**** BEGIN LOGGING AT Tue Jan 05 02:59:56 2010 **** ENDING LOGGING AT Tue Jan 05 05:57:21 2010 **** BEGIN LOGGING AT Tue Jan 05 05:58:14 2010 Jan 05 07:50:59 how does one recompile only a specific a package (if one has built from source) Jan 05 07:51:28 without going through all the packages in the build process Jan 05 07:51:59 (i.e. more specific than make package/x/clean ; make ) Jan 05 07:55:27 I've been doing make package/x/clean && make package/x/compile && make package/x/install && make package/rootfs-prepare && make target/install && make package/index Jan 05 07:55:27 but that's rather a pain...is there an easier way? Jan 05 08:21:08 <[florian]> cshore: recompiling a specific package is done by make package/foo/{clean,compile} Jan 05 08:28:59 [florian], yes but how do you make the image after that, without going through the whole make process (i.e. running through every package) Jan 05 08:30:08 (e.g. if the package is in the firmware image) Jan 05 08:30:42 or like base-files is essential to the image Jan 05 08:32:06 <[florian]> cshore: make target/linux/install should do it Jan 05 08:33:33 ok so I can just do make package/x/compile && make package/x/install && make target/linux/install, and the firmware should be in bin? Jan 05 08:33:40 <[florian]> yes Jan 05 08:33:51 <[florian]> cshore: happy new year by the way ;) ! Jan 05 08:34:07 ok...I was copying from the output of make...I didn't know what was actually essential Jan 05 08:34:21 happy new year :) Jan 05 08:34:35 have you been busy with the new job? Jan 05 08:34:47 <[florian]> yes, quite busy Jan 05 08:34:56 <[florian]> and it's probably not going to calm down that much Jan 05 08:39:07 has there been any news on the ADSL front for 63xx? Jan 05 08:40:03 <[florian]> no news, but it's still in a discussion state with bcm Jan 05 08:43:04 is it possible to dynamically add a host dns record to dnsmasq (i.e. without restart)? Jan 05 08:44:05 <[florian]> I think it polls regularly for updates of /etc/hosts (like 1 second) Jan 05 08:44:13 <[florian]> so there is no need to restart dnsmasq Jan 05 08:44:24 <[florian]> we may however have disabled that in the openwrt config Jan 05 08:44:27 <[florian]> but you can re-enable it back Jan 05 08:44:48 <[florian]> I do not think dnsmasq supports /etc/hosts changes notifications Jan 05 08:44:53 or to add a host record that isn't in /etc/hosts ? (e.g. if you want the lan to see a different address for a dyndns record) Jan 05 08:45:03 ok Jan 05 08:45:29 <[florian]> if it is not in /etc/hosts, you might need to hack a bit the dnsmasq sources I'd say Jan 05 08:46:45 that's what I thought....my problem is that my webserver is accessed via dyndns, which resolves to the router, but from the lan I want it to resolve to the local server (because the lan side of the router is used for LuCI so I can't redirect it) Jan 05 08:53:10 someone had a patch for the -A parameter but it doesn't seem to have been applied yet Jan 05 08:53:25 <[florian]> cshore: can you point me to the ticket? Jan 05 09:12:04 ticket #5017 Jan 05 09:26:37 <[florian]> cshore: thanks Jan 05 09:27:16 <[florian]> cshore did you try this patch? Jan 05 09:31:48 http://wiki.openwrt.org/doc/howto/dmz creates a DMZ interface of type bridge Jan 05 09:31:58 I think this is not needed Jan 05 09:32:20 bridge only applies to the regular LAN to bridge with wifi, or am I wrong? Jan 05 09:44:49 yes, the bridge is not strictly required Jan 05 09:53:55 has anybody a working ar7 with newer kernels (2.6.30 or later)? Jan 05 09:56:08 I have trouble with ar7 networking (cpmac with fixed phy) Jan 05 09:56:54 <[florian]> hi sw Jan 05 09:57:14 <[florian]> I have it working on my wag54g v2 with cpmac.external_switch=1 Jan 05 09:58:18 <[florian]> sw: this whole problem with fixed phy and switches not working should soon disappear Jan 05 09:58:41 <[florian]> sw: as I see no reasons why we should hardcode the MDIO bus number in case of a "fixed phy" or switch like it is done currently Jan 05 11:06:18 florian * r19039 /trunk/tools/mtd-utils/patches/120-cygwin_fixes.patch: [tools] make mtd-utils compile correctly on cygwin Jan 05 11:06:23 florian * r19040 /trunk/scripts/config/.gitignore: [scripts] adjust svn:ignore for cygwin builds Jan 05 11:22:25 Hi there Jan 05 11:22:56 does