**** BEGIN LOGGING AT Mon Dec 30 02:59:58 2013 Dec 30 09:47:27 morning all, on networkmanager_0.9.8.bb, I want to propose a modification : --without-systemdsystemunitdir= -> --without-systemdsystemunitdir, this file is available on meta-openembedded/meta-oe/recipes-connectmanager_0.9.8.0.bb, where should I propose this patch? Dec 30 11:33:51 good morning. Can anybody please kindly point me at any documentation on how to add a binary package for 2 different architectures? Specifically I want to create a java package for armv5 and armv7 **** BEGIN LOGGING AT Mon Dec 30 12:01:50 2013 Dec 30 12:20:15 hi i have been using poky with meta-atmel,i have been tying to use gprs,it is working fine,i have used /etc/ppp script,but it is very slow when download a small file using wget,at the same time i used the same device with buildroot compiled files,it downloaded the same file without stalled,the csq value is 22 and 23,can you help me Dec 30 14:37:03 hi. Can anybody please kindly point me at any documentation on how to add a binary package for 2 different architectures? Specifically I want to create a java package for armv5 and armv7 Dec 30 14:40:26 I think that you need to be more specific about what it is that you are trying to accomplish? Dec 30 14:41:08 T0mW: you are talking to me, right? Dec 30 14:41:15 yes Dec 30 14:41:25 panda84kde: yes, sorry Dec 30 14:42:59 ok. I want to package Oracle Java for ARM in my own layer. I want to create 2 recipes (which I already have) one for Java for ARM5 and one for ARM7. Dec 30 14:43:36 panda84kde: are these recipes each for a different "machine"? Dec 30 14:44:04 panda84kde: for example arm7 would be "foo_machine" and arm5 for "other_machine"? Dec 30 14:45:15 T0mW: sort of. I'm targeting a specific machine on arm7 (Nitrogen6x), but there's nothing against running the same package on another arm7 chip Dec 30 14:47:07 panda84kde: ah, I see, you want to base it upon the ARCH instead of MACHINE. Something like SRC_URI_append_arm Dec 30 14:47:50 COMPATIBLE_HOST may be of interest, either directly or as inspiration, by examining how it's implemented Dec 30 14:49:02 T0mW: kergoth: thanks for the indications Dec 30 14:50:40 panda84kde: if you make your recipes for armv7 / armv5 specific to a machine, there is no reason why someone else couldn't extend your recipe to another machine useing a bbappend + COMPATABLE_MACHINE statement(s). Dec 30 15:03:55 qtwebkit with eglfs, anyone sucessfull? Dec 30 17:00:48 * kergoth tests a tiny little script which checks for signature differences when including a bsp layer but not using a machine it provides, and the same for a distro layer and not using that distro, to check for leaky non-compliant layers that affect things they shouldn't (that is, not using overrides) Dec 30 18:06:04 that would be handy for me kergoth Dec 30 19:49:23 Anyone aware of a recipe for mii-tool Dec 30 19:51:19 or even ethtool Dec 30 19:57:42 found it.. whee .. Dec 30 19:57:44 * WarheadsSE is blind Dec 30 20:08:39 This has me really stumped. I've a custom recipe which creates a directory in the target image and attempts to chown to www-data. The directory is created but is still owned by root. This is in my do_install(): install -m 775 -o www-data -g www-data -d ${D}/binaries Dec 30 20:09:33 I've even tried 'install -m 775 -d ${D}/binaries; chown 33:33 ${D}/binaries' with the same result, root ownership. Dec 30 20:22:05 T0mW: are you looking at the files from the host or on the target? Also did you set your recipe to have base-passwd as a DEPENDS since it will provide the passwd file in the sysroot with the correct uid? Dec 30 20:23:25 sgw_: mounting the ext3 image and looking at that. No, I thought that DEPENDS meant "build the DEPENDs first"? Dec 30 20:24:15 T0mW: is does, base-passwd needs to be available (ie built) in the target sysroot Dec 30 20:24:27 sgw_: does DEPENDS mean to include the listed package, and files, into build environment for referencing? Dec 30 20:25:09 sgw_: the base password is there in the image, and the group + user is in /etc/passwd + /etc/group files. Dec 30 20:26:03 yes, the build system creates a build sysroot (see tmp/sysroots/) which has headers and libraries and some files like passwd Dec 30 20:26:42 edited my DEPENDS and only placed 'base-passwd' in it. Dec 30 20:27:04 so does it still need a postinst to add user/group, or does it happen when the rootfs is built? Dec 30 20:27:05 building rootfs, again, :-/ Dec 30 20:28:01 no, didn't try a postinstall, but was thinking of that. That something in the logic was ignoring the -o and -g install options, and the chown. Dec 30 20:28:44 nerdboy: the group + user is www-data, which is in base-passwd. Dec 