**** BEGIN LOGGING AT Tue May 25 02:59:57 2010 May 25 06:19:54 hi geeks May 25 06:20:46 I am going to write a net bootloader for PIC18 micro-controller.... I dont have any idear how to proceed for it May 25 06:20:53 can anybody suggest me for this May 25 06:52:51 HenryHamdan: why would you go for that? it's a limited microcontroller, besides that it can't execute from ram... May 25 07:00:00 HenryHamdan, plus, you are in the wrong channel. I daresay openembedded isnt useful for PIC chips. May 25 07:03:41 does pic have something 32-bit like atmel have avr32? May 25 07:05:07 Linf, yes PIC32 are based upon the MIPS core May 25 07:05:07 Linf, yeah. http://www.microchip.com/pic32/ May 25 07:09:51 morning May 25 07:10:18 good morning May 25 07:11:18 03Martin Jansa  07org.openembedded.dev * r6b01320f0f 10openembedded.git/recipes/vala-terminal/vala-terminal_git.bb: May 25 07:11:18 vala-terminal: bump SRCREV May 25 07:11:18 Signed-off-by: Martin Jansa May 25 07:11:19 03Martin Jansa  07org.openembedded.dev * redc90bdfa9 10openembedded.git/recipes/linux/ (15 files in 2 dirs): May 25 07:11:19 linux-openmoko-2.6.34: add rebased gdrm and qtmoko patches, use gta02_drm_defconfig again May 25 07:11:19 Signed-off-by: Martin Jansa May 25 07:20:23 denix: did you ever manage to resolve the external toolchain issue you got quite a while back? (http://www.mail-archive.com/openembedded-users@lists.openembedded.org/msg00406.html) I'm having the very same issue at the moment May 25 07:25:47 03Stefan Schmidt  07org.openembedded.dev * ref20f4f572 10openembedded.git/conf/machine/imote2.conf: May 25 07:25:47 imote2.conf: Only define the eraseblock size once May 25 07:25:47 The second definition overlayed the first one. It was consistent, but we still May 25 07:25:47 want to only have the correct one. May 25 07:44:13 03Stefan Schmidt  07org.openembedded.dev * r404e5eea98 10openembedded.git/.gitignore: gitignore: Ignore *.pyc from oeaudit May 25 07:44:15 03Stefan Schmidt  07org.openembedded.dev * rf604aed34f 10openembedded.git/recipes/ibrdtn/ibrdtn.inc: May 25 07:44:15 ibrdtn.inc: Remove DP = -1 May 25 07:44:15 The new versions have been tested enough now to be the new default. May 25 07:44:16 03Stefan Schmidt  07org.openembedded.dev * r4efe6de5a7 10openembedded.git/recipes/images/imote2-image.bb: May 25 07:44:16 imote2-image: The main ibrdtn package is now called ibrdtnd. May 25 07:44:16 The new version is split into different modules. To drag everything in we want May 25 07:44:17 we include ibrdtnd now. May 25 07:47:33 03Steffen Sledz  07org.openembedded.dev * rde05c6b574 10openembedded.git/recipes/u-boot/ (2 files in 2 dirs): May 25 07:47:33 u-boot-2009.03: abortboot behaviour changed for hipox machine May 25 07:47:33 * ignore all chars sent before boot prompt May 25 07:47:33 * accept only '.' to abort boot May 25 07:47:33 Signed-off-by: Mario Schuknecht May 25 07:47:34 Acked-by: Steffen Sledz May 25 07:58:00 good morning May 25 08:00:22 hi florian May 25 08:49:41 Hi, I'm having some trouble with grep-native, QA issues with libform.la, libgettextlib.la and others May 25 08:49:55 Anyone noticed something similar? May 25 08:51:23 not me. dev head? May 25 08:53:02 dev head from e few days ago but I don't see any significant changes to could have anything to do with grep-native May 25 08:53:47 checked grep-native bb file, it just installs grep, egrep and fgrep executables. Where do the libform.la and others come from? May 25 08:54:11 morning May 25 08:54:26 janp find tmp | grep libform May 25 08:54:37 ~seen bluelightning May 25 08:54:38 bluelightning <~bluelight@pdpc/supporter/professional/bluelightning> was last seen on IRC in channel #handhelds, 50d 9h 29m 37s