**** BEGIN LOGGING AT Sun Feb 20 23:59:56 2005 Feb 21 02:41:38 03dereky 07 * r1.2834 10openembedded/packages/openslug-init/openslug-init_0.10.bb: added creation of /initrd for switchbox compatibility Feb 21 02:47:59 03dyoung 07 * r1.2835 10openembedded/packages/openslug-init/openslug-init_0.10.bb: Oops forgot to bump the PR Feb 21 04:14:52 kergoth`zzz: ping Feb 21 04:15:04 so OpenSlug's /sbin/init has the following code in it: Feb 21 04:15:13 #ifndef DEBUG_INIT Feb 21 04:15:14 if (getpid() != 1 Feb 21 04:15:14 #ifdef CONFIG_FEATURE_INITRD Feb 21 04:15:14 && strstr(bb_applet_name, "linuxrc") == NULL Feb 21 04:15:16 #endif Feb 21 04:15:16 ) { Feb 21 04:15:18 bb_show_usage(); Feb 21 04:15:20 } Feb 21 04:15:44 which means that it either has to be PID 1, or invoked as /linuxrc. Feb 21 04:16:04 which means that it can't be used as a target for the exec chroot that happens after a pivot_root Feb 21 04:18:15 so a symlink from /sbin/init to /linuxrc is called for in openslug-init ... Feb 21 04:18:31 ln -s /sbin/init /linuxrc Feb 21 04:20:44 very quiet here tonight .... Feb 21 04:20:57 rwhitby-away: i know and you're wrong Feb 21 04:21:16 rwhitby-away: i mean, you're right, but you just have to make sure it is pid 1 Feb 21 04:21:35 rwhitby-away: and, /sbin/init -i works too (undocumented i guess) Feb 21 04:22:07 rwhitby-away: but you really have to make sure it is pid 1, or else you get a swapper process which eats up your cpu time Feb 21 04:22:15 how do you make sure it is pid 1 when you're pivot_rooting to it? Feb 21 04:22:27 root=/dev/ram0 init=/linuxrc Feb 21 04:22:34 then your linuxrc execs as pid 1 Feb 21 04:22:47 or, with devfs automount: root=/dev/rd/0 Feb 21 04:23:01 yeah, yeah - that's how you do it without pivot_root. Feb 21 04:23:05 no no Feb 21 04:23:09 that is the pivot_root solution Feb 21 04:23:10 I asked how you do it with pivot_root Feb 21 04:23:41 didn't [g2] fix openslug to use that exact command line? Feb 21 04:24:05 (sorry for arguing - you know more about this than me I think) Feb 21 04:24:49 ok, my bad Feb 21 04:25:15 I forgot to change the cmdline_root back to /dev/ram0 .... Feb 21 04:25:30 thx giel - you saved me going down a rathole Feb 21 04:25:50 uhu Feb 21 04:26:08 i've discussed this for a while with the author of initrd Feb 21 04:26:25 np Feb 21 04:27:17 ok, let me try again with the right kernel cmdline Feb 21 04:42:50 hmm : Feb 21 04:42:55 RAMDISK: Compressed image found at block 0 Feb 21 04:42:55 VFS: Mounted root (ext2 filesystem) readonly. Feb 21 04:42:55 Freeing init memory: 72K Feb 21 04:42:55 BusyBox v1.00 (2005.02.21-11:08+0000) Built-in shell (ash) Feb 21 04:42:56 Enter 'help' for a list of built-in commands. Feb 21 04:42:56 # Feb 21 04:43:04 giel: any idea what's happening there? Feb 21 04:43:29 # /linuxrc Feb 21 04:43:29 Preferred root device is jffs2 Feb 21 04:43:29 Root filesystem will be mounted from /dev/mtdblock4 ... Feb 21 04:43:29 Usage: init 0123456SsQqAaBbCcUu Feb 21 04:43:29 # Feb 21 04:44:15 running /sbin/init -i after that boots OpenSlug properly ... Feb 21 04:45:14 root@nslu2:~# ps -ef Feb 21 04:45:14 PID Uid VmSize Stat Command Feb 21 04:45:14 1 root 188 S /bin/sh Feb 21 04:45:14 2 root SWN [ksoftirqd/0] Feb 21 04:45:15 3 root SW< [events/0] Feb 21 04:45:15 4 root SW< [khelper] Feb 21 04:45:17 11 root SW< [kblockd/0] Feb 21 04:45:19 21 root SW [kapmd] Feb 21 04:45:21 23 root SW [pdflush] Feb 21 04:45:23 24 root SW [pdflush] Feb 21 04:45:25 26 root SW< [aio/0] Feb 21 04:45:27 25 root SW [kswapd0] Feb 21 04:45:29 570 root SW [mtdblockd] Feb 21 04:45:31 645 root 532 S init [2] Feb 21 04:45:33 742 root SWN [jffs2_gcd_mtd4] Feb 21 04:45:35 886 root DW [ixp425_csr] Feb 21 04:45:37 887 root SW< [ixp425_eth/0] Feb 21 04:45:39 893 root SW [khubd] Feb 21 04:45:41 954 daemon 456 S /sbin/portmap Feb 21 04:45:43 1653 root 648 S /usr/sbin/dropbear -r /etc/dropbear/dropbear_rsa_host Feb 21 04:45:45 1678 root 584 S /sbin/syslogd -n -C 64 -m 20 Feb 21 04:45:47 1680 root 568 S /sbin/klogd -n Feb 21 04:45:49 1686 root 768 S -sh Feb 21 04:45:51 1692 root 708 R ps -ef Feb 21 04:45:53 root@nslu2:~# Feb 21 04:45:57 that's the process tree at the end of that manual running of /linuxrc and then manual running on /sbin/init -i Feb 21 04:46:08 I guess I should have run exec /sbin/init -i .... Feb 21 04:46:54 ah - cmdline is wrong: Feb 21 04:47:08 root=/dev/ram0 rootfstype=ext2,jffs2 initrd=0x01000000,10M /init=/linuxrc mem=32M@0x00000000 