[05:10] <DanaG> ooh, new omap kernel in repos... cool.
[05:10] <DanaG> Just need to figure out how it does postinst. =P
[05:20] <persia> dpkg -e is your friend
[05:37] <DanaG> dpkg -e does absolutely nothing on that deb.
[05:37] <DanaG> and dpkg --help doesn't list a -e.
[05:38] <persia> Are you sure?
[05:38] <persia> dpkg -e foo.deb BAR should extract all the metacontent in the .deb to BAR
[05:38] <DanaG> dana@beagleboard:~$ dpkg -e linux-image-2.6.33-dl.1_1.0lucid_armel.deb
[05:38] <DanaG> dana@beagleboard:~$
[05:38] <persia> OK
[05:38] <DanaG> er
[05:38] <persia> ls
[05:39] <DanaG> wrong file
[05:39] <DanaG> does it need absolute path?
[05:39] <persia> ls anyway :)
[05:39] <persia> No.
[05:39] <persia> Just enough path to find it.
[05:39] <DanaG>  dpkg -e /var/cache/apt/archives/linux-image-2.6.33-500-omap_2.6.33-500.3_armel.deb -- does nothing.
[05:40] <persia> Are you sure?
[05:41] <DanaG> yeah.
[05:41] <persia> Did you look for new files that may have been created?
[05:41] <DanaG> oh.
[05:41] <DanaG> I see.
[05:41] <DanaG> DEBIAN dir.
[05:41] <DanaG> still, why does --help not show -e?
[05:41] <armin76> fail
[05:42] <persia> Probably because it's really a dpkg-deb command that dpkg just passes through.
[05:42] <persia> It's definitely shown in `man dpkg`
[05:49] <DanaG> init: ureadahead main process (45) terminated with status 5
[05:49] <DanaG> happens on my host system, too...
[05:49] <DanaG> ureadahead seems not to like non-generic kernels.
[05:50] <DanaG> ugh, ubuntu -500-omap kernel can't read mtd devices!
[05:50] <DanaG> er, maybe my mtdparts got broken.
[05:55] <DanaG> argh, u-boot won't see some of the files on my fat32 partition.
[06:02] <DanaG> Unhandled fault: external abort on non-linefetch (0x1018) at 0x402e1000
[06:03] <DanaG> well... the thing now goes into suspend... but how do you wake it?
[06:10] <cooloney> hi, does anybody know how does NetworkManager know this "<info>  (eth0): driver 'fec' does not support carrier detection"?
[06:10] <DanaG> OMAP3 beagleboard.org # mtd erase 260000 20000
[06:10] <DanaG> mtdparts - define flash/nand partitions
[06:10] <DanaG> argh
[06:11] <DanaG> oh
[06:11] <DanaG> right
[06:11] <DanaG> nand erase
[06:11] <DanaG> sorry, thinking out loud about my own stuff.
[06:17] <DanaG> looks like that lucid omap kernel is screwed up somehow.
[07:01] <JaMaWrk> Is there some un-official repository with ubuntu but built for armv4t instead armv5t/armv6/armv7 for jaunty/karmic/lucid? I know I can use debian (as I already do), but I'm interstad in additional ubuntu stuff (except being rebuilt for newer arch). If no, do you know someone who tried and failed to provide this repo?
[07:01] <persia> There isn't such a repository.
[07:02] <persia> There have been a couple people who said they wanted to try, but it takes months to recompile everything, and I don't know of anyone who actually did it.
[07:02] <JaMaWrk> OK fair enough, thanks
[09:14] <cooloney> ogra_cmpc and persia: i already enabled phylib and carrier detection in fsl-imx51 fec driver
[09:15] <cooloney> plars and GrueMaster: could you please help me to test it? ^^^
[09:15] <cooloney> i already did that on my babbage board.
[09:51] <lool> cooloney: Sounds great
[09:52]  * lool doesn't have hardware to test though
[09:55] <cooloney> lool: why not enable that in qemu? and test it? haha -:))
[10:05] <ogra> cooloney, yeah, lool should really hack imx51 support into qemu !!
[10:05] <ogra> and add a virtual 2G :)
[10:06] <ogra> persia, heh, i see the first netwalker user on the ubuntu-users ML ... being desperate, he wants to upgrade
[10:09] <lool> cooloney: Yeah!  Let's do that during the next 2 years!  :-0
[11:27] <dmart> asac: hi there
[14:37] <dimitris> any gps navigation program ? I use ubuntu in arm (HTC diamond armv6)
[14:38] <dimitris> or at least any suggestion to see if gps is working
[14:46] <persia> There are a few, but I don't think any of them work very well at lower resolutions :(
[14:46] <Stskeeps> tangogps
[14:46] <persia> Oh, cool!  I didn't know we had that already.
