[00:00] <{Nico}> groz: yup
[00:00] <groz> just a note for you on the porting question yesterday
[00:00] <groz> example, wpa_supplicant, I didn't actually 'port' anything, started from scratch, and it's necessary for -ng
[00:00] <groz> because of supporting lots of hw environments
[00:00] <groz> that one is very hw specific
[00:01] <groz> and gonna need a lot of tweaking for differences in platforms
[00:01] <{Nico}> groz: my point was just to use the svn cp command to keep cvs/svn history in -ng
[00:02] <{Nico}> i made a quick note in the OpenWrtPackageToDo
[00:02] <groz> ok, but my point was, nothing was copied :)
[00:02] <groz> well, not completely true, i did steal the makefile patch
[00:02] <groz> for ldflags
[00:03] <groz> but will do on more
[00:03] <groz> right now, i'm fighting with some more basic issues, getting x86-2.6 to actually boot and run off a memory stick on my via
[00:04] <groz> for some reason, i cannot for the life of me actually get it to mount and use the usb stick with 2.6
[00:04] <{Nico}> groz: it doesn't mind if you change everything, the important point is keep history
[00:05] <{Nico}> groz: should uml-2.6 work out-of-the-box?
[00:05] <groz> if you do a full clean build
[00:05] <groz> yup
[00:05] <groz> i added a quick readme
[00:05] <groz> that readme gets you to the point of uml up, no network
[00:05] <groz> now, i'm gonna do another change later, maybe today yet
[00:05] <groz> and remove the dependancy on an xterm
[00:06] <groz> so it can be started and run on a headless box
[00:06] <{Nico}> mine is stopping with "mount_devfs_fs(): unable to mount devfs, err: -2"
[00:06] <groz> but, probably somewhere along the line
[00:07] <groz> is it kicking off the xterm for tts0
[00:07] <{Nico}> groz: nope
[00:07] <groz> couple of things, are you running it on a box with x
[00:07] <groz> are you running as a user, or root ?
[00:07] <groz> there may be a permissions problem
[00:07] <groz> just run it as a normal user
[00:07] <{Nico}> ubuntu with compiz as a normal user
[00:07] <groz> ok
[00:08] <groz> i'm finding all sorts of little things when i put the 2.6 stuff onto the via board too
[00:08] <groz> i'll sort my way thru them
[00:08] <groz> one of them, the default config is hard wired to a geode setup
[00:08] <groz> im debating making some wholesale changes in defaults for x86-2.6
[00:09] <groz> so that its very generic
[00:09] <groz> and will come up and run on 'just about anything'
[00:09] <groz> with or without console, with or without serial ports
[00:09] <groz> etc
[00:10] <groz> humor me with that uml build
[00:10] <groz> try this
[00:10] <groz> configure for ramfs
[00:10] <groz> then run it without the reference to the ext filesystem
[00:11] <groz> actually, before you do that
[00:11] <groz> when you run it
[00:11] <groz> are you taking the command line out of the readme I added to it ?
[00:11] <groz> if you dont have the reference to the ubd device on the command line
[00:11] <groz> you will likely get that error
[00:12] <{Nico}> yep, i took the cmdline
[00:12] <groz> hmm thats very indicative of not having the root fs
[00:12] <groz> properly set
[00:13] <groz> i'm gonna fuss with uml some more once i have this via up and going correctly
[00:13] <groz> get rid of a few more issues in there
[00:13] <groz> but, h3 was able to get it going last nite
[00:13] <groz> basically strait out of the box
[00:16] <{Nico}> i added a vanilla genext2fs to the toolchain and use it to build the ext2 image
[00:16] <groz> ahhh, i already had it on my machine
[00:17] <groz> ok, humor me
[00:17] <groz> compile it now using ramfs
[00:17] <groz> run without that reference to the ubd
[00:17] <groz> and see if it comes up ok like that
[00:17] <groz> you should be able to 'just run it' from the command line
[00:17] <groz> you will have to dig down into the build_i386/linux-2.6xxxxx directory to find it
[00:18] <{Nico}> btw, is there a way to stop a panic-ed uml without killing it from another term?
[00:18] <groz> actually, that one is not panicd
[00:18] <groz> its running
[00:18] <groz> just no io on that terminal
[00:18] <groz> and it's trying to talk to an xterm
[00:18] <groz> for tts0
[00:18] <groz> but ya, kill it from another terminal
[00:18] <groz> then, here's another thing
[00:19] <groz> parameters
[00:19] <{Nico}> ok, i'll kill it and try one with ramfs
[00:19] <groz> hang on
[00:19] <groz> wait, i got another way to try something
[00:19] <groz> gimme just a sec
[00:19] <groz> con0=null tts0=fd:0,fd:1
[00:19] <groz> no
[00:19] <groz> i got that wrong
[00:19] <groz> it's not tts0
[00:20] <groz> lemme dig that back up
[00:20] <groz> ssl0=fd:0,fd:1
[00:21] <groz> then the dead display you see goes to null, and the serial console it's looking for comes in the term you started it from
[00:21] <groz> i'm gonna change the defaults so it works that way
[00:23] <{Nico}> same, ends up with "Warning: unable to open an initial console."
