**** BEGIN LOGGING AT Thu Nov 29 02:59:59 2012
Nov 29 07:42:55 <OpenWrt-Commits> juhosg r34402 trunk/ (12 files in 10 dirs) * kernel: update linux 3.6 to 3.6.8
Nov 29 11:10:23 <karlp> what does it take to get oprofile on openwrt?
Nov 29 11:10:37 <karlp> I've done make kernel_menuconfig and added profiling and oprofile there,
Nov 29 11:10:50 <karlp> but the oprofile package still doesn't turn up in regular make menuconfig?
Nov 29 11:11:48 <jow_laptop> you need to enable profiling in the regular menuconfig
Nov 29 11:14:23 <karlp> ahh, global build settings. thanks
Nov 29 11:14:42 <jow_laptop> it will propagate down to the kernel conf
Nov 29 11:26:09 <karlp> is perf in in AA or trunk?
Nov 29 11:26:29 <karlp> (I've never actually used either, but I'm not sure yet whether I can reproduce this on desktop linux)
Nov 29 12:00:50 * dwmw2_gone boots his geos on 3.6.8
Nov 29 12:04:37 <stintel> :)
Nov 29 12:06:07 <dwmw2_gone> looks like AA does work OK with teql. The scripts handle it fine, and netifd doesn't need to do anything
Nov 29 12:13:52 <jow_laptop> dwmw2_gone: btw, br2684ctl is only started if you have an "atm-bridge" section in the network config
Nov 29 12:14:07 <jow_laptop> if you delete this section it will remain disabled on boot
Nov 29 12:14:32 <dwmw2_gone> I had just worked out that I'd gained one of those.
Nov 29 12:14:37 * dwmw2_gone wonders wtf put that there
Nov 29 12:14:44 <jow_laptop> uci-defaults
Nov 29 12:15:28 <jow_laptop> actually network defaults: https://dev.openwrt.org/browser/trunk/target/linux/x86/geos/base-files/etc/config/network
Nov 29 12:15:32 <dwmw2_gone> aha
Nov 29 12:15:50 <dwmw2_gone> it was there in my old (r31574) config which I copied over to the new AA image.
Nov 29 12:15:58 <dwmw2_gone> but it didn't *work* there for some reason. Which was convenient
Nov 29 12:15:58 <jow_laptop> ah ok
Nov 29 12:16:04 <jow_laptop> :)
Nov 29 12:16:06 <dwmw2_gone> ok, if it's just a default and not hard-coded then that's fine.
Nov 29 12:17:25 <dwmw2_gone> I don't seem to have /lib/network/functions.sh in my old image
Nov 29 12:17:40 <jow_laptop> yes, it is a recent adddition
Nov 29 12:18:19 <dwmw2_gone> so can this atm-bridge be configured from luci?
Nov 29 12:18:31 <dwmw2_gone> or is it automatic with a pppoe(oa) network config?
Nov 29 12:28:00 <dwmw2_gone> ah, yes we do
Nov 29 12:32:22 <jow_laptop> yes, should be possible from the network itnerface overview
Nov 29 12:40:12 <dwmw2_gone> got it, thanks
Nov 29 12:54:25 <karlp> can I rebuild just certain packages with debugging information by editing their package makefiles directly? will they get overwritten by antyhing?
Nov 29 12:57:02 <jwendell> a strip is done automatically
Nov 29 12:57:35 <karlp> I guess I can rebuild certain binaries myself manually
Nov 29 12:57:37 <jwendell> unless you setup that in make menuconfig (build options, compile with debug info)
Nov 29 12:59:03 <karlp> won't that rebuild everything with debug? which will be far to big to fit in flash?
Nov 29 12:59:25 <karlp> what's an easy way of checking if the binary is stripped or not?
Nov 29 12:59:30 <jow_laptop> if you use gdbserver + gdb you can deploy stripped binaries on the target and use the local unstripped ones to resolve symbols
Nov 29 12:59:40 <jow_laptop> there is no need to deploy unstripped binaries
Nov 29 12:59:42 <dwmw2_gone> would be lovely to package the debuginfo separately, like Fedora does
Nov 29 12:59:48 <dwmw2_gone> so you can install it for just the packages you need.
