[00:18] <db90h> hehe, that's whack.. defining a variable after its referenced
[00:57] <{Nico}> nbd: are you around?
[04:30] <dbcch> none 1128 0 1128 0% /tmp
[04:30] <dbcch> none 1128 0 1128 0% /mnt
[04:30] <dbcch> any idea wzup with that?
[04:31] <nbd> that's tmpfs
[04:31] <dbcch> this is kamikaze.. did JFFS2 fail to mount?.. maybe I am missing some dependency, but dunno what
[04:31] <dbcch> ramdisk?
[04:31] <nbd> yes
[04:31] <nbd> are you using a squashfs image?
[04:31] <dbcch> yea
[04:32] <nbd> it starts by mounting a tmpfs with a symlink overlay
[04:32] <nbd> when it's done booting, it copies that symlink 'overlay' on the flash
[04:32] <nbd> (when first flashing the image)
[04:33] <dbcch> ic, so its gotten stalled during booting?
[04:33] <nbd> have a look at ps ax
[04:33] <nbd> to see what it's doing
[04:33] <dbcch> yea, the weird think is I don't see it 'hung up' on anything in the init.d stuff
[04:34] <nbd> then the jffs2 should be mounted now
[04:34] <dbcch> "jffs2 unusable; using ramdisk"
[04:35] <dbcch> would not enough space result in such an error?
[04:35] <nbd> yes
[04:35] <dbcch> k, that's it then for sure
[04:35] <[mbm]> well, it's normal to see that message the first time
[04:35] <dbcch> ...this is subsequent times
[04:35] <[mbm]> ah
[04:35] <dbcch> 4th boot or so at least
[04:35] <nbd> then it's probably lack of space
[04:36] <[mbm]> check the size of the jffs2 partition
[04:36] <dbcch> yea, its on a v5, so definitely
[04:36] <[mbm]> oh, then yes
[04:36] <nbd> oh... a v5
[04:36] <nbd> yeah, kamikaze is a bit bigger than whiterussian
[04:36] <nbd> it hasn't been tuned for the small stuff at all
[04:36] <dbcch> sorry I didn't make that clear.. seeing the tmpfs just confused me as I had not seen that mount before
[04:36] <nbd> i should have guessed it from the size of the tmpfs :)
[04:37] <dbcch> this is one of my 'custom' builds with an increasing load of shit
[04:37] <dbcch> ah, well in that case I"m switching back to WR ;)
[04:37] <h3sp4wn> That happened to me once when I rebooted before the convert to jffs2 script had finished and running firstboot didn't help (had to reflash)
[04:39] <[mbm]> h3sp4wn: ? if you interrupt it all that happens is that it does it again the next bootup
[04:41] <h3sp4wn> I didn't interupt it just edited /etc/config/network and then rebooted - it was giving me lots of jffs2 errors (I thought I had messed up the flash but it was ok after reflashing, I may have been wrong however)
[04:42] <nbd> how long ago was this?
[04:42] <h3sp4wn> A few weeks ago (I haven't done it since)
[04:43] <h3sp4wn> I just wait until it has finished now
[04:43] <h3sp4wn> (but it was with buildroot-ng)
[04:44] <[mbm]> what's supposed to happen is that there's a dirty flag to mark jffs2 as unusable
[04:44] <[mbm]> if that's set, then it runs firstboot
[04:45] <[mbm]> and it doesn't clear it until after firstboot runs
[04:45] <[mbm]> so if you ever interrupt firstboot, the flag is still set and it runs firstboot again
[04:46] <nbd> btw. i just noticed that maybe it would be better if we kept the dirty flag an didn't use the jffs2 hack for erasing the jffs2 overlay
[04:46] <[mbm]> :)
[04:46] Action: [mbm] was thinking about that
[04:46] <nbd> :)
[11:17] <malbon> oh no, it's the murb :P
[11:17] Action: malbon runs for cover.
[11:17] <murb> my first attempt at a openwrt kmod worked :p
[11:18] Action: murb now has worky SIP conntracking
[11:18] <murb> und nat
[11:18] <malbon> is it legal sip contracking?
[11:18] <murb> probably some law against it in germany.
[11:18] <murb> i now have telephone and intenret in my new house without cables
[11:19] <malbon> always nice. sipgate?
[11:19] <murb> asterisk box in house 70m away + asterisk box in flat in Wiesbaden
[11:20] <malbon> ah.
[11:20] <murb> oh and some sipgate for some +44207[0-9]*
[11:20] <malbon> the sip contrack is gpl? not-dubious licencing?
