**** BEGIN LOGGING AT Thu Mar 17 02:59:57 2011 Mar 17 05:34:38 build #98 of orion is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/orion/builds/98 Mar 17 06:02:20 build #77 of etrax is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/77 Mar 17 06:44:08 gmorning Mar 17 07:08:14 acinonyx * r26196 /trunk/package/acx/files/lib/wifi/acx.sh: [package] acx: Add newline at end of autodetected wireless config (cosmetic) Mar 17 07:09:41 acinonyx * r26197 /trunk/package/madwifi/files/lib/wifi/madwifi.sh: [package] madwifi: Add newline at end of autodetected wireless config (cosmetic) Mar 17 07:10:23 acinonyx * r26198 /trunk/package/hostap-driver/files/lib/wifi/hostap.sh: [package] hostap-driver: Add newline at end of autodetected wireless config (cosmetic) Mar 17 07:25:57 acinonyx * r26199 /trunk/package/hostap-driver/files/lib/wifi/hostap.sh: [package] hostap-driver: Fix adhoc mode Mar 17 07:26:00 acinonyx * r26200 /trunk/package/hostap-driver/files/lib/wifi/hostap.sh: [package] hostap-driver: Use 'any' for unspecified SSIDs Mar 17 07:26:04 acinonyx * r26201 /trunk/package/acx/files/lib/wifi/acx.sh: [package] acx: Major rework of acx.sh script Mar 17 07:26:08 acinonyx * r26202 /trunk/package/hostap-driver/files/lib/wifi/hostap.sh: [package] hostap-driver: Remove newline at start of config (cosmetic) Mar 17 07:26:11 acinonyx * r26203 /trunk/package/hostapd/patches/350-hostap_multicall_fix.patch: [package] hostapd: Fix hostap driver for multicall binary builds Mar 17 08:17:23 build #83 of uml is complete: Failure [failed compile_10] Build details are at http://tksite.gotdns.org:8010/builders/uml/builds/83 Mar 17 08:18:21 I'm trying to build a custom package with OpenWRT and it's failing to find __ns_initparse which is meant to be part of libresolv Mar 17 08:18:28 I have -lresolv specified Mar 17 08:18:31 but no joy Mar 17 09:32:25 Olipro: what is the message? Mar 17 09:34:59 never mind, fixed it Mar 17 09:35:39 build #73 of mpc52xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/mpc52xx/builds/73 Mar 17 09:58:43 how would I go about compiling a specific package? Mar 17 09:58:49 say... bitlbee Mar 17 09:59:43 `make package/bitlbee/compile` Mar 17 12:08:32 build #84 of x86 is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/x86/builds/84 Mar 17 14:16:57 Olipro: do you mean compile, or do you mean prepare package to be installed? Mar 17 14:17:33 heh, I got it all figured out now Mar 17 14:17:44 Olipro: ok, cool Mar 17 14:17:48 managed to create a package for bitlbee3 Mar 17 14:17:57 I created a ticket and attached it Mar 17 14:18:07 I assume that's the right way to submit a package for inclusion into OpenWRT? Mar 17 14:18:20 You can also submit to the -devel list Mar 17 14:20:24 I think more often people submit to the -devel list, not sure which is better in terms of being seen. Mar 17 15:38:21 jow * r26204 /trunk/package/base-files/ (Makefile files/etc/sysctl.conf): [package] base-files: update sysctl.conf for modern kernels Mar 17 16:52:53 * swalker goes to close #8954 Mar 17 16:57:43 Olipro_: either way tends to be a crap-shoot Mar 17 16:58:06 you mean getting a package inserted? Mar 17 16:58:14 well, I've a mind to ask for commit access Mar 17 16:58:25 since according to the wiki, "ask and ye shall receive" Mar 17 16:58:49 and I pinky swear not to break anything Mar 17 16:58:54 anything basically. Mar 17 17:46:28 libX11 is failing on creating libX11.ps? Mar 17 17:53:17 cshore * r26205 /packages/net/miniupnpd/ (Makefile files/upnpd.config files/upnpd.config.example): Mar 17 17:53:17 [net] IP Addresses and Names: miniupnpd: Added rule to allow high ports be Mar 17 17:53:17 default. This fixes miniupnpd default to