console=ttyS0,115200n8 Feb 21 04:47:16 should just be init=/linuxrc ... Feb 21 04:47:25 not /init= ... Feb 21 04:49:42 yeah Feb 21 04:55:22 ok, unslung still works after the change to switchbox and slingbox to start using /usr/sbin and /usr/bin again (so that the location of chroot matches openslug) Feb 21 04:55:46 (I'll push everything once openslug works too) Feb 21 04:59:39 YES! Feb 21 05:00:02 OpenSlug works with switchbox 4 Feb 21 05:00:13 thx again giel Feb 21 05:00:17 ~worship giel Feb 21 05:00:20 * jbot bows down before giel. "I'm not worthy!" Feb 21 05:00:39 WOW! Feb 21 05:04:05 03rwhitby 07 * r1.2836 10openembedded/ (10 files in 8 dirs): Feb 21 05:04:05 Unslung and OpenSlug both working with switchbox. Updated CVSDATE stuff in Feb 21 05:04:05 nslu2.conf. Changed CMDLINE in openslug kernel to use switchbox. Changed Feb 21 05:04:05 switchbox and slingbox to use /usr/sbin and /usr/bin (just like in OpenSlug Feb 21 05:04:05 busybox) so that the location of executables match across SwitchBox, Unslung and Feb 21 05:04:06 Openslug. Added /initrd directory to OpenSlug rootfs so that it can be used for Feb 21 05:04:08 the pivot_root in switchbox. Feb 21 05:04:24 Phew! Feb 21 05:04:57 I hope everyone feels all unified again now :-) Feb 21 05:05:42 Can I have volunteers to build OpenSlug and Unslung from scratch and make sure they work for you? Feb 21 05:12:32 Don't all step forwards at the same time ;-) Feb 21 05:13:19 no time, today. maybe tommorrow (if I can get my dev. env. to work again) Feb 21 05:13:37 cool, thx perlguru-work Feb 21 05:13:54 still got mu ipkg-native problem Feb 21 05:13:59 mu->my Feb 21 05:14:21 what problem is that? Feb 21 05:15:20 it won't build. libbb problem. full_read already defined Feb 21 05:16:11 build under which environment? Feb 21 05:17:19 bitbake unslung-image Feb 21 05:17:45 do you have the latest bitbake? Feb 21 05:18:06 look at log of 14/2 around 19:00 Feb 21 05:18:25 ( that was also the last time i tried it) Feb 21 05:18:27 <[g2]> someone's been busy :) Feb 21 05:18:28 morning sunshine - I've got good news for you Feb 21 05:18:34 hey, [g2] Feb 21 05:18:36 <[g2]> I read the change logs Feb 21 05:18:46 <[g2]> FILES += Feb 21 05:18:47 <[g2]> :) Feb 21 05:19:03 <[g2]> well changeset logs Feb 21 05:19:28 <[g2]> that's great news not good news Feb 21 05:19:53 <[g2]> hi perlguru-work Feb 21 05:20:10 [g2]: it needs some independent testing, but it seems to work here ... Feb 21 05:20:45 and I haven't tested anything except switching to jffs2 Feb 21 05:22:28 <[g2]> I've pulled and I'm baking now Feb 21 05:23:22 BTW, [g2], it's init=, not /init= :-) Feb 21 05:23:57 rwhitby: you might need to add 'rw' to the command line too Feb 21 05:23:57 <[g2]> hey :) Feb 21 05:24:10 rwhitby: if you're pulling drivers out of the payload or jffs2 Feb 21 05:24:20 rwhitby: 'cause initrd gets mounted ro n ow Feb 21 05:24:23 now* Feb 21 05:24:46 rwhitby: or add a "mount -o remount /" somewhere Feb 21 05:24:58 giel: yeah, looking at that now ... Feb 21 05:25:11 rwhitby: adding rw to the command line pbly does the trick Feb 21 05:25:25 we need to update switchbox anyway to get drivers from jffs2 instead of payload for nfsroot Feb 21 05:26:01 [g2]: does the openslug kernel now have everything in there static for usb disk access? Feb 21 05:26:42 [g2]: can we put a .openslug file in openslug-init to capture the version number? Feb 21 05:28:31 <[g2]> rwhitby, no it's modules Feb 21 05:29:53 <[g2]> .openslug, sure openslug-1.0-rc1 ? Feb 21 05:30:02 <[g2]> OpenSlug-1.0-rc1 ? Feb 21 05:30:05 ok, that presents a problem for switchbox switching to external usb disk rootfs for openslug Feb 21 05:30:30 OpenSlug-1.0-alpha ? Feb 21 05:30:53 [you know how much I love going straight past 1.0 and never looking back ...] Feb 21 05:30:54 <[g2]> -alpha-1 Feb 21 05:31:07 no, next would be 1.1-alpha Feb 21 05:31:38 <[g2]> actually it'll probably be 1.1-beta soon :) Feb 21 05:31:53 one version number (incremented often) and the status after that Feb 21 05:32:06 My money is on 1.5-beta Feb 21 05:32:24 <[g2]> ok -0.9-alpha ? Feb 21 05:32:32 (based on Unslung history, it takes 5 versions from feature freeze to beta release) Feb 21 05:32:55 you're really hooked on that magic 1.0 number, aren't you :-) Feb 21 05:33:04 <[g2]> or 0.5-alpha :) Feb 21 05:33:32 what is it about 1.0 that people never want to get past it ? Feb 21 05:33:53 <[g2]> I'm fine with 1.0 for beta Feb 21 05:34:13 