[00:00] <nbd> btw. maybe i won't have that much time in the next few days. if anybody wants to finish moving the kernel packaging to package/kernel/, feel free to do so
[03:13] <CIA-17> nico * r4161 /packages/libs/libxml2/ (. Config.in Makefile ipkg/): port libxml2 to buildroot-ng
[03:27] <CIA-17> nico * r4162 /packages/libs/libxslt/ (. Config.in Makefile ipkg/): port libxslt to buildroot-ng
[04:41] <BrainSlayer> @florian: i ported the brcm63xx port to 2.6.17.5 now and i integrated also the cfe support which saves alot of space
[08:46] <CIA-17> nico * r4163 /trunk/openwrt/package/axtls/ (Makefile patches/002-strip.patch): don't use the host strip command (closes: #592).
[08:53] <CIA-17> nico * r4164 /packages/libs/axtls/ (. Config.in Makefile ipkg/ patches/002-strip.patch): port axtls to buildroot-ng
[08:59] <CIA-17> nico * r4165 /packages/libs/matrixssl/ (. Config.in Makefile ipkg/): port matrixssl to buildroot-ng
[09:25] <CIA-17> nico * r4166 /trunk/openwrt/package/ (4 files in 3 dirs): fix elinks dependency on openssl (closes: #614). Bump release number.
[10:03] <florian__> BrainSlayer: great, does it work ?
[10:04] <florian__> malbon: sorry for yesterday, I had connection problems
[10:04] <florian__> yes the flash is 8Mb
[10:12] <malbon> florian__: no prob, don't worry about it.
[10:13] <florian__> malbon: yes you are right there are no redboot parsing in this driver
[10:13] <malbon> :) that doesn't happen very often .
[10:13] <florian__> in fact, I was just reading a few drivers, and the mtx1 one is very simple, but seems very static, do you think it can be possible in the same driver to both have a physically mapped flash for CFE devices, and a redboot parsing ?
[10:14] <malbon> yeah why not?
[10:14] <florian__> ok, then it is maybe easier to completely rewrite the map driver for brcm63xx
[10:14] <malbon> if the redboot probe doesn't return any items then use the static?
[10:15] <florian__> yes I think so
[10:15] <florian__> do you have any documentation about how to use the redboot parsing/probing ?
[10:16] <florian__> I am reading the linux cross-reference, it helps a bit
[10:19] <florian__> maybe I had better take the bcm947xx map and add redboot parsing to it
[10:26] <malbon> the sa1100-flash.c has a redboot then if not static map style thing.
[11:09] <florian__> ok thanks, I will take a look
[11:29] <nbd> florian__: the ar7 map driver is a good example of what to do with the values that you get either from redboot or from cfe
[11:32] <groz> nbd I see you have been busy
[11:32] <groz> saw a commit of a bunch of packages in the list
[11:32] <groz> i've been quietly starting to work here with the ng buildroot, and going to get into a few packages later this week
[11:32] <groz> wanna make sure we are not both doing the same ones
[11:33] <nbd> i'm not doing any packaging work right now
[11:33] <nbd> and the latest package commit series was done by nico
[11:33] <groz> i'll be needing (so bringing them over) wpa_supplicant and xsupplicant later this week
[11:33] <groz> oh, sorry, i misread
[11:33] <groz> right, it was nico, just looked again
[11:34] <groz> I've got one new target done, and working on another right now
[11:34] <groz> i used the uml as my first 'lets learn this new buildroot environment' target
[11:35] <florian__> nbd: thanks
[11:35] <nbd> hm... openwrt building uml images - nice idea
[11:35] <groz> yah, it builds a uml kernel, with the root file system right in it
[11:35] <groz> or, it'll build an ext2 with some free space
[11:35] <groz> i have a little cleanup to do
[11:36] <groz> but I thought it's a great way to 'learn to work with it all'
[11:36] <groz> but there are some gotchas, you need to build the uml kernel with HOST tools/libs, not TARGET
[11:36] <groz> then build everything in the rootfs with target cross
[11:37] <groz> i have a little cleanup to do in the makes so it's not quite so 'hackky' the way i build the kernel using host rather than cross compile tools
[11:37] <nbd> so. how do you like buildroot-ng compared to the older stuff?
