**** BEGIN LOGGING AT Thu Jun 16 23:59:57 2005 Jun 17 05:46:47 [g2]: I noticed that someone pushed in a fix for shutdown. DOes that mean it correctly powers off now? That would be kind of useful for my headless robot. Jun 17 06:33:14 <[g2]> VoodooZ_Work, I haven't tried it out but I'm guessing it works fine Jun 17 06:33:25 <[g2]> jbowler pushed the change Jun 17 06:34:22 <[g2]> I've pulled and build but haven't reflashed yet Jun 17 06:45:26 cool. I'll give it a try. Jun 17 07:59:16 VoodooZ_Work: it shuts down now (power but and 'poweroff' go through the same kernel code ;-) Jun 17 08:00:34 <[g2]> jbowler, I just tried the shutdown -h and it powered off Jun 17 08:01:27 Cool. I'm just about to flash it. Thanks mate. Jun 17 08:02:02 In the case of a robot it's not always practical to connect to it and shut it down. Jun 17 08:02:23 [g2]: the soft reboot seems like the best bet, clearly something is missing since it never works... Jun 17 08:02:25 I need it to become like an appliance ie. power button and that's it. Jun 17 08:02:44 <[g2]> jbowler, the ipkg update with the kernel and kernel modules is pretty close Jun 17 08:03:12 <[g2]> it's gonna need a little tweaking though Jun 17 08:03:20 Perhaps I'll add a soft-power button to my robot to make it more like a PC. Jun 17 08:04:28 [g2]: we're close to rootfs full on glibc, if rootfs does fill ipkg doesn't handle it at all well (I ended up with depmod crashing, consistently) Jun 17 08:04:57 <[g2]> nod Jun 17 08:05:22 <[g2]> there's a depmod issue and the proper files to copy Jun 17 08:05:33 The /dev/random (ethernet random) fix seems to work fine, and it even seems to speed up dropbear a bit (might be my imagination) Jun 17 08:06:32 reflash -i works to swap a glibc build to uclibc (with config file preservation) too - just tested it! Jun 17 08:06:55 <[g2]> great! Jun 17 08:07:06 <[g2]> I think reflash -i may be the way to go for now Jun 17 08:16:35 <[g2]> jbowler, can you type "uptime" ? Jun 17 08:20:27 root@lochalsh:/# uptime Jun 17 08:20:27 15:20:09 up 15 min, load average: 0.99, 0.94, 0.60 Jun 17 08:20:47 The load average is bogus, as explained previously... Jun 17 08:22:57 <[g2]> ok.. Jun 17 08:23:06 <[g2]> my uptime was giving an error Jun 17 08:23:39 <[g2]> probably because my system was a little out of kilter Jun 17 08:23:48 That was uclibc, but it worked for me on glibc a couple of days ago Jun 17 08:24:01 <[g2]> mine too a week or so ago Jun 17 08:24:16 What was the error? Jun 17 08:24:22 <[g2]> uptime Jun 17 08:24:22 <[g2]> 15:24:10 up 2 min, load average: 1.18, 0.54, 0.20 Jun 17 08:24:42 <[g2]> looks like a full flash is better Jun 17 08:24:55 <[g2]> I had only updated parts of the system Jun 17 08:26:45 /usr/lib/ipkg/lists should probably be a symbolic link to /var/lib/ipkg/lists Jun 17 08:31:16 root@lochalsh:~# cat /proc/sys/kernel/random/entropy_avail Jun 17 08:31:16 655 Jun 17 08:31:30 Much better ;-) (That's a diskfull system) Jun 17 08:32:45 <[g2]> diskless ? Jun 17 08:33:10 [g2]: how about 'ipkg install dev-environment' - that would make it easy to put all the required tools for native build in. Jun 17 08:33:36 <[g2]> jbowler, that's what I was going to do Jun 17 08:33:46 <[g2]> I think it's pretty close Jun 17 