**** BEGIN LOGGING AT Wed Mar 13 02:59:58 2013 Mar 13 03:56:03 eigma: git pull Mar 13 05:19:35 build #192 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/192 Mar 13 05:22:46 build #192 of ppc44x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/192 Mar 13 07:45:48 build #182 of sibyte is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/182 Mar 13 07:48:16 build #187 of xburst is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/187 Mar 13 12:05:28 build #164 of iop32x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/164 Mar 13 13:24:18 build #160 of gemini is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/160 Mar 13 14:46:47 hi. does anybody knows how config-3.8 for a new target is created? Mar 13 14:47:05 by hand Mar 13 14:47:19 no autogeneration process? Mar 13 14:47:20 you can make it an empty file if you don't have a reference config Mar 13 14:47:27 and then use make kernel_menuconfig Mar 13 14:47:38 which will fill in the relevant options Mar 13 14:48:02 if you do have a reference config, you can copy it to config-3.8 and use make kernel_menuconfig afterwards Mar 13 14:48:12 which will strip it to just the changes relative to the generic config-3.8 Mar 13 14:48:36 afterwards you should take a look at the file and throw out all lines that shouldn't be changed compared to the default config template Mar 13 14:49:48 has anybody here built for an octeon recently ? Mar 13 14:50:03 I just got what appears to be an interesting board, ubiquity edgerouter light Mar 13 14:50:16 and contemplating where to start with it Mar 13 14:51:03 dual core mips processor, couple gigs of flash, 3 gigabit ports Mar 13 14:51:26 it comes with a debian derviative of some type pre-loaded Mar 13 14:52:01 <[florian]> groz: you'd need to resurrect the octeon target to start with :) Mar 13 14:52:21 i'm looking at doing just that, and, just looking at dmesg, I lied, it's 4 gig of flash Mar 13 14:52:30 and it's mapped to usb device Mar 13 14:52:35 very interesting Mar 13 14:52:45 it's /dev/sda via usb storage Mar 13 14:53:10 just unboxed this thing last nite, quite late, starting to poke at it Mar 13 14:53:24 <[florian]> cool, avoids having the need for flash driver :) Mar 13 14:53:37 well, it's got a flash inside from which it boots Mar 13 14:53:47 still unravelling how it's all glued together here Mar 13 14:54:00 but, 3 gigabit ports, with a dual core processor Mar 13 14:54:13 has a bit of 'oomph' and it's a nice small little package Mar 13 14:54:40 can you measure power usage ? Mar 13 14:55:04 hmm, not sure I have the tools for that handy, but can figure it out eventually Mar 13 14:55:27 it has an rj-45 serial console port too, gotta get a cable glued together later today for that Mar 13 14:55:40 I like what I'm reading :) Mar 13 14:55:43 these things are not spendy, sub $100 I believe Mar 13 14:55:57 or right in that range Mar 13 14:56:56 it even has hardware encryption accelerator. very neat Mar 13 14:58:10 http://pastebin.ca/2331789 Mar 13 14:58:48 pity, no Po Mar 13 14:58:50 poE Mar 13 14:59:14 I got one of the toughswitches in the same order Mar 13 14:59:22 5 port switch with management port as well Mar 13 14:59:33 under the covers, it's an ar71xx process seems Mar 13 14:59:42 and just a pretty gui to manage a 5 port switch Mar 13 15:00:01 with the second processor port on an ethernet port all of it's own Mar 13 15:00:08 but not a lot of poop to the processor Mar 13 15:00:17 it has switchable passive poe per port Mar 13 15:01:19 just scrolling thru that dmesg, appears to be 4 meg boot flash, then 4 gig of usb-storage Mar 13 15:01:31 on this little router Mar 13 15:05:11 puh. it is really possible to build OpenWrt on Windows Cygwin. I got a qemu arm build complete. Mar 13 15:07:56 <[florian]> wbx: cool Mar 13 15:09:04 [florian]: i have send a patch this week, which is nearly complete. just this is missing : https://patchwork.kernel.org/patch/1598391/ Mar 13 15:17:09 [florian]: what do you think about the patch? Mar 13 15:17:44 <[florian]> wbx: pretty much the same thing as what the guy commented, I do not understand what it is supposed to fix Mar 13 15:21:27 [florian]: but it fixes the cross-compile of the linux kernel. but I meant my Cygwin.patch for OpenWrt Mar 13 15:44:38 nbd r35996 trunk/package/libs/libubox/Makefile * libubox: update to latest version, adds libjson-script Mar 13 15:46:09 nbd r35997 trunk/package/ubus/Makefile * ubus: update to latest version Mar 13 15:46:14 <[florian]> wbx: ah ok, well yes we can probably commit it Mar 13 15:46:58 jow r35998 trunk/package/network/ config/firewall3/Makefile config/firewall3/files/firewall.init config/firewall3/files/firewall.hotplug * firewall3: update to git head Mar 13 15:47:06 florian, i see you dealt with a few octeon tickets in the past, have you any idea roughly when that target was actually built / used / whatever Mar 13 15:47:34 just looking at what I need to do to try build something here, and wondering if I should be checking out a very old version first, as a starting point Mar 13 15:48:10 <[florian]> groz: just check out the current version of trunk Mar 13 15:48:20 <[florian]> then throw in the old octeon target support and update the kernel to 3.8 Mar 13 15:48:39 <[florian]> the octeon target was fairly minimaly supported, it booted from initramfs only Mar 13 15:48:41 well, on the other targets I have going here, current version of trunk doesn't actually build Mar 13 15:48:52 <[florian]> I then handed my sole octeon board to Kaloz Mar 13 15:48:59 it's failing in toolchain build for ar71xx right now on my ssystem Mar 13 15:49:08 <[florian]> what's the error? Mar 13 15:49:11 ok, well, i've got one now, and, I'd like to experiment Mar 13 15:49:32 I am not sure, I didn't persue it when I saw it, was kind of busy and it went into the 'when i get to it' pile of things to be done Mar 13 15:49:33 lol Mar 13 15:49:43 i think we all have one of those piles dont we ? Mar 13 15:50:12 i dont have a console cable for this router yet, should have that later today, then I'll see if the boot loader is capable of netbooting Mar 13 15:50:32 if so, life will get much easier, cuz I can netboot an initramfs image until all the bits work Mar 13 15:51:28 <[florian]> exactly Mar 13 15:51:29 I'll check out a fresh tree, and see what happens when it's configured for the octeon Mar 13 15:54:44 <[florian]> sounds like a good plan Mar 13 15:55:17 <[florian]> I expect a vanilla openwrt kernel to boot correctly on octeon, but you might need to add a CUST_UBNT* or something like that which defines your specific board hardware Mar 13 15:55:25 build starting, cross fingers, pour a coffee, and off we go Mar 13 15:55:56 i see in the dmesg spew, they have something in for ethernet phy that is ubnt specific apparently Mar 13 15:56:04 but, time will tell, the proof is in the first boot that boots Mar 13 15:56:13 <[florian]> that's usually where the CUST_ thing plays a role Mar 13 15:56:27 and if I can get to netbooting with initramfs, life is much easier sorting things out Mar 13 15:57:10 but th edge lite is an intriguing little router, and, seems they have larger versions in the pipeline Mar 13 15:57:16 with more ethernet ports etc Mar 13 15:58:09 apparently the load that comes on it, is relatively vanilla debian, and one can load packages from debian repositories Mar 13 15:58:29 <[florian]> can you check what userland they support (n32, o32 or n64?) Mar 13 15:58:37 n64 Mar 13 15:58:47 from what I've read Mar 13 15:59:00 <[florian]> ok, that's not what I would have chosen, but why not Mar 13 15:59:25 well, i'm just