**** BEGIN LOGGING AT Thu Aug 27 02:59:57 2009 Aug 27 03:04:49 hi, here... Aug 27 03:05:10 did somebody build directfb in oe ? Aug 27 03:07:18 scheng: there seems to be recipes there so someone have built them at certain point Aug 27 03:11:09 khem: i am starting to use oe now, so maybe the question is simple... but when i run "bitbake directfb" ,there is not directb libary installed in rootfs.. am i missing something ? Aug 27 03:11:49 libraries are only installed when a package needs them Aug 27 03:12:18 So your rootfs recipe needs to depend on directfb-sample-apps or whatever it is Aug 27 03:13:40 Tartarus: did you mean i need to change some file on this ? Aug 27 03:14:08 scheng, doing 'bitbake directfb' doesn't result in making a rootfs Aug 27 03:14:58 Tartarus:oh, sorry.. i'm starting to use oe today.. could u please tell me more about it ? Aug 27 03:15:06 Ah, sec.. Aug 27 03:15:25 Tartarus:okay, thanks advance.. Aug 27 03:15:59 http://docs.openembedded.org/usermanual/usermanual.html Aug 27 03:16:05 search for 'images' Aug 27 03:23:28 anybody familiar with insight debugger Aug 27 03:27:40 mankash: its just a gtk frontend to gdb Aug 27 03:28:25 how do I see the code in debugger window, it show me only assembly code Aug 27 03:28:54 compile with debug info Aug 27 03:29:17 what is the flag Aug 27 03:29:19 g3 Aug 27 03:31:02 -g would be enough Aug 27 03:33:12 that option is already there in CFlags Aug 27 03:37:54 ok Aug 27 03:38:01 then do you have sources in right place Aug 27 03:39:40 yes Aug 27 03:44:45 what is your host machine Aug 27 03:44:49 running Aug 27 03:46:04 ubuntu Aug 27 03:47:29 does it work with gdb Aug 27 03:49:16 yes Aug 27 04:06:10 mankash: there are some settings to set the source path IIRC Aug 27 04:08:17 you can add dir in ~/.gdbinit and retry Aug 27 04:08:35 if it works then somehow insight is not seeing the src paths Aug 27 04:11:05 * * OE Bug 5297 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 27 04:11:07 * * python-pyqt fails do_compile Aug 27 04:11:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5297 Aug 27 04:28:36 khem: hey, do you happen to know how I can detect if soft float or hard float was used in a binary (besides disassembling and search for the ops)? Aug 27 04:33:09 zecke: it should be in the elf header Aug 27 04:38:42 or if you have newer toolchain and EABI then you could see it in .ARM.attribute section Aug 27 04:41:59 arm-none-linux-gnueabi-readelf -A will dump them for you Aug 27 04:44:43 Tag_VFP_arch: VFPv3... okay... hmm Aug 27 04:45:38 hmm but I think its not definite Aug 27 04:45:50 in OABI you could see it in ELF flags Aug 27 04:45:58 but in EABI I dont see it Aug 27 04:47:23 Tag_VFP_arch is your best bet Aug 27 04:48:53 khem: yeah, I remember the ELF flags with OABI Aug 27 04:49:16 * zecke is just optimizing for armv5t and sets the float abi to soft Aug 27 04:49:46 zecke: do you have hardware with armv5t with fpu Aug 27 04:50:09 I remember some philips part vaguely which was armv5t and had fpu Aug 27 04:50:52 khem: no, not at all, I want to see the difference between soft float (as we should use normally) and neon Aug 27 04:55:05 ah ok Aug 27 04:55:16 zecke: soft float will be better Aug 27 04:55:27 unless you test something like mplayer Aug 27 05:10:06 zecke: do you know where I can access gentoo sources online Aug 27 05:10:57 khem: ebuilds or distfiles? Aug 27 05:12:17 f\distfiles? Aug 27 05:13:00 i mean distfiles yes Aug 27 05:13:04 khem: ftp://ftp.osuosl.org/pub/gentoo/distfiles/ Aug 27 05:13:13 khem , i am in the process of moving my work to the latest git and using 4.4.1+uclibc, thanks for the follow-up on #5282 Aug 27 05:15:00 m4t ok Aug 27 05:26:12 radhermit: it has all the sources+patches Aug 27 05:26:46 khem: Didn't you want the distfiles? Aug 27 05:32:01 infact yes thats what I needed my machine is slow in opening them Aug 27 05:33:18 Well, you could try another mirror -- http://www.gentoo.org/main/en/mirrors2.xml Aug 27 05:37:53 or use cmdline ftp Aug 27 05:38:04 firefox has its limits Aug 27 05:40:21 Yeah, firefox lags on long file lists... Aug 27 05:48:13 03Tushar Gohad  07org.openembedded.dev * r0b3b3824dc 10openembedded.git/recipes/tcpdump/ (files/ipv6-cross.patch tcpdump_3.9.7.bb tcpdump_4.0.0.bb): Aug 27 05:48:13 tcpdump: obey distro for the ipv6 support Aug 27 05:48:13 When cross compiling, get rid of the AC_TRY_RUN Aug 27 05:48:13 style checks for ipv6, obey DISTRO_FEATURES. Aug 27 05:48:13 Signed-off-by: Tushar Gohad Aug 27 05:48:17 Signed-off-by: Chris Larson Aug 27 06:06:32 has anyone set up swap on nand flash before? Aug 27 06:07:16 using a raw mtdblock as swap appears to be a bad idea, due to bad blocks Aug 27 06:07:34 and flash filesystems dont usually support swap files Aug 27 06:07:45 I'm looking for other options Aug 27 06:09:08 grg: Maybe a swap file on top of nand? Aug 27 06:09:33 radhermit, on which filesystem? Aug 27 06:09:49 i'm using yaffs2 for root. but that does not allow swap files Aug 27 06:09:59 Ah Aug 27 06:10:46 Haven't used yaffs2 much so I don't know... Aug 27 06:10:48 i don't know for sure, but i doubt jffs2 does either Aug 27 06:11:10 Let me try... Aug 27 06:11:54 on yaffs22, i get: 'swapon: swapfile has holes' Aug 27 06:16:49 Not having too much luck on jffs2 either Aug 27 06:18:04 * * OE Bug 5036 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 27 06:18:06 * * tzdata2007k.tar.gz no longer downloadable Aug 27 06:18:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5036 Aug 27 06:20:27 I might investigate UBI Aug 27 06:20:47 I'm running into the swapon: Invalid argument problem Aug 27 06:27:13 Has anyone run into the issue where the binaries or libraries are placed in different places on different build machines? Aug 27 06:27:25 For example, on one of my build machines cacaoh-native builds fine Aug 27 06:27:55 but on the other it looks in .libs for the binary and its in the main dir instead -- src/cacaoh Aug 27 06:30:52 so of course the one machine fails in the do_stage phase Aug 27 06:35:27 03Khem Raj  07org.openembedded.dev * r4bd9a03ac9 10openembedded.git/ (conf/checksums.ini recipes/diffstat/diffstat_1.39.bb): Aug 27 06:35:27 diffstat_1.47.bb: Add recipe for new release Aug 27 06:35:27 diffstat_1.39.bb is moved out the tars dont exist on debian servers Aug 27 06:35:27 anymore. Aug 27 06:35:27 Signed-off-by: Khem Raj Aug 27 06:35:29 03Khem Raj  07org.openembedded.dev * rb3f1fb663d 10openembedded.git/recipes/tzdata/tzdata_2007k.bb: Aug 27 06:35:32 tzdata_2007k.bb: Update SRC_URI Aug 27 06:35:34 * The tar file is removed from the original url. Aug 27 06:35:36 Signed-off-by: Khem Raj Aug 27 06:35:38 03Khem Raj  07org.openembedded.dev * r813fb1632d 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 27 06:35:43 03Khem Raj  07org.openembedded.dev * r865ed19cd3 10openembedded.git/: Merge branch 'develop' into org.openembedded.dev Aug 27 06:36:07 * kergoth tries to determine if any of our tasks make alterations to the datastore without making a copy first Aug 27 06:37:56 sheesh, elc europe 09 has presentations on OpenEmbedded (2 of them), ptxdist, e2factory, openwrt... people do love their build tools of choice.. :) Aug 27 06:38:04 * * OE Bug 5140 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 27 06:38:06 * * diffstat SRC_URI needs an update Aug 27 06:38:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5140 Aug 27 06:49:21 Hi all Aug 27 06:51:22 I am having one problem with bitbaake Aug 27 06:51:27 bitbake Aug 27 06:52:11 I am able to compile kernel with bitbake virtual/kernel -c compile -f Aug 27 06:52:38 What do I need to do for compiling all modules ? Aug 27 06:53:03 with bitbake ? Aug 27 07:03:57 03Khem Raj  07org.openembedded.dev * r628d887f5c 10openembedded.git/ (3 files in 3 dirs): Aug 27 07:03:57 dash_0.5.2.bb: Add recipe for latest dash. Aug 27 07:03:57 Remove unbuildable dash_0.5.2 recipe. Aug 27 07:03:57 Add checksums for new dash and diffstat Aug 27 07:03:57 Signed-off-by: Khem Raj Aug 27 07:12:52 03Khem Raj  07org.openembedded.dev * r1482876c97 10openembedded.git/conf/checksums.ini: Aug 27 07:12:52 checksums.ini: Sort and remove duplicates. Aug 27 07:12:52 Signed-off-by: Khem Raj Aug 27 07:14:30 03Chris Larson  07org.openembedded.dev * r4e00d90a75 10openembedded.git/recipes/autofs/ (autofs-4.1.4/Makefile.rules-cross.patch autofs_4.1.4.bb): Aug 27 07:14:30 autofs: fix ldflags issue, missing debug info, duplicate opt flags. Aug 27 07:14:30 Signed-off-by: Chris Larson Aug 27 07:15:49 hmm, looks like the main abuses of tasks changing metadata are pstage_set_pkgmanager and the read of the subpackage metadata Aug 27 07:15:57 thus far, anyway Aug 27 07:18:03 * * OE Bug 5044 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 27 07:18:05 * * The source archive for dash 0.5.2 is no longer downloadable Aug 27 07:18:07 * * http://bugs.openembedded.net/show_bug.cgi?id=5044 Aug 27 07:30:13 03Khem Raj  07org.openembedded.dev * rf15dec06d8 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 27 07:30:31 03Khem Raj  07org.openembedded.dev * r706d50df60 10openembedded.git/recipes/iputils/iputils_20071127.bb: Aug 27 07:30:31 iputils: Make iputils_20071127 default. Aug 27 07:30:31 Signed-off-by: Khem Raj Aug 27 07:37:22 morning Aug 27 07:39:06 I've compiled a gtk code using armv4t-angstrom toolchain for mini2440, but when I try to start it, I get this error "Alignment trap: eray-gtk (1913) PC=0x408133c4 Instr=0x280069c0 Address=0xffffffff FSR 0x813 Aug 27 07:39:06 Segmentation fault" Aug 27 07:39:26 sx: what did you google for already? Aug 27 07:39:58 good morning Aug 27 07:39:58 zecke, nothing. I couldn't find anything Aug 27 07:41:37 I wrote the following for debug and I don't see the "init" Aug 27 07:41:37 printf("start\n"); Aug 27 07:41:37 gtk_init(NULL, NULL); Aug 27 07:41:37 printf("init\n"); Aug 27 07:45:04 * * OE Bug 5141 has been RESOLVED (FIXED) by raj.khem(AT)gmail.com Aug 27 07:45:05 sx: alignment trap hmm, one way you could let kernel handle the traps Aug 27 07:45:07 * * iputils does not build Aug 27 07:45:09 * * http://bugs.openembedded.net/show_bug.cgi?id=5141 Aug 27 07:45:46 sx: or you can look at the build logs of gtk and see if there are warnings about conversions Aug 27 07:45:52 and fix them Aug 27 07:51:45 sx: what did you google for? Aug 27 07:53:19 zecke, I couldn't find nothing in google about the error. Aug 27 07:53:37 sx: which terms did you use to google? Aug 27 07:54:13 zecke, Alignment trap Segmentation fault gtk Aug 27 08:04:29 mickeyl: good morning Aug 27 08:05:02 sx: this is not primarily an alignment fault, that's just a red herring. Aug 27 08:05:15 the address 0xffffffff has much worse things wrong with it than being misaligned. Aug 27 08:05:47 morning pb! Aug 27 08:06:17 sx: you just need to debug this like any other segfault. Aug 27 08:07:10 pb___, how? It fails on gtk_init function. Aug 27 08:08:11 sx: run it under the debugger like you would with any such crash Aug 27 08:35:10 good morning Aug 27 08:36:23 Hey. Aug 27 08:37:39 Is there a magic mode in oe where you can directly edit source code that wont compile, and it will figureout the changes,generate the patches, and put it the the right recipe? Aug 27 08:39:14 florian: good morning Aug 27 08:39:27 tardyp: not yet... but I have been thinking about making this possible several times :) Aug 27 08:39:29 tardyp: sadly I don't think so, but that would be an awesome feature to add. Aug 27 08:39:45 tardyp: kergoth has been working on improving oe for interactive use, you might talk to him about it. Aug 27 08:47:44 03Florian Boor  07org.openembedded.dev * r1760484c1f 10openembedded.git/recipes/cairo/cairo-native_1.8.0.bb: cairo: add native 1.8.0 Aug 27 08:47:45 03Florian Boor  07org.openembedded.dev * r5fec88660d 10openembedded.git/conf/checksums.ini: checksums.ini: A lot of maemo checksums and some updated bits. Aug 27 08:47:45 03Florian Boor  07org.openembedded.dev * r33fa619402 10openembedded.git/recipes/gtk+/gtk+-native_2.12.11.bb: gtk+: add simple native gtk+ 2.12.11 with all modules built in. Aug 27 08:47:47 03Florian Boor  07org.openembedded.dev * rbe459b4f9b 10openembedded.git/recipes/pango/ (pango-native-1.22.0/no-tests.patch pango-native_1.22.0.bb): pango: Add native pango 1.22.0 Aug 27 08:47:50 03Florian Boor  07org.openembedded.dev * r9e7085ede3 10openembedded.git/recipes/atk/atk-native_1.27.90.bb: atk: add 1.27.90 native atk Aug 27 08:47:53 03Florian Boor  07org.openembedded.dev * r448bbbf73c 10openembedded.git/recipes/libpixman/libpixman-native_0.12.0.bb: libpixman: Add native 0.12.0 Aug 27 08:52:07 morning Aug 27 08:52:30 florian: merging maemo/oe work? Aug 27 08:53:57 hi, is there somebody get the beagle-board directfb benchmark result with hardware acceleration ? Aug 27 08:54:31 hrw: not yet... but this is for maemo already. Aug 27 09:13:25 03Koen Kooi  07org.openembedded.dev * rf4dd0b7005 10openembedded.git/recipes/julius/ (julius/patch-julius-4.1.2-command-pipe.patch julius_4.1.2.bb): julius: add commandpipe patch from TI to allow commands over a pipe to control apps using your voice Aug 27 09:18:38 morning Aug 27 09:20:17 is it possible to get oe to look first in a predefined dir if a file exists there, and only download it with wget/svn/whatever if it doesn't? Aug 27 09:23:51 th1_: like with DL_DIR? Aug 27 09:24:12 yes but like a DL_DIR overlay, a read-only cache Aug 27 09:26:05 th1_: I assume a http server doesn't do? Aug 27 09:26:22 I suppose I can just copy the read-only cache into DL_DIR before starting the build Aug 27 09:26:30 http server doesn't handle svn and git etc. Aug 27 09:27:02 th1_: says who? Aug 27 09:27:11 how? Aug 27 09:27:45 