**** BEGIN LOGGING AT Wed Mar 23 02:59:57 2011 Mar 23 03:28:58 build #107 of s3c24xx is complete: Failure [failed compile_10] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/107 Mar 23 05:26:26 dumb question... when /etc/uci-default/ gets processed, how far along is the system? have all of the modules been modprobe'd? Mar 23 05:26:52 philpp64: not a dumb questions, and no Mar 23 05:27:16 philipp64: IIRC it happens in /etc/init.d/boot Mar 23 05:27:34 philipp64: i.e. the first init script Mar 23 05:28:23 philipp64: (which is after preinit) Mar 23 05:34:46 hmmm... load_modules gets run WAY after uci_apply_defaults. sigh. Mar 23 05:35:03 I need to get the MAC address of eth0 and clone it onto the DSL bridge. Mar 23 05:35:50 philipp64: how about a hotplug script? Mar 23 05:36:47 but it's platform specific, not device specific. Mar 23 05:37:34 philipp64: target/linux//base-files/etc/hotplug.d/blah Mar 23 05:37:35 ? Mar 23 05:38:42 hmmm... setting a MAC address on the DSL device only makes sense if it's been configured for PPPoE or bridged RFC-2684. Mar 23 05:38:57 for routed RFC-2684 or for PPPoA it's not used. Mar 23 05:40:12 philipp64: that sounds like you need a platform network script Mar 23 05:42:14 philipp64: is this something that can even be reasonably automated? That is, under what circumstances is this needed? Mar 23 05:42:54 philipp64: and what's different from every other DSL bridge device? Mar 23 05:45:13 well... the default mac address that br2684ctl configures is a bogus one. Mar 23 05:45:31 some ISP's might accept it... mine won't. that's how I discovered this. Mar 23 05:45:48 philipp64: ah, that sounds like a general network problem rather than platform Mar 23 05:46:43 philipp64: i.e. the fix is in the network scripts that do br2684ctl, not a platform-specific fix Mar 23 05:46:58 right, but cloning the MAC address.... where do you clone it from? not every device has a eth0. Mar 23 05:47:55 philipp64: hmmm......probably should have a default that can be overriden....maybe use the lan interface (i.e. uci lan turned into the lan interface e.g. eth0/eth1 Mar 23 05:47:56 err... not every platform, rather. Mar 23 05:48:33 hmmm... on my geos, the lan interface is a br-lan which is a bridge of eth0, eth1, and wlan0. Mar 23 05:49:00 but bridges can also be queried for a MAC address, just like a regular Ethernet device. Mar 23 05:49:06 philipp64: well the mac could come from br-lan though couldn't it? Mar 23 05:49:08 yeah Mar 23 05:50:44 hmmm... I'll put it on my todo list, but it's not a high priority. Mar 23 05:50:57 still need to write the leds-geos.c module... Mar 23 05:51:17 and get netdev to accept 4 solos-pci patches I've submitted. Mar 23 05:51:45 if only I didn't have a day job... :-) Mar 23 05:51:58 heh, tell me about it Mar 23 05:52:00 or this was my day job. Mar 23 05:52:51 I've been telling Intel they need to support various F/OSS projects like Openwrt better, but... Mar 23 05:57:24 cshore: since I'm picking your brain... Mar 23 05:57:33 yes? Mar 23 05:57:57 one of the nice things about openwrt is that it reboots quickly... quickly enough that it can do so before a TCP connection times out. Mar 23 05:58:46 but... my ISP will release my DHCP lease if my carrier drops... but they put the address at the end of the unallocated list (round-robin). Mar 23 05:58:46 at least if you don't have major firewall rules....but that will be better once the firewall is in C Mar 23 05:59:01 if I ask for it quickly enough when I come back up, they will give it back to me. Mar 23 05:59:21 problem is... we currently don't