**** BEGIN LOGGING AT Thu May 27 02:59:57 2010 May 27 04:16:48 * kergoth_ thinks about revamping the variable reference tracking code to use dumb nodes in an AST rather than objects that do their own traversing May 27 04:32:20 hmmmm May 27 04:48:01 WHat is the difference between x11-gpe-image and gpe-image? is the first one - x11 and second GPE with X? May 27 04:48:23 thanx khem May 27 04:49:05 I updated the bug about binutils/nm. Thanx 4 looking into it. May 27 05:03:40 IgorK: ok May 27 05:03:46 I will fix it soon May 27 05:09:46 IgorK: so this happens with cross binutils or target binutils May 27 05:10:28 target binutils May 27 05:11:37 IgorK: do you see it when you boot the root file system May 27 05:12:01 I mean when you boot you dont have /usr/bin/nm ? May 27 05:13:49 no, I didn't May 27 05:14:43 unfortunately I removed it from the image, but it installed itself in osme funny path May 27 05:15:21 well the real binary should be something like arm-angstrom-linux-gnueabi-nm May 27 05:15:23 do you know the difference between x11-gpe and gpe? May 27 05:15:24 in /usr/bin May 27 05:15:30 and nm should be a symlink to it May 27 05:15:42 as the name says May 27 05:15:50 its GPE with X server May 27 05:16:17 Understood, but just typing "nm" on my device gives "Unrecognized command" May 27 05:16:40 ah and you have binutils installed on device right ? May 27 05:17:07 I had, but it is removed now May 27 05:17:53 I am building gpe-image. want to give it a try instead of x11-gpe May 27 05:18:01 hence the question of difference May 27 05:19:16 I can quickly rebuild x11-gpe-iimage with binutils just to test if u want to May 27 05:28:14 yes please do May 27 05:28:22 I dont have this problem here May 27 06:06:39 03Khem Raj  07org.openembedded.dev * re1828041e6 10openembedded.git/recipes/openssl/ (openssl-1.0.0/configure-targets.patch openssl_1.0.0.bb): May 27 06:06:39 openssl_1.0.0.bb: Update configure target patch to have correct number of colons (Bug 5429) May 27 06:06:39 * Use linux-uclibcebi for target OS on ARM uclibc. May 27 06:06:39 Signed-off-by: Khem Raj May 27 06:06:50 03João Henrique Freitas  07org.openembedded.dev * rd6608669e7 10openembedded.git/recipes/lksctp-tools/ (lksctp-tools.inc lksctp-tools_1.0.7.bb): May 27 06:06:50 lksctp: Add recipes for lksctp tools May 27 06:06:50 Signed-off-by: João Henrique Freitas May 27 06:06:50 Signed-off-by: Khem Raj May 27 07:11:09 03sledz  07org.openembedded.dev * rcf2bdb96eb 10openembedded.git/recipes/linux/linux-2.6.24/hipox/defconfig: May 27 07:11:09 linux-2.6.24: enable ppp modules for hipox machine May 27 07:11:09 Signed-off-by: Steffen Sledz May 27 07:13:23 morning May 27 07:29:11 03Steffen Sledz  07org.openembedded.dev * ra04c213a1a 10openembedded.git/recipes/jansson/ (jansson.inc jansson_1.2.1.bb): May 27 07:29:11 jansson: new recipe May 27 07:29:11 Jansson is a C library for encoding, decoding and manipulating JSON data. May 27 07:29:11 Signed-off-by: Steffen Sledz May 27 07:29:11 Acked-by: Mario Schuknecht May 27 07:32:16 03Steffen Sledz  07org.openembedded.dev * r816630f2b2 10openembedded.git/recipes/ntp/files/hipox/ (ntp.conf ntpd): May 27 07:32:16 ntp: modified config for hipox machine May 27 07:32:16 * don't use externals NTP servers May 27 07:32:16 * use '-x' option to avoid forward time skips May 27 07:32:16 Signed-off-by: Steffen Sledz May 27 07:34:41 morning May 27 07:38:42 after last round of patches here dropbear now fails do_configure: -> file://allow-nopw.patch;apply=no \ -> not found May 27 07:48:04 03Roger Monk  07org.openembedded.dev * r2ccbd56b86 10openembedded.git/recipes/angstrom/angstrom-uboot-scripts.bb: (log message trimmed) May 27 07:48:04 angstrom-uboot-scripts: Change installdir to /boot and rename configs May 27 07:48:04 * Install the scripts into /boot instead of datadir May 27 07:48:04 * Install the original .txt's files as well as .scr's so that users can May 27 07:48:04 use target mkimage to rebuild/modify/recreate May 27 07:48:05 * Rename configs for consistency May 27 07:48:06 Signed-off-by: Roger Monk May 27 08:01:04 03Koen Kooi  07org.openembedded.dev * rf65670e47a 10openembedded.git/recipes/ti/ti-devshell.bb: ti-devshell: fix patch to TI stuff, spotted by Yaman Umuroglu May 27 08:28:46 Morning all. Dumb question of the day. Can anyone suggest why I'm getting no kernel modules installed in imote2-image? May 27 08:28:59 It's using the linux-zigbee git recipe. May 27 08:29:23 i'll admit I've never really understood how oe knows which modules to install. May 27 08:32:57 jic23, I have a IMAGE_INSTALL_append line in my machine conf to specify which kernel modules to install when creating an image, but this can be done at image level too of course. May 27 08:34:25 ao2, cool, I'll try that. May 27 08:34:49 I had thought the whole rrecomends stuff should pull in modules if they are available, but that's not working. May 27 08:35:19 kernel modules package have names like kernel-module-moto-usb-ipc that is kernel-module-${MODULE_NAME} May 27 08:35:26 packages* May 27 08:35:35 yup. May 27 08:36:56 jic23, there should be also something like kernel-module-all, but I don't recall if that's the exact name May 27 08:38:24 iirc kernel-modules May 27 08:48:37 ao2 and JaMa, Thanks, that fixed it - could have sworn I tried that earlier! May 27 08:49:37 jic23, you're welcome May 27 08:51:10 hey people, here's a post about using the git fetcher with "protocol=file", I found that very useful: http://ao2.it/en/blog/2010/05/27/neat-compilerun-cycle-git-and-openembedded May 27 08:52:03 let me know if there is something incomplete/inaccurate/incorrect May 27 09:04:41 is raj here atm? May 27 09:05:07 aw ok i c he's away May 27 09:27:03 03Stefan Schmidt  07org.openembedded.dev * r7b6e5798eb 10openembedded.git/recipes/images/imote2-image.bb: May 27 09:27:03 imote2-image: Remove ibrdtnd from the default install. May 27 09:27:03 Thinking about it twice it makes not much sense in the image. It is not needed May 27 09:27:03 for the board and people who are interested in delay tolerant networking can May 27 09:27:03 just install the package. May 27 09:41:19 aoe, looks good to me. That's pretty much how I do it. May 27 09:41:29 ao2 that is! May 27 09:42:08 and Stefan just fixed my other bug. Today is proving successful. May 27 09:54:15 bump: after last round of patches here dropbear now fails do_configure: -> file://allow-nopw.patch;apply=no \ -> not found May 27 09:54:29 kergoth is still afk... May 27 09:57:59 ant_work, i'm stuck at the same error May 27 09:58:03 I just spotted this this morning, no time to debug it until this evening...hopefully is recipe-issuel (patch -p1 fails) May 27 09:58:35 if you are there pls check where the patch landed in WORKDIR May 27 10:01:15 no, there's no such patch in workdir (if you're asking me, ant_work :D) May 27 10:01:28 yea, thx May 27 10:01:44 ant_work: quick fix : DISTRO_TYPE=release ;-) May 27 10:01:50 ;) May 27 10:02:02 then we have an issue with file://allow-nopw.patch;apply=no