**** BEGIN LOGGING AT Thu Jun 12 02:59:57 2008 Jun 12 07:10:51 morning Jun 12 07:56:19 morning Jun 12 08:01:32 yo hrw Jun 12 08:02:26 mickey|ICE694: train again? Jun 12 08:07:49 yep Jun 12 08:07:55 braunschweig, hackathon with alphaone, shoragan, stefan_schmidt Jun 12 08:08:35 moin :) Jun 12 08:09:48 ah, he wakes up Jun 12 08:09:48 morning :) Jun 12 08:09:55 03koen 07org.oe.dev * rd84f271b... 10/ (4 files in 3 dirs): traqconfig: add configapplication of the lasertraq suite Jun 12 08:10:35 * mickey|ICE694 in severe need of a shoragan coffe (tm) Jun 12 08:15:44 mickey|ICE694: did someone in openmoko tried to make boot faster? Jun 12 08:17:22 hrw: I took 10seconds off boot time Jun 12 08:17:38 XorA|gone: how? Jun 12 08:17:47 hrw: dont run udevtrigger/udevsettle Jun 12 08:18:22 nice Jun 12 08:18:45 hrw: Tartarus put me onto it, preload /dev with static devices built into kernel, then dont run trigger/settle Jun 12 08:18:58 hrw: only issue is SD driver has to be a module :-( Jun 12 08:20:09 so you create /dev/ by hand then? Jun 12 08:21:25 hrw: yes, I copied one from a booted device Jun 12 08:21:38 hrw: removed devices that were modules Jun 12 08:21:57 hrw: but mmc layer needs to be a module to force it to hotplug the partitions :-( Jun 12 08:22:19 ok, so no option for me it is Jun 12 08:23:00 hrw: well you could technically put mmc devices in as well, but then you have to rely on people using the same partitions you do Jun 12 08:23:20 hrw: or just make dev nodes for all partitions Jun 12 08:23:47 but udev is damn slow on trigger Jun 12 08:23:55 XorA|gone: but I need to have device first Jun 12 08:24:10 even if you remove the udevsettle, the udevtrigger pulls enough cpuload to still extend boot by 10secs Jun 12 08:26:08 some kind of eeepc-fastinit would be great Jun 12 08:26:25 cekay: what it is? Jun 12 08:27:44 the eeepc boots from bios to xserver in 16seconds Jun 12 08:27:49 hrw: some guys in .tw yeah, but there were no concrete patches. i have a couple of ideas to do that, but have more important stuff amt. Jun 12 08:27:54 we need a leaner init + initscripts in OE Jun 12 08:27:58 it doesn't make sense for a phone to boot like a desktop system Jun 12 08:28:02 i wish i could work on that, but there'sno time Jun 12 08:28:05 they seem to have an init replacements which is fast as hell Jun 12 08:28:30 i don't think fastinit is free tho Jun 12 08:28:51 cekay: it is simple idea - do anything needed and boot X. then boot other components Jun 12 08:29:06 its just lots of stuff like udevtrigger adding up, none of it is needed on a device with fixed specs Jun 12 08:29:08 does not help too much on embedded as udev taskes most anyway Jun 12 08:29:26 that's fair Jun 12 08:29:45 what kind of init does oe use? i assume it's a simple sequential thing as most distributions have Jun 12 08:29:54 still sysvinit Jun 12 08:29:55 i'm still thinking for some devices we just need a good linuxrc Jun 12 08:29:57 together with XorA|gone's static dev population Jun 12 08:30:01 that would speed things up a lot Jun 12 08:30:10 mickey|ICE694: not mine, its already in the init script Jun 12 08:30:19 they are a few replacements which try to parallize Jun 12 08:30:24 mickey|ICE694: Tartarus just told me where to look and added a small patch Jun 12 08:30:25 like minit Jun 12 08:30:45 I think on slow devices like arm parrallel wins nothing Jun 12 08:31:06 that was what I tried by removing udevsettle so udev populated in background Jun 12 08:31:24 the fact it still takes 10s means we are fullyloaded throughout boot Jun 12 08:33:38 one thing that kills us is everytime we load a new command from flash, so we need to cut out all those greps etc from init scripts Jun 12 08:33:51 ACK