**** BEGIN LOGGING AT Fri Jun 17 23:59:56 2005
Jun 18 00:00:06 <eno-away> hi ka6sox-away, recently i actually got tightvnc sort of working
Jun 18 00:01:02 <eno-away> need to talk to jp30 about some x11 stuff
Jun 18 00:02:52 <ka6sox-away> eno-away, cool! 
Jun 18 00:03:03 <ka6sox-away> jp30 is on holiday for a bit.
Jun 18 00:03:15 <ka6sox-away> but when he gets back I'm sure he would enjoy talking about it.
Jun 18 00:03:36 <eno-away> i guessed that, when he's back, he'll have a few more pkgs to play with
Jun 18 00:04:00 <ka6sox-away> yes...I see that...very good.
Jun 18 00:06:27 <ka6sox-away> I wonder how many of these unslung packages will just work in openslug...
Jun 18 00:23:58 <dyoung-zzzz> jbowler tried a bunch of unslung packages and they just worked.
Jun 18 00:24:19 <dyoung-zzzz> not building them I dont think, just a curiosity ipkg install type deal.
Jun 18 01:05:02 <jbowler-away> The ones I tried semi-worked - which is to say that I only got one 'shared library not found' (libz.so, needs a symlink) and no cases of missing symbols, or missing dependent objects.
Jun 18 01:05:57 <jbowler-away> I tried which (worked) and ctorrent (ran, but didn't seem to do anything - probably I don't know how to use it.)
Jun 18 01:06:50 <jbowler-away> I've also noticed that glibc packages for the most part work on a uclibc openslug - at the cost of installing glibc...
Jun 18 01:08:38 <ka6sox-away> okay so we really will need to build a complete openslug feed.
Jun 18 01:14:20 <jbowler-away> certainly: the unslung startup scripts won't typically work and the installation ends up in /opt
Jun 18 01:14:55 <jbowler-away> So even though the executables may work fine they're not really packaged right.
Jun 18 01:15:26 <jbowler-away> The complexity with openslug is that there probably should be two feeds - one for glibc and one for uclibc.
Jun 18 01:15:32 <ka6sox-away> so we should expect them to be in "normal" places instead...like /usr/local/bin
Jun 18 01:17:18 <jbowler-away> Well, installing in /opt is fine IMO, but the standard stuff is in /usr (etc), not in /opt and libraries should probably be in /usr/lib to avoid LD_LIBRARY_PATH issues.
Jun 18 01:17:58 <jbowler-away> All the OE stuff goes in /usr
Jun 18 01:18:00 <ka6sox-away> k
Jun 18 02:54:19 <ilg> I tried to build openslug and got up to compiling glibc:
Jun 18 02:56:10 <ilg> elf-init.c:(.text+0x54): undefined reference to `__init_array_start'
Jun 18 02:56:16 <ilg> elf-init.c:(.text+0x58): undefined reference to `__init_array_end'
Jun 18 02:56:26 <ilg> elf-init.c:(.text+0xa8): undefined reference to `__fini_array_start' 
Jun 18 02:56:26 <ilg> elf-init.c:(.text+0xac): undefined reference to `__fini_array_end' 
Jun 18 02:56:26 <ilg> collect2: ld returned 1 exit status 
Jun 18 02:57:23 <ilg> I just did :
Jun 18 02:57:25 <ilg> (cd openembedded ; bk pull ; bk -r co -q) 
Jun 18 02:57:25 <ilg> (cd nslu2-package-symlinks ; bk pull ; bk -r co -q) 
Jun 18 02:57:25 <ilg> (cd build; bitbake -v cfetch $MYIMAGE-image)
Jun 18 02:57:38 <ilg> so I assume I'm running from a fresh copy
Jun 18 02:57:44 <ilg> any suggestions?
Jun 18 02:59:29 <rwhitby-away> dunno, haven't pulled lately
Jun 18 03:00:18 <ilg> I'm duing now -c clean world and I'll retry
Jun 18 03:00:56 <rwhitby-away> do an rm -rf tmp for a real clean
Jun 18 03:10:10 <ilg> the rm -rf tmp took quite some time... build started...
Jun 18 03:14:24 <CIA-12> 03marceln * 10unslung/make/net-snmp.mk: 
Jun 18 03:14:24 <CIA-12> The package build and works.
Jun 18 03:14:24 <CIA-12> TODO: Cleaning up, configuration
Jun 18 04:30:34 <ilg> (while waiting for the long build to run) how is your build environment organized? do you have completely different directories with unslung/openslug or do you share a single directory for both of them?
Jun 18 04:33:53 <NAiL> dyoung-zzzz: the wget died. Can I use your mirror?
Jun 18 04:33:55 <NAiL> :)
Jun 18 04:37:55 <dyoung-zzzz> ok
Jun 18 04:38:03 <dyoung-zzzz> what method do you prefer?
Jun 18 04:40:15 <mr_claus> morning
Jun 18 04:42:27 <NAiL> rsync?
Jun 18 04:44:25 <dyoung-zzzz> mail me a key.
Jun 18 04:55:35 <NAiL> rwhitby-away: ping?
Jun 18 05:02:03 <mr_claus> anybody here is working with openslug sources?
Jun 18 05:02:15 <dyoung-zzzz> whats the question
Jun 18 05:02:47 <mr_claus> i want to build the image from the source because i need ext3 acl support
Jun 18 05:03:12 <mr_claus> in nslu2-linux/packages/linux/openslug-kernel-2.6.9/defconfig the acl-support is disabled
Jun 18 05:03:23 <dyoung-zzzz> thats not the default kernel
Jun 18 05:03:27 <dyoung-zzzz> you want 2.6.11.2
Jun 18 05:03:38 <dyoung-zzzz> or override it in your local.conf if you really do want 2.6.9
Jun 18 05:04:08 <mr_claus> ok, i want 2.6.11.2 but there is no ACL support included too
Jun 18 05:04:14 <mr_claus> so i have to change defconfig?
Jun 18 05:04:26 <dyoung-zzzz> so again I ask, whats is the specific question?
Jun 18 05:04:39 <mr_claus> how i add acl-support
Jun 18 05:04:59 <dyoung-zzzz> I havnt any idea.  If you need to change the defconfig, bk edit it.
Jun 18 05:05:11 <dyoung-zzzz> beyond that, I dunno
Jun 18 05:06:48 <mr_claus> ok, to make the image only go into the main directory and type in make?
Jun 18 05:07:19 <dyoung-zzzz> have you found the wiki page on how to build using oe/bitbake?
Jun 18 05:07:31 <mr_claus> no
Jun 18 05:07:44 <mr_claus> thats why i ask here :)
Jun 18 05:07:45 <dyoung-zzzz> http://www.nslu2-linux.org/wiki/OpenEmbedded/HomePage
Jun 18 05:07:57 <mr_claus> thx
Jun 18 05:08:26 <dyoung-zzzz> the command youre looking for is "bitbake openslug-image" .  But youre going to have to setup some stuff first, so look around that page a bit.
Jun 18 05:08:56 <mr_claus> ok
Jun 18 05:13:16 <dyoung-zzzz> nite all
Jun 18 05:15:12 <NAiL> nite
Jun 18 05:17:15 <mr_claus> whats up with that error? ERROR: myProgressCallback() takes exactly 3 arguments (5 given) while parsing ........
Jun 18 05:17:44 <mr_claus> perhaps i missed something
Jun 18 05:26:02 <NAiL> rwhitby: :)
Jun 18 05:26:34 <rwhitby> ilg: I have /home/unslung and /home/openslug, and /home/bitkeeper.  In bitkeeper I have openembedded and nslu-package-symlinks.  Then I symlink openembedded and nslu2-package-symlinks into both unslung and openslug.
Jun 18 05:27:06 <NAiL> rwhitby: I suddenly became a openslug dev on sf.
Jun 18 05:27:44 <rwhitby> mr_claus: your bitbake version is probably too old
Jun 18 05:28:21 <rwhitby> NAiL: what is your sf.net id ?
Jun 18 05:28:26 <NAiL> orepvik
Jun 18 05:28:46 <rwhitby> NAiL: you want CVS access too?
Jun 18 05:29:29 <mr_claus> rwhitby: if i download the source.tar.gz of the beta then i dont need any other packages?
Jun 18 05:29:37 <mr_claus> rwhitby: there is bitbake included too
Jun 18 05:29:52 <mr_claus> rwhitby: now i downloaded bitbake with svn and got that error
Jun 18 05:29:56 <NAiL> rwhitby: Not a too bad idea
Jun 18 05:29:57 <rwhitby> mr_claus: you will still need all the upstream package sources, but OE will download them
Jun 18 05:30:32 <mr_claus> rwhitby: but i dont need bitkeeper right?
Jun 18 05:30:51 <rwhitby> you will if you want to stay up to date with the absolute latest
Jun 18 05:31:03 <mr_claus> :/
Jun 18 05:31:07 <rwhitby> otherwise, the beta release sources are in svn on berlios.de
Jun 18 05:31:19 <rwhitby> they are updated each beta release
Jun 18 05:32:01 <mr_claus> do you think i should use the latest or should the beta be enough?
Jun 18 05:32:32 <rwhitby> up to you.  openslug is quite dynamic at the moment.
Jun 18 05:32:49 <rwhitby> NAiL: done.
Jun 18 05:33:24 <NAiL> rwhitby: Thanks. I'll start syncing my mirror now
Jun 18 05:33:28 <mr_claus> rwhitby: ok, i will start with the beta for the first testing and will bitkeeper later
Jun 18 05:34:34 <rwhitby> mr_claus: note that you will need to have the correct version of bitbake to match the beta, and it may not be the latest bitbake.
Jun 18 05:35:05 <rwhitby> the svn repo has the correct version listed in the svn:externals property in the releases branch
Jun 18 05:35:28 <mr_claus> rwhitby: there is a bitbake version included with the sources too
Jun 18 05:36:15 <mr_claus> rwhitby: last try was with that version "svn co svn://svn.berlios.de/bitbake/trunk/bitbake"
Jun 18 05:37:25 <rwhitby> if you check out openslug from svn, it will include the correct version of bitbake as an external
Jun 18 05:37:59 <rwhitby> mr_claus: feel free to create a new wiki page which describes how to build the latest beta from svn
Jun 18 05:45:14 <rwhitby> So is there anyone here who needs a MAX3232 level converter board?
Jun 18 05:45:35 <mr_claus> rwhitby: it it the board to get serial console access to nslu2?
Jun 18 05:45:42 <rwhitby> A community member has one spare, and has asked me if there is a developer who would benefit from one.
Jun 18 05:46:06 <rwhitby> Yes, it is the correct chip to convert the levels from RS232 to the onboard serial port on the CPU.
Jun 18 05:46:16 <NAiL> odd... why can I login to shell.sf.net, but not rsync through ssh?
Jun 18 05:46:32 <rwhitby> NAiL: did you recently add a key?
