**** BEGIN LOGGING AT Sat Jan 16 02:59:56 2010 Jan 16 04:18:17 man, why does bitbake -n take so long Jan 16 04:29:00 the initial portion happens whether you do -n or not — it has to parse the metadata (or read the cache), then it has to generate its runqueue Jan 16 04:29:09 bitbake master has a performance improvement for the runqueue Jan 16 04:29:15 pretty sure that one 's on 1.10 too Jan 16 04:30:52 Tartarus: 52127bcebe651cc529118804ff05678749c0d003 on 1.10 Jan 16 04:32:16 yeah, still using 1.8 Jan 16 04:32:33 it's not the parse that takes so long, it's that every task seems to take a good second or two to "run" Jan 16 04:33:32 you should be able to cherry pick 52127bce, though. Jan 16 04:33:39 huh, thats odd.. Jan 16 04:33:55 it could be the event handlers. Jan 16 04:34:01 it may be firing those, not sure Jan 16 04:34:05 would have to double check the code Jan 16 08:32:45 RP: are you there by any chance? Jan 16 08:33:41 RP: I have a little problem with stage populating (not legacy) Jan 16 08:48:49 03Martin Jansa 07org.openembedded.dev * r92d0badcd3 10openembedded.git/recipes/xserver-kdrive-common/xserver-nodm-init.bb: Jan 16 08:48:49 xserver-nodm-init: uncomment DESCRIPTION, commented out by accident Jan 16 08:48:49 * Extra entry to my commit policy: be double carefull after 2AM, sorry Jan 16 08:48:49 Signed-off-by: Martin Jansa Jan 16 09:02:33 I'm building the latest console-image (angstrom) and it fails. The error message I get is: cp: target `/home/jovox/oe2/tmp/staging/x86-angstrom-linux/kernel/include/asm-x86/' is not a directory Jan 16 09:13:02 After manually creating it I passed the task... Jan 16 09:13:42 jovox: who was trying to copy things there? Jan 16 09:14:46 zecke, http://pastebin.com/m66fc9727 Jan 16 09:17:36 zecke: Making any sense? Jan 16 09:19:03 jovox: i'm no expert in this area... there is nothing fancy besides you having a quite recent kernel. *shrug* Jan 16 09:19:23 Yes, it's the dev branch Jan 16 09:19:46 The stable 2009 looks pretty outdated? Jan 16 09:20:38 no idea, I'm always on dev Jan 16 09:20:50 ok Jan 16 09:21:53 Anyway, heading off to IKEA :-) Hopefully the build is ready when I get back to office Jan 16 09:22:15 jovox: you might want to check classes/kernel.bbclass and see why it is failing to copy (besides the dir not being present) Jan 16 09:22:52 e.g. according to your message it does exist but is not a directory Jan 16 09:23:58 it did not exist so I created it Jan 16 09:24:43 I will have a look at the kernel.bbclass Jan 16 09:29:34 hm, when I 'inherit native', then ${WORKDIR}/sysroot-destdir doesn't get filled Jan 16 09:29:35 :/ Jan 16 11:30:37 CoreDump|home: ping.. check bot logging ability please :) Jan 16 11:40:52 hello Jan 16 11:41:16 ipkg update on my ipaq gives 404's Jan 16 11:41:33 angstrom-distribution.org resolves to 85.214.40.226 Jan 16 11:41:45 in the browser on the pc it works Jan 16 11:43:27 ah Jan 16 11:43:30 host file issue Jan 16 12:37:01 hm.. glibc 2.9 build failed Jan 16 12:38:48 http://tinderbox.openembedded.net/public/logs/task/4444948.txt Jan 16 12:42:37 looks like limits.h again.. Jan 16 12:50:40 hehe.. rebuild glibc-initial solves that Jan 16 12:50:46 again and again Jan 16 13:08:42 NOTE: Multiple libraries (libnss_nis.so.2, libnss_hesiod.so.2, libnss_nisplus.so.2) found, but LEAD_SONAME 'libc.so' doesn't match any of them Jan 16 13:08:46 JFYI Jan 16 13:43:39 may i know how to fix this Jan 16 13:43:46 /usr/lib/python2.6/site-packages/bb/COW.py:29: DeprecationWarning: the sets module is deprecated Jan 16 