**** BEGIN LOGGING AT Fri Aug 17 02:59:56 2007 Aug 17 07:16:30 rwhitby: unfortunately, that was 3 a.m. my time, I'll send you an e-mail Aug 17 08:19:04 vlabakje: here again Aug 17 08:50:32 hey Aug 17 08:50:49 I was wondering if you made any progress on the nslu2 front Aug 17 08:51:00 I've been trying with trunk, to no avail Aug 17 08:54:28 vlabakje: I got as far as trying svn trunk on my devslug, and it failed to boot. Aug 17 08:54:52 so I can confirm it's broken, but have had no time to look at it. There are really no significant changes since 7.07 anyway. Aug 17 08:55:09 (especially not in the kernel) Aug 17 08:55:19 it failed just after the kernel? (just before preinit?) Aug 17 08:55:32 yep Aug 17 08:55:50 hmmm, too bad Aug 17 08:55:56 did you see this line from last night? Aug 17 08:56:04 22:13 < thepeople_work> trunk is currently unstable as the buildroot has been reworked some Aug 17 08:56:07 22:14 < thepeople_work> use 7.07 Aug 17 08:56:13 yep, that's very true. Aug 17 08:56:36 that's why I'm not going out of my way to look at it until things settle down a bit. Aug 17 08:56:47 okay, no problems Aug 17 08:57:03 i'll just use a checkout from 7.07 svn to build on then Aug 17 08:57:07 yep Aug 17 08:57:16 i'm trying to implement nbd and cryptoloop on my slug Aug 17 08:57:37 anything you do on 7.07 should port forward without changes. Aug 17 08:57:54 great Aug 17 08:58:50 it's been quite nice to know I'm not the only one with this problem Aug 17 08:59:08 and I wanted to thank you for your pages on nslu2 modifications Aug 17 09:00:05 <[mbm]> what's wrong with trunk? Aug 17 09:00:37 <[mbm]> loaded it on a dsm g600 a few days ago; admittedly not an nslu2 but similar Aug 17 09:00:45 [mbm]: when compiled for the nslu2, it stops somewhere in preinit Aug 17 09:00:54 i tried putting debug prints in the script, to no avail Aug 17 09:01:30 the only output beyond the kernel i got was using init=/bin/blabla which would result in something like no such file Aug 17 09:01:52 <[mbm]> right, you need to go through preinit for proper bootup Aug 17 09:02:27 <[mbm]> do you have a log of one of your failed attempts? Aug 17 09:04:32 no sorry, I'm at work right now, logs are at home. I can give you one in a couple of hours. Or maybe rwhitby has one handy Aug 17 09:05:12 [mbm]: I haven't had a chance to look at it at all. Aug 17 09:05:41 <[mbm]> should just be some trivial misconfiguration Aug 17 09:07:27 could easily be - I'll try to get some time to look at it this weekend. Aug 17 09:08:02 <[mbm]> not qute sure I understand the purpose of cryptoloop in an embedded device with no user input Aug 17 09:08:16 <[mbm]> sort of implies that the key is stored on the device Aug 17 09:08:46 [mbm]: i want to sync my backups to a remote drive (across the internet) Aug 17 09:09:01 but I don't trust the machine the drive is in, so I want encrypted backups Aug 17 09:09:25 but I also want to be able to rsync them, because uploading a complete backup trough my adsl line would take 12 days Aug 17 09:09:25 <[mbm]> ah Aug 17 09:09:58 so, I'll use a network block device, put a cryptoloop ontop of it, format the device ext3, and rsync to it on my slug Aug 17 09:10:22 atleast, that's the idea ;) Aug 17 09:10:53 allready have a more-or-less working Makefile for the nbd userland utils, because they're not in openwrt yet Aug 17 09:11:56 * [mbm] just had the evil thought of nbd + raid Aug 17 09:12:31 <[mbm]> each block device on it's own host Aug 17 09:12:34 that actually works quite nicely, but better if you have enough bandwidth (100Mbit or more) Aug 17 09:13:03 <[mbm]> raid resync would be a bitch Aug 17 09:13:19 but then you're really getting more into the iscsi or esata domain Aug 17 09:13:27 <[mbm]> true Aug 17 09:13:37 it is actually called drbd :) Aug 17 09:14:16 http://www.drbd.org/ < niec :) Aug 17 09:14:17 nice Aug 17 09:24:41 actually, someone has previously done nbd and raid with some slugs. Aug 17 09:29:19 interesting, but i think nbd + filesystem + rsync will be a bit more flexible than raid Aug 17 09:36:24 * nbd + filesystem + rsync? Aug 17 09:38:38 nbd: yes indeed ;) does your nickname have anything to do with the equally-named kernel module/ Aug 17 09:46:08 <[mbm]> only problem I see is that you'll have to pre allocate space for the block device Aug 17 10:13:15 I'm trying to port a library that uses some md5-hashes? What should I use to replace those calls OpenSSL or straight to the linux cryptoapi? Aug 17 10:35:00 [mbm]: i've got an entire harddisk dedicated to it Aug 17 19:29:34 [mbm]: still here? Aug 17 19:29:51 <[mbm]> somewhat Aug 17 19:30:16 would you still like to see the log from the slug? Aug 17 19:30:23 <[mbm]> sure Aug 17 19:30:33 hmmmm Aug 17 19:30:45 just a moment, i've thrown it away, apparantly Aug 17 19:34:12 http://pastebin.com/d40c8d3a8 Aug 17 19:34:28 voila, it's the console log, would you like a .config to go with it? Aug 17 19:36:41 [mbm] ? Aug 17 19:38:08 <[mbm]> nope Aug 17 19:38:35 <[mbm]> the unable to open initial console message is normal Aug 17 19:38:56 <[mbm]> preinit should be creating an initial console and then printing '-preinit-' Aug 17 19:39:03 i know, but after this, it doesn't do anything Aug 17 19:39:13 <[mbm]> although it still may be booting without a console Aug 17 19:40:21 then it would react on the standard ip of 192.168.1.77 Aug 17 19:44:52 [mbm]: you what's even stranger, if I put init=/bin/blabla on the command line, I get this: Aug 17 19:44:58 Failed to execute /bin/blabla. Attempting defaults... Aug 17 19:45:25 but starting anything else just gets to the same message as before, it's as if the preinit script hangs or fails somewhere Aug 17 19:45:46 I allready tried editing the preinit script with some more debug echo's, but that didn't work either Aug 17 19:45:47 <[mbm]> well, you can compile a rootfs with a /dev/console Aug 17 19:46:02 how do I do that? Aug 17 19:47:15 <[mbm]> mknod console c 5 1 Aug 17 19:47:39 yeah ok, but where's the rootfs in the build tree? Aug 17 19:48:10 is it staging_dir/armeb ? Aug 17 19:48:54 <[mbm]> yeah, although I'm not sure that the scripts will let you alter that and rebuild .. you might be able to create a files/dev/console and have it copied into the rootfs Aug 17 19:49:26 <[mbm]> haven't had to do that trick lately and last time I recreated the rootfs manually Aug 17 19:51:07 what if I make it in target/linux/ixp4xx-2.6/base-files/default ? Aug 17 19:53:46 <[mbm]> yeah Aug 17 19:56:25 great, I'll know more in a few mins Aug 17 20:02:46 [mbm]: no difference :( Aug 17 20:02:57 should I report it as a ticket in trac? Aug 17 20:05:30 <[mbm]> look at the build_dir/armeb/root-ixp4xx and make sure that the /dev/console got copied Aug 17 20:06:00 <[mbm]> you won't get an 'unable to open initial console' when the filesystem contains a /dev/console Aug 17 20:06:44 hmmm, it didn't (get copied) Aug 17 20:09:15 <[mbm]> could be a permissions issue; dev entries usually require root permissions to be created Aug 17 20:09:28 created it as root indeed Aug 17 20:09:41 <[mbm]> cp counts as a creation Aug 17 20:09:44 but it should be able