cache client leases. Mar 23 06:00:17 That would have to be a site-local change because we try to avoid writing to the flash on a regular basis Mar 23 06:00:27 such as on every lease event Mar 23 06:00:37 I wonder what would be required to have udhcpc save out leases... and reload them on startup. Mar 23 06:00:54 leases are good for hours, though... Mar 23 06:00:57 it's a rare event. Mar 23 06:01:33 philipp64: probably just change the path from /var/dhcp.leases to something persistant Mar 23 06:02:16 philipp64: I'm not the one to argue with...I personally thing wear-levelling built in ought to take care of any realistic concerns about wearing out the flash Mar 23 06:02:30 that's for leases for which we are the server... Mar 23 06:02:35 for infrequent events Mar 23 06:02:46 I'm talking about DSL, where we're the client of the ISP's DHCP server. Mar 23 06:02:49 philipp64: oh right, sorry Mar 23 06:03:16 easy mistake. Mar 23 06:04:18 philipp64: I don't think it is possible with udhcpc at least in busybox Mar 23 06:04:50 philipp64: you'd have to use a one of the clients in the network menu Mar 23 06:04:53 in menuconfig Mar 23 06:06:37 yeah... I understand why we use busybox, but... this is the sort of tradeoff one has to constantly live with. Mar 23 06:11:51 philipp64: well it's the embedded space that does it...we're working on reducing bloat, and trying to get smaller not bigger. Openwrt is already pretty big for the kind of devices we want to be able to run on as are main target (though of course other uses involves targets with more resources and wanting to have more features) Mar 23 06:15:51 one long term goal is to replace a lot of script-based stuff with C and eventually have the possibility of running without ash (and have command console but it wouldn't a full shell). Mar 23 06:16:43 of course if you want ash you could have it Mar 23 06:28:35 I'm running on an 500MHz Geode, with 512MB of DDR, and 4GB of Flash. Mar 23 06:29:24 I could see myself running Asterisk on it... and terminating several IPsec tunnels as well. Mar 23 06:30:03 philipp64: yeah, the problem of having totally different use cases for the same source tree Mar 23 06:30:18 yup. Mar 23 06:30:44 philipp64: that's why we try not make choice possible (like doing ash if you want) Mar 23 06:30:47 the box is 15cm x 15cm x 4cm. Mar 23 06:31:00 phillip64: sweet Mar 23 06:31:09 phillip64: how much does it cost? Mar 23 06:31:13 miniPCI, but no PCI. Mar 23 06:31:39 phillipp64: usually tiny embedded capability is a cost choice rather size choice Mar 23 06:31:41 it's about $420 USD with CF, case, power supply, and memory. Mar 23 06:32:05 Wifi, pigtails, and antennae are another.... call it $60. Mar 23 06:32:05 philipp64: ah, ok, fairly expensive then Mar 23 06:32:23 well, I'm getting it at full retail price, single quantity. Mar 23 06:32:39 philipp64: as opposed to a $40 CAD router Mar 23 06:32:50 retail price Mar 23 06:33:08 actually it's about the same USD now Mar 23 06:33:08 yeah, but will it double as an IP-PBX and VPN concentrator? Mar 23 06:33:45 philipp64: that's not the point I was making....I was making the point our primary target if $40-120 USD routers Mar 23 06:33:57 no, I get that. Mar 23 06:34:21 I'm just wondering if there's a curve that users get onto where they want the box to do more and more over time. Mar 23 06:34:22 philipp64: it sounds like a sweet box Mar 23 06:34:36 philipp64: probably Mar 23 06:34:51 philipp64: same as all tech really Mar 23 06:35:11 phillipp64: plus business demands that you have something new for the marketplace fairly often