13:44:10 import types, sets Jan 16 13:44:23 ERROR: Please set the 'CACHE' variable. Jan 16 13:45:19 i got this error while running $ bitbake opie-image Jan 16 13:47:19 03Frans Meulenbroeks 07org.openembedded.dev * r277b2e0bd1 10openembedded.git/recipes/mythtv/myththemes_0.22.bb: myththemes: 0.22 fix SRC_URI and ${S} Jan 16 13:48:00 manz: i think your settings are not ok Jan 16 13:48:37 guess this is BBPATH not being set properly Jan 16 13:52:09 is thir anything rong with python version Jan 16 13:53:41 i don't think so Jan 16 13:55:10 manz: if i bake something without setting BBPATH I get ERROR: Please set the 'PERSISTENT_DIR' or 'CACHE' variable. Jan 16 13:56:04 whihave you set BBPATH Jan 16 13:56:09 ? Jan 16 13:56:11 yes that i forgotten to run the export BBPATH Jan 16 13:56:36 ok guess that will work, gl Jan 16 13:56:41 File "", line 18, in check_sanity Jan 16 13:56:50 NameError: global name 'os' is not defined Jan 16 13:57:01 have you set MACHINE ? Jan 16 13:57:10 mx27ads Jan 16 13:57:15 did you make a local.conf ? Jan 16 13:57:39 i copied the sample one from openembedded and edited it Jan 16 13:57:44 what DISTRO in local.conf? Jan 16 13:58:15 TARGET_OS = "linux-uclibc" Jan 16 13:58:36 DISTRO is commented Jan 16 13:58:45 hm, where did you get this from ? Jan 16 13:58:48 i dont know which to use Jan 16 13:58:52 # DISTRO = "angstrom-2008.1" Jan 16 13:58:56 and what do you want to do Jan 16 13:59:20 i want to run poky on imx27 Jan 16 13:59:24 imx27ads Jan 16 13:59:39 which distro do i need to choose Jan 16 14:00:08 sorry i m bit new to these stuff. Jan 16 14:00:36 no idea what distro to use for poky, i have no poky knowledge Jan 16 14:00:46 what web page are you following ? Jan 16 14:00:51 http://wiki.openembedded.net/index.php/Getting_Started Jan 16 14:01:27 i have asked poky people they said u should be able to run using oe Jan 16 14:02:05 manz1: poky doesnt support imx27, but as OE does you can extract rules and things from OE to add support imx27 to poky Jan 16 14:03:00 manz: i think this is a better start: http://www.pokylinux.org/support/ Jan 16 14:03:36 e.g. this has a lot of poky specific info also on adding a new machine ets: http://www.pokylinux.org/doc/poky-handbook.html Jan 16 14:03:45 they said poky does not support imx27 Jan 16 14:03:55 they have support only for imx31 Jan 16 14:04:23 which means you might need to look at this: http://www.pokylinux.org/doc/poky-handbook.html#platdev-newmachine Jan 16 14:04:58 afk for a while Jan 16 14:05:08 DISTRO = "angstrom-2008.1" Jan 16 14:05:19 for that also it is giving same error. Jan 16 14:07:39 hi,I've some issues with gdbserver->gdb that doesn't seem to work anymore: Jan 16 14:07:50 *** glibc detected *** python: free(): invalid pointer: 0x00225fc0 *** <- that is the bug Jan 16 14:08:04 Anyhow thanks for your time eFfeM_home Jan 16 14:08:29 http://pastebin.com/m6f9fbd11 Jan 16 14:29:02 libX11 build failed Jan 16 14:29:12 manz: your bitbake is too old Jan 16 14:29:19 1.8.18 at least Jan 16 14:30:37 03Rolf Leggewie 07org.openembedded.dev * r6f98dd8c18 10openembedded.git/recipes/gimp/ (gimp_2.2.10.bb gimp_2.3.7.bb): gimp: move two old and unfetchable recipes to obsolete directory Jan 16 14:30:38 03Rolf Leggewie 07org.openembedded.dev * r92037c99c1 10openembedded.git/recipes/didiwiki/ (didiwiki.inc didiwiki_0.2.bb didiwiki_0.4.bb): didiwiki: move oldest, unfetchable recipes to obsolete/ and selfhost sources for the rest Jan 16 14:30:39 03Rolf Leggewie 07org.openembedded.dev * r96db6269a5 10openembedded.git/classes/xfce.bbclass: classes/xfce.bbclass: fix SRC_URI Jan 16 14:30:40 03Rolf Leggewie 07org.openembedded.dev * rcc13f5b577 10openembedded.git/recipes/xmame/ (4 files in 2 dirs): xmame: move oldest and unfetchable entries to obsolete/ Jan 16 14:30:43 03Rolf Leggewie 07org.openembedded.dev * r83f66a32dc 10openembedded.git/recipes/zd1211/ (6 files in 3 dirs): zd1211: move oldest and unfetchable entries to obsolete/ Jan 16 14:31:23 03Frans Meulenbroeks 07org.openembedded.dev * r751cbac0d9 10openembedded.git/conf/checksums.ini: conf/checksums.ini: moved one item according to the sorting rules Jan 16 14:31:24 03Frans Meulenbroeks 07org.openembedded.dev * r13e28c208e 10openembedded.git/recipes/mythtv/mythplugins_0.22.bb: mythplugins: fixed SRC_URI and ${S} for 0.22 Jan 16 14:31:26 03Frans Meulenbroeks 07org.openembedded.dev * rf64741b096 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Jan 16 14:37:28 [havoc@Nosgoth ~]$ git clone git://git.openembedded.org/openembedded Jan 16 14:37:29 Initialized empty Git repository in /home/havoc/openembedded/.git/ Jan 16 14:37:29 git.openembedded.org[0: 140.211.169.165]: errno=Connection refused Jan 16 14:37:29 fatal: unable to connect a socket (Connection refused) Jan 16 14:37:41 ? Jan 16 14:38:26 B_Lizzard: me too Jan 16 14:39:01 try to put off firewalls and try Jan 16 14:39:01 Probably an temporary issue. Jan 16 14:39:17 may be idk. Jan 16 14:39:18 I tried the "firewall challenged" command Jan 16 14:39:29 Doesn't quite work either Jan 16 14:39:38 ok Jan 16 14:40:07 oe tree works for me.. but bitbake tree refused and it worked about 3 mins ago Jan 16 14:46:35 Is the dropbear ssh server broken in the dev branch? I get segfault after "Creating Dropbear SSH server host key" Jan 16 14:47:51 Now I get a "the remote end hung up unexpectedly" Jan 16 14:53:44 http://tinderbox.openembedded.net/public/logs/task/4450291.txt Jan 16 14:54:24 :) Jan 16 14:54:25 libx11 fails because of some groff issue as I understand Jan 16 14:57:04 i will check it laibsch Jan 16 14:59:31 /usr/bin/groff -Thtml -e -t -ms -dxV="libX11 1.3.2" -I. ../../specs/macros.t libX11.ms 2> index.libX11.html.raw; echo $? -> 3 Jan 16 15:00:45 /usr/bin/groff: can't find `DESC' file Jan 16 15:00:47 /usr/bin/groff:fatal error: invalid device `html' Jan 16 15:00:50 hmm Jan 16 15:08:31 ah.. host system problem Jan 16 15:08:35 no groff-extra Jan 16 15:27:05 is it a known issue avahi and dropbear doesnt seem to start in the latest version? Jan 16 15:40:46 git pu Jan 16 15:55:46 JaMa: What md5sum do you get for ffalarms-0.2.2.tar.gz? Jan 16 16:01:30 JaMa: Nevermind Jan 16 16:01:59 jovox: did they start previously? Jan 16 16:14:16 adding "--libdir=/usr/lib/nspr --includedir=/usr/include/nspr" to EXTRA_OECONF causes non-legacy staging to skip these directories Jan 16 16:14:20 can I prevent that somehow? Jan 16 16:22:58 how do I extract "15" from a variable value of "2.15" in OE. IOW, get the last two digits. Jan 16 16:23:01 ? Jan 16 16:23:46 split function Jan 16 16:37:01 Laibsch: foo.split(".")[1] would do Jan 16 16:37:15 yes, I'm experimenting with that now Jan 16 16:37:22 hi Laibsch Jan 16 16:37:28 but so far I'm getting an error Jan 16 16:38:15 Laibsch: I'm fixing minimal-uclibc build of ltrace. Any uclibc-hacker around? Jan 16 16:39:07 kergoth: maybe you can take a quick look, I can't find the error and double-checked for unbalanced parantheses Jan 16 16:39:10 ARCHIVE = "${@base_less_or_equal({@bb.data.getVar('PV',d,1).split('.')