**** BEGIN LOGGING AT Fri May 08 02:59:57 2009 May 08 10:39:33 nbd * r15706 /trunk/scripts/feeds: scripts/feeds: ignore virtual packages May 08 12:31:29 nbd * r15707 /trunk/package/madwifi/patches/423-phyerr_handling.patch: madwifi: fix selection of phy errors in the rx filter May 08 12:52:49 xMff:ping May 08 12:57:19 xMff: in default.script... for bridge devices like lan, maybe it will writhe in state/network.lan.device=br-lan May 08 12:58:13 xMff: now It write network.lan.device=eth0.0 wl0 May 08 13:03:53 fofware: don't put wl0 in your network config May 08 13:05:15 hello nbd, that is not present in network config May 08 13:05:46 ah, only in the state? May 08 13:05:48 or where? May 08 13:05:55 yep May 08 13:06:15 ok May 08 13:08:05 nbd: http://openwrt.pastebin.com/d4a582740 May 08 13:09:33 I don't know what is better, but for ip route br-lan is valid for example and others things too May 08 14:25:59 fofware: use network..ifname May 08 14:26:05 don't see the problem... May 08 14:28:14 Hello xMff, i did check if is a bridge interface and change the device name for my script, that is not issue for that, but i did think could be good let you know May 08 14:29:14 xMff: thanks May 08 14:31:56 fofware: ifname will contain the interface that's usually used for ifconfig & co and device contains the physical device(s) May 08 14:32:16 if I use network..ifname maybe I will have issues with ppp(x) devices May 08 14:33:54 so, you mean network..ifname is changed by udhcp? May 08 14:34:32 I can't check that here May 08 14:35:37 no May 08 14:39:04 xMff: no, what? May 08 14:39:55 dhcp script does not set such viariables, that's done by the network setup scripts May 08 14:40:30 ah! thanks May 08 15:21:28 <__Flo__> hi May 08 15:29:34 while applying 100-ncurses-5.6-20080112-urxvt.patch, build stops with "Get file misc/terminfo.src from Perforce with lock? [y]" May 08 15:32:39 oops May 08 15:32:55 i think unsetting variables with export VAR= doesn't work May 08 15:32:58 i'll fix that May 08 15:35:29 nbd * r15708 /trunk/include/toplevel.mk: use unexport instead of export to clear perforce variables May 08 15:35:47 netprince: should be fixed now May 08 15:35:48 <__Flo__> I'm playing with OpenWRT 8.09 and trunk on Ubiquiti Bullet2/Bullet5 sticks since a few days, now May 08 15:35:55 nbd: nice! May 08 15:37:34 <__Flo__> unfortunatly the wireless network throughput is limited to about 1.2 MB/s on OpenWRT (-> ~97% softirq load) May 08 15:39:42 <__Flo__> The original Ubiquiti FW (linux 2.4.27..., madwifi 5.0...) is able to handle about 1.8-2.2MB/s. May 08 15:41:20 __Flo__: routing or bridging? May 08 15:41:21 <__Flo__> do u have any idea what may help to reduce the softirq load of the madwifi/hal on OpenWRT in order to increase the performance in sta (only) mode? May 08 15:41:36 <__Flo__> I currently only tried and tested routing May 08 15:42:34 did you use routing on the ubiquiti firmware as well? May 08 15:42:39 <__Flo__> yes May 08 15:42:48 have you tried a recent trunk version? May 08 15:43:07 actually i think there's a bug in the current version which i still need to fix May 08 15:43:12 but i'll get that done during the weekend probably May 08 15:43:14 then it should be fast May 08 15:43:27 <__Flo__> I'm currently using trunk r15628 May 08 15:44:04 <__Flo__> btw. I've also submitted a patch for OpenWRT madwifi to support the Ubiquiti Bullet2 / Bullet5 signal leds: May 08 15:44:23 <__Flo__> https://dev.openwrt.org/attachment/ticket/5066/openwrt-trunk_ubnt_bullet_signal-leds.patch May 08 15:44:52 __Flo__: hows the range on that bullett2? May 08 15:45:11 <__Flo__> It works for me but as I'm no coder it would be cool if anyone with knowledge of the madwifi code could have a review on it ;-) May 08 15:45:39 <__Flo__> I'm currently testing within the office: distance ~10-15m May 08 15:46:01 <__Flo__> But the stuff should be used outdoor distance ~1-3m l8rz May 08 15:48:33 <__Flo__> The Ubiquiti original firmware seems to use a different Atheros HALs for each type of their equipment: May 08 15:48:55 <__Flo__> mipsisa32-be-elf.inc May 08 15:49:01 <__Flo__> mipsisa32-be-elf-ls2.inc May 08 15:49:08 <__Flo__> mipsisa32-be-elf-ls5.inc May 08 15:49:14 yes, they have to, because they use a variant of the old atheros hal, which hardcodes a lot of crap May 08 15:49:19 <__Flo__> maybe that's the source key for this performance improvement May 08 15:49:53 highly unlikely May 08 15:51:59 <__Flo__> do you have an idea how I could further debug the high softirq load while having wireless network throughput? May 08 15:52:20 do you get any weird kernel messages? May 08 15:52:48 <__Flo__> I already tried disabling the WPA2-AES encryption with no result. May 08 15:53:07 <__Flo__> no May 08 15:53:41 have you tried without the led code? May 08 15:54:21 <__Flo__> yes, I also tested with current 8.09 and release 8.09 May 08 15:54:55 <__Flo__> it does not make any difference May 08 15:55:24 please try the new version after my last commit May 08 15:55:28 and disable softled May 08 15:55:30 <__Flo__> btw. if you need two bullet sticks for dev., leave me a priv. msg? ;) May 08 16:08:35 nbd * r15709 /trunk/package/madwifi/ (Makefile ath_hal-20081002.tgz): upgrade to a new public hal build - contains several minor fixes including ani improvements May 08 16:21:26 nbd * r15710 /trunk/package/madwifi/patches/400-new_hal.patch: madwifi: remove unused functions May 08 16:45:19 ping juhosg: Just got back to looking at creating an image for this wrt400n. Am I correct in thinking I need to find a firmware build utility similar to mkplanexfw for the planex routers? May 08 16:57:43 netprince: yes, if you want to flash it via the factory webinterface May 08 16:58:53 ok, will look into it May 08 17:55:08 ping aoz May 08 18:49:07 nbd * r15711 /trunk/package/madwifi/patches/ (389-autochannel.patch 411-autochannel_multi.patch): madwifi: fix uninitialized variable May 08 19:12:07 lars * r15712 /packages/ (Xorg/lib/wnck/ Xorg/lib/wnck/Makefile libs/wnck/): Update libwnck package to 2.26.1 and move it to the Xorg section. May 08 19:25:32 nbd * r15713 /trunk/package/madwifi/ (Makefile files/lib/wifi/madwifi.sh patches/424-timing.patch): madwifi: clean up handling of various timings such as slot time, ack timeout, eifs time, etc. May 08 19:29:13 <[Fate]> nbd: ping May 08 19:29:18 [Fate]: pong May 08 19:29:56 <[Fate]> nbd: my diploma thesis includes a chapter on openwrt. might you will be willing to read/review it, especially as regards the openwrt history? May 08 19:30:02 sure May 08 19:30:36 <[Fate]> what was your email addy again? May 08 19:30:43 nbd nbd.name May 08 19:34:46 <[Fate]> merci May 08 19:34:51 <[Fate]> on its way May 08 20:44:48 nbd * r15714 /trunk/tools/ (6 files in 6 dirs): clean up / fix various Host/Clean templates in tools/ May 08 22:48:02 nbd * r15715 /branches/8.09/include/prereq-build.mk: remove obsolete prereq checks May 08 22:48:04 nbd * r15716 /branches/8.09/tools/ (6 files in 6 dirs): clean up / fix various Host/Clean templates in tools/ (backport from r15714) May 08 22:48:06 nbd * r15717 /branches/8.09/tools/Makefile: add missing dependency May 08 22:48:12 nbd * r15718 /branches/8.09/target/linux/ (201 files in 30 dirs): restore 2.6.23 support for rb532. 