**** BEGIN LOGGING AT Mon Sep 07 02:59:57 2009
Sep 07 06:17:22 <dkostousov> Hi
Sep 07 06:23:44 <dkostousov> I'm customizing openwrt for one hardware. Can I predefine network configuration?
Sep 07 06:25:22 <thepeople> dkostousov: you can include any files that you want in the image in files/ in the buildroot (you have to make that directory it isn't there by default)
Sep 07 06:30:55 <dkostousov> Does the firmware's firstrun scripts overwrite my files?
Sep 07 06:31:05 <thepeople> no
Sep 07 06:33:45 <dkostousov> Who detects network cards during first firmware start?
Sep 07 06:35:21 <thepeople> depends on the arch, but there are init script that do but they generally will not nuke your files
Sep 07 06:43:17 <dkostousov> What is the init script filename?
Sep 07 08:48:42 <Hopscotch> good morning
Sep 07 10:15:06 <ChoHag> I've worked out what causes my problem but I don't know why.
Sep 07 10:15:18 <ChoHag> libtool is called and inserts -L/usr/lib into the command it eventually runs.
Sep 07 10:15:22 <ChoHag> But I've no idea why.
Sep 07 10:20:09 <_trine> I have found a package that looks very interesting it's supposed to be similar to asterisk but better in many respects. The Package compiles but in total it's 3.2 megs. This is too big to fit in my router I was wondering if anyone here with more/better programing skills could take a look at it with a view to splitting it up into a core and various additions.
Sep 07 10:20:47 <_trine> The package is called Freeswitch
Sep 07 10:21:07 <ChoHag> That's already split in debian iirc.
Sep 07 10:21:12 <ChoHag> Not debian proper, though.
Sep 07 10:21:48 <_trine> yes but I need the .ipk split up into parts
Sep 07 10:25:21 <_trine> The Package can be dowloaded here:-  http://wiki.freeswitch.org/wiki/OpenWrt
Sep 07 10:26:50 <_trine> I had to remove this patch to get it to compile:-  004-pcre.patch
Sep 07 11:03:44 <mb__> what's the new sstrip option for? Is sstrip better than binutils strip?
Sep 07 12:04:30 <nbd> mb__: sstrip isn't new. we've used it for years
Sep 07 12:04:35 <nbd> and yes, it's better than binutils strip
Sep 07 12:05:29 <mb__> hm, ok. Compilation failed yesterday with some error when using it. So I switched the option to strip.
Sep 07 12:33:02 <nbd> mb__: odd. sstrip has been the default for years as well
Sep 07 13:55:04 <CIA-36> nbd * r17537 /trunk/tools/firmware-utils/src/trx.c: firmware-utils: raise the default trx image size limit for something more appropriate for devices with 8M flash
Sep 07 14:06:15 <matteo> anyone here has a new "M" Ubiquiti device?
Sep 07 14:07:15 <OutBackDingo_> matteo: whats up
Sep 07 14:07:24 <matteo> hi
Sep 07 14:07:35 <matteo> i have one here
Sep 07 14:07:39 <OutBackDingo_> matteo: and
Sep 07 14:07:43 <matteo> i can't find the serial pinout
Sep 07 14:08:02 <OutBackDingo_> what series M
Sep 07 14:08:07 <matteo> bullet M
Sep 07 14:08:13 <matteo> M5
Sep 07 14:08:38 <OutBackDingo_> pinoutsmost likely the same as the Bullet%
Sep 07 14:09:04 <matteo> it has only 4 pins
Sep 07 14:09:10 <OutBackDingo_> no shit
Sep 07 14:09:28 <OutBackDingo_> u sure itsd a serial
Sep 07 14:09:34 <matteo> yes
Sep 07 14:09:36 <OutBackDingo_> ive not opened one of these up
Sep 07 14:09:42 <matteo> if I ssh and i dmesg >/dev/ttyS0
Sep 07 14:09:48 <matteo> i see shit in the serial
Sep 07 14:10:07 <matteo> you're right BTW
Sep 07 14:10:17 <matteo> bullet G has also 4 pin
Sep 07 14:10:22 <matteo> so i assume it's the same
Sep 07 14:10:31 <matteo> just the nanostation has 10 pins...
