**** BEGIN LOGGING AT Tue Dec 04 02:59:59 2012
Dec 04 12:07:20 <OpenWrt-Commits> nbd r34470 trunk/package/ network/services/hostapd/files/hostapd.sh mac80211/files/lib/wifi/mac80211.sh * mac80211, hostapd: Fix macfilter for multi bssid setups
Dec 04 15:12:23 <fullstop> I hope to get a serial cable connected today, but the WZR-300HP looks nearly identical to the G300NH(2) models.
Dec 04 15:13:07 <fullstop> DDR chip looks physically different and some capacitors have been moved, but the layout, SoC and nand flash look to be the same.
Dec 04 15:13:27 <fullstop> same wmac too
Dec 04 15:20:30 <OpenWrt-Commits> florian r34471 trunk/target/linux/ brcm63xx/patches-3.6/532-board_96328a-1241N.patch brcm63xx/patches-3.3/532-board_96328a-1241N.patch * [brcm63xx ] Fix Comtrend AR-5381u Power LED
Dec 04 15:25:00 <OpenWrt-Commits> jow r34472 trunk/package/network/ config/firewall/files/reflection.hotplug config/firewall/Makefile * firewall: extend nat reflection support
Dec 04 16:19:48 <OpenWrt-Commits> florian r34473 trunk/include/image.mk * include/image.mk: call Image/Build/{cpiogz,targz}
Dec 04 16:19:51 <OpenWrt-Commits> florian r34474 trunk/target/linux/cns3xxx/Makefile * cns3xxx: advertise pcie usb usbgadget features
Dec 04 16:19:57 <OpenWrt-Commits> florian r34475 trunk/target/linux/ cns3xxx/files/drivers/usb/dwc/otg_pcd.c cns3xxx/patches-3.3/200-dwc_otg_support.patch * cns3xxx: fix dwc_otg driver compat with udc-core
Dec 04 16:20:00 <OpenWrt-Commits> florian r34476 trunk/package/linux-atm/Makefile * linux-atm: factor common descriptions
Dec 04 16:20:03 <OpenWrt-Commits> florian r34477 trunk/package/linux-atm/Makefile * linux-atm: create atm-debug-tools package
Dec 04 16:21:18 <OpenWrt-Commits> florian r34478 packages/libs/libdaq/Makefile * libdaq: dependency fix
Dec 04 16:21:19 <OpenWrt-Commits> florian r34479 packages/libs/libmodbus/Makefile * libmodbus: update to 3.0.3
Dec 04 16:21:20 <OpenWrt-Commits> florian r34480 packages/utils/dialog/Makefile * dialog: update to version 20120706
Dec 04 16:21:21 <OpenWrt-Commits> florian r34481 packages/utils/procps/Makefile * procps: Split package into subpackages
Dec 04 16:21:22 <OpenWrt-Commits> florian r34482 packages/lang/python-crypto/Makefile * python-crypto: Bump to 2.6
Dec 04 16:21:23 <OpenWrt-Commits> florian r34483 packages/libs/gmp/Makefile * gmp: update to 5.0.5
Dec 04 16:21:26 <OpenWrt-Commits> florian r34484 packages/net/radsecproxy/Makefile * radsecproxy: bump to 1.6.1, security bug fix
Dec 04 16:21:31 <OpenWrt-Commits> florian r34485 packages/net/isc-dhcp/ (6 files in 4 dirs)
Dec 04 16:21:32 <OpenWrt-Commits> isc-dhcp-relay: Add init scripts for ipv4 and ipv6 as well as config
Dec 04 16:21:35 <OpenWrt-Commits> file
Dec 04 16:21:37 <OpenWrt-Commits> I cleaned up the patch so line do not overflow and removed some white
Dec 04 16:21:40 <OpenWrt-Commits> space in hope to get it accepted or revised.
Dec 04 16:21:42 <OpenWrt-Commits> As isc-dhcp-relay does not come with an init script, I made one based
Dec 04 16:21:45 <OpenWrt-Commits> on sysntpd init.d file for both ipv4 and ipv6.
Dec 04 16:21:47 <OpenWrt-Commits> If you need more arguments from isc-dhcp-relay to be included in the
Dec 04 16:21:51 <OpenWrt-Commits> config file, I'll be able to add them.
Dec 04 16:21:53 <OpenWrt-Commits> This patch has been applied to trunk and compiles/install on my side.
Dec 04 16:21:56 <OpenWrt-Commits> The script for ipv4 and ipv6 have been tested with the
Dec 04 16:21:58 <OpenWrt-Commits> isc-dhcp-relay-ipv6 binary and work well (start/stop with the right
Dec 04 16:22:01 <OpenWrt-Commits> settings).