[00:23] <groz> ok, i'll dig on that shortly
[00:23] <groz> i want that one fixed up so it 'just works' in all cases
[00:29] <CIA-17> nico * r4220 /packages/libs/libpng/ (. Config.in Makefile ipkg/): port libpng to buildroot-ng
[00:29] <groz> wow, you are really going to town on libs
[00:29] <groz> is libupnp high on your list ?
[00:29] <groz> i'm gonna get to it, but, not till i'm finished with some of these target details
[00:30] <{Nico}> groz: libupnp is in my bunch-of-commits of the day, should be there in a few minutes
[00:31] <groz> oh, i see you did put your name on it
[00:31] <groz> wasn't there last nite when i went to bed
[00:32] <{Nico}> i'm focusing on libs atm, but tell me if you need something else
[00:33] <groz> if you are doing libs, then i'll just stick to making some targets buildable
[00:33] <groz> so they actually build and run
[00:34] <groz> i just did a checkout and started a build of a vanilla uml set
[00:35] <groz> should be able to try replicate your problem here shortly
[00:36] <CIA-17> nico * r4221 /packages/libs/gd/ (. Config.in Makefile ipkg/): port (lib)gd to buildroot-ng
[00:36] <groz> I've actually got another target that i use quite a bit here, but wondering if the folks here can 'appreciate' it
[00:36] <groz> and that's colinux
[00:36] <groz> i use a really 'different' setup on my notebook when i'm travelling
[00:37] <groz> i have it set up so that colinux is the only thing that can actually route thru the wifi adapter
[00:37] <groz> then it routes back to the basic windows setup thru a tap
[00:37] <groz> so i've got windows behind linux, on a single machine
[00:37] <groz> when i connect to public networks
[00:37] <groz> on the road
[00:37] <groz> it's slick, colinux has 2 taps
[00:38] <groz> one of them on the windows side is attached to the bridge, as is the wifi adapter
[00:38] <groz> there is NO ip on the bridge
[00:38] <groz> colinux can then dhcp thru that via the wireless
[00:38] <groz> and create a firewall/set of routes for the windows thru the other tap
[00:39] <nbd> openwrt on colinux could be interesting
[00:39] <groz> i'm doing it right now with a 'hand rolled' setup
[00:39] <groz> but considering just tossing in an openwrt file system
[00:39] <groz> it allows me to do things like 'my own' kind of ipsec setup
[00:39] <groz> etc
[00:40] Action: {Nico} is googling on colinux
[00:40] <groz> essentially i've got windows behind a linux system on a single box that way
[00:40] <groz> colinux = uml for windows
[00:40] <groz> in a nutshell
[00:41] <groz> I use colinux a lot
[00:41] <groz> example, need to put a small lamp setup on a customer site, and they are a very windows centric shop
[00:41] <nbd> if i was using windows, i'd probably use colinux as well
[00:41] <groz> just install it as a colinux on thier server, call it a 'windows service'
[00:41] <groz> and voila, i got a small linux server in thier data center
[00:42] <groz> without forcing a linux computer on them
[00:42] <groz> they just see it as 'yet another service' on a windows box
[00:42] <groz> in the long run, i'm very os agnostic, try use the best tool for any job
[00:42] <nbd> it would be cool to have a package to hand out to windows users so that they can have decent routing even though they're using windows
[00:42] <groz> often, best is defined as 'what the customer is comfortable with'
[00:43] <groz> and if colinux on windows keeps them comfy
[00:43] <groz> i'll go that route
[00:43] <groz> then to use openwrt that way, need to set up some extra taps
[00:43] <groz> but it's not hard
[00:43] <groz> then bridge an internet facing adapter with one tap
[00:43] <groz> remove all protocols from the bridge
[00:43] <groz> voila, colinux can dhcp from the world
[00:44] <groz> and no ip traffic reaches windows without first passing thru the colinux setup
[00:44] <groz> the only real windows exposure left at that point is windows bridge, and the actual device driver for the wifi
[00:45] <CIA-17> nico * r4222 /packages/libs/libupnp/ (. Config.in Makefile ipkg/): port libupnp to buildroot-ng
[00:45] <groz> so here's another thing i'm planning on doing, but have not got to it yet, soon tho
[00:45] <groz> a upnp client for openwrt
[00:45] <groz> works like this
[00:46] <groz> if the router is behind yet another router
[00:46] <groz> it can accept upnp requests from the lan, forward them
[00:46] <groz> then build the passthru rules as required
[00:46] <groz> where that'll be interesting
[00:46] <groz> it will allow one to set up routers internally offsite
[00:46] <groz> and by eploiting upstream upnp
[00:46] <groz> you can build ssh tunnels into them
[00:47] <groz> to manage them
[00:47] <groz> and selectivelly allow clients to do upnp
[00:47] <groz> so you can accomodate brain dead xbox clients
[00:47] <groz> without compromising everything behind that router
[00:48] <groz> ie a methodology of saying 'ok, this client can do upnp, just enough to make an xbox happy'
[00:48] <groz> but exclude it from the rest of them
[00:52] <groz> hmm, looks like i got kmod breakage now in the uml build
[00:53] <CIA-17> nico * r4223 /packages/libs/libmad/ (. Config.in Makefile ipkg/): port libmad to buildroot-ng
[00:56] <CIA-17> nico * r4224 /packages/libs/libnids/ (. Config.in Makefile ipkg/): port libnids to buildroot-ng
[01:00] <CIA-17> nico * r4225 /packages/libs/libtasn1/ (. Config.in Makefile ipkg/): port libtasn1 to buildroot-ng
[01:04] <CIA-17> nico * r4226 /packages/libs/libvorbisidec/ (. Config.in Makefile ipkg/): port libvorbisidec to buildroot-ng
[01:07] <groz> ok, i see the issue
[01:07] <groz> somebody 'ported over nefilter module packaging'
[01:45] <nbd> groz: i did. did i introduce any annoying bugs or it it just because of the config related change?