Jun 12 08:33:52 yep Jun 12 08:33:57 lets get our act togehter and write a small C thingy Jun 12 08:34:00 no runlevels Jun 12 08:34:53 no fancy output Jun 12 08:34:54 mickey|ICE694: you want init from busybox then Jun 12 08:34:54 just setting up the device enough to launch a UI Jun 12 08:34:54 userland can then start services on demand using dbus or whatever Jun 12 08:35:01 people who _need_ a full init can install it afterwards Jun 12 08:35:28 mickey|ICE694: the problem is that there are diferent environments for target devices Jun 12 08:35:54 phone runs other set then devboard or dvbsettopbox Jun 12 08:37:36 or we invent suspend to flash Jun 12 08:37:58 then boot speed is purely controlled by how fast we can uncompress and copy flash Jun 12 08:38:11 no need o invent Jun 12 08:38:18 you can have swap on MTD Jun 12 08:38:26 for sure Jun 12 08:38:26 but that's ok Jun 12 08:38:30 we need minimal customization Jun 12 08:38:32 the C thingy should be generated out of templates Jun 12 08:38:52 mickey|ICE694: your itching to write vala-init :-) Jun 12 08:40:03 * mickey|ICE694 nods and smiles Jun 12 08:40:05 no way i'm doing plain C if I don't need ot Jun 12 08:40:05 s/ot/to/ Jun 12 09:41:18 lips_event-0.1.tar.gz, does somebody have it? Jun 12 09:42:33 Hi! Jun 12 09:42:42 I experience a compilation problem with imagemagick Jun 12 09:42:58 libtoolize: `config/compile' not found in `/home/bot/ademarez/oe/build/tmp/staging/x86_64-linux/usr/share/libtool' Jun 12 09:43:15 autoreconf: libtoolize failed with exit status: 1 Jun 12 09:43:50 can anyone help ? Jun 12 11:16:31 broonie: I have the wm9705 working now after replacing more or less all definitions for switches and regulators. Jun 12 11:18:11 Regulators? Jun 12 11:18:31 * XorA chuckles, bet wolfson didnt know that chip was a PMU Jun 12 11:18:32 But excellent news - please send a patch :) Jun 12 11:18:48 * broonie suspects a language thing and it's actually "volume" Jun 12 11:18:54 broonie: The power management will have to wait... its a little bit complex for now. Jun 12 11:19:12 eh s/regulators/channels Jun 12 11:20:07 that was German English ;) Jun 12 11:22:01 broonie: I guess currrently any call of a SOC_DAPM_* macro is wrong. The wm9705 register set is much closer to the ac97 standard than the one the driver was derived from. Jun 12 11:22:52 ;) Jun 12 11:23:12 florian: dont worry, your German English is much better than my English German Jun 12 11:23:28 Yes, indeed - it doesn't have the PCM interface which is the main reason for having ASoC drivers for the 9712 and 9713; I think most WM9705 users just use the standard AC97 stuff. Jun 12 11:26:27 broonie: right... but since I am using it in a mobile device I will have to care about power consumption soon and looking at the power consumption matrix from the datasheet I see that DAPM is quite important. Jun 12 11:26:48 XorA: :-) Jun 12 11:27:33 Yeah. Anyway, please do send what you've got just now - we may as well keep it in git and perhaps someone will come along and do the DAPM stuff for you if they find it :) Jun 12 11:29:05 broonie: okay :) Jun 12 11:29:34 * hrw prefers French English then German English Jun 12 11:30:01 morning all Jun 12 11:30:14 Nah, the whole doubt/question thing gets on my nerves :) Jun 12 11:31:45 hrw: But you should avoid English French... in France at least ;) Jun 12 11:31:53 hi thesing Jun 12 11:33:16 French English is normally more of an abuse of the language even than American English Jun 12 11:39:23 *g* Jun 12 11:40:24 XorA, I see the UK is re-opening the Tower of London to imprison traitor for up to 42 days Jun 12 11:41:38 03koen 07org.oe.dev * r7ced2522... 