**** BEGIN LOGGING AT Tue Jan 10 02:59:58 2012 Jan 10 07:31:37 how to remove all the debug and development files such as man, info etc to minimize the filesystem in oe Jan 10 07:35:57 snkt: as for me there is no any man in oe. Jan 10 07:36:13 or you have any? Jan 10 07:37:31 reisei, man pages in usr/share Jan 10 07:39:36 snkt: what image are you bitbaking? Jan 10 07:40:17 reisei, x11-image Jan 10 07:41:44 snkt: you can just delete it on your target system. Jan 10 07:42:39 reisei, that I had done..... but is there any feature in oe ..so that we can reduce filesystem size.... Jan 10 07:43:43 snkt: I guess, you should write your own image file that contains only what you need. Jan 10 07:44:14 reisei, ok I vll try.... Jan 10 07:44:50 snkt: It simple and fun :-) Jan 10 07:44:56 *it is Jan 10 07:47:30 reisei, i m newbei to it..... I have tried by modifying the minimal-gpe-image Jan 10 07:48:05 snkt: it's the other way, have you success with that? Jan 10 07:49:12 reisei, yes... I have just add some packages such rdesktop, vnc etc to it ... and its work Jan 10 07:53:05 snkt: I looked in minimal-gpe-image file in find nothing about man pages... What size of final rootfs do you expect? Jan 10 07:53:45 reisei, 30 Mb Jan 10 07:54:14 snkt: It's hard to do with the X-server Jan 10 07:54:33 I think Jan 10 07:54:35 reisei, with rdesktop, x11vnc, wpa_supplicant Jan 10 07:54:56 reisei, with uClibc the ratio is too minimal... Jan 10 07:55:06 just 3 Mb difference Jan 10 08:06:50 snkt: how will x11vnc work without x-server? Jan 10 09:19:55 reisei, i m building image with x-server..... Jan 10 09:27:00 snkt: but minimal-gpe-image doesn't contains it... Jan 10 09:27:26 reisei, it has.... Jan 10 09:27:37 reisei, its not minimal-image Jan 10 09:32:30 snkt: okay then Jan 10 09:43:46 Good Morning Jan 10 10:04:22 florian: morning Jan 10 10:40:46 Hi, I Got the error, When loading kernel in at91sam9260ek board. VFS: Mounted root (jffs2 filesystem) on device 31:0 devtmpfs: error mounting -2 Freeing init memory: 124K Kernel panic - not syncing: No init found. Try passing init= option to kernel Jan 10 10:41:06 pls help anybody to me! Jan 10 10:43:37 amazing Jan 10 10:46:25 -ENOENT, /dev does not exists? Jan 10 10:46:39 good morning Jan 10 11:38:12 I want to build a simple X image , which will have icewm package , Is it possible to build a minimum file system image under 20 Mb with X , Jan 10 11:38:51 i have tried the minimal-gpe-image ,i have managed to get around till 26 Mb ,can it made more less Jan 10 11:43:38 I would have thought so. Jan 10 11:44:42 Depends a bit what xserver and suchlike you want, but you can build a minimal image without X in about 4MB so you'd think it should be possible to get X into the remaining 16MB. Jan 10 11:49:25 pb_, I want X bcoz it in future i will be including xephyr vnc rdesktop etc ,so which options should i enable Jan 10 12:07:29 * otavio having a bad time to find why firefox is segfaulting Jan 10 12:17:10 erwt: I already told you before, start with a small image (e.g. micro-minimal-image or so) and add only those packages that you really need (and make sure your machine config does not state things that you do not want (like e.g. bluetooth) Jan 10 12:17:26 that is the only way to get it under control Jan 10 12:17:42 ya i am doing that Jan 10 12:18:12 you might also consider r/o filesystems like squashfs Jan 10 12:18:26 but it is going large ,i am trying to control it ,sometimes it fails to load even X Jan 10 12:18:52 erwt: welcome to embedded World ;-) Jan 10 12:18:53 i have been sucessful till minimal-gpe-image Jan 10 12:18:57 the squashfs filesystem I just made is 2.5 MB; this excludes the kernel which is another 1.5 MB (powerPC, no X) Jan 10 12:19:17 eFfeM_work, thats really great Jan 10 12:19:23 otavio, thanks Jan 10 12:19:36 also think upfront what you want, e.g. do you need the gpe specifics Jan 10 12:21:03 i removed all of them even splash , Jan 10 12:22:15 then build with keeping the rootfs and go to tmp/rootfs and see what else is unneeded, at best remove the recipe that provides the stuff that you do not need, but if that for whatever reason is impossible in your image recipe add a postprocess command to remove unneeded junk Jan 10 12:23:35 btw just verified. The 2.5 M image I just mentioned: it is actually 2596864 bytes squashfs, but I also build a jffs2 image with the same content that one is 3407872 (and I did not really see if I could tune the compression of squashfs) Jan 10 12:23:54 then again squashfs is of course not persistent Jan 10 12:24:12 so I do have a small jffs2 partition to store some data Jan 10 12:25:44 and to weed out unneeded stuff in your image recipe do things like this Jan 10 12:25:49 IMAGE_PREPROCESS_COMMAND += "rm -rf ${IMAGE_ROOTFS}/boot;" Jan 10 12:26:32 ok, flashing the device has finished, back to work .... Jan 10 12:31:02 eFfeM_work, i will follow your suggestion Jan 10 12:32:28 for absolute minimum work bottom up (and yes that requires to build up knowledge on what needs what etc, and you will undoubtly run into problems e.g. because a recipe DEPENDS fails it list a dependency) Jan 10 12:33:56 another suggestion: do not put all kind of modules in your image that you are never going to load. e.g. some distro's aim to be generic and have lots of modules in their images Jan 10 12:44:26 oh, (not sure if that is feasible with X images), but also use busybox where possible instead of the full blown things, my busybox image is 770 KB or so (again for PPC) and I have 200+ applets in it (actually it could use some pruning) Jan 10 12:49:42 and if you want to go to the last bit: http://www.elinux.org/System_Size, if you master all things on that page you'll probably have a pretty small image and have become an expert in the field Jan 10 12:50:25 it'll take some time to reach that level though ;-) Jan 10 13:07:04 eFfeM_work: there are good news, I've see some 'tiny' project is developed under Yocto. More than that, Peter Chubb (linux-tiny project) seems using OE nowadays Jan 10 13:07:29 ah, nice! Jan 10 13:07:34 :-D Jan 10 13:08:01 hideedledee Jan 10 13:08:05 eFfeM_work: http://linux.conf.au/schedule/49/view_talk?day=wednesday Jan 10 13:08:20 nice Jan 10 13:09:12 but don't think he can get back to a kernel that fits on an old 360k floppy Jan 10 13:09:34 well, I went to 800kb with many filesystems compiled in... Jan 10 13:09:45 and embedded cpio Jan 10 13:09:52 eFfeM_work, minimal image with X 21 Mb now i feel its nearer Jan 10 13:09:57 with kexecboot inside ;) Jan 10 13:10:17 hi ppl :) Jan 10 13:10:18 ant_work: wow Jan 10 13:10:23 eFfeM_work: no network :/ù Jan 