**** BEGIN LOGGING AT Fri Oct 12 03:00:00 2018 Oct 12 04:28:32 d Oct 12 06:26:26 What's the status of rk3066 support on meta-rockchip layer from yocto git? I'm looking to build for marsboard-rk3066 Oct 12 06:35:10 Anybody that got icecc working with yocto ? Oct 12 06:36:27 my remote iceccd's complain about what "yocto" sends them as toolchain Oct 12 06:36:34 LowLander: we've ahd occaional reports that its supposed to work, yes. are you seeing any particular problems, or was this more of a survey style question? Oct 12 06:36:55 and localy it gives compile errors as soon as I enable it Oct 12 06:37:23 the icecc network works with my ptxdist (www.ptxdist.org) setup Oct 12 06:37:39 LowLander: this is probably not super up to date, but a start: https://www.openembedded.org/wiki/Using_IceCC Oct 12 06:38:04 yeah seen taht and that didn't help Oct 12 06:38:27 and https://github.com/openembedded/openembedded-core/commits/master/meta/classes/icecc.bbclass suggests that its seeing continueing love Oct 12 06:38:31 JPEW: ^^^^^^^ Oct 12 06:40:04 hmmm don't think there is anything relevant commited after "sumo" Oct 12 06:40:47 not relevant, but maybe some hints to get it working. lets hope JPEW shows up some time. Oct 12 06:42:00 I am a bit surpriced not more ppl use it Oct 12 06:42:26 it seems the way to improve build times Oct 12 06:42:55 what timezone is JPEW ? Oct 12 06:57:16 LowLander: no idea about his time zoen Oct 12 06:57:32 we just use a bit buildserver and sstate, no need for further tinkering Oct 12 08:22:10 Hi there, I need some help. I used devtool modify and update recipe commands and got my patch. But when i start bitbake i got errors. I found that newly added patch somehow broke .bb file licence part. If i commend the patch name from the SRC_URI its working fine. Also if i turn the license CLOSED its working too. Any idea? Oct 12 08:32:19 EmreT: depends on what the patch is doing... does it touch any file mentioned in LIC_FILES_CHKSUM ? Oct 12 08:35:40 jostor: is it replacing the original config with your new config Oct 12 08:49:59 Its u-boot patch. I just configured u-boot files (rcw, u-boot board config).In patch there is nothing about License or Yocto config. Oct 12 08:50:35 http://lists.openembedded.org/pipermail/openembedded-core/2018-July/153275.html Oct 12 08:50:49 my error same with on this page Oct 12 08:51:23 but my version i guess diffrent and patch is not compatible with my files. Oct 12 09:12:37 LetoThe2nd, how do you mean -> we just use a bit buildserver and sstate, no need for further tinkering Oct 12 09:15:17 LowLander: we have a beefy buildserver and utilize sstate cache as much as possible, so we don't need to go to greater lengths for optimizing the build time. Oct 12 09:17:48 ah ok Oct 12 09:26:04 I changed license file and file path from bb file, after changed it to first license and its working :) so weird. Oct 12 10:44:39 New news from stackoverflow: Unable to install a shared library in yocto Oct 12 12:13:15 mckoan: I think so. From what I understand from the warning it does. The image I built with this isn't booting. When I removed the SRC_URI += "file://defconfig" it is booting. Oct 12 12:15:03 jostor: I'd run a diff on both to see what's happening Oct 12 12:16:35 mckoan: I am sorry. What should I run a diff on? Oct 12 12:40:28 "directories installed but not shipped, /usr/lib/cmst/cmstroothelper" this file is available at this location workdir/image/usr/lib/cmst/cmstroothelper ..... my question is who put this file at this location {workdir}/image Oct 12 12:40:39 recipies' makefile? Oct 12 12:41:28 xtron: the makefile installed it there. you just need to add to FILES_* to put it in the right package Oct 12 12:42:03 if you want those in the main package FILES_${PN} += "${libdir}/cmst/" would work, assuming they're all binaries/modules/etc Oct 12 12:42:27 rburton: there are alot of other files in workdir/image not being handle by FILES_* so why there is no warning for them? Oct 12 12:44:15 because FILES has a default value? Oct 12 12:44:53 see