[01:23] <CIA-17> mbm * r4171 /branches/buildroot-ng/openwrt/package/broadcom-wl/Makefile: fix broadcom-wl dependancies
[08:05] <nbd> [mbm]: ping
[09:40] <[mbm]> nbd: ?
[12:34] <nbd> [mbm]: why did you remove the + from the nvram dependency in broadcom-wl?
[12:37] <[mbm]> because it didn't work as intended
[12:37] <[mbm]> nas was visible on brcm-2.6
[12:37] <[mbm]> which caused a compile error when the wl wouldn't build
[12:38] <nbd> i'll fix it
[12:39] <[mbm]> shouldn't it be += ?
[12:40] <nbd> +=?
[12:41] <nbd> why?
[12:41] <[mbm]> guess I'm thinking c syntax
[12:41] <CIA-16> nbd * r4172 /branches/buildroot-ng/openwrt/package/broadcom-wl/Makefile: fix nas depends
[12:42] <[mbm]> oh, I also plan on reverting that V= thing and just using KBUILD_VERBOSE if you want to override it via environment
[12:42] <nbd> ok
[12:42] <[mbm]> hmm what's the + actually mean on the dependancy line?
[12:43] <[mbm]> I though you were attempting some variable+=value syntax
[12:45] <nbd> no. @ and + are prefixes
[12:46] <nbd> @ means referencing a real menuconfig symbol
[12:46] <nbd> + means turning a depends into a select
[12:47] <[mbm]> oh.. is that documented .. anywhere?
[12:47] <nbd> not yet
[12:47] <[mbm]> :P
[19:15] <CIA-16> mbm * r4173 /branches/buildroot-ng/openwrt/include/verbose.mk: cleanup & revert [4159]
[19:20] <CIA-16> mbm * r4174 /branches/buildroot-ng/openwrt/Makefile: clean up console output
[19:24] <CIA-16> nbd * r4175 /branches/buildroot-ng/openwrt/Makefile: clean up console output even more
[19:25] <[mbm]> can we fix the whole download/source targets to all be 'download' ?
[19:25] <nbd> sure
[19:25] <nbd> can you do it? i'm working on the network scripts right now
[19:28] <CIA-16> mbm * r4176 /branches/buildroot-ng/openwrt/ (4 files in 4 dirs): change 'source' to 'download'
[19:28] <nbd> btw. for the network scripts i'd like to use different names for bridge interfaces
[19:28] <[mbm]> ?
[19:28] <nbd> something like br-lan for a bridge in the lan interface config
[19:29] <nbd> so that you don't have to specify the interface name of the bridge manually
[19:29] <[mbm]> ok
[19:30] <nbd> it'll just be the default. you can always override it if you like
[19:36] <[mbm]> hmm .. why is python a prereq?
[19:36] <nbd> one of the ipkg tools is written in python, iirc
[19:36] <[mbm]> ah
[19:39] <[mbm]> btw, including verbose.mk in prereq.mk is somewhat pointless -- there's no $(MAKE) calls
[19:40] <[mbm]> (it's also included by rules.mk anyways)
[19:40] <nbd> there are NO_TRACE_MAKE calls
[19:40] <nbd> that's why
[19:40] <[mbm]> see point #2 :)
[19:41] <nbd> the reason was because i had a different structure before
[19:41] <nbd> but it doesn't make sense anymore, that's right
[19:41] <[mbm]> I'll clean it up in a sec
[19:41] <nbd> k
[19:46] <nbd> brb
[19:56] <[mbm]> hmm didn't know we had a todo page for the buildroot-ng packages
[19:57] <{Nico}> [mbm]: i started it to keep a list of what's ported and what's not
[19:57] <[mbm]> watchdog was a package? I always used the busybox version
[19:57] <groz> I've got a new target ready for buildroot-ng but dont have info on committing around anymore, kaloz is the person to talk to no ?
[19:57] <[mbm]> groz: I can set that up
[19:57] <[mbm]> (don't you already have a devel account?)
[19:57] <groz> he set it up a while back
[19:57] <groz> i just dont remember the details
[19:58] <groz> and, like a good boy, never wrote down accounts/passwords
[19:58] <groz> my first delve into buildroot-ng, i set up a uml target, works real slick
[19:58] <groz> like this new setup
[19:59] <[mbm]> well, the username is groz .. do you need the password reset (if so send me an password hash -- md5 format prefered)
[19:59] <groz> it's thru svn.openwrt or is it thru the .hu machine ?
[20:00] <groz> i can make a good guess at the password, but think i may be going to the wrong place
[20:00] <[mbm]> just use the instructions on dev
[20:00] <[mbm]> when you commit it'll ask for a username & pass
[20:01] <groz> yah, ok, i had password wrong when i tried a few minutes ago, i'll let you know if i need to get it reset
[20:01] <groz> couple more to try
[20:04] <{Nico}> [mbm]: i started porting libraries mainly, to ease porting packages thats depend on them
[20:04] <[mbm]> {Nico}: good idea
[20:05] <{Nico}> what are the packages you would like to see there?
