11:57 <pgwipeout[m]> Heh, probably was fixed when I changed from pull_down to pull_none. Should have checked that again.

21:48 <digetx> pgwipeout[m]: the lm90 interrupt was broken on all tegra boards anyways, they all were getting the non-stopping interrupt because it should be edge-triggered

21:50 <digetx> that interrupt behaviour is documented in the sensor datasheet

22:08 <DavidHeidelberg> first test of CI for grate repo, not finished yet thou. https://github.com/okias/linux/actions

22:08 <DavidHeidelberg> digetx do you want it only against master branch or also setup another branch?

22:18 <digetx> master only

22:21 <DavidHeidelberg> digetx are u interested in building with multiple compilers matrix or just regular default gcc is enough?

22:22 <DavidHeidelberg> btw. I just realized I'll love one feature of it. And it's always hard to spot, when did you updated git repo from Web UI, since you rebase, so now it'll be nicely visible by pushed release

22:23 <DavidHeidelberg> https://github.com/okias/linux/releases (ignore stupid name linux- , it'll be linux-grate-next after fix)

22:23 <DavidHeidelberg> it gets overwritten everytime it's build

22:38 <DavidHeidelberg> sent PR, just waiting for final CI test finishes on my okias/master

22:45 <digetx> it could be good to have kernel compiled by older stable compiler and the most recent, for example thumb2 kernel build was broken using gcc10+ for a long time

22:55 <DavidHeidelberg> digetx so first options, one standard build, then THUMB2 mode? CONFIG_THUMB2_KERNEL ?

22:55 <DavidHeidelberg> or just give me a list of scenarions with CONFIG_ values and I'll make a build matrix

22:56 <DavidHeidelberg> just thinking not sure about compilers, because the script uses some debian/buntu crosscompilation toolchain, so not sure how clear would be implement that

22:57 <digetx> let's use one config for now

22:59 <DavidHeidelberg> digetx so the output of "current PR" https://github.com/okias/linux/actions/runs/1016664981

22:59 <digetx> it also should be a special config focused on optimizing kernel image size, shouldn't it? will you use the pmOS config?

22:59 <DavidHeidelberg> (dtbs will take longer I guess)

22:59 <DavidHeidelberg> for Nexus 7 2012, for sure at some point

23:00 <DavidHeidelberg> I could eventually create job for N7, curl config from pmOS pmaports and build it with it

23:00 <DavidHeidelberg> (which is very good idea, because it will make pmaports update a piece of cake)

23:01 <DavidHeidelberg> btw. tuxmake builds can be reproducible (if is used same command as saved in build done by CI)

23:06 <digetx> DavidHeidelberg: why is it checking non-tegra dtbs? it should happen using tegra_defconfig

23:07 <DavidHeidelberg> it's always check everything, which is kinda ... pain... but it's still searchable.. meh.. I know, not great

23:20 <DavidHeidelberg> I think it takes whole architecture, the KBUILD_DEFCONFIG is set to "tegra_defconfig"

23:29 <DavidHeidelberg> well, still the kernel gets generated in 22 minutes, dts will take approx more than hour, but I didn't noticed that parallel jobs has any effect on speed of other jobs..

23:30 <DavidHeidelberg> I think limit is 20 concurrent job, so we're at 3/20 :)

23:34 <digetx> the non-tegra dtbs are not interesting, perhaps better to skip dtbs_check then?

23:35 <digetx> it actually checks only tegra dtbs for me locally

23:35 <DavidHeidelberg> digetx that's weird, what you doing differently?

23:35 <DavidHeidelberg> status=0

23:35 <DavidHeidelberg> make -j3 dtbs || status=1

23:35 <DavidHeidelberg> make -j3 dtbs_check || status=1

23:37 <digetx> make defconfig

23:37 <digetx> looks suspicious, shouldn't it be make $KBUILD_DEFCONFIG?

23:37 <DavidHeidelberg> nwm, oh I see "2021-07-09T22:38:15.1143268Z *** Default configuration is based on 'multi_v7_defconfig'"

23:38 <DavidHeidelberg> digetx well, it should use exported KBUILD_DEFCONFIG

23:38 <DavidHeidelberg> but it doesn't use it in this scenario...

23:39 <digetx> it doesn't use KBUILD_DEFCONFIG

23:41 <DavidHeidelberg> oh... btw. what about dt_bindings_check, I assume there it doesn't matter

23:41 <DavidHeidelberg> or should it also get make $KBUILD_DEFCONFIG ?

23:45 <digetx> dt_bindings_check shouldn't be needed

23:45 <DavidHeidelberg> ok, seems to be fixed

23:45 <DavidHeidelberg> pushed changes to ci for merge

23:50 <DavidHeidelberg> I'm little bit playing in my head with idea like -> having one default tegra_defconfig and tegra_defconfig + all debug stuff + some initrd, which gets bundled, kernel gets uploaded to device and it'll do like "self testing" of everything which can be tested by automatization

23:51 <DavidHeidelberg> digetx cool, dtbs done in 7 minutes, much better than 1h+

23:55 <digetx> DavidHeidelberg: is PR ready now?

23:55 <DavidHeidelberg> yup, kernel build is without change and DTS just compiled

23:55 <DavidHeidelberg> so, it's ready to roll :)

23:56 <DavidHeidelberg> digetx cool, thx for merging, first build rolling: https://github.com/grate-driver/linux/actions/runs/1016805424 :)

23:59 <digetx> nice