[11:37] <groz> i LOVE IT
[11:37] <nbd> great
[11:38] <groz> i've been away from all this kind of stuff for quite a while
[11:38] <groz> and just getting back into it seriously over the last couple of weeks, so i just checked it all out
[11:38] <groz> and started playing
[11:38] <groz> i've been building wrt and i386 targets, and fiddling with various things
[11:38] <groz> once i figured my way thru it, it's VERY well done I think
[11:39] <nbd> yeah, it still needs some documentation
[11:39] <groz> so, for my experiment, try this idea
[11:39] <groz> i got a 'front machine' connected to cable modem
[11:39] <groz> it's a very dumb little box, and does not much
[11:39] <nbd> the result machine could be very useful for people who want to try openwrt on their linux box
[11:39] <groz> it sets up a bridge, bridges the external ethernet with a vlan
[11:39] <groz> then
[11:39] <groz> on another machine, i run the uml openwrt
[11:39] <groz> and using umlswitch
[11:40] <groz> stuff it into that vlan
[11:40] <groz> and then it gets dhcp from my cable company
[11:40] <groz> front box doesn't
[11:40] <groz> now i have an openwrt router inside a uml, on a powerful box
[11:40] <groz> where i can play to hearts content
[11:40] <groz> and if i break it
[11:40] <groz> oh well, kill uml, start over
[11:40] <nbd> yeah
[11:40] <nbd> that's nice
[11:40] <groz> I did it as an experiment in leanring buildroot-ng
[11:41] <nbd> btw. for the rootfs you could try to use jffs2 over block2mtd
[11:41] <nbd> like on the rb532 port
[11:41] <groz> but i think, it'll be a great way for folks to learn the concept of 'embedded router'
[11:41] <nbd> then you can always increase the filesystem size by making the fs image larger
[11:41] <groz> yah, i'm trying all sorts of things, it's a very easy thing to experiment with
[11:41] <groz> for now, to get it going in a world of 'known how to do things' i got it up with ext2 uml file systems
[11:41] <groz> and now starting to play with other options
[11:42] <groz> it starts up fine with the ramfs too, but, with no place to write anything
[11:42] <groz> even trivial stuff like dhcp fails to complete
[11:42] <groz> tht could be addressed pretty easy by linking a few things into tmpfs
[11:42] <groz> in the original root fs
[11:43] <groz> but, it's been fun actually
[11:43] <groz> I'm planning to commit all the uml stuff in just as soon as it's 'clean'
[15:16] <CIA-17> nbd * r4167 /branches/buildroot-ng/openwrt/package/libpcap/patches/103-flex_workaround.patch: add workaround for a broken flex version on some systems (e.g. fc4)
[15:25] <CIA-17> nbd * r4168 /branches/buildroot-ng/openwrt/ (Makefile include/prereq.mk rules.mk): add a simple prerequisite check
[15:37] <CIA-17> nbd * r4169 /branches/buildroot-ng/openwrt/Makefile: add .prereq to distclean
[16:26] <BrainSlayer> @florian: yes it does work
[16:26] <BrainSlayer> @florian: i'm just testing right now the userspace. the flash implementation of broadcom is very buggy, so not all squashfs options will work. the inode table cannot be compressed for example
[16:27] <BrainSlayer> @florian: dont ask me why, but if i compress the inode table squashfs will bring a error like as it tried to read somthing out of bounds of the flash
[16:46] <malbon> BrainSlayer: which bit did you get to work? Redboot partitioning?
[16:48] <nbd> BrainSlayer: what does the flash driver have to do with squashfs inode compression? this doesn't make sense at all
[16:49] <nbd> i'd say enable the compression and try to find another way to solve the problem
[16:49] <nbd> otherwise you have to expect more weird filesystem issues later
[18:51] <CIA-17> nbd * r4170 /branches/buildroot-ng/openwrt/include/prereq.mk: check for zlib.h as well
[19:10] <florian_1> nbd: why not make a configure.ac ?
[19:13] <nbd> i hate autoconf
[19:13] <nbd> it's so weird
[19:13] <florian_1> yes but it's simple for things like that no ?
[19:13] <nbd> i don't think so
[19:14] <florian_1> ok
[19:14] <florian_1> AC_CHECK_GCC ?
[19:20] <florian__> nbd: what's the interest of apsta where wds can do almost the same thing ?
[19:21] <Kaloz> ?
