**** BEGIN LOGGING AT Sun Nov 14 02:59:59 2010
Nov 14 03:29:39 <builder> build #30 of brcm63xx is complete: Success [build successful]  Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/30
Nov 14 05:39:09 <CIA-81> jow * r23989 /trunk/package/busybox/patches/244-udhcpc_add_6rd_option.patch: [package] busybox: implement support for 6RD (RFC5969 7.1.1, DHCP option 212) in udhcpc
Nov 14 06:04:54 <CIA-81> jow * r23990 /trunk/package/busybox/ (Makefile patches/244-udhcpc_add_6rd_option.patch): [package] busybox: add 6RD prefix sanity checking as mandated by RFC5969, bump pkg revision
Nov 14 07:21:35 <philipp64|laptop> jow_laptop: you around?
Nov 14 10:26:15 <loswillios> nbd: is nbd.name down?
Nov 14 11:31:11 <nbd> loswillios: not anymore
Nov 14 11:31:52 <nbd> apache was acting up
Nov 14 11:49:43 <CIA-81> acoul * r23991 /trunk/target/linux/generic/patches-2.6.36/992-mips_of_2.6.37_bp.patch: linux/generic: backport device tree support to MIPS from 2.6.37 to 2.6.36
Nov 14 12:42:43 <xMff> philipp64|laptop: now yes
Nov 14 12:58:06 <KanjiMonster> [florian]: in case you are interested (and even if not ;p), this is the error I am getting when I try to compile with a linux-mips git kernel: http://pastebin.com/X3a1ypGB
Nov 14 13:08:12 <[florian]> KanjiMonster: if_packet.h is missing an include <linux/types.h>
Nov 14 13:09:00 <KanjiMonster> [florian]: ah, ok
Nov 14 13:16:58 <KanjiMonster> [florian]: the include is there
Nov 14 13:17:45 <[florian]> is it protected with an #ifdef __KERNEL__ ?
Nov 14 13:20:57 <KanjiMonster> nope
Nov 14 13:21:57 <[florian]> humm
Nov 14 13:23:30 <KanjiMonster> [florian]: perhaps that's the problem? -> http://git.uclibc.org/uClibc/tree/libc/inet/netlinkaccess.h?h=0.9.31#n27 - it seems to just define some types and then pretent to have included linux/types.h
Nov 14 13:24:45 <[florian]> I guess so
Nov 14 13:29:50 <KanjiMonster> [florian]: I wonder why this doesn't work with the git tree, but works with .35 - the _be16 is already there in .35
Nov 14 13:31:10 <[florian]> yes, I am wondering too
Nov 14 15:22:39 <KanjiMonster> [florian]: also happens if I try to compile with 2.6.37-rc1, so its not an external kernel tree problem
Nov 14 15:23:46 <loswillios> xMff: does luci-app-ddns lack the attribute 'day'? I can only see minutes and hours
Nov 14 16:19:43 <sn9> NOR chip too large to fit in mapping. Attempting to cope...
Nov 14 16:19:43 <sn9> Reducing visibility of 32768KiB chip to 8192KiB
Nov 14 16:19:43 <sn9> wtf???
Nov 14 16:29:11 <loswillios> heh
Nov 14 16:29:26 <sn9> loswillios: what gives?
Nov 14 16:29:33 <loswillios> no idea sorry
Nov 14 16:30:10 <sn9> did somebody completely bollocks the ar7 target when i wasn't looking?
Nov 14 16:34:05 <[florian]> sn9: can you be more specific?
Nov 14 16:34:24 <[florian]> sn9: like which device you are running from etc...
Nov 14 16:34:36 <sn9> actiontec pk5000
Nov 14 16:34:54 <[florian]> I am not sure that one was even supported once
Nov 14 16:34:57 <sn9> it has a 32M flash chip
Nov 14 16:35:10 <sn9> it's a relatively new model
Nov 14 16:35:12 <[florian]> then the message is just fine
Nov 14 16:35:30 <sn9> the ar7 chips have infineon markings
Nov 14 16:35:31 <[florian]> the chip select window is certainly not enough to deal with a 32MiB flash without some changes
Nov 14 16:35:54 <sn9> well, that would explain the huge crash
Nov 14 16:36:18 <[florian]> 0x1fc00000 only allows 4MiB of flash to be mapped without setting up an alias
Nov 14 16:37:45 <[florian]> I do not have any ar7 docs, I wish I had, but do you know if there is a way to get the chip-select window size_
Nov 14 16:37:48 <[florian]> ?