bitbake.conf where all the defaults are Oct 12 12:46:54 rburton: you means values not defined for FILES_* in bitbake.conf have to be define in recipe? Oct 12 12:47:05 like additional path? Oct 12 12:47:19 FILES_${PN} has a default value, if you want more files in that package then add them to FILES_${PN} Oct 12 12:47:45 rburton: right, got it Oct 12 12:59:51 LowLander: GMT -5 is my timezone Oct 12 13:00:47 LowLander: What version of Yocto are you using, and what version of iceccd? Oct 12 13:01:48 JPEW tried with sumo Oct 12 13:02:07 and on Centos7 with that icecream Oct 12 13:02:45 which is 1.1 Oct 12 13:03:20 LowLander: Ya that should work. Whats the error you see? Oct 12 13:03:20 I know the setup works, because i use it already for ptxdist builds Oct 12 13:03:46 JPEW the remote complains about the toolchain it gets Oct 12 13:04:00 one moment phone Oct 12 13:04:06 n/p Oct 12 13:15:09 New news from stackoverflow: configure image sensor device on i2c bus in linux Oct 12 13:23:58 jostor: diff old-defconfig new-defconfig Oct 12 13:45:16 New news from stackoverflow: Yocto: INITSCRIPT_PARAMS not works as expected Oct 12 13:47:17 JPEW ping Oct 12 13:47:28 j/s :) Oct 12 13:47:34 pesky clients :-) Oct 12 13:48:15 so yeah the toolchain it tries to push to the remote host is not liked Oct 12 13:48:49 mckoan: I got it to work, with some help from agherzan in https://gitter.im/agherzan/meta-raspberrypi. Renamed the file in my layer from defconfig to can.cfg Now it is working. Looks like when the file was called defconfig it was replacing the original defconfig. I have not quite understood how this is working Oct 12 13:49:48 mckoan: Thanks for your help with this! Oct 12 13:51:15 resulting in errors like Oct 09 00:48:57 eir.home.erwinrol.com iceccd[6780]: /usr/bin/tar: ./bin/true: Cannot create symlink to `..//usr/bin/true': Permission denied Oct 12 13:51:50 JPEW it seems it has some weird path setup Oct 12 13:53:53 Ya, perhaps you can look at the tar file that got generated? Oct 12 13:54:14 rburton: using FILES_${PN} += "${libdir}/cmst/cmstroothelper" still give me warning, and file is not included in packages-split Oct 12 13:55:30 LowLander: tar -tvf tmp/build/ice/ Oct 12 13:56:49 JPEW hang on Oct 12 13:57:24 jostor: so your defconfig contanined only portions of modification Oct 12 13:57:56 LowLander: Oh I remember why this sounds familiar; do you have sellinux enabled? Oct 12 13:58:09 selinux Oct 12 13:58:13 jostor: glad to know it works Oct 12 14:01:20 LowLander: https://bugzilla.redhat.com/show_bug.cgi?id=1479048 Oct 12 14:01:21 Bug 1479048: was not found. Oct 12 14:02:57 xtron: did you get the path right? does $libdir match the path? are there other files you're missing/ Oct 12 14:04:20 JPEW yes default centos7 setup with SElinux Oct 12 14:05:13 LowLander: Ah that would do it. I had to disable selinux, or you can try installing the additional permissions in that bug I filed Oct 12 14:05:14 mckoan: Yes, only the two modifications I wanted to add. I guess defconfig should include all kernel config, and portions should be in a .cfg file. Oct 12 14:06:05 JPEW I'll look into that Oct 12 14:06:08 Perhaps you if you also report it they will fix it upstream instead of just closing the bug :) Oct 12 14:07:08 rburton: that what I'm looking, libdir = /usr/lib64 ... and I need /usr/lib so I've to use prefix Oct 12 14:07:09 JPEW but why is it a problem the same works with ptxdist (www.ptxdist.org if you don't know that) just fine Oct 12 14:07:26 That icecream toolchain probably doens't have symlinks Oct 12 14:07:28 the same iceccd hosts Oct 12 14:08:40 JPEW but anyway thanks for the hints, I'll look into that over the weekend Oct 12 14:08:57 now back to pesky client that didn't call for no reason :-) Oct 12 14:10:51 JPEW PS there is very little info to find about ICE+yocto :-/ Oct 12 14:11:19 LowLander: Ya, I should update the documentation..... I'll throw that on my TODO mountain :) Oct 12 14:13:41 I have a TODO list with things I need to put on my TODO list ;-) Oct 12 14:21:51 xtron: so