10 13:10:39 hi otavio Jan 10 13:11:35 ant_work: that would help quite a lot, but not sure if it is enough Jan 10 13:12:13 eFfeM_work: in my test the major disappointment came comparing uclibc and eglibc builds..almost same size Jan 10 13:12:28 florian: hi, any more luck building? mine fails to configure upower_0.9.14 Jan 10 13:12:35 klibc / dietlibc do make the difference Jan 10 13:12:46 but it all depends how many utils you need Jan 10 13:13:01 ant_work: not sure if I moved from glibc or eglibc to uclibc, but saved quite a lot Jan 10 13:13:07 cs_nbp: not yet... but I'll try again later Jan 10 13:14:26 ant_work: Hi :-D Jan 10 13:14:34 eFfeM_work: I'll surely retest...maybe eglibc got magic optimizations ;) Jan 10 13:14:56 size of images was almost equal Jan 10 13:15:02 ant_work: you can definitely tune some stuff in eglibc, I'm not sure I did Jan 10 13:15:16 ah, yes, neither me, vanilla images Jan 10 13:15:32 eFfeM_work: glibc and eglibc are quite different in this regard Jan 10 13:16:09 otavio: I know Jan 10 13:20:10 otavio: have you looked at the runtime deps of x11-common? What would you ad/remove comparing with xserver-common? Jan 10 13:20:31 one thing is xinput, xinput-calibrator Jan 10 13:20:35 ant_work: I did not Jan 10 13:20:46 ant_work: I am fighting with firefox at this moment Jan 10 13:20:50 ant_work: it segfaults Jan 10 13:21:03 I see. iirc is not the first time ;) Jan 10 13:21:04 that's what it does^^ Jan 10 13:21:15 heh Jan 10 13:21:25 ant_work: unfortunately you're right Jan 10 13:21:47 ant_work: want to share some pain? :-) Jan 10 13:22:01 ant_work: I'll be glad to lend you some firefox' pain Jan 10 13:23:11 * ant_work hides back Jan 10 13:23:23 ^_^ Jan 10 13:23:47 I'll be glad to debug it once I have a working xserver :p Jan 10 13:25:41 ant_work: working xserver? why it is hard? Jan 10 13:25:43 fwiw firefox 9.01 is eating 51MB just for showing an empty page Jan 10 13:25:55 (here in office on M$ Win PC) Jan 10 13:26:21 ant_work: my current recipe is to 3.6.8 but I do want to make a 9.0.1 recipe for it Jan 10 13:26:48 otavio: maybe it's dying because of age :) Jan 10 13:26:57 iirc 3.x was usable with 512MB RAM Jan 10 13:26:59 lol Jan 10 13:27:40 ant_work: I started a meta-mozilla layer with my current recipes Jan 10 13:27:54 ant_work: but they're not working as good as I wish Jan 10 13:28:05 I've seen, unfortunately my devices have 32 or 64 MB RAM Jan 10 13:28:06 JaMa: do you use firefox in any embedded OS? Jan 10 13:29:22 otavio: no, only eve/midori/dillo Jan 10 13:29:35 +microB Jan 10 13:30:12 ah +firefox for embedded on n900 Jan 10 13:30:30 oh! Jan 10 13:30:34 help me out! Jan 10 13:30:36 hah Jan 10 13:30:38 but last 2 are not built by OE Jan 10 13:30:47 :-( Jan 10 13:30:49 shiff Jan 10 13:34:16 Does anyone knows any window manager which is light weighted ,size if around 1 Mb excellent ,i just want it for keyboard working Jan 10 13:34:38 huh? even emacs is 15 MB Jan 10 13:35:04 brb Jan 10 13:35:24 erwt: openbox (iirc not packages in oe-core) Jan 10 13:35:26 yeah but he wanted just wm, not whole system in "text editor" :) Jan 10 13:35:43 erwt: matchbox-wm Jan 10 13:35:59 I just told that because he said size 1MB Jan 10 13:35:59 JaMa: humm you seems to talk as a vi user heh Jan 10 13:36:05 what happened to my nick Jan 10 13:36:07 -.- Jan 10 13:36:23 waita sec Jan 10 13:37:27 aah Jan 10 13:37:36 better Jan 10 13:38:02 otavio, for matchbox which one should i include in my recipe Jan 10 13:38:54 wm Jan 10 13:40:14 don;'t we have a recipe for jwm or so Jan 10 13:46:17 otavio, worse image size shooted to 38 Mb Jan 10 13:47:04 florian: apparently tworaz is traing to build a rootfs, too :D Jan 10 13:47:13 * Jay7 was tried to port fvwm one time Jan 10 13:47:19 and it even was it oe-classic Jan 10 13:47:40 s/it/in/ Jan 10 13:48:05 but fvwm have no xrandr support.. so you should restart it after screen rotation Jan 10 13:48:29 anyway, matchbox is more suitable for small screen devices Jan 10 14:09:33 muriani: ping Jan 10 14:24:09 after bitbaking a menuconfig, does oe automatically use the new config or do I have to tell it to somehow? Jan 10 14:26:15 about kernels, if you bitbake the same recipe it will take the modified .config but next time will be lost Jan 10 14:26:48 you have to copy it out and provide a defconfig for the kernel recipe Jan 10 14:27:02 just asking because bitbake didn't seem to do much, kernel was ready in like 5 secs Jan 10 14:27:52 sounds strange Jan 10 14:29:27 will bitbake take the config if i only do virtual/kernel, or does it have to be the whole image Jan 10 14:29:44 like something-image Jan 10 14:30:07 no diff, it will compile the kernel before doing the rootfs Jan 10 14:30:39 (usually kernel is required in some ways by the machine/image) Jan 10 14:33:12 guess i'll have to provide a defconfig -.- Jan 10 15:03:02 does exist an equivalent of meta-toolchain-qte for qt4-X11 ? Jan 10 15:07:30 maybe is recipes/qt4/qt4-tools-sdk ? Jan 10 15:54:07 cs_nbp: Which device is the CF card? Jan 10 15:58:59 floria: /dev/sda Jan 10 15:59:09 florian: /dev/sda :D Jan 10 15:59:51 cs_nbp: ok great... can you look up the partitions as well? :-) Jan 10 16:00:21 florian: what do u mean ? sda1 vfat sda2 ext2? Jan 10 16:00:47 cs_nbp: ah sorry no, the partitions in flash Jan 10 16:03:34 florian: err sry, what is that? if u have a second i can put the rc9 kernel on and look directly Jan 10 16:04:16 cs_nbp: okay... it should print some stuff about mtd devices Jan 10 16:04:44 hmm, but that's a good point - we should start to use ubi while we are on it Jan 10 16:06:42 florian: i tried to do and menuconf and succeeded, but i have the impression oe ignores the config Jan 10 16:06:52 florian: k rc9 is up im in /dev Jan 10 16:07:16 florian a lot of mtd stuff there Jan 10 16:07:31 very good Jan 10 16:08:00 maybe i just build one manually later on, get thign right and return it to the tree Jan 10 16:08:04 florian: mtd0 mtdoro mtd1 mtd1ro mtdblock0 mtdvlock1 Jan 10 16:08:09 florian: that's it Jan 10 16:08:21 florian and oro is 0ro Jan 10 16:08:53 florian: never understood why o and 0 have to be so close together -.