it would be otherwise set Nov 19 20:03:05 for ath5k i changed device_initcall to subsys_initcall to bring it in line with ath9k Nov 19 20:04:02 there is not much else to be said for the patch Nov 19 20:04:56 blogic: if you have further objections, post to the list and i'll see it after work Nov 19 20:05:10 sfk Nov 19 20:05:13 *afk Nov 19 20:06:06 that reminds me, does anyone know how tunlr exactly work? since you only have to add there dns and the ip you get back from them is the same as you would from google dns for example. and still it bypasses the geo fence Nov 19 20:06:31 wrong window :/ Nov 19 20:54:20 luka: ping Nov 19 20:58:04 blogic r38870 trunk/package/ libs/libubox/Makefile system/ubox/Makefile system/procd/Makefile * libubox: ubox: procd: move md5.{c,h} to libubox Nov 19 20:58:15 blogic r38871 trunk/package/system/mtd/ Makefile src/mtd.c src/Makefile * mtd: add a "mtd verify" call Nov 19 21:05:18 Can we override the KERNELNAME in a profile or not? Nov 19 21:15:17 blogic r38872 trunk/target/linux/ramips/base-files/etc/uci-defaults/02_network * ralink: set wmr-300 mac properly Nov 19 21:56:44 blogic r38873 trunk/package/libs/libubox/Makefile * libubox: fix compile error (#14497) Nov 19 22:10:45 blogic: if you're still up, meeting at work was canceled due to network outage, if you have any comments, feel free Nov 19 22:20:50 When would I want to have a config-default instead of a config-3-* in my target dir? Nov 19 22:26:42 v0n: if you don't plan to support more than one kernel version at the same time Nov 19 22:27:12 ok Nov 19 22:27:36 KanjiMonster: In which case would we want OpenWRT to support multiple kernel? Nov 19 22:27:54 (for a specific board) Nov 19 22:28:58 v0n: usually a target is for a whole family of boards (or rather all boards using a certain chip), not a single board - also e.g. you might want to support a LTS kernel and a non-LTS kernel at the same time Nov 19 22:32:30 ok Nov 19 22:55:13 build #423 of uml is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/423 Nov 19 22:55:57 With OpenWRT, it seems like there is no need to save the .config (or running make defconfig). Am I wrong? Nov 19 22:59:32 its recommended to run scripts/diffconfig and use the output then with make defconfig Nov 19 23:00:33 v0n: do you mean for the kernel? it's part of the target Nov 19 23:02:07 DonkeyHotei: no, for the build system Nov 19 23:02:54 I'm adding a target for a customer board, and in the menuconfig, I'm choosing some options such as uImage, include the kernel in rootfs, etc. Nov 19 23:03:14 you get a .config out of that Nov 19 23:03:22 I was about to save the buildsystem .config file, but it seems like there's another way to provide such options Nov 19 23:03:41 as long as you don't distclean, it stays around Nov 19 23:04:36 scripts/diffconfig.sh is for saving the build system .config file Nov 19 23:04:52 If you have to release the buildsystem to the customer, what would you do? put a .config file somewhere or define some option in the target makefile? Nov 19 23:07:47 decisions, decisions Nov 19 23:08:23 revision number of openwrt.git and packages.git and .config should be enough to replicate an image - except for some build system dependencies that are stated in the wiki Nov 20 00:43:07 build #12 of mxs is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mxs/builds/12 **** ENDING LOGGING AT Wed Nov 20 02:59:58 2013