[20:05] <groz> nico, i saw a big list of packages go in the other day
[20:05] <groz> i'm going to be working on a couple of them later this week
[20:05] <groz> wpa_supplicant and xsupplicant
[20:06] <groz> if you haven't already started on those two, leave em to me
[20:06] <{Nico}> groz: it was my commiting day :)
[20:06] <[mbm]> {Nico}: kismet gets my vote .. one of the few on that todo list that I actually use
[20:06] <groz> looked like a busy one
[20:07] <{Nico}> groz: use https://dev.openwrt.org/wiki/OpenWrtPackageToDo and mark ownership of the one you're working on
[20:07] <[mbm]> less and watchdog are both provided by busybox and nolonger require a package
[20:11] <{Nico}> [mbm]: ok, obsoleted
[20:18] <[mbm]> didn't someone mention replacing $(PKG_CAT) with file?
[20:27] <nbd> [mbm]: yeah, it was florian iirc
[20:30] <groz> mbm, how do you want me to get that hash to you ?
[20:30] <[mbm]> via msg is probably fine
[20:30] <groz> email or just paste it into a pm here
[20:30] <[mbm]> wouldn't paste it publically
[20:30] <groz> you think i'm completely stupid ?
[20:33] <[mbm]> not compeltely
[20:42] <CIA-16> groz * r4177 /branches/buildroot-ng/openwrt/target/linux/ (5 files in 3 dirs): Start adding uml target
[20:42] <groz> wow, it works
[20:42] <nbd> :)
[20:43] <groz> ok, there's going to be a few more changes scattered around before that one is fully buildable
[20:43] <groz> i'll get them in shortly
[20:43] <groz> just wanted to see if it actually worked
[20:43] <florian__> how and in which purpose can we use UML ?
[20:44] <groz> I use it to build up a system without actually having hardware, great place to test a lot of things
[20:44] <nbd> so that you can experiment with openwrt routing on a normal linux machine
[20:44] <groz> example, here's one way i'm using it
[20:44] <groz> i have a 'front router' connected to cable
[20:44] <florian__> ok, why not use qemu instead ?
[20:44] <groz> it has NO ip from cable, but, it bridges a vlan
[20:44] <groz> to the outside world
[20:44] <groz> the uml talks to that vlan
[20:44] <groz> i can test all sorts of packages that way, without ever flashing hardware
[20:45] <groz> different strokes for different folks i guess
[20:46] <groz> qemu requires a fair amount of setup, the uml target using ramfs
[20:46] <groz> build, run
[20:46] <nbd> and uml is faster than qemu
[20:46] <groz> 10 umls, pretend-a-mesh
[21:08] <florian__> ok
[21:08] <florian__> that what I wanted to know
[21:08] <groz> just another tool in the toolbox
[21:09] <groz> and potentially, it's a decent target for folks to work with/on packages, without actually having hardware to work on
[21:09] <florian__> right
[21:09] <groz> so you dont have to fuss with your 'only router' when trying to get a package working with uclibc etc
[21:11] <{Nico}> nbd: how do we deal with "configurable" packages (like openvpn) in buildroot-ng?
[21:12] <nbd> have a look at how it's done with busybox
[21:12] <groz> nico, i dont see an obvious way to 'claim ownership' of a package on that page
[21:12] <{Nico}> nbd: i prefer the -custom package option (like it's done in snort for example)
[21:12] <groz> the todo page
[21:12] <nbd> {Nico}: ah, that...
[21:13] <{Nico}> groz: are you logged in trac?
[21:13] <nbd> {Nico}: well... is there a decent default that we can expect most people to use?
[21:13] <{Nico}> nbd: dunno, not yet an openvpn user
[21:14] <{Nico}> groz: btw, you should add yourself to the people page too
[21:15] <groz> I'll get to it eventually nico
[21:28] <florian__> malbon: what's the problem with your redboot working on ar531x ? is it illegal ?
[21:38] <malbon> florian__: there is no legality problems.
[21:39] <malbon> florian__: I don't want to get loads of 'I've installed redboot how do I make linux work' type questions.
[21:39] <malbon> florian__: are you registered yet?
[21:41] <florian__> not yet
[21:41] <florian__> I did not had the time
[21:42] <malbon> hmm, ok. it only takes 5 mins you know.
[21:43] <florian__> I had network problems for 700 people that needed to be solved *quickly* :)
[21:45] <malbon> fair enough.
[21:47] <malbon> florian__: been reading the forum messages?
[21:48] <florian__> malbon: right, I have a friend who has a couple of dwl2100ap and who would be happy with an opensource firmware ;)
[21:49] <malbon> florian__: I'm working with mats to get it working right now on dwl2100ap but your friend can try it when I am happy.
[21:50] <malbon> florian__: Linux doesn't work today, it's broken because of madwifi hal.
[21:53] <florian__> ah
[21:53] <florian__> what's the problem with the madwifi hal ?
[21:54] <malbon> binutils
[21:54] <[mbm]> they compiled it wrong
[21:55] <[mbm]> you'll basically have a nice kernel oops if you try to use it
[21:58] <florian__> lol
[21:59] <[mbm]> (and that's only if you manage to compile it)
[22:16] <florian__> and I think even compiling it is a real pain :)
[22:16] <[mbm]> yep, especially since it thinks it's a 64bit app
[22:20] <florian__> ah great
[22:24] <florian__> we have to find a better name for buildroot-ng
[22:24] <h3sp4wn> bluemonkey ?
