**** BEGIN LOGGING AT Wed Aug 11 02:59:57 2010 Aug 11 03:24:05 hey, I've been given a binary that I want to add to an image temporarily. It executes if I manually copy it across but I want to add it to the image. I keep getting a QA error during packaging (No GNU_HASH). Anyway to ignore these so that do_package finishes anyway? Aug 11 03:26:14 google keeps telling me to play with LDFLAGS, but that's not really an option. And the executable works... it's just failing checks because it doesn't have a gnu hash table Aug 11 03:53:56 porcoesphino: you dont build it from source ? Aug 11 04:46:13 khem: I can confirm that same toolchain (gcc-4.5, binutils-2.20.1, eglibc-2.12) and now also same distro (SHR), built on same buildhost does have those alignment traps on armv5te (so the same as on armv4t), I'll try to narrow the diff between Angstrom and SHR Aug 11 05:24:59 gh Aug 11 05:25:00 gm Aug 11 05:25:56 gm Aug 11 05:28:43 anyone with autotools knowledge around who can help me with this: http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-August/022614.html Aug 11 05:29:09 i'm pretty stuck with it, probably due to limited knowledge of autotools Aug 11 05:29:52 current solution is to develop a patch which is applied after configure (but that solution does not seem too nice) Aug 11 05:35:41 JaMa: oh, Angstrom does not have them? Aug 11 06:23:57 mrmoku|away: no Aug 11 06:24:51 mrmoku|away: I thought it was because of different arch, but now it seems that's because different CFLAGS Angstrom is using or something like that (as I have Angstrom updated to same toolchain as we use) Aug 11 06:36:26 there is still something wrong with multimachine cross builds :/ see this log http://tinderbox.openembedded.org/packages/682521/ Aug 11 07:09:36 JaMa: LD_RUN_PATH="/lib" Aug 11 07:09:41 thats bogus Aug 11 07:10:19 morning Aug 11 07:13:02 hi hrw Aug 11 07:17:42 khem: but why it used armv6? Aug 11 07:17:55 khem: while building for armv4t I mean Aug 11 07:18:14 khem: and armv6 build from same tmpdir works ok Aug 11 07:34:53 03Martin Jansa  07org.openembedded.dev * ra4c72e350c 10openembedded.git/recipes/ (kexecboot/linux-kexecboot_2.6.35.bb linux/linux_2.6.35.bb): Aug 11 07:34:53 linux(-kexecboot)-2.6.35: upgrade to latest stablepatch 2.6.35.1 Aug 11 07:34:53 Signed-off-by: Martin Jansa Aug 11 07:34:53 03Martin Jansa  07org.openembedded.dev * re6b733c03e 10openembedded.git/recipes/tasks/task-shr-feed.bb: Aug 11 07:34:54 task-shr-feed: add xfsdump Aug 11 07:34:55 Signed-off-by: Martin Jansa Aug 11 07:36:13 03Martin Jansa  07org.openembedded.dev * r1816ba3639 10openembedded.git/recipes/xfsprogs/xfsdump_3.0.4.bb: Aug 11 07:36:13 xfsdump: add recipe Aug 11 07:36:13 Signed-off-by: Martin Jansa Aug 11 07:36:13 03Martin Jansa  07org.openembedded.dev * r8d5d9701a1 10openembedded.git/recipes/xfsprogs/xfsprogs_3.1.2.bb: Aug 11 07:36:14 xfsprogs: stage dev headers, used later by xfsdump Aug 11 07:36:15 Signed-off-by: Martin Jansa Aug 11 08:55:41 03Martin Jansa  07org.openembedded.dev * rd1c85b5333 10openembedded.git/recipes/xfsprogs/xfsdump_3.0.4.bb: xfsdump: add xfsprogs to DEPENDS Aug 11 09:30:53 hi zecke Aug 11 09:31:15 woglinde: do you remember if the modutils were broken recently? Aug 11 09:31:21 no Aug 11 09:31:23 sorry Aug 11 09:31:41 why should they broken? compile or execute? Aug 11 09:31:57 * zecke can't wait to buy a new workstation in berlin... my laptop takes days to compile OE Aug 11 