**** BEGIN LOGGING AT Tue Aug 28 02:59:56 2007 Aug 28 03:33:12 18M Angstrom-x11-image-uclibc-test-20070828-at91sam9263ek.rootfs.tar.gz :) Aug 28 03:34:05 18M glibc/images/at91sam9263ek/Angstrom-x11-image-glibc-2008.1-test-20070826-at91sam9263ek.rootfs.tar.gz Aug 28 03:34:20 * koen shivers at the OMG OPTIMIZED AND SMALLER uclibc rootfs Aug 28 03:35:04 down from what size? Aug 28 03:35:18 I haven't been paying attention :) Aug 28 03:35:51 the uclibc image is the same size as the glibc image Aug 28 03:36:09 ah Aug 28 03:37:04 03koen 07org.oe.dev * r7f7e2abb... 10/ (1 packages/midori packages/midori/midori_0.0.5.bb): disapproval of revision '6b34ae43bf4f59a33f79e42d0cc2ad1b0d5e0c6e' Aug 28 03:37:08 03koen 07org.oe.dev * r160d348d... 10/ (1 classes/rootfs_deb.bbclass classes/rootfs_ipk.bbclass): rootfs*.bbclass: uclibc doesn't provides glibc-locale* and locale-base*, so don't try to install it Aug 28 03:46:40 I don't understand why PATH of scripts executed by udev is "/usr/gnu/bin:/usr/local/bin:/bin:/usr/bin:." Aug 28 03:47:15 "/usr/gnu/bin" ? "." ?! Aug 28 03:47:50 it's not set by init, not by the init scripts Aug 28 03:48:24 and strings /sbin/udevd | grep PATH doesn't return anything... Aug 28 03:48:54 I'll need to UTSL... :/ Aug 28 04:08:14 it uses fork() to run programs... it should keep env... Aug 28 04:11:28 /* clear environment for forked event processes */ Aug 28 04:11:29 clearenv(); Aug 28 04:12:13 ok, it explains why the PATH is not keeped, but not why the PATH in udev env is so strange :) Aug 28 04:22:17 03clement.hermann 07org.oe.dev * rc036ae29... 10/ (5 files in 3 dirs): udev: add sbin into PATH for network.sh, should close #2869 Aug 28 04:26:01 I didn't know I commited anything ;) Aug 28 04:26:12 but thanks Aug 28 04:33:15 FIXED. Aug 28 04:33:37 it would be nice to have an auto-close on changelog feature :) Aug 28 04:33:51 (kind of like the debian BTS) Aug 28 04:34:26 now, let's find out why this PATH is so strange. I suspect bash. Aug 28 04:35:07 yup. it's bash. Aug 28 04:36:54 oups. no, same with busybox. Aug 28 04:36:58 * * OE Bug has been RESOLVED by clement.hermann(AT)free.fr Aug 28 04:37:00 * *  udev does not bring up network interfaces on insertion Aug 28 04:37:02 * * http://bugs.openembedded.org/show_bug.cgi?id= Aug 28 04:40:09 erf. tired. it's bash indeed. Aug 28 04:41:00 it explains why this bug wasn't reported before, busybox sets the default PATH as /usr/local/bin:/usr/bin:/sbin:/bin if the EUID is 0 Aug 28 04:44:14 #ifndef DEFAULT_PATH_VALUE Aug 28 04:44:14 #define DEFAULT_PATH_VALUE \ "/usr/gnu/bin:/usr/local/bin:/bin:/usr/bin:." Aug 28 04:44:14 #endif Aug 28 04:44:18 eeeerk. Aug 28 04:47:33 anyone understand why distro=angstrom-2007.1, target=angstrom-minimal-image might build udev-092 instead of udev-100? Aug 28 04:47:59 Don't see a preferred version in the distro Aug 28 04:48:04 I think this is machine related, from a quick look at the .bb file Aug 28 04:48:09 yes, udev 092 is the default for everything, except slugos Aug 28 04:48:16 ah Aug 28 04:48:32 092 is the latest, tested release in OE Aug 28 04:48:46 read wrong. wonder why I thought it was machine related... Aug 28 04:48:49 OK, is there a reason for using an old release? Aug 28 04:49:13 Newer release just not tested? Aug 28 04:49:30 newer versions sucked (coded by kernel people, etc), so we just got tired of testing and are using a known good one Aug 28 04:49:51 Understood Aug 28 04:50:33 gumstix uses 101 in their buildroot setup, so I was just wondering Aug 28 04:51:08 same story with busybox 1.2.1 Aug 28 04:54:31 I should provide a patch for 2881... but it's 07:00... time to sleep a bit ;) Aug 28 04:54:46 'n8 (or morning) ppl ;) Aug 28 04:55:00 where does version 092 get set? not seeing it! Aug 28 04:55:28 sakoman: all other udev recipes have a negative DEFUALT_PREFERENCE Aug 28 04:56:36 Thanks! right on the first line :-) Aug 28 04:56:50 Still learning "how things are done in OE" Aug 28 05:25:47 * koen takes a photo of his mini-sunflowers in the morning sun Aug 28 05:27:09 that makes me happy enough to delve into blackfin toolchains again Aug 28 06:43:33 ~botmail for lrg: Packaged contents of bfin-uclibc into /media/hda4/OE/build/tmp-new/angstrom/deploy/uclibc/ipk/adsp-bf537-stamp/bfin-uclibc_0.9.29+svnr1757-r0_adsp-bf537-stamp.ipk Aug 28 06:58:38 good morning! Aug 28 07:03:05 hey cyberdeck Aug 28 07:06:33 03koen 07org.oe.dev * r90cf7986... 10/ (5 files in 4 dirs): uclibc 0.9.29: add config files for gumstix Aug 28 07:06:42 03koen 07org.oe.dev * r00076a86... 10/ (1 packages/uclibc/uclibc.inc): uclibc.inc: add support for uclinux-uclibc binaries Aug 28 07:06:47 03koen 07org.oe.dev * rc9d279c1... 10/ (4 files in 3 dirs): openmoko-panel-bt: add patch from http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=639 Aug 28 07:06:51 03koen 07org.oe.dev * r8ada571e... 10/ (4 files in 3 dirs): openmoko-dialer2: apply patch from http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=757 Aug 28 07:17:13 hello. I would like to add a file in /etc/modprobe.d for my machine. Is module-init-tootls the "most natural" package to do it ? Aug 28 07:19:14 I suspect you want to use the module_autoload syntax Aug 28 07:20:57 * * OE Bug 2882 has been created by autobuild(AT)openembedded.org Aug 28 07:20:59 * * u-boot-git-r1-do_fetch Aug 28 07:21:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2882 Aug 28 07:21:42 koen: that is for an alias for module autoloading right Aug 28 07:21:55 yes Aug 28 07:24:03 thanks koen for this pointer, that is exactly what I needed Aug 28 07:24:23 you do need to rebuild the kernel in OE for it to take effect Aug 28 08:04:53 morning all Aug 28 08:05:00 Morning Aug 28 08:06:43 hey RP Aug 28 08:21:26 hi rp, NAbyss_ Aug 28 08:22:12 hey XorA Aug 28 08:23:30 * koen is a bit suspicous of http://lists.linuxtogo.org/pipermail/openembedded-commits/2007-August/008332.html Aug 28 08:24:32 morning Aug 28 08:24:41 morning all Aug 28 08:24:56 hey do13 Aug 28 08:25:13 hey koen, zecke Aug 28 08:25:42 koen: hm, only possible values are linux and linux-gnueabi that provide that? I'd imagine testing for uclibc and excluding that would be smarter.. Aug 28 08:26:31 only linux and linux-gnueabi provide that Aug 28 08:26:58 but my real questions is: WTF is it doing in rootfs_.bbclass??!?! Aug 28 08:27:14 dont ask me, I only speak english Aug 28 08:27:34 IMO it should be image.bbclass with better safeguards Aug 28 08:27:52 or even better, it should be in a meta package Aug 28 08:28:56 hey florian Aug 28 08:30:55 koen: When those classes were abstracted I guess factoring out that fragment of code was a low priority... Aug 28 08:32:23 hi all Aug 28 08:36:02 /nick hrw Aug 28 08:36:04 morning Aug 28 08:36:52 now better Aug 28 08:37:03 hi Marcin Aug 28 08:37:30 steliosk: can you send me that 16M image? Aug 28 08:38:55 hi hrw, florian, Dirk Aug 28 08:39:46 hey Richard Aug 28 08:48:04 does someone has msdos boot floppy image? Aug 28 08:49:01 hrw: http://www.bootdisk.com/ Aug 28 08:49:25 http://www.bootdisk.com/bootdisk.htm Aug 28 08:49:30 ah, too slow :) Aug 28 08:50:56 wow I love this camel/camel-lite mails Aug 28 08:51:45 those ones are self extracting 'insert floppy to write' apps Aug 28 08:52:03 any way I got one - google: win98boot.img ;D Aug 28 08:52:04 zecke: pvanhoof had a nice talk about this topic at Guadec last year :-) Aug 28 08:52:39 I bet the ximian folks fear their incompetence is brought to the light :} Aug 28 08:53:02 zecke: did you see the FUD about threaded display on PGO as well? Aug 28 08:53:24 koen: yes, just read it... Aug 28 08:53:32 OMG EVO WONT SWITCH BECAUSE MODEST CANT DO THREADED VIEWS WTFBBQPONY Aug 28 08:53:55 and there is a another ximian guy writing (the p.g.o got updated) Aug 28 08:54:53 oh btw, it looks like I'm going to be a Troll... Aug 28 08:55:26 http://www.allbootdisks.com is nice Aug 28 08:57:06 zecke: uh? Aug 28 08:58:31 florian: the name of the people getting money from a norwegian toolkit vendor Aug 28 09:01:12 where is the actual mkfs.jffs2 implemented ? I guess in the image class ? can we easily change the block erase size ? Aug 28 09:01:35 vlo: it is from mtd-utils, if you asked for that Aug 28 09:01:47 vlo: and yes, you can easily change the erase size Aug 28 09:02:14 zecke: for stopping to complain about them? ;) Aug 28 09:02:19 * florian *hides* Aug 28 09:02:29 vlo: EXTRA_IMAGECMD_jffs2 = "--pad --little-endian --eraseblock=0x40000 -n" in your machine .conf Aug 28 09:02:37 florian: I hope to not get a paid opinion... Aug 28 09:02:57 zecke: thank you very much Aug 28 09:04:12 zecke: Finally made a decision? :) Aug 28 09:04:42 stefan_schmidt: knut visited us yesterday... Aug 28 09:05:32 zecke: Ah, personal contacts to belief you. :) Aug 28 09:05:49 stefan_schmidt: give mickey|offline some more coffee :) Aug 28 09:06:08 koen: Not in his near right now. Aug 28 09:06:27 koen: I guess he has some offline days between froscon and MDD Aug 28 09:06:50 ah, I thought he would go straight to alborg Aug 28 09:07:41 koen: No, we start thursday morning. 