**** BEGIN LOGGING AT Fri Aug 05 23:59:56 2005 Aug 06 01:06:57 dyoung ? Aug 06 01:27:16 rwhitby-away: peteru is at Latitude: -33.83, Longitude: 151.19 Aug 06 01:28:00 * peteru-home is off to a birthday celebration... Aug 06 02:30:01 morning Aug 06 02:30:17 congrats on the time fixing commit Aug 06 02:30:55 perhaps it should be submitted upstream (given that it is basically an incorrect assumption/value for all IXP425 boards) Aug 06 02:39:37 blaster8: what as wrong with the time? Aug 06 02:39:40 was Aug 06 02:51:27 the fact that it was 1% out, roughly Aug 06 02:51:39 http://www.nslu2-linux.org/wiki/HowTo/SetUpAnNtpServer Aug 06 02:52:12 the first problem hasn't applied for ages but the second problem has only just been fixed Aug 06 02:52:53 Apparently, Linksys hacked the default ticks value, but hacked it the wrong way - doubling the problem rather than fix it Aug 06 02:53:18 (they hacked the ticks value to 9903, rather than putting it to 10101(.010101) Aug 06 02:53:35 probably because they misunderstood what the value actually was Aug 06 02:54:06 Openslug has a 'fresh' kernel which defaults the ticks to 10000 - which is unfortunately still (but less) wrong Aug 06 02:54:18 now jbowler fixed the underlying issue Aug 06 02:57:43 but the kernel source is basically almost unreadable and practically uncommented Aug 06 02:57:55 see this discussion: http://loglibrary.com/show_page/view/177?Multiplier=3600&Interval=6&StartTime=1123260816 Aug 06 03:56:48 peteru-home: coords added Aug 06 03:57:06 morning, rwhitby Aug 06 03:57:20 g;day Aug 06 03:57:32 http://www.apache.de/dist/spamassassin/source/Mail-SpamAssassin-3.0.2.tar.bz2 <- 404 (optware/nslu2) Aug 06 03:58:17 try with 3.0.3 or 3.0.4? :) Aug 06 04:03:40 03rwhitby * 10unslung/make/perl-spamassassin.mk: Updated to latest upstream version (previous version source has disappeared). Aug 06 04:05:32 I'm just bringing unslung native builds into the autobuild fold ... Aug 06 04:05:45 nice Aug 06 04:05:59 I'm starting to hack udev again. It's about time ;) Aug 06 04:07:46 03rwhitby * 10unslung/make/perl-storable.mk: Updated to latest upstream version (previous version source has disappeared). Aug 06 04:21:25 How should I configure debian to look for packages suitable for compiling openslug> Aug 06 04:21:29 i.e. what repository? Aug 06 04:22:38 Debian Etch, then use make setup-host-debian Aug 06 04:23:06 can I swap a debian stable installation over to grab packages from the testing branch? Aug 06 04:25:24 probably. edit source.list somehow I think Aug 06 04:25:39 (or something like that). I'm sure Google would know ... Aug 06 04:25:50 checking google without much success Aug 06 04:26:14 debian installation doco then? Aug 06 04:27:19 which are not exactly friendly (one of the reasons I like gentoo - but I don't currently have the time or the proc power) Aug 06 04:27:25 I think I'm getting there, though Aug 06 04:28:08 aha Aug 06 04:29:24 it's easier than I thought (just change stable to testing in sources.list) Aug 06 04:30:43 03nail 07org.openembedded.nslu2-linux * r3a06ef82... 10/conf/distro/openslug-packages.conf: Add udev Aug 06 04:31:04 NAiL: udev works? Aug 06 04:31:26 no Aug 06 04:31:33 :) Aug 06 04:31:42 If it did, I'd add it to openslug-image.bb ;) Aug 06 04:31:55 any idea what is wrong? Aug 06 04:32:30 no, can't remember. Aug 06 04:32:38 :) Aug 06 04:32:41 It builds though, so there's no problem in getting it in the firmware image Aug 06 04:32:50 but from there on to actually it working Aug 06 04:32:57 tricky Aug 06 04:33:15 Well, it'll require quite a few changes to several systems Aug 06 04:33:23 turnup is one Aug 06 04:34:00 When I(we) get udev working, someone can get leds working properly too Aug 06 04:35:24 ahahaha Aug 06 04:35:37 sometimes, public service ads from google make me laugh Aug 06 04:36:11 gotta love linux - under what was wrong with the devfs system: the devfs maintainer/author disappeared and stoped maintaining the code. Aug 06 04:36:29 "SECURITY Encrypting Root FileSystems With DM-Crypt". And directly below: "Help Brain Tumor Patients" Aug 06 04:37:31 ok, now I've got a firmware image with udev in it Aug 06 04:37:38 now, if I only had a disk! Aug 06 04:38:31 NAiL: how much (in USD$) would it take for you to be able to go out now and buy a new disk? Aug 06 04:39:20 maybe we can make a contribution ... Aug 06 04:39:23 for economy, USB2 2.5" Cases are VERY cheap from eBay as are small 2.5" drives Aug 06 04:39:29 but that takes time, unfortunately Aug 06 04:41:54 rwhitby: I have a small disk that'll do, if I get an enclosure Aug 06 04:42:24 mine cost about £5.5 in the UK (aluminium, decently made IBM job) Aug 06 04:47:44 another bug in the makefile - apparently a default debian install doesn't have sudo either Aug 06 04:48:13 add su - -c apt-get install sudo ;-) Aug 06 04:48:21 fine by me Aug 06 04:48:27 (what I've just done) Aug 06 04:49:47 except of course that the sudoers file has to be properly configured too Aug 06 04:55:23 hmm, time to switch to debian unstable Aug 06 04:55:37 debian testing (etch) only has monotone 0.19 Aug 06 04:56:01 oops, sorry, wrong architecture Aug 06 04:56:11 you need to install 0.21 from venge.net Aug 06 04:57:16 then that setup-host-debian script really needs fixing Aug 06 04:57:57 patches welcome ... Aug 06 04:59:38 Ribbit Aug 06 04:59:55 i bought a used slug off ebay, and it turns out it doesn't recognize hard drives (it could have broken during shipping; seller asserts it was working with Maxtor One Touch drive)... Aug 06 05:00:00 leap :) Aug 06 05:01:08 Bummers. Aug 06 05:01:25 but, i was