10/ (5 files in 3 dirs): linux-omap2 git: bump SRCREV and remove patches that are upstream now Jun 12 11:41:49 Crofton|work: :-( Jun 12 11:42:09 Crofton|work: lieing fuckers making the laws these days Jun 12 11:42:35 universal problem Jun 12 11:42:47 at least they haven't brought back the axe, yet :( Jun 12 11:43:01 Crofton|work: of course Bush will now copy siting evidence from other countries Jun 12 11:43:20 Crofton|work: hopefully the house of lords will bump the law and it will take another year to pass Jun 12 11:43:56 I heard they can delay passage Jun 12 11:44:19 211 days, 16 hours and change until Bush is gone :) Jun 12 11:44:27 yeah, they can deny the law, the government can then re-read it next year Jun 12 11:44:31 ah Jun 12 11:44:40 Lords cant deny forever though :-( Jun 12 11:44:47 at least we know when the next election is :) Jun 12 11:44:58 brb+ Jun 12 11:45:12 we so need to abolish government here Jun 12 11:45:23 all they do these days is spend money and lie about it Jun 12 11:46:10 that sounds familiar ... Jun 12 11:46:17 they cant even do simple addition and subtraction so now we are in trouble with the EU as well :-( Jun 12 11:47:01 I thought UK policy was to always be in troble with the EU? Jun 12 11:47:19 I wonder when they realise if they classify the other parties as terrorists, they can lock them up with no charges :-) Jun 12 11:47:38 Crofton|work: unfortuneately with the xenophobic morons in charge that is true Jun 12 11:48:32 I had fun in .ie pointing out that they had statues and train stations names after "terrorists" Jun 12 11:49:13 there are a million Nelson Mandela (Street/Road/Court/....) in the UK and he is defineately officially a terrorist Jun 12 11:51:13 * Crofton|work wishes the Libertarians weren't all crazy :) Jun 12 11:52:09 * XorA wishes more fights were settled inside a ring Jun 12 11:52:12 Holy crap, the Belgians are trying to buy Budweiser! Jun 12 11:52:19 UFC world politics would rock Jun 12 11:53:06 Damn EU, buying all the American companies Jun 12 11:53:20 Crofton|work: well you guys are so cheap these days :-) Jun 12 11:53:24 some guys I know are getting bought by Finnmechanica Jun 12 11:53:27 yeah Jun 12 11:53:43 Crofton|work: we will revoke your indepdance one way or another :-D Jun 12 11:53:53 :) Jun 12 11:54:03 and teach you to spell Jun 12 11:54:12 * Crofton|work knows how to spell Jun 12 11:54:33 although I actually prefer the amican simplifications Jun 12 11:54:42 make more sense than our insane antiquities Jun 12 11:54:43 * Crofton|work used to get stuff marked wrong in first grade due to use of proper English Jun 12 11:55:08 Crofton|work: where proper mean British? Jun 12 11:55:20 hrw, of course :) Jun 12 11:55:31 it is like driving on the proper side of the road Jun 12 11:55:53 after our .ie trip, some guys stayed a few extra days and rented a car Jun 12 11:56:10 they kept the car for five hours before returning it wrecked Jun 12 11:56:17 hehehehehehehhe Jun 12 11:56:37 you guys were in Dublin? Jun 12 11:57:38 just outside Jun 12 11:57:40 Maynooth Jun 12 11:58:04 only thing its safe to drive inside Dublin is a cheiftan tank Jun 12 12:00:02 there were well outside Dublin Jun 12 12:00:28 apparently the driver struck a parked car, due to sitting on the "wrong" side of the car Jun 12 12:01:06 :-D Jun 12 12:04:03 .ie is left driven? Jun 12 12:14:16 hrw yes, they drive on the left side of the road Jun 12 12:17:19 heh.. I was thinking that they are normal Jun 12 12:18:10 25s udev is insane Jun 12 12:45:41 zecke: Did you see the email to oe-priv about scms? Jun 12 12:46:25 RP: not yet, I didn't have access to my university imap folder... Jun 12 12:47:12 zecke: ok, I think you're only one we're missing a reply from :) Jun 12 12:51:01 RP: just read