**** BEGIN LOGGING AT Fri Jan 08 02:59:57 2021 Jan 08 08:27:17 Unfortunately, the new qemu is hanging on 32 bit musl builds :( Jan 08 08:46:45 v0n: are you sure you're talking about "encrypted" or rather "authenticated"? Jan 08 08:47:04 (i.e. signed and checked) Jan 08 08:48:20 if you have everything authenticated except the bootloader, replacing the bootloader on the medium storage or doing a side-channel attack and your system is NOT secure anymore. Secure boot is either all of it or none of it, compromises aren't really friends with security ;) Jan 08 09:08:22 yo dudX Jan 08 09:10:10 LetoThe2nd: welcome back and happy new year \o/ Jan 08 09:10:41 howdy mckoan \o/ Jan 08 09:11:16 happy (relatively new) year to all of you too Jan 08 09:11:20 happy new year Yoctos :) Jan 08 09:20:20 * qschulz waves Jan 08 09:24:50 yo qschulz Jan 08 09:25:06 already properly selected your kernel? ;-) Jan 08 09:36:59 if you find out why 4.19 is preferred over 5.2 I'm all ears :) Jan 08 09:37:38 i don't care in the least :) Jan 08 10:06:19 Hi Yocto team Jan 08 10:06:40 Can we generate an TEMPFS + SELINUX Jan 08 10:06:48 image using Yocto Jan 08 10:13:55 techically, probably yes. out of the box? probably not. Jan 08 10:15:39 Any pointers how can i try this Jan 08 10:15:59 like any doc / blog or discussion Jan 08 10:16:07 1) set up a standard build 2) make that into a tempfs build 3) add selinux? Jan 08 10:16:16 ok Jan 08 10:16:25 will try this Jan 08 10:16:43 (whatever a tempfs build is anywas :P ) Jan 08 10:17:12 thanks letothe2nd Jan 08 10:21:08 have fun! Jan 08 10:21:45 Hello! Has anyone had issues with changing INITSCRIPTS_PARAMS under zeus? I am trying that for the openssh and keep getting: "Postinstall scriptlets of ['openssh-sshd'] have failed. If the intention is to defer them to first boot, Jan 08 10:21:45 then please place them into pkg_postinst_ontarget_${PN} ()." Jan 08 10:39:02 gsalazar: what's your change? Jan 08 10:41:19 abelloni, hi! I saw it. is it able to run on 512MB RAM + 4x1GZ cpu? Jan 08 10:42:17 Did anybody try to run Firefox or Chrome at 512MB RAM/4x1 GZ cpu ? Jan 08 10:42:47 I guess the main question would be about the GPU, not the CPU Jan 08 10:54:21 qschulz: I have the INITSCRIPT_PARAMS depending on the distro, and either removing it or keeping the original value Jan 08 10:55:12 INITSCRIPT_PARAMS_${PN}-sshd = "${@oe.utils.ifelse(d.getVar('DISTRO') == 'distro-name','remove','defaults 9')}" Jan 08 10:55:57 gsalazar: just use INITSCRIPT_PARAMS_${PN}-sshd_distro-name = "remove" Jan 08 10:57:08 might not fix it but at least it is easier to read/debug Jan 08 10:58:55 qschulz: I don't get the error anymore. Thanks! I'll test it and see if it is working Jan 08 11:02:09 gsalazar: you can check with bitbake -e openssh | grep -e "^INITSCRIPT_PARAMS_openssh-sshd=" (or maybe replace openssh with ${PN} don't remember if it's expanded or not in bitbake -e Jan 08 11:02:25 if it's the correct value for your distro, that's it :) Jan 08 11:03:23 wrong value :( Jan 08 11:03:54 have you selected the correct distro? Jan 08 11:04:00 how's your distro named? Jan 08 11:08:00 yes, its named nbox-distro. but running: bitbake -e openssh | grep -e "^DISTRO" I don't see the variable Jan 08 11:18:00 gsalazar: what's the content of OVERRIDES in bitbake -e? Jan 08 11:18:08 or more specifically DISTROOVERRIDES Jan 08 11:18:20 where did you select your DISTRO also? Jan 08 11:18:22 and how? Jan 08 11:18:59 using devtool modify in dunfell causes ModuleNotfoundError no module named '_sysconfigdata'. Any hints what that is about, bug in my end or dunfell? Jan 08 11:19:42 qschulz: DISTROOVERRIDES="fslc" Jan 08 11:20:42 I set DISTRO = "nbox-distro" in distro conf file Jan 08 11:22:16 but the other variables set in the same way (ie DISTRO_NAME) are present in bitbake -e Jan 08 11:28:33 gsalazar: your DISTRO needs to be set in local.conf or passed from the environment (before bitbake) Jan 08 11:29:03 yocto will search in all layers' conf/distro/ a ${DISTRO}.conf file Jan 08 11:29:31 It is also set in local conf but with DISTRO ?