[14:47] <dmart> asac: ping
[14:54] <asac> dmart: yup
[15:02] <dimitris> tangogps there is not in repository !
[15:10] <dimitris> I find gpsdrive
[15:14] <lool> ogra: Do you have details on why you disabled LDFLAGS in u-boot?
[15:16] <ogra> lool, because we set -Wl
[15:16] <ogra> and the build doesnt like that
[15:17] <lool> ogra: Which cc were using that with?
[15:17] <ogra> default on the babbage board
[15:17] <lool> ogra: So with armel gcc?
[15:17] <ogra> lucid
[15:17] <ogra> yes
[15:17] <lool> native build
[15:17] <ogra> it sets -Wl,-Bsymbolic-functions iirc
[15:17] <lool> ogra: Was it -Bsymbolic-functions?
[15:17] <lool> Ok
[15:17] <lool> ogra: I don't actually understand where the issue is there
[15:18] <ogra> the -Wl made it break (i didnt find it in ld's help output)
[15:18] <lool> ogra: man ld has it
[15:18] <ogra> if you find a fix i wont complain :)
[15:18] <lool> It's in the --help output as well
[15:19] <lool> arm-linux-gnueabi-ld: unrecognized option '-Wl,-Bsymbolic-functions'
[15:19] <lool> That's the failure I get when cross-building
[15:19] <ogra> that was the same i had
[15:19] <ogra> using the native ld
[15:20] <lool>                 cd $(LNDIR) && $(LD) $(LDFLAGS) $$UNDEF_SYM $(__OBJS) \
[15:20] <lool>                         --start-group $(__LIBS) --end-group $(PLATFORM_LIBS) \
[15:20] <lool>                         -Map u-boot.map -o u-boot
[15:20] <lool> I don't understand why it says unrecognized
[15:21] <ogra> well, but it does :)
[15:22] <lool> ogra: It certainly works with all the other packages, doesn't it?
[15:22] <ogra> no idea, i never ran into it so i wouldnt know
[15:22]  * ogra waits for his babbage to come up
[15:22] <ogra> i'm 100% positive there was no -Wl in ld --help
[15:23] <arthur-> ogra: -Wl,foo tells *gcc* to pass foo to ld
[15:24] <arthur-> but if you use ld directly, you just pass foo
[15:24] <ogra> ogra@babbage2:~$ ld --help|grep Wl
[15:24] <ogra> ogra@babbage2:~$
[15:24] <ogra> right, so its wrong in the LDFLAGS in this case
[15:24] <arthur-> $ gcc --help  | grep Wl
[15:24] <arthur->   -Wl,<options>            Pass comma-separated <options> on to the linker
[15:24] <ogra> yeah
[15:24] <ogra> but look at lool's example above
[15:25] <ogra> $(LD) $(LDFLAGS)
[15:25] <ogra> it calls ld directly
[15:25] <lool> arthur-: You think it's because it's passing -Wl to ld?
[15:25] <lool> That might be
[15:25] <arthur-> lool: depends on what $(LD) is
[15:25] <ogra> very likely because the whole source is set up for cross compiling by default
[15:25] <ogra> which we dont do at all
[15:26] <ogra> arthur-, LD is ld :)
[15:26] <lool> arthur-: it's ld
[15:26] <lool> or a cross-ld
[15:26] <lool> I guess the bug is that they should call gcc for ld
[15:27] <arthur-> lool: if this can work, this is what they should do yes
[15:27] <arthur-> lool: what is the full command (given by make output I guess)?
[15:28] <lool> LD      = $(CROSS_COMPILE)ld
[15:28] <lool> CC      = $(CROSS_COMPILE)gcc
[15:28] <arthur-> it's maybe possible to use gcc+objcopy instead of ld
[15:29] <lool> arthur-: http://paste.ubuntu.com/400594/
[15:30]  * lool tries biulding with -B... in LDFLAGS
[15:31] <ogra> -Bsymbolic-functions surely works
[15:31] <lool> ogra: You tried?
