**** BEGIN LOGGING AT Sun Sep 02 02:59:57 2007 Sep 02 12:06:45 any developer currently online? Sep 02 12:07:09 I get kernel oops with my router under heavy load Sep 02 12:07:37 and could need some help to resolv the problem Sep 02 12:19:12 pi: wl500gp? Sep 02 12:19:39 buffalo Sep 02 12:19:49 http://www.physik.uni-wuerzburg.de/~pgeisler/temp/ooops.txt Sep 02 12:20:28 I think line 399 is interesting Sep 02 12:23:14 ack, but i think i'm unable to help you Sep 02 12:23:51 k Sep 02 12:24:30 maybe updating to the latest svn helps (compiling at the moment) Sep 02 12:27:52 *reflashing* bbl Sep 02 12:56:21 any devs here? Sep 02 13:07:49 why do you expect devs in a devs-channel? Sep 02 13:09:08 good question Sep 02 13:09:24 I was wondering if I could draw some attention to a ticket I filed Sep 02 14:15:57 <{Nico}> danchr: what ticket? Sep 02 14:17:26 ok, seems like I found what triggered my kernel oops (or at least what made it occur so soon) Sep 02 14:18:23 nico * r8573 /trunk/target/linux/generic-2.4/patches/801-usb_serial_endpoint_size.patch: add high-speed usb serial fix from Junxion to 2.4 kernel too (closes: #2274) Sep 02 14:18:31 I missconfigured iptables and thatway created a loop Sep 02 14:20:05 nico * r8574 /trunk/target/linux/generic-2.6/patches-2.6.22/ (2 files): rename 2.6 high-speed usb serial fix to match 2.4 one Sep 02 14:21:04 the system shouldn't look up that way but thats propably a problem that needs to get fixed upstream Sep 02 15:05:27 {Nico}: #2156 Sep 02 15:05:39 with patch and all :) Sep 02 15:30:47 I finally got OpenWrt to build on Mac OS X Sep 02 16:39:25 nico * r8575 /trunk/ (29 files in 29 dirs): use a default VERSION field for kmod packages Sep 02 17:09:46 {Nico}: there? Sep 02 17:15:18 nbd, ping Sep 02 17:18:54 <{Nico}> danchr: yep Sep 02 17:19:19 you're an openwrt dev, right? Sep 02 17:19:44 could you perhaps review my portability patch? Sep 02 17:20:37 Sep 02 17:22:17 or perhaps set it to Cc someone who can? Sep 02 17:28:09 <{Nico}> danchr: humm... lot of changes in this patch... Sep 02 17:29:21 Yeah, posting to another OS is rather invasive :) Sep 02 17:36:29 <[florian]> danchr-mobile: your patch sounds good, but I suspect it would break some things on other arches Sep 02 17:54:57 [florian], {Nico}, if you guys get a chance kick nbd into looking at #2178 Sep 02 17:56:49 i see what hes trying to do but its also incrementing the first ppp link (ppp0 ->ppp1) but this isnt needed nor is the interface change passed on to the hotplug scripts Sep 02 17:59:21 <[florian]> Weedy: will poke him Sep 02 17:59:46 thx Sep 02 19:36:03 florian * r8576 /packages/net/openvpn/files/ (openvpn.default openvpn.init): Fix openvpn init script (#2284) Sep 02 19:40:29 florian * r8577 /trunk/package/strongswan/files/ipsec.iface: Do not restart ipsec on hotplug events (#2212) Sep 02 19:51:44 Ironically, aiccu is proving the hardest package to cross-compile :) Sep 02 19:52:50 [florian]: would you be interested in reviewing my portability patch? Sep 02 19:53:25 <[florian]> danchr: I was checking the find options you changed, other than that Sep 02 19:53:27 <[florian]> it is ok Sep 02 19:53:43 find -lname is non-standard Sep 02 19:54:36 <[florian]> precisely Sep 02 19:54:53 <[florian]> I wanted to check that replacing -uid with -user worked even with a uid, and it is ok Sep 02 19:55:16 what was there before definitely didn't work on Mac OS X :) Sep 02 19:55:28 AFAICT is should work on Linux too Sep 02 19:55:51 seems lots of GNU-isms has crept into the build system, but that's to expect :) Sep 02 19:56:58 <[florian]> yes, we have been targeting linux x86 hosts for a long time :) Sep 02 19:57:54 you have an awfully complicated build system, but for a full distro, I guess it's not that bad Sep 02 19:58:23 plus, I find GNU Make rather pleasant to use Sep 02 19:58:31 ./configure is overestimated... Sep 02 19:59:40 <[florian]> well,the build system is way better than oe or buildroot if you have been using them as well Sep 02 19:59:52 <[florian]> at least, ours tend to be ran on non-gnu hosts Sep 02 20:01:55 not really, no Sep 02 20:02:11 last time I compiled the Linux kernel was in my Computer Architecture class Sep 02 20:02:16 and that was v2.2 Sep 02 20:02:46 but I did find out that Apple GCC 4.0 won't compile FSF gcc 3.4 Sep 02 20:03:00 <[florian]> yes, it