wondering... i upgraded the firmware from R21->R63. dunno if there might have been a problem with that. but doesn't matter, i went out and bought a brand new slug last night, so now good to go. :) Aug 06 05:01:27 geojeff: which hard drive does it not recognise? Aug 06 05:01:29 But on the plus side, there are a couple people attempting to make audio input a reality. Aug 06 05:01:46 geojeff: do you still have the other one? Aug 06 05:02:07 if so, flash unslung 5.5 on it and try the disk again Aug 06 05:02:14 bbiab Aug 06 05:02:16 yes, still have other one. Aug 06 05:03:10 no disks were recognized, not even a flash drive. so got nothing on it. Aug 06 05:04:03 dyoung, i bought a Soundblaster 24bit external drive. what is happening regarding audio dev? Aug 06 05:04:20 (just to get it to work, i suppose) Aug 06 05:04:29 working on getting the bits together for input. Aug 06 05:04:43 neato Aug 06 05:05:12 someone has worked out output already, its someplace on the wiki. Aug 06 05:05:24 but output is less interesting than input. Aug 06 05:05:25 yeah, saw that Aug 06 05:05:31 :-) Aug 06 05:05:35 yep Aug 06 05:05:45 rwhitby: easy to fix setup-host-debian --> just run apt-get -t 0.21* install monotone Aug 06 05:06:01 oops Aug 06 05:06:12 apparently it just ignored that flag - nice Aug 06 05:06:33 odoc, pong Aug 06 05:07:19 blaster8, debian instead of gentoo? (i only ask because i arbitrarily took the gentoo plunge earlier in the summer on my laptop) Aug 06 05:07:45 hey dyoung Aug 06 05:07:51 got serial on nslu Aug 06 05:08:24 got kernel oops Aug 06 05:08:27 was it any help in finding whats causing your instability? Aug 06 05:08:29 after klicking in webfrontend Aug 06 05:08:31 hmm Aug 06 05:08:42 http://pastebin.com/330521 Aug 06 05:09:09 geojeff: I am running this in an emulated virtual machine Aug 06 05:09:09 was latest unslung-5.5 Aug 06 05:09:23 odoc, you tried reflashing for good measure? Aug 06 05:09:28 yes Aug 06 05:09:36 first flash was with windows soft Aug 06 05:09:41 blaster8, ok, so a specific type of need. Aug 06 05:09:41 second with upslug Aug 06 05:09:51 geojeff: talk about it Aug 06 05:10:12 resoldered r83 Aug 06 05:10:18 cleaning up a hdd to play a bit Aug 06 05:10:25 blaster8, just curious, as i like gentoo Aug 06 05:10:59 odoc, did putting it back affect the stablity? Aug 06 05:11:09 (but would probably like debian just as well) Aug 06 05:11:18 geojeff: so do I - but running gentoo while compiling 7MB bits of source leads to a tendency to destory itself Aug 06 05:11:21 and if not, did you use a SMT or something like that? Aug 06 05:11:26 have to wait to get the hdd clean Aug 06 05:11:39 smt ? Aug 06 05:11:51 Surface Mount part. Aug 06 05:12:13 i used the original one Aug 06 05:12:29 okay cool. Aug 06 05:12:43 blaster8, ah Aug 06 05:12:48 and i didnt mess around with soldering ... everything was fine ;P Aug 06 05:13:03 tripple checked that first :) Aug 06 05:17:42 blaster8: if you n eed to refer back to that talk about clock timing and 33.000 versus 33.333 time bases; http://loglibrary.com/dialogue/view/85 Aug 06 05:19:31 thanks dyoung Aug 06 05:19:46 I really think we should shout at somekernel maintainers about that one Aug 06 05:20:34 Or linksys for not following the spec? Aug 06 05:21:06 true Aug 06 05:22:32 I think that both are wrong, actually (but linksys hardware was more wrong) Aug 06 05:22:40 perhaps we should submit the changes to them Aug 06 05:35:41 still looking for a ipkg mirror ? Aug 06 05:35:52 yes Aug 06 05:36:54 odoc, did resoldering R83 solve your instability problem? Aug 06 05:39:02 odoc: Do you have 5Gb of space, 50Gb/month of bandwidth, and at least 2Mb/sec upload pipe? Aug 06 05:39:23 ive two sites Aug 06 05:39:26 both 100 mbit Aug 06 05:39:29 both legal ;P Aug 06 05:39:39 one site will be down in around 3 month Aug 06 05:39:56 but is used as dev-box till than ... and has more than enough free traffic Aug 06 05:40:01 (around 300 gb) Aug 06 05:40:17 the other one will rest longer ... but ive to check things Aug 06 05:40:26 ahh Aug 06 05:40:30 both located in germany Aug 06 05:40:56 another european mirror :-) Aug 06 05:41:11 hehe Aug 06 05:41:15 world domination tour :) Aug 06 05:41:22 dyoung: we should mark the mirror machines on http://www.nslu2-linux.org/developers.php Aug 06 05:42:15 wow Aug 06 05:42:16 nice map Aug 06 05:42:23 rwhitby, sorry to bother - the other day you gave me a convining reason to try OpenSlug over Debian. What was that again? Aug 06 05:42:32 convincing Aug 06 05:42:38 cool Aug 06 05:42:44 noone from Holland yet Aug 06 05:42:51 wspraul: ethernet works in openslug Aug 06 05:43:00 hawaii Aug 06 05:43:03 internal ethernet that is Aug 06 05:43:39 wspraul: I don't have a convincing argument not to use DebianSlug Aug 06 05:44:28 Does OpenSlug or DebianSlug swith the NSLU2 CPU to LE now early on? Aug 06 05:45:40 s/LE/Little Endian/ Aug 06 05:51:01 wspraul: debianslug does Aug 06 05:51:59 kolla, does it help? (in terms of portability, bugs) Does OpenSlug plan to do the same thing (I read the comments in the WishList) Aug 06 05:52:29 no idea Aug 06 05:52:50 big-endian in itself is not a problem Aug 06 05:52:51 OpenSlug is Big Endian, and will remain so. Aug 06 05:53:16 the problems is the assumption that some software make.. that arm is litle-endian Aug 06 05:53:39 (unless someone gets the internal ethernet to work, and then we can reconsider the question) Aug 06 05:55:17 thos who develop that software is better to be told that arm can also be big-endian :) Aug 06 05:56:21 kolla: some software doesn't take into the account even the possiblity