**** BEGIN LOGGING AT Tue Oct 11 02:59:57 2011
Oct 11 03:44:36 <builder_> build #60 of mpc52xx is complete: Failure [failed compile_10]  Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/60
Oct 11 07:17:17 <CIA-62> nico * r28409 /branches/packages_10.03.1/lang/ (distribute/Makefile mako/Makefile): [backfire/packages] backport [28402] (distribute) & [28403] (mako) changes
Oct 11 07:28:37 <CIA-62> nico * r28410 /branches/packages_10.03.1/net/freeswitch/Makefile: (log message trimmed)
Oct 11 07:28:37 <CIA-62> [backfire/packages] freeswitch: various build fixes:
Oct 11 07:28:37 <CIA-62> * fix -mod-cdr-pg-csv dependency on libpq
Oct 11 07:28:37 <CIA-62> * fix -mod-cdr-sqlite dependency on libsqlite3
Oct 11 07:28:37 <CIA-62> * mark -mod-erlang-event as broken (fails to build on x86)
Oct 11 07:28:37 <CIA-62> * mark -mod-event-zmq as broken (needs ZeroMQ, not packaged yet)
Oct 11 07:28:38 <CIA-62> * mark -mod-mp4 as broken (needs libmp4v2, not packaged yet)
Oct 11 08:47:27 <builder_> build #84 of ps3 is complete: Failure [failed compile_6]  Build details are at http://buildbot.openwrt.org:8010/builders/ps3/builds/84
Oct 11 09:04:11 <mystica555_> jow: out of curiosity, regarding my bug #10181 should i perhaps file a second bugreport against luci for the specific bit of 'have luci detect invalid htmode options and prevent this from affecting all of networking' ?
Oct 11 09:04:45 <jow_laptop> maybe, but I cannot reproduce the described behaviour in the first place
Oct 11 09:05:14 <jow_laptop> bad wifi config simply causes the wlan0 to stay down / be not created here
Oct 11 09:05:17 <jow_laptop> lan is not affected
Oct 11 09:05:24 <mystica555_> weird crap.
Oct 11 09:20:42 <mystica555_> why is it that i always get the weird un-reproducible bugs
Oct 11 09:21:15 <mystica555_> i bet i must've done something in another life to give me karma that in this life, im the problem solver; but on the other hand, it also gave me insane problems to solve
Oct 11 09:22:53 <CIA-62> florian * r28411 /packages/libs/sctp/Makefile: [packag] sctp: remove binutils dependency.
Oct 11 09:25:03 <mystica555_> some sort of test of humility and patience lol
Oct 11 09:26:36 <jow_laptop> I suggest to use a recent trunk and then try to build various invalid wifi configs
Oct 11 09:26:57 <jow_laptop> luci calls /etc/init.d/network restart internally
Oct 11 09:27:13 <mystica555_> this is currently the sep29 compile on the router in question
Oct 11 09:27:42 <mystica555_> havent upgraded really since then..
Oct 11 09:27:50 <mystica555_> a few new packages, but no opkg upgrade
Oct 11 13:46:37 <CIA-62> jow * r28412 /trunk/package/base-files/ (Makefile files/lib/network/config.sh): [package] base-files: introduce "igmp_snooping" option to toggle IGMP snooping on bridge interfaces
Oct 11 14:29:21 <CIA-62> nico * r28413 /packages/Xorg/lib/hippo-canvas/Makefile: [packages] hippo-canvas: make python-hippo-canvas also depend on @FEATURE_drawing-backend_libX11
Oct 11 14:29:51 <voyo> hey, I suspect I have some issues with endianness on routerstation (MIPS AR7161). is there someone experienced with using this platform ?
Oct 11 14:48:52 <mystica555_> colloquially ive seen a large number of atheros platforms big-endian but as its configurable, i dont know about that specific board
Oct 11 14:49:26 <mystica555_> i've often wondered if there was a performance benefit/penalty for either on that specific MIPS arch
Oct 11 14:54:43 <voyo> mystica555_ : I want to assure, if is enough  to compile binary for BE/LE (gcc switches) , and both objects can run on the system. (or only one of it), or I have to 'switch' endianness somehow (with redboot maybe? or some pin on the board? dont know...) any advice ?
