**** BEGIN LOGGING AT Fri Jul 18 02:59:57 2008
Jul 18 04:01:21 <luke-jr> wonderful
Jul 18 04:01:25 <luke-jr> I have about 2 MB to work with
Jul 18 04:01:26 <luke-jr> :/
Jul 18 05:11:25 <luke-jr> so how do I get b44 to actually do something?
Jul 18 11:43:54 <CIA-8> florian * r11856 /trunk/package/kexec-tools/ (Makefile patches/0004-mips_regdefs.patch): Allow kexec-tools to be built for mipsel, tested on a routerboard 500
Jul 18 13:52:15 <CIA-8> florian * r11857 /packages/admin/monit/ (Makefile patches/001-no_sslv2.patch): Update monit to 4.10.1 (#3776)
Jul 18 13:58:05 <aroscha> hi
Jul 18 13:58:11 <wrt> aroscha?
Jul 18 13:58:42 <aroscha> what is the recommended way to disable the firewall all together in the image building process already?
Jul 18 13:59:19 <aroscha> Should I add a file $TOPDIR/files/etc/init.d/firewall which contains a null body?
Jul 18 14:00:48 <[florian]> hi aroscha
Jul 18 14:00:55 <wrt> [florian]?
Jul 18 14:01:04 <aroscha> hi florian :)
Jul 18 14:01:08 <[florian]> or just remove the executable bit of package/iptables/files/firewall.init
Jul 18 14:01:11 <wrt> welcome to #openwrt-devel!
Jul 18 14:01:21 <aroscha> ah ok
Jul 18 14:01:28 <loswillios> aroscha: beware, NAT depends on it
Jul 18 14:01:37 <aroscha> i know
Jul 18 14:01:49 <loswillios> ok
Jul 18 14:01:55 <aroscha> should I write a nicer /etc/config/* thnig for it?
Jul 18 14:01:56 <aoz> a dirtier hack would be to add 'iptables -F; iptables -t nat -F' to the end of /etc/firewall.user
Jul 18 14:02:12 <aroscha> aoz: yeah that is what I did right now but it seems hmm...
Jul 18 14:02:18 <aroscha> not beautiful somehow
Jul 18 14:02:25 <loswillios> aroscha: I think the firewall is currently being rewritten
Jul 18 14:02:35 <aoz> ugly.  inelegant.  poorly-contrived.
Jul 18 14:02:39 <aroscha> loswillios: ok, then I don't do anything
Jul 18 14:02:59 <aroscha> I guess x-wrt and luci depend on it a bit, no?
Jul 18 14:03:05 <aoz> I'd love to see them use something like HLFL to designate the rules, but that probably won't happen.
Jul 18 14:03:50 <aroscha> ah and one more thing while we are at inet security!
Jul 18 14:04:08 <aroscha> did anybody take a look if dnsmasq does source port randomization?
Jul 18 14:04:20 <CIA-8> florian * r11858 /packages/ipv6/6tunnel/ (Makefile files/): The 6tunnel and 6bridge helpers should be packaged separately from 6tunnel which is another application
Jul 18 14:04:32 <aroscha> I guess it is on *many* DSL routers
Jul 18 14:04:49 <CIA-8> florian * r11859 /packages/ipv6/6scripts/ (7 files in 2 dirs): Move the helper scripts to setup IPv6 connectivity fo the 6scripts package
Jul 18 14:04:58 <aroscha> and we have time until the 6th of Aug until Dan Kaminsky publishes the DNS exploit
Jul 18 14:05:34 <[florian]> the dns exploit is just pure crap
Jul 18 14:05:41 <aroscha> ?
Jul 18 14:05:53 <aroscha> you know it already? :)
Jul 18 14:06:06 <[florian]> of course, I have been upgrading my debians ;)
Jul 18 14:06:25 <aroscha> yes but I mean the way to write the exploit?
Jul 18 14:06:37 <[florian]> a friend of mine explained it to me in understandable way and I do not see it as a major threat
Jul 18 14:06:44 <aroscha> ok
Jul 18 14:07:01 <aroscha> well it is for organisations with big central recursive DNS server such as ISPs
Jul 18 14:07:43 <farnz> There's no easy fix, either - really, you need a big change like DNSSEC to deal with it.