going by what uname -a is telling me righ tnow Mar 13 15:59:30 still hunting to learn more Mar 13 15:59:52 <[florian]> if you have a non-empty /lib64 chances are good that the userland is n64 Mar 13 16:00:16 ubnt@ubnt:/$ ls /lib64 Mar 13 16:00:16 ls: /lib64: No such file or directory Mar 13 16:00:16 ubnt@ubnt:/$ Mar 13 16:00:31 so much for that concept Mar 13 16:00:39 <[florian]> the kernel is 64-bits but the user-land is probably n32 Mar 13 16:00:56 <[florian]> that's the best tradeoff between 64-bits features you want (64-bits pointers, registers ...) and size Mar 13 16:01:04 http://pastebin.ca/2331817 Mar 13 16:01:19 dunno if that bogomips is calculated, or just hardwired number Mar 13 16:01:32 but if it's real, that makes this quite a capable little router for the price Mar 13 16:01:53 <[florian]> it's computed Mar 13 16:02:06 it should be computed Mar 13 16:02:17 but, i've seen folks plug in a hardwired number in the past Mar 13 16:02:20 to make it look better Mar 13 16:02:28 <[florian]> eh eh right Mar 13 16:03:47 cable net connection around here, is all 100mbit these days, with 250 'coming soon' Mar 13 16:04:09 so, inexpensive little router capable of dealing with those speeds, is quite a nice little gadget to get up to speed on Mar 13 16:05:42 gonna stress this thing later today, see what kind of numbers it can do for thruput, with nat in play Mar 13 16:17:10 florian r35999 packages/ net/bridge-utils/patches/010-fix_struct_in6_addr_usage.patch net/bridge-utils/Makefile * bridge-utils: fix a struct in6_addr reference Mar 13 16:28:02 <_trine> python appears to be segfaulting Mar 13 16:56:26 _trine: uhmm nope that would be your brain Mar 13 16:56:36 [florian]: thanks! Mar 13 16:56:58 <[florian]> russell--: yw Mar 13 16:58:20 btw, i noticed recently (we just went to daylight savings here in the states, sigh) that logread is spitting out times in standard time from the kernel and daylight time from userspace ... i don't remember that before. Mar 13 16:59:05 e.g. Mar 13 16:59:08 Mar 13 09:36:33 rag user.notice gateway: [2013-03-13 09:36:33] Displaying mobile splash page to peer 10.11.8.162 Mar 13 16:59:11 Mar 13 08:36:35 rag kern.info kernel: [ 3153.970000] device eth1 left promiscuous mode Mar 13 17:02:22 yes DST is a bug on reality.. dunno where to file it. Mar 13 17:03:34 everyone should just switch to UTC Mar 13 17:03:53 yes. that would be one of the sanest solutions. Mar 13 17:04:26 but i guess getting rid of dst would not help us get rid of timezones that easily. Mar 13 17:05:06 <_trine> OutBackDingo, python is segfaulting Mar 13 17:05:15 timezones are good, they keep many software developers employed Mar 13 17:05:22 lol Mar 13 17:06:27 groz: is this also the reason for the different encodings? so UTF-8 will kill our jobs ;-) Mar 13 17:07:02 i dont think it'll kill our jobs, but, it'll kill them for folks closer to the bottom of the food chain Mar 13 17:07:10 our jobs will be to support utf-8 Mar 13 17:07:19 so they can eliminate those lower down the food chain Mar 13 17:07:26 <_trine> OutBackDingo, Mar 13 17:07:26 <_trine> [0:03:20] starting wpa handshake capture Mar 13 17:07:26 <_trine> [0:02:53] sent 3 deauth packets; no handshake yet Segmentation fault Mar 13 17:07:48 <_trine> boom Mar 13 17:10:59 think about it Hauke, if all websites were UTF and UTC, they still going to move the same amount of data, so, router based jobs are safe, but, if your bread and butter is from coding up web pages and web scripts, lots of time/hassle with those will vanish, and with it goes many many billable hours for folks who make a living doing that stuff Mar 13 17:12:38 groz: yes, that's true Mar 13 17:13:42 I have this same discussion here regularily, people say we shouldn't use paper bags Mar 13 17:14:01 I say bollocks, ask for paper