**** BEGIN LOGGING AT Sat Sep 28 02:59:58 2013 Sep 28 05:19:48 build #298 of ep93xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/298 Sep 28 05:23:10 nbd r38247 trunk/package/kernel/linux/modules/other.mk Sep 28 05:23:10 kernel: disable kmod-iio for linux 3.6 as well (to avoid build issues due to missing config symbol changes) Sep 28 05:27:17 I'm trying to compile the ag71xx ethernet driver as a module and finding that it causes "VFS: Cannot open root device" errors.. any clues how these may be related? Sep 28 05:29:29 eigma: it's highly unlikely that the ethernet driver change caused this, you probably messed up something else Sep 28 05:29:58 eigma: but without further information (e.g. a boot log), there's no way to figure out what went wrong Sep 28 05:29:58 hmm Sep 28 05:30:05 I can post a boot log, one sec Sep 28 05:30:18 actually I narrowed it down to the MDIO driver Sep 28 05:30:30 ag71xx_main actually registers two platform drivers Sep 28 05:30:39 the ethernet driver itself and MDIO Sep 28 05:30:46 neither of which is relevant for filesystem stuff Sep 28 05:30:59 ok, I will post boot logs, one sec Sep 28 05:41:31 sorry, my dev cycle is a bit long Sep 28 05:45:09 is there a make target that will rebuild kernel, modules and rootfs, but nothing else? Sep 28 05:46:40 make target/compile package/linux/compile package/install target/install Sep 28 05:47:39 this takes into account changes in target/linux/ar71xx/files? Sep 28 05:47:59 no Sep 28 05:48:14 but you can edit stuff in the kernel build tree Sep 28 05:48:17 then you can just run make Sep 28 05:48:22 and it won't have to rebuild everything Sep 28 05:48:42 I would like to keep it in git somehow, so i can track exactly what I've changed Sep 28 05:49:07 you could copy target/linux/ar71xx/files/ to the kernel build tree before running make Sep 28 05:49:26 k Sep 28 05:50:35 a few links.. Sep 28 05:50:37 tree: https://github.com/cpatulea/openwrt/tree/c6b97ec7fa7b1f4e93da116a069853b92e94ad9c Sep 28 05:50:53 .config: http://pastebin.com/gbqMAmRg Sep 28 05:51:17 boot log: http://pastebin.com/3gH5HD60 Sep 28 05:51:27 this is the working version, with ag71xx in-kernel Sep 28 05:52:14 next, I will build from cly branch HEAD (just 1 commit ahead) Sep 28 05:56:21 boot log: http://pastebin.com/kWCXw683 Sep 28 05:57:44 i'm not going to go through that tree and look for possibly related changes Sep 28 05:58:22 it's not much, I can show you a diffstat Sep 28 05:59:02 a full diff would be more useful Sep 28 05:59:18 ok Sep 28 05:59:18 a full diff between plain openwrt and whatever it is you're using Sep 28 05:59:59 I've made my own changes and merged from upstream at different points in time, so there's not a clear version to diff against Sep 28 06:00:07 let me try to get as clean a diff as I can, one sec Sep 28 06:02:27 against 299e650: http://sprunge.us/MEUB Sep 28 06:02:56 it's a little old, but I think it should be ok? Sep 28 06:05:35 and you're absolutely sure that this is all of it? Sep 28 06:06:46 I'm pretty sure.. what can I do to double-check? Sep 28 06:07:19 I have a few untracked files but they shouldn't interfere.. http://sprunge.us/dEAP Sep 28 06:07:44 I have some things in files/ Sep 28 06:08:13 but again I don't see a problem: http://sprunge.us/VKEj Sep 28 06:08:49 I find it interesting that the MTD partition listing is mixed up with other inits Sep 28 06:08:54 in the boot log Sep 28 06:09:02 do you know if these run in parallel? Sep 28 06:09:44 no idea Sep 28 06:09:53 I just realized the partition listing doesn't even finish for the broken case Sep 28 06:10:40 actually the only reason I'm doing this is netconsole doesn't seem to work well through AR8316 Sep 28 06:10:51 