Jun 18 05:46:35 <mr_claus> rwhitby: i would be interested in the board perhaps
Jun 18 05:47:19 <NAiL> rwhitby: Yeah. So I can login to the shell. 
Jun 18 05:47:40 <NAiL> but when I try to rsync -e ssh, it can't connect at all
Jun 18 05:47:44 <rwhitby> NAiL: sometimes there is a 6 hour delay for some things on sf.net
Jun 18 05:48:29 <NAiL> rwhitby: Yeah, but why would it matter when it allows me to login manually?
Jun 18 05:48:30 <rwhitby> check the "upload" target in the Unslung packages Makefile for what we use to rsync up to sf.net
Jun 18 05:48:33 * NAiL doesn't understand
Jun 18 05:48:49 <rwhitby> oh, you meant login with ssh ...
Jun 18 05:48:56 <rwhitby> dunno then.
Jun 18 05:50:40 <NAiL> rwhitby: The donation drive won't close for a while will it? 
Jun 18 05:51:05 <NAiL> I'm a student, so the funds aren't that great until the end of the month ;)
Jun 18 05:51:30 <rwhitby> NAiL: you're providing a package mirror - you aren't expected to donate too
Jun 18 05:52:52 <NAiL> rwhitby: heh, well... Donating a few bucks won't hurt
Jun 18 05:53:22 <rwhitby> There are perhaps 100 people who contribute in other ways (packages, wiki pages, developers, infrastructure, doco, etc).  It's the other four thousand or so uses who should be donating :-)
Jun 18 05:53:53 <rwhitby> (assuming that the firmware has some value for them)
Jun 18 05:54:11 <rwhitby> s/uses/users/
Jun 18 05:54:58 <rwhitby> anyway, the donation drive will close maybe a week or so after we have enough money to buy the things we need this time around.
Jun 18 05:55:39 <rwhitby> maybe we should put the paypal addresses in the next version of the README ...
Jun 18 05:56:30 <rwhitby> but I think it's more accountable if we have specified purchases identified before we ask for donations.  then people know the donations are not just going into someone's pocket :-)
Jun 18 05:59:11 <copperbeech> anyone ever tries MS Virtual PC to install linux ?
Jun 18 06:00:16 <rwhitby> not me.  Only VMWare and CoLinux.
Jun 18 06:01:19 <copperbeech> Hmmm - I just downloaded it on my MSDN subscription - it's very easy to use... bit scary really!
Jun 18 06:02:05 <copperbeech> Currently installing Debian on it, but booted Knoppix and Ubuntu from the ISOs with no problems.
Jun 18 06:02:33 <copperbeech> (except the 16bit colour depth and turning off knoppix dhcp.
Jun 18 06:02:44 <copperbeech> heh :-)
Jun 18 06:11:17 <NAiL> hmm
Jun 18 06:11:34 <NAiL> rsync insists on connecting on 873
Jun 18 06:14:30 <rwhitby> you must have something local
Jun 18 06:14:36 <rwhitby> doesn't do that for me
Jun 18 06:18:49 <NAiL> ah, now it works :)
Jun 18 06:21:58 <NAiL> mirror should be up in a couple of hours then
Jun 18 06:25:08 <NAiL> hmm... get ~3mbit from sf
Jun 18 06:27:53 <rwhitby> consider yourself lucky :-)
Jun 18 06:33:52 * NAiL considers himself lucky!
Jun 18 08:25:31 <NAiL> wrote 16 bytes  read 339801 bytes  29549.30 bytes/sec
Jun 18 08:25:31 <NAiL> total size is 1567523385  speedup is 4612.85
Jun 18 08:25:48 <NAiL> Mirror is synced with sf.net now.
Jun 18 08:27:59 * NAiL goes off to party :D
Jun 18 09:18:56 <CIA-12> 03bzhou * 10unslung/ (make/coreutils.mk sources/coreutils/rc.coreutils): /usr/bin/env symlink to /opt/bin/env on slug
Jun 18 09:23:07 <ka6sox-zzzz> eno is a machine!
Jun 18 09:24:47 <eno> this is a small change, and a bit controvertial
Jun 18 09:25:23 <eno> i took a safe approach, and conditionally symlink /usr/bin/env only for the slug
Jun 18 09:26:12 <ka6sox-zzzz> being safe is always a good plan.
Jun 18 09:27:40 <eno> see http://groups.yahoo.com/group/nslu2-linux/message/6657
Jun 18 09:28:53 <ka6sox-zzzz> let jeanfabrice or a wl500g dev add the changes.
Jun 18 09:29:29 <eno> they can't, because on wiley they don't have a writable /usr
Jun 18 09:30:49 <ka6sox-zzzz> they could patch it for /opt/bin/env
Jun 18 09:33:14 <eno> i guess for scripts that we have control, we should generally use /opt/bin/env; this change improves the portability of ad hoc scripts a little bit for slug
Jun 18 09:33:27 <ka6sox-zzzz> k
Jun 18 09:34:00 <ka6sox-zzzz> for OpenSlug it would have to use /usr/bin/env again..
Jun 18 09:34:14 <ka6sox-zzzz> so for 2 distro's this current method would work.
Jun 18 09:34:49 <eno> right
Jun 18 09:35:43 <eno> can you usually run unslung packages on openslug?
Jun 18 09:36:32 <ka6sox-zzzz> well...the answer I got lastnight is that usually they *almost* work.
Jun 18 09:38:33 <eno> i c
Jun 18 09:39:56 <ka6sox-zzzz> while Unslung has a rich series of libraries available, OpenSlug is more of a blank canvas "What do want your Slug to REALLY be?"
Jun 18 09:42:38 <eno> i'm considering moving one of my two slugs to OpenSlug
Jun 18 09:43:00 <eno> then when i build a package, i can test for both
Jun 18 09:43:13 <ka6sox-zzzz> good plan.
Jun 18 10:05:02 <ilg> I'm experimenting with openslug and although loseup is available (via busybox), it seems that kernel loop support is missing. 
Jun 18 10:05:08 <ilg> losetup
Jun 18 10:09:32 <[g2]> ilg what are you trying to do ? and how do you like openslug so far ?
Jun 18 10:10:17 <ilg> I'm still waiting it to compile, now I'm using the binary version
Jun 18 10:10:40 <[g2]> excellent
Jun 18 10:10:51 <[g2]> you're building with OE ?
Jun 18 10:10:58 <ilg> I'm trying to switch root fs to my flash disk
Jun 18 10:11:10 <ilg> yes
Jun 18 10:11:30 <[g2]> before switching your root fs I'd wait for you compile to finish
Jun 18 10:11:45 <[g2]> there have been lots of changes since the last beta in that area
Jun 18 10:11:51 <ilg> ok
Jun 18 10:11:55 <[g2]> things are much improved
Jun 18 10:12:13 <[g2]> you can play around on the jffs2 
Jun 18 10:12:37 <[g2]> are there any specific packages you are interested in ?
Jun 18 10:12:58 <ilg> regarding usb support, I was surprised to see that my usb flash appears as sda1 or sdb1 although I insered in in the same port
Jun 18 10:13:20 <[g2]> it's just like normal linux
Jun 18 10:13:31 <[g2]> first device plugged in is sda
Jun 18 10:13:36 <[g2]> next sdb
Jun 18 10:13:45 <ka6sox-zzzz> ilg: great suggestion for reorganizing the Development index
Jun 18 10:14:06 <ilg> even if I remove it and insert in the same port it is normal to increase index?
Jun 18 10:14:15 <[g2]> Linksys had a patch for forcing the ports to specific devices and limited the funcitionality that way
Jun 18 10:14:24 <[g2]> not normally
Jun 18 10:14:44 <[g2]> but it depends on how fast you do it probably
Jun 18 10:15:21 <[g2]> the hotplug on the re-insertion my occur before the old unplug cleaned up the drivers
Jun 18 10:15:42 <[g2]> that's just a guess on my part though
Jun 18 10:15:54 <jacques> he has a point - I can't find loop module or builtin in latest openslug
Jun 18 10:16:40 <[g2]> ipkg list | grep loop
Jun 18 10:16:40 <[g2]> kernel-module-loop - 2.6.11.2-r12 - loop kernel module
Jun 18 10:16:57 <[g2]> that's yesterdays build
Jun 18 10:17:14 <jacques> ah heh
Jun 18 10:17:15 <jacques> lol
Jun 18 10:17:28 <jacques> I keep forgetting I need to install packages for kernel modules
Jun 18 10:17:41 <[g2]> nod
Jun 18 10:17:52 <jacques> I'm used to having them all there (I would just build the kernel and copy all the modules I'd built over to the image)
Jun 18 10:17:53 * [g2] keeps forgetting *lots* of stuff :)
Jun 18 10:18:37 <[g2]> yeah that poor 5-6MB jffs2 budget
Jun 18 10:18:39 <ilg> do you think it is reasonable to use rootfs on a file mounted via loop from the vfat flash?
Jun 18 10:18:52 <[g2]> no
Jun 18 10:19:01 <[g2]> well maybe
Jun 18 10:19:23 <ilg> I see no other way. I do not want to reformat my flash
Jun 18 10:19:45 <jacques> heh, that's a pretty big restriction
Jun 18 10:19:45 <ilg> ut I can reserve a large part of it as a single file and format this area 
Jun 18 10:20:14 <[g2]> I think you might have to tweak the /linuxrc and boot scripts
Jun 18 10:20:26 <ilg> I don't mind this
Jun 18 10:20:53 <[g2]> and hopefully you're loopmount is not vfat
Jun 18 10:21:46 <[g2]> i'd imagine that might make wear-leveling work a little differently
Jun 18 10:21:50 <[g2]> but dunno
Jun 18 10:22:30 <ilg> I think my flash has 100.000 cycles, if not even 1.000.000
Jun 18 10:23:10 <ilg> and anyway I wanted to format the loopmounted file as linux (jffs2 or ext)
Jun 18 10:23:10 <[g2]> nod, but if you are always writing inside a given vfat file I'm not sure how the wear leveling is setup
Jun 18 10:23:32 <[g2]> I don't think it can be jffs2 it's got to be ext
Jun 18 10:23:44 <jacques> agreed
Jun 18 10:24:20 <[g2]> jacques, you didn't get samba running btw did you ?
Jun 18 10:24:41 <jacques> no, I have 0 interest in samba
Jun 18 10:24:52 <jacques> maybe you're confusing me with someone else ?
Jun 18 10:25:28 <[g2]> I think we were just talking about versions yesterday with jbowler 
Jun 18 10:25:52 <jacques> yeah I just made some crack about /etc/init.d/samba start which turned out to be totally wrong :-)
Jun 18 10:26:03 <jacques> sorry if I gave the wrong impression
Jun 18 10:26:10 <[g2]> Oh... NP
Jun 18 10:26:19 <[g2]> I'm not into samba either, but I'
Jun 18 10:26:29 <[g2]> I'd like to see if it runs
Jun 18 10:26:42 <[g2]> beewoolie was curious
Jun 18 10:27:01 <[g2]> NFS client and server run and samba was next on the list
Jun 18 10:27:01 <ilg> what method do you recommend at linuxrc/rc time to reliably detect the usb flash is inserted?