[01:45] <groz> well, give me a bit
[01:46] <groz> i just went back to my 4202 revision
[01:46] <groz> that did build clean
[01:46] <groz> and then i'll see
[01:46] <groz> i've been pounding on a lot of stuff here, and i want a baseline to compare to here first
[01:46] <groz> before I scream :)
[01:46] <groz> but what i see happening, it's not even building iptables right now
[01:47] <groz> then complainging and puking when a pile of modules are missing
[01:47] <groz> so let me figure out if that's an ARCH related thing
[01:47] <groz> or if it really did break
[01:47] <groz> the uml build actually does use ARCH variables a little different than some others
[01:48] <groz> host and target are NOT interchangeable
[01:48] <groz> and, TARGET and KARCH are also not interchangeable
[01:49] <groz> i got a sneaking suspicion this may get messy
[01:50] <groz> but i got a question for you ndb
[01:50] <groz> totally different subject
[01:50] <groz> the x86-2.6 vanilla out of the box setup
[01:50] <groz> is specifically targetted to a geode
[01:50] <groz> wont even boot on a via itx board
[01:51] <groz> should that target be cleaned up to be more generic, and boot on just about anything ?
[01:51] <groz> i'm thinking it should, but, ther may be a reason i'm not aware of that it's specifically geode targetted
[02:06] <crazy_imp> groz: ever tried to use the uml files with xen? or isn't that possilbe?
[02:06] <groz> i've never looked at xen
[02:06] <groz> not even sure whaqt it does :)
[02:07] <{Nico}> groz: x86-2.6 used to be soekris-2.6, that explains why it's geode centric
[02:07] <groz> yah, i have a routerboard that's a geode too
[02:07] <groz> but i have a considerable number of targets here
[02:07] <groz> what i'm considering doing is this
[02:07] <groz> rework the configs in the x86-2.6 to be very generic
[02:07] <groz> then, add in board specific options
[02:07] <crazy_imp> groz: virtual machines, i would like to feed them with openwrt :)
[02:08] <groz> basically a 'boards' submenu
[02:09] <groz> it should be relatively trivial to build a xen target
[02:09] <groz> does it bascially take a disk image, then boot it in a virtual machine ?
[02:10] <crazy_imp> yepp, it needs a kernel with some xen things inside and it takes a img file as disk or a real disk
[02:10] <groz> ok, then we shyould be able to create a xen target just like the uml target
[02:10] <groz> and add the xen patches to it's kernel
[02:11] <groz> ok, nbd, my 4202 revision 'just builds and goes'
[02:11] <groz> now i'll start looking at what the differences are
[02:16] <groz> so Nico, if i commit a bunch of changes that convert it from 'soekris centric' to 'general purpose' that would probably be ok then
[02:18] <{Nico}> groz: i guess
[02:18] <groz> what i would _like_ to end up with
[02:18] <groz> is a board submenu at configure time
[02:18] <groz> so there is a 'general purpose' that'll boot on just about anything
[02:18] <groz> and then board optimized variants
[02:19] <{Nico}> groz: sounds good
[02:19] <groz> not quite sure the 'right' way to get there yet
[02:19] <groz> but give me a little time
[02:20] <groz> probably something along this line
[02:20] <groz> a 'master' config file
[02:20] <groz> then some board specific deltas that get merged in
[02:21] <groz> ok, for your question on the uml earlier
[02:21] <groz> i've just rebuilt it, without the recent commits on kmod stuff
[02:21] <groz> and it 'just ran'
[02:21] <groz> but, if you are not getting the xterms
[02:21] <groz> just add this to the command line
[02:22] <groz> con0=null ssl0=fd:0,fd:1
[02:22] <groz> and once it starts , you should get the 'press enter to activate this console'
[02:23] <{Nico}> i'm rebuilding my uml tree atm
[02:23] <groz> I'm not sure how you folks manage multiple trees, but i'm currently working with 4
[02:24] <groz> what i did, i checked out packages into yet another spot
[02:24] <groz> then wrote a little script to soft link aqll the packages into each tree
[02:24] <groz> works well for me, i svn up the packages tree after your commits, run that script
[02:24] <groz> and like magic
[02:24] <groz> they show up in all my trees
[02:25] <groz> is thast kinda how you are dong it too, or are you getting fresh copies of packages into each tree
[02:25] <{Nico}> groz: nope, symlinks like you do but i only port in one tree atm
[02:26] <groz> well as soon as I've got this via booting clean, and the routerboard
[02:26] <groz> i'm gonna have a whole big mess of hardware to test various things on
[02:27] <groz> I've got every variant of wrt54g ever made here
[02:27] <groz> and a big stack of various x86 target boards
[11:37] <groz> kaloz, are you around ?
