**** BEGIN LOGGING AT Mon Sep 12 02:59:57 2011 Sep 12 09:28:00 RP__: morning and ping Sep 12 09:33:12 dongxiao: morning and pong Sep 12 09:34:42 RP__: For some RDEPENDS that is dynamically set, how can we ensure that those RDEPENDS packages are built out before rootfs? Sep 12 09:35:17 dongxiao: Do you have an example? Sep 12 09:36:00 yes, in connman.inc, ofono wpa-supplicant bluez4 are dynamiccaly set as RDEPENDS Sep 12 09:36:37 RP__: in our sato image we don't have problem because there is "3g" set as DISTRO_FEATURES, so those packages are built out. Sep 12 09:37:32 RP__: but currently I am testing multilib case of a normal qemux86-64 image containing lib32 version of connman, we wil find lib32-ofono, lib32-wpa-supplicant, and lib32-bluez4 are requires by rootfs, but they are not built out. Sep 12 09:40:28 dongxiao: Well, we consider it necessary to list those dependencies elsewhere other than connman in the normal target case. We therefore just have to accept this is the case in the multilib case too? Sep 12 09:42:39 RP__: Hi, disabling PSEUDO in 72252799e8c51a633a231a2cd1fe797b7faae713 breaks chown in do_install in resolvconf recipe Sep 12 09:43:57 JaMa: latest bitbake or not? Sep 12 09:44:54 last 2 commits missing, I have 24272dae15ccf641ece11ef5a6e2bfa3ebb6f5f9 Sep 12 09:45:33 will retest with HEAD Sep 12 09:46:17 RP__: hmm, I see your point. currently we set MULTILIB_IMAGE_INSTALL = "lib32-connman" in local.conf to let a normal qemux86-64 containing lib32-connman, so we need to let user also setting lib32-task-base-3g, lib32-task-base-wifi, and lib32-task-base-bluetooth in MULTILIB_IMAGE_INSTALL? Sep 12 09:48:17 dongxiao: I think so Sep 12 09:48:52 RP__: ok thanks:) I have several multilib rpm fixes sent out for review. Sep 12 09:50:06 dongxiao: I saw, thanks Sep 12 09:50:32 btw, if people want my attention, please cc rpurdie@rpsys.net Sep 12 09:50:54 My LF accounts are not usable at this time :( Sep 12 09:51:45 RP__: seems that LF website down also influent our normal build Sep 12 09:52:10 dongxiao: The autobuilders and other servers are connected with the LF racks Sep 12 10:01:59 RP__: resolvconf do_install works fine with HEAD, thanks! Sep 12 10:05:10 JaMa: great :) Sep 12 12:53:41 Hi, assuming a recipe is written such that it downloads source from a git server, say git.kernel.org but that server is down, what is the suggest work-around? should I change the udev recipe and change it back once the servers are back up, for example? Sep 12 13:14:44 holgerwrs: We should have MIRRORS support which handles this Sep 12 13:15:39 holgerwrs: See MIRRORS in mirrors.bbclass Sep 12 13:16:46 k Sep 12 13:52:08 RP__: should mirrors.bbclass support git: SRC_URIs ? Sep 12 13:52:44 holgerwrs: It should be able to yes Sep 12 14:49:01 any word on when the bugzilla is going to come back? Sep 12 14:54:52 is wiki.pokylinux.org down? Sep 12 15:00:16 hollisb: yes, it's connected with the current linuxfoundation outage Sep 12 15:04:12 oh, the kernel.org thing? Sep 12 15:06:07 there's a notice on the yoctoproject.org website, if you read that you know as much as I do :) Sep 12 15:06:53 well... not so much ":)" but anyway Sep 12 15:07:05 ok, thanks! Sep 12 17:55:00 ugh.. I really need to get onto the yocto bugzilla.. (even read-only) Sep 12 18:11:37 fray, i know an rpm multilib bug you could work on… =) Sep 12 18:36:04 it looks like Dongxiao