Jun 18 10:27:25 <[g2]> ilg look at the linuxrc sripts
Jun 18 10:27:29 <[g2]> ilg look at the linuxrc scripts
Jun 18 10:27:35 <[g2]> and I think the /boot scripts
Jun 18 10:28:45 <ilg> it seems I have to do it again, more carefully
Jun 18 10:30:14 <[g2]> heh no smb.conf could be a key to starting samba properly :)
Jun 18 10:33:59 <ilg> I'm looking at /linuxrc, but I do not see where you test if the device is plugged. all I see is a test of a file like /.sd[ab][12]root which is different. am I 'usb-test' blind?
Jun 18 10:34:38 <ilg> I see an attempt to do a mount
Jun 18 10:37:24 <ilg> ok, based on the filename  /.sd[ab][12]root it tries to mount /dev/sd[ab][12]  on /mnt
Jun 18 10:41:20 <ilg> how complicated would be to run openslug with the root jffs2 fs mounted as read only?
Jun 18 10:48:38 <[g2]> not super hard
Jun 18 10:49:40 <ilg> for production environments (as I'm considering) that would be an interesting option
Jun 18 10:50:18 <[g2]> I'm making hardware what kind of production environment are you looking at ?
Jun 18 10:51:33 <ilg> I'm looking at a device to be installed in a datacenter, to provide network boot and network install support to a remote server
Jun 18 10:52:06 <[g2]> nod. I've been thinking trying that at home plus the WOL stuff
Jun 18 10:52:17 <[g2]> I currently can boot via PXE
Jun 18 10:52:33 <[g2]> I just haven't setup openslug to feed the process
Jun 18 10:53:01 <ilg> that's what I'm looking or.
Jun 18 10:53:03 <ilg> for
Jun 18 10:53:21 <ilg> dhcp/tftp/etc
Jun 18 10:53:35 <[g2]> nod. it's pretty trivial with OpenSlug
Jun 18 10:54:04 <[g2]> I just have setup the external CDROM with the media :)
Jun 18 10:54:11 <ilg> great! but for this I need the rootfs ro.
Jun 18 10:54:25 <[g2]> oops.. that's s/have/haven't/
Jun 18 10:55:36 <ilg> not mandatory, I can install suse from a remote ftp/http server
Jun 18 10:55:57 <ilg> I just need a device on the same net with the server to assist it in booting
Jun 18 10:56:01 <[g2]> yeah I know what you mean
Jun 18 10:56:26 <[g2]> sure it needs to hand out the IP and inform the device where the boot server lives
Jun 18 10:56:32 <ilg> and perhaps add a usb serial device, to get access tot he sever console
Jun 18 10:56:57 <[g2]> sure and intelligent front end
Jun 18 10:57:11 <[g2]> actually I think you could do keyboard and mouse
Jun 18 10:57:48 <[g2]> that'd be a nifty little device :)
Jun 18 10:58:06 * [g2] note to self talk to ep1220
Jun 18 10:58:10 <ilg> probably, but again not mandatory, my (Dell) server gives me access to the bios setup via console
Jun 18 10:58:22 <[g2]> serial console
Jun 18 10:58:32 <ilg> right
Jun 18 10:59:04 <[g2]> I've got a DIY Gigabit Server I've been playing with
Jun 18 10:59:23 <[g2]> It's based on an AMD64 and Nforce4 ultra mobo
Jun 18 10:59:33 <[g2]> I'm quite pleased with it
Jun 18 10:59:48 <ilg> and with an additional device allowing remote powercycling of the server, I have (almost) toatl remote control on the server
Jun 18 11:00:08 <[g2]> ilg, extactly.. that's the setup to have
Jun 18 11:00:57 <[g2]> we a single Maxtor SATA drive I can server up nearly 400Mbs
Jun 18 11:01:07 <[g2]> from memory over 600MBs
Jun 18 11:01:36 <[g2]> throught the IP stack the processor/memory can do nearly 2.4G
Jun 18 11:01:58 <[g2]> so it's not breathing real hard @ 400Mbs
Jun 18 11:02:18 <ilg> in the past I was involved in designing a pci board to do exactly this kind of remote control. now I want an external solution, and I'm evaluating how can the nslu2  be used.
Jun 18 11:02:25 <[g2]> I think the mobo, processor and memory set me back like $405 US
Jun 18 11:02:53 <[g2]> cool is that your primary function ?
Jun 18 11:04:40 <ilg> more or less. now I'm working on a project with an m2m gateway, and I want to have full remote control on this server
Jun 18 11:05:01 <[g2]> mobile to mobile ?
Jun 18 11:05:10 <ilg> machine 2 machine
Jun 18 11:05:15 <[g2]> ah
Jun 18 11:05:29 <[g2]> good to have control of the gate :)
Jun 18 11:05:47 <ilg> a dedicated server where a lot of gprs devices connect
Jun 18 11:06:02 <[g2]> GGSN ?
Jun 18 11:06:16 <ilg> ?
Jun 18 11:06:32 <[g2]> That's a 3G data gateway device
Jun 18 11:06:44 <ilg> that's different
Jun 18 11:14:25 <[g2]> ilg, you can always just remount the / fs as ro after boot
Jun 18 11:14:45 <[g2]> depends on what you're trying to protect from
Jun 18 11:15:09 <[g2]> BTW you may be interested in the APEX boot loader also
Jun 18 11:15:34 <ilg> I'd sleep better to know I can always have the nslu2 available :-)
Jun 18 11:15:44 <[g2]> that'll remove many or most of the security issues with the default Redboot bootloader
Jun 18 11:16:33 <ilg> I'll check this too. on my previous project I had my own bootloader, built on eCos too
Jun 18 11:17:00 <[g2]> We're booting the kernel from with the jffs2 partition with APEX 
Jun 18 11:17:51 <[g2]> within
Jun 18 11:19:02 <ilg> that's interesting, but except having a single file with everything, what are the other advantages?
Jun 18 11:19:42 <[g2]> you mean single filesystem ?
Jun 18 11:20:11 <ilg> I mean maintaining a single file instead of two (kernel/initrd)
Jun 18 11:20:43 <ilg> this avoids cases when the initrd and the kernel do not match
Jun 18 11:20:44 <[g2]> for non APEX users the kernel is in its own partition
Jun 18 11:21:13 <[g2]> mtdblock2 
Jun 18 11:21:30 <ilg> I saw this on my device
Jun 18 11:22:13 <[g2]> so if the kernel is in the rootfs, then there are a couple of advantages
Jun 18 11:22:20 <[g2]> 1) no fragmentation
Jun 18 11:22:44 <[g2]> 2) fewer size constrainsts
Jun 18 11:23:04 <[g2]> 3) kernel and modules are easier to update 
Jun 18 11:23:40 <ilg> my experience with embedded devices also showed it is more convenient to have a single file with all components already in (kernel/initrd/packages/modules/etc).
Jun 18 11:24:01 <ilg> even more, for availability/reliability reasons, this should be ro
Jun 18 11:24:54 <[g2]> well availability/reliability is driven mostly by a) moving parts, b) good sw
Jun 18 11:25:18 <[g2]> ppl usually need to store some data so the whole device usually isn't ro
Jun 18 11:25:43 <ilg> of course during development it is convenient to load/unload packages/modules, but the production release must e as ro as possible
Jun 18 11:26:07 <[g2]> it's helpful
Jun 18 11:26:34 <[g2]> Openslug only has like < 1110 files on the entire rootfs
Jun 18 11:26:46 <ilg> my previous device also had a rw part for configuration, but it was designed such that nothing could prevent it from booting 
Jun 18 11:26:53 <[g2]> I think 70+ are in /dev
Jun 18 11:27:05 <[g2]> that's a good practice
Jun 18 11:27:21 <[g2]> fault tolerant is important
Jun 18 11:27:37 <[g2]> we've still got an issue with the reset from external devices
Jun 18 11:27:54 <[g2]> actually that's our main issue right now
Jun 18 11:28:28 <[g2]> I've got to run... 
Jun 18 11:28:31 <[g2]> nice chatting
Jun 18 11:28:41 <ilg> bye
Jun 18 11:28:45 <[g2]> pls let me know how you like OpenSlug
Jun 18 11:28:53 <[g2]> cheers
Jun 18 11:34:06 <ilg> I tryed to download the kernel-module-loop package from ipkg.nslu2-linux.org, but the 'Download' link is broken
Jun 18 11:35:03 <ilg> it seems the ipkgfind needs an update
Jun 18 12:31:58 <dyoung-zzzz> ilg, I think you probably want to use the one you just built.
Jun 18 12:32:17 <dyoung-zzzz> tmp/deploy/ipk/kernel-foo.ipk
Jun 18 12:32:58 <ilg> that i'm just ... building (not ready yet)
Jun 18 12:33:22 <ilg> btw I downloaded the loop module from the proper place and it works
Jun 18 12:39:47 <dyoung-zzzz> cool, so OE/bitbake kinda works for you now.
Jun 18 12:40:38 <ilg> compiling dropbear. in the mean time I'm playing with the 'stock' 1.12
Jun 18 12:46:52 <Christopher> anyone here?
Jun 18 12:48:21 <ilg> ocasionaly
Jun 18 12:50:25 <dyoung-zzzz> Christopher, Morning.  Did you hear some news from the Google folk
Jun 18 12:50:26 <dyoung-zzzz> ?
Jun 18 12:50:50 <Christopher> no, not yet
Jun 18 12:51:01 <Christopher> I am having trouble with unslung 5.5 though
Jun 18 12:51:06 <Christopher> Root filesystem will be mounted from /dev/sdb1 ...
Jun 18 12:51:06 <Christopher> Can't chroot to sdb1, using jffs2 ...
Jun 18 12:51:33 <Christopher> I keep getting that message when I boot up and then it sits there saying that /bin/sh is respawning to quickly
Jun 18 12:52:05 <dyoung-zzzz> do you get that message when booting the thing without the flash stick plugged?
Jun 18 12:52:27 <Christopher> it's a hard drive, but no
Jun 18 12:52:58 <Christopher> it also says that it cannot open /dev/console
Jun 18 12:52:59 <dyoung-zzzz> tried unslinging the drive again?
Jun 18 12:53:02 <Christopher> yep
Jun 18 12:53:09 <dyoung-zzzz> Ok thats a good clue.
Jun 18 12:53:49 <Christopher> ok, i think I just got it to work
Jun 18 12:54:10 <Christopher> i had to copy the contents of /dev  to /share/flash/data/dev
Jun 18 12:54:29 <dyoung-zzzz> I wonder why it didnt get copied in the unsling.