that... I want darcs Jun 12 12:51:26 zecke: I really hope you're joking ;-) Jun 12 12:51:28 hey guys.. how can i switch from normal kernel branch to andy branch when bitbaking an openmoko kernel? Jun 12 12:54:30 cekay: you can build openmoko-kernel-devel... Jun 12 12:54:58 this pulls the andy branch from git? Jun 12 12:57:27 cekay: this is building a rev from andy's branch... Jun 12 12:57:42 ok thanks :) Jun 12 13:02:19 cekay: you use want to use autorev or update the srcrev for the branch though :) Jun 12 13:07:10 a successfull build would be enough for me today ;-)) Jun 12 13:32:22 zecke: thanks, I've sent an RFC now :) Jun 12 14:37:08 has anyone managed to compile packages found using the code sourcery toolchain as an external toolchain Jun 12 14:37:36 your board has a phoenix bios? Jun 12 14:37:39 dammit Jun 12 14:37:41 sorry Jun 12 14:37:52 pardon me? Jun 12 14:39:31 hehehe Jun 12 14:53:33 rootfs_ipk_do_rootfs: command not found Jun 12 14:53:36 balls thats annoyuing Jun 12 14:53:40 also annoying Jun 12 15:05:02 i'm getting this when trying to build an image where the distro inherits "package_ipk" Jun 12 15:05:08 rootfs_ipk_do_rootfs: command not found Jun 12 15:05:14 any ideas anyone? Jun 12 15:06:27 works for me Jun 12 15:06:41 devel branch? Jun 12 15:08:58 ok - it worked for me Jun 12 15:09:08 my last .dev build was over week ago Jun 12 15:10:24 install the command in your ram image? Jun 12 15:16:20 mattface: started qemuarm/angstrom-2008.1 build now Jun 12 15:25:08 hrw: noted Jun 12 15:30:30 Hi thesing, the initramfs image I build some time ago does not do anything. The machine does not even turn on Jun 12 15:31:08 did you check with a serial cable? Jun 12 15:34:26 mattface, what distro? Jun 12 15:35:41 Fedora 9? Jun 12 15:35:49 bbl Jun 12 15:36:31 03jeremy_laine 07org.oe.dev * r0adb4216... 10/ (1 packages/libosip2/libosip2_3.1.0.bb): libosip2: add libosip2 version 3.1.0 Jun 12 15:36:36 03jeremy_laine 07org.oe.dev * r3fbd6047... 10/ (1 packages/libexosip2/libexosip2_3.1.0.bb): libexosip2: add libexosip2 version 3.1.0 Jun 12 15:59:06 Crofton|work: Debian Lenny Jun 12 16:38:48 bye Jun 12 16:52:05 * * OE Bug 4373 has been created by udovdh(AT)xs4all.nl Jun 12 16:52:07 * * /etc/asoundrc must have different name Jun 12 16:52:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4373 Jun 12 17:01:23 I am using SRC_URI="git://git.kernel.org/pub/scm/linux/kernel/git/acme/python-ethtool.git;protocol=git", and when I bitbake my package, it tries to do an HTTP get from the angrstrom website, any reason why? Jun 12 17:05:57 Fidget: Its trying the tarball stash first? Jun 12 17:06:26 RP: why would there be a tarball stash? Jun 12 17:06:33 Can I disable that? Jun 12 17:07:05 disabling it would make fetching really slow Jun 12 17:07:24 it uses the hostname in the filename anyway Jun 12 17:07:27 Only want to on a per package basis. Jun 12 17:07:35 Fidget: SRC_TARBALL_STASH = "" or CVS_TARBALL_STASH = "" probably Jun 12 17:07:47 I don't know which one will be set for you Jun 12 17:07:56 Obviously if I am using some kernel modules that are unique to me, I'd like to skip the stashes and go straight to the source Jun 12 17:08:16 RP: In the package.bb or on the command line to bitbake? Jun 12 17:08:33 I don't want to globally disable tarball stash checking Jun 12 17:08:41 I want to disable it on a per package basis Jun 12 17:21:37 Fidget: Add it to the specific recipe then Jun 12 17:21:48 RP: thank you Jun 12 19:18:06 thesing: No Jun 12 19:18:18 But the lights don't come up, there is absolutely no reaction Jun 12 19:18:31 Do you have an image for me that I can test? Jun 12 19:18:43 althought it will probably be next week Jun 12 19:21:08 Laibsch: the