May 27 10:02:15 yesterday apply=yes was changed... May 27 10:02:36 su I suspect base.bbclass or patch.bbclass May 27 10:03:48 well, i guess i'll just edit the file in the meantime May 27 10:04:17 he he May 27 10:05:08 pls let us know if this is the only recipe breaking after the last kergoth's cleanup May 27 10:05:49 is there an easy way of disabling parallel building for a particular recipe? May 27 10:06:05 (lowpan-tools fails currently due I think to a bork upstream) May 27 10:06:12 PARALLEL_MAKE = "" in recipe May 27 10:06:24 hrw, ah. I guess that was obvious May 27 10:06:38 ok, ant_work May 27 10:06:43 brb May 27 10:10:25 ericben: btw, about psplash logo, wich size is it? May 27 10:10:34 do we have vga and qvga? May 27 10:11:06 ant_work: only one (I think there was only one befor no ?) May 27 10:11:30 ant_work: size : 220x177 May 27 10:11:38 fits well in a QVGA screen May 27 10:12:01 I ask because I did 2 versions of it as kernel bootlogo May 27 10:12:14 see path: root/recipes/kexecboot/files May 27 10:12:29 ant_work: ok, it would be possible to tine psplash recipe to generate the bootlogo depending on the screen size May 27 10:13:10 I tried to have 'similar if not same' logos for kernel and bootsplash May 27 10:13:19 ericben: define screen size for whole arch? May 27 10:13:30 ericben: what about devices with video output? May 27 10:13:32 hrw: no for machine May 27 10:13:51 beagleboard looks different with 800x480/1024x768/1280x800 etc May 27 10:13:58 still one machine May 27 10:14:08 hrw: yes but actually you have only one logo on beaghleboard ? May 27 10:14:19 which is the small one May 27 10:14:34 probably May 27 10:14:39 now I have ubuntu on bb May 27 10:14:47 so maybe we could have one for wvga and one for VGA which can be used for "bigscreen" May 27 10:14:47 jo May 27 10:14:55 hi woglinde May 27 10:15:07 hrw: last time I checked ubuntu on i.MX51 is was quite slow May 27 10:15:23 ~seen zecke May 27 10:15:24 zecke <~ich@dsl-62-220-14-162.berlikomm.net> was last seen on IRC in channel #oe, 2d 2m 20s ago, saying: 'mckoan: is it quilt or patch?'. May 27 10:15:28 is it fast on BB ? May 27 10:15:52 hi woglinde May 27 10:16:06 ericben: I use it as headless for some experiments May 27 10:16:20 hi ant May 27 10:17:00 hrw: ok I had a big screen connected : http://www.eukrea.com/en/news/101-122009-ubuntu-sur-imx515 ;-) May 27 10:17:53 ericben maybee you had only xorg-fdev running May 27 10:18:08 hi djwillis May 27 10:18:15 ericben: technically speaking I have 20" lcd with 1280x800 connected. but most of time turned off May 27 10:18:55 damn musb error with bb May 27 10:19:15 I cannt connect anything via usb to it May 27 10:19:21 with the latest kernel May 27 10:19:30 woglinde: maybe I didn't spent some time to optimize it, I expect to integrate the accelerated drivers to OE soon, ubuntu is to big for what I need May 27 10:19:44 right May 27 10:19:50 to many deps May 27 10:20:18 hrw: here it was in 1280x1024 (when I said big screen, it was big resolution ;-) May 27 10:21:19 woglinde: yes and the parts provided by Freescale on their site are split in several places which don't makes it easy to use them and to use a customized kernel/config May 27 10:21:34 ericben haha sounds like ti May 27 10:22:28 okay lunch with zecke now May 27 10:30:58 03Martin Jansa  07org.openembedded.dev * re9aacc4ce4 10openembedded.git/recipes/ (8 files in 2 dirs): May 27 10:30:58 shr-apps: bump SRCREV of autorev packages May 27 10:30:58 Signed-off-by: Martin Jansa May 27 10:31:08 03Martin Jansa  07org.openembedded.dev * re70f21a140 10openembedded.git/recipes/openmoko-3rdparty/emtooth_svn.bb: May 27 10:31:08 emtooth: bump SRCREV May 27 10:31:08 Signed-off-by: Martin Jansa May 27 10:31:10 03Martin Jansa  07org.openembedded.dev * r892b0fc752 10openembedded.git/recipes/eglibc/eglibc_2.12.bb: May 27 10:31:10 eglibc_2.12: add es_CR.ISO-8859-1 to GLIBC_BROKEN_LOCALES May 27 10:31:10 * details: http://sourceware.org/bugzilla/show_bug.cgi?id=11258 May 27 10:31:10 Signed-off-by: Martin Jansa May 27 10:31:11 03Martin Jansa  07org.openembedded.dev * rb3f4268d41 10openembedded.git/recipes/shr/phoneui-shr-theme-neo_git.bb: May 27 10:31:11 phoneui-shr-theme-neo: bump SRCREV (fixed quick-settings theme) May 27 10:31:12 Signed-off-by: Martin Jansa May 27 10:31:12 03Martin Jansa  07org.openembedded.dev * ra8409e9153 10openembedded.git/recipes/shr/phoneui-shr-theme-o2_git.bb: May 27 10:31:13 phoneui-shr-theme-o2: bump SRCREV (fixed quick-settings theme) May 27 10:31:13 Signed-off-by: Martin Jansa May 27 10:51:16 What the environment variables enaMonExt in u-boot is used for ? May 27 11:03:18 re May 27 11:04:39 hi, anyone by any chance tried to bake gcc 4.1.2 (for any target) on ubuntu 10.04? I'm getting gcc-4.1/gcc/crtstuff.c:323: internal compiler error: Segmentation fault May 27 11:06:34 effem where do you need gcc .41? May 27 11:07:52 woglinde: i'm trying to get the nios2 toolchain to build under oe, that toolchain is make with a 4.1.2-like gcc (from windriver, not exactly the 4.1.2 from gnu) May 27 11:08:08 ieeehks May 27 11:08:14 wanted to get things working on 4.1.2 before I proceed to a newer version May 27 11:09:09 tried also to use it as an external toolchain in oe but failed miserably, bitbake helloworld or zip works but for soem reason with external toolchain bitbake bash or busybox still want to compile gcc-cross May 27 11:10:31 * eFfeM_work is open to any suggestion May 27 11:10:42 hi mickyl May 27 11:10:46 hi coredump May 27 11:12:04 eFfeM_work hm run bitbake -DDD and bitbake -g to see why bash or busybox not depends on external toolchain May 27 11:12:25 http://www.ouaza.com/wp/?p=271 May 27 11:12:30 Raphaël Hertzog: Rewriting update-alternatives in C May 27 11:13:42 woglinde, did bitbake -g, this line didn't make me happy: May 27 11:13:43 "libgcc" [label="libgcc(gcc-cross) 0:4.3.3-r12.1\n/home/frans/workspace/nios2/openembedded.git/recipes/gcc/gcc-cross_4.3.3.bb"] May 27 11:14:00 that was with external toolchain May 27 11:14:04 hi GNUtoo|oeee May 27 11:14:17 hrw: great news! u-a-cworth ie after upgrading busybox is quite slow :) May 27 11:15:22 eFfeM_work, hi May 27 11:15:28 eFfeM_work hm could you try libgcc_provided May 27 11:15:39 hi gnutoo May 27 11:18:12 * woglinde wonders why nobody packed libxvp yet May 27 11:18:13 woglinde, you mean ASSUME_PROVIDED += "libgcc" ? (or would that be virtual/libgcc) Or do I misunderstand you May 27 11:19:49 effem yeah thats what I meant May 27 11:20:29 hm you could grep recipes if it is virtual/ or purelibgcc May 27 11:20:31 ok, will try May 27 11:20:35 dont know it offhand May 27 11:23:48 satisfy_dependencies_for: Cannot satisfy the following dependencies for task-base: May 27 11:23:49 | * alsa-utils-alsactl * alsa-utils-alsamixer May 27 11:24:18 anyone familiar with that? May 27 11:24:57 hm either remove alsa-complet May 27 11:25:07 or correct the dep May 27 11:25:11 in the recipe May 27 11:25:35 or try to add both packages to the image May 27 