search

JeffyCN / meta-rockchip

Yocto BSP layer for the Rockchip SOC boards
Other
120 stars 93 forks source link

weston cannot resume after idle #17

Closed SquallATF closed 2 years ago

SquallATF commented 2 years ago

After weston resume from the idle black screen, only the background image is displayed, and the newly launched application is not displayed. but if set idle-time to 0, it seems to work fine all the time.

JeffyCN commented 2 years ago

couldn't repro, it works well on my side(3588)

JeffyCN commented 2 years ago

the weston supports using drm atomic api and legacy api(default is atomic), so maybe try legacy api: export WESTON_DISABLE_ATOMIC=1

and maybe also check the kernel's dmesg for error logs and check /sys/kernel/debug/dri/0/summary for drm driver status

JeffyCN commented 2 years ago

any luck? i found there might be a problem for 356x and 3588 when using atomic mode with kernel drm fbdev emulation enabled: https://github.com/JeffyCN/meta-rockchip/commit/bd9b31e11b4f508464d124c88a8c789e3ca296d8

SquallATF commented 2 years ago

any luck? i found there might be a problem for 356x and 3588 when using atomic mode with kernel drm fbdev emulation enabled: bd9b31e

Unlucky export WESTON_DISABLE_ATOMIC=1 didn't help to solve my problem, and new patch apply failed. my poky version is https://git.yoctoproject.org/poky/tree/meta/recipes-graphics/wayland?h=kirkstone-4.0.1

ERROR: weston-10.0.0-r0 do_patch: Applying patch '0005-backend-drm-Bind-Nth-primary-plane-to-Nth-crtc.patch' on target directory '/home/squallatf/yocto/build/tmp/work/armv8a-poky-linux/weston/10.0.0-r0/weston-10.0.0'
CmdError('quilt --quiltrc /home/squallatf/yocto/build/tmp/work/armv8a-poky-linux/weston/10.0.0-r0/recipe-sysroot-native/etc/quiltrc push', 0, 'stdout: Applying patch 0005-backend-drm-Bind-Nth-primary-plane-to-Nth-crtc.patch
patching file libweston/backend-drm/drm.c
Hunk #1 FAILED at 842.
1 out of 1 hunk FAILED -- rejects in file libweston/backend-drm/drm.c
Patch 0005-backend-drm-Bind-Nth-primary-plane-to-Nth-crtc.patch does not apply (enforce with -f)

stderr: ')
ERROR: Logfile of failure stored in: /home/squallatf/yocto/build/tmp/work/armv8a-poky-linux/weston/10.0.0-r0/temp/log.do_patch.55784
ERROR: Task (/home/squallatf/yocto/build/../poky/meta/recipes-graphics/wayland/weston_10.0.0.bb:do_patch) failed with exit code '1'

there have bouth 0005-backend-drm-Bind-Nth-primary-plane-to-Nth-CRTC.patch and 0005-backend-drm-Bind-Nth-primary-plane-to-Nth-crtc.patch, should remove 0005-backend-drm-Bind-Nth-primary-plane-to-Nth-crtc.patch

after remove 0005-backend-drm-Bind-Nth-primary-plane-to-Nth-crtc.patch weston build successful, but still unlucky. If move the mouse for a short time (about 1-2 seconds) after the black screen, it can resume normally, but it will not work for a long time.

JeffyCN commented 2 years ago

patches re-uploaded.

did the app crashed? or just not shown?

SquallATF commented 2 years ago

app will not shown and not crash, run glmark2-es2-wayland will show score in ssh terminal. and If restart weston, nothing will be displayed on screen.

dmesg 

[  813.641099] dw-mipi-dsi fe060000.dsi: [drm:dw_mipi_dsi_encoder_enable] final DSI-Link bandwidth: 420 x 4 Mbps
[  813.749044] rockchip-vop2 fe040000.vop: [drm:vop2_crtc_atomic_disable] Crtc atomic disable vp1
[  813.829554] rockchip-vop2 fe040000.vop: [drm:vop2_crtc_atomic_enable] Update mode to 800x1280p56, type: 16 for vp1
[  813.829964] dw-mipi-dsi fe060000.dsi: [drm:dw_mipi_dsi_encoder_enable] final DSI-Link bandwidth: 420 x 4 Mbps
JeffyCN commented 2 years ago

nothing displayed or just background image? maybe try to check screenshot: weston --debug --idle-time=2& then: weston-screenshooter

and if you are using kernel 4.19 or 5.10, it might be old(i've been warned not to sync the newest kernel versions), and might cause issues like blank screen