09:32:04 woglinde: -lc missing when linking Aug 11 09:32:04 hehe Aug 11 09:32:10 6core amd? Aug 11 09:32:17 or intel? Aug 11 09:32:34 woglinde: 128 Atom CPUs? Aug 11 09:32:51 needs to much power Aug 11 09:32:59 10xbeagle :) Aug 11 09:33:06 problem is anyway the disks Aug 11 09:33:22 see this often with my phenom2 Aug 11 09:35:46 let's hope new SSD will be bigger and cheaper enough Aug 11 09:37:02 oe builds would be a good stress test Aug 11 09:37:10 hi ant btw. Aug 11 09:37:26 sure, I burned two disks this summer... Aug 11 09:37:51 cheap hdd shouldn't be so stressed Aug 11 09:40:04 which vendor? Aug 11 09:40:06 2? Aug 11 09:40:24 one was a Seagate Barracuda Aug 11 09:40:29 othe a Samsung Aug 11 09:41:02 imagine I have still some SCSI hdd from 2000 on a backup server Aug 11 09:41:07 hm samsung was good enough for me the last years Aug 11 09:41:38 I even got a refurbished 70gig disk, just 2 weeks before the warranty was over Aug 11 09:41:44 for my borken Aug 11 09:42:09 now I got 2 Seagate Barracuda 7200.12 - Hard disk - 500 GB Aug 11 09:42:15 one platte Aug 11 09:42:24 hopefully more resistent... Aug 11 09:42:49 hm the trick with scsi disk is Aug 11 09:42:52 dont turn them off Aug 11 09:45:12 I've replaced the server , now we have 2 fusion mpt SAS in raid1. It's a little server Aug 11 09:45:32 rit's up since october Aug 11 09:45:59 this SAS seems good, expensive though Aug 11 09:47:34 hm yes Aug 11 09:48:18 btw is Siemens now definitely out of this server business? Aug 11 09:48:40 I mean with Fujitsu Aug 11 09:49:08 woglinde: which phenom you have? Aug 11 09:49:21 phenom2x4 the smalltes one Aug 11 09:49:25 2,4 ghz Aug 11 09:49:28 * JaMa model name : AMD Phenom(tm) II X4 965 Processor Aug 11 09:49:40 for next machine (but in ~May 2011) I am starting to checking which way to go Aug 11 09:49:48 mine should be 8xx something Aug 11 09:49:55 cpu MHz : 3400.000 Aug 11 09:50:04 hrw sure intel should be have something new than Aug 11 09:50:10 1. am3 + 4/6 core + 8-16 gb ddr3 Aug 11 09:50:32 2. intel i5/i7/xeon 4/6 core + 8-16gb ddr3 Aug 11 09:50:57 xeon is to expensive Aug 11 09:51:16 woglinde: xeon 3430 is ~i7 860 or sth like that Aug 11 09:51:44 xeon profits from larger chaches Aug 11 09:52:03 ~curse the crappy inet connection here Aug 11 09:52:04 May you be reincarnated as a Windows XP administrator, the crappy inet connection here ! Aug 11 09:57:18 woglinde: x3430 ~= i5 760 (2.4 contra 2.8GHz but 8MB L2 instead of 1MB) Aug 11 09:59:20 I updated my old 775 witha xeon X3220 (2,4Ghz) and this latest stepping goes well beyond 3.2 ghz Aug 11 10:00:20 ant_work: I have q6600 (2.4GHz) now on 775 Aug 11 10:00:33 very similar Aug 11 10:00:52 33-40°C with low load Aug 11 10:02:25 I got it to 2.8GHz or more but then goodbye cpufreq Aug 11 10:08:11 I reached 58 C° but mine is almost fanless (out was 39 C°) Aug 11 10:08:32 then HDD broke :p Aug 11 10:09:18 I have mugen2 with 140mm fan blowing into it. then 140mm which pulls from case (behind cpu radiator) + 120mm fan in psu (over cpu radiator) Aug 11 10:09:39 I have old tuniq tower Aug 11 10:11:10 hrw: temps went down when I exchanged PSU Aug 11 10:11:23 I can really suggest Enermax EMG500AWT MODU 87 500W Aug 11 10:11:35 cold and silent up to 300w Aug 11 10:11:47 i.e. almost always (I'm not a gamer) Aug 11 10:12:21 I got one for free.... Aug 11 10:12:22 I have corsair 550W Aug 11 10:19:39 * hrw -> reboot Aug 11 10:33:36 re Aug 11 10:34:19 