Nov 29 12:59:56 <dwmw2_gone> locally or for use with gdbserver
Nov 29 13:00:01 <karlp> currently using oprofile on the device, so think it does need unstripped
Nov 29 13:00:28 <jow_laptop> in this case  make package/foo/{clean,compile} V=99 CONFIG_DEBUG=y STRIP=/bin/true
Nov 29 13:01:37 <karlp> thanks, building that now.
Nov 29 13:03:35 <dwmw2_gone> what am I missing to make kallsyms actually work for modules?
Nov 29 13:04:33 <dwmw2_gone> I rebuilt with CONFIG_KALLSYMS and even CONFIG_DEBUG_INFO and still only get EIP for crashes in modules
Nov 29 13:05:07 <jow_laptop> did you reflash the kernel ?
Nov 29 13:05:13 <dwmw2_gone> yeah
Nov 29 13:06:34 <dwmw2_gone> I'll try it all again now the daystar is up
Nov 29 13:06:37 <dwmw2_gone> sometimes that helps :)
Nov 29 13:07:06 <jow_laptop> maybe also run a "make target/linux/clean" to really force recompiling all of thek ernel
Nov 29 13:07:08 <dwmw2_gone> and now I've removed that *huge* lost+found directory from my boot partition, I can fit an extra bzImage on there :)
Nov 29 13:07:15 <jow_laptop> not sure if enabling KALLSYMS alone does it
Nov 29 13:07:18 <dwmw2_gone> I actually did it manually this time
Nov 29 13:26:07 <dwmw2_gone> hm, the list of associated stations in the wifi status page doesn't make it clear which radio they're attached to.
Nov 29 13:26:24 <dwmw2_gone> It gives the SSID, but I have the same SSID on both 2.4G and 5G networks so clients can roam between them
Nov 29 14:27:13 <rektide> jow_laptop: that patch http://pastebin.com/JwVP7gMj indeed fixed my suite of lzma errors (ex: lmza_properties_decode@XZ_5.0 undefined), re: https://dev.openwrt.org/ticket/12539#ticket
Nov 29 14:28:54 <rektide> i have commented on the issue saying as much https://dev.openwrt.org/ticket/12539#comment:2
Nov 29 14:28:57 <rektide> thanks so much for the help
Nov 29 14:29:07 <[florian]> rektide: it has been committed, thanks for testing
Nov 29 14:30:48 <rektide> i saw, i pulled, i compiled.
Nov 29 15:22:14 <OpenWrt-Commits> juhosg r34403 trunk/target/linux/ (37 files in 37 dirs) * remove common symbols from 3.3 configs
Nov 29 15:22:17 <OpenWrt-Commits> juhosg r34404 trunk/target/linux/ (11 files in 11 dirs) * remove common symbols from 3.6 configs
Nov 29 15:33:54 <karlp> any other suggestions for cpu profiling? oprofile's non-stripped symbols worked, but I need more symbols than I could fit in flash, I can fiddle around with remote filesystems, but are there better ways?
Nov 29 15:37:05 <[florian]> initramfs?
Nov 29 15:37:32 <karlp> I'm afraid I need more explanation of that.
Nov 29 15:37:47 <[florian]> ramdisk/initramfs is a rootfs embedded in your kernel
Nov 29 15:37:53 <[florian]> which runs entirely from ram
Nov 29 15:38:33 <[florian]> if your problem is that oprofile produces too much data, then a remote filesystem is probably the key
Nov 29 15:38:57 <[florian]> if your problem is more, how can I get my rootfs with unstripped kernel to fit in flash, then I'd try ramdisk/initramfs
Nov 29 15:40:03 <karlp> ok, remote filesystems are probably better for me now, but are there other options that work on owrt other than oprofile?
Nov 29 15:40:16 <[florian]> which target are you working with?
