**** BEGIN LOGGING AT Thu Aug 18 02:59:56 2011 Aug 18 05:11:08 khem: I'm pretty sure I was using 4.6 before holidays and n900 image was building fine Aug 18 05:11:40 khem: so maybe 4.6 revision was changed or it's new dependency and wasn't built in my image before Aug 18 06:34:30 JaMa: I would doubt that. if you look at the bug http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50104 Aug 18 06:34:52 you will see that its a regression caused by a patch thats in gcc 4.6 branch in FSF Aug 18 06:37:32 khem: I was using gcc-4.6 since 2011-05-23 http://git.shr-project.org/git/?p=meta-smartphone.git;a=commit;h=8b58481b11bbfb4bf0200565f13ca8737e27a02b Aug 18 06:41:40 khem: but it can be new in 0.15 and 0.12 didn't fail http://git.openembedded.org/cgit.cgi/openembedded-core/commit/?id=0d240f56ea0128c1228f7cfe1471039a7b602a88 Aug 18 07:31:30 hello Aug 18 07:31:43 can anyone help me how to update oe Aug 18 09:01:29 hi ant_work Aug 18 09:01:32 hi bluelightning, all Aug 18 09:01:43 good morning Aug 18 09:01:44 hi pb Aug 18 09:01:47 hm intressting Aug 18 09:01:54 many people wants the pre3 Aug 18 09:05:33 can anyone help me how to update oe Aug 18 09:05:57 hi all Aug 18 09:06:02 sanket: "git pull" Aug 18 09:08:05 sanket depends what oe you are using Aug 18 09:08:19 pb I am always using git pull --rebase Aug 18 09:09:02 woglinde, i didn't get you... and how can I find what oe I m using? Aug 18 09:09:17 sanket what did you do to get oe? Aug 18 09:14:17 woglinde, I had used this usermanual http://www.google.com/url?sa=t&source=web&cd=1&ved=0CBgQFjAA&url=http%3A%2F%2Fdocs.openembedded.org%2Fusermanual%2Fusermanual.pdf&rct=j&q=openembedded%20pdf&ei=59ZMTs-YHMusrAfMmuyMAw&usg=AFQjCNHVjFrSuFMgp5Wd5Uw3rjcCFOfTXA&cad=rja Aug 18 09:14:31 google.com? Aug 18 09:14:38 okay Aug 18 09:14:42 so its classic Aug 18 09:14:45 git pull Aug 18 09:15:32 woglinde, I had used that ..... but doesnt shows anything .... Aug 18 09:16:32 ? Aug 18 09:17:07 it has stuck at Aug 18 09:17:31 than check the network connection Aug 18 09:17:59 network connection is working fine Aug 18 09:20:51 when I used this ..... ................... at the end I m getting .......refusing to pull with rebase: your working tree is not up-to-date Aug 18 09:21:40 git status Aug 18 09:21:47 you changed something locally Aug 18 09:21:52 either commit it Aug 18 09:21:56 or use git stash Aug 18 09:21:57 git pull Aug 18 09:22:00 git stash apply Aug 18 09:24:30 hi esben Aug 18 09:35:38 hi Aug 18 09:58:58 heh.. today GMail runs at moto-speed (GPRS ?) Aug 18 09:59:40 doh, now is better Aug 18 10:55:06 hi mickeyl Aug 18 11:40:15 Hello; I am building a sdk and noticed that runqemu uses 'uname -i' to look for libGL.so but this seems to fail in some cases Aug 18 11:40:42 otavio there is a fix already Aug 18 11:40:50 which uses a glob instead of uname Aug 18 11:40:55 search the oe-core ml Aug 18 11:45:03 woglinde: cool, nice Aug 18 12:08:27 Hi Aug 18 12:11:18 I have a custom embedded board that is very similar to TI's beagle board. I also have a quite old kernel. Since there is some trouble with it I'd like to switch to a new one. I fetched it ("tlind"-tree), compiled it (omap2plus_defconfig) and flashed it. When starting, I get no console. Aug 18 12:11:49 However, if I use a prebuilt image from web, I do get a console with the very same boot args. Aug 18 12:12:45 aks4: serial port name switched from ttyS to ttyO this may explain your problem Aug 18 12:12:58 I also noticed that I get only "Uncompressing Linux... done, booting the kernel." when it fails. When using a kernel for which it succeeds, I get a lot of further dots before it says "done": Aug 18 12:13:16 he ericben Aug 18 12:13:20 hi woglinde Aug 18 12:13:30 aks as ericben said Aug 18 12:14:10 aks4 and how new your kernel should be? Aug 18 12:14:26 you can use the beagleboard kernel and patches from the meta-ti layer Aug 18 12:14:29 Doesn't help. Aug 18 12:15:05 aks4: without knowing the version, the command line, the bootargs, the bootlog, it's difficult to help you Aug 18 12:15:13 Ok, here they come: Aug 18 12:15:22 aks4: also maybe you have a custom board id thus using a beagle kernel won't work