hm I am thinking about water cooling Aug 11 10:34:26 in the next time Aug 11 10:42:57 hello Aug 11 10:43:16 i will try to build cairo Aug 11 10:43:51 the problme is task do_rootfs failed Aug 11 10:47:59 no Aug 11 10:48:09 you do not try to build cairo then Aug 11 10:48:21 do_rootfs tells that you are building image Aug 11 10:51:51 ok thanks Aug 11 10:56:56 03Koen Kooi  07org.openembedded.dev * r56f7be7645 10openembedded.git/recipes/powervr-drivers/ (2 files in 2 dirs): omap3-sgx-modules: add patch to fix texture width detection Aug 11 11:28:25 khem: confirmed, O2 doesn't do alignment traps, Os does Aug 11 11:53:25 Hello Aug 11 11:55:14 I have a Gumstix Overo Fire with a Chestnut43 board. I bitbaked a omap3-console-image. Everything was fine. System is booting up properly. But if I try to install a program using opkg then it fails. Aug 11 11:55:34 Also an opkg update && opkg upgrade fails with following error: Aug 11 11:55:42 /usr/lib/opkg/info/hicolor-icon-theme.postinst: line 7: can't create /etc/gtk-2.0/gdk-pixbuf.loaders: nonexistent directory Aug 11 11:56:01 /usr/lib/opkg/info/hicolor-icon-theme.postinst: line 7: gdk-pixbuf-query-loaders: not found Aug 11 11:56:08 /usr/lib/opkg/info/hicolor-icon-theme.postinst: line 13: gtk-update-icon-cache: not found Aug 11 11:57:49 I checked my bitbake recipe and I can't anything that is pointing to gdk or gtk. Also googling didn't point to the right path. Does anybody know anything about why this error occurs? Aug 11 12:03:32 was it here we talked about the beer-bike in amsterdam (about a week ago)? The court has ruled it is not allowed Aug 11 12:03:41 Taalas remove the hicolor-icon-theme package Aug 11 12:13:56 woglinde thanks for the point. angstrom-zeroconf-audio had some dependencies which included hicolor-theme...After forcing to remove all depending packages everything worked fine Aug 11 12:14:20 seems hicolor-package needs some more deps Aug 11 12:16:41 I'm trying to get ruby on rails on a gumstix. But my ruby could not be installed because of this error... Aug 11 12:17:24 03Koen Kooi  07org.openembedded.dev * rfe8ec48505 10openembedded.git/recipes/upnp/gupnp_0.13.4.bb: gupnp: add 0.13.4 Aug 11 12:17:35 03Koen Kooi  07org.openembedded.dev * rb5ef0accda 10openembedded.git/recipes/rygel/rygel_0.6.0.bb: rygel: add 0.6.0 Aug 11 12:17:36 03Koen Kooi  07org.openembedded.dev * r9f75b210f2 10openembedded.git/recipes/upnp/gupnp-av_0.5.9.bb: gupnp-av: add 0.5.9 Aug 11 13:47:00 03Martin Jansa  07org.openembedded.dev * r6f16935793 10openembedded.git/recipes/bash/bash_3.2.bb: Aug 11 13:47:00 bash: bump PR after ncurses upgrade Aug 11 13:47:00 Signed-off-by: Martin Jansa Aug 11 13:47:12 03Martin Jansa  07org.openembedded.dev * r7751138384 10openembedded.git/recipes/tasks/task-shr-feed.bb: Aug 11 13:47:12 task-shr-feed: add numberx and neomis Aug 11 13:47:12 Signed-off-by: Martin Jansa Aug 11 13:48:17 03Koen Kooi  07org.openembedded.dev * r6d5bcdc04d 10openembedded.git/recipes/lmbench/lmbench_2.5.bb: lmbench: fix build Aug 11 13:50:41 re zecke Aug 11 15:32:43 morning Aug 11 15:33:48 hi Aug 11 15:34:02 hmmm khem is not there Aug 11 15:34:08 hi kergoth Aug 11 15:34:08 I wanted to ask him that: Aug 11 15:34:20 where to learn more about elf,binutils etc... Aug 11 15:34:31 and gcc Aug 11 15:34:39 learn what? Aug 11 15:34:59 basically the goal will be to be able to get a lot of infos Aug 11 15:35:12 from the binaries,intermediate source