- Jan 10 16:09:04 hmm... Jan 10 16:09:24 ah yes and v is b .,. Jan 10 16:09:53 ur lucky u can't hear me play piano :D Jan 10 16:10:31 cs_nbp: In the boot log there should be something like "Creating 2 MTD partitions on..." after that it lists ranges and names Jan 10 16:12:14 florian: theres nothing in var log and dmesg kinda scrolls fast ill pipe it to more Jan 10 16:13:27 florian: you mean the mtd partitions on nand? Jan 10 16:14:01 cs_nbp: yes Jan 10 16:14:18 florian: 0x000000400000-0x000001c00000 LX nand 1 Jan 10 16:14:53 florian: 0x000001c00000-0x000002000000 LX nand 2 Jan 10 16:16:48 florian: that's what you meant with in flash (i'm always talkin of it as nand ;)) Jan 10 16:17:28 cs_nbp: yes... we have nand flash only on this machine anyway Jan 10 16:18:26 florian: my build stopped at upower config btw Jan 10 16:19:17 * florian tries to remember how he fixed this Jan 10 16:21:06 ah Jan 10 16:21:23 cs_nbp: it looks like I had to add gobject-introspection-native to its dependencies Jan 10 16:21:41 florian: in the upower recipe i assume Jan 10 16:22:47 cs_nbp: yes Jan 10 16:24:25 florian: k done, but will take a while as i cleared out everything in tmp again Jan 10 16:25:43 florian: don't tell me it runs through after that ;) Jan 10 16:26:42 cs_nbp: unlikely Jan 10 16:26:54 florian: here a citation from tworaz "The patches I Jan 10 16:26:55 have for xserver-kdrive work in terms of, it starts and you can see Jan 10 16:26:55 stuff on the screen, but everything is slow. I don't know enough about EXA to port old stuff from KAA" Jan 10 16:26:58 cs_nbp: just do not kill tmp for any change :) Jan 10 16:27:26 florian: i just didn't remember if i cleared it out after tworaz changed to glibc Jan 10 16:27:47 florian: need dates in my logs.,. Jan 10 16:28:21 err history Jan 10 16:29:00 cute, this device has 16k erase blocks? Jan 10 16:30:40 florian: like your gta01 ;) Jan 10 16:33:01 cs_nbp: but the gta01 supports dfu :-) Jan 10 16:33:18 florian: yeh that's a plus Jan 10 16:33:34 definitely Jan 10 16:38:21 While my machine is busy compiling Android I can take a little rest and build a kernel with matching configuration and initfs. Jan 10 16:38:37 yeeh Jan 10 16:39:14 i tried to make a rootfs with multistrap but that needs the nbp to be able to export nfs first Jan 10 16:40:33 oh... we need change the kernel configuration anyway. it seems mtd is built as a module per default.. Jan 10 16:42:25 florian: task do_rootfs: Failed Jan 10 16:42:47 florian: that's good, roght? it means it's one of the last tasks to fail Jan 10 16:43:16 cs_nbp: yes indeed Jan 10 16:43:22 4067 of 4071 Jan 10 16:43:42 ill try to catch where he put the log Jan 10 16:44:27 bah too long -.- can't scroll up to it Jan 10 16:45:34 Collected errors: Jan 10 16:45:34 * extract_archive: Cannot create symlink from ./var/run to 'volatile/run': File exists. Jan 10 16:45:34 * extract_archive: Cannot create symlink from ./var/run to 'volatile/run': File exists. Jan 10 16:45:45 florian: that's all errors? Jan 10 16:47:31 florian: yes that's all, apart libgpg-error0 but i assume that's right ;) Jan 10 16:49:36 anybody knows how to create a meta-toolchain for qt/x11 ? Jan 10 16:50:39 can you not just adapt meta-toolchain-qte.bb? Jan 10 16:51:20 pb_: that's what I needed to know Jan 10 16:51:29 thx Jan 10 16:52:11 throw in the needed -dev packages on top of meta-toolchain-qte Jan 10 16:52:20 cs_nbp: interesting... I never ran into that one Jan 10 16:52:35 florian: sounds easy to fix though Jan 10 16:52:53 florian: trying to find where he tries to do that Jan 10 16:57:26 khem: looks not so easy, I don't understand which -dev packages to add Jan 10 16:57:55 thats another story Jan 10 16:58:25 you asked "how" not "with what" Jan 10 16:58:39 cant use meta-qt3? Jan 10 17:07:05 cs_nbp: the kernel configuration really needs some attention in order to become useful Jan 10 17:07:52 florian: ah ok, only enabled devtmpfs so far and checked that initfs is supported Jan 10 17:08:04 I changed the most important bits and added a hack to get initfs into the kernel... so now I can search for a CF card before it finishs Jan 10 17:08:26 *has 4 CF cards in front of him Jan 10 17:08:56 I suppose we have to provide a more useful command line and include mtd in the kernel itself Jan 10 17:11:00 florian: and currently when I type reboot I end up booting wince Jan 10 17:12:37 cs_nbp: this would mean the image ends up in ram... iirc boost is able to run images in ram as well. it makes sense somehow during development Jan 10 17:17:14 where do i find the definition of task-base-extended, doesn't seem to be a bb like task-base Jan 10 17:17:38 i think that's where it goes wrong Jan 10 17:18:08 yes, the oecore one Jan 10 17:31:53 cs_nbp: kernel boots... need to check the wrapper because it changes the cmdline Jan 10 17:32:28 florian: so you carved that together manually? Jan 10 17:32:55 cs_nbp: mostly... built with oe, but the bianry initfs is a bad hack Jan 10 17:33:10 florian: ok, don't know about that Jan 10 17:33:18 flroian: only read about it Jan 10 17:34:26 making one partition from nand might be a good idea as well Jan 10 17:35:23 later :) Jan 10 17:35:26 laterZ Jan 10 17:35:37 ah ok misunderstood -.- Jan 10 17:35:40 :D Jan 10 17:37:07 03Dongxiao Xu  07master * r96c307b987 10bitbake.git/lib/bb/command.py: Jan 10 17:37:07 command.py: add parseConfigurationFiles API Jan 10 17:37:07 The parseConfigurationFiles API calls the related function in Jan 10 17:37:07 cooker.py to parse config files. Jan 10 17:37:07 Signed-off-by: Dongxiao Xu Jan 10 17:37:07 Signed-off-by: Richard Purdie Jan 10 17:37:22 florian: all this sound much 'Zaurish' Jan 10 17:38:37 ant_work: close :-) Jan 10 17:38:48 ant_work: its a netbook pro Jan 10 17:38:52 heh Jan 10 17:38:56 task-base-extended is defined right IN task-base, right? Jan 10 17:39:47 cs_nbp: I think so yes Jan 10 17:41:03 ah, is pxa255 Jan 10 17:41:25 well can't see anything wrong with task-base Jan 10 17:41:31 03Paul Eggleton  07master * r379b12107e 10bitbake.git/bin/bitbake-layers: (log message trimmed) Jan 10 17:41:31 bitbake-layers: flatten: allow specifying layers to flatten Jan 10 17:41:32 You can now optionally specify two or more layers to flatten into the Jan 10 17:41:32 output, rather than flattening all of the layers in the current Jan 10 17:41:32 configuration (but this is still the default behaviour if no layers are Jan 10 17:41:32 specified). Note that this means the output layer may still contain Jan 10 17:41:32 bbappends where the corresponding recipes are not present in the list of Jan 10 17:41:36 cd ~/oe Jan 10 17:41:41 03Paul Eggleton  07master * r8e4dc97614 10bitbake.git/bin/bitbake-layers: (log message trimmed) Jan 10 17:41:41 bitbake-layers: flatten: warn the user if output structure is incorrect Jan 10 17:41:41 If you flatten layers that have different directory structures you may Jan 10 17:41:41 not end up with a usable layer in the output directory - some files Jan 10 17:41:41 won't be picked up by BitBake. Jan 10 17:41:42 To try to avoid this problem, once flattening has completed, get the Jan 10 17:41:42 