**** BEGIN LOGGING AT Tue Feb 04 03:00:52 2020 Feb 04 12:39:12 https://marcin.juszkiewicz.com.pl/2020/02/04/fosdem-2020/ Feb 04 12:39:33 OE workshop was great! Feb 04 14:57:03 hi Feb 04 15:02:26 is anyone using meta-rockchip with yocto 3.0? I tried a clean build today, actually for the first time for this target and it fails in u-boot do_deploy - not finding some "idbloader.img" which according to google was indeed some binary in the past but I understand these days it should be generated during the build? Feb 04 15:03:15 not getting much hints on this, does anyone have an idea? Feb 04 15:13:42 Jin^eLD: I've recently done some work in meta-rockchip. What machine are you building? Feb 04 15:14:07 tinker-board Feb 04 15:14:30 MACHINE="tinker-board" Feb 04 15:14:30 trying meta-rockchip master now insead of zeus Feb 04 15:15:45 I see that conf/machine/include/rk3288.inc sets SPL_BINARY ?= "idbloader.img" Feb 04 15:15:54 but I still wonder where the file is supposed to be coming from Feb 04 15:16:29 poky-build/build/work/tinker_board-poky-linux-gnueabi/u-boot/1_2019.07-r0/build/idbloader.img': No such file or directory Feb 04 15:17:13 only .img files I see are: Feb 04 15:17:15 ./build/u-boot-dtb.img Feb 04 15:17:15 ./build/u-boot.img Feb 04 15:18:02 and there are no files called "idbloader*" at all in the u-boot build directory Feb 04 15:19:44 I did not add any cutom configuratoins or overrides or whatever so I'm on a pretty vanilla version of poky/3.0 Feb 04 15:22:31 found this http://opensource.rock-chips.com/wiki_Boot_option#What_is_idbloader from an IRC log conversation between khem and tlwoerner Feb 04 15:23:10 still a bit confused if one is supposed to organize that file before building the layer or if meta-rockchip is supposed to handle it somehow Feb 04 15:29:21 Jin^eLD: idbloader comes from U-Boot Feb 04 15:29:41 Jin^eLD: it's U-Boot's SPL, simply renamed Feb 04 15:30:25 tlwoerner: but should it be built automatically and if yes, why am I missing it? Feb 04 15:30:27 Jin^eLD: let me try a "fresh" build, but my nightly master build of tinker-board is green... Feb 04 15:30:31 oh Feb 04 15:31:42 OK thank you, curious what results you will get... I hope it'll be reproducible Feb 04 15:31:44 Jin^eLD: yes, it is built automatically. the U-Boot defconfig tells U-Boot to build an SPL, the MACHINE file from meta-rockchip tells it to rename it to idbloader.img Feb 04 15:34:04 hmm, I wonder what happened then Feb 04 15:40:44 Jin^eLD: Perhaps you are building the wrong u-boot for some reason? Feb 04 15:45:01 JPEW: I did not change any of it, I simply selected the machine conf provided by the meta-rockchip layer and that's it, it should have selected the proper u-boot and so on Feb 04 15:45:34 let's see if tlwoerners clean build works or not... Feb 04 15:46:09 Jin^eLD: OK Feb 04 15:50:27 brb 30min Feb 04 16:27:11 Jin^eLD: my clean build for tinker-board succeeds, and idbloader.img, and friends, are found in the deploy directory Feb 04 16:27:42 Jin^eLD: very basic build, core-image-minimal, nodistro Feb 04 16:50:55 re Feb 04 16:51:15 tlwoerner: "nodistro" as in literally "nodistro"? I used "poky" Feb 04 16:51:38 Jin^eLD: yes, i tend to use nodistro a lot Feb 04 16:51:46 ok let me try that Feb 04 16:52:11 Jin^eLD: i think JPEW uses poky a lot, so i'm guessing he has tested with that Feb 04 16:52:27 question is why does it work for you guys, but not for me hmm Feb 04 16:52:38 can the build host be an issue? I'm on Fedora Feb 04 16:52:45 conf/local.conf: DISTRO = "nodistro" Feb 04 16:52:58 i'm on openSUSE Feb 04 16:53:20 I'll try with