Nov 14 16:38:00 <[florian]> that would allow the size passed to the physmap-flash driver to match the real flash size
Nov 14 16:38:38 <sn9> no, but blogic might
Nov 14 16:43:47 <blogic> nope
Nov 14 16:45:09 <sn9> still no ar7 docs from infineon?
Nov 14 17:02:26 <blogic> nope
Nov 14 17:02:29 <blogic> but i will ask
Nov 14 17:02:35 <blogic> i am visiting them soonish
Nov 14 17:02:42 <blogic> so i will do it face 2 face
Nov 14 17:41:10 <KanjiMonster> actiontec has a gpl tar for the pk5000 (at least they claim it's for it too), I would expect the info on how to acccess the whole 32 MB somewhere in it
Nov 14 17:41:35 <sn9> good idea
Nov 14 17:42:02 <sn9> it's a 2.4 kernel, though, but built in 2009
Nov 14 17:45:57 <danage> blogic: can i pm you right quick?
Nov 14 17:46:11 <danage> will only take a minute
Nov 14 17:46:58 <builder> build #25 of ps3 is complete: Failure [failed compile_3]  Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/25
Nov 14 18:02:23 <builder> build #27 of pxcab is complete: Failure [failed compile_3]  Build details are at http://tksite.gotdns.org:8010/builders/pxcab/builds/27
Nov 14 18:49:18 <CIA-81> obsy * r23992 /packages/net/transmission/ (Makefile files/transmission.config files/transmission.init): [packages] transmission: update to 2.12, remove obsolete options
Nov 14 19:04:32 <_trine> who maintains asterisk ?
Nov 14 19:05:16 <_trine> asterisk14 has been seg faulting for a bit now
Nov 14 19:19:54 <blogic> sure
Nov 14 19:29:06 <_trine> blogic, do you have any idea why asterisk should start to seg fault considering nothing has been change on it for quite a while
Nov 14 19:31:26 <blogic> magic
Nov 14 19:31:45 <crow> lol :)
Nov 14 19:32:07 <_trine> black magic
Nov 14 19:32:54 <_trine> I am trying to get this all working for a friend of mine who is taking it abroad with him
Nov 14 19:33:30 <_trine> I have everything working except for asterisk damnit
Nov 14 19:44:20 <crow> is this normal that b43 is included in image, where ath5k is selected and b43 not?? http://openwrt.pastebin.com/p8XjRr4R
Nov 14 19:45:40 <xMff> if your config is not fresh then yes, its normal
Nov 14 19:46:12 <crow> xMff is fresh.. i think let me post config
Nov 14 19:46:52 <xMff> fresh as in rm .config and not a single save before selecting the profile?
Nov 14 19:46:59 <philipp64|laptop> why do I keep needing to select CONFIG_L2TP_V3=n manually?
Nov 14 19:48:44 <crow> http://openwrt.pastebin.com/3w18rjdW
Nov 14 19:48:46 <xMff> maybe because kmod-l2tp is enabled and kmod-l2tp-eth not
Nov 14 19:49:22 <crow> xMff i rmed it, selected profile, target and did my stuff, and Exit (to save)
Nov 14 19:49:30 <xMff> odd
Nov 14 19:49:47 <crow> xMff i actualy did it twice as i was thinking i did prob..