10 hours train ride. He is not that happy about it. (Old man..) Aug 28 09:08:04 he'll have internet for at least half of the ride afaik Aug 28 09:08:10 and can sleep the other half ;) Aug 28 09:09:13 stefan_schmidt: was your funding worked out? Aug 28 09:09:28 koen: No sleep. I guess he will have to prepare his slides, like me. :) Aug 28 09:09:55 zecke: I don't know. mickey|offline cared about this. Aug 28 09:10:14 stefan_schmidt: I will stay at GMIT, am allowed to do OE work (well Qt,Qtopia should play a role) and am asked to travel to conferences Aug 28 09:10:16 zecke: At least I hope it will. To much money just for an OpenEZX talk. Aug 28 09:10:41 zecke: OSiM?! Aug 28 09:11:10 zecke: That sounds nice. So you can stay at your beloved office, kick Qtopia asses and travel a lot. :) Aug 28 09:11:12 OSiM? what is that? Aug 28 09:11:26 zecke: Open Source in Mobile Aug 28 09:11:32 zecke: www.osim-conference.com iirc Aug 28 09:11:55 stefan_schmidt: and hopefully just wake up and do only one other thing besides studying... (just like at the camp...) Aug 28 09:12:06 zecke: :D Aug 28 09:12:37 lol TT is the drinksponsor Aug 28 09:12:55 zecke: After some more thinking it really sounds like the best for you. Congrats. Aug 28 09:19:13 some thinking? we have to see how things work out but I'm excited :) Aug 28 09:20:31 zecke: I was not really sure what would be the best option for you after our talks at the camp. But this one gives you all you like it seems. Aug 28 09:20:55 money to buy mate? right Aug 28 09:22:28 zecke: Exactly. That the reason why I work for OM. Aug 28 09:24:11 koen: Who cares about the haircut of your pigs? ;) Aug 28 09:25:23 PIGIIIIEEES Aug 28 09:26:22 koen: why did you shoot bob? what did he do? Aug 28 09:26:28 how to autostart pand in recent bluez? Aug 28 09:27:04 zecke: being too cute Aug 28 09:27:19 hrw: afaik by clicking 'networking' in bluez-gnome Aug 28 09:27:46 koen: on headless? Aug 28 09:28:04 hrw: dont be stupid, no linux box is headless :-D Aug 28 09:28:50 ok, so on x11-less Aug 28 09:29:22 by some config param, you'd have to poke holtman or bmidgley about it Aug 28 09:30:27 hiya folks Aug 28 09:30:59 koen: or check in debian and fix in OE? Aug 28 09:31:22 sounds like a plan Aug 28 09:31:25 torpor: Aug 28 09:31:46 torpor: bluez-gnome cvs seems to have 90% of the bits needs for doing PAN with a gui Aug 28 09:32:00 ah cool! Aug 28 09:32:09 i'll check it out .. desperate to get a bluetooth connection on my neo .. Aug 28 09:32:15 now if right-click worked on the neo instead of break due to excessive jitter.... Aug 28 09:32:17 * hrw need iperf for neo to test bt2.0edr speed Aug 28 09:33:01 bt2 <> bt1 is 580-750-930kbps Aug 28 09:33:18 ~lart logitech bt dongles Aug 28 09:33:18 * ibot gets a hotmal account and SPAMs logitech bt dongles Aug 28 09:34:35 koen: thx for adding my patch Aug 28 09:36:24 ~lart neos batteries Aug 28 09:36:24 * ibot whips out a sword and chops neos batteries in half Aug 28 09:38:13 only bateries? :p Aug 28 09:38:39 stefan_schmidt: and I was not asked to stop doing OM stuff... Aug 28 09:40:25 zecke: ... you really feared this? I mean you have spare time and your own opinion. Aug 28 09:40:54 can someone tell me the location of the latest buildhost stuff? for a new kernel image, i mean .. Aug 28 09:40:59 stefan_schmidt: no Aug 28 09:43:11 hello all Aug 28 09:43:17 zecke: No opinion you mean? ;) Aug 28 09:44:20 stefan_schmidt: I'm a zombie with paid opinion, please insert disk Aug 28 09:44:41 zecke: which slot :-) Aug 28 09:45:31 zecke: Insert: FOSS enthusiast Vol. 12 Aug 28 09:45:37 XorA: haha, the one you prefer :} Aug 28 09:46:29 * stefan_schmidt grumbles about using disk in the area of blue-ray. FOSS enthusiast Vol. 12 is 142 disks... Aug 28 09:50:04 03stefan 07org.oe.dev * r2bc8a3eb... 10/ (3 files in 3 dirs): linux-gta01: Add 2.6.22.5 Aug 28 09:51:25 argh.. Aug 28 09:51:54 I need to charge neo batteries and finally put something more fresh on bv4 and bv3 Aug 28 09:53:00 hrw: I recomend Angstrom 2010.1 :-D Aug 28 09:53:05 ;D Aug 28 09:53:13 stefan_schmidt: does 2.6.22.5 works ok on neo? Aug 28 09:54:00 hrw: It should. Shoragan runs it a long time already. I like some more testing and Haralds ok to make it default. Aug 28 09:54:16 stefan_schmidt: thx Aug 28 09:54:32 i havn't noticed any problems specific to 2.6.22.5 yet Aug 28 09:54:47 Shoragan: Fast guy. :) Aug 28 09:55:14 and i've done most of my testing for bug 677 with 22.5 Aug 28 09:55:50 Shoragan: btw, with this patch my microSDHC card works like a charm. :) Aug 28 09:56:01 i read your report Aug 28 09:56:20 did you try my debian chroot yet? :) Aug 28 09:56:42 Shoragan: No. Where to find? Aug 28 09:56:44 speaking of chroot - no news on gliin/eabi? Aug 28 09:56:55 planet.o.o or my blog Aug 28 09:57:17 hrw: Sadly not. Sean promised to bug GL/broadcom every day. Aug 28 09:59:03 Shoragan: hmm, your last post did not contain a link to your chroot. Or do you mean building myself by the wiki instructions? Aug 28 09:59:17 no, i've updated the post Aug 28 09:59:50 Shoragan: Don't update blog posts. Mosts feedreaders don't like it. :) Aug 28 10:00:25 planet does ;) Aug 28 10:01:27 Shoragan: Using a web browser for rss feeds? You are oldschool sometimes. :) Anyway, I'll have a break now. Later. Aug 28 10:01:52 :) Aug 28 10:11:12 grrr, the build logs are not attached to bugs from sepekku Aug 28 10:11:59 Crofton|home: right, buy me mate and I will take a look. I'm currently busy preparing stuff for a trade show... Aug 28 10:12:26 which show? Aug 28 10:12:45 Crofton|home: IFA and IBC Aug 28 10:16:06 those acronyms do not goole well :) Aug 28 10:16:42 zecke: btw, it works fine again on the OM bugzilla: http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=763 Aug 28 10:17:18 Crofton|home: IBC == International Broadcast Convention Aug 28 10:17:28 stefan_schmidt: hmm... Aug 28 10:17:46 strange conf for you ? Aug 28 10:18:19 stefan_schmidt: I would not know what did change... or is OpenMoko distro including angstrom as well? Aug 28 10:18:24 hi Aug 28 10:19:29 zecke: Yes, we base on it. I guess it was koens change from yesterday suggested by you. Aug 28 10:20:01 stefan_schmidt: hopefully Aug 28 10:21:36 http://www.engadget.com/2006/11/22/ps1-ps2-ps3-ebay-scam-ftw/ this is the best thing ever :-) Aug 28 10:22:27 zecke: were you looking for me yesterday? Aug 28 10:22:52 oh, that's right it was about bitbake 1.8 take and I said I wasn't fussed either way. Aug 28 10:25:16 rwhitby: right Aug 28 10:25:20 03florian 07org.oe.dev * rf1f1679b... 10/ (1 conf/machine/palmtx.conf): palmtx.conf: Include more modules, makes sound, battery and pcmcia work. Aug 28 10:25:25 03florian 07org.oe.dev * raceb7774... 10/ (1 conf/machine/include/palm.conf): palm.conf: Autoload the TX battery module. Aug 28 10:25:30 03florian 07org.oe.dev * r8d20b30b... 10/ (1 packages/linux/linux-hackndev-2.6/palmtx/defconfig): linux-hackndev-2.6: Update Palm TX defconfig from upstream defconfig: Build bluetooth, border and pcmcia modules, some configuration changes, reduce number of PTYs. Aug 28 10:25:36 03florian 07org.oe.dev * r4dafb02c... 10/ (1 packages/linux/linux-hackndev-2.6_svn.bb): linux-hackndev-2.6: PR++ Aug 28 10:34:59 PAN alix<>desktop works. PAN alix<>neo does not ;( Aug 28 10:35:10 l2ping works Aug 28 10:36:23 ideas? Aug 28 10:36:57 * * OE Bug 2882 has been RESOLVED (INVALID) by philip(AT)balister.org Aug 28 10:36:59 * *  u-boot-git-r1-do_fetch Aug 28 10:37:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2882 Aug 28 10:46:17 hi mickeyl Aug 28 10:47:10 hey hrw Aug 28 10:52:27 modules from MACHINE_EXTRA_RRECOMMENDS do not get installed in Angstrom minimal image .... Aug 28 10:52:55 because they are EXTRA Aug 28 10:53:02 minimal get ESSENTIAL ones Aug 28 10:57:12 heh Aug 28 10:57:34 anyone know offhand how to display the public key portion of a monotone key? Aug 28 10:57:37 task-boot gets ESSENTAIL Aug 28 10:57:38 it's been a while here Aug 28 10:57:55 task-? gets extra Aug 28 10:58:31 mickeyl: cat ~/.monotone/keys/key? Aug 28 10:59:57 isn't that only the private portion? Aug 28 11:00:05 ah yes Aug 28 11:00:50 aah Aug 28 11:00:58 i think mtn pubkey >/file Aug 28 11:01:50 koen: do you have a moment to inject a key? Aug 28 11:02:59 anyways, sending it to you via mail Aug 28 11:09:05 mickeyl: what does neod do currently? Aug 28 11:10:20 hrw: launching apps/menus on key presses, touch click sound, welcome sound Aug 28 11:10:29 hrw: dimming display, turning off etc. Aug 28 11:10:34 thx Aug 28 11:15:53 neod -- the thing that shouldn't be Aug 28 11:15:54 :D Aug 28 11:16:06 looking forward to just use one of the alternatives eventually Aug 28 11:16:12 re Aug 28 11:16:33 wb riot Aug 28 11:16:34 03xora 07org.oe.dev * r95e88a18... 10/ (1 packages/openmoko2/openmoko-dialer2_svn.bb): openmoko-dialer2_svn.bb : patch is now applied upstream Aug 28 11:17:21 * koen returns from lunch Aug 28 11:17:24 mickeyl: key added Aug 28 11:17:47 mickeyl: I do advice that those people use --author in conjunction with that key Aug 28 11:18:13 koen, the gumstix module installations has been due to ESSENTIAL_EXTRA versus plain EXTRA ... Aug 28 11:18:15 I hope Aug 28 11:18:25 koen: oe@company key? Aug 28 11:18:39 my keyboard is in writelikegerman mode Aug 28 11:20:08 hrw: company@oe key Aug 28 11:20:13 (same difference) Aug 28 11:21:19 mickeyl: will openmoko-today get a --no-phone mode? I would like to use it as mb-desktop-1 replacement for X images Aug 28 11:21:28 hi mallum Aug 28 11:21:37 hey florian Aug 28 11:21:48 hey mallum Aug 28 11:21:52 koen: heh, I thought about this too :-) Aug 28 11:22:32 florian: I'm digging out the gpe-theme-industrial to put mb-panel in the title bar as well Aug 28 11:23:52 koen: interesting idea. what would you like to see changed in such a mode? Aug 28 11:23:54 hey koen Aug 28 11:23:56 koen: Using matchbox-keyboard as the default would be good for most devices as well, the only drawback is that it needs much more time to start up... Aug 28 11:23:57 hi mallum Aug 28 11:24:12 hey mickeyl Aug 28 11:24:23 mickeyl: leave out hte 'missed calls' and dialer icon Aug 28 11:24:24 re Aug 28 11:24:25 well, -today is much more sexy, so i think a --no-phone mode makes sense Aug 28 11:24:41 koen: right. ultimately (and this i wanted to discuss with O-Hand soon), i want to see -today plugin-based Aug 28 11:24:48 then missed calls would just be a plugin Aug 28 11:25:24 mickeyl: in the mean time I'm working on a patch that makes only -today use the openmoko theme Aug 28 11:25:37 (ripped from gpe-login) Aug 28 11:26:06 koen: surely a patch is overkill for that? how about just presetting GTK_PARSE_RC2_FILES? Aug 28 11:26:18 ah, hmm Aug 28 11:26:22 didn't think of that :) Aug 28 11:26:32 it's one of the least documented env variables in gtk Aug 28 11:26:51 aren't all vars undocumented? Aug 28 11:26:59 heh Aug 28 11:27:01 kind of Aug 28 11:27:24 sorry, GTK2_RC_FILES even Aug 28 11:27:26 no PARSE Aug 28 11:29:40 first I need to have a kernel that has networking on my h2200 Aug 28 11:41:05 say, why is the font-size in mrxvt so huge? its set to the minimum for ttf, 8, but still that extremely large. I have the same problem in all my python-apps :( Aug 28 11:42:28 riot: you should patch it like we do in OE Aug 28 11:42:45 minimal font size is configurable Aug 28 11:43:08 8 is pretty large since we report our DPI to the xserver Aug 28 11:45:38 is the rootfs_tar class gone?? Aug 28 11:45:56 don't see it in classes/ Aug 28 11:46:24 and get errors like Could not inherit file classes/rootfs_tar.bbclass while parsing /org.openembedded.dev/packages/images/angstrom-minimal-image-with-mtd-utils.bb Aug 28 11:46:30 just did a pull/update Aug 28 11:46:34 any ideas? Aug 28 12:08:08 Jin^eLD: it was there 5 days ago Aug 28 12:09:46 03crofton 07org.oe.dev * r4990ba85... 