**** BEGIN LOGGING AT Sun Apr 19 02:59:57 2009 Apr 19 03:26:36 Hm. If I try to install openjdk-langtools-native, bitbake complains that there's no such package, despite my having built it. Apr 19 03:26:53 Any suggestions? I need native Java tools. :c Apr 19 03:36:37 Wait. Hm. Apr 19 03:36:59 The "-native" convention is for host utils, not target utils? Ah. Apr 19 03:37:20 So what do I build to give me javac, javah, etc. on the target? Apr 19 03:58:26 Or is there no such recipe? Apr 19 04:01:33 I mean, I hate to be a bother, but if there's no such recipe for target Java compiler, I'm gonna have to port one by hand, and that would make everybody sad all around. :C Apr 19 04:57:11 Oh, sweet. There's a jikes bb. Wonder how I missed that before. Sorry for being silly. Apr 19 07:13:06 morning Apr 19 07:13:35 what software can I use to view openstreet maps? Apr 19 07:14:04 I have no GPS but I will just view and search such maps offline Apr 19 07:16:11 both tangogps and navit also work without a gps Apr 19 07:18:16 valhalla: thanks :) Apr 19 08:31:30 from where i can get latest dss drivers/pacthes Apr 19 09:46:05 so guys. I have a valid /etc/localtime, I also tried setting the TZ in /etc/defaults/rcS, and regardless I'm still in UTC... any pointer on how to change the angstrom's timezone? Apr 19 10:16:23 nbd 2.8.7 is borken in stable/2009 . Upstream version 2.9.11 fixes the problem. Pushed the recipe and checksum.ini patch into git://repo.or.cz/openembedded/mini2440.git head "for-upstream" Apr 19 10:17:02 tinc 1.0.2 is borken in stable/2009 . Upstream version 1.0.9 fixes the problem. Pushed the recipe and checksum.ini patch into git://repo.or.cz/openembedded/mini2440.git head "for-upstream" Apr 19 10:17:42 directfb 1.2.7 is borken in stable/2009 on ARM (runtime). Upstream version 1.3.0 fixes the problem. Pushed the recipe and checksum.ini patch into git://repo.or.cz/openembedded/mini2440.git head "for-upstream" Apr 19 11:37:47 03Koen Kooi  07org.openembedded.dev * ra6f5ece4e7 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Apr 19 11:37:57 03Koen Kooi  07org.openembedded.dev * rf7915fe7e9 10openembedded.git/conf/distro/include/angstrom.inc: angstrom: start removing FEED_ARCH Apr 19 11:56:41 hi Apr 19 13:40:02 anyone here know uboot ? Apr 19 13:40:07 or can direct me to a channel that does Apr 19 13:42:57 rajkohaxor: #u-boot Apr 19 13:47:46 i use omap3evm with tomi's tree, i get the message: "omapfb omapfb: can't get display 1" and i dont see anything on lcd, any suggestions why this happens? Apr 19 14:50:47 i am using 2.6.29 with Open Embedded on a evm and am not able to get display working it i get the messages omapfb omapfb: can't get display 1 and dont see any OMAP DSS interrupts at all Apr 19 15:36:28 woglinde: I had to do the same thing to libiconv as you had me do to the expat. I'm still working on the build so there may be more. Should I give anyone a list of packages that need the ARM_INSTRUCTION_SET = "arm" Apr 19 15:36:42 woglinde: or is there no need? Apr 19 15:39:27 a lot of stuff seems to break in thumb. I'm currently fighting with libcurl on stable. crashes systematicaly here Apr 19 15:39:30 could someone shows me his BBMASK ? Apr 19 15:46:12 BusError: I'm starting to think uClibc is not worth the hassle. The only reason I wanted to use it was to reduce our rootfs image size so programming the board was quicker. We have tons of flash so flash space is not an issue. Apr 19 15:46:58 I use glibc here, and there are still problems. I think the interworking is a bit fragile Apr 19 15:49:19 svolpe, note that some recipe use ancient code, and the more recent upstream version works. I've been adding them as I go along Apr 19 15:52:35 BusError: When I use glibc I have not had any issues, I'm using AT91SAM9260 which is set up in OE using thumb. Although I'm using it only a web server so I'm not using a lot of packages. Apr 19 15:59:40 here libcurl dies in curl_easy_perform, and I have no idea why. no stack in gdb