[22:24] <malbon> pan galactic gargle blaster?
[22:24] <nbd> florian__: it'll get folded back into trunk
[22:24] <groz> bill
[22:24] <nbd> florian__: then it's just kamikaze again
[22:25] <groz> a simple name, fits most languages
[22:25] <florian__> ok
[22:25] <florian__> what is the plan for kamikaze releases ?
[22:26] <[mbm]> new-improved-super-buildroot-next-gen ?
[22:26] <florian__> lol sounds good
[22:27] <[mbm]> well, if we ever finished the whole rc6/0.9 nonsense maybe we'd think about another release
[22:29] <florian__> sure
[22:42] <groz> question for you guys on kernel-build.mk
[22:42] <groz> is there a reason for objcopy to be TARGET_CROSS rather than KERNEL_CROSS ?
[22:43] <groz> reason i ask, when making uml, you gotta do the kernel with host tools, but the target uses the uclibc toolset
[22:43] <groz> so i set KERNEL_CROSS to nothing
[22:43] <groz> it's a special case, the kernel cross is indeed different from target
[22:44] <groz> wondering if i'll break anything committing this
[22:51] <florian__> well, can't you test ?
[22:53] <groz> it does things fine for the arch's im building
[22:53] <groz> but i'm wondering if it'll break something more generically in a pile of other arches is all
[22:53] <groz> doesn't hurt me here for x86 and a wrt targetted build
[22:54] <groz> it _seems_ right to me, but, i'm still learning some of the tricks in this whole setup
[22:54] <florian__> ok, I will test with au1000 which makes use of objcopy
[22:55] <groz> it's a trivial thing to back out if it breaks stuff
[22:55] <[mbm]> groz: yo9u just wnat to change the objcopy line in kernel-build.mk ?
[22:55] <groz> yah
[22:55] <groz> uml is special case
[22:55] <[mbm]> should be fine
[22:55] <groz> build kernel with host tools and target
[22:56] <groz> then build the target file system with target tools
[22:56] <[mbm]> normally TARGET_CROSS and KERNEL_DROSS point at the same thing
[22:56] <groz> yes, and this is the one place they wouldnt that i can think of
[22:56] <[mbm]> right
[22:56] <groz> cuz the kernel you build runs locally and becomes the target
[22:57] <groz> i've actually found a few places where that kind of minor change was needed
[22:57] <groz> should help keep things cleaner in the long run
[22:57] <groz> couple places where make needed the KARCH params added
[22:58] <groz> now, on another subject, is there any interest in some automated test scripts ?
[22:58] <groz> example, i've written some scripts that run on a wrt
[22:59] <[mbm]> certainly
[22:59] <groz> set it into client mode, then do things like exercise the dhcp server on the target
[22:59] <groz> by chaning mac addresses
[22:59] <groz> grabbing yet another ip
[22:59] <[mbm]> we actually need to come up with some proceedure to certify a release rather than just 'it boots? => ship it'
[22:59] <groz> then after using them all up, wait for lease time to expire, then ensure they will recycle onto new macs
[22:59] <groz> yah, that's EXACTLY what i've been working on lately, stuff to do that
[23:00] <groz> it's partly why i wanted a uml target, easy to do automated testing against it
[23:00] <groz> without even flashing hardware
[23:00] <groz> to test packages etc
[23:00] <[mbm]> I test what I can but I don't use ppoe and some of the other stuff
[23:00] <groz> for some of my other stuff, it's gotten out of hand, so we started writign automated test scripts
[23:00] <groz> and i could easily drop some of them into this build somehow
[23:01] <groz> the test env i use, is a wrt in client mode to exercise one in ap mode
[23:02] <groz> when it gets scripted, it becomes soooo much easier to run drudgery test suites that take a few hours
[23:02] <groz> start the scripts, then look at the reports they generate
[23:06] <CIA-16> groz * r4178 /branches/buildroot-ng/openwrt/ (9 files in 5 dirs): UML target completed. Added ext2 target fs for use with uml images
[23:07] <groz> there we go, it just builds and runs
[23:07] <groz> when you start it up using the generated ext2 image, you get serial port on an xterm, and openwrt running inside the uml
[23:09] <groz> mbm probly thinks hell has frozen over or something, with me actually working on this stuff again
[23:09] <[mbm]> you marked uml as x86 .. doesn't uml work on any platform?
