**** BEGIN LOGGING AT Mon Jan 26 02:59:59 2015 Jan 26 05:19:12 build #51 of netlogic is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/netlogic/builds/51 Jan 26 06:47:48 are the buildbot configs posted somewhere? interested what make targets the ar71xx build builds Jan 26 06:50:42 ah nm I can see the commands on the buildbot page Jan 26 07:30:08 build #320 of x86_64 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/x86_64/builds/320 Jan 26 09:08:37 build #318 of malta is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/malta/builds/318 Jan 26 09:26:43 build #851 of ramips is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/851 Jan 26 10:33:17 build #822 of rb532 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/822 Jan 26 14:53:56 i'd like to add ssl support to opkg Jan 26 14:54:16 would -nossl/-openssl/-polarssl packages be appropriate? Jan 26 14:54:27 the latter two would use the curl instead of the wget backend Jan 26 14:54:37 curl should work with either Jan 26 15:00:59 any suggestions, ideas? Jan 26 15:22:50 lgierth: why not simply install a full wget? Jan 26 15:23:51 jow_laptop: i dunno, can i build wget with polarssl? i'm trying to fit on 4mb and have only a few 100kb free Jan 26 15:26:53 no, apparently not Jan 26 15:29:50 if i can make it fit, wget+openssl is an option Jan 26 15:30:10 i should probably also check whether polarssl supports the particularities of our ssl certs Jan 26 16:32:58 karlp: you may remember that I said I'm going to push a JSON-C lua binding eventually, I did it now Jan 26 16:36:00 what's ugpraded? Jan 26 16:36:22 there's a new package luci-lib-jsonc in the luci feed Jan 26 16:37:23 you can use it like require "luci.jsonc"; t = luci.jsonc.parse('{"some": [ "json", "data" ]}'); s = luci.jsonc.stringfy(t); print(s)' Jan 26 16:37:46 *stringify Jan 26 16:37:49 is it as wkard to install locally as luci is? Jan 26 16:39:24 mkdir luci; gcc -O2 -Wall -shared -o luci/jsonc.so jsonc.c -ljson-c Jan 26 16:39:51 thats how I tested it locally Jan 26 16:40:24 what's the advantage of this again? vs dkjson? Jan 26 16:40:41 or is this going to be the json lib used in C land as well, instead of json-c ? Jan 26 16:41:10 oh,. this is _for_ libjson-c. Jan 26 16:42:15 yeah. I want to get rid of luci.json (the lua variant) Jan 26 16:42:24 and the custom lua to json formatting Jan 26 16:43:08 we've just been using dkjson for a while now, and we're on BB now, not rushing onto CC, but thanks for lett me know. Jan 26 16:43:09 anyhow, I just remembered that you had some issues with luci.json a while back (probably why you use dkjson now) so I thought you might be interested Jan 26 16:43:29 yeah, using a mix of dkjson and libjson-c Jan 26 16:44:38 thanks for the update, it's appreciated Jan 26 16:45:23 what is this "upgraded" from r44134 ? Jan 26 16:48:27 The called action terminated with an exception: Jan 26 16:48:27 stack traceback: Jan 26 16:48:27 [C]: in function 'assert' Jan 26 16:48:28 /usr/lib/lua/luci/dispatcher.lua:430: in function 'dispatch' Jan 26 16:48:30 /usr/lib/lua/luci/dispatcher.lua:171: in function Jan 26 16:48:55 */usr/lib/lua/luci/cbi.lua:311: Unable to read UCI data: openvpn Jan 26 16:49:18 DonkeyHotei: either /etc/openvpn has syntax errors or it does not exist Jan 26 16:49:25 */etc/config/openvpn Jan 26 16:49:59 karlp: on nand systems, all running processes must be terminated to release the rootfs, including pid 1 Jan 26 16:50:34 karlp: upgraded is a tiny pid 1 server which procd exec's into in order to move itself away from the rootfs Jan 26 16:50:56 release rootfs = ubidetach Jan 26 16:51:05 alright. Jan 26 16:51:24 I made its build optional because it does not link on avr32 Jan 26 