[15:32] <lool> arthur-: yeah that's the issue
[15:33] <lool> So upstream should have CCLD = $(CC) or something like that
[15:34] <ogra> lool, no, but i know that -Bsymbolic-functions is supported by ld ... while -Wl isnt
[15:35] <ogra> lool, feel free to discuss with sarkoman in #beagle :) he is upstream
[15:35] <ogra> though i think we're pretty unusual in the light that we build everything natively ... nobody usually does that for x-loader or uboot
[15:39] <lool> It's not a problem of building natively
[15:39] <lool> it's a problem of setting the same type of LDFLAGS for all software you build
[15:39] <lool> I have the same problem cross-building u-boot
[15:39] <ogra> ah
[15:40] <arthur-> lool: you can get ride of dpkg's ldflags here anyway, I guess there is no shared lib in u-boot
[15:42] <lool> arthur-: Well it builds some tools, but I prefer not dropping LDFLAGS entirely
[15:42] <lool> I added a snippet to convert them
[15:42] <lool> I thing it makes sense to turn on -O1 for all packages for instance
[15:42] <ogra> feel free to add it to the package
[15:43] <arthur-> lool: it it built from a regular debian process? (debian/rules, etc)
[15:44] <ogra> arthur-, yes
[15:44] <lool> arthur-: built and cross-built, yes
[15:44] <ogra> https://launchpad.net/ubuntu/+source/u-boot-omap3/2010.3git20100315-0ubuntu3
[15:44] <arthur-> then at the beginning of dbeian/rules, you can add something like
[15:44] <arthur-> define unsetenv
[15:44] <arthur->   unexport $(1)
[15:44] <arthur->   $(1) =
[15:44] <arthur-> endef
[15:44] <arthur-> $(foreach v, CPPFLAGS CFLAGS CXXFLAGS FFLAGS LDFLAGS, $(if $(filter environment,$(origin $(v))),$(eval $(call unsetenv, $(v)))))
[15:45] <arthur-> with a comment which says ubuntu flags are not made for building a boot loader :)
[15:45] <lool> arthur-: I did LDFLAGS := $(patsubst -Wl$(comma)%,%,$(LDFLAGS))
[15:45] <ogra> well, it currently has unexport LDFLAGS at the top of rules
[15:45] <lool> Which I think is nicer
[15:45] <ogra> and the binary works fine
[15:46] <lool> ogra: I didn't like unexporting LDFLAGS completely
[15:46] <ogra> indeed
[15:46] <ogra> there is always room for improvement :)
[15:46] <lool> The first thing I did was replacing this with LDFLAGS := $(filter-out -Wl$(comma)-Bsymbolic-functions, $(LDFLAGS))
[15:46] <ogra> i just needed a binary package for image building for now
[15:47] <dmart> asac: hi again... sorry, I got distracted
[15:49] <dmart> asac: see http://pastebin.ubuntu.com/400604/ for some code which adds some additional bx in the ARM-specific code.  It's probably best to have this for completeness, though it looks like the build flags for klibc and everything built against it may make this unnecessary.  We might want to protect with #ifndef __ARM_ARCH_4__ but we can probably consider <=armv3 obsolete for now.
[15:51] <ogra> dmart, did you see https://bugs.launchpad.net/ubuntu/+source/linux-fsl-imx51/+bug/457878/comments/8 ?
[15:51] <ubot4`> Launchpad bug 457878 in linux-fsl-imx51 (Ubuntu Lucid) (and 1 other project) "imx51 on board ethernet plug/unplug events not detected (affects: 2)" [Medium,In progress]
[15:51] <ogra> please test with your hub
[15:51] <ogra> it seems to solve a bunch of other issues so i'm hopeful
[15:52] <dmart> ogra: yes, currently in the process of testing it
[15:52] <dmart> I broke my boot SD card somehow, so I'm installing today's daily-live
[15:52] <ogra> ah
[15:53] <dmart> I'm hoping this will fix the problem
[15:59] <ogra> lool, you are such a nitpicker !
[15:59] <ogra>  * rules: minor whitespace cleanups.
[15:59] <ogra> heh
[15:59] <lool> ogra: Well my eyes bleed when I see \t\t before each command in rules
[15:59] <lool> (instead of \t)
[16:00] <lool> Not that it breaks anything but still
[16:00] <ogra> well, my vim is set up for python :)
[16:00] <ogra> using four spaces for tabs
[16:00] <lool> You should use your python setup *when editing python*
[16:00] <ogra> pfft details :)
[16:00] <lool> I hate usage of tabs outside of makefiles  :-(
[16:01] <ogra> thanks for the fixes !