th1_: pre mirrors are checked before doing a svn/cvs co, for git a repo will be downloaded and then git fetched' Aug 27 09:28:11 ok Aug 27 09:28:18 then it'd work Aug 27 09:28:32 thanks Aug 27 09:32:36 I guess you can even set file:// in PRE_MIRRORS. I can't think of any reason why that shouldn't work. Aug 27 09:36:02 pb__: probably, I never tried that :) Aug 27 09:43:24 RP: ping? is opstart from you? did you push it upstream? Aug 27 09:45:44 doh..a new Zaurus... http://sharp-world.com/corporate/news/090827.html Aug 27 09:46:18 it has only the Z in common Aug 27 09:48:15 hm.. i.MX51 Freescale Aug 27 09:50:52 nice Aug 27 09:51:07 Perhaps someone might upstream the FSL drivers then :/ Aug 27 09:51:25 Ubuntu keep claiming that FSL are trying but the evidence is... thin on the ground. Aug 27 10:01:26 hrw: the atom based are even smaller...http://apcmag.com/tiny-mbook-netbook-now-shipping.htm Aug 27 10:04:44 03Thomas Zimmermann  07shr/import * r98cef122df 10openembedded.git/recipes/openmoko-3rdparty/om-neon_svn.bb: Add neon svn reciepe, renamed neon package to om-neon because another neon already exists in OE Aug 27 10:04:53 03Thomas Zimmermann  07shr/import * rea450a7797 10openembedded.git/recipes/aceofpenguins/aceofpenguins-launcher_0.2.bb: Updated aceofpenguins-launcher Aug 27 10:05:15 omg Aug 27 10:05:19 hi rkirti Aug 27 10:05:27 hello hrw Aug 27 10:05:32 morning everyone Aug 27 10:12:29 hi rkirti Aug 27 10:12:37 rkirti: you have mail Aug 27 10:13:08 I have managed to build a mameo image... and forgot the target hardware at home! Aug 27 10:14:26 florian: :) Aug 27 10:17:20 florian: for the gpe- apps, did you check my latest commit on github ? So I have to change it from maemo extras-devel to svn is what you are saying ? Aug 27 10:39:04 * * OE Bug 5309 has been created by javiroman(AT)kernel-labs.org Aug 27 10:39:06 * * minimal distro with alix.conf machine: gcc-cross-initial-4.2.4-r8.1 build problem (geode) Aug 27 10:39:08 * * http://bugs.openembedded.org/show_bug.cgi?id=5309 Aug 27 10:40:38 rkirti: not yet... let me see Aug 27 10:51:50 rkirti: hum well... you can use these too, but the code is the same :-) You need to do two things: Tell the gpe components to actually build the hildon ui and you should get red of the old recipes in your tree. Aug 27 10:54:12 why is it that the pstage files for native are generally directory specific but the ones for target aren't? Aug 27 11:04:43 e.g. in buildroot it fiddles the pkgconfig files etc., so they are not path dependent Aug 27 11:14:13 could it not be more clever like check if any of the files are actually using the path? Aug 27 11:30:48 03Stanislav Brabec  07org.openembedded.dev * r13090db2af 10openembedded.git/recipes/glibc/glibc-package.bbclass: Aug 27 11:30:48 glibc-package.bbclass: Reverted 056a67f6070c5ea561acb65049b551e24f3b7eb6. Aug 27 11:30:48 * see http://lists.linuxtogo.org/pipermail/openembedded-devel/2009-August/013429.html Aug 27 11:59:09 what all changes needs to be done to change oe toolchain Aug 27 11:59:30 actually it is giving an error "C compiler could not find executables" Aug 27 11:59:52 i think that is because oe is trying to make its own toolchain Aug 27 11:59:59 and so the build is taking time too Aug 27 12:00:00 ? Aug 27 12:20:28 anish, did you read http://docs.openembedded.org/usermanual/html/commonuse_prebuilt_toolchain.html ? Aug 27 12:27:00 Hi, I have problem about installing libqt embedded packages compiled on OE, can anybody help me? Aug 27 12:28:45 munzur: welcome to irc, general rule is ask and see who is awake.. Aug 27 12:35:56 Ok, I've compiled libqtguie4, libqtnetworke4 and libqtcoree4 in OE, then transferred them to BeagleBoard Angstrom. Those are of version 4.4.3. On BeagleBoard same packages of version 4.4.1 has already been installed. I tried to install each of the new packages using `opkg install ./´. For each package, it required other two to be at version 4.4.3. Then I removed the installed packages, and tried to install those 4.4.3 (compiled on OE) . For each Aug 27 12:37:40 zecke: about prebuilt toolchains, these expect the files in pre-built/cross/bin....and packaged-staging recreates them in ../cross/bin. So why are the binaries placed in e.g. /cross/armv5te/bin on first build? This results in the funny 'compiler cannot create executables' :) Aug 27 12:37:58 ~curse glibc changes Aug 27 12:37:59 May you be reincarnated as a Windows XP administrator, glibc changes ! Aug 27 12:43:51 munzur: try to provide all these package files to cmd line Aug 27 12:44:50 morning Aug 27 12:45:33 ant_work: I don't follow. pre-built is whatever toolchain you use Aug 27 12:45:34 You mean `opkg install ./libqtguie.ipk ./libqtnetworke.ipk ./libqtcoree.ipk´ ? Aug 27 12:45:55 Let me print the error report here. Aug 27 12:46:07 munzur: not here please Aug 27 12:46:09 pastebin it Aug 27 12:47:03 zecke: I thought I found a bug in packaged-staging, it recreates the CROSS_DIR in ${PSTAGE_TMPDIR_STAGE}/cross Aug 27 12:47:19 but it is expected in ${PSTAGE_TMPDIR_STAGE}/cross/${BASE_PACKAGE_ARCH} Aug 27 12:47:52 now I see external-toolchain doesn't use BASE_PACKAGE_ARCH Aug 27 12:48:06 so..where is exactly the issue...? Aug 27 12:48:30 on first build 'incorrectly' using BASE_PACKAGE_ARCH ? Aug 27 12:49:08 that's default Aug 27 12:50:36 booxter: here's the report http://pastebin.com/m123509a0 Aug 27 12:50:39 ant_work: I have no idea what you are talking about. pre-built has nothing to do with packaged staging Aug 27 12:50:58 zecke: they have the same path for CROSS_DIR Aug 27 12:51:18 which differs from the one built from scratch Aug 27 12:51:31 I've got a weird problem with e2fsprogs-native_1.41.5.bb: it fails to configure the first time with "error: possibly undefined macro: AM_GNU_GETTEXT", but if I run bitbake again after finishing my build otherwise (I used -k), it works the 2nd time round Aug 27 12:51:46 munzur: and what if you install core, then gui? Aug 27 12:52:07 ant_work: I'm the wrong person! I have no idea what you mean when refering to "pre-built"... I'm refering to the user manual... Aug 27 12:54:34 boxter: This is what I get http://pastebin.com/m4b55bbd4 Aug 27 12:55:04 Maybe it is because I couldn't completely remove the core package of version 4.4.1 Aug 27 12:59:36 zecke:np, thx. I found it in bitbake.conf CROSS_DIR = "${TMPDIR}/cross/${BASE_PACKAGE_ARCH}" Aug 27 12:59:44 boxter: I listed installed packages, libqtguie and libqtnetworke appear installed. I remove them now. Aug 27 13:07:44 booxter: After removing, I checked installed package list, none of them is installed. I tried to install libqtcoree separately. This is what I got: http://pastebin.com/m66ac818. Aug 27 13:37:31 leads to issues: external-toolchains and packaged-staging blindly rebuild in /bin/cross Aug 27 13:38:25 ops..sorry Aug 27 14:05:34 