Oct 11 14:55:06 <mystica555_> i believe its set in the boot loader
Oct 11 14:55:27 <mystica555_> i've tried to run different endian things on my router, mistakenly, it doesnt like it so much (wndr3700;same CPU)
Oct 11 14:56:10 <mystica555_> i would think that even if the cpu could run both endians at the same time, the OS itself would have to support this
Oct 11 14:57:46 <KanjiMonster> voyo: even if the chip supports both endianess, it is usually locked to one; and iirc you can't really create binaries that run on both (unless you create some sort of fat binary that contains both the LE and the BE object code, making it twice as big)
Oct 11 14:59:28 <voyo> ok.. in kernel_menuconfig, for this platform I can choose only "Endianess selection (Big endian) " . is that safe to assume that EVERYTHING, whole system, and also CPU is running BE ?
Oct 11 14:59:39 <[florian]> not everything
Oct 11 14:59:46 <[florian]> usually some access to hardware might be little-endian
Oct 11 14:59:49 <[florian]> especially pci/pcie
Oct 11 15:00:19 <KanjiMonster> right
Oct 11 15:00:37 <voyo> hm.... maybe thats my issu... but not sure how to prove it :-.
Oct 11 15:00:40 <voyo> :-/
Oct 11 15:00:59 <KanjiMonster> voyo: what exactly is your issue?
Oct 11 15:01:06 <KanjiMonster> or your symptoms
Oct 11 15:01:52 <voyo> Im usig mono,I created package for owrt. works fine on x86, on arm. on this mips platform - only simply 'helloworld'. anything more complicated is creating exceptions
Oct 11 15:01:59 <mystica555_> the 'fat binary' approach would indeed be an extension to the 'os has to support it' bit i mentioned :)
Oct 11 15:02:13 <voyo> with my other developer we are suspecting endinness problems..
Oct 11 15:02:17 <mystica555_> like apple has done oh so long with their m68k/ppc, then ppc/x86 binaries..
Oct 11 15:03:26 <voyo> it might be related also to byte alignment..
Oct 11 15:03:50 <mystica555_> being that mono runs things compiled to bytecode...hrm..
Oct 11 15:04:07 <KanjiMonster> voyo: ah, okay, yes; that is quite likely that it makes assumptions about the byte order when reading from the filesystem - reading 4 bytes as a 32bit integer will give you a different result on BE and LE
Oct 11 15:04:49 <voyo> KanjiMonster : yes, exatcly
Oct 11 15:10:20 <KanjiMonster> voyo: then you'll probably have to do an endianess audit on everything loading from the filesystem (sprinkling printfs and comparing the output with a LE run can help tracing differences ;) - maybe the vm even does assumtions about the byteorder of values stored in ram
Oct 11 15:11:43 <voyo> KanjiMonster:  yeah, I tought about this.. but I had hope for some less time consuming tests.. ;)
Oct 11 15:14:08 <KanjiMonster> voyo: depending on how good you know mono, you might also try to first reduce the program failing to a minimal test case, then try to backtrace to the source of the error by going up the call chain
Oct 11 15:19:03 <voyo> ok.. tnx guys
Oct 11 15:24:12 <CIA-62> nico * r28414 /branches/packages_10.03.1/Xorg/lib/hippo-canvas/Makefile: [backfire/packages] hippo-canvas: merge [28413]
Oct 11 15:26:26 <voyo> btw.. is there a way to force 'make package/ddddd/compile' to NOT recompress package, but use files which are already in build_dir/dddddd ?