[16:01] <lool> np
[16:01] <lool> It cross-debuilds now
[16:03] <lool> ogra: Did you push an x-loader package for bb?  and do you plan to add udebs to u-boot / x-loader?
[16:03] <ogra> i didnt plan udebs, no
[16:04] <ogra> lool, https://launchpad.net/ubuntu/+source/x-loader/
[16:11] <lool> ogra: thanks, seems to suffer from similar issues
[16:11] <ogra> yup
[16:11] <ogra> could need the same fixes, though note that there is a quilt patch to override the hardcoded CROSS_COMPILE in the Makefile
[16:13] <lool> Why didn't you put signGP.c in debian/>
[16:14] <ogra> seemed appropriate as a patch
[16:15] <ogra> i might replace it by gpsign.c anyway
[16:27] <lool> ogra: what's the license of signGP.c?
[16:28] <ogra> lool, GPL2 the prob is that it is not added to it anywhere
[16:28] <ogra> lool, which is the reason i'm considering gpsign.c instead
[16:28] <lool> ogra: Where is it stated?
[16:28] <ogra> lool, jkrindner stated that to me on IRC
[16:29] <lool> ...
[16:29] <ogra> but its not historically recoverable
[16:29] <ogra> http://github.com/nmenon/omap-u-boot-utils/blob/master/src/gpsign.c
[16:29] <ogra> its the same author (and accorting to jkrinder the same license)
[16:30] <ogra> but apparently it got checked in to the TI git without any further info
[16:30] <lool> ogra: What's $(sh) for in rules?
[16:30] <ogra> lool, i'll switch to gpsign but i havent found the time to test if it gets the same MLO for us yet
[16:31] <ogra> $(sh) execs a shell command, no ?
[16:31] <ogra> in a way that i dont need to convert it to make
[16:32] <lool> ogra: it's $(SHELL), but every line in a makefile target is run through SHELL
[16:32] <ogra> i had issues with that in the past
[16:32] <lool> ogra: $(sh) expands to nothing and is ignored, see your build log
[16:32] <ogra> hrm
[16:32] <ogra> well, then it can as well go away
[16:33] <ogra> if you touch the package feel free to do so
[16:41] <hrw> ehlo
[16:42] <prpplague> oh lord, who let hrw in here
[16:42] <hrw> prpplague: haha
[16:43] <hrw> prpplague: I am surprised seeing you here
[16:44] <prpplague> hrw: hehe indeed
[16:45] <prpplague> hrw: i'm not a big ubuntu fan, but my current contract has ubuntu requirements
[16:45] <ogra> sensible customer then :)
[16:45] <hrw> prpplague: ubuntu on arm?
[16:45] <ogra> prpplague, why dont i see you in #ltsp anymore ?
[16:45] <prpplague> ogra: yea you could say that
[16:46] <prpplague> hrw: yea
[16:46] <hrw> prpplague: nice.
[16:46] <hrw> prpplague: omap3 or other?
[16:46] <prpplague> ogra: just never do any LTSP items any more
[16:46]  * ogra will work on full omap integration for ltsp in 10.10 
[16:46] <prpplague> hrw: omap
[16:46] <prpplague> ogra: dandy
[16:47] <hrw> omaps are everythwere
[16:47] <hrw> next week I will find one in fridge and start screaming
[16:47] <prpplague> ogra: you using beagle boards for that?
[16:47] <ogra> well, beagles didnt make good thin clients in the past due to the missing NIC
[16:47] <ogra> but that might change :)
[16:47] <hrw> ogra: beware - prpplague will sell you zippy2
[16:47] <prpplague> ogra: hehe, *cough*
[16:48] <prpplague> ogra: you can always use a zippy or zippy2 with the revC boards, and the new beagleXM has usb ethernet
[16:48] <prpplague> main problem with the beagle board is that there isn't a good case for it
[16:49] <ogra> i think i saw some nice cases
[16:49]  * prpplague had planned to do the DogHouse, but backed out on it
[16:49] <ogra> lool, dont you have a beagle with case ?
[16:49] <lool> ogra: I do
[16:49] <prpplague> ogra: http://www.elinux.org/DogHouse
[16:50] <prpplague> lool: that the plexiglass case from special computing?
