**** BEGIN LOGGING AT Wed Aug 29 02:59:56 2007 Aug 29 05:28:43 neat -- look what I found, how handy Aug 29 05:28:43 packages/images/initramfs-image.bb:PACKAGE_REMOVE = "kernel-image-* update-modules" Aug 29 05:28:58 I've been wondering how to force-remove a package Aug 29 05:38:23 Hi Craig! Aug 29 05:38:30 Had any luck? Aug 29 05:40:21 sakoman: yes getting there Aug 29 05:40:26 got udev 115 working looks like Aug 29 05:40:33 well, compiled anyway Aug 29 05:40:38 uploading the flash image now Aug 29 05:40:47 fingers crossed Aug 29 05:41:36 I've also fixed the two bugs from your bug webpage I think Aug 29 05:42:14 woo-hoo! Aug 29 05:42:33 I updated it so as not to forget anything Aug 29 05:46:38 List of issues not yet fixed in head of tree: http://www.sakoman.net/oe/gumstix-issues.txt Aug 29 05:48:46 Missing anything we've discussed? Aug 29 05:54:37 I've got alingment traps fixed; size fixed (still to be seen if udev 115 actually works); uImage in /boot fixed Aug 29 05:54:55 the modprobe.conf should be easy, there's actually not a ton of suff to go in there Aug 29 05:55:50 How did you fix the uImage in /boot issue? Aug 29 05:57:31 http://pastebin.com/d27985763 Aug 29 05:58:00 actually it still has a dangling symlink in rootfs/boot Aug 29 05:58:08 but the big file's gone Aug 29 05:58:31 Nice & simple Aug 29 05:59:07 yup Aug 29 05:59:41 later it'll want to be fixed so that the kernel-image package runs flashcp into /dev/mtd2 or something.... Aug 29 05:59:50 so that kernel updates can just be done using ipkg Aug 29 05:59:56 but I'll worry about that later Aug 29 06:00:37 I'm really puzzled why modules.dep, etc. aren't being generated/populated Aug 29 06:00:45 It was working last night Aug 29 06:01:45 hmm, still no /lib/modules/2.6.21/modules.dep Aug 29 06:01:46 But I did a pull/update so perhaps something changed and broke it Aug 29 06:01:59 but alignment stuff's fixed and no udev errors starting up... Aug 29 06:02:11 let me depmod -a then reboot... Aug 29 06:02:48 alignment fix is just the silent fixup (2 > /proc/cpu/alignment ?) Aug 29 06:04:29 yup Aug 29 06:05:09 Makes a big difference on the noise level doesn't it? Aug 29 06:05:41 http://pastebin.com/d20f05c36 Aug 29 06:06:26 Yup, that's what I've been doing too Aug 29 06:09:28 so looks like there's some udev rules in 115 which alleviate some of the modprobe.conf stuff Aug 29 06:10:18 nice! did the depmod -a / reboot work? Aug 29 06:11:01 yes, nicely Aug 29 06:11:16 cool, so does networking work? Aug 29 06:11:25 actually, I'm getting some errors about some missing modules, but it looks like it's due to the pxa module renaming in iirc 2.6.23 Aug 29 06:11:46 ie udev 115 appears to know the new names already, but my kernel still has the old names Aug 29 06:12:04 maybe you need to back up a rev or 2 Aug 29 06:13:15 http://pastebin.com/d23208996 Aug 29 06:13:22 that's the weird stuff I get at boot Aug 29 06:13:34 the FATAL stuff might not actually be from udev Aug 29 06:13:39 I can't figure it out yet Aug 29 06:13:54 I don't know what "Ive" is either Aug 29 06:14:45 Think I've seen that even on buildroot boots Aug 29 06:15:06 It varies sometimes, like some sort of conflict for the serialport Aug 29 06:18:33 gonna crash now -- good luck! Aug 29 07:08:18 good morning Aug 29 07:08:57 * * OE Bug 2883 has been created by autobuild(AT)openembedded.org Aug 29 07:08:59 * * uboot-openmoko-1.2. 0+git3e66c078003607a7d1d214c15a5f262bc1b4032f+svn20070829-r2-do_fetch Aug 29 07:09:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2883 Aug 29 07:09:57 moin Aug 29 07:15:34 good moring phroton! Aug 29 07:19:57 * * OE Bug 2884 has been created by autobuild(AT)openembedded.org Aug 29 07:19:59 * * openmoko-panel-mainmenu-0.1.0+svnr2567-r1-do_compile Aug 29 07:20:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2884 Aug 29 07:20:23 good morning all Aug 29 07:23:58 * * OE Bug 2883 has been marked as DUPLICATE of bug 2882 by Aug 29 07:23:59 * *  uboot-openmoko-1.2. 0+git3e66c078003607a7d1d214c15a5f262bc1b4032f+svn20070829-r2-do_fetch Aug 29 07:24:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2883 Aug 29 07:50:51 morning all Aug 29 08:16:30 morning Aug 29 08:18:15 hey XorA Aug 29 08:18:48 ha! Aug 29 08:18:58 I fixed tap-n-hold in openmoko Aug 29 08:19:13 ~lart openmoko dudes for not loading libgtkstylus Aug 29 08:19:13 * ibot takes out openmoko dudes with the trash for not loading libgtkstylus Aug 29 08:19:21 bluez-gnome is now working :) Aug 29 08:20:19 koen: cool Aug 29 08:20:33 Ill have to get round to actually flashing a working image on terrier Aug 29 08:20:42 and bv3 Aug 29 08:21:42 hmm, getting HD off bitorrent backups is still too big for my bandwidth :-) Aug 29 08:32:57 * * OE Bug 2885 has been created by autobuild(AT)openembedded.org Aug 29 08:32:59 * * qt4-x11-free-4.3.0-r6-do_populate_staging Aug 29 08:33:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2885 Aug 29 08:38:05 morning Aug 29 08:43:37 03koen 07org.oe.dev * rfd5dcb8c... 10/ (3 files in 3 dirs): openmoko-session2: fix tap-n-hold Aug 29 08:48:26 morning Aug 29 08:48:41 Crofton|home: you wanted something from me? Aug 29 08:49:11 Architecture: geode Aug 29 08:49:24 left it as geode or switch to geodelx... Aug 29 08:50:26 hrw: crofton wants your task-{base.boot} expertise Aug 29 08:50:53 and I am rebuilding angstrom-minimal-image now Aug 29 08:53:03 found. update-modules lack depmod dependency ;( Aug 29 08:55:34 http://pastebin.ca/674480 is proposed fix Aug 29 09:01:11 03koen 07org.oe.dev * r309de2f5... 10/ (7 files in 4 dirs): uclibc 0.9.29: add templates for different ARM generations and use iwmmxt for gumstix-verdex Aug 29 09:18:31 ~lart cyberdeck for spamming the channel. Aug 29 09:18:32 * ibot readies the nuke launcher and fires some rounds at cyberdeck for spamming the channel. Aug 29 09:18:37 hehe. Aug 29 09:18:38 cool. Aug 29 09:23:44 * koen is getting a bit tired of all the DBUS fud on gsmd-devel Aug 29 09:29:00 koen: which thread? Aug 29 09:31:01 the robustness one (among others) Aug 29 09:31:34 summary: dbus will solve problems, but harald doesn't understand it so it won't go in Aug 29 09:41:32 no, it's good design decision to have it layered Aug 29 09:41:53 so you can choose --with-dbus if you like to Aug 29 09:43:25 but it should work correctly without dbus also Aug 29 09:44:45 it doesn't work correctly without dbus Aug 29 09:45:06 and since it doesn't use dbus that equates to "broken piece of crap" Aug 29 09:45:10 yes many thinks don't work now Aug 29 09:45:21 ~curse openmoko for shit called kernel recipe Aug 29 09:45:22 May the fleas of a thousand camels infest your most sensitive regions, openmoko for shit called kernel recipe ! Aug 29 09:45:37 using dbus won't fix the issues Aug 29 09:48:27 hrw: maybe we should just clean up the kernel recipe and put the patches in OE Aug 29 09:48:54 and then they will fsck it again Aug 29 09:49:19 koen: That really needs to be agree with mickeyl and the mystery omoko maintainer Aug 29 09:49:52 koen: Its funny you say that as hrw just suggested doing that in poky too :) Aug 29 09:50:10 "mystery omoko maintainer" is mickeyl second^Wfifth life Aug 29 09:50:24 RP: because that recipe really stinks Aug 29 09:50:42 hrw: I agree Aug 29 09:50:48 1. using of full kernel source archives for each stable release Aug 29 09:50:58 2. not using stable branch patchsets Aug 29 09:51:03 3. do_prepatch Aug 29 09:51:17 the ezx recipe has a script to convert patch repos to SRC_URIs :) Aug 29 09:51:30 4. not using linux.inc Aug 29 09:52:54 linux-rp doesn't use linux.inc :} Aug 29 09:52:57 03hrw 07org.oe.dev * r28c0b471... 10/ (1 packages/update-modules/update-modules_1.0.bb): update-modules: depend on depmod (Take from Poky - r2589) Aug 29 09:54:38 03koen 07org.oe.dev * r5cb43bc2... 10/ (4 files in 2 dirs): Aug 29 09:54:38 angstrom: add more images: Aug 29 09:54:38 angstrom-dtp-x11-image: gives you abiword, gnumeric, claws-mail and cups Aug 29 09:54:38 angstrom-gpe-x11-image: x11 image with gpe PIM Aug 29 09:54:38 angstrom-pimlico-x11-image: x11 image with pimlico PIM Aug 29 09:55:16 koen: s/angstrom-dtp-x11-image/angstrom-x11-office-image/ Aug 29 09:55:21 abiword is not DTP Aug 29 09:55:33 ssssh! don't tell :) Aug 29 09:55:38 koen: can you also use angstrom-x11-task-image naming? Aug 29 09:56:16 sure Aug 29 09:58:14 thx Aug 29 09:58:49 ~lart openmoko for linux-gta01 (archives, prepatch, bogus depends, uboot dependency) Aug 29 09:58:50 * ibot takes openmoko to the vet for a "special" visit for linux-gta01 (archives, prepatch, bogus depends, uboot dependency) Aug 29 09:59:28 ~change 25 gbp to pln Aug 29 09:59:34 i got some error trying that Aug 29 10:03:35 03koen 07org.oe.dev * rf14dc76b... 10/ (4 files in 2 dirs): fic-gta01, fic-gta02, smdk2443: set UBOOT_ENTRYPOINT in preparation of kernel recipe cleanup Aug 29 10:03:43 03koen 07org.oe.dev * rddcd2103... 10/ (7 files in 2 dirs): angstrom-x11-images: follow angstrom-x11--image naming (thanks hrw) Aug 29 10:03:51 03koen 07org.oe.dev * r3737e891... 10/ (1 conf/machine/dht-walnut.conf): dht-walnot: set UBOOT_ENTRYPOINT in preparation of kernel recipe cleanup Aug 29 10:05:26 hmm. when i checkout org.openembedded.dev, i get this from mtn: "branch org.openembedded.dev has multiple heads:" ... Aug 29 10:05:46 should i choose the later one? Aug 29 10:06:02 you should pull again Aug 29 10:06:06 (and is this a transient issue, or should it be updated in the wiki?) Aug 29 10:06:16 hmm, just pulled 5 minutes ago. i'll try again. Aug 29 10:06:59 bernard_: select one of them Aug 29 10:07:24 hrw: http://rafb.net/p/p34p8o85.html Aug 29 10:07:29 or mtn pull again Aug 29 10:07:58 pull again worked. i was just unlucky with my timing? Aug 29 10:08:42 kind of Aug 29 10:09:09 koen: alix is no graphics for now Aug 29 10:09:23 koen: I did not tried x11 on it yet Aug 29 10:09:30 hi all Aug 29 10:09:36 hey florian Aug 29 10:09:49 koen: and one more thing - I am changing arch for it so it can wait for some time Aug 29 10:10:30 hrw: btw, you only checked in the blktool dir, not the recipes Aug 29 10:10:35 * hrw -> out Aug 29 10:10:41 koen: it was Richard not me Aug 29 10:11:01 RP: btw, you only checked in the blktool dir, not the recipes Aug 29 10:13:11 exp93xx -> ep93xx ;D Aug 29 10:21:50 koen: oh cool... more images! Aug 29 10:22:42 koen: doh! Aug 29 10:22:58 * * OE Bug 2886 has been created by autobuild(AT)openembedded.org Aug 29 10:22:59 * * squid-2.6.STABLE14-r0-do_configure Aug 29 10:23:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2886 Aug 29 10:23:06 koen: ... sdk-native-image, tablet-image, xterminal-image... Aug 29 10:23:22 feel free to add them :) Aug 29 10:24:48 03koen 07org.oe.dev * rde6041c4... 