**** BEGIN LOGGING AT Fri Jun 20 02:59:56 2008 Jun 20 03:51:22 Anyone here build the linux-rp for hx2000? I can't seem to get it to build. The kernel generates numerous errors Jun 20 07:11:27 morning Jun 20 08:33:32 what gui do you recommend for git? Jun 20 08:36:03 xterm :-) Jun 20 08:56:06 * * OE Bug 4251 has been RESOLVED (WONTFIX) by Jun 20 08:56:08 * * evaluate hg as replacement for monotone Jun 20 08:56:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4251 Jun 20 09:43:59 good morning Jun 20 09:48:37 florian: good morning Jun 20 09:55:24 hey pb_! Jun 20 10:24:51 * pb_ working on CIFS implementation today Jun 20 10:24:54 not much fun Jun 20 10:33:55 XorA, rofl on the xterm remark Jun 20 10:54:11 pb_: I feel with you ;) Jun 20 10:54:58 * florian fights buggy sm501 hardware Jun 20 11:30:09 03koen 07org.oe.dev * r90bc3bcd... 10/ (7 files in 4 dirs): Jun 20 11:30:09 libusb: add libusb1, a rewrite of libusb 0.1.x Jun 20 11:30:09 libusb-compat: add a compat layer for libusb 0.1.x apps Jun 20 11:30:09 hal: use libusb-compat, not libusb Jun 20 11:30:09 lcd4linux: use libusb-compat, not libusb Jun 20 11:30:09 This will give increased powersavings because of this new feature: Jun 20 11:30:11 * libusb-0.1 wakes up the CPU every millisecond when waiting for bulk/interrupt data. The replacement of libusb-1.0 combined with libusb-compat-0.1 sleeps efficiently, only waking up when data has arrived. Jun 20 11:51:52 /nick hrw Jun 20 11:51:53 morning Jun 20 11:52:00 /nick hrw Jun 20 11:52:38 gm Jun 20 12:02:45 03koen 07org.oe.dev * rcbc5b6db... 10/ (3 files in 2 dirs): bluez-utils: use libusb-compat Jun 20 12:28:49 03koen 07org.oe.dev * rc06692d7... 10/ (3 files in 3 dirs): linux omap2 git: add patch to surpress spurious timer reprogramming Jun 20 13:13:03 ~curse sharp for kernel size limit Jun 20 13:13:04 May the fleas of a thousand camels infest your most sensitive regions, sharp for kernel size limit ! Jun 20 13:25:23 hi Jun 20 13:25:27 ~seen sakoman Jun 20 13:25:30 sakoman is currently on #oe (4d 12h 42m 6s). Has said a total of 35 messages. Is idling for 18h 2m 43s, last said: 'hate seeing those red bars on failed packages though'. Jun 20 13:59:02 hi woglinde! Jun 20 13:59:45 hi sakoman Jun 20 14:00:58 sakoman, I have oe on another machine now Jun 20 14:01:08 what image is giving you troubale for beagle Jun 20 14:01:12 sakoman for libsdl-x11 replace libsdl.inc in do_stage in the oe_install lines src and src/main with build Jun 20 14:01:43 sakoman and you can delete the install -m 0644 src/main/libSDLmain.a ${STAGING_LIBDIR} line Jun 20 14:02:56 * hrw wants git... Jun 20 14:03:18 sakoman the "ln -sf libSDL.so ${STAGING_LIBDIR}/libSDL-1.2.so" you can delete too Jun 20 14:03:22 hrw hehe Jun 20 14:05:35 * Crofton|work wonders if there will be a flood of commits when we can commit to git .... Jun 20 14:05:43 I had libsdl-x11 fix locally but do not even want to push it into mtn Jun 20 14:05:52 Crofton: libsdl-x11 and gimp were failing Jun 20 14:05:54 Crofton|work: I would do more Poky merges probably Jun 20 14:06:04 hrw I have fixed it to in better way Jun 20 14:06:23 Crofton|work: git format-patch + git am == good combo Jun 20 14:06:27 but I will commit it with libsdl-directfb and -qpe Jun 20 14:12:13 woglinde: not quite sure what you mean by " replace libsdl.inc in do_stage in the oe_install lines src and src/main with build" Jun 20 14:13:42 Crofton: are you getting the same failures? Jun 20 14:13:52 Or is it just me that is unlucky? Jun 20 14:15:55 gnumerics is failing too: Jun 20 14:15:57 application.c:26:44: error: pixmaps/gnumeric-stock-pixbufs.h: No such file or directory Jun 20 14:15:57 application.c:169:2: warning: #warning "FIXME: 1: this doesn't belong here. 2: it is not multihead safe." Jun 20 14:16:17 I jsut started libsdl build Jun 20 14:16:55 Crofton: with woglinde changes, or stock .dev? Jun 20 14:17:12 stock .dev Jun 20 14:17:17 I want to see it fail Jun 20 14:18:14 did you understand what changes woglinde is proposing above? Jun 20 14:18:18 I need to set that machine to do || make Jun 20 14:18:24 not yet Jun 20 14:18:36 I'll scroll back when I see the failure Jun 20 14:18:55 I understand the part about deleting two lines :-) Jun 20 14:19:06 I'm going to go ride my bike for a bit Jun 20 14:19:12 Enjoy! Jun 20 14:19:38 sakoman the sdllibs arent build in src or src/main, instead they are build in build Jun 20 14:19:49 and oe_libinstall do not honor this yet Jun 20 14:20:05 ah, OK! Jun 20 14:20:52 now I get it :-) Jun 20 14:24:14 ~9o~8~~~~~~~~~~~~~~~~~~~~ Jun 20 14:24:16 ~~~+ Jun 20 14:25:30 * mwester gets woglinde a new keyboard Jun 20 14:26:48 can I get one too? MS CM4000 Jun 20 14:28:20 woglinde: with those changes