of bigendian. They just didn't think of it. Aug 06 05:57:01 NAiL: right Aug 06 05:58:00 woohoo Aug 06 05:58:00 repvik@maimedslug:~$ ping 10.0.0.4 Aug 06 05:58:00 ping: permission denied. (are you root?) Aug 06 05:58:08 I can't even *ping* as a user? Aug 06 05:58:39 secureslug :) Aug 06 06:00:48 alias ping="sudo /bin/ping" :) Aug 06 06:01:52 hehe, yes Aug 06 06:03:04 what on earth is going on with my serial NOW? Aug 06 06:03:05 FLASH: 0|50000000 - 0|50800000, 64 +Ä+_ Ä° 0|00020000 <+_ #. Aug 06 06:03:13 parts get through ok Aug 06 06:04:05 ah, now it got it back Aug 06 06:08:14 rwhitby ownz NAiLs slug - part of the plans for world domination from down under ;-) Aug 06 06:08:53 iputils wont build here, hm Aug 06 06:23:54 peteru-home: what makes you think it's just NAiL's ? Aug 06 06:27:53 hehe Aug 06 06:33:07 one slug at a time, I suppose - sounds like kolla was next :-O Aug 06 06:33:41 kolla knows too much ;) Aug 06 06:33:41 folks: how do I best power-off my slug? does halt -p work? Aug 06 06:33:45 * kolla puts out a decoy... Aug 06 06:33:48 Norway doesn't have an extradition agreement with Australia - so he was the obvious next candidate .... Aug 06 06:34:52 so, did you see my kinky issue.net yet? :) Aug 06 06:35:16 or is this best done via thw webinterface? Aug 06 06:37:55 why do you want to turn it off? :) Aug 06 06:38:02 poor thing.. Aug 06 06:38:11 because i'm leaving for a fortnight Aug 06 06:38:25 it's not ready yet to substitute my server alas Aug 06 06:38:50 and i want it to be in good shape again when I reboot Aug 06 06:38:52 Anyone know what the easy way to view a PNG file from within X is? Aug 06 06:39:12 dyoung: display? Aug 06 06:39:28 Cool, thanks! Aug 06 06:39:28 almost any program will do :) Aug 06 06:39:43 I dont know many. Aug 06 06:40:00 a GUI is a way to have many console windows at the same time. :-) Aug 06 06:41:13 heh Aug 06 06:41:15 xv xnview display Aug 06 06:41:26 firefox, ie, safari :-) Aug 06 06:41:45 dyoung: you missed out the "just" Aug 06 06:42:04 just d Aug 06 06:42:26 heh Aug 06 06:43:26 'k bye Aug 06 06:48:28 03rwhitby * 10unslung/make/python.mk: Added a /usr/bin/python symlink (just like /usr/bin/perl), mainly so that ipkg-make-index will work for the native builds. Aug 06 06:49:30 I'm determined to get Unslung native autobuild working tonight ... Aug 06 06:50:38 03rwhitby 07org.nslu2-linux.dev * rbeab5448... 10/Makefile: Inserted conditional clauses for native compilation. Aug 06 06:56:10 dyoung: gqview is nice Aug 06 06:56:22 if you have GNOME installed Aug 06 06:58:28 optware gtk is not building right. I noticed it has many instances of include directives pointing to the host FS (like -I/opt/include/glib-2.0) Aug 06 06:58:52 anyone maintaining that package, or at least familiar with it? Aug 06 07:00:22 looks like Christopher (bluck2) then giel, but nothing serious since January Aug 06 07:03:26 * peteru-home wonders why the hell something like the slug needs to build gtk of all things. Aug 06 07:04:06 peteru thanks for the tip. I'll look at that tommrow. I'm satisifed that this thing is doing what its usp posed to. Aug 06 07:05:13 peteru-home: apparently, for atk, bitlbee, ctrlproxy, giftcurs, gkrellm, irssi, pango, vte and xchat Aug 06 07:07:26 * peteru-home shrugs and wonders how long before someone will try to run Open Office on the slug. Aug 06 07:09:45 Great idea! ;) Aug 06 07:10:50 peteru-home: well, trying to run a desktop system these days without loading either KDE or GNOME is getting extremely difficult .. Aug 06 07:11:05 * rwhitby remembers the days when using Xaw was considered cutting edge .. Aug 06 07:23:31 is there a way to integrate busybox1 into the unslung firmware ? Aug 06 07:23:51 no Aug 06 07:24:09 :-( Aug 06 07:24:18 the linksys firmware has modifications to the built-in busybox for which there is no source released. Aug 06 07:24:27 ah k Aug 06 07:35:04 03rwhitby * 10unslung/make/unslung-devel.mk: Added python. Aug 06 07:42:21 woohoo, python :) Aug 06 07:42:33 Woohoo! make autobuild now works on gastro. Unslung native feed is being updated now ... Aug 06 07:43:07 geojeff: it was already there, i just added it as a native build dependency (it's needed for ipkg-make-index) Aug 06 07:43:30 ah, well woohoo anyway. or double. :) Aug 06 07:46:52 native compilation on the unslung slug should now be as easy as: 1) Install Unslung, with a big disk, 2) ipkg install unslung-devel, 3) mkdir /home/slug ; cd /home/slug ; wget http://www.nslu2-linux.org/Makefile ; make Aug 06 07:48:14 Now to just get openslug native to the same level of automation .... Aug 06 07:51:39 ... but that's a job for another day ... Aug 06 07:54:41 which package should I install to get the unix 'top' command? Aug 06 07:55:17 ipkgfind.nslu2-linux.org Aug 06 07:55:53 and #nslu2-general :-) Aug 06 07:56:26 oh ok, sorry ;-) Aug 06 08:06:47 <[cc]smart> how is openslug supposed to be set up wrt dnsdomainname. do we use /etc/dnsdomainname ? Aug 06 08:50:25 rwhitby: how's that (modifying and "shipping" busybox without releasing sources) in terms of GPL2? Aug 06 08:56:00 seems like libstdc++ is missing from the feeds (http://ipkg.nslu2-linux.org/feeds/unslung/unstable/libstdc++_5.0.7-3_armeb.ipk) Aug 06 09:10:56 03ingeba * 10unslung/sources/ncftp/configure2.patch: Patch to make configure use 64 bit integers for dealing with file size Aug 06 09:11:55 03ingeba * 10unslung/make/ncftp.mk: Added large file support and upped IPKG number Aug 06 09:21:26 ingeba-zzz: can you add 2.5, 2.6 and 2.7 as valid OpenSlug firmware version numbers? Aug 06 09:23:01 jbowler-zzz: Yes - can do. Aug 