**** BEGIN LOGGING AT Wed Jan 23 02:59:59 2013 Jan 23 05:37:22 build #165 of at91 is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/165 Jan 23 06:22:56 build #154 of cobalt is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/154 Jan 23 06:23:39 build #164 of brcm47xx is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/164 Jan 23 07:17:35 build #153 of lantiq is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq/builds/153 Jan 23 07:22:35 build #123 of iop32x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/123 Jan 23 07:50:42 build #142 of avr32 is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/142 Jan 23 07:54:24 build #120 of gemini is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/120 Jan 23 08:16:30 cyrus r35302 trunk/package/network/ipv6/ipv6-support/ Makefile files/network-enable-ipv6.defaults * ipv6-support: don't overwrite existing setting across reflashs Jan 23 08:18:02 cyrus r35303 trunk/package/network/ipv6/6distributed * ipv6: remove empty directory Jan 23 10:01:47 build #135 of ixp4xx is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/ixp4xx/builds/135 Jan 23 10:13:30 jogo r35304 trunk/target/linux/ brcm63xx/profiles/104-No-WiFi.mk brcm63xx/profiles/200-GW6X00.mk brcm63xx/modules.mk brcm63xx/config-3.6 * bcm63xx: move bcm63xx-spi into the kernel Jan 23 10:13:45 jogo r35305 trunk/target/linux/ (18 files in 8 dirs) * generic: add b53 swconfig switch driver Jan 23 10:13:48 jogo r35306 trunk/target/linux/brcm63xx/patches-3.6/447-bcm63xx_enet_add_b53_support.patch * bcm63xx: expose the internal switch to b53 Jan 23 10:13:51 jogo r35307 trunk/ (26 files in 6 dirs) * bcm63xx: switch to b53 and swconfig Jan 23 10:24:52 build #140 of xburst is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/140 Jan 23 12:49:51 hi, is openwrt-devel@lists.openwrt.org closed to non-subscribers? Jan 23 12:50:30 yes Jan 23 12:51:28 fborges: registration requires no moderator approval, and you can disable the delivery of emails, so you can easily create a "write-only" account Jan 23 12:51:47 KanjiMonster, that sounds like a good solution Jan 23 12:52:25 KanjiMonster, I am subscribed using GMail but I was not able to use git-send-mail due to the 2-factor authentication setup Jan 23 13:43:21 luka r35308 packages/net/nginx/Makefile * [packages] nginx: update to 1.2.6 and revert r35201 Jan 23 13:48:27 I have some questions w.r.t source package from CVS, GIT, HG, SVN, etc., sourcing. Jan 23 13:49:14 Is it possible add an xz compression support when tarring the source package? Jan 23 13:49:53 fborges: it should work with application-only passwords no? Jan 23 13:50:09 just create a password for git-send-mail in the google webinterface Jan 23 13:51:20 loswillios, I did try that, and used gsmtp as application name (perhaps that was the mistake). Next thing I know, Google thinks my account is hacked, and required me to change password/confirm phone number etc Jan 23 13:51:36 mazilo: xz already works Jan 23 13:51:48 mazilo: ah you mean when compressing Jan 23 13:51:51 hmm Jan 23 13:52:01 cyrusff: Exactly. Jan 23 13:52:10 loswillios, the application name to use for git-send-email is "git-send-email"? (or the application name does not matter?) Jan 23 13:52:24 it doesn't matter Jan 23 13:54:06 loswillios, I may try that again later. But honestly, I already did it, and Google forced me into a big dance of security reset... Jan 23 13:55:19 Also, I notice in the section "DownloadMethod/git" of include/download.mk (on line #93), the ".git" subdir is removed before the downloaded source gets packaged. I understood this is to shrink the file size. However, bandwidth has already been committed to download the whole package. So, why not just NOT to remove the ".git" before packaging the source? Jan 23 13:55:42 loswillios, fwiw, I work for Red Hat, and decided to use the company's SMTP to send the patches. Since I am submitting FOSS code for a FOSS project, the company is OK with it (even though I am doing this on my own time and is not related to my job). Jan 23 13:55:59 mazilo: the git clone is shallow the .git is almost worthless (no history) Jan 23 13:56:04 *so the Jan 23 13:57:03 fborges: ok. I noticed a few hickups with 2-factor auth as well, but nothing like your Jan 23 13:57:13 loswillios, hence BTW the usage of a @gmail.com address as author/committer instead of mine @redhat.com