[16:50] <hrw> prpplague: usb ethernet + omap3... arhg combo
[16:50] <lool> prpplague: Yes
[16:50] <nosse1> ogra: I ran into a issue where I need to run qemu for a openembedded target and it complains about mmap_min_addr not set to zero. I see from a changelog in qemu-kvm that it is set to 4097. Is there any dangers/consequences for me to override it to 0?
[16:50] <jaustin> is it 'normal' for the moment that I'm seeing dependency issues with xubuntu-desktop in lucid? Seems to come down to libgtk2.0 not being at 2.20 yet
[16:51] <prpplague> ogra: you still doing alot of LTSP work?
[16:51] <ogra> nosse1, yes
[16:52] <ogra> nosse1, if you can avoid 0 you should
[16:52] <ogra> prpplague, sadly not
[16:52] <prpplague> ogra: ahh
[16:52]  * prpplague was big into LTSP from around 1999 to 2002
[16:53]  * ogra wrote most of ltsp5 until hardy
[16:53] <ogra> since then i moved over to mobile and got sucked into other work
[16:53] <prpplague> ogra: hehe i even had LTSP running on ......... SCO OpenServer
[16:53] <hrw> nosse1: qemu 0.12 does not need that setting anymore
[16:54] <ogra> hrw, for the syscall translation layer ?
[16:54] <ogra> lool, ^^^ did you know that ? we should drop it if its true
[16:55] <prpplague> btw, if anyone wants to provide feed back on things that they don't like about the beagle, i'd appreciate it
[16:55] <nosse1> ogra, thanks. It's probably an old security check which I can override
[16:55] <prpplague> (as well as wish list accessories)
[16:56] <lool> ogra: Did you try it out?
[16:56] <hrw> prpplague: I would like to have ftdi serial+jtag on miniusb
[16:57] <hrw> prpplague: like in sheevaplug
[16:57] <ogra> lool, nope
[16:57] <prpplague> hrw: hehe , flyswatter isn't good enough for you eh?
[16:57] <hrw> prpplague: and 4port usb hub integrated
[16:57] <suihkulokki> and a pony?
[16:57] <prpplague> hrw: the 4port hub will be in beagleXM
[16:57] <dmart> asac: ping?
[16:57] <hrw> suihkulokki: two of them
[16:58] <ogra> suihkulokki, sounds more like a herd of ponies :)
[16:58] <hrw> prpplague: I do not have it
[16:58] <prpplague> hrw: beagleXM or the flyswatter?
[16:59] <hrw> both
[16:59] <prpplague> hrw: why the heck not?
[17:00] <prpplague> hrw: anything else you'd like to see on the beagle?
[17:00] <hrw> prpplague: beaglexm looks like TI internal so far, flyswatter (or other jtagger) was not needed for me so far, except neo1973 but for this I have debugboard
[17:00] <prpplague> hrw: flyswatter is pretty handy to have around
[17:01] <nosse1> Coming back to my AM3517-EVM ubuntu bringup: I have now a working qemu running. And I am capable of booting and mounting the ubuntu root (either from NFS or SD). However it crashes during starting (i.e. after init has started but before login) with the error "init: procps main process terminated with status 255"
[17:01] <prpplague> hrw: yea, the XM isn't public yet
[17:01] <prpplague> nosse1: oh you have a AM3517 board?
[17:01] <nosse1> Yes
[17:01] <hrw> prpplague: my next home project is beagleboard + wifi + bt + usb hub + serial + hdmi
[17:01] <ogra> nosse1, sounds like your kernel is missing something
[17:01] <nosse1> trying to get it up now
[17:01] <prpplague> hrw: when are you planning to do that?
[17:01] <jaustin> the IGEPv2 has Ethernet+Wifi+Bluetooth on board, makes things very tidy
[17:02] <prpplague> nosse1: dandy, i'd love to get one of those up and running with socketcan for the canbus stuff
[17:02] <hrw> prpplague: next month, have all parts available so it will be size of two beagleboards side by side
[17:02] <hrw> jaustin: I have 3 beagleboards here
[17:02] <nosse1> ogra, yes it's probable the kernel is foo, because I'm using on delivered with the evalkit. How do you in debian world compile the kernel? (...given that I dont have a running target yet)
[17:03] <hrw> nosse1: tried OpenEmbedded kernels?
[17:03] <ogra> nosse1, there are docs for this somewhere on the ubuntu wiki and the guys in #ubuntu-kernel can help i guess
[17:03] <prpplague> nosse1: going to do anything automobile related with your board?