config so that it works as expected Mar 17 17:53:17 after installed and enabled. Also added comments for the UI, and removed the Mar 17 17:53:17 example config file as it is no longer needed. Mar 17 17:54:20 swalker: hey now, I know how to use svn >_> Mar 17 18:47:02 cshore: r26205: i understand wanting the default "allow" to restrict ext_ports to 1024-65535, but is there any reason to restrict int_ports too? the router shouldn't care what port some internal system wants to use. Mar 17 18:49:34 markmentovai: it makes it harder for a malicious client on the internal host to affect the host when running as a user process Mar 17 18:50:37 ok, i guess i'll buy that for a default rule. Mar 17 18:50:46 cshore: btw. just a quick heads up: the overlay stuff is going to work much better in the future. i managed to get overlayfs working as a replacement for mini_fo in 2.6.37 Mar 17 18:51:04 it already feels much more stable than mini_fo Mar 17 18:51:14 markmentovai: the only reason for the commit was that I convinced him that the default reject is bad for ootb experience Mar 17 18:51:25 nbd: excellent! thank you. Mar 17 18:51:48 markmentovai: in particular since it deviates from how it used to be (miniupnd works after enabling) Mar 17 18:52:07 so the compromise was > 1024 Mar 17 18:52:25 jow_laptop: understand. practically, as long as secure_mode is on, i don't think it's necessary to be restrictive at all about int_port, but i can understand why you might want it to be >= 1024. Mar 17 18:57:31 markmentovai: I prefer to be paranoid about firewalls, especially since there are sometimes Windows hosts connected to my internal guest network. Mar 17 18:58:22 i can respect that. thanks. Mar 17 18:59:16 markmentovai: thank Jo...as he says my preferred default was deny all and only allow if the user knows what they're doing. Mar 17 19:00:19 when i thanked jo for sysctl.conf, he blew me off, should i give him a second chance? :) Mar 17 19:00:43 blew you off? Mar 17 19:00:51 I don't understand that term Mar 17 19:01:01 sorry, "ignored me" Mar 17 19:01:20 ah, sorry for that. I'm at a conference here and get distracted a lot Mar 17 19:01:59 not a problem, my feelings weren't hurt. thanks for both sysctl.conf and the miniupnpd default. Mar 17 19:09:07 cshore * r26206 /packages/net/miniupnpd/files/miniupnpd.init: [net] IP Addresses and Names: Fixed using single IP addresses instead of always required n.n.n.n/n even for single IP's. Mar 17 19:12:20 nbd * r26207 /trunk/package/base-files/files/lib/ (3 files in 2 dirs): base-files: add initial support for mounting overlayfs Mar 17 19:12:27 nbd * r26208 /trunk/package/base-files/files/lib/firstboot/99_10_with_fo_cleanup: base-files: overlayfs whiteouts depend on a specific xattr entry, which does not get preserved by the tmpfs->overlay copy. work around this by cleaning all copied "(overlay-whiteout)" symlinks after the switch to jffs2 Mar 17 19:12:28 nbd * r26209 /trunk/target/linux/generic/ (15 files in 2 dirs): kernel: replace mini_fo with overlayfs for 2.6.37 Mar 17 19:12:42 :) Mar 17 19:13:24 * nbd recommends make target/linux/clean Mar 17 19:14:45 * jow_laptop tries Mar 17 19:29:44 cshore * r26210 /packages/net/yate/Makefile: [net] Telephone: Yate: Removed @BROKEN from Yate. It appears to work, except for the modules marked as broken. Mar 17 20:03:44 nbd: does a 'sysupgrade' to >r26209 work? Mar 17 20:04:04 i don't see why not Mar 17 20:04:10 haven't tested it though Mar 17 20:04:17 ok Mar 17 20:11:24 nbd * r26211 /trunk/target/linux/ar71xx/ (config-2.6.32 patches-2.6.32/): ar71xx: remove 2.6.32 support Mar 17 20:19:12 * loswillios tries too Mar 17 20:20:15 hi guy Mar 17 20:20:17 guys* Mar 17 