06 09:23:59 Many thanks Aug 06 09:24:16 I can under NAiL's udev bug ;-) Aug 06 09:25:12 jbowler-zzz: Done. Can I retire 1.12 as active version for filing of bugs? Aug 06 09:26:29 Yes Aug 06 09:26:41 Anyone who files against 1.12 will get laughed at. Aug 06 09:27:59 jbowler-zzz: Good attitude. Obsoletion by intimidation :-) Aug 06 09:32:43 you guys work very well in your sleep :) Aug 06 10:00:16 geojeff: It is hard to believe, but most creative leaps in the history of man kind were conceived in sleep and motivated from laziness. Aug 06 10:01:22 <[cc]smart> i don't find this hard to comprehend Aug 06 10:01:25 yes, i've heard that. Aug 06 10:01:48 <[cc]smart> i think this fact is purely natural Aug 06 10:02:25 <[cc]smart> and suits me very well :) Aug 06 10:35:00 just saw this: rwhitby the linksys firmware has modifications to the built-in busybox for which there is no source released. Aug 06 10:35:21 well - busybox is GPL - http://www.busybox.net/license.html Aug 06 10:45:54 <[cc]smart> do we have a tool in openslug with which changing nvram configuration settings is possible ? Aug 06 10:49:49 blaster8: Yes - and so? Aug 06 10:50:38 03ccsmart 07org.openembedded.nslu2-linux * r27caeec0... 10/packages/postfix/ (files/main.cf_2.0 postfix_2.0.20.bb): Polishing maildir based install defaults. Aug 06 10:50:51 well, they should release the changed source code (if it was BSD licensed, for example, they wouldn't have to) Aug 06 10:50:59 blaster8: I've asked the BB people - they didn't give a shit (didn't even bother answer). Asked Linksys - they did't give a shit. Aug 06 10:51:15 ingeba-zzz: How nice Aug 06 10:51:29 well done for trying anyway :) Aug 06 10:51:31 blaster8: The point is: In theory (and legally) they should, but they don't and there is no way to make them short of threatening Aug 06 10:51:50 and if _I_ do they will call my bluff :-( Aug 06 10:52:38 Talk to these ppl: http://gpl-violations.org/ Aug 06 10:53:24 Yes - I saw that a little while back - this issue with BB was a long time ago. Without even the Busybox people behind us, we'll get nowhere. They supposedly have good relations with Linksys. Aug 06 10:54:46 ridiculous Aug 06 10:55:08 I don't care whether busybox like Linksys or not - it's us who need and have a right to use the code Aug 06 10:56:13 blaster8: Yes. Feel free. The proof is basically strings from the binary. The suspected mods are minor - calling some external stuff to halt the system and so on. It is still hard to tell how much they have changed. Maybe even also the insmod applet may have been changed. Aug 06 10:57:51 technically they need to provide the source even if they haven't changed it Aug 06 10:58:02 at least, on request! Aug 06 11:01:17 Hmm - I'll try and get something going tonight if I have time Aug 06 11:10:36 03ccsmart 07org.openembedded.nslu2-linux * r68aa7712... 10/packages/cyrus-imapd/ (cyrus-imapd_2.2.12.bb files/cyrus.conf_2.2): Fixed paths for lmtp and notify. Aug 06 11:13:34 hmm - to any devs - could we create a 'trivial' http debian repository with our desired version of monotone (and any other tools) in it? - All we would need is some space (~18MB) of space on an 'official' web server Aug 06 11:14:07 it's just it is hard to 'temporarily use unstable repositories for monotone then go back to testing for everything else with apt-get Aug 06 11:14:16 (and I though gentoo was hard) Aug 06 11:46:22 blaster8: hard? what's hard about apt-get install -t unstable monotone? Aug 06 12:05:32 well, the fact that you need an unstable reposity in sources.list Aug 06 12:05:51 doesn't apt-get automatically install the newest version of everything? Aug 06 12:06:45 so you add an unstable repository, run apt-get update install monotone, remove the unstable repository, run apt-get update again and continue Aug 06 12:06:56 not exactly an elegant solution Aug 06 12:08:14 also just using -t unstable means that if monotone release 0.22 which is incompatible with 0.21 servers and NSLU2 don't upgrade, anyone running setup-host-debian after that are a bit stuffed Aug 06 12:10:43 whereas if we had a repository of NSLU2's preferred monotone version, we could easily keep it in sync with the server's version Aug 06 12:29:53 blaster8: yes, if you don't use pinning Aug 06 12:31:30 true, but then we would have to undo the pin when NSLU2 wanted the upgrade to happen (ugly, ugly, ugly) Aug 06 12:31:56 I have to admit, gentoo does a much better job of this currently... Aug 06 12:35:27 obviously if monotone didn't keep breaking everything between versions this wouldn't be a problem, but it isn't exactly mature software, so what can you expect Aug 06 12:48:22 BTW, I have an interim cleanup of the setup-host script in the makefile if anyone with commit privs is around Aug 06 12:48:41 blaster8: shoot Aug 06 12:48:48 * NAiL *just* came back Aug 06 12:50:06 NAiL: can you have a look at my minor rant above re: debian and monotone Aug 06 12:50:36 blaster8: in a little while, trying to catch up on several fronts here ;) Aug 06 12:52:23 cool - it's not a priority but I'm willing to set it up as I have a couple of days spare here Aug 06 12:56:17 anyway - about the patch http://people.pwf.cam.ac.uk/mlj28/makefilechanges Aug 06 12:57:25 sounds like a good idea to me Aug 06 12:58:24 the only thing I would like is an easy way of telling what architecture debian is running on, from a script Aug 06 12:59:03 that's not hard Aug 06 12:59:48 how do I do it (it's for whether to grab the i386 monotone binary or compile it from source if on another architecture, in case you're wondering) Aug 06 13:00:37 