03Pratheesh Gangadhar  07org.openembedded.dev * r1eb89aacbc 10openembedded.git/recipes/glib-2.0/ (files/gatomic_armv6.patch glib-2.0_2.21.4.bb): (log message trimmed) Aug 27 14:05:34 glib-2.0 2.21.4: add armv6 atomic ops, this fixes 100% utilization hangs when doing DSP based GStreamer playback Aug 27 14:05:34 GLIB by default uses ARMV5 instructions for emulating atomic operations and this Aug 27 14:05:34 patch (derived from Aug 27 14:05:35 https://stage.maemo.org/svn/maemo/projects/haf/trunk/glib/debian/patches/25-gatomic.patch) Aug 27 14:05:37 make use ARMV6 atomic instructions instead. CPU goes into "sched_yield" while Aug 27 14:05:39 loop during hang - due to a race condition triggered by high rate context Aug 27 14:33:11 /home/hrw/devel/build/angstrom/tmp/cross/armv5te/lib/gcc/arm-angstrom-linux-gnueabi/4.3.3/include-fixed/limits.h:122:61: error: limits.h: No such file or directory Aug 27 14:33:17 ideas? Aug 27 14:38:04 hrw: read few days back in mailing list Aug 27 14:40:37 http://git.uk.xensource.com/git/xenclient/toolstack.git/tree/xenvm/xenvm.readme Aug 27 14:41:03 sorry wrong chan Aug 27 14:43:45 XorA: remember subject? Aug 27 14:44:04 hrw: no, but zecke was posting to the thread Aug 27 14:44:22 hrw: something about that header removed from recent kernels Aug 27 14:59:14 I will rebuild whole toolchain Aug 27 15:01:31 hrw: while doing that, make a new hwcap patch :-) Aug 27 15:07:04 heh.. another mips device at home Aug 27 15:10:49 hrw: oh? Aug 27 15:10:57 morning Aug 27 15:11:33 florian: T-Com Speedport w500v Aug 27 15:12:36 hrw: I've been working with similar devices recently Aug 27 15:12:48 If you do anything out of the ordinary with it, let me know Aug 27 15:13:00 There are already alternative fw out there for it Aug 27 15:13:09 Laibsch: I have bitswitcher on it Aug 27 15:13:14 hehe Aug 27 15:13:16 me too Aug 27 15:13:20 hi kergoth Aug 27 15:13:22 but on the 500V Aug 27 15:13:34 Laibsch: I just got it with BS on it Aug 27 15:13:39 hrw: I think bitswitcher used OE to generate their toolchain Aug 27 15:13:44 took me a while to find login/pass Aug 27 15:14:01 tardyp: boo Aug 27 15:14:12 pratheesh needs to word wrap his commits Aug 27 15:17:37 Laibsch: the fun is that I only need voip functionality from w500v. dsl/wlan/dhcp/dns got already disable Aug 27 15:17:41 d Aug 27 15:18:05 yes, that's the awesome thing about bitswitcher Aug 27 15:18:18 and the reason that saved my 500V from the dumpster ;-) Aug 27 15:18:34 but bitswitcher is still very rough around many edges Aug 27 15:18:41 see all the bugs I filed Aug 27 15:19:24 first I want to get it running Aug 27 15:19:41 Laibsch: and the problem with that device is that it lacks any english docs Aug 27 15:20:48 Laibsch: normal phone I should connect to T-Net connector (rj-11) or use that adapter with rj11 connector on one end and strange connector on other? Aug 27 15:23:09 I had to experiment myself a bit Aug 27 15:23:27 I got it working with an RJ-11 to TAE adapter in the end Aug 27 15:23:35 not sure that was really necessary Aug 27 15:27:10 Laibsch: bitswitcher gives just one SIP account Aug 27 15:27:29 I hope that can be changed Aug 27 15:27:37 but I don't care that much anymore Aug 27 15:27:47 You may use asterisk for consolidation Aug 27 15:28:00 I'm in the middle to consolidate all telco Aug 27 15:28:09 so that I can be reached when in Asia Aug 27 15:28:24 and avoid high local tariff when dialing out Aug 27 15:28:44 pb__: one quick question... is -fomit-framepointer enabled by default on arm? Aug 27 15:29:00 Laibsch: thats what my cellphone is capable already - doing voip calls over wifi Aug 27 15:29:18 kergoth hello Aug 27 15:29:34 hrw: that's the SIP client part Aug 27 15:29:44 I was always interested in the SIP server part Aug 27 15:29:53 I am not Aug 27 15:29:53 and the connection to ISDN Aug 27 15:30:37 I want to use my home telephone just as if I was at home - even when thousands of miles away Aug 27 15:32:09 I do not have home phone Aug 27 15:32:22 hehe Aug 27 15:32:25 that helps Aug 27 15:32:43 zecke: Hi. Yes, I wrote opstart iirc Aug 27 15:32:49 Laibsch: we use cellphones for all calls and I want to add voip at home to pay less Aug 27 15:32:55 zecke: I did try for upstream and gave up in the end Aug 27 15:34:55 RP: okay, opstart does not do what I expected to do :} Aug 27 15:35:05 RP: opcontrol -s is closer to what I expected Aug 27 15:35:29 RP: have you recently tried "timer" debug in oprofile? Aug 27 15:36:34 zecke: From what I remember opstart is a low overhead version of opcontrol which stops your traces being screwed around by loading busybox ;-) Aug 27 15:36:40 zecke: Not recently Aug 27 15:37:59 RP: the problem is opstart didn't start the tracing at all :} Aug 27 15:37:59 tardyp: see http://thread.gmane.org/gmane.comp.handhelds.openembedded/26156 , it may help with the question you had earlier Aug 27 15:43:43 RP: upstream oprofile seems rather slow... and my tasks now include finding out why userspace backtraces are broken and fixing the timer based profiling Aug 27 15:43:44 hey, can someone remind me where copysignl.h resides? i think i need to move a few header files over to get uclibc to build Aug 27 15:44:01 eg: /home/matt/devel/tmp/cross/ppc405/lib/gcc/powerpc-angstrom-linux-uclibc/4.4.1/../../../../powerpc-angstrom-linux-uclibc/lib/libgcc_s.so: undefined reference to `copysignl` Aug 27 15:44:06 er not copysignl.h Aug 27 15:44:23 i forget how i fixed it, i think it involved moving header files though Aug 27 15:44:38 zecke: They probably changed the interface since I wrote it :/ Aug 27 15:44:44 m4t: yes copysign is not there did you link with libm Aug 27 15:44:52 khem yea exactly Aug 27 15:45:01 zecke: Assuming this is arm, it is not missing frame pointers or debug info? Aug 27 15:45:14 hmm Aug 27 15:46:14 RP: I have debug infos, I have to find out about fram pointers Aug 27 15:46:23 m4t: which gcc version ? Aug 27 15:46:24 RP: I don#t compile with -fomit-frame-pointer... Aug 27 15:46:25 hmm, i should really experiment with oprofile again ,haven't touched it since i was at ti Aug 27 15:46:42 RP: but it might be enabled in gcc -O2 Aug 27 15:47:02 4.4.1 Aug 27 15:47:13 i totally remember running into this before, and fixing it Aug 27 15:47:28 i searched for it and all the links were purple heh Aug 27 15:47:32 yeah the correct fix would be to fix gcc Aug 27 15:47:43 m4t: and whats your target Aug 27 15:47:53 is it one of soft-float ones? Aug 27 15:48:11 yes its ppc405 with soft fpu Aug 27 15:48:21 hmm ok Aug 27 15:48:27 kergoth, thanks :) Aug 27 15:48:40 I think dirty fix is to link libgcc with lm Aug 27 15:49:07 yea Aug 27 15:51:02 file://powerpc-link-with-math-lib.patch;patch=1 \ Aug 27 15:51:09 i had already fixed it, duh ;/ Aug 27 15:51:18 in my old git tree Aug 27 15:52:08 