Jul 18 14:08:33 <aroscha> and if you can write an exploit which just takes let's say 10 packets (I am guessing now) then also cache poisoning many simple DSL boxes' dnsmasq might pay off
Jul 18 14:08:49 <aroscha> but maybe I misunderstood something of course
Jul 18 14:11:38 <CIA-8> florian * r11860 /packages/ipv6/dibbler/ (4 files in 2 dirs): Upgrade dibbler to 0.7.1 (#3778)
Jul 18 14:30:56 <aroscha> hm, seems like madwifi has some problems again. Anybody seen this already?
Jul 18 14:30:57 <aroscha> wifi0: ath_bstuck_tasklet: Stuck beacon; resetting (beacon )
Jul 18 14:30:57 <aroscha> wifi0: ath_bstuck_tasklet: Stuck beacon; resetting (beacon miss count: 11)
Jul 18 15:06:59 <aroscha> guys, I am seeing a beacon and gone again and beacon and gone on madwifi openwrt/trunk on an atheros CM9
Jul 18 15:07:11 <aroscha> like the network is constantly appearing and gone again
Jul 18 15:07:20 <aroscha> standard AP mode
Jul 18 15:07:35 <aroscha> nbd: ping
Jul 18 15:07:43 <nbd> pong
Jul 18 15:07:51 <aroscha> nbd: :)
Jul 18 15:08:03 <nbd> which trunk rev?
Jul 18 15:08:13 <aroscha> nbd: what version do you recommend as halfways stable in pure AP mode?
Jul 18 15:08:21 <aroscha> I just used the most current openwrt trunk version
Jul 18 15:08:25 <nbd> try this:
Jul 18 15:08:30 <nbd> echo 0 > /proc/sys/dev/wifi0/intmit
Jul 18 15:08:32 <aroscha> is there any debug info that give you BTW?
Jul 18 15:08:35 <aroscha> ok
Jul 18 15:08:38 <nbd> echo 0 > /proc/sys/dev/wifi0/ofdm_weak_det
Jul 18 15:09:15 <aroscha> same
Jul 18 15:09:33 <aroscha> do I need to reload anything?
Jul 18 15:09:37 <nbd> run 'wifi' again
Jul 18 15:09:41 <aroscha> ok
Jul 18 15:10:07 <aroscha> works!
Jul 18 15:11:04 <aroscha> nbd: I have around 15 foneras here which I could set up as testbed in case it is needed...
Jul 18 15:12:07 <nbd> please check if this bug reappears after some usage
Jul 18 15:14:58 <mb___> hm, trunk compile fails with
Jul 18 15:15:00 <mb___> cp: cannot stat `/home/mb/develop/svn/openwrt_4710/trunk/build_dir/mipsel/wpa_supplicant-0.6.3/wpa_cli': No such file or directory
Jul 18 15:15:19 <mb___> I tried make clean, but didn't help
Jul 18 15:21:13 <mb___> seems to work after disabling wpa_cli support
Jul 18 15:26:45 <nbd> oops
Jul 18 15:26:51 <nbd> that one's probably my fault
Jul 18 15:26:52 * nbd fixes
Jul 18 15:26:55 <mb___> :)
Jul 18 15:28:27 <CIA-8> nbd * r11861 /trunk/package/wpa_supplicant/Makefile: fix wpa_cli compile
Jul 18 15:28:32 <nbd> there you go ;)
Jul 18 15:29:19 <mb___> :)
Jul 18 15:32:45 <CIA-8> cyrus * r11862 /trunk/package/dnsmasq/ (Makefile patches/101-ipv6.patch patches/200-backports.patch): Updated Dnsmasq to 2.43
Jul 18 15:34:14 <mb___> nbd: works
Jul 18 16:14:18 <luke-jr>  13:55:07 up 2 days, 13:55, load average: 0.00, 0.00, 0.00
Jul 18 16:14:28 <luke-jr> (powered on this morning)
Jul 18 16:14:49 <luke-jr> nbd: hey, these bcm63xx irq patches all svn blame as you ;)
Jul 18 16:14:54 <aoz> someone's clock is running fast?