submitted some patches Sep 12 20:58:29 fray: ping Sep 12 21:46:29 anyone notice hob requires MACHINE to be defined? Sep 12 21:46:42 seems counter intuitive Sep 12 21:54:30 is there a latest poky git somewhere else? Sep 12 21:54:33 github or something? **** ENDING LOGGING AT Tue Sep 13 01:52:38 2011 **** BEGIN LOGGING AT Tue Sep 13 19:40:15 2011 Sep 13 20:38:50 ok, looking for assistance w/ QEMU.. Sep 13 20:39:23 first I'm on a shared machine adn getting access to being able to sudo was "difficult".. we really need a mode where you don't need root access to use qemu Sep 13 20:39:34 second, the machine I'm on doesn't have kvm support.. so I had to modify the recipe.. Sep 13 20:39:52 and third (where I'm stuck now), the machine doesn't support the tun interface.. so um, how do I run QEMU? Sep 13 20:40:02 (I don't need network support, just to be able to run) Sep 13 20:41:17 any suggestions short of "find another machine"? Sep 13 20:42:34 'er.. make that the system doesn't support tap.. but does support tun.. (I think) Sep 13 20:50:57 fray, I just copy the qemu command line and edit it manually Sep 13 20:51:08 iirc tun is the only thing you really need sudo for anyway Sep 13 20:51:27 ya.. I really hate the hardcoded -enable-gl as well.. Sep 13 20:51:29 does anyone know how hob determines which images are available? just the ones in image_types class? Sep 13 20:51:48 in reference to tar.gz tar.bz2 ext2.gz, etc Sep 13 20:51:52 We should really have some disable options for that script Sep 13 20:51:55 I don't think I have a single host where -enable-gl works Sep 13 20:52:20 msm, I'm guessing here, I think those are defined by the machine conf Sep 13 20:52:21 msm: it uses the list in image_types.bbclass Sep 13 20:52:38 msm: You can append to the list whereever you define extra ones Sep 13 20:52:41 msm, weigth RPs reply much heavier than mine Sep 13 20:53:05 ah, there is a list.. ok Sep 13 21:01:16 ok, running qemu here I'm not geting any output.. I don't know if the system is crashing, qemu is running or not.. Sep 13 21:01:34 is there something I can pass to the kernel command line or elsehwere to re-enable the kernel boot messages? Sep 13 21:02:06 try the --no-gui or similar Sep 13 21:02:21 this is a core-image-minimal.. I'm simply not getting the kernel console messages.. Sep 13 21:02:35 -nographics Sep 13 21:02:38 right Sep 13 21:02:38 I suspect the default kernel has verbose turned off or something.. but I don't remember the equivlaent command to enable it Sep 13 21:02:47 this will use the console instead of an SDL console Sep 13 21:02:49 already using that Sep 13 21:02:54 and you get nothing? Sep 13 21:02:56 huh Sep 13 21:02:57 anyone know if kergoth is still on vacation? Sep 13 21:02:59 Running qemu... Sep 13 21:03:08 /msp-lpggp12/mhatle/git/oss/oe-core/build-x86/tmp-eglibc/sysroots/i686-linux/usr/bin/qemu -kernel /msp-lpggp12/mhatle/git/oss/oe-core/build-x86/tmp-eglibc/deploy/images/bzImage-qemux86.bin -hda /msp-lpggp12/mhatle/git/oss/oe-core/build-x86/tmp-eglibc/deploy/images/core-image-minimal-qemux86.ext3 -show-cursor -usb -usbdevice wacom-tablet -no-reboot -nographic -m 128 --append "root=/dev/hda rw mem=128M oprofile.timer=1 " Sep 13 21:03:20 that is the last thing I see.. and it hangs util I do the ctl-a x Sep 13 21:04:10 you can add "earlyprintk=vga debug" Sep 13 21:04:37 msm: got back last wed, spent thurs & fri getting caught up and building momentum back up. what's up? Sep 13 21:04:38 I haven't tried vga with -nographics... not sure if that is plumbed correctly Sep 13 21:04:53 kergoth: err, i swear i jus checked for your nick Sep 13 21:05:11 evice wacom-tablet -no-reboot -nographic -m 128 --append "root=/dev/hda rw mem=128M oprofile.timer=1 " Sep 13 21:05:13 qemux86 login: Sep 13 21:05:20 thats what I get when it works.. I'm still not getting kernel messages Sep 13 21:05:36 try appending console=tty0 Sep 13 21:06:41 alternatively, try eliminating -nographics and seeing if earlyprintk=vga works there Sep 13 21:07:45 console does nothing Sep 13 21:07:54 I can't run with graphics, the machine doesn't have X Sep 13 21:08:15 ah Sep 13 21:08:41 the boot stuff is disabled, but I can't for the life of me remember how to reenable it Sep 13 21:10:38 ah, -nographic redirects output to the emulated serial console Sep 13 21:10:54 so you would need: console=ttyS0 earlyprintk=ttyS0 Sep 13 21:11:03 it's possible you'll have to tweak baud rates, etc Sep 13 21:11:11 maybe using the -serial argument Sep 13 21:11:21 one of those did it.. console or earlyprintk Sep 13 21:11:29 fray: When you get it working can you write it in the wiki or something please? Sep 13 21:11:43 RP__ I have this -very- hacked up.. Sep 13 21:11:43 or file a bug and put the details in it Sep 13 21:11:54 I had to remove all of the networking and graphics support in order to get this far.. Sep 13 21:11:59 fray: Having some kind of hack documented somewhere is better than nothing Sep 13 21:11:59 maybe "-serial stdio" Sep 13 21:12:05 I'll file a bug on it and we can do it "better" again Sep 13 21:12:54 what item is qemu under? Infrastructure, Poky or metadata layers? Sep 13 21:13:14 fray, I'd try "-nographics -serial=stdio --append 'console=ttyS0,115200 earlyprintk=ttyS0,115200'" Sep 13 21:13:56 fray: don't know :/ Sep 13 21:16:50 RP_ 1474 -- enhancement request to avoid sudo/tap Sep 13 21:19:06 1475 -- no kernel boot messages w/o passing additional params Sep 13 21:21:25 RP/sgw -- FYI.. I finally have the prelink reproducer.. I'm working on narrowing down the problem (on x86 at least) Sep 13 21:21:56 it's sad the only 32-bit host I have access to is an ancient RHEL5 Sep 13 21:26:19 fray: ouch Sep 13 22:06:06 If I want to add dropbear to my current minimal image build (or rather rebuild with dropbear installed in the image) what is the simplest way to do that? Sep 13 22:06:17 can I just add it to EXTRA_IMAGE_FEATURES in my local.conf? Sep 13 22:07:25 afaik that should work Sep 13 22:07:33 is dropbear not in minimal already? Sep 13 22:07:43 it's currently building.... but I was impatient :) Sep 13 22:07:54 for somereason it started building the kernel again Sep 13 22:07:57 I havn't tried it, but Richard mentioned POKY_EXTRA_INSTALL += "package" Sep 13 22:08:13 huh Sep 13 22:08:44 the core-image.bbclass has references to POKY_EXTRA_INSTALL (and a few other things) Sep 13 22:09:20 I'm a bit confused by what, if anything, loads core-image.bbclass.. Sep 13 22:09:32 http://www.yoctoproject.org/docs/1.1/poky-ref-manual/poky-ref-manual.html#usingpoky-extend-customimage-imagefeatures Sep 13 22:09:34 any of our core-image-x Sep 13 22:09:44 this suggests the *IMAGE_FEATURES I think Sep 13 22:10:03 POKY_EXTRA_INSTALL is a Poky specific EXTRA_IMAGE_INSTALL variable, iirc it's just