Sep 07 14:38:50 <mb__> SQUASHFS error: Major/Minor mismatch, older Squashfs 3.0 filesystems are unsupported
Sep 07 14:38:54 * mb__ scratches head
Sep 07 14:40:06 <mb__> I get this on 2.6.30. Is this known? How to fix it?
Sep 07 14:40:11 <nbd> make oldconfig
Sep 07 14:40:16 <nbd> in the build system dir
Sep 07 14:46:51 <mb__> nbd: thanks :)
Sep 07 15:01:18 <mb__> Look at this guys. I tested the BFS scheduler on bcm47xx: http://www.nopaste.org/p/a5eTYRv5f
Sep 07 15:01:28 <mb__> simple iperf test with iperf server running on the bcm47xx
Sep 07 15:02:15 <blogic> mb__: speaks for itself really
Sep 07 15:02:27 <mb__> That's really impressive
Sep 07 15:02:37 <blogic> its like 40-50%
Sep 07 15:02:50 <nbd> whoa
Sep 07 15:02:55 <blogic> mind boggingly impressive i would even say
Sep 07 15:03:10 <blogic> nbd: whoa is another way of putting it :)
Sep 07 15:03:11 <nbd> let's merge it
Sep 07 15:03:14 <nbd> :D
Sep 07 15:03:24 <blogic> nbd: whe is the mainline scheduler so much slower ?
Sep 07 15:03:30 <blogic> nbd: scaled for larger systems ?
Sep 07 15:03:31 <maligor> but C is a larger letter than B! :P
Sep 07 15:03:37 <nbd> the mainline scheduler is designed to scale to systems with 4096 CPUS
Sep 07 15:03:54 <nbd> mb__: let's join in on the lkml flamewar :D
Sep 07 15:04:16 <nbd> lkml people are asking for hard numbers on BFS vs CFS
Sep 07 15:04:18 <nbd> so let's give them some
Sep 07 15:04:29 <mb__> yeah
Sep 07 15:04:39 <blogic> brainfuck
Sep 07 15:04:40 <blogic> lol
Sep 07 15:04:48 <mb__> BFS is 8% faster for kernel compiles on my 4way PowerMAC
Sep 07 15:05:49 <maligor> tempted to try that on my desktop
Sep 07 15:06:04 <blogic> maligor: haha
Sep 07 15:06:30 <mb__> BFS throws out a lot of stuff and is optimized for small systems. It doesn't know about NUMA, HT and big SMP.
Sep 07 15:06:39 <maligor> large amounts of cpu's would suffer but I only have 2 cores
Sep 07 15:06:59 <nbd> the BFS author says it should theoretically scale up to 16 cores
Sep 07 15:07:28 <maligor> 1. Intermittent boot failures on some hardware.
Sep 07 15:07:35 <maligor> I wonder if this faq is uptodate
Sep 07 15:08:07 <nbd> imho it shouldn't be hard to rework the patch to make the scheduler selectable at compile time
Sep 07 15:08:18 <nbd> then we can test and switch our targets individually
Sep 07 15:09:19 <maligor> For openwrt on NUMA?
Sep 07 15:09:32 <nbd> no, for BFS
Sep 07 15:10:11 <maligor> actually... I think I'll try it on my NAS
Sep 07 15:10:34 <maligor> it could do with every bit of performance increase it can get
Sep 07 15:15:14 <mb__> nbd: should I CC you?
Sep 07 15:15:40 <nbd> yeah
Sep 07 15:44:13 <acoul> @mb__: have you tried BFS on brcm47xx/2.6.31 ?
Sep 07 15:44:26 <mb__> no
Sep 07 15:44:37 <mb__> We don't have a bcm47xx patchset for .31, yet
Sep 07 15:48:07 <acoul> I am working on it that's why I asked
Sep 07 15:48:51 <mb__> cool. Lemme know when you're done so I can run the test :)
Sep 07 15:49:08 <mb__> con just released a BFS for .31
Sep 07 15:49:44 <acoul> yes I saw it
Sep 07 17:02:06 <_trine> if openwrt-ar71xx-wrt160nl-squashfs.bin is 3.1MB in size what's the biggest package size I could install with opkg on a router with 8MB flash
Sep 07 17:02:37 <_trine> specifically a WRT160NL
Sep 07 17:03:45 <_trine> I have a package that is 3.2MB in size should it fit if I was to try it?