[1]}, '15', 'archive', '', d)}" Jan 16 16:39:39 if PV ends in a value less or equal than 15 Jan 16 16:39:48 then ARCHIVE should be "archive" Jan 16 16:39:53 and "" otherwise Jan 16 16:39:57 | gdkscreen-x11.c:43:37: error: X11/extensions/Xinerama.h: No such file or directory Jan 16 16:40:06 hi Jay7 Jan 16 16:40:26 http://tinderbox.openembedded.net/public/logs/task/4450967.txt Jan 16 16:40:49 Laibsch: uh? Jan 16 16:40:49 l Jan 16 16:41:03 Laibsch: you have a pretty obvious useless, syntactically invalid bit of python there Jan 16 16:41:09 Laibsch: equal({@bb.data Jan 16 16:41:14 looks like a copy/paste error Jan 16 16:41:25 remnant of a previous ${@, i assume Jan 16 16:41:36 * Laibsch doesn't know much about python Jan 16 16:41:44 ${@} is OE, not python Jan 16 16:41:45 remove the second @? Jan 16 16:41:47 it tells oe that anything inside it is python Jan 16 16:41:49 no Jan 16 16:41:52 remove {@ Jan 16 16:41:58 kergoth: base_version_less_or_equal breaks if the var contains '-' or '+' apparently Jan 16 16:42:04 lovely Jan 16 16:42:07 using bb.vercmp Jan 16 16:42:11 ant__: hey Jan 16 16:43:18 I tried to use bb.utils.vercmp_part for similar things Jan 16 16:44:32 anybody have any ideas about Xinerama.h absence? Jan 16 16:45:17 Jay7: check xorg/xserver in zaurus-2.6.inc Jan 16 16:45:45 and if it's xorg, ask JaMa ! Jan 16 16:45:46 ant__: I've pulled today Jan 16 16:46:54 PREFERRED_PROVIDER_virtual/xserver_akita = "xserver-kdrive-1300" Jan 16 16:46:55 XSERVER_akita = "xserver-kdrive-1300-fbdev" Jan 16 16:47:28 ah, right, xorg only for spitz atm Jan 16 16:48:19 hmm Jan 16 16:49:15 I suspect lost xineramaproto in dependencies Jan 16 16:50:00 even xinerama Jan 16 16:50:23 iirc Stanislav tested kdrive-1300 Jan 16 16:56:08 good morning Jan 16 16:57:04 Laibsch: the opie-taskbar problem is actually a problem in libqpe Jan 16 16:58:27 khem: hello, I new uclibc issue Jan 16 16:58:29 http://tinderbox.openembedded.net/public/logs/task/4450408.txt Jan 16 16:58:32 good morning, khem Jan 16 16:58:32 there is a constructor/destructor pair for LnkProperties which is conditionalized under QTOPIA_INTERNAL_FSLP Jan 16 16:58:44 khem: do you have a fix handy or is it complicated? Jan 16 16:58:53 but if I copy the definition in trace.c to events.c it compiles... Jan 16 16:59:10 this define is defined in lnkproperties.cpp but not in moc_lnkproperties.cpp I am not qpe expert but this seems wrong Jan 16 17:00:05 Laibsch: a fix would be to add #define QTOPIA_INTERNAL_FSLP at top of moc_lnkproperties.cpp in libqpe/library sources Jan 16 17:00:10 must be different headers ? sys/ptrace.h? Jan 16 17:00:56 but there seems to be a deliberate reason not to define it and if they were using it in someway it was a wrong assumption Jan 16 17:01:00 ant__: hello Jan 16 17:01:07 hi ;) Jan 16 17:01:28 me and Laisch are jumping on you like sharks.... Jan 16 17:01:36 sorry and TIA Jan 16 17:01:39 ah ltrace Jan 16 17:02:33 trace.c has a guard, lacking in events.c Jan 16 17:03:01 but in glibc and eglibc no issues... Jan 16 17:03:19 khem: Maybe you can make a comment in the bug tracker and we will wait for bluelightning's reaction Jan 16 17:03:22 yes I think its a uclibc problem Jan 16 17:03:42 it does not seem to have synced its version of ptrace.h Jan 16 17:04:00 If it's not obvious how to fix this properly, he should probably have a look at how he wants it for the future Jan 16 17:04:13 khem: the first reaction was to copy and paste #ifndef PTRACE_EVENT_FORK... Jan 16 17:04:21 but is obviously just an hack Jan 16 17:05:03 Laibsch: yeah it was wrong to start with but we got by it