[19:21] <[mbm]> florian__: not everything supports wds
[19:21] <nbd> apsta works with normal aps that don't support wds
[19:22] <Kaloz> apsta and wds are pretty different for different purposes
[19:22] <[mbm]> you could setup a repeater by using sta to connect and then ap to repeat
[19:22] <[mbm]> and it would work for any wireless network, regardless of if it supported wds
[19:22] <florian__> yes
[19:31] <florian__> ok
[19:43] <BrainSlayer> @florian: brcm63xx works now
[19:43] <BrainSlayer> @florian: i had to fix some things in the broadcom specific sources. i will give out the patches to nbd for integrating
[00:00] --- Thu Jul 20 2006
[03:13] <CIA-17> nico * r4161 /packages/libs/libxml2/ (. Config.in Makefile ipkg/): port libxml2 to buildroot-ng
[03:27] <CIA-17> nico * r4162 /packages/libs/libxslt/ (. Config.in Makefile ipkg/): port libxslt to buildroot-ng
[04:41] <BrainSlayer> @florian: i ported the brcm63xx port to 2.6.17.5 now and i integrated also the cfe support which saves alot of space
[08:46] <CIA-17> nico * r4163 /trunk/openwrt/package/axtls/ (Makefile patches/002-strip.patch): don't use the host strip command (closes: #592).
[08:53] <CIA-17> nico * r4164 /packages/libs/axtls/ (. Config.in Makefile ipkg/ patches/002-strip.patch): port axtls to buildroot-ng
[08:59] <CIA-17> nico * r4165 /packages/libs/matrixssl/ (. Config.in Makefile ipkg/): port matrixssl to buildroot-ng
[09:25] <CIA-17> nico * r4166 /trunk/openwrt/package/ (4 files in 3 dirs): fix elinks dependency on openssl (closes: #614). Bump release number.
[10:03] <florian__> BrainSlayer: great, does it work ?
[10:04] <florian__> malbon: sorry for yesterday, I had connection problems
[10:04] <florian__> yes the flash is 8Mb
[10:12] <malbon> florian__: no prob, don't worry about it.
[10:13] <florian__> malbon: yes you are right there are no redboot parsing in this driver
[10:13] <malbon> :) that doesn't happen very often .
[10:13] <florian__> in fact, I was just reading a few drivers, and the mtx1 one is very simple, but seems very static, do you think it can be possible in the same driver to both have a physically mapped flash for CFE devices, and a redboot parsing ?
[10:14] <malbon> yeah why not?
[10:14] <florian__> ok, then it is maybe easier to completely rewrite the map driver for brcm63xx
[10:14] <malbon> if the redboot probe doesn't return any items then use the static?
[10:15] <florian__> yes I think so
[10:15] <florian__> do you have any documentation about how to use the redboot parsing/probing ?
[10:16] <florian__> I am reading the linux cross-reference, it helps a bit
[10:19] <florian__> maybe I had better take the bcm947xx map and add redboot parsing to it
[10:26] <malbon> the sa1100-flash.c has a redboot then if not static map style thing.
[11:09] <florian__> ok thanks, I will take a look
[11:29] <nbd> florian__: the ar7 map driver is a good example of what to do with the values that you get either from redboot or from cfe
[11:32] <groz> nbd I see you have been busy
[11:32] <groz> saw a commit of a bunch of packages in the list
[11:32] <groz> i've been quietly starting to work here with the ng buildroot, and going to get into a few packages later this week
[11:32] <groz> wanna make sure we are not both doing the same ones
[11:33] <nbd> i'm not doing any packaging work right now
[11:33] <nbd> and the latest package commit series was done by nico
[11:33] <groz> i'll be needing (so bringing them over) wpa_supplicant and xsupplicant later this week
[11:33] <groz> oh, sorry, i misread
[11:33] <groz> right, it was nico, just looked again
[11:34] <groz> I've got one new target done, and working on another right now
[11:34] <groz> i used the uml as my first 'lets learn this new buildroot environment' target
[11:35] <florian__> nbd: thanks
[11:35] <nbd> hm... openwrt building uml images - nice idea
[11:35] <groz> yah, it builds a uml kernel, with the root file system right in it
[11:35] <groz> or, it'll build an ext2 with some free space
[11:35] <groz> i have a little cleanup to do
[11:36] <groz> but I thought it's a great way to 'learn to work with it all'
[11:36] <groz> but there are some gotchas, you need to build the uml kernel with HOST tools/libs, not TARGET
[11:36] <groz> then build everything in the rootfs with target cross
[11:37] <groz> i have a little cleanup to do in the makes so it's not quite so 'hackky' the way i build the kernel using host rather than cross compile tools
[11:37] <nbd> so. how do you like buildroot-ng compared to the older stuff?