[11:38] <groz> are you working on the aetheros stuff ?
[11:39] <Kaloz> yeah
[11:40] <groz> have you got a 2 line summary for me of where it's at ?
[11:40] <groz> dont wanna be a pest, but, i'm curious
[11:40] <groz> there are some very interesting boards using that
[11:40] <Kaloz> two lines? let's see, what about
[11:40] <Kaloz> it's still
[11:40] <Kaloz> work in progress? ;)
[11:40] <Kaloz> btw, msg
[12:04] <murb> more reliable if you don't use the virtual ssid stuff.
[12:09] <groz> hey
[12:09] <groz> is anything reliable with the virtual ssid stuff ?
[12:11] <murb> well i switch it on and it tends to lock up my wgt634u
[12:12] <murb> probably i need to try a newer madwifi snapshot
[12:20] <groz> ok folks i gotta go get some zzzz time
[12:20] <groz> will be back later tomorrow
[13:49] <CIA-17> nico * r4227 /packages/libs/glib/ (. Config.in Makefile ipkg/): port glib to buildroot-ng
[14:11] <CIA-17> nico * r4228 /packages/libs/gnutls/ (. Config.in Makefile ipkg/): port gnutls to buildroot-ng
[14:14] <CIA-17> nico * r4229 /packages/libs/flac/ (. Config.in Makefile ipkg/): port flac to buildroot-ng
[14:16] <CIA-17> nico * r4230 /packages/libs/gsm/ (. Config.in Makefile ipkg/): port gsm to buildroot-ng
[14:59] <CIA-17> nico * r4231 /packages/libs/id3lib/ (. Config.in Makefile ipkg/): port id3lib to buildroot-ng
[15:03] <CIA-17> nico * r4232 /packages/libs/libid3tag/ (. Config.in Makefile ipkg/): port libid3tag to buildroot-ng
[15:05] <CIA-17> nico * r4233 /packages/libs/libmcrypt/ (. Config.in Makefile ipkg/): port libmcrypt to buildroot-ng
[15:08] <CIA-17> nico * r4234 /packages/libs/libol/ (. Config.in Makefile ipkg/): port libol to buildroot-ng
[15:11] <CIA-17> nico * r4235 /packages/libs/uclibc++/Makefile:
[15:11] <CIA-17> fix missing shared lib in package and lib path,
[15:11] <CIA-17> remove unneeded PKG_INSTALL_DIR, indent package desc.
[16:21] <CIA-17> nbd * r4236 /branches/buildroot-ng/openwrt/package/kernel/ (Makefile modules.mk): bugfixes for the kernel module packaging
[17:16] <CIA-17> nico * r4237 /packages/libs/ustl/ (. Config.in Makefile ipkg/): port ustl to buildroot-ng
[17:24] <CIA-17> nico * r4238 /packages/libs/libdaemon/Makefile: fix $$ escape
[19:11] <CIA-17> nico * r4239 /packages/libs/avahi/ (. Config.in Makefile ipkg/): port avahi to buildroot-ng
[19:16] <CIA-17> nico * r4240 /packages/libs/howl/ (. Config.in Makefile ipkg/): port howl to buildroot-ng
[19:19] <CIA-17> nico * r4241 /packages/libs/net-snmp/ (. Config.in Makefile ipkg/): port net-snmp to buildroot-ng
[19:20] <nbd> {Nico}: ping
[19:42] <CIA-17> nico * r4242 /packages/libs/radiusclient-ng/ (. Config.in Makefile ipkg/): port radiusclient-ng to buildroot-ng
[19:43] <{Nico}> nbd: pong
[19:45] <db90h> don't play games in here
[19:45] <db90h> or at least not ping pong, you might hit someone with the ball
[19:47] <{Nico}> db90h: it won't hurt, just wake up those sleepy channel lurkers :)
[19:48] <CIA-17> nico * r4243 /packages/libs/speex/ (. Config.in Makefile ipkg/): port speex to buildroot-ng
[19:49] <db90h> hehe
[19:52] <nbd> {Nico}: do avahi, howl and net-snmp really belong in /packages/libs?