libsdl-x11 build completes successfully. I will try the image build now. Thanks! Jun 20 14:28:46 mwester you buy me a new son? fine Jun 20 14:52:50 building meta-toolchain in dev branch gives me error in qa_staging action - http://www.pastebin.org/44921 - what's this all about? Jun 20 14:53:56 hm your machine has an fpu? Jun 20 14:54:26 the target does yes Jun 20 14:54:53 freescale imx31 Jun 20 14:54:58 ah Jun 20 14:55:09 hm only QA issue and in i686 Jun 20 14:55:15 not such a big problem Jun 20 14:55:25 not sure why ncurses is being built for meta-toolchain anyway, but also not sure what this 'qa_staging' action is all about Jun 20 14:56:27 tharvey hm some libs need it Jun 20 14:56:47 but its complaining about ncurses built for host (i686) right? Jun 20 14:57:25 hms Jun 20 14:57:32 where was this option Jun 20 15:00:22 tharvey hm I suggest you try sdk Jun 20 15:00:38 what do you mean? Jun 20 15:01:19 tharvey or what is your build goal? Jun 20 15:01:24 to have a toolchain? Jun 20 15:01:52 my goal is to have my project build an external toolchain that I could provide to others, then use that toolchain to build an image Jun 20 15:02:12 yes Jun 20 15:02:17 thats exactly what sdk is for Jun 20 15:03:12 I'm listening.... Jun 20 15:03:56 hm Jun 20 15:04:12 meta-toolchain is part of sdk right? Jun 20 15:04:15 okay unfornatly only for openmoko there is an sdk-image yet Jun 20 15:04:21 thervey no it isnt Jun 20 15:05:05 hm try bitbake task-sdk Jun 20 15:07:26 what you want in sdk-image? Jun 20 15:08:11 hrw so I am right he has to write his own sdk-image.bb? Jun 20 15:08:13 just trying to build an external toolchain and use it... Jun 20 15:13:15 is richard purdie around? Jun 20 15:13:51 ~seen RP Jun 20 15:13:54 rp is currently on #maemo (8d 21h 36m 28s) #handhelds (8d 21h 36m 28s) #oe (8d 21h 36m 28s) #openmoko (8d 21h 36m 28s). Has said a total of 39 messages. Is idling for 2d 3h 1m 7s, last said: 'You will have to untar and retar though'. Jun 20 15:14:20 cool Jun 20 15:16:11 cbrake: awake? Jun 20 15:18:28 dbtid: yes, kicking a little :-) Jun 20 15:20:07 do you know of any way i can get nptl for arm? Jun 20 15:20:15 would oe have that somehow? Jun 20 15:21:20 dbtid: yes, OE has been using nptl for some time Jun 20 15:21:56 dbtid: so if you build the angstrom distro for about any recent ARM processor, you should get a nptl setup Jun 20 15:22:33 dbtid: http://wiki.openembedded.net/index.php/Category_talk:FAQ#9._What_is_NPTL_and_how_do_I_use_it_in_OE.3F Jun 20 15:22:44 dbtid: we have not finished cleaning up the OE faq yet ... Jun 20 15:25:36 so whats the diff between meta-toolchain and task-sdk? Jun 20 15:26:24 task-sdk is metapackage Jun 20 15:26:40 meta-toolchain generates archive with task-sdk Jun 20 15:27:01 is that to say meta-toolchain is a subset of task-sdk? Jun 20 15:27:14 no Jun 20 15:27:17 vice versa Jun 20 15:27:23 reverse I mean Jun 20 15:27:30 * hrw needs another big coffee Jun 20 15:27:55 hrw: I would think toolchain is part of SDK logically Jun 20 15:28:07 SDK would be toolchain + dev libs Jun 20 15:28:17 I'm definately confused... is any of this written up in documentation? Jun 20 15:28:36 Poky Handbook? Jun 20 15:30:25 brb Jun 20 15:37:24 hrw: In OE I see that meta-toolchain depends on task-sdk-bare as well as task-sdk depends on task-sdk-bare but meta-toolchain also generates a deployable toolchain is my understanding correct ? Jun 20 15:38:42 khem: task-* only generate packages. meta-toolchain generates deployable toolchain Jun 20 15:39:01 yeah thats what I also thought Jun 20 15:45:46 hello Jun 20 15:46:46 ok, so help me understand poky vs oe... is poky a 'stable' oe? Jun 20 15:47:26 no, it's a commercially supported subset of oe Jun 20 15:47:52 wow mickeyl is alive :-D Jun 20 15:48:02 o mickeyl Jun 20 15:48:06 ~poky Jun 20 15:48:07 it has been said that poky is an open source platform build tool based on OpenEmbedded. It is a complete software development environment for the creation of Linux devices. It enables you to design, develop, build, debug, and test a complete, modern, software stack using Linux, the X Window System and GNOME Mobile based application frameworks for both ARM and x86 based platforms. Website: http://pokylinux.org/ Jun 20 15:48:14 marketing ;D Jun 20 15:48:24 ya... marketing is useless... Jun 20 15:48:33 * ant|work thinks that the move to git would be the ideal point for freezing a new stable 2008.x Jun 20 15:48:35 I got that info from the website... Jun 20 15:48:37 I keep trying to get them all shot :-) Jun 20 15:50:00 jffs2 suxx Jun 20 15:50:14 cbrake: thanks for the info Jun 20 15:50:17 i'm going to check it out Jun 20 15:50:20 53MB on c760 mounts in 4.7s (with summary info) Jun 20 