Apr 19 15:59:59 oh and with gdbserver not working, that doesn't help :/ Apr 19 16:19:50 svolpe depends Apr 19 16:20:02 svople mostly its a workaorund for now Apr 19 16:20:19 but weh should find the root of the errors Apr 19 16:21:40 svolpe when you use glibc libiconv is provided from glibc Apr 19 16:22:02 and glibc for now has the instruction set to arm too Apr 19 16:23:05 03Jan Lübbe  07fso/milestone5.5 * r192099546a 10openembedded.git/ (5 files in 5 dirs): frameworkd: bump frameworkd and drop applied patch Apr 19 16:29:13 woglinde, you work on stable/2009 right ? Apr 19 16:30:04 I got a few recipes for packages that are borken in stable, and work with a more recent upstream version -- git://repo.or.cz/openembedded/mini2440.git head "for-upstream" Apr 19 16:31:04 BusError not really Apr 19 17:00:45 jo kergoth Apr 19 17:06:49 tiff build fail/join #linuxpmi Apr 19 17:15:13 03Koen Kooi  07org.openembedded.dev * r6ea387ca31 10openembedded.git/recipes/gnome/gnome-media_2.26.0.bb: gnome-media: fix packaging and make it more determistic Apr 19 17:18:05 gremlin[it]: hi, I retried building bluez-utils with stable/2009 but it always fails Apr 19 17:18:47 gremlin[it]: are you sure that you tried with srable instead of org.opernembedded.dev ? Apr 19 17:19:21 gremlin[it]: because with org.opernembedded.dev bluez-utils builds successfully, but x11-mage fails further Apr 19 17:20:07 * mckoan can't guess what happened to bluez-utils with stable/2009 because seems not-too-stable :-P Apr 19 17:21:18 mckoan: bluez4? Apr 19 17:21:39 seems that it should be tested before in .dev Apr 19 17:22:14 Jay7: http://tinderbox.openembedded.net/public/logs/4503894.txt Apr 19 17:22:58 mckoan: at first try w/o -j 4 :) Apr 19 17:23:31 stable use bluez-utils-3.36-r4 Apr 19 17:23:43 yes, I see already Apr 19 17:24:04 hello,I don't understand what is the svn module Apr 19 17:24:22 is it the directory to download? Apr 19 17:24:43 Gnutoo: in cvs it was record in 'modules' file that point to some directory Apr 19 17:24:51 Jay7: retrying w/o -j4 Apr 19 17:24:56 ah ok thanks a lot Apr 19 17:25:04 Gnutoo: kind of alias Apr 19 17:25:12 ok thanks Apr 19 17:25:32 I get it...so I have to dowload the whole svn and look for this module Apr 19 17:25:43 maybe in . svn dirs Apr 19 17:26:17 hm.. in cvs there was possibility to fetch only CVS 'metadata' with settings Apr 19 17:26:21 CVSRoot iirc Apr 19 17:26:30 imho in svn should be something like Apr 19 17:27:09 hi, does anyone know what to do when this comes up: Apr 19 17:27:10 Error, DISTRO_PR has changed ( to .1) which means all packages need to rebuild. Please remove your TMPDIR so this can happen. For autobuilder setups you can avoid this by using a TMPDIR that include DISTRO_PR in the path. Apr 19 17:27:21 ok thanks Apr 19 17:28:11 mrec__: 'Please remove your TMPDIR so this can happen.' Apr 19 17:28:42 the question is where is TMPDIR Apr 19 17:29:06 I read it worked for someone to remove the tmp directory from the build dir, I don't have it there Apr 19 17:29:40 mrec__: look in local.conf Apr 19 17:31:35 Jay7: same problem http://tinderbox.openembedded.net/packages/492201/ Apr 19 17:32:18 mckoan: can you grep for HCI_SECMGR in bluez sources? Apr 19 17:32:27 or right in workdir Apr 19 17:32:43 I don't know what do do, I tried to solve the error HCI_SECMGR but I'm not sure it is the best thing to do Apr 19 17:33:02 I can't figure out why in main branch it works Apr 19 17:33:05 imho, this coming from configure Apr 19 17:33:24 may be some autoconf variables Apr 19 17:33:38 a diff of two bluez branches shows nod diffs Apr 19 17:33:44 Jay7: I overlooked that thanks, I overlooked that since the getting started doc didn't say it has to be edited in the conf file Apr 19 17:35:18 mrec__: iirc there was some point about 'get local.conf.sample and edit it according to your environment' :) Apr 19 17:39:07 Jay7: ya I'm still learning Apr 19 17:39:23 * Jay7 too.. Apr 19 17:39:25 NOTE: package shasum-native-1.0-r1: task