initramfs image tries to kexec a new kernel from / or /boot (not sure). the normal images don't have a kernel there, thats probably your problem. Jun 12 19:29:40 03mickeyl 07org.oe.dev * r16fe73fb... 10/ (1 packages/images/fso-image.bb): fso-image: ship sounds, add info 'bout state files Jun 12 19:30:47 03mickeyl 07org.oe.dev * rdc2ea45e... 10/ (1 packages/freesmartphone/fso-sounds.bb): add fso-sounds, notification sounds for the freesmartphone image Jun 12 20:21:47 thesing: probably Jun 12 20:21:59 Maybe you can write a small HOWTO in the wiki Jun 12 20:22:25 I'd like to get this stuff usable more mainstream Jun 12 20:22:32 I'm a good test ;-) Jun 12 20:23:19 I think one only needs to change one line in the linux-rp recipe to get zImages included. Jun 12 20:28:30 http://butterfeet.org/?p=57 Jun 12 20:29:06 Garmin using gtk and that file list looks very suspicous :) Jun 12 20:49:03 RP: So when do the garmin goodies start rolling in :) Jun 12 20:49:26 hi hvontres|work: how do you do? Jun 12 20:50:37 ant_: not too bad Jun 12 20:51:13 now that RP is getting billionaire he'll forget us poor Zaurus people... Jun 12 20:55:01 helgeD_: I meant to write back to you earlier, but those 2.6.23 modules in the 2.6.24 image ar3e being pulled in by a RRECOMMENDS some place. I think those modules were re-named between 2.6.23 and 2.6.24 and we need to figure out which config file needs updateing. Jun 12 20:55:51 hvontres|work: hairy issue! I cowardly jumped off 2.6.23 wagon ;-) Jun 12 20:55:52 hvontres|work: ok. Jun 12 20:57:03 helgeD_: I need to check when I get home, but I think at least the rtc module was fixed in .dev. Have you noticed any issues with the clock not working btw? Jun 12 20:57:04 hvontres|work: out of curiosity: is udev_118 "working" on your Z ? Jun 12 20:57:29 ant_: i haven't had a chance to test that lately Jun 12 20:57:31 hvontres|work: yes, I saw the clock drifting a lot. Jun 12 20:57:45 ok, it's the .dev version Jun 12 20:57:50 * hvontres|work fires up laptop to take a look Jun 12 20:57:51 hvontres|work: I use NTP, so it was not so important for me, but it's annyoing Jun 12 20:58:03 hvontres|work: esp. after wake-up Jun 12 21:04:29 helgeD_: did you create a valid /etc/localtime ? Jun 12 21:05:52 03koen 07org.oe.dev * rfcd08145... 10/ (3 files in 2 dirs): gcc-cross-kernel: add 4.2.4 and csl-2007q3 Jun 12 21:07:11 helgeD_: hmm. it looks like the rtc driver is compiled into the kernel in 2.6.24 Jun 12 21:07:54 helgeD_: and I think the other module was left over from the whole WPA-crypto mess Jun 12 21:07:59 ant_: does a /etc/localtime influences the drifting? Jun 12 21:08:21 at resume, I had strange issues w/out this file. GPE bug? Jun 12 21:08:26 hvontres|work: yeah, they are not needed at least Jun 12 21:08:39 ant_: I'm using opie, not gnome Jun 12 21:09:24 ok, just check because the file is not provided (los of programs rely on it, at least for servers) Jun 12 21:10:56 regarding the drift, I know about it.. my old server (supermicro - dual BX) suffering from a broken rtc... Jun 12 21:13:14 ant_: yes, it's missing. Jun 12 21:13:23 How to create it best ? copy over from PC? Jun 12 21:13:57 copy from /usr/share/zoneinfo/yourzone Jun 12 21:14:04 copy, not link Jun 12 21:17:37 helgeD_: is it drifting or does the clock stop during suspend? IIRC when the rtc module was missing in 2.6.23, the time just didn't update Jun 12 21:18:14 helgeD_: BTW, it looks like for 2.6.24, the rtc driver is compiled in again Jun 12 21:20:10 hmm... my flash is full. not even space for /etc/localtime Jun 12 21:20:55 there are drawbacks at boot using a symlink, but it could work Jun 12 21:22:52 helgeD_: http://www.redhat.com/archives/rhl-devel-list/2006-April/msg00023.html