10/ (1 conf/machine/include/gumstix.inc): gumstix.inc : Make sure RECOMMENDED modules get into minimal image. Aug 28 12:13:36 03crofton 07org.oe.dev * r4990ba85... 10/ (1 conf/machine/include/gumstix.inc): gumstix.inc : Make sure RECOMMENDED modules get into minimal image. Aug 28 12:13:45 03mickeyl 07org.oe.dev * re90a6a8f... 10/ (6 files in 2 dirs): Aug 28 12:13:45 sshfs-fuse: update to 1.8 Aug 28 12:13:45 fuse: remove 2.5.3 and 2.6.0, 2.6.5 proved to be pretty stable. may add 2.7.0 soon Aug 28 12:13:51 Crofton: uboot failed on git for me as well Aug 28 12:13:57 03mickeyl 07org.oe.dev * r01c82ba2... 10/ (1 packages/libsdl/libsdl-ttf-native_2.0.3.bb): libsdl-ttf-native: catch up STAGING_BINDIR changes Aug 28 12:14:10 03openmoko 07org.oe.dev * r715411a2... 10/ (1 packages/tasks/task-sdl.bb): task-sdl: add more packages Aug 28 12:14:45 mickey|lunch: please use --author when using the openmoko key Aug 28 12:48:36 steliosk: you have mail Aug 28 12:49:09 03koen 07org.oe.dev * r17c1f582... 10/ (1 packages/linux/linux-handhelds-2.6/h2200/defconfig): h2200: fix defconfig Aug 28 12:49:47 time to read the national geographic Aug 28 13:09:13 hi likewise Aug 28 13:09:24 likewise: well.. can you confirm it is still there? Aug 28 13:09:31 or is it gone for you too? Aug 28 13:27:19 morning Aug 28 13:27:52 koen, I think the u-boot repo comes and goes Aug 28 13:28:02 I rebuild it and it succeeded Aug 28 13:28:23 Aug 28 13:31:39 koen is rootfs_tar.bbclass gone? Aug 28 13:32:16 it's not in my tree Aug 28 13:32:23 hmm Aug 28 13:32:35 but pull && update should be enough to get everything, right? Aug 28 13:32:41 yes Aug 28 13:32:48 ok I will try again.. Aug 28 13:33:43 nope, nothing Aug 28 13:34:19 http://pastebin.ca/673487 Aug 28 13:34:25 nothing there.. how can this be? Aug 28 13:38:17 I wanted to suggest using bonsai, but that's giving error: 500 now :( Aug 28 13:41:14 what is bonsai? :) or were you not talking to me? Aug 28 13:42:02 mhm. does anyone have a rootfs_tar.bbclass file for me? :) Aug 28 13:42:14 I really wonder where it went and why I can not get it by pulling/updating Aug 28 13:44:42 http://openembedded.org/bonsai/ Aug 28 13:46:35 mtn list missing? Aug 28 13:47:22 Crofton: does not return anything Aug 28 13:47:36 so it does not think its missing... Aug 28 13:47:44 I really don't get it Aug 28 13:48:24 Crofton: that's actually the reason I was not able to try the davinci yet ;) failed to build anything because of the above and want to solve that first Aug 28 13:48:25 Hmm... www.openembedded.org suggests that it is unable to connect ot database. Aug 28 13:48:43 koen: bonsai url returned a 404 Aug 28 13:49:25 crofton: ping Aug 28 13:49:49 ...and then it suggest that I proceed to the install page. Being that sort of person, I did so. It's installed apparently, and now suggests that I proceed to the update script. I think perhaps I better stop clicking links before I make it worse. Aug 28 13:51:00 Jin^eLD: http://www.openembedded.org/bonsai/query/file/rootfs_/?page=2&scm_id=3 Aug 28 13:51:15 sakoman, pong Aug 28 13:51:38 koen: thanks Aug 28 13:51:48 Started a build last night with MACHINE_EXTRA_RRECOMMENDS Aug 28 13:51:55 Now see it was a waste :-) Aug 28 13:51:59 sorry Aug 28 13:52:08 But I did create the list of all modules Aug 28 13:52:19 So I'll edit & rerun Aug 28 13:52:20 it is one of those things that slowly became painfully obvious :) Aug 28 13:52:29 Yup Aug 28 13:52:32 there are a lot of modules Aug 28 13:52:38 Yes :-) Aug 28 13:52:40 koen: wait... rootfs_tar.bbclass is NOT listed there Aug 28 13:52:46 I saw the answer yesterday evening in task-boot Aug 28 13:52:47 It took a while even with scripts! Aug 28 13:52:57 i.e. - on the summary by the link you gave me Aug 28 13:53:05 Jin^eLD: right, like it never existed :( Aug 28 13:53:13 does gumstix build root install all the modules? Aug 28 13:53:20 or rather, hasn't been edited in a few years Aug 28 13:53:25 yes, they aren't really too big Aug 28 13:53:43 ok.. so... where did it go then? :) Aug 28 13:53:44 And they make support automatic for most of their add-ons Aug 28 13:54:05 I am thinking modules that support add-ons should be essential Aug 28 13:54:12 and other non-essential Aug 28 13:54:33 can someone who has a local copy check it in again or make it available somehow? or was it removed on purpose? Aug 28 13:54:35 sakoman: I'm going to wait for the scanner to arrive: http://www.flickr.com/photos/koenkooi/1256668324/ :) Aug 28 13:55:19 Jin^eLD: rootfs_tar doesnt exists as its impossible Aug 28 13:55:34 you can RRECOMMENDS = "kernel-modules" to get them all Aug 28 13:55:36 Jin^eLD: tar files contain no dependency data to build a rootfs from Aug 28 13:55:43 Nice photo! I even like the effect of the ghetto 'scan' :-) Aug 28 13:56:12 XorA: huh? ok why did I get that error message then that rootfs_tar was missing? or did I screw somethng up completely? :) Aug 28 13:56:14 koen: sheesh wish I'd known that! Aug 28 13:56:24 would have saved some work! Aug 28 13:56:29 and who/what does create bla_rootfs.tar.gz image then? Aug 28 13:56:32 ooho Aug 28 13:56:39 I get it Aug 28 13:56:47 Jin^eLD: because the strings are auto generated, so you can ask for a tar package based rootfs, but you cant really have one Aug 28 13:56:57 Have meetings early today so it will be a while before I can report success Aug 28 13:57:03 ok now I finally understand :) so the mistake is somewhere in the config Aug 28 13:57:05 Back later! Aug 28 13:57:06 doh.. me stupid Aug 28 13:57:08 thanks XorA :) Aug 28 13:57:23 Jin^eLD: I only know because lrg fell over this Aug 28 13:57:25 koean, can we name ESSENTIAL modules by name and just recommends all? Aug 28 13:57:27 :) Aug 28 13:58:52 Jin^eLD: if you need tar packages then INHERIT = "package_ipk package_tar" should "fix" the problem Aug 28 13:59:08 XorA: roger that Aug 28 13:59:28 btw, it would be good to add that to site/common-glibc too: ac_cv_func_realloc_0_nonnull=${ac_cv_func_realloc_0_nonnull=yes} Aug 28 13:59:36 the malloc stuff is already there but the realloc stuff is missing Aug 28 13:59:46 that helps against the rpl_realloc stuff Aug 28 13:59:58 I want to deploy package foo into a special ipkg feed separate from all, armv4t, . how can this be done? Aug 28 14:01:44 (reason: packages in my private tree (which are not in OE yet) should be placed separate from all the packages that have their BBs in OE) Aug 28 14:03:36 03hrw 07org.oe.dev * ra8eed604... 10/ (3 files in 3 dirs): nfs-utils: do not check for nfsd kernel module as it can be built-in into kernel. (Taken from Poky - r1395) Aug 28 14:03:44 solution: add your package tree into OE Aug 28 14:03:44 03hrw 07org.oe.dev * rb2b1b0bc... 10/ (3 files in 3 dirs): disapproval of revision 'a8eed6047c4764f6f475a519b127e6c00f07442e' Aug 28 14:03:50 03rwhitby 07org.oe.dev * r6fc40d7a... 10/ (1 packages/openocd/openocd_svn.bb): openocd: Enabled ft2232_libftdi so it can be used on a slug to devirginate a neo Aug 28 14:03:54 03hrw 07org.oe.dev * r54f440d5... 10/ (3 files in 3 dirs): nfs-utils: do not check for nfsd kernel module as it can be built-in into kernel. (Taken from Poky - r1395) Aug 28 14:04:00 03hrw 07org.oe.dev * r70b02d13... 10/ (3 files in 3 dirs): nfs-utils: Add patch to work with uclibc (Taken from Poky -r1796) Aug 28 14:04:05 03hrw 07org.oe.dev * r893542e4... 