2.6.24 is 100% broken and was only added accidentally May 08 22:48:21 nbd * r15719 /branches/8.09/package/ (88 files in 19 dirs): backport libnl-tiny, wprobe, madwifi May 08 22:55:41 nbd * r15720 /branches/8.09/target/linux/rb532/config-2.6.23: rb532: refresh kernel config, disable the automatic root device selection, as the kernel command line patching takes care of that May 08 23:06:45 nbd * r15721 /branches/8.09/target/linux/rb532/base-files/etc/config/network: remove rb532 specific network config May 08 23:08:41 nbd * r15722 /trunk/target/linux/rb532/base-files/etc/config/network: rb532: bring the target specific default /etc/config/network closer to the global default by setting it to static instead of dhcp May 08 23:08:45 nbd * r15723 /trunk/target/linux/rb532/config-2.6.28: rb532: disable the automatic rootfs selection - it breaks the cmdline based selection May 08 23:27:47 nbd * r15724 /trunk/target/linux/rb532/patches-2.6.28/007-cmdline_hack.patch: rb532: restore command line patching functionality May 08 23:27:51 nbd * r15725 /trunk/target/linux/rb532/ (config-2.6.28 image/Makefile): rb532: add missing drivers and adjust the kernel command line for the right device name May 08 23:41:59 nbd * r15726 /branches/8.09/package/wprobe/src/kernel/linux/math64.h: wprobe: add a header file that is missing in older kernel versions May 08 23:46:16 nbd * r15727 /branches/8.09/target/linux/rb532/config-2.6.23: rb532: also disable the rootfs split (cosmetic fix) May 08 23:51:31 nbd * r15728 /branches/8.09/package/libipfix/ (8 files in 3 dirs): add libipfix May 09 00:25:42 nbd * r15729 /branches/8.09/target/linux/generic-2.6/patches-2.6.23/921-gpio_spi_driver.patch: rb532: remove the gpio spi driver patch May 09 01:54:11 jow * r15730 /trunk/package/base-files/files/lib/upgrade/common.sh: [package] base-files: show reboot notification is sysupgrade May 09 01:57:47 jow * r15731 /branches/8.09/package/base-files/files/lib/upgrade/common.sh: Merge r14784 to 8.09 (#4747) May 09 02:01:56 jow * r15732 /branches/8.09/package/base-files/files/ (lib/upgrade/common.sh sbin/sysupgrade): Merge r15672 to 8.09 ([PATCH] Add option in sysupgrade to flash gzipped images) May 09 02:03:08 jow * r15733 /branches/8.09/package/base-files/files/lib/upgrade/common.sh: Merge r15730 to 8.09 May 09 02:21:28 for the image on this linksys wrt400n, it appears the header should be GMTKRT400N, but addpattern requires a 4 character header? May 09 02:22:15 netprince_: how does linksys generate it in their gpl tarball? May 09 02:22:52 yes, I should figure that out, but I didn't have luck locating it May 09 02:23:07 grep was not my friend May 09 02:23:16 is there a good search term or file name? May 09 02:23:34 '.bin' ? May 09 02:23:37 ;) May 09 02:24:15 I'll pulling down the gpl tarball now, I look at it when I get home (if I don't fall asleep first) May 09 02:25:24 I will look for .bin, :) May 09 02:25:48 maybe '.BIN' if the official images are in all caps. May 09 02:29:27 hmm, there is a FW-Generator.o without source, probably not a good sign **** ENDING LOGGING AT Sat May 09 02:59:57 2009