[17:04] <nosse1> hrw: Indirectly I guess, since TI uses Aragon, which is based on Ångström (which is based on OpenEmbedded) *puh*
[17:05] <hrw> Arago
[17:05] <dmart> ogra: Looks like Bryan's kernel fixes my slow Ethernet \ü/
[17:05] <nosse1> ogra: Thanks, I'll check it out. Principally I can cross compile the kernel (with no modules) and then perhaps build a native ubuntu kernel when the target it running
[17:05] <lool> dmart: Are you on a German keyboard?  :)
[17:06] <dmart> no, I just got a bit creative...
[17:06]  * lool finds it very delicate to send umlauts to ogra
[17:06] <armin76> lol
[17:06] <nosse1> prpplague: Not automotive, but semi-medical. I'd be happy to talk about it, but I think my boss it not...
[17:07] <dmart> ü is the nearst I can get to a vertial smiley face in iso8859-1
[17:07] <dmart> graphical dingbats would obviously be cheating...
[17:08] <prpplague> nosse1: hehe, np, i have the same restrictions
[17:08] <ogra> lool, lol
[17:08] <ogra> dmart, yay !
[17:08] <ogra> i *knew* it !
[17:09] <dmart> I'll wait for the real .deb, but it looks good
[17:10] <nosse1> Is there a "set -x" or debug setting in upstart for it to output every stage of the init/boot process?
[17:11] <ogra> i dont think so
[17:11] <dmart> nosse1: upstart init has a -v or --verbose switch
[17:11] <ogra> right, but hosw do you call that
[17:11] <ogra> *how
[17:11] <dmart> When I need this I typicall boot with init=/bin/bash and then exec init manually
[17:11] <ogra> oh, indeed ;)
[17:12] <dmart> init=/sbin/init -v  may also work if it's at the end of the kernel command line
[17:12]  * ogra was wondering if init= can take arguments
[17:12] <nosse1> I can try that
[17:12] <dmart> I also find it useful to pass --debug to mountall in /etc/init/moutall.conf, if you're getting filesystem mounting problems
[17:12] <ogra> yeah
[17:12] <dmart>  /mountall.conf
[17:13] <dmart> init= _can_ take arguments ... init=/bin/bash blarg  ->
[17:13] <jaustin> How can I debug random restarts when using 91.0 and 2.6.33.1?
[17:13] <dmart> bash: Cannot execute blarg
[17:13] <dmart> PANIC
[17:13] <dmart> (etc.)
[17:13] <ogra> ah, i never tried that
[17:13] <jaustin> I don't see anything special over the serial console, it just restarts
[17:14] <ogra> i would have expected you need some magic quoting
[17:14] <jaustin> I'm using an IGEPv2
[17:14] <plars> dmart: this was the one where it was slowing down when going through a hub, but not a switch?
[17:14] <dmart> plars: yes
[17:14] <plars> dmart: good to hear :)
[17:14] <nosse1> dmart: Yes! I'm in.
[17:14] <dmart> cool
[17:16] <nosse1> eh? what is the default init string from the kernel. isn't it "/sbin/init"... What I try to run /sbin/init manually, it complains about missing runlevel
[17:16] <ogra> lool, debian/clean is that new ?
[17:16]  * ogra never heard of it
[17:17] <dmart> nosse1: init must run as pid=1, so you must type exec /sbin/init -v (not just /sbin/init -v)
[17:17] <dmart> ...otherwise init tries to be initctl
[17:17] <dmart> or telinit or something
[17:18] <dmart> any luck
[17:18] <dmart> ?
[17:18] <nosse1> d'oh
[17:18] <nosse1> thanks
[17:18] <dmart> no probs
[17:20] <hrw> nosse1: /sbin/init then /bin/init then /linuxrc iirc
[17:24] <nosse1> hrw: You asked if I have tried openembedded kernel. Does this imply that I possibly can use a kernel compiled from OE?
