**** BEGIN LOGGING AT Wed Feb 28 03:00:01 2007 Feb 28 03:23:24 anyone built a 2.6.18-4 or 2.6.20 image w/ intel driver support? Feb 28 03:24:49 rwhitby, You alive? Feb 28 03:25:15 yep Feb 28 03:25:33 I've gotten to the kernel selection for Debian, what should I choose? Feb 28 03:25:36 My options are: Feb 28 03:25:41 linux-image-ixp4xx Feb 28 03:25:50 linux-image-2.6.18-3-ixp4xx Feb 28 03:25:57 linux-image-2.6-ixp4xx Feb 28 03:26:06 Or, rather confusingly, 'none'. Feb 28 03:26:34 the third one Feb 28 03:26:44 Ok, and might I ask why? Feb 28 03:26:58 cause then it will upgrade to -4 later Feb 28 03:27:02 Makes sense. Feb 28 03:27:15 Now it wants to know what to generate the boot initrd with, yaird or initramfs-tools? Feb 28 03:27:25 initramfs-tools Feb 28 03:27:39 I've never been asked that before, what does it mean? Feb 28 03:28:09 rwhitby: any idea whne 2.6.18-4 is going to be ready for nslu2? Feb 28 03:28:22 jason0_: it's already in testing Feb 28 03:28:31 s/testing/unstable/ maybe Feb 28 03:28:44 rwhitby: when you upgrade the kernel do you have to reflash the slug or is it just a simple apt-get Feb 28 03:28:50 I think it's an apt-get. Feb 28 03:28:59 apt-get calls flash-kernel which reflashes Feb 28 03:29:03 Oh. Feb 28 03:29:24 hm.. I'd really like to to get 2.16.18-4 so rtorrent works. Feb 28 03:29:26 Either way. Feb 28 03:30:09 jason0_: what do you currently have installed? Feb 28 03:30:24 2.6.18-3 Feb 28 03:31:02 I thought Martin posted an email on how to upgrade from there ... Feb 28 03:31:29 in nslu2-linux/general on yahoo groups? Feb 28 03:31:39 rwhitby, Did I tell you about how to use RC1 (at least temporarily) and still have it work? Feb 28 03:31:50 theblue: yes Feb 28 03:32:04 Ok, I wasn't sure if I said in this channel or somewhere else. Feb 28 03:32:06 jason0_: I forget where Feb 28 03:33:44 jason0_: have you checked the unstable debian repo? Feb 28 03:34:20 Yeah I know it's in there, but when I asked about how to upgrade in #debian everyone advised against it. I'm willing to take a gamble. Feb 28 03:34:56 Someone has to be, otherwise there's no development. Feb 28 03:35:10 * theblue runs off to update the wiki. Feb 28 03:35:24 take a backup of your flash with cat /dev/mtdblock* > backup.bin, and then you can always upslug2 it back again if things go horribly wrong. Same with the disk (take a backup first) Feb 28 04:28:33 03bzhou * r5651 10optware/trunk/ (4 files in 3 dirs): added and promoted launchtool for non-uclibc platforms Feb 28 04:40:33 rwhitby: do the debian repo's include the onboard ethernet adapter support? Feb 28 04:42:34 jason0_, Nope. Feb 28 04:42:45 YAY! Feb 28 04:43:06 I hereby christen thee 'chadwick'! Feb 28 04:43:15 * theblue breaks a bottle of champagne on the side of the rack! Feb 28 04:45:38 So if they don't include the ethernet support, wouldn't people be hozed when they did an apt-get upgrade? Feb 28 04:46:24 No, since the microcode is included in the image. Feb 28 04:47:00 so the debian repo "images" should include the microcode. Feb 28 04:47:22 I guess. Feb 28 04:49:09 there are no "debian repo images" Feb 28 04:49:15 only slug-firmware.net images Feb 28 04:49:35 the microcode cannot be distributed by Debian, cause it's not DFSG compliant Feb 28 04:52:12 When I try to build the cross compile env. I get an error: make[1]: *** [configure-arm-linux-gnu-stamp] Error 127. Can't seem to resolve it otherwise I'd build the image myself. Feb 28 04:57:09 jason0_: why not just install the RC1 like theblue did, then upgrade the kernel from there? Feb 28 04:58:11 I did install RC1 a few weeks ago. Feb 28 04:58:59 so point your apt/sources.list to an unstable arm repo and upgrade Feb 28 05:01:46 