Nov 14 19:50:22 <xMff> Hauke: ^
Nov 14 19:55:21 <Hauke> crow: scripts/ipkg-make-index.sh (the script with "Generating index for package..") goes over every *.ipk file in the packages dir an adds it into the index
Nov 14 19:56:19 <Hauke> crow: when you build an image with kmod-b43 the ipk file will be placed in the packages dir
Nov 14 19:57:14 <Hauke> crow: in the next build, even if xou have not selected kmod-b43 the file will still be in the packages dir and it will be added into the index
Nov 14 19:57:49 <Hauke> crow: you should remove bin/brcm47xx/ and build the image again and kmod-b43 should not be there any more
Nov 14 19:58:08 <crow> Hauke i did "svn -R revert ." then "make target/linux/clean" then make menuconfig selected profile, target, my stuff and exited to save .conf
Nov 14 19:58:10 <Hauke> crow: and I will have a look at CONFIG_L2TP_V3
Nov 14 19:58:54 <crow> ok let me remobe bin/brcm47xx directory and rebuild image
Nov 14 19:59:19 <crow> remove*
Nov 14 20:00:11 <sn9> [florian]: if i just use the first 8M of the chip for now, shouldn't it work?
Nov 14 20:02:48 <crow> Hauke seems that fixed the problem :)
Nov 14 20:06:18 <[florian]> sn9: certainly yes
Nov 14 20:06:36 <sn9> well, it isn't
Nov 14 20:57:55 <CIA-81> hauke * r23993 /trunk/ (4 files in 2 dirs):
Nov 14 20:57:56 <CIA-81> kernel: fix l2tp kernel packages
Nov 14 20:57:56 <CIA-81> l2tp-ip and l2tp-eth should depend on l2tp_netlink
Nov 14 20:57:56 <CIA-81> fix pppol2tp for kernel >= 2.6.35
Nov 14 21:15:40 <builder> build #22 of xburst is complete: Failure [failed compile_4]  Build details are at http://tksite.gotdns.org:8010/builders/xburst/builds/22
Nov 14 21:18:26 <philipp64|laptop> [florian]: not sure what's going on the GEODE_LX and the L1 cache size.
Nov 14 21:19:57 <[florian]> philipp64|laptop: try removing L1_CACHE_SHIFT fromthe kernel configuration and see if that helps
Nov 14 21:36:23 <philipp64|laptop> it still puts them back to 6.
Nov 14 21:41:10 <philipp64|laptop> [florian]: found some weirdness.
Nov 14 21:41:41 <philipp64|laptop> in target/linux/x86/config-default, CONFIG_X86_GENERIC is turned on. that's forcing it to 6.
Nov 14 21:42:23 <philipp64|laptop> but also, looking at it I see CONFIG_X86_L1_CACHE_BYTES=64 / CONFIG_X86_L1_CACHE_SHIFT=4  ?????
Nov 14 21:42:40 <philipp64|laptop> how can it be BYTES=64 but SHIFT=4 ????
Nov 14 21:46:33 <philipp64|laptop> doing a patch...  should X86_GENERIC be turned off in target/linux/x86/config-default but turned on in target/linux/x86/generic/config-default?
Nov 14 21:53:49 <philipp64|laptop> [florian]: just sent a patch.  can you please apply it?
Nov 14 22:01:30 <philipp64|laptop> sent an alternate patch that only affects Geode-based platforms.
Nov 14 22:03:16 <philipp64|laptop> either way, it's hard to get the correct values for target/linux/x86/generic/profiles/* (i.e. net4801, alix, etc)
Nov 14 22:06:35 <CIA-81> jow * r23994 /trunk/package/hotplug2/files/hotplug2.rules:
Nov 14 22:06:35 <CIA-81> [PATCH] add input subsystem to hotplug rules
Nov 14 22:06:35 <CIA-81> This patch adds the input subsystem to the hotplug rules, making it possible to
Nov 14 22:06:35 <CIA-81> react on the addition of input devices (e.g. USB numpads).
Nov 14 22:06:35 <CIA-81> Signed-off-by: Stefan Tomanek <stefan.tomanek+openwrt@wertarbyte.de>
Nov 14 22:07:14 <CIA-81> jow * r23995 /branches/backfire/package/hotplug2/files/hotplug2.rules: [backfire] merge r23994
Nov 14 22:40:12 <sn9> [florian]: ever have problems using pspboot's ftp server to flash?