Jun 18 12:55:02 <CIA-12> 03marceln * 10unslung/make/net-snmp.mk: 
Jun 18 12:55:02 <CIA-12> Correction description
Jun 18 12:55:02 <CIA-12> Changed configure flags for native build
Jun 18 12:55:07 <Christopher> i've been having some odd issues with my drive being read only when it boots up that may have caused problems
Jun 18 12:55:21 <dyoung-zzzz> Ah, that could be it.
Jun 18 12:55:35 <dyoung-zzzz> read only on boot usually means its not done running the quotas yet.
Jun 18 12:56:02 <Christopher> i should wipe it and start over, but I don't have any place with enough room to store the data that's currently on the drive, because I only have my laptop right now
Jun 18 12:56:10 <Christopher> yeah, that's what seems to be happening
Jun 18 12:56:58 <dyoung-zzzz> Cool.
Jun 18 12:57:49 <Christopher> hmmm it makes it further, but still complains about /bin/sh respawning too quickly
Jun 18 12:58:06 <Christopher> mmm, now it's making it further on reboot
Jun 18 12:58:15 <Christopher> but then the same error
Jun 18 13:00:04 <ilg> is there anything similar to unsling for openslug? 
Jun 18 13:00:14 <dyoung-zzzz> turnup.
Jun 18 13:00:31 <dyoung-zzzz> although thats not a really fair comparison.
Jun 18 13:00:48 <dyoung-zzzz> the turnup script is quite comprehensive and capable of lots of stuff.
Jun 18 13:01:19 <dyoung-zzzz> Christopher, I'm not sur what else would lcause that.  is there something funky in your hard drive's inittab or something?
Jun 18 13:01:36 <Christopher> i'll look into some more
Jun 18 13:36:59 <jacques> how do I build switchbox?
Jun 18 13:37:21 <jacques> I thought it was switchbox-image, but I get ERROR: Nothing provides switchbox-image
Jun 18 13:40:27 <jacques> yes, it definitely used to be switchbox-image  - where did all the switchbox stuff go in oe ???
Jun 18 13:44:05 <jacques> ChangeSet@1.3212.1.1, 2005-05-04 14:41:13+00:00, rwhitby@home.(none)
Jun 18 13:44:05 <jacques>   Unslung/OpenSlug: Removed all trace of switchbox
Jun 18 13:44:22 <jacques> does that mean removed all trace from openslug and unslung, or removed all trace from oe
Jun 18 13:44:29 <jacques> if the latter I think it was an extremely bad idea
Jun 18 13:46:57 <jbowler-away> jacques: it's there in the BitKeeper directory, a deleted file.  I also have an old clone which contains it.
Jun 18 13:47:11 <jacques> yes, but why the heck was it removed?
Jun 18 13:47:44 <jbowler-away> good question
Jun 18 13:48:07 <jacques> it was a cannonical example of a tiny initrd
Jun 18 13:48:28 <jacques> oe devs had expressed interest in it as useful for other projects (with changes of course)
Jun 18 13:48:41 <jacques> and now I need it, and it's gone for no good reason
Jun 18 13:49:42 <jbowler-away> bk clone bk://jonslug.bkbits.net/jonslug and look at 'bootbox'
Jun 18 13:50:01 <jbowler-away> It's half the size of switchbox (i.e. it fits in one block)
Jun 18 13:50:21 <jacques> why should I have to maintain a custom oe repo just to get functionality that was there and working fine a little over a month ago
Jun 18 13:51:26 <jacques> jbowler-away, OK thanks, I will look at that, but what I need to build will end up being about 2x the size of switchbox
Jun 18 13:52:14 <jacques> God, I'm frustrated lately with all the perfectly working things inexplicably removed from oe - like strace 4.5.9
Jun 18 13:52:25 <jacques> what does it hurt to leave an older version around?
Jun 18 13:52:42 <jacques> or something maybe your project no longer needs but would be a great example for some other project to use
Jun 18 13:53:00 <jbowler-away> It doesn't, except that it slows down the build because some people don't do .bb file selection/masking
Jun 18 13:54:31 <jbowler-away> bootbox and switchbox were both just static busybox's - nothing else really, bootbox is smaller cause I removed things like telnet (replaced it with nc) and I fine tuned the ext2 size.
Jun 18 13:55:49 <jbowler-away> The ideal way is to make a compressed cpio archive and have that loaded into the early user space rootfs - removes the ext2 overhead which is significant, but I never managed to get that to work.
Jun 18 13:56:49 <jacques> what I'm trying to build is an initrd that will set things up for booting from pcmcia
Jun 18 13:57:08 <jacques> last time I did this was about two years ago, and I used uclibc buildroot
Jun 18 13:57:36 <jacques> this is for an ipaq handheld
Jun 18 13:57:49 <jbowler-away> bootbox has a static uclibc with various bits of size reduction - e.g. no LFS
Jun 18 13:58:18 <marceln-away> jeanfabrice: Hi. Could you help me one moment
Jun 18 13:58:33 <jacques> yeah, I don't think static will work for me in this because I need the pcmcia-cs stuff in there too
Jun 18 13:59:15 <jacques> I tried the pivotboot-image in oe but it came out 1.5 MB - waaay too big - because of glibc mostly
Jun 18 13:59:37 <jacques> this should end up around 600kB
Jun 18 13:59:50 <jacques> gzipped ext2 image
Jun 18 14:02:30 <jbowler-away> I build extra things into busybox - busybox is the only executable on the system.  Depending on the number of pcmcia things it might still be a win to use static.  Anyway, the shared variant should work too.
Jun 18 14:03:18 <jeanfabrice> marceln: Hi. If I can, I will. what's on ?
Jun 18 14:04:37 <marceln> jeanfabrice: I am working on net-snmp wish was on the wish list
Jun 18 14:05:05 <marceln> jeanfabrice: But there is a small problem with configure flags.
Jun 18 14:06:06 <marceln> jeanfabrice: If i do a cross compile i have to specifie a flag --with-endianness=big.
Jun 18 14:06:14 <jbowler-away> jacques: do you know what the translation table base register (the top of the MMU page tables) is on power on?
Jun 18 14:06:37 <marceln> jeanfabrice: In the native compile mode this flag will give an error.
Jun 18 14:06:56 <marceln> jeanfabrice: So i made a nice workoround and made this flag dynamic.
Jun 18 14:07:28 <marceln> jeanfabrice: The real question is if i did this alread for the willey as well.
Jun 18 14:08:12 <marceln> jeanfabrice: In the make file stood that the willey is a "mipsel" so i think its a mips in litle endian mode.
Jun 18 14:09:49 <jacques> jbowler-away, sorry, I don't
Jun 18 14:10:18 <jeanfabrice> marceln: Didn't oleg add netsnmp support in -CR6 ?
Jun 18 14:10:40 <jbowler-away> fscking XScale manual doesn't say anywhere I can see...
Jun 18 14:11:07 <marceln> jeanfabrice: I could't find a makefile in the directorie?
Jun 18 14:11:34 <jacques> jbowler-away, but pb or joshua might
Jun 18 14:13:47 <jeanfabrice> marceln : In what directory are you searching for a Makefile ?
Jun 18 14:14:47 <marceln> jeanfabrice: The standard package makefile directorie.
Jun 18 14:39:45 <CIA-12> 03marceln * 10unslung/make/net-snmp.mk: BUG FIX: litle --> little
Jun 18 14:58:42 <rwhitby-away> jacques: switchbox removal was me - didn't realise anyone else apart from nslu2-linux had even looked at it, let alone was interested in it.
Jun 18 14:59:07 <rwhitby-away> It's all still in BK deleted, and could probably be got back by the Unslung-4.20-beta tag
Jun 18 15:00:29 <rwhitby-away> if you *really* want it back undeleted I can look into how to do that.  I deleted it to save confusion (i.e. people with no history with nslu2-linux looking in the busybox package and saying "what the hell is this stupid switchbbox thing?"
Jun 18 15:02:01 <jeanfabrice> 'night
Jun 18 15:02:18 <marceln> By
Jun 18 15:04:17 <jf-logger> 
ACTION 

Jun 18 15:39:38 <marceln> Bye
Jun 18 16:07:19 <ilg> I am succesfully booting and pivot-ing root to an ext3 fs loopmounted from a vfat file on a flash device. 
Jun 18 16:08:35 <jacques> cool, tho some ppl say ext3 and reiserfs are not great ideas on flash (due to the journal writing)
Jun 18 16:08:47 <rwhitby-away> did anything have to change in the defconfig?
Jun 18 16:09:05 <rwhitby-away> or did we have everything as modules already?
Jun 18 16:09:34 <ilg> nope. all I had to do is to add the loop0 module and insert a short patch in /linuxrc
Jun 18 16:10:01 <jacques> ant you should also mount noatime if you aren't already
Jun 18 16:10:56 <rwhitby-away> ilg: can you suggest a new file in /boot to handle your new boot scenario?
Jun 18 16:11:42 <ilg> I used .sd1a-vfat
Jun 18 16:11:49 <ilg> in /
Jun 18 16:12:37 <rwhitby-away> are you using the beta linuxrc or the latest system that jbowler added with multiple files in /boot ?
Jun 18 16:13:13 <rwhitby-away> (cause that has changed significantly since the last beta release)
Jun 18 16:13:14 <ilg> I'm using the binary 1.12, the compiled one just finished
Jun 18 16:13:37 <ilg> I'll update it to the new version and fix mount details tomorrow. no I have to go to sleep
Jun 18 16:13:43 <ilg> it's 2am here :-)
Jun 18 16:13:45 <rwhitby-away> ok, when you retrofit your changes into the latest, have a think about how it can be made into a generic loop mount boot option
Jun 18 16:14:13 <jacques> only 2am? wimp
Jun 18 16:14:37 <rwhitby-away> (jacques is nocturnal)
Jun 18 16:14:58 <jacques> well I have a revolving schedule
Jun 18 16:15:17 <ilg> bye
Jun 18 16:15:18 <jacques> I stay up so late that I end up going to sleep at normal times every once in a while
Jun 18 16:15:23 <rwhitby-away> ilg: we'll probably consider making the loop mount boot an option to turnup script
Jun 18 16:15:27 <jacques> bye ilg 
Jun 18 16:15:47 <jacques> I do wonder about the performance implications tho
Jun 18 16:15:59 <rwhitby-away> ilg: so keep that in mind as you forward migrate your changes
Jun 18 16:16:20 <ilg> will do
Jun 18 16:16:23 <crh> Hmmm...  Okay, I'm a newbee (sort of).  I just upgraded to unslung 5.5.  I thought I'd try installing the native build tools...
Jun 18 16:16:38 <crh> Is it normal for everything to install in the root of the hard drive?
Jun 18 16:17:00 <rwhitby-away> crh: what do you mean?
Jun 18 16:17:02 <jacques> crh, I'm not sure what you mean
Jun 18 16:17:04 <jacques> lol
Jun 18 16:17:11 <crh> :)
Jun 18 16:17:26 <dyoung-zzzz> ... "In the root of the hard drive..."  Youre going to have be more specific.