11:26:01 i'll be needing alsa though May 27 11:29:29 woglinde: seems to improve things,let's see how far we get with that May 27 11:32:58 yeah May 27 11:35:17 i don't get it - how would the alsa-utils-alsactl dep be satisfied ? May 27 11:35:59 cmoh look if there is a recipe or packge for it May 27 11:36:00 i mean there is an alsa-utils package, but no alsa-utils-alsactl May 27 11:36:23 the alsa-utils package does build FILES_alsa-utils-alsactl = "${sbindir}/alsactl" May 27 11:36:34 yes May 27 11:36:50 the reason i'm asking is that a simple grep through recipes provides a whole list of recipes using this dep May 27 11:39:18 woglinde, what should i do then? make alsa-utils PROVIDE alsa-utils-alsactl? May 27 11:41:04 cmoh is there no line PACKAGES += "alsa-utils-alsactl" ? May 27 11:41:07 woglinde: added ASSUME_PROVIDED for libgcc and a few friends, still bitbake -g says it wants to build gcc-cross-initial and gcc-cross-intermediate (but not gcc-cross). no idea why, there is no package in depends.dot that has something with gcc-cross* at the right side of the -> May 27 11:41:22 effem strange May 27 11:42:00 woglinde, there is that in the alsa-utils recipe May 27 11:42:25 woglinde: see http://www.pastie.org/979717 May 27 11:44:17 and I already have virtual/${TARGET_PREFIX}gcc-cross-initial virtual/gcc-cross-initial gcc-cross-initial in ASSUME_PROVIDED May 27 11:44:26 at this stage better have too much in it than too little May 27 11:44:36 right May 27 11:46:33 re ericben May 27 11:46:57 re woglinde May 27 11:48:04 trying to build 4.1.2 for beagleboard now, to find out if it is an issue with gcc or with my machine (or the nios changes I have) May 27 12:39:40 03Roger Monk  07org.openembedded.dev * r9b6418d9f3 10openembedded.git/recipes/angstrom/angstrom-uboot-scripts/linuxtagdemo.cmd: May 27 12:39:40 angstrom-uboot-scripts: Add linuxtag demo script May 27 12:39:40 Signed-off-by: Roger Monk May 27 12:39:40 Signed-off-by: Koen Kooi May 27 12:41:52 03Martin Jansa  07org.openembedded.dev * rea69e240f8 10openembedded.git/recipes/shr/elementary-theme-gry_git.bb: May 27 12:41:52 elementary-theme-gry: bump SRCREV (fixed #1115) May 27 12:41:52 Signed-off-by: Martin Jansa May 27 13:18:45 i'm trying to use a debian nfs rootfs with my custom kernel. At first i had unable to open initial console problem, that i resolved by creating the node ttyS0 but now, after the routt file system has been mounted i get error 'nfs: server 192.168.0.2 not responding, still trying'. What could that be ? May 27 13:21:11 another quick question: how can i control the cxxflags and the cflags globally for a bitbake run? May 27 13:33:54 03Martin Jansa  07org.openembedded.dev * r7ce6d567e4 10openembedded.git/recipes/xinput-calibrator/ (4 files in 3 dirs): May 27 13:33:54 xinput-calibrator: add pointercal.xinput data for om-gta02 and install it by default May 27 13:33:54 * is there someone else then SHR using xinput_calibrator_once.sh in May 27 13:33:54 Xsession.d? Then you should provide valid pointercal.xinput for your May 27 13:33:54 devices. May 27 13:33:55 Signed-off-by: Martin Jansa May 27 13:34:03 03Martin Jansa  07org.openembedded.dev * rfa32ebe810 10openembedded.git/recipes/xserver-common/ (2 files in 2 dirs): May 27 13:34:03 xserver-common: use only xinput_calibrator for calibration (used only for SHR) May 27 13:34:03 Signed-off-by: Martin Jansa May 27 13:54:12 03Martin Jansa  07org.openembedded.dev * r53d71738a2 10openembedded.git/recipes/mesa/ (mesa-dri-7.8.1/glamo.patch mesa-dri_7.8.1.bb): May 27 13:54:12 mesa-dri_7.8.1: update glamo patch a bit May 27 13:54:12 Signed-off-by: Martin Jansa May 27 13:54:22 03Martin Jansa  07org.openembedded.dev * re04e059b26 10openembedded.git/recipes/xorg-driver/xf86-video-glamo_git.bb: May 27 13:54:22 xf86-video-glamo: bump SRCREV for newer libdrm compatibility May 27 13:54:22 Signed-off-by: Martin Jansa May 27 13:54:23 03Martin Jansa  07org.openembedded.dev * r00d250c39a 10openembedded.git/recipes/dri/glamo-dri-tests_git.bb: May 27 13:54:23 glamo-dri-tests: bump SRCREV for newer libdrm compatibility May 27 13:54:23 Signed-off-by: Martin Jansa May 27 13:54:24 03Martin Jansa  07org.openembedded.dev * rc6c42567d9 10openembedded.git/recipes/dri/libdrm_git.bb: May 27 13:54:24 libdrm_git: bump SRCREV for 2.4.20 May 27 13:54:25 Signed-off-by: Martin Jansa May 27 15:16:57 hi all. i had a question for people working professionally with SD-based rootfs's May 27 15:17:12 i work for bug labs and we've had trouble finding a company to program our SD cards in any volume May 27 15:17:40 we used a company a while ago but they would send us back FAT16-formatted cards with /bug-image-production.ext3 residing on the card May 27 15:17:50 anyone have suggestions or prior experience with a reliable company? May 27 15:17:58 we've been doing it in house but obviously it doesn't scale May 27 15:34:04 jconnolly, I'll ask a friend May 27 15:38:41 hi JC May 27 15:41:19 Crofton|work: much obliged May 27 15:41:22 hi hrw May 27 15:44:39 Hi. Trying to build boost for omap3530, bitbake wants first to build libxt. How can I find where this dependency is? I generated .dot files from bitbake for boost, but libxt does not show up. May 27 16:20:44 bitbake -g (or similar, check the help text from bitbake) May 27 16:22:40 anyone recall if we still have the original wiki pages from the original oe brainstorming sessions? May 27 16:22:44 in a backup somewhere, perhaps? May 27 16:23:34 cbrake: ping May 27 16:28:28 mwester-laptop: Thanks for the suggestion. I tried bitbake -g but the resulting .dot files contain no libxt May 27 16:42:50 hoj: libxt PROVIDES xt, but still you should see it in .dot files - grep libxt *.dot May 27 16:43:46 denix: grep libxt *.dot | wc -l = 0 I have very old bitbake; woudl that matter? May 27 16:49:53 hoj: it shouldn't. let me try... May 27 16:58:38 hoj: bitbake boost -g; grep libxt *.dot|wc -l = 164 May 27 16:58:58 hoj: depends.dot:"tk" -> "libxt" May 27 16:59:36 BitBake Build Tool Core version 1.8.19, bitbake version 1.8.19 May 27 17:01:22 denix: thanks for trying that. Must be old recipe files or some other change I never picked up. I'll try again in newer sources. May 27 17:07:06 good morning May 27 17:07:39 can someone please look at this compilation failure: http://pastebin.com/J5r6RmAP May 27 17:10:22 denix: updated arago tree gives better results. Thanks May 27 17:10:40 hoj: welcome :) May 27 17:21:19 kergoth_: hello May 27 17:25:45 cbrake: i was wondering if you happened to recall whatever happened to the really, really old wiki content. specifically, i'm wondering if we still have the original wiki pages where the brainstorming about the original design of oe were done May 27 17:25:52 or if you knew who might know May 27 17:26:17 kergoth_: koen was the one who switched from the really old wiki to drupal May 27 17:26:22 ah, okay, thanks May 27 17:26:45 i'm looking to try writing up some background / history and perhaps some high level conceptual info on oe that i