anyone is using the tor packages on kamikaze 8.02 with hardwarelike fonera1/2 ? Jan 05 11:23:07 it seems that tor is using much memory Jan 05 11:27:12 and make it quite unsuable on foneras Jan 05 11:54:57 [florian]: no I haven't tried it yet Jan 05 12:14:03 [florian]: I applied the changes from the patch directly to the device (e.g. not to a source tree, but, except for removing "local" I made all the changes in the patch) Jan 05 12:14:30 [florian]: I'm not sure about removing local, but for the rest it works fine Jan 05 12:17:21 <[florian]> cshore: cool Jan 05 12:25:41 ping xMff Jan 05 12:32:13 cshore: pong Jan 05 12:32:31 just a couple of quick things for the NIU Jan 05 12:33:37 I think for some things like the DHCP/DNS (dnsmasq) you should put the the 'techie' name for the task in parentres ( e.g. Manage address (DHCP)) so that those of us who are used to that name can fine what we're looking for Jan 05 12:34:01 also, NIU seems to be unauthenticated. Is this intentional? Jan 05 12:34:41 parentres = parenthesis () Jan 05 12:35:13 unauthenticated access is not intentional ;) It's a one-liner to change that once it is done Jan 05 12:35:38 ok Jan 05 12:35:46 cool Jan 05 12:35:57 overall it's great though Jan 05 12:37:09 will look into the labeling too, but might take a while - not much spare time atm Jan 05 12:37:13 right Jan 05 12:37:18 how is the firewall going Jan 05 12:38:35 [florian]: can you look at the patch I posted to the devel list for entering failsafe on the 63xx? CTRL-C doesn't work Jan 05 12:39:14 hm firewall needs more time, I have a working poc but not much more Jan 05 12:40:10 [florian] ticket #5516 Jan 05 14:39:01 hello all Jan 05 14:39:39 I don't understand how to install an init script Jan 05 14:39:55 I have a package with a script in init.d Jan 05 14:40:12 and I would like to have the links in rc.d Jan 05 14:40:33 it should be done by calling my script with the parameter enable Jan 05 14:41:19 Is there a place where it should be done automatically during install ? Jan 05 14:54:38 the buildroot does it automatically if the package is included (*) in the image Jan 05 14:54:53 if you install the package with opkg the init script is not enabled by default Jan 05 15:02:02 nico * r19041 /trunk/target/linux/olpc/: [target] also remove empty directories after [19035] Jan 05 15:03:34 <[florian]> {Nico}: thanks! git-svn does not do that for you Jan 05 15:05:17 jow_laptop: it doesn't work for me Jan 05 15:05:34 jow_laptop: do you know where it is done ? Jan 05 15:06:56 tmonjalo: if it does not work during build time then your script is faulty. This usually happens if your script is not able to run on the host system, make sure to put everything target specific into the start() procedure Jan 05 15:08:10 within the build system only /etc/functions.sh and /etc/rc.common are available, the sdk will invoke your init script with enable as argument which must succeed Jan 05 15:09:50 maybe put the script in pastebin and I can take a look Jan 05 15:10:18 jow_laptop: I haven't redefined the enable function Jan 05 15:10:34 tmonjalo: this is okay, it is provided by rc.common Jan 05 15:10:56 my point is, if you init script exits in the sdk with an unsucessful error code then enabling it won't work Jan 05 15:11:23 jow_laptop: ah ok Jan 05 15:12:04 jow_laptop: but I'd prefer if it was unrelated... Jan 05 15:12:15 no way Jan 05 15:12:33 the script must be evaluated in order to determine START and STOP Jan 05 15:12:34 my init script cannot run on the host Jan 05 15:12:49 yes I agree Jan 05 15:12:49 it can, just protect the target specific stuff Jan 05 15:12:54 it is Jan 05 15:13:02 show me please Jan 05 15:13:25 yep Jan 05 15:15:02 jow_laptop: http://pastebin.com/d55da32e8 Jan 05 15:15:34 basic... Jan 05 15:15:55 ah Jan 05 15:16:01 your first line is wrong Jan 05 15:16:07 you forgot to source rc.common Jan 05 15:16:13 #!