**** BEGIN LOGGING AT Sat Dec 04 02:59:59 2010 Dec 04 03:16:24 kergoth: now I am tryin the git merge release-2010.12 Dec 04 03:16:28 and its not fun Dec 04 03:17:41 kergoth: I did git pull --rebase after the merge Dec 04 03:17:51 and I still resolving conflicts for last hour Dec 04 03:19:10 zenlinux: yes its still WIP Dec 04 03:20:21 khem: if you know that master has everything the release does, you dont' need to do an actual merge Dec 04 03:20:37 kergoth: ok Dec 04 03:20:38 you can do a fake one of sorts, by telling merge to just use the current code Dec 04 03:20:49 git checkout master; git merge --ours release-2010.12 Dec 04 03:20:50 or so Dec 04 03:20:55 see git help merge Dec 04 03:21:07 hmm merge went fine Dec 04 03:21:25 its when I am trying to push the merge up it wants me to rebase my master Dec 04 03:21:43 don't see what you mean. Dec 04 03:21:53 it sounds like you're tring to rebase the merge Dec 04 03:21:57 thats way more trouble than its worth Dec 04 03:22:14 hmm yes Dec 04 03:22:17 either reset to before the merge, merge from upstream, then merge from the branch and push, or just pull with merge and push Dec 04 03:22:30 would be my advice Dec 04 03:23:12 hmmm I just finished Dec 04 03:30:24 kergoth: how to pull with merge i did not get it Dec 04 03:34:21 pull defaults to merge unless you pull --rebase or have the git config option set for it Dec 04 03:34:24 either way, pull isn't magic Dec 04 03:34:30 pull only does git fetch, then git merge or git rebase Dec 04 03:34:36 git fetch; git merge origin/master Dec 04 03:34:39 is equivalent Dec 04 03:35:10 ah i see Dec 04 03:35:31 so git merge origin/release-2010.12; git pull Dec 04 03:35:35 lets see Dec 04 03:51:53 kergoth: I just committed a merge Dec 04 03:52:06 kergoth: can you see if that makes sense at all Dec 04 04:10:55 I've updated my kernel config, but when I run ' KERNEL_IMAGETYPE="uImage" bitbake -f -c compile linux-omap-2.6.32 ' nothing happens. Do I have to manual poke a version number or something? Dec 04 04:11:56 to update my kernel, I ran ' KERNEL_IMAGETYPE="uImage" bitbake -c menuconfig linux-omap-2.6.32 ' Dec 04 04:12:13 my config I mean Dec 04 04:15:29 coreyfro: this should be set in local.conf Dec 04 04:23:11 khem: I'm trying not to be a complete newb, but I can't figure out what you mean. Is there a variable I set up in the local.conf saying "new version, please build again" etc? Dec 04 04:24:20 ok you can do is bitbake -c clean virtual/kernel; bitbake virtual/kernel Dec 04 04:24:47 awesome, thanks! Dec 04 05:09:58 khem, i can help you with git stuff :) Dec 04 05:10:31 hiya HopsNBarley Dec 04 05:10:36 crazy week. Dec 04 05:25:25 Is there a document for working with collections? I am using http://docs.openembedded.org/usermanual/usermanual.html#special_bitbake_collections but that doesn't cover machine/conf changes Dec 04 05:25:53 Is this where bblayers.conf comes in to play? Dec 04 07:27:37 03Khem Raj  07master * r73085615de 10openembedded.git/recipes/gcc/ (6 files in 3 dirs): Dec 04 07:27:37 gcc-4.5: Apply linaro updates and upstream updates Dec 04 07:27:37 Signed-off-by: Khem Raj Dec 04 07:27:49 03Khem Raj  07master * ra15c8fc323 10openembedded.git/recipes/uclibc/uclibc_git.bb: Dec 04 07:27:49 uclibc_git.bb: Bump SRCREV to latest git Dec 04 07:27:49 Signed-off-by: Khem Raj Dec 04 07:49:07 HopsNBarley: thanks, I think I made it finally :) Dec 04 10:17:23 morning Dec 04 10:17:39 morning Dec 04 10:18:18 congrats with release :) Dec 04 10:18:56 cbrake: good idea to merge release to testing and testing-next Dec 04 10:19:21 :) it's amazing it finally happened Dec 04 10:19:33 this is a good sign for all potential contributors Dec 04 10:21:17 it is nice to see a new release. Dec 04 10:25:26 gm Dec 04 10:25:37 good morning eFfeM Dec 04 10:25:58 morning eFfeM Dec 04 10:26:26 eFfeM, I"m hunting down that source that failed to be found in ipkg-utils-native Dec 04 10:31:34 oh, let me check here too, what was the name again Dec 04 