10/ (1 packages/blktool/blktool_4.bb): blktool: add 4 (from poky) Aug 29 10:24:54 03koen 07org.oe.dev * r5265cf50... 10/ (1 packages/squid packages/squid/squid_2.6.STABLE14.bb): squid: add 2.6-stable14 Aug 29 10:25:22 koen: thanks, I was just about to try adding that, then hit a merge conflict :) Aug 29 10:25:29 koen: I just started to play wit a native SDK image target :) Aug 29 10:26:00 florian: Do look at what we did in poky Aug 29 10:26:13 RP: ok Aug 29 10:27:03 florian: Its in the form of the meta-toolchain.bb file but in theory you can just extend the packages list and get more SDK packages Aug 29 10:47:50 what i love about openembedded: by the time you do "make update all" its time to make update again .. Aug 29 10:48:02 koen, I see we are having git fetch issues in openmoko? Aug 29 10:57:38 hrw|gone, you around? Aug 29 10:58:35 * RP worries about these makefiles meaning certain developers now have no idea what people are talking about :/ Aug 29 10:59:03 anyone know why modules listed in MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS are not installed properly Aug 29 10:59:30 as in having to run depmod after booting Aug 29 10:59:35 Oh great... icecc breaks building gcc :-/ Aug 29 10:59:42 RP what makefiles? Aug 29 10:59:46 RP, they just use a different "ontology" Aug 29 10:59:54 Crofton|home: See hrw|gone's last checkin Aug 29 11:00:20 http://en.wikipedia.org/wiki/Ontology_%28computer_science%29 Aug 29 11:00:24 rp thanks! Aug 29 11:00:54 Crofton|home: one that means I don't understand half of whats been said now though :/ Aug 29 11:02:53 RP: just download the makefile and learn it - get with the program! ;-) Aug 29 11:03:28 rwhitby: time for a mokomakefile.bb :-D and recurse the universe away Aug 29 11:03:29 I mean gcc/libtool/automake/make/bitbake/OE already has so many layers, what's another one between friends. Aug 29 11:04:22 rwhitby: I'm just not sure wrapping around bitbake itself is a good thing in the long term. Its not as if bitbake's commandline itself is that scary Aug 29 11:04:35 03koen 07org.oe.dev * r3a325d4a... 10/ (1 packages/angstrom/angstrom-x11-office-image.bb): angstrom x11 office: add gphoto2 Aug 29 11:05:01 RP: I always ensure that people can use bitbake commands directly if they so choose. Aug 29 11:05:03 rwhitby: I agree setting up an OE environment and keeping it up to date needs something like the makefile Aug 29 11:05:09 * XorA screams as hosting company stupidity Aug 29 11:06:04 we need an eclipse plugin for OE Aug 29 11:06:16 its taken me a lot of work to catch up with a working build, but for the last 3 days its been pretty smooth Aug 29 11:06:22 Crofton|home: Yes :) Aug 29 11:06:47 but i have to admit, as an average "make/vim/cscope" style developer, all this layering just looks like cruft. there is a liability to all this convenience-wrapper thinking. Aug 29 11:06:57 Crofton|home: This is part of the reasons for the changes in bitbake trunk so layers can be built on top of bitbake Aug 29 11:07:04 anyoen know what the "right" way to keep uImage from a root file system is? Aug 29 11:07:24 and it sonds like the gusmtix guy is going to send me some boards Aug 29 11:07:38 Crofton|home: There are three ways I can think of with varying "rightness" Aug 29 11:07:55 ok Aug 29 11:08:11 Crofton|home: The proper way will be to break the REPDENDS in the kernel image with something like RDEPENDS_kernel-image = "" to override wherever is set in kernel.bbclass Aug 29 11:08:17 http://pastebin.com/d27985763 is what the gumstix guys came up Aug 29 11:08:27 Crofton|home: You can also do what linux-rp does: FILES_kernel-image = "" Aug 29 11:08:31 that stops the file and the symlink? Aug 29 11:08:57 Crofton|home: I want to break the RDEPENDS in due course so the kernel packe still exists Aug 29 11:09:15 so the RDEPENDS method will break? Aug 29 11:09:37 Crofton|home: No, I want to use that method as its better than creating an empty package Aug 29 11:09:45 ok Aug 29 11:10:05 At some point, I will have a list of patches the gumstix guys have worked out Aug 29 11:10:05 "break the RDEPENDS" as in "override the RDEPENDS" Aug 29 11:10:12 and need to feed them back Aug 29 11:10:15 ok Aug 29 11:10:26 I may have questions then :) Aug 29 11:10:47 I am sort of in the middile, between the EU and California crowd :) Aug 29 11:10:50 timewise Aug 29 11:11:13 brb Aug 29 11:18:38 re Aug 29 11:19:35 gm hrw, thansk for the modules fix Aug 29 11:19:41 Crofton|home: no problem Aug 29 11:20:09 sounds like it should see some gumstix hw soon Aug 29 11:20:22 that should help get our support in better shape :) Aug 29 11:20:36 yes Aug 29 11:20:53 Crofton|home: I see that you will get nice hw to play with ;) Aug 29 11:20:58 yeah Aug 29 11:21:22 I have a couple of things to try that are even work related Aug 29 11:21:32 I still want to build a helicopter + gumstix uav Aug 29 11:22:11 even a RC plane for airborne ad hoc network research Aug 29 11:26:17 koen: prpplague in #openjtag is looking for people to do cool projects in exchange for free hardware of http://www.elinux.org/Hammer_Board Aug 29 11:26:37 morning all Aug 29 11:26:40 he offered me that board a few weeks ago :) Aug 29 11:26:47 hey do13 Aug 29 11:26:53 hi koen Aug 29 11:27:51 * Crofton|home is waiting until he has some real time to look at the hammer :) Aug 29 11:33:57 * * OE Bug 2887 has been created by autobuild(AT)openembedded.org Aug 29 11:33:59 * * xserver-kdrive-imageon-1.3.0-r0-do_fetch Aug 29 11:34:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2887 Aug 29 11:55:03 * ynezz wonders what or who the Hardy Haron is Aug 29 11:55:24 its ubuntu release name iirc Aug 29 11:55:36 gutsy+1 Aug 29 11:55:46 i know, just read the announce Aug 29 11:56:01 but couldn't find the reason behind the name :) Aug 29 11:56:07 koen, you have anuy ideas why the git fetch of u-bbot fails? Aug 29 11:56:25 because git is evil and wolfgang in idiot Aug 29 11:56:33 :) Aug 29 11:57:50 ynezz: I thought it was Inpiring Word + Random Animal, for the suits Aug 29 11:59:10 Adjective Animal Aug 29 11:59:47 hi Aug 29 12:06:59 03koen 07org.oe.dev * ra20b5285... 10/ (1 conf/machine/c7x0.conf): c7x0: use imageon xserver Aug 29 12:11:56 03hrw 07org.oe.dev * re37c01e1... 10/ (1 packages/linux/linux-gta01.inc): linux-gta01: use KERNELORG_MIRROR instead of german mirror Aug 29 12:14:58 * * OE Bug 2887 has been RESOLVED (INVALID) by Aug 29 12:15:00 * *  xserver-kdrive-imageon-1.3.0-r0-do_fetch Aug 29 12:15:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2887 Aug 29 12:15:05 gmm the gphone Aug 29 12:15:16 sure glad I did not run out and but an iPhone :) Aug 29 12:20:47 * koen heads out for a few hours while it is stlll sunny Aug 29 12:22:48 hi, Aug 29 12:23:20 is there a way to avoid building a Linux kernel while building an image (in a distribution context for example) ? Aug 29 12:23:56 here is my patched kernel building reciepe : http://rafb.net/p/eud9yc79.html Aug 29 12:23:56 * * OE Bug 2888 has been created by autobuild(AT)openembedded.org Aug 29 12:23:58 * * xserver-kdrive-imageon-1.3.0.0-r0-do_patch Aug 29 12:24:00 * * http://bugs.openembedded.org/show_bug.cgi?id=2888 Aug 29 12:25:31 (I used the FILES_kernel-image = "" directive to try not to build the kernel... but it builds anyway...) Aug 29 12:27:39 bitbake bla -g Aug 29 12:27:57 to see the dependency graph as .dot files Aug 29 12:36:10 ok, so to stop the installation of the kernel into the fs image Aug 29 12:36:21 for machines that store the kernel in another flash partition Aug 29 12:36:31 I put this in the kernel bb file? Aug 29 12:36:40 DEPENDS_kernel-image = "" Aug 29 12:36:58 can I add a machine qualifier Aug 29 12:37:07 Crofton|home: RDEPENDS Aug 29 12:37:14 Crofton|home: yes, you can Aug 29 12:37:24 right, bad cut ans paste :) missed the R Aug 29 12:37:31 examples in linux-rp? Aug 29 12:37:45 Crofton|home: No, I've not switched it to use it yet :/ Aug 29 12:37:48 I just realized the kernel is in the OSK image :) Aug 29 12:38:06 RDEPENDS_kernel-image_omap5912osk = "" Aug 29 12:38:11 will that work? Aug 29 12:38:42 this is in linux-omap.inc right now (I suspect all the machines that use this .inc need it) Aug 29 12:38:51 but I want to make sure I can move to linux.inc Aug 29 12:39:21 Crofton|home: Sorry, it needs to be RDEPENDS_kerne-base = "" Aug 29 12:39:25 kernel Aug 29 12:39:40 heh Aug 29 12:39:41 Crofton|home: You can see the line you've overriding in kernel.bbclass Aug 29 12:39:45 ok Aug 29 12:39:53 but yes, the override should work Aug 29 12:39:55 and how do I add the machine qualifier? Aug 29 12:40:05 like you did above Aug 29 12:40:17 ok, I'll give that a try Aug 29 12:40:53 I think that line should be in the machine.conf file ultimately if the kernel image file is never going to be wanted in the images Aug 29 12:41:14 Crofton|home : me also Aug 29 12:41:19 rp agreed Aug 29 12:41:34 To work in machine.conf files the entry in kernel.bbclass needs to be a ?