**** BEGIN LOGGING AT Sat Jul 30 23:59:56 2005 Jul 31 00:18:52 where abouts is the kernel stored on the device itself? Jul 31 02:25:31 unbelievable, how many people have done all the packaging work for openslug? Jul 31 02:25:43 amazing.... Jul 31 03:17:51 landslide: a cast of many Jul 31 03:18:40 good afternoon all Jul 31 03:19:00 g'day Jul 31 04:11:07 i'm still compiling. Jul 31 04:11:09 it's up to bind. Jul 31 04:11:15 i hope that's towards the end ;) Jul 31 04:58:32 I just installed the image, that I build myself, the slug is up and running, but there are some problems with the portmap server and rpc Jul 31 05:00:33 http://pastebin.com/326011 is what the dmesg says, when I boot up my slug Jul 31 05:07:45 well, /etc/init.d/portmap works with some tools called pmap_set and pmap_dump, but they aren't included, I tried a ipkg install -force-reinstall portmap for the case, that only my portmap-package is corrupted, but still nothing to see from these binaries. Jul 31 05:07:53 that looks like localhost doesn't get set up Jul 31 05:08:18 * malfi ponders. Jul 31 05:11:20 oh, I needed to do a ipkg install portmap-utils, I wonder why they aren't included by default, now nfs works Jul 31 05:14:48 well, do you agree with me, that portmap should depend on portmap-utils? Jul 31 05:15:14 if it's useless without them, yes Jul 31 05:16:16 RDEPENDS = "portmap-utils" in the portmap .bb Jul 31 05:16:30 (R for Runtime) Jul 31 05:16:58 openembedded/packages/meta/openslug-image.bb doesn't mention these utils, does it trust that they will be included through this dependency? Jul 31 05:17:57 putting the rdepends in portmap will make portmap depend on the portmap-utils package Jul 31 05:18:18 so if openslug-image.bb somehow includes portmap, it will pull in portmap-utils Jul 31 05:18:24 ah I misunderstood you, I thought this RDEPENDS is there Jul 31 05:18:42 but I should add this line, right? Jul 31 05:18:56 yes Jul 31 05:19:04 I'll file a ticket for that Jul 31 05:24:04 FYI: #229 Jul 31 06:52:38 DaKa2: ping Jul 31 06:55:08 hmm Jul 31 06:55:14 monotone propagate org.openembedded.dev org.openembedded.nslu2-linux Jul 31 06:55:14 monotone: misuse: branch 'org.openembedded.nslu2-linux' is not merged Jul 31 06:55:14 make: *** [propagate-from-oe] Error 1 Jul 31 06:55:48 neat Jul 31 06:56:07 Now I've got one of them tinderbox nasty three way ones Jul 31 06:56:07 NAiL: it's merged here ... Jul 31 06:56:32 http://monotone.vanille.de/viewmtn/revision.psp?id=100af48176006080bc6eaf59f5e30c1da29c86c3 Jul 31 06:57:06 I think I pulled *just* before your last rev Jul 31 06:57:24 then two quick commits and make publish-openembedded Jul 31 06:57:37 s/pulled/make update/ Jul 31 06:57:48 NAiL: you don't have the latest Makefile for a start. Jul 31 06:57:58 propagate-from-oe doesn't exist as a target any more Jul 31 06:58:10 odd Jul 31 06:58:11 and the replacement has extra merge checks added Jul 31 06:58:16 I'll try again Jul 31 06:58:25 what's mt status say in /home/slug ? Jul 31 06:58:47 * rwhitby thinks it's time to add a "make status" target Jul 31 06:58:48 new_manifest [9f04bffb1797ad89c2b195a809e102353b27bb90] Jul 31 06:58:58 it updated more stuff this time it seems Jul 31 06:59:04 but I still get that three-way merge Jul 31 06:59:10 how do I get ediff? Jul 31 06:59:17 I use xxdiff Jul 31 06:59:26 I assume that's graphical? Jul 31 06:59:43 yep, but doesn't require gnome or kde Jul 31 07:00:05 I'm still looking for a good non-graphical three way merge program Jul 31 07:00:08 That box is currently cli only Jul 31 07:02:00 urgh Jul 31 07:02:10 Why can't I just apt-get install emacs? Jul 31 07:05:12 and why on earth doesn't monotone find any ext 3-way merge util? Jul 31 07:05:21 It used to find something earlier ;) Jul 31 07:08:12 maah Jul 31 07:09:41 