[11:20] <murb> it'll be going into 2.6.18
[11:21] <malbon> ah cool.
[11:21] <malbon> buildroot or buildroot-ng?
[11:24] <malbon> murb: can you send me a svn diff, I'll check it in.
[11:31] <murb> oh bugger someone else has already done this work :(
[11:34] <malbon> :) SNAFU
[11:35] <CIA-17> florian * r4101 /packages/libs/uclibc++/ (. Makefile files/ files/config.i386 files/config.mips): Add uclibc++ 0.2.0
[12:15] <CIA-17> florian * r4102 /packages/libs/libpthread/ (. Makefile): Add libpthread
[12:25] <CIA-17> florian * r4103 /packages/libs/cgilib/ (5 files in 2 dirs): Add cgilib
[12:36] <florian___> nbd: ping ?
[12:39] <nbd> pong
[12:42] <florian___> is there anything special to worry about when using $(call Build/Configure/Default
[12:42] <florian___> can I put my options with \ at the end ?
[12:44] <florian___> is this valid ? http://pastebin.ca/76278
[12:45] <nbd> yes, although i wanted to move these ac_foo=bar things into 'site files'
[12:46] <florian___> site files ?
[12:46] <nbd> files that define these variables globally for any configure script
[12:46] <florian___> ok that would be a lot easire
[12:46] <florian___> easier
[12:46] <nbd> openembedded is doing that already
[12:47] <nbd> maybe you can copy some stuff from them
[12:47] <florian___> 1 site files per package ?
[12:47] <nbd> no, one global site file
[12:47] <nbd> (for every target arch)
[12:48] <nbd> things like the ac_cv_sizeof_* variables are probably used in many packages
[12:48] <florian___> right
[12:48] <florian___> many are package specific
[12:50] <florian___> so we have to concatenate this arch variables with the package specific options specified in the package Makefile right ?
[12:50] <nbd> maybe it would be easier if we did it like this: add a template Configure/sitedefs/default
[12:50] <nbd> and put that into Configure/sitedefs by default
[12:51] <nbd> (allowing the user to override it, reusing the defaults)
[12:52] <florian___> ok
[12:54] <nbd> the build system should then write the contents Build/Configure/sitedefs to a file in PKG_BUILD_DIR
[12:54] <nbd> and include that before running configure
[12:54] <nbd> so that it won't generate an awfully long command line later
[12:55] <nbd> writing the stuff to a file might be a bit tricky, you can see from the postinst/preinst/... templates how it's done
[12:57] <florian___> yes it's not that easy
[12:58] <florian___> what about package specific sitedefs should we still define them as second argument for Build/Configure/Default?
[12:58] <nbd> overriding Build/Configure/Sitedefs is probably better
[12:58] <nbd> i want to get away from using lots of non-obvious argument lists
[12:59] <florian___> ok
[13:08] <florian___> for the arch specifi sitedefs, should we put them in in a file ?
[13:08] <nbd> the build system should merge everything and then put it in a file
[13:08] <florian___> how would he determine the architecture specifi values ?
[13:08] <nbd> and maybe instead of loading it from the shell it should set the appropriate env var
[13:08] <nbd> the
[13:09] <nbd> just a sec..
[13:10] <nbd> http://pastebin.ca/76286
[13:11] <florian___> that's good in a package makefile
[13:11] <florian___> what about the definition in include/package.mk ?
[13:12] <nbd> they define Build/Configure/Sitedefs with the contents of .../default
[13:13] <florian___> I was thinking of smthg like that http://pastebin.ca/76288
[13:15] <nbd> hmm... yeah
[13:15] <nbd> probably better
[13:15] <nbd> what's with the $(1)
[13:15] <nbd> ah, wait
[13:15] <florian___> your custom sitedefs
[13:15] <florian___> but your method is better
[13:15] <nbd> we should combine both
[13:16] <florian___> lol that's exactly what I wanted to write :)
[13:16] <nbd> use $(TOPDIR)/include/site/
[13:16] <nbd> sorry
[13:16] <nbd> use $(TOPDIR)/include/site/$(ARCH).def
[13:16] <nbd> and then add the contents of Build/Configure/Sitedefs
[13:16] <nbd> which will only get added if the user overrides it
[13:17] <nbd> and the contents of that file should be simple name=value lines without \ or indentation
[13:17] <nbd> s/indentation/forced indentation/
[13:18] <florian___> right
[13:26] <florian___> does this look good : http://pastebin.ca/76290
[13:28] <florian___> I forgot a trainling slash in Build/Configure/Sitedefs
[13:35] <florian___> nbd: is this ok to you ?