ago, saying: 'mzb: cya'. May 25 08:54:44 hm... May 25 08:55:00 I've seen one libopie2 issue May 25 08:55:15 headers are not staged May 25 08:55:44 and recipes are failing (first one is opie-button-settings) May 25 08:56:19 recipes RIP since 2009, so must be som eother change... May 25 08:57:11 which broke libopie2 May 25 09:00:20 eFfeM_work: they all are from gettext-native and ncurses-native May 25 09:01:24 janp, i'd say something is wrong there, you could try a bitbake -cclean for those pacakges to see if a rebuild fixes things, if not probaly pastebin a log somewhere May 25 09:02:56 eFfeM_work: As far as I can tell, the QA issue seems to be that my workdir is somewhere in these files and that is indeed the case May 25 09:03:33 janp, does your TMPDIR var have a / at the end ? May 25 09:04:34 if so, remove May 25 09:04:42 eFfeM_work: indeed: TMPDIR = ${OEDIR}/build/tmp/${FEED_ARCH}/ May 25 09:05:04 janp, remove the trailing /, rm tmpdir and rebuild from scratch May 25 09:05:48 not too sure about the details any more, but your error msg will have something like // in it which in the end triggers the error May 25 09:06:59 yes, some sed goes wrong with that slash May 25 09:07:14 eFfeM_work: yes, that error message has indeed // in it. I'll rebuild from scratch and let you know if it works May 25 09:15:42 * mckoan is trying to build Apache+PHP5 May 25 09:18:27 mckoan: i suggest not to, tried that too in the past, eventually moved to lighttpd May 25 09:18:34 apache is a can of worms May 25 09:20:21 eFfeM_work: actually looks like Apache2 is working May 25 09:20:38 PHP lacks of config files May 25 09:21:04 mckoan: ah ok (it might be XorA|gone looked into php in the past) May 25 09:21:45 Iam working on overo desktop image. I wish to intergrate the gstreamer (with ti dsp support) plugin in openembedded May 25 09:22:54 Will any one guide me the steps need to modify and include needded in the open-embedded recepes ? May 25 09:23:33 Ananth: recipes/gstreamer is already present in OE May 25 09:35:26 zecke: AFAYK is it php working? May 25 09:36:32 no idea May 25 09:37:21 zecke: from git log seems that you sent a lot of patches recently May 25 09:39:17 zecke: so I supposed you knew the package May 25 09:41:22 actually there is a weird bug in php_5.2.13.bb May 25 09:41:47 ...php-5.2.13/.pc/CVE-2010-0397.patch/ext/xmlrpc/tests/bug51288.phpt: Permission denied May 25 09:42:18 due to missing file properties ---------- May 25 09:42:56 I don't know how a patch cold create a file with mode 0x0 May 25 09:43:04 s/cold/could May 25 09:46:37 mckoan: I had one similar issue May 25 09:47:08 ant_work: with php recipe or another one? May 25 09:47:11 file was there but 000 0b May 25 09:47:15 other May 25 09:47:24 parallel build May 25 09:47:34 ? May 25 09:47:34 Hi all, Will you any one guide me integrating the gstreamer (with ti dsp support) into omap-desktop-image (overo board from gumstix) May 25 09:47:45 mckoan: I hav ethe log at home May 25 09:48:12 ant_work: how did you solve it May 25 09:48:21 just rebuilding May 25 09:48:39 seems like a race issue, file is accesssed before write is finished? May 25 09:48:46 * mckoan is trying May 25 09:49:33 ant_work: which would be the proper solution to this kind of issue? May 25 09:53:22 best would if others were able to reproduce May 25 09:53:31 ant_work: then I suspect a matter of bitbake multithread management May 25 09:53:35 I build with -j5 and 4 threads May 25 09:54:06 ant_work: I just tested launching bitbake php twice, unsuccessfully May 25 09:54:22 I have to chmod 666 ...php-5.2.13/.pc/CVE-2010-0397.patch/ext/xmlrpc/tests/bug51288.phpt May 25 09:54:33 then relaunch it May 25 09:57:12 mckoan: I found the