[23:09] <groz> i've never tried compiling it on another platform
[23:09] <groz> but it's got a lot of x86 asm stuff in it
[23:10] <groz> it may, wouldn't be hard to change that if i had a target to try it with
[23:11] <groz> as an fyi, i've not had ANY success compiling and running uml with 64 bit target
[23:11] <groz> i gave up on that a while back, and just make it all 32
[23:11] <[mbm]> hmm.. I'll take a look later
[23:12] <[mbm]> I reinstalled my amd64 and forgot to leave ssh running so I can't get to it right now
[23:12] <groz> lol
[23:13] <groz> i had a lot of issues with various things on my amd64, finally put the 32 bit debian back on it
[23:13] <groz> i needed to get work done, not fight with things not working
[23:13] <[mbm]> I've been running 64bit debian on mine since I got it a few years ago .. rock solid
[23:14] <florian__> as a server 64bit no mutlilib works fine :)
[23:14] <groz> building some stuff causes issues, example
[23:14] <[mbm]> just recently upgraded it to ubuntu's 64bit because it was better maintained than the debian stuff
[23:14] <groz> i needed partimage server running
[23:14] <groz> and that wouldn't run no matter what on 64
[23:14] <Kaloz> only stuff i didn't see working is wine and flash
[23:14] <Kaloz> :p
[23:14] <groz> build or run
[23:15] <florian__> Kaloz: with qemu you can get it running :)
[23:15] <florian__> that's almost what I do for flash on linux ppc :p
[23:15] <Kaloz> i don't care about either really
[23:15] <Kaloz> :p
[23:15] <groz> but that's niether here nor there, what i got set up right now is getting the job done
[23:15] <groz> that's what matters
[23:15] <Kaloz> to be hones,t if oyu have to do any windows-related stuff
[23:16] <florian__> you use windows
[23:16] <Kaloz> you're better with dual-booting and real windows anyways
[23:16] <Kaloz> :p
[23:16] <florian__> especially on an ibook ? :p
[23:16] <groz> or do like i do, one computer running windows, then the rest running various flavours of linux
[23:16] <[mbm]> upgrade and run xen :P
[23:17] <Kaloz> florian__: buy a macbook :PP
[23:17] <groz> i find the multi monitor support in windows makes it a great terminal for the rest of the boxes
[23:17] <groz> i got a desk full of monitors, half of them are nothing more than vnc or ssh to linux boxes
[23:17] <Kaloz> groz: xfce + xinerama
[23:18] <groz> gonna try that soon kaloz, getting another one of the amd64 motherboards i got my raid running on, it's got a nice dual head adapter on it, and room to add a couple more in pci slots
[23:18] <groz> but for now, what i got works
[23:43] <[mbm]> groz: is there much point to a squashfs/jffs2 uml or can I just lock it to ext2?
[23:45] <groz> somebody was suggesting using a jffs over block driver
[23:45] <groz> to get jffs inside the uml file system
[23:46] <groz> i haven't looked at / tried yet
[23:46] <groz> one of the things i'm thinking here with this, ultimately, it's emulating a wrt54gr quite nicely
[23:46] <groz> and i am planning on targettign that router real soon
[23:46] <groz> have you looked at it at all ?
[23:46] <[mbm]> there's a gr model now?
[23:47] <groz> yah, and, it's i386 using some chipset
[23:47] <nbd> gr == RDC x86 core with ralink wifi
[23:47] <groz> kinda way cool
[23:47] <[mbm]> weird.
[23:47] <groz> ralink system on chip
[23:47] <groz> that runs i386 instruction set
[23:47] <groz> it's got a mini pci for the radio
[23:47] <groz> and it costs like 80 bucks
[23:48] <nbd> rdc != ralink
[23:48] <nbd> afaik
[23:48] <groz> not sure how much flash/ram on it, it uses redboot if i remember correctly
[23:48] <groz> thought they were the same company
[23:48] <groz> just different divisions
[23:48] <nbd> somebody told me otherwise
[23:48] <groz> well, ok, i do know this, sourcing the chips to make boards, both come from the same place :)
[23:48] <nbd> but they usually work together
[23:49] <groz> and there is one coming down the pipe
[23:49] <groz> that combines the radio with the x86 core
[23:49] <groz> all in one package
[23:49] <nbd> cool
[23:49] <groz> i have a customer contemplating using it
[23:50] <groz> so, for a prrof of concept, i'm gonna try get openwrt onto the gr
[23:50] <groz> i think the gr is a way cool piece of hardware actually
[23:50] <groz> x86 core, with a mimo radio
[23:50] <groz> i have no clue how good that radio is tho
[23:50] <groz> and how good/bad thier drivers are
[23:51] <malbon> are those roc things buyable?
[23:51] <groz> not yet
[23:51] <Kaloz> groz: when linksys relaeses the bootloader sources
[23:52] <Kaloz> groz: the current redboot on the gr is simply unusable
[23:52] <groz> i was talking to folks at a tradeshow a couple weeks ago, they kinda spilled some beans over dinner
[23:52] <Kaloz> hardcoded crap, no flash mapping
[23:52] <malbon> Kaloz: if it's a 386 then there probably isn't much to do to make the redboot to work on it.
[23:52] <groz> hmmm kaloz, same hardware comes from different vendors too
[23:52] <Kaloz> malbon: i want the tftpd inside ;)
[23:53] <groz> and from what i understand , isn't the flash mapped thru the ide ?
[23:53] <groz> to make it look 'just like a pc'
[23:53] <malbon> kaloz: yes, that feature would be nice.
[23:53] <Kaloz> groz: no
[23:53] <Kaloz> groz: on the gr redboot loads the kernel and the ramdisk froma hardcoded flash area
[23:53] <groz> from what i';ve read, that chipset is a real pc emulation, pretty much all pc oriented stuff should work on it
[23:53] <Kaloz> nothing else
[23:54] <groz> hmm, ok
[23:55] <groz> i dont have one yet, just took a cursory look at the gpl sources
[23:55] <groz> but, was planning on getting one, and getting detailed on it 'real soon
[23:57] <Kaloz> i have a device like that
[23:57] <Kaloz> but only the linksys one has tftpd
[00:00] --- Fri Jul 21 2006
[08:05] <nbd> [mbm]: ping
[09:40] <[mbm]> nbd: ?