[19:57] <{Nico}> nbd: i put those here because their implementations are based on libraries
[19:58] <{Nico}> look at net-snmp, i put the client utils and daemons in Network
[20:00] <{Nico}> nbd: i'm not sure how it must be done, feel free to move them around
[20:07] <nbd> no idea. i guess we can leave them there for now
[00:00] --- Mon Jul 24 2006
[00:00] <groz> just a note for you on the porting question yesterday
[00:00] <groz> example, wpa_supplicant, I didn't actually 'port' anything, started from scratch, and it's necessary for -ng
[00:00] <groz> because of supporting lots of hw environments
[00:00] <groz> that one is very hw specific
[00:01] <groz> and gonna need a lot of tweaking for differences in platforms
[00:01] <{Nico}> groz: my point was just to use the svn cp command to keep cvs/svn history in -ng
[00:02] <{Nico}> i made a quick note in the OpenWrtPackageToDo
[00:02] <groz> ok, but my point was, nothing was copied :)
[00:02] <groz> well, not completely true, i did steal the makefile patch
[00:02] <groz> for ldflags
[00:03] <groz> but will do on more
[00:03] <groz> right now, i'm fighting with some more basic issues, getting x86-2.6 to actually boot and run off a memory stick on my via
[00:04] <groz> for some reason, i cannot for the life of me actually get it to mount and use the usb stick with 2.6
[00:04] <{Nico}> groz: it doesn't mind if you change everything, the important point is keep history
[00:05] <{Nico}> groz: should uml-2.6 work out-of-the-box?
[00:05] <groz> if you do a full clean build
[00:05] <groz> yup
[00:05] <groz> i added a quick readme
[00:05] <groz> that readme gets you to the point of uml up, no network
[00:05] <groz> now, i'm gonna do another change later, maybe today yet
[00:05] <groz> and remove the dependancy on an xterm
[00:06] <groz> so it can be started and run on a headless box
[00:06] <{Nico}> mine is stopping with "mount_devfs_fs(): unable to mount devfs, err: -2"
[00:06] <groz> but, probably somewhere along the line
[00:07] <groz> is it kicking off the xterm for tts0
[00:07] <{Nico}> groz: nope
[00:07] <groz> couple of things, are you running it on a box with x
[00:07] <groz> are you running as a user, or root ?
[00:07] <groz> there may be a permissions problem
[00:07] <groz> just run it as a normal user
[00:07] <{Nico}> ubuntu with compiz as a normal user
[00:07] <groz> ok
[00:08] <groz> i'm finding all sorts of little things when i put the 2.6 stuff onto the via board too
[00:08] <groz> i'll sort my way thru them
[00:08] <groz> one of them, the default config is hard wired to a geode setup
[00:08] <groz> im debating making some wholesale changes in defaults for x86-2.6
[00:09] <groz> so that its very generic
[00:09] <groz> and will come up and run on 'just about anything'
[00:09] <groz> with or without console, with or without serial ports
[00:09] <groz> etc
[00:10] <groz> humor me with that uml build
[00:10] <groz> try this
[00:10] <groz> configure for ramfs
[00:10] <groz> then run it without the reference to the ext filesystem
[00:11] <groz> actually, before you do that
[00:11] <groz> when you run it
[00:11] <groz> are you taking the command line out of the readme I added to it ?
[00:11] <groz> if you dont have the reference to the ubd device on the command line
[00:11] <groz> you will likely get that error
[00:12] <{Nico}> yep, i took the cmdline
[00:12] <groz> hmm thats very indicative of not having the root fs
[00:12] <groz> properly set
[00:13] <groz> i'm gonna fuss with uml some more once i have this via up and going correctly
[00:13] <groz> get rid of a few more issues in there
[00:13] <groz> but, h3 was able to get it going last nite
[00:13] <groz> basically strait out of the box
[00:16] <{Nico}> i added a vanilla genext2fs to the toolchain and use it to build the ext2 image
[00:16] <groz> ahhh, i already had it on my machine
[00:17] <groz> ok, humor me
[00:17] <groz> compile it now using ramfs
[00:17] <groz> run without that reference to the ubd
[00:17] <groz> and see if it comes up ok like that
[00:17] <groz> you should be able to 'just run it' from the command line
[00:17] <groz> you will have to dig down into the build_i386/linux-2.6xxxxx directory to find it
[00:18] <{Nico}> btw, is there a way to stop a panic-ed uml without killing it from another term?
[00:18] <groz> actually, that one is not panicd
[00:18] <groz> its running
[00:18] <groz> just no io on that terminal
[00:18] <groz> and it's trying to talk to an xterm
[00:18] <groz> for tts0
[00:18] <groz> but ya, kill it from another terminal
[00:18] <groz> then, here's another thing
[00:19] <groz> parameters
[00:19] <{Nico}> ok, i'll kill it and try one with ramfs
[00:19] <groz> hang on
[00:19] <groz> wait, i got another way to try something
[00:19] <groz> gimme just a sec
[00:19] <groz> con0=null tts0=fd:0,fd:1
[00:19] <groz> no
[00:19] <groz> i got that wrong
[00:19] <groz> it's not tts0
[00:20] <groz> lemme dig that back up
[00:20] <groz> ssl0=fd:0,fd:1
[00:21] <groz> then the dead display you see goes to null, and the serial console it's looking for comes in the term you started it from
[00:21] <groz> i'm gonna change the defaults so it works that way
[00:23] <{Nico}> same, ends up with "Warning: unable to open an initial console."