haven't seen written down in many places May 27 17:26:49 kergoth_: I seem to recall it got pulled out from under us for some reason, so not sure if he has a copy of the content May 27 17:26:51 * kergoth_ needs to practice his writing anyway May 27 17:26:55 hmm, okay May 27 17:27:13 well, i can probably remember most of it, but i'll check with koen May 27 17:28:52 anyone? gpe-mixer can't be compiled.... May 27 17:29:24 cbrake: do you recall the irc channel he's usually in, offhand? May 27 17:29:26 * kergoth_ alays forgets May 27 17:30:28 kergoth_: try #beagle May 27 17:32:02 i get build failure of dropbear_0.51 because allow-nopw.patch No such file or directory May 27 17:32:19 wroberts1: check ML May 27 17:37:57 denix: devel, commits, issues, users,...? May 27 17:38:30 cbrake: koen says we had a backup when we were on linuxtogo, so i wonder what happenedt o it when we moved from there May 27 17:40:37 though, he says the really really old stuff was on hh.o.. May 27 17:40:41 * kergoth rolls eyes May 27 17:47:49 meh May 27 17:48:12 i have openembedded/recipes/dropbear/allow-nopw.patch, but it seems do_configure wants it to be in //work/armv5te-angstrom-linux-gnueabi/dropbear-0.51-r1.01 May 27 17:48:16 think the info i wanted is either gone entirely, or on hh.o's machine May 27 17:48:17 heh May 27 17:58:09 wroberts1: apply this patch http://patchwork.openembedded.org/patch/2106/ and http://patchwork.openembedded.org/patch/2104/ May 27 17:58:33 and http://patchwork.openembedded.org/patch/2105/ May 27 17:58:39 good morning all May 27 17:59:03 http://thread.gmane.org/gmane.comp.handhelds.openembedded/32984/focus=32989 May 27 18:00:15 wroberts1: google is your friend :) May 27 18:00:53 Hi denix May 27 18:01:35 JaMa: around ? May 27 18:01:39 khem: gm May 27 18:02:01 JaMa: I have figured out the gcc bug while compiling kernel 2.6.34 May 27 18:02:12 JaMa: its underwork to fix gcc May 27 18:03:08 meanwhile you can add a workaround and test if you want to the workaround is to add -fno-ipa-sra option to gcc you can use append it to FULL_OPTIMIZATION var May 27 18:04:12 denix: where in US are you ? May 27 18:04:23 khem: DC area May 27 18:04:34 its been raining in nothern CA May 27 18:04:38 this whole week May 27 18:05:14 one I am happy that I dont have to water the plants and garden but sad that I have to mow it often May 27 18:05:45 I didn't know rains are possible in the Valley... :) May 27 18:06:19 yeah this year its been raining since december May 27 18:06:55 denix: how is DC I was there in 2007 summer May 27 18:06:59 its nice in summer May 27 18:07:01 khem: great news, should I rebuild everything, or just kernel for test? May 27 18:07:13 JaMa: try kernel only first May 27 18:07:17 ok May 27 18:07:30 it should boot a root file system built with 4.4.4 ok May 27 18:08:05 khem: usually too humid in the summer here May 27 18:08:13 denix: thats right May 27 18:08:15 feels like a sauna :) May 27 18:08:20 heh May 27 18:09:58 but this year's spring wasn't too hot, sometimes even cold, which is unusual for the area May 27 18:10:30 yeah I have my sweater on and its beginning of June May 27 18:10:50 we just got hot weather (mid 90s F) few days back, which is 1-2 months later than usual :) May 27 18:10:55 global warming? :) May 27 18:11:07 yeah :) May 27 18:11:27 I think the fluctuations are high May 27 18:11:45 it varies from 55 to 80 here in couple of days May 27 18:12:47 right, same here May 27 18:13:12 denix: do you work out of home ? May 27 18:15:49 any suggestions on fixing gpe-mixer.patch so it become unified? anybody? May 27 18:16:03 hello pb_ May 27 18:17:25 pb_: I would like to create separate staging for intermediate gcc and glibc steps in order to avoid the sysroot conflicts do you think its ok approach May 27 18:19:06 IgorK: whats wrong with gpe-mixer.patch May 27 18:19:07 khem: sometimes, not right now May 27 18:20:14 khem: http://pastebin.com/J5r6RmAP May 27 18:22:26 03Martin Jansa  07org.openembedded.dev * r475aedbc9b 10openembedded.git/recipes/dri/libdrm_2.4.18.bb: May 27 18:22:27 libdrm_2.4.18: add missing dependency libpthread-stubs May 27 18:22:27 Signed-off-by: Martin Jansa May 27 18:26:51 IgorK: it works ok here May 27 18:27:00 IgorK: I would say clean it and rebuild May 27 18:27:16 bitbake -b gpe-mixer_0.42.bb -c clean May 27 18:27:20 bitbake -b gpe-mixer_0.42.bb -c patch May 27 18:27:25 see if that works May 27 18:27:47 IgorK: btw. it will be good if you updated to latest May 27 18:29:13 vim-7.2 also seems still broken after last .patch .diff changes (;apply=no is used there), seen some patch on ML about it May 27 18:30:33 khem fixed that yesterday May 27 18:31:00 Hi kergoth_ May 27 18:31:06 I'm pretty sure I have HEAD and it's broken again May 27 18:31:20 kergoth_: hmmm the patches are now in WORKDIR too May 27 18:31:53 kergoth: may be quilt should be using it from workdir and not creating a patches dir as it does May 27 18:32:18 kergoth_: that will be good because sometime some packages have patches as a sourcedir May 27 18:32:24 well, there's a patch on the list to fix an issue with unpacked patches combined with apply=no May 27 18:33:10 khem: personally, i think we could drop quilt entirely as our default patch application method. it's only really useful when debugging or developing, and you can set PATCHTOOL yourself to override it in that case May 27 18:34:50 kergoth: I would do the opposite i.e. set PATCHTOOL to something else if we are not developing and only doing build and release May 27 18:35:04 but thats a developer's hat May 27 18:35:16 pretty sure quilt also takes longer to apply them May 27 18:35:21 since it backs up all the files at each patch May 27 18:35:23 putting on user's hat I think I would say I dont care May 27 18:36:08 should profile it May 27 18:36:11 kergoth: although it would be cool if we could use PATCHTOOL like DISTRO or MACHINE May 27 18:36:19 in that case it will be good mix May 27 18:36:29 its just another variable May 27 18:36:38 i.e. as a developer I can start of my build with PATCHTOOL being patch May 27 18:36:41 you can May 27 18:36:58 if the recipe sets it, it'll override you, of course, unless you set PATCHTOOL_local or something May 27 18:37:02 but it works fine May 27 18:37:07 kergoth_: cool so it wont reparse again the whole thing ? May 27 18:37:19 ? May 27 18:37:26 if its in a .conf, of course it reparses May 27 18:38:13 the parse cache is only used to store package name, version, and dependency information for the runqueue generation May 27 18:38:17 + provides, etc of course May 27 18:38:29 any other variable could almsot certainly be switched to an env var with extrawhite safely May 27 18:38:36 kergoth_: say I was building a console image with default patchtool which was say patch. Now it breaks in vim and I want to debug why vim compile failed so now I would clean vim and the want to use quilt as patchtool and workout a patch for vim May 27 18:38:37 (as long as that var isnt' used indirectly in those) May 27 18:38:47 