**** BEGIN LOGGING AT Tue Jun 09 02:59:57 2009
Jun 09 03:33:07 <swalker> shouldn't the dependency system bail on external package DEPENDS not prefixed with +?
Jun 09 05:38:15 <russell_> Weedy, i have some, but i haven't done much with it
Jun 09 05:38:28 <russell_> like, i have a routerstation sitting on my desk
Jun 09 05:38:43 <russell_> and some nanostations
Jun 09 05:47:16 <russell_> what is the canonical way to add your own configuration and other files to an image?  i recall there was a magical directory in the buildroot you could put your files and they'd get overlaid or something.
Jun 09 05:48:00 <[florian]> you should put your files in trunk/files/ for instance
Jun 09 05:48:19 <[florian]> they will be copied in the final rootfs
Jun 09 05:52:39 <russell_> okay, cool
Jun 09 05:53:12 <cshore> what if you want to override a package's settings?
Jun 09 05:54:21 <cshore> (because AIUI packages do their settings after trunk/files (or am I wrong?)
Jun 09 05:54:39 <Bartman007> cshore: files/ is copied after packages are installed, directly before the image itself is created.
Jun 09 05:54:51 <cshore> cool, thanks
Jun 09 05:55:07 * russell_ clarifies that "trunk" in not a literal here?
Jun 09 05:55:15 <Bartman007> trunk == your checkout dir.
Jun 09 05:55:18 <Bartman007> $(TOPDIR)
Jun 09 05:55:18 <russell_> right
Jun 09 05:55:53 <cshore> so it doesn't exist by default, correct
Jun 09 05:56:07 <Bartman007> right
Jun 09 05:56:27 <russell_> rsync -v -a -H /jffs/ buildhost:/src/openwrt/files/
Jun 09 05:57:00 <russell_> (and then clean up the magic files)
Jun 09 05:57:34 <Bartman007> I used to do that, but I'm moving back to keeping my configs in jffs on any platform supported by sysupgrade
Jun 09 05:58:02 <russell_> another idea i had was to have a local package with the files
Jun 09 05:58:44 <Bartman007> that's the "cleaner" solution for deploying to multiple devices, but not worth the effort for most individual users.
Jun 09 05:58:53 <russell_> yeah
Jun 09 05:59:20 <cshore> can you control what order the package is placed in the image?
Jun 09 05:59:29 <russell_> is there a way to ensure the order the packages are unpacked?  so that ... what cshore said!
Jun 09 05:59:34 <Bartman007> though you'll have to resolve the issue w/ package file collisions.
Jun 09 06:00:06 <russell_> uninstalling could be a bitch if there are collisions
Jun 09 06:01:10 <cshore> Hmmm...a colleague of mine is saying that /files/etc/config gets overwritten by some packages
Jun 09 06:02:46 <cshore> Does that get modified at runtime when the image is first copied to jffs2?
Jun 09 06:03:08 <Bartman007> cshore: if the settings are done via /etc/uci-defaults/ they would overrride files/
Jun 09 06:03:29 <Bartman007> don't know of any other instances it would occur in, but some packages may do things weirdly.
Jun 09 06:03:33 <cshore> Is there anyway to override that?
Jun 09 06:04:03 <cshore> modify /etc/uci-defaults/?
Jun 09 06:04:19 <cshore> rather files/etc/uci-defaults?
Jun 09 06:28:55 <cshore> ok, here's a question: why have uci-defaults at all?  Why not unify to all being done by files in the image?
Jun 09 07:13:06 <danage> salut [florian], sorry i had to log off unexpectedly. did you see my last paste? oh and another thing: are you even french?