20:20:26 anyone have time to try building libmysql? Mar 17 20:20:56 Bartman007: :) Mar 17 20:24:09 nbd * r26212 /trunk/package/ (hostapd/Config.in hostapd/Makefile mac80211/Makefile): hostapd: add a separate config symbol for 802.11n driver support Mar 17 20:25:54 nbd * r26213 /trunk/target/linux/generic/ (15 files in 2 dirs): kernel: replace mini_fo with overlayfs for 2.6.38 Mar 17 20:27:15 nbd: how do we handle the block-mount/extroot stuff for mini_fo vs. overlay fs since it's kernel version dependant....there there a way to do a different package make depending on kernel version? Mar 17 20:27:36 cshore: look at what i did in base-files Mar 17 20:27:41 ok Mar 17 20:28:18 I can't do it right now, I've got to work (so I keep having to remind myself)I just noticed your commit. Mar 17 20:28:35 but will look at when I get a chance. Mar 17 20:28:55 since only the mount command needs to be changed and the rest is pretty much the same, you can just use grep -q overlay /proc/filesystems Mar 17 20:29:02 to check for overlayfs Mar 17 20:29:31 ok Mar 17 20:30:34 btw, do you know if Florian is working on updating brcm63xx kernel version...I know we've both been pretty busy. Mar 17 20:34:36 no idea Mar 17 20:34:59 * nbd is happy to finally have a decent overlay filesystem Mar 17 20:35:17 * cshore cheers nbd Mar 17 20:35:21 only took more than half a decade to get there ;) Mar 17 20:35:26 heh Mar 17 20:36:18 nbd: how does overlayfs store the changes? Mar 17 20:39:29 same as mini_fo Mar 17 20:39:34 only difference is whiteout handling Mar 17 20:39:40 mini_fo creates weird META_ files Mar 17 20:39:51 ok...yeah, those are replaced by and xattr? Mar 17 20:39:52 cshore: 2.6.37.2 works for me (on bcm63xx); 2.6.38 is missing lzma support for squashfs Mar 17 20:39:59 overlayfs creates symlinks to an (overlayfs-whiteout) with a special xattr set Mar 17 20:40:08 ok Mar 17 20:41:09 with 2.6.38 we'll probably switch to xz once i figure out how to handle compression parameters Mar 17 20:41:14 i think i'll also add xz support to jffs2 Mar 17 20:41:23 that'll reduce the bloat a bit Mar 17 20:41:33 because we don't need two decompressors for the same thing Mar 17 20:41:33 KanjiMonster: ok, thanks for letting me know Mar 17 20:43:32 nbd: what do you think of removing the blkid and swap-utils from busybox and only installing them for block-extroot. The extra functionality of 'real' blkid could be nice, and 30k or so was the difference IIRC. (or maybe it was 15k, it's been a while). Since we're talking about reduce bloat, and blkid/swap-utils are only needed if you have a block device and therefore are more likely to have extra space. Mar 17 20:44:04 makes snse Mar 17 20:44:06 sense Mar 17 20:44:46 ok, I'll do that after I update for overlayfs Mar 17 20:48:28 Someday I will have a 4/16 47xx router to test with....I want to make sure that a firmware with a reasonable amount of stuff still works on 4/16 without extroot. Mar 17 20:48:55 and I still don't have 47xx for testing Mar 17 20:49:11 and I don't want to buy a Broadcom router again Mar 17 20:50:11 i'm not too worried about the size on 4/16 devices Mar 17 20:50:16 trunk is already in better shape there than backfire Mar 17 20:50:48 I work on it because I've got 63xx equipment and that's what my job has been, but for my personal stuff I want to switch to devices that are better supported by openwrt Mar 17 20:50:58 nbd: cool Mar 17 20:51:12 nbd: are there still people trying 2/8? Mar 17 20:51:28 probably not at the moment Mar 17 20:51:49 once the openwrt user space rework progresses some more, this will become more realistic Mar 17 20:52:01 because at some point you'll be able to remove the shell from