Jul 18 16:15:01 <luke-jr> aoz: very
Jul 18 16:15:10 <luke-jr> aoz: but I'm just happy it's running at all
Jul 18 16:16:01 <aoz> :D
Jul 18 16:16:13 <luke-jr> now I'm trying to figure out how to get the ethernet working
Jul 18 16:16:18 <luke-jr> and/or making the kernel smalle
Jul 18 16:16:19 <luke-jr> r
Jul 18 16:19:52 <nbd> luke-jr: svn blame doesn't tell you the whole story :P
Jul 18 16:20:14 <nbd> luke-jr: after running svn blame you should use svn log to figure out the rest ;)
Jul 18 16:20:24 <nbd> (which btw shows that i'm not the one who did this)
Jul 18 16:21:45 <luke-jr> r11178 | nbd | 2008-05-18 12:31:25 -0500 (Sun, 18 May 2008) | 7 lines
Jul 18 16:21:52 <nbd> read on
Jul 18 16:22:16 <luke-jr> Signed-off-by: Axel Gembe <ago@bastart.eu.org>
Jul 18 16:22:23 <nbd> :)
Jul 18 16:22:55 <luke-jr> (means nothing to me)
Jul 18 16:23:07 <nbd> means Axel Gembe submitted the patch
Jul 18 16:23:10 <nbd> and i applied it
Jul 18 16:23:43 <luke-jr> i c
Jul 18 16:27:34 <luke-jr> anyhow, I figured out that this bootloader has a 2 MB cap on how much it can tftp
Jul 18 16:27:46 <luke-jr> so I think I might need to reverse engineer its compression format(s)
Jul 18 16:29:28 <nbd> is it the standard broadcom 63xx format?
Jul 18 16:29:48 <nbd> because that one uses a slightly non-standard lzma
Jul 18 16:29:58 <nbd> which can be created from regular lzma with a shell script, iirc
Jul 18 16:32:30 <luke-jr> nbd: no idea what 63xx uses; got more info?
Jul 18 16:32:40 <luke-jr> I know this thing's "LZMA" isn't normal LZMA
Jul 18 16:53:51 <luke-jr> nbd: …
Jul 18 16:54:38 <nbd> i don't have a link
Jul 18 16:54:49 <nbd> but it should be possible to find it with google
Jul 18 16:58:20 <luke-jr> meh, useless google
Jul 18 18:47:19 <mb___> hm, a mips kernel doesn't report the function name in an oops correctly, if the function is static. it will report the next non-static function in the binary.
Jul 18 18:47:39 <nbd> yeah
Jul 18 18:47:45 <nbd> static functions are not in the symbol table iirx
Jul 18 18:47:46 <nbd> iirc
Jul 18 18:48:25 <mb___> very confusing
Jul 18 18:49:18 <nbd> yeah
Jul 18 18:49:22 <nbd> annoys me every time i encounter it
Jul 18 18:49:25 <nbd> is this different on x86?
Jul 18 18:49:39 <mb___> well, x86 is crap anyway :)
Jul 18 18:49:50 <mb___> PPC has the best oops reports, as far as I can see.
Jul 18 18:50:03 <mb___> They are almost always correct, unless you have huge memory corruption
Jul 18 18:50:37 <mb___> but x86 does prefix a ? if it's not sure about the symbol. That helps to easily filter out the bogus ones.
Jul 18 18:50:46 <nbd> interesting
Jul 18 18:51:19 <mb___> I wonder if that can also be done on mips
Jul 18 18:52:25 <nbd> and one thing i think would also help is to print out BadVAddr on data bus error or bogus pointer deref
Jul 18 18:52:38 <nbd> iirc that should always give us the faulting address
Jul 18 18:53:09 <mb___> what is BadVAddr?
Jul 18 18:53:23 <nbd> a cp0 register, iirc
Jul 18 18:53:28 <mb___> Oh btw, there seems to be something wrong with 47xx kernel config.
Jul 18 18:53:49 <mb___> if I do   make target/linux-{clean,prepare} QUILT=1 V=99
Jul 18 18:53:56 <mb___> it will ask for
Jul 18 18:53:58 <mb___>   BCM47xx flash device (MTD_BCM47XX) [N/m/y/?] (NEW)
Jul 18 18:54:05 <nbd> run make kernel_oldconfig
Jul 18 18:54:14 <nbd> and commit the config change afterwards
Jul 18 18:54:16 <mb___> Which will result in a fatal failure :)
Jul 18 18:54:39 <mb___> oh I see. I need to be careful. I also have some other config changes..
Jul 18 18:54:55 <nbd> it splits up the config again just like make kernel_menuconfig
Jul 18 18:57:03 <CIA-8> nbd * r11863 /trunk/package/madwifi/patches-r3776/300-napi_polling.patch: fix some unused variables
Jul 18 18:57:59 <mb___> I'm redoing the gpiommc stuff and trying another upstream submission. I added API so platforms can also register a device statically.