BBFILES entries that correspond to the layer from which the output Jan 10 17:41:42 03Paul Eggleton  07master * r470a160813 10bitbake.git/bin/bitbake-layers: Jan 10 17:41:43 bitbake-layers: close files in apply_append() Jan 10 17:41:44 It's recommended practice to close files when finished with them and the Jan 10 17:41:44 code in this function was not doing this. Jan 10 17:41:44 Signed-off-by: Paul Eggleton Jan 10 17:41:45 Signed-off-by: Richard Purdie Jan 10 17:41:45 03Richard Purdie  07master * r9943bad611 10bitbake.git/lib/bb/ui/knotty.py: Jan 10 17:41:46 bitbake/knotty: Show summary of warning/error messages shown Jan 10 17:41:46 Show a summary count of warning/errors messages shown to the user during the Jan 10 17:41:58 spam? :p Jan 10 17:42:03 florian: I suppose we could easily add support for it in meta-handheld Jan 10 17:42:11 build and make it clear when an error exit code is being set. Jan 10 17:42:11 [YOCTO #1540] Jan 10 17:42:11 Signed-off-by: Richard Purdie Jan 10 17:42:11 03Richard Purdie  07master * reac9249b40 10bitbake.git/lib/bb/ (__init__.py command.py providers.py): (log message trimmed) Jan 10 17:42:11 bitbake: Add BBHandledException exception class Jan 10 17:42:11 We have a problem knowing when to show the user debug information and Jan 10 17:42:12 when not to since the code has already shown the user suitable information Jan 10 17:42:12 about why a failure is occurring. Jan 10 17:42:13 This patch adds a bb.BBHandledException exception class which can be used Jan 10 17:42:13 to identify those exceptions which don't need further explanation to Jan 10 17:42:36 smone is piping their stdout to irc? -.- Jan 10 17:42:52 cs_nbp: normal CIA bot activity Jan 10 17:43:01 ant_work: yes... actually there are soem layers to support it already. i'm just trying to get it running while my machine compiles... Jan 10 17:43:24 ant_work: din't see onr of those yet Jan 10 17:43:30 cs_nbp, its extremely helpful being able to keep an eye on what other developers are doing. http://cia.vc/ Jan 10 17:43:50 florian: cs_nbp: the BSP should be kept together with the other old devices Jan 10 17:43:56 yeah, bit of a shame it doesn't seem to work for anything other than bitbake at the moment Jan 10 17:44:06 huh, wonder what happened there Jan 10 17:44:18 pb_: we can set it up Jan 10 17:44:50 maybe the last server migration? Jan 10 17:44:51 pb_: I dont remember I think it was something like its reporting to CIA but not to IRC Jan 10 17:45:05 * khem takes a note of it Jan 10 17:45:14 yeah, dunno. I'm not sure if it has ever worked for oe-core. Jan 10 17:46:07 see, I monitor openembedded-commits and there are both Jan 10 17:46:24 meta-oe and oe-core Jan 10 17:46:55 but not bitbake it sems Jan 10 17:47:09 I think bitbake has its own mailing list Jan 10 17:47:27 bitbake commits go the bitbake-devel list Jan 10 17:47:31 not separated from the discussion Jan 10 17:47:33 afaik anyway Jan 10 17:48:01 all that said, since (most) oe-core patches are posted on the list anyway, having cia notifications of them is not all that important I suppose Jan 10 17:48:02 still CIA-9 reported bitbake commits Jan 10 17:48:22 here Jan 10 17:49:09 pb_: yea, having all commits here could make irc annoying Jan 10 17:50:03 I've somehow settled keeping an eye on openembedded-commits Jan 10 17:50:24 it was helpful in the past, but at a certain point the activity is high enough that it can consume the channel Jan 10 17:50:25 * kergoth shrugs Jan 10 17:51:04 we could maybe put this in topic Jan 10 17:52:36 who has topic change access for this channel? eFfeM_work pointed out the bugtracker and tinderbox links probably aren't very useful the other day Jan 10 17:53:11 hmm, not sure, I might still have access to chanserv, will check after i take the dog out Jan 10 17:53:20 good question. it used to be unlocked but I think you need ops nowadays. Jan 10 17:53:22 perhaps someone knoes about this?: * extract_archive: Cannot create symlink from ./var/run to 'volatile/run': File exists. Jan 10 17:53:29 while do_rootfs task Jan 10 17:53:44 -e+w Jan 10 17:54:36 on another subject, is it expected that qt4-embedded-demos.ipk contains a bunch of source code? Jan 10 17:54:47 seems slightly odd, but maybe that's the way things are done in qt-land Jan 10 17:55:12 maybe it is a rootkit ;) Jan 10 17:55:28 pb_: if it's examples, possibly Jan 10 17:55:48 yeah examples Jan 10 17:56:34 cs_nbp: maybe we can make one image that runs from ram only and uses sd / cf for filesystem. this would be perfect for testing... and one that installs to flash Jan 10 17:57:09 pb_: speaking of rights, are you able to give me access to delete branches for meta-opie / meta-handheld git repos on openembedded.org? or should I ask someone else about that? (I already have push access) Jan 10 17:57:13 florian: you mean the chance to brick the device?^^ Jan 10 17:57:52 florian: or copy rootfs to flash? Jan 10 17:57:56 bluelightning: dunno. apparently I need official blessing from the TSC to do that sort of thing nowadays. Jan 10 17:58:08 florian: then i need to buy the 256mb mb Jan 10 17:58:11 so you might need to get them to tell me to do it, or something. Jan 10 17:58:27 pb_: oh... well, you have mine at least ;) I'll raise it at the next meeting, it's not urgent Jan 10 17:58:30 cs_nbp: well... not only, mostly for people who do want to try out Jan 10 17:58:58 florian: ah you're targeting consumers already, havn't thought so far Jan 10 17:59:02 bluelightning: usually one of oe admins should be informed to delete branched Jan 10 17:59:17 send an email to oe-devel on what branches needs to go Jan 10 17:59:18 pb_: really? I wonder if this is really desirable... TSC should have "more important" things to do :-/ Jan 10 17:59:32 bluelightning: sorry, what do you mean about meta-handheld? Jan 10 17:59:37 khem: hmm, personally I think the layer admin ought to have the ability to do it at will Jan 10 17:59:39 moving it away? Jan 10 18:00:02 ant_work: no, I made a branch in meta-opie and now I can't get rid of it Jan 10 18:00:02 bluelightning: it probably is ok when you have pull models Jan 10 18:00:08 but with push it does not work so well Jan 10 18:00:17 ant_work: same might apply for meta-handheld so I figured would get the access there at the same time Jan 10 18:00:25 plus deleting public branches need notice Jan 10 18:00:35 you can delete stuff in contrib trees Jan 10 18:00:39 florian: isn't there some problem with system in flash? when i add packages won't they be lost at reboot? Jan 10 18:00:48 bluelightning: urg...I see Jan 10 18:01:08 khem: perhaps I ought to be using another repo for these branches in future, if that's the recommended method I'm happy with that Jan 10 18:01:53 florian: yeah, I do keep thinking that I should ask the tsc to consider whether it really wants to micro-manage things in that way Jan 10 18:02:24 yes, if the author asks it is unnecessary painful Jan 10 18:02:39 I agree it would make sense for layer maintainers to be able to delete branches in their own repos. Jan 10 18:02:41 Has anyone worked on a bash completion for oe-core/meta/recipes/....?? Jan 10 18:03:01 cs_nbp: no... its persistent, only CE looses some data because it is kept in RAM. But you'll loose CE writing the fs to nand Jan 10 18:03:28 pb_: yes really... Jan 10 18:05:22 florian: oooh but i need ce soo badly Jan 10 18:05:44 florian: can't even install a driver to it without a 36€ cable it seems Jan 10 18:06:18 florian: btw apple picked that up and sells 40€ display adapter cables -.