so I'd like to build the net drivers as modules so I can hack on them more easily Sep 28 06:11:07 maybe there's some better way to do this? Sep 28 06:19:05 something's happening in a different order Sep 28 06:19:31 "Creating 5 MTD partitions" and "NET: Registered protocol family 17" are swapped Sep 28 06:19:54 a few others as well Sep 28 06:20:23 is there an easy way to print the stack in kernel? Sep 28 06:22:30 dump_stack() Sep 28 06:23:30 do you have a debugging rig? Sep 28 06:28:16 ? Sep 28 06:32:33 an in-circuit debugger for your hardware Sep 28 06:32:50 i do most debugging with printk Sep 28 06:33:13 if there's something where the hardware locks up or really weird stuff happens where i need to inspect the state, i use jtag Sep 28 06:35:11 nbd do you work with hardware development as well or only driver/software? Sep 28 06:35:21 only driver/software Sep 28 06:35:53 "only" hehe Sep 28 06:36:11 at some point i'll probably take some time to learn more about hw stuff as well Sep 28 06:36:41 qualcomm should hire you, if they didn't already Sep 28 06:37:02 they tried to Sep 28 06:37:08 but i don't want to work full time for a big company Sep 28 06:37:50 fund your own company Sep 28 06:38:04 Fietkau Co., Sep 28 06:38:07 i'm already working just like i want to Sep 28 06:38:28 choosing my own projects (based on fun, not money) Sep 28 06:38:34 is it contracts? Sep 28 06:38:53 sometimes contracts, sometimes other things Sep 28 06:39:02 yeah, that's how I want to live Sep 28 06:39:10 puzzles, fun puzzles Sep 28 06:40:12 how long have you been doing this? Sep 28 06:40:31 let's just say, i never had a normal job Sep 28 06:40:34 in my entire life Sep 28 06:41:10 i've been doing openwrt as a hobby since around 2004 Sep 28 06:41:28 and gradually started getting paid for it Sep 28 06:41:33 until it became my main source of income Sep 28 06:42:19 that's cool Sep 28 06:42:48 either I'm dumb for a major or too smart Sep 28 06:42:50 I would like that kind of job.. I think :) Sep 28 06:43:01 I can't get a college degree :/ Sep 28 06:43:10 i didn't get mine either Sep 28 06:43:25 at some point i decided that my projects are more important than my degree Sep 28 06:43:32 and you are doing just fine Sep 28 06:43:39 you give me hope.. Sep 28 06:43:58 a degree matters for getting jobs in many big (and sometimes smaller) companies Sep 28 06:44:09 but not for working on your own Sep 28 06:46:29 the main thing that allowed me to work the way that i do is that i made my stuff open source Sep 28 06:46:48 and i wrote enough code that it became very visible Sep 28 06:47:05 that way i never had to go around looking for contracts or companies to work with Sep 28 06:48:22 exactly Sep 28 06:48:25 do you find the city you're in matters much? for meeting clients for example? Sep 28 06:48:58 i've never worked for companies in the city that i live in Sep 28 06:49:11 I wish I was younger, today 12 year old can code in 20 languages Sep 28 06:49:39 how old are you? Sep 28 06:49:44 29 Sep 28 06:49:49 that's young enough Sep 28 06:49:54 younger than me actually Sep 28 06:50:07 you must have 35 tops Sep 28 06:50:12 31 Sep 28 06:50:38 at that age, learning is still a very easy thing to do Sep 28 06:50:42 yep, and you have way more patience than me :) Sep 28 06:51:22 I plan to learn C, but the experience is what counts.. and that will take awhile Sep 28 06:52:11 best thing in my opinion is to force yourself to use it Sep 28 06:52:35 the next small script you will write, do it in C instead of whatever you're comfortable with Sep 28 06:52:49 not sure i'd agree with that Sep 28 06:52:58 yes, I'm formatting all my computers to linux, some dual boot Sep 28 06:53:02 i like to always