Oct 11 15:33:08 <nbd> voyo: if you've unpacked the package with QUILT=1 it'll keep files around even if the package dir changes
Oct 11 15:43:12 <voyo> tnx nbd
Oct 11 15:59:12 <CIA-62> nico * r28415 /packages/libs/libxapian/patches/004-xapian_config_no_ltlibs.patch:
Oct 11 15:59:12 <CIA-62> [packages] libxapian: fix xapian-config M4 macro
Oct 11 15:59:12 <CIA-62> don't check for xapian using 'xapian-config --ltlibs', since .la files are not installed anymore
Oct 11 16:05:57 <CIA-62> nico * r28416 /packages/lang/xapian-bindings/ (Makefile patches/010-build.patch):
Oct 11 16:05:57 <CIA-62> [packages] xapian-bindings: cleanup
Oct 11 16:05:57 <CIA-62> provide PYTHON_LIB, instead of patching configure
Oct 11 16:43:53 <CIA-62> nico * r28417 /branches/packages_10.03.1/ (3 files in 3 dirs): [backfire/packages] merge [28415] & [28416]
Oct 11 16:53:44 <CIA-62> nico * r28418 /packages/multimedia/ffmpeg/Makefile: [packages] ffmpeg: install libswscale headers (closes: #9447), cleanup
Oct 11 18:18:41 <CIA-62> jow * r28419 /trunk/include/autotools.mk: include: autotools.mk: do not fail in patch-libtool fixup if patch is already applied
Oct 11 18:33:14 <CIA-62> nico * r28420 /branches/packages_10.03.1/net/l2tpv3tun/: [backfire/packages] l2tpv3tun: remove, needs linux kernel 2.6.33+
Oct 11 19:31:07 <mystica555_> jow_laptop: when you mentioned you could not replicate the issue, did you even get the 'brctl: bridge br-lan: Operation not supported brctl: bridge br-lan: File exists  errors?
Oct 11 20:32:28 <builder_> build #85 of pxcab is complete: Failure [failed compile_10]  Build details are at http://buildbot.openwrt.org:8010/builders/pxcab/builds/85
Oct 11 20:36:01 <jow_laptop> mystica555_: nope
Oct 11 20:36:09 <mystica555_> strange..
Oct 11 20:36:20 <jow_laptop> mystica555_: however it looks like the network config got inconsistent
Oct 11 20:36:29 <jow_laptop> see whats in /var/state
Oct 11 20:36:29 <mystica555_> in what way?
Oct 11 20:39:00 <mystica555_> ...i need new glasses; i just tried 6 times logging into a router as roote
Oct 11 20:39:45 <mystica555_> so do you want me to try this when i induce the change that breaks it?
Oct 11 20:39:54 <mystica555_> (looking in /var/state)
Oct 11 20:41:00 <jow_laptop> yes
Oct 11 20:41:21 <jow_laptop> also /var/run/hostapd-phy0.conf
Oct 11 20:41:39 <mystica555_> phy0 would be 2.4; the change was made in 5ghz
Oct 11 20:41:55 <jow_laptop> 1 then
Oct 11 20:46:58 <mystica555_> http://paste.pocoo.org/show/CCCMAKtNNEqtoqXFOrlL/
Oct 11 20:47:42 <mystica555> what the hell
Oct 11 20:48:56 <builder_> build #80 of ppc44x is complete: Failure [failed compile_4]  Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/80
Oct 11 21:01:11 <jow_laptop> interesting
Oct 11 21:02:14 <jow_laptop> whats the ifconfig and route situation afterwards?
Oct 11 21:02:27 <jow_laptop> and does it change anything if you run wifi down; wifi
Oct 11 21:16:30 <mystica555> sec ill try that
Oct 11 21:21:07 <mystica555_> http://paste.pocoo.org/show/gc8Gz3D5ZvIA2MMlvFKm/
Oct 11 21:26:09 <jow_laptop> hm, no it recovers gracefully here
Oct 11 21:29:01 <mystica555_> and i just tried running it through strace... am i barking up the wrong tree expecting brctl to also run through strace when called via wifi?