Jun 12 21:29:20 ant_: thx Jun 12 21:31:00 03koen 07org.oe.dev * rc95a34ec... 10/ (1 conf/distro/angstrom-2008.1.conf): Jun 12 21:31:00 angstrom 2008: use gcc 4.3.1 for armv7a now that NEON works, use a seperate gcc (sadly a CSL one) for kernels Jun 12 21:31:00 * this is still missing a mechanism to build gcc-cross-kernel automagically when needed Jun 12 21:38:56 for a new machine type that requires custom kernel patches is it most appropriate to put those in packages/linux/linux-/? I see a few conventions used for kernel patches Jun 12 23:02:21 Does angstrom currently build in OE? home/diefool/oe/build/tmp/cross/lib/gcc/arm-angstrom-linux-gnueabi/4.2.4/../../../../arm-angstrom-linux-gnueabi/bin/ld:.libs/libgdk_pixbuf-2.0.v-2.0.ver:2: syntax error in VERSION script Jun 12 23:13:03 03koen 07org.oe.dev * r0b3b2489... 10/ (5 files in 5 dirs): omap3evm: add omap3evm machine and make linux-omap2 kernel the default for that. Based on work from Steve Sakoman Jun 12 23:16:17 having some trouble trying to follow how oe builds a kernel. I see that task-boot.bb has a 'DEPENDS="virtual/kernel"' not sure how 'virtual/kernel' relates to any recipe in packages/linux Jun 12 23:33:16 tharvey: other recipes simply claim to satisfy the dependency for virtual/kernel -- hopefully those recipes do, in fact, build a kernel! Jun 12 23:33:59 Specifically which kernel recipe is selected depends on other criteria in the environment, such as "PREFERRED_PROVIDER" vars. Jun 12 23:35:47 which machine you choose in your local.conf also picks which kernel is used Jun 12 23:38:10 I think I'm missing how a package tells bb that it provides something. For example ok... that makes sense. So for example, conf/machine/mx31ads defines 'PREFERRED_PROVIDER_virtual/kernel = ${MACHINE}-kernel' and I would assume this would be provided by packages/linux/mx31ads-kernel_2.6.19rc6.bb but I'm not sure what in that file tells bb that virtual/kernel is satisfied by it Jun 12 23:39:37 I expected to see something like 'PROVIDES=virtual/kernel' in packages/linux/mx31ads-kernel_2.6.19rc6.bb Jun 12 23:41:42 tharvey: a lot of that stuff is in the bbclass files that are "inherited" by the specific bb file -- check the includes and inherits, and if you decode the logic you should find a PROVIDES in there (although it may be done at a low level, so grepping for "PROVIDES" probably won't do it) Jun 12 23:41:51 duh... it has an 'inherit kernel' which has the PROVIDES += "virtual/kernel" Jun 12 23:42:28 yeah... I was looking for something I could grep to see what is provided where... Jun 12 23:43:00 Somewhere in there is also the logic that automagically breaks the kernel up into packages, so that each kernel module is a separate package that can be "RDEPENDS" by other software, etc. Jun 12 23:44:07 now I understand why bb has to scan every bb file in the path.... Jun 12 23:44:45 so the word 'virtual' is not special at all Jun 12 23:46:05 ok, so I'm still trying to understand how various kernels have handled patches specific to them. I guess ultimately its completely up to the recipe but if I'm going to have to add custom kernel patches that relate to a specific machine is there a suggested place to put the patches in the oe tree? Jun 12 23:49:07 03koen 07org.oe.dev * r15fcb1e4... 