**** BEGIN LOGGING AT Wed Oct 25 03:00:03 2017
Oct 25 07:01:30 <Klopilot> Hello
Oct 25 07:13:28 <Klopilot> For yocto 2.2, in etc/rcS setting ROOTFS_READ_ONLY to yes is the only thing to do for setting the image to read-only? And the IMAGE_FEATURES read-only-rootfs is for testing the recipes for post-installation correctness and is only setting ROOTFS_READ_ONLY to yes?
Oct 25 07:18:32 <styler2go_> Good day everyone! Can i somehow delete all image files except for the "final" one? I usually have 4 files: iso, hddimg, ext4 and wic. All have approx 2gb. Can i somehow keep only the "final" wic file?
Oct 25 07:23:51 <nayfe> styler2go_: hey maybe a do_deploy_append to remove those files ?
Oct 25 07:50:20 <Klopilot> For yocto 2.2, in etc/rcS setting ROOTFS_READ_ONLY to yes is the only thing to do for setting the image to read-only? And the IMAGE_FEATURES read-only-rootfs is for testing the recipes for post-installation correctness and is only setting ROOTFS_READ_ONLY to yes?
Oct 25 08:28:18 <Klopilot> the bash in my yocto 2.2 image does not know source! Do i have to enable it in a recipe?
Oct 25 08:29:34 <cornel> hello
Oct 25 08:29:50 <Klopilot> hello again cornel!
Oct 25 08:29:55 <Klopilot> For yocto 2.2, in etc/rcS setting ROOTFS_READ_ONLY to yes is the only thing to do for setting the image to read-only? And the IMAGE_FEATURES read-only-rootfs is for testing the recipes for post-installation correctness and is only setting ROOTFS_READ_ONLY to yes?
Oct 25 08:30:19 <cornel> is there a table somewhere documenting what variables are local to a package and what are global (like in require _${PN} to make them local)?
Oct 25 08:32:14 <cornel> styler2go, try to not generate all
Oct 25 08:32:28 <styler2go> cornel, wic needs them
Oct 25 08:32:34 <cornel> ah
Oct 25 08:32:35 <cornel> sorry
Oct 25 08:33:08 <cornel> then nayfe is probably right, you need to remove them at the end
Oct 25 08:33:28 <cornel> Klopilot, i don't know about rc.S , sorry
Oct 25 08:33:50 <Klopilot> thx
Oct 25 08:34:24 <cornel> styler2go, maybe you can write a special cleaning function to remove those
Oct 25 08:35:27 <nayfe> Klopilot: what do you want to do ?
Oct 25 08:35:51 <ChrysD> nrossi : Hi, do you remember the problem of strace yesterday ? You will laugh at me when you will know why it wasn't working...
Oct 25 08:36:46 <Klopilot> i am porting a system to yocto. And now i would like to set read-only-rootfs via kernel cmdline
Oct 25 08:37:44 <nayfe> what system do you use ?
Oct 25 08:37:53 <nayfe> which machine.conf i mean
Oct 25 08:37:56 <ChrysD> nrossi : i've plugged an USB into my computer, and he takes the "X:\ " which i usually use for my sdcard... as i rarely plug the usb, i forgot that i put it there and didn't notice he took the X:\ of my sdcard... Sorry for losting your time.
Oct 25 08:38:54 <nayfe> Klopilot: for example, in my case, i have UBOOT_EXTLINUX_KERNEL_ARGS += "rootfstype=ext4 video=...."
Oct 25 08:39:26 <Klopilot> system? this is a arm system, include conf/machine/include/tune-arm926ejs.inc
Oct 25 08:40:34 <Klopilot> the kernel cmdline is not the problem! : rootfstype=ubifs ro ...
Oct 25 08:41:24 <Klopilot>  ro // rw should be scanned for read-only-filesystem or not
Oct 25 08:42:42 <Klopilot> I think i can change ROOTFS_READ_ONLY to yes and then i locked the file system forever, but it is much more convinient setting this via kernel cmdline
Oct 25 08:43:17 <Klopilot> ...much more convenient
Oct 25 08:51:09 <fberg> Hello guys | I have a question. I have developed my yocto enviroment that generates a .sdimg file with two partitions: boot+rootfs. Now I want to have an img partitioned as (bootloader + (rootfs1) +(rootfs2)+(data)). This is to allow OTA operations and/or failure recovery an the golden image on, say, rootfs1. Does anybody know how to do that and/or have any useful link ?