khem: same error May 27 18:38:53 you could do patchtool via the environemtn with BB_ENV_EXTRAWHITE May 27 18:38:55 or you could use amend.inc May 27 18:39:04 which lets you amend to recipes, so only that recipe would be reparsed May 27 18:39:29 kergoth_: yeah BB_ENV_EXTRAWHITE should be fine May 27 18:39:37 so you have my ack for it :) May 27 18:39:49 * kergoth_ 'll have to experiment with it first :) May 27 18:40:17 but still it would be nice if quilt used WORKDIR for patches May 27 18:40:18 may be May 27 18:40:53 IgorK: do you have a ~/.quiltrc May 27 18:41:11 would be nice to make it apply a patch from an external location without storing it locally in patches at all May 27 18:41:21 even better May 27 18:41:22 then it could apply them directly from their file:// location May 27 18:42:07 and yeah when I do quilt new then the patch could be created in FILESPATH too May 27 18:42:40 and quilt add would also mean git add :) May 27 18:43:00 and another script would append it to SRC_URI :) May 27 18:44:08 QUILT_PATCHES var can be used to set patches dir to whatever we want May 27 18:44:10 hehe May 27 18:44:12 thatd be cute May 27 18:44:16 hmm May 27 18:45:14 problem is when it collects patches from more than one location May 27 18:45:52 Well we could use relative paths in series file prolly May 27 19:21:01 03Martin Jansa  07org.openembedded.dev * rc9362d96d7 10openembedded.git/recipes/dri/glamo-dri-tests_git.bb: glamo-dri-tests: bump SRCREV a bit more May 27 19:25:42 03Michael 'Mickey' Lauer  07org.openembedded.dev * rc69ff6d9c0 10openembedded.git/recipes/htcdream/htcdream-gps_git.bb: htcdream-gps: new recipe; gps test program for the HTC Dream May 27 19:26:59 GNUtoo: please try with this recipe May 27 19:27:13 GNUtoo: if that works, i'll create an fsodeviced plugin May 27 19:27:28 ok I'll have to install something and then I'll do that May 27 19:27:31 thanks a lot!!! May 27 19:46:18 03Enrico Scholz  07org.openembedded.dev * r311bed0b40 10openembedded.git/classes/packaged-staging.bbclass: (log message trimmed) May 27 19:46:18 packaged-staging.bbclass: added missing space in find call May 27 19:46:18 Old find statement failed with something like May 27 19:46:18 $ find . \( -name XXX\) May 27 19:46:18 find: invalid expression; I was expecting to find a ')' somewhere but did not see one. May 27 19:46:19 Signed-off-by: Enrico Scholz May 27 19:46:20 Acked-by: Chris Larson May 27 19:55:12 re effem May 27 19:57:11 03Enrico Scholz  07org.openembedded.dev * r0d270547fd 10openembedded.git/lib/oe/utils.py: (log message trimmed) May 27 19:57:11 oe.utils: added param_bool() method May 27 19:57:11 This new function works like dict's get() method but converts the May 27 19:57:11 returned value to a boolean. May 27 19:57:11 It is to be used to interpret e.g. 'apply=yes' parameters in SRC_URI. May 27 19:57:12 Moved from base.bbclass into lib/oe/utils.py -kergoth May 27 19:57:13 Signed-off-by: Enrico Scholz May 27 19:57:14 03Enrico Scholz  07org.openembedded.dev * r7cb990de93 10openembedded.git/classes/patch.bbclass: May 27 19:57:14 patch.bbclass: use param_bool() May 27 19:57:14 This patch replaces a complicated if statement with a more simple May 27 19:57:15 code. May 27 19:57:15 Slightly altered to use oe.utils -kergoth May 27 19:57:16 Signed-off-by: Enrico Scholz May 27 19:57:16 Signed-off-by: Chris Larson May 27 19:57:17 03Enrico Scholz  07org.openembedded.dev * reb8ad2af3a 10openembedded.git/classes/base.bbclass: (log message trimmed) May 27 19:57:17 base.bbclass: fixed 'apply=no' behavior May 27 19:57:53 This patch evaluates content of 'apply' parameter and copies files to May 27 19:57:54 workdir when it is not a patch or the parameter is false. May 27 19:58:22 ensc: thanks May 27 20:01:17 CIA-1: my patch.bbclass changes are broken and the other two need rework (moving into util.bbclass or lib/oe) May 27 20:01:26 did not had time to change them May 27 20:01:43 eeek.. that's a bot :( May 27 20:02:24 broken? how so? May 27 20:02:38 already moved them into lib/oe/utils.py and adjusted the other two to use it May 27 20:02:45 if there's a bug in the function, let me know May 27 20:03:00 and yes, CIA reports commits pushed to the repository May 27 20:03:03 * kergoth_ hugs CIA-1 May 27 20:03:04 * CIA-1 hugs kergoth_ May 27 20:03:24 * denix slaps CIA-1 May 27 20:03:27 kergoth_: it fails on stuff like file://foo.patch.gz May 27 20:03:45 oh, you didn't bring the extension mangling over into the function May 27 20:04:08 * kergoth_ was quite careful to test nearly all the cases when he made the original change, but must've missed one when he adjusted unpack.bbclass May 27 20:04:19 * kergoth_ grumbles May 27 20:05:04 er.. hmm May 27 20:05:16 wait...https://www.cvg.de/people/ensc/x.patch May 27 20:05:35 what's broken about that? May 27 20:05:38 would be my new patch but I did not tested it yet May 27 20:05:46 * kergoth_ looks May 27 20:05:57 hi woglinde, all May 27 20:06:24 ensc: that damned logic has way too many variations May 27 20:06:29 it hurt my head getting it the way it was originally May 27 20:06:34 heh May 27 20:06:42 this is why we need actual unit tests May 27 20:06:48 yes May 27 20:07:15 I built a complex image with that new version and it suceeded May 27 20:07:31 ensc: btw, do you have commit rights? May 27 20:07:38 denix: no May 27 20:07:56 cbrake: ping May 27 20:08:18 denix: May 27 20:08:22 denix: yes May 27 20:10:39 ensc: pretty sure teh logic in base.bbclass did function properly, so i'm going to revert and leave that alone for the time being, and adjust the unpack function to check for if not patch in parm and apply != 'yes' May 27 20:10:46 that should take care of it, as far as i can tell May 27 20:10:52 cbrake: I thought someone already proposed to give ensc commit rights, but I maybe mistaken... May 27 20:10:54 can then start putting together unit tests to validate May 27 20:11:15 kergoth_: no, base.bbclass was broken (in 'apply=no' case) May 27 20:11:31 worked fine in all my tests May 27 20:11:34 which case was broken? May 27 20:11:40 vim-7.2 May 27 20:11:40 denix: could have slipped through the cracks -- are there any messages in oe-dev? May 27 20:11:43 er, i mispoke May 27 20:11:46 it broke dropbear (the nopw patch) May 27 20:11:47 i meant the patch.bbclass changes were fine May 27 20:13:01 kergoth_: the old version (without my patch) was ok. But I wanted to remove the code duplication (interpretation of 'apply=XXX') with my patch May 27 20:13:10 i agree May 27 20:13:20 but i think we should make unit tests before making further changes May 27 20:14:23 03Chris Larson  07org.openembedded.dev * r5ff088e9df 10openembedded.git/classes/base.bbclass: May 27 20:14:23 base.bbclass: fix issue with unpack of patches with apply=no May 27 20:14:23 Signed-off-by: Chris Larson May 27 20:14:25 03Chris Larson  07org.openembedded.dev * rac5931ec67 10openembedded.git/classes/base.bbclass: May 27 20:14:25 Revert "base.bbclass: fixed 'apply=no' behavior" May 27 20:14:25 This reverts