10:32:46 eFfeM, I can't remember the exact name Dec 04 10:32:52 its on hte ML Dec 04 10:33:26 I asked mwester to look (as he maintains the slugOS branch) and I'm looking in the nslu2-linux source repo too. (but ididn't see it there) Dec 04 10:33:45 ah ok, not got that far, let me peek into it Dec 04 10:34:04 I have an older repo that is currrently offline I'm bringing that back online next week..maybe its there. Dec 04 10:38:14 already have it Dec 04 10:38:56 eFfeM, good. please let me know where I can get it and put it in the repo. Dec 04 10:39:05 yes, working at it Dec 04 10:39:13 (along with the correct md5sum that the recipe calls for :D Dec 04 10:39:16 ) Dec 04 10:40:07 some dns issue, can't get to our opensource repo from bheind the firewall Dec 04 10:40:11 and ip did not work either Dec 04 10:40:47 okay, when you have time..you can send it to the list or me directly. Dec 04 10:41:08 thanks. Dec 04 10:41:14 ftp://opensource.axon.nl/mirror/ipkg-utils_anoncvs.handhelds.org__20060106.tar.gz Dec 04 10:41:34 md5 file is not there, can't upload to that mirror, but will stuff it somewhere too, one sec Dec 04 10:42:17 eFfeM, you can post it here Dec 04 10:43:24 frans@linux-suse:~/oe> cat downloads/ipkg-utils_anoncvs.handhelds.org__20060106.tar.gz.md5 Dec 04 10:43:24 ccd585eb98e78c29bbbf91445b5ad732frans@linux-suse:~/oe> Dec 04 10:43:32 no linebreak in the file Dec 04 10:43:36 afk for 15 mins Dec 04 10:44:00 http://sources.nslu2-linux.org/sources/ Dec 04 10:44:09 have that ipkg-utils too Dec 04 10:44:13 google for that file :) Dec 04 10:44:42 I've found at least 3 locations, 2 are from osuosl :) Dec 04 10:44:46 I must have missed that. Dec 04 10:44:57 http://ftp-osl.osuosl.org/pub/nslu2/sources/ Dec 04 10:45:03 http://ftp-nyc.osuosl.org/pub/nslu2/sources/ Dec 04 10:45:20 oh I kept forgetting we put them up there...thought that was offline. Dec 04 10:45:32 :) Dec 04 10:45:56 (that was next weeks project.) Dec 04 10:53:31 ccd585eb98e78c29bbbf91445b5ad732 ipkg-utils_anoncvs.handhelds.org__20060106.tar.gz Dec 04 10:53:51 good matches Dec 04 10:59:58 eFfeM, I keep forgetting we have like 100GB of source on the nslu2-linux sourcemirror. Dec 04 11:00:13 and all of its for embedded stuffs. Dec 04 11:00:18 it is Dec 04 11:00:46 ka6sox: there was mail in ML about sources that fails to fetch Dec 04 11:01:14 may be you have some other :) Dec 04 11:01:18 about 24hrs ago eFfeM posted this one. Dec 04 11:01:37 no, I mean older Dec 04 11:01:59 there was script for checking downloads also Dec 04 11:02:52 oh, I didn't see that. Dec 04 11:03:17 I just inherited the job of trying to keep sources around and make sure we have them. Dec 04 11:06:22 http://article.gmane.org/gmane.comp.handhelds.openembedded/39885 Dec 04 11:06:24 this one Dec 04 11:06:50 some was fixed already Dec 04 11:07:30 re Dec 04 11:10:24 * eFfeM is reading the news, always had the impression that US wa the country defending free speech: http://gawker.com/5705492/library-of-congress-is-latest-government-institution-to-block-wikileaks Dec 04 11:11:51 anyway... back to work Dec 04 11:13:14 Jay7, thats quite a laundry list Dec 04 11:14:22 hi all Dec 04 11:14:31 evening Dec 04 11:14:48 hi bluelightning Dec 04 11:15:45 wrt the list from Klaus, my preference would be to start removing all recipes that do not fetch and are not the last version or pinned Dec 04 11:17:04 then again the list is not sound, it mentions both flex recipes that we have, can't imagine that both do not fetch Dec 04 11:17:15 that might be a good proposal for the ML. I wonder how many are using non-pinned recipies with bad sources. Dec 04 11:17:53 eFfeM, up until recently I didn't have good opie sources Dec 04 11:18:19 and with hh.org being down that was a major pain too. Dec 04 11:19:42 so I could take Klaus' list (or rerun the script) and start looking to bring those sources that are on the list inhouse. Dec 04 11:19:56 ka6sox: removing old versions is highly controversial even if they are old