= Jan 08 11:30:06 gsalazar: ok, look in bitbake -e | less and look for the line starting with DISTRO= Jan 08 11:30:16 read the lines above Jan 08 11:30:24 you'll understand why ?= does not work for you Jan 08 11:30:48 (spoiler alert, someone else sets it for you with = or ?= but before local.conf is parsed) Jan 08 11:31:27 Try with DISTRO="nbox-distro" in your local.conf and if that still does not do it, you'll need to punch someone in the face probably Jan 08 11:32:07 virtually, because social distancing... and we're non-violent Jan 08 11:34:25 qschulz: Strange. There isn't a single line starting with DISTRO= in bitbake -e... Jan 08 11:37:12 gsalazar: I see `unset DISTRO` Jan 08 11:37:20 and it has the info I wanted to go for this :) Jan 08 11:38:04 I do see unset DISTRO, I'll take a deeper look. Thanks Jan 08 11:38:27 if DISTRO = "nbox-distro" in local.conf does not work, please upload the few lines between # $DISTRO and unset DISTRO to a pastebin and give us the link so we can help further Jan 08 11:41:28 It did not help. Here it is: https://pastebin.com/nGGqkktx Jan 08 11:42:42 you did not set DISTRO = "nbox-distro" in local.conf Jan 08 11:42:50 you set it to DISTRO ?= "nbox-distro" Jan 08 11:43:28 I created the paste before, I can rerun it Jan 08 11:43:32 or you tried to pass it from the environment maybe? Don't know if that is supported (I do it for MACHINE, so expected it to work OOTB for DISTRO?) Jan 08 11:43:36 gsalazar: yes please Jan 08 11:44:51 https://pastebin.com/ntYP0d7c Jan 08 11:48:05 how to write recipe for python3 package that has setup_requires? I get "ERROR: Do not try to fetch `my-python-app~=0.2.0' for building. Please add its native recipe to DEPENDS". So I made recipe for my-python-app and put it to DEPENDS, but error remains Jan 08 11:50:24 oh nevermind, the keyword was "native" recipe Jan 08 11:59:40 gsalazar: can you send me the few lines above your ^INITSCRIPT_PARAMS line? Jan 08 12:00:01 (and check the output of DISTROOVERRIDES too while at it) Jan 08 12:00:38 (and also the few lines before DISTROOVERRIDES) Jan 08 12:01:07 qschulz: I tried setting it from the environment and it still does not show up on the bitbake -e Jan 08 12:01:39 the output from bitbake -e? Jan 08 12:05:24 yes, bitbake -e output for the lines above DISTROOVERRIDES= and INITSCRIPT_PARAMS_.*= Jan 08 12:07:02 tig Jan 08 12:11:22 qschulz: what I find strange is that the initial error was not present in thud (I am upgrading from thud to zeus) Jan 08 12:17:57 gsalazar: nothing standing out in the release notes :/ Jan 08 12:18:10 can you still send me the bitbake -e output? Jan 08 12:20:56 qschulz: sure https://pastebin.com/tzCxu2pL Jan 08 12:21:28 gsalazar: UGH FUCKING FREESCALE DOING IT AGAIN Jan 08 12:21:37 told you we were about to punch someone Jan 08 12:22:17 something in meta-freescale is overriding the DISTRO? Jan 08 12:22:44 ok, so can you set DISTROOVERRIDES = "nbox-distro" in your nbox-distro.conf file? Jan 08 12:23:42 still confused as to why it works this way... since fslc distro isn't actually used... but maybe it being parsed fucks up everything else? Jan 08 12:24:02 I tried it but the unset for the DISTRO is still present.... Jan 08 12:24:10 this is anoying... Jan 08 12:24:48 but DISTROOVERRIDES was set to the proper value Jan 08 12:24:56 ok, that's enough for us Jan 08 12:25:11 yeah, unset DISTRO is also there for me, i woyuldn't worry about it Jan 08 12:25:35 if DISTROOVERRIDES is now correctly set, your INITSCIPT_PARMAS_${PN}_nbox-distro will work Jan 08 12:26:31 minus all the typos :p Jan 08 12:29:37 :P thanks for all the help! Jan 08 12:38:33 my pleasure, now I have a question for others :) Jan 08 12:40:04 gsalazar: wait