use the right tool for the task Sep 28 06:53:14 but i like to pick tasks where the right tool is something i haven't worked with yet ;) Sep 28 06:53:22 :) Sep 28 06:53:38 there languages I don't know but I have empathy, like Python Sep 28 06:53:45 I guess I find myself writing little scripts pretty often, so that's why I chose that example Sep 28 06:53:54 I find Python so powerful Sep 28 06:54:05 it was easy to learn Python that way, and indeed it's more suited for scripting than C Sep 28 06:54:15 I think it would be hard to find day to day things that are best written in C Sep 28 06:55:02 my current plan is C, Java (or probably Scala), Python, HTML/CSS Sep 28 06:55:50 I need some OO and HTML to make front-ends Sep 28 06:57:41 nbd, you wrote the original AR8216 driver? Sep 28 06:58:07 yes Sep 28 06:58:17 with docs from Atheros? Sep 28 06:58:22 yes Sep 28 06:59:44 nbd are you having fun in #openwrt? Sep 28 07:00:35 yep Sep 28 07:00:52 you must be Sep 28 07:02:45 build #338 of ixp4xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/338 Sep 28 07:04:07 heh Sep 28 07:04:25 "too many people out there are full of crap and write random stuff" :) Sep 28 07:05:32 oh well, trying to find a way to study at Goethe Institute.. thousands of Euro Sep 28 07:05:57 let's see if I still have Merkel's number.. Sep 28 07:09:04 do you ever use kgdb? Sep 28 07:09:44 no Sep 28 07:09:58 never tried it Sep 28 07:10:18 if i do jtag debugging, i use normal gdb to debug the kernel Sep 28 07:10:50 i have a jtag device that speaks jtag on one side and gdb over ethernet (+ telnet) on the other Sep 28 07:11:13 but i only need it in rare cases Sep 28 07:11:37 which jtag interface do you have? unbuffered one? have you seen diy universal jtag from Tornado (dd-wrt contributor)? Sep 28 07:12:04 openocd? Sep 28 07:12:18 abatron bdi3000 Sep 28 07:12:23 openocd does similar stuff though Sep 28 07:12:50 Devastator: that stuff is only usable for flashing Sep 28 07:12:56 i use jtag for debugging Sep 28 07:13:00 jtag over ethernet? Sep 28 07:13:06 that's pretty neat :> Sep 28 07:13:11 it's a standalone box Sep 28 07:13:17 speaks gdb over ip on an ethernet port Sep 28 07:13:21 i mean the native gdb protocol Sep 28 07:13:38 hah Sep 28 07:13:41 that's excellent Sep 28 07:14:15 do you have the jtag pinout for buffalo wzr-600dhp? :) Sep 28 07:14:27 it's the normal mips ejtag layout Sep 28 07:15:11 compatible with BDI3000 I guess? Sep 28 07:15:21 right Sep 28 07:15:26 should work with any mips capable jtag tool Sep 28 07:15:39 not sure how well openocd does mips these days Sep 28 07:15:43 there's some support Sep 28 07:15:45 heh, BDI3000 is $2.3k :) Sep 28 07:15:49 but i never tested it Sep 28 07:16:02 let's see that stuff :) Sep 28 07:16:13 eigma: naturally i didn't pay for it myself ;) Sep 28 07:16:40 hahaha Sep 28 07:16:43 lucky you :P Sep 28 07:17:35 maybe I'll stick to my improvised serial link for now Sep 28 07:18:55 * [mbm] remembers building a crappy jtag out of a parallel port and a few diodes Sep 28 07:19:29 I have one, actually I have a USB Blaster clone, but it hinges on openocd support like nbd said Sep 28 07:20:30 [mbm] I still have one of these :) Sep 28 07:20:50 and it's pretty slow.. I forget whether openocd uses the fast USB Blaster mode yet.. I remember hacking on that driver but couldn't get it working quite right Sep 28 07:22:01 * russell-- has one of these: https://www.olimex.com/Products/ARM/JTAG/ARM-USB-OCD/ ... hasn't been plugged in for several years Sep 28 07:23:09 I need to build a buffered one, it's way faster Sep 28 07:23:18 but i did managed to dump flash off a peplink winti board with it Sep 28 07:23:52 what is that? Sep 28 07:24:16 what is