Dec 04 16:22:04 <OpenWrt-Commits> Signed-off-by: Jérôme Poulin <jeromepoulin@gmail.com>
Dec 04 16:22:06 <OpenWrt-Commits> Signed-off-by: Florian Fainelli <florian@openwrt.org>
Dec 04 16:22:08 <OpenWrt-Commits> florian r34486 packages/multimedia/ (9 files in 4 dirs) * multimedia/tvheadend: new package
Dec 04 16:22:12 <OpenWrt-Commits> florian r34487 packages/utils/collectd/ (6 files in 2 dirs) * collectd: 4.10.8 version bump
Dec 04 16:22:15 <OpenWrt-Commits> florian r34488 packages/libs/libffi/Makefile * libffi: bump version to 3.0.11
Dec 04 16:22:18 <OpenWrt-Commits> florian r34489 packages/libs/libffi/Makefile * libffi: enable the host build, needed for newer glib
Dec 04 16:22:21 <OpenWrt-Commits> florian r34490 packages/libs/ glib2/patches glib2/Makefile * glib2: bump version to 2.34.3
Dec 04 16:22:24 <OpenWrt-Commits> florian r34491 packages/admin/ zabbix/files/zabbix_agentd.conf zabbix/Makefile * zabbix: preserve conffile over sysupgrade
Dec 04 16:22:27 <OpenWrt-Commits> florian r34492 packages/admin/ zabbix/patches/010-change-agentd-config.patch zabbix/files/zabbix_agentd.init * zabbix: correct pidfile und update initscript
Dec 04 16:22:31 <OpenWrt-Commits> florian r34493 packages/net/autossh/files/autossh.init * autossh: pass env variables on init properly
Dec 04 16:22:34 <OpenWrt-Commits> florian r34494 packages/net/tinyproxy/files/tinyproxy.config * tinyproxy: fix default config binary flags
Dec 04 16:22:37 <OpenWrt-Commits> florian r34495 packages/net/nginx/ patches/101-feature_test_fix.patch Makefile * nginx: upgrade to 1.2.3
Dec 04 16:22:42 <Chocks> woo
Dec 04 16:22:49 <Chocks> lots of commits
Dec 04 16:28:14 <dwmw2_gone> [florian]: any thoughts on the patch I posted to build a single ext4 image instead of two partitions?
Dec 04 16:28:33 <[florian]> dwmw2_gone: not yet, let me look at it
Dec 04 16:28:37 <nbd> dwmw2_gone: was it verified with sysupgrade?
Dec 04 16:28:47 <dwmw2_gone> does sysupgrade even work for ext4?
Dec 04 16:28:55 <dwmw2_gone> (that's a no, in case you hadn't guessed)
Dec 04 16:29:05 <nbd> i thought somebody had done some work on that at some point
Dec 04 16:29:13 <dwmw2_gone> last I knew, sysupgrade for ext4 was still unimplemented and there was a bounty for it
Dec 04 16:29:20 <nbd> i don't know much about the current state of x86 outside of the specific things that i worked on
Dec 04 16:29:30 <nbd> but i heard at some point that somebody made a proper sysupgrade fix
Dec 04 16:29:33 <dwmw2_gone> makes me grumpy to call it x86 :)
Dec 04 16:29:43 <dwmw2_gone> block-device support isn't just for x86; some other platforms are that stupid too :)
Dec 04 16:29:50 <dwmw2_gone> I'll take a look
Dec 04 16:30:52 <dwmw2_gone> looks like it might have been implemented. I'll play with it
Dec 04 16:32:18 <dwmw2_gone> hm, sysupgrade uses the separate boot partition for storing config during upgrade
Dec 04 16:32:34 <dwmw2_gone> that's... interesting.
Dec 04 16:35:35 * dwmw2_gone tries to work out what flash-based platforms do
Dec 04 16:35:44 <dwmw2_gone>                 mtd jffs2write "$CONF_TAR" rootfs
Dec 04 16:40:22 <Chocks> moo
Dec 04 16:40:54 <Chocks> yes. and then lots of "changing underpants whilst running" tricks to be in a position so that can work
Dec 04 16:40:57 <OpenWrt-Commits> florian r34496 trunk/package/network/ (5 files in 2 dirs) * dropbear: update to 2012.55 and refresh patches
Dec 04 16:45:08 <OpenWrt-Commits> florian r34497 packages/net/nginx/Makefile * nginx: update to 1.2.5.
Dec 04 16:46:09 <dwmw2_gone> they write the *tarball* raw image, to the mtd partition which will contain the rootfs?