= Aug 29 12:42:10 linux.inc will fill up with these ..... Aug 29 12:42:26 I haven't been strapped for space on teh OSK so I ahd not noticed Aug 29 12:43:47 ALLOW_EMPTY_kernel-base = "1" Aug 29 12:44:07 makes sure the build continues, even though there is not a kernel base package? Aug 29 12:44:52 Crofton|home: No, that just made sure the package was built even if it was empty Aug 29 12:45:12 Crofton|home: All the modules used to depend on that package before I rearranged kernel.bbclass Aug 29 12:46:37 hrw: your update-modules commit seems to be bringing in depmod.26 for unslung which has a 2.4 kernel Aug 29 12:47:44 hrw: scrub that - looks like MACHINE_FEATURES needs to be over-ridden for that DISTRO (since unslung uses a 2.4 kernel, but nslu2 machine features is 2.6 by default) Aug 29 12:48:03 Crofton|home: I've just committed changes which in theory remove kernel-image from the zaurus images. Its untested ;-) Aug 29 12:48:19 I'm goign to look on the OSK now Aug 29 12:50:26 can anyone suggest how, for a distro which uses a 2.4 kernel (and must do so, for vendor compatibility, even though the machine supports a 2.6 kernel), can remove kernel26 from MACHINE_FEATURES ? Aug 29 12:52:13 rwhitby: Which modutils do you install onto the device and how do you know which to use? Aug 29 12:53:20 RP: with hrw's latest commit, for a machine that supports a 2.6 kernel, you get a 2.6 depmod, irrespective of whether you're actually including a 2.6 kernel or a 2.4 kernel in the rootfs Aug 29 12:53:26 (see update-modules.bb) Aug 29 12:54:08 what does 'kernel26' really mean? does it mean "this machine *can* use a 2.6 kernel", or does it mean "this machine *must* use a 2.6 kernel" ? Aug 29 12:54:13 rwhitby: I know the change. I didn't realise we had machines advertising MACHINE_FEATURES = "kernel26" whilst building a 2.4 kernel :/ Aug 29 12:54:29 isn't kernel version a DISTRO decision, not a MACHINE decision? Aug 29 12:54:55 (or some combination of the two, not a MACHINE-only decision) Aug 29 12:55:01 I'd taken "kernel26" to mean "uses a 2.6 kernel" Aug 29 12:55:32 but surely that not based on the machine only. Or do I now need to create an 'nslu2be-2.4' machine? Aug 29 12:55:33 rwhitby: I did make comment earlier that it perhaps should DEPEND on virtual/kernel and then look at the version of the kernel built in staging Aug 29 12:55:58 RDEPENDS_kernel-base_omap5912osk = "" Aug 29 12:56:02 didn't seem to work Aug 29 12:56:39 rwhitby: The root of the problem is that update-modules had incorrect RDEPENDS. How we fix that is the question Aug 29 12:57:11 RP: since it's for unslung, which feature-frozen and in maintenance mode only, I'm happy to put a special case in update-modules for unslung rather than force OE to handle a machine being able to support distros with different kernel major versions. Aug 29 12:57:11 Crofton|home: :/ Aug 29 12:57:45 as long as I don't get grief from people for putting RDEPENDS_unslung = ... in update-modules.bb Aug 29 12:57:58 rwhitby: Not ideal but I can see the attraction of doing that Aug 29 12:58:26 rwhitby: You agree/understand why the RDEPENDS need fixing? Aug 29 12:58:36 yes, indeed. Aug 29 12:58:54 I'm not complaining, just pointing out a special case that the fix didn't encompass. Aug 29 12:59:08 (and am happy to special case a work-around on that basis) Aug 29 12:59:58 rwhitby: Add the special case and send any grief my way ;-) (Perhaps comment it to say why its a special case) Aug 29 13:00:10 yep, writing that very comment right now Aug 29 13:01:45 rwhitby: and thx for showing such situation Aug 29 13:03:51 rp are you sure it is kernel-base and not kernel-image? Aug 29 13:04:16 There is a kernel-image ipk Aug 29 13:05:05 Crofton|home: There should also be a kernel-base.ipk Aug 29 13:05:40 no ..... Aug 29 13:05:52 but that is what I disabled Aug 29 13:06:14 does this need anything from teh last 24 hours updating? Aug 29 13:06:34 and how much would ned rebuilding ... Aug 29 13:07:21 Crofton|home: You need the commit I just made to OE.dev if you're setting this in the machine.conf file and you should undo any FILES_kernel-image or ALLOW_EMPTY variables you added Aug 29 13:07:28 Crofton|home: You would just need to repackage the kernel Aug 29 13:07:39 upadting now Aug 29 13:07:39 03rpurdie 07org.oe.dev * r20ab813e... 10/ (1 classes/kernel.bbclass): kernel.bbclass: Allow machines to override the kernel-base runtime dependencies Aug 29 13:07:52 That change :) Aug 29 13:08:27 03rpurdie 07org.oe.dev * r3388a823... 10/ (1 conf/machine/include/zaurus-2.6.inc): zaurus-2.6.inc: Zaurus images shouldn't include kernel-image Aug 29 13:11:16 rp, so I can set the override in the machine now? Aug 29 13:11:33 Crofton|home: In theory and it shouldn't need to be an override Aug 29 13:12:56 RDEPENDS_kernel-base = "" in machine omap5912osk Aug 29 13:13:10 right Aug 29 13:31:00 rp no joy Aug 29 13:36:22 hrw: is there a way I can only update the change you just made for the update-modules? Aug 29 13:39:08 hrw: do I have to do a mtn update? just wondering if that was the only thing I wanted to update, if I can update just that one Aug 29 13:45:32 Crofton|home: I'll have to look at it later then Aug 29 13:47:05 morning Aug 29 13:55:03 thanks rp Aug 29 13:55:28 RDEPENDS_kernel-base = "" Aug 29 13:55:47 in summary, this line in machine.conf shoudl stop the kernel from appearing in the image in /boot? Aug 29 13:56:33 all: in using angstrom distro, I built a x11 image and ran a gtk problem. when I did, I got two warnings, one that pango could not find /etc/pango/pango.modules and the other from gdk that it could not find gdk-pixbuf.loaders. After looking at it, it turns out that the bb files for the current versions no longer put those files there but older ones do. Is there a distro file that has all versions that work well together (vs latest Aug 29 13:56:33 and greatest?) Aug 29 13:56:34 christopher: mtn pull/update and then build update-modules Aug 29 13:56:57 hrw: yeah, I was going to do that, just didn't know if I wanted whatever other changes have been made. Aug 29 13:57:06 hrw: the computer I have is slow so these builds take hours Aug 29 13:57:22 rp, I nukes tmp and am startig from scratch Aug 29 13:57:34 christopher, my computer is fast and builds take hours :) Aug 29 13:58:09 Crofton|home: :) hehe. yeah, every time i see a build like now where it says 760/2600, I gasp. :) Aug 29 13:58:31 Crofton|home: but I have to say, it works so I can put up with it. you all have done a nice job with it all. Aug 29 14:00:33 * florian shoots apr-util Aug 29 14:09:48 ah.. apt-util remind me one thing Aug 29 14:11:09 good morning! Aug 29 14:11:46 Crofton|home, RP: could one of you can paste the kernel.bbclass modification in a web nopaste area ? Aug 29 14:12:11 I could not pull with monotone and the commit mailing list seems to be down Aug 29 14:12:38 and whould like to test this commit Aug 29 15:12:58 sakoman, I will be not working hard from Th-Mon Aug 29 15:13:17 if you guys have anything I should commit, the sooner I see it the better Aug 29 15:13:38 I think craig should have some stuff today Aug 29 15:13:44 we think we fixed the modules in ESSENTIAL_EXTRA_RRECOMMENDS issue Aug 29 15:13:46 ok Aug 29 15:13:58 actuulayy, I may be able to work in it in mornings Aug 29 15:14:26 bbiab, gotta run to lunch Aug 29 15:14:30 Our list is at: http://www.sakoman.net/oe/gumstix-issues.txt Aug 29 15:14:32 these clowns have class at 1230 Aug 29 15:14:46 bbi an hour Aug 29 15:14:49 how inconsiderate! Aug 29 15:14:54 yeah Aug 29 15:15:03 trying the depmod fix now Aug 29 15:19:00 sakoman: so, what line did you put in? did you put it in your machine conf file? MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS += kernel-modules ?? Aug 29 15:20:27 At the moment I'm using: Aug 29 15:20:31 MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS = " \ Aug 29 15:20:31 kernel-modules \ Aug 29 15:20:31 module-init-tools \ Aug 29 15:20:31 update-modules \ Aug 29 15:20:31 sysfsutils \ Aug 29 15:20:34 ipkg \ Aug 29 15:20:36 " Aug 29 15:20:52 in gumstix.inc Aug 29 15:21:15 oh, ok. hmmm. Aug 29 15:22:30 gumstix.inc is included in the machine conf file Aug 29 15:22:52 ok. yeah, I couldn't find that in mine. Aug 29 15:23:13 just having kernel-modules in there didn't do the trick...so, must need the others with update-modules etc Aug 29 15:24:51 yes, module-init-tools and update-modules seem to be required to make modules work Aug 29 15:26:00 sakoman: I'm gonna give that shot now by rebuilding task-base and then the image. Aug 29 15:26:19 sakoman: I'm still having trouble getting pango and gtk things to populate in the rootfs. Aug 29 15:26:42 not using either, so I have no advice Aug 29 15:27:06 sakoman: not sure what is going on there. thought I found that the version spec'ed in the distro file had a bb file that didn't reference it so I changed versions to one that did, but still nothing. Aug 29 15:27:18 sakoman: yep, you mentioned that yesterday Aug 29 15:27:57 with this morning's update-modules change I have high hopes that the whole kernel module thing will now work properly on gumstix Aug 29 15:28:12 sakoman: update-modules are not needed to be listed (dependency of kernel modules), module-init-tools are not needed (busybox ones are ok, depmod goes from update-modules) Aug 29 15:28:17 building right now, so we'll see soon enough Aug 29 15:29:09 hrw: my build would fail without update-modules being included Aug 29 15:29:47 sakoman: strange Aug 29 15:30:00 I can try again if you like Aug 29 15:30:11 hrw: I tried doing just update-modules in the extra line in the machine file and it was empty in /lib/modules/2.6.20 Aug 29 15:31:18 hrw: after doing the mtn pull and update, I don't see anything new in update-modules_1.0.bb ... what line was changed? Aug 29 15:31:54 christopher: update-modules r5+ depends on depmod now Aug 29 15:32:26 hrw: so, if the line in mine says ' PR = "r4" Aug 29 15:32:36 hrw: I should change that to r5 ? Aug 29 15:33:26 christopher: no - you need to really update metadata Aug 29 15:34:00 hrw: ok. is that not done with mtn pull or mtn update? those are the only two I did Aug 29 15:34:42 christopher: when you did it? Aug 29 15:35:11 hrw: about 30 min to an hour ago Aug 29 15:36:00 ten show me 'mtn head' output Aug 29 15:36:31 e24780d148b92ff2ae1162d9886e4a7befaf60da rwhitby@nslu2-linux.org 2007-08-29T13:07:28 Aug 29 15:36:33 here Aug 29 15:36:53 e6dba568668063d79e9f4d1843c87e113ea51be3 mickeyl@openembedded.org 2007-08-29T14:52:10 Aug 29 15:37:01 I have that Aug 29 15:37:34 and still r4 update-modules? Aug 29 15:37:41 yes Aug 29 15:38:00 do I need to do something else with the command? I just typed in 'mtn pull' Aug 29 15:38:08 and 'mtn update' before that Aug 29 15:38:21 first pull, then update Aug 29 15:38:43 hrw: whoops. not sure if I did the update before or not. I'll try that Aug 29 15:38:48 hrw: again Aug 29 15:39:23 christopher: add to aliases: alias mtpull="mtn pull && mtn update" Aug 29 15:39:41 ok. that's a good idea Aug 29 15:40:32 or even: mtpull="mtn pull && mtn update -r `mtn automate heads|head -n1`" Aug 29 15:40:48 which will autoselect first head if there will be more of them Aug 29 15:42:51 hrw: still showing r4 here Aug 29 15:43:48 so you need to 'mtn update' Aug 29 15:43:51 * hrw -> out Aug 29 15:43:52 by Aug 29 15:43:57 hrw: I did. Aug 29 15:44:00 hrw: ok.bye Aug 29 15:45:30 christopher: I did a pull/update and my update-modules has