commit eb8ad2af3a59dc7b4c64814815fc13977efaec6b. May 27 20:14:26 03Chris Larson  07org.openembedded.dev * r8c04a982e9 10openembedded.git/classes/patch.bbclass: May 27 20:14:26 Revert "patch.bbclass: use param_bool()" May 27 20:14:26 This reverts commit 7cb990de933f22331cd701ecb45d1e30dd8f3c11. May 27 20:14:45 i left the param_bool function in lib/oe/utils, just reverted the use of it for the time being May 27 20:16:40 denix: yes; Tom Rini wanted to bring me in (15.04.2010) May 27 20:19:21 ensc: was it on the mailing list? trying to find it... May 27 20:19:41 denix: I think ant reported it on IRC May 27 20:21:29 denix: http://article.gmane.org/gmane.comp.handhelds.openembedded/31985 May 27 20:22:09 khem: ls: cannot access /home/arm/.qui*: No such file or directory May 27 20:22:28 IgorK: whats your distro on host May 27 20:22:34 do I have to have it in order to apply non-unified patches? May 27 20:22:42 Gentoo May 27 20:23:24 IgorK: It applies fine here May 27 20:23:33 I use ubuntu 10.04 May 27 20:23:52 so I dont know what the problem really is on your end unless I can reproduce it May 27 20:24:03 btw. hmmm May 27 20:24:11 which quilt is it using May 27 20:24:25 do you have quilt in your build sysroot May 27 20:24:38 iow did it install quilt native May 27 20:38:51 khem: http://pastebin.com/BNQPQ462 May 27 20:44:03 IgorK: I am using 0.48 here May 27 20:44:11 IgorK: thats the difference May 27 20:45:08 so the best is to do "git pull" from the main OE directory, right? May 27 20:46:20 IgorK: do you have 0.46 in PREFERRED_VERSION in you local.conf ? May 27 20:46:49 IgorK: before that I want to find out why 0.46 is being pulled in May 27 20:47:26 paste your local.conf May 27 20:54:23 khem: here we go: http://pastebin.com/dYVsADRG May 27 20:57:34 re pb May 27 20:57:38 IgorK: hmm are you using org.openemededded.dev branch or something else May 27 20:59:16 IgorK: btw TARGET_OS = "linux" is not needed in your local.conf May 27 21:01:41 woglinde: hody May 27 21:01:51 insert w somewhere May 27 21:02:12 *g* May 27 21:05:26 khem: Im using git tree of mini2440 device: git://repo.or.cz/openembedded/mini2440.git May 27 21:05:42 khem: tried kernel with that flag.. but now it complains about missing init, but init is there (sysvinit), I'll try to rebuild whole image with that flag if it changes May 27 21:07:51 and using this instructions: http://www.electronics.diycinema.co.uk/embedded/mini2440/bitbaking-the-kernel-angstrom-and-qt4-embedded-all-at-once/ May 27 21:11:49 IgorK: is it tracking .dev I have no clue what that repo is May 27 21:12:09 JaMa: missing init ? May 27 21:12:22 khem: like missing /sbin/init May 27 21:12:27 JaMa: hmmm May 27 21:12:34 JaMa: 2.6.34 ? May 27 21:12:51 khem: http://paste.pocoo.org/show/219088/ May 27 21:13:08 yes 2.6.34, eglibc-2.12, gcc-4.5.0 etc May 27 21:14:11 JaMa: and same setup works otherwise ? May 27 21:14:26 you sure the commandline has init=/sbin/init in it May 27 21:14:30 same except gcc version May 27 21:15:13 JaMa: hmm thats strange May 27 21:15:19 I did not run into that May 27 21:15:45 khem: maybe it's faulty uSD... May 27 21:15:49 JaMa: did you patch makefile to add this option May 27 21:16:31 khem: because even after mkfs I had to untar new image to different partition because the one I used for gcc-4.5 tests before always ended with IO errors May 27 21:16:44 khem: which makefile? May 27 21:17:14 kernel makefule May 27 21:17:18 makefile May 27 21:18:20 khem: I would say yes, but not sure. the person to ask would be florian, but he is not here @ the moment May 27 21:19:28 khem: no, but I'm using kexecboot, mmt checking serial output May 27 21:22:31 anyway to check if it is? May 27 21:24:10 hi florian May 27 21:26:53 re ant May 27 21:27:20 hey May 27 21:28:09 pb_: hello there May 27 21:34:58 florian: is git://repo.or.cz/openembedded/mini2440.git tracking .dev? May 27 21:35:08 JaMa: your setup is different May 27 21:38:05 mickey|zzZZzz, 2 things on gps: May 27 21:38:26 1)be able to kill the program in order to stop gps May 27 21:38:44 2)change /dev/oncrpc/3000005b:90380d3d to /dev/3000005b:90380d3d: May 27 21:38:47 khem: FYI - this repo is for mini2440 computer based on the Samsung S3C2440 chip May 27 21:38:47 IgorK: alright this explains it http://repo.or.cz/w/openembedded/mini2440.git/tree/HEAD:/recipes/quilt May 27 21:38:58 IgorK: its a weird fork May 27 21:39:13 unless its merged upstream OE you wont find much support in this channel May 27 21:39:16 mickey|zzZZzz, ifyou're there let's talk in #openmoko-cdevel May 27 21:39:23 because not many devs might be using it May 27 21:39:39 :( May 27 21:39:50 IgorK: you should always mention what repo/branch you are using May 27 21:40:01 so to set expectations right May 27 21:40:17 OK, will do May 27 21:40:29 So what do u suggest? May 27 21:41:26 whatever you prefer. you can fix that repo with whatever problems you encounter or promote the changes it has into upstream oe May 27 21:42:13 many times such forks are done to maintain a given thing and you cant stretch it too much May 27 21:42:25 like it may be good for a given machine and given image May 27 21:42:38 but if you try to play around it might break May 27 21:42:59 best is to ask the maintainer of that fork what the plans are May 27 21:43:45 ok, will do, thx May 27 21:52:31 khem: about that udev_151 hack for kernels <2.6.27 May 27 21:53:00 some packaging doubts May 27 21:53:22 first: where to pull in the udev-static dependency May 27 21:54:03 second: why are udevadm and udevd packaged in dbg _? May 27 21:54:18 (thats' for udev-static_124) May 27 21:55:21 hi ant__ May 27 21:56:21 khem: I understand we don't want to overwrite the 151 files, just a parallel installation May 27 21:56:38 but how to do it properly? May 27 21:58:49 basically udev-static doesn't seem to package anything :/ May 27 22:01:39 GNUtoo: 2) ok, will do that. 1) what should be executed on SIGTERM? May 27 22:01:44 i.e. how do i turn it off? May 27 22:02:04 mickey|zzZZzz, where did you took the sources from? May 27 22:02:17 GNUtoo: your tarball May 27 22:02:43 it seem not the same that I have May 27 22:02:54 did you modify the sources? May 27 22:04:22 just removed some bogus includes and added some more to make it compile May 27 22:05:41 -> #openmoko-cdevel May 27 22:05:58 ok May 27 22:08:37 ant__: hmm let me check May 27 22:09:01 see logs on screen about files installed May 27 22:11:07 I think is do_install May 27 22:11:45 ant__: udevadm should be in udev-utils May 27 22:12:13 hm.. I see it in workdir after do_compile May 27 22:12:22 NOTE: /sbin/.debug/udevd-124-static May 27 22:12:23 NOTE: /sbin/.debug/udevadm-124-static May 27 22:12:24 ^^ May 27 22:12:27 ?? May 27 22:13:03 see packages-split/ May 27 22:13:10 but we set PACKAGES = "${PN}" May 27 22:13:28 packages-split/udev-utils/sbin/ May 27 22:16:45 ant__: hmmm oh you are talking about udev-static_124.bb May 27 22:16:46 ok May 27 22:16:51 now I am with you May 27 22:17:02 :D May 27 22:18:27 probably we should void