10/ (1 conf/distro/angstrom-2008.1.conf): angstrom 2008: remove bogus preferred version Jun 12 23:49:30 for example, if I were to add a machine type for gateworks avila boards, which are ixp4xx but require some kernel patches to mainline, I guess I would create packages/linux/avila which contained a recipe and patches for various kernel versions I want to support right? Jun 12 23:51:46 close Jun 12 23:51:55 tharvey: You would want to speak with Rod Whitby, who maintains the ixp4xx kernels in OE. Jun 12 23:52:05 packages/linux/linux-your-recipie-i-assume-is-avila-2.6.2N/avila/ Jun 12 23:52:14 er, misread :) Jun 12 23:52:26 it would be linux-ixp4xx-2.6.N/avila/ Jun 12 23:52:49 then modify the existing recipe to patch in your stuff, on avila Jun 12 23:53:03 is SRC_URI_avila += "file:///patch1.patch;patch=1" Jun 12 23:53:03 etc Jun 12 23:53:58 03koen 07org.oe.dev * rd9f9c022... 10/ (17 files in 4 dirs): linux-omap2 git: update evm defconfig (thanks Steve!), clean up beagle patch dir, apply smartreflex and cpuidle patches to evm as well Jun 12 23:54:06 ah... I see what your saying... as a generic ixp4xx kernel recipe already exists, use the infrastructure its setup to deal with patches for various variants Jun 12 23:55:10 yeap Jun 12 23:55:18 it should already do it for some, even Jun 12 23:55:25 if not, there's other examples that do (ie "tosa" board) Jun 12 23:55:38 Coordination with rwhitby (I'll be happy to help as well, i handle the SlugOS ixp4xx kernels) would be valuable to ensure that patches of general interest are not duplicated, and don't get missed. Jun 12 23:57:07 tharvey: g'day Jun 12 23:58:02 ok so back on the virtual/kernel question: I see how packages/linux/mx31ads-kernel_2.6.19rc6.bb now provides 'virtual/kernel' but I'm not sure I understand how PREFERRED_PROVIDER_virtual/kernel = "${MACHINE}-kernel" in conf/machine/mx31ads.conf makes bb choose that recipe b/c mx31ads-kernel != mx31ads-kernel_2.6.19rc6rc6' to it? Jun 12 23:58:53 rwhitby, greets! I may be interested in adding gateworks avila support in oe - specifically there are some kernel patches required - so I was asking how I would go about doing so Jun 12 23:59:09 tharvey: there are two options: Jun 12 23:59:17 1) Do it in OE, and manage the patches there. Jun 12 23:59:45 2) Get an nslu2-linux svn client cert, and manage the patches in our kernel svn repo whilst they're waiting to go upstream. Jun 13 00:00:34 As you will notice, the ixp4xx-kernel recipe has no local patches, cause we're methodically pushing all the patches upstream. As on 2.6.27, we should only have about 2 patches left outstanding. Compare this to the other kernel recipes in OE ;-) Jun 13 00:01:04 ah yes... I've seen (and used) the nslu2-linux patches for some previous kernels (outside of oe) and I find that nslu2 uses oe (right?). A lot of the patches have been waiting to get upstream for many kernel versions though... why not put them into oe? Jun 13 00:01:41 no reason other than they'll probably get forgotten in OE, since they're not with all the rest of the ixp4xx patches out there ... Jun 13 00:02:06 i.e. if you're going to have a set of non-upstream patches for ixp4xx, then it's probably a good idea to keep them all in one place if you can. Jun 13 00:02:10 so is it true that nslu2 uses oe? Jun 13 00:02:31 nslu2 has five different firmware distributions, three of which use OE. Jun 13 00:02:49 yes... thats where I've gotten very confused Jun 13 00:03:10 what causes the confusion? Jun 13 00:03:33 5 different distros with names that have changed over time... Jun 13 00:03:59 i never realized that the nslu2-linux kernel patches were generic and possibly used by multiple distros Jun 13 00:05:23 makes sense now that I think about it - nslu2 is a hardware device are generic to it and distro agnostic Jun 13 00:09:11 I notice 'COMPATIBLE_MACHINE' being defined in kernel recipes - perhaps this is whats used with PREFERRED_PROVIDER? What I'm trying to do is create a new machine that is based off a mx31ads so I want it to use the mx31ads kernel and I'm geting 'multiple providers are available for virtual/kernel' so my