because ld did not merge symbol attributes until now for weak symbols but not the problem surfaces Jan 16 17:05:18 s/not/now Jan 16 17:05:27 kergoth: It's still not working. The error I get is Jan 16 17:05:27 NOTE: :float() argument must be a string or a number while evaluating: Jan 16 17:05:27 ${@base_less_or_equal(2, 15, 'archive', '', d)} Jan 16 17:05:41 I tried quoting and unquoting at various places Jan 16 17:05:55 but 2 and 15 do look like numbers to me Jan 16 17:06:03 python must think otherwise ;-) Jan 16 17:06:23 I'd suggest looking at what base_less_or_equal is doing Jan 16 17:06:26 haven't used it myself Jan 16 17:06:30 Laibsch: which var are you evaluating? how does th estring look like? Jan 16 17:06:44 just like above Jan 16 17:06:51 ${@base_less_or_equal(2, 15, 'archive', '', d)} Jan 16 17:07:02 that's what it's choking on Jan 16 17:07:15 you should give a var and not a value Jan 16 17:07:24 2 is bad Jan 16 17:07:37 ant__: you're not marching in the right direction Jan 16 17:07:47 the evaluations before that are obviously fine Jan 16 17:07:49 Laibsch: see linux-kexecboot.inc Jan 16 17:07:54 see my notes about this Jan 16 17:07:54 OK Jan 16 17:08:17 ah, my 'fix' just happens to works Jan 16 17:08:21 if this is about + or -, that's got nothing to do with it Jan 16 17:08:26 I'm absolutely python newb Jan 16 17:08:40 was my first failure Jan 16 17:08:49 setting the string inline Jan 16 17:08:59 JaMa: enlightened me :) Jan 16 17:09:51 Laibsch: first argument to base_less_or_equal should be a variable Jan 16 17:10:06 ant__: there is no comment except "do our own thing until the other thing is fixed" Jan 16 17:10:13 what's broken isn't explained at all Jan 16 17:10:15 Laibsch: see the usage in distro/angstrom-2008.1.conf Jan 16 17:10:36 I made a bit of buzz with RP Jan 16 17:10:44 khem: I'm sure the function would be fine with comparing numbers Jan 16 17:10:49 but since both of you insist Jan 16 17:10:52 RL_VER = "${@bb.data.getVar('PV',d,1).split('.')[0]}" Jan 16 17:10:58 that was the call Jan 16 17:11:30 Laibsch: well thats a different expectation :) Jan 16 17:11:42 not really Jan 16 17:11:43 this function says give me a variable and I will get the value Jan 16 17:11:48 myself Jan 16 17:11:56 that evaluates to ${@base_less_or_equal(2, 15, 'archive', '', d)} Jan 16 17:12:00 it does getvar Jan 16 17:12:07 now you could argue in many ways but thats how it is now Jan 16 17:12:09 at least that is what bitbake tells me Jan 16 17:12:12 and that fails Jan 16 17:12:28 Simpler question is better Jan 16 17:12:37 so doing @base_less_or_equal(2, 15, 'archive', '', d) is definitely not adhering to functions expected inputs Jan 16 17:12:42 so, I post the problem when parts have been evaluated Jan 16 17:13:04 khem: I never did that function. Jan 16 17:13:13 its like a c function foo (char*, int) and passing foo (char**, int) to it Jan 16 17:13:14 But bitbake'ing the recipe in question gives that error Jan 16 17:13:31 Laibsch: yes it needs to be fixed then Jan 16 17:13:37 the recipe Jan 16 17:14:00 http://paste.debian.net/56864/, just to end the misunderstandings Jan 16 17:14:07 well, obviously Jan 16 17:14:20 RL_VER is obviously just a variable for testing what breaks Jan 16 17:14:26 divide and conquer Jan 16 17:14:52 which is the same reason I gave 2 and not "${@bb.data.getVar('PV',d,1).split('.')