08:34:11 <[g2]> right now there's an issue with some header files Jun 17 08:34:22 <[g2]> and a minor edit or two Jun 17 08:34:30 <[g2]> but I think it's close to rolling Jun 17 08:36:23 Sounds great - easy (but maybe very slow ;-) builds on glibc or uclibc! Jun 17 09:15:30 <[g2]> jbowler, (sorry I've afk) native compiling is pretty slow, but useful in many cases and easy Jun 17 09:24:07 [g2]: Agreed. I compiled samba natively earlier.. I couldn't get the lastest version to compile cross... Jun 17 09:24:45 <[g2]> NAiL, which version are you running ? Jun 17 09:25:04 <[g2]> 3.14 ? Jun 17 09:25:14 hang on :) Jun 17 09:26:32 ah, right now I'm running 3.10. Came with openslug 1.12pre, haven't updated it. Jun 17 09:26:44 I was running 3.14 when I ran uClibc Jun 17 09:27:07 samba-3.0.10 is in the feed (cross compiled) Jun 17 09:28:17 yeah, but I tried compiling 3.0.14 (IIRC), and failed. Jun 17 09:29:02 So I compiled it natively on the slug. Jun 17 09:29:53 Took overnight, but I got it up and running. Jun 17 09:36:54 [g2], did you install coreutils? that uptime doesn't work I think Jun 17 09:37:08 gives some "proc file not found" error Jun 17 09:37:33 <[g2]> I did install coreutils Jun 17 09:38:53 I'm currently using the procps uptime Jun 17 09:39:10 tho it seems just like the bb one (no extra features) Jun 17 09:40:33 OK, well the procps uptime does give number of users Jun 17 09:40:42 09:15:04 up 4 days, 4:13, 2 users, load average: 1.00, 1.00, 1.00 Jun 17 09:40:50 09:16:39 up 4 days, 4:14, load average: 1.16, 1.03, 1.01 Jun 17 09:41:09 What utils is needed to completely replace busybox? coreutils+procms... anything more? Jun 17 09:41:15 procps, even Jun 17 09:41:22 (bash, of course) Jun 17 09:43:01 I don;t know - that's why I don't like to remove busybox, I just remove symlinks as I add more packages Jun 17 09:43:39 diffutils and findutils are already in openslug base install Jun 17 09:45:32 jbowler, I'll feel pretty bad if I was trashing dropbear for being so much slower than openssh and it was because it wasn't getting enough entropy Jun 17 09:51:17 <[g2]> I just had a need idea for APEX Jun 17 09:51:50 <[g2]> copy the kernel from jffs2 then run an MD5 or CRC32 sum on it and compare it with a file in the jffs2 Jun 17 09:52:19 <[g2]> if it doesn't match, then boot from an alternate place like the kernel in the kernel partition Jun 17 09:56:18 <[g2]> jbowler, ping Jun 17 09:57:03 * [g2] had another crazy idea --- must've been that banana I ate... brain sugar and all Jun 17 09:57:52 <[g2]> jbowler, I was thinking on turnup we might like the option of turing up to a directory on the drive (since we're just chrooting anyway) Jun 17 09:58:10 <[g2]> then we could easily have multiple instances on the same drive partition Jun 17 10:11:02 [g2]: I don't think that works because / is not a mount point, so the checkroot.sh will fail (and possibly other things). Jun 17 10:11:20 loopback mounts should be ok. Jun 17 10:12:14 <[g2]> jbowler, nod. good point Jun 17 10:12:23 <[g2]> hey what's up with ipkg ? Jun 17 10:12:44 <[g2]> I changed my src but it's still inflating packages from nslu2-oe Jun 17 10:13:44 Check in /etc/ipkg/* Jun 17 10:14:19 <[g2]> yeah I just saw that Jun 17 10:18:23 I