would you be more likely to try a new program if it is marked 1.0 or marked 1.6 ? Feb 21 05:34:36 * [g2] doesn't care too much about the number, I try to look at the substance Feb 21 05:34:51 to me, 1.0 screams "this is flakey and you should wait till the real release with the bug fixes" :-) Feb 21 05:35:09 <[g2]> google uses beta for years Feb 21 05:35:41 <[g2]> call it 1.1-alphs then :) Feb 21 05:36:21 * rwhitby is looking up the database to see what the current version is called, and then will just increment taht Feb 21 05:36:34 <[g2]> 1.x.1 Feb 21 05:36:39 <[g2]> 1.y Feb 21 05:37:05 1.1-alpha it is then Feb 21 05:38:26 siddy-away: msg me when you get back and have a couple of mins to spare Feb 21 05:40:25 [g2]: it's good to all be working on the same thing again. takes me back to the old days of nslu2-linux .... Feb 21 05:41:05 <[g2]> rwhitby, it's very fun Feb 21 05:43:27 <[g2]> rwhitby, I don't think the kernel pr number got bumped Feb 21 05:44:41 yeah, about to push the rw change, and I bumped it then Feb 21 05:45:01 just flashing it now to make sure the version got in there Feb 21 05:47:35 can't push cause the repo is locked ... Feb 21 05:49:52 <[g2]> someone doing a pull ? Feb 21 05:50:19 someone else doing a push it seems ... Feb 21 05:51:11 hmm - that's a long time for a push. I'm going to remove the lock Feb 21 05:51:12 <[g2]> or a pull for oe-devel maybe ? Feb 21 05:51:33 <[g2]> can you see who's doing it ? Feb 21 05:51:40 03rwhitby 07 * r1.2836.1.1 10openembedded/packages/ (2 files in 2 dirs): OpenSlug-1.1-alpha, versioned, with rw initrd. Feb 21 05:51:49 there ya go Feb 21 05:52:18 stupid ssh connection died on me. Feb 21 05:52:27 rwhitby: a very late PONG. Feb 21 05:52:57 VoodooZ_log: I should be getting a bluetooth dongle (Linksys USBBT100) tomorrow in the post. You know you are going to get some questions, right :-) Feb 21 05:53:10 hehehe NO problem mate. Feb 21 05:53:34 <[g2]> Applying 1 revisions to packages/kdepimpi/kdepimpi_2.0.9.bb Feb 21 05:53:39 Check my badly written wiki first though and please make corrections if you find I forgot some important stuff. Feb 21 05:53:51 <[g2]> I'd guess that was the auto oe-devel pull Feb 21 05:54:19 rwhitby: Do you know how to use quilt to maintain my large private pool of changes to my repo? Feb 21 05:54:43 now [g2] knows the reason for getting switchbox working for openslug ... Feb 21 05:54:48 VoodooZ_log: nope Feb 21 05:54:53 ok Feb 21 05:55:09 How would you handle a bunch of private changes though? Feb 21 05:55:31 I don't have to - I make the decision to include them all ;-) Feb 21 05:56:01 yeah. That's what I though so. I can't include mine as I usually remove stuff or add weird stuff. Feb 21 05:56:31 If I had to, I would create a new distro, and conditionalise stuff based on that. Feb 21 05:56:47 ouch. Feb 21 05:56:48 conf/distro/voodooslug.conf Feb 21 05:56:57 hehehe Feb 21 05:57:04 or is that voodoozlug Feb 21 05:57:06 It those sound good! Feb 21 05:57:10 srx2_robot Feb 21 05:57:52 Is that simple though? Just creating a distro and configuring my local.conf accordingly? Feb 21 05:57:59 A new distro isn't that much of a deal - we did it for switchbox Feb 21 05:58:10 nothing else to do? Feb 21 05:58:30 But that doesn't help Feb 21 05:58:54 as I still coulnd't push my private changes Feb 21 05:59:04 they are useless to the community Feb 21 05:59:20 if they are conditionalise with your distro, then you can push them. Feb 21 05:59:28 ok. i c Feb 21 05:59:50 <[g2]> VoodooZ_log, most of your changes are localized to the kernel right ? Feb 21 06:00:02 I guess eventually it could become useful to other robot hobbyist. Feb 21 06:00:16 you just have to prove that they don't affect the other distros. Just like we prove to kergoth that our nslu2-linux stuff doesn't stuff up the other distros in openembedded. Feb 21 06:00:21 well, no. I had to change base-files, udev, bluez, etc... Feb 21 06:00:52 Well, most changes were actually fixes to other package's scripts to run under a busybox shell. Feb 21 06:01:12 btw, did somebody push my grep-o-less.patch from the wiki? Feb 21 06:01:30 I guess it's a moot point now that we got rid of hotplug. Feb 21 06:01:34 <[g2]> don't think so, Feb 21 06:01:38 <[g2]> nod. on the hotplug-linux Feb 21 06:02:22 I guess I'll try to figure out what kergoth suggested using quilt to make patch-sets Feb 21 06:02:36 simpler for now and it prevents contention with others. Feb 21 06:03:07 * [g2] needs to learn how to do that also Feb 21 06:03:46 Unless I can just do it with a normal patch command? Feb 21 06:03:46 <[g2]> although I think the changes to use busybox may be valid changes to go upstream Feb 21 06:03:48 VoodooZ_log: I really think a new robot-friendly distro would be easier in the long run Feb 21 06:04:04 definitely. Feb 21 06:04:41 I just feel lazy and overwhelmed right now unfortunately as I have to redo all those changes. Some I forgot about. Feb 21 06:05:15 perfect time to get them into a new distro .... you know you won't do it later .... Feb 21 06:05:48 hehehe damn right Feb 21 06:06:13 ok, but don't be surprised if I ask you questions! :) Feb 21 06:06:36 2.06 hours on average between nslu2-linux commits since day 1, 5.5 months ago Feb 21 06:06:58 how would I deal with distro specific changes like creating a /flash dir though? I previously modified base-files to do it. Feb 21 06:07:01 http://cia.navi.cx/stats/project/nslu2-linux Feb 21 06:07:05 wow Feb 21 06:08:28 Or even the interfaces file? do I just have my distro overwrite that file? How do you conditionalize? Feb 21 06:08:56 VoodooZ_log: look for nslu2, unslung and openslug directories in the repo ... Feb 21 06:09:01 e.g. in base-files Feb 21 06:09:24 FILES_PATH is the key Feb 21 06:10:14 cool. I actually noticed that and put my fstab changes in there. Feb 21 06:10:20 <[g2]> /etc/rcS.d/S35mountall.sh: 23: test: not found Feb 21 06:10:20 <[g2]> /etc/rcS.d/S35mountall.sh: 42: test: not found Feb 21 06:10:20 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:20 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:21 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:23 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:25 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:27 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:29 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:10:31 <[g2]> etc .... Feb 21 06:10:51 [g2]: oh - did you have an old version of switchbox-4.8 ? Feb 21 06:11:19 <[g2]> it got pulled this moring or maybe last night Feb 21 06:11:38 <[g2]> I can delete it and rebuild Feb 21 06:11:40 remove switchbox-4.8.tar.gz* from your downloads dir Feb 21 06:11:45 <[g2]> nod. Feb 21 06:11:48 so it gets re-downloaded Feb 21 06:12:04 it was never mentioned in a .bb, so we didn't bother bumping the version Feb 21 06:12:54 heh, in 3 weeks, jp30 has an average of 8.5 hours between commits Feb 21 06:13:17 my average has dropped to 4.25 hours Feb 21 06:13:26 slacker! ;) Feb 21 06:13:37 yeah, RLB Feb 21 06:14:06 http://cia.navi.cx/stats/author/jp30 Feb 21 06:14:45 I wish I could use a longer nick so I could be called VoodooZ_Pretend2Work Feb 21 06:15:40 [g2]: let me know when it works, cause then I'll go to bed Feb 21 06:15:50 * [g2] is rebuilding now Feb 21 06:16:14 should be quick for just a switchbox change Feb 21 06:17:12 <[g2]> actually it didn't dl switchbox I think I need to remove the stamps also Feb 21 06:17:18 Hmm - Gastro and Nudi are still offline .... Feb 21 06:17:26 yeah remove nslu2-switchbox-firmware Feb 21 06:24:07 http://slugsite.us/ Feb 21 06:24:34 <[g2]> rwhitby, same thing with 242a308a53818f9d7887b5c4cb86ac36 ../sources/switchbox-4.8.tar.gz Feb 21 06:25:17 hmm Feb 21 06:25:31 that's the right md5sum Feb 21 06:25:42 did the image get rebuilt? Feb 21 06:26:00 <[g2]> sure Feb 21 06:26:31 changes were to openslug-init, the kernel, and switchbox. Feb 21 06:26:51 <[g2]> I pulled those Feb 21 06:27:11 <[g2]> I've obviously got the kernel changes or I'd be booting to mtd4 Feb 21 06:27:14 I have no outstanding changes or local mods here ... Feb 21 06:27:34 did chroot run? Feb 21 06:27:51 (before all those errors) Feb 21 06:27:52 <[g2]> dunno switchbox is quiet Feb 21 06:28:10 quiet? Feb 21 06:28:28 <[g2]> NET: Registered protocol family 1 Feb 21 06:28:28 <[g2]> RAMDISK: Compressed image found at block 0 Feb 21 06:28:28 <[g2]> VFS: Mounted root (ext2 filesystem). Feb 21 06:28:28 <[g2]> Freeing init memory: 72K Feb 21 06:28:28 <[g2]> Preferred root device is jffs2 Feb 21 06:28:29 <[g2]> Root filesystem will be mounted from /dev/mtdblock4 ... Feb 21 06:28:34 <[g2]> maybe not so quiet Feb 21 06:28:37 <[g2]> :) Feb 21 06:28:50 what's the next three lines after that? Feb 21 06:29:06 <[g2]> modprobe: FATAL: Could not load /lib/modules/2.6.9/modules.dep: No such file or directory Feb 21 06:29:06 <[g2]> modprobe: FATAL: Could not load /lib/modules/2.6.9/modules.dep: No such file or directory Feb 21 06:29:06 <[g2]> INIT: version 2.86 booting Feb 21 06:29:06 <[g2]> modprobe: FATAL: Could not load /lib/modules/2.6.9/modules.dep: No such file or directory Feb 21 06:29:06 <[g2]> modprobe: FATAL: Could not load /lib/modules/2.6.9/modules.dep: No such file or directory Feb 21 06:29:07 <[g2]> /etc/rcS.d/S35mountall.sh: 23: test: not found Feb 21 06:29:09 <[g2]> /etc/rcS.d/S35mountall.sh: 42: test: not found Feb 21 06:29:11 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:29:13 <[g2]> /etc/init.d/rcS: 92: [: not found Feb 21 06:29:17 <[g2]> the modprobe's are normal Feb 21 06:29:51 <[g2]> then we are off to the other issue Feb 21 06:29:55 well that's weird. Feb 21 06:30:07 I've kicked off full rebuilds now .. Feb 21 06:30:16 <[g2]> ok Feb 21 06:31:16 <[g2]> Filesystem 1k-blocks Used Available Use% Mounted on Feb 21 06:31:16 <[g2]> /dev/mtdblock4 955 395 560 41% /initrd Feb 21 06:31:16 <[g2]> /dev/mtdblock4 6400 4680 1720 73% / Feb 21 06:31:16 <[g2]> /dev/ram0 6400 4680 1720 73% / Feb 21 06:31:16 <[g2]> tmpfs 15324 8 15316 0% /var Feb 21 06:31:18 <[g2]> tmpfs 15324 0 15324 0% /media/ram Feb 21 06:31:30 ah - base-files Feb 21 06:32:01 I must have forgotten to bump the PR on base-files - I added jbowler's /etc/fstab fix Feb 21 06:32:10 and that's what causes those errors. Feb 21 06:32:28 <[g2]> I'll check the /etc/fstab Feb 21 06:32:48 <[g2]> /dev/ram0 / ext2 defaults 1 1 Feb 21 06:32:50 <[g2]> :) Feb 21 06:33:32 ok, blow away base-files and you should be sweet Feb 21 06:35:01 03rwhitby 07 * r1.2839 10openembedded/packages/base-files/base-files_3.0.14.bb: Bumped the base-files PR for nslu2 Feb 21 06:35:04 <[g2]> that's the issue, patching the files in jffs2 works fine, I'll clean and rebuild Feb 21 06:35:14 just pull and rebuild now :-) Feb 21 06:35:15 <[g2]> actually, I'll just pull and rebuild Feb 21 06:38:00 <[g2]> rwhitby, with initrd mounted we could make a little tool that enables switchbox changes from the OpenSlug cmd line Feb 21 06:38:18 <[g2]> and it can be umounted and freed normally Feb 21 06:39:18 initrd is an ext2 fs from ram0 - you can't change it and make it stick Feb 21 06:39:49 <[g2]> you can make a tarball from it Feb 21 06:40:05 <[g2]> gzip it, and flash that Feb 21 06:40:07 <[g2]> right ? Feb 21 06:40:21 <[g2]> add header the header too, Feb 21 06:40:33 you need genext2fs, don't you? Feb 21 06:41:35 <[g2]> actually we'd need to create a ram partition and copy it stuff to there Feb 21 06:41:41 <[g2]> then dd that Feb 21 06:41:54 right Feb 21 06:42:01 that'd work Feb 21 06:42:31 <[g2]> I think all the stuff is there or should be there Feb 21 06:42:34 we need to get openslug's init scripts (maybe linuxrc) to umount /initrd to free up the ram Feb 21 06:49:09 <[g2]> ~emulate rwhitby Feb 21 06:49:10 Another Satisfied Customer! Feb 21 06:49:25 <[g2]> that's the ticket Feb 21 06:52:34 cool Feb 21 06:52:47 <[g2]> I think ppl will be happy :) Feb 21 06:53:02 <[g2]> umount /initrd works Feb 21 06:53:12 don't forget initrd is still mounted and using up ram - can you deal with that in openslug-init ? Feb 21 06:53:41 <[g2]> sure, for now ppl maybe interestd in looking around in the ramdisk Feb 21 06:54:03 yep Feb 21 06:54:55 <[g2]> soon, after booting directly to the hd works (we just need to load drivers) we can free it up Feb 21 06:55:06 <[g2]> for the beta release Feb 21 06:55:45 yep. Are you going to compile the drivers into the kernel for hd ? Feb 21 06:56:32 <[g2]> I'm 50/50 on the issue, 50 compile them in, 50 load from the switchbox linuxrc like the ethernet driver Feb 21 06:57:24 <[g2]> I'll probably first try with them compiled in, and see how close we are to the 1MB limit Feb 21 06:57:34 cool Feb 21 06:57:36 night all Feb 21 06:57:40 ~emulate rwhitby Feb 21 06:57:41 Another Satisfied Customer! Feb 21 06:57:47 ~emulate rwhitb Feb 21 06:57:51 ~emulate rwhitby Feb 21 06:57:52 Another Satisfied Customer! Feb 21 06:58:00 <[g2]> we're using up space in the jffs2 partition cause the zImage is there Feb 21 06:58:02 ~insult jbot Feb 21 06:58:13 <[g2]> thx for the good work Feb 21 06:58:44 np - now OpenSlug is ready for my bluetooth dongle when it arrives tomorrow ... Feb 21 06:58:49 Great work guys Feb 21 06:58:57 <[g2]> rwhitby, which one did you get ? Feb 21 06:59:12 Linksys USBBT100 - based on good reports with Treo650 Feb 21 06:59:17 They are all pretty much supported so it should smooth. Feb 21 