log, was: May 25 09:57:14 grep: /oe/build/tmp/work/i686-linux/bzip2-full-native-1.0.5-r1/bzip2-1.0.5/.pc/bzip2-1.0.5-autoconfiscated.patch/configure.ac: Permission denied May 25 09:57:55 http://pastebin.com/ckhvd8WA May 25 10:03:08 mckoan: You had to chmod on the patch? May 25 10:10:42 zecke: I manually chmod and relaunch, yes a chmod on the pach would solve the problem May 25 10:12:49 mckoan: do you know which app is having an issue with the patch? May 25 10:13:04 mckoan: is it quilt or patch? May 25 10:14:49 zecke: recipes/php/php_5.2.13.bb SRC_URI += "file://pear-makefile.patch;patch=1 \ file://imap-fix-autofoo.patch;patch=1 \ file://CVE-2010-0397.patch;patch=1 \ May 25 10:15:05 file://CVE-2010-0397.patch;patch=1 May 25 10:20:15 bb in 1h May 25 12:05:41 re May 25 12:52:03 lool: I got gcc built with dh_isntall instead of dh_movefiles May 25 12:53:11 * JaMa gives hrw coffee May 25 12:53:58 rebuilding again now May 25 12:55:42 ops,wrong window indeed May 25 14:09:13 hi May 25 14:10:05 while building an image I noticed psplash's OE logo is the old one : is there a high (320x240 ;-) resolution logo available somewhere ? May 25 14:10:21 sorry : a high resolution file of the new logo May 25 14:11:49 morning May 25 14:12:16 hi kergoth May 25 14:12:22 ericben: I think not May 25 14:12:54 mckoan: ok I'll do with the better on I found, it's a little bit smaller than previous one but it's nice on a QVGA screen May 25 14:15:28 contrib/artwork/oe.svg May 25 14:15:35 generate any size you need from it May 25 14:15:43 hrw: thanks May 25 14:15:50 and prepare/send patch for psplash to update it with new logo May 25 14:17:59 hrw: patch in process ;-) May 25 14:18:48 or in progress maybe, don't know May 25 14:20:54 morning all May 25 14:35:58 hrw: should I push the patch or send it to the list for ack ? May 25 14:41:24 if it builds then I would push May 25 14:43:30 hrw: OK I just sent it to the list, will push in a few hours May 25 14:44:51 khem: thanks for ACKs May 25 15:03:10 hey RP are you in bay area these days ? May 25 15:03:15 JaMa: np May 25 15:04:24 khem: I was at one point but back in the UK now May 25 15:05:55 ok May 25 15:11:33 what are the preconditions for a "preferred version" to be available? i'm trying to build glib-2.0-native as 2.16.3, i see the bb files, but it tells me that version is not available as a warning May 25 15:11:50 bitbake 1.8? May 25 15:12:17 oh, i'm a moron, didn't see the -native bb files May 25 15:37:00 eFfeM_work: external toolchain support has been improved dramatically since then - it's the primary mode of operation for my distro and many other people use it in theirs too May 25 15:55:09 RP: next time you come to bay area let me know. We can meet and have a meal together May 25 15:56:12 login: no valid shadow password May 25 15:56:21 and then it kicks me out May 25 15:56:25 how to fix it May 25 15:56:51 khem: will do although I don't know when that will be! :) May 25 15:57:54 RP: ok. not this summer ? May 25 15:58:09 Do you come to Intel Santa Clara Bldg 3 :) May 25 15:58:25 Bldg 12 on Juliette Lane May 25 15:58:34 thats where I was when I was working for Intel May 25 16:37:53 khem: I was in SC12, yes May 25 16:38:06 khem: Are you anywhere near there? May 25 16:39:03 03Koen Kooi  07org.openembedded.dev * re3157aad81 10openembedded.git/recipes/alsa/ (alsa-state.bb alsa-state/beagleboard/asound.state): alsa-state: update beagleboard state file for 2.6.32-psp May 25 16:47:39 With the latest recepie of omap3-desktop-image the gui image is not loading May 25 16:47:44 And in reboot produces Unhandled fault: external abort on non-linefetch (0x1028) at 0xfa050040 May 25 16:47:51 ] (dss_select_dispc_clk_source+0x20/0x40) from [] May 25 16:47:58 (omapdss_dpi_display_disable+0x20/0x50) May 25 16:48:04 [] (omapdss_dpi_display_disable+0x20/0x50) from [] May 25 16:48:12 (samsung_lte_panel_disable+0x28/0x34) May 25 16:48:17 Did any one came across this problem ? May 25 16:48:39 In the booting sequence iam able to observe loading progress bar but after end of that gpe-dm ends with error unable to open display 0 May 25 18:16:09 RP: Yes I am close probably 3 miles from there is my office May 25 18:16:16 and 10 miles away is my home May 25 19:10:00 gm May 25 19:10:02 re May 25 19:11:02 florian: does Nils have IRC presence? Would like to ask him about TK71 PCIe MSI support. May 25 19:13:04 likewise: usually not, but i can ping him... he forgot to answer your mail? May 25 19:14:59 florian: think so, or I just missed it somehow May 25 19:15:30 I had an email box full moment, most servers would retry though May 25 19:15:52 likewise: he had a full mailbox after his vacations :) May 25 19:16:15 (no quota involved here) May 25 19:16:20 florian: still thinking about an PCIe module that would fit the board perfectly, we have the project on-hold because we haven't seen MSI interrupts work on the Kirkwood yet. May 25 19:17:16 likewise: but apart from this PCIe is working for you? May 25 19:18:37 Hi likewise howdy May 25 19:18:39 I wonder how much hardware specific bits are involved in MSI... sounds like it is not much. May 25 19:18:49 But I have to admit I have no idea... May 25 19:19:41 anyone attends ARM DevCon in Santa Clara this october ? May 25 19:19:49 florian: zero, sometimes the message goes into the interrupt dispatcher hardware, but that's about the only stuff h/w related May 25 19:20:04 khem: hi. googl'ing... May 25 19:20:54 hi zecke May 25 19:21:57 likewise: sounds like it should not cause problems indeed... May 25 19:23:56 khem: not sure, I'ld rather go to more Linux related stuff instead. May 25 19:26:00 likewise: it has open source technology tracks May 25 19:26:15 likewise: I am proposing a talk on embedded lnx :) May 25 19:26:24 likewise: if I could prepare an abstract soon May 25 19:26:36 hello zecke May 25 19:27:07 likewise: From what we know the PCIe works pretty well now. But no idea if someone tried hardware that needs MSI. May 25 19:27:49 * khem is using python in gdb these days May 25 19:27:59 its damn powerful May 25 19:28:06 03Eric Bénard  07org.openembedded.dev * rfab0af3aef 10openembedded.git/recipes/psplash/ (files/psplash-poky-img.h psplash.inc): May 25 19:28:06 psplash: update OE logo May 25 19:28:06 * use the shiny new logo May 25 19:28:06 Signed-off-by: Eric Bénard May 25 19:28:06 Acked-by: Khem Raj May 25 19:29:11 florian: would love to hear what device you tested (or the dmesg output when the driver was loaded) May 25 19:29:28 florian: I'll bother Nils again by email :-) May 25 19:29:38 khem: python IN gdb? May 25 19:29:51 yep May 25 19:30:14 khem: you mean, gdb interprets python? May 25 19:30:19 sure May 25 19:30:38 you just type python on gdb prompt May 25 19:30:43 * likewise has been away from debugging using GDB for too long. printk is my favorite for device drivers :-) May 25 19:30:46 and it goes into python shell May 25 19:30:47 likewise: ok... May 25 19:31:02 likewise: I like it for conditional break points May 25 19:31:08 likewise: do you know about a device that would be useful for testing? May 25 19:31:14 like I want to break only when a function is called from a given function May 25 19:31:21 say memcpy call May 25 19:32:07 cond 1 $(gdb.current_frame().get_prev().get_name() == "my_func_who_calls_memcpy") May 25 19:32:38 then it will only hit bkpt 1 if calling function is my_func_who_calls_memcpy May 25 19:32:44 what a