do_unpack: failed Apr 19 17:39:25 ERROR: TaskFailed event exception, aborting Apr 19 17:39:38 hope this is something temporary... I'll retry Apr 19 17:41:34 Jay7: I start commenting out the line if (hci_test_bit(HCI_SECMGR, &di->flags)) just to see what happens Apr 19 17:42:32 mckoan: it will fail somewhere else ;) Apr 19 17:43:09 Jay7: so what would be the best way to proceed in this case? Apr 19 17:43:22 bad luck it fails again Apr 19 17:43:23 mckoan: I'm looking for this define in includes Apr 19 17:43:27 NOTE: package shasum-native-1.0-r1: task do_unpack: failed Apr 19 17:43:27 ERROR: TaskFailed event exception, aborting Apr 19 17:43:47 Jay7: it is in kernel include/net/bluettoth Apr 19 17:43:48 is there anything that can be done? Apr 19 17:45:14 mrec__: which distro/machine/image triplet? Apr 19 17:45:15 mckoan, sure i'm on stable/2009 and check i have no custom change on recipes or strange things in local.conf :( don't know Apr 19 17:45:35 gremlin[it]: could you please enable tinderbox? Apr 19 17:46:31 distro: ångström; machine: davinci-dvevm Apr 19 17:46:39 activated yesterday, you should see a good compilation of bluez ... my name in "gremlin.it" Apr 19 17:49:41 gremlin[it]: sorry, I'm unable to find it Apr 19 17:53:17 mckoan, true, isn't listed ... i'm not my build machine now ... i'll do later probably Apr 19 17:53:21 mrec__: may well I won't be able to help you but, do you have more log info about your error? Apr 19 17:53:51 gremlin[it]: no prob, thank you :-D Apr 19 17:54:46 Jay7: commenting that line I get this : http://tinderbox.openembedded.net/public/logs/4506719.txt Apr 19 17:56:32 mckoan: still looks like lack of some includes Apr 19 17:56:41 may coming from wrong configure Apr 19 17:58:21 Jay7: in bluez-utils_3.36.bb and bluez-utils3.inc ? Apr 19 17:58:48 mckoan: try to use devshell and configure it by hands Apr 19 17:58:57 then check what is in config.in Apr 19 17:59:27 Jay7: I don't know how to use devshell, sorry Apr 19 17:59:42 mmm...strange there is no activity since that message: "At revision 8867." since a couple of minutes Apr 19 18:00:53 ah maybe it's gzipping the svn Apr 19 18:05:01 * mckoan added INHERIT += "oestats-client devshell" Apr 19 18:06:48 Jay7: what should happen when devshell is set in INHERIT ? I get the same error? Apr 19 18:06:52 maybe it was trackerd that made it so slow... Apr 19 18:07:15 I mean that I get the same error. Apr 19 18:07:26 mckoan: try to bitbake -c devshell your_bluez_recipe Apr 19 18:07:39 it should falls to devshell Apr 19 18:07:50 aha! Apr 19 18:08:00 hi, all! Apr 19 18:10:35 to support blackfin builds I need to build several versions of toolchain, one for target bfin-uclinux and another for target bfin-linux-uclibc, what is the best way to achieve this in OE? primary toolchain for applications should be linux-uclibc one (since it supports shared libs), but uclinux one is needed too. Any ideas? Apr 19 18:10:55 Jay7: same error I faced tome ago: devshell doesn't work if you work in a remote screen shell :-( Apr 19 18:11:00 ERROR: function do_devshell failed Apr 19 18:11:07 | cannot open display: Apr 19 18:11:47 I can't go further here, maybe tomorrow...(I already spent all the weekend on this bug) Apr 19 18:12:25 mckoan: I ran it a 3rd time, without changing any configuration and now it works for some reason... maybe the download was corrupted Apr 19 18:12:30 ok, good luck :) Apr 19 18:13:08 mrec__: good to know, I would like to be in you shoes with my problem :-D Apr 19 18:13:18 mckoan: what's your problem? Apr 19 18:13:19 Jay7: thank you anyway Apr 19 18:13:43 bluez-utils does not build with stable Apr 19 18:19:13 * Jay7 need more time and new buildbox :) Apr 19 18:21:13 mmm...I have no response in #buglabs since several hours...so I'm trying here...does someone knows the kernel build system...I've :drivers/video/bugfb.c:39:31: error: mach/mx31bug_cpld.h: No such file or directory Apr 19 18:22:07 I bet the kernel use its own headers(not the linux-libc-headers) Apr 