SquallATF commented 2 years ago

nothing displayed, and weston-screenshooter not return

my kernel based on 

commit 5bebb638902ab6ebe6a73a6628adb703a819e077
Author: Chen Shunqing <csq@rock-chips.com>
Date:   Thu Apr 14 09:04:15 2022 +0000
JeffyCN commented 2 years ago

it looks like weston hang somewhere...maybe you can try to use gdb to check it, or "echo t > /proc/sysrq-trigger" to dump the stacks in kernel's dmesg

SquallATF commented 2 years ago

gdb dump

Using host libthread_db library "/lib/libthread_db.so.1".
0x0000007f9f647644 in epoll_pwait () from /lib/libc.so.6
(gdb) info thread
  Id   Target Id                                       Frame
* 1    Thread 0x7f9f7a7460 (LWP 309) "weston"          0x0000007f9f647644 in epoll_pwait () from /lib/libc.so.6
  2    Thread 0x7f9b898f40 (LWP 311) "mali-mem-purge"  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
  3    Thread 0x7f9b088f40 (LWP 312) "mali-utility-wo" 0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
  4    Thread 0x7f9a878f40 (LWP 313) "mali-utility-wo" 0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
  5    Thread 0x7f9a068f40 (LWP 314) "mali-utility-wo" 0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
  6    Thread 0x7f99858f40 (LWP 315) "mali-utility-wo" 0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
  7    Thread 0x7f99048f40 (LWP 316) "mali-cmar-backe" 0x0000007f9f63da20 in poll () from /lib/libc.so.6
  8    Thread 0x7f986d2f40 (LWP 317) "weston"          0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f647644 in epoll_pwait () from /lib/libc.so.6
#1  0x0000007f9c79d82c in wl_event_loop_dispatch () from /usr/lib/libwayland-server.so.0
#2  0x0000007f9c79ae98 in wl_display_run () from /usr/lib/libwayland-server.so.0
#3  0x0000007f9f72bf44 in wet_main (argc=<optimized out>, argv=0x7fe2dc1b98, test_data=0x0) at ../weston-10.0.0/compositor/main.c:3615
#4  0x0000007f9f58b230 in ?? () from /lib/libc.so.6
#5  0x0000007f9f58b30c in __libc_start_main () from /lib/libc.so.6
#6  0x000000556c1d7770 in ?? ()
(gdb) thread 2
[Switching to thread 2 (Thread 0x7f9b898f40 (LWP 311))]
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
#1  0x0000007f9f5e8310 in ?? () from /lib/libc.so.6
#2  0x0000007f9d018c10 in gles_vertexp_bb_neon_transform_and_produce_clip_bits () from /usr/lib/libmali.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) thread 3
[Switching to thread 3 (Thread 0x7f9b088f40 (LWP 312))]
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
#1  0x0000007f9f5e8310 in ?? () from /lib/libc.so.6
#2  0x0000007f9cffab0c in gles_vertexp_bb_neon_transform_and_produce_clip_bits () from /usr/lib/libmali.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) thread 4
[Switching to thread 4 (Thread 0x7f9a878f40 (LWP 313))]
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
#1  0x0000007f9f5e8310 in ?? () from /lib/libc.so.6
#2  0x0000007f9cffab0c in gles_vertexp_bb_neon_transform_and_produce_clip_bits () from /usr/lib/libmali.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) thread 5
[Switching to thread 5 (Thread 0x7f9a068f40 (LWP 314))]
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
#1  0x0000007f9f5e8310 in ?? () from /lib/libc.so.6
#2  0x0000007f9cffab0c in gles_vertexp_bb_neon_transform_and_produce_clip_bits () from /usr/lib/libmali.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) thread 6
[Switching to thread 6 (Thread 0x7f99858f40 (LWP 315))]
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
#1  0x0000007f9f5e8310 in ?? () from /lib/libc.so.6
#2  0x0000007f9cffab0c in gles_vertexp_bb_neon_transform_and_produce_clip_bits () from /usr/lib/libmali.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) thread 7
[Switching to thread 7 (Thread 0x7f99048f40 (LWP 316))]
#0  0x0000007f9f63da20 in poll () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f63da20 in poll () from /lib/libc.so.6
#1  0x0000007f9cffad30 in gles_vertexp_bb_neon_transform_and_produce_clip_bits () from /usr/lib/libmali.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) thread 8
[Switching to thread 8 (Thread 0x7f986d2f40 (LWP 317))]
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
(gdb) bt
#0  0x0000007f9f5dc74c in ?? () from /lib/libc.so.6
#1  0x0000007f9f5df308 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000007f9d046af0 in ?? () from /usr/lib/libmali.so.1
#3  0x0000007fe2dbecb8 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb)

kernel dump stack.txt

JeffyCN commented 2 years ago

could be something wrong in the mali. maybe try to run without gpu:

WAYLAND_DEBUG=1 weston --use-pixman --idle-time=2

SquallATF commented 2 years ago

Same result without gpu. first time.log at [4264897.792] wl_callback@24.done(43764) screen black and [4277943.126] -> wl_seat@11.capabilities(3) mouse move

second time.log this is stop weston and restart log, screen always black.