http://gcc.gnu.org/ml/gcc-patches/2008-10/msg00237.html Aug 27 15:56:00 bye Aug 27 15:59:50 tardyp: np, let me know how it goes, if you end up using it. i don't think anyone has messed with it other than me, just yet Aug 27 16:00:07 anyway, good night! Aug 27 16:01:00 kergoth, ok, I'll send may have to push it on poky, which is the oe setup I have now. Aug 27 16:01:44 tardyp: just drop the classes/srctree.bbclass (and classes/gitver.bbclass if you need that) into classes/ at toplevel in poky, or under one of the meta areas. should be fine Aug 27 16:02:29 ok. Aug 27 16:04:12 03acid-burn  07org.openembedded.dreambox * rad2d1a2891 10openembedded.git/packages/images/dreambox-image.bb: dreambox-image.bb: add CleanupWizard to OPENDREAMBOX_EXPERIMENTAL_R Aug 27 16:04:22 03acid-burn  07org.openembedded.dreambox * r8184b48798 10openembedded.git/packages/enigma2/ (enigma2-plugins.bb enigma2.bb): Aug 27 16:04:22 enigma2.bb, enigma2-plugins.bb: -push srcdate to include new cleanupwizard. Aug 27 16:04:22 - add cleanupwizard description to enigma2.bb Aug 27 16:33:48 03Koen Kooi  07org.openembedded.dev * r98d2f53a2b 10openembedded.git/ (conf/checksums.ini recipes/ti/ti-codec-engine_2.24.01.bb): ti-codec-engine: make 2.24.1 default and fix the build Aug 27 16:46:15 Has anyone started work on getting OpenEmbedded to produce LSB4 compliant filesystems? Aug 27 16:46:24 xserver-xorg_1.5.3.bb failing for anyone else? Aug 27 16:46:33 Maybe I just need to rm -Rf tmp Aug 27 16:47:35 FWIW, tinderbox looks OK Aug 27 16:49:38 03Sebastian Krzyszkowiak  07shr/import * r6f685c6fe5 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: add om-neon Aug 27 16:49:39 03shr build user  07shr/import * red64eb1eb4 10openembedded.git/ (5 files in 2 dirs): navit: set to AUTOREV, bump PV Aug 27 16:49:40 03Sebastian Krzyszkowiak  07shr/import * rd43883a9ba 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: add aceofpenguins-launcher Aug 27 16:58:49 03Brijesh Singh  07org.openembedded.dev * r4923ae44c9 10openembedded.git/recipes/ti/ (3 files): cmem, sdma, lpm: add 2.24.01 release Aug 27 17:12:30 what can cause glibc-localedata-i18n.ipk not to be built? Aug 27 17:12:53 th1_: if you have disabled locale generation Aug 27 17:14:49 khem, but e-image-core.bb tries to include it regardless? Aug 27 17:23:17 --------- Hi: sombody have experience with NX ? Aug 27 17:52:32 i want bash/zsh completion of providers/tasks for bitbake Aug 27 17:52:36 someday Aug 27 17:58:21 any thoughts on passing tasks a fresh instance of the metadata store, so their changes wouldn't persist? Aug 27 17:58:38 I'm thinking it would be a good, if minor, improvement in the direction of deterministic builds Aug 27 18:04:19 hmm, we need a tool as capable or more capable than moblin-image-creator-2 Aug 27 18:06:52 does anyone know where '_GLIBCXX_LONG_DOUBLE_COMPAT' is defined, as used by libstdc++v3/src/c++locale.cc? Aug 27 18:10:49 /* Define if compatibility should be provided for -mlong-double-64. */ Aug 27 18:11:59 03Chris Larson  07org.openembedded.dev * r42dc52e38d 10openembedded.git/classes/task-metadata-track.bbclass: (log message trimmed) Aug 27 18:11:59 Add task-metadata-track.bbclass, for tracking task changes to metadata. Aug 27 18:11:59 This class uses events to capture the state of the datastore when the task Aug 27 18:11:59 starts, and after it completes. It diffs those captured states, and emits Aug 27 18:12:00 messages showing which variables changed, and what their values were changed Aug 27 18:12:02 to. Aug 27 18:12:04 It provides a mechanism to blacklist variables you expect to change, both Aug 27 18:12:58 that happens after libgcc is linked to libm, i guess Aug 27 18:19:58 m4t: are you enabling long-double-128 in gcc or not ? Aug 27 18:20:36 no i dont think so Aug 27 18:20:48 it says its deprecated, i dont see that define used anywhere else Aug 27 18:20:55 oh and you have uclibc right Aug 27 18:21:28 yea im testin it again with 4.4.1 Aug 27 18:21:48 it built after i removed those lines at the end of c++locale.cc Aug 27 18:22:42 maybe its a bit of dangling code that only ends up being called if that libgcc -libm patch is applied Aug 27 18:23:17 Its unlikely Aug 27 18:23:20 let me look Aug 27 18:23:34 thanks Aug 27 18:23:37 re Aug 27 18:24:38 th1_: yeah it seems for that image you need to enable locale generation Aug 27 18:25:57 m4t: what is exact error you are seeing Aug 27 18:26:26 let me pastebin this, one sec Aug 27 18:26:52 http://pastebin.com/d4fa58d34 Aug 27 18:35:14 florian: ping Aug 27 18:35:24 Laibsch: pong Aug 27 18:35:46 m4t: it seems like a config error Aug 27 18:36:04 florian: Would you be so kind to deactivate laibsch@openembedded.org? Aug 27 18:36:13 I've never received anything but spam Aug 27 18:36:16 --------- Aug 27 18:36:22 And I use my own address for commits now Aug 27 18:36:29 I cant compile gperf-native Aug 27 18:36:34 Laibsch: I have no idea about OE mail. Aug 27 18:36:36 help Aug 27 18:36:56 florian: who does? It comes from ltg Aug 27 18:37:18 help: http://pastebin.com/d493ee388 Aug 27 18:37:33 I have a basic idea how it might be routed but no idea how to deactivate it for a single person. Aug 27 18:37:33 th1_: help? Aug 27 18:37:44 khem: help Aug 27 18:37:47 oops Aug 27 18:37:47 :-) Aug 27 18:37:58 looks like I just created the master branch in the OE repo Aug 27 18:38:03 I can take a look when I'm finished with the new machine... but no free minute till then. Aug 27 18:38:26 mickey|sports: Can you please delete that (master branch in OE repo) when you have the time? ^^ Aug 27 18:38:37 florian: OK. Thank you. Aug 27 18:41:40 sushisan: line 37 looks suspicious... does this reflect your directory layout Aug 27 18:41:46 ? Aug 27 18:41:59 khem: I'm having difficulty compiling xserver-xorg version 1.5.3. Could be a 64bit-host issue. Maybe an effect of the eglibc introduction? Would be nice if you had a quick look: http://tinderbox.openembedded.net/public/logs/task/2633743.txt and http://tinderbox.openembedded.net/packages/xserver-xorg/ Aug 27 18:42:32 Laibsch: don't you have the power to take a look? i remember something like this :) Aug 27 18:42:53 Laibsch: yw Aug 27 18:43:01 florian: No, of course I don't. Mamie told me not to ;-) Aug 27 18:44:09 sushisan: your build failed with signal 4 which is SIGILL I think btw I have gperf-native building file on x86-64 kubuntu box Aug 27 18:44:16 Has anyone written a class to add instrumentation to the tasks, capturing the task execution times? Aug 27 18:44:31 Laibsch: don't listen to parents and k. ;) Aug 27 18:45:25 florian: Spaß beiseite. I didn't break it and I have no time or skill to fix it. Aug 27 18:46:30 khem: There've been some recent changes to xorg-xserver/, too. It may be those that did it. tinderbox has only me trying to build the most recent version after introduction of INC_PR. Aug 27 