19 18:22:37 mckoan, why do you inherit devshell instead of bitbaking it? Apr 19 18:23:05 http://wiki.openembedded.net/index.php/Interactive_debugging Apr 19 18:26:16 I'll build it with V=1 Apr 19 18:29:44 03Marcin Juszkiewicz  07org.openembedded.dev * r6abf408410 10openembedded.git/recipes/tracker/tracker_0.5.4.bb: tracker: added libgnomecanvas to dependencies Apr 19 18:29:46 03Marcin Juszkiewicz  07org.openembedded.dev * r8543e3ed28 10openembedded.git/conf/checksums.ini: checksums.ini: added checksum for util-linux-ng 2.13.1 Apr 19 18:29:52 03Marcin Juszkiewicz  07org.openembedded.dev * r8972f6f429 10openembedded.git/recipes/util-linux-ng/ (util-linux-ng.inc util-linux-ng_2.13.1.bb): util-linux-ng: depend on virtual/libintl to pass building with uclibc Apr 19 18:30:29 ??? Apr 19 18:30:54 it seems that it doesn't include the arch where there is the header Apr 19 18:31:26 Gnutoo: because I read this http://bec-systems.com/oe/html/usage_devshell.html Apr 19 18:31:48 Gnutoo: BTW now I understand that is a bitbake option :-) Apr 19 18:31:57 ok Apr 19 18:32:45 Gnutoo: BTW I added a bug #5084 about this Apr 19 18:33:06 Gnutoo: (about devshell) Apr 19 18:33:23 mckoan, mmm...the bug reporting system is kind of not very used...prefer reporting it to the mailing lists Apr 19 18:34:06 Gnutoo: I'm having trouble with SL-5500 USB ethernet bridging. Can someone here help me, or is this OT? Apr 19 18:34:32 Gnutoo: ok thanks I'll post there too Apr 19 18:34:33 Balrog_, ask #angstrom ..we're a build system Apr 19 18:34:49 ok ... no one seems to be there much of the time :/ Apr 19 18:34:56 but ok. Apr 19 18:39:09 Balrog_: retry in #angstrom tomorrow after 9:00 CET Apr 19 18:39:26 I'm ok. CET is? Apr 19 18:40:05 Central European Time (GMT+2 with daylight saving) Apr 19 18:40:50 well, I can be online after 18:00 CET or so Apr 19 18:41:27 Balrog_: should be good, there are a few ppl here in the weekend ;-) Apr 19 18:43:39 is it possible to do a git bisect wth oe (in an easy way) ? Apr 19 18:46:16 re Apr 19 18:53:02 good morning Apr 19 18:53:37 Gnutoo: you're right, OE ML seems down Apr 19 19:10:42 I've a problem when I use my kernel recipe in my overlay dir Apr 19 19:11:13 I can build my overlay recipe with bitbake -b local/recipes/linux/linux-omap_local.bb Apr 19 19:11:51 but when I do a bitbake virtual/kernel, oe goes mad Apr 19 19:12:19 :| Apr 19 19:12:24 goes mad ? Apr 19 19:12:32 ERROR: '[]' RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'virtual/arm-angstrom-linux-gnueabi-depmod-' but it wasn't found in any PACKAGE or RPROVIDES variables Apr 19 19:12:37 hi Apr 19 19:12:44 hi hrw Apr 19 19:13:33 flameman: and it starts building linux-2.6.25.20-r2 Apr 19 19:13:49 which makes no sense to me... Apr 19 19:13:58 not to me, too Apr 19 19:14:20 anyway i got a bit of strange these days ... trying to compile kexecboot Apr 19 19:14:46 03Leon Woestenberg  07org.openembedded.dev * r26c2d38f3c 10openembedded.git/recipes/lighttpd/lighttpd_1.4.22.bb: lighttpd: Added 1.4.22 with DEFAULT_PREFERENCE -1. Apr 19 19:14:56 03Leon Woestenberg  07org.openembedded.dev * r6eb7e1f420 10openembedded.git/recipes/linux/linux-rt_2.6.29.bb: linux-rt: Add 2.6.29 with DEFAULT_PREFERENCE -1. Apr 19 19:15:36 the kernel recipe in my overlay is here: http://pastebin.com/d51d776ac Apr 19 19:31:18 mpoullet|home: did you copy the linux.inc over to your overlay? Apr 19 19:31:35 mpoullet|home: is your overlay higher priority? Apr 19 19:32:00 likewise: yes and yes Apr 19 19:33:02 mpoullet|home: why DEFAULT_PREFERENCE = "-1"? Apr 19 19:33:26 mpoullet|home: you select the version elsewhere? show how pls Apr 19 19:35:01 likewise: I thought that way I can select it in my local.conf Apr 19 19:35:15 likewise: haven't tried it yet though Apr 19 19:35:44 mpoullet|home: if you have DEFAULT_PREFERENCE -1 you must either override that, or specifically set a version Apr 19 19:36:01 mpoullet|home: bitbake warns about this, but will commence. Apr 19 19:37:42 likewise: I'd like to use the scheme proposed here: https://lists.berlios.de/pipermail/bitbake-dev/2009-April/000432.html Apr 19 19:38:14 likewise: so far it works with my own packages Apr 19 19:38:44 likewise: it's the first time I try to duplicate something in the oe repo in my overlay Apr 19 19:39:45 mpoullet|home: the scheme proposed is not using overlays... Apr 19 19:39:57 mpoullet|home: How is your PREFERRED_VERSION_linux-omap ? Apr 19 19:40:35 likewise: PREFERRED_VERSION_linux-omap = "local" in my local.conf Apr 19 19:40:48 mpoullet|home: and what's the name of your bb? Apr 19 19:41:18 likewise: local/recipes/linux/linux-omap_local.bb Apr 19 19:42:04 mpoullet|home: Looks good so far. I would do a "bitbake -D -v linux-omap" and carefully see the (first) bitbake out. Apr 19 19:43:02 likewise: ok, trying this Apr 19 19:46:02 likewise: here is the log: http://pastebin.com/d3d4c3254 Apr 19 19:47:50 mpoullet|home: ok, this is just a guess: by changing the version number to "local", some recipe cannot find out whether it needs 2.4 or 2.6 depmod. Apr 19 19:48:11 mpoullet|home: suggestion: in your layout, change -local.bb back to the version number Apr 19 19:48:45 mpoullet|home: I'm quite sure this solves it Apr 19 19:49:04 likewise: ok, trying this Apr 19 19:49:14 mpoullet|home: if your local overlay has higher preference, the .bb file may be of the same name. Apr 19 19:53:17 03Leon Woestenberg  07org.openembedded.dev * rd40ae0dced 10openembedded.git/recipes/wt/wt_2.2.3.bb: Apr 19 19:53:17 wt-2.2.3: ARM_INSTRUCTION_SET = "arm" to workaround toolchain issue. Apr 19 19:53:17 Works around this: Apr 19 19:53:17 | /tmp/ccGh2Qre.s: Assembler messages: Apr 19 19:53:17 | /tmp/ccGh2Qre.s:2339: Error: selected processor does not support `swp r4,r8,[r5]' Apr 19 19:55:00 likewise: bitbake linux-omap seems to work, thanks a lot! :)! Apr 19 19:55:27 likewise: but does this work with virtual/kernel too? Apr 19 19:55:45 mpoullet|home: so, to summarize, there is some magic involved finding out if depmod 2.4 or 2.6 must be built, but as you removed the kernel version in recipe name, this failed. Apr 19 19:55:57 hm likewise wt has to be fixed Apr 19 19:56:10 woglinde: in what way? Apr 19 19:56:21 likewise: and thanks for explaining me the details Apr 19 19:56:31 mpoullet|home: no problem Apr 19 19:56:32 swp r4,r8,[r5] Apr 19 19:56:36 mpoullet|home: you're welcome Apr 19 19:56:38 likewise, nice patch for wt, didn't know about ARM_INSTRUCTION_SET Apr 19 19:56:41 so it its thumb compatible Apr 19 19:56:53 woglinde: isn't that a toolchain fault rather than a wt fault? Apr 19 19:57:17 eFfeM: arm switched to thumb instruction set by default, but not all packages go well. Apr 19 19:57:35 likwise I will look at wt Apr 19 19:57:36 eFfeM: rather, "arm in angstrom in oe" switched ... Apr 19 19:57:49 likewise with which machine did you test? Apr 19 19:58:00 woglinde: pls explain how you can change that in wt itself? Apr 19 19:58:04 woglinde: sheevaplug Apr 19 19:58:40 swp shouldnt be there on sheeva Apr 19 19:59:06 hm oh wt needs boost Apr 19 19:59:11 on what cores is swp present? Apr 19 19:59:37 swp should be used on arm cores < 6 Apr 19 20:00:14 mpoullet|home: so at last what was the magic? how did you set your variables? Apr 19 20:00:39 woglinde: from the datasheet: "Compliant with v5TE architecture" Apr 19 20:00:56 for armv5 there is on linux a kernelhack which can be used Apr 19 20:02:44 (I only build-tested BTW) Apr 19 20:03:30 #Is ARM_INSTRUCTION_SET dependent on machine, i.e. can I do ARM_INSTRUCTION_SET_ Apr 19 20:03:51 oops, not quite finished typing, but you get the question :-) Apr 19 20:05:11 ah right the workaround is Apr 19 20:05:16 what's wt? Apr 19 20:05:18 wesnoth? Apr 19 20:05:20 mckoan: I've renamed the recipe linux-omap_local.bb back to linux-omap_2.6.28.bb and that fixes the problem, see likewise's explanation Apr 19 20:05:27 that the specifig file is compiled wihtout thumb support Apr 19 20:05:42 ah no...it's related to the commit