Oct 25 08:51:12 <fberg> TY !
Oct 25 08:53:26 <nrossi> ChrysD: ahh the old auto drive letter assignment confusion... :)
Oct 25 08:53:41 <ChrysD> nrossi : :(
Oct 25 09:10:04 <Klopilot> ls
Oct 25 10:29:33 <osse> using meta-virtualization I get a checksum error for protobuf-3.1.0. can I leave that package out of the image?
Oct 25 10:35:43 <aurele> hi everyone
Oct 25 10:37:49 <aurele> I'm using gsoap, and while using gsoap you need to call some code generators, today code generators are sent to my target and nothing in the sdk native part. I tried to set the generators in the -dev package, but I still have nothing in the sdk native arch
Oct 25 10:38:22 <aurele> (I hope I was clear)
Oct 25 10:41:38 <aurele> I think my problem is related to "nativesdk" but can't figure out how I can get gsoap in the nativesdk
Oct 25 11:06:13 <DStorey> Hi all, Hopefully have a quick question for you. I need to remove Dropbear SSH from our Freescale i.MX Release Distro 3.14.52-1.1.1 OS and apt-get etc isn't working. Is there another command I can use which will let me perform the removal?
Oct 25 11:11:34 <FenrirJk> DStorey: I'm a Yocto newbie, so excuse me if I'm wrong, but isn't apt-get supposed to manage packages from the host machine? I would edit the image recipe to remove a package on the target distribution.
Oct 25 11:13:58 <DStorey> Unfortunately, it's not quite so simple. The equipment isn't owned by us, we are running a dev version of Yocto with SSH enabled using Dropbear. We need to shut this off on the devices, so were hoping there was a simple way of doing it. We can stop the service, but that will just start again when it is rebooted or the PoE switch interface downs and comes back up. The vendor doesn't know how to shut it off
Oct 25 11:17:14 <aurele> DStorey does it use systemd or sysVinit
Oct 25 11:17:21 <FenrirJk> I see, what's the apt-get error? Can't it be disabled with configuration files?
Oct 25 11:18:02 <aurele> DStorey just remove startup script or with systemd use systemctl disable dropbear
Oct 25 11:18:10 <DStorey> How woul dI find otu which it uses?
Oct 25 11:18:12 <DStorey> out**
Oct 25 11:18:23 <zero_note> DStorey: if you're on systemd, stop and mask the socket : systemctl stop dropbear.socket && systemctl mask dropbear.socket
Oct 25 11:19:08 <zero_note> DStorey: simply type 'systemctl' in you console
Oct 25 11:19:34 <DStorey> -sh: systemctl: command not found
Oct 25 11:19:47 <zero_note> DStorey: or check if you have some dirs like /lib/systemd/system or /etc/systemd
Oct 25 11:20:02 <DStorey> systemtap - this is there
Oct 25 11:20:07 <zero_note> DStorey: so you're on init.d
Oct 25 11:20:23 <DStorey> yep I can see that on the list
Oct 25 11:22:06 <aurele> DStorey, cat /proc/1/comm
Oct 25 11:23:07 <DStorey> shows: cat /proc/1/comm init
Oct 25 11:24:35 <zero_note> DStorey: if you have update-rc.d you can try with "update-rc.d -f dropbear remove"
Oct 25 11:25:56 <zero_note> DStorey: if not, delete (or comment the "Start" code of ) /etc/init.d/dropbear
Oct 25 11:26:24 <aurele> zero_note, DStorey just start with find /etc/rc* -name "*dropbear*" -exec rm {} \;
Oct 25 11:26:52 <aurele> and then you can delete or comment the start code /etc/init.d/dropbear
Oct 25 11:27:35 <zero_note> aurele: of course, much better ;)
Oct 25 11:28:12 <DStorey> Awesome, I will give that a whirl. Thanks for your help so far all
Oct 25 11:28:15 <aurele> DStorey a more verbose one : find /etc/rc* -name "*dropbear*" -exec rm -v {} \;
Oct 25 11:29:51 <DStorey> This command doesn't bring anything back: find /etc/rc* -name "*dropbear*" -exec rm -v {} \;
Oct 25 11:30:34 <aurele> DStorey find /lib/systemd/system/ -name  "*dropbear*"
Oct 25 11:32:49 <aurele> Does anyone knows if it would be possible to install package only at compile time but not send it to the target rootfs?