the FILES_* included from udev.inc May 27 22:18:40 FILES_${PN} = "${base_sbindir}/*-${PV}-static" May 27 22:18:44 should be doing that May 27 22:19:03 but only for udev May 27 22:19:11 yeah try to void that May 27 22:19:18 we included: FILES_${PN}-dbg += "${usrbindir}/.debug ${usrsbindir}/.debug" May 27 22:21:51 hm..no May 27 22:22:02 FILES_${PN}-dbg = "${base_sbindir}/.debug/*-${PV}-static" May 27 22:22:28 add this to udev-static_124.bb May 27 22:24:26 still the do_install is strange... May 27 22:26:23 khem: still not packaged May 27 22:26:37 ant__: what is not packaged May 27 22:26:44 mom May 27 22:27:17 http://fr.pastebin.ca/1873150 May 27 22:28:16 but then... May 27 22:28:16 NOTE: package udev-static-124-r28.0: task do_package_write_ipk: Started May 27 22:28:17 NOTE: Not creating empty archive for udev-static-151-r13.5 May 27 22:28:23 :/ May 27 22:30:08 khem: ^^ why 151? must be 124! May 27 22:31:39 I dont have 151 appearing here May 27 22:31:53 it's default in my distro May 27 22:32:28 but you must be doing bitbake udev-static May 27 22:32:31 I'm tee'ing the log May 27 22:33:03 after -c clean May 27 22:33:09 do we need to package May 27 22:33:16 these remaining files too I wonder May 27 22:34:56 http://filebin.ca/wyggbf May 27 22:37:13 ant__: I can not open that file May 27 22:37:17 it says it bin file May 27 22:37:28 hm..text May 27 22:37:35 just d/l it May 27 22:37:44 stupid mime May 27 22:39:49 03Khem Raj  07org.openembedded.dev * rf692b724d9 10openembedded.git/recipes/linux/ (linux-2.6.34/qemuarm/defconfig linux_2.6.34.bb): May 27 22:39:49 linux_2.6.34.bb: Make it to support qemuarm May 27 22:39:49 Signed-off-by: Khem Raj May 27 22:40:36 khem: btw this was the patch (well, the alternative patched init) http://filebin.ca/pwosh May 27 22:41:29 ant__: I cant download on this computer May 27 22:41:39 so unless you paste something I can view in browser May 27 22:41:49 ok, I'll poste it on pastebin, np May 27 22:41:50 its useless for me atm May 27 22:41:59 blame .ca ! May 27 22:42:26 dont use that crap May 27 22:42:42 http://fr.pastebin.ca/1873162 May 27 22:42:52 I meant filesbin.ca ;) May 27 22:43:05 filebin, even May 27 22:47:06 ant__: patch looks ok May 27 22:47:18 hi May 27 22:47:54 now we have to deploy DAEMON="/sbin/udevd-124-static" UDEVADM="/sbin/udevadm-124-static" May 27 22:48:10 but udev-static seems to fail here May 27 22:48:21 I don't have those 2 in my image May 27 22:49:06 khem, I suspect the problems I face on htcleo are kernel related May 27 22:49:48 khem, but before I stop I would like to do one more thing and complete the test without neon. as you have seen the build failed May 27 22:51:54 ant__: hmm you have install it DISTRO_DEV_MANAGER = "udev-static" may be May 27 22:52:30 ant__: or you just have to add it to IMAGE_INSTALL May 27 22:52:32 I'm looking in run.do_install May 27 22:53:11 I've added it as EXTRA_IMAGEDEPENDS += May 27 22:53:34 but perhaps udev itself could check kernel version May 27 22:53:47 and decide if install this static May 27 22:54:16 $ ar p /scratch/oe/deploy/eglibc/ipk/armv5te/udev-static_124-r27.0_armv5te.ipk data.tar.gz | tar -tvz May 27 22:54:19 drwxr-xr-x root/root 0 2010-05-27 15:23 ./ May 27 22:54:22 drwxr-xr-x root/root 0 2010-05-27 15:23 ./sbin/ May 27 22:54:24 -rwxr-xr-x root/root 538292 2010-05-27 15:23 ./sbin/udevd-124-static May 27 22:54:27 -rwxr-xr-x root/root 548316 2010-05-27 15:23 ./sbin/udevadm-124-static May 27 22:54:57 ah, btw it's not machine specific May 27 22:55:03 nope May 27 22:55:09 not so far May 27 22:55:16 well, udev is May 27 22:55:47 ant__: no May 27 22:55:52 here... May 27 22:56:07 probably because of some machine overrides May 27 22:56:10 ant__, my I ask why you are building that specific version of udev and for which device? May 27 22:56:15 ant__: thats possible May 27 22:56:29 udev 151 fails on <2.6.27 May 27 22:56:36 dcordes: I doubt if its related to kerne May 27 22:56:53 dcordes: if kernel can boot other root file systems it could do that for OE too May 27 22:57:13 well the busybox isn't really demanding.. May 27 22:57:16 until we move to 2.6.34 we have to live with this hack (or use other distro) May 27 22:57:19 dcordes: one more thing you could try though May 27 22:57:38 dcordes: you could use armvt5t userspace and try to boot into that May 27 22:57:48 dcordes: this is for zaurus but is generic issue May 27 22:57:54 dcordes: it wont have any confusion about neon or vfp May 27 22:58:23 khem, I tried to use builds from htcraphael. these machines use the armv6-novfp tunefile May 27 22:58:27 khem, they fail May 27 22:58:43 whats htcraphael ? May 27 22:58:52 msm7201A based machine May 27 22:59:01 ah they fail too May 27 22:59:15 similar to OE ? May 27 22:59:48 ant__: why does udev 151 fails on <2.6.27 May 27 22:59:55 lacking headers May 27 23:00:03 is it ppoll issue May 27 23:00:18 khem: pls try to prefer udev_151 in your build May 27 23:00:26 ant__: I have it May 27 23:00:29 the issue is not apparent if you use 124 possibly May 27 23:00:32 and it works well on qemu May 27 23:00:37 I don't have the static... May 27 23:00:42 s@t May 27 23:00:49 khem: here is boot output http://pastebin.ca/1873165 image was whole built with this patch applied http://gitorious.org/~jama/angstrom/jama-shr-experimental/commit/788e17eb98cf76c5edbe2f3e0129027d8f8e8329 (Added it also to BUILD_OPTIMIZATION hoping it won't harm..) May 27 23:00:51 I am using 151 and 2.6.34 on qemu arm May 27 23:01:38 but now I've to go sleep.. May 27 23:03:16 khem: see here :/ http://fr.pastebin.ca/1873166 May 27 23:03:17 JaMa: before you sleep May 27 23:03:26 JaMa: did you compile kernel with this option May 27 23:03:28 btw you build minimal/eglibc, isnt? May 27 23:03:29 prolly not May 27 23:03:43 JaMa: here is deal May 27 23:03:50 compile the kernel manually May 27 23:04:12 unless you can make sure that the option was used to compile the kernel too May 27 23:04:37 ant__: yes May 27 23:04:59 here is Angstrom/glibc fwiw May 27 23:05:24 somehow 151 steals the task :/ May 27 23:06:25 r13.5 May 27 23:06:27 khem: ah I see what you mean.. sorry I was already sleepy before.. May 27 23:07:02 I modifies the toplevel Makefile May 27 23:07:04 khem: you were talking about this flag added to kernel makefile (not about init= :)) May 27 23:07:14 JaMa: yes May 27 23:07:49 sorry I misunderstood before.. rebuilding now May 27 23:07:56 ifdef CONFIG_CC_OPTIMIZE_FOR_SIZE May 27 23:07:56 KBUILD_CFLAGS += -Os May 27 23:07:56 else May 27 23:07:56 KBUILD_CFLAGS += -O2 #-fno-ipa-sra May 27 23:07:56 endif May 27 23:08:08 and i think you build with CONFIG_CC_OPTIMIZE_FOR_SIZE May 27 23:08:23 so add -fno-ipa-sra to first KBUILD_CFLAGS May 27 23:08:39 ie KBUILD_CFLAGS += -Os -fno-ipa-sra May 27 23:08:49 or to both it does not matter May 27 23:09:45 even if its passed in thru CC , KBUILD_CFLAGS will be appended to commandline May 27 23:09:57 building.. May 27 23:09:58 which means the effect will be nullified once you specify -Os or -O2 again May 27 23:10:09 