PREFERRED_PROVIDER_virtual/kernel = mx31ads-kernel is apparently not enough Jun 13 00:09:37 do I need to add my machine name to COMPATIBLE_MACHINE In packages/linux/mx31ads-kernel_2.6.19rc6.bb? Jun 13 00:11:33 gotta run... will be back tomorrow with questions :) thx for the help Jun 13 01:31:13 i cant build freaking any OE distributions Jun 13 01:31:22 even generic fails on freetype Jun 13 01:31:33 ./configure: line 6637: syntax error near unexpected token `build_libtool_libs,' Jun 13 01:31:33 ./configure: line 6637: ` _LT_DECL(build_libtool_libs, enable_shared, 0,' Jun 13 01:56:02 try Angrstaom 2008 Jun 13 01:56:10 what machine are you building for? Jun 13 01:59:57 Poodle Jun 13 02:00:02 I tried 2008, fails on gdk Jun 13 02:00:09 2008? or 2008.1? Jun 13 02:00:45 Crofton|work: Have you had success building angrstaom? Jun 13 02:01:08 console-image builds for me on the beagleboard earlier today Jun 13 02:01:24 MACHINE ="beagleboard" Jun 13 02:01:29 DISTRO = "angstrom-2008.1" Jun 13 02:01:48 I've been trying to build gpe-image Jun 13 02:01:54 Crofton|work: err can you buy a beagleboard yet? Jun 13 02:02:28 not yet Jun 13 02:02:41 you could also try omap5912osk Jun 13 02:03:19 I'm going to run an OSK build on my laptop over night Jun 13 02:03:22 I have looked at the hardware once and it is incredible what they put on that small board Jun 13 02:03:32 yeah Jun 13 02:03:57 *rm -rf's tmp and tries again* Jun 13 02:03:58 I think more will be available over the next few months, but you will need a good story to get one Jun 13 02:04:10 df00z, what machine and distro? Jun 13 02:04:16 and what are you building? Jun 13 02:04:28 * dcordes makes up a good story Jun 13 02:04:40 :) Jun 13 02:04:54 Crofton|work: i'm using org.openembedded.dev, tried angstrom-2008.1, gpe-image failed building GDK Jun 13 02:05:04 http://beagleboard.nl Jun 13 02:05:04 tried "generic", failed building freetype Jun 13 02:05:11 start with console-image Jun 13 02:05:20 generic has known issues Jun 13 02:05:29 console-image prolly will build...i can build nano and all prereqs Jun 13 02:05:38 df00z, what machine? Jun 13 02:05:43 poodle Jun 13 02:05:59 I have the PXA255 so i wanted to compile a custom kernel Jun 13 02:06:02 I don't build gpe Jun 13 02:06:06 plus its nice to be able to compile my own packages Jun 13 02:06:06 ah Jun 13 02:06:11 right Jun 13 02:06:53 try filing a report in bugzilla, also try and catch the .eu crowd earlier in the day Jun 13 02:07:02 ah Jun 13 02:07:23 I'll prolly try angstrom 2008.1 again..i don't remember if i tried 2008 or 2008.1... Jun 13 02:07:37 2008.1 Jun 13 02:07:57 if you complain about generix, we will tell you to build Angstrom Jun 13 02:08:15 then all admit we need to change generix to minimal.distro and fix it :) Jun 13 02:08:16 allright. and i have to use the dev tree, right? Jun 13 02:08:25 most likely Jun 13 02:08:34 I am not a poodle guy Jun 13 02:08:43 ah Jun 13 02:08:48 meh..all the same Jun 13 02:08:54 df00z: you can of course also work with the stable metadata Jun 13 02:09:07 I'm more a black lab kinda guy, m'self. Jun 13 02:09:13 :p Jun 13 02:09:27 dcordes: i cant build the x86 emulator w/ that Jun 13 02:09:36 gcc 4.2.3 Jun 13 02:09:43 not bochs..the other one Jun 13 02:09:50 cant think of the name of the one OE requires Jun 13 02:10:06 qemu Jun 13 02:10:11 yeah Jun 13 02:11:41 I need a brilliant and scientific project for the beagleboard Jun 13 02:14:54 what the heck, it's got a 3d accelerator? Jun 13 02:16:44 df00z: what? Jun 13 02:18:12 that beagleboard Jun 13 02:18:16 df00z: seems quite a few mobile procs are using the PowerVR Jun 13 02:18:26 df00z: i.MX31 includes a PowerVR I think **** ENDING LOGGING AT Fri Jun 13 02:59:57 2008