Oct 11 21:29:06 <mystica555_> (it = wifi)
Oct 11 21:30:29 <jow_laptop> strace -f
Oct 11 21:30:46 <jow_laptop> however its easier to do  sh -x /sbin/wifi up
Oct 11 21:31:58 <mystica555_> just did my wonderful 'bypass kmods and upgrade everythingelse' oneliner... lets try after that...
Oct 11 21:32:00 <mystica555_> r file in `opkg list-installed |grep -v kmod |awk '{print $1}'` ; do opkg upgrade $file; done
Oct 11 21:32:03 <mystica555_> *for
Oct 11 21:34:29 <mystica555_> + brctl addif br-lan wlan1
Oct 11 21:34:29 <mystica555_> seems to give me the 'brctl: bridge br-lan: Operation not supported' error
Oct 11 21:34:32 <mystica555_> when it doesnt exist
Oct 11 21:34:44 <mystica555_> do you want to have the output of that entire sh -x ? x.x
Oct 11 21:35:28 <jow_laptop> no
Oct 11 21:35:56 <mystica555_> i didn't think so...(its massive)
Oct 11 21:36:02 <jow_laptop> not existing is different
Oct 11 21:36:03 <jow_laptop> root@OpenWrt:~# brctl addif br-lan foobar
Oct 11 21:36:04 <jow_laptop> brctl: iface foobar: No such device
Oct 11 21:36:17 <mystica555_> hm
Oct 11 21:36:35 <jow_laptop> it seems to me that hostapd puts the wifi into a weird state
Oct 11 21:37:19 <mystica555_> http://paste.pocoo.org/show/491245/
Oct 11 21:38:59 <jow_laptop> whats in brctl show?
Oct 11 21:40:00 <mystica555_> bridge name     bridge id               STP enabled     interfaces
Oct 11 21:40:00 <mystica555_> br-lan          8000.e0469a5eefe6       no              eth0.1
Oct 11 21:40:00 <mystica555_>                                                         wlan0
Oct 11 21:40:23 <jow_laptop> the bridge is still defunct?
Oct 11 21:40:33 <mystica555_> yes, it cant ping anything
Oct 11 21:40:36 <mystica555_> on the lan
Oct 11 21:40:45 <jow_laptop> brctl delif br-lan wlan0
Oct 11 21:40:49 <jow_laptop> what happens then?
Oct 11 21:41:29 <mystica555_> br-lan          8000.e0469a5eefe6       no              eth0.1
Oct 11 21:41:35 <mystica555_> is all thats listed, and still no lan ping
Oct 11 21:42:25 <jow_laptop> odd
Oct 11 21:48:31 <mystica555_> yes..
Oct 11 23:08:40 <CIA-62> jogo * r28421 /trunk/target/linux/generic/files/drivers/net/phy/ar8216.c: ar8216: use mdiobus_{read,write} to ensure proper locking
Oct 11 23:08:43 <CIA-62> jogo * r28422 /trunk/target/linux/ar71xx/patches-2.6.39/920-mdio_bitbang_ignore_ta_value.patch:
Oct 11 23:08:43 <CIA-62> ar71xx: add a workaround for ar8316 not always driving the TA bit to low
Oct 11 23:08:43 <CIA-62> AR8316 behind a GPIO bitbanged MDIO bus fails to drive the turnaround bit
Oct 11 23:08:43 <CIA-62> to low despite returning a valid value. Ignore it and just use the
Oct 11 23:08:44 <CIA-62> returned value anyway.
Oct 11 23:08:46 <CIA-62> jogo * r28423 /trunk/target/linux/ar71xx/ (6 files in 4 dirs): linux: ar71xx: add support for Mikrotik Routerboard RB493G
Oct 11 23:11:00 <CIA-62> jow * r28424 /packages/lang/distribute/Makefile: [packages] distribute: make the host build depend on python to ensure that python-package.mk is available, fixes random distribute build failures during parallel build runs
Oct 11 23:14:28 <CIA-62> jow * r28425 /branches/packages_10.03.1/lang/distribute/Makefile: [backfire/packages] merge r28424
**** ENDING LOGGING AT Wed Oct 12 02:59:57 2011