Nov 14 22:41:39 <crow> philipp64|laptop is that for "AMD Geode LX800 " to?
Nov 14 22:55:10 <KanjiMonster> acoul: ping
Nov 14 22:57:39 <[florian]> sn9: always :)
Nov 14 22:58:15 <sn9> [florian]: how do i minimize that?
Nov 14 22:58:19 <[florian]> sn9: using the tcp_wmem options here solved it: https://forum.openwrt.org/viewtopic.php?pid=71139
Nov 14 22:58:32 <[florian]> sn9: I am off to bed now, see you later
Nov 14 22:58:38 <KanjiMonster> [florian]: I just tried 2.6.37 for ar71xx (since there are already some patches in trunk for that), and I hit the same error, so I assume this is a generic uclibc + 2.6.37 problem, at least on mips
Nov 14 22:59:08 <KanjiMonster> [florian]: and good night :)
Nov 14 23:22:21 <philipp64|laptop> sn9: looking
Nov 14 23:22:43 <philipp64|laptop> it should be for all Geode LX processors, I believe.
Nov 14 23:32:52 <philipp64|laptop> this says it's a 32-bit cache line: http://7-cpu.com/cpu/GeodeLX.html
Nov 14 23:34:09 <philipp64|laptop> which concurs with ./arch/x86/Kconfig.cpu's defaults for X86_L1_CACHE_SHIFT
Nov 14 23:35:39 <philipp64|laptop> errr...  32 Byte.
Nov 14 23:38:02 <philipp64|laptop> xMff: ping
Nov 14 23:43:45 <philipp64|laptop> hmmm... Geode NX is also 32 bytes... though it's Athalon based and not Cyrix based.
Nov 14 23:48:46 <philipp64|laptop> except that it's not. Hmm.  https://patchwork.kernel.org/patch/58052/
Nov 14 23:50:04 <crow> philipp64|laptop i have Alix 3D3 board and in its configuration i see 256 KB cache (64K data + 64K instruction + 128K L2)
Nov 14 23:50:52 <philipp64|laptop> crow: yeah, but how wide are the individual cache lines?
Nov 14 23:51:26 <philipp64|laptop> there are many ways to divide up 64KB
Nov 14 23:51:52 <philipp64|laptop> what do the last 5 lines of /proc/cpuinfo say?
Nov 14 23:52:01 <crow> cant help you with that i have this mobo http://www.pcengines.ch/alix3d3.htm
Nov 14 23:52:05 <crow> jus a sec
Nov 14 23:53:03 <crow> here is complete /proc/cpuinfo http://openwrt.pastebin.com/TFgFujTq
Nov 14 23:53:30 <crow> its current runing *buntu distro
Nov 14 23:53:36 <philipp64|laptop> look at clflush_size
Nov 14 23:53:42 <philipp64|laptop> that's what we're talking about.
Nov 14 23:54:13 <crow> :)
Nov 14 23:57:41 <CIA-81> jow * r23996 /trunk/package/6to4/ (. Makefile files/ files/6to4.hotplug files/6to4.sh): [package] add 6to4 - provides full automatic 6to setup with optional configuration in /etc/config/network
Nov 15 00:23:15 <sn9> what am i doing wrong? http://www.gimpelevich.san-francisco.ca.us/danielg4/booting.log
Nov 15 00:35:30 <sn9> anyone? bueller? bueller?
Nov 15 00:38:56 <sn9> bueller?
Nov 15 00:41:25 <xMff> sn9: looks like init=/etc/preinit is missing
Nov 15 00:41:46 <xMff> either through forced commandline stuff or bootloader or preinit-as-init patch
Nov 15 00:42:10 <sn9> looks to be present at build_dir/target-mipsel_uClibc-0.9.31/root-ar7/etc/preinit
Nov 15 00:42:45 <sn9> oh, you mean the arg
Nov 15 00:42:52 <xMff> ack
Nov 15 00:43:03 <sn9> the preinit-as-init patch is in use
Nov 15 00:43:09 <xMff> hm
Nov 15 00:43:17 <xMff> rootfs fuckup then maybe
Nov 15 00:43:32 <sn9> yes, but how?