you need to tell the makefiles to respect $libdir instead of just throwing stuff in $prefix/lib and hoping for the best Oct 12 16:42:17 LowLander: it seems you got that idea from hollowchain project Oct 12 16:42:36 which is blockchain of blockchains Oct 12 17:17:54 at some poine we really need to fix it so bitbake can use -e against skipped recipes Oct 12 19:05:07 kergoth, do you know if there is a bug opened for that? Oct 12 19:50:53 denix: around ? Oct 12 19:51:25 khem: kind of :) Oct 12 19:53:37 denix: I was trying master on bbb and it does not boot Oct 12 19:53:39 known ? Oct 12 19:54:09 khem: from meta-ti? Oct 12 19:54:14 yes meta-ti Oct 12 19:55:33 khem: hmm, should be the same as rocko that we currently productize. the only difference is gcc version... can you try 7.3? and I can check here as well Oct 12 19:55:53 ugh Oct 12 19:56:17 7.3 lets try Oct 12 20:07:00 denix: which u-boot should we use for bbb from meta-ti Oct 12 20:07:29 denix: its using u-boot-ti-staging_2018.01.bb Oct 12 20:29:09 khem: that's correct Oct 12 20:31:46 denix: I think there are issues with gcc8 with earlier versions of u-boot Oct 12 20:32:17 denix: I have recipes where it uses precompiled toolchains for older u-boot/kernels Oct 12 20:34:25 khem: yeah, officially we use Linaro 7.x for our releases this year Oct 12 20:35:04 khem, Tartarus: are there u-boot patches to work with gcc 8 that we can back-port to 2018.01? Oct 12 20:35:43 denix: Depends on target Oct 12 20:35:50 the xscale thing is hard :( Oct 12 20:35:53 ateml has a patch Oct 12 20:35:58 (nand) Oct 12 20:36:04 x86 has a patch too Oct 12 20:36:41 Tartarus: bbb at least Oct 12 20:36:49 denix: Well, what's the build error? Oct 12 20:37:14 https://travis-ci.org/trini/u-boot/builds/438800860 Oct 12 20:37:19 is what fails out of the box atm Oct 12 20:37:39 Tartarus: it builds, but won't boot Oct 12 20:37:50 can't say then Oct 12 20:37:53 khem: do you have more input? ^^^ Oct 12 20:37:54 Haven't done boot test Oct 12 20:41:21 Is this an appropriate channel for general yocot/bitbake questions? Oct 12 20:41:31 yocto* Oct 12 20:45:15 In the context of a .bbappend file, is "SRC_URI_machine += "file://myfile.diff"" the same as "SRC_URI_append_machine = "file://myfile.diff""? Oct 12 20:45:54 bb88: They are very similar Oct 12 20:46:29 bb88: Although you would want SRC_URI_append_machine = " file://myfile.diff" (note the space) Oct 12 20:48:28 JPEW: Would I need that space if I used the += method? Oct 12 20:48:43 bb88: No, it is added implicitly Oct 12 20:49:10 bb88: https://www.yoctoproject.org/docs/1.6/bitbake-user-manual/bitbake-user-manual.html#basic-syntax Oct 12 20:50:08 JPEW: Great thanks. Oct 12 20:56:46 I added the SRC_URI_machine += "file://myfile.dff", but it complains: "can't find file to patch at input line 5" Oct 12 20:57:15 It appears that it does not know what to append to, is there another variable to utilize when using SRC_URI_machine? Oct 12 21:01:11 bb88: FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" Oct 12 21:11:54 denix: I am using "FILESEXTRAPATHS_prepend := "${THISDIR}/diff:"" and I have a diff/ directory containing my patch Oct 12 21:18:07 bb88: "can't find file to patch at input line 5" - apparently it can't find source file. do you have correct patch level? Oct 12 21:43:53 denix: Strange findings... Apparently SRC_URI_machine += "file://myfile.diff" is different than SRC_URI_append_machine = " file://myfile.diff" Oct 12 21:44:14 Because the latter worked, whereas the former gave me the error Oct 12 21:44:39 Does "+=" prepend the .bb file instead of appending? Oct 12 21:48:51 Tartarus: it build ok, but boot doesnt seem to go past MLO messages, Oct 12 21:49:11 I am building with gcc7 and see if that boots Oct 12 21:49:32 that will atleast make it a compiler related issue Oct 12 21:50:58 khem: 7 is OK Oct 12 21:51:11 7.3.0 from kernel.org is the normal one and I boot my am335x_evm with each push Oct 12 21:51:18 which is ~== BBB for what we're doing here Oct 12 21:52:53 Tartarus: OK let me see where I endup Oct 