**** BEGIN LOGGING AT Tue Feb 13 03:00:02 2018 Feb 13 03:01:30 spierepf: firstly you need to check if its enabled in your kernel .config if it is then its is being generated as a separate ipk already and you need to add that ipk to your MACHINE_EXTRA_RRECOMMENDS in your local.conf Feb 13 03:02:06 kernel-module-foo or something like that Feb 13 03:02:27 I think you need to enable NETFILTER_XT_MATCH_ADDRTYPE = "m" Feb 13 03:02:44 may be via a netfliter.cfg or somesuch Feb 13 03:03:01 :) Feb 13 03:03:41 and IIRC we had a session in YPDD om this in past Feb 13 03:03:43 https://www.yoctoproject.org/sites/yoctoproject.org/files/elc2013-kernel-lab.pdf Feb 13 03:04:05 it might be dated by now though Feb 13 03:04:12 * moto-timo intends to redo the materials for that at some point Feb 13 03:04:24 it was updated by tom zanussi not too long ago Feb 13 03:05:10 We should just create living documents with branches in git Feb 13 03:05:15 ah, it was krogoth Feb 13 03:05:16 https://www.yoctoproject.org/training/kernel-lab Feb 13 03:05:25 ah ok Feb 13 03:05:27 +1 khem Feb 13 03:05:55 we do have living documents, but maybe we could make it more agile Feb 13 03:16:36 khem: I've definitely enabled it in my kernel recipe. I've got an xt_addrtype.cfg file that is being included by a SRC_URI and everything. Feb 13 03:17:06 khem: the .ko is getting built. Feb 13 03:17:21 ok then whats remaining is adding it to image Feb 13 03:17:27 khem: just not included in the rootfs image, or tarball. Feb 13 03:17:32 Indeed. Feb 13 03:18:10 That is the bit I am missing. Feb 13 03:18:14 so you add it IMAGE_INSTALL via a bbappend or Feb 13 03:18:25 which distro are you using and machine ? Feb 13 03:19:19 I am using yocto/poky/pyro, and the adzs-sc589-ezlite machine layer I created myself. Feb 13 03:22:55 I added an IMAGE_INSTALL += "xt_addrtype.ko" entry to my linux-yocto_%.bbappend file. but it had no effect. Feb 13 03:26:12 not the bare filename Feb 13 03:26:13 I tried changing it to IMAGE_INSTALL += "not-a-thing.ko" and it didn't barf, so I don't know what to do with that. Feb 13 03:26:23 you need to add ipk/package name Feb 13 03:26:28 where this goes into Feb 13 03:26:48 How can I determing the ipk/package name? Feb 13 03:27:11 yeah IMAGE_INSTALL is an image thing so a package recipe like kernel wont entertain it at all Feb 13 03:27:19 thats why you dont see it Feb 13 03:27:36 may be you can first find out to which .ipk is it going to Feb 13 03:28:39 Is there a document somewhere that identifies which kernel modules are assigned to which .ipks? Feb 13 03:30:08 they follow a pattern: kernel-module- Feb 13 03:31:04 so like kernel-module-xt_addrtype? Feb 13 03:31:53 IIRC, it is a dash instead of _ Feb 13 03:32:01 so kernel-module-xt-addrtype Feb 13 03:32:29 classes/kernel-module-split.bbclass takes care of it, I was checking the code, but the regex’s aren’t easy to read :D Feb 13 03:34:16 Cool, at least is barfed on kernel-module-xt_addrtype. Feb 13 03:35:02 Nope. "No package kernel-module-xt-addrtype available." Feb 13 03:35:07 head into /tmp/work//linux-yocto//pkgdata/runtime Feb 13 03:36:03 or better: /tmp/work//linux-yocto//deploy-ipks// Feb 13 03:36:09 or deploy-rpms if you are using rpms Feb 13 03:36:19 and you’ll see all the modules that were built and packaged. Feb 13 03:36:27 Yes. I have kernel-module-xt-addrtype-...-yocto-standard Feb 13 03:36:30 and .packaged Feb 13 03:38:07 there’s been a bunch of patches that have been merged recently for alternate kernel names, suffixes, and prefixes .. I wonder if they busted something. what ifyou use that entire name as the install name ? Feb 13 03:38:26 if you look at: recipes-core/packagegroups/packagegroup-base.bb in oe-core Feb 13 03:38:40 you’ll see it should just be: kernel-module- for the install Feb 13 03:39:15 but those are just RRECOMMENDS, so if something broke, I don’t think it would throw an error. so something may be busted. Feb 13 03:39:15 with or without the .packaged? Feb 13 03:39:18 without Feb 13 03:40:20 are you using rpm or ipk ? you can always peek at the packaged module to confirm that it RPROVIDES what the install might be looking for. Feb 13 03:40:38 * zeddii_home is not a packaging guy though, so I get out of my depth :D Feb 13 03:42:05 I don't suppose there is a flag for "Install all the kernel modules you are