10/ (3 files in 3 dirs): nfs-utils: start/stop also statd so clients will be able to do locking (Taken from Poky -r2578) - close #2799 Aug 28 14:04:32 RP: angstrom-x11-image builds against uclibc nowadays Aug 28 14:17:57 * * OE Bug 2799 has been RESOLVED (FIXED) by Aug 28 14:17:59 * *  NFS Utils are broken - statd is not started so no locking Aug 28 14:18:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2799 Aug 28 14:39:07 koen: neat :) Aug 28 14:39:29 koen: Now the overrides problem is fixed, I should fixup poky too... Aug 28 14:51:56 CosmicPenguin: hi Aug 28 14:52:16 nfs-utils 1.1.0 req many extra libs ;( Aug 28 14:52:20 hi Aug 28 14:53:03 CosmicPenguin: got answer from vendor. they cant change timers locking in awardbios, but can in tinybios. and there is someone working on linuxbios Aug 28 14:55:00 wow is my old bitbake-1.5 snapshot slow :} Aug 28 14:55:16 hrw: humph... award could change them if they wanted to Aug 28 14:55:30 * zecke thinks Richard and me did a good job removing bogus loops from bitbake Aug 28 14:55:44 fruit loops? Aug 28 14:56:22 well, we made sure to keep the lovely and fruity loops in Aug 28 14:58:26 hi koen zecke Aug 28 15:03:00 hey chouimat Aug 28 15:38:36 long lunch mickeyl ? Aug 28 15:39:04 torpor: nah, he is normally days in his lunching Aug 28 15:39:54 you know - that business lunches with breakfast, girls and lots of drinking.. Aug 28 15:40:01 w00t! Aug 28 15:40:05 girls are lunch. Aug 28 15:41:30 :D Aug 28 15:43:08 nfsv4 is space expensive ;( Aug 28 15:43:29 ~150KB extra Aug 28 15:44:52 can avahi run on ARM based linux? Aug 28 15:45:23 I just installed the new cups 1.3 which can advrtize the printers to a client Aug 28 15:45:42 and connecting to a printer is so easy on the client side Aug 28 15:45:57 niv_one_three: angstrom defaults with avahi included Aug 28 15:46:07 sweet Aug 28 15:46:13 niv_one_three: I see it start on my devices, but I have never used it Aug 28 15:46:17 did anyone actually tried it? Aug 28 15:46:23 avahi is great Aug 28 15:46:36 ssh alix.local (nevermind which IP alix got) Aug 28 15:46:40 ya hrw, your blog pointed me to this :-) Aug 28 15:46:45 ;D Aug 28 15:46:56 where is koen? Aug 28 15:47:36 in the Netherlands ... Aug 28 15:49:23 do I need to run the deamon on each pc? Aug 28 15:49:35 or does the client does it all Aug 28 15:50:19 need Aug 28 15:59:58 cbrake: so, I got the image built and it loads up to the gpe login. Aug 28 16:00:22 christopher: excellent :-) Aug 28 16:00:24 cbrake: of course, that's the good news. bad news is that no text is shown. just boxes. guessing it has to do with pango-modules that are in tehre Aug 28 16:00:35 cbrake: I have tons of messages about no modules.dep being found Aug 28 16:01:04 christopher: are you using the kernel built with OE? Aug 28 16:01:19 cbrake: I am not. I have a kernel that's already working with the unit. Aug 28 16:01:33 cbrake: it has the nand flash changes etc. Aug 28 16:01:48 christopher: the kernel version probably does match exactly, so that is why it can't find the right modules directory then Aug 28 16:02:29 cbrake: yeah, that's true as I checked for that... Aug 28 16:02:37 cbrake: but I still don't see a modules.dep file. Aug 28 16:02:43 cbrake: do I copy that over? Aug 28 16:03:07 christopher: modules.dep is generated Aug 28 16:03:22 cbrake: and put on the fs right? Aug 28 16:03:35 christopher: but, if you don't have the right /lib/modules directy, it can't be generated Aug 28 16:04:06 cbrake: oh, so generated at run time. gotcha. Aug 28 16:04:27 cbrake: so, I can change that. I know I put in a special code for it so, I can just build 2.6.20 instead of 2.6.21 Aug 28 16:04:34 cbrake: I didn't think it would be a big deal Aug 28 16:04:49 cbrake: but you know what they say about ASSuME'ing Aug 28 16:05:17 cbrake: so, that would explain pango right? Aug 28 16:05:28 christopher: I don't know. I've not loaded GPE lately Aug 28 16:05:38 cbrake: what do you use? Aug 28 16:06:16 christopher: my customer projects mostly just run a single GTK app on top of X -- no window manager Aug 28 16:06:52 christopher: and my openmoko phone -- whatever they put together ... Aug 28 16:07:09 cbrake: I think that is how we'll want to do it in the end, but I am not sure of how to start with that either. are there any sample apps in oe for a gtk demo? Aug 28 16:07:34 christopher: you can run any GTK app. Just start Xwindows, and then run the GTK app Aug 28 16:07:35 cbrake: I am a little concerned that the graphics may not be 'graphical' enough for our customers and that is a concern of mine Aug 28 16:07:50 cbrake: ok. Aug 28 16:08:33 cbrake: so, when you build an image, do you just use the minimum image or have you just cut out of the x11 image? Aug 28 16:09:03 christopher: I typically start with angstrom-console-image, and then add X/GTK and whatever else I need Aug 28 16:09:50 cbrake: ok. Aug 28 16:10:16 cbrake: we will probably end up with something like that. figured it would be nice to see something like gpe or matchbox run so I can get an idea of what the processor can handle Aug 28 16:10:25 cbrake: so far, not so good. :) Aug 28 16:11:00 christopher: yes, that would make a nice wizbang demo Aug 28 16:14:35 cbrake: X-Window, not XWindows Aug 28 16:15:32 time to go home Aug 28 16:15:34 bye bye Aug 28 16:16:04 cbrake: are most of the products you do for industrial type use where it is function over form? that is, just some boxes for touchscreen etc? Aug 28 16:16:29 hrw: Hmm, and I've been saying it wrong all these years ... Aug 28 16:16:40 03mickeyl 07org.oe.dev * r266f6beb... 10/ (1 packages/tasks/task-openmoko-feed.bb): add task-openmoko-feed, things we want to have in the feed Aug 28 16:19:16 christopher: yes, most of my customers are industrial. These systems tend to only run one application. But, I'm starting to get requests to add web browsers and such once people learn what all you can do with a Linux platform. Aug 28 16:19:49 cbrake: ok. yeah, that's neat. I don't have enough experience with it to know what we can or can't do with our hardware. Aug 28 16:20:00 cbrake: most of our customers will want more graphics and whizbang stuff... Aug 28 16:20:23 cbrake: and I'm concerned that I may not either a.) be able to find something that is already done or b.) be able to do it myself. :) Aug 28 16:21:57 christopher: well, hopefully webkit will eventually be stable, and you can do a lot with GTK as far as applications Aug 28 16:22:52 cbrake: I hadn't heard of webkit. I heard about things like FLTK with nano-x. what does it give to you? widgets etc? Aug 28 16:23:30 christopher: webkit is a browser engine Aug 28 16:24:10 christopher: these days, I would use GTK+X for graphics on embedded Linux systems. It is main stream and very well supported. Aug 28 16:24:27 christopher: although a little slow, but works well enough Aug 28 16:25:25 cbrake: ok...that sounds good. I am concerned about the slowness of it...we have an arm9 so it should be fairly good Aug 28 16:25:53 christopher: a lot depends on your screen size, and processor Aug 28 16:29:40 cbrake: yeah, we have 15inch and 7 inch up to 1024x768 Aug 28 16:41:25 cbrake: thanks for the help. I'll build the other version of linux and give it a whirl Aug 28 16:47:14 bleh Aug 28 16:47:27 running a powerpc photoshop on an intel machine is no fun Aug 28 16:50:52 ~lart gimp for being 8bit Aug 28 16:50:52 * ibot calls gimp on the phone ... the lights are on but nobody's home for being 8bit Aug 28 16:54:16 sakoman: now with a piece of white plastic as backing: http://www.flickr.com/photo_zoom.gne?id=1258783265&size=l Aug 28 16:55:25 koen: nice! new scanner there yet? Aug 28 16:55:38 sakoman: no, delayed a few days Aug 28 16:55:56 * koen mumbles something about stores not keeping a good stock inventory Aug 28 16:56:02 bummer Aug 28 16:58:51 yeah Aug 28 17:04:51 bye Aug 28 17:13:58 * * OE Bug 2882 has been REOPENED by Aug 28 17:13:59 * *  u-boot-git-r1-do_fetch Aug 28 17:14:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2882 Aug 28 17:14:41 christopher: 1024x768 is getting pretty large to drive with low end ARM procs Aug 28 17:32:59 koen, now I can reproduce the error :( Aug 28 17:35:53 RP: ping Aug 28 17:38:44 gm khem Aug 28 17:38:53 hey Crofton Aug 28 17:39:32 have I committed all the kernel stuff you posted? Aug 28 17:39:45 Crofton: I was looking at some of dependency stuff do u know about this ? Aug 28 17:39:50 Crofton: so far yes Aug 28 17:39:55 what kind of dependency stuff? Aug 28 17:42:03 +#PREFERRED_VERSION_glibc = "2.6.1" Aug 28 17:42:04 +#PREFERRED_VERSION_glibc-intermediate = "2.6.1" Aug 28 17:42:04 +#PREFERRED_VERSION_glibc-initial = "2.6.1" Aug 28 17:42:25 if set it creates problem when you chose eglibc Aug 28 17:42:26 heh, this I do not understand well Aug 28 17:42:47 you are working in areas I have not had to figure out yet Aug 28 17:42:48 I was wondering why is glibc being pulled in eglibc based build Aug 28 17:43:16 so I wanted to understand how these variables are treated by bb Aug 28 17:43:26 heh ok Aug 28 17:43:40 at some point you have to start looking in the code ... Aug 28 17:43:56 I have been working out tasks that way Aug 28 17:44:07 at bb code you mean ? Aug 28 17:44:34 and classes in org....dev/classes Aug 28 17:44:48 I am not sure where that stuff lives Aug 28 17:45:05 I did try making the dot file of dependencies and looking at it Aug 28 17:45:20 but the drawing was really huge and did not work well :( Aug 28 17:45:44 tell me this Aug 28 17:46:00 if I set PREFERRED_PROVIDER_virtual/arm-angstrom-linux-gnueabi-libc-for-gcc = "glibc-intermediate" Aug 28 17:46:09 and then later on set it again Aug 28 17:46:14 PREFERRED_PROVIDER_virtual/arm-angstrom-linux-gnueabi-libc-for-gcc = "eglibc-intermediate" Aug 28 17:46:18 which one is honored Aug 28 17:46:26 I just don't know Aug 28 17:50:32 I see some packages have RDEPENDS += "glibc- ....." Aug 28 17:50:44 ah Aug 28 17:50:54 thats probably not correct Aug 28 17:51:04 they should depend on virtual package Aug 28 17:51:06 needs a virtual? Aug 28 18:03:31 good evening Aug 28 18:05:32 khem: pong Aug 28 18:05:36 hi pH5 Aug 28 18:06:06 RP: I am having some problem with glibc being pulled even when I select ANGSTROM_MODE="eglibc" Aug 28 18:06:24 RP: the problem is realted to unmet deps Aug 28 18:06:57 RP: I search for RDEPENDS and I see some packages mentioning REDEPENDS += "glibc" Aug 28 18:06:59 khem: At a guess I'd say eglibc isn't providing everything glibc provides and it therefore pulls in glibc for those things Aug 28 18:07:15 hi ph5 Aug 28 18:07:39 khem: Right an RDEPENDS on glibc will cause glibc to get built... Aug 28 18:07:42 RP: I think it provides everything except the locales which I have disabled Aug 28 18:07:43 hi woglinde, RP Aug 28 18:08:01 RP: So I will make them virtual would that be ok Aug 28 18:08:40 khem: "virtual" entities in the runtime namespace cause other weird problems :/ Aug 28 18:08:58 hmm Aug 28 18:09:02 khem: Which package has this RDEPENDS? Aug 28 18:09:34 RP: few Aug 28 18:11:36 khem: I'd be tempted to have the packaging names for eglibc match that for glibc or at least RPROVIDE their glibc counterparts Aug 28 18:11:51 RPROVIDE and RCONFLICT I guess Aug 28 18:12:32 RP: Can I do it in bb file Aug 28 18:12:39 No distro matinainer is ever going to have two different glibcs in their feed (or if the do, they need their sanity checking :) ) Aug 28 18:12:59 RP: Right Aug 28 18:13:13 khem: Sure RPROVIDES_eglibc = "glibc" Aug 28 18:13:16 * mwester makes a note to question his sanity... Aug 28 18:13:19 thats why I want to make sure no traces of glibc are left if eglibc is used Aug 28 18:13:51 RP: ah that means eglibc provides glibc Aug 28 18:14:10 sakoman, did you get the ethernet working on your connex? Aug 28 18:14:18 khem: Right, so anything RDEPENDS = "glibc" will then use eglibc instead Aug 28 18:14:26 RP: cool Aug 28 18:14:30 RP: Iwill try that Aug 28 18:14:34 khem: Since its a direct and totally compatible replacement, that should be accetable Aug 28 18:14:43 RP: I totally agree Aug 28 18:14:46 khem: I just hope I taught bitbake about RPROVIDES ;-) Aug 28 18:15:02 We will see Aug 28 18:15:12 If not, thats a bitbake bug :/ Aug 28 18:16:06 RP: whats difference between PROVIDES and R variant Aug 28 18:16:33 R for runtime Aug 28 18:16:36 hehe Aug 28 18:16:41 oh man ! Aug 28 18:17:02 R is when system is running you mean ? Aug 28 18:17:29 I though we only has buildtime Aug 28 18:17:34 no Aug 28 18:17:55 we also have runtime deps and suggestions Aug 28 18:17:59 khem: The r* runtime namespace maps into the world of package managers Aug 28 18:18:12 hmmm I see Aug 28 18:18:14 khem: RDEPENDS becomes a debian Dependcies field Aug 28 18:18:28 RPROVIDES becomes debian Provides: Aug 28 18:18:29 etc Aug 28 18:18:31 RP: I see thats whats failing for me righr now Aug 28 18:18:41 and PROVIDES? Aug 28 18:18:47 buildtime Aug 28 18:18:58 think about java compiler Aug 28 18:19:02 khem: PROVIDES is a buildtime issue and just something bitbake cares about Aug 28 18:19:34 RP: OK so I think I should set both Aug 28 18:20:12 khem: Do you really need to? The virtual/libc that both provide already should take care of PROVIDES Aug 28 18:20:41 khem: People are going to get really annoyed if they bitbake glibc and eglibc builds ;-) Aug 28 18:21:05 You need to think about the different namespaces very differently and carefully Aug 28 18:21:18 RP: oops yes I usually have the virtuals there so prolly dont need DEPEND Aug 28 18:22:04 RP: I think in OE it will be eglibc and no glibc soon Aug 28 18:22:30 khem: Maybe, maybe not. glibc has managed well enough for a long time ;-) Aug 28 18:22:44 khem: It certainly won't get removed outright Aug 28 18:22:48 slowly but surely :) Aug 28 18:23:14 khem: THe locale generation doesn't take long if you limit the number of locales it builds Aug 28 18:23:32 khem: The length of time "make install" takes is a much bigger problem :/ Aug 28 18:23:38 RP: I will use the cross locale generation soon Aug 28 18:31:30 RP: inside eglibc.bb I will just need RDEPENDS = "glibc" right ? Aug 28 18:32:12 khem: No, that wasn't what I said Aug 28 18:33:16 hmmm Aug 28 18:33:44 khem: That would mean that when you run eglibc binarys, they need glibc to run Aug 28 18:33:52 yeah realised that Aug 28 18:34:32 may be RPROVIDES = "glibc" Aug 28 18:34:38 is the right think Aug 28 18:34:44 (s/think/thing/) correction: is the right thing Aug 28 18:35:07 khem which package sets RDEPENDS = glibc? Aug 28 18:35:26 khem: close. You want RPROVIDES_eglibc = "glibc" as I said before Aug 28 18:35:27 abiword has some glibc locale Aug 28 18:35:41 khem: You might also want things like RPROVIDES_eglibc-utils = "glibc-utils" Aug 28 18:35:59 yeah it will be some more actually all of provides Aug 28 18:36:16 that glibc has needs to be provided by eglibc Aug 28 18:36:23 03rpurdie 07org.oe.dev * r70f25943... 10/ (1 packages/blktool): Add blktool 4 (from Poky) Aug 28 18:37:33 RPROVIDES_eglibc-initial = "glibc-initial" inside eglibc-initial.bb Aug 28 18:37:41 is that ok ? Aug 28 18:37:56 I thught it would pick up the package name automatically Aug 28 18:40:04 khem: No, what RDEPENDS on glibc-initial? Aug 28 18:40:23 just an example I was asking Aug 28 18:40:54 khem: RPROVIDES and all the other R* variables behave like RDEPENDS. They don't pick up names automatically since there is no way of knowning what you're trying to do... Aug 28 18:41:55 khem: If eglibc-initial produces a package called eglibc-initial, yes. Since it has PACKAGES = "", its not something you'd do though Aug 28 18:42:26 yes ok Aug 28 18:42:29 got it Aug 28 18:45:06 RP: what is PACKAGES_DYNAMIC Aug 28 18:47:12 asmola: not yet, but getting close! Aug 28 18:47:18 hopefully later today Aug 28 18:47:40 lots of multi-tasking today :-( Aug 28 18:51:28 RP: will regexps work with RPROVIDES? Aug 28 18:52:00 * khem thinks probably not Aug 28 18:54:29 sakoman, ok, i am working on it here as well Aug 28 19:00:33 khem: no, they won't Aug 28 19:00:58 khem: You can inject the RPROVIDES into the code when the packages are generated though Aug 28 19:01:21 how do I do that Aug 28 19:01:45 khem: You'll have to play around with split_packages and see if it provides a way and if not, add it Aug 28 19:03:20 ok Aug 28 19:03:23 will do Aug 28 19:03:40 for now I am doing a clean build with the new RPROVIDES added Aug 28 19:06:11 sounds good. Its best to take these things one issue at a time so you understand the changes Aug 28 19:06:21 yes Aug 28 19:06:34 thats been my approach Aug 28 19:06:56 I have shot my own foot so many times with doing multiple things in one go Aug 28 19:11:49 I have a newbie Q: I've got everything checked out, and set myself up a local.conf, and can bitbaked angstrom-minimal-image Aug 28 19:12:00 this is targetting gumstix-verdex using latest HEAD of the .dev branch Aug 28 19:12:17 with some mods to gumstix-verdex brought in from the gumstix buildroot stuff Aug 28 19:12:20 anyway Aug 28 19:12:45 the rootfs has a boot/uImage inside it, but does not have /lib/modules/x.y.z Aug 28 19:12:58 the modules are getting tarred up in build/tmp/deploy Aug 28 19:13:06 I have 2 Qs, basically Aug 28 19:13:33 1. How can I get the modules to actually be installed (looks like it's also making about a hundred .ipkgs out of them) in the rootfs image? Aug 28 19:14:04 2. How can I get the uImage to not be installed in the rootfs, but just copied as a separate file into tmp/deploy Aug 28 19:14:05 hughescr: You need an image which includes the modules in it. The minimal image only includes things absolutely essential to boot Aug 28 19:14:45 RP: ah, ok -- so I then tried angstrom-console-image and that seemed to want to include X, gnome, and all kinds of stuff Aug 28 19:14:51 which was not expected, given the "console" in there Aug 28 19:15:03 hughescr: It would build them, that doesn't mean they'd end up in the image Aug 28 19:15:05 so do I need to build my own image .bb? Aug 28 19:15:12 RP: ah Aug 28 19:15:15 hughescr: Probably not Aug 28 19:15:27 I expected it'd be somewhat conservative about building only what was needed for an image Aug 28 19:15:40 hughescr: To copy to a file in deploy, the kernel needs a do_deploy task like the ones lots of other kernels have Aug 28 19:16:01 yeah, so it is copying it Aug 28 19:16:08 but it's also sticking it in rootfs/boot Aug 28 19:16:19 I can't figure out where it's doing that 2nd bit, to stop that Aug 28 19:16:29 hughescr: The problem was talked about recently, I think we've enabled options in some of the tools which end up pulling in X. If you don't want to do that, you could disable the options that are pulling that in Aug 28 19:17:07 by "disable options" does that mean change MACHINE_FEATURES or whatever that var is? Aug 28 19:17:09 hughescr: It packages the kernel then installs the package. Some machines stop it doing that like the zaurus ones. I think the magic to do it is somewhere in linux-rp.inc Aug 28 19:17:36 hughescr: No, you'd have to change the DEPENDS and configure commandlines for the tools in quesiton. I don't know which they are offhand though Aug 28 19:18:27 ah, gotcha, so basically angstrom-console-image is doing something like including package foo, which DEPENDS or maybe just build-depends on some X thing Aug 28 19:18:37 hughescr: The magic in linux-rp.inc is the FILES_kernel-image = "" line. We have a better way of doing that brewing but not applied yet Aug 28 19:18:52 I haven't been able to figure out also, where the mapping of image name -> included packages happens Aug 28 19:19:06 RP: ok, thx I'll check it out Aug 28 19:19:33 * koen has a dejavu :) Aug 28 19:19:49 s/RP/hrw/ and s/hughesrc/crofton/ :) Aug 28 19:19:59 koen: I wasn't paying attention earlier :) Aug 28 19:20:22 I just went unafk as well :) Aug 28 19:22:30 * RP finds two minutes to spend on linux-rp... Aug 28 19:22:43 koen: Did you try bb-trunk btw? Aug 28 19:25:59 Hi Craig! Aug 28 19:28:29 sakoman, were you able to build from clean today for the connex? mine died out at the very end complaing about an 'update-modules' dependency for 'task-boot' Aug 28 19:28:36 (building for verdex) Aug 28 19:28:38 hi steve Aug 28 19:29:11 asmola: yes I get a clean build, but have a modified gumstix.inc Aug 28 19:29:14 btw, does monotone let one have multiple remote repos like git does? Aug 28 19:29:31 ie so others can pull from my repo onto a repo checked out from the main OE one? Aug 28 19:29:39 programming flash with the image right now Aug 28 19:30:08 sakoman: just emailed you my local.conf and minor mods for better verdex/connex optimization Aug 28 19:30:22 can you forward that to me? :) Aug 28 19:30:23 sakoman: and MACHINE_FEATURES Aug 28 19:30:26 OK, I'll take a look Aug 28 19:31:48 asmola: where should I send them? Aug 28 19:31:54 adam.smola@gmail.com Aug 28 19:32:45 thanks craig, i have been one of the other people on here trying out stuff on the verdex and providing feedback to the guys doing the real work :) Aug 28 19:34:18 comrades Aug 28 19:34:31 where can I define which locales du in/exclude? Aug 28 19:34:36 during build Aug 28 19:36:34 re: iwmmxt, the general feeling here was against require tune-wmmxt because it causes runtime issues with some