bags at the grocery store, not plastic, dont bring your re-cycled bags Mar 13 17:14:14 puts a logger back to work, and lord knows, we have enough of them out of work right now Mar 13 17:17:49 too many trees eating our microwaves! Mar 13 17:18:18 trees are by definition, a weed Mar 13 17:18:31 if you leave a piece of land untended here for a few years, it grows back as trees Mar 13 17:18:36 which is the definition of a weed Mar 13 17:19:00 where is here? Mar 13 17:19:08 vancouver island, british columbia Mar 13 17:19:11 russell--: can you try whether running "date -k" fixes the time discrepancy? Mar 13 17:19:15 west coast of canada Mar 13 17:19:31 <_trine> who's going to do revision 36000 Mar 13 17:19:40 <_trine> we are at 35999 Mar 13 17:19:56 omg, we are gonna break the system with a rollover Mar 13 17:20:53 jow_laptop: /me waits for kernel to spit out something to log Mar 13 17:21:28 russell--: you could run tcpdump or ifup lan, that should recreate the bridge and/or toggle promisc Mar 13 17:23:27 yeah Mar 13 17:24:16 jow_laptop: didn't fix it Mar 13 17:25:10 ok Mar 13 17:25:35 I guess we need to ammend the kernel time setting to copy over the dst offset as well Mar 13 17:25:38 -m Mar 13 17:27:55 groz: /me is further down the coast a few hundred miles Mar 13 17:28:04 where abouts ? Mar 13 17:28:15 portland Mar 13 17:28:26 ok, I know that area a bit Mar 13 17:29:08 * russell-- took a ferry once from port hardy to prince rupert Mar 13 17:29:34 that's a nice trip, when we have relatives over from europe, I usually try convince them to take that trip Mar 13 17:29:52 then drive thru the interior to come back, makes a really nice big loop thru the province Mar 13 17:30:34 one of the screws was broken, so we didn't do much inside-passage, but pitched about a bit in a storm and felt queezy Mar 13 17:30:50 hehe, musta been during the winter Mar 13 17:30:59 it's always beautiful in the summer (giggle) Mar 13 17:31:20 yeah, we did that loop, prince rupert to jasper and then down through banff and glacier nat'l park in the states Mar 13 17:31:28 it was early september Mar 13 17:31:42 jasper is nice in early september, not cold yet Mar 13 17:31:52 yeah, smithers was a little chilly Mar 13 17:31:53 and the vast majority of tourists, have left Mar 13 17:31:56 yes Mar 13 17:32:11 when we travel in bc, try do it after the labour day long weekend Mar 13 17:32:16 it's much more sane Mar 13 17:32:26 i was on a motorcycle with my dad Mar 13 17:32:53 <_rmk_> bah, can't send this draytek router back under the distance selling regulations because it was a business-to-business purchase :( Mar 13 17:33:04 russell--: good time of the day. the problem with the ping6 packets seems to be resolved, the 6in4 tunnel provider had problems Mar 13 17:33:11 <_rmk_> so looks like I'm stuck with a fscked [AV]DSL router. Mar 13 17:33:27 BasicXP: still seems weird that it panic'd Mar 13 17:33:53 _rmk_: ok Mar 13 17:34:01 _rmk_: sooo .... have you opened it ? Mar 13 17:34:02 well the panic was probably due to configuration problem, don't know Mar 13 17:34:11 imho draytek uses lantiq SoC Mar 13 17:34:11 I'm talking about only the first packet going through Mar 13 17:34:19 yeah Mar 13 17:34:21 <_rmk_> blogic: not yet... Mar 13 17:34:25 _rmk_: so in theory we could have a go at cracking it open Mar 13 17:34:39 try to open it up and make a photot Mar 13 17:34:46 chances are its brcm or lantiq Mar 13 17:34:54 whats the model number ? Mar 13 17:35:00 <_rmk_> It's the 2850Vn Mar 13 17:36:03 anyway, thank you very much for help! I hope this issue won't repeat anymore Mar 13 17:36:34 this has a good ending - configured dual access properly this time, without hotplug scripts Mar 13 17:37:41 _rmk_: http://www.simet.com.tr/index.cfm?fuseaction=objects2.detail_product&pid=841 Mar 