- Jan 10 18:06:30 florian: try to put a BSP in meta-handheld, we'll share the pain together afterwards ;) Jan 10 18:07:42 so really I'm the only one ever getting Cannot create symlink from ./var/run to 'volatile/run': File exists ? Jan 10 18:10:02 normal oe-core task-base Jan 10 18:10:25 cd .. Jan 10 18:11:19 bbl Jan 10 18:13:29 grep -ri 'run/volatile' . run in the overlay dir returned nothing, so it's not from the overlay Jan 10 18:14:53 cs_nbp: I think I had this volatiles problem before Jan 10 18:14:54 cs_nbp: then do not try to write to flash :) Jan 10 18:15:15 this symlink thing Jan 10 18:15:31 yes it bothers me alot atm Jan 10 18:15:35 Its likly to be something Jlime specific... Jan 10 18:15:51 well I had this problem in OE Jan 10 18:15:57 I just built this lxde image but using distro defaults instead of Jlime Jan 10 18:16:08 can't find string 'volatile' in jlime dir Jan 10 18:16:15 ah ok -.- Jan 10 18:16:21 u mean angstrom? Jan 10 18:16:29 no, nothing Jan 10 18:16:33 I had it with an angstrom based distro Jan 10 18:16:35 ah ok Jan 10 18:18:15 the android build system sucks... I want Android in OE ;-) Jan 10 18:18:18 https://gitorious.digitalstrom.org/dss-oe/dss-oe/commit/44b7b46fc97ab5942812f26c71b7561cad06b409 Jan 10 18:18:22 cs_nbp: search the ml this was discussed in some threads in past Jan 10 18:18:47 khem: thanks, ill do that just now Jan 10 18:18:52 florian: I had this idea sometimes :) Jan 10 18:18:55 cs_nbp: have a look at the dbus recipe and at angstroms .bbappend for dbus (in case you are using it) Jan 10 18:19:32 Jin^eLD: srey using Jlime distro atm Jan 10 18:19:40 takin a look anyways ;) Jan 10 18:21:19 khem: actually it should not be that hard... but its a lot of effort. but splitting the stuff into useful packages and having a single device configuration would rock... not speaking about the makefile dependency hell of Android :) Jan 10 18:22:35 yeah android is a bit bsdlike Jan 10 18:22:43 in terms of build system Jan 10 18:27:34 well grep can' find a match if i put stuff in the wrong order, can it -.- Jan 10 18:28:16 it's in udev/init Jan 10 18:53:15 so if udev/init makes a volatile/tmp dir, it should remove it at the end? Jan 10 18:53:52 or the other way round /var/run/ has one too much? that would be strange though Jan 10 18:54:44 cs_nbp: I think in my case the problem was that dbus was installing (touching) /var/run/dbus Jan 10 18:55:23 that means that in the image /var/run became a directory Jan 10 18:55:36 but in my configuration of base_files I wanted /var/run to be volatile Jan 10 18:55:55 so at some point during image generation some step tried to create the /var/run link - which of course failed Jan 10 18:56:04 because /var/run was already there in the image and was a directory Jan 10 18:56:31 similar happening here Jan 10 18:56:39 just hard to trace where it's coming from Jan 10 18:56:44 to solve this problem I had to move directory creation in the dbus recipe from do_install() to some other place (I do not remember if I did it in the init script or in postinst) Jan 10 18:57:16 cs_nbp: go to the place where the workdir of your rootfs is and see what is already in var or var/run Jan 10 19:00:03 it's in work/sysroots, right? Jan 10 19:00:33 I think so, I never remember it... Jan 10 19:00:50 ls sysroots/i686-linux/var/run/ Jan 10 19:00:53 dbus Jan 10 19:00:54 -.- Jan 10 19:01:32 lol Jan 10 19:01:40 so exactly the same problem actually ;) Jan 10 19:01:47 seems so Jan 10 19:02:35 did you reconfigure your var/run to be volatile? I do not remember what it is by default, I changed some stuff in base_files and I thought that this was the reason why this happened Jan 10 19:03:19 I didn't configure anything :D Jan 10 19:04:58 think i got it in dbus-1.init Jan 10 19:05:12 PIDDIR=/var/run/dbus Jan 10 19:05:18 well it's OK if it is in the init script, the problem is if it's in do_install() Jan 10 19:05:19 mkdir -p $PIDDIR Jan 10 19:05:28 ah k Jan 10 19:05:34 the init scripts are not run during rootfs generation Jan 10 19:05:51 so if its in pkg_postinst() or something or in some init script - that should not be a problem Jan 10 19:05:53 dammit :) Jan 10 19:06:18 well, with one addition: it's ok in postinst if it tests the parameter and is not running it during offline image genration Jan 10 19:08:07 where do I find the dbus install? not much in meta/recipes-core/dbus Jan 10 19:08:40 ah i see i was one dir too far down Jan 10 19:11:05 guess it's that then: echo "d messagebus messagebus 0755 ${localstatedir}/run/dbus none" Jan 10 19:11:26 and that mkdir -p ${D}${localstatedir}/run/dbus ${D}${localstatedir}/lib/dbus Jan 10 19:13:19 what should I change ${localstatedir}/run/dbus do, any suggestions? Jan 10 19:14:02 well, in my bbappend that I posted I simply rm -rf ${D}${localstatedir}/run Jan 10 19:14:15 since /var/run is created by the init script anyway we do not need it in the recipe Jan 10 19:14:29 ah sounds logical Jan 10 19:15:38 k now lets see some magic happen -.