code etc... Aug 11 15:35:15 for instance Aug 11 15:35:28 you've got a runtime issue Aug 11 15:35:37 like an alignement issue Aug 11 15:35:53 and with the binutils you go to inspect the elf binary Aug 11 15:36:05 you find the initial gcc added code for loading the binary Aug 11 15:36:10 .ctrl0 or something like that Aug 11 15:36:14 you disassemble that Aug 11 15:36:17 http://en.wikipedia.org/wiki/Executable_and_Linkable_Format appears to give a fair bit of info, and links to a fair bit more ;) Aug 11 15:36:19 you see the issue Aug 11 15:36:20 ok Aug 11 15:36:25 thanks a lot Aug 11 15:36:37 I'll look at it when my internet connection will be faster Aug 11 15:36:59 I tried to download the 2.6.35 kernel....1k/s Aug 11 15:37:14 when I open something else than irc...the irc lags... Aug 11 15:37:23 and I get no more data from irc Aug 11 15:37:34 (far wifi+vpn) Aug 11 15:37:47 maybe with links.... Aug 11 15:37:55 Has anybody got ruby with socket support into work? My OE generated ruby.ipk does not have included it. Therefore rubygems is failing on setup. Aug 11 15:38:06 * GNUtoo|laptop tries wikipedia Aug 11 15:39:15 * kergoth keeps meaning to read Linkers and Loaders, but its low on his todo list Aug 11 15:40:24 ok Aug 11 15:40:57 of course, maybe it has something to do with the fact that there's 172 books on my 'to read' list, bit hard to prioritize that many :) Aug 11 15:41:00 * kergoth rolls eyes Aug 11 15:41:04 lol ok Aug 11 15:41:23 in my todo list for books there are only 2 books left Aug 11 15:41:42 the new rms biography and code 2.0 Aug 11 15:42:00 nice Aug 11 15:42:12 many of mine are just classic works in the cs arena that i never got around to Aug 11 15:42:54 ok Aug 11 15:42:55 peopleware, code complete, the dragon book, SAICP, etc Aug 11 15:43:03 oh god, and the art of computer programming Aug 11 15:43:12 i've never gotten more than like 40 pages into that before halting Aug 11 15:43:16 :) Aug 11 15:43:19 donald Aug 11 15:43:30 I've already read embedded linux primer... Aug 11 15:43:33 its interesting, but .. awfully dense Aug 11 15:43:42 + other books Aug 11 15:44:00 they cover a bit binutils Aug 11 15:44:32 time to go home Aug 11 15:44:43 ok Aug 11 15:46:18 anyone that knows python feel like taking a quick look at http://github.com/kergoth/bitbake/compare/master...logging and giving an opinion on whether this seems a worthwhile way to go? Aug 11 15:56:20 the dragon book rules. Aug 11 16:31:35 03Enrico Scholz  07org.openembedded.dev * r02b2067989 10openembedded.git/recipes/ncurses/ncurses_5.7.bb: Aug 11 16:31:35 ncurses: swapped installation of widec and narrowc headers Aug 11 16:31:35 Widec headers contain additional definitions and shall be installed Aug 11 16:31:35 last hence. To ensure compatibility of both headers a sanity check Aug 11 16:31:35 was added. Aug 11 16:31:36 Thanks to Martin Jansa for noticing this issue. Aug 11 16:31:37 Signed-off-by: Enrico Scholz Aug 11 17:05:26 hmm. Aug 11 17:06:04 when using the 'none' server in bitbake, i wonder if my logging patch should continue to let the server's messages be logged directly, or if they should instead go via the server's event queue, the way they are in the xml/rpc setup Aug 11 17:09:50 old 'computer related' movies are funny Aug 11 17:10:21 just watched 'the net' from 1995 with that '.345.200' IP addr Aug 11 17:10:58 That's hardly an old movie error. Aug 11 17:11:10 1995 - so old Aug 11 17:11:16 that