Mar 23 06:35:21 true enough. Mar 23 06:37:42 philipp64: and cost is always a factor....and there's a lot companies were doing iP-PBX (or up to 24 phones was the claim) in brcm96xxx with 32MB RAM and 200 MHz processors Mar 23 06:38:20 realistically they were aimed at average 2-10 users Mar 23 06:39:34 not sure what the retail price on the box was though Mar 23 06:40:30 considering it was three years ago, probably the same as the Geode now Mar 23 06:40:48 maybe five Mar 23 06:41:23 that type of product is EOL now Mar 23 06:44:54 yup... so $400-500 isn't that much to pay if you get a fair amount of flexibility, functionality, and 2-3 years (or more!) of shelf-life. Mar 23 06:45:30 philipp64: if you need the functionality yeah Mar 23 06:45:55 wish my DSL loop wasn't so crappy. sigh. Mar 23 06:46:11 1800k/450k... sigh. Mar 23 06:46:39 philipp64: heh, I have good signal, but I complain about 6000k/800k Mar 23 06:46:53 sucks Mar 23 06:47:05 yours I mean Mar 23 06:47:31 mine only sucks in comparison to what Bell doesn't let wholesalers have Mar 23 06:47:49 (20000/2000 Mar 23 06:47:58 yup... by the way, if I've just built a new image, what's the best way to blast it into flash on a running system? Mar 23 06:48:06 on x86? Mar 23 06:48:10 not sure Mar 23 06:48:12 yeah. Mar 23 06:48:34 I'm not sure what type of rootfs x86 sysupgrade accepts Mar 23 06:48:47 I discovered dd is bad on x86 Mar 23 06:48:53 on a running system Mar 23 06:49:05 I tried doing a straight dd of the squashfs.img onto /dev/sda, but that occasionally panics the box because the jiffs gets corrupted in the process. Mar 23 06:49:12 yeah Mar 23 06:49:45 maybe JOW will get uppity and write a few new chapters for the Wiki. :-) Mar 23 06:50:32 heh, I wish I had more time to contribute to the wiki Mar 23 06:50:45 not that know as much as jow Mar 23 06:50:50 but he can't do everything Mar 23 06:51:00 though he sure puts a lot in Mar 23 06:51:26 yeah... pretty proficient. Mar 23 06:55:28 and works hard on openwrt Mar 23 06:56:47 phillipp64: know anything about zaptel? Mar 23 06:58:12 I know that I hate maintaining it... Mar 23 06:58:19 philipp64: or maybe telephony in general....my analog handset notices dtmfs for numbers other than 9, 8, 6, Mar 23 06:58:22 well, getting it to cross-compile on Astlinux. Mar 23 06:58:36 philipp64: I did some zaptel stuff recently for openwrt Mar 23 06:58:56 philipp64: haven't done dahdi yet Mar 23 06:59:26 last time I banged on dadhi was 2.2.1 Mar 23 06:59:31 philipp64: but the driver itself was already there, I just added some support stuff Mar 23 07:00:21 back up... you mean your analog card will recognize digits other than 6,8,9 but not those? Mar 23 07:00:26 right Mar 23 07:00:41 sounds like you have the wrong country code set. Mar 23 07:00:48 that's what I was wondering Mar 23 07:01:30 but my expertise was ISDN and SIP... a little SS7... and that was 5-10 years ago... Mar 23 07:01:36 I could be way off base. Mar 23 07:03:04 I suspect the problem is the country code in my telephony app - it defaults to itu, so I probably need to find and change a setting for country Mar 23 07:03:16 it was written primarily by russians Mar 23 07:03:31 (Yate if you've heard of it) Mar 23 07:07:02 no, doesn't ring any bells. Mar 23 07:07:16 philipp64: what country codes does zaptel accept? Mar 23 07:07:40 most of the major ones as I remember. Mar 23 07:07:48 maybe not Togo. Mar 23 07:09:08 speaking togo, I have to go... bed time. Mar 23 07:09:15 well, bed time was an hour ago... Mar 23 07:09:26 night Mar 23 07:09:36 I'd try #asterisk for help with that issue... some