[17:27] <hrw> nosse1: personally I would try that at least
[17:27] <hrw> nosse1: usually our kernels works with our targets
[17:28] <nosse1> Yes. Compiling a crossed OE kernel is easier. I just need to find what kernel resources ubuntu requires. I can probably ask on #ubuntu-kernel
[17:29] <hrw> nosse1: it is probably linux-omap-psp 2.3.32 tree
[17:30] <ogra> hrw, note that ubuntu userspace often uses kernel features from the ubuntu kernel
[17:30] <ogra> i.e. OE might not have everything enabled the ubuntu userspace expects to exist
[17:31] <nosse1> Perhaps it would be easier by looking at the source package of another Cortex-A8 kernel to find these core linux kernel settings
[17:32] <jaustin> ogra: any advice on how to best get a kernel with the features Ubuntu userpsace expects if the board vendor's kernel isn't mainline? (but is ina git repo)
[17:32] <ogra> just take the imx51 or dove kernel config
[17:32] <hrw> ogra: thats a matter of setting.
[17:32] <ogra> jaustin, well, compare the configs is the best i can recommend
[17:33] <nosse1> pardon my naivity: Where can I find the imx51 config?
[17:33] <jaustin> ogra: right, so you're talking config and not features missing from the source?
[17:34] <hrw> nosse1: in linux-image package for imx51 board
[17:34] <ogra> jaustin, well, config but also features like apparmor and friends for example
[17:35] <dmart> Would it be worth diffing the ubuntu versatile config against mainline?  That might be useful to highlight non-platform-specific options turned on for Ubuntu
[17:35] <jaustin> ogra:  I see. the main issue I ran in to was udev and 2.6.28 - which is why I'm now trying to use 2.6.33
[17:35] <ogra> ah, yeah udev and kernel go hand in hand
[17:35] <dmart> My experience is that using a kernel <2.6.31 with lucid is a waste of effort
[17:35] <ogra> same for other stuff
[17:35] <ogra> i.e. mountall
[17:36] <ogra> it expects devtmpfs to exist in lucid
[17:36] <dmart> You can work around mountall by putting "ignore" entries in fstab, or /dev ... tmpfs (but it's not really worth the work)
[17:37] <ogra> yeah
[17:37] <ogra> or by just properly using an initramfs
[17:37] <ogra> lucids initramfs takes care if devtmpfs doesnt exist
[17:37] <jaustin> In Karmic I had everything I wanted except pulsaudio with 2.6.28. I got sound using the jaunty udev rules, but pulseaudio never got onboard.
[17:38] <dmart> I think now /lib/init/fstab has /dev ... devtmpfs,tmpfs (so it will fall back to tmpfs if needed)
[17:38] <dmart> ogra: is the devtmpfs code in mainline?  I wasn't sure whether that was ubuntu-specific
[17:39] <ogra> yes
[17:39] <ogra> its mainline
[17:39] <dmart> oh, ok
[17:39] <ogra> but only in .32
[17:39] <ogra> and later indeed
[17:39] <ogra> for imx51 we had to backport it
[17:39] <ogra> speeds up booting significantly
[17:39] <dmart> did we backport this to .31 for imx51?  I seem to remember traffic about it
[17:39] <ogra> yes
[17:39] <dmart> makes sense
[17:47] <nosse1> In ubuntu, what do you do in initramfs before heading into main boot (apart for loading any required modules)?
[17:52] <ogra> initializing things like frambuffer, running early udev ... getting your password in case you got any encrypted partionions, setting up dm-raid or LVM etc etc
[17:53] <nosse1> So I guess you use it on the imx51 target as well then. Thanks
[17:55] <ogra> there is no ubuntu without initramfs so yes, we use it in imx51 too
[17:56] <ogra> there are a bunch of packages that rely on initramfs bits
[18:03] <DanaG> hmm, but how do you get u-boot to USE initramfs?
[18:04] <hrw> /var/run/dhclient.wlan0.pid: interface name too long (max 27)
[18:04] <hrw> someone saw: "/var/run/dhclient.wlan0.pid: interface name too long (max 27)" on debian/ubuntu?
[18:05] <ogra> DanaG, you load it to the appropriate address and attach that address to the bootm command
[18:08] <hrw> have  a nice rest of day
[18:08] <ogra> gah, crap
[18:08] <ogra> u-boot can only fatload from partitioned devices
[18:09] <ogra> that makes image building trickier :/
[18:10] <ogra> which is a bit funny since x-loader apparently loads u-boot.bin form that unpatitioned SD
[18:11]  * ogra will go on with that tomorrow
[18:33] <nosse1> ogra: I'm having trouble starting qemu with tap using the description on RootfsFromScratch. I'm getting could not configure /dev/net/tun: no virtual network emulation. Perhaps the /etc/qemu-ifup isn't being run?