I must be totally off base about the microcode thing. I thought if I used the repo then that means no microcode Feb 28 05:02:35 and therefore no ethernet support Feb 28 05:03:47 the microcode is likely on your disk already - it should have been installed with RC1 I believe Feb 28 05:04:34 So the microcode isn't used internally in the kernel, but delegated to at startup? Feb 28 05:04:50 I thought the microcode had to be compiled into the kernel, thats why I was confused Feb 28 05:05:10 the microcode cannot be compiled into the kernel - that would be a *huge* intel license violation Feb 28 05:05:34 and GPL license violation Feb 28 05:06:02 Okay, that makes it clear then. When I was reading the BuildImage directions I misunderstood that it had to be built it, hence my confusion. Feb 28 05:06:14 *in Feb 28 05:07:36 jacques: yeah, but I fear Intel's lawyers more than Richard Stallman :-) Feb 28 05:13:46 Yikes! Feb 28 05:14:08 rwhitby, Were you the one who found the microcode? Feb 28 05:14:52 you download the microcode from intel's site when you build an image Feb 28 05:16:41 I see. Feb 28 05:17:02 Thanks once again for the help, chadwick's running beautifully now. Feb 28 05:23:13 upgrading now. Feb 28 05:26:45 03bzhou * r5652 10optware/trunk/make/python24.mk: python24: reuse the same host build if we can; a couple of small improvements Feb 28 05:26:57 03bzhou * r5653 10optware/trunk/make/python25.mk: python25: reuse the same host build if we can; a couple of small improvements Feb 28 05:27:08 03bzhou * r5654 10optware/trunk/make/python.mk: python: python-stage includes -stage -host-stage for python24 & python25 Feb 28 05:37:25 theblue: your slug is in a rack? Feb 28 05:39:26 rwhitby: reboot worked, running 18-4, thanks for your patience. Feb 28 05:39:53 jason0_: glad to hear it. which wiki pages can you update to make it easier for the next person? Feb 28 05:40:50 I'll browse nslu2-linux and see where it's appropriate. Feb 28 05:46:08 under http://www.nslu2-linux.org/wiki/Debian/HomePage somewhere. Feb 28 12:48:32 rwhitby: How does a cross compiler find the non-system (in this case opencv) libraries? it can't use the host's ldconf.so? Feb 28 12:48:51 dunno Feb 28 12:48:56 darn. Feb 28 12:49:11 I guess I'll try to build it outside of OE then. Feb 28 15:02:39 03bzhou * r5655 10optware/trunk/make/ntop.mk: ntop: added pcre to DEPENDS and -stage Feb 28 15:16:33 03nail * r5656 10optware/trunk/sources/buildroot/buildroot.ts101: Update ts101 buildroot config Feb 28 15:19:50 03nail * r5657 10optware/trunk/sources/buildroot/uclibc.ts101: Update ts101 uclibc config Feb 28 15:21:06 eno: ping? Feb 28 15:21:23 eno: I can't get it to actually *use* that config file. Not quite sure why. Feb 28 15:21:54 03nail * r5658 10optware/trunk/sources/ts101-kernel-modules/kernel.defconfig: ts101-kernel-modules: Update kernel config Feb 28 15:35:06 hey! Feb 28 15:52:59 eno: vxwidgets built if I replaced binutils ;-) Feb 28 16:13:45 03nail * r5659 10optware/trunk/make/buildroot.mk: buildroot: Hack around missing uclibc.config Feb 28 16:22:15 NAiL: ts101 toolchain is building here, good you get wxbase working Feb 28 16:23:37 will be in a long meeting this whole morning Feb 28 16:25:04 eno: cool Feb 28 16:25:11 I'm still building toolchain too Feb 28 16:28:49 I think most of the failed packages now are either very obscure errors or glib failing Feb 28 16:41:13 hey NAiL! Feb 28 16:41:59 thanks for the file yesterday, .. even though, I was unable to get it to function :) It reported a different schema which my monotone did not like.. so i've spent a day migrating the db myself. It actually finished like 1 hour ago Feb 28 16:44:09 odd... which version are you