[13:38] <nbd> i think that would not work
[13:39] <nbd> because you're mixing the cat command with the sitedefs
[13:49] <florian___> does "use" exist ?
[14:04] <nbd> use?
[14:06] <florian___> well I could I read settings from a file ?
[14:07] <florian___> include <file> ?
[14:07] <nbd> you can't arbitrarily put the contents of a file at a certain place
[14:08] <florian___> I need to parse it right ?
[14:11] <nbd> you should do this: $(CP) $(TOPDIR)/include/site/$(ARCH).def $(PKG_BUILD_DIR)/site.def
[14:11] <nbd> and then append the contents of Build/Configure/Sitedefs
[14:12] <nbd> and finally set the environment variable that tells configure to use this file
[14:19] <florian___> how long can it take for jffs2 to erase all blocks after the end markers ?
[14:21] <florian___> oh it's finished :)
[14:29] <florian___> there is no more /etc/preinit in buildroot-ng images ?
[14:40] <florian___> nbd: is this better ? http://pastebin.ca/76315
[14:45] <nbd> /etc/preinit is now target-specific
[14:45] <florian___> ok, I will adapt it for au1000
[14:45] <nbd> that's not quite what i meant. i wanted to get rid of the extra configure parameter entirely
[14:45] <nbd> (pastebin)
[14:46] <nbd> i wanted the user to put in the extra definitions inside the Build/Configure/Sitedefs template directly
[14:48] <florian___> ok, so I just $(call Build/Configure/Sitedefs) inside Build/Configure/Default without arguments ?
[14:48] <nbd> yes
[14:48] <nbd> well, not exactly
[14:49] <florian___> ah :)
[14:49] <nbd> the contents of Build/Configure/Sitedefs should be appended to the site.def file
[14:50] <florian___> is not $(1) >> $(PKG_BUILD_DIR)/site.def doing it ?
[14:55] <nbd> no
[14:55] <nbd> i have to go now... have a look at how it's done through the environment with preinst/postinst/etc.
[14:55] <florian___> ok
[15:23] <malbon> printfprintf/win 4
[15:23] <malbon> oops
[23:15] <crazy_imp> has someone tried to use a cd-drive in a usb-box with kamikaze? which lines i have to add in the kernel-config?
[00:00] --- Sun Jul 2 2006
[00:57] <{Nico}> nbd: are you around?
[04:30] <dbcch> none 1128 0 1128 0% /tmp
[04:30] <dbcch> none 1128 0 1128 0% /mnt
[04:30] <dbcch> any idea wzup with that?
[04:31] <nbd> that's tmpfs
[04:31] <dbcch> this is kamikaze.. did JFFS2 fail to mount?.. maybe I am missing some dependency, but dunno what
[04:31] <dbcch> ramdisk?
[04:31] <nbd> yes
[04:31] <nbd> are you using a squashfs image?
[04:31] <dbcch> yea
[04:32] <nbd> it starts by mounting a tmpfs with a symlink overlay
[04:32] <nbd> when it's done booting, it copies that symlink 'overlay' on the flash
[04:32] <nbd> (when first flashing the image)
[04:33] <dbcch> ic, so its gotten stalled during booting?
[04:33] <nbd> have a look at ps ax
[04:33] <nbd> to see what it's doing
[04:33] <dbcch> yea, the weird think is I don't see it 'hung up' on anything in the init.d stuff
[04:34] <nbd> then the jffs2 should be mounted now
[04:34] <dbcch> "jffs2 unusable; using ramdisk"
[04:35] <dbcch> would not enough space result in such an error?
[04:35] <nbd> yes
[04:35] <dbcch> k, that's it then for sure
[04:35] <[mbm]> well, it's normal to see that message the first time
[04:35] <dbcch> ...this is subsequent times
[04:35] <[mbm]> ah
[04:35] <dbcch> 4th boot or so at least
[04:35] <nbd> then it's probably lack of space
[04:36] <[mbm]> check the size of the jffs2 partition
[04:36] <dbcch> yea, its on a v5, so definitely
[04:36] <[mbm]> oh, then yes
[04:36] <nbd> oh... a v5
[04:36] <nbd> yeah, kamikaze is a bit bigger than whiterussian
[04:36] <nbd> it hasn't been tuned for the small stuff at all
[04:36] <dbcch> sorry I didn't make that clear.. seeing the tmpfs just confused me as I had not seen that mount before
[04:36] <nbd> i should have guessed it from the size of the tmpfs :)
[04:37] <dbcch> this is one of my 'custom' builds with an increasing load of shit
[04:37] <dbcch> ah, well in that case I"m switching back to WR ;)
[04:37] <h3sp4wn> That happened to me once when I rebooted before the convert to jffs2 script had finished and running firstboot didn't help (had to reflash)
[04:39] <[mbm]> h3sp4wn: ? if you interrupt it all that happens is that it does it again the next bootup
[04:41] <h3sp4wn> I didn't interupt it just edited /etc/config/network and then rebooted - it was giving me lots of jffs2 errors (I thought I had messed up the flash but it was ok after reflashing, I may have been wrong however)
[04:42] <nbd> how long ago was this?
