**** BEGIN LOGGING AT Sat Apr 21 03:00:00 2012 Apr 21 03:08:16 what is this 2.6.38 kernel you speak of Apr 21 03:33:30 jow * r31390 /trunk/toolchain/gcc/patches/ (4 files in 4 dirs): [toolchain] gcc: add patch to make the getenv() spec function nonfatal if requested environment variable is unset Apr 21 05:05:21 got a curious question about ath9k and the wndr3700; i notice that anything below 5dBm seems to be *greater* signal than 5dBm Apr 21 05:05:35 what could cause this? Apr 21 05:05:56 clients scanning report higher RSSI from the AP when i set it from 0-4dBm than if its at 5dBm Apr 21 06:29:52 Hi Apr 21 06:30:03 in uboot, how do I load firmware image directly to sdram and boot Apr 21 06:30:09 I am on serial console Apr 21 08:38:13 mirko * r31391 /trunk/target/linux/lantiq/base-files/etc/uci-defaults/network: (log message trimmed) Apr 21 08:38:13 [target/linux/lantiq] change /etc/config/network according to actual VLAN config to enable LAN port 4 on the ARV4520PW board Apr 21 08:38:13 The VLAN mapping on the ARV4520PW board is the following: Apr 21 08:38:13 port on Router: | VLAN in software: Apr 21 08:38:13 LAN 1 | 3 Apr 21 08:38:14 LAN 2 | 2 Apr 21 08:38:14 LAN 3 | 1 Apr 21 10:29:23 mirko * r31392 /trunk/toolchain/gcc/ (20 files in 2 dirs): [toolchain/gcc] add support for GCC 4.7-linaro (based on GCC 4.7.1) Apr 21 10:54:41 mirko * r31393 /trunk/toolchain/gcc/patches/4.7-linaro/400-fix-incorrect-optimization.patch: [toolchain/gcc] remove screwed up patch for gcc 4.7-linaro which got committed by accident Apr 21 11:17:57 juhosg * r31394 /trunk/target/linux/generic/patches-3.3/322-ppc4xx-crypto-compile-fix.patch: linux/3.3: fix crypto4xx build failure Apr 21 11:19:18 build #141 of avr32 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/141 Apr 21 13:01:23 juhosg * r31395 /trunk/target/linux/ramips/ (rt288x/config-3.2 rt305x/config-3.2 rt3883/config-3.2): ramips: remove hardcoded console parameter from kernel configs Apr 21 13:01:33 juhosg * r31396 /trunk/target/linux/ramips/ (14 files in 11 dirs): Apr 21 13:01:33 ramips: rt305x: add initial support for Tenda W306R V2.0 Apr 21 13:01:33 More device info at: http://wiki.openwrt.org/toh/tenda/w306r Apr 21 13:01:33 Signed-off-by: David Pearce Apr 21 13:01:34 [juhosg: fix checkpatch errors, remove kmod-phy and luci-ssl packages Apr 21 13:01:34 from the profile, change broken GPIO number] Apr 21 13:01:35 juhosg * r31397 /trunk/target/linux/ramips/image/Makefile: ramips: build initramfs image for the Fonera 2.0N Apr 21 13:01:35 juhosg * r31398 /trunk/target/linux/ramips/Makefile: ramips: add swconfig to the default package list Apr 21 13:01:36 juhosg * r31399 /trunk/target/linux/ramips/ (image/Makefile rt3883/profiles/asus.mk): ramips: rt3883: add profile and build sysupgrade image for the RT-N56U Apr 21 13:01:37 juhosg * r31400 /trunk/target/linux/ramips/files/arch/mips/ (include/asm/mach-ralink/rt305x.h ralink/rt305x/rt305x.c): ramips: rt305x: use different memory size setup for RT3352 Apr 21 13:01:39 juhosg * r31401 /trunk/target/linux/ramips/files/arch/mips/ (include/asm/mach-ralink/rt305x_regs.h ralink/rt305x/clock.c): ramips: rt305x: fix CPU clock detection on RT3352 Apr 21 13:01:39 juhosg * r31402 /trunk/target/linux/ramips/files/arch/mips/ (2 files in 2 dirs): ramips: rt305x: add OHCI/EHCI registration code for RT3352 Apr 21 13:01:42 juhosg * r31403 /trunk/target/linux/ramips/ (17 files in 4 dirs): ramips: add support for 3.3 Apr 21 13:01:43 juhosg * r31404 /trunk/target/linux/ramips/Makefile: ramips: switch to 3.3.2 Apr 21 13:01:44 juhosg * r31405 /trunk/target/linux/ramips/ (4 files in 4 dirs): ramips: nuke 3.2 suport Apr 21 13:07:36 juhosg * r31406 /trunk/target/linux/mpc52xx/config-3.3: mpc52xx: add kernel config for 3.3 Apr 21 13:07:38 juhosg * r31407 /trunk/target/linux/mpc52xx/Makefile: mpc52xx: switch to 3.3.2 Apr 21 13:07:39 juhosg * r31408 /trunk/target/linux/mpc52xx/config-3.2: mpc52xx: nuke 3.2 config Apr 21 14:57:26 juhosg * r31409 /trunk/target/linux/generic/patches-3.3/833-gpio_buttons.patch: Apr 21 14:57:26 linux/3.3: nuke the gpio_buttons patch Apr 21 14:57:26 It is superseded by the in-kernel gpio-keys-polled driver. Apr 21 15:08:48 hm,. I installed luci-app-openvpn to use it as inspiration for luci-app-openconnect. Apr 21 15:08:55 but can't actually see any evidence of it in the ui Apr 21 15:09:06 I can't add an openvpn type interface, which is what I'd have expected. Apr 21 15:09:10 and can't see it anywhere else either. Apr 21 15:09:29 rm /tmp/luci-indexcache Apr 21 15:09:50 does opkg not support post-install scripts? Apr 21 15:10:18 it does Apr 21 15:10:27 I still don't see it as a protocol when adding an interface Apr 21 15:10:38 jow_laptop: shouldn't luci apps automatically remove that file then? Apr 21 15:10:39 openvpn is not implemented as protocol Apr 21 15:10:48 Hi All Apr 21 15:10:53 ah, I see it under services. Apr 21 15:10:54 in fact luci-app-openvpn is not useful at all Apr 21 15:11:08 heh, I shall uninstall it and not try to use it as a template for an openconnect app then :) Apr 21 15:11:20 is there a better one to use as an example, or should I just try to work it out from the ppp support? Apr 21 15:11:27 in uboot, how do I load image and boot from ram? Apr 21 15:11:33 it provides very little abstraction, merely maps config options 1:1 to the ui Apr 21 15:11:54 a good ui module would somewhat consolidate that all and offer only the most needed stuff wrapped up in some wizzard or something Apr 21 15:12:30 I thought about making openvpn a protocol a few times but never came to a conclusion Apr 21 15:12:41 problem is that you have both per iface and global config Apr 21 15:12:42 for a vpn client that seems to make most sense. Apr 21 15:12:53 and for a server setting it does nto fit well Apr 21 15:12:55 yeah, although you could consider that to be an implementation detail Apr 21 15:13:04 it's the job of the code to sort that out and hide it from the user. Apr 21 15:13:19 can't you run multiple openvpn dæmons with different settings? Apr 21 15:13:22 ahcp tried to do something similar with the outcome that the client was implemented as proto and the server in another config, which confused both the implementors and the users Apr 21 15:13:52 where we provide an ethernet "server" with dhcp and radvd, etc., that's also just an interface with a protocol. Apr 21 15:14:29 anyway, we don't have an openconnect server so that's not strictly relevant for what I was doing. openconnect is client-only for now Apr 21 15:14:42 (I *do* have a test hack as a server, for easy IPv6 testing, but it's not something you'd use in anger) Apr 21 15:14:50 then a proto is probably fine. Model it after pptp then Apr 21 15:15:09 ok, thanks Apr 21 15:15:18 may do l2tp after that too. Apr 21 15:15:23 as client Apr 21 15:15:51 ideally the /e/c/network settings should just cover a few basic aspects and for the rest reference some external config Apr 21 15:16:07 otherwise it might get bloated quite a lot with all sorts of stuff, like certificates etc. Apr 21 15:16:20 yeah Apr 21 15:17:19 for openvpn I envisioned something like "config interface vpn; option proto vpn; option instance example" which would