Jul 18 18:58:23 <mb___> Not really added, but reveiled
Jul 18 18:58:33 <mb___> It always was there
Jul 18 19:04:34 <luke-jr> sigh
Jul 18 19:04:38 <luke-jr> this is probably hopeless
Jul 18 19:04:48 <luke-jr> I found http://atheros.openwrt.net/trac.cgi/browser/branches/valxdater/tools/firmware-utils/src/lzma2eva.c
Jul 18 19:04:52 <luke-jr> but it doesn't match up
Jul 18 19:06:36 <nbd> that one is for something entirely else
Jul 18 19:06:41 <nbd> look for 63xx specific stuff
Jul 18 19:08:01 <luke-jr> I did
Jul 18 19:09:21 <luke-jr> tried: 63xx lzma script, 63xx lzma non-standard, 63xx lzma firmware non-standard, 63xx lzma firmware modified, 63xx lzma firmware, 63xx lzma
Jul 18 20:32:01 <luke-jr> haha, turned out to just omit one of the header fields
Jul 18 20:32:06 <luke-jr> or not
Jul 18 20:32:09 <luke-jr> now I get ……..CodeReal: invalid data
Jul 18 20:32:19 <luke-jr> but decompression was ok
Jul 18 20:32:21 <luke-jr> sigh
Jul 18 20:53:57 <CIA-8> mb * r11864 /trunk/ (7 files in 6 dirs): Update GPIO-based MMC driver
Jul 18 22:03:45 <aroscha> nbd: still here?
Jul 18 22:03:49 <nbd> yes
Jul 18 22:04:27 <aroscha> nbd: the echo "1" > /proc/sys/dev/wifi0/intmit etc commands worked very nice
Jul 18 22:04:42 <aroscha> I am in pure AP mode and no complaints so far
Jul 18 22:04:48 <aroscha> streaming videos over the link
Jul 18 22:05:02 <aroscha> I havent tested in adhoc mode yet
Jul 18 22:05:09 <nbd> cool
Jul 18 22:05:49 <aroscha> what does the echo "1" >  do?
Jul 18 22:06:29 <nbd> so you're just doing echo 1 now? because the previous command that i gave you had 0
Jul 18 22:06:41 <aroscha> oops
Jul 18 22:06:52 <nbd> no, it's fine
Jul 18 22:06:53 <aroscha> yes
Jul 18 22:06:54 <aroscha> sorry
Jul 18 22:06:56 <nbd> i just want to know if you did 1 or 0
Jul 18 22:07:00 <aroscha> I actually did echo "0"
Jul 18 22:07:00 <nbd> to put the test results in perspective
Jul 18 22:07:08 <aroscha> so it was ok
Jul 18 22:07:25 <nbd> it disables automatic noise immunity
Jul 18 22:07:41 <nbd> and the 0 with ofdm_weak_det changes an noise immunity parameter
Jul 18 22:07:42 <aroscha> meaning?
Jul 18 22:07:49 <nbd> so that it reduces sensitivity
Jul 18 22:07:55 <nbd> (as was default in much older hals)
Jul 18 22:08:10 <nbd> costs about 10-12 db of ofdm sensitivity
Jul 18 22:08:20 <aroscha> ah , hm that is a lot
Jul 18 22:08:25 <nbd> i know
Jul 18 22:08:32 <nbd> but in many setups it's the only way to get reliable wifi
Jul 18 22:08:33 <aroscha> but stability goes first
Jul 18 22:08:44 <nbd> btw. did you use the sam hal that is in openwrt svn or did you use a different one?
Jul 18 22:08:45 <aroscha> the rest you can do with proper antennas
Jul 18 22:08:53 <aroscha> i use openwrt trunk
Jul 18 22:08:55 <aroscha> as of today
Jul 18 22:09:07 <aroscha> reflashed the box  completely
Jul 18 22:49:32 <CIA-8> nbd * r11865 /trunk/package/madwifi/ (4 files in 2 dirs): fix some madwifi-testing bugs
Jul 18 22:50:50 <CIA-8> nbd * r11866 /trunk/package/madwifi/patches-r3776/ (41 files): refresh madwifi-testing patches
Jul 18 22:51:12 <CIA-8> nbd * r11867 /trunk/package/madwifi/patches-r3776/333-hal_init_msg.patch: fix hal module load msg
Jul 18 22:57:33 <CIA-8> nbd * r11868 /trunk/package/madwifi/ (90 files in 3 dirs): rename patches-r3776 to patches-testing
Jul 18 22:58:09 <CIA-8> nbd * r11869 /trunk/package/madwifi/patches-r3776/: remove empty directory
Jul 18 23:56:51 <NAiL> Is there a regular bugday still?
**** ENDING LOGGING AT Sat Jul 19 02:59:57 2008