PR="r6" Aug 29 15:46:00 sakoman: really? wth. ok. let me look and see what could be going on then. Aug 29 15:46:28 are you doing the pull/update from the right directory :-) Aug 29 15:46:40 from /oe/org.openembedded.dev Aug 29 15:47:31 seems OK Aug 29 15:48:13 Have you edited the bb file previously? Aug 29 15:48:31 mtn might not update if you have Aug 29 15:48:45 sakoman: I am not sure. I am wondering if it is a permissions thing. Aug 29 15:49:05 do a mtn revert on the bb file and see what you get Aug 29 15:49:15 where it may have been created at one time with sudo or root priv. and then I am udating on my account. Aug 29 15:49:22 ok. Aug 29 15:54:04 so, same thing. it didn't write to it. shows the same date which was back in july Aug 29 15:54:20 permissions? Aug 29 15:54:39 i changed that file to 777 so there wouldn't be an issue Aug 29 15:54:54 mtn status Aug 29 15:55:39 it doesn't show up there. just shows a couple files that i did change Aug 29 15:57:05 What's the time stamp on your OE.mtn file in /OE? Aug 29 15:57:37 -rw-r--r-- 1 christopher root 180568064 2007-08-29 10:17 OE.mtn Aug 29 15:58:28 I'm at a loss, looks right to me Aug 29 15:58:38 sakoman: yeah, it's weird. Aug 29 15:58:50 I'm a mtn newbie though, so what do I know! Aug 29 15:58:54 I don't know what to do. I guess I could delete that file and have the database update it Aug 29 15:59:56 revert should have done that Aug 29 16:00:14 but it's worth a trry Aug 29 16:00:48 deleted it. came back with "r4" Aug 29 16:01:32 Exactly what file are you looking at? Perhaps we're looking at different files :-) Aug 29 16:02:08 update-modules_1.0.bb ? Aug 29 16:03:13 yup, just checking:-) Aug 29 16:03:51 ok. hmmm. if that file isn't right, I wonder about the others and if that is the issue with pango, for example. Aug 29 16:04:05 Perhaps time for a fresh start Aug 29 16:04:47 yeah, I hate the idea of that, but it may be. it takes a long, long time to build everything. I guess I could save some time and copy over the download directory. Aug 29 16:15:33 bk Aug 29 16:19:11 crofton: no joy on depmod even with pull/update/clean build :-( Aug 29 16:20:04 urh Aug 29 16:20:07 sakoman: is there a quick way with vimdiff to just take the changes from the most recent file or, when doing a mtn pull to force all the files to get udpated? Aug 29 16:20:24 christopher: no clue! Aug 29 16:20:30 sakoman: it is showing the files that are different and I'm ok with just taking the files from the database Aug 29 16:20:31 ok. Aug 29 16:20:47 it is showing 3 files and I can't remember the command to copy over the diffs Aug 29 16:21:07 :diffget 3 maybe... Aug 29 16:21:15 sakoman, I need to reproduce this on the OSK Aug 29 16:21:34 crofton: hrw says I shouldn't need to specify update-modules and module-init-tools Aug 29 16:21:43 right Aug 29 16:21:53 these should be in task-boot Aug 29 16:21:56 so I'll try it that way to duplicate his method Aug 29 16:22:19 what hrw says should work is the correct answer Aug 29 16:22:27 they aren't specifically, but come as deps (I thnk that's what he said) Aug 29 16:22:33 we may need his help fixing it though :) Aug 29 16:22:38 gr Aug 29 16:22:52 my build at home failed in mtd-utils Aug 29 16:23:03 that sucks! Aug 29 16:23:12 yeah, it is coming from a git repo Aug 29 16:23:46 RP says there is a better way to suppress the uImage on the rootfs, but what you have is OK for now Aug 29 16:24:21 Makes re-flashing faster :-) Aug 29 16:24:27 yeah Aug 29 16:24:48 So what is the safe way to rebuild after removing those two packages from the machine conf? Aug 29 16:25:14 Please don't begin with rm -rf :-) Aug 29 16:25:55 what is the question? Aug 29 16:26:18 basically, try bitbake -c clean task-boot Aug 29 16:26:23 then nrebuild image Aug 29 16:26:34 I removed update-utils and module-init-tools from the RRECOMMENDS in gumstix.inc Aug 29 16:27:26 cleaning task-boot should be enough Aug 29 16:27:34 you are working from minimal image? Aug 29 16:27:35 OK, underway Aug 29 16:27:40 re Aug 29 16:27:45 Yes always angstrom-minimal Aug 29 16:29:22 crofton: sigh still no joy -- kernel modules are all in the image, but not modules.dep, etc that shoulds be there as the result of a depmod -a Aug 29 16:29:51 ok, we need to take this up with hrw Aug 29 16:29:53 I'll reflash anyway and manually depmod after boot Aug 29 16:29:57 he is in .pl Aug 29 16:30:41 I am going to set up the osk with some ESSENTIAL_RRECOMEND Aug 29 16:30:48 so I can check locally also Aug 29 16:31:14 I need to get an imge together and do some memory usage measurements Aug 29 16:31:24 I forgot to make the image needed last night :) Aug 29 16:31:40 My changes for RRECOMEND: Aug 29 16:31:40 I may have to get it running today and do measurement from a campground in WV :) Aug 29 16:31:43 MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS = " \ Aug 29 16:31:43 kernel-modules \ Aug 29 16:31:43 sysfsutils \ Aug 29 16:31:43 ipkg \ Aug 29 16:31:43 " Aug 29 16:32:06 why doesn't ipkg get into minimal? Aug 29 16:32:21 don't know, just never has been there Aug 29 16:32:32 by design perhaps? Aug 29 16:32:56 * Crofton wonders how many people have actually flashed minimal-image .... Aug 29 16:33:09 I know at least 3 :-) Aug 29 16:33:15 :) Aug 29 16:33:30 many, many times! Aug 29 16:33:50 ipkg is in there Aug 29 16:34:04 I nfsboot serval times minimal-image Aug 29 16:34:23 hm but maybe this has changed Aug 29 16:34:33 well, I'll try again to verify Aug 29 16:34:35 this should get much easier once I can test on a gumstix :) Aug 29 16:34:54 Sounds like it will happen now, you got a commit from Craig himself! Aug 29 16:34:55 since I get up early enough to speak first hand with hrw :) Aug 29 16:34:59 yeah Aug 29 16:35:18 I'll try and get a gumstix platform in front of these guys Aug 29 16:35:24 Badger helps, but personal contact always cinches the deal Aug 29 16:35:27 www.maniacchallenge.org Aug 29 16:35:31 yeah Aug 29 16:36:19 looks like fun Aug 29 16:36:35 yeah, I am interested in to many things Aug 29 16:36:43 same here Aug 29 16:36:46 I have been adding pieces of SW in my spare time Aug 29 16:37:11 I ahve talked with the profs and I sm sure they would really like the SW running on a gumstix Aug 29 16:37:27 anyway, I need to budget tiem carefully today Aug 29 16:37:38 leaving town at 8AM tomorrow and need to pack Aug 29 16:37:50 Then ignore the OE stuff and get to work! Aug 29 16:37:55 lol Aug 29 16:37:59 heh Aug 29 16:38:09 sakoman is right Aug 29 16:38:13 I need an ossie image on the OSK so I can get some number Aug 29 16:38:20 And a laptop at a campground is kind of sick :-) Aug 29 16:38:27 need to write a draft of a paper if possibel Aug 29 16:38:33 I have an image to maintian Aug 29 16:38:38 chief nerd Aug 29 16:38:51 Hey I thought that was my job! Aug 29 16:39:11 crofton look at the caos computer camp they had laptops Aug 29 16:39:18 +all Aug 29 16:39:27 so many cool events, so little money :) Aug 29 16:39:38 get a job Aug 29 16:39:40 *duck* Aug 29 16:39:43 :) Aug 29 16:39:43 so many cool events, so little time! Aug 29 16:39:48 that too Aug 29 16:39:58 I do recommend FOSDEM Aug 29 16:40:10 two days, lots of people Aug 29 16:40:11 I thought I would have more time after leaving Apple Aug 29 16:40:16 as it is free Aug 29 16:40:18 heh Aug 29 16:40:24 Managed to fill it up completely though! Aug 29 16:40:30 there is never enough time for al the cool stuff Aug 29 16:40:36 sakoman oh you were at apple Aug 29 16:40:39 Yup Aug 29 16:40:44 crofton right Aug 29 16:41:44 hmm, so it is true, you are a shape shifting reptilian alien? Aug 29 16:41:54 pictures don't lie! Aug 29 16:42:02 :) Aug 29 16:42:16 they caught me mid-shift Aug 29 16:42:21 christopher: it looks like the postinsts don't get run in your image Aug 29 16:42:39 (since you're missing both gtk and pango module files in /etc) Aug 29 16:42:54 re koen Aug 29 16:43:32 koen: ok. what is the best way to check that? I saw that the bb files for older versions of pango, for example, speced out doing a cat of that file to put it in the fs, but not the latest one called out by angstrom? Aug 29 16:45:02 how do you boot the image? Aug 29 16:45:33 koen: I copy it to flash and boot off of a /dev/mtdblock Aug 29 16:45:37 is that what you mean? Aug 29 16:45:47 hmm Aug 29 16:45:54 that should run the postinsts on first boot Aug 29 16:46:10 does it depend on modules.dep? Aug 29 16:46:11 some people chroot into an image and complain about similar breakge :) Aug 29 16:46:26 because I get errors about it not finding that at bootup Aug 29 16:46:30 when it is running all of the scripts Aug 29 16:46:40 try doing 'pango-querymodules > /etc/pango/pango.modules' Aug 29 16:46:44 is there a way to not have it depend on modules? Aug 29 16:48:58 koen: because I build the kernel outside of OE and wouldn't want it thinking or depending on something built as a module when the kernel wasn't if that makes any sense Aug 29 16:58:02 koen, how does the depmod -a happen on images? during assmebly, or after booting? Aug 29 17:01:24 Crofton: with each module postinsts Aug 29 17:01:49 ok, so you should not need to run it after booting Aug 29 17:02:05 03rwhitby 07org.oe.dev * r288504c0... 10/ (1 packages/update-modules/update-modules_1.0.bb): Aug 29 17:02:05 update-modules: Insert special case work-around for a legacy corner case of a Aug 29 17:02:05 machine which supports a 2.6 kernel but is used in a distro which must use a 2.4 Aug 29 17:02:05 kernel. It's ugly, just look away now. It has the blessing of RP and hrw, Aug 29 17:02:05 however. Aug 29 17:02:10 03mickeyl 07org.oe.dev * r44d05006... 10/ (3 files in 3 dirs): openmoko-terminal2: use more appropriate icon, bump PR Aug 29 17:02:16 03mickeyl 07org.oe.dev * reb765fb1... 10/ (1 classes/efl_base.bbclass): efl_base.bbclass: presource task is no longer necessary Aug 29 17:02:23 03daniel 07org.oe.dev * r324c11dd... 10/ (3 files in 3 dirs): Aug 29 17:02:23 linux-gta01_2.6.22.5.bb: Modify fix-EVIOCGRAB-semantics patch for 2.6.22 and Aug 29 17:02:23 enable it again. Aug 29 17:02:28 03koen 07org.oe.dev * r898c984d... 10/ (3 files in 2 dirs): bluez: update to 3.16 Aug 29 17:02:33 03koen 07org.oe.dev * r51c38800... 