Nov 15 00:44:08 <sn9> wait a sec ---
Nov 15 00:44:15 <sn9> i think i know
Nov 15 00:47:22 <swalker> updated openwrt/upstream, https://home.comcast.net/~sdwalker/uscan/uscan.shtml
Nov 15 00:48:02 <sn9> how do i cross-ldd, again?
Nov 15 00:48:30 <xMff> ./staging:dir/toolchain-*/bin/*-ldd ?
Nov 15 00:48:49 <sn9> that's a mips executable
Nov 15 00:49:17 <xMff> hm?
Nov 15 00:49:34 <xMff> should be a cross toolchain util
Nov 15 00:49:47 <sn9> $ file staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.31/bin/ldd
Nov 15 00:49:47 <sn9> staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.31/bin/ldd: ELF 32-bit LSB executable, MIPS, MIPS32 version 1, dynamically linked (uses shared libs), with unknown capability 0xf41 = 0x756e6700, not stripped
Nov 15 00:50:00 <xMff> k
Nov 15 00:55:46 <CIA-81> jow * r23997 /trunk/package/6to4/files/6to4.sh: [package] 6to4: remove leftover peeraddr stuff
Nov 15 01:23:59 <CIA-81> jow * r23998 /trunk/package/6to4/files/6to4.sh: [package] 6to4: use revert instead of delete, avoids aggregating deltas in /tmp/.uci over time
Nov 15 01:31:14 <CIA-81> jow * r23999 /packages/ipv6/radvd/files/radvd.init: [packages] radvd: fix a config_get_bool call in initscript, fixes "sh: bad number" during restart
Nov 15 01:47:20 <sn9> xMff: i accidentally the libc config option, but now that i put it back, the error is different: http://www.gimpelevich.san-francisco.ca.us/danielg4/booting.log
Nov 15 01:48:03 <xMff> there's a word missing
Nov 15 01:48:11 <xMff> accidentally what?
Nov 15 01:48:14 <sn9> in the log?
Nov 15 01:48:16 <sn9> oh'
Nov 15 01:48:24 <sn9> deleted
Nov 15 01:48:28 <xMff> I see
Nov 15 01:48:58 <xMff> attempted to kill init... hmm that could happen if the userland is compiled without softfloat
Nov 15 01:49:05 <sn9> ahh
Nov 15 01:49:24 <sn9> maybe that should be a default
Nov 15 01:49:26 <philipp64|laptop> xMff: can you have a look at the X86_GENERIC patch(es)?
Nov 15 01:49:39 <xMff> the cacheline stuff?
Nov 15 01:51:30 <sn9> xMff: CONFIG_SOFT_FLOAT=y
Nov 15 01:51:40 <sn9> so that's not it
Nov 15 01:53:44 <sn9> is there anything else critical i might have left out? options nothing will work without really should not be deselectable without at least a warning
Nov 15 01:54:12 <xMff> well base-files and mtd should be there
Nov 15 01:54:17 <xMff> and busybox obviously
Nov 15 01:54:36 <xMff> I have no real clue what it could be, sorry. Can you pass a kernel cmdline?
Nov 15 01:54:44 <sn9> mtd would not kill init, and the other two are definitely there
Nov 15 01:54:46 <xMff> If yes, maybe try init=/bin/sh
Nov 15 01:54:59 <xMff> weill missing mtd can kill preinit
Nov 15 01:55:05 <sn9> and no, there is nowhere to put the cmdline that i can see
Nov 15 01:55:10 <xMff> erm nvm
Nov 15 01:55:12 <xMff> not kill
Nov 15 01:55:17 <xMff> just no unlock
Nov 15 01:56:21 <sn9> mtd is there anyway
Nov 15 02:31:24 <philipp64|laptop> xMff: yes, the cache line stuff.
Nov 15 02:38:39 <sn9> xMff: ok, i forced init=/bin/sh and still got attempt to kill init
Nov 15 02:39:32 <sn9> clearly, i'm leaving out something busybox needs
**** ENDING LOGGING AT Mon Nov 15 02:59:58 2010