Jun 18 16:17:29 <jacques> things should install in the "normal" places in openslug
Jun 18 16:17:36 <jacques> like /bin /usr/bin
Jun 18 16:17:42 <rwhitby-away> jacques: he's running unslung
Jun 18 16:17:47 <jacques> oh damn
Jun 18 16:17:53 <jacques> sorry
Jun 18 16:17:57 <jacques> :-(
Jun 18 16:18:07 <crh> I mean I ran ipkg install unslung-devel and all of the new stuff seems to be going into /share/hdd/data
Jun 18 16:18:20 <dyoung-zzzz> directly?
Jun 18 16:18:20 <crh> I was hoping it'd go into a subdirectory thereof.  :)
Jun 18 16:18:34 <crh> Not directly.  That's where the trees are being created.
Jun 18 16:18:36 <dyoung-zzzz> ls -ld /opt
Jun 18 16:19:01 <crh> Here's what I'm seeing:
Jun 18 16:19:04 <crh> # ls /share/hdd/data
Jun 18 16:19:04 <crh> bin             crh             dev             etc             home            initrd          lib             linuxrc         lost+found      mnt             opt             proc            public          quota.user      quota.user~     sbin            share           tmp             unslung         upload          usr             var
Jun 18 16:19:12 <crh> (well, sort of like that...)
Jun 18 16:19:16 <rwhitby-away> crh: that's correct
Jun 18 16:19:53 <crh> Hmmm...  Okay.
Jun 18 16:20:14 <jacques> well well well, 2.6.12 has been released
Jun 18 16:20:24 <crh> I'm interested in taking up the challenge of getting an iSCSI target running on this puppy.
Jun 18 16:20:35 <rwhitby-away> jacques: any relevant usb fixes
Jun 18 16:20:46 <dyoung-zzzz> jacques, I think I'm gonna let someone else take care of that one. :-)
Jun 18 16:20:55 <jacques> rwhitby-away, 2.6.12 is going to be a long slog for arm
Jun 18 16:21:05 <jacques> rmk broke lots of stuff
Jun 18 16:21:18 <rwhitby-away> is he going to fix it?
Jun 18 16:21:19 <dyoung-zzzz> I'm sure that jbowler will have a bb up within hours.
Jun 18 16:21:22 <jacques> we will need to wait for those things to get fixed
Jun 18 16:21:40 <jacques> rwhitby-away, no, he's not - he's leaving that asn an exercise to the peons (i.e. everyone else)
Jun 18 16:21:44 <dyoung-zzzz> ... Hence my waiting for someone else to take care of it.
Jun 18 16:21:48 <rwhitby-away> jbowler is an OpenSlug machine
Jun 18 16:22:05 <jacques> rmk "doesn't think about xscale anymore"
Jun 18 16:22:22 <rwhitby-away> jacques: perhaps we should send rmk a slug to play with ;-)
Jun 18 16:22:51 <jacques> rwhitby-away, if I thought it would help I'd pay for it myself, but ... well I don't think it would
Jun 18 16:23:26 <rwhitby-away> does he just have a lack of time, or is it ideological?
Jun 18 16:23:37 <dyoung-zzzz> the $80 would better be spent on jacques hiking boots. :-)
Jun 18 16:23:58 <jacques> rwhitby-away, he is now and always has been a *()&%&^#$
Jun 18 16:24:17 <rwhitby-away> jacques: don't hold back - tell us how you really think ....
Jun 18 16:24:30 <rwhitby-away> :-)
Jun 18 16:25:10 <rwhitby-away> so rmk is the armel maintainer now.  is there going to be a new armeb maintainer?
Jun 18 16:25:34 <jacques> rwhitby-away, if you get a chance you should read the lak archives, or talk to RP in oe about it
Jun 18 16:25:49 <jacques> RP has been fixing all the things lately that rmk has been breaking
Jun 18 16:25:58 <rwhitby-away> got a starting thread in lak?
Jun 18 16:26:20 <jacques> not offhand, you can pretty much look at any post by rmk tho - well give me a few minutes
Jun 18 16:27:40 <rwhitby-away> crh: so the slug will be the iSCSI target, providing files to something else?
Jun 18 16:31:57 <crh> rwhitby-away:  That'd be my plan, yes.
Jun 18 16:32:16 <crh> I'm trying to avoid kernel modules, for now.
Jun 18 16:32:30 <crh> I can probably hack the UNH code to read/write from local container files.
Jun 18 16:32:46 <crh> Eventually, I'll have to go the OpenSlug route...
Jun 18 16:33:05 <rwhitby-away> yeah, that's what I was just thinking
Jun 18 16:34:32 <jacques> rwhitby-away, this is one of the recent famous threads: http://article.gmane.org/gmane.linux.ports.arm.kernel/18779
Jun 18 16:34:53 <crh> For a target to use native disk, it kinda needs to be partly in the kernel.  Likewise, the initiator needs to have a kernel part in order to present the storage as local disk.
Jun 18 16:35:19 <rwhitby-away> crh: this you? http://www.elx.com.au/item/013047116X#AUTHORBIO
Jun 18 16:36:16 <crh> rwhitby-away:  Yeah, that be me.  What it means is that I'm usually too busy to actually get anything done, though.  :)
Jun 18 16:37:13 <rwhitby-away> heh.  join the club :-)
Jun 18 16:37:33 <crh> ...I'm a member in good standing.  ;)
Jun 18 16:37:39 <rwhitby-away> you should go straight to openslug.
Jun 18 16:37:49 <crh> Could do...  
Jun 18 16:38:09 <rwhitby-away> are you bk-compatible?
Jun 18 16:38:27 <crh> I was toying with that.  The UNH code is aimed at the 2.4 kernel.  What's OpenSlug based on (from the comments here, I'm guessing 2.4).
Jun 18 16:38:36 <rwhitby-away> openslug is based on 2.6
Jun 18 16:38:50 <rwhitby-away> 2.6.11.something
Jun 18 16:38:57 <crh> That should work too.  Okay.  I'm good at taking advice from folks who know...
Jun 18 16:39:28 <rwhitby-away> easiest is to svn co from svn.berlios.de/openslug/trunk/openslug
Jun 18 16:39:29 * dyoung-zzzz looks around
Jun 18 16:39:45 <rwhitby-away> that'll get you close to the latest stable, and will set up your env automatically
Jun 18 16:40:19 <rwhitby-away> then replace the openemebedded and nslu2-linux directories with symlinks to openembedded and nslu-package-symlinks repos from BK
Jun 18 16:40:23 <jacques> OK, from talking to RP, it looks like 2.6.12 is worth a go
Jun 18 16:40:54 <crh> rwhitby-away:  starting...
Jun 18 16:41:30 <jbowler-away> [g2] tried 2.6.12-rc5, I think without any NSLU2 patches - ethernet failed.
Jun 18 16:41:32 <rwhitby-away> crh: jacques and dyoung (and others) can help.  I need to go out and about for a while.
Jun 18 16:41:52 <rwhitby-away> crh: jbowler and [g2] are the current openslug gurus
Jun 18 16:42:12 <jbowler-away> 2.6.12-rc6 fixed patches were easy, only head.S failed (and the fix is obvious)
Jun 18 16:42:31 <jbowler-away> 2.6.12 (release) works with my fixes - i.e. the patches apply
Jun 18 16:42:48 <jacques> jbowler-away, pleae try 2.6.12 when you get a chance - apparently there's a copypage fix which is showing massive speedups on the pza
Jun 18 16:42:49 <crh> I'll dig into the howtos before I start bugging folks.  Thanks for the guidance.
Jun 18 16:42:50 <jacques> pxa
Jun 18 16:42:57 <jbowler-away> Don't know whether it will compile, let alone boot, yet
Jun 18 16:43:33 <jacques> it's running on the zaurus (pxa le)
Jun 18 16:43:37 <rwhitby-away> crh: questions from developers are always welcome here
Jun 18 16:44:02 <jbowler-away> Someone with either JTAG or APEX, or both, is going to need to look at the reset problem - I can't debug into the RedBoot code and my disassembly of that code leaves me wondering how it can possibly work...
Jun 18 16:44:30 <rwhitby-away> jbowler-away: you know we have the RedBoot source, right?
Jun 18 16:44:52 <jbowler-away> Do we have the LinkSys hacked source - like where they removed the page table setup?
Jun 18 16:44:58 <dyoung-zzzz> yes
Jun 18 16:45:21 <rwhitby-away> jbowler-away: check out "sluggo" cvs module on nslu.sf.net
Jun 18 16:45:29 <rwhitby-away> and ask dyoung about it :-)
Jun 18 16:45:45 <dyoung-zzzz> I have a script in sluggo that pulls down the linksys hacked patches and redboot version.
Jun 18 16:46:06 <dyoung-zzzz> and builds a linksys compatible hacked redboot.
Jun 18 16:46:12 <jbowler-away> Ah, ok, I pulled the CVS head of eCos
Jun 18 16:46:14 <rwhitby-away> dyoung-zzzz: did anyone ever test that one?
Jun 18 16:46:19 <dyoung-zzzz> other than me?
Jun 18 16:46:28 <rwhitby-away> did you flash it?
Jun 18 16:46:28 <dyoung-zzzz> I think [g2] did...
Jun 18 16:46:32 <dyoung-zzzz> of course I flashed it.
Jun 18 16:46:43 <rwhitby-away> ok, that's good enough for me :-)
Jun 18 16:46:55 <dyoung-zzzz> I think ka6sox flased it too; but that was on a fatslug of unknown working state.
Jun 18 16:47:18 <dyoung-zzzz> It booted fine o a 32MB slug.
Jun 18 16:47:22 * rwhitby-away has a fading memory of those halcyon days ..
Jun 18 16:47:22 <jbowler-away> What would be nice is one which does write to the 7 segment display but instead writes to GPIO
Jun 18 16:47:27 <jbowler-away> s/does/doesn't/
Jun 18 16:47:59 <dyoung-zzzz> jbowler-away, thats howh I was looking at the code for the upgrade modes and stuff and gleaned that bit about the mac squirting mode...
Jun 18 16:48:09 <rwhitby-away> bbiab
Jun 18 16:51:27 <jbowler-away> dyoung-zzzz: what seems to be happening is that there is a hang very soon after I do mov pc, #0 - before the LED gets set to amber.  I know the mov executes (because I can write to physical GPIO and set the leds in the preceding instruction).
Jun 18 16:52:14 <jbowler-away> 2.6.12 compiles/builds.
Jun 18 16:52:49 <jacques> yay
Jun 18 16:59:51 <jbowler-away> 2.6.12 reaches the end of /etc/rc5.d/S??* and the ethernet is up, but dropbear seems to be broken (no connection refused, no timeout)
Jun 18 17:00:12 <jbowler-away> It may be that my 'ostest' build is bogus, I'll probably try it with a uclibc build.
Jun 18 17:01:06 <jbowler-away> Or, maybe, by SA_RANDOM_SAMPLE fix to the eth driver isn't 2.6 compatible, because I didn't recompile the drivers, duh.