Aug 18 12:15:27 bootargs: console=ttyO2,115200n8 ramdisk_size=8192 root=/dev/ram0 rw rootfstype=ext2 initrd=0x81600000,8M Aug 18 12:15:37 Image Name: Linux-3.0.0 Aug 18 12:16:40 ericben I read you have played with fsl_udc on bareboot Aug 18 12:16:44 The board is very similar to the beagle board. And, using a beagle kernel I can at least see the kernel booting. It stops somewhere later when mounting the rootfs. But that is ok, since I assume it has a different ethernet PHY. Aug 18 12:16:51 woglinde: yes I did Aug 18 12:16:56 woglinde: for which cpu ? Aug 18 12:17:01 is the serial console stuff now fixed? Aug 18 12:17:13 I would need it for tegra Aug 18 12:17:37 but maybee bring fsl_udc with the tegra patches to uboot is easier Aug 18 12:18:24 woglinde: from memory I only tried DFU Aug 18 12:18:52 woglinde: and was using netconsole for the console over eth Aug 18 12:19:07 woglinde: is tegra using fsl_udc ? Aug 18 12:19:28 woglinde: by bareboot, you mean barebox ? Aug 18 12:20:19 woglinde: btw with DM3730 in recent u-boot usb serial console is broken (it's loosing some chars) Aug 18 12:20:46 hey hiii .... is it possible to integrate ... MFC's application libraries into MPlayer .... Aug 18 12:21:17 I assume the kernel is not built correctly.Before "done, booting the kernel" the kernel itself should not matter, right? It should uncompress, even if built for a totally different system, right? If so, why does my own kernel produce just three "dots" and the beagle one I fetched from net some 100 or so? They are similar in size: 3.3 MB for my failing one and 3.1 MB for the beagle one. Aug 18 12:21:34 since I cant able to run H.264 videos on mini6410 Aug 18 12:22:02 aks4: no if you board has a different board id than beagleboard, kernel won't boot Aug 18 12:22:19 aks4: enable low level debugging and maybe you will get the error message Aug 18 12:25:11 ericben: Thanks, I'll check the number Aug 18 12:29:41 ericben: Should be ok. "bdinfo" says "arch_number = 0x0000060A". That is beagle board according to "arch/arm/tools/mach-types". And, beagle board support is included in .config: CONFIG_MACH_OMAP3_BEAGLE=y Aug 18 12:30:11 ericen yes baerbox sorry Aug 18 12:30:24 ericben yes tegra is using with fsl_udc with some patches Aug 18 12:30:53 ericben: Rebuilding now with low level debugging enabled (and early_printk) Aug 18 12:34:53 ericben: No messages even with lowlevel debug Aug 18 12:37:53 ericben: I'll try a different toolchain now... Aug 18 12:42:12 woglinde: I can't be of big help on this now sorry Aug 18 12:44:53 ericben only wanted to knwo the status Aug 18 13:01:42 ericben: Different toolchain didn't help. I'll now try to investigate why compressing is so "fast" Aug 18 13:02:44 aks4 whats the size of your kernel? Aug 18 13:03:01 3.1 MB Aug 18 13:03:37 I guess that somehow the size passed to do_decompress() is wrong Aug 18 13:04:12 woglinde: I am just adding code to print it to the screen to find out. Aug 18 13:10:29 output_data: 2147516416 Aug 18 13:10:29 free_mem_ptr: 2157433032 Aug 18 13:10:29 free_mem_end_ptr: 2157498568 Aug 18 13:10:29 arch_id: 1546 Aug 18 13:10:29 output_start: 2147516416 Aug 18 13:10:29 input_data_end: 2157428842 Aug 18 13:10:29 input_data: 2154217524 Aug 18 13:11:19 That is 3,211,318 bytes input data. Sounds ok. Aug 18 13:18:41 yes Aug 18 13:20:53 Ok, the decompression is done using gunzip in decompress_inflate.c and it is successfully called. Aug 18 13:26:52 I remember miscompilation of decompress code some months ago when optimize-for-size is enabled but it went away and I could never reproduce it Aug 18 13:28:17 perhaps a pattern in the binary triggered a bug in the uncompress algorithm, perhaps it was a compiler bug Aug 18 13:31:39 Hm, that would explain it. But I tried several kernels and three toolchains in total.... I'll try to switch to LZO compression or so. That should give at least a somehow different behavior. Aug 18 13:41:41 Switched to LZMA. Now the kernel image is smaller (2.3 MB) and decompression takes longer (2 or 3 seconds). But still the same: No dots are printed, no messages after it returned to misc.c and said it would boot the kernel now. Aug 