competent people there. Mar 23 07:10:04 ok, yawn... night! Mar 23 07:10:17 * philipp64|laptop stumbles off to meet the sandman Mar 23 08:26:25 hcg * r26274 /trunk/target/linux/omap35xx/patches-2.6.36/004-nand_subpage_align.patch: [omap35xx] Add patch to correct sub-page alignment Mar 23 08:40:49 gmorning Mar 23 08:41:11 good morning Mar 23 10:45:46 ping KanjiMonster Mar 23 10:50:25 Good morning Mar 23 10:50:46 Could someone please give any feedback on this one http://www.mail-archive.com/openwrt-devel@lists.openwrt.org/msg07794.html Mar 23 10:52:53 build #102 of cobalt is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/cobalt/builds/102 Mar 23 10:53:19 lauri: it sounds good, but could you make a separate update patch? Mar 23 10:53:59 you mean diff for each package? Mar 23 10:55:44 these diffs should be against the packages feed svn? Mar 23 10:55:57 yes Mar 23 10:56:28 good point :) Mar 23 11:35:10 hcg * r26275 /trunk/target/linux/omap35xx/patches-2.6.36/005-add_cti_usbids.patch: [omap35xx] kernel: add CTI usb-id's to FTDI driver Mar 23 11:35:52 xMff: pong (for some minutes ;) Mar 23 11:36:06 KanjiMonster: you did the znc config stuff? Mar 23 11:36:15 yes Mar 23 11:36:45 someone whined that it does not use the original config, maybe we should add an [ -f /etc/znc.conf ] || { do uci stuff } Mar 23 11:37:35 xMff: yeah, already thought about that (or having a uci option external_config) Mar 23 11:38:42 thought it sounds like a luxury problem to me, it isn't that hard to add znc -c /etc/znc.conf to rc.local Mar 23 11:39:44 xMff: just doing -c can be a security problem though, since znc drops from root through znc module, and if the user config doesn't include it then ... ugh.. ah, znc should abort, so it isn't one Mar 23 11:40:07 KanjiMonster: wouln't worry about that Mar 23 11:40:20 deviate from the path of openwrt and you're on your own :P Mar 23 11:40:47 no seriously, I think checking for an /etc/znc.conf and then not shipping one by default should be enough Mar 23 11:41:06 O.o Mar 23 11:41:16 xMff: znc refuses to start as root as long as you don't explicitly allow running as root Mar 23 11:41:38 but yeah, its the user's problem then ;) Mar 23 11:42:01 KanjiMonster: yeah but why should we care? someone whines that he does not want uci so I expect him to be pro enough to sort out that stuff on his own Mar 23 11:43:52 uci is ugly Mar 23 11:43:54 ;) Mar 23 11:44:13 xml-style config for an irc bouncer is too Mar 23 11:44:18 i do plan to remove that from my system eventually hehe Mar 23 11:44:23 have fun Mar 23 11:44:45 irc bouncers are ugly Mar 23 11:44:47 :) Mar 23 11:44:53 but so is xml Mar 23 11:45:40 xMff: when just using an external config the stopping of znc will most likely blow unless the user specifies in his config that znc should write its pid file where the init script expects it Mar 23 11:46:25 KanjiMonster: [ -f /etc/znc.conf ] && killall znc || { do sophisticated kill } Mar 23 11:48:04 xMff: [ -f /etc/znc/configs/znc.conf ], but apart from that it looks okay ;) Mar 23 11:50:08 yeah, I killed that from the init.d script Mar 23 11:50:41 giving ZNC an OpenWRT is a terrible idea unless you make something to translate the working script back Mar 23 11:50:47 *an OpenWRT config Mar 23 11:52:24 so don't use it Mar 23 11:52:48 and start it yourself Mar 23 11:53:24 bbl Mar 23 11:53:33 znc has a drop-root plugin Mar 23 11:54:21 yeah, that I wondered too, wtf should a daemon that dropped root be able to alter a system wide config Mar 23 11:54:31 that is what sounds like a terrible idea to me Mar 23 11:54:41 * ermo