Dec 04 16:46:18 <dwmw2_gone> and where do they put the rootfs? :)
Dec 04 16:47:02 <Chocks> it depends upon the platform.  But that's not how I've seen it done.  there is a tarball, but that just contains changed stuff in /etc usually
Dec 04 16:47:43 <dwmw2_gone> yeah. And then a platform-specific function is called to store that tarball
Dec 04 16:47:53 <nbd> dwmw2_gone: jffs2write generates jffs2 data containing the tarball, starting at the block where it finds the end-of-filesystem marker
Dec 04 16:48:10 <nbd> dwmw2_gone: works with squashfs and jffs2
Dec 04 16:48:14 <dwmw2_gone> nbd: ah, right. So it writes it *after* the original rootfs has been stored?
Dec 04 16:48:21 <nbd> yes
Dec 04 16:48:27 <dwmw2_gone> basically it's the same as mounting it and extracting the tarball?
Dec 04 16:48:41 <nbd> not extracting it
Dec 04 16:48:47 <nbd> it writes the tarball
Dec 04 16:48:52 <nbd> after the reboot it will be extracted
Dec 04 16:48:56 <dwmw2_gone> ok, But in the rootfs.
Dec 04 16:49:26 <dwmw2_gone> ah, great. So I could just change the block-device sysupgrade to do the same thing? Put it in the rootfs, not the boot partition
Dec 04 16:50:03 <nbd> that's fragile
Dec 04 16:50:12 <dwmw2_gone> but it's what 99% of platforms do...
Dec 04 16:50:36 <nbd> with mtd it's safe to erase and rewrite a few blocks
Dec 04 16:50:43 <nbd> with block devices it isn't
Dec 04 16:50:59 <nbd> because of caching issues on a mounted fs
Dec 04 16:51:14 <nbd> that's the reason why the boot partition was used
Dec 04 16:51:27 <nbd> remounting the rootfs after rewriting the image didn't work
Dec 04 16:51:45 <dwmw2_gone> because we hadn't properly unmounted it and weren't running from a ramfs?
Dec 04 16:52:10 <Chocks> on one of my platforms, I use a whole other small partition for the config tarball
Dec 04 16:52:14 <nbd> some processes cannot be easily restarted
Dec 04 16:52:16 <dwmw2_gone> if we had issues like that, it sounds like things were broken anyway, and we just got lucky
Dec 04 16:52:25 <nbd> with /boot it's reliable
Dec 04 16:52:28 <nbd> because it doesn't have to be mounted
Dec 04 16:52:29 <dwmw2_gone> I thought we kill everything for a sysupgrade and reboot?
Dec 04 16:52:44 <Chocks> almost everything.  it's staged.  The process is quite complex
Dec 04 16:53:03 <dwmw2_gone> a separate small partition for the config tarball may make sense
Dec 04 16:53:19 <dwmw2_gone> at least it's not as wasteful as a separate ext4 filesystem just for kernel+grub.conf
Dec 04 16:53:36 <Chocks> I've written sysupgrade style stuff for x86, which does stuff in quite a different way. You still have to be exceptionally careful about which filesystems you handle when and which processes you kill in which order
Dec 04 16:53:45 <dwmw2_gone> or perhaps add 'incremental write' support to genext2fs or something like that.
Dec 04 16:54:03 <dwmw2_gone> which could add the file in a similar way.
Dec 04 16:56:39 <dwmw2_gone> I take it sysupgrade doesn't attempt to upgrade from grub to grub2 :)
Dec 04 16:57:30 <dwmw2_gone> actually it doesn't seem to upgrade the kernel at all.
Dec 04 16:57:35 <dwmw2_gone> I must be missing something
Dec 04 17:32:25 <DonkeyHotei> dwmw2_gone: i lost track: which part did you say you'll do wrt the atm stuff?
Dec 04 17:33:05 <DonkeyHotei> Delboy: i can't seem to get a serial console on the sx762; what did you do?
Dec 04 19:16:18 <fullstop> assuming the serial pinout is the same as the G300NH, u-boot output is crippled on the WZR-300HP
Dec 04 19:16:39 <fullstop> accept_open_rt_fmt=1, so that is good
Dec 04 19:16:53 <fullstop> There's also this variable: tmp_uart=1
Dec 04 19:18:07 <Chocks> moo
Dec 04 20:26:59 <Delboy> DonkeyHotei: do you get bootlog on serial?
Dec 04 20:27:12 <DonkeyHotei> no, nothing
Dec 04 20:32:13 <Delboy> DonkeyHotei: hmm, maybe you swiched tx and rx
Dec 04 20:41:19 <DonkeyHotei> Delboy: tried that
Dec 04 22:59:47 <OpenWrt-Commits> florian r34498 trunk/include/image.mk * Revert "include/image.mk: call Image/Build/{cpiogz,targz}"
**** ENDING LOGGING AT Wed Dec 05 02:59:58 2012