I am trying to get access to a framebuffer using DRM/KMS ioctl requests. I can’t get it to work.
What works:
CREATE_DUMB/MAP_DUMB— buffer allocation works- Read-only DRM queries — connector, mode, resource enumeration all work
What doesn’t work:
DRM_IOCTL_SET_CLIENT_CAPreturnsENOTSUPfor bothDRM_CLIENT_CAP_UNIVERSAL_PLANESandDRM_CLIENT_CAP_ATOMIC, even after acquiring DRM master. YetDRM_IOCTL_GET_CAPreportsDRM_CAP_ATOMIC=1. This is contradictory — advertising atomic support but refusing to enable it.- Zero planes visible to any client regardless of master status, because
SET_CLIENT_CAP UNIVERSAL_PLANESfails. DRM_CAP_ATOMIC=1is misleading — the driver cannot actually perform atomic modesetting from userspace via standard KMS interfaces.
Fortunately i found out that the good old /dev/fb0 works, so i’ll focus on that for the time being.
Has there been any progress in DRM/KMS in the later jetpacks? Why is the Jetson system so non-standard?
Jetson Orin Nano
JetPack 6.1 / L4T R36.4.