Jun 18 17:02:21 <jbowler-away> Nope, broken image - I don't think I even managed to pull the kernel in...
Jun 18 17:09:49 <crh> okay... so if I'm not BK-compatible (and I just want to be able to work in a stable build environment) what are recommendations?
Jun 18 17:14:00 <jbowler-away> For OpenSlug the SVN 1.12 source tarball
Jun 18 17:15:22 <crh> jbowler-away:  I'm going through the recommended package install now, and I'll grab the 1.12 tarball...  Thanks.
Jun 18 17:16:42 <rwhitby-away> 8:51 to 9:12 - jbowler had 2.6.12 built within half an hour - much less than dyoungs "within hours" prediction :-)
Jun 18 17:16:56 <rwhitby-away> crh: svn trunk on berlios
Jun 18 17:17:20 <rwhitby-away> that's 1.12, but with jbowler's boot structure change included
Jun 18 17:17:48 <crh> rwhitby-away:  I grabbed that, and it whined at me.  You asked earlier, and I don't yet have bk.  
Jun 18 17:17:50 <rwhitby-away> and we can always update that svn snapshot if you need something more recent
Jun 18 17:18:01 <crh> I'm just trying to fill in the missing pieces.
Jun 18 17:18:19 <rwhitby-away> the svn trunk whined?
Jun 18 17:18:58 <crh> No...  I need to figure out what else I'm missing.  It seems to want the jit compiler.  I've just installed ccache and I'm working on finding out what else is missing.
Jun 18 17:19:10 <rwhitby-away> oh, ok.
Jun 18 17:19:36 <crh> Samba's all nicely in userspace, so I can ignore some details...  :)
Jun 18 17:20:09 <rwhitby-away> crh: yeah, if you're ignoring boot or rc changes, then 1.12-beta binary will be fine
Jun 18 17:20:51 <crh> I'll go through the steps and work with the svn version.  I just need to get the environment set up...
Jun 18 17:21:26 <rwhitby-away> be aware that the wiki is probably a bit out of date on the openslug build front, but you'll get the idea ...
Jun 18 17:22:46 <crh> I already got that idea.  :)
Jun 18 17:22:50 <jbowler-away> The trunk used to build fine for me (a couple of weeks ago), but I've got psyco installed.
Jun 18 17:25:06 <crh> I may have to download it...
Jun 18 17:26:57 <crh> ERROR: Nothing provides openslug-packages   ...Is this just a setup goof on my part or do I need to find and download these as well?
Jun 18 17:29:22 <jbowler-away> That's a setup problem
Jun 18 17:29:58 <jbowler-away> The 'make' should run bitbake and find around 300 packages (symlinks/packages/*.bb)
Jun 18 17:30:45 <crh> It started finding packages...  it stopped with that error.  What should I look at fixing?
Jun 18 17:31:40 <rwhitby-away> crh: did you let the Makefile from svn create your local.conf, or did you create it yourself?
Jun 18 17:31:51 <crh> I let it do the work...
Jun 18 17:32:07 <rwhitby-away> (the svn one uses nslu2-linux for the symlinks dir, the bk uses nslu-package-symlinks for the name of the same dir)
Jun 18 17:32:33 <crh> Ah.  So I need to change that in the local.conf?
Jun 18 17:32:50 <crh> I'm doing an svn update and it's updating bitbake...
Jun 18 17:33:52 <jbowler-away> It won't work after that, you need -r154 of bitbake (it's in the 'externals' or something like that...)
Jun 18 17:34:04 <rwhitby-away> crh: it has to match the name of the dir.  if you mix and match svn and bk, then you'll need to change it to suit
Jun 18 17:34:26 <rwhitby-away> I gotta go.  good luck.  the guys here will sort it out.
Jun 18 17:34:50 <jbowler-away> No the SVN trunk Makefile should correctly create local.conf from local.conf.template and it should 'just work'.
Jun 18 17:35:35 <crh> jbowler-away:  I'm just typing 'make'.  Should that 'just work' or should I be changing or creating anything before hand?
Jun 18 17:35:42 <crh> thanks, rwhitby-away
Jun 18 17:36:18 <crh> jbowler-away: here's the first error I get:  ERROR: [Errno 2] No such file or directory: '/home/crh/work/code/NSLU2/openslug/nslu2-linux/packages/*/*.bb' while parsing /home/crh/work/code/NSLU2/openslug/nslu2-linux/packages/*/*.bb
Jun 18 17:36:19 <jbowler-away> 'make' should just work.
Jun 18 17:37:48 <crh> I sure wish it did...  I'm looking for that dumb newbie thing I'm doing...
Jun 18 17:38:16 <crh> Note that psyco isn't installed.  Dunno if that'll have any impact.
Jun 18 17:39:52 <jbowler-away> You don't have a populated nslu2-linux directory tree
Jun 18 17:41:23 <crh> I checked out the tree via SVN as rwhitby suggested.  What must I do to get the missing parts?  Should I start with the tgz and then update via SVN?
Jun 18 17:43:18 <crh> Urq!  Being called to dinner (and the 10-year-old will not take no for an answer).  I am currently installing psyco.  Will return...
Jun 18 17:43:29 * crh is away: gotta go to dinner...
Jun 18 17:43:37 <jbowler-away> svn co should propulate the whole thing
Jun 18 17:44:27 <crh> svn co svn://svn.berlios.de/openslug/trunk/openslug openslug
Jun 18 17:44:36 <crh> That's what I ran.  Did I get the wrong module?
Jun 18 17:46:36 * crh is away: gotta go to dinner...
Jun 18 17:51:34 <jbowler-away> Try an svn up in 'openslug', I think that will pull in what is missing and I  think the latest bitbake will be fine in this case.
Jun 18 17:54:43 <jbowler-away> rwhitby-treo - how do you sync up the SVN trunk?  (Is there a script to remove the deleted bitbake stuff?)
Jun 18 17:55:32 <rwhitby-treo> I've just done it manually so far
Jun 18 17:56:12 <rwhitby-treo> just delete SCCS and Bitkeeper dirs and then svn add and commit
Jun 18 17:56:33 <jbowler-away> What about all the svn dels?  (That was what caused me problems)
Jun 18 17:57:05 <rwhitby-treo> hmm - did nothing for them :-)
Jun 18 17:57:31 <jbowler-away> Ah ;-) maybe switchbox is still in there then :)
Jun 18 17:58:21 <jbowler-away> No, it isn't, but all the gcc stuff which pb removed is there - that might be a problem.
Jun 18 17:58:35 * crh is back (gone 00:11:59)
Jun 18 17:59:10 <jbowler-away> crh: run svn up in the openslug directory, I just tried it and it seems to do the right thing.
Jun 18 17:59:15 <crh> jbowler-away:  svn update in openslug gives the same result
Jun 18 18:00:51 <jbowler-away> from 'openslug' what does "ls -l nslu2-linux/packages/meta' produce?
Jun 18 18:01:21 <crh> -rw-r--r--  1 crh users 37 2005-06-18 19:59 nslu2-linux/packages/meta
Jun 18 18:02:10 <rwhitby-treo> should be a symlink over to openembedded
Jun 18 18:02:15 <crh> the contents look wrong, though...
Jun 18 18:02:43 <jbowler-away> Can you 'cat' it and copy the output?
Jun 18 18:02:57 <jbowler-away> (It should be 32 bytes long)
Jun 18 18:03:46 <crh> link ../../openembedded/packages/meta
Jun 18 18:04:49 <jbowler-away> Ah, cool.  What operating system is this?
Jun 18 18:05:25 <crh> SuSEPro 9.1
Jun 18 18:05:52 <jbowler-away> What file system is 'openslug' on (i.e. what file system type)?
Jun 18 18:06:14 <crh> it's ext3
Jun 18 18:06:53 <jbowler-away> That's weird, try 'rm nslu2-linux/packages/meta; ln -s ../../openembedded/packages/meta nslu2-linux/packages/meta'
Jun 18 18:07:07 <rwhitby-treo> hmm - I wonder if I stuffed up the svn repo on the last update - that should just be a normal symlink
Jun 18 18:07:33 <jbowler-away> I just did an svn up and it's a normal link on my (reiser4) file system.
Jun 18 18:07:53 <jbowler-away> Oh, and I deleted nslu2-linux before doing it too.
Jun 18 18:08:07 <crh> bingo.
Jun 18 18:08:21 <jbowler-away> ?  did that output an error message
Jun 18 18:08:37 <crh> I'm getting very different results.  Dunno why it would come through as a file instead of a link.  My svn version?
Jun 18 18:08:41 <crh> no error messages.
Jun 18 18:08:50 <crh> wait... yes error message...
Jun 18 18:09:04 <jbowler-away> Yes, but did the 'ln -s' actually produce a symbolic link?
Jun 18 18:09:12 <crh> yes, it did.
Jun 18 18:09:31 <crh> I'm guessing there are other links to be fixed.  I'm now getting errors on ipkg:
Jun 18 18:09:34 <crh> NOTE: build 200506182007: started
Jun 18 18:09:34 <crh> OE Build Configuration:
Jun 18 18:09:34 <crh> TARGET_ARCH   = "armeb"
Jun 18 18:09:34 <crh> TARGET_OS     = "linux"
Jun 18 18:09:35 <crh> MACHINE       = "nslu2"
Jun 18 18:09:37 <crh> DISTRO        = "openslug"
Jun 18 18:09:39 <crh> TARGET_FPU    = "soft"
Jun 18 18:09:44 <jbowler-away> Hum.  Delete (rm -rf) nslu2-linux and try another svn up - see if you still have files in there (they should all be symbolic links).
Jun 18 18:09:45 <crh> ERROR: Nothing provides ipkg-utils-native
Jun 18 18:09:47 <crh> ERROR: dependency ipkg-utils-native (for openslug-packages) not satisfied
Jun 18 18:09:49 <crh> NOTE: no buildable providers for openslug-packages
Jun 18 18:09:51 <crh> make: *** [openslug-firmware] Error 1
Jun 18 18:11:11 <crh> Done.  Here's what the meta looks like: -rw-r--r--  1 crh users 37 2005-06-18 20:10 nslu2-linux/packages/meta
Jun 18 18:12:06 <crh> My (wild) guess is that svn 1.0.4 isn't handling symlinks properly.
Jun 18 18:12:16 <crh> Weird.
Jun 18 18:12:17 <jbowler-away> So there seems to be a problem with SVN and SuSEPro
Jun 18 18:12:34 <crh> Should be manually fixable.  I think.
Jun 18 18:13:01 <jbowler-away> Maybe - it's weird that SVN should generate the correct link target prefixed by 'link '
Jun 18 18:13:30 <crh> Holy bananafish, batman.  All of the files are like that.  Looks like an easy thing to fix with a shell script.
Jun 18 18:13:42 <jbowler-away> If symbolic links don't work reliably there will be problems generating an image - the root file system contains a number of links.
Jun 18 18:14:28 <jbowler-away> The only symbolic links in the trees are those within nslu2-linux/packages.