/bin/sh /etc/rc.common Jan 05 15:16:17 then it'll work Jan 05 15:16:28 ha Jan 05 15:16:38 jow_laptop: really thanks Jan 05 15:16:51 np yw Jan 05 15:16:57 hard return from holidays... Jan 05 15:53:15 I'm looking for a simple package makefile skeleton Jan 05 15:53:45 just to place 2 files included with the package Jan 05 15:54:17 what package can I use for example? Jan 05 15:54:58 package/firewall Jan 05 15:55:52 exactaly what I was looking Jan 05 16:08:03 after I place the package folder under /feeds/packages/utils/utils/, do I need to do anything else to have it on make menuconfig? Jan 05 16:12:14 I have to put it under /packages/ to appear, but to make the patch I should first place it under /feeds/packages? Jan 05 16:13:18 ./scripts/feeds update packages Jan 05 16:13:30 this should pick up your new package Jan 05 16:14:16 great, thanks Jan 05 16:23:50 jow_laptop: What is the best way to disable a startup script at build time ? Jan 05 16:24:05 hm Jan 05 16:24:20 you can't I believe Jan 05 16:24:33 you could define a postinstall section Jan 05 16:24:44 which removes / disables he symlink Jan 05 16:24:52 ok Jan 05 16:25:00 just rgrep for "postinst" in the package makefiles Jan 05 16:25:14 bbl Jan 05 16:25:29 tmonjalo: if the script returns 1 @ build time it will be disabled Jan 05 16:26:00 solca: yes I experimentaded this Jan 05 16:26:37 solca: if it's the official way to have a disabled package in the firmware... Jan 05 16:27:03 tmonjalo: I think not making it executable works also Jan 05 16:27:33 Maybe that an explicit variable should be better in the build system Jan 05 16:28:10 loswilios: I like your solution Jan 05 17:01:37 jow_laptop: "./scripts/feeds update packages" didn't work Jan 05 17:02:22 Is there anybody who is aware of a madwifi bug when connected to a WEP AP while being a WEP AP on another interface ? Jan 05 17:02:58 jow_laptop: sorry, I forgot to do feeds install Jan 05 17:13:00 xMff: luci included in trunk is broken Jan 05 17:15:53 checkout trunk, add luci-kamikaze, make Jan 05 17:15:57 compiles OK Jan 05 17:16:07 make again, and I get a error Jan 05 18:07:27 nico * r19042 /feeds/efl/python-etk/Makefile: [packages] python-etk: update svn url Jan 05 18:37:16 nbd * r19043 /trunk/package/mac80211/patches/540-ath9k_tx_fix.patch: mac80211: update ath9k tx fix with changes from upstream Jan 05 18:37:23 nbd * r19044 /trunk/target/linux/ar71xx/image/Makefile: ar71xx: fix ubntxm image builds Jan 05 19:14:29 nbd * r19045 /trunk/package/mac80211/patches/550-queue_stop_fix.patch: mac80211: merge johill's tx queue stop fix patch from the linux-wireless list - fixes serious latency spike and tcp throughput drop issues Jan 05 19:47:57 nico * r19046 /trunk/target/linux/ixp4xx/config-2.6.30: [target] ixp4xx: add missing config symbol for 2.6.30 Jan 05 20:12:14 juhosg * r19047 /trunk/target/linux/ar71xx/files/arch/mips/ar71xx/ (Kconfig mach-wrt400n.c): ar71xx: init PCI on the WRT400N board (thanks to Ben Pfountz for testing) Jan 05 20:35:28 xMff: ping Jan 05 22:17:20 I created a package and put it under feeds/packages/utils/, what command can I use to create the patch? Jan 05 23:05:25 on Tue, 27 Oct 2009 13:26:00 I submitted to openwrt-devel a patch, proposing that the init.d/led should run after init.d/done to avoid led settings being override Jan 05 23:06:01 that's because otherwise I can not set the led being use to signal the system boot on /etc/config/system Jan 05 23:06:38 that led is changed on /etc/init.d/done, so led configuration should be the last thing running in the boot sequence Jan 05 23:06:46 I didn't get any comments Jan 06 00:38:14 hi Jan 06 00:38:55 I was having this problem https://forum.openwrt.org/viewtopic.php?pid=99856#p99856 and it turned out that ppp(oe) scripts just didn't do anything with the dns option, so I made these changes which work: http://fpaste.org/OHRp/ Jan 06 00:39:04 is that the right way to do it? Jan 06 00:53:26 humm Jan 06 00:53:53 and I don't even want to guess how [ -n "$demand" ] && echo "nameserver 1.1.1.1" > /tm\ Jan 06 00:53:56 p/resolv.conf.auto Jan 06 00:53:58 is supposed to work... Jan 06 01:08:02 nico * r19048 /trunk/package/kernel/modules/block.mk: [package] kernel: fix IDE modules on 2.4 Jan 06 01:11:32 nico * r19049 /trunk/package/kernel/modules/fs.mk: [package] kernel: remove more calls to CompareKernelPatchVer and use 2.4 exceptions instead Jan 06 02:34:22 nbd * r19050 /trunk/package/base-files/files/etc/init.d/led: Jan 06 02:34:22 run led init after /etc/init.d/done to keep the configured led settings active after the system has finished booting Jan 06 02:34:22 Signed-off-by: Nuno Goncalves **** ENDING LOGGING AT Wed Jan 06 02:59:57 2010