[18:54] <nosse1> I'm sorry, I seem to have network problems. I don't know if ogra has answered my Q
[18:56] <armin76> nosse1: he hasn0t
[18:56] <armin76> hasn't
[20:51] <lool> ogra: it's not any different for beagle than for dove?
[20:52] <lool> Wee!
[20:52] <lool> x-loader built and is now cross-buildable too
[20:53] <dcordes> lool, dove ?
[21:00] <lool> dcordes: mach-dove, the marvell image we output
[21:00] <lool> to support the soc of the same name
[21:00] <lool> dcordes: ARMADA
[21:05] <dcordes> marvell..
[22:27] <dcordes> I'm looking forward to run ubuntu on my htc hd2
[23:30] <DanaG> rcn-ee: interesting stuff with the new 33-500-omap kernel.
[23:31] <DanaG> One: mtd doesn't seem to work.  /dev/mtd stuff doesn't exist.
[23:31] <DanaG> Another: I sometimes get an oops and a backtrace upon initializing twl4030_usb
[23:31] <rcn-ee> Hey DanaG, sorry been gone the last two days.. ;)  which 500.X? ;)
[23:31] <DanaG> linux-image-2.6.32-500-omap, in lucid.
[23:32] <rcn-ee> yeah, the package has a 500.x number on it.. uname -a might show.. ;)
[23:32] <DanaG> ah
[23:32] <rcn-ee> 500.1 broken, 500.2 will boot.. ;)
[23:33] <DanaG> [    0.000000] Linux version 2.6.33-500-omap (buildd@korlan) (gcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu2) ) #3-Ubuntu Tue Mar 23 20:29:18 UTC 2010
[23:33] <rcn-ee> humm... crap...
[23:35] <rcn-ee> DanaG, i submited a patch to ubuntu-kernel for the usb problems. ubuntu would like to first try to have it loaded as an external module.. I'd like to bring it internal.. ;) this patch: http://bazaar.launchpad.net/~beagleboard-kernel/%2Bjunk/lucid-ti-omap/annotate/head%3A/patches/beagle/0001-UBUNTU-Config-Enable-musb-ehci-on-beagleboard.patch
[23:36] <DanaG> unfortunately, I can't figure out how to grab all the output through "screen"
[23:36] <rcn-ee> use cutecom, it has a log to disk feature..
[23:37] <rcn-ee> other than that it's not that good, so i'm constantly switching back and forth..
[23:38] <DanaG> ah, screen -L /dev/ttyUSB0 115200
[23:38] <DanaG> -L for log.
[23:39] <DanaG> oh, and something weird: the only place I've seen the gpio-keys input thingy appear and work, is in the Angstrom kernel.
[23:40] <rcn-ee> hum, must have mis did one of the gpio settings in the config..
[23:41] <DanaG> www.csc.calpoly.edu/~dgoyette/screenlog.0
[23:42] <DanaG> weird... last time I rebooted the thing, it gave an oops and hung.
[23:42] <DanaG> this time it rebooted fine.
[23:47] <DanaG> anyway, the host sees "device descriptor read error"
[23:47] <rcn-ee> sorry DanaG, was quickly writing a howto.. with ubuntu's kernel we need to make an initramfs...
[23:47] <rcn-ee> take a look at: http://elinux.org/Talk:BeagleBoardUbuntu#Conversion_to_Ubuntu_ti-omap_kernel
[23:48] <DanaG> cool
[23:48] <rcn-ee> it's a little of a mess at this point... ;)
[23:49] <rcn-ee> 1. boot with no initrd.. 2. mount fat16 partition (fails) 3. run setup.sh script 4. reboot 5. some things work better..
[23:51] <rcn-ee> i have a patch for 2. in my tree, but i don't think ubuntu will exept that one.. i think they are planning to generate their own uIntrd's on release... it's just a chicken and egg problem for us developers..
[23:51] <DanaG> hmm, why does it copy to boot.ini?
[23:51] <rcn-ee> IGEPv2's boot loader looks for that. .;)
[23:51] <rcn-ee> I'm hacking on three boards at once.. bx, IGEPv2 and xm..
[23:52] <DanaG> http://www.csc.calpoly.edu/~dgoyette/screenlog.0.txt
[23:53] <rcn-ee> yeap that's the musb module...
[23:53] <DanaG> "unable to read boot.scr
[23:53] <DanaG> "
[23:53] <DanaG> weird... u-boot doesn't see the file!
[23:55] <rcn-ee> i'm still suprised it mounted your sd card... mine didn't like that at all..