Jun 09 07:51:16 <[florian]> danage: no problems, the last log seems encouraging and I think I found the culprit
Jun 09 07:51:45 <danage> you can make something of the register dump? wow
Jun 09 07:52:31 <[florian]> what seems to happening is that it received an early interrupt, not sure why CFE does not globaly disable them though
Jun 09 07:52:45 <[florian]> and I recall seeing in broadcom's code that they cleared the irq mask
Jun 09 07:52:49 <[florian]> early while booting
Jun 09 07:54:48 <CIA-29> florian * r16392 /trunk/package/iptables/Makefile: [package] fix typo in iptables makefile resulting in patches not being applied (#5311)
Jun 09 07:55:54 <[florian]> danage: so basically I think we are missing one line and it should be booting
Jun 09 08:04:53 <CIA-29> florian * r16393 /trunk/ (6 files in 5 dirs):
Jun 09 08:04:53 <CIA-29> [brcm63xx] rework imagetag firmware generator and MTD partition parser to accomodate with all known
Jun 09 08:04:53 <CIA-29> imagetag versions from Broadcom and vendors, patch from Daniel Dickinson (fixes #4987).
Jun 09 11:16:34 <danage> [florian]: that would be really nice. let me know when you have results, i will probably be home at around 6
Jun 09 12:43:23 <Weedy> russell_: anyidea how to build a image for picostation2 hp?
Jun 09 18:33:40 <fofware> xMff:ping
Jun 09 18:33:54 <xMff> .
Jun 09 18:34:47 <fofware> Hello xMff, the patch that i did send about pack in lua, work but realy it's not needed
Jun 09 18:35:09 <xMff> okay
Jun 09 18:36:11 <fofware> I write all in lua, because i didn't understand very well how that patch work
Jun 09 18:37:34 <fofware> Now i have a loginpage of chilli all in lua without any line in js, I'm happy :o)
Jun 09 19:26:06 <yassine> anyone can explaings why is it saying "enable multicast routing by default" in http://kamikaze.openwrt.org/release.txt while # tcpdump  -s 60 -tt -n -l -v udp and ip multicast does not capture a signle packge?
Jun 09 19:29:49 <yassine> am i supposed to configure something aditional ??
Jun 09 19:30:19 <xMff> maybe you need some additional firewall rules
Jun 09 19:31:17 * russell_ fervently wishes there was some guidance on translating uci-style configuration of snmpd into something familiar like the native configuration format.  the /etc/config/snmpd that ships with the package is a mismash of confusion.
Jun 09 19:31:34 <russell_> or rather, the other way around
Jun 09 19:34:14 * russell_ has a perfectly understandable snmpd.conf, wants a corresponding /etc/config/snmpd
Jun 09 19:47:13 <yassine> xMff, im unable to find any option that would add/activate multicast routing :s
Jun 09 19:47:35 <xMff> yassine: it's now activated in the kernel, not more
Jun 09 19:47:42 <xMff> no idea what else is needed to actually use it
Jun 09 19:48:11 <xMff> but maybe the firewall blocks such traffic by default that's what I meant
Jun 09 19:48:19 <xMff> maybe forwarding whatever
Jun 09 19:48:24 <yassine> ok ill continue my search, but if you have a hint for me its always welcome
Jun 09 19:48:33 <xMff> russell_: have you tried to just create /etc/snmpd.conf ?
Jun 09 19:48:59 <xMff> russell_: for dnsmasq this works, it will always search it's standard confg regardless of what has been specified in uci
Jun 09 19:55:02 <noz> What determines whether a package build tries to apply patches? I have a new Makefile, almost identical to another package, neither has a 'patches' dir, but mine fails with "ls: cannot access ./patches: No such file or directory" during 'make package/name/prepare V=99'
Jun 09 19:56:22 <xMff> nothing determines it
Jun 09 19:56:35 <noz> !
Jun 09 19:56:40 <xMff> the ls ./patches is the part where the build system decides whether to apply patches or not
Jun 09 19:57:06 <xMff> as long as it does not interrupt the build, all is well
Jun 09 19:57:07 <noz> But the other package has only files/ and Makefile, and builds.
Jun 09 19:57:43 <noz> Doh! Sorry - you mean that it has actually prepared?
Jun 09 19:59:59 <russell_> xMff: oh, yeah.
Jun 09 20:00:11 <xMff> noz: yes
Jun 09 20:00:28 <noz> Thanks. (Hangs head!)
Jun 09 20:00:54 <russell_> still, if i could understand what all that uci converting stuff was doing, i could use all their hard work
Jun 09 20:01:40 <noz> Now I shall tackle the automake mess, but won't trouble the list with that.