- Jan 10 19:16:55 ah yes i already removed the var/run/dbus i found there so it can't fail because of that Jan 10 19:17:37 *doing_rootfs ***************............................................... Jan 10 19:19:11 *doing_rootfs *************************...fail.............................. Jan 10 19:20:24 rogntuudjuu same error again Jan 10 19:21:45 i guess dbus dir in var/lib is ok Jan 10 19:21:54 yes, then continue searching Jan 10 19:22:04 I think there was some other package doing the same stuff too Jan 10 19:22:19 well it's harder this time there is nothing in var anymore Jan 10 19:22:30 ah wait theres oher sysroots :D Jan 10 19:26:34 they are all empty Jan 10 19:26:51 i mean if they have var/run it's empty Jan 10 19:27:07 cd Jan 10 19:27:45 uhm.. Jan 10 19:28:27 nostromo and lmbench seem to have a similar command Jan 10 19:28:51 in the install Jan 10 19:29:31 the rm -f goes as last line of install {} right? Jan 10 19:30:18 where? in your .bbappend or are you editing the original recipe? Jan 10 19:30:38 err as a noob im of course editing the original recipes :D Jan 10 19:31:01 then don't forget to bump PR Jan 10 19:31:09 ? Jan 10 19:31:31 well if you edit a recipe for a package that has already been built, then it will not be rebuilt automatically, you will have to increase the PR Jan 10 19:32:03 or rebuild/clean it manually Jan 10 19:32:04 sry I don't get 'pr' Jan 10 19:32:10 PR="r1" Jan 10 19:32:27 id do a bitbake -ccleanall dbus, and any other package that bugs me Jan 10 19:32:27 where 1 is a number that gets incremented, you see that in all packages Jan 10 19:32:34 ok that will be fine then Jan 10 19:32:40 don't bother in that case Jan 10 19:32:42 k Jan 10 19:33:13 txhbtw Jan 10 19:33:47 np Jan 10 19:34:20 yeh now i get pr Jan 10 19:34:49 PR - package revision Jan 10 19:36:41 *doing_rootfs ***************............................................... Jan 10 19:37:59 *doing_rootfs *************************...fail.................typo......-.- Jan 10 19:38:31 still the same? Jan 10 19:38:45 same error that is? Jan 10 19:38:55 believe it or not i put a # at the begining of the line Jan 10 19:38:59 .,. Jan 10 19:40:08 Have all machines been migrated from oe-classic to some layer for oe-core? If so, how should I find the right layer, for example atmel's at91 machines Jan 10 19:40:50 peahonen: I do not think that all machines have been migrated Jan 10 19:41:48 Jin^eLD: is layerindex page in the wiki the best place to look for what has been migrated? Jan 10 19:42:14 I guess, I do not know any other source, I did look there, yes Jan 10 19:44:34 *doing_rootfs ***************............................................... Jan 10 19:44:53 rechecked all sysroot folders var/run again Jan 10 19:45:09 btw there's only pne in i686-linux anyways Jan 10 19:45:12 Jin^eLD: thanks. to continue using it, I should either stay with oe-classic or attempt creating my own layer for it, right? Jan 10 19:46:18 I am using AT91SAM9G20-EK and I did create an own layer for it Jan 10 19:47:51 Jin^eLD: have you shared it anywhere? Jan 10 19:48:25 yes Jan 10 19:48:30 * extract_archive: Cannot create symlink from ./var/run to 'volatile/run': File exists. Jan 10 19:48:35 it's like milking an ant Jan 10 19:48:35 one sec Jan 10 19:49:32 peahonen: https://gitorious.digitalstrom.org/dss-oe/dss-oe/trees/master Jan 10 19:50:11 the oe classic stuff is in org.openembedded.dev, some older snapshot, the oe core stuff is in OE and my own layers are in dS Jan 10 19:50:46 Jin^eLD: ok, thank you. I'll take a look. Jan 10 19:50:46 configurations are in targets/* (where -legacy is for classic configs) Jan 10 19:51:10 cs_nbp: well some package is creating the dir :P do more grepping Jan 10 19:52:30 Jin^eLD: thing is i don't see any dir there any more Jan 10 19:52:47 Jin^eLD: before there at least was var/run/dbus Jan 10 19:53:00 Jin^eLD: grep running Jan 10 19:53:34 well grep for localstatedir as well as var Jan 10 19:53:58 i grep -r '${D}${localstatedir}/run/' Jan 10 19:54:01 combined with "install", I think there was one more package that did some eveil stuff Jan 10 19:54:06 but I do not remember which one it was Jan 10 19:54:17 just have it in memory that I had to fix more than dbus alone Jan 10 19:55:00 i fixed nostromo and lmbench too Jan 10 19:55:31 I do not use these so I did not have to touch them... Jan 10 19:56:31 florian: and? makin some progress? I'm grepping for wrong var/run writes atm -.- Jan 10 19:57:18 cs_nbp: sort of... for some reason rebuilding the kernel in oe fails but apart from this I make progress Jan 10 19:57:25 but need to go home first Jan 10 19:57:27 bbiab Jan 10 20:01:23 florian: here's something to test :D http://www.tworaz.net/downloads/NetBookPro/lxde-image-2012.0-pre0-netbookpro-20120110.rootfs.tar.bz2 Jan 10 20:01:40 actually for everybody with a netbook pro Jan 10 20:30:31 I am using oe-core to build a beaglebone image. I had been working with it for a few weeks with good success. I then tried to install a new setup for a friend and have run into a problem building readline-native Jan 10 20:30:42 I keep getting the error "bbfatal 'oe_libinstall: unable to locate shared library'" Jan 10 20:31:02 I then reinstalled oe on my machine in a chroot and am now getting the same error in that environment Jan 10 20:31:38 I have followed all of the instructions to set up my system(or I think that I have) - could I have missed something that would cause this? Jan 10 20:32:37 readline-native-6.2-r0 Jan 10 20:34:01 bah Kernel panic - not syncing: VFS Unable to mount root fs on unknown-block(0,0) Jan 10 20:34:33 florian: here's something to test :D http://www.tworaz.net/downloads/NetBookPro/lxde-image-2012.0-pre0-netbookpro-20120110.rootfs.tar.bz2 Jan 10 20:34:49 florian but I only get Kernel panic - not syncing: VFS Unable to mount root fs on unknown-block(0,0) Jan 10 20:34:55 florian re-hi btw Jan 10 20:36:32 hm. Jan 10 20:36:41 Jin^eLD: grep gave no further insight, only strange "meta-angstrom/recipes-angstrom/angstrom/omap3-mkcard/omap3-mkcard.sh: No such file or directory" in addition to the files I already fixed Jan 10 20:38:18 sounds like it can't find the rootfs. Possible the boot args beed to be modified before being packaged in the nbkProOs.img? Jan 10 20:38:48 the image i compiled does about the same Jan 10 20:39:05 right, it would Jan 10 20:39:19 did you use tworaz's kernel for his rootfs? Jan 10 20:39:23 or a new one from oe? Jan 10 20:39:27 his Jan 10 20:39:29 and mine Jan 10 20:39:31 hm Jan 10 20:39:52 in 2 secs I can look into menuconfig Jan 10 20:39:57 or maybe 20 Jan 10 20:41:22 cs_nbp: hi... ok, not too bad. the default kernel commandline is defined in one of the tools recipes... Jan 10 20:41:55 florian: kernel panic not too bad? o_O Jan 10 20:42:07 never heard those together :) Jan 10 20:43:19 well I can see the (0x0) in the boot options Jan 10 20:43:35 question is if it's that from the kernel panic Jan 10 20:43:36 cs_nbp: right - for this reason it is not too bad :) Jan 10 20:43:58 cs_nbp: might be the default Jan 10 20:44:15 florian: should be changed? but seems bootloader default, too Jan 10 20:46:36 florian: btw the build failure of rootfs is tracked down to something writing into var/run/ Jan 10 20:47:22 florian: only apart dbus, lmbench and nostromo I can't find anything else - error stays after bitbake -cclean