Oct 25 11:35:08 <hamis> I have a licence scanning utility which should be run on each package. I have created a class with a task to perform this scanning and recipe to fetch sources and populate sysroot so that it can be used from there inside the class.
Oct 25 11:35:31 <hamis>  Class made dependent on this recipe. While trying to build I am getting circular dependency error
Oct 25 11:35:43 <hamis> How can I solve such an issue.
Oct 25 11:35:56 <hamis> Is there some other way to avoid this extra recipe which is only required for pulling utility source code
Oct 25 11:42:06 <nayfe> aurele: for me, DEPENDS does not include corresponding package to target?
Oct 25 11:50:05 <aurele> nayfe, you rock!  now I know the difference between DEPENDS and RDEPENDS
Oct 25 11:50:32 <nayfe> aurele: np :)
Oct 25 11:54:20 <nayfe> anyone using swupdate here ? :) got "[extract_files] : Found file: filename u-boot.img size 347328 Not required: skipping" not sure why
Oct 25 11:55:02 <aurele> nayfe, I am
Oct 25 11:56:32 <aurele> nayfe, can you pastebin or something your swdescription?
Oct 25 11:59:54 <aurele> nayfe, as far as I remember there is a "version" file you can put on your filesystem and a "version" string in the sw-description and swupdate can compare them to tell if it is relevant to update
Oct 25 12:00:15 <aurele> I will check if I can find where it is written
Oct 25 12:02:06 <nayfe> its /etc/sw-versions
Oct 25 12:02:31 <aurele> nayfe, https://sbabic.github.io/swupdate/sw-description.html#checking-version-of-installed-software
Oct 25 12:03:50 <nayfe> https://pastebin.com/Mj6q9k8j
Oct 25 12:05:13 <nayfe> what is strange is that i don't have that sw-version file, and the other images (rootfs and SPL) are set to "required"
Oct 25 12:13:45 <aurele> nayfe, this is strange you don't set any version so it should write the booloader
Oct 25 12:16:06 <aurele> nayfe, why your u-boot and SPL are not with type raw?
Oct 25 12:17:23 <nayfe> aurele: i took example from https://sbabic.github.io/swupdate/sw-description.html#images, in previous log, i have raw handle set for it
Oct 25 12:17:25 <nayfe> [parse_images] : Found Image : u-boot.bin in device : /dev/mmcblk0boot0 for handler raw
Oct 25 12:22:39 <aurele> nayfe, I can't tell you why it tells you that it isn't required... my sw-description is more simple but yours seems to be ok to me
Oct 25 12:23:57 <aurele> nayfe, I had some issues with corrupted file when using http download (with "-d" option) and changing the files order in the cpio file changed the behaviour
Oct 25 12:24:36 <aurele> nayfe, maybe to o debug this you could try to make an archive without SPL and see what happend after
Oct 25 12:27:28 <aurele> nayfe, it could be because they share the same device but it would be strange
Oct 25 12:27:32 <nayfe> aurele: i'm trying to regenerate it by hand :)
Oct 25 12:33:23 <nayfe> aurele: hum, now it fails with u-boot and rootfs ... i'll try to update locally
Oct 25 12:34:02 <aurele> nayfe, you try with "-d"?
Oct 25 12:34:40 <nayfe> with -w
Oct 25 12:53:55 <nayfe> grrrrr, [cpio_scan] : Found file: filename u-boot.img not required // SWUPDATE failed [0] ERROR core/swupdate.c : check_provided : 170 : Requested file not found in image: u-boot.bin
Oct 25 12:54:01 <osse> I updated BBMASK but it tries to run a recipe anyway. Is there a cache I need to clear?
Oct 25 12:56:44 <aurele> nayfe, can you list the files in your sw (cpio -t < your_file.swu)
Oct 25 12:57:09 <nayfe> aurele: I decompressed it, u-boot.bin is here
Oct 25 12:57:25 <nayfe> i'm looking into swupdate code now ..