was an error back then too :P Aug 11 17:11:27 Yeah, but the error is just an error rather than a datedness thing. Aug 11 17:12:30 * ensc|w wonders why hollywood does not have the 555.0.0.0/8 net Aug 11 17:12:52 (however you pack 555 into 8 bits...) Aug 11 17:12:55 Of course the 555 phone numbers are only invalid in the US :( Aug 11 17:12:57 they did not? Aug 11 17:13:36 ensc|w: they should try to get 555:0:0:0:0:0 one then Aug 11 17:20:44 hmm, the message indicating that importing cPickle failed and its falling back to pickle, which is slower.. should that be a warnings.warn(..., ImportWarning) rather than a bb.msg.note? Aug 11 17:22:02 is it possibe to auto remove the workidr? Aug 11 17:22:09 to save space Aug 11 17:22:18 INHERIT += "rm_work" Aug 11 17:22:52 btw why do we keep it? Aug 11 17:23:29 some of us likes to play with contents of it Aug 11 17:23:54 as example on gentoo we remove it and keepit only with --keep IIRC Aug 11 17:24:51 likely for debugging, i could see an argument for making it default, but there's no way to -= a variable :) Aug 11 17:24:53 in OE we do things different then gentoo does Aug 11 17:25:27 kergoth: INHERIT += "no_rm_work" with do_rm_work() { : } maybe? Aug 11 17:25:39 touch classes/rm_work.bbclass seems easier :) Aug 11 17:25:54 but either would do, yeah Aug 11 17:26:35 kergoth: or add USER-DISABLE-RM-WORK = "yes" and check for it in rm_work.bbclass Aug 11 17:26:53 INHIBIT_RM_WORK or something perhaps Aug 11 17:26:55 but on the other hand Aug 11 17:26:59 we have too many variables as is :) Aug 11 17:27:15 heh.. when I remind how simple rm_work.bbclass was when I wrote it... Aug 11 17:28:21 heh Aug 11 17:28:33 it's sad how it takes us years to implement something that we've known was a good idea Aug 11 17:28:40 and it turns out you can throw it together in like 2 days Aug 11 17:28:43 it makes me sad Aug 11 17:29:47 tell me something about it... Aug 11 17:29:55 heh Aug 11 17:30:03 I had to write bootstrapped cross compiler for ubuntu Aug 11 17:30:06 kergoth, that is a really common problem :) Aug 11 17:30:21 one gvim had ubuntu packaging files, second had openembedded recipes/classes Aug 11 17:31:07 of course, sometimes it takes me a while to work out the design issues and increase my knowledge so I can write the code in a few days Aug 11 17:31:34 Crofton: I guess its understandable, until you actually sit down and figure out what needs to be done, it may seem like it's much more work than it is.. or vice versa, but it seems like the former is more common for some reason Aug 11 17:31:36 * kergoth nods Aug 11 17:32:47 there is no way I could have done the work I did yesterday a couple of months ago in the same time Aug 11 17:54:54 hey khem, does --sysroot always work for gcc/etc for the versions we have still? Aug 11 17:55:04 * Tartarus wants to change from -isystem to --sysroot= Aug 11 18:59:26 Hi. I synced today, which caused gcc-4.3.3 to get rebuilt. Now when I try to build recipes, autoconf fails with: "Fatal error: Invalid -march= option: `armv7-a'". I'm using tune-coretexa8.inc. What could cause a build to fail in that way? Aug 11 19:01:42 Xobs, what recipe does that? Aug 11 19:02:16 It's dying on jpeg-8b-r0. Aug 11 19:17:19 hm did a git pull about an hour ago, now building mythtv and by sheer coincidence jpeg_8b just build fine for me a few minutes ago Aug 11 19:18:07 Alright. Perhaps I'll try doing a clean of binutils and gcc and