**** BEGIN LOGGING AT Mon Oct 30 03:00:02 2017 Oct 30 05:53:01 New news from stackoverflow: Trying to use Linux menuconfig with yocto/morty Oct 30 08:24:13 good morning Oct 30 08:25:49 good morning Oct 30 11:10:27 hey guys! could anyone give a hint on how to instuct Yocto to create SSH host keys during image generation? Oct 30 11:13:02 something like ROOTFS_POSTPROCESS_COMMAND += ssh-keygen ? Oct 30 11:20:04 luneff: I think host keys are created at first boot Oct 30 11:20:29 kanavin, yeah, but if you go read-only root, they are volatile and regenerated every boot Oct 30 11:21:58 which is not that great Oct 30 11:22:12 found this: http://book.yoctoprojectbook.com/page/code-chapter-7 , ROOTFS_POSTPROCESS_COMMAND += "configure_sshd;" Oct 30 11:23:33 luneff: read meta/classes/rootfs-postcommands.bbclass, I think it has the right answer :) Oct 30 11:24:17 yeah, been there. That's where I've found a way for SSH to accept pregenerated keys :-) testing already Oct 30 11:32:51 luneff: the yoctobook snippet should work too, just append it to your image recipe Oct 30 11:35:48 oh. in my case, read_only_rootfs_hook seemed to trigger before my script was done. so I get wrong /etc/default/ssh contents :-( Oct 30 11:38:06 what can I do instead of += there? Oct 30 11:38:37 oh. =+ it seems Oct 30 11:49:22 how do you guys deal with per-device specific configuration with yocto? Oct 30 11:50:22 I mean when there's like 20+ devices running the same firmware and you need to pack some keys with each one Oct 30 12:03:14 luneff: the subject of configuration management is a regular one :) Oct 30 12:03:30 luneff: there was a thread on one of the ml lately, but I can't find it quickly Oct 30 12:06:43 ;-( Oct 30 12:09:44 luneff: I suggest you ping fray, he knows all about it Oct 30 12:09:52 (american timezone though) Oct 30 12:18:07 kanavin, thanks :-) Oct 30 14:29:40 Is bitbake -e doing the same thing as bitbake --parse-only with respect to the internal parsing state? Oct 30 14:51:10 Anyone using overlayfs with yocto ? Oct 30 14:53:02 nayfe, yes, I used this class: https://patchwork.openembedded.org/patch/138100/ Oct 30 14:55:27 micka> I use that class too (very usefull), but it's not the same as overlayfs filesystem, to mount for example "/" in read-only, and add a rw layer on top of it in another partition Oct 30 14:57:44 It's possible to mount partition in rw (fstab), but I'm not sure is that you want. sorry Oct 30 15:00:35 micka: np :) in fact, i'm trying to understand existing read-only/volatile initscripts to integrate that kind of read-only with persistance system without a lots of changes Oct 30 15:01:43 that topic https://lists.yoctoproject.org/pipermail/yocto/2015-July/025704.html talks about it, and point to volatile-bind recipe, but it's for systemd only Oct 30 15:02:58 there is also https://github.com/rdm-dev/meta-jens/tree/jethro-auesee example Oct 30 15:11:09 apparently there is difference between bitbake -e and bitbake --parse-only, as the latter removes a taskhash mismatch in the next bitbake run, while the former does not Oct 30 15:16:22 Our system is haunted with an occational taskhash mismatch error, but at the same time too elusive to pinpoint. And since bitbake is very sparse on hinting to the location for it, we've learned that running bitbake --parse-only a single time before the real production build takes all taskhash errors away. Oct 30 15:44:54 Can anybody recommend an easy way to replace the SRC_URI for a handful of recipes that are pointing at a server that my company no longer has access to? Oct 30 15:44:57 I tried this: Oct 30 15:44:59 PREMIRRORS_append = " \ git://inaccessible.server.com/path/to/teh-codz/.* git://our.shiny.forks.com/repos/BASENAME;protocol=ssh \n \ " Oct 30 15:45:06 in local.conf Oct 30 15:47:51 When I try putting this in local.conf: Oct 30 15:47:53 PREMIRRORS_append = " \ git://inaccessible.server.com/path/to/teh-codz/.