13 17:37:44 looks ugly Mar 13 17:37:47 <_rmk_> if you're interested... he's my writeup of the issue I've been having with it - it'll make you laugh or cry: http://www.home.arm.linux.org.uk/~rmk/misc/draytek.txt Mar 13 17:38:05 ok Mar 13 17:39:53 <_rmk_> well, it is packed with two sockets for pots phones, pots line, and can do voip from those pots phones or route them over the pots line. Mar 13 17:40:23 <_rmk_> and has separate ADSL and VDSL stuff Mar 13 17:40:32 why so ? Mar 13 17:40:53 i mean vdsl has a adsl fallback unless oyu have a really crappy realtek / ikanos chipset Mar 13 17:41:01 looks like a badly designed unit Mar 13 17:41:22 <_rmk_> http://www.draytek.co.uk/products/vigor2850.html Mar 13 17:41:28 if i see PCBs with fat piggy back boards stuck on top i know something is wrong ;) Mar 13 17:41:35 <_rmk_> because it supports simultaneous ADSL and VDSL Mar 13 17:41:54 VDSL with compatible BT Infinity™ Mar 13 17:42:03 BT Infinty is etherne and not vdsl Mar 13 17:42:09 ?! Mar 13 17:42:30 <_rmk_> BT Infinity is FTTC, with VDSL in the cabinet. Mar 13 17:42:33 the home hub 2 has a generic gbit phy for the infinity part Mar 13 17:42:43 not for the home hub 2 Mar 13 17:43:15 i have a home hub 2 here with a port that says infinity and its a ar9/adsl chipset Mar 13 17:43:18 weird Mar 13 17:43:26 <_rmk_> http://www.trefor.net/wp-content/uploads/2010/03/cabinet-with-kit_small.jpg Mar 13 17:43:27 maybe they changed the product but not the marketing name Mar 13 17:43:33 <_rmk_> that's what's in the box. Mar 13 17:43:44 hehe Mar 13 17:44:56 <_rmk_> the scarey thing is that BT supply the 2850Vn to businesses for use with their Infinity stuff. Mar 13 17:45:06 :-D Mar 13 17:45:40 back in the days where i lived in the UK i was on superjanet Mar 13 17:45:49 i miss those days Mar 13 17:46:11 full 10mbit both directions Mar 13 17:46:17 and napste just came out Mar 13 17:46:21 *napster Mar 13 17:46:29 <_rmk_> anyway... I'll give them a bit longer to come back - I have both Draytek UK (who are apparantly just support+distributors) and Draytek worldwide (I think China) a bit more time. Mar 13 17:46:30 it was like a digital orgasm Mar 13 17:46:36 * groz cant imagine going back to 10mbit Mar 13 17:46:45 groz: i am on 6/1 just now Mar 13 17:46:47 :) Mar 13 17:46:59 i am on 100/5 and soon will be on 250/10 Mar 13 17:47:15 <_rmk_> My jaw dropped when Draytek China asked for my Linksys router login details... the one which is currently my primary router. Umm... no. :) Mar 13 17:47:24 wife may be transferring for work in a couple of months, we were looking at houses in the new area last nite Mar 13 17:47:27 _rmk_: :-D Mar 13 17:47:28 <_rmk_> So I sent them all the config and status pages instead :) Mar 13 17:47:43 but, I put my foot down, we will NOT go into an area where the cable company doesn't have net service Mar 13 17:47:57 she found a really nice house, but, can only get 6/1 there Mar 13 17:47:59 so, I wont go Mar 13 17:48:04 <_rmk_> blogic: btw, that's a Linksys AG241. been doing good for some 10+ years :) Mar 13 17:48:10 if in closer to town, we can have 100/5 Mar 13 17:48:39 _rmk_: running stock os or id you flash it with something Mar 13 17:49:08 <_rmk_> running stock - if it works I leave it :) Mar 13 17:49:32 <_rmk_> the only thing I've had to do over that time is replace some of the caps once. Mar 13 17:50:43 hehe Mar 13 17:53:20 <_rmk_> once I get something sorted out with fttc, the AG241 will become the backup router, and I'll see about flashing it then. Mar 13 18:11:28 blogic r36000 trunk/target/linux/ar71xx/base-files.mk * [ar71xx] make sure the console works when using procd as our pid 1 Mar 13 18:11:35 blogic r36001 trunk/package/procd/ files/hotplug.json Makefile files/hotplug-preinit.json * [procd] update to latest git head Mar 