see if that fixes it. Aug 11 19:18:43 i fyou have the time, I suggest doing it thorougly and remove all of tmp Aug 11 19:19:15 Sure. I've got time. Is that generally a good idea when pulling? Aug 11 19:22:19 When the toolchain changes, yes, it's best to blow away your tmp dir and start over again. Aug 11 19:24:05 Actually, when any critical dependency changes in a significant way, you need to forcibly rebuild the things that depend on it -- but it is sufficiently difficult to determine what is critical, and what depends upon it, that it is usually easier to blow away everything and start anew. I usually rename my tmp dir, and change Aug 11 19:25:15 my settings in local.conf to not inherit rm_work, so that I can compare build dirs to see what went wrong with the new state of the world... it's about 50/50 if things will build again after a toolchain change, in my experience... Aug 11 19:25:24 hey, would it be interesting to add support for localversion in linux recipes? Aug 11 19:25:44 if so i can produce a patch otherwise i'll just keep it in my own Aug 11 19:26:18 etrunko: The best thing with stuff like this is usually to just propose whatever patch you're thinking of. Aug 11 19:28:48 broonie: if you have CONFIG_LOCALVERSION_AUTO set, kernel configuration it will look in a file named localversion to append to kernel release Aug 11 19:28:52 there aren't many recipes using it atm Aug 11 19:51:32 kergoth, I looked at the wikipedia article...and looked at the arm specs for ELF....but that are specs....not binutils usage Aug 11 19:52:06 if you know how elf works, operating objdump/readelf/objcopy based on the man pages is not particularly difficult Aug 11 19:52:43 ok Aug 11 19:52:51 so I should read the manpages+specs? Aug 11 19:52:55 * kergoth guess he just isn't clear on what information you're looking for Aug 11 19:53:00 or is there a howto that skip the specs Aug 11 19:53:25 basically I'm looking at a mastering binutils like thing for debugging purposes Aug 11 19:54:23 I am watching Tron Aug 11 19:54:32 CGI effects are awesome ;D Aug 11 19:57:27 hrw, the original Tron? :) Aug 11 19:57:35 yes Aug 11 19:57:38 198x one Aug 11 19:57:56 hehe! rather dated, I think. But I recall quite enjoying that movie many years ago. Aug 11 19:58:41 I'm not against old movies, often they're better than the remakes Aug 11 19:59:44 altough I exchanged "looking at movies" with "computer things like oe,htcdream liberation,bug device things..." Aug 11 20:00:49 mwester: those old movies reminds how fun computers were years ago Aug 11 20:01:43 hrw: the new tron? :) Aug 11 20:01:57 buZz: no, 198x one Aug 11 20:02:03 ah k Aug 11 20:02:29 1982 widescreen edition Aug 11 20:04:47 khem: ping Aug 11 20:05:15 indeed it seem that computer were funier before.... Aug 11 20:05:23 like in the 60's Aug 11 20:06:04 they invented things like WYSIWYG editors,terminals using computer screens etc... Aug 11 20:06:39 graphical serial connected terminals... Aug 11 20:08:42 indeed Aug 11 20:32:04 khem: alignment traps are on armv4t and armv5te ONLY when it's compiled WITHOUT -ggdb3 in FULL_OPTIMIZATION (ie sane-toolchain-eglibc.inc), WITH -ggdb3 you're fine (ie angstrom-eglibc.inc), Now I have 3 builds (SHR with -Os, without ggdb3, ANG with -Os and ggdb3, ANG with -Os without ggdb3). libc.6.so and ld-2.12.1.so for all 3 variants are available here: http://build.shr-project.org/tests/jama/ Aug 11 20:46:57 03Martin