10/ (1 packages/bluez/bluez-gnome_0.14.bb): bluez-gnome: update to 0.14 Aug 29 17:02:42 urg, python is not build on my x86_64 Aug 29 17:04:29 koen: yes, running pango-querymodules>etc/pango/pango.modules worked after loading it up Aug 29 17:06:10 that means something went wrong running postinsts on first boot Aug 29 17:06:21 (no idea what, though) Aug 29 17:06:41 does it print anything out that I could take a look for ? Aug 29 17:06:48 is that part of initscripts? Aug 29 17:07:07 part of the packagemanager Aug 29 17:07:10 root@gumstix-connex:~$ ping openembedded.org Aug 29 17:07:10 PING openembedded.org (85.214.40.226): 56 data bytes Aug 29 17:07:10 64 bytes from 85.214.40.226: icmp_seq=0 ttl=51 time=247.8 ms Aug 29 17:07:14 :-) Aug 29 17:07:22 something should tell the package manager to configure stuff Aug 29 17:07:31 ipkg install a initscript for that IIRC Aug 29 17:07:55 crofton: as you can see from above, the depmod -a after boot did the trick Aug 29 17:07:57 sakoman, did you have to fidlle, or did it do it automatically? Aug 29 17:08:00 ok Aug 29 17:08:14 Had to modprobe smc91x Aug 29 17:08:17 ok Aug 29 17:08:35 in the linux-omap.inc there is a line I use to force loading at boot Aug 29 17:08:39 for a module Aug 29 17:08:47 Also had to patch by hand for the size and alignment issues Aug 29 17:09:13 I should be able to get the fiddle stuff in easily Aug 29 17:09:37 module_autoload_pxamci = "smc91x ? Aug 29 17:09:48 I just have to be careful not to break a lot of peopls stuff :) Aug 29 17:09:54 something like that Aug 29 17:10:04 I'll try it Aug 29 17:10:26 pxamci, may need to be the machine Aug 29 17:10:28 not sure Aug 29 17:11:07 trial build running Aug 29 17:11:21 need to step away for a half hour Aug 29 17:11:38 getting close! Aug 29 17:16:32 mickey|zzZZzz, ping Aug 29 17:17:14 basically it is module_autoload_ = "" Aug 29 17:17:30 (and rebuild the kernel to have it take effect) Aug 29 17:18:02 it's on the the thing the "sucky thing to improve in OE that no-one has time for"-list Aug 29 17:18:09 :) Aug 29 17:19:18 * Crofton curses him self for not keeping an ossie-image handy Aug 29 17:28:05 koen: ah, OK kernel rebuild required Aug 29 17:28:15 starting a *new* trial build :-) Aug 29 17:32:00 yeah, the info is put in the kernel-module- postinst Aug 29 17:32:20 not optimal, but as I said: it's on the the thing the "sucky thing to improve in OE that no-one has time for"-list Aug 29 17:33:33 sakoman: good morning Aug 29 17:33:42 got a bunch of stuff working last night Aug 29 17:34:17 sakoman: I think hrw and I fixed the bottom two points on your list Aug 29 17:34:21 Crofton: should I send patches to you? Aug 29 17:34:35 hughescr: or attach them to bugzilla Aug 29 17:34:58 koen: ok -- and "mtn diff" is what you basically want? Aug 29 17:35:10 yes Aug 29 17:35:20 preferably one bug per issue Aug 29 17:35:32 yup Aug 29 17:35:46 (so 'mtn diff ' or 'mtn diff ') Aug 29 17:36:20 mega bugs like "this patches fixes 50 problems" tend to rot away because everyone is too lazy to apply the 40 good fixes and discard the 10 bad ones Aug 29 17:36:31 koen: yup, I know that drill Aug 29 17:37:08 some people don't and get all pissy when we don't apply the 40 good fixes :( Aug 29 17:37:32 good morning hughescr! Aug 29 17:37:43 networking is up on my connex! Aug 29 17:38:16 sakoman works on verdex too Aug 29 17:38:24 excellent! Aug 29 17:38:26 including the new unlaunched board I'm planning to send to Crofton Aug 29 17:38:50 I think my next step is cleaning up the config/startup scripts to match what starts on buildroot a little more closely Aug 29 17:39:28 and then build some angstrom-derived package groupings to have a "4MB", "16MB" and maybe "32MB" standard target list Aug 29 17:39:48 plus adding a package or two from buildroot not yet in OE, like the marvell wifi driver and stuff Aug 29 17:40:36 I vote for early Marvel addition :-) Aug 29 17:40:51 Since my device needs it Aug 29 17:41:16 shouldn't be too hard I'm guessing... Aug 29 17:41:27 the pcmcia layer appears to already work OK Aug 29 17:42:09 Oh, I found some new odd bug though -- when I build/install/boot angstrom-console-image I get segfaults from some of the things which try to start themselves at boot Aug 29 17:42:25 haven't spent time tracking it down more, but the console gets a bunch of "segfault" msgs Aug 29 17:42:58 console-image just has way too much stuff in it for this early in the debug! Aug 29 17:42:58 keesj: are you the openprotium guy, or did I mix something up? your name seems familiar but I am not sure :) Aug 29 17:43:02 it might have been related to some changes I made half way through building to uclibc though, so I've kicked off a clean rebuild to check Aug 29 17:43:16 sakoman: it builds out to about 13MB Aug 29 17:43:41 yeah, takes forever to reflash! Aug 29 17:44:17 I've set my deploy dir to be served by tftp Aug 29 17:44:25 then just grab images over ethernet instead of serial Aug 29 17:44:36 hughescr, go ahead and send Aug 29 17:44:44 or buzilla, more eyes there Aug 29 17:45:48 ok caught up Aug 29 17:46:45 hughescr, does gumstix support th eomilex jtag thingy? Aug 29 17:47:17 I have the 20 pin version, need to make an adapter to 14 Aug 29 17:48:14 Crofton: not familiar with that adapter; our JTAG board has a connector which expects a wiggler thingie to plug in Aug 29 17:49:01 I picked one up, just haven't had time to play with it yet Aug 29 17:49:40 after today I will be difficult to reach until Tuesday Aug 29 17:49:46 Horror day and everything Aug 29 17:50:26 post the stuff on bugzilla for sure Aug 29 17:51:36 hughescr: which marvell driver? Aug 29 17:52:00 hughescr: hrw is working on one based on the olpc driver, and it should be in OE Aug 29 17:52:39 hughescr, if I have time I will try and get the sw for www.manicchallenge.org running on a gumstix Aug 29 17:52:51 koen: when did your fixes for the depmod stuff go in? Aug 29 17:52:54 I have been slowly adding pieces into OE Aug 29 17:53:07 I did a pull/update and got changes to modules-update Aug 29 17:53:33 But still no joy with modules.dep making it into the rootfs Aug 29 17:54:16 koen: It's marvell's one which I convinced them to release Aug 29 17:54:19 for the 8385 Aug 29 17:54:28 sakoman: http://lists.linuxtogo.org/pipermail/openembedded-commits/2007-August/008370.html and http://lists.linuxtogo.org/pipermail/openembedded-commits/2007-August/008369.html Aug 29 17:55:01 hughescr: yes, that's the one: http://projects.linuxtogo.org/projects/marvell8385/ Aug 29 17:55:50 has anyone tried or is anyone using an OE built version of imagemagick? I can't get all the formats to work Aug 29 17:55:54 koen: I've swapped some emails/code with Holger Schurig on that Aug 29 17:55:56 marvell had multiple releases for it, but most of them lacked copyright info and/or licenses Aug 29 17:56:21 koen: yes indeed -- I got them to clean that up for a specific release which seemed to be their most recent one Aug 29 17:56:28 Jin^eLD: maybe the recipe needs more DEPENDS (e.g. tiff, libpng) for it to work as expected Aug 29 17:56:30 worked it past their lawyers and everything Aug 29 17:56:35 hmm, I have a modules.dep on the OSK Aug 29 17:56:40 koen: no, its built with those libs, I checked configure output Aug 29 17:56:56 and ldd shows its linked against those too, so it must be something else Aug 29 17:56:57 oh, but it is running ossie-image, which comes from task-base Aug 29 17:56:59 crofton: not fair! Aug 29 17:57:01 not task-boot Aug 29 17:57:08 ahh Aug 29 17:57:17 hi all Aug 29 17:57:28 koen, I think we are debugging task-base versus task-boot issus? Aug 29 17:57:31 steliosk, gm Aug 29 17:57:39 Crofton: most likely Aug 29 17:57:41 when is your trip? Aug 29 17:58:03 good, that shows I am beginning to gain some understanding of how to make images Aug 29 17:58:44 koen: I got a more recent update-modules with PR = "r6" Aug 29 18:01:13 sakoman: mtn log packages/update-modules/ --no-merges --diffs --no-graph| more Aug 29 18:03:04 Crofton : hi ! Aug 29 18:03:26 I sent my advisor ch1-3, now to finish 4-6 Aug 29 18:03:36 and paper for SDR Forum tech conference Aug 29 18:03:44 aka the "Conference in Denver: Aug 29 18:04:10 so late September I will start trying to figure out PPC in Xilinix Aug 29 18:05:53 koen: lots of output! What am I looking for? Aug 29 18:06:18 sakoman: that shows a history of changes to update-modules Aug 29 18:07:37 hrw added a line that says, if this is a 2.6 machine and module-init-tools-depmod and modutils-depmod? Aug 29 18:08:57 if 2.6 ; then module-init-tools, else modutils Aug 29 18:09:11 is there an equivalent in "mtn log" of "svn log"'s --limit Aug 29 18:09:29 or do you always have to use a pager Aug 29 18:09:47 --last Aug 29 18:09:53 or --to Aug 29 18:10:12 so, eg --last 3 shows most recent 3 updates? Aug 29 18:10:23 where can be a magic selector ("1 week ago") Aug 29 18:10:32 hufnus: the last 3 revisions, yes Aug 29 18:10:43 ah, nice -- I'd also not been able to tell from docs how to do "1 week ago" and such Aug 29 18:10:47 ehm, hughescr Aug 29 18:11:30 gotta change to a new secret location -- bbiab Aug 29 18:11:34 --to l:"1 week ago" Aug 29 18:12:01 * Crofton needs to read the mtn FM Aug 29 18:12:04 see contrib/mtn2cl in OE Aug 29 18:12:17 hughscr is worse than my advisor .... Aug 29 18:15:04 Crofton : well you got 3 weeks :) Aug 29 18:15:24 Crofton: http://rafb.net/p/6Y0lko92.html Aug 29 18:15:42 21 days x 20 hours a day should be enough to get things done :) Aug 29 18:16:00 not for me any more Aug 29 18:16:07 I blow up after a week or so :) Aug 29 18:18:39 koen: so, when I do that command to see the history of update-modules, the first that shoes up is from 2006! any idea of what I need to do in order to clear out a cache or regenerate something to get the latest files? Aug 29 18:18:59 mtn pull && mtn up Aug 29 18:19:10 koen: yeah, I did both of those earlier. Aug 29 18:19:14 koen: but no update to that file Aug 29 18:19:26 the revision from 2005 shows I wrote OE in one day :) Aug 29 18:20:03 koen: that was probably a long day Aug 29 18:20:46 Powerbook-2:~/Projects/OpenEmbedded/org.openembedded.dev koen$ mtn ls vars database Aug 29 18:20:46 database: default-exclude-pattern Aug 29 18:20:47 database: default-include-pattern {org.openembedded.dev,org.openembedded.packaged-staging,org.openembedded.documentation,org.openembedded.oz354x} Aug 29 18:20:47 database: default-server monotone.openembedded.org Aug 29 18:21:01 what does your default-server look like? Aug 29 18:21:20 christopher@linuxbox:/oe/org.openembedded.dev$ mtn