programs and that it should use the standard tune by default, with that being said, iwmmxt support is making it into the kernel so an application will be able to make use of it if it was built that way Aug 28 19:37:00 verdex in buildroot's been using iwmmxt (including the kernel option) since day 1 Aug 28 19:37:02 im largely echoing the sentimments of others here, i'm far too green to have any experience of my own Aug 28 19:37:59 I've not noticed any problems, but then again, there's a lot more pkgs in OE than buildroot Aug 28 19:38:22 but my inclination would be to fix any problems, rather than waste the advantage of the 270 over 255 Aug 28 19:39:21 or if needed, disable iwmmxt on a package-by-package basis Aug 28 19:39:52 i think that's the key, knowing which apps have issues and flagging them as such Aug 28 19:39:57 but at least, the tune include should be in the board-specific file, in case, say, we release a gumstix which is not an xscale :) Aug 28 19:40:07 :) Aug 28 19:40:20 is that a pre-announcement? Aug 28 19:40:23 :-) Aug 28 19:40:27 nah Aug 28 19:40:30 haha Aug 28 19:41:58 hughescr: Crofton has been a major hero in getting the gumstix build as far along as it is Aug 28 19:42:24 i couldn't agree more Aug 28 19:42:30 I sent you his address, he's been working without hardware :-) Aug 28 19:42:35 :) Aug 28 19:43:14 sakoman, let me known how your flash goes Aug 28 19:43:34 doing katinstall right now Aug 28 19:43:37 sakoman: yup -- in the works Aug 28 19:43:38 awesome Aug 28 19:43:58 He'll be happy and I think you will be too :-) Aug 28 19:44:00 sorry to bother here, but I changed the kernel version I had on a distro from 2.6.21 to 2.6.20. however, in the rootfs, it is still creating the lib/modules with a 2.6.21 directory. I checked the tmp/work directory and even removed the 2.6.21 directory. does anyone know what I may need to rebuild or check config on for this? Aug 28 19:45:52 asmola, hughescr: no joy Aug 28 19:46:05 sakoman: failure mode? Aug 28 19:46:20 being bitten by the mount bug: option size not supported Aug 28 19:46:54 This time the boot doesn't complete so I can't patch /etc/init.d/udev and reboot Aug 28 19:47:13 sakoman: what if you pass init=/bin/sh in bootargs? Aug 28 19:47:35 How do I do that :-) Aug 28 19:47:45 In u-boot Aug 28 19:47:57 set bootargs $bootargs init=/bin/sh;bootd Aug 28 19:48:12 OK, will try Aug 28 19:48:49 aka "you don't need my password" mode Aug 28 19:49:16 Good suggestion. I get a prompt now Aug 28 19:49:27 Let me patch and reboot Aug 28 19:49:34 is probably still mounted ro Aug 28 19:49:46 but at least you should be able to poke around a bit Aug 28 19:49:55 '/' got eaten Aug 28 19:49:57 stupid irc Aug 28 19:52:34 didn't seem to be ro, let me edit Aug 28 19:53:16 OK did a "normal" boot with the patch Aug 28 19:53:45 seems to be suffering from the lack of the modules I mentioned in my email to you Craig Aug 28 19:53:56 I'll post the boot log in a moment Aug 28 19:56:08 asmola, hughescr: http://www.sakoman.net/oe/boot3.txt Aug 28 19:56:08 in buildroot, the modules.alias etc stuff gets built at kernel "install" time by calling the busybox depmod.pl Aug 28 19:56:25 then depmod -a is also run on boot just in case Aug 28 19:56:30 it should rebuild all that stuff, I think Aug 28 19:56:41 Thanks! Gives me a place to start Aug 28 19:57:05 bk Aug 28 19:57:09 I think I had to patch depmod.pl at some point -- check the buildroot busybox package patches Aug 28 19:57:22 Hey Crofton! Aug 28 19:57:43 this global economy is crazy ,,,,, Aug 28 19:58:27 Ouch! you made me look at stock prices :-( Aug 28 19:58:35 urg Aug 28 19:58:41 * Crofton is afraid Aug 28 19:59:00 short scox Aug 28 19:59:04 that is my advice Aug 28 19:59:05 Crofton, meet Craig (hughescr) from gumstix Aug 28 19:59:15 * Crofton weeps at his portfolio Aug 28 19:59:17 You are both my heros :-) Aug 28 19:59:23 hufnus, good to meet you Aug 28 19:59:26 er Aug 28 19:59:31 hughescr, hello Aug 28 19:59:39 er craigh that is Aug 28 19:59:52 Crofton is so upset he can't type! Aug 28 20:00:24 my angstrom-minimal-image build for gumstix-verdex is dying right near the end Aug 28 20:00:25 http://pastebin.com/m10e5329 Aug 28 20:00:27 ok, I get stuck in a meeting for two hours Aug 28 20:00:34 Now I am too Aug 28 20:00:59 koen, ping? Aug 28 20:01:03 pong Aug 28 20:01:26 can you take a quick look at asmola's pastebin Aug 28 20:01:36 it seems like we are missing a dependency? Aug 28 20:01:43 bitbake update-modules -c rebuild? Aug 28 20:01:52 yes Aug 28 20:01:55 that makes sense Aug 28 20:01:59 this was from a clean build Aug 28 20:02:09 that i got the error Aug 28 20:02:21 first lets see if rebuilding update-modules then making image works Aug 28 20:02:24 I'm gonna head into the office -- laptop battery is about to croak at the cafe here bbiab Aug 28 20:02:34 then we can fix build files Aug 28 20:02:41 Crofton|work, sounds good Aug 28 20:02:47 good to meet you craig Aug 28 20:03:37 Crofton: I'm real close now Aug 28 20:03:52 I'll bet Aug 28 20:04:02 you are keeping your list of hacks up to date? Aug 28 20:04:07 How do I get depmod to run to create modules.dep? Aug 28 20:04:14 Very few hacks Aug 28 20:04:22 good, the goal is zero Aug 28 20:04:47 module_autoload_ohci-hcd_omap5912osk = "ohci-hcd" Aug 28 20:04:49 Actually, zero hacks Aug 28 20:04:53 is a line in linux.inc Aug 28 20:05:01 Just add this to gumstix.inc: Aug 28 20:05:06 MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS = " \ Aug 28 20:05:06 sysfsutils \ Aug 28 20:05:06 module-init-tools \ Aug 28 20:05:06 update-modules \ Aug 28 20:05:06 kernel-modules \ Aug 28 20:05:06 " Aug 28 20:05:14 Crofton|work, that worked Aug 28 20:05:51 so task-boot needs update-modules Aug 28 20:06:07 Yup, could go there :-) Aug 28 20:06:29 I suspect no one has fed task-boot ESSSENTIAL_RRECOMMENDS Aug 28 20:06:30 But still not generating modules.dep & its bretheren Aug 28 20:06:37 I think hrw knows alot about that Aug 28 20:06:42 brb Aug 28 20:07:09 ok, I am guessing the problem lies in task-boot not handling module installation properly Aug 28 20:07:25 any suggestions? Aug 28 20:07:31 thinking Aug 28 20:07:38 I am betting that task-base does Aug 28 20:07:41 sakoman: I'm haivng the same issue with the modules.dep Aug 28 20:07:54 Misery loves company :-) Aug 28 20:07:57 christopher, on what machine/image? Aug 28 20:08:22 this is sarge-at91, angstrom-x11-image Aug 28 20:08:49 but I may have messed something up at one point. I know that I didn't have the kernel image the same but now have that, but it is still making the 2.6.21 directory instead of 2.6.20 Aug 28 20:09:02 I'm looking around for update-modules to be sure everything is set correctly Aug 28 20:09:52 christopher: I think ipkg will have preferred the "newer" ipks to the 2.6.20 ones. You need to remove the 2.6.21 files from tmp/deploy/ipk and then rebuild the kernel Aug 28 20:10:23 RP: oh, didn't remove those, only removed the directories in tmp/work ...etc Aug 28 20:14:33 crofton|work: I had taken out update-modules from the angstrom conf file. I'm going to rebuild along with removing the ipks and see how that affects it. So, it may be a different thing altogether Aug 28 20:15:01 Crofton: a clue -- when I put the kernel-modules in RDEPENDS in angstrom-minimal-boot.bb the depmod stuff got doen properly Aug 28 20:15:03 yeah Aug 28 20:15:22 I wish hrw was awake :) Aug 28 20:16:04 So why wouldn't a minimal boot *want* to have kernel modules included? Aug 28 20:16:17 so they need to go in ESSENTIAL_EXTRA_RRECOMMENDS to get on to the image, but tha modules are not processed properly Aug 28 20:16:38 well, I think you can split between essential to running modules Aug 28 20:16:45 and modules that would be nice to have :) Aug 28 20:16:55 yeah, you're right Aug 28 20:17:02 like how many people really want the netfilter modules Aug 28 20:17:09 :) Aug 28 20:17:12 seven Aug 28 20:17:30 well, the adhoc networking guys would be interrested Aug 28 20:17:32 :) Aug 28 20:17:37 sakoman: yeah, I'd have thought a minimal boot would want to have anything from linux.config in it Aug 28 20:17:57 well, there are two issues Aug 28 20:18:08 one getting the stuff we need in to the minimal boot Aug 28 20:18:22 and what the minimal set is :) Aug 28 20:18:29 right Aug 28 20:18:29 working is step one Aug 28 20:18:31 :) Aug 28 20:18:37 working is good! Aug 28 20:18:54 Crofton: yeah, I suppose so Aug 28 20:19:03 the best thing to do, might be bugzilla this and let me talk to hrw in the morning Aug 28 20:19:14 I think he did the task-boot/task-base stuff Aug 28 20:19:23 We really need to figure out why mount won't take size as a valid option for tmpfs Aug 28 20:19:24 in the gumstix case, "minimal" is particularly messy, since "gumstix" doesn't specify what combo of daughtercards might be connected Aug 28 20:19:33 he is in poland, so is up earlier Aug 28 20:19:35 it's a bit like "minimal" for x86, more than embedded Aug 28 20:19:39 craigh, exactly Aug 28 20:19:41 craigh, i was just going to mention something along those lines Aug 28 20:19:57 my suggestion is all modules that provide hw support go in minimal Aug 28 20:20:02 sakoman: that rings a bell if it's tmpfs Aug 28 20:20:14 iirc that might be cos you're using a different tmpfs or something Aug 28 20:20:17 angstrom pretty much needs tmpfs Aug 28 20:20:28 hard coded in Aug 28 20:20:38 istr that tmpfs used to be shmfs and had fewer options available or something Aug 28 20:20:48 which tmpfs are you using craigh Aug 28 20:20:54 craigh, that sounds about right Aug 28 20:21:08 sakoman: the full-blown one Aug 28 20:21:08 I had to flip one of the config options from M to Y Aug 28 20:21:20 tmpfs is compiled in Aug 28 20:21:50 guys, I need to do some editting now, just lost two hours to a meeting Aug 28 20:22:04 I will chat with hrw in the AM Aug 28 20:22:44 OK, later! Aug 28 20:22:49 ciao Aug 28 20:22:51 later