changed it to use FEED_URIS (as on unslung). There's still a bug - it needs to be FEED_URIS_append_linux (to avoid ending up with the uclibc openslug downloading everything from glibc) Jun 17 11:10:12 <[g2]> jacques, ping Jun 17 11:14:57 [g2], what's up? Jun 17 11:15:44 <[g2]> FYI I'm writing up my HowTo on the native compile for OpenSlug Jun 17 11:15:52 <[g2]> Python is building now Jun 17 11:16:14 <[g2]> it should be done in 15-20 minutes (the howto that is) Jun 17 11:18:56 great! :-) Jun 17 11:22:25 [g2]: Great :D Jun 17 11:30:46 Are there any grave outstanding problems with openslug except the nobody cared issue? Jun 17 11:31:23 Openslug's KnownProblems is empty, thought I'd do something useful today, and fillit up ;) Jun 17 11:31:48 <[g2]> NAiL, nobody cared is a performance issue and may cause crashes.. I think enabling nodebugirq may fix those issues but not really solve the root issue Jun 17 11:32:22 <[g2]> That and the reset issues is better from external devices but not perfect Jun 17 11:32:50 ok, so the reset issue wasn't fixed with the patch jbowler(?) submitted? Jun 17 11:33:04 <[g2]> shutdown -r was Jun 17 11:33:08 aha Jun 17 11:33:15 <[g2]> DOH... Jun 17 11:33:20 <[g2]> shutdown -h was Jun 17 11:33:27 poweroff works, but reboot won't? Jun 17 11:33:49 <[g2]> poweroff works, but reboot doesn't always Jun 17 11:33:54 yeah Jun 17 11:34:03 <[g2]> from an external device Jun 17 11:34:12 ok Jun 17 11:36:58 [g2], I have seen nothing to indicate wwhat you said: "nobody cared is a performance issue and may cause crashes." Jun 17 11:37:10 how do you draw that conclusion? Jun 17 11:37:30 <[g2]> well certainly the logging is a performance issue to the console Jun 17 11:37:47 I never get more than 1 every 5-10 seconds Jun 17 11:38:05 <[g2]> well that's pretty often Jun 17 11:38:18 but not enough to cause anything to block Jun 17 11:38:20 <[g2]> not to mention filling up the log Jun 17 11:39:13 jacques: samba seems to "pause" when I transfer large amount s of data to the slug. This coincides with irq26 Jun 17 11:40:01 I seem to be the only one this happens to :( Jun 17 11:40:56 <[g2]> NAiL, it'd be interesting to see what happens with nodebugirq enabled on the kernel command line Jun 17 11:40:58 I haven't noticed any crashes because of irq26 though. Only thing that has crashed my slug was the x1206-issue Jun 17 11:40:59 yeah I remember you reporting the pause NAiL, I haven't seen it myself Jun 17 11:41:22 yeah (knock) wood I don't remember the last time one of mine crashed Jun 17 11:41:36 even with uptimes of 100+ days doing native devel Jun 17 11:41:51 unionfs crashed mine (well, kernel oops then a hang) Jun 17 11:42:08 I'm running 1.12-pre now. Going to update it soon. Jun 17 11:42:17 [g2]: Is 1.20 anywhere in the vincinity? :) Jun 17 11:42:32 It should probably be called '2.0' ;-) Jun 17 11:42:44 <[g2]> NAiL, I think the next release is getting close.. and I'm all for 2.0 Jun 17 11:42:49 Cool Jun 17 11:43:06 this reminds me - I meant to kick off a build with the latest fixes :-) Jun 17 11:43:19 We just need to fix the soft reboot Jun 17 11:43:25 jbowler-away, so you have it shutting down power now eh? congrats Jun 17 11:43:39 jbowler-away: So that's the roadblock that