and not pinned. Dec 04 11:20:31 There is always this mythical user that has an out-of-tree overlay that uses this recipe Dec 04 11:20:55 eFfeM, I met one this week. Dec 04 11:21:03 not mythical. Dec 04 11:21:06 but this mythical user is very silent, as he never posts to the list or otherwise contributes or makes known that he has an interest in that recipe Dec 04 11:21:11 ka6sox: ah ok Dec 04 11:21:41 actually @work we use quite some older recipes, but we also keep the recipes in an overlay if needed Dec 04 11:22:06 for some older products we want to keep things as stable as possible, so we're not moving to newer versions Dec 04 11:22:28 what I'd really like to do is to be able to rsync to the larger repos that I've seen out there. Dec 04 11:22:35 and pull in what we are missing Dec 04 11:23:13 the recipies are only as good as the sources. Dec 04 11:23:38 yes Dec 04 11:24:28 actually one of the things I like about yocto is that they typically have one well-maintained version instead of several ones of which the status is fuzy Dec 04 11:24:59 for core recipies thats going to be good. Dec 04 11:25:08 the corner cases are still going to be tough Dec 04 11:26:40 okay I'll rerun the script and see whats missing still/now Dec 04 11:28:05 Jay7, are you doing any hardware characterizations for the build testing you are doing? Dec 04 11:32:37 ka6sox: I have a fair number of things archived but typically only those that are most recent versions or pinned (as build quite a lot and hardly ever clean downloads Dec 04 11:35:41 ka6sox: what do you mean under 'hardware characterizations'? Dec 04 11:36:04 I've not tried images on real hw if you mean this Dec 04 11:36:41 I mean CPU Performance/Memory Performance/HDD Array Performance Dec 04 11:36:45 ah Dec 04 11:37:12 have not done this long time Dec 04 11:38:01 what I'm trying to come up with is a way to characterize build servers so that we can maximize the utilization without causing thrashing (due to either Disk saturation/memory Swapping/Disk thrashing) mostly Dec 04 11:38:22 https://spreadsheets.google.com/a/jay-tech.ru/ccc?key=0AhY8lmfkCabTdFlPQ1ozR2ZNM0Z4ckpLZ3VOLVE2dnc&authkey=CKy8lbkH Dec 04 11:38:27 this was last time Dec 04 11:38:34 vmstat Dec 04 11:40:48 ka6sox: this need lot of testing on different systems Dec 04 11:41:01 s/testing/accounting/ Dec 04 11:41:01 Jay7, yes, I agree Dec 04 11:41:20 my testbuilder can account ps, df and vmstat Dec 04 11:41:37 ps == # of processes Dec 04 11:42:32 I think we either need to create a test suite or find one that we can use. then feed the output into bb and let the scheduler use that information to maximize utilization without "thrashing" Dec 04 11:43:31 btw, now I'm using 6/6 threads Dec 04 11:43:50 but I've lowered cpu.shares value for my OE lxc Dec 04 11:44:11 the interrupts/context switches are very high. Dec 04 11:44:40 one thing I would like to gather is time in state. Dec 04 11:44:43 RAM usage is too high as well Dec 04 11:45:21 linking and autotools seem to be the worst offenders. Dec 04 11:45:26 my 4Gb is not enough.. Dec 04 11:45:41 desktop tasks are using about 1-1.5Gb of RAM Dec 04 11:46:55 staying out of swap is very important to prevent thrashing. Dec 04 11:47:32 but your wait state numbers are quite good. Dec 04 11:49:41 it would be interesting to know, in your case where you are limited in RAM as to which task are forcing you into swap Dec 04 11:49:57 I suspect either linking or autotools. Dec 04 11:50:26 it's hard to find what tasks are doing what load.. Dec 04 11:50:51 we should do accounting right from bitbake Dec 04 11:51:15 or at least change log format Dec 04 11:52:12 being able to identify task costs would help with lowering build times. Dec 04 11:55:57 Jay7, in your next runs could you gather load averages too? Dec 04 11:56:22 last time it was about 7-8 :) Dec 04 11:56:35 ouch Dec 04 11:56:40 but afaik, LA is # of processes in runqueue Dec 04 11:56:50 i.e. vmstat's first column Dec 04 11:57:23 