**** BEGIN LOGGING AT Fri Oct 28 02:59:59 2005 Oct 28 08:13:48 hi, wanted to add my pegasus patch to the archive as agreed last saturday, but found two bb files, don't know to which one to add my patch, the bb files I found are nslu2-kernel_2.6.14-rc5-mm1.bb nslu2-kernel_2.6.14.bb Oct 28 08:13:53 anyone any idea Oct 28 08:14:55 I think jbowler made 2.6.14 default this week Oct 28 08:18:12 koen, tnx. Apparently I did something wrong; I did a mt update;mt pull; after that there was no sign of these bb files so I made my change in the 2.6.14-rc5 bb file but when I wanted to commit I got a complaint on a 3-way merge Oct 28 08:18:33 try pulling before update Oct 28 08:18:41 did an mt update and pull again and got the new version, I now disapproved my changes Oct 28 08:18:53 pull connect to the server, update uses the local database to update the files Oct 28 08:18:53 ah, I thought it had to be the other way around Oct 28 08:19:02 I see Oct 28 08:19:17 I'll redo the changes and submit Oct 28 08:22:52 koen, I'm fully lost. I did mt pull; mt update; then bb openslug-kernel and it tries to build openslug-kernel-2.6.13.2-r1.1 Oct 28 08:22:56 ?????? Oct 28 08:23:20 no idea on that Oct 28 08:23:29 I'm away from a slug sourcetree at the moment Oct 28 08:23:54 anyone else ? Oct 28 09:23:31 koen, still puzzling why I get this 2.6.13 kernel whereas I got a .14 version earlier; I expect openslug-kernel_2.6.14-rc5.bb to be considered as a higher version than openslug-kernel_2.6.13.2.bb. right? both bb files reside in packages/linux; shouldn't bb openslug-linux have picked the .14-rc5 version? Oct 28 09:23:51 also there is a nslu2-kernel_2.6.14.bb file, but no openslug equivalent Oct 28 09:26:01 I guess openslug specifies a specific version Oct 28 09:26:29 any idea where this happens? Oct 28 09:27:42 openslug-kernel_2.6.14-rc5.bb only says include nslu2-kernel_${PV}.bb Oct 28 09:43:13 jbowler, can you perhaps help me out, I try to build the head for openslug and I get 2.6.13.2 iso 2.6.14 (and last weekend I did build 2.6.14) Oct 28 09:43:24 see also my text above Oct 28 10:41:44 eFfeM: if you build openslug from the head you will get nslu2-kernel_2.6.14, openslug-kernel isn't used any more. Oct 28 10:42:00 And I will delete the .bb files soon... Oct 28 10:48:46 ah ok Oct 28 10:48:55 tnx Oct 28 10:51:02 You probably need to at least remove tmp/deploy/ipk/kernel* and tmp/deploy/ipk/ixp* to be sure to get a consistent rootfs - I'd suggested rebuilding the whole thing now we have 2.6.14 Oct 28 10:53:49 jbowler, I'm starting it up now, cleared out all of tmp except cross and staging Oct 28 10:54:18 You need to remove staging too to be safe - it may contain incorrect linux headers. Oct 28 10:54:41 done Oct 28 10:54:43 It does not harm to remove cross - bitbake doesn't know it is correct so will just rebuild it. Oct 28 10:54:50 thought it was for the cross compiler Oct 28 10:55:04 Yes, but that gets built close to the start of the build. Oct 28 10:55:30 ok, just removed all of tmp just to be sure Oct 28 10:55:47 I'll have to go in a few mins anyway, so the sys has a few hrs to rebuilt anyway Oct 28 10:55:52 Everything in 'tmp' can be deleted, 'downloads' (which is in ..) is the only thing you need to retain (to avoid downloads...) Oct 28 10:56:01 what exactly will bb openslug give me? kernel and image? I found it quite convenient to have a way to only build the kernel Oct 28 10:57:20 Try "bitbake virtual/kernel" Oct 28 10:57:33 ok, learned something new Oct 28 10:57:58 I think that works, but "bitbake openslug-image" builds just kernel and rootfs (no packages) and that definately works. Oct 28 10:58:14 ok, that is already fine with me Oct 28 10:58:31 i'll try the virtual/kernel later this weekend Oct 28 10:58:39 seems there is also some wiki updating needed Oct 28 10:58:55 Oh yes, certainly true. Oct 28 10:59:12 is LE already fully operational? I did not completely understand that from your msg in nslu2-developers Oct 28 11:01:15 I'll report results of the build later today or tomorrow; gotta leave now Oct 28 