Jan 08 12:40:17 what is the content of you distro configuration file (nbox-distro)? Jan 08 12:40:30 do you happen to include or require another distro configuration file? Jan 08 12:53:28 I'm pretty sure you included another distro or distro include configuration file that pulls this DISTROOVERRIDES Jan 08 12:53:51 so it makes sense, I just naively thought you were using a completely different distro than fslc but you probably based yours on that one Jan 08 12:54:02 so all good from Yocto's perspective :) Jan 08 12:58:09 abelloni, GPU is very-very basic... Jan 08 13:08:47 qschulz: yes, I require fslc-base Jan 08 13:10:18 gsalazar: there you have it :) Jan 08 13:10:33 everything makes sense now :) Jan 08 14:21:55 Hello everyone. Please I need help understanding the function of the TEMPDBGSRCMAPPING variable. Jan 08 14:25:58 idadel: it's a variable that is added in RP's branch, so he'll be the only one able to help probably Jan 08 14:26:41 except if you have a more specific question and you put a bit of context, maybe Yocto veterans will be able to guide you (probably not me :) ) Jan 08 14:28:34 idadel: its a temporary internal variable storing the information returned by oe.utils.multiprocess_launch(splitstaticdebuginfo, staticlibs, d, extraargs=(dvar, debugstaticdir, debugstaticlibdir, debugstaticappend, debugsrcdir, d)) Jan 08 14:32:53 idadel: BTW, you probably want to pass the licenses from the recipes through canonical_license() from license.bbclass, that would eliminate the false positive you have in your mail Jan 08 14:38:35 RP: Got it Jan 08 14:38:47 qschulz: Okay thanks. Jan 08 14:49:02 idadel: might want to use https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/lib/oe/license.py#n10 license_ok() and pass the licenses from the file as dont_want and if it returns false, it means the license from your filee is in your package licenses Jan 08 14:49:51 anyway, meta/lib/oe/license.py and meta/classes/license.bbclass are worth a read for inspiration at least and avoid reiventing the wheel if you can (no idea what you're going for or if it's a good hint I'm giving you, so take with a grain of salt :) ) Jan 08 14:55:28 qschulz: Thanks. Jan 08 15:17:11 qschulz, so there are around binutils_2.32 patches for gcc10 in these zeus-based distros Jan 08 15:17:39 and fun enough I can't fix a silly .bbappend issue Jan 08 15:18:47 nativesdk-binutils won't find the patch..under /files Jan 08 15:20:36 I am suing the usual tricks for suffixes/classes/overrides but here the prefix seems tricking me Jan 08 15:22:30 dv: be aware that building chromium takes an enormous amount of RAM for the final linking; 16GB or RAM is recommended Jan 08 15:22:41 https://github.com/OSSystems/meta-browser/blob/master/recipes-browser/chromium/README Jan 08 16:11:04 ant__: the prefix? Jan 08 16:11:31 can you give us more context on what you're trying to do and give us some recipe "code" and some bitbake -e output Jan 08 16:44:19 What is the best way to disable coredumps on yocto? Jan 08 17:40:22 rob_gries: fix the code that produces them? Jan 08 17:47:30 kanavin: found the weird musl qemu issue, or at least a much better understanding of it... Jan 08 18:16:17 RP: cheers :) I'm on a holiday this week, and halstead still hasn't installed my new ssh key :-/ Jan 08 18:16:49 RP: there was a progressive harddrive failure in typical SSD fashion Jan 08 18:17:04 I copied everything out to USB sticks when I realized what's about to happen Jan 08 19:05:14 qschulz, https://github.com/OpenPLi/openpli-oe-core/tree/develop/meta-openpli/recipes-devtools/binutils Jan 08 19:05:56 RP: some magic for the .bbappend when extending BBCLASSEXTEND with nativesdk ? Jan 08 19:05:57 ^ Jan 08 19:06:52 btw the best ting would be to backport this in zeus (if it is really needed, both for native and cross?) Jan 08 19:09:38 (gcc10 fixes see c41fe48f in dunfell) Jan 08 