[04:42] <h3sp4wn> A few weeks ago (I haven't done it since)
[04:43] <h3sp4wn> I just wait until it has finished now
[04:43] <h3sp4wn> (but it was with buildroot-ng)
[04:44] <[mbm]> what's supposed to happen is that there's a dirty flag to mark jffs2 as unusable
[04:44] <[mbm]> if that's set, then it runs firstboot
[04:45] <[mbm]> and it doesn't clear it until after firstboot runs
[04:45] <[mbm]> so if you ever interrupt firstboot, the flag is still set and it runs firstboot again
[04:46] <nbd> btw. i just noticed that maybe it would be better if we kept the dirty flag an didn't use the jffs2 hack for erasing the jffs2 overlay
[04:46] <[mbm]> :)
[04:46] Action: [mbm] was thinking about that
[04:46] <nbd> :)
[11:17] <malbon> oh no, it's the murb :P
[11:17] Action: malbon runs for cover.
[11:17] <murb> my first attempt at a openwrt kmod worked :p
[11:18] Action: murb now has worky SIP conntracking
[11:18] <murb> und nat
[11:18] <malbon> is it legal sip contracking?
[11:18] <murb> probably some law against it in germany.
[11:18] <murb> i now have telephone and intenret in my new house without cables
[11:19] <malbon> always nice. sipgate?
[11:19] <murb> asterisk box in house 70m away + asterisk box in flat in Wiesbaden
[11:20] <malbon> ah.
[11:20] <murb> oh and some sipgate for some +44207[0-9]*
[11:20] <malbon> the sip contrack is gpl? not-dubious licencing?
[11:20] <murb> it'll be going into 2.6.18
[11:21] <malbon> ah cool.
[11:21] <malbon> buildroot or buildroot-ng?
[11:24] <malbon> murb: can you send me a svn diff, I'll check it in.
[11:31] <murb> oh bugger someone else has already done this work :(
[11:34] <malbon> :) SNAFU
[11:35] <CIA-17> florian * r4101 /packages/libs/uclibc++/ (. Makefile files/ files/config.i386 files/config.mips): Add uclibc++ 0.2.0
[12:15] <CIA-17> florian * r4102 /packages/libs/libpthread/ (. Makefile): Add libpthread
[12:25] <CIA-17> florian * r4103 /packages/libs/cgilib/ (5 files in 2 dirs): Add cgilib
[12:36] <florian___> nbd: ping ?
[12:39] <nbd> pong
[12:42] <florian___> is there anything special to worry about when using $(call Build/Configure/Default
[12:42] <florian___> can I put my options with \ at the end ?
[12:44] <florian___> is this valid ? http://pastebin.ca/76278
[12:45] <nbd> yes, although i wanted to move these ac_foo=bar things into 'site files'
[12:46] <florian___> site files ?
[12:46] <nbd> files that define these variables globally for any configure script
[12:46] <florian___> ok that would be a lot easire
[12:46] <florian___> easier
[12:46] <nbd> openembedded is doing that already
[12:47] <nbd> maybe you can copy some stuff from them
[12:47] <florian___> 1 site files per package ?
[12:47] <nbd> no, one global site file
[12:47] <nbd> (for every target arch)
[12:48] <nbd> things like the ac_cv_sizeof_* variables are probably used in many packages
[12:48] <florian___> right
[12:48] <florian___> many are package specific
[12:50] <florian___> so we have to concatenate this arch variables with the package specific options specified in the package Makefile right ?
[12:50] <nbd> maybe it would be easier if we did it like this: add a template Configure/sitedefs/default
[12:50] <nbd> and put that into Configure/sitedefs by default
[12:51] <nbd> (allowing the user to override it, reusing the defaults)
[12:52] <florian___> ok
[12:54] <nbd> the build system should then write the contents Build/Configure/sitedefs to a file in PKG_BUILD_DIR
[12:54] <nbd> and include that before running configure
[12:54] <nbd> so that it won't generate an awfully long command line later
[12:55] <nbd> writing the stuff to a file might be a bit tricky, you can see from the postinst/preinst/... templates how it's done
[12:57] <florian___> yes it's not that easy
[12:58] <florian___> what about package specific sitedefs should we still define them as second argument for Build/Configure/Default?