local editor = os.getenv("EDITOR") Jul 31 07:09:41 if editor ~= nil then editor = string.lower(editor) else editor = "" end Jul 31 07:09:41 if program_exists_in_path("kdiff3") then Jul 31 07:09:41 cmd = merge3_kdiff3_cmd (left_path, anc_path, right_path, merged_path, lfile, afile, rfile, outfile) Jul 31 07:09:41 elseif program_exists_in_path ("xxdiff") then Jul 31 07:09:43 cmd = merge3_xxdiff_cmd (left_path, anc_path, right_path, merged_path, lfile, afile, rfile, outfile) Jul 31 07:09:44 elseif string.find(editor, "emacs") ~= nil or string.find(editor, "gnu") ~= nil then Jul 31 07:09:46 if string.find(editor, "xemacs") and program_exists_in_path ("xemacs") then Jul 31 07:09:48 cmd = merge3_emacs_cmd ("xemacs", lfile, afile, rfile, outfile) Jul 31 07:09:50 elseif program_exists_in_path ("emacs") then Jul 31 07:09:52 cmd = merge3_emacs_cmd ("emacs", lfile, afile, rfile, outfile) Jul 31 07:09:54 end Jul 31 07:09:56 elseif string.find(editor, "vim") ~= nil then Jul 31 07:09:58 if os.getenv ("DISPLAY") ~= nil and program_exists_in_path ("gvim") then Jul 31 07:10:00 cmd = merge3_vim_cmd ("gvim", lfile, afile, rfile, outfile) Jul 31 07:10:02 elseif program_exists_in_path ("vim") then Jul 31 07:10:04 cmd = merge3_vim_cmd ("vim", lfile, afile, rfile, outfile) Jul 31 07:10:06 end Jul 31 07:10:08 elseif program_exists_in_path ("meld") then Jul 31 07:10:10 tbl.meld_exists = true Jul 31 07:10:12 io.write (string.format("\nWARNING: 'meld' was choosen to perform external 3-way merge.\n".. Jul 31 07:10:14 "You should merge all changes to *CENTER* file due to limitation of program\n".. Jul 31 07:10:16 "arguments.\n\n")) Jul 31 07:10:18 cmd = merge3_meld_cmd (lfile, afile, rfile) Jul 31 07:10:20 end Jul 31 07:10:22 that's what it does. Jul 31 07:12:32 yeah Jul 31 07:12:52 but how do I get the diff thing that I actually know how to use? Jul 31 07:13:08 it doesn't want to install emacs Jul 31 07:17:47 *what* doesn't want to install emacs? Jul 31 07:18:00 apt-get install emacs Jul 31 07:18:24 so that's a problem with your desktop distro, not openslug Jul 31 07:18:25 how do I figure out which package actually contains what I want? Jul 31 07:18:35 packages.debian.org Jul 31 07:18:37 try emacs21 Jul 31 07:18:47 I did. That apparently installed the x version Jul 31 07:19:22 it's the same thing these days isn't it? Jul 31 07:19:46 emacs -nw Jul 31 07:20:03 or DISPLAY not set Jul 31 07:21:05 aha Jul 31 07:21:08 emacs21-nox Jul 31 07:21:13 might be the solution Jul 31 07:25:23 yes Jul 31 07:25:29 if I set EDITOR as well Jul 31 07:26:30 thanks :) Jul 31 08:00:38 NAiL: That RDEPENDS is wrong! Jul 31 08:00:44 huh? Jul 31 08:00:47 (I.e. slugbug 229) Jul 31 08:00:53 portmap is a firmware package Jul 31 08:01:11 There's very little space in the flash. Jul 31 08:01:15 aha Jul 31 08:01:22 alternative suggestions? Jul 31 08:01:32 Uh, remove it... quickly? Jul 31 08:01:50 The bug malfi reports does not happen on any of my standard systems. Jul 31 08:02:04 And if you look at the nature of the bug it can't happen... Jul 31 08:02:47 And nfs is certainly works or hanjo and others would be jumping up and down saying 'hey' (or similar) by now. Jul 31 08:05:22 I have even tried /etc/init.d/portmap restart in the past and I thought it worked, but it doesn't matter if it doesn't for the firmware image. Jul 31 08:09:27 If this is new it's probably happening because of the IPV6 stuff. Jul 31 08:10:17 'restart' works fine - it just says 'command not found'. That's fine. Jul 31 08:11:16 Yeah Jul 31 08:13:53 I can't reproduce it on my current system, but it looks like it might be as old as Jul 18. Jul 31 08:14:21 Hopefully I'll get the disk from dyoung very soon, so I can set up a proper testing slug Jul 31 08:14:36 right now I can't test stuff directly after