look at the master Makefile for a start ;) Aug 06 13:00:56 aah, got it Aug 06 13:01:08 I'll work it out from there Aug 06 13:05:49 On #8 in the unslung README, it says that after doing #7 (reboot), telnet in and make sure the root filesystem was copied. Perhaps it is too much hand-holding, but you must revist 192.168.1.77/Management/telnet.cgi to reenable telnet again. i guess ppl can figure that out, as they had just done it a few steps back. Aug 06 13:06:43 (it doesn't say to enable telnet again, in order to telnet in) Aug 06 13:10:12 that statement on #5 to change your password in /share/hdd/conf/passwd should be taken out of the NOTE: bullet and put as an explicit item to do. i missed it :( Aug 06 13:11:30 geojeff: change it :) Aug 06 13:11:50 ~nslu2-rules Aug 06 13:11:51 from memory, nslu2-rules is 1) You must search the wiki before asking questions on IRC or mailing list. 2) You must follow the README steps precisely when flashing firmware. 3) Those who ask the questions, update the wiki with the answers. 4) Those who complain about doco, update the wiki to improve it. 5) Friends don't let friends flash custom firmware without ... Aug 06 13:12:00 4) :) Aug 06 13:12:26 k Aug 06 13:12:35 On that point I have been changing http://www.nslu2-linux.org/wiki/HowTo/UseTheMasterMakefile Aug 06 13:13:29 well, doesn't the README need to be changed in several places, web and in svn? Aug 06 13:14:04 (and i'm not there yet -- to svn) Aug 06 13:15:14 not sure, this is unslung, isn't it? Aug 06 13:15:49 yes, unslung Aug 06 13:25:08 Unslung-5.x Family Release README INSTALLATION Aug 06 13:25:08 You must follow the steps in this README *exactly*. Do not skip any steps, and Aug 06 13:25:12 do not skip ahead thinking you know what you are doing and don't need to follow Aug 06 13:25:14 That one? Aug 06 13:25:27 yes Aug 06 13:29:49 what's the url to the readme on the wiki? Aug 06 13:30:57 actually, i was reading it on sf: http://puzzle.dl.sourceforge.net/sourceforge/nslu/Unslung-5.5-beta-README.txt Aug 06 13:31:13 that's where the wiki link takes you Aug 06 13:31:47 I guess you should bug some unslung devs about this a bit... Aug 06 13:31:51 ingeba-zzz: ping? Aug 06 13:32:06 NAiL, ok. thanks. Aug 06 13:33:16 geojeff, http://www.nslu2-linux.org/wiki/Unslung/SuggestedChangesToTheReadme Aug 06 13:35:57 caplink811_log, aha, someone's been there and done that :) Aug 06 14:35:51 NAiL - when you come back can you have a look at this Aug 06 14:35:51 http://people.pwf.cam.ac.uk/mlj28/makefilechanges Aug 06 14:36:35 urgh Aug 06 14:36:37 guess I can now Aug 06 14:36:38 ? Aug 06 14:36:44 not that bad is it Aug 06 14:36:53 I was going to sleep, but I remembered there was stuff I had to do first ;) Aug 06 14:37:18 fair enough, it is an hour later where you are Aug 06 14:38:21 one thing I would like to add is a grep to check if 'testing' appears in the sources.list file and have a prompt if it doesn't Aug 06 14:46:34 looks ok Aug 06 14:47:26 how about this time (if you are still around) Aug 06 14:47:34 just reload Aug 06 14:47:58 give me a sec Aug 06 14:48:04 no probs Aug 06 14:50:41 03nail 07org.openembedded.nslu2-linux * rb8873177... 10/ (3 files in 3 dirs): Added watchdog package Aug 06 14:50:44 03nail 07org.openembedded.nslu2-linux * raa7c47af... 10/ (3 files in 3 dirs): disapproval of revision b887317745e76c913d221d462248b9555eaf7c4d Aug 06 14:55:53 03nail 07org.openembedded.nslu2-linux * rae3124f8... 10/ (3 files in 3 dirs): disapproval of revision aa7c47af4a1d20ac50212af0be57a0ab4b597184 Aug 06 14:55:56 03nail 07org.openembedded.nslu2-linux * rcaf4fea9... 10/ (3 files in 3 dirs): disapproval of revision ae3124f810e521f4a710f11319d143e188e3fee4 Aug 06 14:56:01 03nail 07org.openembedded.nslu2-linux * r233263c0... 10/packages/watchdog/watchdog_5.2.bb: Added watchdog package. Without adding anything else this time Aug 06 14:56:03 03nail 07org.openembedded.nslu2-linux * r30035fbe... 10/ (2 files in 2 dirs): Add watchdog to openslug Aug 06 14:56:34 WTF? Aug 06 14:56:40 ? Aug 06 14:56:55 What are the two other disapprovals? Aug 06 14:57:49 you seem to have disapproved the disapproval of the disapproval Aug 06 14:57:51 impressive Aug 06 14:58:16 hahaha Aug 06 15:00:52 03nail 07org.openembedded.nslu2-linux * r69894987... 10/ (22 files in 17 dirs): Aug 06 15:00:53 propagate from branch 'org.openembedded.dev' (head 5180c5e9083e9a3074c95abbc09e11cc69b721a9) Aug 06 15:00:53 to branch 'org.openembedded.nslu2-linux' (head 30035fbe7ead3cb3777531f88b76f38232f617eb) Aug 06 15:01:07 oh well Aug 06 15:01:15 the end result was adding watchdog anyhow ;) Aug 06 15:01:35 exactly Aug 06 15:04:50 anyway, I have a sample debian repository (with i386 binary and source of monotone) set up and an updated (but not fully tested) setup-host-debian script up at http://people.pwf.cam.ac.uk/mlj28/makefilechanges Aug 06 15:06:33 Yeah, I don't think there'll be any changes until tomorrow Aug 06 15:06:46 want to check with the rest of the core devs Aug 06 15:07:00 * NAiL tries out watchdog Aug 06 15:07:02 I'm not in a hurry Aug 06 15:07:48 --if-- you have time can you check the script for sanity, I'm basically new to this sort of thing, but I need to learn somehow Aug 06 15:08:27 I didn't realise that board had a watchdog sensor Aug 06 15:08:30 it *looks* sane to me. But what happens if the sources have already been added to sources.list? ;) Aug 06 15:08:38 aha Aug 06 15:08:42 you've updated again Aug 06 15:08:45 damn, should have set up that grep thing Aug 06 15:08:48 hold on Aug 06 15:11:41 should fix it Aug 06 15:11:51 argh Aug 06 15:12:15 