included in the latter Sep 13 22:10:32 EXTRA_IMAGE_INSTALL vs. EXTRA_IMAGE_FEATURES ? Sep 13 22:10:36 this is a problemt hat we have and we're currently discussing elsewhere.. Sep 13 22:10:43 what is the right way to do this.. Sep 13 22:10:52 to me EXTRA_IMAGE_FEATURES is not a "package" Sep 13 22:11:03 it's a feature.. like diretfb, x11, eglibc, uclibc, etc.. Sep 13 22:11:08 agreed, non-intuitive Sep 13 22:11:16 but that docs say to use ssh-server-dropbear Sep 13 22:11:19 not "dropbear" Sep 13 22:11:20 so if that is the right way to add a package, I think it's poorly named at the least Sep 13 22:11:45 thats the other issue IMHO with this approach.. people, generally, have a knowledge of the name of the recipes and may not have the knowledge of the produced package name.. Sep 13 22:11:47 EXTRA_IMAGE_INSTALL is not noted int he ref manual anywhere Sep 13 22:11:52 all of these things require the name of the produced package Sep 13 22:12:08 I've used IMAGE_INSTALL += "dropbear" before Sep 13 22:12:14 in a custom image recipe Sep 13 22:12:24 but perhaps that has changed recently Sep 13 22:12:52 * dvhart goes to pick up a kindergartner... Sep 13 22:12:55 brb Sep 13 22:13:57 fray: we are looking at re-orging core-image. Sep 13 22:14:08 fray: to get the x stuff out of it. Sep 13 22:14:24 underlying problem is.. how do I tell the system (the right way) to add a specific package [or packages] to my image.. Sep 13 22:15:54 sorry, yes - POKY_EXTRA_INSTALL is included in IMAGE_INSTALL Sep 13 22:16:06 EXTRA_IMAGEDEPENDS += seems still on duty Sep 13 22:16:29 IMAGE_INSTALL is the right way to add a package (the output of a recipe) Sep 13 22:16:52 will conf/local.conf: IMAGE_INSTALL += "bash" work? Sep 13 22:17:05 fray: yes, but that will affect every image Sep 13 22:17:09 not usually desirable Sep 13 22:17:20 if it's in my conf/local.conf, I would expect it to affect all images Sep 13 22:17:28 * incandescant has known of people wondering why they can't stop an item being included Sep 13 22:17:38 fray: you would expect that, yes ;-) Sep 13 22:17:51 I understand, that is where you need to do custom tasks and images right now Sep 13 22:18:41 POKY_EXTRA_INSTALL only affects images which inherit core-image.bbclass, which may or may not be useful to some Sep 13 22:43:30 adding ssh-server-dropbear to EXTRA_IMAGE_FEATURES appears to work Sep 13 22:43:41 grrr... it's rebuilding the kernel again though Sep 13 22:44:07 dvhart: do you have the kernel-dev (or whatever) layer enabled? Sep 13 22:44:30 I do have meta-kernel-dev enabled Sep 13 22:45:04 I expect that's you 'problem' Sep 13 22:45:18 anything in particular about it that you suspect triggers a rebuild? Sep 13 22:45:20 AUTOREV? Sep 13 22:45:37 I'd assume so but I've *never* looked at that layer Sep 13 22:46:09 * incandescant would disable it and see if the kernel is still rebuilt Sep 13 22:46:15 I guess that's something to add to the things to investigate list Sep 13 23:33:47 umm Sep 13 23:34:10 should bash DEPEND on ncurses? or RDEPEND? Sep 13 23:34:15 i think there is some nasty circular dep Sep 13 23:34:25 or its time for me to go home Sep 13 23:35:17 which was causing bash and ncurses to always be built Sep 13 23:38:30 well lets just rephrase everything. how can I get a recipe to not depend on libc? Sep 13 23:40:54 INHIBIT_DEFAULT_DEPS = "1" **** ENDING LOGGING AT Wed Sep 14 02:59:57 2011