On Tue, 7 Jun 2022 21:19:54 +0900 Vincent MAILHOL <mailhol.vincent@xxxxxxxxxx> wrote: > In his CAN327 driver, Max manages to bring the can0 device without the > need of dedicated user space daemon by using line discipline > (ldattach): > https://lore.kernel.org/linux-can/20220602213544.68273-1-max@xxxxxxxxx/ > > Isn't the same feasible with slcan so that we completely remove the > dependency toward slcand? > Max what do you think of this? I think it is a good idea to move this into the kernel driver. I don't have a slcan device, but a quick peek at its protocol suggests that it can be done much easier and cleaner than in can327. Fun fact: The use of a userspace "ignition" tool is a possible use case baked into can327's design ;) There is only one use case I can see for it though: Probing the ELM327's baud rate, and/or setting it before attaching the ldisc. This is the single thing that the kernel driver cannot do, since it is attached to an already running TTY, with a fixed speed. Everything else is configurable via "ip link". slcand could be used to auto-probe and set the baud rate as well. Then again, these stripped-down tools could likely be implemented as shell scripts calling stty and ldattach... Max