ah yes, i see that now Dec 04 11:58:16 7-8 is too high to be efficient. Dec 04 11:58:22 ka6sox: run on graphs was with lover bb/make threads values Dec 04 11:59:39 seems like later runs lowered the # of processes waiting. Dec 04 11:59:40 # of processes in runqueue should be equal to CPUs afaik Dec 04 11:59:51 ideal case Dec 04 12:00:06 do you count hyperthreaded cores? Dec 04 12:00:22 i.e. for my 6 cores this is good to have LA=6 Dec 04 12:00:29 but I should ensure that Dec 04 12:00:59 my understanding of load avg is that its processes *waiting* to be run Dec 04 12:01:05 not actually executing. Dec 04 12:01:57 may be.. I'll ask people about Dec 04 12:03:31 03Melissa Watkins  07org.openembedded.dev * r66dd7b3575 10openembedded.git/recipes/ti/ti-pru-sw-edma-library_1.00.00.bb: (log message trimmed) Dec 04 12:03:31 ti-pru-sw-edma-library: first addition of PRU sw edma library Dec 04 12:03:31 * Add PRU sw edma library recipe for the omapl138 and am181x-evm machines Dec 04 12:03:31 This recipe builds and stages the edma user space libraries for use by Dec 04 12:03:31 the PRU sw. Because the edma APIs available by the kernel can only be Dec 04 12:03:40 03Melissa Watkins  07org.openembedded.dev * re8d5610aa1 10openembedded.git/ (5 files in 3 dirs): (log message trimmed) Dec 04 12:03:40 linux-davinci_git: Patch omapl138 kernel Dec 04 12:03:40 * Update the linux recipe for linux-davinci_git to patch kernel for Dec 04 12:03:40 omapl138 machines Dec 04 12:03:40 * Add uio_pruss kernel patches for omapl138 platform Dec 04 12:03:47 This recipe builds the PRU application loader that is used to load a Dec 04 12:03:47 binary to the PRU's memory and manage the code executed in the PRU from Dec 04 12:03:48 the user space. The library and include files from the application Dec 04 12:03:48 loader are staged for use by the ti-pru-sw-examples recipe. Details Dec 04 12:03:49 03Melissa Watkins  07org.openembedded.dev * rddee5c623c 10openembedded.git/recipes/linux/linux-davinci/omapl138/defconfig: Dec 04 12:03:49 linux-davinci: enable UIO support in l138 defconfigs Dec 04 12:03:50 Signed-off-by: Melissa Watkins Dec 04 12:04:01 Signed-off-by: Koen Kooi Dec 04 12:04:03 03Melissa Watkins  07org.openembedded.dev * r672f472b08 10openembedded.git/recipes/ti/ti-pru-sw-examples_1.00.00.bb: (log message trimmed) Dec 04 12:04:04 ti-pru-sw-example: first addition of PRU sw examples recipe Dec 04 12:04:04 * Add PRU sw examples recipe for the omapl138 machines Dec 04 12:04:04 The ti-pru-sw-example recipe builds example codes that demonstrate use Dec 04 12:04:04 of the AM180x-evm PRU, compiles the PRU binaries, and saves example Dec 04 12:05:37 Jay7, you are correct in your understanding Dec 04 12:05:55 so an ideal number for 6 cores is 6.0 Dec 04 12:06:16 thats maximum utilization without any waiting processes. Dec 04 12:06:38 (waiting on CPU) Dec 04 12:15:11 yes Dec 04 12:15:42 having LA <= # of cores is ok Dec 04 12:49:19 03Koen Kooi  07org.openembedded.dev * r65d9c71b42 10openembedded.git/recipes/xorg-lib/ (24 files in 2 dirs): (log message trimmed) Dec 04 12:49:19 pixman 0.21.2: updates Dec 04 12:49:19 * add 12 more NEON patch Dec 04 12:49:19 * add 1 C fast path Dec 04 12:49:19 * 1 autofoo fix Dec 04 13:15:05 Where are xorg.conf and asound.state files stored for specific devices in oe? Dec 04 13:20:48 nm I found them Dec 04 13:56:28 ka6sox: I have ipkg-utils_anoncvs.handhelds.org__20050404.tar.gz (along with numerous other verrsions of that repo) Dec 04 13:56:40 Still need that file? Dec 04 13:57:09 * mwester has not been paying attention to his downloads dir, and it's up to 10GBytes now :/ Dec 04 14:02:05 xload is MLO Dec 04 14:17:47 Are there any ixp4xx-based machines that qemu can emulate? Dec 04 14:41:15 gm Dec 04 14:43:15 I'm noticing some issues with native compiles on OMAP Dec 04 14:43:37 and have tracked it down to the libgcc-dev package Dec 04 14:44:39 It contains a broken link - /usr/dev/libgcc_s.so Dec 04 14:47:11 the