JeffyCN commented 2 years ago

those logs look fine to me. so the weston seems fine(not sure why screenshot hang).

maybe something wrong in the DSI driver or hardware? could you try to use HDMI?

root@RK3588:/# echo off > /sys/class/drm/card0-DSI-1/status # turn off DSI and you can dump drm driver status here: root@RK3588:/# cat /sys/kernel/debug/dri/0/summary root@RK3588:/# cat /sys/kernel/debug/dri/0/state

SquallATF commented 2 years ago

If DSI is turned off, the screen connected to hdmi always black. /sys/kernel/debug/dri/0/summary

Video Port0: ACTIVE
    Connector: HDMI-A-1
        bus_format[100a]: RGB888_1X24
        overlay_mode[0] output_mode[f] color_space[0]
    Display mode: 1920x1080p60
        clk[148500] real_clk[148500] type[48] flag[5]
        H: 1920 2008 2052 2200
        V: 1080 1084 1089 1125
    Smart1-win0: ACTIVE
        win_id: 1
        format: XR24 little-endian (0x34325258) SDR[0] color_space[0] glb_alpha[0xff]
        rotate: xmirror: 0 ymirror: 0 rotate_90: 0 rotate_270: 0
        csc: y2r[0] r2y[0] csc mode[0]
        zpos: 1
        src: pos[0, 0] rect[1920 x 1080]
        dst: pos[0, 0] rect[1920 x 1080]
        buf[0]: addr: 0x0000000001149000 pitch: 7680 offset: 0
Video Port1: ACTIVE
    Connector: DSI-1
        bus_format[100a]: RGB888_1X24
        overlay_mode[0] output_mode[0] color_space[0]
    Display mode: 800x1280p56
        clk[68000] real_clk[68000] type[48] flag[a]
        H: 800 872 892 912
        V: 1280 1310 1314 1326
    Smart0-win0: ACTIVE
        win_id: 0
        format: RG16 little-endian (0x36314752) SDR[0] color_space[0] glb_alpha[0xff]
        rotate: xmirror: 0 ymirror: 0 rotate_90: 0 rotate_270: 0
        csc: y2r[0] r2y[0] csc mode[0]
        zpos: 0
        src: pos[0, 0] rect[654 x 270]
        dst: pos[73, 505] rect[654 x 270]
        buf[0]: addr: 0x000000007df00000 pitch: 1308 offset: 397312