18 13:43:20 hm Aug 18 13:49:02 Is there something to verify if the kernel decompression was successful? Some magic number for example? Aug 18 13:55:42 aks4: https://groups.google.com/forum/#!topic/beagleboard/N7zIEYQCpKI/discussion maybe you can start by testing this 3.0 kernel for BB mentioned in this email Aug 18 13:56:39 aks4: and does a beagleboard kernel ever worked on your board ? Aug 18 14:04:30 How I can pass extra kernel options to runqemu Aug 18 14:04:31 ? Aug 18 14:09:29 ericben: Yes, I downloaded a beagleboard kernel from web and it booted. It hung later on with a panic saying it can't mount the root filesystem but that was ok. Aug 18 14:12:08 hi gnutoo Aug 18 14:27:02 Interesting. The kernel presented in the googlegroups-posting you mentioned has the same issue. Just like mine. This one works: http://www.angstrom-distribution.org/demo/beagleboard/ Aug 18 14:27:21 (ignoring the rootfs panic) Aug 18 14:49:38 hi woglinde Aug 18 14:52:00 gnutoo: http://wiki.shr-project.org/trac/wiki/Building%20SHR#Cannotparserecipesordo_rootfsafteropensslupgrade Aug 18 14:54:18 ok I'll look thanks to have made a wiki page Aug 18 14:54:51 because while I know how to cope with that kind of failures not everybody does..... Aug 18 14:54:55 gnutoo: I've added it when nshle asked the same :) Aug 18 14:55:49 gnutoo: btw also initscripts-shr are gone.. so you have to rebuild task-core-boot and keymaps again after next update Aug 18 14:57:01 ok Aug 18 14:58:28 btw is there other alternatives from making a branch of oe-core and oe when you do a bsp? Aug 18 14:59:03 what could go wrong if you do not lock the revisions? Aug 18 15:00:41 .bbappends which doesn't append anymore etc Aug 18 15:00:56 gnutoo: for simple bsp won't happend so often Aug 18 15:01:13 ok Aug 18 15:01:14 gnutoo: but from meta-smartphone I'm moving stuff to upstream layers quite often Aug 18 15:01:25 ok Aug 18 15:03:16 JaMa: pls think about moving kexecboot in meta-oe and adding a .bappend with EXTRA_OECONF_zaurus Aug 18 15:03:22 btw is there some docs about .bbappend Aug 18 15:03:23 hi ant_work Aug 18 15:03:29 hello gnutoo Aug 18 15:04:04 about kexecboot: can I pass /proc/atags of the first kenrel to the second one? Aug 18 15:04:18 or is that idea broken? Aug 18 15:04:21 ant_work: I've sent pull request for kexec-tools today.. you can do whatever you want with kexecboot :) Aug 18 15:04:22 for arm there was long ago a patch for passing atags Aug 18 15:04:29 ok Aug 18 15:04:29 I think was mwester worls Aug 18 15:04:33 *works Aug 18 15:04:42 and indeed it should work Aug 18 15:05:14 JaMa: it is to minimize the meta-zaurus Aug 18 15:05:49 ok,so I find that patch and refresh it, is it a kenrel patch only? or is there an --option of kexec userspace program too? Aug 18 15:05:51 kexecboot is LF/CELF sponsored so could even belong to oe-core Aug 18 15:06:01 gnutoo: there isn't much to document, really. make sure the bbappend is in bbfiles (most layers already do that, if you're using oe-core or yocto), and make sure the filename is exactly the same as the .bb you're applying it to, other than extension Aug 18 15:06:02 ant_work: I know.. I needed kexec package for ie n900 outside meta-zaurus.. but I don't need kexecboot outside yet.. so you can move it if you need.. Aug 18 15:06:42 kergoth: some bbappend only append, some have FILEPATH and simillar stuff.... Aug 18 15:06:51 ? Aug 18 15:06:57 bbappend is just recipe metadata, period Aug 18 15:06:59 it's not magic Aug 18 15:07:01 ok Aug 18 15:07:18 some add to filespath, yes, to make sure the dir the bbappend is in is also in filespath, so patches local to it are found Aug 18 15:07:19 then I'll ask another question in a moment, let me find the correct .bbappend Aug 18 15:07:29 ok Aug 18 15:07:34 thanks a lot Aug 18 15:07:34 ericben: Ok, I'm giving up for now. Running out of options. I guess that somewhere shortly after 2.6.32 a change was made that works with beagle boards but not with mine. Aug 18 15:07:48 but its just metadata, not magic. recipes happen to automatically get the dir they live in put into filespath, so they don't have to do it explicitly, is all Aug 18 15:07:51 that's the