13 18:11:41 blogic r36002 trunk/target/linux/generic/base-files/init * [base-files/generic] update the /init file to start /sbin/init Mar 13 18:11:47 blogic r36003 trunk/package/base-files/ (57 files in 14 dirs) * [base-files] make basefiles aware of procd Mar 13 18:11:53 blogic r36004 trunk/package/network/ services/hostapd/Makefile services/hostapd/files/wps-hotplug.sh * [hostapd] use new button scheme Mar 13 18:11:59 blogic r36005 trunk/package/ busybox/Makefile busybox/config/init/Config.in busybox/config/sysklogd/Config.in * [busybox] make init and logread depend on !PROCD_INIT Mar 13 18:29:36 blogic r36006 trunk/package/procd/Makefile * [procd] fix date and dependency Mar 13 18:45:11 hi blogic, now you have your hands dirty, can you consider applying http://patchwork.openwrt.org/patch/3428 ? Mar 13 18:45:13 :) Mar 13 21:00:21 fresh checkout from trunk, getting failure here in building gcc it seems Mar 13 21:00:30 just got a V=s build running to try see where Mar 13 21:00:37 configured for ar71xx Mar 13 21:01:01 had no problems building this from a fresh checkout a few weeks ago Mar 13 21:01:11 <[florian]> groz: can you show me the log please? Mar 13 21:01:27 i'm waiting for the log to finish, just re-running it with V=s now Mar 13 21:02:03 <[florian]> ok Mar 13 21:02:40 will a simple 'make' leave a log file behind somewhere ? Mar 13 21:02:47 or do I need params for that ? Mar 13 21:03:35 or should I just redirect it to a file and re=run with V=s ? Mar 13 21:03:47 <[florian]> you need to enable BUILD_LOG if you want it to log somewhere Mar 13 21:03:56 <[florian]> or use V=s Mar 13 21:04:00 lemme re-do that with a log Mar 13 21:04:04 makes it easier Mar 13 21:05:56 re-building with logs enabled Mar 13 21:06:00 will take a little bit Mar 13 21:07:01 first round I thought it was the octeon target, so, just to be sure, I retargetted to ar71xx Mar 13 21:07:09 and got same failure Mar 13 21:10:02 hmmm, now it looks like it may have gone thru Mar 13 21:10:03 go figger Mar 13 21:11:11 <[florian]> did you run parallel/non parallel or make any change? Mar 13 21:11:26 non parallel both times, only change is, enabled logging this time Mar 13 21:11:38 i just blew it away completely, and restarted the build from scratch Mar 13 21:11:47 will see if I can re-create Mar 13 21:12:16 i have logs enabled now, and just 'make' going Mar 13 21:12:22 nothing special Mar 13 21:20:41 is parallel building safe these days, my experience in the past was, do the first build serial in a tree, then parallel works ok after that Mar 13 21:20:49 but, is problematic on the first pass thru Mar 13 21:43:30 <[florian]> it's mostly safe except for a couple of eglibc versions which miserably fail Mar 13 21:45:59 this build is as far as gcc initial, so, I'll just let it run it's course at this point Mar 13 21:46:05 it's building serially Mar 13 21:53:58 build #212 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/212 Mar 13 22:00:02 florian, it went thru this time Mar 13 22:00:10 only difference, I have build logs enabled Mar 13 22:00:32 it's building packages now Mar 14 00:05:05 so I just opened up the edge router, to see what it is inside Mar 14 00:05:21 much to my surprise, the usb storage is actually in the form of a usb jumpstick type device Mar 14 00:05:31 removes, and plugs into desktop, shows up 2 partitions Mar 14 00:05:38 one has the kernel, one the rootfs on it Mar 14 00:05:57 this will be an easy device to work with, no worries about putting an unrecoverable broken build into flash Mar 14 00:12:23 cool Mar 14 00:12:43 /dev/sda1 contains vmlinux.64 Mar 14 00:12:54 it's fat32 formatted Mar 14 00:12:59 how hard is that to 'reflash' Mar 14 00:13:39 now to see if I can get a kernel built and running for it **** ENDING LOGGING AT Thu Mar 14 02:59:58 2013