16:51:30 and avr32 does not need it Jan 26 16:51:54 there's a whole pile of this procd stuff that is "use the source" documented only, and the source has no comments, or even a description. Jan 26 16:55:19 is there any chance of this being regenenerated? http://luci.subsignal.org/api/luci/index.html Jan 26 16:55:26 there seems to be lots missing these days Jan 26 16:55:32 will do it soon Jan 26 17:02:26 do you know why this sort of thing happens? http://pastebin.com/fneeSkMa Jan 26 17:13:02 tftpboot 0x83000000 openwrt-ar71xx-generic-rme-eg200-kernel.bin; bootm 0x8300000 boots the kernel ok. Jan 26 17:14:06 (except it fails, as it doesn't have disk that way. Jan 26 17:22:45 is it possible my kernel is too big now? Jan 26 17:31:31 hrm, tftpboot 80060000 blah-sysypgrade.bin; imi 8006000 verifyes checksum ok. Jan 26 17:31:42 must be in the eraseing and writing to flash Jan 26 17:38:00 bleh, rerunning the same commands and it works now. Jan 26 17:38:15 * karlp does not approve of this sort of behaviour :) Jan 26 17:53:00 karlp: interesting. i'm using dkjson too, committed a package to packages.git Jan 26 17:53:11 shortly after the BB freeze, so it's CC only Jan 26 17:57:16 yeah, we were just carrying it locally, it's a single file, so easy to manage Jan 26 18:30:05 hrm, I wonder if this is failing to write to flash properly if the sd card I'm trying to get working is inserted. Jan 26 18:31:50 ahha! other sdcard works! Jan 26 18:32:04 ok, progress. home time instead Jan 26 18:58:28 build #382 of mvebu is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/mvebu/builds/382 Jan 26 18:59:21 build #298 of sunxi is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/sunxi/builds/298 Jan 26 19:13:35 build #276 of omap is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/omap/builds/276 Jan 26 19:37:28 build #666 of mpc52xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/666 Jan 26 20:00:04 build #688 of adm5120 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/688 Jan 26 20:03:55 build #703 of ep93xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/703 Jan 26 20:17:18 build #295 of mpc83xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mpc83xx/builds/295 Jan 26 20:44:12 kaloz r44153 trunk/target/ linux/mvebu/base-files/sbin linux/mvebu/base-files/etc/crontabs/root linux/mvebu/base-files/sbin/fan_ctrl.sh linux/mvebu/base-files/etc/crontabs * mvebu: temporary el cheapo' fan control script Jan 26 21:17:21 jow_laptop: ich hab mal ein bisschen probiert, re: opkg/ssl Jan 26 21:17:23 http://paste.debian.net/142421/ Jan 26 21:17:39 s/german/english/ Jan 26 21:17:52 i sketched a bit of code re: opkg/ssl -- not tested yet, but builds Jan 26 21:18:10 do you see any obvious problem with this approach? Jan 26 21:18:32 it just adds another variant, opkg-polarssl Jan 26 21:29:12 lgierth: you probably could remove in a separate patch: $(if $(CONFIG_GCC_VERSION_4_3)$(CONFIG_GCC_VERSION_4_4), Jan 26 21:29:30 openwrt oldest gcc is 4.6 afaik Jan 26 21:31:55 plntyk: yeah :P Jan 26 21:31:57 * 32d6f1b - [toolchain/gcc]: drop 4.5 support (1 year, 10 months ago) Jan 26 21:32:15 and you seem to be right about 4.6 too Jan 26 21:32:35 (i'm just skimming through the trunk log) Jan 26 21:46:03 is there any uci support for vpnc? Jan 26 23:45:15 build #792 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/792 Jan 27 00:09:07 nbd r44154 trunk/package/kernel/linux/modules/other.mk * kernel: fix regression in building kmod-serial-8250 on linux 3.8 (#18821) Jan 27 00:10:42 in desperation to get a switch to configure i just wrote a monstrosity that loops over all registers, logs them, resets the switch, logs them all again and prints those that are