of dbus, nostromo and lmbench Jan 10 20:47:30 It basically needs to be pointed to the second partition on the card, looks like it's looking for thr first one (which is where the /boot is, but not rootfs) Jan 10 20:47:37 installing vbox now, whee. Jan 10 20:47:53 install the right one for usb passthrough ;) Jan 10 20:48:25 yeah Jan 10 20:48:30 already have that set up Jan 10 20:48:45 linux almost done downloading Jan 10 20:48:52 oh bugger, I got the wrong one Jan 10 20:48:53 shte Jan 10 20:49:59 florian: -ccleanall of course Jan 10 21:08:34 I see no way around it: gotta clear out tmp/ again :DD Jan 10 22:09:57 florian: so I have to change the (0x0) to something else indeed? Jan 10 22:10:25 cs_nbp: yes... but some more Jan 10 22:11:34 florian: and lil sidequestion: setting variables to stuff in var/run shouldn't matter (like PIDFILE) for the var/run var/volatile mount error, should it Jan 10 22:12:15 florian: ill go into menuconfig. btw I havn't succeeded in making oe use the altered config Jan 10 22:13:56 florian: tworaz wrote back he doesn't have the mount problem of course Jan 10 22:15:13 cs_nbp: well... its a question of what we do want to achieve Jan 10 22:15:48 the situation is that some bits in the kernel are built as a module which might want to get inti the kernel itself Jan 10 22:15:53 e.g. mtd Jan 10 22:17:22 my idea currently is to keep the flash like it is and boot the kernel from cf into ram Jan 10 22:17:51 ah so the kernel would be kind of in the rootfs Jan 10 22:17:53 and have an initrd which takes the "rootfs" parameters from cmdline Jan 10 22:18:09 so we have a really tiny fs in the kernel Jan 10 22:18:40 but for a usable filesystem we currently need some additional storage Jan 10 22:18:45 this could be Jan 10 22:18:46 sry misunderstood, kernel still comes from sda1 Jan 10 22:19:43 a second cf partition, pcmcia, usb, or sd Jan 10 22:20:15 well 2nd cf partition sounds easiest or not Jan 10 22:20:27 florian: sounds like you could try linux-kexecboot on mtd, as we do for Zaurus Jan 10 22:21:03 then you put a modular kernel in /boot on removable device Jan 10 22:21:26 ant____: the kernel is on cf anyway Jan 10 22:22:13 I see, well, then you can boot from sd, cf, mtd, ... Jan 10 22:23:02 ant____: the bootloader understand cf only... but we could make a nice initfs script that searches for a filesystem if cmdline is wrong Jan 10 22:23:53 once launched, kexecboot passes the mtdparts to the kexec'ed kernel so you don't care about that and have always sane mtd Jan 10 22:24:30 the drawback is that most people are likely not to use mtd Jan 10 22:24:59 yes, ideally this kind of second-stage bootloaders should reside in nand Jan 10 22:25:13 right Jan 10 22:25:27 btw you can just runb kexecboot binary from shell Jan 10 22:25:37 (unmount all before!) Jan 10 22:26:16 so you can try to boot from elsewhere Jan 10 22:26:44 i'm still not really used to these newfangled devices that boot off sd and the like. *shakes head* :) Jan 10 22:27:04 kergoth: heh Jan 10 22:27:25 wha doo you use? 5.25" floppy? Jan 10 22:27:31 :p Jan 10 22:27:45 no.. devices with onboard flash connected to chip selects and most of the setup there Jan 10 22:28:00 one or two stage, not three stage, boot loading :) Jan 10 22:28:11 cs_nbp: you can boot from any device having a kernel driver (and a known filesystem) Jan 10 22:28:15 sounds nice;) , but the nbp is not really 'new' Jan 10 22:29:31 ant__: kernel rc9 can boot to initfs, if I have kexecboot on cf I could copy it over and try to boot my filesystem? Jan 10 22:31:14 yes, just kexec and kexecboot binaries (we ship static binaries) Jan 10 22:31:48 and i'd have to have a copy of the unwrapped kernel in the rootfs? Jan 10 22:32:15 yes, a zImage or uImage is expected in /boot Jan 10 22:32:40 and a corresponding boot.cfg (syslinux style) Jan 10 22:32:56 yeah i'd have to extract that forst, that kernel is only available wrapped Jan 10 22:33:10 ill give it a try Jan 10 22:35:30 when i find the time Jan 10 22:35:34 :D Jan 10 23:27:04 cs_nbp: I managed to boot into a filesystem on SD Jan 10 23:27:20 * florian waits for the first boot to finish Jan 10 23:27:27 florian: k I give up for today no luck in neither resolving var/run mount problem nor kernel recompile, allthough I now understand where I need to move the config to I can't find where menuconf put it Jan 10 23:27:32 florian: kewl Jan 10 23:27:47 florian: I got plenty sd's too Jan 10 23:28:07 :-) Jan 10 23:28:09 florian: what are you booting into? shell? Jan 10 23:28:39 cs_nbp: its the lxde image... but it is still configuring stuff Jan 10 23:29:16 florian: I came to understand this irc stuff has file transfer capabilities ^^ Jan 10 23:31:23 they're weak and requires you to punch a hole through the nat, iirc. you're bette roff with other mechanisms Jan 10 23:32:01 cs_nbp: kergoth is right... I can upload stuff to a server. Jan 10 23:32:09 * florian sees LXDE Jan 10 23:32:22 florian: yeeeh, nive boot time Jan 10 23:32:37 florian: but we don't plan on that every time we turn it on anyways :D Jan 10 23:32:41 I'm playing with it already Jan 10 23:32:54 florian: what? hunt the wumpus? Jan 10 23:33:05 florian: try going left Jan 10 23:33:09 Touchscreen sort of works... but not the keyboard Jan 10 23:33:18 who needs kbd Jan 10 23:33:54 I need one... no free USB cable at hand to connect the laser keyboard ;) Jan 10 23:34:20 try bluetooth keyboard remote from your phone :D Jan 10 23:37:09 android haxxor Jan 10 23:38:02 or build onboard screen kbd in the meantime Jan 10 23:39:14 since touchscreen is working, maybe we can munchhausen us out that way Jan 10 23:39:25 no android device in this room :-) Jan 10 23:39:34 only sdk :D Jan 10 23:39:37 :p Jan 10 23:40:09 so it's onboard Jan 10 23:40:12 or literki Jan 10 23:40:17 or whatever Jan 10 23:40:41 preferable the illume kbd from raffael Jan 10 23:46:51 florian: tworaz thinks "It's possible that rootdelay value is too small. You can change the defaults in boost-bootcode recipe." Jan 10 23:47:17 cs_nbp: use rootwait Jan 10 23:47:49 right Jan 10 23:48:59 I'll try to improve kernel configuration and initfs a little bit... Jan 10 23:50:27 so I simply put "rootwait" without further arguments or should i put a number also? Jan 10 23:51:52 k Jan 10 23:52:09 add where you want to have your rootfs Jan 10 23:52:28 yeah that was already set right to /dev/sda2 Jan 10 23:52:40 so there was ne error there I guess Jan 10 23:54:31 k i'll recompile a kernel and test it, any spontaneous short changes to config or should I let the fingers off to just test rootwait? Jan 10 23:56:03 florian: does the kbd come through the '2nd CPU'? Jan 10 23:56:28 cs_nbp: I think so yes... Jan 11 00:03:03 gawddammit I shopuldn't have cleared the kernel source too -.