here Apr 19 20:05:50 likewise please look at glib there we do the same Apr 19 20:06:01 pb fixed it some month ago Apr 19 20:08:01 Gnutoo: wt is like Qt but for web pages Apr 19 20:08:08 hrms Apr 19 20:08:12 boost sucked Apr 19 20:08:20 Gnutoo: it allows you to program a GUI in C++ for web pages. Apr 19 20:08:29 yes I just looked at the recipe Apr 19 20:08:47 Gnutoo: really nice approach, there was a presentation during FOSDEM 2009 Apr 19 20:08:52 yes there was a talk about that at fosdem...but I didn'a attend to it Apr 19 20:09:04 Gnutoo: you visited the OE booth? Apr 19 20:09:05 woglinde, problem on boost? Apr 19 20:09:12 woglinde, s/on/with/ Apr 19 20:09:15 which version? Apr 19 20:09:20 do you need help? Apr 19 20:09:28 gnutoo I compiled my uclibc without the SUS3 macros Apr 19 20:09:43 this unlcloacked many nasty stuff Apr 19 20:09:57 where the programm thinks uclibc has only timeb.h header Apr 19 20:10:00 woglinde, ah ok uclibc...never tried boost with uclibc...yet Apr 19 20:10:13 and I don'tn know what SUS3 is Apr 19 20:10:34 ~seen hrw Apr 19 20:10:38 hrw was last seen on IRC in channel #oe, 58m ago, saying: 'hi'. Apr 19 20:10:44 woglinde: looking at glib-2.0 latest recipe doesn't tell me what the thumb fix is? Can you point me into some direction? Apr 19 20:11:09 ah I had problems with glib-2.0 but it was related to 2.4 kernel Apr 19 20:11:22 too much thing was staged Apr 19 20:11:49 mom Apr 19 20:12:23 does someone know if kergoth's collections.inc patch is going into stable/2009 someday? Apr 19 20:12:40 (uclibc staged things related to inotify...but 2.4 didn't have inotify) Apr 19 20:14:24 likwiese -> SRC_URI_append_arm = " file://atomic-thumb.patch;patch=1" Apr 19 20:14:47 woglinde: tnx, looking Apr 19 20:17:35 woglinde: it just sets "no-thumb". that's not really a proper fix IMHO Apr 19 20:18:06 woglinde: the toolchain generates instructions that are not supported, that is the real problem AFAICS Apr 19 20:18:34 woglinde: disabling thumb (either per package, or per part of it) is a workaround, AFAIK Apr 19 20:18:49 likewise thumb never has swp Apr 19 20:19:02 so there is no other way Apr 19 20:19:12 woglinde: yes, but swp is generated by the toolchain Apr 19 20:19:31 woglinde: I mean, wt is written in C++, not assembler Apr 19 20:19:41 yes Apr 19 20:19:50 then let us look into the code Apr 19 20:20:00 woglinde: yup ok Apr 19 20:28:14 hrms Apr 19 20:28:28 vlc needs clock_nanosleep.c Apr 19 20:28:39 which is only in uclibc-nptl branch for now Apr 19 20:45:13 03Leon Woestenberg  07org.openembedded.dev * rae7125e0e7 10openembedded.git/conf/checksums.ini: lighttpd: Added 1.4.22 to checksums.ini to prevent punishment. Apr 19 20:45:51 likewise we should update openssl Apr 19 20:48:15 * Jay7 is away for 3 days :) Apr 19 20:49:42 woglinde: any idea how I can see the output of a cmake build. I'm suspecting not all flags are properly propagated into the Wt build. Apr 19 20:49:54 woglinde: why openssl? Apr 19 20:51:57 security issuses Apr 19 20:52:22 hm there was a flag to cmake Apr 19 20:53:47 gb Apr 19 20:54:07 Set CMAKE_VERBOSE_MAKEFILE to ON Apr 19 20:56:43 hm uh Apr 19 20:56:53 uclibc with linuxthread has the same problem Apr 19 20:58:45 but there is the assembler hardcoded Apr 19 21:06:40 woglinde: I just editing the WCalendar.o gcc call in src/CMakeFiles/wt.dir/build.make line 703 Apr 19 21:07:11 edited. It is a name-mangled function, doesn't make much sense to me, it mentions boost in the mangled function name Apr 19 21:08:42 hm pspinlock.c needs some work on uclibc linuxthread Apr 19 21:09:20 woglinde: assembler output shows "#APP" before every swap. THis might be a boost thing? Apr 19 21:09:22 #APP Apr 19 21:09:24 swp r2, r3, [r5] Apr 19 21:10:09 *g* yes boost suffters the same Apr 19 21:12:18 likewise mightbe so Apr 19 21:14:14 woglinde: why suffers? boost build fine here, not with you? Apr 19 21:14:27 likewise atomic stuff Apr 19 21:14:30 like swp Apr 19 21:37:00 likewise: "#APP" is just an instruction