crofton Aug 28 20:23:00 thanks for debugging all this stuff! Aug 28 20:23:32 sakoman: just added your MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS to gumstix.inc and re-bitbaked and still get empty rootfs/lib/modules/2.6.21 Aug 28 20:23:40 do I need to force something to rebuild? Aug 28 20:23:51 what did you change? Aug 28 20:24:44 forcing rebuilds after changin stuff at this level is tricky Aug 28 20:24:50 craigh: I haven't yet figured out what is safe to change without a clean build :-) Aug 28 20:24:53 added some stuff like kernel-modules to MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS in gumstix.inc, having previously built ang-min Aug 28 20:25:08 bitbake -c clean task-boot Aug 28 20:25:15 the make the image Aug 28 20:25:45 I then bitbake angstrom-minimal-image again, and it didn't seem to do anything with the kernel-modules Aug 28 20:26:06 it does seem to have at least done something with the sysfsutils, module-init-tools, etc Aug 28 20:26:07 task-boot is sucking in the modules :) Aug 28 20:26:29 nice! Aug 28 20:26:32 read angstrom-minimal-image.bb and tasks/task-boot :) Aug 28 20:26:54 alright, now I must minimize this window Aug 28 20:26:58 haha Aug 28 20:27:05 bye :) Aug 28 20:27:10 bye Aug 28 20:28:00 bye crofton Aug 28 20:28:47 I made a little guide for moving from gpe to mokoui2 on angstrom - where can I post this? Aug 28 20:29:13 craigh: I'm trying another build with a hacked angstrom-minimal-image in order to get the modules.dep done properly Aug 28 20:30:12 After we see if I can get a good boot image we can work with Crofton to do it the proper way Aug 28 20:30:31 hmm, verdex isn't booting anymore with angstrom-minimal-image :( Aug 28 20:30:46 although it looks like the network stuff may be working??? Aug 28 20:30:57 two steps forward, one back Aug 28 20:30:59 asmola: log? Aug 28 20:31:35 hughescr: any further thoughts on the tmpfs issue? Aug 28 20:31:42 lhttp://pastebin.com/m100b2c16 Aug 28 20:31:43 sakoman: just a sec Aug 28 20:31:51 sakoman: did you have to define kernel-modules somewhere? Aug 28 20:32:01 no Aug 28 20:32:23 http://pastebin.com/m453399d9 Aug 28 20:32:35 it will pick up all ipkg's that start with kernel-module and put them in the rootfs Aug 28 20:33:06 oh, neat Aug 28 20:33:11 so why's it not? Aug 28 20:33:13 :) Aug 28 20:33:34 Don't know! It works for me :-) Aug 28 20:33:45 What's your gumstix.inc look like? Aug 28 20:36:07 hang on Aug 28 20:36:45 http://pastebin.com/m495068be Aug 28 20:37:07 also tried with RRECOMMENDS instead of RDEPENDS Aug 28 20:37:15 in line 24 Aug 28 20:37:39 it is processing sysfsutils, module-init-tools and update-modules ok Aug 28 20:37:45 just not doing anything with kernel-modules Aug 28 20:38:48 Hmmm . . . I use MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS Aug 28 20:41:09 aha, bitbake task-boot -c rebuild Aug 28 20:41:16 seems to have forced it to rebuild the right bits Aug 28 20:41:29 when followed by bitbake angstrom-minimal-image Aug 28 20:41:36 got all the modules in there now looks like Aug 28 20:41:53 Cool! Including modules.dep? Aug 28 20:42:06 nope Aug 28 20:42:13 need to figure out why it's not doing a depmod Aug 28 20:42:23 is that what update-modules is supposed to do? Aug 28 20:42:49 That's the issue we were starting to work when you joined us :-) Aug 28 20:43:21 Seems something is broken in angstrom-minimal-boot with the kernel-modules installation Aug 28 20:44:22 it doesn't seem to have done update-modules *after* kernel-modules Aug 28 20:44:35 ie when it did the update-modules, the modules weren't there, looks like Aug 28 20:44:45 seems to be so Aug 28 20:45:01 Thats what crofton was going to research later Aug 28 20:45:02 I did a bitbake update-modules -c rebuild Aug 28 20:45:05 still didn't work Aug 28 20:52:17 Is there some way to *remove* a package from a pre-existing RDEPENDS? Aug 28 20:52:29 ie task-boot brings in kernel to the rootfs, but I don't want it there Aug 28 20:52:45 I want to build kernel, and get the uImage in /deploy but not in rootfs/boot Aug 28 20:53:22 That's another thing on the to do list :-) Aug 28 20:54:03 yep, the kernel is in /boot on the OSK also Aug 28 20:54:14 You might also notice that that uImage is 2x the size of the one in deploy! Aug 28 20:54:21 * Crofton heads for the coffee ship wo the laptop! Aug 28 20:54:46 yeah, but that just means jffs2 will compress it better ;) Aug 28 20:56:09 Just think, if you had delayed another day or two this stuff might be fixed :-) Aug 28 20:57:00 I hate to do this, but I'm gonna try an experiment that probably needs a clean build Aug 28 20:57:34 Plenty of time for a coffee now :-) Aug 28 20:57:42 actually I think the ipk I don't want is kernel-image Aug 28 20:57:49 kernel seems to be some tiny thing Aug 28 20:58:26 What do you want to do? Aug 28 20:59:24 clean OE build is pushing me towards accelerating my buy-some-faster-disk concept Aug 28 20:59:32 if you want to flash your verdex, I use the uImage in deploy Aug 28 20:59:40 I thought buildroot clean builds were painful! Aug 28 20:59:55 About the same actually Aug 28 21:00:07 sakoman: it's not getting the image, it's the extra ~1MB I have to transfer/flash that I'm trying to avoid :) Aug 28 21:00:18 sakoman: except for *-native Aug 28 21:00:30 Yes Aug 28 21:00:39 *-native seems to take about 1/2 of the total time for me Aug 28 21:01:30 I tried cutting that down by just installing all the right versions of the various *-native on my build host and then using ASSUME_PROVIDED Aug 28 21:01:38 but I ended up getting lots of random build errors Aug 28 21:01:52 probably different patches between OE's builds and my host distro's Aug 28 21:02:22 autoconf/automake/libtool/etc seem to be particularly sensistive Aug 28 21:03:10 Yeah, probably want to use the ones generated after your initial clean build Aug 28 21:04:53 re Aug 28 21:13:55 hughescr: autoconf and libtool are patched in OE and you need OE's versions. Other things shouldn't be too problematic... Aug 28 21:23:16 RP: there were a bunch of problems with random stuff which I forget -- rather than going through individual packages at a time turning ASSUME_PROVIDED off, I just removed the whole ASSUME_PROVIDED Aug 28 21:23:19 hughescr: IMAGE_PREPROCESS_COMMAND += "rm ${IMAGE_ROOTFS}/boot/uImage*;" Aug 28 21:23:31 christopher: aha, thanks Aug 28 21:23:42 hughescr: I had the same question a day or so ago. Aug 28 21:24:08 hughescr: just put that in the bb file where you build the image. there may already be a post process command there Aug 28 21:24:27 christopher: but it'll still show the ipkg as being installed then, even though it's not Aug 28 21:24:52 hughescr: it will be built because of the modules, but you won't have to carry that extra 2 meg into the rootfs Aug 28 21:25:08 do the modules depend on kernel or on kernel-image? Aug 28 21:25:39 The FILES_kernel-image = "" I mentioned earlier is a bit neater than that ;-) Aug 28 21:26:08 RP: that does sound cleaner. I think someone else showed me the preprocess command. Aug 28 21:26:12 RP: :) Aug 28 21:27:23 sakoman: so, to get modules.dep built, you just added in the MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS for update-modules? Aug 28 21:29:11 so, I am assuming that once you have that modules.dep file, the support has to be built into the kernel for those modules. right now, I am using a different kernel because there are hw dependencies Aug 28 21:29:31 christopher: that allowed the build to succeed, and populated the modules Aug 28 21:29:42 but sadly it did not generate modules.dep Aug 28 21:30:07 sakoman: ok, I have the build finished but didn't see modules.dep. I thought you were saying it worked for you Aug 28 21:30:15 sakoman: so I was trying to see what I left out. :) Aug 28 21:30:26 wish I knew :-) Aug 28 21:30:33 sakoman: what hw are you working with? Aug 28 21:30:37 gumstix Aug 28 21:30:41 connex Aug 28 21:31:46 christopher: are you building angstrom-minimal-image? Aug 28 21:32:01 sakoman: I am building angstrom-x11-image Aug 28 21:32:20 of course, with it, it is bloated and runs all sorts of scripts and other things I don't need. so, I need to try and track down those Aug 28 21:32:40 and it looks like the text on the gpe-login is just boxes...which I am assuming has to do with pango and how it is setup Aug 28 21:32:45 but I don't know why it wouldn't work? Aug 28 21:32:49 we're trying to get minimal working first before worrying about all the apps and stuff :) Aug 28 21:32:52 angstrom-minimal here, trying to keep it simple till I can get the basics working Aug 28 21:33:02 great minds Aug 28 21:33:20 oh, I don't blame you...that was the first build I did. Aug 28 21:33:36 but I'd like to have some of the graphics etc. I had a rootfs with buildroot... Aug 28 21:33:47 but once you get tiny-x going, buildroot just sort of stops IMO Aug 28 21:34:29 christopher: yeah, I've been getting a bit frustrated with the broken-ness of all the gfx apps in buildroot Aug 28 21:34:56 hughescr: yeah, there are so many people in there making changes...you never know who really knows what's going on. Aug 28 21:35:03 and for a newbie like me, that's not good. Aug 28 21:35:42 are both of you using the kernel built by oe or in a sep tree? Aug 28 21:35:53 OE kernel Aug 28 21:36:32 ok. we have a custom board so I guess I would just need to make a patch to overlay on, but I haven't done that. Aug 28 21:36:51 christopher: patched kernel, or just different .config? Aug 28 21:37:19 hughescr: well, patched kernel. I mean, we are similar to a at91rm9200-ek board, but different hw out there Aug 28 21:37:38 different nand chip, not a bga so that's different...etc Aug 28 21:38:02 probably not a whole lot in the end now that I think about it, but certainly some changes had to be made. adding in sound for the SSC... Aug 28 21:39:13 btw, I'd rather the kernel have the functionality