select "config openvpn example" from /etc/config/openvpn Apr 21 15:17:30 *option proto openvpn Apr 21 15:17:56 but then I was too lazy to figure out how to extract the runtime tun/tap iface name from a particular openvpn instance Apr 21 15:18:05 woulb probably need to be enforces somehow then Apr 21 15:18:26 like passing --dev-type tun/tap --dev openvpn-foo to the instance Apr 21 15:18:34 *need to be enforced Apr 21 15:18:35 openconnect takes a '-i' arg to *tell* it the iface name to use :) Apr 21 15:19:04 thats good Apr 21 15:19:15 I usually make the ifaces in the form $proto-$name Apr 21 15:19:24 but "openconnect" is very long Apr 21 15:19:28 I'd noticed :) Apr 21 15:19:31 vpn- Apr 21 15:19:35 or oc- Apr 21 15:19:41 yeah oc- is better Apr 21 15:22:28 when we've connected and got IP addresses, etc. ... Apr 21 15:22:36 Do we have to call uci to tell it that stuff Apr 21 15:22:49 or just set it up on the interface manually (as vpnc-script does) Apr 21 15:23:21 in the case of netifd you would tell the daemon to do it Apr 21 15:49:45 juhosg * r31410 /trunk/target/linux/ (5 files in 5 dirs): linux: add broken flag to targets which are using 2.6.30 Apr 21 15:49:47 juhosg * r31411 /trunk/target/linux/generic/ (config-2.6.30 patches-2.6.30/): linux/2.6.30: R.I.P. Apr 21 15:49:49 juhosg * r31412 /trunk/target/linux/generic/ (config-2.6.31 patches-2.6.31/): linux/2.6.31: R.I.P. Apr 21 16:39:57 juhosg * r31413 /trunk/tools/Makefile: tools: add xz dependency to automake Apr 21 16:48:38 juhosg * r31414 /trunk/target/linux/ixp4xx/ (5 files in 4 dirs): ixp4xx: remove 2.6.3{2,8} and 3.0 support Apr 21 16:48:39 juhosg * r31415 /trunk/target/linux/x86/ (config-default patches-2.6.32/): x86: remove 2.6.32 support Apr 21 17:01:33 juhosg * r31416 /trunk/target/linux/au1000/ (5 files in 3 dirs): au1000: remove 2.6.39 support Apr 21 17:03:54 jow_laptop: yeah, the way vpnc and openconnect works, it makes that fairly simple. They set environment variables with the information, then spawn a vpnc-script to set it up. Apr 21 17:04:10 which can either set it up directly, or can pass the information back to netifd/connman/networkmanager/etc. Apr 21 17:07:43 juhosg * r31417 /trunk/target/linux/au1000/patches-2.6.39/: au1000: remove empty dir Apr 21 17:12:56 juhosg * r31418 /trunk/target/linux/cns3xxx/patches-2.6.31/: cns3xxx: remove 2.6.31 support Apr 21 17:16:24 juhosg * r31419 /trunk/target/linux/octeon/ (config-2.6.30 patches-2.6.30/): octeon: remove 2.6.30 support Apr 21 17:22:51 juhosg * r31420 /trunk/target/linux/mpc83xx/ (config-2.6.32 config-default patches-2.6.32/): mpc83xx: remove 2.6.32 support Apr 21 17:27:06 juhosg * r31421 /trunk/target/linux/malta/ (patches-2.6.39/ patches-3.0/): malta: remove 2.6.39 and 3.0 support Apr 21 17:31:50 build #171 of s3c24xx is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/s3c24xx/builds/171 Apr 21 17:34:02 juhosg * r31422 /trunk/target/linux/xburst/ (config-2.6.36 patches-2.6.36/): xburst: remove 2.6.36 support Apr 21 17:36:23 juhosg * r31423 /trunk/target/linux/versatile/: linux: remove the empty versatile target Apr 21 17:40:24 juhosg * r31424 /trunk/target/linux/uml/ (5 files in 5 dirs): uml: remove empty patch dirs Apr 21 17:41:39 juhosg * r31425 /trunk/target/linux/rdc/patches-2.6.30/: rdc: remove empty patch dir Apr 21 17:47:32 juhosg * r31426 /trunk/target/linux/omap24xx/ (config-2.6.38 patches-2.6.38/): omap24xx: remove 2.6.38 support Apr 21 18:10:40 jow_laptop: hm, the luci-proto-pptp package contains just one, very simple, file. Apr 21 18:10:52 it doesn't seem to actually register the pptp