better Aug 06 15:14:13 looks ok by me Aug 06 15:14:49 does anyone know a way I can lock up the slug hard? :) Aug 06 15:15:19 forkbomb Aug 06 15:15:34 hmm Aug 06 15:16:04 That'll take too long :P Aug 06 15:16:27 besides, the watchdog only needs a little time during a minute not to reset the slug Aug 06 15:16:36 true Aug 06 15:16:46 So that might take forever ;) Aug 06 15:18:17 try echo "1" > /dev/watchdog Aug 06 15:19:08 NAiL: did you do something to allow the watchdog to be set from user land? Aug 06 15:19:59 Because there is an XScale firmware problem which means it doesn't work. Aug 06 15:20:07 jbowler-zzz: mknod /dev/watchdog c 10 130 Aug 06 15:20:10 oh? heh Aug 06 15:20:10 s/firmware/hardware/ Aug 06 15:20:23 *sigh* Aug 06 15:20:24 That's why shutdown -r didn't work. Aug 06 15:20:53 The flash driver has to put the flash chip into a non-write state or the reboot will hang. Aug 06 15:20:57 blaster8: the slug shut down 1 min after I did that Aug 06 15:21:10 jbowler-zzz: and the reboot worked ;) Aug 06 15:21:24 technically, that should have started the timer - hold on Aug 06 15:21:30 Yes - well, then you know that the flash wasn't in a write state when the watchdog hit. Aug 06 15:22:04 jbowler-zzz: define "write state". You mean, *writing* to the flash? Aug 06 15:22:17 blaster8: yes, it started the timer Aug 06 15:22:34 No, I mean the registers used to access the flash have a 'write mode' and this can be set by the kernel. Aug 06 15:22:49 NAiL: so, at least part of it is working Aug 06 15:23:19 Remember I did too fixes - one was to remount /initrd 'ro' in the bootstrap, the other to add a call to the shutdown code to the flash driver. Aug 06 15:23:53 Either is sufficient - remount -o ro on /dev/mtdblock4 leaves it without write mode set so the watchdog will work. Aug 06 15:24:09 I'm running from flash only Aug 06 15:24:15 However if you do a reflash -k then the system will hang (IIRC) without the shutdown fix. Aug 06 15:24:37 uh Aug 06 15:24:38 no I'm not Aug 06 15:24:45 I'm running from disk on that slug Aug 06 15:25:03 Running from flash also seems to be safe - but that's probably because the kernel does mount -o ro / Aug 06 15:25:10 (on reboot). Aug 06 15:25:28 /initrd is mounted ro Aug 06 15:25:35 that happens automagically, doesn' it Aug 06 15:26:03 yes, you said so Aug 06 15:26:04 actually Aug 06 15:26:15 /etc/rcS.d/umountinitrd.sh Aug 06 15:26:47 well, so why shouldn't it work then? Aug 06 15:27:32 reflash -k, anything which writes to the flash, if the user rmdirs /initrd... Aug 06 15:28:05 if the user runs from flash probably (since the watchdog doesn't do a shutdown) Aug 06 15:28:13 Is the kernel resetting the timer? Aug 06 15:28:23 Or is this a user-land process? Aug 06 15:28:28 userland Aug 06 15:28:39 Ok, then anything which actually hangs the process for 60s Aug 06 15:28:50 Or anything which causes OOM Aug 06 15:28:54 It's a realtime process Aug 06 15:28:59 with locked mem pages Aug 06 15:29:34 It's not an interruptible kernel though. Aug 06 15:30:49 btw, to crash the slug just compile a kernel module which is just an infinite loop and insmod it Aug 06 15:30:58 The better way to do it is to use something other than the hardware watchdog and to call the kernel to do a reboot. Aug 06 15:31:20 the kernel soft watchdog? Aug 06 15:31:34 What is the kernel soft watchdog? Aug 06 15:31:46 a software watchdog Aug 06 15:31:54 does what the hardware watchdog is supposed to do Aug 06 15:32:06 hmm, a piece of software that can detect if it has crashed and shut itself down Aug 06 15:32:07 you still have a userspace app though Aug 06 15:32:28 well, that works if it actually calls the kernel reboot - because that shuts down the MTD. Aug 06 15:33:14 What I'm saying is that the hardware problems means that if the kernel is not able to do stuff (shut down the MTD) there is no reliable recovery. Aug 06 15:33:30 lousy hardware :( Aug 06 15:34:14 Right, and they apparently don't regard it as serious enough to fix - which I find amazing. Aug 06 15:35:08 and, of course, there is no workaround... Aug 06 15:35:56 Well, the reset line could be wired to cause a power off/on sequence. Aug 06 15:36:10 I believe reset is always asserted. Aug 06 15:36:51 ok, but that does require that I figure out how to actually do that ;) Aug 06 15:37:45 For the watchdog? Is it really setting the hardware timeout? Aug 06 15:38:04 probably not Aug 06 15:38:12 the watchdog defaults to 1 minute Aug 06 15:38:16 which is "standard" Aug 06 15:38:46 Doing the hardware watchdog is very hardware specific... Let me check. Aug 06 15:40:23 hmm Aug 06 15:41:29 ok, I mounted the flash read/write and opened /dev/watchdog. That means my slug will hang in about half a minute, right? Aug 06 15:42:43 Maybe - remember shutdown -r sometimes worked, I think the driver just leaves the MTD in the state it was in on the last operation. Aug 06 15:43:25 So I observed a hang on a 'fixed' (ro initrd) system after a reflash -k - my assumption was that this was because the last thing done to the MTD was a write of part of the kernel. Aug 06 15:44:07 (And reflash -k does a complete read through the whole of /dev/mtdblock2 in a separate process after the last block is written and after an fdatasync, so this is a bit weird!) Aug 06 15:44:30 hmm Aug 06 15:44:30 Anyway. I don't think you will see a hang because I don't think you are using the hardware watchdog. Aug 06 15:44:41 why not? Aug 06 15:44:50 is the kernel watchdog compiled in? Aug 06 15:44:57 The only code in the whole kernel which uses IXP4XX_OSW is "arch_reset". Aug 06 15:45:04 odd Aug 06 