30 20:28:48 nerdboy: no, it should not since www-data is part of the base-passwd, if you are adding a new user, you should use the useradd.bbclass and USERADD* Dec 30 20:29:17 yeah, useradd is pretty slick. Dec 30 20:29:18 and no more postinst stuff? Dec 30 20:29:44 * nerdboy needs to migrate off of classic... Dec 30 20:30:42 nerdboy: nope, it does it for you automagically Dec 30 20:31:23 uh, I have pkg_postinst_${PN} in a package I migrated from Bernard and the postinst is still working. Dec 30 20:31:55 hmm, maybe that is a consideration too. Dec 30 20:32:05 i just meant "no longer requires useradd/groupadd in postinst" Dec 30 20:32:20 oh Dec 30 20:33:21 heh, well, I'm jumping about 5 revisions of Yocto (OE) and find a lot has changed (improved as well). I thought that postinst had been deprecated. Dec 30 20:33:57 * T0mW still waiting on do_rootfs Dec 30 20:34:49 nerdboy: useradd does the magic as a preinst, but it can happen during rootfs creation time, and does not have to happen on the target Dec 30 20:34:55 nope, no good, still owned by root. Dec 30 20:35:00 T0mW: postinst is not deprecated Dec 30 20:35:07 k Dec 30 20:35:29 T0mW: I will check something, brb Dec 30 20:36:12 do_install says: Dec 30 20:36:17 install -d -m 775 ${D}/configs Dec 30 20:36:39 chown -h 33:33 ${D}/configs Dec 30 20:36:48 and still no change Dec 30 20:36:56 gotit Dec 30 20:38:00 weird thing is that I extended lighttpd (bbappend) and can chown /www from do_install_append. Dec 30 20:38:31 I'll pull Dora and make sure I'm up to date. Dec 30 20:38:32 T0mW: so you are seeing it work someplace? Dec 30 20:38:41 yes Dec 30 20:38:56 sgw_: yes, in my lighttpd.bbappend Dec 30 20:39:14 But not in a standalone recipe on your layer? Dec 30 20:39:15 sgw_: same owener.group Dec 30 20:39:19 correct Dec 30 20:39:43 T0mW: ok, now you got me, I was going to do a local check Dec 30 20:39:51 sgw_: it is a simple recipe too. Dec 30 20:40:54 Dora is up-to-date Dec 30 20:41:50 sgw_: I've even blown everything away (cache, sstate-cache, tmp, etc...) and rebuild the entire project. same result. Dec 30 20:43:43 Wonder if this is a 'race' condition as I have -j 8 on make and bitbake Dec 30 20:44:43 T0mW: willing to share your recipe for a quick examination? (you can pm me and/or pastebin) Dec 30 20:46:21 sgw_: http://pastebin.com/J4xCPFTt Dec 30 20:46:41 sgw_: simple recipe, that is what i'm building now Dec 30 20:48:41 brb, need coffee Dec 30 20:53:26 sgw_: still no good, that recipe in pastebin is the one that just failed. Dec 30 20:56:34 T0mW: I am doing a local build, unfortunatly I have other changes that caused more to rebuild, it would have had to happen eventually! Dec 30 20:56:52 lol Dec 30 20:57:16 sgw_: maybe you should go for coffee too? Dec 30 20:57:45 as fast as Yocto (OE) has become, it still gives you time for coffee. Dec 30 21:00:17 sgw_, can you fix the foo in the latest db commit message? Dec 30 21:00:32 otherwise I can resend Dec 30 21:02:50 Crofton|work: I can tweak that for you! Dec 30 21:05:16 T0mW: I have to head out for an hour or so, I will look at this again when I get back, sorry to delay this. Dec 30 21:05:44 sgw_: thanks, I've more to keep me busy.. Dec 30 21:12:59 thanks, sorry about missing the sdk booboo Dec 30 21:27:30 what's with the foo thing? Dec 30 21:28:17 and next time i think i'll clean up what i send you first, since it might just go "public" ;) Dec 30 21:28:57 yeah, no not-nice language in the commits Dec 30 21:29:16 i meant clean up the recipe hacks... Dec 30 21:29:31 that too Dec 30 21:29:35 i would never commit bad language... or would i? Dec 30 21:29:58 things that make you go "hmm..." Dec 30 21:30:28 heh, my boss and I collaborate on a project, I try to be careful in my commit messages about his foul^H^H^H^H mistakes Dec 30 21:32:15 this is really strange, the pkg_postinst script didn't change the ownership either. Dec 30 21:59:52 Crofton|work: are you paying attention? you're taking all the fun out of harassing you... Dec 30 22:06:48 heh Dec 30 22:06:49 afk Dec 30 22:07:00 foo is what I put in commits I will turn around and rebase Dec 30 22:56:53 sgw_: around? Dec 30 22:57:26 I just checked AB and fsl-arm is still failing; but the failure does not seem to be BSP specific ... am I right? Dec 31 01:46:02 T0mW: any chance your still around? I think you found a genuine bug! I looked in my log file for rootfs and found that the user was not being found by the smart package manager, well it turns out that it tries to read the rootfs/etc/passwd, which does not exist yet and then falls back to /etc/passwd on the host, instead of the sysroot//etc/passwd! I will file a bug **** ENDING LOGGING AT Tue Dec 31 02:59:58 2013