15:50:53 we can hope one day yaffs2 enters kernel and has sensible tools Jun 20 15:51:11 hrw: never noticed real improvements with/without summary (udev initscript...) Jun 20 15:51:26 ant|work: udev has nothing to summary Jun 20 15:51:35 I have 3s udev on Zaurus Jun 20 15:51:39 no, I mean it takes a long time Jun 20 15:52:02 udev taking time is nothing to do with jffs2 Jun 20 15:52:06 even if you'd save 5 sec by using summary you'll lose afterward Jun 20 15:52:10 but kernel needs 8.9s to start init Jun 20 15:52:16 after init Jun 20 15:52:24 ant|work: as I said - on my Z udev takes ~3s Jun 20 15:52:45 nice, please give out the patches Jun 20 15:53:02 first need to check lot of things Jun 20 15:53:26 btw please do "du -h /dev" on c7x0 Jun 20 15:53:56 so you could think of poky as a 'stable subset' of oe? Jun 20 15:54:01 got udev is less than a second on gta02 Jun 20 15:54:22 ant|work: 32k due to /dev/.udev taking 28k Jun 20 15:54:28 tharvey: yes Jun 20 15:54:31 udev 115 right? Jun 20 15:54:42 ant|work: 115 or 124 - both give same results Jun 20 15:54:47 now 115 Jun 20 15:55:05 ah, interesting (oebug 4118) Jun 20 15:55:52 hrw: I got 4,8kb ;-( Jun 20 15:57:24 so task-sdk builds for me... but not clear what it did. In deploy I see ipk dir and pstage dir... no sdk Jun 20 15:59:21 hrw: are you using linux-rp_2.6.24 on c7x0? Jun 20 15:59:29 or newer? Jun 20 16:00:09 * ant|work imagines hrw hard working on the next Poky's release Jun 20 16:00:42 ant|work: 2.6.26-rc4 Jun 20 16:00:52 hmm... Jun 20 16:04:34 I wonder if I should stop further testing on 2.6.24 and jump on the new wagon... Jun 20 16:05:24 I already abandoned the rotten 2.6.223 Jun 20 16:06:07 (2.6.23 doesn't resume on c7x0...) Jun 20 16:09:05 hrw: last question before I go: have you a special defconfig for 2.6.26_c7x0? The one in Poky? Jun 20 16:10:58 poky one but had to disable oabi Jun 20 16:11:20 size contraints I guess... Jun 20 16:13:11 yes\ Jun 20 16:13:59 hrw: 2008-06-02: looks like the same defconfig is in oe .dev Jun 20 16:14:17 thanks for your time Jun 20 16:14:23 have a nice weekend! Jun 20 16:14:26 bye Jun 20 16:25:11 great manual for poky.... should be required reading for anyone using oe Jun 20 16:25:35 thx Jun 20 16:29:25 ade|desk: You wanted me? Jun 20 16:44:46 ~seen methril Jun 20 16:44:50 methril is currently on #oe (1d 8h 1m 22s) #openmoko (1d 8h 1m 22s). Has said a total of 30 messages. Is idling for 9h 33m 18s, last said: 'morning'. Jun 20 16:45:04 hi Jun 20 17:09:29 hello Jun 20 17:22:41 hi udo Jun 20 17:32:40 which was the machtbox packet to show as a filemanager ? Jun 20 17:34:30 there is no matchbox-filemanager Jun 20 17:34:37 Poky use pcmanfm Jun 20 17:35:50 cool thanks .. Jun 20 17:35:58 then i need a gui for mplayer or similar Jun 20 17:39:51 do not know any Jun 20 17:40:09 you can try mplayer with gtk Jun 20 17:47:41 irt to the operator, it's currently ?= Jun 20 17:47:47 it should be += ? Jun 20 17:48:05 (oops, wrong channel, sorry :/ ) Jun 20 17:51:47 hi kgilmer Jun 20 17:54:26 hello woglinde :) Jun 20 17:58:00 sorry if the is distro specific, but I am failing my bitbake base-image for angstrom x86-presoctt (org.openembedded.dev) build with a 'duplicate src declaration'. Anyone any thoughts? Jun 20 17:59:06 waite please provide the error.log as pastebin Jun 20 17:59:10 ~pastebin Jun 20 17:59:10 [~pastebin] A "pastebin" is a web-based service where you can paste anything over 3 lines without flooding the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://www.rafb.net/paste Jun 20 17:59:56 woglinde, http://pastebin.com/d26344b13 Jun 20 18:01:01 updated with actual error msg Jun 20 18:01:07 hm Jun 20 18:01:08 /home/bwaite/Projects/CommonOS/OE3/tmp/rootfs/usr/lib/opkg/info/udev.postinst: line 10: /etc/udev/mount.blacklist: Permission denied Jun 20 18:01:26 is this oe-dev or stable? Jun 20 18:02:20 oe-dev Jun 20 18:02:46 waite did you pull and update? Jun 20 18:03:02 I did as of yesterday. Let me see if there is anything new Jun 20 18:03:18 which udev version is this? Jun 20 18:03:40 Downloading file:/home/bwaite/Projects/CommonOS/OE3/tmp/deploy/glibc/ipk/i686/udev_118-r3_i686.ipk Jun 20 18:03:53 I was already up to date Jun 20 18:05:17 Sounds like pkg problem. I'll hunt it downm Jun 20 18:06:19 hm could ypu please put /home/bwaite/Projects/CommonOS/OE3/tmp/work/x86-prescott-angstrom-linux/base-image-1.0-r0/temp/log.do_rootfs.2185 on pastebin too? Jun 20 18:10:25 morning Jun 20 18:10:57 hi cyrilRomain Jun 20 18:12:16 <|dl9pf|> hi ! Jun 20 18:12:52 <|dl9pf|> i've some problems with gcc-4.2.2 in target-compilation stage Jun 20 18:13:43 bye Jun 20 18:13:50 bye Jun 20 18:13:53 have a nice weekend Jun 20 18:14:03 you too Jun 20 18:14:34 dl9pf