Jun 09 20:18:02 <xMff> russell_: yeah, ... the documentation problem :)
Jun 09 20:51:25 <yassine> is there a place i can find docs related to the newly inserted multicast routing support for openwrt?
Jun 09 21:09:56 <[florian]> yassine: no there is not one, apart the changelog
Jun 09 21:11:54 <yassine> :((
Jun 09 21:12:25 <[florian]> this is just general linux multicast routing support
Jun 09 21:12:29 <yassine> [florian], im unable to set it to work
Jun 09 21:13:15 <[florian]> check your firewall settings
Jun 09 21:13:19 <[florian]> they are very likely to block igmp
Jun 09 21:13:58 <danage> ping [florian], news?
Jun 09 21:14:27 <[florian]> danage: almost, hold on :)
Jun 09 21:14:34 <danage> :)
Jun 09 21:19:32 <yassine> [florian], what should i exactly check in my firewall ? i mean its the default configuration shipped with openwrt 8.09.1
Jun 09 21:19:42 <yassine> any further hints maybe?
Jun 09 21:20:24 <[florian]> yassine: you should check wether igmp is blocked or not
Jun 09 21:22:27 <[florian]> danage: can you test that please: http://openwrt.pastebin.com/m5ceed61f ?
Jun 09 21:27:36 <danage> sec.
Jun 09 21:28:16 <CIA-29> florian * r16394 /trunk/target/linux/brcm63xx/ (config-2.6.27 files/drivers/watchdog/bcm63xx_wdt.c): [brcm63xx] fix the watchdog driver and enable it by default
Jun 09 21:31:50 <danage> do i need that?
Jun 09 21:32:30 <[florian]> danage: no, this is an extra driver
Jun 09 21:33:50 <danage> compiling...
Jun 09 21:33:57 <[florian]> good
Jun 09 21:34:05 <[florian]> I have not even compile tested this patch
Jun 09 21:34:14 <danage> i applied the patch manually, let's hope i made no mistakes. this p4 is a little slow, it might take 20 minutes or so
Jun 09 21:34:37 <danage> i _am_ leaving all of yesterday's patches in (i.e. the second one)
Jun 09 21:34:43 <danage> correct?
Jun 09 21:35:21 <[florian]> correct
Jun 09 21:35:53 <CIA-29> florian * r16395 /trunk/target/linux/brcm63xx/files/drivers/watchdog/bcm63xx_wdt.c: [brcm63xx] cleanup the watchdog driver a bit
Jun 09 21:38:06 <[florian]> danage: the patch is not exactly correct for all devices, but anyway, I will fix it up later if it happens to work ;)
Jun 09 21:39:12 <danage> let's see what happens
Jun 09 21:40:00 <[florian]> yup
Jun 09 21:40:24 <[florian]> the current patch should work for your device since I take it from broadcom's bsp, but it could break other devices ;)
Jun 09 21:51:01 <[florian]> danage: if you prefer that I compile images and put them for you on a server let me know
Jun 09 21:51:49 <danage> done for now... flashing...
Jun 09 21:52:34 <[florian]> cool
Jun 09 21:54:56 <danage> http://openwrt.pastebin.com/m2a2c0b11
Jun 09 21:55:23 <danage> they seem to have different values
Jun 09 21:55:26 <[florian]> crap
Jun 09 21:55:38 <danage> hang on, double checking i patched correctly
Jun 09 21:57:00 <danage> [florian]: http://openwrt.pastebin.com/m59958bdc
Jun 09 21:57:06 <danage> i think i did...
Jun 09 21:57:52 <[florian]> danage: I also think you did
Jun 09 21:57:59 <danage> rats
Jun 09 21:58:23 <[florian]> danage: could you send me your System.map in build_dir/linux-brcm63xx/linux-2.6.27.24/System.map ?
Jun 09 22:02:25 <danage> it exceeds pastebin size
Jun 09 22:04:50 <danage> http://openwrt.pastebin.com/m329dd7e1
Jun 09 22:04:53 <danage> here is parts of it
Jun 09 22:51:04 <russell_> hmm.  /me working on a package, $(INSTALL_BIN) seems to resolve a symlink (-> /sbin/iptables) and as a result gives the ipkg a copy of the build-system's iptables instead of a symlink to the target's iptables.