can't Sep 02 20:03:36 <[florian]> I have been building openwrt on my osx something like 1000 revisions ago Sep 02 20:04:08 well, lots of things broke since then :) Sep 02 20:04:19 <[florian]> right, I will merge your patch soon Sep 02 20:04:30 so you won't get an Intel-native 2.4-targetting toolchain, but who cares Sep 02 20:04:40 if you've got Intel, you've got cycles to spare :) Sep 02 20:04:58 <[florian]> mine is a g4 :) a real machine Sep 02 20:05:45 1.25GHz eMac here Sep 02 20:05:55 it looks like a vacuum cleaner, and sounds like one too Sep 02 20:06:18 heh Sep 02 20:06:44 I wish I had an Intel to run VMware on... Sep 02 20:07:26 <[florian]> yes, kind of convenient Sep 02 20:10:19 but having a nice little mipsel-linux box is an ok substitute, now that I can target it Sep 02 20:12:24 <[florian]> it took me quite a few hours to build an openwrt toolchain on my qube2, but worked :) Sep 02 20:13:29 [florian]: ooo, qubes are fun. Sep 02 20:13:55 <[florian]> Bartman007: oh yeah Sep 02 20:23:43 the NeXT cube? Sep 02 20:24:07 <[florian]> it is pretty old actually Sep 02 20:24:17 danchr: nope, http://en.wikipedia.org/wiki/Cobalt_Qube Sep 02 20:24:37 the NeXT cube was 68k based, not MIPSel :-) Sep 02 20:24:47 <[florian]> Bartman007: x86, not 68k ;) Sep 02 20:24:55 <[florian]> 68k was pretty old at that time already Sep 02 20:25:29 [florian]: Wikipedia says different, http://en.wikipedia.org/wiki/NeXTcube Sep 02 20:25:53 well, I definitely won't use my buffalo for compiling anything, that's for sure... Sep 02 20:26:33 [florian]: the Cube's I've used/disassembled say differently too :-) Sep 02 20:27:39 <[florian]> Bartman007: ah, the NeXTcube, ok, thouhgt you meant qube3, which is definitively x86 Sep 02 20:28:28 [florian]: oh, ok. Sep 02 20:28:59 <[florian]> anyway, good hardware in both cases Sep 02 20:30:09 erm, if __res_ninit is undefined when linking, what does that mean? :) Sep 02 20:30:56 the mac cube is also quite popular among hardware connoisseurs :) Sep 02 20:31:16 <[florian]> danchr: which one ? Sep 02 20:31:34 mac or symbol? Sep 02 20:31:52 <[florian]> mac, and which program do you compile ? Sep 02 20:31:58 http://en.wikipedia.org/wiki/Power_Mac_G4_Cube Sep 02 20:32:02 aiccu Sep 02 20:32:15 I've succeeded to get it to compile, now it fails linking Sep 02 20:33:33 I had to put LDFLAGS before the objects to get it to compile Sep 02 20:46:23 does `uname -s` output GNU/Linux on Debian? Sep 02 20:48:43 <[florian]> just Linux Sep 02 20:56:30 good :) Sep 02 21:20:31 does -pthread work on Linux 2.6 too? Sep 02 21:22:33 <[florian]> oh yes Sep 02 21:23:29 good :) Sep 02 21:27:40 is it possible to control the lights from a switch on a router? or are the completely independed and hardwaredriven without a chance to influence them by software? Sep 02 21:28:51 yes, it is, via /proc Sep 02 21:29:45 /proc/diag/led Sep 02 21:30:03 hmm, i'll take a look :). but maybe you got me wrong, i want to turn on/off the activity/connection lights there Sep 02 21:30:45 err, in the dir you meant are only the leds which are connected over gpios, iirc Sep 02 21:31:45 yes? Sep 02 21:31:56 ah, sorry, missed the switch part Sep 02 21:32:29 my best guess would be that they're independent, but I really wouldn't know :) Sep 02 21:35:20 maybe, hope there is something written in the datasheet i have about one switch ic :) Sep 02 21:35:59 intersting, .la = laos, think i never saw a domain from there :D Sep 02 21:38:23 <[florian]> crazy_imp: not all leds on a switch can be software controlled Sep 02 21:38:40 <[florian]> crazy_imp: some adm5120 devices (most) allow you to do so, not sure for broadcom Sep 02 21:40:05 [florian]: ok, i'll try to find the datasheet for the switch in the wrt54g, iirc the one i have was for one i've found in an usr device Sep 02 21:40:25 <[florian]> crazy_imp: search in google for bcm5365p.pdf Sep 02 21:40:37 <[florian]> should provide you some interesting infos Sep 02 21:45:59 "The resistors are not mapped into the internal bus address space and are defined outside the memory address space." <- *confused* Sep 02 21:54:34 [florian]: page 141, MII PORT STATE OVERRIDE REGISTER, might it's possible to set the link status with it. so it would be possible to turn the led from one port on and off :) **** ENDING LOGGING AT Mon Sep 03 02:59:56 2007