[12:34] <nbd> [mbm]: why did you remove the + from the nvram dependency in broadcom-wl?
[12:37] <[mbm]> because it didn't work as intended
[12:37] <[mbm]> nas was visible on brcm-2.6
[12:37] <[mbm]> which caused a compile error when the wl wouldn't build
[12:38] <nbd> i'll fix it
[12:39] <[mbm]> shouldn't it be += ?
[12:40] <nbd> +=?
[12:41] <nbd> why?
[12:41] <[mbm]> guess I'm thinking c syntax
[12:41] <CIA-16> nbd * r4172 /branches/buildroot-ng/openwrt/package/broadcom-wl/Makefile: fix nas depends
[12:42] <[mbm]> oh, I also plan on reverting that V= thing and just using KBUILD_VERBOSE if you want to override it via environment
[12:42] <nbd> ok
[12:42] <[mbm]> hmm what's the + actually mean on the dependancy line?
[12:43] <[mbm]> I though you were attempting some variable+=value syntax
[12:45] <nbd> no. @ and + are prefixes
[12:46] <nbd> @ means referencing a real menuconfig symbol
[12:46] <nbd> + means turning a depends into a select
[12:47] <[mbm]> oh.. is that documented .. anywhere?
[12:47] <nbd> not yet
[12:47] <[mbm]> :P
[19:15] <CIA-16> mbm * r4173 /branches/buildroot-ng/openwrt/include/verbose.mk: cleanup & revert [4159]
[19:20] <CIA-16> mbm * r4174 /branches/buildroot-ng/openwrt/Makefile: clean up console output
[19:24] <CIA-16> nbd * r4175 /branches/buildroot-ng/openwrt/Makefile: clean up console output even more
[19:25] <[mbm]> can we fix the whole download/source targets to all be 'download' ?
[19:25] <nbd> sure
[19:25] <nbd> can you do it? i'm working on the network scripts right now
[19:28] <CIA-16> mbm * r4176 /branches/buildroot-ng/openwrt/ (4 files in 4 dirs): change 'source' to 'download'
[19:28] <nbd> btw. for the network scripts i'd like to use different names for bridge interfaces
[19:28] <[mbm]> ?
[19:28] <nbd> something like br-lan for a bridge in the lan interface config
[19:29] <nbd> so that you don't have to specify the interface name of the bridge manually
[19:29] <[mbm]> ok
[19:30] <nbd> it'll just be the default. you can always override it if you like
[19:36] <[mbm]> hmm .. why is python a prereq?
[19:36] <nbd> one of the ipkg tools is written in python, iirc
[19:36] <[mbm]> ah
[19:39] <[mbm]> btw, including verbose.mk in prereq.mk is somewhat pointless -- there's no $(MAKE) calls
[19:40] <[mbm]> (it's also included by rules.mk anyways)
[19:40] <nbd> there are NO_TRACE_MAKE calls
[19:40] <nbd> that's why
[19:40] <[mbm]> see point #2 :)
[19:41] <nbd> the reason was because i had a different structure before
[19:41] <nbd> but it doesn't make sense anymore, that's right
[19:41] <[mbm]> I'll clean it up in a sec
[19:41] <nbd> k
[19:46] <nbd> brb
[19:56] <[mbm]> hmm didn't know we had a todo page for the buildroot-ng packages
[19:57] <{Nico}> [mbm]: i started it to keep a list of what's ported and what's not
[19:57] <[mbm]> watchdog was a package? I always used the busybox version
[19:57] <groz> I've got a new target ready for buildroot-ng but dont have info on committing around anymore, kaloz is the person to talk to no ?
[19:57] <[mbm]> groz: I can set that up
[19:57] <[mbm]> (don't you already have a devel account?)
[19:57] <groz> he set it up a while back
[19:57] <groz> i just dont remember the details
[19:58] <groz> and, like a good boy, never wrote down accounts/passwords
[19:58] <groz> my first delve into buildroot-ng, i set up a uml target, works real slick
[19:58] <groz> like this new setup
[19:59] <[mbm]> well, the username is groz .. do you need the password reset (if so send me an password hash -- md5 format prefered)
[19:59] <groz> it's thru svn.openwrt or is it thru the .hu machine ?
[20:00] <groz> i can make a good guess at the password, but think i may be going to the wrong place
[20:00] <[mbm]> just use the instructions on dev
[20:00] <[mbm]> when you commit it'll ask for a username & pass
[20:01] <groz> yah, ok, i had password wrong when i tried a few minutes ago, i'll let you know if i need to get it reset
[20:01] <groz> couple more to try
[20:04] <{Nico}> [mbm]: i started porting libraries mainly, to ease porting packages thats depend on them
[20:04] <[mbm]> {Nico}: good idea
[20:05] <{Nico}> what are the packages you would like to see there?