Nov 29 15:40:27 <karlp> atheros, ar2317
Nov 29 15:40:29 <[florian]> there is the perf tool which is also available and packaged
Nov 29 15:40:37 <karlp> in what release?
Nov 29 15:40:48 <[florian]> it needs to be built from the packages feed
Nov 29 15:40:59 <karlp> I'm in backfire at the moment, I can try this again on AA and see if it still happens.
Nov 29 15:41:04 <karlp> but I didn't see perf when I built oprofile
Nov 29 15:41:25 <[florian]> perf needs the BROKEN flag to be enabled
Nov 29 15:42:05 <karlp> is it actually broken?
Nov 29 15:42:22 <[florian]> its build was not ok with all targets, but it works
Nov 29 15:42:42 <[florian]> it was marked broken more due to underlying build issues in its dependencies rather than it not being functionnal
Nov 29 15:44:21 <karlp> oh, it requires libelf to be manually selected first, instead of selecting libelf which it needs
Nov 29 15:44:49 <karlp> is there a motivation for package makefiles that do that vs just selecting libs that are required?
Nov 29 15:44:59 <[florian]> no, this was a "bug"
Nov 29 15:45:15 <[florian]> it is fixed in trunk/packages probably not in the backfire branch
Nov 29 15:45:44 <karlp> not in AA either.
Nov 29 15:46:27 <[florian]> it's marked broken anyway
Nov 29 15:46:30 <karlp> perf's still going to want the unstripped symbols though right?
Nov 29 15:46:36 <[florian]> of course
Nov 29 15:46:56 <karlp> right, well I'd better get that up again first.
Nov 29 17:37:47 <OpenWrt-Commits> juhosg r34405 trunk/target/linux/ (6 files in 3 dirs) * ramips: rt305x: add kernel support for the Asus RT-N13U Wireless-N300 Router
Nov 29 17:37:48 <OpenWrt-Commits> juhosg r34406 trunk/target/linux/ (6 files in 6 dirs) * ramips: rt305x: add user-space support for the Asus RT-N13U
Nov 29 17:37:50 <OpenWrt-Commits> juhosg r34407 trunk/target/linux/ ramips/rt305x/profiles/n13u.mk ramips/image/Makefile * ramips: rt305x: add profile and build image for the Asus RT-N13U
Nov 29 17:37:51 <OpenWrt-Commits> juhosg r34408 trunk/ include/image.mk Config.in * build: Add TARGET_ROOTFS_RESERVED_PCT option to reduce wasted space
Nov 29 17:37:52 <OpenWrt-Commits> juhosg r34409 trunk/target/linux/generic/config-3.6 * kernel: Add missing CONFIG_BATTERY_DS2781 option to generic 3.6 config
Nov 29 17:37:54 <OpenWrt-Commits> juhosg r34410 trunk/target/linux/ generic/patches-3.3/080-prot-release-cb.patch generic/patches-3.7/131-atm-fixes.patch generic/patches-3.6/131-atm-fixes.patch generic/patches-3.3/131-atm-fixes.patch * kernel: Add ATM fixes pending upstream merge (queue reduction, race fixes)
Nov 29 17:37:59 <OpenWrt-Commits> juhosg r34411 trunk/target/linux/ x86/geos/target.mk x86/config-3.6 x86/geos/config-3.6 * kernel: Update Geos to 3.6.8
Nov 29 17:39:27 <Chocks> busy
Nov 29 17:47:27 <karlp> perf has a --kallsyms= argument, what file does it want from my buildroot?
Nov 29 17:48:47 <karlp> --kallsyms=/tmp/remote/build_dir/linux-atheros/linux-3.3.8/vmlinux but it still says it's not ok...
Nov 29 17:50:34 <OpenWrt-Commits> juhosg r34412 branches/attitude_adjustment/ include/image.mk Config.in * AA: build: Add TARGET_ROOTFS_RESERVED_PCT option to reduce wasted space
Nov 29 17:50:39 <OpenWrt-Commits> juhosg r34413 branches/attitude_adjustment/target/ linux/generic/patches-3.3/080-prot-release-cb.patch linux/generic/patches-3.3/131-atm-fixes.patch * AA: kernel: Add ATM fixes pending upstream merge (queue reduction, race fixes)
Nov 29 17:50:44 <karlp> ah, might be that the kernel wasn't compiled with debug.