12 23:41:13 denix: should wic image work for bbb sd-card ? Oct 13 00:19:19 bb88: i believe it's a bad idea to mix += and overrides Oct 13 01:44:59 khem: yes, if it boots, wic should work Oct 13 01:50:57 denix: So I switched u-boot to one from oe-core and its working, I also added copy wks file from meta-yocto-bsp Oct 13 01:51:10 denix: so it seems something is hosed in TI u-boot Oct 13 01:51:23 bb88: VAR_machine += "val" is not the same as VAR_append_machine = "val"! first one is 2 step process - 1. evaluate += and add to VAR_machine first (which basically sets it, as it was empty) and 2. evaluate VAR_machine by overriding VAR if MACHINE=machine Oct 13 01:51:30 gcc8 or gcc7 is not an issue Oct 13 01:52:19 khem: ok, I'll dig further. someone was saying beaglebone config had issues, while am335x was fine Oct 13 01:52:52 Tartarus: see above ^ - bbb seems to be slightly different than am335x Oct 13 01:53:04 should I send my patch Oct 13 01:53:16 khem: which patch? Oct 13 01:53:38 the one I am using for switching u-boot Oct 13 01:53:47 denix: oh? Oct 13 01:56:03 denix: I will send them to ml for reference Oct 13 01:56:31 while you debug the issue in ti-staging version, this can unblock master Oct 13 01:56:41 khem: using latest u-boot from oe-core? Oct 13 01:57:44 yes Oct 13 01:57:50 denix: patches are on ml now Oct 13 01:58:07 Tartarus: this is 2018.01 base. I had a report someone was getting much smaller SPL when building "beaglebone" compared to "am335x" Oct 13 01:58:16 Oh yes Oct 13 01:58:19 They're configured differently Oct 13 01:58:24 But ti shouldn't matter wrt runtime failure Oct 13 01:58:27 s/ti/it/ Oct 13 01:58:39 if you want all the changes then look here https://github.com/YoeDistro/meta-ti/commits/yoe/mut Oct 13 01:59:42 denix: bbb should probably use upstream u-boot since its a community board Oct 13 01:59:53 may be even kernel Oct 13 02:00:30 khem: use meta-yocto-bsp Oct 13 02:01:01 khem: that was specifically configured to be all upstream Oct 13 02:01:42 denix: I am interested to get it from meta-ti Oct 13 02:01:51 since there are more TI boards of interest Oct 13 02:02:11 yocto BSP is not of that interest to me Oct 13 02:03:04 OK now everything is working, let me for once switch the PROVIDER to check if it was some other config that fixed it Oct 13 02:03:09 khem: when I was setting up reference BSP for bbb in meta-yocto-bsp, we wanted all pure upstream, while meta-ti provides TI BSP with few additional pieces Oct 13 02:03:31 khem: beagleboard.org community uses TI kernel and u-boot, FYI Oct 13 02:03:39 I dont think yocto bsps should provide h/w support, Oct 13 02:03:48 it should be purely emulators Oct 13 02:03:54 but thats my opinion Oct 13 02:04:07 it just confuses people Oct 13 02:04:12 khem: pure emulation is oe-core, not meta-yocto-bsp Oct 13 02:05:15 yeah even that, moment people fork BSPs chaos ensues for new comers Oct 13 02:06:02 I wasnt happy with meta-beagleboard either Oct 13 02:11:06 denix: there are enough beaglebone variants now to have a meta-bone similar to meta-rpi Oct 13 02:11:56 sure, lets fragment it even further... Oct 13 02:12:26 khem: pretty much they are all covered by different dtbs, the rest is the same Oct 13 02:14:03 denix: yeah I see that Oct 13 02:15:45 denix: I can confirm that only when I comment the highlighted lines https://github.com/YoeDistro/meta-ti/blob/yoe/mut/conf/machine/beaglebone.conf#L33-L34 Oct 13 02:15:53 MLO breaks Oct 13 02:16:10 so definitely ti-staging-uboot needs to be looked at Oct 13 02:17:23 * khem goes back to reading animal farm Oct 13 02:30:46 khem: yes, understood, looks like same SPL issue someone else mentioned recently... Oct 13 02:33:54 khem: ah, just realized, since am335x-evm machine config can be booted on bbb, the simplest fix for now would be to change UBOOT_MACHINE :) Oct 13 02:34:24 khem: btw, am335x-evm is tested on a daily basis, so known to work Oct 13 02:35:00 khem: but I'll look into figuring out why beaglebone u-boot config breaks SPL **** ENDING LOGGING AT Sat Oct 13 02:59:59 2018