Jun 18 18:14:47 <jbowler-away> You don't need it anyway - just change 'local.conf' so it says 'openembedded/packages/*/*.bb'
Jun 18 18:15:02 <jbowler-away> (i.e. replace the nslu2-linux by openembedded)
Jun 18 18:15:48 <rwhitby-treo> and wait a little longer for the parsing :-)
Jun 18 18:16:03 <jbowler-away> Thsi is a security thing.  The process is creating a link pointing below it's own current directory.
Jun 18 18:16:52 <rwhitby-treo> selinux?
Jun 18 18:17:20 <jbowler-away> rwhitby-treo: maybe I should check the 'freeze' script in to the svn trunk.  It can work out the exact list of bb files, and, for a source tarball, that's faster than nslu2-links...
Jun 18 18:17:32 <crh> Not selinux...  At least, I don't think so.
Jun 18 18:17:56 <jbowler-away> SuSEPro 9.1
Jun 18 18:18:33 <jbowler-away> Easy experimit - repeat the ln -s but cd to nslu2-linux first
Jun 18 18:18:59 <crh> From dmesg: SELinux:  Disabled at boot.
Jun 18 18:19:11 <jbowler-away> It must be an OS/SVN interaction of some kind, unless they've broken SVN and you have a newer version.
Jun 18 18:19:59 <crh> jbowler-away, My version of svn is probably older (current is 1.2.0).
Jun 18 18:20:33 <crh> changing local.conf seems to have worked.
Jun 18 18:21:11 <jbowler-away> I have: svn, version 1.1.3 (r12730)
Jun 18 18:21:30 <jbowler-away> If yours is older I would suggest upgrading...
Jun 18 18:21:44 <crh> Newer than 1.0.4, certainly.  Yes... I'll bring down 1.2.0
Jun 18 18:21:55 <crh> Haven't had a problem until now...
Jun 18 18:23:06 <rwhitby-treo> yeah - it seems weird that such a basic thing isn't working
Jun 18 18:24:52 <crh> Well... you'll know what it is the next time someone falls in the same hole.
Jun 18 18:25:08 <jbowler-away> Not many people check symbolic links into source control systems - they're not portable (at least not to older SysV systems - pre BSD ffs)
Jun 18 18:25:49 <crh> It might be better to include a setup script that would create them.
Jun 18 18:26:26 <rwhitby-treo> heh - wasn't a problem with bk ;-)
Jun 18 18:26:32 <jbowler-away> 2.6.12 is checked in in changeset 1.3569, to use it change:
Jun 18 18:26:39 <jbowler-away> PREFERRED_PROVIDER_virtual/kernel ?= "nslu2-kernel"
Jun 18 18:27:10 <jbowler-away> PREFERRED_VERSION_nslu2-kernel = 2.6.12
Jun 18 18:27:37 <jbowler-away> It boots, but the ethernet doesn't come up and it doesn't get to the end of /etc/init.d/rcS
Jun 18 18:27:55 <jbowler-away> (BTW, this is openslug only)
Jun 18 18:27:57 <rwhitby-treo> senneth must be tested irssi ...
Jun 18 18:30:11 <rwhitby-treo> back later
Jun 18 19:28:16 <crh> jbowler-away, rwhitby-away:  Confirmed.  Using svn 1.2.0 the links appear as real symbolic links (not as files with the word "link" in them).
Jun 18 19:28:47 <crh> All is happy... :)  Now I'll start building again...
Jun 18 19:33:52 <[g2]> crh you looking to run an iSCSI block device on the slug ?
Jun 18 19:46:08 <[g2]> the MAN is back!
Jun 18 19:48:00 <kergoth> heh.
Jun 18 19:49:11 <[g2]> kergoth, how goes my friend ?
Jun 18 19:49:39 <[g2]> or should I just call you OE master
Jun 18 19:51:39 <jacques> 691.21K/s from kernel.org - now that's how you run a site :-)
Jun 18 19:52:03 <[g2]> jacques, are you testing 2.6.12 ?
Jun 18 19:52:40 <jacques> yes
Jun 18 19:52:46 <[g2]> slug or PC ?
Jun 18 19:53:30 <jacques> slug, I don;t see a need to upgrade my PCs
Jun 18 19:54:07 <[g2]> pls let me know how it goes :)
Jun 18 19:54:19 <jacques> 2.6.11.11 was good enough for my grandfather's PC, it's good enough for mine  :-)
Jun 18 19:54:35 <[g2]> I saw the changesets and I'm building mysql now
Jun 18 19:54:53 <jacques> [g2], what's up with jbowler-away not having a serial port? we gotta get that boy serial-enabled!
Jun 18 19:58:11 <[g2]> ain't that the truth!
Jun 18 19:58:51 <[g2]> I'd send him a fat slug, but I haven't verified the JTAG (well made it work yet)
Jun 18 19:59:31 <jacques> heh, I just tried rebooting my about a week old openslug and it locked hard
Jun 18 19:59:35 <jacques> had to pull power
Jun 18 20:03:07 <[g2]> I've been using shutdown -h now :)
Jun 18 20:04:46 <[g2]> DOH! mysql builds the -native first :)
Jun 18 20:13:56 <jacques> dammit
Jun 18 20:14:09 <jacques> something wrong with console args or something, or they changed the serial ports again
Jun 18 20:14:31 <jacques> because I don't get anything on serial after: ... done, booting the kernel.
Jun 18 20:15:28 <jacques> oh, it just took a really long time and disn't start printing anything til it reached sysvinit
Jun 18 20:15:48 <jacques> hmm, it's up
Jun 18 20:16:08 <jacques> Linux LKG0FB08F 2.6.12 #1 Sat Jun 18 19:54:38 PDT 2005 armv5teb unknown
Jun 18 20:17:33 <jacques> seems to work, I am ssh'd in
Jun 18 20:19:58 <jacques> well the cmdlins is identical to the older one, so something else must have changed with the serial during boot
Jun 18 20:21:40 <[g2]> jacques, the serial stuffs been broke since -rc5
Jun 18 20:21:47 <[g2]> probably before :(
Jun 18 20:22:03 <jacques> [g2], oh I wasn't aware of that
Jun 18 20:22:19 <jacques> is that another think rmk broke ?
Jun 18 20:22:26 <[g2]> that's what happended when I ran my version
Jun 18 20:22:36 <[g2]> I dunno
Jun 18 20:23:46 <[g2]> that should be pretty easy to track down :)
Jun 18 20:27:01 <jacques> well, hdparm numbers are pretty much the same, and I still get irq26
Jun 18 20:27:20 <jacques> at least nothing (other than serial on boot) appears worse yet
Jun 18 20:27:23 <[g2]> you should try noirqdebug
Jun 18 20:28:20 <[g2]> where's jbowler got the 2.6.12 ?
Jun 18 20:30:20 <jacques> <jbowler-away> 2.6.12 is checked in in changeset 1.3569, to use it change:
Jun 18 20:30:20 <jacques> <jbowler-away> PREFERRED_PROVIDER_virtual/kernel ?= "nslu2-kernel"
Jun 18 20:30:20 <jacques> <jbowler-away> PREFERRED_VERSION_nslu2-kernel = 2.6.12
Jun 18 20:31:29 <[g2]> I've got a build for -rc5 and I'm wondering if it's conflicting
Jun 18 20:31:59 <[g2]> that's in my packages/linux/openslug-kernel_2.6.12.bb
Jun 18 20:32:40 <[g2]> he's got it under nslu2-linux I think
Jun 18 20:34:35 <jacques> OK, I turnuped a spare partition and am booted to it now
Jun 18 20:35:20 <[g2]> The lack of serial messages is quite annoying!
Jun 18 20:35:50 <jacques> yeah for me it's 90 seconds of nothing
Jun 18 20:36:00 <[g2]> nod
Jun 18 21:19:13 <jacques> how does reflash work? why can't I just do "reflash -k kernel_image"  ?
Jun 18 21:25:40 <jacques> Results of search for  devio: 0 pages found out of 491 pages searched.
Jun 18 21:25:49 <jacques> :-|
Jun 18 21:48:50 <jbowler-away> devio.sourceforge.net
Jun 18 21:50:17 <jbowler-away> reflash should accept just the -k, but I didn't think of that.  At present I use this, in a shell script:
Jun 18 21:50:21 <jbowler-away> devio '<<zImage' '>>/dev/mtdblock2' 'wb $16+,4;fb 12,0;cp $;fb #t-,255'
Jun 18 21:51:37 <jacques> jbowler-away, thanks! I was hoping you would see my plea
Jun 18 21:52:11 <jacques> jbowler-away, did you see 2.6.12 appears to be working for me (except for the serial console during boot thing)
Jun 18 21:53:52 <jacques> heh, devio don't exactly talk your ear off :-)
Jun 18 21:54:31 <jbowler-away> You mean while writing?  No - it's very basic.
Jun 18 21:55:02 <jbowler-away> I could write it in blocks, in a loop - the 'cp $' just copies the whole kernel in at once.
Jun 18 21:55:18 <jacques> ok I got hard lock while rebooting with disk attached (in case you had changed anything)
Jun 18 21:55:47 <jbowler-away> What do the LEDs say?
Jun 18 21:56:19 <jacques> oops sorry
Jun 18 21:56:27 <jacques> I already power cycled
Jun 18 21:56:35 <jbowler-away> (They should display something for about 4s, then just go blank when the watchdog kicks in.)
Jun 18 21:56:56 <jacques> was testing my attempt at getting the serial port to work during boot - which did not work :-(
Jun 18 21:57:07 <jacques> so now I am blind for 90 seconds
Jun 18 21:57:32 <jbowler-away> This 2.6.12 is mysterious, mine definately had no ethernet - might be DHCP/afpacket broken.
Jun 18 21:57:48 <jacques> yeah, I use static addressing
Jun 18 21:58:13 <jacques> I haven't noticed anything different other than the serial thing
Jun 18 21:58:59 <jbowler-away> Maybe that's it - I don't have the serial port, maybe /dev/console opens in 2.6.12 then blocks when the first message goes out.
Jun 18 21:59:19 <jbowler-away> I've got a brike, so I'll have to reflash.
Jun 18 21:59:59 <jacques> what I'm seeing is "done, booting the kernel. " and then nothing til "boot: rootfs: mount -t reiserfs -o noatime,notail /dev/sda2"
Jun 18 22:00:49 <jacques> man the clock is sooo bad
Jun 18 22:03:41 <jbowler-away> That 'boot rootfs' is from /boot/disk and it's just an echo into stdout (fd 1).
Jun 18 22:05:55 <jbowler-away> Unless I'm missing something fd1 just comes from the kernel, so this doesn't make sense - apparently kprintf's are disappearing somewhere but /dev/console is fine.