main difference there Aug 18 15:08:11 kergoth: btw it seems that using something like THISDIR := "${@os.path.dirname(bb.data.getVar('FILE', d, True))}" is quite common in bbappends Aug 18 15:08:12 the convention is to use a FILESEXTRAPATHS variable rather than manipulating FILESPATH directly, though Aug 18 15:08:12 ahhh now I understand Aug 18 15:08:27 kergoth: today I had issue with it when I had 3 .bbappend files for one recipe Aug 18 15:08:29 (with the fact that it works by default) Aug 18 15:08:52 JaMa: the bit added to filespath has to be force expanded, not the variable it references, otherwise you hit tha tproblem Aug 18 15:08:52 kergoth: and only last THISDIR won, so I've renamed THISDIR in every bbappend file and it works fine Aug 18 15:08:56 each bbappend is overwriting the others thisdir Aug 18 15:08:57 right Aug 18 15:09:00 just do this Aug 18 15:09:02 FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" Aug 18 15:09:06 that, yes Aug 18 15:09:13 :) Aug 18 15:09:21 JaMa: don't set THISDIR with oe-core, it's done automatically Aug 18 15:09:25 th so something like FILESPATH =. "${@base_set_filespath(["${THISDIR-NOKIA}/${P}"], d)}:" is the cause of my problems here, right? Aug 18 15:09:34 JaMa: yep Aug 18 15:09:45 ok I'll update it Aug 18 15:09:47 thanks Aug 18 15:09:57 kergoth: what would you say to having this sort of thing done automatically for bbappends? Aug 18 15:10:10 slightly unorthodox, but could reduce pain Aug 18 15:11:02 heh, back when using collections.inc, i used a snippet that added the recipe dir in every layer to filespath, but nowadays we try not to bloat filespath like that Aug 18 15:11:15 not sure how else you'd do it, unless bbappends show up in the recipe __depends, guess you could search that Aug 18 15:16:40 Bye & Thanks! Aug 18 15:37:11 is there any way to pass extra kernel args to runqemu script? Aug 18 16:56:08 otavio: anything you pass with <> will pass as argument Aug 18 16:58:16 khem: but how I add more boot params into --append? Aug 18 16:59:22 otavio: SCRIPT_KERNEL_OPT = "options you want" Aug 18 17:00:24 otavio: thats with oe-core runqemu script thats what you are asking for I assume Aug 18 17:00:42 khem: yes, oe-core Aug 18 17:01:54 khem: seem to not work using SCRIPT_KERNEL_OPT Aug 18 18:06:22 whats the preferred usage between ARCH and TARGET_ARCH? Aug 18 18:06:28 when dealing with recipes Aug 18 18:28:19 msm: TARGET_ARCH most likely, but what's the context exactly? what are you trying to achieve? Aug 18 18:38:49 bluelightning: to pass to build scripts Aug 18 18:39:04 i just noticed that recently ARCH stopped working so I moved to TARGET_ARCH Aug 18 18:39:15 im just curious what is correct/why one stopped working Aug 18 18:49:05 hi stefan_schmidt Aug 18 18:53:08 msm: not sure, perhaps someone else could advise Aug 18 18:53:57 bluelightning: np Aug 18 18:54:45 well ARCH should not be there to use Aug 18 18:59:40 khem: right, its not anymore… i want to say it *was* Aug 18 18:59:48 because it worked at some point... Aug 18 18:59:54 did it recently go away? Aug 18 19:00:05 or maybe something else was wrong with my env Aug 18 20:23:58 Do someone has any clue why my init do not receive sigusr1 from X but if it is run in a shell it receives? Aug 18 20:55:39 msm: IIRC it never was there Aug 18 21:00:56 khem: ok im not sure what was going on then… maybe it was a kernel.bbclass thing, since that was the recipe I was dealing with Aug 18 21:01:15 yes in kernel and kernel headers there is ARCH Aug 18 21:03:45 JaMa: ping Aug 18 21:05:04 JaMa: the 'wrong' device_table-minimal.txt is used, the one in oe-core :/ Aug 18 21:27:06 khem: what is the purpose of adding ARCH just for these? why dont they just use TARGET_ARCH? Aug 18 21:31:31 msm: since target arch is not always same as kernel arch Aug 18 21:35:44 kergoth: as in ppc vs. powerpc? Aug 18 22:19:03 gm Aug 18 22:19:43 I'm having issues with my images failing due to a conflict between libgcc-dev and libgcc-s-dev both suppling the same file Aug 18 22:20:15 is there a simple way to blacklist one or the other? Aug 18 22:20:22 hi Aug 18 22:20:40 hi lumag **** ENDING LOGGING AT Fri Aug 19 02:59:57 2011