which one? Jun 20 18:14:43 dl9pf did you check the bugtracker? Jun 20 18:15:19 <|dl9pf|> its bailing out at libiberty Jun 20 18:17:18 ~pastebin Jun 20 18:17:19 [~pastebin] A "pastebin" is a web-based service where you can paste anything over 3 lines without flooding the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://www.rafb.net/paste Jun 20 18:18:17 <|dl9pf|> checked the bugtracker, but didn't find a maching entry Jun 20 18:18:41 please pastebin the error Jun 20 18:19:47 <|dl9pf|> will do ... just a minute - tnx Jun 20 18:25:17 <|dl9pf|> http://pastebin.ca/1052059 Jun 20 18:27:01 hm seems you mixing up some archs Jun 20 18:27:10 <|dl9pf|> ? Jun 20 18:27:11 could please provide your local.conf Jun 20 18:27:18 <|dl9pf|> jap Jun 20 18:27:22 and which gcc ist your host-gcc Jun 20 18:29:10 <|dl9pf|> local.conf http://pastebin.ca/1052061 Jun 20 18:29:23 <|dl9pf|> gcc version 4.2.1 (SUSE Linux) Jun 20 18:30:15 hm okay Jun 20 18:30:47 coul please try DISTRO = "angstrom-2008.1" Jun 20 18:31:01 generic is not used so much Jun 20 18:32:03 <|dl9pf|> ok ... Jun 20 18:32:04 in local.conf Jun 20 18:32:20 <|dl9pf|> machine/lire-i586.conf http://pastebin.ca/1052067 Jun 20 18:33:38 <|dl9pf|> ok .. .rerunning Jun 20 18:34:43 <|dl9pf|> should i remove the tmpdir to get a clean build ? Jun 20 18:34:50 why not use a "generic" or minimal? Jun 20 18:35:17 dl9pf better clean tmpdir Jun 20 18:35:30 <|dl9pf|> methril: generic or minimal for WHAT ? Jun 20 18:35:54 methril distro sets which glibc and gcc and other toolchain tools is used Jun 20 18:36:14 image decide what is build for an image and whats in the image Jun 20 18:36:27 then you didn' t depend of angtrom distro? Jun 20 18:36:36 ok Jun 20 18:36:57 but angstrom with uclibc is not working, isn't it? Jun 20 18:37:15 its working Jun 20 18:37:19 why it shouldnt? Jun 20 18:37:23 for oabi its works Jun 20 18:37:31 because I have it here Jun 20 18:37:54 hm okay its compile Jun 20 18:40:15 i had some troubles with it Jun 20 18:41:07 then speak up and I will fix it if I can Jun 20 18:41:33 well, now i'm not on that Jun 20 18:41:43 right now Jun 20 18:41:52 it was a test Jun 20 18:41:59 we forget the uclibc :) Jun 20 18:47:52 *g* I like it Jun 20 18:48:06 because I dont have to wait to build glibc twice Jun 20 18:50:53 anyone know if there is a place to buy stuff with OE logo? like tshirts and mugs etc.? Jun 20 18:55:29 kgilmer nope Jun 20 18:55:32 there isnt one Jun 20 18:55:33 kgilmer: at FOSDEM 2007 there was tshirts. No sure there is stuff to bye outside such events. Jun 20 18:55:54 OE plan (is already maybe?) a German registered Jun 20 18:55:56 I hope we can improve the situation after the e.V. is real Jun 20 18:55:56 association (e.V., non-profit, not for business) Jun 20 18:56:13 cyrilRomain, saw that in the blog... what's the eta? Jun 20 18:56:26 i want to have an OE sticker right next to my gentoo sticker on my thinkpad :) Jun 20 18:56:53 cyril the documents arent yet at the court Jun 20 18:56:55 i know there are sites that will sell stuff with a custom logo on it. cafepress i think is what gentoo uses. Jun 20 18:56:57 kgilmer: hehe. But tbh I don't know the status Jun 20 18:56:57 gentoo on thinkpad .. i have a highlight on that ;) Jun 20 18:57:02 woglinde: ah thanks Jun 20 18:57:11 than it will least half an year if it is finished Jun 20 18:58:46 well i could look at setting up a cafepress thing for oe...but i wonder if there is a high quality copy of the logo somewhere... Jun 20 18:59:10 kgilmer you could ask koen or mickeyl Jun 20 18:59:29 ok, i'll check with them. thx woglinde. Jun 20 18:59:49 kgilmer: yep there is a svg logo Jun 20 19:00:15 kgilmer: yeah mickeyl used it even for the last FOSDEM iirc Jun 20 19:01:12 cyrilRomain, great, thx. I'll email him and ask. Jun 20 19:01:22 kgilmer: it is in org.openembedded.dev/contrib Jun 20 19:01:49 kgilmer: in org.openembedded.dev/contrib/artwork/oe.svg even Jun 20 19:02:55 i see it Jun 20 19:03:00 but...it's pretty small Jun 20 19:03:05 283x213 Jun 20 19:03:27 kgilmer: but it's svg, so you can scale it at will without quality loss Jun 20 19:03:40 ah :) I feel silly Jun 20 19:03:54 yeah it's a simple set of ellipses Jun 20 19:04:35 kgilmer: friday evening, you're forgiven ;) Jun 20 19:05:14 <|dl9pf|> 40 tasks to go ... Jun 20 19:05:26 |dl9pf| ;) Jun 20 19:05:28 *whew* Jun 20 19:06:06 I wonder if anyone will get bent out of shape if I setup a cafepress account for OE... Jun 20 19:07:39 woglinde, sorry I got pulled away. THe do_rootfs.log is what I posted. Jun 20 19:07:42 it looks i can do this without having to setup any public account, so it should be fine. Jun 20 19:13:21 kgilmer post it to devel list Jun 20 19:13:41 waite hm is it