11:01:18 Yes Oct 28 11:01:40 (I.e. LE is fully operational.) Oct 28 11:01:46 I assume the yes was for LE, probably try that one as well this weekend, Oct 28 11:01:51 our messages crossed ... Oct 28 11:02:10 I assume LE is more efficient wrt networking, right? Oct 28 11:02:56 thanks alot for your info and also for your work! Oct 28 11:02:59 Yes - the 'yes' was for the LE question, BE should be more efficient for networking but the difference will be very slight, if any. Oct 28 11:05:09 <_snow8> is there a dyndns updater for zoneedit avail for nslu2? Oct 28 14:23:17 jbowler, managed to do the build (using bb openslug-image); the resulting img boots if the hd is disconnected but not if the hd is connected (and I already booted without the hd, then copied the modules from flash to hd and rebooted, don't want to loose everying on the disk) Oct 28 14:29:40 hm, turnup also does not want to do the job: Oct 28 14:29:42 oot@NSLU2:/# turnup disk /dev/sda5 Oct 28 14:29:42 /usr/sbin/turnup: umounting any existing mount of /dev/mtdblock4 Oct 28 14:29:42 sed: /tmp/rootfs.1251/etc/fstab: No such file or directory Oct 28 14:30:48 eFfeM: it's a kernel upgrade so you need to follow the upgrade instructions on the wiki Oct 28 14:30:58 ok, tnx Oct 28 14:31:18 i was trying to do the same thing as I did when going from rc2 to rc3 to rc5 Oct 28 14:34:14 jbowler, I'm looking at http://www.nslu2-linux.org/wiki/OpenSlug/UpgradingOpenSlugToANewRelease but that does not help me as I already flashed with the latest image (that I just have build) Oct 28 14:34:28 I was expecting that turnup would bring me back in business again Oct 28 14:35:23 "If you forgot to upgrade a root paritition" Oct 28 14:35:48 However, you're system is in a VERY bad state - turnup shouldn't be executing a sed on fstab!!! Oct 28 14:36:07 Someone has been hacking your turnup. Oct 28 14:38:00 Hum, it's changed... I didn't think setup_fstab got run without -i, but it apparently does... Oct 28 14:38:03 hm, I'm just using the head version Oct 28 14:39:16 there is this line in my turnup Oct 28 14:39:18 sed -i '\@[ ]/var[ ][ ]*tmpfs[ ]@d' "$1"/etc/fstab Oct 28 14:40:02 Yes, so apparently there is no /etc/fstab in your root partition, so it isn't surprising it doesn't boot... Oct 28 14:40:33 I probably changed it to put in modified -o mount options. Oct 28 14:40:53 should there also be a /etc/fstab on the hd root fs? Oct 28 14:41:10 e.g. /mnt/etc/fstab for me Oct 28 14:41:12 That's the partition I meant - i.e. the one you are trying to boot to. Oct 28 14:41:57 ah, ok, I know what I did wrong. Because I wanted to avoid that my scripts in /etc are overwritten I did mv etc etc- before turnup Oct 28 14:43:01 just copied it back and now turnup works Oct 28 14:43:14 expected turnup to scribble a completely fresh /etc Oct 29 02:26:40 jbowler-zzz, got everything running again Oct 29 02:27:30 i think the root cause of my problem was that in etc/modprobe.conf the alias for eth0 was changed from ixp425_eth to ixp400_eth Oct 29 02:28:17 ipkg -o did not resolve my problems so I ended in renaming etc, dev, usr etc on my hd and do a turnup disk -f -i /dev/sda5 Oct 29 02:28:23 this gave me a booting system again Oct 29 02:29:38 I probably should have removed the ipkg database as well as I had to use -force-downgrade when installing additional modules afterwards (I was on rc5 and apparently versionwise this is considered to be higher Oct 29 02:29:54 anyway, got it working again & wanted to let you know. Oct 29 02:30:03 thanks again for your help Oct 29 02:31:14 (btw, one of the other reasons ipkg -o did not help was because I am on the head and so the unstable feed, but this still contained the rc5 versions of all packages) Oct 29 02:44:14 wrt the release the only suggestion I have is to give a msg in the beginning of the build process about the intel files that need to be downloaded (I just started to run make, examined that it started correctly and later discovered that it stalled on its way due to the intel files) **** ENDING LOGGING AT Sat Oct 29 03:00:00 2005