[00:23] <groz> ok, i'll dig on that shortly
[00:23] <groz> i want that one fixed up so it 'just works' in all cases
[00:29] <CIA-17> nico * r4220 /packages/libs/libpng/ (. Config.in Makefile ipkg/): port libpng to buildroot-ng
[00:29] <groz> wow, you are really going to town on libs
[00:29] <groz> is libupnp high on your list ?
[00:29] <groz> i'm gonna get to it, but, not till i'm finished with some of these target details
[00:30] <{Nico}> groz: libupnp is in my bunch-of-commits of the day, should be there in a few minutes
[00:31] <groz> oh, i see you did put your name on it
[00:31] <groz> wasn't there last nite when i went to bed
[00:32] <{Nico}> i'm focusing on libs atm, but tell me if you need something else
[00:33] <groz> if you are doing libs, then i'll just stick to making some targets buildable
[00:33] <groz> so they actually build and run
[00:34] <groz> i just did a checkout and started a build of a vanilla uml set
[00:35] <groz> should be able to try replicate your problem here shortly
[00:36] <CIA-17> nico * r4221 /packages/libs/gd/ (. Config.in Makefile ipkg/): port (lib)gd to buildroot-ng
[00:36] <groz> I've actually got another target that i use quite a bit here, but wondering if the folks here can 'appreciate' it
[00:36] <groz> and that's colinux
[00:36] <groz> i use a really 'different' setup on my notebook when i'm travelling
[00:37] <groz> i have it set up so that colinux is the only thing that can actually route thru the wifi adapter
[00:37] <groz> then it routes back to the basic windows setup thru a tap
[00:37] <groz> so i've got windows behind linux, on a single machine
[00:37] <groz> when i connect to public networks
[00:37] <groz> on the road
[00:37] <groz> it's slick, colinux has 2 taps
[00:38] <groz> one of them on the windows side is attached to the bridge, as is the wifi adapter
[00:38] <groz> there is NO ip on the bridge
[00:38] <groz> colinux can then dhcp thru that via the wireless
[00:38] <groz> and create a firewall/set of routes for the windows thru the other tap
[00:39] <nbd> openwrt on colinux could be interesting
[00:39] <groz> i'm doing it right now with a 'hand rolled' setup
[00:39] <groz> but considering just tossing in an openwrt file system
[00:39] <groz> it allows me to do things like 'my own' kind of ipsec setup
[00:39] <groz> etc
[00:40] Action: {Nico} is googling on colinux
[00:40] <groz> essentially i've got windows behind a linux system on a single box that way
[00:40] <groz> colinux = uml for windows
[00:40] <groz> in a nutshell
[00:41] <groz> I use colinux a lot
[00:41] <groz> example, need to put a small lamp setup on a customer site, and they are a very windows centric shop
[00:41] <nbd> if i was using windows, i'd probably use colinux as well
[00:41] <groz> just install it as a colinux on thier server, call it a 'windows service'
[00:41] <groz> and voila, i got a small linux server in thier data center
[00:42] <groz> without forcing a linux computer on them
[00:42] <groz> they just see it as 'yet another service' on a windows box
[00:42] <groz> in the long run, i'm very os agnostic, try use the best tool for any job
[00:42] <nbd> it would be cool to have a package to hand out to windows users so that they can have decent routing even though they're using windows
[00:42] <groz> often, best is defined as 'what the customer is comfortable with'
[00:43] <groz> and if colinux on windows keeps them comfy
[00:43] <groz> i'll go that route
[00:43] <groz> then to use openwrt that way, need to set up some extra taps
[00:43] <groz> but it's not hard
[00:43] <groz> then bridge an internet facing adapter with one tap
[00:43] <groz> remove all protocols from the bridge
[00:43] <groz> voila, colinux can dhcp from the world
[00:44] <groz> and no ip traffic reaches windows without first passing thru the colinux setup
[00:44] <groz> the only real windows exposure left at that point is windows bridge, and the actual device driver for the wifi
[00:45] <CIA-17> nico * r4222 /packages/libs/libupnp/ (. Config.in Makefile ipkg/): port libupnp to buildroot-ng
[00:45] <groz> so here's another thing i'm planning on doing, but have not got to it yet, soon tho
[00:45] <groz> a upnp client for openwrt
[00:45] <groz> works like this
[00:46] <groz> if the router is behind yet another router
[00:46] <groz> it can accept upnp requests from the lan, forward them
[00:46] <groz> then build the passthru rules as required
[00:46] <groz> where that'll be interesting
[00:46] <groz> it will allow one to set up routers internally offsite
[00:46] <groz> and by eploiting upstream upnp
[00:46] <groz> you can build ssh tunnels into them
[00:47] <groz> to manage them
[00:47] <groz> and selectivelly allow clients to do upnp
[00:47] <groz> so you can accomodate brain dead xbox clients
[00:47] <groz> without compromising everything behind that router
[00:48] <groz> ie a methodology of saying 'ok, this client can do upnp, just enough to make an xbox happy'
[00:48] <groz> but exclude it from the rest of them
[00:52] <groz> hmm, looks like i got kmod breakage now in the uml build
[00:53] <CIA-17> nico * r4223 /packages/libs/libmad/ (. Config.in Makefile ipkg/): port libmad to buildroot-ng
[00:56] <CIA-17> nico * r4224 /packages/libs/libnids/ (. Config.in Makefile ipkg/): port libnids to buildroot-ng
[01:00] <CIA-17> nico * r4225 /packages/libs/libtasn1/ (. Config.in Makefile ipkg/): port libtasn1 to buildroot-ng
[01:04] <CIA-17> nico * r4226 /packages/libs/libvorbisidec/ (. Config.in Makefile ipkg/): port libvorbisidec to buildroot-ng
[01:07] <groz> ok, i see the issue
[01:07] <groz> somebody 'ported over nefilter module packaging'
[01:45] <nbd> groz: i did. did i introduce any annoying bugs or it it just because of the config related change?