nodistro now and will see what happens Feb 04 16:53:24 Jin^eLD: I would hope that the host distro wouldn't matter Feb 04 17:00:01 OK thats really weird... nodistro build fails in "attr" Feb 04 17:00:03 | gmake[1]: o: Command not found Feb 04 17:00:08 what the hell is going on there? Feb 04 17:00:19 o? Feb 04 17:00:22 what the deuce Feb 04 17:01:46 yeah, confused as well, let me clean out everything again and make sure I have no ccache or whatever Feb 04 18:20:33 tlwoerner: u-boot is at do_compile, so now it's getting interesting Feb 04 18:20:49 wiping ccache etc seems to have solved the gmake "o" build error no idea what it was Feb 04 18:30:02 build/idbloader.img': No such file or directory Feb 04 18:30:05 that's with "nodistro" Feb 04 18:30:39 so we get different results for some reason... I guess I should be looking at the u-boot makefiles closely? Feb 04 18:31:22 only remaining differnce is the host distro, which - I agree with JPEW - should not matter... can there be something else? Feb 04 18:31:57 my machine conf: https://paste.ee/p/gTFKP Feb 04 18:32:07 Jin^eLD: Can you post log.do_compile in a paste bin or something? Feb 04 18:32:24 sure, let me get it Feb 04 18:35:54 to big for paste.ee Feb 04 18:35:57 http://deadlock.dhs.org/jin/log.do_compile.txt Feb 04 18:36:23 actually there is not a single mention of idbloader in the do_compile log Feb 04 18:39:48 if I grep in temp of the u-boot build, then the only logs mentioning idbloader are do_deploy (thats where it fails) and do_install Feb 04 18:47:16 Jin^eLD: Hmm, I'm building u-boot 2020.01 on master... Feb 04 18:47:43 I could switch from 3.0 to master of course... Feb 04 18:47:52 and see what happens Feb 04 18:48:33 Jin^eLD: I recall at one point I had to backport a patch to get idbloader.img to work, but I think I was able to drop it because oe-core had a new enough u-boot Feb 04 18:48:56 oe core on what branch? 3.0 or master? Feb 04 18:51:08 I would have been master, but a while ago (1-2 months) Feb 04 18:51:21 I suspect the patch is needed on 3.0 Feb 04 18:51:35 yeah I think 3.0 is older than 1-2 months, that could explain it Feb 04 18:52:22 I switched all layers to master now, so let's see Feb 04 18:52:50 JPEW: so you're playing around with the tinker board specifically? or just rockchip in general? Feb 04 18:53:04 Jin^eLD: 78af73efa is the commit needed from u-boot I think Feb 04 18:53:26 If you want to make a patch to meta-rockchip to backport it in u-boot.bbappend :) Feb 04 18:53:39 Jin^eLD: Both Feb 04 18:53:47 I have a RockPi4 (rk3399) also Feb 04 18:54:24 I have a weak spot for raspberry pi clones... I have a stack of 5 on my desk right now :) Feb 04 18:54:24 that 78af73efa - on what repo is it? Feb 04 18:54:31 heh Feb 04 18:54:33 u-boot Feb 04 18:54:53 oh, its not something in the layer but in the actual u-boot tree? got it Feb 04 18:55:21 I am helping out a friend to set up an swupdate based double-copy-with-fallback rootfs update thingie on a tinkerboard Feb 04 18:56:29 did not do much with this hardware before Feb 04 19:14:18 Jin^eLD: oh yes, i should have thought of that before, i'm using all master, here's my config/hases: https://pastebin.com/AusqA6hQ Feb 04 19:15:28 that'll probably be it :) its still building... Feb 04 19:15:44 JPEW: which 5? Feb 04 19:35:13 tlwoerner: Raspberry Pi 3, RockPi4, Tinker board, Pine H64, IMX8 Pico Pi Feb 04 20:02:03 https://marcin.juszkiewicz.com.pl/fosdem-videos.html Feb 04 20:07:19 JPEW, tlwoerner: looks like u-boot build ran through on master, so that was indeed it, thank you! **** ENDING LOGGING AT Wed Feb 05 03:00:08 2020