libgcc_s.so link is broken -- it expects the target libgcc_s.so.1 in /usr/lib, but gcc packages that library in /lib Dec 04 14:47:44 does anyone know the proper way to fix this? Dec 04 14:49:12 It's simple enought to correct the link on the target system manually, but I'd like to get the recipe corrected properly Dec 04 14:49:59 Searching the mailing list I see quite a few messages referring to issues with libgcc_s.so, so I expect I am not the only one encountering this Dec 04 15:07:02 ~seen khem Dec 04 15:07:07 khem is currently on #oe (1d 23h 28m 27s) #edev (1d 23h 28m 27s) #uclibc (1d 23h 28m 27s). Has said a total of 129 messages. Is idling for 7h 17m 57s, last said: 'HopsNBarley: thanks, I think I made it finally :)'. Dec 04 15:08:51 please ignore my comments above :-) Dec 04 15:09:24 I just did a pull and see that khem has addressed this :-) Dec 04 15:23:21 03Koen Kooi  07org.openembedded.dev * rf0688a49eb 10openembedded.git/recipes/gcc/gcc-4.5.inc: Dec 04 15:23:21 gcc 4.5: comment out patch that doesn't apply Dec 04 15:23:21 * This is temporary till a proper fix lands Dec 04 15:23:21 Signed-off-by: Koen Kooi Dec 04 15:57:54 03Koen Kooi  07org.openembedded.dev * re3b1e3fb89 10openembedded.git/recipes/powervr-drivers/ (3 files in 2 dirs): Dec 04 15:57:54 omap3-sgx-modules: some more for fixes for 2.6.37rc Dec 04 15:57:54 Signed-off-by: Koen Kooi Dec 04 16:25:49 hi obi Dec 04 16:26:57 03Khem Raj  07master * r3fd62b8f49 10openembedded.git/recipes/gcc/ (6 files in 2 dirs): Dec 04 16:26:57 gcc-4.5: Refresh the lately added patches with quilt Dec 04 16:26:57 Signed-off-by: Khem Raj Dec 04 16:27:30 he khem Dec 04 16:28:09 khem vs koen :) Dec 04 16:28:26 hehe Dec 04 16:28:41 he mickey Dec 04 16:29:01 hi woglinde Dec 04 16:29:03 how are things? Dec 04 16:30:09 hi Dec 04 16:30:09 I am bit behind my plan Dec 04 16:30:22 but in 7 days is my thesis talk Dec 04 16:31:05 ah, cool Dec 04 16:31:08 so end in sight? Dec 04 16:31:41 hm yes Dec 04 16:31:44 3 oral exams Dec 04 16:32:26 and one report Dec 04 16:33:04 about partical work Dec 04 16:33:28 not too bad. so definitely end in 2011. Dec 04 16:33:36 yes Dec 04 16:36:53 hello woglinde Dec 04 16:36:59 long time no see Dec 04 16:37:20 jo khem Dec 04 16:37:27 khem I have a problem with .la files Dec 04 16:37:35 hmm Dec 04 16:37:40 which package Dec 04 16:38:10 the problem is when compiling openjdk Dec 04 16:38:26 woglinde: hmmm Dec 04 16:38:31 openjdk configures a version of cacao its self Dec 04 16:38:38 k Dec 04 16:39:00 and the libtool inside this cacao uses the .la from sysroot as dep Dec 04 16:39:18 hmmm Dec 04 16:39:20 and so -lz is converted too /usr/lib/libz.so as final call Dec 04 16:39:40 is it with libtool 2.4 ? Dec 04 16:39:51 unfornatly the same happens when I run autreconf with libtool 2.5 Dec 04 16:39:55 args 2.4 Dec 04 16:40:04 ok Dec 04 16:40:07 removing the libz.la solves it Dec 04 16:40:21 removing from where ? Dec 04 16:40:25 fro sysroot ? Dec 04 16:40:38 I investigated that libstdc++.la has the sysroot path instead of /usr/lib in sysroot Dec 04 16:40:42 form sysroot yes Dec 04 16:41:02 hmm Dec 04 16:41:18 should not be the case Dec 04 16:41:55 did u try to upgrade the libtool macros for openjdk/cacao Dec 04 16:42:11 yes Dec 04 16:42:26 how can I reproduce it Dec 04 16:42:34 build openjdk Dec 04 16:42:35 bitbake openjdk ? Dec 04 16:42:38 k Dec 04 16:42:54 bitbake openjdk-6-jre Dec 04 16:47:07 woglinde: I will try to reproduce it Dec 04 16:48:35 thanks Dec 04 16:49:02 I discovered that openjdk cannt build with vpu=hard Dec 04 16:49:07 args fpu Dec 04 16:49:15 okay have to go now for a while Dec 04 16:54:28 hm, gcc crashes on the autobuilder when building samba Dec 04 16:54:46 ICE Dec 04 16:59:08 trying a clean on git head to verify things Dec 04 17:00:00 eFfeM: I've seen this too Dec 04 17:00:16 http://tinderbox.openembedded.net/packages/1178390/ Dec 04 17:00:18 e.g. here Dec 04 17:01:49 hm, seems a regression is gcc, this worked for me a while ago