ls vars database Aug 29 18:21:20 database: default-exclude-pattern Aug 29 18:21:20 database: default-include-pattern org.openembedded.dev Aug 29 18:21:20 database: default-server monotone.openembedded.org Aug 29 18:21:20 christopher@linuxbox:/oe/org.openembedded.dev$ Aug 29 18:22:11 is there a way to force the mtn pull even if there are changes to files? like a command that says no matter what files i have, I want the latest Aug 29 18:23:07 'mtn up' Aug 29 18:23:40 and if vimdiff comes up, can I force it to take the server's files? Aug 29 18:23:46 what does 'mtn heads' show? Aug 29 18:24:04 if vimdiff comes up, just merge the changes Aug 29 18:24:05 christopher@linuxbox:/oe/org.openembedded.dev$ mtn heads Aug 29 18:24:05 mtn: branch 'org.openembedded.dev' is currently merged: Aug 29 18:24:05 324c11ddb442f19bd1e4d37d80db586af377552e daniel@openembedded.org 2007-08-29T16:39:19 Aug 29 18:24:05 christopher@linuxbox:/oe/org.openembedded.dev$ Aug 29 18:24:16 or 'mtn revert ' before mtn up Aug 29 18:24:33 (I use meld instead of vimdiff) Aug 29 18:25:07 the last vimdiff that came up had 3 files and I could tell that the file on the right was the newest, but I didn't know how to just take those changes other than line by line with diffget3 Aug 29 18:25:41 speaking of vimdiff, is there an easy way after selecting the merged bits you'd like to just say "I'm all done now" instead of :wq!,:q,:q Aug 29 18:26:24 hughescr: :qa! ? Aug 29 18:26:55 cbrake: aha thx Aug 29 18:28:43 the problem was that the first file wasn't merged properly so it didn't get the others. back in business now. I'll try the update-modules fix and see if that gets in the rootfs. Aug 29 18:28:59 koen: thanks for being patient. Aug 29 18:29:48 now if i can only get the postinsts to work. Aug 29 18:30:07 I'd be able to use this like it was supposed to be used :) Aug 29 18:32:11 http://www.angstrom-distribution.org/unstable/images/gumstix-connex/20070829/ Aug 29 18:32:27 and later tonight http://www.angstrom-distribution.org/unstable/images/gumstix-verdex/20070829/ Aug 29 18:34:02 ok Aug 29 18:36:48 * koen learns that 'powerdoppler' is a legit scientific term Aug 29 18:42:03 time to start powerOE :) Aug 29 18:44:33 hmm, weird disconnection Aug 29 18:44:44 can anyone tell me the last msg I sent as hughescr which got through? Aug 29 18:45:00 heh Aug 29 18:45:04 "And I will release bluez-utils-3.17 with a bunch of bug fixes tomorrow." Aug 29 18:46:05 koen: is that answering me? I wasn't talking about bluez... Aug 29 18:48:11 koen: your http://www.angstrom-distribution.org/unstable/images/gumstix-connex/20070829/ console-image seems to be about 4MB smaller than the uclibc based one I built for verdex last night Aug 29 18:48:30 koen: oh I know why, I turned on stuff probably with my MACHINE_FEATURES Aug 29 18:53:10 craigh, at some point we will need to se about building uboot for gumstic in oe Aug 29 18:53:31 Crofton: yeah, np Aug 29 18:53:33 should be easy ;) Aug 29 18:54:07 are there any patches? Aug 29 18:58:38 CosmicPenguin, ping Aug 29 18:59:50 i know this is probably a lingering issue, but the build just failed for me due to qemu native requestion gcc3.x do you guys get around it by disabling the check or not compiling it or what? Aug 29 19:00:06 anyone know what jordan uses to create a dos partition table in a disk image, for image creation for x86? Aug 29 19:01:08 Crofton: yes there are lots of uboot patches Aug 29 19:01:22 upstream .... Aug 29 19:01:23 against something which is probably not quite the current HEAD of the uboot tree Aug 29 19:01:30 koen: problems I had were related to the imagemagick bb file :P I submitted a patch http://bugs.openembedded.org/show_bug.cgi?id=2892 Aug 29 19:01:38 actually, it's against the 1.2.0 release I think that my patches go Aug 29 19:01:44 we can figure it out though Aug 29 19:01:58 better if it is against a fixed rev :) Aug 29 19:02:01 Crofton: just added 4 bugzilla bugs for my current diffs against OE head Aug 29 19:02:05 cool Aug 29 19:02:16 I'll skim b4 I bug out Aug 29 19:02:29 you have my information? Aug 29 19:04:18 Crofton: I think so Aug 29 19:04:49 ok, I should be checking email Aug 29 19:05:25 Crofton: oh we need a phone number for the UPS shipment Aug 29 19:05:38 just email me one and I can send you out some stuff today Aug 29 19:05:47 should reach you by monday or so Aug 29 19:06:21 Hello all. Aug 29 19:07:51 whats eglibc? Aug 29 19:08:02 embedded glibc Aug 29 19:08:24 based on glibc code? sounds.. icky Aug 29 19:08:25 apparently binary compat with glibc, but you can split bits off or soemthign Aug 29 19:08:34 khem, is the dude Aug 29 19:08:35 I'm burning the Angstrom-boostrap-*.jffs2 to NAND in a kwikbyte 9202 running u-boot on an Arm. I've tried a couple of variants, but my kernel does not see the mtd0 block (or any mtd blocks). Does anyone know offhand what I may be doing wrong? Aug 29 19:09:49 kergoth: pong Aug 29 19:10:15 CosmicPenguin, was curious what you used to create the partition table and stuff for disk images, but it seems like sfdisk will do Aug 29 19:10:38 * kergoth 's playing around with xen Aug 29 19:10:50 I use my own home grown tools Aug 29 19:10:54 hi kergoth Aug 29 19:10:59 hey chouimat Aug 29 19:11:05 ah Aug 29 19:11:48 russf: what is in the /dev directory for mtds ? mtdblock0? and how is the nand partitioned? do you know which mtds are assigned? Aug 29 19:14:33 kergoth: nuts, I thought I had a tarball on my server at home, but I guess I don't Aug 29 19:16:24 * chouimat wonders how is angstrom on the nslu2 Aug 29 19:16:57 morning Aug 29 19:17:09 gm Aug 29 19:17:11 hi hrw Aug 29 19:17:39 thats nifty, xen automatically saves the state of vms on shutdown and restores them on bootup Aug 29 19:18:04 kergoth: xen req virtualization support in cpu? Aug 29 19:18:17 in order to boot an unmodified guest, yeah Aug 29 19:18:23 ify ou dont mind patching the guest kernel, itll work without it Aug 29 19:19:17 thx Aug 29 19:19:41 course still need qemu for arm, but this seems nice for x86 Aug 29 19:19:57 * * OE Bug 2889 has been created by craig(AT)gumstix.com Aug 29 19:19:59 * * Remove kernel image from rootfs for gumstix Aug 29 19:20:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2889 Aug 29 19:20:10 kergoth: it will do you good to play with x86 more.. :) Aug 29 19:20:13 christopher: I had only erased the nand, nothing more. And I'll need to get the kernel to find its root before there will be a /dev directory - I'm getting a panic because I'm initing root=/dev/mtdblock0 Aug 29 19:20:37 CosmicPenguin, indeed.. my job is all x86_64 now, so far Aug 29 19:20:53 re Aug 29 19:21:02 * RP has yet to touch a 64bit box Aug 29 19:21:06 kergoth: good for you.. :) Aug 29 19:21:19 christopher: I have been able to boot to an nfs mounted root, which was encouraging, although there were a few issues related to ndb. Aug 29 19:21:23 "someone" should mess with it and multilib in oe.. but it probably wont be pleasant Aug 29 19:21:55 I think thats a certainty Aug 29 19:21:57 * * OE Bug 2890 has been created by craig(AT)gumstix.com Aug 29 19:21:57 org.openembedded.dev has 2 heads now, is one main and one experimental? Aug 29 19:21:59 * * Prevent flood of alignment fault messages on gumstix Aug 29 19:22:00 russf: ok. well, from uboot, I'll use the nand erase clean and then nand write.jffs2 for a jffs2 image. I assume you did that part. Aug 29 19:22:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2890 Aug 29 19:22:05 last time we tried to do some x86_64 images, it ended badly Aug 29 19:22:10 * * OE Bug 2891 has been created by craig(AT)gumstix.com Aug 29 19:22:11 * * Provide .bb for udev-115 Aug 29 19:22:13 * * http://bugs.openembedded.org/show_bug.cgi?id=2891 Aug 29 19:22:20 russf: but you should be able to take a look at the rootfs on your computer and look in the dev directory to see what devs are created. Aug 29 19:22:21 * * OE Bug 2892 has been created by jin(AT)mediatomb.org Aug 29 19:22:23 * * ImageMagickis not correctly staged, module libraries not correctly installed Aug 29 19:22:25 * * http://bugs.openembedded.org/show_bug.cgi?id=2892 Aug 29 19:22:26 RP: I tripped over some build problems when using a 64 bit machine that would not manifest on 32 bits. Aug 29 19:22:47 russf: just making sure that you would have a mtdblock0 or whatever it is, and that was how it was partitioned via the kernel Aug 29 19:22:54 russf: (the simple first things) Aug 29 19:22:57 * * OE Bug 2893 has been created by craig(AT)gumstix.com Aug 29 19:22:59 * * Fixups for gumstix machines Aug 29 19:23:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2893 Aug 29 19:23:04 christopher: yes, I used the nand write.jffs2 (actually tried *both* ;) ). Aug 29 19:23:33 russf: and you are creating a rootfs ? that is, you can see all of the directories under tmp/rootfs ? Aug 29 19:23:55 Crofton: so what problems left? Aug 29 19:24:05 russf: and you have the extra command depending on the erase size? for nand, for example, you tell jffs2 not to use clean markers from what I've read Aug 29 19:24:29 ask craigh Aug 29 19:24:32 russf: tell mkfs.jffs2 to be more exact. so, there is an input for that to be sure is being called when generating the image Aug 29 19:24:36 christopher: mmmh. Not sure on that. But I'll follow up tomorrow. Aug 29 19:24:44 russf: That doesn't surprise me. 64 bits builds are ever improving but probably not up to the standard of the 32 bit ones :/ Aug 29 19:24:47 I am preparing for holiday weekend by collecting data :) Aug 29 19:25:05 russf: what failed? Aug 29 19:25:26 russf: Please do report the problems so we can try and address them Aug 29 19:25:48 craigh: no diff attached to 2889! Aug 29 19:25:54 RP: I posted a question on OE-users that touched on this issue and gave my work around. But I was asking how to formalize the fix. I'd like to contribute what I learned, so we don't all have to deal with it. Aug 29 19:25:59 hrw bugs 2890, 2891, 2893 Aug 29 19:26:15 2889 Aug 29 19:26:19 sakoman: hang on Aug 29 19:26:27 russf: I think I remember reading that, it was renaming as iirc? Aug 29 19:26:36 sakoman, have you guys changed tuning for verdex? Aug 29 19:26:41 to iwmmxt? Aug 29 19:26:56 russf: I don't know what the proper fix for that is... Aug 29 19:26:57 koen fears that flag for creating random seg faults Aug 29 19:27:01 RP. right. Aug 29 19:27:06 * Crofton heads