reads the back log to understand the context Mar 23 11:55:05 but I don't actually use it so what do I know. I'm really gone for now Mar 23 11:55:19 but hey, I use znc and sudo, so I'm not exactly in the 'served-by-the-vanilla-OpenWrt-config' dept Mar 23 11:55:49 in any case, a 'works ootb znc' configuration would be nice. Mar 23 12:00:18 ermo: I'm using the drop-root plugin to avoid a dependency to su or sudo (it's always added to the generated config) Mar 23 12:02:15 okay, gotta work, bbl Mar 23 12:30:22 build #87 of uml is complete: Failure [failed compile_10] Build details are at http://tksite.gotdns.org:8010/builders/uml/builds/87 Mar 23 12:36:02 jow_laptop: About two days ago, you pastebin an instruction to me on how to truncate '$(STAGING_DIR)" from "$(STAGING_DIR)/usr/lib/python2.6" using the "echo ..." method. Apparently, the link (http://paste.openwrt.org/d3cb5ecda) is now empty. So, can you please re-upload the instruction again? Mar 23 12:51:10 jow_laptop: so, what is it that determines the maximum number of connection the router can handle Mar 23 12:52:03 since I notice it shot up to 0x4000 recently Mar 23 14:11:57 hcg * r26276 /trunk/target/linux/omap35xx/patches-2.6.36/005-add_cti_usbids.patch: [omap35xx]: Correct CTI patch Mar 23 14:31:10 build #102 of orion is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/orion/builds/102 Mar 23 14:37:16 mb__: ping Mar 23 14:42:39 KanjiMonster: pong Mar 23 14:42:45 http://www.youtube.com/watch?v=5sakN2hSVxA Mar 23 14:44:11 mb__: I wondered if you intend to push the b43-fwcutter update for openwrt's brcm-wl Mar 23 14:45:14 hm? Mar 23 14:45:34 posted on b43dev some time ago Mar 23 14:46:05 ah Mar 23 14:47:10 and yeah, that's quite an interesting choice of symbolism in the video Mar 23 15:00:43 I'll look at that fwcutter thing Mar 23 15:04:00 mb__: I can verify it applies and extracts stuff; I haven't been able to get neither a bcm4321 or a 4322 to work though ;) (I was probably missing some ssb updates) Mar 23 15:35:00 xMff: ping Mar 23 15:36:53 KanjiMonster: Released fwcutter-014 Mar 23 15:37:29 mb__: cool, thanks Mar 23 15:37:56 then I don't have to keep a local pseudo release anymore ;) Mar 23 15:41:07 build #83 of ixp4xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/ixp4xx/builds/83 Mar 23 15:56:46 build #87 of ar7 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ar7/builds/87 Mar 23 16:17:41 xMff: unping Mar 23 16:31:52 build #88 of x86 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/x86/builds/88 Mar 23 16:45:44 build #80 of etrax is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/etrax/builds/80 Mar 23 17:08:47 build #83 of kirkwood is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/kirkwood/builds/83 Mar 23 17:53:25 jow * r26277 /packages/lang/perl/Makefile: [packages] perl: make sure that staging_dir/*/usr/lib/perl5 is in host perl's @INC (#9111) Mar 23 20:05:27 acoul * r26278 /packages/net/iputils/Makefile: net/iputils: fix ping & arping under brcm47xx. Mar 23 20:09:15 acoul * r26279 /packages/net/iputils/Makefile: net/iputils: include a missing option (thanks Hauke) Mar 23 20:20:18 nbd * r26280 /trunk/package/mac80211/patches/ (2 files): ath9k: merge a pending patch for fixing a stopped queue issue (mostly for client mode) Mar 23 20:20:23 nbd * r26281 /trunk/package/mac80211/patches/ (10 files): ath9k: add a few de-bloating and optimization patches Mar 24 01:05:52 jow_laptop: ping Mar 24 02:12:50 build #87 of avr32 is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/avr32/builds/87 **** ENDING LOGGING AT Thu Mar 24 02:59:58 2011