06:59:29 <[g2]> is that a Class 1 device ? Feb 21 06:59:33 I checked bluez compatibility before buying it Feb 21 06:59:36 yep, class 1 Feb 21 06:59:41 Just build and install bluez-nodbus and remove the pcmcia stuff. Feb 21 06:59:54 class1? cool. mine is class2 Feb 21 06:59:58 some guy said he got 1000 feet with his Treo650 and the USBBT100 Feb 21 07:00:00 plenty for a small robot. Feb 21 07:00:13 sucks more power though. Feb 21 07:00:15 and that was through walls Feb 21 07:00:30 <[g2]> yeah my IOGear goes quite far Feb 21 07:00:41 [g2]: is it a class1 too? Feb 21 07:00:45 <[g2]> nod. class 1 Feb 21 07:00:58 <[g2]> rwhitby-asleep, sweet dreams Feb 21 07:01:03 Mine is rated 35 meters but only takes like 30mAh Feb 21 07:01:13 30mA even Feb 21 07:34:54 good morning sir kergoth Feb 21 07:38:19 s/good // Feb 21 07:43:08 I'm starting to look in using quilt for my private stuff as you suggested as I'm f'n tired of having to redo them everytime my repo craps out. Feb 21 07:52:32 wow. I must be dumb as I as feel I need a quilt howto. :( Feb 21 07:53:55 it has a lot of commands, definately go through a tutorial Feb 21 07:53:58 quilt new patchname Feb 21 07:54:00 quilt add filename Feb 21 07:54:04 Feb 21 07:54:05 quilt refresh Feb 21 07:54:20 quilt pop # to unapply Feb 21 07:54:22 quilt push # to apply Feb 21 07:54:35 quilt edit filename # to do a quilt add on the file, then spawn $EDITOR Feb 21 07:54:41 yeah, but I'm having a problem understanding the concepts first. LIke the stack. Feb 21 07:54:57 you've never applied a patchset before? Feb 21 07:55:02 The actual syntax is not hard after you understand it. Feb 21 07:55:13 its a set of patches that must be applied in a specific order. thats all. Feb 21 07:55:20 well, I guess I need an actual example to get my terminology right Feb 21 07:55:25 yeah ok Feb 21 07:55:52 But in the context of OE: if I modify base-files_x.bb for example. Feb 21 07:56:11 the first ;patch=1 entry gets 'quilt import'd, then quilt pushed. Feb 21 07:56:13 then the second does Feb 21 07:56:15 then the third Feb 21 07:56:18 applied in order Feb 21 07:56:30 quilt import pulls an external patch into the local patchset. quilt push applies it Feb 21 07:57:17 I'm trying to fit all this with my private changes so I'll have to try it first. Feb 21 07:57:36 cd packages Feb 21 07:57:40 quilt new mybase-files-changes Feb 21 07:57:46 bk edit base-files/base-files_x.bb Feb 21 07:57:51 quilt edit base-files/base-files_x.bb Feb 21 07:57:57 Feb 21 07:57:59 quilt refresh Feb 21 07:58:04 ta da, the patch file is in patches/ Feb 21 07:58:09 ok. Feb 21 07:58:12 quilt pop to unapply it Feb 21 07:58:14 quilt push to apply it again Feb 21 07:58:34 ANd how easy is it to port over my patchset if my repo ever dies again? Feb 21 07:58:39 port? Feb 21 07:58:45 well, more like move. Feb 21 07:58:48 cp -a patches /somewhere/safe Feb 21 07:58:53 cd /path/to/new/packages Feb 21 07:58:58 cp -a /somewhere/safe/patches patches Feb 21 07:59:00 quilt push -a Feb 21 07:59:04 Feb 21 07:59:09 ok. sounds straightforward enough. Feb 21 07:59:18 the 'patches' directory contains a 'series' file Feb 21 07:59:25 the series file describes the patch application order Feb 21 07:59:29 quilt will obey that Feb 21 07:59:29 Perfect. That's a great howto and in only 5 minutes. Feb 21 07:59:55 a hell of a lot easier than having to cp the file to a .orig or something so you can manually do the diff, thats for sure Feb 21 07:59:57 I'll make sure to copy this whole conversation to my notes though. so I won't have to ask again. Feb 21 08:00:03 damn right! Feb 21 08:00:31 one annoying quilt "feature" is that it does a chmod u+w on files when you add them to a patch Feb 21 08:00:46 so if you havent done a bk edit on the file first, it'll screw up bk's perms on the file Feb 21 08:00:49 and make it most unhappy Feb 21 08:01:35 hehehe. I did that a few times to my repo. no wonder it's broken. Feb 21 08:10:33 rwhitby-away, around? Feb 21 08:11:19 <[g2]> perlguru, sleeping Feb 21 08:11:27 ok Feb 21 08:11:31 <[g2]> connection reset Feb 21 08:11:50 still got my problem with bitbaking ipkg-native Feb 21 08:12:05 <[g2]> what the repo access ? Feb 21 08:12:26 ? Feb 21 08:12:39 using the normal repo Feb 21 08:12:41 <[g2]> what's the ipkg-native issue ? Feb 21 08:12:50 | all_read.c:45: error: conflicting types for 'full_read' Feb 21 08:12:50 | libbb.h:119: error: previous declaration of 'full_read' was here Feb 21 08:12:50 | all_read.c:45: error: conflicting types for 'full_read' Feb 21 08:12:50 | libbb.h:119: error: previous declaration of 'full_read' was here Feb 21 08:12:50 | Feb 21 08:13:11 <[g2]> hmmm Feb 21 08:13:12 ipkg-native-0.99.140-r0: Feb 21 08:13:23 <[g2]> that's the new ipkg ? Feb 21 08:13:32 it does this, even after a clean build Feb 21 08:13:45 what do you mean? Feb 21 08:13:57 <[g2]> clean build as in rm -f tmp Feb 21 08:14:07 | ccache gcc -DHAVE_CONFIG_H -I. -I/home/maurice/nslu2/unslung2/build/tmp/work/ipkg-native-0.99.140-r0/ipkg/C/libbb -I.. -I/home/maurice/nslu2/unslung2/build/tmp/staging/x86_64-linux/include -g -O -Wall -DHOST_CPU_STR=\"x86_64\" -DBUILD_CPU=x86_64 -I/home/maurice/nslu2/unslung2/build/tmp/staging/x86_64-linux/include -O2 -MT libbb_la-all_read.lo -MD -MP -MF .deps/libbb_la-all_read.Tpo -c all_read.c -fPIC -DPIC -o .libs/libbb_la-all_r Feb 21 08:14:07 ead.o Feb 21 08:14:26 yes, rm -rf build/tmp Feb 21 08:14:38 also did a new clone Feb 21 08:14:55 <[g2]> did you flush ccache ? Feb 21 08:15:13 flush? Feb 21 08:15:26 <[g2]> clear the ccache Feb 21 08:15:37 how? Feb 21 08:16:14 <[g2]> I think there may be options or you can just delete the directory if you don't mind losing the cached stuff Feb 21 08:17:13 if i delete tmp, the cache dir is also deleted. and i did this several times now. Feb 21 08:18:02 <[g2]> hmmm, I don't know if OE override the local ccache Feb 21 08:18:14 <[g2]> I know I've got a ~/.ccache dir Feb 21 08:18:44 running ccache -c now Feb 21 08:19:03 <[g2]> that was the option I saw :) Feb 21 08:19:14 no luck Feb 21 08:19:19 still get the error Feb 21 08:20:25 <[g2]> then it's probably x86_64 specific Feb 21 08:20:50 <[g2]> can you ship me your system for a little testing ? :) Feb 21 08:21:05 sure, come and get it ;) Feb 21 08:22:03 ccache -C now... Feb 21 08:23:26 <[g2]> that and tmp should be *everything* Feb 21 08:23:52 ok, removing tmp Feb 21 08:24:30 <[g2]> you could also rm the ipkg code from sources so it gets pulled again, but I doubt that's it Feb 21 08:24:44 (this could take some time, 'cause i'm emerging also (glibc)) Feb 21 08:25:00 <[g2]> nod. Feb 21 08:25:21 <[g2]> Am I rushing you along ? Feb 21 08:25:26 * perlguru bitbaking Feb 21 08:25:35 <[g2]> :) Feb 21 08:25:58 * perlguru busy waiting Feb 21 08:27:01 hi [g2], concerning point 2) of your mail: installing "modutils-initscripts" solved it for me. Feb 21 08:27:36 <[g2]> siddy-away, thx I'll check what that point was Feb 21 08:27:53 the update-modules.postinst message Feb 21 08:28:11 <[g2]> ah... yeah Feb 21 08:28:42 <[g2]> I'm not sure whether modutils-initscripts is 2.4 specific Feb 21 08:29:01 dunno, works quite well for me. Feb 21 08:29:09 I think it is. Feb 21 08:29:14 the init script modprobes all the modules listed in /etc/modules. dont recall if it does anything else Feb 21 08:29:21 dont see any reason why it wouldnt work for both Feb 21 08:29:29 <[g2]> most excellent Feb 21 08:29:56 <[g2]> kergoth, THX again for adding hotplug-ng, I don't know if you noticed but I pushed it for OpenSlug baseline Feb 21 08:30:03 yep. thanks. Feb 21 08:30:13 <[g2]> it makes a *huge* difference Feb 21 08:30:43 that could also help with point 5) (loading ixp425_eth and ehci-hcd). It could be done the "normal way" (via /etc/modultils/whatever) Feb 21 08:31:24 <[g2]> I'll test it right now Feb 21 08:34:59 <[g2]> siddy-away, which repo are you pointing at nslu2-linux or oe-devel ? Feb 21 08:37:07 [g2], nslu2-linux Feb 21 08:37:35 <[g2]> you are aware of all the changes that went it recently ? Feb 21 08:38:01 you mean the switchbox-stuff? Feb 21 08:38:11 <[g2]> yeah Feb 21 08:38:55 I've read that, but haven't had the time to take a look at it. Feb 21 08:39:00 <[g2]> it should be in a good state now, but I think it was a little hosed earlier Feb 21 08:39:32 And I won't have the time in the near future.... :-\ Feb 21 08:40:55 <[g2]> just FYI Feb 21 08:41:10 <[g2]> and thx for all the help Feb 21 08:41:12 will booting to usb-disk work with that switchbox out of the box? Feb 21 08:41:33 <[g2]> that's what will be testing very soon Feb 21 08:41:50 <[g2]> it does for unslung, but there kernel has all the modules built in Feb 21 08:42:13 <[g2]> we'll either change the default kernel config or explicitly load the modules from switchbox via the jffs2 partition Feb 21 08:42:30 oh, not the problem with my current kernel. I already have all thet usb-stuff in there. Feb 21 08:42:43 <[g2]> and