address in the 2 patches I sent Jan 23 13:57:38 jow_laptop: That is true, but it contains some information that one can use to update the source to the latest git HEAD on-the-fly when compiling the source. Jan 23 13:58:28 how_laptop: without the ".git", one has to redownload the whole source package if wanted to compile with the latest git HEAD. Jan 23 13:59:04 sounds like micro-optimization to me Jan 23 13:59:18 if you actually develop on a package use the git-src override mechanism Jan 23 14:01:55 jow_laptop: Is git-src part of git? Jan 23 14:02:12 echo CONFIG_SRC_TREE_OVERRIDE=y >> .config Jan 23 14:02:33 git clone some_project; ln -s /patch/to/git/clone/.git package/whatever/git-src Jan 23 14:02:33 jow_laptop: OK. Thanks. Jan 23 14:02:53 then do you commits, updates whatever in the seaprate checkout Jan 23 14:03:06 and simpyl rebuild with package/whatever/{clean,compile} Jan 23 14:03:11 no makefile changes needed Jan 23 14:03:28 if you're done change the makefile to the commit you want to target and remove the git-src symlink Jan 23 14:03:48 jow_laptop: Thanks. Jan 23 14:08:49 huh, that's yet another way of doing it. Jan 23 14:08:50 neat Jan 23 14:54:46 does this look reasonable? libjpeg-turbo is currently broken otherwise ... http://patchwork.openwrt.org/patch/3189/ Jan 23 15:39:22 build #122 of mcs814x is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/mcs814x/builds/122 Jan 23 15:43:36 build #111 of octeon is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/111 Jan 23 16:30:41 nbd r35309 trunk/target/linux/ar71xx/image/Makefile * ar71xx: sync kernel sizes where possible to fix images for some devices (#12874) Jan 23 16:32:50 nbd r35310 trunk/target/linux/ generic/patches-3.8/604-netfilter_cisco_794x_iphone.patch generic/patches-3.7/604-netfilter_cisco_794x_iphone.patch * kernel: remove the cisco sip phone patch for 3.7 and 3.8 as well Jan 23 16:34:42 nbd, hi! in netifd, if I have a iface with protocol 'none', but it has an ip, if I call /etc/init.d/network restart, it cleans up the ip for that interface. is there an way to prevent that? Jan 23 16:35:18 jwendell: how do you usually set the ip? Jan 23 16:35:28 openvpn does that for me :) Jan 23 16:35:35 ah Jan 23 16:35:36 hm Jan 23 16:36:02 i think i'll add something to netifd to prevent it from removing addresses by default on specific devices Jan 23 16:36:49 one approach would be: don't touch an interface with proto 'none' Jan 23 16:37:16 not sure yet Jan 23 16:37:51 we have in /etc/config/network: Jan 23 16:37:53 config interface 'vpn' Jan 23 16:37:54 option ifname 'tun0' Jan 23 16:37:54 option proto 'none' Jan 23 16:38:49 we need that to be notified by hotplug when that interface is up/down Jan 23 16:39:30 right Jan 23 16:40:54 does that make sense? Jan 23 16:41:08 yes Jan 23 16:41:14 I'd like to be notified, but don't want its IP to be touched Jan 23 16:41:21 but i think i'll add an extra option instead of relying on proto none Jan 23 16:41:47 that's another approach I was thinking too Jan 23 16:42:12 it should be a device option, and not an interface option Jan 23 16:42:21 because multiple interfaces can bind to the same device Jan 23 17:45:10 nbd, is it easy to do that? if you give me a hint where to look/touch I could try to cook a patch... Jan 23 17:45:50 the code is the easy part, i still haven't decided where to put it exactly Jan 23 17:46:31 an uci option 'donottouch' = 1 Jan 23 17:46:32 :D Jan 23 18:11:24 nbd: how would a proto handler react if it is bound to a device which has this "do not touch" flag set? will netifd simply discard any config update requests? Jan 23 18:12:07 i wouldn't call it do not touch, the flag that i'm thinking about would only prevent the initial address purge Jan 23 18:12:22 I understand Jan 23 18:13:30 what about config device; option name tun0; option external true Jan 23 18:14:19 or if you prefer options that describe the action, what about "bool deconfig" with default 1 Jan 23 18:14:36 yeah, i guess we could even reuse the existing internal 'dev->external' flag Jan 23 18:14:44 or at least trigger setting it Jan 23 18:18:21 build #126 of adm5120 is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/126 Jan 23 18:32:26 build #114 of mpc52xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/114 Jan 23 18:39:32 hi! Jan 23 18:39:43 what is the magic to rebuild a kernel package module