/sys/kernel/debug/dri/0/state

plane[57]: Smart1-win0
        crtc=video_port0
        fb=167
                allocated by = weston
                refcount=2
                format=XR24 little-endian (0x34325258)
                modifier=0x0
                size=1920x1080
                layers:
                        size[0]=1920x1080
                        pitch[0]=7680
                        offset[0]=0
                        obj[0]:(null)
        crtc-pos=1920x1080+0+0
        src-pos=1920.000000x1080.000000+0.000000+0.000000
        rotation=1
        normalized-zpos=0
        color-encoding=ITU-R BT.601 YCbCr
        color-range=YCbCr limited range
plane[73]: Smart0-win0
        crtc=video_port1
        fb=164
                allocated by = kworker/2:2
                refcount=1
                format=RG16 little-endian (0x36314752)
                modifier=0x0
                size=654x270
                layers:
                        size[0]=654x270
                        pitch[0]=1308
                        offset[0]=397312
                        obj[0]:(null)
        crtc-pos=654x270+73+505
        src-pos=654.000000x270.000000+0.000000+0.000000
        rotation=1
        normalized-zpos=0
        color-encoding=ITU-R BT.601 YCbCr
        color-range=YCbCr limited range
plane[89]: Esmart1-win0
        crtc=(null)
        fb=0
        crtc-pos=0x0+0+0
        src-pos=0.000000x0.000000+0.000000+0.000000
        rotation=1
        normalized-zpos=0
        color-encoding=ITU-R BT.601 YCbCr
        color-range=YCbCr limited range
plane[103]: Esmart0-win0
        crtc=(null)
        fb=0
        crtc-pos=0x0+0+0
        src-pos=0.000000x0.000000+0.000000+0.000000
        rotation=1
        normalized-zpos=0
        color-encoding=ITU-R BT.601 YCbCr
        color-range=YCbCr limited range
plane[117]: Cluster0-win0
        crtc=(null)
        fb=0
        crtc-pos=0x0+0+0
        src-pos=0.000000x0.000000+0.000000+0.000000
        rotation=1
        normalized-zpos=0
        color-encoding=ITU-R BT.601 YCbCr
        color-range=YCbCr limited range
plane[131]: Cluster1-win0
        crtc=(null)
        fb=0
        crtc-pos=0x0+0+0
        src-pos=0.000000x0.000000+0.000000+0.000000
        rotation=1
        normalized-zpos=0
        color-encoding=ITU-R BT.601 YCbCr
        color-range=YCbCr limited range
crtc[71]: video_port0
        enable=1
        active=1
        planes_changed=1
        mode_changed=0
        active_changed=0
        connectors_changed=0
        color_mgmt_changed=0
        plane_mask=1
        connector_mask=2
        encoder_mask=2
        mode: 0:"1920x1080" 60 148500 1920 2008 2052 2200 1080 1084 1089 1125 0x48 0x5
crtc[87]: video_port1
        enable=1
        active=1
        planes_changed=1
        mode_changed=0
        active_changed=0
        connectors_changed=0
        color_mgmt_changed=0
        plane_mask=2
        connector_mask=4
        encoder_mask=4
        mode: 0:"800x1280" 56 68000 800 872 892 912 1280 1310 1314 1326 0x48 0xa
connector[150]: Writeback-1
        crtc=(null)
connector[152]: HDMI-A-1
        crtc=video_port0
connector[163]: DSI-1
        crtc=video_port1

If connected bouth HDMI and DSI, the HDMI display is incorrect image wayland-screenshot before idle wayland-screenshot-2022-03-17_14-44-14 wayland-screenshot after resume from idle (when hdmi connected, DSI screen keep black,but HDMI screen will resume like before.) wayland-screenshot-2022-03-17_14-44-49

when connected HDMI, stop weston and restart, some time work but most time not work.

JeffyCN commented 2 years ago

so it looks like a driver or hardware issue. (the drm driver state looks well)

maybe: 1/ try to disable DSI in the kernel's dts 2/ try to disable kernel's logo(in dts) 3/ try to confirm the drm's displaying content is correct: kmsgrab.zip root@RK3588:/# /kmsgrab Usage: /kmsgrab [--crtc|--connector|--plane|--fb] OBJ_ID Valid crtc: 85 119 root@RK3588:/# /kmsgrab --crtc 85 > /tmp/85.rgb32 fb=0x17545310 id=253 fd=4 handle=1 size=1920x1080(7680) bpp=32(24) first pixel: ff292726

JeffyCN commented 2 years ago

after disabling DSI and kernel logo(delete logo related nodes in the dts), the /sys/kernel/debug/dri/0/summary should be clean(only HDMI actived) if the dumped HDMI content is correct too, but the display is still abnormal, then it's a hardware/driver issue

JeffyCN commented 2 years ago

and try to disable the kernel's CONFIG_DRM_FBDEV_EMULATION too

JeffyCN commented 2 years ago

FYI, i did find some issues when testing multi-head, but might not related...patches uploaded just now.

SquallATF commented 2 years ago

running weston-screenshooter does not hang with the latest version, but screen still black after idle. I think maybe there is some wrong with the screen initialization, because if run echo 1 > /sys/class/graphics/fb0/blank sleep screen before run weston has the same effect.

JeffyCN commented 2 years ago

that's why i think this is a hardware/driver issue. a few tests to confirm would be: 1/ disable fbdev emulation, to get a pure drm enviroment 2/ disable kernel logo, this logic is rockchip's custom hack, might cause lots of unexpected behaviors 3/ disable DSI to rule out DSI hardware/driver issue 4/ dump HDMI screen content and drm summary/state to confirm that userspace(weston) is doing nothing wrong

anyway, everything works well on my side(5.10 kernel+3588 evb)

SquallATF commented 2 years ago

Thank you for your suggestion. In the next period of time, I will transfer the development board to other colleagues for application development. I cannot debug this problem for the time being, so I will close this issue first, and I will open it later if necessary.

JeffyCN commented 2 years ago

ok.

and if you can access to rockchip's redmineissue system, i would sugguest to report this issue there...because i don't know much about dsi driver or hardware