Oct 25 12:59:38 <LetoThe2nd> kanavin: ping concerning bc license
Oct 25 13:01:01 <kanavin> LetoThe2nd: yeah?
Oct 25 13:01:10 <FenrirJk> What is the default behavior of do_install?
Oct 25 13:03:42 <LetoThe2nd> kanavin: the gplv3 & gplv2+ comes from the original license checking. they do verify the sublicenses of bc itself, the lib, dc and the number library. in 1.06 they all were gplv2+ / lgplv2.1 - now bc and the lib are gplv3, and the number lib stuck at gplv2
Oct 25 13:04:20 <aurele> nayfe, I had some corrupted files while 'uncompressing' the cpio, changing the order of files inside the cpio, to achive this I had to shange the name of my files (from u-boot.bin to sw-u-boot.bin)
Oct 25 13:04:33 <LetoThe2nd> stuck at gplv2+, hence this licensing string. so one could shorten that into gplv3+ and not explicitly name number lib license, you mean?
Oct 25 13:05:10 <aurele> nayfe, but anyhow "may the force be with you"
Oct 25 13:05:23 <LetoThe2nd> kanavin: or should i just clarify that in the patch?
Oct 25 13:05:29 <kanavin> LetoThe2nd: yes. if you combine gplv2+ and gplv3 code, the result is gplv3. Do explain this in detail in the commit message, with links to the code tree.
Oct 25 13:06:35 <LetoThe2nd> kanavin: ok, will do. thanks for the review, just felt like taking this question to irc will be faster
Oct 25 13:07:05 <nayfe> aurele: that sounds sooooo secure!
Oct 25 13:07:08 <kanavin> LetoThe2nd: also https://www.gnu.org/software/bc/bc.html clearly says it's v3
Oct 25 13:07:38 <kanavin> LetoThe2nd: so is anything still licensed under lgpl?
Oct 25 13:07:56 <osse> Is there a recommendeded way of versioning yocto configuration files?
Oct 25 13:08:17 <nayfe> aurele: thanks, i'll put some logs everywhere
Oct 25 13:09:35 <LetoThe2nd> kanavin: i see. just stuck to the original license testing. well they explicitly have COPYING.LIB in-tree which states lgpl
Oct 25 13:10:04 <kanavin> LetoThe2nd: that is lgpl3 :)
Oct 25 13:10:18 <kanavin> so please take a look at what the actual source files say in the various directories
Oct 25 13:12:26 <LetoThe2nd> kanavin: yeah the sources i saw state gplv2+ or gplv3+, depending on where you look at. probably they didn't update all notices
Oct 25 13:13:04 <kanavin> LetoThe2nd: yeah, but looks like intent is v3+
Oct 25 13:13:39 <LetoThe2nd> kanavin: agreed an that coincides with the web site
Oct 25 13:14:30 <LetoThe2nd> kanavin: so you suggest to boil it down to gplv3+, as there seems to be no document that states to which te copying.lib license applies?
Oct 25 13:15:01 <kanavin> LetoThe2nd: no, as that would 'contaminate' anything that links to the library, right?
Oct 25 13:15:50 <kanavin> LetoThe2nd: the 'document' is the copyright headers of various source files, it's annoying, but that's the only way to establish what is lgpl in the tree and what is plain gpl
Oct 25 13:17:35 <LetoThe2nd> kanavin: this is where it gets fishy. copying.lib says lgpl, yet the only source file in /lib that has an explicit license statement says gplv2+
Oct 25 13:19:19 <kanavin> LetoThe2nd: /lib/number.c says gplv3+
Oct 25 13:20:21 <LetoThe2nd> kanavin: dang misread again. right. so as i understand it, the library cannot be lgpl anyways.
Oct 25 13:20:54 <kanavin> LetoThe2nd: yeah, looks like there's actually nothing in there that is lgpl
Oct 25 13:21:11 <kanavin> LetoThe2nd: what recipes in oe-core (and beyond) link to this?
Oct 25 13:21:23 <LetoThe2nd> kanavin: not the slightest clue, TBH
Oct 25 13:21:39 <kanavin> LetoThe2nd: so what was your reason to update?