stops 2.0? :) Jun 17 11:43:43 Yes - it simply wasn't implemented! Jun 17 11:43:47 heh Jun 17 11:43:48 <[g2]> Yes congrats! Jun 17 11:43:51 haha Jun 17 11:44:13 Everything relevant should now be in slugbug Jun 17 11:44:50 Think I'll wait with upgrading until 2.0 comes as a binary. Save my poor cpu cycles :P Jun 17 11:45:41 So there are three bugs: 132(unconfirmed) 145(must fix - shutdown -r) 109 (irq26 - would be nice to fix...) Jun 17 11:45:51 132? Jun 17 11:45:57 haven't seen that one... Jun 17 11:46:00 NAiL: samba-3.0.14a cross-builds just fine for me (I cp'ed samba_3.0.10.bb) - want me to check in the bb? Jun 17 11:46:41 <[g2]> jbowler is AN ANIMAL! Jun 17 11:46:56 <[g2]> go jb, go jb code it up Jun 17 11:47:11 jbowler-away: Been a long time since I tried. Sure, check it in :) Jun 17 11:47:21 It's only a cp of 3.0.10.bb... Jun 17 11:47:21 damn, where is the query bug by number ??? Jun 17 11:47:39 Do a query for 'openslug firmware' (state assigned) Jun 17 11:48:22 NAiL: could you take a look at 132 and comment - I can't reproduce it (it's uclibc 1.12, but I don't have 1.12) Jun 17 11:49:14 jbowler-away: I'd have to rebuild and reflash then. Switched to glibc a long while ago Jun 17 11:49:30 jbowler-away: For all I care, that bug can be closed Jun 17 11:50:12 Ok, I tried glibc and uclibc on the latest builds, I don't remember ever seeing the problem Jun 17 11:50:49 Ok. Just two bugs ;-) Jun 17 11:51:09 The processes were killed off eventually. After an hour or so I'd guess Jun 17 11:52:00 Ah, I've seen something like that - sometimes if one end dies in a bad way the other end doesn't notice. Jun 17 11:52:14 <[g2]> http://www.nslu2-linux.org/wiki/HowTo/OpenSlugNativeCompileEnvironment Jun 17 11:52:27 <[g2]> DOH! Jun 17 11:53:21 even openssh will keep processes alive if the link dies in certain ways Jun 17 11:53:24 jbowler-away: Yeah, except when it happened to me it was easily reproducible. Every time. Jun 17 11:53:39 <[g2]> there we go better Jun 17 11:54:06 <[g2]> I had wanted to thank jacques on the native compile stuff Jun 17 11:54:18 Hum, ok, I assert it's fixed in the latest builds ;-) Jun 17 11:55:24 jbowler-away: ;-) Jun 17 11:56:53 [g2], on ipkg, I made a /etc/ipkg/local-feed.conf Jun 17 11:57:08 and removed the other feed.conf - I didn't have to change main ipkg.conf Jun 17 11:57:18 <[g2]> jacques, cool Jun 17 11:57:29 root@LKG0FB07F:/etc/ipkg# cat local-feed.conf Jun 17 11:57:29 src/gz local http://192.168.0.12/feeds/openslug/local Jun 17 11:57:38 Add something like this to local.conf: FEED_URIS = "${DISTRO}##ftp://marsco/${BUILDNAME}/deploy/ipk" Jun 17 11:58:13 so 'openslug##ftp://myserver/openembedded/tmp/deploy/ipk' Jun 17 11:59:09 <[g2]> OH.. yeah.. and coreutils does break the uptime Jun 17 11:59:19 <[g2]> uptime command Jun 17 12:00:32 [g2], is bison working for you? I think oe-built is broken Jun 17 12:00:38 if I need it I build native Jun 17 12:00:44 <[g2]> dunno Jun 17 12:00:55 <[g2]> I think perl is broken too Jun 17 12:01:03 well yeah Jun 17 12:01:16 bison perl file Jun 17 12:01:18 <[g2]> but I don't there are dependencies with auto* Jun 17 12:01:50 <[g2]> this still needs work, but I think we can then build perl natively Jun 17 12:02:04 I have built