what? Sep 28 07:24:22 peplink winti Sep 28 07:25:08 i got them out of the inside of a SkyPilot SkyExtenderDualBand Sep 28 07:25:25 it's just an arm-based single board computer with a mini-pci slot Sep 28 07:25:48 we have a bunch, so i'm motivated to make them work ... eventually Sep 28 07:26:40 http://skypilot.trilliantinc.com/ Sep 28 07:27:16 http://skypilot.trilliantinc.com/products/skypilot_extender.php Sep 28 07:29:42 well, bed time for me.. Sep 28 07:30:19 tomorrow I will try to work out why sysupgrade image for router that uses seama doesn't build with proper header Sep 28 07:30:58 not header, magic :P Sep 28 07:31:17 or is it the same thing? I'm guessing not.. Sep 28 07:31:39 if i'm lucky, i'll decided to give up instead. actually hoping to sucker a college student into porting u-boot for me. Sep 28 07:32:16 build #286 of gemini is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/286 Sep 28 07:33:10 bed time for me too.. good night/good morning as the case may be Sep 28 07:36:04 yep :) Sep 28 08:02:11 * russell-- also has a couple epi jeeni jtag probes in a box, never fired up Sep 28 09:12:56 nbd: hi, will the group ownership of the ubus socket be implemented? Sep 28 09:32:44 build #343 of xburst is complete: Failure [failed compile_9] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/343 Sep 28 10:43:42 Hauke: btw. wouldn't it make sense to move bgmac patches to generic? Sep 28 10:43:52 yes Sep 28 12:14:40 nbd: Toolchain now compiles OK. But, then the compilation crashes on package gmp as shown here (http://pastebin.com/F1wTKFTM). The reason is the extra white space `str r1,[ r0 ]'. I also included the patch in the link and now package gmp compiles OK. I hope you can include that patch to the package gmp. Sep 28 12:55:22 mazilo: feel free to submit a patch for that Sep 28 13:13:49 jow r38248 trunk/package/system/ubox/Makefile Sep 28 13:13:49 ubox: fix kmod loading for modules with varying dashes & underscores in names (#14182, #14189) Sep 28 13:27:50 jow_laptop: thx! Sep 28 13:35:53 Hauke: hey Sep 28 13:36:01 Hauke: have you ever tried enabling 5GHz in b43? Sep 28 13:36:10 Hauke: just register 5GHz channels in mac80211? Sep 28 13:36:13 zajec: hi Sep 28 13:36:22 no I haven't tried 5 GHz on b43 Sep 28 13:36:56 i wonder if it requires much work Sep 28 13:37:02 maybe just registering channels ;) Sep 28 13:37:07 does something work? Sep 28 13:37:23 i'll try later Sep 28 13:37:27 didn't start yet Sep 28 13:37:33 I had some problems with beaconing in brcmsmac on 5GHz Sep 28 13:38:01 the problem is still there ;-) Sep 28 13:40:04 Hauke: could it be data rate related? IIRC 11b/g defines some rates that aren't present in 11a, so it might cause issues if you try to send beacons in 5 GHz with a 2.4 GHz legacy rate Sep 28 13:40:32 KanjiMonster: yes I assume something like that Sep 28 14:17:33 nbd: Done. Ticket #14231 Sep 28 14:22:38 btw. patches should usually go to the mailing list Sep 28 14:22:44 see https://dev.openwrt.org/wiki/SubmittingPatches Sep 28 14:23:02 nbd: OK. Sep 28 15:26:58 nbd: on ticket #14111, how do I change the status to 'close'? Sep 28 15:27:47 are you logged in? Sep 28 15:28:11 nbd: Yes. And on the page, I only have "leave as new" option on modify ticket. Sep 28 15:28:41 odd Sep 28 15:28:53 you should ask mirko to check your permissions Sep 28 15:28:59 nbd: OK. Sep 28 15:42:01 good morning :) Sep 28 15:42:13 almost night, but ok Sep 28 20:43:10 build #412 of at91 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/412 Sep 29 00:15:10 build #335 of kirkwood is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/kirkwood/builds/335 Sep 29 01:42:00 mirko: ping **** ENDING LOGGING AT Sun Sep 29 02:59:59 2013