18:47:03 Does Michael Smith hang around here? Aug 27 18:47:04 also, any opinions on docbook vs reStructuredText for the bitbake manual? I'm leaning toward converting it to ReST+Sphinx Aug 27 18:47:17 Laibsch: yeah, he was on yesterday Aug 27 18:47:26 What's his nick? Aug 27 18:47:32 good question Aug 27 18:47:33 * kergoth thinks Aug 27 18:47:53 ~seen msmith_ Aug 27 18:47:54 msmith_ is currently on #oe. Has said a total of 74 messages. Is idling for 15h 52m 22s, last said: 'that's true, since the source tree is there, you could even parse it out of a header file if you had to'. Aug 27 18:47:56 :) Aug 27 18:47:59 Thank you Aug 27 18:48:02 np Aug 27 18:48:11 msmith_: ping Aug 27 18:48:53 hmm Aug 27 18:49:10 Laibsch: http://cgit.freedesktop.org/xorg/xserver/commit/?id=ae38151ddda9984effca5bb7c582540061201dce Aug 27 18:49:11 I'm seeing recent breakage in xserver-xorg. You've been the last to touch the directory. Would you please be so kind to try and compile xserver-xorg_1.5.3.bb and let me know if that works for you? Aug 27 18:49:15 Laibsch: http://cgit.freedesktop.org/xorg/xserver/commit/?id=efb723e166e5fa89e90c7b400fb4c7979b1f50c0 Aug 27 18:49:31 can you try applying these two patches to xserver ? Aug 27 18:50:09 I'll give it a shot locally Aug 27 18:50:20 I'll let you make the commit to make sure it's proper Aug 27 18:50:23 Laibsch: thx Aug 27 18:50:50 Laibsch: ok let me know if it helps Aug 27 18:54:38 03Rolf Leggewie  07org.openembedded.dev * rb1185de8a9 10openembedded.git/recipes/intltool/ (16 files): Aug 27 18:54:38 intltool: move dependency on libxml-parser-perl-native to inc file Aug 27 18:54:38 * introduce INC_PR while we're at it Aug 27 18:55:00 kergoth: the directories by bitbake trunk are created with 0777 perms where to change it if I want to Aug 27 18:55:19 what do you mean? Aug 27 18:55:58 bitbake is in git, first of all, it's master, not trunk, and the permissions are correct in the repository, so if you're seeing something different when you check out, its somethign on your end Aug 27 18:56:10 oh yes master I meant Aug 27 18:56:24 I mean when bitbake is used Aug 27 18:56:35 and the directories in the tmp are created by it Aug 27 18:56:40 ahh Aug 27 18:56:41 hmm Aug 27 18:56:45 that's odd, you'd think it would obey umask Aug 27 18:56:55 it seems not to Aug 27 18:57:03 maybe its a consequence of the ui client/server split, maybe the daemonization loses it? Aug 27 18:57:04 $ umask Aug 27 18:57:05 0022 Aug 27 18:57:20 I'm rusty, i need to check the daemonization section of advanced programming in the unix environment Aug 27 18:58:24 :-) Aug 27 18:58:53 kergoth: I was trying to add upstream patches the lazy way like http://paste.debian.net/45162/ That doesn't work. I assume because of the weird filenames. Any idea how to make this work? Aug 27 18:59:07 from the man page of umask, forked children should bei nheriting the parent umask Aug 27 18:59:12 khem: can you open a bug on it? Aug 27 18:59:21 $rev.patch would be nice. Aug 27 18:59:34 hmm. what's the behavior? Aug 27 18:59:37 some mangling in bitbake/Oe maybe Aug 27 18:59:50 kergoth: ok. Do we use bugs.openembedded.org Aug 27 18:59:56 it could be picking up the id= in its grab of the options like patch=1 Aug 27 19:00:03 might not be passing it along Aug 27 19:00:12 we should make sure it only pulls out the options we know about, and pass the rest as is Aug 27 19:00:14 hmm Aug 27 19:00:22 kergoth: http://paste.debian.net/45163/ is what it looks like Aug 27 19:00:23 fetch should really be switched to use the urlparse module and stuff Aug 27 19:00:45 oh, its not getting the localpath right Aug 27 19:00:50 so it thinks it didn't get it Aug 27 19:01:00 must be ripping off the options when generating the localpath Aug 27 19:01:18 fetch needs to be revamped, no magic guessing of the localpath, instead run the fucking fetch, and pass back the localpath.. Aug 27 19:01:35 should open a bug on that issue, if you would Aug 27 19:01:41 khem: yeah, at least for now Aug 27 19:02:03 kergoth: OK, I will Aug 27 19:02:07 Thanks for taking a look Aug 27 19:02:46 np. i know exactly where the broken code is. as a workaround, we could do what i said above, only rip off the options we use, and leave the rest.. long term we should rewrite bb.fetch Aug 27 19:02:59 kergoth: it's a bitbake bug, right Aug 27 19:03:01 ? Aug 27 19:03:08 yeah, in the wget fetcher Aug 27 19:03:12 OK Aug 27 19:03:13 thanks Aug 27 19:03:16 it has a localpath method that returns what it should end up with, locally Aug 27 19:03:37 basically does a basename and concats with DL_DIR Aug 27 19:03:39 hack, at best Aug 27 19:05:14 now that we have persist_data, we might be able to switch to passing back the localpath in the fetch.. the only issue i could think with that would be a fresh from scratch build with an existing DL_DIR, it might try to fetch anyway. so either we store the persistent data about the fetch in DL_DIR, or we leave it as is, and just make localpath() smarter Aug 27 19:06:11 kergoth: here you go http://bugs.openembedded.org/show_bug.cgi?id=5310 Aug 27 19:06:29 Laibsch: for now, you could append to do_fetch and grab it yourself, but then you'd have to append to do_patch too :) hmm... Aug 27 19:07:01 i was thinking about changing do_patch to not go by SRC_URI directly, but instead have a SRC_LOCAL or SRC_PATCHES or something Aug 27 19:07:09 then you could manipulate that directly Aug 27 19:07:24 that's pretty low priority on the task list though :) Aug 27 19:17:04 * * OE Bug 5306 has been RESOLVED (WORKSFORME) by Aug 27 19:17:06 * * x11-gpe-image do_rootfs failed Aug 27 19:17:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5306 Aug 27 19:17:41 CoreDump|home: the bot seems to be reporting resolution of a bug fine. new bugs don't see to be reported here. Can you take a look? Aug 27 19:17:46 kergoth: it's Aug 27 19:17:51 !oebug 5311 Aug 27 19:17:53 * * Bug 5311, Status: CONFIRMED, Created: 2009-08-27 19:09 Aug 27 19:17:54 * * : fetch error for unconventional URI Aug 27 19:17:55 * * http://bugs.openembedded.org/show_bug.cgi?id=5311 Aug 27 19:18:14 khem: found it Aug 27 19:18:24 khem: the daemonize module we're using forces the umask to 0 Aug 27 19:18:27 why, i have no fucking clue Aug 27 19:18:27 Laibsch: will have a look next week Aug 27 19:18:32 stupid Aug 27 19:18:51 # We probably don't want the file mode creation mask inherited from Aug 27 19:18:51 # the parent, so we give the child complete control over permissions. Aug 27 19:18:51 os.umask(UMASK) Aug 27 19:19:01 i just love seeing "probably" in code Aug 27 19:19:05 assumptions.. Aug 27 19:19:06 :) Aug 27 19:23:54 kergoth: less ugly than other hacks :) Aug 27 19:24:12 03Chris Larson  07master * r30d00e1a55 10bitbake.git/lib/bb/daemonize.py: Aug 27 19:24:12 daemonize: Kill the forcing of umask to 0 in our