help May 25 19:34:00 likewise: is there a website where lnx based conferences are listed May 25 19:34:13 likewise: I always know when it has finished :) May 25 19:34:26 or rather come to know May 25 19:35:05 khem: I never check those sites, I only know http://elinux.org/Events May 25 19:35:18 khem: usually I find out through the usual spam emails :-) May 25 19:35:27 likewise: cool May 25 19:35:33 I even dont get span May 25 19:35:35 spam May 25 19:35:41 khem: Ah damn, just missed this one :-) Embedded Linux Conference 2010 - San Francisco, California, USA, April 12-14, 2010 May 25 19:35:43 google filters are too strong May 25 19:35:52 heh May 25 19:35:57 I missed it too May 25 19:36:02 I was in India May 25 19:36:18 khem: seems very incomplete though for 2010... May 25 19:36:36 Linux Plumbers Conference will take place in Cambridge, MA on November 3-5, 2010. May 25 19:37:46 http://elinux.org/Events doesnt list OLS May 25 19:38:11 yeah its not that populated May 25 19:38:47 Also in Cambridge (different side of the ocean) CELF Embedded Linux Conference Europe, October 27 & 28, 2010, Cambridge, UK May 25 19:39:57 right. heh May 25 19:44:04 khem: http://elinux.org/Events does list OLS now :-) May 25 19:44:46 likewise: heh cool May 25 19:44:49 khem: at least, if you meant http://www.linuxsymposium.org/2010/ May 25 19:44:52 I wish you win the netbook May 25 19:45:15 yes ols = ottawa lnx sympsm May 25 19:46:27 likewise: how about http://linux.org.au/LCA May 25 19:48:26 likewise: http://events.linuxfoundation.org/ lists some too May 25 19:49:49 likewise: http://www.socallinuxexpo.org/scale8x/ May 25 19:49:59 oh its becoming like ATP calendar May 25 19:50:07 with few grand slams May 25 19:52:41 likewise: this one is close to you http://lac.linuxaudio.org/2010/ May 25 19:55:43 03Chris Larson  07org.openembedded.dev * rbf7d0467a0 10openembedded.git/ (2759 files in 1086 dirs): May 25 19:55:43 Rename url params patch=/pnum= to apply={yes,no}/striplevel= May 25 19:55:43 I think this makes the behavior rather more clear. May 25 19:55:43 Signed-off-by: Chris Larson May 25 19:55:43 Acked-by: Denys Dmytriyenko May 25 19:55:44 Acked-by: Khem Raj May 25 19:56:18 03Chris Larson  07org.openembedded.dev * r6fe7cef270 10openembedded.git/ (2726 files in 1078 dirs): May 25 19:56:18 Make the do_patch apply=yes param implicit if extension is .diff/.patch May 25 19:56:18 For .diff/.patch you need to apply manually, you can specify apply=no. May 25 19:56:18 Signed-off-by: Chris Larson May 25 19:56:19 Acked-by: Denys Dmytriyenko May 25 19:56:19 Acked-by: Khem Raj May 25 19:57:48 * kergoth bets that'll bite some people doing merges, sorry May 25 20:03:04 Fight fight fight! Bite bite bite! The Itchy and Scratchy Show! May 25 20:05:32 kergoth: I am also affected but its ok :) May 25 20:05:44 I have to merge my gettext stuff May 25 20:05:46 easy merge, thankfully May 25 20:06:04 kergoth: yeah May 25 20:06:30 kergoth: zap_root_password did not help May 25 20:06:36 with my login woes May 25 20:06:40 huh May 25 20:07:55 login: no valid shadow password May 25 20:08:06 thats what I get with console-image May 25 20:08:23 with native-sdk-image it even does nt say that May 25 20:09:02 I booted into shell changed the password on nfs mounted rfs but to no avail May 25 20:15:41 hi khem, do you have /etc/shadow ? May 25 20:17:55 ebenard: yes May 25 20:19:35 khem: I had the same problem a few weeks ago, I'm searching how I solved it May 25 20:20:43 ebenard: ok. Let me know how you got away with it May 25 20:24:20 khem: you are trying to log as root over serial port ? May 25 20:27:07 ebenard: yes May 25 20:27:21 here is what I did but it should be solved in OE : http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-March/018394.html May 25 20:32:01 ebenard: sounds on similar lines May 25 20:32:17 but I am building like yesterdays snapshot May 25 20:32:37 khem: try to modify inittab as I did May 25 20:32:50 to put the right patch to login May 25 20:33:02 yeah I am trying that May 25 20:37:36 ebenard: my /bin/login is already symlinked to /bin/login.shadow May 25 20:37:50 ebenard: which is same as /us/bin/login May 25 20:38:06 and changing the way you suggested did not work for me May 25 20:38:11 it seems something else is messed up May 25 20:38:27 khem: ok which distro/image are your using ? May 25 20:38:39 minimal/native-sdk-image May 25 20:41:44 anyone tried 720p H.264 video with using the DSP gstreamer plugin or anything using the DSP? May 25 20:41:46 khem: ok I'll launch the compilation here to test. One customer told me he had a strange thing with passwords & shadow so this may have something to see with what you meet May 25 20:44:21 ebenard: minimal-image seems to work fine May 25 20:44:34 ebenard: because it does not use shadow May 25 20:49:11 tom__: I guess you should ask on #beagle May 25 20:49:28 tom__: assuming that's your target device :-) May 25 20:49:42 khem: do you have IMAGE_LOGIN_MANAGER = "shadow" set somewhere ? I have this in my image May 25 20:49:46 how's the weather in california? May 25 20:50:34 likewise: weather is not usual its cold May 25 20:51:35 ebenard: I dont have it May 25 20:52:25 which means IMAGE_LOGIN_MANAGER ?= "tinylogin" May 25 20:52:28 it seems when this is not set tinylogin is installed as a default (maybe I'm wrong here) May 25 20:52:29 is set for me May 25 20:52:37 you are right May 25 21:00:29 khem: http://www.raiden.net/shows/ May 25 21:02:31 likewise: thats cool. Is that your website btw ? May 25 21:04:18 khem: nope. mine is www.sidebranch.com May 25 21:05:50 likewise: you have improved it a lot May 25 21:09:28 khem: did i? May 25 21:09:45 last time I checked was like a year ago May 25 21:09:54 khem: I s*ck at maintaining web sites. May 25 21:10:43 now that you are independent you need it even more :) May 25 21:10:48 where is jobs section :) May 25 21:12:58 re May 25 21:13:34 hey pb_ May 25 21:14:33 hi pb_ May 25 21:20:47 khem: you have to apply through IRC :-) May 25 21:21:03 likewise: heh May 25 21:21:12 khem: and you don't get a job, only work :-) May 25 21:21:30 even better May 25 21:50:30 khem, I have setup a working busybox initrd. but the plan doesn't work out. soon as I try to chroot to the minimal rootfs on SD the system will. May 25 21:50:33 die. May 25 21:55:06 die yourself :-) May 25 21:55:18 dcordes: oh, never mind :-) May 25 21:55:59 :-) May 25 21:56:00 RP: pls have a look at the packaged-staging pending patch. May 25 21:56:39 likewise, is this supposed to be funny in any way or just rude? May 25 21:57:32 dcordes: the former May 25 21:57:50 dcordes: hmmm do you get any messages when it dies May 25 21:58:09 khem, ok I will post the dmesg with user debug enabled May 25 21:58:24 dcordes: I admire your grit May 25 21:58:55 dcordes: Can you check one more thing May 25 21:59:09 khem, of course May 25 21:59:14 dcordes: if all OE software is compiled without -mthumb switch May 25 21:59:29 its just for ease May 25 22:00:47 on a site note regarding switches: the busybox flags that work well are "-mlittle-endian -marm -mabi=aapcs-linux -mno-thumb-interwork -march=armv7-a -msoft-float -Uarm -g" May 25 22:00:58 (directly taken from busybox .config) May 25 22:01:25 -mno-thumb-interwork vs -mthumb ?? May 25 22:02:37 if you have both disabled is ok too May 25 22:03:05 ie. use -mno-thumb-interwork -marm May 25 22:04:21 khem, where to set it? in the cortex-a8.inc ? May 25 22:06:48 dcordes: you are using minimal right ? May 25 22:06:56 it should already be set for you May 25 22:07:10 but that said yes cortex-a8.inc could be used May 25 22:07:32 so I'm running with these options already. nothing to try right May 25 22:07:39 ..sorry yes I am using minimal. May 25 22:07:47 yeah May 25 22:08:07 check log.do_compile of some package to make sure May 25 22:16:16 khem, so you expect -mno-thumb-interwork and -marm to appear in the compile logs ? May 25 22:17:09 dcordes: I just want to see accidently if thumb is enabled May 25 22:18:11 no thumb found in compile logs May 25 22:18:19 dcordes: ok May 25 22:18:40 dcordes: we might even disable -march=armv7-a May 25 22:18:52 and instead use -march=armv5te May 25 22:19:02 dcordes and disable neon May 25 22:19:04 armv7-a is enabled in the known working busybox build May 25 22:19:14 how about doing a run with only neon disabled ? May 25 22:19:28 dcordes: ok yeah step by step May 25 22:19:31 disable neon first May 25 22:19:43 only neon or also the vfp stuff ? May 25 22:19:54 don't they depend on each other? May 25 22:20:04 sorry I don't have a good understanding of these things May 25 22:20:49 dcordes: I mean to use -msoft-float May 25 22:20:59 so yes disable vfp to May 25 22:25:49 ah ok http://gcc.gnu.org/onlinedocs/gcc-4.4.4/gcc/ARM-Options.html is very useful May 25 22:27:39 -msoft-float Equivalent to -mfloat-abi=soft May 25 22:28:02 yes May 25 22:30:29 -TARGET_CC_ARCH = "-march=armv7-a -mtune=cortex-a8 -mfpu=neon ${ARM_FP_OPT}" May 25 22:30:29 +TARGET_CC_ARCH = "-march=armv7-a -mtune=cortex-a8 ${ARM_FP_OPT}" May 25 22:30:32 khem, like this? May 25 22:30:52 yes May 25 22:33:49 I will compile minimal-image with that plus the gcc-4.4.4 instead of gcc-4.4.1 patch May 25 22:34:16 ok May 25 22:41:16 mplayer makes heavy use of the neon optimizations right ? May 25 22:42:35 mplayer can be compiled to use neon optimizations ./configure --enable-neon May 25 22:44:42 tom__, so without adding this configure switch I will get no neon code compiled when running bitbake mplayer ? (no matter if built for a machine using neon) May 25 22:45:32 I haven't looked at the recipe, it might have changed, should be clear looking at the .bb file though May 25 22:45:48 right May 25 22:47:00 but I believe when I do a ./configure --help on mplayer, the --enable-neon is automatic, so that might be a clue May 25 22:47:28 EXTRA_OECONF_append_armv7a = " --enable-armv6 --enable-neon" May 25 22:47:35 tom__: dcordes is fighting a different var May 25 22:47:47 war May 25 22:47:56 its not about mplayer May 25 22:47:58 freudian slip May 25 22:48:40 I see. I'd really like to see benchmarks of using armv7 vs armv6 ffmpeg May 25 22:48:40 khem, yes.. I was just curious if mplayer I'm running is making use of neon May 25 22:48:57 mplayer -v help May 25 22:48:57 dcordes: ah I see May 25 22:49:00 khem, cause it was going so fast May 25 22:49:05 should give you the compiled options May 25 23:02:19 ok it was built using neon. It can be looked up in config.mak in workdir May 25 23:16:12 khem, assuming everything will work, what can we take from that ? May 26 00:06:44 dcordes: if it works then we have a lot of possibilities May 26 00:06:53 dcordes: compiler generating wrong code with neon May 26 00:06:59 kernel bug executing it May 26 00:07:04 etc. etc. May 26 00:08:21 khem, compiler generating wrong code with neoen > doesn't oe work for other neon devices? May 26 00:08:54 dcordes: it does May 26 00:09:05 but you have a new device May 26 00:09:13 and it might have its own story May 26 00:11:04 I should examine the buildsystem used for htc nexus one (same qsd8250 chip) May 26 00:11:23 that wud help