directed to build"? Feb 13 03:42:14 yup. “kernel-modules" Feb 13 03:42:29 AWESOME Feb 13 03:42:40 that’s basically a package that has nothing in it, but has a RDEPENDS on every individual module package Feb 13 03:42:45 so when you install it, you get them all. Feb 13 03:43:01 Perfect. Let me try that. Feb 13 03:43:02 * zeddii_home almost always uses it .. since I’m working with more than really small systems. Feb 13 03:47:32 Thanks so much zeddi_home! Feb 13 04:11:16 zeddi_home: and khem: if there is any way I can buy you guys some beers, just let me know. Feb 13 04:20:49 heh zeddii_home is taking you into the crawl space to show you all plumbing Feb 13 04:21:16 but we have some porcelain stuff too Feb 13 04:21:20 see https://gist.github.com/d699532ee252c004d476ba7ef69aab68 Feb 13 04:21:32 oe-pkgdata-util is quite handy tool Feb 13 04:21:44 I wish more people knew about it Feb 13 05:49:03 Hi all, is there a list if open issues I can go though (I am interested in contributing) Feb 13 05:49:12 Hi all, is there a list of open issues I can go though (I am interested in contributing) Feb 13 06:28:58 khem: There are a list and **short** explanation of all oe prodivded tools? I'm using yocto/oe with some timeouts for more than 7 years. But I don't know this tool. ^^ Feb 13 08:57:31 New news from stackoverflow: Bitbake: "The metadata is not deterministic and this needs to be fixed" Feb 13 09:03:08 hi all, Doubt is regarding wic image creation. In source plugins (like rootfs.py), class pluginbase.py provides ways to prepare the partition using --source option. There are also ways to do some operations prior to creating the partition (like stage/configure). Feb 13 09:04:06 But I would like to do some specific operations after the partition is created. Say for example, I want to ass signature to the end of my rootfs partition. Feb 13 09:05:08 Are there any ways to define some works after "do_prepare_partition" is called? How can I add signature to each partition after it has been created? Feb 13 09:11:39 was is the reason to put systemd files in /lib/systemd instead of /usr/lib/systemd ? Feb 13 09:21:53 s/was/what Feb 13 12:43:31 http://lists.openembedded.org/pipermail/openembedded-core/2014-December/099483.html Feb 13 12:43:35 guess that is why Feb 13 13:18:36 RP: hey, any plan to bump the -rt patch to latest in linux-yocto-4.9 ? Feb 13 13:19:17 RP: it's currently on -rt18 while the latest is -rt61 and with -rt18 I'm triggering a BUG() in scheduler Feb 13 13:19:23 RP: with -rt61 the bug isn't there Feb 13 13:28:21 New news from stackoverflow: kernel fragment not applied while other fragments are applied Feb 13 13:39:29 hello everyone! Could anyone help me with tricking udev into thinking about my touchscreen as a keyboard as well? I need to trigger specific part of wayland/weston that enables focus, and it happens only with a keyboard attached :-) And I can't really have one. Saw adding ID_INPUT_KEYBOARD=1 as a workaround for such issues, but really not sure where it is an appropriate place to put that given we have rules chain. Please :-D Feb 13 13:40:56 Is there some rfkill unblock systemd service in yocto? Feb 13 13:50:36 marex: you need input from zeddii/zeddii_home Feb 13 13:56:36 architekt, just use connman for that; enable wifi :-) Feb 13 14:02:54 RP: oh, that's Bruce's nick, thanks Feb 13 14:04:05 * zeddii reads. Feb 13 14:04:27 yah. I have updates to the various -rt integrations under test now. Feb 13 14:04:37 4.14 and 4.9 mainly, but some of the older ones as well. Feb 13 14:05:10 I had to survive all of the -stable (CVE fun), 4.14 and 4.15 before I had enough time to loop back around and refresh -rt. I'll follow things closely from this point on. Feb 13 14:05:39 If all goes well, should be something in a day or so, I have to run some boot + latency tests. Feb 13 14:07:07 zeddii: super ! Feb 13 14:07:19 zeddii: Im in no haste, for now I have a patch locally anyway Feb 13 14:08:16 zeddii: while I have you here, are you aware of the problem with kernel-devsrc? Feb 13 14:08:26 cool. the ping was well timed, since I was at it. Feb 13 14:08:35 RP: about the size blowing out some of the images. yah. Feb 13 14:08:47 I was looking into stripping it down to just enough to build modules. Feb 13 14:08:52 zeddii: right, trying to figure out what to do about