Dec 04 17:05:09 mwester, found it in 3 different repos (including the nslu2-linux ones) Dec 04 17:07:48 Jay7: it is not the same error as Iget, but mine is not on tinderbox Dec 04 17:10:30 triying also on the local system, will report later, afk for a while Dec 04 18:47:23 is patchwork going to be revived someday? what's the problem with it? is anyone investigating? Dec 04 18:47:32 it is being worked on Dec 04 18:48:03 ok, thanks Dec 04 19:44:23 03Khem Raj  07master * rf7b3b6a287 10openembedded.git/recipes/gcc/gcc-4.5.inc: Dec 04 19:44:23 gcc-4.5.inc: Re-enable linaro/gcc-4.5-linaro-r99419.patch Dec 04 19:44:23 Signed-off-by: Khem Raj Dec 04 19:45:07 eFfeM: I am aware of the samba issue on arm Dec 04 19:45:22 eFfeM: I have the file in question somewhere, I just need to reduce it Dec 04 19:49:57 ericben|away: around ? Dec 04 19:56:06 re Dec 04 19:57:37 khem something new? Dec 04 19:58:48 woglinde: regarding openjdk ? Dec 04 19:58:55 not yet. Its still buildin Dec 04 19:59:06 my machine is bit old takes some tie Dec 04 20:00:03 okay Dec 04 20:00:18 I wonder if we need .la files at all Dec 04 20:01:32 In theory with some changes we can get rid of them Dec 04 20:02:08 but static and .so selection etc might not work as well Dec 04 20:02:26 * khem 's son thrashed the 'm' key on the keyboard Dec 04 20:04:40 khem, ok good, thanks for the info (and it saves me the time to write a problem report) Dec 04 20:06:27 khem, btw the problme is in rpc_client/cli_reg.c Dec 04 20:07:05 eFfeM: hi Dec 04 20:07:10 yes Dec 04 20:07:21 if you compile just that file with -O0 you will get thru Dec 04 20:10:35 03Klaus Kurzmann  07master * r8ed56c77b9 10openembedded.git/ (6 files in 3 dirs): Dec 04 20:10:35 glib-2.0: add 2.26 with DEFAULT_PREFERENCE = "-1" and prefer it for SHR Dec 04 20:10:35 Signed-off-by: Klaus Kurzmann Dec 04 20:15:29 davidlt: , hi Dec 04 20:15:56 khem, I'm in no urgent need for the file, still try to get a good build on our autobuilder Dec 04 20:26:03 eFfeM: ok Dec 04 20:26:25 eFfeM: I have updated gcc if we are sheer lucky we might have a fix already Dec 04 20:26:34 but thats like 0.1% chance Dec 04 20:27:24 khem: talking about gcc... how good is the arm implementation of atomic? Dec 04 20:28:04 mrmoku: I think its good but you might still beat it Dec 04 20:28:14 mrmoku: they are intrinsics Dec 04 20:28:21 ok, because glib-2.0 2.26 is using that Dec 04 20:28:23 so someone smart has written them Dec 04 20:29:50 mrmoku: are you seeing some perf problems Dec 04 20:29:54 wrt to them Dec 04 20:30:02 so far not Dec 04 20:30:13 did only try it on armv7 (nokia n900) Dec 04 20:30:28 khem, I'll give it a try Dec 04 20:31:04 mrmoku: ok Dec 04 20:31:21 khem: next device will be the good old freerunner :) Dec 04 20:31:36 heh Dec 04 20:32:17 gcc is using the kernel stuff I think Dec 04 20:32:24 for atomics Dec 04 20:32:58 hmm Dec 04 20:34:11 hi all Dec 04 20:34:34 hi reactor Dec 04 20:35:03 ah some atomic conversations Dec 04 20:35:11 hi Dec 04 20:35:28 hm not really Dec 04 20:35:32 I've some atomic problems : http://pastebin.com/yZVUdn44 Dec 04 20:36:25 hm armv4t? Dec 04 20:36:41 re-insert the control rods!!! (I saw that movie!) Dec 04 20:36:43 armv7 Dec 04 20:36:56 gnutoo hm Dec 04 20:37:03 which package is this? Dec 04 20:37:08 fsogsmd Dec 04 20:37:17 from autorev Dec 04 20:37:25 on n900? Dec 04 20:37:31 yes Dec 04 20:37:39 which kernel? Dec 04 20:37:56 2.6.28 Dec 04 20:38:01 hm yes Dec 04 20:38:09 2.6.28 might be too old Dec 04 20:38:14 ah ok Dec 04 20:38:23 what should I do? Dec 04 20:38:32 and what is the issue exactly? Dec 04 20:38:37 we should try to go for the newer kernel again ;) Dec 04 20:38:50 mrmoku, it's not finished Dec 04 20:38:54 look into the kernel stuff which patches are applied since 2.6.28 for atomic stuff Dec 04 20:39:06 for arm Dec 04 20:39:15 why does it have to do with the kernel for armv7 Dec 04 20:39:19 for armv4 I understand Dec 04 20:39:30 but armv7 has a lots of atomic instructions right? Dec 