only? Jan 23 18:39:53 kmod-crypto-core in particular Jan 23 18:43:02 make target/compile package/kernel/compile Jan 23 18:43:29 have to do them all, then? Jan 23 18:43:39 yes Jan 23 18:43:43 aw. Jan 23 18:44:04 adding more shortcuts for rebuilding individual modules does not make any sense at all Jan 23 18:44:31 sure. Anyway, I think the issue was something else, that's since been fixed Jan 23 18:44:42 crypto-core wasn't being built at all Jan 23 18:45:03 it is no longer being built as module on 3.7 Jan 23 18:45:40 so I see from SVN. but on 3.38 here Jan 23 18:45:43 3.3.8 Jan 23 18:45:49 ah Jan 23 18:58:36 * Chocks tries the magic Jan 23 18:59:12 * Chocks pokes wifi detect Jan 23 19:19:00 hm, usb stuff missing my USB IDs for Wifi Jan 23 19:20:11 * Chocks tries newer compat-wireless Jan 23 19:40:50 nbd: blah. gotta add some more numbers to compat-wireless to detect chipset Jan 23 19:41:08 expect incoming patch, if I have enough time and it all works, etc, etc Jan 23 19:41:28 what chipset, what numbers? Jan 23 19:41:37 [ 6.930000] phy0 -> rt2800_init_eeprom: Error - Invalid RT chipset 0xd664 detected. Jan 23 19:41:43 Qcom Jan 23 19:41:58 [ 2.390000] usb 1-2.3: New USB device found, idVendor=18e8, idProduct=6229 Jan 23 19:42:04 ok. i don't want to carry more random non-upstream rt2x00 patches Jan 23 19:42:09 so please submit that stuff to linux-wireless@ Jan 23 19:42:15 aw, ok Jan 23 19:42:20 and i'll put it into openwrt from there Jan 23 19:42:40 that kind of stuff belongs upstream and not in local distro patches Jan 23 19:42:47 will carry locally for moment. I'm not even sure customer wants to stick with openwrt. they have a hard-on for Angstrom Jan 23 19:42:50 yeah, understood Jan 23 19:43:27 heh, angstrom Jan 23 19:43:31 righ Jan 23 19:43:39 yes, various comedy there Jan 23 19:43:42 i've worked with some companies that had some experience with that Jan 23 19:43:53 but once their guys tried openwrt they figured out that it was much easier to maintain than OE based stuff Jan 23 19:43:56 ;) Jan 23 19:43:59 I only have superficial knowledge of it, but I can make a bunch of guesses Jan 23 19:44:23 it's what they know, but I suspect the package base is much much smaller Jan 23 19:44:35 and they have some stuff they want to integrate with ours, so we'll see Jan 23 19:44:45 angstrom/OE does have a lot more packages than openwrt Jan 23 19:44:51 hm, ok Jan 23 19:45:04 downsides of it? Jan 23 19:45:13 and in some ways that's part of the problem, because it makes it harder for them to keep stuff clean and buildable, or make changes to their infrastructure Jan 23 19:45:19 right Jan 23 19:45:28 OE typically is a lot more complex than OpenWrt Jan 23 19:45:56 right. I haven't used OE in 15 years. back in the very early days when it was more like OpenWrt is now. Jan 23 19:45:57 so when starting a new project, it typically takes quite a bit longer to get *something* running Jan 23 19:46:06 all python driven builds Jan 23 19:47:53 we'll push them to OpenWrt of course, but if they want Anstrong and they want to pay for it, we'll do it Jan 23 19:48:10 it's more for us to maintain Jan 23 20:31:00 florian r35311 packages/libs/ (7 files in 2 dirs) * elfutils: update to 0.155 Jan 23 20:34:56 luka, ping Jan 23 21:49:07 build #157 of ppc40x is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/ppc40x/builds/157 Jan 23 22:29:32 nbd: is it true that the kernel extra symbols strip is more thorough in recent openwrt? Jan 23 22:35:35 nbd: seems to be the case. no matter :o Jan 24 00:45:59 build #150 of x86 is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/150 Jan 24 00:48:10 build #158 of ramips is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/ramips/builds/158 Jan 24 00:52:05 build #149 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/149 Jan 24 00:55:46 build #149 of ppc44x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/149 Jan 24 01:15:29 build #166 of at91 is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/166 Jan 24 01:55:53 anyone want to checkin the xz dl_pack'ing mazilo asked about earlier? Jan 24 02:28:19 build #165 of brcm47xx is complete: Failure [failed shell_14] Build details are at http://buildbot.openwrt.org:8010/builders/brcm47xx/builds/165 Jan 24 02:59:39 build #154 of uml is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/154 **** ENDING LOGGING AT Thu Jan 24 02:59:58 2013