it... Feb 13 14:08:57 zeddii: btw linux-yocto-4.14 repo is not out yet ? Feb 13 14:09:14 zeddii: that would help a lot if we could figure out how to do it. I was toying with the idea of binning the drivers/ directory Feb 13 14:09:37 zeddii: We have some confusion about what that code is actually meant to do, its uses are ill defined :/ Feb 13 14:10:10 zeddii: Juro was suggesting we remove the used arch/* != $ARCH directories and the generated files, I was nervous about the latter Feb 13 14:10:20 Marex. It is there, just in a different format this time. I was getting complaints over time about all of the kernel downloads, so I put 4.14 and 4.15 into a single repo and made them available. v4.14/* and v4.15/* along with an old v4.8/* that I did about two years ago. All in a linux-yocto.git repo. Feb 13 14:10:36 RP: yah. let me poke at it a bit. There's a forumla for stripping it down into just what you need. Feb 13 14:10:54 I used to do just that with the ancient LDAT build system, I just have to look and see how it applies to the modern kernel now. Feb 13 14:11:02 zeddii: ok, its blocking the other kernel changes getting in right now Feb 13 14:11:24 zeddii: I say that not as pressure, just a statement on where we are! Feb 13 14:11:40 you really want two packages. one that is "all the source" and one that is "enough to build modules", but as you know, the entire kernel packaging is chaos and a tangled mess. so I'll be less ambitious and just get it working. Feb 13 14:12:15 RP: no worries. I'm letting -rt kernels build and run in the background today, so I can poke at it. Feb 13 14:12:35 zeddii: on the plus side, kernel-devsrc is now a separate recipe so we can more easily tweak it (or create two of them for different purposes if we really had to) Feb 13 14:12:56 zeddii: I'd be tempted to strip back the current set of files, see who screams ;-) Feb 13 14:13:20 that was my plan as well. go for 'just enough' and let other use cases lazy initialize :D Feb 13 14:13:44 zeddii: good news is creating a new recipe like kernel-devsrc should be easy for people too Feb 13 14:26:17 zeddii: oh Feb 13 15:53:48 RP: what sizes are you seeing for the kernel-devsrc package at the moment ? mine come in around 109M for rpm. I just want to make sure I'm starting from the same baseline when I start tearing it apart. Feb 13 15:55:50 zeddii: compressed rpm is around 110MB, yes. Uncompressed is obviously larger and where the problem comes in Feb 13 15:56:18 ok. cool. so my baseline is comparable. digging in now. Feb 13 15:57:12 zeddii: drivers alone is 500MB uncompressed Feb 13 15:58:58 ./amd/include/asic_reg is 89M !! Feb 13 15:59:09 * zeddii starts running the right clean targers Feb 13 15:59:12 targets even Feb 13 16:11:03 shame we don't know anyone who likes deleting ancient crap drivers.... Feb 13 17:00:16 Does anyone know if it's possible to create a disto based on fslc-base (freescale i.mx) and poky-tiny? Feb 13 17:22:32 seppe: the concept of distro is totally NOT linked to any machine so i guess yes. Feb 13 17:22:54 find/write the meta implementing the BSP for your board Feb 13 17:23:21 and simply set MACHINE= in build/conf/local.conf Feb 13 17:23:42 that's true unless the distro creator did bad things, yes :) Feb 13 17:24:14 :) Feb 13 17:24:46 seppe: i would add that you mustn't use fslc-base but https://github.com/Freescale/meta-freescale which is the meta that have the BSP Feb 13 18:29:01 Hi Feb 13 18:29:32 I am in need of some assistance regarding an issue with user permissions Feb 13 18:29:44 is this the right place to ask Feb 13 18:29:54 ? Feb 13 18:29:56 :-) Feb 13 18:30:33 I am using the meta-skeleton meta layer Feb 13 18:30:56 in this I am using the useradd-example.bb recipe Feb 13 18:31:19 I can add users but cannot change ownership of files Feb 13 18:32:21 the created files are with correct ownership at the do_install() time but revert to root:root at run time on the target Feb 13 18:32:37 has anyone encountered this issue ? Feb 13 18:55:14 I keep approaching success asymptotically. :D Feb 13 18:55:30 Does anyone recognize this error: https://pastebin.com/cPmJ6ckJ Feb 13 18:55:48 I believe that it has to do with a missing kernel config. Feb 14 01:00:32 New news from stackoverflow: Add packages to yocto host sdk **** ENDING LOGGING AT Wed Feb 14 03:00:01 2018