[01:45] <groz> well, give me a bit
[01:46] <groz> i just went back to my 4202 revision
[01:46] <groz> that did build clean
[01:46] <groz> and then i'll see
[01:46] <groz> i've been pounding on a lot of stuff here, and i want a baseline to compare to here first
[01:46] <groz> before I scream :)
[01:46] <groz> but what i see happening, it's not even building iptables right now
[01:47] <groz> then complainging and puking when a pile of modules are missing
[01:47] <groz> so let me figure out if that's an ARCH related thing
[01:47] <groz> or if it really did break
[01:47] <groz> the uml build actually does use ARCH variables a little different than some others
[01:48] <groz> host and target are NOT interchangeable
[01:48] <groz> and, TARGET and KARCH are also not interchangeable
[01:49] <groz> i got a sneaking suspicion this may get messy
[01:50] <groz> but i got a question for you ndb
[01:50] <groz> totally different subject
[01:50] <groz> the x86-2.6 vanilla out of the box setup
[01:50] <groz> is specifically targetted to a geode
[01:50] <groz> wont even boot on a via itx board
[01:51] <groz> should that target be cleaned up to be more generic, and boot on just about anything ?
[01:51] <groz> i'm thinking it should, but, ther may be a reason i'm not aware of that it's specifically geode targetted
[02:06] <crazy_imp> groz: ever tried to use the uml files with xen? or isn't that possilbe?
[02:06] <groz> i've never looked at xen
[02:06] <groz> not even sure whaqt it does :)
[02:07] <{Nico}> groz: x86-2.6 used to be soekris-2.6, that explains why it's geode centric
[02:07] <groz> yah, i have a routerboard that's a geode too
[02:07] <groz> but i have a considerable number of targets here
[02:07] <groz> what i'm considering doing is this
[02:07] <groz> rework the configs in the x86-2.6 to be very generic
[02:07] <groz> then, add in board specific options
[02:07] <crazy_imp> groz: virtual machines, i would like to feed them with openwrt :)
[02:08] <groz> basically a 'boards' submenu
[02:09] <groz> it should be relatively trivial to build a xen target
[02:09] <groz> does it bascially take a disk image, then boot it in a virtual machine ?
[02:10] <crazy_imp> yepp, it needs a kernel with some xen things inside and it takes a img file as disk or a real disk
[02:10] <groz> ok, then we shyould be able to create a xen target just like the uml target
[02:10] <groz> and add the xen patches to it's kernel
[02:11] <groz> ok, nbd, my 4202 revision 'just builds and goes'
[02:11] <groz> now i'll start looking at what the differences are
[02:16] <groz> so Nico, if i commit a bunch of changes that convert it from 'soekris centric' to 'general purpose' that would probably be ok then
[02:18] <{Nico}> groz: i guess
[02:18] <groz> what i would _like_ to end up with
[02:18] <groz> is a board submenu at configure time
[02:18] <groz> so there is a 'general purpose' that'll boot on just about anything
[02:18] <groz> and then board optimized variants
[02:19] <{Nico}> groz: sounds good
[02:19] <groz> not quite sure the 'right' way to get there yet
[02:19] <groz> but give me a little time
[02:20] <groz> probably something along this line
[02:20] <groz> a 'master' config file
[02:20] <groz> then some board specific deltas that get merged in
[02:21] <groz> ok, for your question on the uml earlier
[02:21] <groz> i've just rebuilt it, without the recent commits on kmod stuff
[02:21] <groz> and it 'just ran'
[02:21] <groz> but, if you are not getting the xterms
[02:21] <groz> just add this to the command line
[02:22] <groz> con0=null ssl0=fd:0,fd:1
[02:22] <groz> and once it starts , you should get the 'press enter to activate this console'
[02:23] <{Nico}> i'm rebuilding my uml tree atm
[02:23] <groz> I'm not sure how you folks manage multiple trees, but i'm currently working with 4
[02:24] <groz> what i did, i checked out packages into yet another spot
[02:24] <groz> then wrote a little script to soft link aqll the packages into each tree
[02:24] <groz> works well for me, i svn up the packages tree after your commits, run that script
[02:24] <groz> and like magic
[02:24] <groz> they show up in all my trees
[02:25] <groz> is thast kinda how you are dong it too, or are you getting fresh copies of packages into each tree
[02:25] <{Nico}> groz: nope, symlinks like you do but i only port in one tree atm
[02:26] <groz> well as soon as I've got this via booting clean, and the routerboard
[02:26] <groz> i'm gonna have a whole big mess of hardware to test various things on
[02:27] <groz> I've got every variant of wrt54g ever made here
[02:27] <groz> and a big stack of various x86 target boards
[11:37] <groz> kaloz, are you around ?