the config if you want to ;) Mar 17 20:52:01 nbd: do you think we should have a -mircro version for that? Mar 17 20:52:23 nbd: heh, fully configurable Mar 17 20:53:01 at the moment a -micro version does not make much sense, because i don't think we'll be able to get current trunk down to 2/8 with reasonable amounts of effort Mar 17 20:53:11 nbd: ok Mar 17 20:53:22 nbd: what would be needed to get there in the future? Mar 17 20:53:34 not sure Mar 17 20:53:41 making busybox optional would free up lots of space Mar 17 20:53:47 we also need to find more ways to de-bloat the kernel Mar 17 20:53:57 nbd: yeah, the kernel is big Mar 17 20:54:53 nbd: what would you use for a shell without busybox? (i.e. that's small enough)? Mar 17 20:55:01 no shell at all Mar 17 20:55:23 nbd: ah, so total custom init/execution of stuff Mar 17 20:55:27 at some point i want to have a fully functional userland with cli and web configuration working without a shell Mar 17 20:55:40 right now i'm rewriting all the network init stuff Mar 17 20:55:59 i'm writing a c daemon that manages all this Mar 17 20:57:01 nbd: ok. If I ever get the time, I'll look the luci2 init stuff and see about moving preinit stuff I've done to it in C Mar 17 20:57:25 i don't think you need to look at the current stuff, i intend to rework quite a bit of it Mar 17 20:57:39 nbd: ok, I'll wait then Mar 17 20:58:40 nbd: unless there is something I can help with (though I suspect this is a case of too many cooks spoil the broth) Mar 17 20:59:24 well, you could look into the init stuff and see what you think about the design or if you can find some ways to make it better Mar 17 20:59:47 i didn't really get around to working out the design for the init stuff myself yet, the network stuff is more important to me right now Mar 17 21:00:17 nbd: ok...I'll look into then...I've been putting off doing that already because I've been expecting the luci2 changes Mar 17 21:00:21 you could also take a look at ubus and tell me what you think about that Mar 17 21:00:29 in the future i want to use that for all of our core services Mar 17 21:00:36 nbd: ok Mar 17 21:07:07 nbd: oh what format does ubus communicate with (e.g. JSON ?) Mar 17 21:07:54 internally it uses a format that i created (blobmsg), parse and generator are part of libubox Mar 17 21:08:04 the cli automatically converts that to and from JSON Mar 17 21:08:49 nbd: ok, so things that would link would use libubox blobmsg, while other stuff would use JSON through the cli ? Mar 17 21:08:57 yes Mar 17 21:10:57 ok, until I read the source I probably don't have any other non-spoonfeed-me questions, so I'll get back to work for now. Mar 17 21:14:11 t Mar 17 21:14:13 oops Mar 17 23:11:39 ok, overlayfs definitely isn't bug free yet Mar 17 23:11:49 rmdir still causes deadlocks Mar 17 23:11:57 but i've just made a fix for that ;) Mar 17 23:12:12 unlike mini_fo, overlayfs is easy to read and fix, even without deep vfs knowledge Mar 17 23:15:31 cool Mar 17 23:16:53 nbd * r26214 /trunk/Config.in: add a kernel config option for magic sysrq Mar 17 23:17:02 nbd * r26215 /trunk/target/linux/generic/ (2 files in 2 dirs): kernel: fix an overlayfs deadlock on rmdir Mar 17 23:20:25 build #101 of ubicom32 is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/101 Mar 17 23:20:35 nbd: Cool, thanks a lot for working on the fo mess Mar 17 23:23:22 i wanted to do this for a long time now, but simply hadn't found a suitable replacement until now Mar 18 02:18:34 nbd * r26216 /trunk/package/hostapd/patches/560-disable_ctrl_iface_mib.patch: hostapd: add a patch to remove some unnecessary control interface functionality **** ENDING LOGGING AT Fri Mar 18 02:59:57 2011