built in than having modules to be loaded but maybe that is the wrong philosophy Aug 28 21:42:52 sakoman: I get your size error Aug 28 21:42:52 ramfs: Bad mount option size Aug 28 21:42:57 ramfs is the wrong one Aug 28 21:43:01 it wants to be shmfs I think Aug 28 21:43:05 and thus the problem Aug 28 21:43:41 I'm also getting all kinds of alignment traps happening -- clearly something's not grokking that EABI is on Aug 28 21:43:57 same here Aug 28 21:44:39 also, looks like the uImage is compressing an uncompressed kernel image Aug 28 21:44:52 ie it's using the u-boot mkimage compression option rather than using the linux kernel compression Aug 28 21:45:04 u-boot's decompression is *way* slower, so we'll want to fix that at some point Aug 28 21:45:24 my two issues at boot: http://www.sakoman.net/oe/gumstix-issues.txt Aug 28 21:45:28 also explains why /boot/uImage is so much bigger Aug 28 21:45:46 Ah, nice to have a linux expert helping :-) Aug 28 21:46:43 Any ideas on how to fix the mount size option issue? Aug 28 21:49:00 sakoman: I'll dig into it -- looks like the mount command thinks we're using ramfs, but kernel built shmfs maybe? Aug 28 21:49:09 sakoman: might also be a modprobe.conf alias thing Aug 28 21:59:42 hughescr: no /etc/modprobe.conf in the build Aug 28 21:59:58 sakoman: yeah, I noticed that when modprobe pcmcia failed to load the pxa stuff Aug 28 22:00:11 is that auto generated or done by hand? Aug 28 22:00:19 I don't think the tmpfs thing is modprobe related though Aug 28 22:00:21 since it's builtin Aug 28 22:00:30 right Aug 28 22:00:49 mostly hand created Aug 28 22:01:41 you can put a copy of what you want in packages/base-files/base-files/gumstix-xxxxx Aug 28 22:02:04 then it will get populated Aug 28 22:03:04 I think :-) Aug 28 22:04:04 I thought wrong :-) Aug 28 22:04:08 nite Aug 28 22:05:05 (gqview:214): GdkPixbuf-WARNING **: Cannot open pixbuf loader module file '/etc/gtk-2.0/gdk-pixbuf.loaders': No such file or directory Aug 28 22:05:09 looks like the "size" arg comes in from HUGETLBFS Aug 28 22:05:17 which isn't supported on ARM Aug 28 22:05:23 anyone here know what didn't get created here for me? Aug 28 22:05:35 that is, that file gdk-pixbuf.loaders is not in the rootfs Aug 28 22:06:50 i see it in librsvg Aug 28 22:07:01 ramfs by itself only supports "mode" as an option, looks like Aug 28 22:07:08 Strange! Wonder why this doesn't break other OE arm builds Aug 28 22:07:31 well,i think crofton is getting the same error on his arm device Aug 28 22:08:16 didn't know that Aug 28 22:10:05 the change is in /etc/init.d/udev Aug 28 22:10:13 Yes Aug 28 22:10:17 just remove the size=$tmp_size bit from the mount Aug 28 22:10:36 That's what I had as the workaround in the link I posted above Aug 28 22:10:51 So why don't you have this issue with buildroot? Aug 28 22:11:11 The udev script comes from the udev package Aug 28 22:11:19 did you patch udev? Aug 28 22:11:36 no, but maybe I have a newer version Aug 28 22:11:55 yes, you use 101, OE is at 092 Aug 28 22:12:04 yeah 092 is ancient I think Aug 28 22:12:06 I tried 100 as an experiment, same issue Aug 28 22:12:16 let me check... Aug 28 22:13:15 yeah, looks like buildroot just straight up replaces the udev init script Aug 28 22:13:31 so there you go :-) Aug 28 22:14:49 the standard line around here is we understand 092 :) Aug 28 22:15:37 there you are, was i off base regarding the 'size option not supported' on the osk? Aug 28 22:16:13 I think the size is fine on the OSK Aug 28 22:16:22 I can check when I get home Aug 28 22:16:28 Crofton: iirc there was some reason I moved from 092 to 101 for gumstix Aug 28 22:16:53 we can set a prefered version for udev in the gumstix machine Aug 28 22:17:10 people here have had various udev issues with later versions Aug 28 22:17:29 and no one has stuck there neck out to try later versions :) Aug 28 22:17:40 I did Aug 28 22:17:43 I tried 100 Aug 28 22:18:29 oh actually, looks like I jumped from 079 to 101 Aug 28 22:18:39 heh Aug 28 22:18:45 same time as I moved pcmciautils up a version -- probably it was pcmcia related then Aug 28 22:25:46 send me some stuff so I can hack at this in the AM :) (AM becasue by PM I need to do school stuff :) Aug 28 22:27:17 ok, back to thesis Aug 28 22:53:48 I'm getting this with just about every package I try to bitbake: Aug 28 22:53:49 NOTE: Applying patch 'noasmlinkage.patch' Aug 28 22:53:49 NOTE: Task failed: Error: /home/craig/code/gumstix/oe/build/tmp/work/iwmmxt-angstrom-linux-uclibcgnueabi/udev-114-r1/udev-114/patches/noasmlinkage.patch not found. Aug 28 22:53:56 for somepatch.patch Aug 28 22:54:04 everything looks OK in the .bb Aug 28 22:54:11 anyone have any thoughts? Aug 28 22:54:37 the patch file is in udev_114/ Aug 28 22:56:38 udev-114 Aug 28 22:56:39 :) Aug 28 22:56:52 actually, I just noticed there's -115 now Aug 28 22:57:01 but it's happening for packages I didn't do too Aug 28 22:57:06 _ is used to split package name from version Aug 28 22:57:08 like lrzsz, gzip,.... Aug 28 22:57:17 basically anything with a SRC_URI="foo;patch=1" Aug 28 22:57:49 this is for bb files you create? Aug 28 22:57:58 for all bb files I can try Aug 28 22:58:02 hmmm Aug 28 22:58:10 was working just fine a moment ago... Aug 28 22:58:23 what did you change ? Aug 28 22:58:29 good Q Aug 28 22:59:02 sorry if I seem a bit cranky, trying to finish editting a thesis chapter Aug 28 22:59:30 feel free to ignore me and get one with your own work :) Aug 28 23:00:06 mtn status says all I've done is some changes to gumstix.inc and adding udev-115.bb Aug 28 23:00:15 I do :), Aug 28 23:00:21 what about env vars? Aug 28 23:00:25 no changes Aug 28 23:00:42 or do you mean env vars inside the .bbs Aug 28 23:00:44 and maybe something you slipped in a bb file has global influence .... Aug 28 23:00:46 yeah Aug 28 23:00:51 I am wondering about that Aug 28 23:00:52 I have +PREFERRED_VERSION_udev = "115" Aug 28 23:01:01 +FILES_kernel-image = "" Aug 28 23:01:10 in gumstix machine file? Aug 28 23:01:13 yeah Aug 28 23:01:19 and +MACHINE_ESSENTIAL_EXTRA_RDEPENDS = " somestuff " Aug 28 23:01:30 what kind of stuff? Aug 28 23:01:47 sysfsutils, module-init-tools, update-modules, kernel-modules Aug 28 23:02:13 you undersand why kernel modules in depends are bad? Aug 28 23:02:22 * Crofton understand why they are there for now? Aug 28 23:02:29 er remove the ? mark Aug 28 23:03:03 why are they bad in depends? Aug 28 23:03:08 basically, if someone changes the kernel config and a RDEPENDS goes away, someone will spend a day figuring out why the build fails :) Aug 28 23:03:15 ah right Aug 28 23:03:26 apparently this has happened more than once :) Aug 28 23:03:58 ideally I'd have DEPENDS get built up dynamically based on what's in linux .config Aug 28 23:04:28 but I don't see how any of that stuff's breaking patching Aug 28 23:04:51 it sounds like you have globally broke something Aug 28 23:04:53 maybe I should blow away my tmp and rebuild Aug 28 23:04:59 what did you start with for udev_115.bb? Aug 28 23:05:06 udev_100.bb Aug 28 23:05:12 I am going to duck out again Aug 28 23:05:26 good luck Aug 28 23:05:31 ok Aug 28 23:05:32 I'll try and check back later Aug 28 23:05:33 thx for the help Aug 28 23:05:35 np Aug 28 23:05:40 Did you notice the first 2 lines? Aug 28 23:06:26 the first one makes sure everyone uses the 092 version Aug 28 23:06:53 the second lets slugos use version 100 Aug 28 23:07:19 ah Aug 28 23:08:03 retrying changing that slugos to gumstix-verdex Aug 28 23:08:32 Then you'll at least not break slugos :-) Aug 28 23:08:45 If it were to ever get checked in :-) Aug 28 23:09:03 I've also fixed the udev-115/init to remove the size thing Aug 28 23:09:13 a good thing! Aug 28 23:13:03 hughescr: ducking out myself now, back in a couple of hours Aug 28 23:13:07 ok cu Aug 28 23:30:51 03stefan 07org.oe.dev * r902224f7... 10/ (3 files in 3 dirs): Aug 28 23:30:51 uboot-openmoko_svn.bb: Update to actual u-boot HEAD, adjust path to newest OM Aug 28 23:30:51 patches and add gta02v2 to the machines. Aug 28 23:30:59 03stefan 07org.oe.dev * rfb319522... 10/ (3 files in 3 dirs): openmoko-panel-bt_svn.bb: Drop panel-bluetooth.patch, as it was applied upstream Aug 28 23:42:28 well crap -- rm -rf /oe/build/tmp didn't resolve the patch problem Aug 28 23:43:00 can you pastebin the entire build log? Aug 28 23:43:35 I would suggest backing out your changes until it works again Aug 28 23:43:51 I just commited my edits for the day Aug 28 23:46:22 ok, I am going home, I'll check in l8r Aug 29 00:41:48 Doh. I had QUILT_PATCHES set Aug 29 00:41:56 that'll teach me to work on two things at once Aug 29 00:42:02 I was typing in the wrong shell window Aug 29 00:43:06 heh Aug 29 00:44:03 lol Aug 29 00:45:01 kind of evil that having QUILT_PATCHES set in the parent shell is causing failure Aug 29 00:45:35 though I guess it kind of makes sense if you're baking only one package at a time or something Aug 29 00:46:03 ok, well heading home while this über rebuild runs Aug 29 00:46:22 Crofton: hopefully I'll have a bunch of patches for you when this finishes Aug 29 00:46:31 to target sakoman's main issues Aug 29 00:46:52 ie the tmpfs size thing, and the alingment fault msgs Aug 29 00:47:06 well, the alignment one is just cosmetic fixing, not fixing the underlying problem Aug 29 00:47:19 just stopping the underlying problem from filling the console with warnings :) Aug 29 00:47:56 Crofton: on hardware, I'll wait till tomorrow to ship you some stuff. New board launches tomorrow which I want to send you ;) Aug 29 00:48:24 ethernet+wifi+microSD on a tiny little board Aug 29 00:48:45 I'll toss in a verdex, a console-LCD board and one of these nice samsung screens we use now Aug 29 00:49:30 that pretty much covers all the various hardware bits since the wifi sits on the pcmcia bus Aug 29 00:49:34 ok ttyl **** ENDING LOGGING AT Wed Aug 29 02:59:56 2007