different, this will take quite some reading Jan 27 00:10:44 build #412 of mxs is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/mxs/builds/412 Jan 27 00:36:23 meh, opkg-polarssl keeps pulling in openssl instead Jan 27 00:36:56 http://paste.debian.net/142421/ Jan 27 00:51:31 lgierth: I got an error in line 69 of patch when trying "git apply " Jan 27 00:51:31 (when did upstream opkg get polarssl support?) Jan 27 00:51:59 swalker: curl has Jan 27 00:52:18 the dependency on libpolarssl is purely to trigger libcurl picking it up Jan 27 00:52:34 ./scripts/diffconig.sh supports my idea that it should work Jan 27 00:52:40 plntyk: hrm? Jan 27 00:52:53 let me pull, should actually be a fresh CC trunk Jan 27 00:54:10 plntyk: ah that's just an output of git diff Jan 27 00:54:16 let me paste a proper patch Jan 27 00:55:39 that approach is flawed Jan 27 00:55:43 curl has multiple build variants Jan 27 00:55:48 for different ssl implementations Jan 27 00:55:52 if I read configure.ac correctly the the curl thing still pulls in openssl Jan 27 00:55:58 see http://git.yoctoproject.org/cgit/cgit.cgi/opkg/tree/configure.ac Jan 27 00:57:00 libssl-curl -> NEED_SSL_LIBS="yes" -> openssl tests Jan 27 00:57:15 nbd: not variants, it switches based on config vars Jan 27 00:57:40 or, mmh. Jan 27 00:58:19 right, my mistake. it's config options. it still doesn't work this way though Jan 27 00:58:21 i might have added CONFIG_LIBCURL_POLARSSL=y to the .config by hand... Jan 27 00:58:42 nbd: got any idea how it *might* work? Jan 27 00:59:15 well, in the interest of having something small, it might be a good idea to look into uclient-fetch (which uses ustream-ssl) Jan 27 00:59:18 and add the missing features Jan 27 00:59:21 and make opkg use it Jan 27 00:59:35 that way it'll get small enough to let us enable it by default Jan 27 00:59:40 curl is bloated Jan 27 00:59:59 i'll have to write c, eh :/ Jan 27 01:00:34 yeah Jan 27 01:00:36 bbl Jan 27 01:02:00 build #51 of ramips.mt7621 is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/ramips.mt7621/builds/51 Jan 27 01:02:30 plntyk: i had the impression it should only pick up openssl if it is available (i.e. libopenssl package), and that something depends on the libopenssl package Jan 27 01:05:05 grr, after a soft reset, this switch refuses to go into the state that the bootloader puts it in :( must be something more than the immediately available registers, because i've copied them 1:1 Jan 27 01:06:54 is it common for there to be some kind of converter / phy between switch and SoC? Jan 27 01:08:34 guess i'll put it apart and find out Jan 27 01:09:12 well there is the normal "AC_ARG_ENABLE(curl,"#56 and there is "AC_ARG_ENABLE(ssl-curl," #89 - the last one explicitly sets need_ssl_libs which sets the same var as the openssl swich (#78-85) Jan 27 01:09:54 that http://git.yoctoproject.org/cgit/cgit.cgi/opkg/tree/configure.ac?id=9c97d5ecd795709c8584e972bfdf3aee3a5b846d should be the version used in openwrt Jan 27 01:10:41 i dont know how the curl and ssl-curl codepaths differ Jan 27 01:12:44 "opkg info opkg" shows "Depends: libc" for me - this means my opkg cannot fetch over https? Jan 27 01:13:05 i think it works now Jan 27 01:13:18 s/works/builds/ Jan 27 01:13:46 my previous config was echo CONFIG_PACKAGE_opkg-polarssl=y\nCONFIG_LIBCURL_POLARSSL=y >> .config Jan 27 01:14:17 i nuked that, then disabled opkg and enabled opkg-polarssl via menuconfig Jan 27 01:14:40 will be able to test in a bit Jan 27 01:16:49 no wait. damnn it Jan 27 01:17:03 i need to have another look at this when i'm actually awake Jan 27 02:16:13 nbd: using uclient-fetch might be easier than i thought **** ENDING LOGGING AT Tue Jan 27 02:59:58 2015