15:45:26 And arch reset causes an immediate reboot - there is absolutely no support I can find for setting the timer! Aug 06 15:45:31 when I run udev, /dev/IXPwhateverWatchdog is created Aug 06 15:46:12 IXP4xx Watchdog, to be exact Aug 06 15:46:13 configs/ixp4xx_defconfig:CONFIG_IXP4XX_WATCHDOG=y Aug 06 15:47:04 I guess the code could have hard-coded the hex address, not used an IXP4XX_OSW define. Aug 06 15:49:35 Ah ha, the code is in drivers/char/watchdog, I just searched arch/arm Aug 06 15:52:09 Yes, that uses the watchdog. It will hang under some circumstances. Aug 06 15:52:35 hmm Aug 06 15:52:43 do you know a way to provoke it to hang? Aug 06 15:54:43 Remounting /initrd rw and writing something to it seemed sufficient. Aug 06 15:55:34 Try while true; sleep 1; do date >/initrd/date; sync; done Aug 06 15:56:10 ok, then it should die soon Aug 06 15:56:13 But I'm just guessing - I really don't understand why it is sometimes inthe 'bad' mode and sometimes now. Aug 06 15:56:20 s/now/not/ Aug 06 15:56:51 I suspect the driver could be fixed up too. Aug 06 15:57:12 yeah Aug 06 15:57:13 that hung Aug 06 15:57:37 too bad Aug 06 15:57:45 would've been nice with a hardware watchdog Aug 06 15:58:05 mtd-shutdown.patch - something in the same files is leaving the flash in the bad state, and something in the shutdown code I added fixes it. Aug 06 15:58:57 Something in dirvers/mtd/maps/ixp4xx.c I think. Aug 06 15:59:16 Only 321 lines in there... Aug 06 15:59:47 *IXP4XX_EXP_CS0 &= ~IXP4XX_FLASH_WRITABLE; Aug 06 16:01:04 And it is only set on line 222 Aug 06 16:04:48 I've got *no* idea what you're talking about right now ;) Aug 06 16:05:13 The code in the kernel which actually does the IXP4XX specific parts of writing to flash memory. Aug 06 16:05:45 Hm Aug 06 16:06:13 Anyway, kernel expert required - I believe some of the writes are in generic code therefore to bracket them with enable/disable of the write may mean hacking the higher levels. Aug 06 16:06:37 I'll slugbug it I think. Aug 06 16:06:41 Correct me if I'm wrong though.. If a user is running from disk, and the flash has already been mounted writeonly.. Will the flash ever (except in reflash situations) be touched? Aug 06 16:07:11 or might it already be in a "bad state" then? Aug 06 16:09:12 If /initrd is remounted readonly then apparently the write mode is cancelled (I don't understand how this happens), reset is then safe. Aug 06 16:09:57 If some other flash partition is accessed the mode can be left in write mode - even after the close of the device - and a simple umount of /initrd does the same thing (i.e. leaves the flash in write mode). Aug 06 16:10:04 Both of those things seem to be bugs. Aug 06 16:10:13 oh Aug 06 16:10:24 If the flash is actually being written at the moment of reset then I'm sure there is no work round. Aug 06 16:10:34 no, there probably isn't Aug 06 16:10:38 But that's pretty unlikely - writes don't take that long... Aug 06 16:10:59 Anyway, it's slugbug 249 and I'm going to copy this log in. Aug 06 16:11:07 go ahead :) Aug 06 16:11:18 Might be worth porting 'softdog' though... Aug 06 16:11:53 I'll get right to it ;) Aug 06 16:14:46 urgh Aug 06 16:14:48 include mess Aug 06 16:14:49 softdog.c: In function `main': Aug 06 16:14:49 softdog.c:128: warning: implicit declaration of function `MKDEV' Aug 06 16:14:49 softdog.c: In function `proc_misc_check_device': Aug 06 16:14:51 softdog.c:282: warning: implicit declaration of function `strcmp' Aug 06 16:15:13 softdog.o(.text+0x39a): In function `main': Aug 06 16:15:13 : undefined reference to `MKDEV' Aug 06 16:15:13 collect2: ld returned 1 exit status Aug 06 16:15:52 MKDEV is a macro Aug 06 16:17:22 if (mknod(DEVICE, S_IFCHR | 0600, MKDEV(MY_MAJOR,MY_MINOR))) <-- offending line. There's no other references to MKDEV Aug 06 16:17:32 NAiL: do you know what the state of the ucslugc feeds is? Aug 06 16:17:53 no idea, thought you kept an eye on that? Aug 06 16:18:19 I don't even know where they are... Aug 06 16:18:48 http://ipkg.nslu2-linux.org/feeds/ucslugc/ Aug 06 16:18:57 That's from my mirror. Looks ok by me. Aug 06 16:19:07 the native feed is empty Aug 06 16:19:14 which is probably to be expected? :) Aug 06 16:19:26 Ah, ok - need to make a release so there is a stable feed :) Aug 06 16:20:30 Are these things on mgabe or whatever the thing is called? Aug 06 16:21:58 I think this is built on nudi, then pushed to nugabe, then pushed to the mirrors from there Aug 06 16:22:29 the problem at the moment is that you have to create the feed dirs manually on each mirror I think Aug 06 16:22:42 Hum... I don't seem to have tickadj on my system. Aug 06 16:25:49 * NAiL puts off fixing softdog until tomorrow Aug 06 16:25:57 softdog requires kernel driver btw Aug 06 16:26:13 hmm Aug 06 16:26:20 Hum... better to fix the MTD driver if possible. Aug 06 16:26:47 Strange, I have 'ntp-tickadj' 'install ok installed' in /usr/lib/ipkg/status Aug 06 16:27:34 that is strange Aug 06 16:27:42 But info/ntp-tickadj.list is empty Aug 06 16:31:07 ipkg remove ntp ntp-tickadj; ipkg install ntp fixes the problem. I suspect this is another side effect of slugbug 248 Aug 06 16:31:09 WTF?! Aug 06 16:31:27 my syslog has stopped working again Aug 06 16:31:35 and now I have "buffer" in /etc/syslog.conf Aug 06 16:32:26 This was fixed in -r30 of busybox, right? Aug 06 16:32:34 That's difficult to track down, but you can cat the current buffer contents (/proc/mesg?) Aug 06 16:32:57 or use logread Aug 06 16:33:04 The death of syslog wasn't fixed - that happens when the kernel hits OOM conditions, the overwrite of syslog.conf