May 26 00:12:58 as opposed to htcleo, nexus comes with linux/android. we ran the htcleo minimal rootfs on nexus and it failed (rebooted, don't know exact error) May 26 00:14:18 android is using linux kernel for sure but userspace is different May 26 00:14:24 they dont use OE like rfs May 26 00:14:28 so assuming this was caused by the same errors I observed on htcleo it is a general problem of the snapdragon platform May 26 00:14:32 its quite different May 26 00:14:52 yeah you could conclude that May 26 01:13:57 khem, http://tinderbox.openembedded.net/packages/582326/ May 26 01:14:33 make[3]: *** [math_vfp_asm.lo] Error 1 May 26 01:17:38 question: if I am compiling something that uses GStreamer for teh beagleboard, which headers do I include in the include path for the compiler? May 26 01:26:28 dcordes: hmm its liboil bug May 26 01:29:33 #if defined(__VFP_FP__) && !defined(__SOFTFP__) May 26 01:29:51 dcordes: somehow its getting that part wrong May 26 01:31:12 dcordes: somehow your toolchain did not configure with softfp May 26 01:32:49 dcordes: problem is -mfloat-abi=softfp should be removed from the commandline May 26 01:38:32 aditya_111: what do you mean by compiler header ? May 26 01:39:02 aditya_111: depends on what headers you use May 26 01:41:54 you might have to add -I /usr/include/gstreamer-/ May 26 01:43:22 so adding headers to gstreamer on my host is fine? May 26 01:43:47 I am confused as to whether I need to have some special version of gstreamer headers that are made for beagleboard/oe May 26 01:44:36 aditya_111: you need the staged headers ofcource May 26 01:44:48 not host ones May 26 01:45:08 ah ok May 26 01:45:10 that makes more sense May 26 01:45:22 but cross gcc should look into sysroot automatically May 26 01:45:28 ah May 26 01:45:45 so if you have -I /usr/include/gstreamer- it should append the sysroot to it May 26 01:45:58 but OE does not rely upon that May 26 01:46:10 it adds explicit paths to -I and -L May 26 01:46:12 mmk May 26 01:46:16 which is ok too May 26 01:46:17 I guess I might as well May 26 01:46:20 just to be safe May 26 01:46:36 You dont have to May 26 01:46:41 try to experiment with sysroo May 26 01:46:51 you can preprocess May 26 01:47:00 and see if right headers are being pulled in May 26 01:47:02 when you are at it May 26 01:47:32 preprocess? May 26 01:47:41 you use OE build words that go over my head :P May 26 02:35:27 I add to RDEPENDS but it does not get installed in root file system May 26 02:54:36 Hi kergoth_ May 26 02:54:44 hey May 26 02:54:59 kergoth_: udev 151 needs depmod on target May 26 02:55:17 kergoth_: so I added it to RDEPENDS May 26 02:55:25 but it still does not show up on target May 26 02:55:50 RDEPENDS = "module-init-tools-depmod" May 26 02:55:54 this is what I added May 26 02:55:56 to recipe May 26 02:56:05 first of all, thats wrong, its global, that'll even make the -dev package require depmod May 26 02:56:26 second, i'd suggest verifying that everything is the way you think it is. look inside the udev ipk and confirm the Depends is correct May 26 02:56:32 when I add it to IMAGE_INSTALL then it shows up there May 26 02:56:34 ar x foo.ipk; tar zxvf control.tar.gz May 26 02:56:52 i'm sure you rebuilt udev after making the change, of course ;) May 26 02:56:57 yes May 26 02:57:10 bitbake -c clean udev minimal-image; bitbake minimal-image May 26 02:57:29 1) check the contents of the udev ipk, 2) check the do_rootfs log of the image creation, see if opkg mentions depmod at all May 26 02:58:43 log.do_rootfs.23871 does not show depmod May 26 02:59:10 well, if its in Depends of the ipk and doesn't end up in the image, i'd blame opkg :) **** ENDING LOGGING AT Wed May 26 02:59:57 2010