kcxtFeb 21
probably a sign i should sleep when i start writing comments like this 😵‍💫
Show threadkcxtFeb 21

crazy thing is if you want to have a .remove callback that's called before we jump to the OS and it's for a device which is a parent of the vidconsole device then u-boot just free's data and everything breaks

but .remove is the correct place to like, disable the DMA fetching of your display controller so Linux doesn't shit the bed

nobody doing this??

Show threadVal Packett 🧉
@cas woah wait is this related to the vidconsole weirdness i mentioned in the chat?
Feb 21 at 4:49amWeb
Show threadkcxtFeb 21

@valpackett not sure i saw that? vidconsole (and USB serial gadget for that matter) have a whole host of issues that either need much more careful state management to lazily de-register them if they break, and most likely a try/catch mechanism to handle all the edge cases

like if you have any logging in the dwc3 core and have usbacm enabled then u-boot immediately deadlocks 🙃

Show threadVal Packett 🧉Feb 21
@cas you responded (mostly with just "hmm weird") in u-boot-qcom re: "vidconsole does not actually show anything until setenv stdout" a few days ago :)