working now? Jun 20 19:14:07 kgilmer, check with mickey, once we have the eV going, we could create an OE store Jun 20 19:14:21 ok will do woglinde. gimp doesn't seem to respect the svg'ness of svg. Jun 20 19:14:27 * Crofton|work wonders of cafepress can deal with non-US bank accounts Jun 20 19:14:57 Crofton with the e.v thats what I said Jun 20 19:15:13 Crofton|work sounds good. cafepress looks like it will let you do one-offs. Jun 20 19:15:23 save the artwork :) Jun 20 19:15:35 not sure about international but i assume they do, as the gentoo foundation uses them. Jun 20 19:16:35 maybe i'll wait so that the money goes to the e.v Jun 20 19:17:30 "...CafePress.com offers multiple shipping options for delivery within the USA, plus international shipping around the globe..." Jun 20 19:25:34 :( now it fails even earlier - Jun 20 19:26:25 http://pastebin.ca/1052097 Jun 20 19:27:31 waha my wif si watching soccer and claims to much badnwidth Jun 20 19:28:22 dl9pf hm sorry I never saw glibc failing with this Jun 20 19:28:31 dl9pf please write a bugreport Jun 20 19:29:04 ok will do Jun 20 19:29:20 i'll retry and if it fails i'll report Jun 20 19:31:14 woglinde: tnx Jun 20 19:33:22 dl9pf you could try uclibc at least Jun 20 19:33:30 setting in local.conf Jun 20 19:33:49 ANGSTROM_MODE= "uclibc" Jun 20 19:35:33 woglinde: need glibc here Jun 20 19:40:08 dl9pf hm why? Jun 20 19:40:53 not my decision ;) Jun 20 19:41:28 ah okay Jun 20 19:41:38 sorry I thought is was personal work Jun 20 19:42:34 its a part i need for my thesis Jun 20 19:44:10 ui nice Jun 20 19:44:27 whats your thesis about? Jun 20 19:45:13 its related to embedded realtime linux in automatation / for robots Jun 20 19:45:46 so i'd bring xenomai and some realtime bits into openembedded Jun 20 19:46:05 but before that i need a _running_ build Jun 20 19:50:00 yes I see Jun 20 19:50:14 sorry I dont know why glibc cant find the missing symbols Jun 20 19:50:28 no problem we'll see Jun 20 19:50:33 it looks like some x86 optimisations Jun 20 19:51:00 __sync_bool_compare_and_swap_4 Jun 20 19:51:21 dl9pf where are you from? Jun 20 19:52:27 dl9 suggests .de :) Jun 20 19:52:55 Crofton: hehe - right Jun 20 19:52:59 i'm from hannover Jun 20 19:53:03 * Crofton is k4ep Jun 20 19:53:27 Crofton: gd dr om ;) Jun 20 19:53:39 freifunker Jun 20 19:53:55 crofton that he is from germany I knowed Jun 20 19:54:01 knew Jun 20 19:54:03 damnit Jun 20 19:55:01 dl9pf: for your problem, it's as if the CFLAGS -march was not (or badly) set Jun 20 19:56:24 hmmm wrong in local.conf or ./configure issue ? Jun 20 19:57:56 dl9pf hm maybee wrong in your machine defines Jun 20 19:58:00 dl9pf: maybe in machine/lire-i586.conf Jun 20 19:58:01 let me see Jun 20 19:59:05 i change to i586-generic for a test Jun 20 19:59:28 machine/lire-i586.conf http://pastebin.ca/1052067 Jun 20 20:00:19 TARGET_CC_ARCH = "-mcpu=pentium" Jun 20 20:00:20 hm yes Jun 20 20:00:22 seems Jun 20 20:00:29 no one updated this Jun 20 20:00:54 the question is why the march flag is not set for any x86 machine .. Jun 20 20:01:16 the lire-i586.conf similar to the i586-generic.conf, just the kernel is changed Jun 20 20:01:37 perhaps march flag is not supported in old version of gcc ? .. Jun 20 20:01:40 can anyone tell me the correct variable for the kernel headers I should use in a recipe? I tried STAGING_KERNEL_DIR but that turns out to be invalid... Jun 20 20:02:02 is my config messed up somehow or am i using the wrong variable? Jun 20 20:02:53 so its an issue with tune-pentium.inc ? Jun 20 20:03:13 dl9pf hm yes Jun 20 20:03:34 dl9pf what processor is your board really? Jun 20 20:03:48 gcc-4 has more different options Jun 20 20:04:00 the board are ~p2 Jun 20 20:04:13 mostly pentium class Jun 20 20:04:24 x86, 32bit Jun 20 20:04:34 *g* Jun 20 20:04:37 gcc hass pentium2 Jun 20 20:04:39 okay Jun 20 20:04:53 copy tune-pentium.inc to tune-pentium2.inc Jun 20 20:05:06 and replace -mcpu with -march Jun 20 20:05:14 and pentium with pentium2 Jun 20 20:05:29 I bet internal gcc will not differ Jun 20 20:05:32 but who knows Jun 20 20:06:37 ok ... it failed also with i586-generic Jun 20 20:06:53 changing tune-pentium now and retrying Jun 20 20:08:23 ok. fresh run Jun 20 20:08:28 mcpu is deprecated (at least in gcc4.3 man page) Jun 20 20:09:01 one should use mtune now, or march that implies mtune Jun 20 20:09:42 cyrilRomain hm for x86 its not deprecated Jun 20 20:11:04 or the debian man page is outdated Jun 20 20:11:25 args Jun 20 20:11:39 got the wrong page Jun 20 20:13:12 well, knowing which version of gcc support which options and per architecture is a science in itself ! Jun 20 20:15:40 yeah yiu are right Jun 20 20:16:19 -mcpu=cpu-type A deprecated synonym for -mtune. Jun 20 20:16:19 -mfpmath=unit Jun 20 20:16:21 ups Jun 20 20:16:33 hmm maybe khem knows why OE only set mcpu=pentium and why there is no march=ix86 in use ... Jun 20 20:16:58 * cyrilRomain guess mcpu=pentium is the least common denominator Jun 20 20:18:43 http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gcc/i386-and-x86_002d64-Options.html#i386-and-x86_002d64-Options Jun 20 20:18:59 dl9pf: so in your machine/lire-i586.conf you could try not to include the tune-pentium.inc, and set TARGET_CC_ARCH = "-mcpu=pentium -march=i586" Jun 20 20:19:40 dl9pf: not sure it would solve your problem though Jun 20 20:19:52 ok ... i'll try in in the next round Jun 20 20:19:56 cyrilRomain not really Jun 20 20:20:06 -march=pentium2 is enough Jun 20 20:20:16 it sets mtune=pentium2 Jun 20 20:20:30 woglinde: you're right Jun 20 20:21:30 with 586 you have no mmx support Jun 20 20:21:40 dont know if this is wanted Jun 20 20:21:46 with yur realtime Jun 20 20:23:05 i need to check, but it wasn't specified iirc Jun 20 20:24:44 hm but for some dsp tasks the mmx register could produce faster results Jun 20 20:25:52 i need to check if it was used in the old environment Jun 20 20:26:02 hm I will fix the include now Jun 20 20:26:06 wasn't on my radar Jun 20 20:27:16 woglinde: hm wait, are you sure the generation of available instruction for machine type does not do harm in certain case ? Jun 20 20:27:55 woglinde: I'm still wondering the effects of setting march in addition to mcpu/mtune Jun 20 20:28:56 cyril its only for x86 valid Jun 20 20:29:05 for arm its different Jun 20 20:29:11 yeah sure Jun 20 20:29:43 but hmm Jun 20 20:30:09 hm? Jun 20 20:30:15 what should it break Jun 20 20:30:22 we only use gcc-3 and gcc-4 Jun 20 20:30:35 and for both the docu says march implies mtune Jun 20 20:31:08 well it's ok. I though tune-pentium.inc was included by several .conf files Jun 20 20:31:25 Is http://www.openembedded.org hosed today? I don't seem to be getting much down from it. And the one page I did get down was illegible. Jun 20 20:31:47 cyril it shouldnt Jun 20 20:32:01 and then machine.conf should be fixed Jun 20 20:33:08 machine.conf ? Jun 20 20:33:09 yep. I'm confident now for the fix :) Jun 20 20:33:35 dl9pf: woglinde was speaking about the machine conf files in general Jun 20 20:34:21 ok Jun 20 20:35:34 woglinde: so you are going to s/mcpu/march/ in tune-pentium.inc ? Jun 20 20:35:59 cyril yes Jun 20 20:36:12 and pentiumpro Jun 20 20:36:14 yep Jun 20 20:36:48 but not s/pentium/pentium2/ because I guess some i586 machine would not like MMX instructions .. (?) Jun 20 20:37:25 right Jun 20 20:37:32 but dl9pf is using a p2 Jun 20 20:37:38 so why he sdhouldnt use mmx Jun 20 20:37:56 * cyrilRomain wonders if MMX instructions are used only if they exists, or if it can make a program fail to run if they are not available Jun 20 20:38:43 woglinde: dl9pf can use pentium2, yes Jun 20 20:40:29 cyril: fails to run if not available. Jun 20 20:40:41 dl9pf: so as woglinde suggested earlier, setting TARGET_CC_ARCH = "-march=pentium2" should be fine for you Jun 20 20:40:57 teamnoir: thanks Jun 20 20:41:07 cyril: I believe they can be trapped into illegal instructions, which could be handled like machines with floating point libraries instead of instructions. But generally, they just fail with sigill. Jun 20 20:41:42 03woglinde2 07org.oe.dev * rf86fbd2f... 10/ (3 files in 2 dirs): Jun 20 20:41:42 *fix -m options now using -march which is fine with gcc-3 and gcc-4 Jun 20 20:41:42 according to the documentation Jun 20 20:42:22 3 files Jun 20 20:42:24 hms Jun 20 20:43:29 cyrilRomain could please show which 3 file I pushed? Jun 20 20:43:37 lookup I meant Jun 20 20:43:38 woglinde: yes Jun 20 20:43:51 linuxtog seems to hang again Jun 20 20:45:00 hm silly me Jun 20 20:45:05 can look it my self Jun 20 20:45:49 woglinde: fatal: The remote end hung up unexpectedly :-/ Jun 20 20:45:53 on which ML can i post patches ? Jun 20 20:45:54 m Jun 20 20:46:05 dl9pf attach it to the bugtracker Jun 20 20:46:15 but its down I think Jun 20 20:46:26 ah no its fine Jun 20 20:46:28 http://bugs.openembedded.net/ Jun 20 20:46:45 ok Jun 20 20:46:56 hm log shows me only 2 files Jun 20 20:47:08 where the heck cia the third cames Jun 20 20:47:13 if i have a do_compile() { oe_make } .. shouldnt the package be compiled by the compiler for the target architecture? Jun 20 20:47:56 cekay hm yes Jun 20 20:48:02 hm mom Jun 20 20:48:09 cekay what you want achieve Jun 20 20:48:23 and why you need to overwrite do_compile? Jun 20 20:48:28 my first bitbake receipt Jun 20 20:49:21 hm Jun 20 20:49:23 oe_runmake Jun 20 20:50:20 yea, runmake. sorry, typo Jun 20 20:50:36 cekay look at acpid.inc Jun 20 