to the assembler to enable syntax checking, it doesn't have any effect on the code output. Apr 19 21:37:23 gcc does that for all inline assembler. Apr 19 21:38:32 he pb Apr 19 21:38:42 hello Apr 19 21:38:59 pb do you have any thoughts for pspinlock.c file and thumb? Apr 19 21:39:00 hi woglinde, ant Apr 19 21:39:11 woglinde: dunno, what's the issue? Apr 19 21:39:20 swp instruction Apr 19 21:39:46 strange problem here..I suppose header have changed (sdp_extract_pdu in bluez-utils)...http://tinderbox.openembedded.net/public/logs/4506615.txt Apr 19 21:40:01 bluez 3 vs. bluez 4 ? Apr 19 21:40:16 http://sources.busybox.net/index.py/trunk/uClibc/libpthread/linuxthreads/sysdeps/arm/pspinlock.c?revision=13779&view=markup Apr 19 21:41:20 woglinde: yeah, you just need to compile that one file as arm code. Apr 19 21:41:36 pb no way to workaround in thumb modus? Apr 19 21:41:52 there's no swp in thumb, and since pspinlock.c doesn't really do anything apart from call SWP, there's no point trying to make it work inline Apr 19 21:42:41 woglinde:, pb__: same issue with "swp" in boost macro's: /tmp/work/armv5te-angstrom-linux-gnueabi/boost-1.36.0-r6$ grep -rn -e 'swp ' * Apr 19 21:42:50 mmm...I've done make ARCH=arm xconfig with a defconfig in .config and it configured me as another arm arch than the one in defconfig Apr 19 21:43:15 gnutpp try make oldconfig first Apr 19 21:43:25 pb___: no easy way to implement the atomics in thumb otherwise? Apr 19 21:43:31 ant__: I also think that bluez-utils have an issue with headers Apr 19 21:43:32 likewise: no Apr 19 21:43:35 woglinde, ok thanks Apr 19 21:43:51 gnutoo dont forget the ARCH=arm Apr 19 21:43:58 yes I didn't forget that Apr 19 21:44:00 pb___: so falling back to nothumb instructions is the proper solution for us? Apr 19 21:44:05 not just no easy way, I don't think there is even any difficult way. Apr 19 21:44:15 mckoan: indeed strange session.c:257: error: too many arguments to function 'sdp_extract_pdu' Apr 19 21:44:23 thumb1 simply doesn't have any atomic instructions. Apr 19 21:44:24 lol thumb+atomic...that's *very difficult* Apr 19 21:44:26 likewise: for that one file, yes Apr 19 21:44:43 yes that's the problem...no support for swp and similar Apr 19 21:44:50 obviously you can still compile the rest of the library as thumb Apr 19 21:45:09 see glib, for example, which has a similar issue with gatomic.c Apr 19 21:45:12 but maybe you could use woglinde's lwn.net link that is: http://lwn.net/Articles/314144/ Apr 19 21:45:22 pb I already showed him Apr 19 21:45:29 (Posted Jan 12, 2009 19:58 UTC (Mon) by npitre (subscriber, #5680) [Link] ) Apr 19 21:45:30 ok Apr 19 21:45:31 gnutoo nope Apr 19 21:45:37 its not about thumb Apr 19 21:45:51 ah ok Apr 19 21:45:57 pb___: boost has "spinlock_gcc_arm.hpp" which is re-used by other apps (such as wt) which now fail. Apr 19 21:45:59 didn't read well what he wrote Apr 19 21:46:42 woglinde: wow there are just too many packages that are not compiling with uclibc, I guess I will switch back to glibc. Apr 19 21:46:56 svolpe???? Apr 19 21:47:14 svolpe: fail how? Apr 19 21:47:31 svolpe I only hit errors because I disabled SUSv3 macros and using linuxthread instead of linuxthread.old Apr 19 21:48:03 hm okay the nanosleep think is beside from this Apr 19 21:48:14 but can be ported from the nptl branch Apr 19 21:48:28 which hopefully will be mergede this year Apr 19 21:48:38 mckoan: I found in the web one example with 2 arguments...http://svn.muiline.com/viewvc.py/obex-data-server/trunk/src/ods-bluez.c?revision=989&view=markup&sortby=author&pathrev=1156 Apr 19 21:49:15 likewise: I'm up to 9 packages I have had to fix already, I just hit the 10th and gave up as I'm running out of time. Apr 19 21:49:29 pb workaround in uclibc will be harder then in glib Apr 19 21:49:44 likewise: I will try it at a later date when I have more time but I need to get a clean build to do some testing. Apr 19 21:49:48 because they are using plain Makefiles mostly Apr 19 