[0]}" Jan 16 17:14:58 just easier to understand Jan 16 17:15:05 easier to understand -> easier to fix Jan 16 17:15:17 error message is misleading I agree Jan 16 17:15:31 so, now that we needlessly backtracked, where is the real problem? Jan 16 17:16:28 bb chokes on Jan 16 17:16:31 ARCHIVE = "${@base_less_or_equal(${RL_VER}, 15, 'archive', '', d)}" Jan 16 17:16:48 I tried "15" and '15', it's all the same Jan 16 17:17:12 and the base_less_or_equal seems to always return an unquoted value anyway Jan 16 17:17:38 but apparently that value is not interpreted as a string or number Jan 16 17:18:05 Laibsch: I think the problem is that float is passed ('') becuase getVar for '2' will return null from metadata Jan 16 17:18:22 and thats why the python error sounds reasonable Jan 16 17:18:49 RL_VER would have type float? Jan 16 17:18:57 how would I change that? Jan 16 17:19:46 but then it should have give valueError not TypeError Jan 16 17:19:46 wasn't it that python converts those types automatically (unlike C)? Jan 16 17:21:00 Laibsch: see the syntax in linux-kexecboot.inc Jan 16 17:21:07 perhaps works Jan 16 17:21:09 ant__: I did Jan 16 17:21:11 um Jan 16 17:21:15 no, thank you Jan 16 17:21:19 RD_COMPRESSION = '${@kernel_version_less_or_equal("PV", "2.6.30", "", "CONFIG_RD_LZMA=y", d)}' Jan 16 17:21:26 see ' and " Jan 16 17:21:28 I've similar things working just fine elsewhere Jan 16 17:21:51 filip.zyzniewski@tefnet ~/pda/oe/build_dir $ bitbake -c stage -b ../openembedded/recipes/nss/nss-native_3.12.5.bb Jan 16 17:21:55 NOTE: Preparing runqueue Jan 16 17:21:55 ARCHIVE = "${@base_less_or_equal(${RL_VER}, 15, 'archive', '', d)}" Jan 16 17:21:55 wtf? Jan 16 17:21:58 ERROR: Task do_stage does not exist for target nss-native Jan 16 17:22:00 filip.zyzniewski@tefnet ~/pda/oe/build_dir $ grep do_stage ../openembedded/recipes/nss/nss-native_3.12.5.bb Jan 16 17:22:03 do_stage() { Jan 16 17:22:39 Laibsch: swap the quotes / double q Jan 16 17:23:10 bbl Jan 16 17:23:34 asking kergoth as python guru. Given PV is 2.11, why does the following evaluation fail (see http://paste.debian.net/56864/) when evaluating the ARCHIVE variable? Jan 16 17:23:34 RL_VER = "${@bb.data.getVar('PV',d,1).split('.')[0]}" Jan 16 17:23:34 ARCHIVE = "${@base_less_or_equal(${RL_VER}, 15, 'archive', '', d)}" Jan 16 17:26:10 Laibsch: RL_VER is a shell variable and when you pass ${RL_VER} it gets expanded Jan 16 17:26:44 so essentially what base_less_or_equal sees as first argument is value stored in RL_VER Jan 16 17:27:13 which it again tries to dereference from metadata by doing getVar Jan 16 17:27:49 hi Jan 16 17:33:26 ant__: RD_COMPRESSION = '${@kernel_version_less_or_equal("PV", "2.6.30", "", "CONFIG_RD_LZMA=y", d)}' Jan 16 17:33:31 how does that work Jan 16 17:33:39 2.6.30 is not a valid number Jan 16 17:34:04 re Jan 16 17:34:15 khem: it's using bb.utils.vercmp_part Jan 16 17:34:22 and not bb.vercmp Jan 16 17:34:54 I still don't know why there are two vercmp Jan 16 17:34:56 in bitbake Jan 16 17:35:24 oh I was looking at base_less_or_equal :) Jan 16 17:35:32 that's broken :) Jan 16 17:35:32 not at kernel_version_less_or_equal Jan 16 17:35:55 sorry ,was about difference bwtween shell and python variable Jan 16 17:36:46 03Rolf Leggewie 07org.openembedded.dev * r6bff7a52ea 10openembedded.git/conf/checksums.ini: checksums.ini: update checksum URLs for change in XFCE source hosting Jan 16 17:37:09 woglinde_: hi Jan 16 17:39:15 khem: building minimal-uclibc for armv5te, apart ltrace , e2fsprogs Jan 16 17:39:18 gen_uuid.c:(.text+0x70): undefined reference to `__aeabi_read_tp' Jan 16 17:40:18 ant__: ltrace builds ok for mips uclibc Jan 16 17:40:25 ant__: I think its arm specific Jan 16 17:40:29 kernel