running now? Feb 28 17:06:56 the lastest stuff from Etch, monotone 0.31 (base revision: 1bd1fe1e811dce82bee09b9f0effca3225bd1cee) Feb 28 17:08:20 also, how much disk space will a full slugosbe build consume? cause my poor p2 450, which im using as a buildserver, only has about 6 gb available Feb 28 17:13:37 Blastur: my sources directory is currently 670MB (could probably be pruned quite a bit) and the actual slugos stuff is at 4.9GB (including all the extra packages I've built) Feb 28 17:14:52 should be ok then Feb 28 17:18:26 joshin, i'm planning to build several different images though, like slugosbe and slugosle .. i guess they have a lot in common, will those files overlap, or will I need the double in diskspace? Feb 28 17:18:46 i mean, it downloads a lot of source, i guess that can be used to compile a toolchain for both Feb 28 17:18:54 Double the fun, double the disk space Feb 28 17:18:57 =) Feb 28 17:19:16 At least the source will be shared. :) Feb 28 17:20:48 i guess i have to compile it, remove, and re-compile =) Feb 28 17:22:20 just to be sure i'm not doing a lot of work in vain here, there is no place to download built images for all different targets built from the latest source, right? like, a system which builds images every night from the repository Feb 28 17:22:47 i'm specifically looking for slugosbe, slugosle Feb 28 17:24:39 Not that I know of. At least not yet. Feb 28 17:25:20 * joshin ponders if he's interested in doing that... Feb 28 17:26:47 I wonder how much net bandwidth that would take... Feb 28 17:29:40 i think the community would love it,.. there are a lot of people sitting out there, just waiting for stuff on slug-firmware.net to update :) and they don't have the resources/knowledge to build themself Feb 28 17:30:09 actually, i could supply with a mirror if someone sets up a build-thingie Feb 28 17:32:06 In that case, you will need a click-through licence Feb 28 17:32:17 ssince the image contains intel-licenced code Feb 28 17:32:43 * joshin starts to get a headache Feb 28 17:32:58 i think most people can live with a click-through licence compared to building an image from scratch Feb 28 17:33:15 Blastur: btw, seen http://hentges.net/tmp/snapshots/SlugOS/ ? Feb 28 17:34:03 It only has LE thus far Feb 28 17:34:10 but it has nightlies Feb 28 17:34:27 hey, that's exactly what I meant :) Feb 28 17:35:05 I like that solution to the click-through license problem. Feb 28 17:37:28 who runs that site? Feb 28 17:38:01 CoreDump Feb 28 17:41:08 maybe i can convince him to do BE aswell :) Feb 28 17:43:08 NAiL: any experience with fetching source code from svn repo via http for oe? Feb 28 17:43:31 hey, while snooping around at hentges.net, i found this: http://hentges.net/tmp/photos/Akita/Casing/Bild(1).jpg .. anyone know the name of that device? looks sweet! Feb 28 17:44:05 oh right, it actually says "ZAURUS" with bold letters on the next picture =) Feb 28 18:01:04 03nail * r5660 10optware/trunk/platforms/packages-ts101.mk: add broken packages Feb 28 18:02:41 osas: no Feb 28 18:02:54 ok ... Feb 28 18:03:35 I was thinking og getting asterisk 1.4 to work on openslug ... but I got stuck at getting the source :p Feb 28 19:32:26 hi any reliable howtos on creating your own jffs2 image out there...? Feb 28 20:01:38 03bzhou * r5661 10optware/trunk/scripts/report-broken.sh: report-broken.sh: still defaults to COMMON_CROSS_PACKAGES, but can use other PKGS_VAR Feb 28 21:25:26 hey.. is it true that Debian Installer will actually flash the bootloader with APEX during install? is that not _insane_ to do, in case something goes wrong and the user does not have a JTAG cable? Also, does APEX still have the feature of going into "upgrade mode"? Feb 28 22:42:30 Blastur: Debian installs Apex as a secondary bootloader, which is called by the existing redboot. Feb 28 22:42:53 ah..why? Feb 28 22:43:20 i mean, why use apex at all? Feb 28 22:44:33 because redboot has too many limitations Feb 28 22:44:37 eg. <1mb kernel Feb 28 22:45:09 Blastur: debian kernel is larger than 1MB Feb 28 22:45:27 that is the only real reason at the moment. Feb 28 22:45:33 Blastur: you can use 'INHERIT += "rm_work"' in your local.conf to conserve disk space Feb 28 22:46:02 we *never* overwrite the primary bootloader Feb 28 22:49:17 ah, thats what i thought. Kinda scared me there for a moment, that someone overwritten redboot Feb 28 22:49:42 hi Feb 28 22:49:46 hey coredump Feb 28 22:50:05 heya, I saw you found the LE image ;) Feb 28 22:50:11 i did :) Feb 28 22:50:18 with some help from my friends! Feb 28 22:50:46 i like the solution, with regularly built images from the repository Feb 28 22:50:50 ;) Feb 28 22:51:01 semi regulary actually Feb 28 22:51:18 well, anything appears regular compared to slug-firmware.net =) Feb 28 22:51:29 anyway, do you have any plans to extend it to include BE images aswell? Feb 28 23:00:12 Blastur: I could kick off a BE build, yeah. You'd have to wait for that until tomorrow this time tho Feb 28 23:00:48 i've started building one myself, been working on it for the past 6 or 7 hours now Feb 28 23:00:57 but i guess it wont be done until tomorrow Feb 28 23:00:58 CoreDump|home: multimachine should work, but not much is shared cause of the endianness. Feb 28 23:01:28 rwhitby: what exactly is multimachine? Feb 28 23:01:50 im just saying, pre-built BE and LE (and possibly other) images would be nice :) (from a somewhat updated source) .. built perhaps once a week Feb 28 23:02:39 NAiL: bitbake class Feb 28 23:03:13 allows you to build multiple MACHINEs in a single tmp dir. Feb 28 23:03:19 and shares what can be shared Feb 28 23:03:29 aha Feb 28 23:04:45 rwhitby: NAiL it has been my _personal_ experience that multimachine builds can screw up badly from time to time. If you have the HDD space, just compile w/ completely different "tmp" directories for each machine (just my 2 cent) Feb 28 23:05:12 * CoreDump|home has been doing the latter in the past Feb 28 23:05:33 CoreDump|home: yeah, I still keep them separate. haven't used the multimachine much yet Feb 28 23:06:10 rwhitby: but we still suffer from the retarded koen deploy patch Feb 28 23:06:36 which is a *real* PITA Feb 28 23:07:19 mwester: indeed. I intend to atleast fix up meta/package-index to create a working single "Packages" file out of the messed up deploy dir Feb 28 23:08:18 as all my testing is done by adding a feed to my build machine, which as a web server that serves the feeds for my local copies of unslung, optware, slugosle, and slugosbe. Now I have to cut and paste and use wget to manually fetch the packages. Feb 28 23:08:53 mwester: yeah. worst .dev change in a _long_ time IMO Feb 28 23:09:13 Well, the most disruptive, anyway. Feb 28 23:10:01 right Feb 28 23:10:34 and the complete rejection of even adding a way to revert to the old way Feb 28 23:13:48 * mwester wonders what happens if he just creates the correct symlinks in the deploy/ipks directory, all pointing to the same real directory... Feb 28 23:14:23 what is this retarded koen deploy patch? :) Feb 28 23:14:27 koen submits a patch that breaks symlinks in package dir :-P Feb 28 23:14:51 who is koen? and does the repository not support rollback? Feb 28 23:15:00 you no longer can use a compiled OE feed as an actual feed w/o manual fixing Feb 28 23:25:20 compiled OE feed? do you mean i cannot use the OE MasterMakefile to make an slugosBE