* git://our.shiny.forks.com/repos/BASENAME;protocol=ssh \n \ " Oct 30 15:48:08 I'm still getting this error: Oct 30 15:48:09 ERROR: ExpansionError during parsing /jenkins/workspace/bsp/build/../layers/meta-virtualization/recipes-extended/xen/xen_git.bb: Failure expanding variable SRCPV, expression was ${@bb.fetch2.get_srcrev(d)} which triggered exception FetchError: Fetcher failure: Fetch command failed with exit code 128, output: fatal: https://inaccessible.server.com/path-to-teh-codz/xen.git/info/refs not valid: is this a git repository? Oct 30 15:48:54 I'm wondering: does SRCPV bypass the (pre)mirrors mechanism? Or... do I just have the syntax wrong? Oct 30 15:49:24 evadeflow: you can replace it in all reciplys using sed Oct 30 15:49:44 *recipes Oct 30 15:50:04 @hmwel, you mean, modify the original recipes? Oct 30 15:50:08 yes Oct 30 15:50:18 Oh. Oct 30 15:50:37 Well... they're from a vendor, and they're still sending us updates. Oct 30 15:50:52 They've just made it impossible to access their servers from behind our corporate firewall, so... Oct 30 15:51:14 I've been manually sync'ing once in awhile to a set of repos I *do* have access to. Oct 30 15:51:28 a oke other option is to ad de server to the host list of the pc Oct 30 15:51:34 build pc Oct 30 15:51:51 Ooh---sneaky! Oct 30 15:52:01 I might have to do that. Oct 30 15:52:12 Thanks for the suggestion. Oct 30 15:52:18 NP Oct 30 15:52:59 Still curious why my PREMIRRORS isn't working for this, though. This thread is where I got the idea from: https://patchwork.openembedded.org/patch/123771/ Oct 30 15:53:07 I mean, it's 'supposed' to work, isn't it? Oct 30 15:54:01 premirrora append should work Oct 30 15:54:34 Any advice on how to debug why it's failing? Oct 30 15:56:37 evadeflow: sounds like they recipes are using AUTOREV maybe? if so, as suggested in that patch, use BB_FETCH_PREMIRRORONLY Oct 30 15:58:10 Yeah, @nrossi, I think they're using AUTOREV. I'll try BB_FETCH_PREMIRRORONLY. So... recipes that use AUTOREV don't run `git ls-remote` (or whatever) through the mirrors, then? Oct 30 15:59:05 evadeflow: yer, AUTOREV recipes do some git ls-remote commands before fetch. BB_NO_NETWORK can also be used to disable those remote checks Oct 30 15:59:37 Perfect, thanks for pointing me in the right direction! Oct 30 16:54:44 New news from stackoverflow: Yocto imx6 runlevel Oct 30 17:29:19 BBMASK = "imx-gst" is set in my local.conf but bitbake-layers show-recipes still has imx-gst1.0-plugin shown... tried the full recipe name and it also doesn't mask it out. No matter what I try the recipe shows up in bitbake-layers show-recipes Oct 30 17:38:46 also if I do a cleanall on an image recipe, will it also clean all the recipes that make up the image, or just things like rootfs, sysroots etc? Oct 30 18:33:35 Something just caught my eye in a recipe (not mine) that makes me want to ask a question. Given this: Oct 30 18:33:37 SRC_URI = "git://code.acme.com/gitlab/kernel.git;protocol=https;branch=${BRANCH}" SRC_URI_remove = " \ file://bsp3.3.1.rc3_linux.patch \ file://bsp3.3.1.rc4_linux.patch \ " Oct 30 18:33:54 (Geesh, formatting fail, I'll pastebin that if needed) Oct 30 18:34:49 Anwyay, the gist of my question is: does setting SRC_URI directly (as opposed to using SRC_URI_append or SRC_URI_remove) replace *all* SRC_URI elements? Oct 30 18:37:51 Here's a better-formatted example: https://gist.github.com/evadeflow/8581c9401753d932d175010b4072cb3a Oct 30 18:38:48 I've been assuming that saying something like, say, SRC_URI = "foo", would replace *all* of SRC_URI. But that recipe seems to suggest that my assumption is incorrect. Oct 30 18:39:13 (Otherwise, why bother with those SRC_URI_remove lines?) Oct 30 20:40:03 Hello Oct 30 20:41:19 I'm creating a recipe and in do_configure ti grives me an error "configure: error: ps not found" Oct 30 20:42:09 what provides "ps"? Oct 30 20:42:24 ghostscript, probably Oct 30 20:42:41 oh thanks checking Oct 30 21:12:32 TurBoss[m]: so we limit the commands you can run from the host, I assume ps isn't on the list Oct 30 21:12:47 really though why is the configure step needing to examine the running processes? **** ENDING LOGGING AT Tue Oct 31 03:00:02 2017