Jun 09 22:51:29 <russell_> i guess i should just use $(CP) instead
Jun 09 23:04:04 <russell_> gah.  trying the $(TOPDIR)/files/ thing, failure when I try to overwrite a symlink with a real file
Jun 09 23:04:34 <russell_> cp: not writing through dangling symlink `/aux/src/openwrt/build_dir/target-mipsel_uClibc-0.9.30.1/root-brcm47xx/./etc/snmp/snmpd.conf'
Jun 09 23:12:08 <Bartman007> russell_: does adding --remove-desination instead of -f behave as desired?
Jun 09 23:12:32 <Bartman007> adding it in the definition of $(CP) in rules.mk that is.
Jun 09 23:13:20 <russell_> wait, /me just described two problems, which one are you referring to?
Jun 09 23:13:51 <Bartman007> the second problem
Jun 09 23:14:27 * russell_ looks
Jun 09 23:15:13 <Bartman007> wrt. the first, as far as I know install cannot be instructed to not dereference the file, whereas cp can.
Jun 09 23:15:39 <russell_> those two problems might work at cross-purposes?
Jun 09 23:35:14 <russell_> Bartman007: this fixes the second problem, http://pastebin.ca/1454560
Jun 09 23:39:53 <cshore> How would I have the router boot from the flash, but instead of using jffs2 use USB as root?
Jun 09 23:54:22 <cshore> is luci's svn (http) feed down?
Jun 09 23:54:35 <xMff> hm no?
Jun 09 23:54:58 <cshore> bother, I can't connect to the server via svn
Jun 09 23:55:15 <xMff> which url?
Jun 09 23:55:18 <Bartman007> cshore: worksforme
Jun 09 23:55:30 <cshore> svn.luci.subsignal.org
Jun 09 23:55:33 <xMff> fine here too
Jun 09 23:55:33 <Bartman007> russell_: cross-purposes?
Jun 09 23:56:10 <xMff> cshore: svn co http://svn.luci.subsignal.org/luci/trunk   woks here
Jun 09 23:56:22 <cshore> svn checkout http://svn.luci.subsignal.org/luci/trunk/contrib/packages
Jun 09 23:56:41 <cshore> wtf....ok must be something here then
Jun 09 23:56:48 <xMff> the s in the end is wrong
Jun 09 23:56:55 <xMff> it's ".../package"
Jun 09 23:57:28 <cshore> still doesn't work (can't connect to server); must be something here that needs fixing; maybe reseting the router will help
Jun 09 23:57:43 <xMff> ah
Jun 09 23:58:37 <xMff> there's also http://svn2.luci.freifunk-halle.net/luci/trunk/contrib/package/   does this work for you?
Jun 10 00:04:34 <russell_> Bartman007: i just didn't want to shoot myself in the foot.  didn't want to modify $(CP) semantics, as it's used so widely.
Jun 10 00:05:51 <russell_> anywho, nm about that remark
Jun 10 00:06:51 <russell_> i solved problem #1 by switching from $(INSTALL_BIN) back to $(CP) and problem #2 with the paste above
Jun 10 00:21:21 <Bartman007> russell_: I
Jun 10 00:21:52 <Bartman007> I'll do the foot shooting myself then, and decide if that is what we actually intend to do.
Jun 10 00:50:13 <cshore> it was squeeze issue
Jun 10 00:50:42 <Bartman007> cshore: heh, more squeeze issues. yay!
Jun 10 00:51:28 <cshore> Bartman007: At least someone else already reported it and there was an answer; I hate waiting
Jun 10 00:51:54 <Bartman007> have the bug number on hand?
Jun 10 00:52:51 <cshore> Barman007: 531338
**** ENDING LOGGING AT Wed Jun 10 01:42:29 2009

**** BEGIN LOGGING AT Wed Jun 10 01:55:15 2009
**** ENDING LOGGING AT Wed Jun 10 02:59:57 2009