Oct 25 13:22:41 <nayfe> aurele: lol, i was blind, it's u-boot.img not u-boot.bin ...
Oct 25 13:22:54 <LetoThe2nd> kanavin: at OEDEM, there was the call for "newbies" to pick up low hanging fruit, like version bumps. coincidentially, sunday evening the upgradeable recipes list came out.... and stupid as i am, i thought i'd jsut pick bc for no specific reason
Oct 25 13:23:19 <kanavin> LetoThe2nd: right, well I guess we can settle for gplv3+ and move on
Oct 25 13:23:23 * LetoThe2nd now goes to the chalkboard and writes "this did backfire" a 100 times
Oct 25 13:24:08 <kanavin> LetoThe2nd: version bumps attempts can unravel badly, depending on what you pick
Oct 25 13:24:27 <kanavin> LetoThe2nd: want to try python 3.5 -> 3.6? :)
Oct 25 13:24:36 <LetoThe2nd> kanavin: i know. and given the list, i deemed bc the most innocent looking option
Oct 25 13:25:01 <LetoThe2nd> kanavin: anyways, thanks for the input.... i'll rework the patch and description, and resend then. probably in the evening, though
Oct 25 13:25:34 <kanavin> LetoThe2nd: cheers
Oct 25 13:25:59 <LetoThe2nd> kanavin: exactly that!
Oct 25 13:27:12 <kanavin> LetoThe2nd: btw, previous version of bc came out 17 years ago
Oct 25 13:27:27 <kanavin> which means there might be more surprises in store as we throw this on the autobuilder
Oct 25 13:28:53 <kanavin> LetoThe2nd: I recommend you try to find things in oe-core that depend on bc (for example, rename the recipe for a moment and see what breaks in world build), and try to build them with the new version
Oct 25 13:29:06 <kanavin> LetoThe2nd: better yet, just build world, if your machine can take it
Oct 25 13:30:42 <kanavin> the API incompatibility issues *will* be found by the AB, and ross will point you to the build failure log on the oe-core list, so it's better to insure yourself against this :)
Oct 25 13:34:28 <joshuagl> especially as ross is afk for a few days :-)
Oct 25 13:37:42 <kanavin> is he at elc somehow?
Oct 25 13:42:35 <joshuagl> nope, just a coincident that ELC-E falls during half term
Oct 25 13:47:27 <aurele> would it be possible to pick files from another package while unpacking a recipe (or while configuring, before starting to compile)
Oct 25 13:47:37 <aurele> I need a file in the source tree
Oct 25 13:47:44 <aurele> of gsoap
Oct 25 13:48:25 <kanavin> aurele: what file and where it comes from?
Oct 25 13:50:03 <aurele> I have a recipe with a onvif application, I need to pick some files from gsoap source tree (those files are part of the package)
Oct 25 13:50:25 <aurele> kanavin, gsoap has its own recipe
Oct 25 13:50:52 <kanavin> aurele: please provide more specifics, which are the files, what do they contain, and why are they needed
Oct 25 13:52:32 <aurele> files are includes and cpp source files, they are installed in the sysroot, I need to add them in my application source tree, I would like to know a way to get them from my recipe.
Oct 25 13:52:58 <aurele> lets say I have files in "tmp-glibc/sysroots/gip2/usr/share/gsoap" (gip2 is my machine name)
Oct 25 13:53:17 <aurele> kanavin, I want to get a file from this directory
Oct 25 13:54:06 <aurele> kanavin, (to me this is crap like I never did, but if anyone knows a better way using gsoap please let me know)
Oct 25 13:54:21 <kanavin> aurele: maybe you just need to link with gsoap library?
Oct 25 13:56:43 <aurele> kanavin, this is what I told the developper... and then he showed me... and I cried (and bought him a beer... don't do any gambling when gsoap is used)
Oct 25 14:00:12 <aurele> kanavin, I just need to copy some files. I think integration could be better, there is work to do for gsoap, but for now I can't see how to get plugin code compiled as libraries in gsoap configure.ac
Oct 25 14:03:20 <kanavin> aurele: you can add the files to SRC_URI, but I do not condone such ugly hacks
Oct 25 14:04:14 <aurele> I just found "STAGING_DATADIR" I think this is the path I need
Oct 25 14:56:54 <eduardas_m> Hello. What is the best way to figure out why and how various Python packages are being pulled into my final sdcard image? Using Yocto Pyro
Oct 25 15:00:58 <FenrirJk> I'm working on porting our bsp from krogoth to pyro, our image recipe have alsa-lib in the IMAGE_INSTALL variable, it was building fine before but I have now the error: couldn't find anything to satisfy 'alsa-lib' do you have any idea of what could be won't?