was. Aug 06 16:33:18 logread requires the "circular buffer" - syslog sets that up. Aug 06 16:33:49 yeah, but the death was because I rebooted. Why was the syslog.conf file back to buffer? Aug 06 16:34:38 Does the syslog have the ROTATE lines, and are they commented out if so? Aug 06 16:34:45 they are commented out Aug 06 16:35:03 The kernel messages (alone) are in /proc/kmsg - you can cat it directly (^C to terminate) Aug 06 16:35:29 So your syslog.conf is the new one from the upgrade - you must have chosen the 'overwrite' CONFFILE option (Y or I) Aug 06 16:35:44 I've restarted the syslog daemon after fixing it to file again Aug 06 16:35:48 so no kmesg Aug 06 16:35:55 besides, I use dmesg for that :) Aug 06 16:51:45 yep, feeds are built on nudi, and pushed locally to nugabe. then sources is asked to mirror them around the world. this is done very 15 minutes (assuming there is nothing to build, or at the end of each individual distro build. Aug 06 16:53:00 jbowler-zzz: do we need a ucslugc native build machine? Aug 06 16:55:47 03jbowler 07org.openembedded.nslu2-linux * r51b6958c... 10/conf/distro/ucslugc.conf: Aug 06 16:55:47 UcSlugC: make (significantly) smaller by removing the kernel from the root Aug 06 16:55:47 file system - the size of the rootfs is reduced by 35% of the previous Aug 06 16:55:47 size! Aug 06 16:56:22 What's the current size? Aug 06 16:56:28 rwhitby: not so far Aug 06 16:57:28 NAiL: 100/35  kernel size (original) 65/35  kernel_size (now) Aug 06 16:57:46 Hum, insert a * either side of the bold bit. Aug 06 16:59:51 2621440 before, 1703936 after (that's the size of the jffs2 image). There is a total of 4.28MByte free on the flash file system initially. Aug 06 17:00:23 http://www.nslu2-linux.org/wiki/UcSlugC/HomePage Aug 06 17:02:00 yes, that was a question I wondered about earlier... why is the kernel still in the flash image? Aug 06 17:04:58 cause that's what OE did by default and no-one changed it until now ... Aug 06 17:19:24 because APEX needs it to boot openslug Aug 06 17:22:04 but that can't be a justification for keeping it in UcSlugC when it increases the size of the image so much. Aug 06 17:22:26 guys, i'm getting an error installing jikes through ipkg Aug 06 17:22:47 it's when it's installing libstdc++, which jikes needs Aug 06 17:24:12 How did you get it to build? I thought there was some problem - it's not in the current builds (openslug or ucslugc) Aug 06 17:24:41 sorry i didn't build it Aug 06 17:25:07 i was just trying to install jikes, which tried to install libstdc++ Aug 06 17:25:23 An error ocurred, return value: 22. Aug 06 17:25:23 Collected errors: Aug 06 17:25:23 ipkg_download: ERROR: Command failed with return value 1: `wget --passive-ftp -q -P /tmp/ipkg-X6jlIX http://ipkg.nslu2-linux.org/feeds/unslung/cross/libstdc++_5.0.7-3_armeb.ipk' Aug 06 17:25:23 Failed to download libstdc++. Perhaps you need to run 'ipkg update'? Aug 06 17:25:28 Airw0lf: did you see the nslu2-linux mailing list message about that problem? Aug 06 17:25:51 sorry i didn't...i need to subscribe Aug 06 17:26:03 It's a known problem with the redirection script Aug 06 17:26:07 http://groups.yahoo.com/group/nslu2-linux/message/7905 Aug 06 17:26:18 Tiersten: ping? Aug 06 17:26:27 any other mod_rewrite gurus here? Aug 06 17:27:33 aha right...i shall get the package myself Aug 06 17:28:29 or someone who knows the PHP code required to take $_GET['f'] and do the right %foo; stuff quoting to make it a suitable value for the Location: header in the presence of punctuation characters in the URL ... Aug 06 17:28:56 $redirect_url = $mirrors[$mirror_number] . $_GET['f']; Aug 06 17:28:56 header("Location: " . $redirect_url); Aug 06 17:30:42 i should have guessed myself that the ++ would have caused a problem Aug 06 17:35:08 can i delete the ipk file after installing it? Aug 06 17:35:26 yes Aug 06 17:35:32 ok Aug 06 17:40:04 very nice...i can compile and run java code now Aug 06 17:41:14 I... I... You could.. nah.. let me sober up first :) Aug 06 17:41:56 kolla: re busybox GPL, yeah, but the copyright owners have been notified, and have chosen to do nothing about it Aug 06 17:42:25 so, I went to stockholm, sweden... met the moosebrothers and maretha franklin and got drunk... happy now :) Aug 06 17:42:44 rwhitby: ah, ok Aug 06 17:43:22 maretha? Is that a mix between merete and aretha? ;-) Aug 06 17:43:31 reminds me.. I asked on the pine mailing list how they would react to someone releasing a binary patch... Aug 06 17:44:20 NAiL: probably.. she was damn good (and imo quite a bit .. more.. like.. atractive .. than aretha :) ) Aug 06 17:44:28 and linksys have been asked and have chosen to do nothing about it. So the only people who can do anything about it is the original copyright holders (busybox authors) or the FSF (on busybox's behalf). The GPL doesn't allow just anyone to take action on it's behalf. Aug 06 17:44:34 tinier :) Aug 06 17:46:00 oh crap...do we have 'nohup' for unslung? Aug 06 17:46:27 Airw0lf: ipkgfind.nslu2-linux.org, and #nslu2-general Aug 06 17:46:36 kill nohup? Aug 06 17:46:53 I'd assume so Aug 06 17:47:40 nope there's no nohup Aug 06 17:57:57 hi Aug 06 17:58:09 hi Aug 06 18:21:54 Airw0lf: See #nslu2-general for nohup answer Aug 06 20:07:46 LinkSys: not only that (that it is the copyright holders right alone to persue infringements of a license), but unless the busybox guys actually say that LinkSys licensed the software under GPL, who knows? "bb" is free to arrange a different license (with the consent of all the contributors). **** ENDING LOGGING AT Sat Aug 06 23:59:57 2005