to copy it, or at least fail the build if it wasn't able to Aug 17 20:10:26 <[mbm]> did you take my advice and put it in files/dev/console at the top level, or did you put in in target/linux/ixp4xx/base-files/default ? Aug 17 20:10:49 <[mbm]> if you put it under base-files you'll have to do a package/base-files-clean first Aug 17 20:10:53 last one, there's no files/ in the root level Aug 17 20:11:15 <[mbm]> you can create a files directory, anything in it gets blindly dumped into the rootfs Aug 17 20:11:36 oh, I didn't know that Aug 17 20:13:09 will it work if i put it in build_dir/armeb/root-ixp4xx/dev and do a make target/linux-install ? Aug 17 20:13:17 or is that really dumb? Aug 17 20:14:00 <[mbm]> it won't Aug 17 20:14:23 <[mbm]> it ignores the root-ixp4xx directory since it's generally deleted when the package selection changes Aug 17 20:15:31 okay, I stubbornly did it anyway (it was allmost done) and now I didn't get the warning, doesn't do anything else either though Aug 17 20:15:38 I'll try the files thing now Aug 17 20:17:36 strange thing is, even a init=/bin/sh doesn't show anything now Aug 17 20:29:21 i'm giving up Aug 17 20:29:34 it doesn't give anything beyond the kernel Aug 17 20:29:52 even with a /dev/console in there, and even giving different init= settings Aug 17 20:30:13 do you have any other ideas [mbm] ? Aug 17 20:32:05 <[mbm]> huh? how can you still be getting the 'unable to open initial console' error? Aug 17 20:32:11 no no Aug 17 20:32:13 that's gone Aug 17 20:32:17 but other than that Aug 17 20:32:21 i'm getting nothing Aug 17 20:32:28 VFS: Mounted root (squashfs filesystem) readonly.Freeing init memory: 76K Aug 17 20:32:45 that's the last 2 lines (somehow the linebreak doesn't get copied over here) Aug 17 20:36:33 <[mbm]> should be: Aug 17 20:36:34 <[mbm]> VFS: Mounted root (squashfs filesystem) readonly. Aug 17 20:36:34 <[mbm]> Freeing init memory: 76K Aug 17 20:36:34 <[mbm]> Warning: unable to open an initial console. Aug 17 20:36:34 <[mbm]> - preinit - Aug 17 20:37:14 i know.. but it doesn't Aug 17 20:37:37 anyway, it's getting late here so I'm giving up for today Aug 17 20:37:52 I'll put some more effort into it tomorrow Aug 17 20:38:08 let me know if you want me to put this into a trac ticket Aug 17 20:38:15 and thanks for your help :) Aug 17 20:39:22 <[mbm]> yeah might aswell open a ticket since I probably won't have time to look into it myself Aug 17 22:17:47 rsc in #openwrt just experienced a mini_fo related oops when trying to run ipkg remove in case anyone is interested. Aug 17 22:18:27 <[mbm]> saw it; null pointer deref Aug 17 22:18:35 <[mbm]> should be filed as ticket Aug 17 22:18:40 [mbm]: related to the brcm47xx-2.6 random segfaulting issue? Aug 17 22:19:01 [mbm]: I mainly asked because I wasn't sure what info you would want him to post in the ticket. Aug 17 22:19:02 <[mbm]> could be memory corruption Aug 17 22:19:37 and because there was a chance that a dev might get more insight into the problem if they can poke around before rebooting. Aug 17 22:20:10 <[mbm]> well, the oops message and information about which kernel it was from so the memory addresses can be matched (in other words did he compile openwrt, or is it one of the standard images) Aug 17 22:25:41 <[mbm]> only see two places in the mini_fo code which could trigger an oops like that Aug 17 23:28:36 hello lucaferr Aug 17 23:28:43 welcome to #openwrt-devel! Aug 17 23:29:59 hi! Aug 17 23:30:03 sup Aug 17 23:31:47 hanging out with some friends **** ENDING LOGGING AT Sat Aug 18 02:59:56 2007