Oct 25 15:02:31 <neverpanic> eduardas_m: If you have a package manager on the target, you can ask that. E.g. for opkg, use opkg whatdepends $package or opkg whatdependsrec $package
Oct 25 15:04:56 <eduardas_m> neverpanic: thank you, just started using opkg in my builds... did not know it could do that
Oct 25 17:02:15 <bodangly> If I want to make BUILDHISTORY_COMMIT = "1" and INHERIT += "buildhistory" part of my image, can I just put those in my custom "images" recipe's .bb file?
Oct 25 17:02:28 <bodangly> Rather than rely on local.conf for it which isn't put into git
Oct 25 17:41:56 <rauji> Hello, I want to copy file to folder & then want to apply patch on top of it whats the best way ?
Oct 25 17:43:03 <lsandov> rauji: SRC_URI is the way to do it
Oct 25 17:43:25 <lsandov> rauji: place the file first then the patch to ti
Oct 25 17:43:27 <lsandov> *it
Oct 25 17:43:42 <rauji> lsandov: SRC_URI is not allowing me to copy file, but it does allow me to apply patch..
Oct 25 17:44:07 <rauji> lsandov: getting this error  unparsed line: 'SRC_URI_append_
Oct 25 17:45:00 <lsandov> rauji: you need to move the file to S somewhere, perhaps do_patch_prepend ()
Oct 25 17:45:32 <lsandov> rauji: or have two patches, one for the file and the other to the patch to it
Oct 25 17:46:43 <rauji> lsandov: thank you for reply let me give try.
Oct 25 18:37:19 <rauji> lsandov: gave try with do_patch_prepend() {  	cp -rf ${WORKDIR}/foo/* ${S}/drivers/foo/ } but didnt work out
Oct 25 18:52:45 <brianm_> Dumb question: how do I generate the "run.do_compile" scripts without actually running them?
Oct 25 18:57:18 <lsandov> rauji: what failed?
Oct 25 18:57:38 <lsandov> brianm_: bitbake has a dry-run command but not sure if that is produced
Oct 25 18:57:58 <lsandov> s/command/command line argument/
Oct 25 19:04:16 <rauji> lsandov: getting error : unable to parse, guess it didnt like do_patch_prepend
Oct 25 19:04:57 <lsandov> rauji: bibtake is not able to parse your metadata, so it not even trying the patch task
Oct 25 19:05:19 <lsandov> rauji: pastebin somewhere so you get better help
Oct 25 19:36:23 <rauji> lsandov: here is pastebin - https://pastebin.com/h6SZhzPQ Note, this is part of .inc file which gets included in .bb file at last..
Oct 25 19:38:38 <lsandov> rauji: i believe this task do_defconfig_patch  needs some leading spaces
Oct 25 19:39:30 <lsandov> rauji:  you can try snippets and quickly check if these work with bitbake -e
Oct 25 19:45:50 <rauji> lsandov: sure thanks
Oct 25 20:11:56 <brianm_> lsandov: Thanks, but unfortunately dry-run doesn't create those files
Oct 25 20:12:07 <brianm_> The real problem I'm after is "how do I find out what ${FOO} will be in recipe X"?
Oct 25 20:14:05 <lsandov> brianm_: bitbake -e X | grep ^FOO
Oct 25 20:22:00 <brianm_> lsandov: Thanks! That's exactly what I wanted
Oct 25 20:22:51 <lsandov> brianm_: thanks grep, without it we were lost!
Oct 25 20:23:16 <lsandov> brianm_: just kidding, np
Oct 25 22:17:45 <rauji> lsandov: this works for me do_setup_additional_sources & do copying folder here & then call do_patch
Oct 25 22:17:52 <rauji> lsandov: thank you for help
**** ENDING LOGGING AT Thu Oct 26 03:00:01 2017