21:49:56 svolpe: for arm? Apr 19 21:49:58 but sed will be my friend Apr 19 21:50:23 we should try starting down-configuring eglibc as a midway inbetween uclibc and glibc Apr 19 21:50:51 likewise: it is for the AT91SAM9260 which is an arm9 although in OE it seems to have thumb enabled. Apr 19 21:50:53 likewise ask khem about eglibc Apr 19 21:51:16 svolpe: are your issues thumb related? Apr 19 21:51:30 svolpe: what happens if you set ARM_INSTRUCTION_SET = "arm" globally? Apr 19 21:51:36 hm I still unsure about which packages should be compiled with arm and which not Apr 19 21:51:44 if libc isnt better with arm support Apr 19 21:53:09 woglinde: at least anything matching: grep -rn -e DEPENDS openembedded.git/recipes/ | grep boost Apr 19 21:53:20 woglinde: or a subset thereof, due to that macro :-) Apr 19 21:53:23 likewise: I have not tried it globaly but all the packages I have added it to that would not compile compiled after I added it. Apr 19 21:53:47 ant did some profiling on it Apr 19 21:53:53 but others should do Apr 19 21:54:16 too Apr 19 21:54:17 woglinde: I should have th elink around Apr 19 21:54:21 likewise: what happens if you set it globally, what are the ramifications? Apr 19 21:56:51 svolpe: hmm, could be there are libraries with header files that leak into applications. if the library has arm instructions in assembly whihc leak into applications, you run into errors. Apr 19 21:57:06 woglinde: http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2008-December/002832.html Apr 19 21:57:19 svolpe: example: boost has arm header file with nothumb instructions, which probably make many dependent packages fail. Apr 19 21:57:33 likewise: ^^^ Apr 19 21:58:10 ant ram perforamnce is better with uclibc? Apr 19 21:58:16 cool Apr 19 21:58:25 seems from that single benchmark Apr 19 21:58:47 floating point is poisoned... Apr 19 21:58:47 ant__: yes, I know the reasons for thumb (smaller hit on the code cache etc) Apr 19 21:59:17 but we need to fix all the breakage... Apr 19 21:59:27 hm but why textrendering and gfx is slower Apr 19 21:59:42 distinctly slower Apr 19 22:00:09 btw opie uses framebuffer, no ATI W100 acceleration Apr 19 22:00:23 ant yes I know Apr 19 22:01:03 that's a bit unfortunate, a regression... Apr 19 22:01:06 ant could you please try the test with the newer uclibc version? Apr 19 22:01:34 yes, I can fire a build overnight Apr 19 22:02:00 newer = angstrom standard? Apr 19 22:02:04 when I fixed the linuxthread error Apr 19 22:02:28 ant gcc-4.3.3 binutils_cvs uclibc=0.9.30.1 Apr 19 22:02:37 otherwise we have no thumb support Apr 19 22:02:39 woglinde: and th egettext issue? Apr 19 22:02:41 because ldso Apr 19 22:02:58 are you mean the libintl only Apr 19 22:03:01 not whole gettext Apr 19 22:03:05 well, your branch Apr 19 22:03:12 yeah Apr 19 22:03:26 is it merged? Apr 19 22:03:30 nope Apr 19 22:03:40 I will about it this week Apr 19 22:04:03 ok, fine Apr 19 22:04:19 it needs some more testing on the RDEPENDS Apr 19 22:05:25 I understand...I must have forgetten ARCH=arm at a point Apr 19 22:05:35 the last defconfig was x86...ouch Apr 19 22:09:39 if it is any help I can give a list of all the packages that I had to add the ARM_INSTRUCTION_SET="arm" Apr 19 22:10:10 svolpe yes Apr 19 22:11:09 ant ah my gettext.bbclass rework isnt finished Apr 19 22:11:25 woglinde: I will keep chugging through the build (as time permits), I have a txt file that I have been keeping a list of all the files and will send it to you when I finally get a clean build. Apr 19 22:11:41 okay thanks Apr 19 22:13:42 gn Apr 19 22:18:08 woglinde: I'll wait and keep an eye on your commits Apr 19 22:27:52 good nite Apr 19 22:28:09 'nite Apr 19 23:13:48 Hello I am trying to bitbake netsurf, however I am receiving http://pastebin.com/d197b50a6 , any ideas that can help? Thanks Apr 19 23:25:20 The errors pertains to the QA handling of the packaging. Any help is appreciated. Apr 19 23:54:36 Anyone? **** ENDING LOGGING AT Mon Apr 20 02:59:57 2009