[BPI-R2] Kernel Development

frank-w · January 17, 2020, 5:10pm

is @DeadMeat s 4.16-lima working? imho no public repo had working lima…but you can test it and report if it’s working and compare to 5.4-lima

eros · January 17, 2020, 6:37pm

It is not working and wlan&BT stuff need to disable because compile error. Lima dosn’t detect, same error than 5.3-lima:

[    1.238006] lima 13040000.gpu: get core clk failed -517
[    1.243301] lima 13040000.gpu: clk init fail -517
[    1.247977] lima 13040000.gpu: Fatal error during GPU init
[    3.415826] lima 13040000.gpu: get core clk failed -517
[    3.421189] lima 13040000.gpu: clk init fail -517
[    3.425974] lima 13040000.gpu: Fatal error during GPU init
[   12.891693] lima 13040000.gpu: get core clk failed -517
[   12.891708] lima 13040000.gpu: clk init fail -517
[   12.891715] lima 13040000.gpu: Fatal error during GPU init
[   12.907644] lima 13040000.gpu: get core clk failed -517
[   12.907658] lima 13040000.gpu: clk init fail -517
[   12.907665] lima 13040000.gpu: Fatal error during GPU init
[   12.975271] lima 13040000.gpu: get core clk failed -517
[   12.975288] lima 13040000.gpu: clk init fail -517
[   12.975294] lima 13040000.gpu: Fatal error during GPU init
[   13.004238] lima 13040000.gpu: get core clk failed -517
[   13.004255] lima 13040000.gpu: clk init fail -517

Were it try to get gpu clock, from dts??

It should give lines like:

[    8.492918] lima 13040000.gpu: bus rate = 500500000
[    8.492930] lima 13040000.gpu: mod rate = 500500000

frank-w · January 17, 2020, 6:43pm

please make separate lima-thread (maybe any exists already).

where did you get the lines? as i said, there is no working lima-kernel, so you can start with 5.4-main and my above patches and start debugging there…clocks and other setting are selected by dts (read and assigned by driver)

eros · January 17, 2020, 7:25pm

These lima lines are from Alex first post Mali-450 support by lima

So need to dig dts and put somewhere 50050000 and 500500000. They are maybe hex or converted decimal by driver? Maybe in mt7623n-bananapi-bpi-r2.dts? or mt7623.dtsi?

frank-w · January 17, 2020, 7:54pm

i guess there are hex…and imho it’s better to continue this thread and asking alex…

DeadMeat · January 18, 2020, 6:20am

No, pls, see Mali-450 support by lima thread

frank-w · January 18, 2020, 8:06am

Which compile-errors and which compiler/version? Imho it should compile till gcc 8 without errors

eros · January 18, 2020, 12:27pm

BPI-R2-4.14-4.20-lima

drivers/gpu/drm/lima/lima_drv.c: At top level: ./include/drm/drm_ioctl.h:159:16: error: ‘DRM_IOCTL_LIMA_GEM_MOD’ undeclared here (not in a function); did you mean ‘DRM_IOCTL_LIMA_GEM_VA’? [DRM_IOCTL_NR(DRM_IOCTL_##ioctl) - DRM_COMMAND_BASE] = { \

And gcc 7.3 CONFIG_COMMON_CLK_MT2701_G3DSYS=y CONFIG_DRM_LIMA=y

frank-w · January 18, 2020, 1:30pm

i meant why you need to disable wifi/bt…in my tree there is no working lima…only tests…so just start with 5.4-main, and further discuss in lima-thread

xpk · January 20, 2020, 10:24am

Lost more more time, of try to compile kernel from sources under MacOS Catalina… Maybe anybody need for notes how to do it: In my MacOS, by default system volume AFPS with case insensitive (WTF???), when receive from git or unzip, I received answer of need to rewrite some files: xt_connmark.h xt_dscp.h xt_mark.h xt_rateest.h xt_tcpmss.h ipt_ecn.h ipt_ttl.h ip6t_hl.h xt_dscp.c xt_hl.c xt_rateest.c xt_tcpmss.c but agree to rewrite. Them I don`t understand why received errors while try to compile kernel… WHAT TO DO: create virtual volume(dmg file) 5Gb size with CASE SENSITIVE from Disk Utility, and work with kernel sources in this Volume. Start compile from docker container under MacOS: Mount your created volume(dmg file) and run:

docker build /Volumes/$NAME_YOUR_VOLUME/$KERNEL_SOURCES/utils/docker --tag bpi-cross-compile:1
mkdir /Volumes/$NAME_YOUR_VOLUME/$KERNEL_SOURCES/SD
docker run -it -v “/Volumes/$NAME_YOUR_VOLUME/$KERNEL_SOURCES/SD”:/SD -v “/Volumes/$NAME_YOUR_VOLUME/$KERNEL_SOURCES”:/srv --tag bpi-cross-compile:1 /bin/bash
And in docker run: ./build.sh importconfig || ./build.sh

frank-w · February 11, 2020, 7:08am

i started work on 5.6-rc1 have wifi working, hdmi works too, but there is a warning in drm_atomic_helper_async_commit

github.com

frank-w/BPI-R2-4.14/blob/3285e1bccfd82748a9d55cb4f123a53e3df5a086/drivers/gpu/drm/drm_atomic_helper.c#L1735


		struct drm_framebuffer *old_fb = plane->state->fb;


		funcs = plane->helper_private;
		funcs->atomic_async_update(plane, plane_state);