[20:05] <groz> nico, i saw a big list of packages go in the other day
[20:05] <groz> i'm going to be working on a couple of them later this week
[20:05] <groz> wpa_supplicant and xsupplicant
[20:06] <groz> if you haven't already started on those two, leave em to me
[20:06] <{Nico}> groz: it was my commiting day :)
[20:06] <[mbm]> {Nico}: kismet gets my vote .. one of the few on that todo list that I actually use
[20:06] <groz> looked like a busy one
[20:07] <{Nico}> groz: use https://dev.openwrt.org/wiki/OpenWrtPackageToDo and mark ownership of the one you're working on
[20:07] <[mbm]> less and watchdog are both provided by busybox and nolonger require a package
[20:11] <{Nico}> [mbm]: ok, obsoleted
[20:18] <[mbm]> didn't someone mention replacing $(PKG_CAT) with file?
[20:27] <nbd> [mbm]: yeah, it was florian iirc
[20:30] <groz> mbm, how do you want me to get that hash to you ?
[20:30] <[mbm]> via msg is probably fine
[20:30] <groz> email or just paste it into a pm here
[20:30] <[mbm]> wouldn't paste it publically
[20:30] <groz> you think i'm completely stupid ?
[20:33] <[mbm]> not compeltely
[20:42] <CIA-16> groz * r4177 /branches/buildroot-ng/openwrt/target/linux/ (5 files in 3 dirs): Start adding uml target
[20:42] <groz> wow, it works
[20:42] <nbd> :)
[20:43] <groz> ok, there's going to be a few more changes scattered around before that one is fully buildable
[20:43] <groz> i'll get them in shortly
[20:43] <groz> just wanted to see if it actually worked
[20:43] <florian__> how and in which purpose can we use UML ?
[20:44] <groz> I use it to build up a system without actually having hardware, great place to test a lot of things
[20:44] <nbd> so that you can experiment with openwrt routing on a normal linux machine
[20:44] <groz> example, here's one way i'm using it
[20:44] <groz> i have a 'front router' connected to cable
[20:44] <florian__> ok, why not use qemu instead ?
[20:44] <groz> it has NO ip from cable, but, it bridges a vlan
[20:44] <groz> to the outside world
[20:44] <groz> the uml talks to that vlan
[20:44] <groz> i can test all sorts of packages that way, without ever flashing hardware
[20:45] <groz> different strokes for different folks i guess
[20:46] <groz> qemu requires a fair amount of setup, the uml target using ramfs
[20:46] <groz> build, run
[20:46] <nbd> and uml is faster than qemu
[20:46] <groz> 10 umls, pretend-a-mesh
[21:08] <florian__> ok
[21:08] <florian__> that what I wanted to know
[21:08] <groz> just another tool in the toolbox
[21:09] <groz> and potentially, it's a decent target for folks to work with/on packages, without actually having hardware to work on
[21:09] <florian__> right
[21:09] <groz> so you dont have to fuss with your 'only router' when trying to get a package working with uclibc etc
[21:11] <{Nico}> nbd: how do we deal with "configurable" packages (like openvpn) in buildroot-ng?
[21:12] <nbd> have a look at how it's done with busybox
[21:12] <groz> nico, i dont see an obvious way to 'claim ownership' of a package on that page
[21:12] <{Nico}> nbd: i prefer the -custom package option (like it's done in snort for example)
[21:12] <groz> the todo page
[21:12] <nbd> {Nico}: ah, that...
[21:13] <{Nico}> groz: are you logged in trac?
[21:13] <nbd> {Nico}: well... is there a decent default that we can expect most people to use?
[21:13] <{Nico}> nbd: dunno, not yet an openvpn user
[21:14] <{Nico}> groz: btw, you should add yourself to the people page too
[21:15] <groz> I'll get to it eventually nico
[21:28] <florian__> malbon: what's the problem with your redboot working on ar531x ? is it illegal ?
[21:38] <malbon> florian__: there is no legality problems.
[21:39] <malbon> florian__: I don't want to get loads of 'I've installed redboot how do I make linux work' type questions.
[21:39] <malbon> florian__: are you registered yet?
[21:41] <florian__> not yet
[21:41] <florian__> I did not had the time
[21:42] <malbon> hmm, ok. it only takes 5 mins you know.
[21:43] <florian__> I had network problems for 700 people that needed to be solved *quickly* :)
[21:45] <malbon> fair enough.
[21:47] <malbon> florian__: been reading the forum messages?
[21:48] <florian__> malbon: right, I have a friend who has a couple of dwl2100ap and who would be happy with an opensource firmware ;)
[21:49] <malbon> florian__: I'm working with mats to get it working right now on dwl2100ap but your friend can try it when I am happy.
[21:50] <malbon> florian__: Linux doesn't work today, it's broken because of madwifi hal.
[21:53] <florian__> ah
[21:53] <florian__> what's the problem with the madwifi hal ?
[21:54] <malbon> binutils
[21:54] <[mbm]> they compiled it wrong
[21:55] <[mbm]> you'll basically have a nice kernel oops if you try to use it
[21:58] <florian__> lol
[21:59] <[mbm]> (and that's only if you manage to compile it)
[22:16] <florian__> and I think even compiling it is a real pain :)
[22:16] <[mbm]> yep, especially since it thinks it's a 64bit app
[22:20] <florian__> ah great
[22:24] <florian__> we have to find a better name for buildroot-ng
[22:24] <h3sp4wn> bluemonkey ?
[22:24] <malbon> pan galactic gargle blaster?