my rootdisk died Jul 31 08:16:41 NAiL: did you bump the PR again for that back out of the RDEPENDS ? Jul 31 08:16:47 yes Jul 31 08:16:57 but that commit is wrong Jul 31 08:17:05 cool. that means the autobuild will fix it in the feed Jul 31 08:17:06 gah Jul 31 08:17:09 yeah Jul 31 08:17:25 night all Jul 31 08:17:28 nite Jul 31 08:18:27 sorry for the chaos with this portmap stuff :-( Jul 31 08:23:29 That's ok - 229 will almost certainly come back to you because I can find no way of even creating /var/run/portmap.upgrade-state Jul 31 08:23:47 (I.e. it isn't in the .c files - only the init script) Jul 31 08:25:15 It's apparently a debian thing done when the package is upgraded Jul 31 08:25:21 (On debian) Jul 31 08:26:06 But OE doesn't use the debian install/remove scripts Jul 31 08:30:24 malfi: you also have no lockd installed, therefore something is wrong with your build, because a standard build has /lib/modules/2.6.12.2/kernel/fs/lockd/lockd.ko and this gets invoked automagically by the kernel. Jul 31 08:45:40 jbowler-away: Well, I installed the portmap utils, did a /etc/init.d/portmap restart and no is everything working. lockd is listed on lsmod Jul 31 08:46:04 jbowler-away: I will reflash my broken image, to see, what it it's loaded the firsttime Jul 31 08:46:41 jbowler-away: ... restart and noW is everthing working Jul 31 08:47:43 Yeah... but where did that file come from? I can't see anything which can create it. Jul 31 08:50:54 I don't have the slightiest idea. Jul 31 08:52:28 ...something you did created it, or the steps you gave to reproduce the bug in the bug report were incommplete. (The file must exist for pmap_set to be executed). Jul 31 08:55:47 I'm now trying the very same steps, I did some hours ago... Jul 31 08:57:10 I don't understand this. I reflashed the very same image, ssh'ed into it, did a turnup nfs host:/path, then a reboot and now everything is working ponder Jul 31 08:57:56 That sequence of steps doesn't require lockd because the root mount is done -o nolock Jul 31 08:58:50 well, this makes me look very stupid. I'm really confused now. Jul 31 08:59:51 well, I ran /etc/init.d/portmap restart, and then pmap_dump is executed even though there is no /var/run/portmap.upgrade-state Jul 31 09:02:30 As I'm not able to reproduce this, can you forgive any inconvenience made by me and close this bug? Jul 31 09:05:15 To explain it, probably I just saw the pmap_dump in restart), thought that it's not good, if a initd-file complains about a missing program, installed the portmap-utils and enjoyed a flawless portmap restart. It probably has never run pmap_set or created this /var/run/portmap.upgrade-state Jul 31 09:09:28 jbowler-away: Why do you think, there might be a /var/run/portmap.upgrade-state involved? I checked my log file and the ticket. There is nothing that indicates that, at least I do not see anything... Jul 31 09:24:01 That bug is already resolved 'works for me'. Jul 31 09:24:51 The /etc/init.d/portmap script only invokes pmap_set or pmap_dump if. either, that file exsts or '/etc/init.d/portmap restart' is executed. Jul 31 09:26:22 So far as I can remember, when the package is reinstalled the most ipkg does is 'stop' followed by 'start' - not 'restart', you said you saw a failure from the absence of the commands and you didn't say you did a 'portmap restart', so I deduced the file had to exist. Jul 31 11:54:29 rwhitby-away: pong (a bit late :) Jul 31 14:35:15 woohooo.. openslug is built! Jul 31 14:35:41 congrats Jul 31 14:39:26 3 days later ;) Jul 31 14:41:08 3 days? ow.. a complete rebuild takes like 3-5 hours for me.. Jul 31 14:43:13 and now with this master_makefile_downloading_initial_db_for_you thing monotone doesn't take that long to set up either Jul 31 15:08:27 an initial make