back to measurign Aug 29 19:27:20 crofton: only building connex here! Aug 29 19:27:42 Crofton: might explain my random segfaults I've seen when running console-image Aug 29 19:27:48 minimal-image seems to be good Aug 29 19:27:53 hrw: it ended up using the CROSS assembler instead of the host. Aug 29 19:28:16 hrw. When I renamed the cross assm out of the way, it worked fine. Aug 29 19:28:51 russf: exact bitbake command please, distro/machine values and log of failed build. otherwise it will be ignored Aug 29 19:29:04 hrw: Its on the OE users list Aug 29 19:29:05 hrw: OK. Aug 29 19:29:10 craigh, that is what I was wondering Aug 29 19:29:24 hrw: is this enough: http://projects.linuxtogo.org/pipermail/openembedded-users/2007-August/000227.html Aug 29 19:29:41 ok Aug 29 19:30:08 sakoman: ok patch in there now on 2889 Aug 29 19:30:18 no idea why the attach didn't work first time Aug 29 19:30:31 -Qy option.. we had it few times.. Aug 29 19:30:50 Jin^eLD>: what worked on imagemagick before your change? I tried the display command but may not have used it correctly as I never saw a jpg. not sure if that was what you saw or not Aug 29 19:31:18 hrw: also some references in that message to previous OE users with similar problem, and related stuff from GCC tracker. Aug 29 19:31:58 khem: your patch got lost, couldyou attach it to the bug tracker? Aug 29 19:32:01 craigh: Also, on 2893 I'm told you don't really need to include module-init-tools and update-modules Aug 29 19:32:17 I verified this with a test build Aug 29 19:32:46 hrw: yes. When you see the Qy option, you are using the wrong asm. Would this normally be something for the bug-tracker, since I'm not up-to-speed to fix this myself, ATM? Aug 29 19:33:12 russf: bugs not in bugtracker are getting lost easily Aug 29 19:33:27 russf: please also note host distro and host cpu type Aug 29 19:33:56 hrw: so you think I should post this in the tracker. (will be tomorrow.) Aug 29 19:34:06 And I'll add those details. Aug 29 19:34:39 Crofton: The kernel in the image issue seems to work ok here for me with the change I suggested :/ Aug 29 19:34:59 RP, I will work on it next week Aug 29 19:35:19 I will need to fix for ompa5912osk and gumstix Aug 29 19:35:20 Crofton: Can you have a look at the kernel module entries in Packages file and see what Depends on kernel-image? Aug 29 19:35:43 RP, not right now, doing exmap runs for paper :) Aug 29 19:35:57 * * OE Bug 2890 has been RESOLVED (FIXED) by Aug 29 19:35:59 Crofton: Ok, that is the next step... Aug 29 19:35:59 * *  Prevent flood of alignment fault messages on gumstix Aug 29 19:36:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2890 Aug 29 19:36:10 hmm. udev 115 wit DEF_PREF=-1 is ok for me. but DEF_PREF=1 for gumstix does not look ok for me Aug 29 19:36:19 no Aug 29 19:36:26 udev version is policy Aug 29 19:36:31 exactly Aug 29 19:36:45 Aug 29 19:36:46 -PACKAGE_EXTRA_ARCHS += " iwmmxt " Aug 29 19:36:46 +require conf/machine/include/tune-iwmmxt.conf Aug 29 19:36:53 that breaks and is policy as well Aug 29 19:37:23 isn't there is def pref =1 for slugos in one of the udev's? Aug 29 19:37:34 yes -- 100 Aug 29 19:37:34 and since I get ignored everytime when I asked for proper iwmmxt benchmarks, I declare it to be marketing and hence bogus :) Aug 29 19:37:36 koen, I think craigh is seeing that breakage Aug 29 19:37:50 Crofton: right, and siugos is a distro Aug 29 19:37:56 and distros set policy Aug 29 19:38:09 koen, I am trying to find a lackey to write a report on iwmmxt :) Aug 29 19:38:12 ah Aug 29 19:38:16 udev 115 does not build for me Aug 29 19:38:32 do we understand what craigh is looking for in 115? Aug 29 19:38:42 I don't :) Aug 29 19:39:32 last time I asked someone for "show me the precise bugs you are experiencing and proof they are fixed in a more recent udev", that guys pasted the changelog... Aug 29 19:39:44 heh Aug 29 19:40:21 but no reason we cannot see about getting the bb file commited Aug 29 19:40:28 so that we can try and figure this out Aug 29 19:40:38 see hrw's comment Aug 29 19:40:38 by all means add the .bb Aug 29 19:40:46 I saw it too Aug 29 19:40:47 "hmm. udev 115 wit DEF_PREF=-1 is ok for me. but DEF_PREF=1 for gumstix does not look ok for me" Aug 29 19:40:51 RP: first let it build Aug 29 19:41:06 I can look at some of these next week Aug 29 19:41:42 * Crofton mutters python is not build on his x86_64 atm Aug 29 19:42:16 koen: I will need to open a bug too then right ? Aug 29 19:42:46 Crofton: builds for me - for alix/geode, fic-gta01/armv4t Aug 29 19:43:23 hrw, I need to go back and double check l8r Aug 29 19:43:33 koen: re iwmmxt benchmarks; I did some Aug 29 19:43:43 it helps a bit, especially with IPP Aug 29 19:43:50 IPP? Aug 29 19:44:20 sakoman: did you ever get modules.dep to show up under lib/modules/ ? Aug 29 19:44:22 I'm hoping we can get asmola_work to tell us about iwmmxt :) Aug 29 19:44:34 Crofton: I need not necessarily 115, but depend on udev a lot more than most embedded archs, since you don't know at compile time what hardware is going to be there Aug 29 19:44:35 sakoman: I still don't have anything there, no moudles.dep Aug 29 19:44:35 christopher: after first boot? Aug 29 19:44:44 :) Aug 29 19:44:52 hrw: no, just checking after the build. Aug 29 19:44:53 christopher: only by manually running depmod -a after boot Aug 29 19:44:54 And udev keeps moving every time there's a kernel update Aug 29 19:45:05 christopher: during first boot it is created on device Aug 29 19:45:06 sakoman: ok, so I need to run that then? Aug 29 19:45:07 got it Aug 29 19:45:17 IPP = "Intel Performance Primitives" Aug 29 19:45:39 craigh: so iwmmxt speedup is shown in intel benchmark? Aug 29 19:45:40 khem: right Aug 29 19:45:44 it's basically a library of primitives for things like mpeg codec, jpeg, etc to use xscale/iwmmxt instructions Aug 29 19:45:53 hrw: will it automatically run the depmod using a script? is that how it works for the first time? Aug 29 19:46:01 hrw: no, shown on actual code when you tune for it Aug 29 19:46:02 IPP rocks Aug 29 19:46:04 craigh, od FIR filter and other things used in software defiend radio :) Aug 29 19:46:06 christopher: it should Aug 29 19:46:13 hrw: There was something ont he gumstix list a while back, let me see if I can find it Aug 29 19:46:13 we use it at the uni to speed up our biometric algos Aug 29 19:46:14 hrw: is that the same with gtk file and pango files? koen had me type in a command by hand to get pango working Aug 29 19:46:27 hrw: someone doing some scientific numeric processing using iwmmxt iirc Aug 29 19:46:31 fftw or something Aug 29 19:46:32 christopher: run 'ipkg configure' on board Aug 29 19:46:39 made like a 3x difference for him Aug 29 19:46:57 hrw: in most stuff I suspect you'll not notice any difference though Aug 29 19:47:12 With media apps though like audio/video it'll be noticeable Aug 29 19:47:45 someone here mentioned it breaking things at runtime in some apps? is there any more info on that? Aug 29 19:47:54 craigh: I do not have any iwmmxt hardware so I do not comment does iwmmxt for whole system has sense. Aug 29 19:47:57 see http://bugs.openembedded.org/show_bug.cgi?id=2893#c1 Aug 29 19:48:17 hrw: several apps randomly segfault :( Aug 29 19:48:17 asmola: I've noticed runtime problems, but they seemed when I've had any time to look at them to be EABI issues rather than specifically iwmmxt issues Aug 29 19:48:26 though the two are related I think Aug 29 19:48:42 that's why we selectively build iwmmxt apps Aug 29 19:49:15 and every app using iwmmxt wil give a slowdown as well due to all the register saving and restoring needed for it Aug 29 19:49:29 fftw with iwmmxt would rock Aug 29 19:50:02 hrw: hey Aug 29 19:50:05 hi zecke Aug 29 19:52:07 hrw: could you do me a favor and use vitamin b and ask tim bird if my proposal was okay? Aug 29 19:53:08 zecke: I do not have contact with him Aug 29 19:53:50 okay :) Aug 29 19:53:59 koen: re register save/restore -- istr seeing something (maybe linux-arm mailing list?) about that, where it had been heavily mitigated recently Aug 29 19:54:53 craigh, my guess is most apps do not benefit from iwwmmxt Aug 29 19:55:09 and ones that do can add the flag in the bb file Aug 29 19:55:58 Crofton but how will they know when to add the flag? Aug 29 19:56:11 seems like the machine should advertise somehow "I can do iwmmxt" Aug 29 19:56:15 03craig 07org.oe.dev * r1c7226c8... 10/ (6 files in 5 dirs): initscripts: lower alignment value for gumstix boards - close #2890 Aug 29 19:56:25 and that advertisement logically would be done by including machine/include/tune-iwmmxt.conf Aug 29 19:56:34 it could be that file has the wrong contents, currently Aug 29 19:56:42 but it seems like including it is the right thing to do Aug 29 19:56:52 does anyone know why an angstrom minimal image built yesterday wouldn't boot on the verdex as of yesterday? Aug 29 19:57:04 asmola: nope -- worked for me Aug 29 19:57:46 we can do CFLAGS_gumstix-verdex += " -iwmmxt " in individiual bb files Aug 29 19:57:49 ok, i just started a fresh build from scratch after an mtn pull / update, i'll let you know how it goes Aug 29 19:58:16 Crofton: that seems utterly wrong -- why should each package need to know about some random piece of hardware? Aug 29 19:58:43 the current problem is we suspect random seg faults when we use the flag Aug 29 19:58:58 so we pick a conservative defautl Aug 29 19:59:21 Crofton: seems to me each package which cares should have something along the lines of { if iwmmxt_is_supported then implement_iwmmxt_support } and not { if machine_abc or machine_xyz or machine_123 then implement_iwmmxt_support } Aug 29 19:59:33 since the latter means each new machine must search all packages and change them Aug 29 19:59:43 the former means each machine just says "I support iwmmxt" Aug 29 19:59:45 I'm looking to get a nice handheld for Linux hacking, I was looking at the N800 how is that one, is there a better one that I should look at instead? Aug 29 19:59:48 lets get the basic support going Aug 29 20:00:00 then we can work though this issue Aug 29 20:00:04 yup Aug 29 20:00:13 svolpe_gerrath: hp ipaq hx4700 is nice supported. hp 2210 is another Aug 29 20:00:39 basically, we do what we can do with the time and resources we have available :) Aug 29 20:01:09 as koen noted earlier about something else, this