ERROR:gbm_wrapper.cc(253) Failed to export buffer to dma_buf: No such file or directory

sdetweil

@dathbe said in ERROR:gbm_wrapper.cc(253) Failed to export buffer to dma_buf: No such file or directory:

export ELECTRON_DISABLE_GPU=1

the default now is disabled…

and the env var has changed to
export ELECTRON_ENABLE_GPU=1
to turn it on

dathbe

@sdetweil
No luck. with DISPLAY=:0 npm start, I get the long list of errors, but everything seems to work ok with the mirror. with

WAYLAND_DISPLAY=wayland-1   
   npm run start:wayland

It throws other errors and gets stuck in what seems to be a restart loop.

0|mm  | [3351:0415/211257.942494:ERROR:wayland_connection.cc(199)] Failed to connect to Wayland display: No such file or directory (2)
0|mm  | [3351:0415/211257.942729:ERROR:ozone_platform_wayland.cc(276)] Failed to initialize Wayland platform
0|mm  | [3351:0415/211257.942774:ERROR:env.cc(257)] The platform failed to initialize.  Exiting.
0|mm  | /home/mmuser/MagicMirror/node_modules/electron/dist/electron exited with signal SIGSEGV

sdetweil

@dathbe yeh, thats what I found too…
maybe its labwc display now?

from the labwc doc,
https://labwc.github.io/labwc-config.5.html

it sounds like DISPLAY is the correct variable

The autostart file is executed as a shell script after labwc has read its configuration and set variables defined in the environment file. Additionally, the environment variables WAYLAND_DISPLAY and (when labwc is built with Xwayland support) DISPLAY will be defined

dathbe

@sdetweil

I tried all three options in raspi-config > Advanced > Wayland (those being 1) X11, 2) Wayfire, and 3) Labwc), and was getting the same error on all of them. I went back to Labwc because my understanding is that is the current recommended.

I’ve tried many different combinations. Anything having to do with start:wayland seems to truly fail and get stuck in a restart loop. npm run start:wayland technically works, but throws about 20 ugly error messages on start.

Various options I’ve tried:

# Work with error messages 
DISPLAY=:0 npm start

export ELECTRON_DISABLE_GPU=1
DISPLAY=:0 npm start

export ELECTRON_ENABLE_GPU=1
DISPLAY=:0 npm start 

# Cause endless start loop

npm run start:wayland

export ELECTRON_DISABLE_GPU=1
npm run start:wayland

export ELECTRON_ENABLE_GPU=1
npm run start:wayland

WAYLAND_DISPLAY=wayland-1
npm run start:wayland

DISPLAY=wayland-1
npm run start:wayland

sdetweil

@dathbe restart loop is pm2, its job is to keep the app running

dathbe

@sdetweil Ok. When I have some time, I’m going to test an install with 32-bit PiOS.

sdetweil

@dathbe 32bit is x11 only . as i recall

dathbe

@sdetweil It appears to load labwc by default, but either way it didn’t do anything different – fails to start with start:wayland and multiple phantom errors with start. I guess I just have to close my eyes and not read the logs on startup. Though weird it’s doing it with my Pi 4B, but not my Pi 2B.

sdetweil

@dathbe yeh some config setup, i think something in /boot/config. txt

dathbe

@sdetweil In case for some reason it’s helpful, I’ve reviewed the contents of /boot/config.txt from my two devices (4B gets errors, 2B does not).

In 2B, not in 4B:

#dtoverlay=vc4-kms-v3d
# see https://forums.raspberrypi.com/viewtopic.php?p=1507622#p1507247
dtoverlay=vc4-fkms-v3d

In 4B, not in 2B:

dtoverlay=vc4-kms-v3d

# Run in 64-bit mode
arm_64bit=1

[cm5]
dtoverlay=dwc2,dr_mode=host