we should be able to boot to .sda[1|2] and .sdb[1|2] Feb 21 08:43:55 hm... of course I need sdd1, so once again I need switchbox-tuning! ;) Feb 21 08:44:17 <[g2]> currently the initrd is on /initrd Feb 21 08:44:54 <[g2]> I think we might be able to setup the necessary stuff (if it's not already there) to be able to modify it from the OpenSlug cmdline for the brave and bold Feb 21 08:50:43 <[g2]> siddy-away, cool that works for both Feb 21 08:50:55 <[g2]> now I've just got one more question Feb 21 08:51:02 ? Feb 21 08:51:35 <[g2]> I've manually added the two modules to the /etc/modules file but the comments in there say "don't edit" this file it's generated by update-modules Feb 21 08:51:50 <[g2]> what's the proper way of doing this ? Feb 21 08:51:57 base-files? Feb 21 08:52:02 you'll have to create a new file in /etc/modutils and add the two modules there. Feb 21 08:52:22 the /etc/modules is created by update-modules iirc Feb 21 08:52:53 (by copying all files in /etc/modutils/) Feb 21 08:53:41 <[g2]> that's simple enough Feb 21 08:54:33 <[g2]> and we may or may not have to run the update-modules earlier in the init sequence Feb 21 08:55:39 yes, we'll especially have to take care of not braking the first boot, so the ordering may be important. Feb 21 08:56:38 perlguru: I just tried a fresh installtion of OE on a brand new Sun quad opteron and it crashes at ipkg-native too. Feb 21 08:57:01 there goes my chance to "borrow" this nice new machine. Feb 21 08:57:17 <[g2]> root@nslu2:/etc# cat /etc/modutils Feb 21 08:57:17 <[g2]> # Adding modules to be automatically loaded via update-modules Feb 21 08:57:17 <[g2]> ipx425_eth Feb 21 08:57:17 <[g2]> ehci-hcd Feb 21 08:57:35 <[g2]> siddy-away, I ran update-modules after that but it didn't update /etc/modules.conf Feb 21 08:57:43 <[g2]> well /etc/modules Feb 21 08:57:45 What about ohci-hcd? Feb 21 08:58:14 <[g2]> VoodooZ_Work, I think every thing gets loaded via hotplug Feb 21 08:58:21 VoodooZ_Work, good to know I am not the only one... Feb 21 08:58:27 err... you need a directory /etc/modutils/ and place some file in there. Feb 21 08:58:28 cool. Feb 21 08:58:39 perlguru: still. it sucks. Feb 21 08:58:45 and both using AMD64 Feb 21 08:58:53 yes, indeed Feb 21 08:58:55 perlguru: What else did you find out about this problem. Feb 21 08:58:57 <[g2]> VoodooZ_Work, ok ohci doesn't get loaded Feb 21 08:59:06 <[g2]> but everything else Feb 21 08:59:12 [g2]: could be useful for USB 1.1 support. Feb 21 08:59:31 <[g2]> hotplug works for usb Feb 21 08:59:33 perlguru: did you google for this? Feb 21 08:59:40 [g2]: good. Feb 21 08:59:41 <[g2]> except for net Feb 21 08:59:42 no Feb 21 09:00:21 I was looking forward to using the quad opteron to cut my build times in half so I could try more stuff in the same time. Feb 21 09:01:10 VoodooZ_Work, the funny part is: it used to work Feb 21 09:01:44 perlguru: what linux distro are you using ? Feb 21 09:01:51 Gentoo Feb 21 09:01:51 really? Feb 21 09:01:56 <[g2]> siddy-away, THAT works better :) Feb 21 09:01:58 I'm using fedora core 3. Feb 21 09:02:21 When did it work? did you find out what the change was? Feb 21 09:02:38 if we can pinpoint it then we can ask the OE guys perhaps. Feb 21 09:02:47 no, not yet. i have problem since XMas Feb 21 09:03:05 last succesfull build was just before xmas Feb 21 09:03:10 ok. I'll check it out soon. Feb 21 09:03:19 I have to run now. Lunch time. bbl Feb 21 09:04:07 VoodooZ_lunch, last image was of 20050128 Feb 21 09:04:43 Frankenslug was purchased on Saturday... Feb 21 09:08:27 Frakenslug = my test slug Feb 21 09:28:16 <[g2]> siddy I've got a chicken and egg issue with the update Feb 21 09:28:31 <[g2]> Starting portmap daemon: portmap. Feb 21 09:28:31 <[g2]> Configuring kernel-module-ehci-hcd Feb 21 09:28:31 <[g2]> /etc/modutils/modutils.txt: 2: ixp425_eth: not found Feb 21 09:28:31 <[g2]> /etc/modutils/modutils.txt: 3: ehci-hcd: not found Feb 21 09:28:31 <[g2]> /etc/modutils/modutils.txt: 4: ohci-hcd: not found Feb 21 09:28:32 <[g2]> Error while executing /etc/modutils/modutils.txt, aborting Feb 21 09:28:34 <[g2]> Configuring kernel-module-ext3 Feb 21 09:28:36 <[g2]> /etc/modutils/modutils.txt: 2: ixp425_eth: not found Feb 21 09:28:38 <[g2]> /etc/modutils/modutils.txt: 3: ehci-hcd: not found Feb 21 09:28:40 <[g2]> /etc/modutils/modutils.txt: 4: ohci-hcd: not found Feb 21 09:28:42 <[g2]> Error while executing /etc/modutils/modutils.txt, aborting Feb 21 09:32:04