Jun 18 22:07:04 <jacques> I've seen this behavior before when the console= is wrong in the kernel cmdline
Jun 18 22:07:50 <jacques> I tried ttySAC0 after reading some posts on lak, but it made no difference
Jun 18 22:08:09 <jacques> heck it's probably ttyPXAC0 or some crap now
Jun 18 22:08:30 <jbowler-away> Might be, also there's a new entry in the config (I think it's new): General Setup --> Configure standard kernel features... --> [*] Enable support for printk
Jun 18 22:08:46 <jbowler-away> I don't remember seeing that before (but it is enabled).
Jun 18 22:09:07 <jacques> hmm, yeah they could have changed some configs, definitely
Jun 18 22:09:46 <jbowler-away> Yep, it's new - it isn't in menuconfig for 2.6.11.2
Jun 18 22:10:27 <jacques> take a look at this (only two post thread)
Jun 18 22:10:32 <jacques> http://news.gmane.org/navbar.php?group=gmane.linux.ports.arm.kernel&article=19026&next=19027&prev=19029&newsrc=,18779-18780,18781,18793,18809,18812,18855,18859,18864,18870,18872,18880,18882,18899-18900,18904,18905,18906,18908,18912,18932,18935,18936,18951,18954,18957,18966,19000,19007,19026,19031,19034
Jun 18 22:11:18 <jacques> (dunno if it's even relevant - I hope it's just a new config we're missing)
Jun 18 22:12:41 <jbowler-away> Oh, there's 10k of help built in to devio BTW (devio -h)
Jun 18 22:13:01 <jacques> ah cool
Jun 18 22:13:29 <jacques> what do you think of the reflash script being able to just do kernel tho?
Jun 18 22:14:21 <jbowler-away> I want it to do that too, having used it a bit (with a warning about making sure /lib/modules is good for the kernel)
Jun 18 22:14:54 <jacques> :-)
Jun 18 22:17:13 <jacques> whoops
Jun 18 22:17:24 <jacques> somehow my change did not "take"
Jun 18 22:17:36 <jacques> my /proc/cmdline still shows ttyS0
Jun 18 22:18:22 <jbowler-away> Did you change it in the defconfig or the .bb?
Jun 18 22:18:39 <jbowler-away> Or did you use menuconfig (which I find to be faster, though less permanent...)
Jun 18 22:18:45 <jacques> defconfig
Jun 18 22:18:51 <jacques> did I do wrong?
Jun 18 22:19:07 <jbowler-away> Nope, you have to change it in a .bb of conf file...
Jun 18 22:19:33 <jacques> so I did do it wrong
Jun 18 22:19:43 <jbowler-away> CMDLINE_CONSOLE ?= "ttyS0,115200n8" (in nslu2-kernel.inc)
Jun 18 22:19:53 <jacques> can I put it in my local.conf ?
Jun 18 22:20:10 <jbowler-away> I think that will work
Jun 18 22:20:30 <jacques> ok, I will test in a few minutes - BIAB
Jun 18 22:21:43 <jbowler-away> Blurgh...  in kernel.bbclass: export CMDLINE_CONSOLE = "console=${@bb.data.getVar("KERNEL_CONSOLE",d,1) or "ttyS0"}"
Jun 18 22:25:41 <jbowler-away> I am in a maze of twisty bb files, all different.  machine/nslu2.conf:KERNEL_CONSOLE = "ttyS0,115200n8"
Jun 18 22:26:49 <jbowler-away> I think that last one should be ?= and the local.conf should specify KERNEL_CONSOLE to override.
Jun 18 22:41:40 <jacques> that sounds right
Jun 18 22:42:18 <jbowler-away> Or even just hard-wire it in the .bb file, since it seems to be version specific.
Jun 18 22:42:28 <jbowler-away> (I.e. set it in nslu2-kernel_2.6.12.bb)
Jun 18 22:44:09 <jbowler-away> Um, that change alone seems to allow the network to come up.  Bogus.
Jun 18 22:44:30 <jacques> which change ?
Jun 18 22:44:45 <jbowler-away> Oh, in recent builds the Disk1 light flashes while the dropbearkey is being generated.
Jun 18 22:44:52 <jacques> neat
Jun 18 22:45:14 <jbowler-away> pinging the machine speeds it up.
Jun 18 22:45:23 <jacques> LOL 
Jun 18 22:46:01 <jacques> so if I set KERNEL_CONSOLE in nslu2-kernel_2.6.12.bb it should take?
Jun 18 22:46:36 <jbowler-away> I believe so - the check is to look at the .config in the 2.6.12 source (S) directory after do_configure has run.
Jun 18 22:46:43 <jacques> yah
Jun 18 22:46:51 <jacques> this time I will before flashing and rebooting :-)
Jun 18 22:47:07 <jacques> just that last time I messed with kernel args, the defconfig was the place to do it
Jun 18 22:47:43 <jbowler-away> Hum.  I think I've got a dropbear which reads from /dev/random.  Hope I didn't check that in...
Jun 18 22:50:04 <jbowler-away> Yes, not checked in, apart from that the tty fix seems to be working.
Jun 18 22:50:19 <jbowler-away> Is this a candidate for OpenSlug 2.0?
Jun 18 22:51:26 <jacques> you mean because of the kernel change?
Jun 18 22:51:44 <jacques> or all the other changes that have gone in recently ?
Jun 18 22:51:53 <jbowler-away> I mean make the next release use 2.6.12 rather than 2.6.11.2
Jun 18 22:52:47 <jbowler-away> All the other changes are in, though the reboot problem still needs to be fixed (and I think turnup is slightly broken on uclibc sed)
Jun 18 22:53:29 <jacques> sounds reasonable to me
Jun 18 22:55:08 <jacques> I guess we really should do some additional testing on 2.6.12, tho what that should be I dunno offhand
Jun 18 22:56:12 <jbowler-away> The reboot changes I checked in (they are in the build you have) set the led to red just before the flash is remapped, switch on disk1 and disk2 just after, mov pc, #0 then fail (the watchdog cuts in about 4s later and kills the LEDs)
Jun 18 22:56:41 <jacques> ok, I did reboot, and it locked with: ready:off ethernet:green
Jun 18 22:56:43 <jbowler-away> I'm interested in knowing whether RedBoot manages to put anything up on the serial
Jun 18 22:57:14 <jacques> last thing I see on serial is:
Jun 18 22:57:17 <jacques> Deactivating swap...
Jun 18 22:57:17 <jacques> Unmounting local filesystems...
Jun 18 22:57:17 <jacques> Rebooting... Restarting system.
Jun 18 22:58:17 <jacques> hmg, I still have no serial boot messages :-\
Jun 18 22:58:32 <jacques> will check /proc/cmdline after boot to amke sure I didn't do anything stupid
Jun 18 22:59:14 <jacques> ok that was really weird - it seemed to come up to a login: prompt really fast
Jun 18 22:59:33 <jbowler-away> Hum.  I just did it on flash-only and got a successfull reboot, soft, no LEDs.  Worrying - my hacks might be not there.
Jun 18 22:59:37 <jacques> ah, it booted from flash for some reason
Jun 18 23:00:45 <jacques> Warning: unable to open an initial console.
Jun 18 23:00:45 <jacques> ReiserFS: sda2: found reiserfs format "3.6" with standard journal
Jun 18 23:00:48 <jacques> yada yada yada
Jun 18 23:00:54 <jacques> but it didn't mount the disk
Jun 18 23:01:31 <jbowler-away> Look at /linuxrc - is it a symlink to /boot/flash?
Jun 18 23:03:11 <jacques> nope, linuxrc.sav is
Jun 18 23:03:34 <jacques> linuxrc is as I would expect - I just flashed back my old kernel
Jun 18 23:04:00 <jacques> ttySAC0 appears to be a bust :-(
Jun 18 23:04:53 <jacques> what does the beep and change from yellow to red mean ?
Jun 18 23:05:02 <jbowler-away> Was there a /.recovery on /dev/sda2?
Jun 18 23:05:20 <jacques> dunno , already rebooting
Jun 18 23:05:41 <jbowler-away> The beep means that nslu2-io.c init has just run - amber is 'redboot' and most of the kernel reboot, 'red' happens near the end.
Jun 18 23:06:19 <jacques> ok it's up and it booted from the disk fine
Jun 18 23:06:30 <jacques> the bad console must have confused it before
Jun 18 23:07:56 <jbowler-away> Do you have /lib/modules/2.6.12/... - on the flash paritition (containing reiserfs)?
Jun 18 23:08:26 <jbowler-away> 'cause if you just reflashed the kernel and you have both module trees the partition is going to be pretty full.
Jun 18 23:09:27 <jacques> I built this image with 2.6.12
Jun 18 23:09:43 <jacques> I only started reflashing the kernel to try the different cmdlines
Jun 18 23:10:04 <jacques> as soon as I reflashed with the original 2.6.12, it booted to the disk
Jun 18 23:11:09 <jbowler-away> Hum, then there is something really weird with the console handling.
Jun 18 23:11:29 <jacques> I tend to agree
Jun 18 23:15:12 <jbowler-away> user.warn kernel: ttyS0 at MMIO 0xc8000000 (irq = 15) is a XScale
Jun 18 23:16:41 <jacques> sounds right, maybe those posts I found on the list are relevant then and there is a bug
Jun 18 23:16:43 <jbowler-away> Nothing else in my log apart from one of the irq26's
Jun 18 23:18:41 <jbowler-away> Same problem for me - no reiserfs boot
Jun 18 23:23:13 <jacques> heh, we both user reiserfs
Jun 18 23:24:26 <jbowler-away> echo foo >/dev/console gives no such device, echo foo >/dev/ttyS0 is fine.
Jun 18 23:26:13 <jacques> ah well, it was a red herring then
Jun 18 23:32:24 <jbowler-away> yes, complete ruby sardine
Jun 18 23:32:37 <jbowler-away> I just changed the command line back and it booted.
Jun 18 23:33:47 <jbowler-away> No it didn't, change didn't stick.
Jun 18 23:43:18 <jbowler-away> With the change I can boot to my reiserfs partition
Jun 18 23:43:32 <jbowler-away> (Well, without the change - with ttyS0 as the console device)
Jun 18 23:43:42 <jacques> hmm
Jun 18 23:44:08 <jbowler-away> I'm going to test NFS and if it works I'll do a complete, from scratch, set of builds.
Jun 18 23:45:28 <jacques> this is really minor compared to the problems I was expecting with 2.6.12
Jun 18 23:47:11 <jbowler-away> yes, it may still just be a .config issue - I'm testing an incremental build
Jun 18 23:47:37 <jbowler-away> Still I just got a potential incompatibility: user.warn kernel: NFS: mount program didn't pass remote address!
Jun 18 23:48:36 <jbowler-away> But it's a warning - turnup seems to be copying stuff fine!
Jun 18 23:52:05 <jbowler-away> Hum, the nslu2-kernel.inc command line lack the 'reboot=s' parameter necessary to use my soft reboot code.
Jun 18 23:56:20 <jbowler-away> NFS boot works fine.
Jun 18 23:56:33 <jacques> cool
**** ENDING LOGGING AT Sat Jun 18 23:59:56 2005