is a sucy bit of OE no one has had time to fix :) Aug 29 20:01:33 MACHINE_Fsounds like time for MACHINE_FEATURES = "iwmmxt" Aug 29 20:01:41 and a proper backspace Aug 29 20:02:36 Ideally, the compiler option should be fixed so packages don't break with it, then none of the above is needed... Aug 29 20:03:55 svolpe_gerrath: I think the NEO1973 phone is pretty decent for hacking, but might be awhile before you can get one. Aug 29 20:04:02 ideally, I would have a lackey write this paper for so I could go to the conference and ski Aug 29 20:04:05 :) Aug 29 20:04:09 koen: done . http://bugs.openembedded.org/show_bug.cgi?id=2894 Aug 29 20:04:33 sending patches through bugs is preferred method ? Aug 29 20:05:14 khem: yes Aug 29 20:05:29 khem: Less easy to get forgotten about Aug 29 20:05:29 koen: MACHINE_FEATURES sounds nice Aug 29 20:05:54 khem: we 'have' bug squash weekends each first weekend of month Aug 29 20:05:55 It means every package you touch with that has to be marked as machine specific Aug 29 20:05:57 * * OE Bug 2894 has been created by raj.khem(AT)gmail.com Aug 29 20:05:59 * * eglibc should provide glibc packages when chosen as default libc Aug 29 20:06:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2894 Aug 29 20:06:26 * khem makes a note to send patches through bugs future Aug 29 20:07:03 RP: or we can create MACHINE_BUILD_FEATURES = "iwmmxt"? Aug 29 20:09:11 re Aug 29 20:09:26 uh, mtn won't notice that the patch in the repo is the same as one already applied and just silently merge? Aug 29 20:09:44 it will notice and silenty merge Aug 29 20:10:08 maybe it wasn't quite identical? Aug 29 20:10:09 koen: http://pastebin.com/d5402824 Aug 29 20:10:31 Crofton: maybe I guess Aug 29 20:10:41 but then I'd expect it to pop vimdiff to ask me to merge or something Aug 29 20:10:47 instead of just bailing with crypticity Aug 29 20:10:57 * * OE Bug 1744 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 29 20:11:00 * *  omap2420h4 does not include kernel configuration Aug 29 20:11:01 * * http://bugs.openembedded.org/show_bug.cgi?id=1744 Aug 29 20:11:03 * Crofton does not heart mtn some days Aug 29 20:11:19 craigh: you had files in local working copy which were added to repository Aug 29 20:11:20 craigh, you can try reverting the local copy Aug 29 20:11:23 craigh: ah right, that's one annoying thing about mtn Aug 29 20:11:25 ah Aug 29 20:11:31 craigh: remove files and mtn will create it on update Aug 29 20:11:31 it looks at node ids before content Aug 29 20:11:41 'mtn drop ' before mtn up Aug 29 20:11:57 (or plain rm if they weren't versioned) Aug 29 20:12:13 revert/update got it to work Aug 29 20:12:17 rm in that case Aug 29 20:12:48 it's one of the side effect that we don't use the distributed stuff for 'external' devs Aug 29 20:13:25 koen: I was just going to ask if one could do git-like remote pulls from external repos Aug 29 20:13:43 craigh, do gumstix use a standard serial cable? Aug 29 20:14:28 we could, but are too lazy Aug 29 20:15:08 null modem Aug 29 20:15:16 db9? Aug 29 20:15:28 3.3v rs232? Aug 29 20:16:27 Crofton: no Aug 29 20:16:35 urg Aug 29 20:16:36 Crofton oh the PC end is DB9 yes Aug 29 20:16:40 the gumstix end is non-std Aug 29 20:16:46 did you include a cable ? Aug 29 20:16:50 Crofton: yes Aug 29 20:16:56 :) Aug 29 20:17:16 I have made to many cables in my lifetime Aug 29 20:17:29 speaking of gumstix, how does one hook the lcd to the verdex without the golaith? Aug 29 20:17:34 craigh: most of your patches look nice, only some of them don't follow the OE machine/distro split Aug 29 20:17:47 craigh: which is proably the hardest thing in OE for newcomers :) Aug 29 20:22:11 yep Aug 29 20:22:33 craigh: basically we want to have any machine buildable with any distro Aug 29 20:22:57 craigh: so machine cant depend on non-machine stuff Aug 29 20:23:09 which is why asmola_ can build my sdr stuff on the gumstix :) Aug 29 20:23:33 Crofton: thats too Aug 29 20:24:09 hi likewise Aug 29 20:24:20 hi hrw, everyone else Aug 29 20:24:48 likewise: you worked on pc engines WRAP support? Aug 29 20:24:55 hrw: yes Aug 29 20:25:05 hi likewise Aug 29 20:25:09 not in oe I think. www.voyage.hk Aug 29 20:25:23 likewise: will you switch to alix boards which replace wrap ones? Aug 29 20:25:41 likewise: voyage linux is distro which I booted on alix when unpacked it Aug 29 20:25:42 hrw: I am not going to buy an alix board. Aug 29 20:26:01 didn't v8jlene use WRAP boards as well? Aug 29 20:26:02 hrw: I worked on voyage linux for WRAP before I learned OE Aug 29 20:26:23 hrw: the wrap works fine for me and I bought a complete airtight case with it. Aug 29 20:26:36 hrw: why? Aug 29 20:26:55 likewise: just wanted to ask as I found you in many wrap docs Aug 29 20:27:15 Yeah, i wrote some wiki stuff over there. Aug 29 20:27:16 btw, iRex is looking for a developer Aug 29 20:27:29 koen: I saw. software dev. Aug 29 20:27:44 koen: but not for 2.6 porting rather Aug 29 20:27:59 shit... after keyboard cleaning 'h' gives problems ;( Aug 29 20:28:25 thats why you should never clean anything :) Aug 29 20:28:31 so use 'poo' instead Aug 29 20:28:41 ;D Aug 29 20:28:53 or Polish curses Aug 29 20:28:54 and start using a squaky french accent Aug 29 20:29:27 so, what's cooking these days? Aug 29 20:29:49 likewise: alix stuff here Aug 29 20:30:20 * hrw also waits for mistic openmoko OE maintainer to show Aug 29 20:30:27 x86 stuff here, can't get x86_64 targets to build though. Anyone tried that before? Aug 29 20:30:47 ~curse openmoko-devel-image for mixing om-2007.1/2007.2 stuff Aug 29 20:30:47 May the fleas of a thousand camels infest your most sensitive regions, openmoko-devel-image for mixing om-2007.1/2007.2 stuff ! Aug 29 20:30:56 oh and powerpc. did someone take the job of submitting a EFIKA2 project? Aug 29 20:31:08 likewise: not yet (amd64 targets) Aug 29 20:31:16 amd64=x86_64 Aug 29 20:31:22 I know Aug 29 20:31:31 ah ok Aug 29 20:31:32 likewise, no, this is the team one? Aug 29 20:31:33 amd64 is easier to type - no need to use shift Aug 29 20:31:53 * chouimat|busy is pissed off again today Aug 29 20:32:08 Crofton: yeah we should really scribble something Aug 29 20:32:12 yeah Aug 29 20:32:19 steliosk, is going to meet them Aug 29 20:32:34 we also need to figure out the powerpc c++ issue one day Aug 29 20:32:37 does efika2 is still custom size instead of more standard ones? Aug 29 20:32:50 but the gumstix guys are bribing me harder :) Aug 29 20:32:52 efika2?? Aug 29 20:33:14 ok, must go pack Aug 29 20:33:17 Crofton: yes, I have an alternative toolchain ready on my machine but I'm not sure how to get OE working with that one yet (it's in the wiki somewhere). No time yet. Aug 29 20:33:24 Crofton: ok, cu around Aug 29 20:33:31 cbrake, has some notes Aug 29 20:33:37 on his web Aug 29 20:33:37 I can send the url Aug 29 20:33:40 efika2 is miniitx Aug 29 20:33:41 yeah Aug 29 20:33:42 will find it Aug 29 20:33:47 it is named funny Aug 29 20:33:51 bec-systems Aug 29 20:33:53 or soemthing Aug 29 20:34:01 not, how to use an external tool chain :) Aug 29 20:34:10 I'll send l8r Aug 29 20:34:34 I'll likely make occasional irc appearances and delete email Aug 29 20:34:58 watch out if my friends hijack my accoutn ...... Aug 29 20:35:00 or Aug 29 20:35:07 if I have been drinking :) Aug 29 20:35:11 l8r Aug 29 20:40:05 do 2 and 4 gig sd cards work under spitz? Aug 29 20:40:18 2gb for sure Aug 29 20:40:41 i heard sdhd was supposed to be backwards compat, but haven't seen verification Aug 29 20:42:49 someone from openmoko use microsdhc card in neo - maybe we can get him to test it in Zaurus too Aug 29 20:57:38 bye Aug 29 20:57:58 nite hrw Aug 29 20:58:18 koen: any hope to get the mainsite fixed again? Aug 29 20:59:06 he zecke Aug 29 20:59:11 that would be appriciated :) Aug 29 21:00:22 koen: or was it a rootkit that messed up the site? ;) Aug 29 21:03:51 i found a post of koen's that says 4gig sdhc work fine Aug 29 21:20:37 is anyone here familiar with what is needed to get altboot to work on spitz? Aug 29 21:32:07 I had to rebuild my rootfs and now it is sized at exactly 16MB which makes me think something is capping it at that. Can someone tell me where the jffs2 generation is done and where it may be capping it? I see some of the config in bitbake.conf, but I am assuming there is another place as well with the command? Aug 29 21:33:33 i see a rootfs size for ext2, but not for jffs2 Aug 29 21:34:25 ROOT_FLASH_SIZE = "16" Aug 29 21:35:42 mine is set to 256 in bitbake.conf...is that what you are saying? Aug 29 21:36:18 i did a mtn pull and update and now it builds differently. just not sure what changed . I am sure I did something wrong, just not sure where to look. :) Aug 29 21:36:31 christopher: it's usually in the machine.conf file ...you have the ROOT_FLASH_SIZE and the EXTRA_IMAGECMD_jffs2_machine variables Aug 29 21:37:04 vlo|zzz: ok. I checked there and only have the EXTRA_IMAGECMD_jffs2 define in there Aug 29 21:37:15 for erase block and no clean markers input Aug 29 21:37:31 should I redefine ROOT_FLASH_SIZE in there to be sure it is set high? Aug 29 21:37:53 christopher: yes, this could be an option I guess Aug 29 21:38:18 since this is typically machine defined variables Aug 29 21:38:22 vlo|zzz: I just didn't know which one took priority. probably the machine as this would make sense Aug 29 21:38:48 yes probably Aug 29 21:47:55 gn Aug 29 22:00:17 03freyther 07org.oe.dev * r376819ac... 10/ (1 packages/python/python-openmoko_0.1.bb): Aug 29 22:00:17 python-openmoko: Apply Marcin's patch until today as this is going to be applied upstream Aug 29 22:00:17 I decided to use the 29th and risking to break the AMD64 builds instead of Aug 29 22:00:17 breaking the build for everyone tomorrow. I will need to poke mickeyl and Aug 29 22:00:17 ask to use a specific revision. Aug 29 23:14:55 03openmoko 07org.oe.dev * rfa056f4e... 10/ (1 packages/gsm/files/gsmd packages/gsm/libgsmd_svn.bb): gsmd: rewrite start script, use start-stop-daemon. fixes OM bugs #704, #751 Aug 29 23:15:00 03mickeyl 07org.oe.dev * r02f5d80d... 10/ (5 files in 3 dirs): openmoko-mediaplayer2: ship images, remove patches that have been applied upstream Aug 30 01:56:51 03rwhitby 07org.oe.dev * r1d0439f6... 10/ (1 packages/tasks/task-openmoko-feed.bb): task-openmoko-feed: added openssh packages **** ENDING LOGGING AT Thu Aug 30 02:59:57 2007