04 20:39:36 does it lack theses Dec 04 20:39:36 ? Dec 04 20:39:43 as I said earlier If I remember correctly gcc is using linux kernel stuff Dec 04 20:39:53 ah ok Dec 04 20:39:54 woglinde: yeah, but GNUtoo|laptop is not yet using 2.26 glib Dec 04 20:39:55 and on 2.6.28 the kernel stuff might have some problems Dec 04 20:40:08 I'll look which glib I'm using Dec 04 20:40:55 libglib-2.0-0 - 2.24.1-r2.6 Dec 04 20:43:18 hm intressting Dec 04 20:43:23 still no sqlit 3.7 Dec 04 20:49:13 woglinde, what are my options? backport some kernel patches to 2.6.28? where should I look in arch/arm ? Dec 04 20:49:31 yeah think so Dec 04 20:49:33 only option Dec 04 20:49:37 indeed Dec 04 20:49:42 as 2.6.35 is not finished Dec 04 20:49:49 lacks lot of stuff Dec 04 20:49:59 require a lots of userland changes Dec 04 20:50:05 that include frustrating stuff Dec 04 20:50:22 where you don't have usbnet and touchscreen Dec 04 20:50:33 I'll look then Dec 04 20:50:36 but where to start? Dec 04 20:52:48 GNUtoo|laptop: that segfault in __sync_... Dec 04 20:52:59 is without your plugin via shr-settings? Dec 04 20:53:06 yes of course Dec 04 20:53:15 ok Dec 04 20:53:21 as you have it too Dec 04 20:53:25 and don't have my plugin Dec 04 20:53:33 anyway I tested a clean version Dec 04 20:53:39 from oe Dec 04 20:53:42 it has it Dec 04 20:53:48 (the segfault) Dec 04 21:01:42 hm Dec 04 21:01:49 will update nspr and nss Dec 04 21:01:56 and then try building openjdk with nss Dec 04 21:02:03 and update sqlite too Dec 04 21:04:21 hm intressting python starts to segfault here again Dec 04 21:09:21 my kernel get segfault on shutdown :) Dec 04 21:09:31 Nothing is strange for me already Dec 04 21:09:35 hm thats okay Dec 04 21:09:41 bios shit Dec 04 21:10:03 I just compiled SQLite3 3.7.3, created a database with WAL Dec 04 21:10:08 or some strange driver Dec 04 21:10:16 And I get I/O Unkown Eror Dec 04 21:10:21 yeah you need PRAGMA command Dec 04 21:10:39 It happens after few mmap() and trying to get data Dec 04 21:10:52 Now I have another program with mallocs Dec 04 21:11:05 And It fails with glibc realloc() stuff Dec 04 21:11:08 http://www.sqlite.org/wal.html Dec 04 21:11:15 While it does work on my computer Dec 04 21:11:37 It works on my main computer, but on the chip it doesn't work Dec 04 21:11:45 It kinda looks like memory related errors Dec 04 21:11:55 faulty ram on the board? Dec 04 21:11:56 I am thinking maybe something wrong with OS? Dec 04 21:12:03 I don't know Dec 04 21:12:28 try some other stress tests Dec 04 21:12:31 or benchmarks Dec 04 21:12:49 My Gumstix now has limited memory to 128 from 256, because of initramfs not working correctly or smth. Dec 04 21:13:23 Would be nice to do that, but know have knowledge much. Any suggestion on exact test utilities? Dec 04 21:13:41 hm mbench? Dec 04 21:13:47 I think we have it in oe Dec 04 21:14:21 hm no Dec 04 21:14:29 Unknown package 'mbench' Dec 04 21:14:48 lmbench Dec 04 21:14:49 sorry Dec 04 21:15:29 okay I will go to sleep Dec 04 21:15:31 good nite Dec 04 21:15:48 night and thanks Dec 04 21:38:12 03Frans Meulenbroeks  07org.openembedded.dev * r0b0a6b43ad 10openembedded.git/recipes/cdparanoia/cdparanoia_10.2.bb: Dec 04 21:38:12 cdparanoia_10.2.bb: fixed LICENSE Dec 04 21:38:12 Signed-off-by: Frans Meulenbroeks Dec 04 21:43:27 hello. I have a software that i want oe frnedly but I have header issue. I have a #include that need to be deferent for OE than on my host. is there a #define that could help me ? Dec 04 21:51:23 GarthPS, I think that the question is rather how to handle the different requirement between oe and ubuntu, that is to say: Dec 04 21:51:27 #include on ubuntu but #include on OE Dec 04 21:51:35 as you said in #openmoko-cdevel Dec 04 21:51:53 GNUtoo|laptop: yop Dec 04 21:51:59 thx Dec 04 21:52:03 you could make an autotools test maybe Dec 04 21:52:15 do you use autotools? Dec 04 21:52:22 GNUtoo|laptop: nop I use qmake Dec 04 21:52:27 ah ok Dec 04 21:52:33 maybe look in qmake's