19:42:19 When I set KBULD_DEFCONFIG this copies the config from arch/${ARCH}/configs/${KBULD_DEFCONFIG} to ${WORKDIR}/defconfig. However in kernel-yocto.bbclass:341 if there is a defconfig it looks like it defaults to using "allnoconfig". Is this correct? Usually the kernel defconfigs assume default configs are enabled. Jan 08 21:22:20 kanavin: fonts in weston/wayland arent much different see https://i.imgur.com/46Z4vsy.png Jan 08 21:27:01 root@qemux86:~# opkg list-installed | grep liberation-fonts Jan 08 21:27:02 liberation-fonts - 1:2.00.1-r0.10 Jan 08 22:01:59 Do people do stuffs like compiling two kernels with yocto, one minimal to kexec a second one? Jan 08 22:03:36 I tried that a long time ago, but we gave up on it because the vendor kernel support for kexec (at the time) was a broken Jan 08 22:06:26 JPEW: ok. Ideally I would like to protect the kernel+dtb+initramfs, before switching root to the customer rootfs. Any idea how I can do that? Jan 08 22:07:11 I was thinking about having kernel+dtb+initramfs in an encrypted partition that u-boot or barebox decrypt and boot from there, but I'm not even sure they can do that. Jan 08 22:17:29 v0n, well, we abused of kexec the other way around...the kexecboot-kernel could boot self-built images from various media Jan 08 22:17:51 ..to circumvent the vendor bootloader-kernel... Jan 08 22:18:46 indeed it was a solution for obsolete/crapped bootloaders Jan 08 22:32:32 Anyone using KBUILD_DEFCONFIG? Jan 08 22:32:56 Seems like it setup to not be a defconfig but a full config. Jan 08 23:25:00 ant__: I don't know how to parse that Jan 08 23:25:19 WARNING: /oe/openpli-oe-core/openembedded-core/meta/recipes-devtools/binutils/binutils_2.32.bb: Unable to get checksum for nativesdk-binutils SRC_URI entry 0001-binutils-program-name-multiply-defined.patchfile: file could not be found Jan 08 23:26:03 vs $ cat ./binutils_2.32.bbappend Jan 08 23:26:04 FILESEXTRAPATHS_prepend := "${THISDIR}/${BPN}:${THISDIR}/files:" Jan 08 23:26:04 SRC_URI += "file://0001-binutils-program-name-multiply-defined.patch" Jan 08 23:26:34 tlwoerner, wow! 16GB for linking... impressive. What sbout Firefox? on PC I prefer Fox. But Which one will work better in ARM A7 4x1GZ + 512 MB ram? Jan 08 23:26:38 the sad story is that thy are moving to gcc10 with one feet in zeus Jan 08 23:27:15 RP: it was simple to fix -cross and -native Jan 08 23:27:46 ant__: ".patchfile: " suggests a missing space somewhere Jan 08 23:28:09 I've seen, think is the output Jan 08 23:28:18 kanavin: key should get sorted ;-) Jan 08 23:28:36 kanavin: ssd sounds painful :( Jan 08 23:29:15 I've checked ${THISDIR} and it points to the binutil dir in the overlay, the search paths seem correct Jan 08 23:36:53 ant__: humour me and add spaces around the file:// in the bbappend? Jan 08 23:37:15 ant__: worst case it doesn't help ;-) Jan 08 23:38:15 omg..seems workin', let me reproduce Jan 08 23:38:47 ant__: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=69e3b6c80ea7cab7f8687ee223dfd29dcc2c0d75 Jan 08 23:39:32 my bad then :) Jan 08 23:39:44 many thanks for the pointers! Jan 08 23:41:47 I should have converted the += in _append as I normally do Jan 08 23:42:14 (now I don't tell you about the tons of _append += in these layers...) Jan 08 23:42:22 np, I think I had faint memories of that patch Jan 08 23:42:32 you have a crazy ram Jan 08 23:42:40 nvram indeed Jan 09 00:05:38 RP: many thanks again, gn Jan 09 02:18:22 dv: sorry, i only have experience with chromium-x11 on arm64 (i think it had 1GB of ram). give both a whirl; tell us what happens ;-) Jan 09 02:19:43 dv: i felt compelled to give the warning because the chromium build only fails after you've downloaded the sources for 4 hours, and compiled for 2 hours… *THEN* it tells says you don't have enough RAM on the build machine ;-) Jan 09 02:20:15 lol **** ENDING LOGGING AT Sat Jan 09 02:59:57 2021