it natively multiple times in the last week Jun 17 12:02:13 even with glibc 2.3.5 Jun 17 12:02:24 <[g2]> I have not doubt you have Jun 17 12:02:29 100% test success (if you put hostname in /etc/hosts) Jun 17 12:02:32 <[g2]> I"m hoping everyone can now :) Jun 17 12:02:51 <[g2]> that's excellent! Jun 17 12:03:11 heck, did I mention it appears the libc.so and libpthread.so thing is even fixed now Jun 17 12:03:17 no need to manually edit those file Jun 17 12:03:18 s Jun 17 12:03:33 <[g2]> they were not in todays build Jun 17 12:03:59 then it must just be in glibc cvs - because it's fixed in glibc 2.3.5 Jun 17 12:04:03 <[g2]> maybe I've got to do a full rebuild or misse the cutoff Jun 17 12:04:04 I confirmed with pb Jun 17 12:04:19 <[g2]> in 2.3.5 possibly Jun 17 12:04:21 but I didn't get whether it should be fixed in 2.3.2 Jun 17 12:04:43 he said he committed fixes - but not where he committed them - to cvs or oe Jun 17 12:05:12 (glibc cvs) Jun 17 12:05:23 <[g2]> I'd like to set that list automated and whittled down a bunch Jun 17 12:05:32 <[g2]> I think it's managable now though Jun 17 12:06:11 <[g2]> I'd like to see ppl with native development going then we can resolve the issues for so broken packages and get them fixed in OE Jun 17 12:06:27 <[g2]> but that's going to take some time and learning Jun 17 12:06:35 * [g2] still has a bunch to learn Jun 17 12:06:37 [g2]: Will set up native dev during the weekend. Jun 17 12:06:58 <[g2]> NAiL, cool.. Lets me know how it works for you Jun 17 12:08:30 <[g2]> we also need to nail the cvs date for gnu_config Jun 17 12:09:07 <[g2]> We'll I'm off for a bit Jun 17 12:09:10 <[g2]> see you guys later Jun 17 12:09:11 <[g2]> cheers Jun 17 12:14:16 14400 -rw-r--r-- 1 root root 14723336 May 18 20:56 log.nmbd Jun 17 12:14:17 47080 -rw-r--r-- 1 root root 48154971 May 18 20:58 log.smbd Jun 17 12:14:17 623432 -rw-r--r-- 1 root root 637759812 May 17 22:25 log.smbd.old Jun 17 12:14:30 Ok... *that* is a bit too much logging.... Jun 17 12:15:54 heh Jun 17 12:17:11 I was wondering where all the space on my root fs went :P Jun 17 12:46:08 logfiles are currently not being rotated on openslug Jun 17 12:52:10 yeah Jun 17 13:00:46 The logs shouldn't be enough to fill a CD-ROM anyways ;) Jun 17 13:01:19 Not in a month. Jun 17 13:01:40 In fact, that's just a day, 22 hours Jun 17 13:03:23 I'm going to recompile samba natively anyways. It'll probably reduce the log output by a tad. Jun 17 13:03:52 hmm.. is anyone else experiencing this btw? Jun 17 13:04:18 A lot of the output is simply samba complaining about not finding some ANSI .so Jun 17 13:05:34 dammit Jun 17 13:06:11 it looks like openslug-image wants to build gcc4 Jun 17 13:06:29 how nay times is thos going to be fixed and break again Jun 17 13:06:49 NAiL: the .bb installation is broken, it fails to install either /usr/lib/charsets/*.so or /usr/lib/vfs/*.so Jun 17 13:06:56 I'm fixing it ;-) Jun 17 13:07:33 Still 14a doesn't build ANSI.so - it does CP437.so and CP850.so, maybe they changed it. Jun 17 13:07:56 are other openslug builders putting a gcc version in their local.conf ? Jun 17 13:08:11 shouldn't it be somewhere else? like distro conf ? Jun 17 13:08:13 Oh, and I'm moving /usr/bin/swat to a separate