[11:38] <groz> are you working on the aetheros stuff ?
[11:39] <Kaloz> yeah
[11:40] <groz> have you got a 2 line summary for me of where it's at ?
[11:40] <groz> dont wanna be a pest, but, i'm curious
[11:40] <groz> there are some very interesting boards using that
[11:40] <Kaloz> two lines? let's see, what about
[11:40] <Kaloz> it's still
[11:40] <Kaloz> work in progress? ;)
[11:40] <Kaloz> btw, msg
[12:04] <murb> more reliable if you don't use the virtual ssid stuff.
[12:09] <groz> hey
[12:09] <groz> is anything reliable with the virtual ssid stuff ?
[12:11] <murb> well i switch it on and it tends to lock up my wgt634u
[12:12] <murb> probably i need to try a newer madwifi snapshot
[12:20] <groz> ok folks i gotta go get some zzzz time
[12:20] <groz> will be back later tomorrow
[13:49] <CIA-17> nico * r4227 /packages/libs/glib/ (. Config.in Makefile ipkg/): port glib to buildroot-ng
[14:11] <CIA-17> nico * r4228 /packages/libs/gnutls/ (. Config.in Makefile ipkg/): port gnutls to buildroot-ng
[14:14] <CIA-17> nico * r4229 /packages/libs/flac/ (. Config.in Makefile ipkg/): port flac to buildroot-ng
[14:16] <CIA-17> nico * r4230 /packages/libs/gsm/ (. Config.in Makefile ipkg/): port gsm to buildroot-ng
[14:59] <CIA-17> nico * r4231 /packages/libs/id3lib/ (. Config.in Makefile ipkg/): port id3lib to buildroot-ng
[15:03] <CIA-17> nico * r4232 /packages/libs/libid3tag/ (. Config.in Makefile ipkg/): port libid3tag to buildroot-ng
[15:05] <CIA-17> nico * r4233 /packages/libs/libmcrypt/ (. Config.in Makefile ipkg/): port libmcrypt to buildroot-ng
[15:08] <CIA-17> nico * r4234 /packages/libs/libol/ (. Config.in Makefile ipkg/): port libol to buildroot-ng
[15:11] <CIA-17> nico * r4235 /packages/libs/uclibc++/Makefile:
[15:11] <CIA-17> fix missing shared lib in package and lib path,
[15:11] <CIA-17> remove unneeded PKG_INSTALL_DIR, indent package desc.
[16:21] <CIA-17> nbd * r4236 /branches/buildroot-ng/openwrt/package/kernel/ (Makefile modules.mk): bugfixes for the kernel module packaging
[17:16] <CIA-17> nico * r4237 /packages/libs/ustl/ (. Config.in Makefile ipkg/): port ustl to buildroot-ng
[17:24] <CIA-17> nico * r4238 /packages/libs/libdaemon/Makefile: fix $$ escape
[19:11] <CIA-17> nico * r4239 /packages/libs/avahi/ (. Config.in Makefile ipkg/): port avahi to buildroot-ng
[19:16] <CIA-17> nico * r4240 /packages/libs/howl/ (. Config.in Makefile ipkg/): port howl to buildroot-ng
[19:19] <CIA-17> nico * r4241 /packages/libs/net-snmp/ (. Config.in Makefile ipkg/): port net-snmp to buildroot-ng
[19:20] <nbd> {Nico}: ping
[19:42] <CIA-17> nico * r4242 /packages/libs/radiusclient-ng/ (. Config.in Makefile ipkg/): port radiusclient-ng to buildroot-ng
[19:43] <{Nico}> nbd: pong
[19:45] <db90h> don't play games in here
[19:45] <db90h> or at least not ping pong, you might hit someone with the ball
[19:47] <{Nico}> db90h: it won't hurt, just wake up those sleepy channel lurkers :)
[19:48] <CIA-17> nico * r4243 /packages/libs/speex/ (. Config.in Makefile ipkg/): port speex to buildroot-ng
[19:49] <db90h> hehe
[19:52] <nbd> {Nico}: do avahi, howl and net-snmp really belong in /packages/libs?
[19:57] <{Nico}> nbd: i put those here because their implementations are based on libraries
[19:58] <{Nico}> look at net-snmp, i put the client utils and daemons in Network
[20:00] <{Nico}> nbd: i'm not sure how it must be done, feel free to move them around
[20:07] <nbd> no idea. i guess we can leave them there for now
[00:00] --- Mon Jul 24 2006