protocol. proto_ppp.lua does that? Apr 21 18:11:35 so I'm playing with something like http://david.woodhou.se/proto_openconnect.lua Apr 21 18:11:39 any pointers would be appreciated. Apr 21 18:38:15 ok, the registration bit goes in /usr/lib/lua/luci/model/network/proto_openconnect.lua Apr 21 18:38:22 not in /usr/lib/lua/luci/model/cbi/admin_network/proto_openconnect.lua Apr 21 19:23:35 there is a problem building busybox tune2fs on 3.3, http://git.busybox.net/busybox/commit/?id=caf5768154301697430f0a3718838967d0f55f95 fixes it Apr 21 19:24:25 build #134 of iop32x is complete: Failure [failed compile_10] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/134 Apr 21 20:13:23 mirko * r31427 /trunk/package/firewall/files/firewall.config: Apr 21 20:13:23 [package/firewall] minor change: adjust formatting of firewall.config Apr 21 20:13:23 - remove trailing whitespaces (s/\ $//g) Apr 21 20:13:23 - replace spaces with tabs between options and values Apr 21 20:13:44 jow_laptop: ok, I have this kind of working. 'ifup foo' and 'ifdown foo' are starting and stopping openconnect. Apr 21 20:14:11 but in luci, the 'Stop' action is always there, even when it's already stopped and it should be a 'Start' button. Apr 21 20:26:40 hm, doh. That's because there's a 'Connect' button next to it :) Apr 21 20:27:02 I expected one button to be start/stop according to the current state. Apr 21 20:27:31 ok, that kind of works. But it insists on the interface being named 'openconnect-$cfg' rather than what's returned by my proto.ifname() function. Apr 21 20:53:00 it's as if my own proto.ifname() function isn't being called. the generic one in /usr/lib/lua/luci/model/network.lua is used instead. Apr 21 20:53:32 is there a problem with this URL? http://downloads.openwrt.org/sources/xz-5.0.3.tar.bz2 Apr 21 20:54:03 philipp64|laptop: 404 Apr 21 20:54:05 can't build because it's missing. Apr 21 20:54:09 yup. Apr 21 20:57:35 what bumped the xz version? Apr 21 20:58:22 oh, here we go: Apr 21 20:58:24 r31172 | florian | 2012-04-02 10:25:25 -0600 (Mon, 02 Apr 2012) | 3 lines Apr 21 20:58:42 [florian]: can you please upload xz-5.0.3? Apr 21 21:05:26 nbd * r31428 /trunk/package/mac80211/patches/ (2 files): ath9k: merge a tkip related stability fix Apr 21 21:05:29 nbd * r31429 /trunk/package/mac80211/patches/ (3 files): mac80211: suppress 3-address multicast frames on WDS APs without non-WDS stations Apr 21 21:05:33 nbd * r31430 /trunk/package/netifd/files/lib/network/config.sh: netifd: make type a local variable in /lib/network/config.sh to avoid clashes with an internal config_foreach variable Apr 21 21:05:36 nbd * r31431 /trunk/package/base-files/files/lib/functions.sh: base-files: rename internal variables in config_foreach to make namespace conflicts more unlikely Apr 21 21:05:40 nbd * r31432 /trunk/package/netifd/files/lib/network/config.sh: netifd: improve error handling in legacy script compatibility library, do not fill the log with error messages for interfaces that could not be found Apr 21 21:06:04 philipp64|laptop: xz 5.0.3 is on mirror2 now Apr 21 21:06:58 thanks. Apr 21 21:07:41 I'm wondering how that could be broken for nearly 3 weeks without the buildbot detecting it... Apr 21 21:07:57 maybe they downloaded it early enough Apr 21 21:08:03 my build had also downloaded it without issues Apr 21 21:15:00 hm, something *removes* the Legacy IP address from my vpn-foo device after vpnc-script sets it up. Apr 21 22:26:09 build #170 of at91 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org:8010/builders/at91/builds/170 **** ENDING LOGGING AT Sun Apr 22 03:00:00 2012