setup takes 9 minutes on the slug ^^, really nice Jul 31 15:16:28 and, building a simple package of mine (bb -b openembedded/packages/vlan/vlan_1.8.bb) takes 32 secs :) Jul 31 15:20:06 native bitbake compilation seems to be working ok, just a few more steps to automate and it would be as simple as: 1. ipkg install openslug-native, 2. wget www.nsul2-linux.org/Makefile && make setup && cd openslug && make setup-env-native conf/local.conf-native, 3. cd openslug && bb package Jul 31 15:23:08 DaKa2: I intend to make the master makefilre aware of which arch it is running on, and do the -native select automatically. Jul 31 15:23:14 things to be fixed: 1. openslug-native.conf distro, or something like that, 2. conf/local.conf use that distro, 3. add/fix some packages so they install without force: ipkg-utils libtool quilt pkgconfig, 4. A set of freezed packages, running on all 600+ packages in openslug is not fun. Jul 31 15:23:21 rwhitby-away: sounds good Jul 31 15:24:06 After that there should be testing of what packages should be in ASSUME_PROVIDED Jul 31 15:24:25 and add the packages there and in openslug-native.bb Jul 31 15:30:21 DaKa2: can you do the initiak openslug-native.conf? that would move us forward ... Jul 31 15:30:51 rwhitby-away: ok, in distro/ ? Jul 31 15:32:31 yep Jul 31 15:32:42 ok, done in a couple of minutes Jul 31 15:32:55 no hurry Jul 31 15:33:22 :) Jul 31 15:33:48 and then a local.conf.template.native or something like that, which I will eventually merge into the real template when I put the arch-aware code in the master makefile Jul 31 15:34:43 basically, if you can get an openslug-native.conf and corresponding local.conf ready in the next 13 hours, then I can do the makefile stuff after that Jul 31 15:35:05 it would probably be easier just to sed it in the makefile, change the distro from openslug to openslug-native Jul 31 15:35:22 I think that is the only change in local.conf Jul 31 15:35:43 yes, if that is the only change then it will be simple. let me know if you confirm that :-) Jul 31 15:37:14 need to talk to jbowler-away aout how openslug-native interacts with freeze/unfreeze. does it share with openslug, or have it's own files. I'd prefer the former if possible Jul 31 15:37:35 But I'm not fussed if it requires the latter Jul 31 15:37:47 after a quick check DISTRO is the only thing that could matter in local.conf Jul 31 15:39:59 I dont really think openslug-native will get enough packages to need freeze/unfreeze, but.. maybe Jul 31 15:42:15 actually, that's exactly the reason *to* use freeze - to reduce the parse time for a small number of packages Jul 31 15:42:57 yes, but you could do it manually.. just add the required things to BBFILES Jul 31 15:45:19 yeah Jul 31 15:45:52 "NOTE: Couldn't find shared library provider for libc.so.6" <-- that could be a problem.. Jul 31 15:46:41 Im guessing the automatic dependancy on libc6 gets wrong on the generated packages.. have to check that.. Jul 31 15:50:21 hmm Jul 31 15:50:41 how do I check what target arch is used? Jul 31 15:54:58 rwhitby-away: initial openslug-native.conf commited, It works for me when just changing to that one in distro in local.conf Jul 31 16:02:10 ok, I'm off to work. back later Jul 31 16:02:53 oh right, Im back at work tomorrow, maybe should get some sleep.. Jul 31 16:04:01 dammit, native packages don't get the automatic depends on libc6 and things... oh well, I need something to do tomorrow night too Jul 31 16:04:41 bbl Jul 31 16:27:31 rwhitby-away: the frozen package/bbfile lists are in ${DISTRO}-package.conf and ${DISTRO}-bbfiles.conf, so if you change distro it changes automatically. Jul 31 16:33:17 * NAiL is doing three builds simultaneously on three hosts. Great fun... **** ENDING LOGGING AT Sun Jul 31 23:59:56 2005