Jansa  07org.openembedded.dev * r0dee8eb345 10openembedded.git/ (2 files in 2 dirs): Aug 11 20:46:57 eglibc: build with -ggdb3 as workaround for alignment traps Aug 11 20:46:57 * I've seen it only with eglibc-2.12 built with gcc-4.5 that's why I bumped PR only in 2.12 Aug 11 20:46:57 * can be removed as soon as gcc bug is fixed Aug 11 20:46:57 Signed-off-by: Martin Jansa Aug 11 20:52:38 Tartarus: yes sysroot should work for gcc 4.1 onwards and I dont think we care of about 3.4s Aug 11 20:53:28 JaMa: thanks I will see if I can spare some time Aug 11 20:58:40 JaMa: armv6/7 are safe? Aug 11 21:00:27 hrw: I have images for both, but no device to try it here Aug 11 21:00:40 hrw: probably not, but Angstrom is safe Aug 11 21:01:15 I can give you account on beagleboard (running ubuntu) Aug 11 21:04:16 I'm going to sleep soon.. so it's not worth it today and tomorrow I'll check with htcdream guys Aug 11 21:04:47 ok Aug 11 23:11:50 ant_: around ? Aug 11 23:38:18 03Andrea Adami  07org.openembedded.dev * rf502472a26 10openembedded.git/recipes/klibc/ (15 files in 2 dirs): Aug 11 23:38:18 klibc: remove old unused 1.5 and 1.5.15 recipes. Aug 11 23:38:18 * Keep 1.5.18 which builds vs. 2.6.2x and 2.6.3x kernels. Aug 11 23:38:18 * Reintroduce do_stage (hard to die!). Aug 11 23:38:18 * WIP in order to get finally rid of legacy staging Aug 11 23:38:19 * For the moment, refactor the recipe and bump PR. Aug 11 23:52:28 03Khem Raj  07org.openembedded.dev * r41f4059656 10openembedded.git/recipes/klibc/ (7 files in 2 dirs): Aug 11 23:52:28 klibc: Remove do_stage Aug 11 23:52:28 * Some cleanups Aug 11 23:52:28 Signed-off-by: Khem Raj Aug 11 23:52:40 03Khem Raj  07org.openembedded.dev * rb3f908c861 10openembedded.git/recipes/xorg-lib/libx11_1.3.2.bb: Aug 11 23:52:40 libx11_1.3.2.bb: Use Os in DEBUG_OPTIMIZATION for thumb Aug 11 23:52:40 * This is to avoid a bug in GCC 4.5 where it ICEs with Aug 11 23:52:40 -O1 -mthumb -fno-omit-frame-pointer Aug 11 23:52:40 Signed-off-by: Khem Raj Aug 11 23:56:54 khem: thx, I'll fire a rebuild now Aug 11 23:57:32 ant_: sure Aug 11 23:57:38 hope not having crossed too much your tests ;) Aug 11 23:57:54 Dont know :) Aug 11 23:58:07 but klibc and kexec-tools-static built fine here now Aug 11 23:58:10 so should be fine Aug 11 23:58:23 same here before pull. now rebuilding gcc Aug 11 23:59:16 khem, one thing more. I've noticed gzip is not listed in RDEPENDS Aug 11 23:59:30 gzip for what Aug 11 23:59:40 the utils Aug 12 00:00:16 FILES_${KLIBC_UTILS_PKGNAME}-gzip = "${base_bindir}/gzip ${base_bindir}/gunzip ${base_bindir}/zcat" Aug 12 00:01:42 hmm it packages them ? Aug 12 00:01:51 why not let gzip package them Aug 12 00:02:03 igzip is listed in FILES Aug 12 00:02:21 and has pkgname Aug 12 00:02:45 ok but my question is if this gzip is built external to klibc ? Aug 12 00:02:56 or is it built from klibc bundle Aug 12 00:03:05 it seems budled, mom Aug 12 00:03:43 then you dont need dep Aug 12 00:03:46 yep, in the src Aug 12 00:04:15 is it static? Aug 12 00:04:21 hm Aug 12 00:04:50 I never needed this gzip so have never tried to install the shared binary Aug 12 00:06:03 i suppose the normal version needs klibc.so Aug 12 00:06:43 to run Aug 12 00:06:54 ok Aug 12 00:07:32 but hey, it's like this since the beginning..I was wondering whether there is a reason or not... Aug 12 00:09:34 anyway, many thanks Aug 12 00:09:39 good night **** ENDING LOGGING AT Thu Aug 12 02:59:57 2010