image? Feb 28 23:26:34 that you can. However, in the past you could point ipkg.conf to the OE-generated deploy dir and use it as a feed for every package compiled w/ OE Feb 28 23:27:45 but that didn't work for koen for some reason (whilst it worked for everyone else in the world) and so he decided to break it for everyone else just so it worked for him. Feb 28 23:30:39 this koen guys seems like a lovely fellah! Feb 28 23:31:50 guy* Feb 28 23:35:02 he is... Feb 28 23:35:36 I'm sure he's a decent guy -- some folks here are just peeved about the ramifications of this particular recent change. Feb 28 23:36:25 does anyone have write-access to this repository? Feb 28 23:36:26 mwester: his response to my email on the ML showed his usual unfriendly face Feb 28 23:36:37 Blastur: we all do Feb 28 23:37:10 Y'know, I complain about it, but I can be easily placated if there was just some easy way to consolidate all the packages once again and ready them for deployment. Feb 28 23:38:17 rwhitby: would it be the wrong solution to add a target to the master makefile to consolidate the packages and prepare the index? Feb 28 23:38:22 mwester: I didn't look into it yet but I believe package-index could be reworked to do just that Feb 28 23:38:35 mwester: better to do it in package-index.bb as CoreDump|home suggests Feb 28 23:38:39 Oops - late to pick up daughter from dance class - bbiab Feb 28 23:38:56 if that can't be done, then we would do it in the mmaster makefile. Feb 28 23:39:47 right Feb 28 23:49:48 rwhitby: it's not easily fixable due to limitations of ipkg-make-index. One could, however, create a new "combined" directory w/ symlinks from all ARCH'es and generate Packages from it Feb 28 23:51:27 tmp/deploy/everything ? Feb 28 23:52:24 would be misleading as it doesn't contain the images (and hence "everything"). That's what koen would rant anyway Feb 28 23:53:29 deploy/ipk-feed ? Mar 01 00:00:49 sure Mar 01 00:01:14 koen will rant that "OE was never, and will never, be a feed" but we can ignore that. Mar 01 00:01:53 =) Mar 01 00:04:18 rwhitby: muhaha, so deploy/oe-feed it shall be then Mar 01 00:04:36 * CoreDump|home is looking forward to the flamefest Mar 01 00:28:14 omg, the "other" friendly OE guy is back as well Mar 01 00:46:32 * mwester is looking forward to being able to bitbake and install easily, once again. :) Mar 01 00:47:48 mwester: I'm almost done Mar 01 00:48:01 thinking of a matching commit message atm =) Mar 01 00:49:31 qtopia is as open as openmoko is Mar 01 00:49:33 hehe Mar 01 01:35:03 mwester: pushed Mar 01 01:42:18 CoreDump|home: love the checkin message :-) Mar 01 01:42:35 rwhitby: heh Mar 01 01:46:02 hehe Mar 01 01:49:47 Somebody is feeling feisty and it ain't no fawn... Mar 01 01:53:17 joshin: you don't even want to see my first draft of that message Mar 01 01:54:09 Well I might since I very rarely send even my third draft of something when ticked off. :) Mar 01 01:55:01 joshin: http://rafb.net/p/GNilso56.html ;) Mar 01 01:55:35 That's not too bad. :) Mar 01 01:56:00 hehe Mar 01 01:56:26 * joshin goes to enable it Mar 01 02:00:34 The one OE feature I'd like to see would be to be able to do a "-=" in my local.conf to not build/install packages that I don't use (thinking of szrz and reiserfs stuff that has been in previous slug builds). Mar 01 02:01:43 joshin: you could always ceate your own "joshin-slugos-image" by inheriting slugos-image and overwriting some package vars Mar 01 02:02:24 oh joy, bbimage is b0rked Mar 01 02:02:35 That's true. Of course when I last mentioned it, MickeyL thought it a great idea (but probably a pain in the neck to implement safely). **** ENDING LOGGING AT Thu Mar 01 02:59:57 2007