		/*
		 * ->atomic_async_update() is supposed to update the
		 * plane->state in-place, make sure at least common
		 * properties have been properly updated.
		 */
		WARN_ON_ONCE(plane->state->fb != new_fb);
		WARN_ON_ONCE(plane->state->crtc_x != plane_state->crtc_x);
		WARN_ON_ONCE(plane->state->crtc_y != plane_state->crtc_y);
		WARN_ON_ONCE(plane->state->src_x != plane_state->src_x);
		WARN_ON_ONCE(plane->state->src_y != plane_state->src_y);


		/*
		 * Make sure the FBs have been swapped so that cleanups in the
		 * new_state performs a cleanup in the old FB.
		 */
		WARN_ON_ONCE(plane_state->fb != old_fb);

and line 1745 (bootlog)

currently i have no idea how to fix it (without removing the WARN_ON-lines)

i noticed a usb-disconnection-problem while testing wifi…maybe some more users can help testing 5.6-rc/5.6-hdmi/5.6-wifi branch

DeadMeat · February 11, 2020, 9:10pm

Is there a merged tree available?

Correct me if i wrong:

5.6-rc - general patches only (dts, defconfig,mmc and serial ports order)
5.6-hdmi - 5.6-rc +hdmi patches
5.6-wifi - 5.6-rc + wifi

frank-w · February 12, 2020, 5:47am

You are right,there is no merged tree yet,because i want to check for regressions first on features alone.

On wifi-branch i experienced usb disconnects. they happen on my display (touchscreen hid-device), but it looks like this happens always if no hdmi is available (display switches also off and on again when it happens…also while hdmi connected and no touchscreen). watchdog-patch on hdmi does not break anything…seems like i should create a merged tree

On hdmi i got some warnings. Got a patch from mtk which fixes this warning, pushed it. there is also a strange cursor-display (on 5.5-rc5 there is no issue, so it should be only on 5.6)

maybe

920fffcc891276a855cb3ce1e7361d2e9cb72581 “drm/mediatek: update cursors by using async atomic update” (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=920fffcc891276a855cb3ce1e7361d2e9cb72581)

can’t be reverted so easily found 2 commits on top of https://github.com/ckhu-mediatek/linux.git-tags/commits/mediatek-drm-fixes-5.6 cursor-related…try to add them and compare cursor behaviour => no success

created merged-tree and no more usb-disconnects…seems like i had not tried older kernels with display connected without hdmi working

frank-w · February 23, 2020, 10:51am

updated 5.4-main to .21 and included patches from @Ilya_Murav_jov for building debian-packages.

https://github.com/frank-w/BPI-R2-4.14/releases/tag/CI-BUILD-20200223_103817-0b9b9e3c9

wondering why binary deb for r2 is only ~20MB instead of ~49MB (my deb 1 release before)…i hope all modules are packed.

maybe anyone using debs can test it? uenv.txt is not changed in these debs…needs to be done manually

DeadMeat · March 3, 2020, 9:28pm

Just wondering… How about to make a 5.4-main branch default in your tree?

Ilya_Murav_jov · March 5, 2020, 1:35pm

It is relevant for BPI-R64 too.

frank-w · March 5, 2020, 4:44pm

Have updated it…i hope most things working on 5.4-main on both devices

frank-w · March 26, 2020, 2:46pm

i merged 5.6-rc7 and noticed that some timeval/ktime-functions/macros are removed because of year2038-issues

i removed additions by me from timekeeping32.h (which are using the removed functions)

struct timeval => timespec64
timespec64 has no tv_usec => tv_nsec/1000
do_gettimeofday => ktime_get_real_ts64
get_monotonic_boottime => ktime_get_boottime_ts64

maybe this can be done in more effective way (especially the usec-calculation), but as first start it compiles

anybody can help me testing?

i did a quick test and wifi seems to work…also noticed that trgmii-mode is broken on 5.6…changed to rgmii

pushed some more commits, but it seems they do not fix the problem…so if you have issues with ethernet while testing just go back to this commit:

05f0f82aeda4 arm: dts: trgmii-mode broken, so move to rgmii

btw. with the author of musb-driver i got otg finally working on r2 (5.4). only one issue exists…if device stays connected to otg-port on poweroff, r2 prints WARNING and do not poweroff. if this one is solved, i merge changes to 5.4-main. if anyone will try it, i have cleaned up to have only 2 commits (dts+defconfig):

frank-w · April 6, 2020, 10:33am

john posted hwnat-patch for openwrt

https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/mediatek/patches-5.4/0999-hnat.patch;h=ad04e355e62ade948c701b8ff9f7165f4ce347b3;hb=8021652478db50f5739bc2d8d8425cabc6e5d957

cioby23 · April 6, 2020, 11:25am

Yes that good to hear. I saw this was pushed to OpenWRT trunk.

Also 4.14 support in Mediatek was dropped and they made 5.4 kernel the default.