[22:24] <nbd> florian__: it'll get folded back into trunk
[22:24] <groz> bill
[22:24] <nbd> florian__: then it's just kamikaze again
[22:25] <groz> a simple name, fits most languages
[22:25] <florian__> ok
[22:25] <florian__> what is the plan for kamikaze releases ?
[22:26] <[mbm]> new-improved-super-buildroot-next-gen ?
[22:26] <florian__> lol sounds good
[22:27] <[mbm]> well, if we ever finished the whole rc6/0.9 nonsense maybe we'd think about another release
[22:29] <florian__> sure
[22:42] <groz> question for you guys on kernel-build.mk
[22:42] <groz> is there a reason for objcopy to be TARGET_CROSS rather than KERNEL_CROSS ?
[22:43] <groz> reason i ask, when making uml, you gotta do the kernel with host tools, but the target uses the uclibc toolset
[22:43] <groz> so i set KERNEL_CROSS to nothing
[22:43] <groz> it's a special case, the kernel cross is indeed different from target
[22:44] <groz> wondering if i'll break anything committing this
[22:51] <florian__> well, can't you test ?
[22:53] <groz> it does things fine for the arch's im building
[22:53] <groz> but i'm wondering if it'll break something more generically in a pile of other arches is all
[22:53] <groz> doesn't hurt me here for x86 and a wrt targetted build
[22:54] <groz> it _seems_ right to me, but, i'm still learning some of the tricks in this whole setup
[22:54] <florian__> ok, I will test with au1000 which makes use of objcopy
[22:55] <groz> it's a trivial thing to back out if it breaks stuff
[22:55] <[mbm]> groz: yo9u just wnat to change the objcopy line in kernel-build.mk ?
[22:55] <groz> yah
[22:55] <groz> uml is special case
[22:55] <[mbm]> should be fine
[22:55] <groz> build kernel with host tools and target
[22:56] <groz> then build the target file system with target tools
[22:56] <[mbm]> normally TARGET_CROSS and KERNEL_DROSS point at the same thing
[22:56] <groz> yes, and this is the one place they wouldnt that i can think of
[22:56] <[mbm]> right
[22:56] <groz> cuz the kernel you build runs locally and becomes the target
[22:57] <groz> i've actually found a few places where that kind of minor change was needed
[22:57] <groz> should help keep things cleaner in the long run
[22:57] <groz> couple places where make needed the KARCH params added
[22:58] <groz> now, on another subject, is there any interest in some automated test scripts ?
[22:58] <groz> example, i've written some scripts that run on a wrt
[22:59] <[mbm]> certainly
[22:59] <groz> set it into client mode, then do things like exercise the dhcp server on the target
[22:59] <groz> by chaning mac addresses
[22:59] <groz> grabbing yet another ip
[22:59] <[mbm]> we actually need to come up with some proceedure to certify a release rather than just 'it boots? => ship it'
[22:59] <groz> then after using them all up, wait for lease time to expire, then ensure they will recycle onto new macs
[22:59] <groz> yah, that's EXACTLY what i've been working on lately, stuff to do that
[23:00] <groz> it's partly why i wanted a uml target, easy to do automated testing against it
[23:00] <groz> without even flashing hardware
[23:00] <groz> to test packages etc
[23:00] <[mbm]> I test what I can but I don't use ppoe and some of the other stuff
[23:00] <groz> for some of my other stuff, it's gotten out of hand, so we started writign automated test scripts
[23:00] <groz> and i could easily drop some of them into this build somehow
[23:01] <groz> the test env i use, is a wrt in client mode to exercise one in ap mode
[23:02] <groz> when it gets scripted, it becomes soooo much easier to run drudgery test suites that take a few hours
[23:02] <groz> start the scripts, then look at the reports they generate
[23:06] <CIA-16> groz * r4178 /branches/buildroot-ng/openwrt/ (9 files in 5 dirs): UML target completed. Added ext2 target fs for use with uml images
[23:07] <groz> there we go, it just builds and runs
[23:07] <groz> when you start it up using the generated ext2 image, you get serial port on an xterm, and openwrt running inside the uml
[23:09] <groz> mbm probly thinks hell has frozen over or something, with me actually working on this stuff again
[23:09] <[mbm]> you marked uml as x86 .. doesn't uml work on any platform?