children. Aug 27 19:24:12 It forces the daemon umask to the module level UMASK variable, which defaults Aug 27 19:24:12 to 0, so the bitbake daemons end up with a umask of 0. Fixed by setting UMASK Aug 27 19:24:12 to None and making it only run umask when it's not None, so our children will Aug 27 19:24:13 inherit the parent umask Aug 27 19:24:15 Signed-off-by: Chris Larson Aug 27 19:25:08 !oebug 5310 Aug 27 19:25:09 * * Bug 5310, Status: RESOLVED (FIXED), Created: 2009-08-27 19:05 Aug 27 19:25:10 * * raj.khem(AT)gmail.com: bitbake does not seem to honor umask when creating directories in tmp area Aug 27 19:25:11 * * http://bugs.openembedded.org/show_bug.cgi?id=5310 Aug 27 19:28:26 khem: the patches don't apply cleanly Aug 27 19:28:27 Laibsch: is it possible to add a convenience link on a bug page to take the bug? i.e. reassign to me, given the current account I'm logged in under, and add the original assignee to the CC list? Aug 27 19:28:35 hmmm Aug 27 19:28:50 kergoth: I'll take a look Aug 27 19:29:18 not a big deal, just think that would be a common thing to want to do, when you start work on it, eh? Aug 27 19:29:27 * kergoth shrugs, could be he's misunderstanding the workflow Aug 27 19:29:41 nobody can help me' Aug 27 19:31:38 kergoth: looks like all we may need to do is update the bugzilla installation: https://bugzilla.mozilla.org/show_bug.cgi?id=35195 Aug 27 19:31:42 it's fixed upstream Aug 27 19:31:48 ah, nice Aug 27 19:32:11 I think it may be missing the "add original assignee to cc", though Aug 27 19:32:18 http://pastebin.com/d24ebd6fa Aug 27 19:32:21 well, thats not terribly difficult to do manually Aug 27 19:32:22 In our case that may not be too dramatic Aug 27 19:32:36 That and most of the time the assignee is the OE ml Aug 27 19:32:52 In any case, the assignee will get mail that he has been unassigned Aug 27 19:32:55 do we have a separate list for the bugzilla activity? if so, there's no real point in moving to the cc Aug 27 19:33:00 * kergoth nods Aug 27 19:33:10 * Laibsch tries to see when the functionality was added Aug 27 19:34:07 Laibsch: hmmm Aug 27 19:38:04 * * OE Bug 5096 has been UNCONFIRMED by Aug 27 19:38:06 * * Testbug for cdbot2 Aug 27 19:38:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5096 Aug 27 19:40:38 03Koen Kooi  07org.openembedded.dev * r0d12f0b648 10openembedded.git/ (3 files in 2 dirs): Aug 27 19:40:38 erlang: start updating erlang to R13B01 Aug 27 19:40:38 * now it's Crofton's turn to work on it :) Aug 27 19:40:39 03Koen Kooi  07org.openembedded.dev * rabff228959 10openembedded.git/recipes/erlang/erlang.inc: erlang: disable parallel make Aug 27 19:40:40 03Koen Kooi  07org.openembedded.dev * r87b8c170f8 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 27 19:40:41 kergoth: there are ^M in the diff you committed Aug 27 19:41:05 khem: the file is dos format, it's consistent with the existing line endings. Aug 27 19:41:17 ugh Aug 27 19:41:23 why not change it Aug 27 19:41:41 can look at fixing it later, but its not particularly important Aug 27 19:41:45 a sane editor doesn't care Aug 27 19:41:52 * kergoth didn't even notice until he saw the diff Aug 27 19:42:04 git show 30d00e1a550931cb7f62e6574a7dd2d7b33f182b Aug 27 19:42:08 cares Aug 27 19:42:20 yes sane editor does not Aug 27 19:44:21 we should fix it at some point, but there are more important things :) Aug 27 19:44:36 * khem agrees Aug 27 19:54:02 Laibsch: can you try the patches below Aug 27 19:54:13 http://www.uclibc.org/~kraj/0035-cherry-pick-the-rest-of-glcore-prepare-for-dynamic.patch Aug 27 19:54:22 http://www.uclibc.org/~kraj/0036-configure.ac-Add-GLX_SYS_LIBS-for-Xvfb-and-Xnest.patch Aug 27 19:58:20 Looking for GUI on device running linux(a carpc and a DVR), checked with Qt,Nano-X,Wxwidgets,andriod,still confused Aug 27 20:08:04 * * OE Bug 5096 has been RESOLVED (INVALID) by Aug 27 20:08:06 * * Testbug for cdbot2 Aug 27 20:08:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5096 Aug 27 20:18:18 Maybe we're lucky and the patch in https://bugzilla.mozilla.org/show_bug.cgi?id=37613 still works without too much trouble Aug 27 20:24:49 03Jeremy Puhlman  07org.openembedded.dev * rf0b67d21ab 10openembedded.git/recipes/iptables/ (files/cross-iptables.diff iptables_1.3.8.bb): Aug 27 20:24:49 iptables: Add patch to use cross ar and ranlib. Aug 27 20:24:49 Signed-off-by: Jeremy Puhlman Aug 27 20:24:49 Signed-off-by: Chris Larson Aug 27 20:24:50 03Jeremy Puhlman  07org.openembedded.dev * r34dee0fced 10openembedded.git/recipes/lsof/lsof_4.78.bb: Aug 27 20:24:53 lsof: Fix misuse of system ranlib. Aug 27 20:24:55 Signed-off-by: Jeremy Puhlman Aug 27 20:24:57 Signed-off-by: Chris Larson Aug 27 20:24:59 03Jeremy Puhlman  07org.openembedded.dev * r2bc39e0c7d 10openembedded.git/recipes/groff/ (groff_1.19.1.bb groff_1.19.2.bb): Aug 27 20:25:02 groff: disable X11 support. Aug 27 20:25:04 Signed-off-by: Jeremy Puhlman Aug 27 20:25:06 Signed-off-by: Chris Larson Aug 27 20:25:08 03Brian Pomerantz  07org.openembedded.dev * rf707d17d8d 10openembedded.git/recipes/net-tools/net-tools_1.60.bb: Aug 27 20:25:11 net-tools: no parallel make Aug 27 20:25:13 Signed-off-by: Brian Pomerantz Aug 27 20:25:15 Signed-off-by: Chris Larson Aug 27 20:25:22 hello Aug 27 20:25:36 can anyone point me to some doc saying what "jffs2_do_read_inode_internal: Truncating ino ..." means? Aug 27 20:26:04 looking in the web I find lots of messages like that Aug 27 20:26:17 but non of them seems to be related to that message Aug 27 20:28:05 * * OE Bug 5312 has been created by  Aug 27 20:28:07 * * it would be nice if bugzilla made it easier for a dev to accept a bug Aug 27 20:28:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5312 Aug 27 20:28:40 CoreDump|home: looks like at least some new bugs make it here Aug 27 20:29:11 I noticed, yeah Aug 27 20:29:33 If if get the chance, I'll have a look next week Aug 27 20:34:18 yes, no rush Aug 27 20:34:19 thanks Aug 27 20:35:15 hm.. can I change events generation speed on /dev/event* anyhow? Aug 27 20:35:43 e.g. slowdown keyboard events Aug 27 20:36:06 or just process it slower? :) Aug 27 20:37:04 * * OE Bug 5305 has been RESOLVED (FIXED) by chrisfriedt(AT)gmail.com Aug 27 20:37:06 * * x11-gpe-image do_rootfs opkg permission denied Aug 27 20:37:08 * * http://bugs.openembedded.net/show_bug.cgi?id=5305 Aug 27 20:50:09 kergoth parsing in 1.8 is lot faster than on 'master' Aug 27 20:50:26 prolly some patches need forward port Aug 27 20:56:18 yeah, richard knows :) Aug 27 20:57:04 need to hammer at a profile of it, try to fix more bottlenecks. he's done more on the perf side than me, I've been more on the cleanup, refactoring, etc end lately Aug 27 20:58:37 * khem has switched to using master lately Aug 27 20:58:44 i wonder if people would be interested in my abiversions.bbclass Aug 27 