[11:37] <groz> i LOVE IT
[11:37] <nbd> great
[11:38] <groz> i've been away from all this kind of stuff for quite a while
[11:38] <groz> and just getting back into it seriously over the last couple of weeks, so i just checked it all out
[11:38] <groz> and started playing
[11:38] <groz> i've been building wrt and i386 targets, and fiddling with various things
[11:38] <groz> once i figured my way thru it, it's VERY well done I think
[11:39] <nbd> yeah, it still needs some documentation
[11:39] <groz> so, for my experiment, try this idea
[11:39] <groz> i got a 'front machine' connected to cable modem
[11:39] <groz> it's a very dumb little box, and does not much
[11:39] <nbd> the result machine could be very useful for people who want to try openwrt on their linux box
[11:39] <groz> it sets up a bridge, bridges the external ethernet with a vlan
[11:39] <groz> then
[11:39] <groz> on another machine, i run the uml openwrt
[11:39] <groz> and using umlswitch
[11:40] <groz> stuff it into that vlan
[11:40] <groz> and then it gets dhcp from my cable company
[11:40] <groz> front box doesn't
[11:40] <groz> now i have an openwrt router inside a uml, on a powerful box
[11:40] <groz> where i can play to hearts content
[11:40] <groz> and if i break it
[11:40] <groz> oh well, kill uml, start over
[11:40] <nbd> yeah
[11:40] <nbd> that's nice
[11:40] <groz> I did it as an experiment in leanring buildroot-ng
[11:41] <nbd> btw. for the rootfs you could try to use jffs2 over block2mtd
[11:41] <nbd> like on the rb532 port
[11:41] <groz> but i think, it'll be a great way for folks to learn the concept of 'embedded router'
[11:41] <nbd> then you can always increase the filesystem size by making the fs image larger
[11:41] <groz> yah, i'm trying all sorts of things, it's a very easy thing to experiment with
[11:41] <groz> for now, to get it going in a world of 'known how to do things' i got it up with ext2 uml file systems
[11:41] <groz> and now starting to play with other options
[11:42] <groz> it starts up fine with the ramfs too, but, with no place to write anything
[11:42] <groz> even trivial stuff like dhcp fails to complete
[11:42] <groz> tht could be addressed pretty easy by linking a few things into tmpfs
[11:42] <groz> in the original root fs
[11:43] <groz> but, it's been fun actually
[11:43] <groz> I'm planning to commit all the uml stuff in just as soon as it's 'clean'
[15:16] <CIA-17> nbd * r4167 /branches/buildroot-ng/openwrt/package/libpcap/patches/103-flex_workaround.patch: add workaround for a broken flex version on some systems (e.g. fc4)
[15:25] <CIA-17> nbd * r4168 /branches/buildroot-ng/openwrt/ (Makefile include/prereq.mk rules.mk): add a simple prerequisite check
[15:37] <CIA-17> nbd * r4169 /branches/buildroot-ng/openwrt/Makefile: add .prereq to distclean
[16:26] <BrainSlayer> @florian: yes it does work
[16:26] <BrainSlayer> @florian: i'm just testing right now the userspace. the flash implementation of broadcom is very buggy, so not all squashfs options will work. the inode table cannot be compressed for example
[16:27] <BrainSlayer> @florian: dont ask me why, but if i compress the inode table squashfs will bring a error like as it tried to read somthing out of bounds of the flash
[16:46] <malbon> BrainSlayer: which bit did you get to work? Redboot partitioning?
[16:48] <nbd> BrainSlayer: what does the flash driver have to do with squashfs inode compression? this doesn't make sense at all
[16:49] <nbd> i'd say enable the compression and try to find another way to solve the problem
[16:49] <nbd> otherwise you have to expect more weird filesystem issues later
[18:51] <CIA-17> nbd * r4170 /branches/buildroot-ng/openwrt/include/prereq.mk: check for zlib.h as well
[19:10] <florian_1> nbd: why not make a configure.ac ?
[19:13] <nbd> i hate autoconf
[19:13] <nbd> it's so weird
[19:13] <florian_1> yes but it's simple for things like that no ?
[19:13] <nbd> i don't think so
[19:14] <florian_1> ok
[19:14] <florian_1> AC_CHECK_GCC ?
[19:20] <florian__> nbd: what's the interest of apsta where wds can do almost the same thing ?
[19:21] <Kaloz> ?
[19:21] <[mbm]> florian__: not everything supports wds
[19:21] <nbd> apsta works with normal aps that don't support wds
[19:22] <Kaloz> apsta and wds are pretty different for different purposes
[19:22] <[mbm]> you could setup a repeater by using sta to connect and then ap to repeat
[19:22] <[mbm]> and it would work for any wireless network, regardless of if it supported wds
[19:22] <florian__> yes
[19:31] <florian__> ok
[19:43] <BrainSlayer> @florian: brcm63xx works now
[19:43] <BrainSlayer> @florian: i had to fix some things in the broadcom specific sources. i will give out the patches to nbd for integrating
[00:00] --- Thu Jul 20 2006