[23:09] <groz> i've never tried compiling it on another platform
[23:09] <groz> but it's got a lot of x86 asm stuff in it
[23:10] <groz> it may, wouldn't be hard to change that if i had a target to try it with
[23:11] <groz> as an fyi, i've not had ANY success compiling and running uml with 64 bit target
[23:11] <groz> i gave up on that a while back, and just make it all 32
[23:11] <[mbm]> hmm.. I'll take a look later
[23:12] <[mbm]> I reinstalled my amd64 and forgot to leave ssh running so I can't get to it right now
[23:12] <groz> lol
[23:13] <groz> i had a lot of issues with various things on my amd64, finally put the 32 bit debian back on it
[23:13] <groz> i needed to get work done, not fight with things not working
[23:13] <[mbm]> I've been running 64bit debian on mine since I got it a few years ago .. rock solid
[23:14] <florian__> as a server 64bit no mutlilib works fine :)
[23:14] <groz> building some stuff causes issues, example
[23:14] <[mbm]> just recently upgraded it to ubuntu's 64bit because it was better maintained than the debian stuff
[23:14] <groz> i needed partimage server running
[23:14] <groz> and that wouldn't run no matter what on 64
[23:14] <Kaloz> only stuff i didn't see working is wine and flash
[23:14] <Kaloz> :p
[23:14] <groz> build or run
[23:15] <florian__> Kaloz: with qemu you can get it running :)
[23:15] <florian__> that's almost what I do for flash on linux ppc :p
[23:15] <Kaloz> i don't care about either really
[23:15] <Kaloz> :p
[23:15] <groz> but that's niether here nor there, what i got set up right now is getting the job done
[23:15] <groz> that's what matters
[23:15] <Kaloz> to be hones,t if oyu have to do any windows-related stuff
[23:16] <florian__> you use windows
[23:16] <Kaloz> you're better with dual-booting and real windows anyways
[23:16] <Kaloz> :p
[23:16] <florian__> especially on an ibook ? :p
[23:16] <groz> or do like i do, one computer running windows, then the rest running various flavours of linux
[23:16] <[mbm]> upgrade and run xen :P
[23:17] <Kaloz> florian__: buy a macbook :PP
[23:17] <groz> i find the multi monitor support in windows makes it a great terminal for the rest of the boxes
[23:17] <groz> i got a desk full of monitors, half of them are nothing more than vnc or ssh to linux boxes
[23:17] <Kaloz> groz: xfce + xinerama
[23:18] <groz> gonna try that soon kaloz, getting another one of the amd64 motherboards i got my raid running on, it's got a nice dual head adapter on it, and room to add a couple more in pci slots
[23:18] <groz> but for now, what i got works
[23:43] <[mbm]> groz: is there much point to a squashfs/jffs2 uml or can I just lock it to ext2?
[23:45] <groz> somebody was suggesting using a jffs over block driver
[23:45] <groz> to get jffs inside the uml file system
[23:46] <groz> i haven't looked at / tried yet
[23:46] <groz> one of the things i'm thinking here with this, ultimately, it's emulating a wrt54gr quite nicely
[23:46] <groz> and i am planning on targettign that router real soon
[23:46] <groz> have you looked at it at all ?
[23:46] <[mbm]> there's a gr model now?
[23:47] <groz> yah, and, it's i386 using some chipset
[23:47] <nbd> gr == RDC x86 core with ralink wifi
[23:47] <groz> kinda way cool
[23:47] <[mbm]> weird.
[23:47] <groz> ralink system on chip
[23:47] <groz> that runs i386 instruction set
[23:47] <groz> it's got a mini pci for the radio
[23:47] <groz> and it costs like 80 bucks
[23:48] <nbd> rdc != ralink
[23:48] <nbd> afaik
[23:48] <groz> not sure how much flash/ram on it, it uses redboot if i remember correctly
[23:48] <groz> thought they were the same company
[23:48] <groz> just different divisions
[23:48] <nbd> somebody told me otherwise
[23:48] <groz> well, ok, i do know this, sourcing the chips to make boards, both come from the same place :)
[23:48] <nbd> but they usually work together
[23:49] <groz> and there is one coming down the pipe
[23:49] <groz> that combines the radio with the x86 core
[23:49] <groz> all in one package
[23:49] <nbd> cool
[23:49] <groz> i have a customer contemplating using it
[23:50] <groz> so, for a prrof of concept, i'm gonna try get openwrt onto the gr
[23:50] <groz> i think the gr is a way cool piece of hardware actually
[23:50] <groz> x86 core, with a mimo radio
[23:50] <groz> i have no clue how good that radio is tho
[23:50] <groz> and how good/bad thier drivers are
[23:51] <malbon> are those roc things buyable?
[23:51] <groz> not yet
[23:51] <Kaloz> groz: when linksys relaeses the bootloader sources
[23:52] <Kaloz> groz: the current redboot on the gr is simply unusable
[23:52] <groz> i was talking to folks at a tradeshow a couple weeks ago, they kinda spilled some beans over dinner
[23:52] <Kaloz> hardcoded crap, no flash mapping
[23:52] <malbon> Kaloz: if it's a 386 then there probably isn't much to do to make the redboot to work on it.
[23:52] <groz> hmmm kaloz, same hardware comes from different vendors too
[23:52] <Kaloz> malbon: i want the tftpd inside ;)
[23:53] <groz> and from what i understand , isn't the flash mapped thru the ide ?
[23:53] <groz> to make it look 'just like a pc'
[23:53] <malbon> kaloz: yes, that feature would be nice.
[23:53] <Kaloz> groz: no
[23:53] <Kaloz> groz: on the gr redboot loads the kernel and the ramdisk froma hardcoded flash area
[23:53] <groz> from what i';ve read, that chipset is a real pc emulation, pretty much all pc oriented stuff should work on it
[23:53] <Kaloz> nothing else
[23:54] <groz> hmm, ok
[23:55] <groz> i dont have one yet, just took a cursory look at the gpl sources
[23:55] <groz> but, was planning on getting one, and getting detailed on it 'real soon
[23:57] <Kaloz> i have a device like that
[23:57] <Kaloz> but only the linksys one has tftpd
[00:00] --- Fri Jul 21 2006