Nov 29 17:52:24 <karlp> oh, no, I was meant to be using -k, not --kallsyms. duh
Nov 29 17:54:18 <OpenWrt-Commits> juhosg r34414 trunk/target/linux/ar71xx/Makefile * ar71xx: switch to 3.6.8
Nov 29 17:55:12 <stintel> neat
Nov 29 17:59:04 <OpenWrt-Commits> juhosg r34415 trunk/target/linux/ generic/patches-3.6/065-8139cp-fixes.patch generic/patches-3.3/065-8139cp-fixes.patch generic/patches-3.7/065-8139cp-fixes.patch * kernel: Fix 8139cp ring buffer initialisation, tx timeout recovery, add BQL
Nov 29 18:03:06 <OpenWrt-Commits> juhosg r34416 branches/attitude_adjustment/target/linux/generic/patches-3.3/065-8139cp-fixes.patch * AA: kernel: Fix 8139cp ring buffer initialisation, tx timeout recovery, add BQL
Nov 29 18:15:08 <karlp> heh, "perf help" includs perf archive, but perf archive doesn't exist :)
Nov 29 19:13:56 <danitool> in bcm63xx, what do mean the registers MPI_CSBASE_SIZE_xxM?
Nov 29 19:14:15 <danitool> what is the purpose of them?
Nov 29 20:14:40 <OpenWrt-Commits> cyrus r34417 trunk/package/base-files/files/etc/sysctl.conf * Remove default sysctl-entry for IPv6 here (races)
Nov 29 20:14:45 <OpenWrt-Commits> cyrus r34418 trunk/package/network/config/netifd/files/etc/hotplug.d/iface/10-sysctl * Remove old IPv6 sysctl-config (deprecated)
Nov 29 20:14:49 <OpenWrt-Commits> cyrus r34419 trunk/package/network/ ipv6/6relayd/files/6relayd.init ipv6/6relayd/Makefile * 6relayd: Fix some RFC 6204 compatibility issues
Nov 29 20:14:54 <OpenWrt-Commits> cyrus r34420 trunk/package/network/ ipv6/6distributed/Makefile ipv6/6distributed ipv6/6distributed/files/6distributed.init ipv6/6distributed/files * 6distributed: add local IPv6 prefix distributor
Nov 29 20:14:58 <OpenWrt-Commits> cyrus r34421 trunk/package/network/ ipv6/odhcp6c ipv6/odhcp6c/Makefile * odhcp6c: Add new embedded DHCPv6 client
Nov 29 20:15:04 <OpenWrt-Commits> cyrus r34422 trunk/package/network/ (7 files in 3 dirs) * ipv6-support: Add new IPv6-support meta-package
Nov 29 20:33:53 <OpenWrt-Commits> cyrus r34423 trunk/package/network/ipv6/ipv6-support/Makefile * ipv6-support: Add conffiles section
Nov 29 21:43:29 <OpenWrt-Commits> florian r34424 trunk/target/linux/ ar7/config-3.3 ar7/config-3.6 * [ar7] use a default kernel command-line to mount squashfs and jffs2
Nov 29 21:43:34 <OpenWrt-Commits> florian r34425 trunk/toolchain/gcc/patches/4.5-linaro/200-musl.patch * [toolchain] gcc/4.5-linaro: backport musl patch
Nov 29 21:43:37 <OpenWrt-Commits> florian r34426 trunk/target/linux/ malta/patches-3.3 malta/config-3.3 * [malta] remove Linux 3.3 support
Nov 29 21:43:41 <OpenWrt-Commits> florian r34427 trunk/Config.in * [buildroot] disable sstrip when using musl
Nov 30 01:45:10 <RealOpty> Why does collectd depend on rrdtool version 1.0 ? :)
**** ENDING LOGGING AT Fri Nov 30 02:59:58 2012