20:51:31 ah, ok Jun 20 20:52:18 * cekay crosses his fingers Jun 20 20:52:43 yay! Jun 20 20:52:48 thanks woglinde! Jun 20 20:57:29 hmm looks better ... but not completed Jun 20 21:02:16 hmmm, still having path issues with KERNEL_STAGING_DIR. is there a way I can see the values of all variables from within my recipe? Jun 20 21:03:00 kgilmer: bitbake -e Jun 20 21:03:42 hm strange Jun 20 21:03:48 why cia says 3 files Jun 20 21:03:59 oh wow Jun 20 21:04:08 bitbake is no longer a black box Jun 20 21:04:19 thanks cyrilRomain :) Jun 20 21:04:49 kgilmer you could use bitbake shell too Jun 20 21:05:04 oh interesting ... Jun 20 21:05:17 need to read a bit more about options for bitbake Jun 20 21:05:22 bitbake --help ;) Jun 20 21:05:28 bitbake -i Jun 20 21:05:31 help Jun 20 21:05:43 amazing... a developer tool that... helps developers :) Jun 20 21:05:56 peek Jun 20 21:06:32 kgilmer: looks like you want '${STAGING_KERNEL_DIR}/include/' Jun 20 21:07:02 cyrilRomain, well, my STAGING_KERNEL_DIR points to a path that does not exist Jun 20 21:07:12 i'm doing a fresh full build Jun 20 21:07:57 kgilmer: hmm wait STAGING_KERNEL_DIR use your host system headers Jun 20 21:08:26 ok, i'm looking for my target kernel headers Jun 20 21:08:32 kgilmer: STAGING_INCDIR should be enough then Jun 20 21:09:18 sweet, let me give that a shot Jun 20 21:10:16 btw, perhaps it's just me but the oe wiki is exteremly slow to load Jun 20 21:10:48 whoa, interactive bitbake Jun 20 21:10:59 this is like a whole new level Jun 20 21:12:28 kgilmer: http://wiki.openembedded.net is not slow here Jun 20 21:13:05 kgilmer no the server has problems Jun 20 21:13:32 I'm in nyc...and this one is fast. I was using the linuxtogo.org wiki. that was slow for me. Jun 20 21:13:32 kgilmer for bitbake shell you could hail mickeyl Jun 20 21:15:38 STAGING_INC dir seems to be pointing to the right place. With -e and -i I have a lot more visibility into what's going on when my recipe executes. thanks cyrilRomain and woglinde. Jun 20 21:16:24 ~hail mickeyl for bitbakeshell Jun 20 21:16:25 * ibot bows down to mickeyl for bitbakeshell and chants, "I'M NOT WORTHY!!" Jun 20 21:16:30 *g* Jun 20 21:16:36 haha Jun 20 21:18:15 'nite all Jun 20 21:19:13 :)) Jun 20 21:19:15 heya Jun 20 21:19:25 woglinde: wow ... semms like that was the solution ... Jun 20 21:19:47 dl9pf cool Jun 20 21:20:59 some more tasks to go, but glibc is done Jun 20 21:21:46 dl9pf I hope I can read your thesis some day Jun 20 21:23:12 i'll post some patches soon Jun 20 21:23:30 new packages and so on Jun 20 21:23:41 I hope I have some time to integrate them Jun 20 21:26:04 woglinde: whare are you located ? Jun 20 21:26:10 berlin Jun 20 21:26:42 were you at linuxtag ? Jun 20 21:27:07 yes Jun 20 21:27:13 but only on friday Jun 20 21:27:48 i was parttime at the opensuse booth that day Jun 20 21:30:17 * teamnoir thinks www.openembedded.org is unusable today. Jun 20 21:31:06 * teamnoir notes that http://oe.linuxtogo.org/wiki/GettingStarted isn't legible today. Jun 20 21:34:35 teamnoir what about this? http://wiki.openembedded.net/index.php/Getting_Started Jun 20 21:46:18 ~hail woglinde for saving my weekend ;) Jun 20 21:46:19 * ibot bows down to woglinde for saving my weekend ;) and chants, "I'M NOT WORTHY!!" Jun 20 21:58:27 Hi, I know this #oe channel is for "OpenEmbedded Developer Lounge" chat, but I am just a Zaurus user. Jun 20 21:59:08 I don't have broadband. I am trying to find out how to tweak my installation so that I don't need to connect to the Internet to complete my installation (ie. to have everything downloaded and prepared on a flash card). What exactly do I have to do for this? Can someone explain? Jun 20 22:00:26 kgilmer: looks better, thanks. Jun 20 22:04:12 Presumably there's a point in the installation where you have to connect to the Net. what do I have to do to steer things to continue installing from files on the flash card? Jun 20 22:06:20 Hurro? Jun 20 23:03:39 Hurro?? Jun 20 23:04:13 "Secrecy for NY Fed's derivatives bailout scheme challenged" :- Jun 20 23:04:29 WASHINGTON -- The New York Federal Reserve's closed-door rule making with top players in the massive $60 trillion credit default swaps market came under legal fire on Sunday, as a fair finance activist filed a complaint questioning why it was done in the dark. Jun 20 23:04:56 "The Federal Reserve seems to think it can engage in rule making in secret only with the industry," said Matthew Lee, executive director of the New York-based non-profit group Inner City Press/Community on the Move. Jun 20 23:05:23 Lee filed the administrative complaint on Sunday with both the New York Fed and the Federal Reserve Board in Washington. In the complaint, he demanded