tests Dec 04 21:52:34 then Dec 04 21:55:36 GarthPS, maybe everybody is sleeping Dec 04 21:55:36 ? Dec 04 21:55:45 woglinde went to sleep Dec 04 21:56:48 GNUtoo|laptop: :) Dec 04 21:57:05 GarthPS, btw what are you writing? Dec 04 21:57:36 GNUtoo|laptop: https://gna.org/projects/myopencoach it has been a while without any significant update Dec 04 21:58:00 wow nice Dec 04 21:58:25 it's the kind of software that makes sports attractive for geeks :) Dec 04 21:58:33 openstreetmap is another one Dec 04 21:58:42 GNUtoo|laptop: lol yep Dec 04 21:58:52 or geocaching(I never tried geocaching) Dec 04 21:59:04 GNUtoo|laptop: so i need qwt to plot accelerometers Dec 04 21:59:17 then gps. speed etc.. Dec 04 21:59:17 ah ok Dec 04 21:59:55 is there some old screenshots? Dec 04 21:59:57 or new ones? Dec 04 22:00:38 unfortunately milan is not great for doing outside sports Dec 04 22:00:59 but maybe during my summer vacations in france.... Dec 04 22:01:13 what hardware does your software requires Dec 04 22:01:27 is it more advanced than tangogps? Dec 04 22:01:28 GNUtoo|laptop: nop until 0.1 published, nothing. Dec 04 22:01:29 it seem so Dec 04 22:01:57 GNUtoo|laptop: nop until 0.1 published, nothing. was for the screenshots? Dec 04 22:02:22 GNUtoo|laptop: yop Dec 04 22:02:24 http://svn.gna.org/viewcvs/myopencoach/trunk/TODO?view=markup Dec 04 22:02:34 ok Dec 04 22:03:14 but my list for th 0.1 is quit un clear because I am actualy trying to do 0.2 before 0.1 is completefd :) Dec 04 22:03:23 lol Dec 04 22:03:23 ok Dec 04 22:18:40 GNUtoo|laptop: yes! thx qgpslog! Dec 04 22:18:49 GNUtoo|laptop: linux-gnueabi-oe-g++ is defined on OE Dec 04 22:19:46 * GNUtoo|laptop is digging again in __kernel_cmpxchg Dec 04 22:21:14 but it's so strange that it works that way for armv7 Dec 04 22:28:49 anyone know when the embedded devroom cfp will go out? Dec 04 22:28:56 for fosdem Dec 04 22:30:17 nope Dec 04 22:30:30 Crofton|work: which hotel do we book this year? Dec 04 22:30:41 not sure Dec 04 22:30:47 we need to figure that out :) Dec 04 22:33:53 where did people stay last yar? Dec 04 22:40:19 Astrid Dec 04 22:40:51 no wifi, breakfast = extra 8euro but good connection to fosdem place and easy to get there from beer event Dec 04 22:55:33 hmmm Dec 04 22:55:47 NEEDS_SYSCALL_FOR_CMPXCHG is for what arm versions Dec 04 22:55:48 ? Dec 04 22:56:19 v7 don't need it right? Dec 04 22:56:33 cmpxchg is not x86 only instruction? Dec 04 22:56:43 no Dec 04 22:56:46 look in: Dec 04 22:57:11 arch/arm/entry-armv.S Dec 04 22:57:34 there is even an lwn article about it: http://lwn.net/Articles/314235/ Dec 04 22:57:48 that reminds me things because I already digged in it for armv4 Dec 04 22:57:58 but here I'm on armv7 Dec 04 22:58:40 http://lxr.free-electrons.com/source/arch/arm/kernel/entry-armv.S?a=arm Dec 04 22:58:55 sorry it's arch/arm/kernel/entry-armv.S Dec 04 22:59:04 anyway the link is correct Dec 04 22:59:08 but not that: Dec 04 22:59:11 arch/arm/entry-armv.S Dec 04 23:00:01 ah I read too fast again,you said instruction Dec 04 23:52:40 03Frans Meulenbroeks  07org.openembedded.dev * r8ea3c122ac 10openembedded.git/recipes/gob2/gob2_2.0.17.bb: (log message trimmed) Dec 04 23:52:40 gob2: updated to 2.0.17 Dec 04 23:52:40 When trying to build gmpc on our autobuilder (64 bit ubuntu 10.04) I got compile errors Dec 04 23:52:40 because gob2 apparently corrupted some data it generated causing type names to be changed. Dec 04 23:52:40 On opensuse 11.2, 32bit it worked like a charm. Dec 05 00:25:54 Crofton|work, you still around? Dec 05 00:28:46 yeah Dec 05 00:29:13 do we have an email account: patches@openembedded.org? Dec 05 00:29:30 I do not know Dec 05 00:29:38 do we need one? Dec 05 00:29:47 khem, what do we need? Dec 05 00:33:43 oe email is done on ltg and I am not an admin there Dec 05 00:36:01 Crofton|work: are you admin on old pw Dec 05 01:13:23 eFfeM_work: try git log -SAUTOTOOLS_STAGE_PKGCONFIG classes **** ENDING LOGGING AT Sun Dec 05 02:59:58 2010