Re: [RFC PATCH v5 2/2] can: reject CAN FD content when disabled on CAN XL interfaces

[Vincent Mailhol <mailhol@kernel.org>]

On 10/09/2025 at 17:48, Oliver Hartkopp wrote:
> On 10.09.25 09:40, Vincent Mailhol wrote:
>> On 10/09/2025 at 16:27, Oliver Hartkopp wrote:
> 
>>>>      /* CAN XL is allowed on virtual interfaces if it fits the MTU */
>>>>         if (!priv)
>>>>             return dev->mtu == CANXL_MTU;
>>>
>>>          return can_is_canxl_dev_mtu(mtu);
>>>
>>> The MTU of CAN XL interfaces might vary.
>>
>> Maybe this is something that we discussed before, I do not remember, but how is
>> it that the MTU can vary?
>>
>> MTU is the *Maximum* Transmission Unit. I understand that the size of a CAN XL
>> frame is variable, but the MTU should be constant, right? Why can it vary?
> 
> Depending on the realtime requirements the length of the CAN frames (and
> therefore the time the bus is blocked) the MTU can be reduced. This is (like the
> bitrate settings) a network architects decision which is enforced by setting the
> MTU accordingly.

Is this an extension we offer in Socket CAN? The standard says nothing about
having the MTU configurable.

For CAN FD, we forcefully set the MTU in netlink.c

https://elixir.bootlin.com/linux/v6.16/source/drivers/net/can/dev/netlink.c#L228

I will have to think of what are the implication for CAN XL.


Yours sincerely,
Vincent Mailhol