package, since none of the swat text or locale stuff was installed it was probably broken too. Jun 17 13:08:56 jacques: must be recent, my build last night (synced about 12 hours ago) was fine. Jun 17 13:09:06 dammit Jun 17 13:09:34 I don't define anything like that - it's all in openslug.conf/nslu2.conf or nowhere. Jun 17 13:11:15 looking there - havent found it yet Jun 17 13:11:20 NAiL: there's no ANSI module in 14a Jun 17 13:12:27 it doesn't even fscking configure Jun 17 13:13:14 we need to do something Jun 17 13:13:32 either put it in openslug.conf like the other distros have or I don't know what the other optionsa re Jun 17 13:14:03 It hasn't changed and I've got gcc-3.4.4, the only gcc 4 thing I haven't got is "2 revisions to new file packages/gcc/gcc-4.0.0/ldflags.patch" Jun 17 13:14:52 And nslu2-linux.bkbits.net still has this: "DEFAULT_PREFERENCE_nslu2 = "-1"" in gcc_4.0.0.bb Jun 17 13:15:27 Of course, if you removed that hack... Jun 17 13:15:34 I cleared tmp and .ccache Jun 17 13:15:41 and it tried to build gcc-snapshot-cross-20050617-r0: Jun 17 13:15:50 after bui,lding gcc-cross-initial-3.4.4-r0/ Jun 17 13:18:10 OK I think I know what the problem is Jun 17 13:18:28 on unslung we can get away (well used to) with a 4-5 line local.conf Jun 17 13:18:41 I modified my openslug.conf to be similar, and I think it needs the Jun 17 13:18:53 PREFERRED_PROVIDERS stuff tho Jun 17 13:19:13 Yes, I have that (the three lines which are in the SVN local.conf.template) Jun 17 13:19:34 yeah Jun 17 13:19:45 added those back in and trying again Jun 17 13:23:50 OK that fixed it Jun 17 13:24:05 sorry for the false alarm :-\ Jun 17 13:24:30 I think those lines should probably be moved up to openslug.conf Jun 17 13:24:45 yea, in looking for this I looked at the other distro confs Jun 17 13:24:55 and lots fo them have it, along with specifying gcc versions Jun 17 13:25:11 the familiar ones are a good example Jun 17 13:28:18 Ok, done it - I just add, so the stuff will be in there twice, but I think that's ok. Jun 17 13:28:42 thanks jbowler-away :-) Jun 17 13:46:30 hmm, openslug is using a new mtd snap: mtd-snapshot-20050615.tar.bz2 Jun 17 13:46:39 should be on the lookout for any fallout from that Jun 17 13:50:05 cvs.sourceforge.net appears to be not responding Jun 17 13:50:14 which means no upslug checkout :-\ Jun 17 13:51:20 NOTE: package upslug-native-1.0-r0: task do_fetch: failed Jun 17 13:51:20 ERROR: TaskFailed event exception, aborting Jun 17 13:51:37 looks like all of sourceforge is not responding - website isn't either Jun 17 14:09:02 jacques: conf/distro/openslug.conf is now pushed, changeset 1.3558 Jun 17 14:10:24 jbowler-away, thanks will pull Jun 17 14:28:27 NAiL: samba 3.0.14a is checked in now too. Jun 17 19:03:39 <[g2]> you guys know how to start samba ? Jun 17 19:08:08 'samba' Jun 17 19:09:16 /etc/init.d/samba start Jun 17 19:09:21 heh Jun 17 19:10:18 Nope, it's not there in the .bb Jun 17 19:11:02 That's certainly one thing which is missing from the OE version. Jun 17 19:11:17 :-( Jun 17 19:12:35 Ok, I remembered wrong anyway, is smbd for the file service daemon and nmbd for the name daemon, after creating the conf files. **** ENDING LOGGING AT Fri Jun 17 23:59:56 2005