that the central bankers explain why the meetings this month were private and requested copies of all communications and details about the New York Fed-sponsored talks. Jun 20 23:05:45 Officials at the Federal Reserve could not immediately be reached for comment. Jun 20 23:06:16 The meetings were held with more than a dozen companies led by investment bank Goldman Sachs Group Inc. The companies -- which account for the bulk of business in the $60 trillion market -- met to help set new rules for credit default swaps trading, including the establishment of a clearinghouse. Jun 20 23:06:37 Credit default swaps are privately negotiated transactions used by companies to hedge against default risks. Over the past decade, the market has grown exponentially, from about $1 trillion to $60 trillion. Jun 20 23:07:10 Lee, referring to the Fed-led rescue of investment bank Bear Stearns by JPMorgan Chase & Co, said, "It was one thing to bail out Bear Stearns without any comments from the public. Now the Fed is trying to bail out or benefit 17 of the largest financial institutions behind closed doors." Jun 20 23:07:26 Citing the federal Administrative Procedures Act, he said it was illegal to have conducted the meetings. Jun 20 23:07:41 HURRO????? Jun 20 23:07:44 "We aim to stop it," said Lee, a fair finance and housing activist whose group has been instrumental in shaping how the Federal Reserve approves bank mergers for more than a decade. Jun 20 23:08:04 Also present at the meetings were representatives from derivatives and securities industry trade groups who were directed to make no public statements or disclosures about the talks, according to information obtained by Reuters. Jun 20 23:08:16 "This closed-down, top-heavy process is unacceptable and, Inner City Press hereby timely contends, is contrary to law," Lee wrote in the complaint. Jun 20 23:08:31 The complaint comes a week after New York Federal Reserve Bank President Tim Geithner announced the initiatives being developed with the companies, including the central bank's endorsement of the use of a clearinghouse for trading the instruments. The companies that attended the meeting own a centralized clearinghouse called Clearing Corp. Jun 20 23:08:51 Among those present at the private meetings with the New York Fed were: Goldman Sachs, Merrill Lynch & Co. Inc., Morgan Stanley, the Royal Bank of Scotland Group PLC, Societe Generale, UBS AG and Wachovia Corp, Bank of America Corp, JPMorgan Chase & Co., and Citigroup, according to the New York Fed. Jun 20 23:09:09 Also attending were representatives from the buy-side firms and hedge funds AllianceBernstein, BlueMountain Capital Management LLC, and Citadel Investment Group LLC. Jun 20 23:09:44 Congressional interest Jun 20 23:09:46 Lee's complaint comes as key lawmakers including House Financial Services Committee Chairman Barney Frank, a Massachusetts Democrat, are beginning to probe whether an overhaul of financial markets regulation is needed. Jun 20 23:10:11 [_WHETHER_???!!!] Jun 20 23:10:31 Next month Frank's committee is expected to hold a series of hearings with regulators, including the New York Fed's Geithner, and industry leaders. Jun 20 23:10:44 There is also concern in the Senate. The Senate Banking Committee this week will hold a hearing with top financial market regulators during which the oversight of off-exchange derivatives is expected to come up. Jun 20 23:11:01 Committee members have raised concerns about the role of credit default swaps in the credit crisis. Jun 20 23:11:24 "While I do not question that the creation of these structured products has delivered measurable benefit to the American consumer by lowering borrowing costs, it is becoming clear that a lack of transparency in the pricing and trading of these instruments has contributed to the credit crisis," committee member Sen. Elizabeth Dole, a Republican from North Carolina, said this month. Jun 20 23:12:07 Dole is not alone in her [PHONEY] concerns, Joint Economic Committee Chairman Charles Schumer, a New York Democrat and also a member of the Senate Banking Committee, has urged the Fed and other key regulators to weigh the benefits of an electronic trading system for these complex, privately negotiated markets, in an effort to increase transparency. Jun 20 23:12:31 (by Joanne Morrison Reuters Sunday, June 15, 2008) Jun 20 23:13:21 HURRRO? Is there anyone alive in this dead fish called "Western Civilization"? Jun 20 23:15:53 hands up if you read all that anyone Jun 20 23:16:02 .... Jun 20 23:28:21 A_A: ever heard of a pastebin? Jun 20 23:30:06 he's spamming in both #oe and #angstrom Jun 21 01:50:44 y **** ENDING LOGGING AT Sat Jun 21 02:59:56 2008