20:59:02 what is it about Aug 27 20:59:24 lets you set ABIVERSION = "1.0" in a lib, then if you change it to 1.1, automatically rebuilds everything that depends on that lib Aug 27 21:00:06 doesn't care what the contents of the var are, just does the rebuilds when it changes Aug 27 21:00:21 so you could show optional --enable/--disable type options in there, and itd rebuild things that depend on it when the options change, or whatever Aug 27 21:00:26 s/show/throw/ Aug 27 21:02:04 do you mean static libraries ? Aug 27 21:02:10 no Aug 27 21:02:12 for shared libs it should not matter Aug 27 21:02:15 what? Aug 27 21:02:33 there are lots of cases where a shared lib's ABI can change on you Aug 27 21:02:58 note: ABI, not API Aug 27 21:03:05 yes I misunderstood Aug 27 21:03:20 I thought its some internal versioning Aug 27 21:03:40 ah, no, this just is a var in the recipe, when the var changes, anything that deps on that recipe is rebuilt Aug 27 21:03:50 but to rebuild deps on say libc.so.6 to libc.so.7 is useful Aug 27 21:04:05 yes thats nice I think Aug 27 21:04:05 indeed Aug 27 21:04:10 k Aug 27 21:05:01 will it detect the change automatically Aug 27 21:05:30 if the var changes, it'll do the rebuilds. Aug 27 21:05:36 don't need to do anything else, just change the ABIVERSION Aug 27 21:06:05 kergoth now the umask think is working correctly thx Aug 27 21:06:05 could write scripts to automatically put things in the var, extract sonames & library ABI & stuff, if you really wanted to Aug 27 21:06:08 great, np Aug 27 21:06:23 course, there's the whole parse time vs run time deal then Aug 27 21:06:38 * kergoth thinks itd be nice if bitbake could write a recipe back out with changes Aug 27 21:07:23 bitbake preprocessor :) Aug 27 21:07:42 hehe Aug 27 21:07:58 well, zecke's AST stuff for the parser is a step in the right direction Aug 27 21:08:18 yes Aug 27 21:23:20 FUCK Aug 27 21:23:31 srctree's magic appears to be failing to pull a rabbit out of its hat Aug 27 21:23:32 hm whats wrong? Aug 27 21:23:41 * kergoth sighs, took days to get that working, and now it fails for no reason Aug 27 21:24:31 errrrr... now it works again Aug 27 21:24:33 aiiieee Aug 27 21:24:37 uhu Aug 27 21:24:41 race condition Aug 27 21:25:50 oh, haha, i broke it, but didn't think that was it due to the cache Aug 27 21:25:54 it kept the broken version Aug 27 21:26:04 ????? Aug 27 21:26:10 * kergoth was testing giving tasks their own local copies of the datastore, but it made srctree explode Aug 27 21:26:20 it cached the incorrect data Aug 27 21:26:27 ah the old problem Aug 27 21:26:28 then i put it back, and it was still broken :) Aug 27 21:26:41 i really dont like tasks being able to change the data Aug 27 21:26:56 it means that what tasks you run will affect the metadata, a lack of consistency Aug 27 21:27:06 anonymous python, sure, but tasks... meh Aug 27 21:28:01 this is strange, i don't see why itd break it Aug 27 21:28:02 hrmph Aug 27 21:35:54 hm I am reading about threading now Aug 27 22:33:47 Does anyone know why cacaoh-native would fail on the do_stage phase on one machine and build fine on another? Aug 27 22:33:58 It appears that on the working machine it places the cacaoh binary where the do_stage phase expects it to be (specifically in the .libs subdirectory) while on the other machine it doesn't place it in the .libs subdirectory so the do_stage phase errors out Aug 27 22:35:03 radhermit nope sorry Aug 27 22:35:10 always works for me Aug 27 22:35:52 It happens with a couple other packages too, actually I think they be all be *-native packages which might be a hint to where the problem lies Aug 27 22:36:35 Yeah it works fine for me on my x86_64 build machine, but breaks on an i686 build machine Aug 27 22:36:56 hm build for me on both Aug 27 22:38:00 Hmm, what's even stranger is that it used to work fine Aug 27 22:38:51 I just don't want to have to git bisect through the oe repo to figure out if it's something in OE causing it Aug 27 22:39:03 That would be somewhat painful Aug 27 22:39:33 hm maybee I have tomorrow some time Aug 27 22:39:36 to test it Aug 27 22:40:30 does anyone have the contact info for the author of the original apache2_2.2.3.bb recipe? Aug 27 22:41:15 nope Aug 27 22:41:22 try to use qgit Aug 27 22:41:31 woglinde: ok thanks! Aug 28 00:37:33 *curses apache 2.2.13 for ignoring CONFIG_SITE* Aug 28 00:52:33 03Khem Raj  07org.openembedded.dev * rf7fb3b7e00 10openembedded.git/recipes/uclibc/ (5 files in 2 dirs): Aug 28 00:52:33 uclibc: Make 0.9.30.1 as default if not specified Aug 28 00:52:33 Add C99 double wrappers. Aug 28 00:52:33 Set the default preferences such that Aug 28 00:52:33 0.9.30.1 is chosen if not specified. Aug 28 00:52:34 Signed-off-by: Khem Raj Aug 28 00:52:43 03Khem Raj  07org.openembedded.dev * r2b183dca44 10openembedded.git/recipes/gcc/ (3 files in 2 dirs): (log message trimmed) Aug 28 00:52:43 gcc-4.4.1: Disable libdecnumber for uclibc target. Aug 28 00:52:43 We do not set FENV in uclibc so as a result we can Aug 28 00:52:45 not build libdecnumber. We disable it for gcc 4.4.1 Aug 28 00:52:47 uclibc targets. Aug 28 00:52:49 copysignl is in libm in uclibc so gcc complains about Aug 28 00:52:51 this function in libgcc. We see if it is uclibc powerpc Aug 28 00:54:03 anybody knows about USB programming Aug 28 01:09:17 minimal-image, which includes gnutls, seems to need 'TLS' from uclibc, aka thread local storage Aug 28 01:09:50 which results in the following: ../powerpc-angstrom-linux-uclibc/lib/nof/libstdc++.so: undefined reference to `__tls_get_addr' Aug 28 01:13:32 m4t: fix the tls detection in gnutls or build gcc without tls support (search the mailinglist) Aug 28 01:15:26 zecke i am thinking about trying a newer version of gnutls Aug 28 01:15:51 m4t: well, they will not have patched out tls Aug 28 01:16:34 zecke alright, i might try gcc without tls Aug 28 01:17:39 m4t: otherwise see d894b5a78eeb5509fe817808234727b86660d430 for an example of changing the autotools macro Aug 28 01:18:08 okay, thanks Aug 28 01:18:11 i found your recent thread Aug 28 01:22:06 m4t: i prefer configuring gcc, some prefer patching tls :) Aug 28 01:23:09 the macro workaround seems like i'd have to comeup with a patch for every program Aug 28 01:34:49 i dont see any reference to tls in gnutls-2.4.2 Aug 28 01:35:03 m4t: yupp, that is why I don't like it too much, the cool thing is the origin took these changes... so they might end up in downstream code Aug 28 01:35:11 m4t: arg, so the tls usage is really in stdc++? Aug 28 01:35:22 maybe Aug 28 01:39:25 I have heard about patches, or at least the need to change gcc's tls detection for stdc++ as well... *no idea* Aug 28 01:48:07 i think i got the gcc*.inc patch applied Aug 28 01:50:32 * m4t rebuild gcc Aug 28 02:25:50 hey **** ENDING LOGGING AT Fri Aug 28 02:59:56 2009