headers? Jan 16 17:40:34 yes Jan 16 17:41:53 ant__: __aeabi_read_tp ? TLS Jan 16 17:42:00 wont work Jan 16 17:42:07 with uclibc Jan 16 17:42:14 unless you are using uclibc-nptl Jan 16 17:42:32 let me see how was configured Jan 16 17:42:56 see if gcc was specifically set to --disable-tls Jan 16 17:43:20 I assume you are using uclibc 0.9.30.1 Jan 16 17:43:50 yes Jan 16 17:44:11 my guess is that there is some TLS usage in e2progfs Jan 16 17:44:32 and gcc compiles it fine because it is configured to accept __thread Jan 16 17:44:32 hm? Jan 16 17:44:35 checking for thread local storage (TLS) class... __thread Jan 16 17:44:39 didnt stumple over it Jan 16 17:44:46 there you go Jan 16 17:45:09 hm false tls macro again? Jan 16 17:45:09 ant__: did the above test succeed ? Jan 16 17:45:35 woglinde_: seems like the case Jan 16 17:45:44 Building uuidd by defaul Jan 16 17:45:55 it must be doing compile time test only Jan 16 17:52:53 khem: good catch, right on the money Jan 16 17:53:16 now I wonder what problems will rear their heads when eliminating that unnecessary helper var Jan 16 17:57:17 kind of the same problems as before Jan 16 18:09:52 Laibsch: you could just use cmp () standard python function Jan 16 18:10:02 you have the two numbers already Jan 16 18:12:50 filip: you want "-c populate_staging", not "-c stage" Jan 16 18:12:57 do_stage() is a function, but it isn't a task Jan 16 18:14:26 he pb Jan 16 18:23:58 hm, trying to build u-boot but get: fatal: Not a valid object name 1 in fetch Jan 16 18:25:05 apparently http://sources.openembedded.org//git_www.denx.de.git.u-boot.git_1.tar.gz Jan 16 18:25:05 does not exist Jan 16 18:27:11 Laibsch: something like this ${@['archive', ''][cmp(${RL_VER}, 15)]} Jan 16 18:27:45 Laibsch: might work, you have to check which condition should set archive which not Jan 16 18:27:50 should really use d.getVar() rather than ${} to get RL_VER Jan 16 18:28:15 kergoth: its a shell variable Jan 16 18:28:24 and? Jan 16 18:28:38 you can't get to a shell variable from a python snippet, only bitbake variables Jan 16 18:28:45 so i really don't know what you're hoping to accomplish Jan 16 18:28:51 yeah another way is to setVar is first Jan 16 18:28:55 it Jan 16 18:29:17 well he wants to compare major version of a package Jan 16 18:29:31 there is no utility function directly to do that Jan 16 18:30:13 so he gets the PV and munges it to extract MAJOR_VERSION and now needs to compare it to a given number for decision Jan 16 18:32:44 may be in bitbake syntax every variable declared should become a python var and added to metadata as a semantic thing Jan 16 18:33:15 then we can avoid playing the shell var and python var games Jan 16 18:34:43 re Jan 16 18:41:53 ${@['archive', ''][cmp(bb.data.getVar('PV',d,1).split('.')[0], 15)]} would be better Jan 16 18:43:31 right kergoth :) Jan 16 18:45:43 does someone remember option to stop shared library loading at randomized memory Jan 16 18:46:35 I need to debug this crash and every now and then libc is relocated to different address Jan 16 18:46:39 all my breakpoints are useless thereafter Jan 16 18:47:43 got it /proc/sys/kernel/randomize_va_space Jan 16 18:47:54 set this to 0 Jan 16 18:52:06 khem: comparison is not sufficient Jan 16 18:52:23 I need <= Jan 16 18:52:49 RL_VER is just a variable I use now for breaking things down Jan 16 18:52:59 to see where bitbake/python bombs out Jan 16 18:53:11 RL_VER is LAIBSCH_VER Jan 16 18:54:23 Laibsch: you can compare a number greater <= 15 is same as < 16 isnt it Jan 16 18:54:37