last -O rules May 27 23:10:24 if you have -O2 -O3 -O1 -O0 May 27 23:10:28 then gcc takes in O0 May 27 23:10:55 and O2 implied -fipa-sra May 27 23:11:08 or even -Os does same May 27 23:12:15 hmm night backup already started, build is getting slower.. May 27 23:12:29 in nutshell -fno-ipa-sra should appear in the very end of compiler commandline May 27 23:12:40 JaMa: ok it take few mins May 27 23:12:58 JaMa: are you GMT +2 May 27 23:13:43 must be late there May 27 23:14:04 1:12 AM.. May 27 23:14:32 but I promised boss to arrive in time tomorrow for presentation... so I'll wake up around 6am :/ May 27 23:15:36 khem: I found udev bot in workdir/armv5te and workdir/c7x0 what a mess... May 27 23:15:44 s/bot/both/ May 27 23:16:32 khem, sorry. major 3g outage May 27 23:17:50 khem, are you still there? May 27 23:19:22 khem: boot looks the same, I'll recheck if that flag was applied right (tomorrow) May 27 23:20:03 JaMa: ok May 27 23:20:09 good night May 27 23:20:19 bye JaMa May 27 23:20:28 gnight May 27 23:26:35 khem: http://paste.pocoo.org/show/219129/ is part of log.do_compile with make V=1 showing it was applied.. May 27 23:28:36 khem, you think it's worth the try completing the 'cortex-a8 no-neon' build ? May 27 23:28:42 khem, I did not understand why it failed May 27 23:30:12 dcordes: prolly yes May 27 23:30:18 tell me waht was error May 27 23:31:24 khem, I will dig it out. 1minute May 27 23:32:34 http://tinderbox.openembedded.org/public/logs/task/5923332.txt May 27 23:33:22 ah ok May 27 23:33:28 get rid of this -mfloat-abi=softfp May 27 23:34:35 machine/include/tune-cortexa8.inc May 27 23:34:54 machine/include/tune-armv7.inc actually May 27 23:34:59 TARGET_CC_ARCH May 27 23:35:19 make it TARGET_CC_ARCH = "-march=armv7" May 27 23:36:10 and in machine/include/tune-cortexa8.inc May 27 23:36:24 make TARGET_CC_ARCH = "-march=armv7-a -mtune=cortex-a8" May 27 23:36:54 this should fix it hopefully May 27 23:37:17 right now I have this May 27 23:37:20 -TARGET_CC_ARCH = "-march=armv7-a -mtune=cortex-a8 -mfpu=neon ${ARM_FP_OPT}" May 27 23:37:20 +TARGET_CC_ARCH = "-march=armv7-a -mtune=cortex-a8 ${ARM_FP_OPT}" May 27 23:37:46 ARM_FP_OPT is no good May 27 23:38:48 should I keep that? May 27 23:38:59 khem: after one beer I see my issue May 27 23:39:09 udev-static is bad split May 27 23:39:37 khem, hold on.. I think this is leading nowhere. I bet I will end up getting something confused. can you perhaps post a patch? May 27 23:40:04 dcordes: hmm follow the steps I said above May 27 23:40:23 you need to modify two files machine/include/tune-cortexa8.inc and machine/include/tune-armv7.inc May 27 23:40:33 and what you need to change is there May 27 23:40:37 khem: my workdir seems insane. May 27 23:40:44 ant__: heh May 27 23:40:48 I'll rebuild from scratch and let you know May 27 23:40:57 ant__: sure May 27 23:41:08 ok right you put the full lines sorry May 27 23:41:11 I have a bogus udev-151-r13 in /armv5te May 27 23:41:13 but I think we generate -static automatically May 27 23:41:26 so udev-static may be a bad choice May 27 23:41:31 for the recipe May 27 23:42:20 although it fits the purpose May 27 23:42:46 I have even udev-151-r11 in /armv5te :/ May 27 23:42:56 /scratch/oe/work/armv5te-oe-linux-gnueabi/udev-151-r11/packages-split May 27 23:43:03 I have udev-static :) May 27 23:43:20 along with others like udev-dev aand udev-doc etcx. etc. May 27 23:44:03 khem, http://privatepaste.com/b29f722536 <- this is the diff with all the relevant changes. can you take a look? May 27 23:44:22 ant rename the recipe to something like udev-nodynamic May 27 23:44:52 khem, gonna build minimal-image with this in f692b724d9cb01cfa6f3bc6a5f171084b1cc503c May 27 23:44:55 dcordes: perfect May 27 23:46:29 ant__: http://pastebin.com/tkj3tRbu potential problems :) May 27 23:46:32 have another beer May 27 23:48:10 now is renamed udev-hack_124.bb ;) May 27 23:49:13 khem: about those xy-static recipes....I'm guilty ;) May 27 23:49:48 probably one should create the generic recipe and hack around the dev, dbg, static May 27 23:50:02 but here it was clearly a klibc static binary May 27 23:50:43 ant__, why are you building udev static? is that also something kernel related? May 27 23:51:09 hm...really we just want the old udevadm and udevd May 27 23:51:28 same toolchain May 27 23:56:43 ant__: I would say patch kernel to work with new udev May 27 23:57:28 recipes/linux-libc-headers/linux-libc-headers-2.6.31/0001-implement-TIF_RESTORE_SIGMASK-support-and-enable-the.patch May 27 23:57:32 is the patch you need May 27 23:57:57 khem: hm, interesting May 27 23:58:11 Are the madwifi drivers finally in any recent Linux kernels yet? May 27 23:58:12 was just temporarly until all devices move to 2.6.3x May 27 23:58:42 (the out-of-kernel madwifi-ng modules won't build with gcc 4.4.2, and gcc 4.2.4 won't build anymore since we did the sysroot thing :( ) May 27 23:58:45 ant__: this patch is not merged in upstream yet I think May 27 23:58:53 so even 2.6.34 will need it May 27 23:59:20 mwester-laptop: whats the error May 27 23:59:38 We can fix the module May 28 00:00:18 I didn't even look at the error in madwifi -- I'd bet lunch that it's the closed-source hal.o binaries. May 28 00:00:53 Boost also failed with 4.4.2, but I don't use that so that's somebody else's problem, I think. :D May 28 00:02:24 I was hoping that the madwifi drivers made it into the kernel finally, which would resolve this issue for me. I can't fix the toolchain problem, and I'd rather not waste more time on those silly madwifi-ng drivers if I can spend it on something more productive, like moving the NSLU2 to a newer kernel instead. :) May 28 00:02:28 khem, I really hope this will build and lead to some useful new conclusion. need to get somewhere with this.. thanks again for the efforts May 28 00:03:01 hitting the sheets May 28 00:03:03 good night May 28 00:04:26 ant__: oh the patch I mentioned has made upstream I am seeing it in 2.6.34 so you will be good May 28 00:04:46 good, we have 2.6.34 on some machines May 28 00:05:10 ant__: and make sure you use 2.6.31 for linux-libc-headers May 28 00:06:44 now I'm battling with the renamed recipe...sob the inc sets too much May 28 00:07:05 cp: cannot stat `/oe/openembedded/recipes/udev/files/c7x0/flags.patch': No such file or directory May 28 00:07:19 seems have to edit FILESPATHPKG or alike May 28 00:08:14 initscripts has an example that sounds sorta like what you need to fix that. May 28 00:09:20 mwester-laptop: no idea about madwifi-ng May 28 00:09:28 my guess is its not merged yet May 28 00:11:16 my bad..did not rename the dir :P May 28 00:13:31 bingo May 28 00:13:41 khem: once renamed all is fine May 28 00:14:03 packaging task was mixed up.. May 28 00:14:47 now, renaming it to udev-compat-static? May 28 00:21:55 ok, good night May 28 00:53:43 khem: do u have wright rights to that git repository? I fixed the patch file and wants to make sure it is updated.... **** ENDING LOGGING AT Fri May 28 03:00:14 2010