[12:58] <nbd> overriding Build/Configure/Sitedefs is probably better
[12:58] <nbd> i want to get away from using lots of non-obvious argument lists
[12:59] <florian___> ok
[13:08] <florian___> for the arch specifi sitedefs, should we put them in in a file ?
[13:08] <nbd> the build system should merge everything and then put it in a file
[13:08] <florian___> how would he determine the architecture specifi values ?
[13:08] <nbd> and maybe instead of loading it from the shell it should set the appropriate env var
[13:08] <nbd> the
[13:09] <nbd> just a sec..
[13:10] <nbd> http://pastebin.ca/76286
[13:11] <florian___> that's good in a package makefile
[13:11] <florian___> what about the definition in include/package.mk ?
[13:12] <nbd> they define Build/Configure/Sitedefs with the contents of .../default
[13:13] <florian___> I was thinking of smthg like that http://pastebin.ca/76288
[13:15] <nbd> hmm... yeah
[13:15] <nbd> probably better
[13:15] <nbd> what's with the $(1)
[13:15] <nbd> ah, wait
[13:15] <florian___> your custom sitedefs
[13:15] <florian___> but your method is better
[13:15] <nbd> we should combine both
[13:16] <florian___> lol that's exactly what I wanted to write :)
[13:16] <nbd> use $(TOPDIR)/include/site/
[13:16] <nbd> sorry
[13:16] <nbd> use $(TOPDIR)/include/site/$(ARCH).def
[13:16] <nbd> and then add the contents of Build/Configure/Sitedefs
[13:16] <nbd> which will only get added if the user overrides it
[13:17] <nbd> and the contents of that file should be simple name=value lines without \ or indentation
[13:17] <nbd> s/indentation/forced indentation/
[13:18] <florian___> right
[13:26] <florian___> does this look good : http://pastebin.ca/76290
[13:28] <florian___> I forgot a trainling slash in Build/Configure/Sitedefs
[13:35] <florian___> nbd: is this ok to you ?
[13:38] <nbd> i think that would not work
[13:39] <nbd> because you're mixing the cat command with the sitedefs
[13:49] <florian___> does "use" exist ?
[14:04] <nbd> use?
[14:06] <florian___> well I could I read settings from a file ?
[14:07] <florian___> include <file> ?
[14:07] <nbd> you can't arbitrarily put the contents of a file at a certain place
[14:08] <florian___> I need to parse it right ?
[14:11] <nbd> you should do this: $(CP) $(TOPDIR)/include/site/$(ARCH).def $(PKG_BUILD_DIR)/site.def
[14:11] <nbd> and then append the contents of Build/Configure/Sitedefs
[14:12] <nbd> and finally set the environment variable that tells configure to use this file
[14:19] <florian___> how long can it take for jffs2 to erase all blocks after the end markers ?
[14:21] <florian___> oh it's finished :)
[14:29] <florian___> there is no more /etc/preinit in buildroot-ng images ?
[14:40] <florian___> nbd: is this better ? http://pastebin.ca/76315
[14:45] <nbd> /etc/preinit is now target-specific
[14:45] <florian___> ok, I will adapt it for au1000
[14:45] <nbd> that's not quite what i meant. i wanted to get rid of the extra configure parameter entirely
[14:45] <nbd> (pastebin)
[14:46] <nbd> i wanted the user to put in the extra definitions inside the Build/Configure/Sitedefs template directly
[14:48] <florian___> ok, so I just $(call Build/Configure/Sitedefs) inside Build/Configure/Default without arguments ?
[14:48] <nbd> yes
[14:48] <nbd> well, not exactly
[14:49] <florian___> ah :)
[14:49] <nbd> the contents of Build/Configure/Sitedefs should be appended to the site.def file
[14:50] <florian___> is not $(1) >> $(PKG_BUILD_DIR)/site.def doing it ?
[14:55] <nbd> no
[14:55] <nbd> i have to go now... have a look at how it's done through the environment with preinst/postinst/etc.
[14:55] <florian___> ok
[15:23] <malbon> printfprintf/win 4
[15:23] <malbon> oops
[23:15] <crazy_imp> has someone tried to use a cd-drive in a usb-box with kamikaze? which lines i have to add in the kernel-config?
[00:00] --- Sun Jul 2 2006