Sep 07 17:12:44 <_trine> I have tried including the package into the compile but the compile always fails saying the image is too large
Sep 07 17:23:37 <swalker> yet another openwrt/upstream package comparison, http://pastehtml.com/view/090907gS9yxjzU.html, generally up-to-date
Sep 07 17:40:45 <nbd> you can remove madwifi, it will never be updated
Sep 07 18:17:47 <CIA-36> florian * r17538 /trunk/target/linux/brcm63xx/files/arch/mips/bcm63xx/boards/board_bcm963xx.c: [brcm63xx] fix typos against the gpio buttons driver (#5804)
Sep 07 20:29:24 <OutBackDingo> nbd: you alive
Sep 07 20:32:03 <nbd> OutBackDingo: yes
Sep 07 20:32:55 <OutBackDingo> nbd: is there anything required to access ttyS0
Sep 07 20:33:08 <OutBackDingo> in ath2k on Ubiquiti
Sep 07 20:33:34 <nbd> access ttyS0 for what and with what?
Sep 07 20:34:58 <OutBackDingo> im trying to reqad serial data from the port
Sep 07 20:34:58 <OutBackDingo> heh read
Sep 07 20:34:58 <OutBackDingo> i connected a ttl converter to it, and the other end is connected to a serial port
Sep 07 20:35:20 <nbd> maybe you should move the kernel console away from that port
Sep 07 20:35:25 <nbd> i think there are utilities for that somewhere
Sep 07 20:35:29 <nbd> but i don't know
Sep 07 20:35:59 <OutBackDingo> hrmmm i was sure disabllling it in inittab did that
Sep 07 20:36:04 <nbd> no
Sep 07 20:36:09 <OutBackDingo> ahhh
Sep 07 20:36:39 <OutBackDingo> so theres a kernel console set to run on the port, which would essentially prevent anything else from using it
Sep 07 20:36:48 <OutBackDingo> ok
Sep 07 20:36:56 <OutBackDingo> ill research this
Sep 07 20:36:57 <nbd> theoretically it shouldn't
Sep 07 20:37:11 <nbd> but if you want to use the port, it is generally a good idea to disable it anyway
Sep 07 20:37:22 <OutBackDingo> any fast way to disable it
Sep 07 20:37:44 <OutBackDingo> kernel recompile ?
Sep 07 20:53:05 <benoar> Hi there
Sep 07 20:54:11 <benoar> I am trying to create packages for some python apps
Sep 07 20:54:30 <benoar> I spotted what may be a bug, but I am not sure :
Sep 07 20:55:28 <benoar> the dynamic library zlib.so is not present in the staging dir in usr/lib/python2.6/lib-dynload/
Sep 07 20:55:45 <benoar> while it is in the build_dir and on the target
Sep 07 20:55:56 <benoar> and this is the only missing library
Sep 07 20:55:57 <benoar> strange
Sep 07 20:56:15 <benoar> did you guys already remarked that ?
Sep 07 20:57:49 <benoar> I think this is one of the reasons setuptools is disabled in many existing python packages (setuptools uses zlib and doesn't work due to this bug)
Sep 07 20:58:00 <benoar> any idea ?
Sep 07 21:44:52 <OutBackDingo> nbd: wierdness opkg install olsrd_0.5.6-r5-2_brcm-2.4.ipk Collected errors: * Packages were found, but none compatible with the architectures configured
Sep 07 21:44:52 <OutBackDingo>  
Sep 07 21:45:02 <OutBackDingo> its a linksys hahhahah
Sep 07 21:45:35 <OutBackDingo> what i gotta build my own now
Sep 07 21:47:44 <thepeople> OutBackDingo: the architectures for the backages changed to the branch name awhile back instead of the arch
Sep 07 21:48:26 <thepeople> OutBackDingo: you can add a /etc/opkg/arch.conf file to override the arches opkg looks for
Sep 07 21:49:27 <OutBackDingo> thepeople: how can i do this for broadcom?
Sep 07 21:49:50 <OutBackDingo> err wait
Sep 07 21:50:13 <thepeople> http://pastebin.ca/1557737
Sep 07 21:56:42 <OutBackDingo> thepeople: thanks that worked
Sep 08 00:58:11 <kupesoft> Bumping my apf and libv4l patches :)
Sep 08 01:32:19 <OutBackDingo_> tonsofpcs: oh well that theory didnt work quite so well
**** ENDING LOGGING AT Tue Sep 08 02:59:57 2009