- Jan 11 00:03:55 tworaz changed quite a lot today... I should rebase my changes Jan 11 00:04:12 argh... but not before I go some sleep Jan 11 00:04:15 onoz? more recent than 18 hours? Jan 11 00:04:54 Unable to fetch URL git://github.com/tworaz/linux.git;protocol=git;branch=v3.1.8-nbpro0 from any source.' wth Jan 11 00:05:01 looks like sleep for me too Jan 11 00:05:33 yeah... he moved some stuff around Jan 11 00:05:59 again? Jan 11 00:06:21 anyone else seen this effect at you clean a target but when you build it again only a pile of setscene tasks runs? Jan 11 00:06:43 yyyupp rings a bell :D Jan 11 00:08:50 btw repo seems to be completely broken atm better don'e check out anything Jan 11 00:09:19 package linux-dummy-1.0-r1: task do_fetch: Started Jan 11 00:09:23 o_O Jan 11 00:13:20 * florian dreams of a large SSD drive for the build server Jan 11 00:14:27 40x 4GB CF RAID0 Jan 11 00:15:39 cs_nbp: rather http://www.golem.de/1201/88917.html Jan 11 00:21:09 hrm... left the USB adapter at the office Jan 11 00:21:13 florian: kind of reminds me a bit of my amiga, i had 4MB RAM of which 2MB were battery powered and lived through a reboot Jan 11 00:21:27 :-) Jan 11 00:21:37 florian: if you put the system in there Jan 11 00:21:40 woosh Jan 11 00:22:50 btw they don't say anything about the price at your link Jan 11 00:22:54 ;) Jan 11 00:24:03 for a good reason ;) Jan 11 00:24:14 yeh not enough space :D Jan 11 00:26:51 cs_nbp: http://www.linuxtogo.org/~florian/files/nBkProOs.img_3.1.1-nbpro0-00627-gedbc7ec Jan 11 00:27:40 hey txh, it's on the way Jan 11 00:28:13 cs_nbp: It boots from first SD partition (tested) and *should* try some other if this fails (not tested yet). Jan 11 00:28:26 SD? really? Jan 11 00:28:39 or you mean sd(x)? Jan 11 00:28:41 It expects an ext4 filesystem on mmcblk0p1 Jan 11 00:28:45 oh Jan 11 00:28:45 nice Jan 11 00:28:57 ah 1st partition Jan 11 00:29:05 for a change Jan 11 00:29:12 crap, now I have no use for the 8GB cf I just bought :P Jan 11 00:29:17 think microsd in adapter should work Jan 11 00:29:22 should Jan 11 00:29:26 if this fails to mount it would try sda2, sdb1... Jan 11 00:29:27 I have one of those handy, actually Jan 11 00:29:29 ah Jan 11 00:29:32 you think it would get recognized? Jan 11 00:29:43 well Jan 11 00:29:49 I doubt sdhc support Jan 11 00:30:06 muriani: I would... I have a camera wich suports CF only :) Jan 11 00:30:15 haha Jan 11 00:30:48 it'd probably cost more to post it to you than I paid for it >_< Jan 11 00:31:04 my 3 IGEL mainboards have only slots for CF hd Jan 11 00:31:22 I doubt that, in germany they are pretty expensive Jan 11 00:31:37 and letter post is comparably cheap Jan 11 00:31:52 right, and cf is cheap here. Jan 11 00:32:00 put it in a bottle, put a gps tag on it and throw it into the ocean. then email the gps tag id Jan 11 00:32:05 hence, possibly more to post than I paid Jan 11 00:32:08 hahaha Jan 11 00:32:09 nice Jan 11 00:32:30 that's either a very slow, or very expensive delivery method Jan 11 00:32:37 * florian patiently and tired waits for the filesystem build to finish Jan 11 00:32:52 whut you can finish it? Jan 11 00:33:00 what am I doin the whole time :D Jan 11 00:33:38 yeah... somehow it finished her one time. Just trying to repeat that Jan 11 00:34:12 just don't update anything or you end up w linux-dummy Jan 11 00:34:24 i already feel like linux-dummy Jan 11 00:35:19 tworaz sent me a patch for upower Jan 11 00:36:05 now it''s too late I put gobject-introspection-native already in there Jan 11 00:36:16 yeeh dload complete Jan 11 00:36:23 ready to brick my book Jan 11 00:37:07 ok since you tried from sd ill try from cf first Jan 11 00:37:40 need to format to ext4 though Jan 11 00:38:31 fat is not really a good idea for linux filesystems ;) Jan 11 00:39:49 was used to it from earlier gta02 Jan 11 00:39:55 android Jan 11 00:40:23 k got kernel and rootfs in place Jan 11 00:41:22 dam Jan 11 00:41:26 no kernel panic : Jan 11 00:41:28 p Jan 11 00:41:46 but like 400000 uncorrectable error Jan 11 00:41:54 -74 ecc error Jan 11 00:42:02 ubi error unable to read Jan 11 00:42:30 is that normal or should I stop that Jan 11 00:42:56 so far its fine... the kernel tries to attach ubi first. but this is likely to fail... Jan 11 00:43:02 ah i know Jan 11 00:43:12 i got a vfat sdcard in the slot Jan 11 00:43:13 no ubi in nand Jan 11 00:43:15 -.- Jan 11 00:43:40 maybe we should make a kernel without nand support so that there is no danger to damage ce Jan 11 00:44:37 dam I pulled th sdcard out and the book bricked w black screen Jan 11 00:44:48 no resetbutton workin Jan 11 00:44:53 takin out batteries now Jan 11 00:45:13 removing the sd card should be harmless Jan 11 00:46:16 uh oh Jan 11 00:46:32 k so i should let a couple o mins of the uncorrectable error pass Jan 11 00:47:01 as admin naturel reflex is to do filesystem check :D Jan 11 00:47:21 but makes no sense -.- Jan 11 00:47:34 heh Jan 11 00:48:27 still uncorrectable error Jan 11 00:50:33 I'll make a kernel image without NAND support if you remind me tomorrow... Jan 11 00:50:48 np i guess^^ Jan 11 00:51:29 err up to which block will that error count up to Jan 11 00:51:47 its quite easy and then there is no danger someone damages ce or whatever is installed in nand Jan 11 00:52:39 k that's 7 straight minutes of uncorrectable error -.- really normal? Jan 11 00:52:46 or should I try SD Jan 11 00:53:27 hmm no... it does not make sense. its because the driver does not understand the ecc data ce uses Jan 11 00:53:45 I do not see this because my nand was formatted using Linux Jan 11 00:53:54 formatted/written Jan 11 00:54:07 k, so it may take a loooong time to boot :D Jan 11 00:55:02 cs_nbp: http://www.linuxtogo.org/~florian/files/jlime-lxde-image-netbookpro.tar.gz Jan 11 00:55:08 but wait Jan 11 00:55:16 its still uploading for ~10min Jan 11 00:55:26 need sleep Jan 11 00:55:30 Good night! Jan 11 00:55:36 well waiting is what I'm doin anyways Jan 11 00:55:38 gnZ Jan 11 00:55:54 'night Jan 11 00:56:02 gn Jan 11 00:58:18 *waiting *******************....................................................................................................................................................... Jan 11 00:59:07 yay I can see lxde Jan 11 01:01:16 but can't login Jan 11 01:01:24 but anyways am amazed Jan 11 01:03:36 just need to edit away login screen and install onscreen kbd Jan 11 01:29:54 k cyaZall tomorrow zZzzZZZZZzz **** ENDING LOGGING AT Wed Jan 11 02:59:57 2012