Re: [PATCH 1/2] can: Add Loongson CAN-FD controller support

[Vincent Mailhol <mailhol@kernel.org>]

On 21/05/2026 at 05:24, Binbin Zhou wrote:
> Hi Vincent:
> 
> On Thu, May 7, 2026 at 1:50 AM Vincent Mailhol <mailhol@kernel.org> wrote:

(...)

>> Use the CAN TDC framework to get the SSP value (c.f. struct can_tdc,
>> struct can_tdc_const and can_fd_tdc_is_enabled())
> 
> Last question:
> 
> In the general framework, the calculation condition for tdco is
> (dbt->brp == 1 || dbt->brp == 2), which does not seem to correspond to
> the current condition (dbt->bitrate > 1000000).

Where does the current condition comes from? Is it some empirical test?

FYI, the

  (dbt->brp == 1 || dbt->brp == 2)

check comes from ISO 11898.

> Although for CANFD, rates below 1 Mbit/s would negate the primary
> advantages of FD, from a controller hardware design perspective, I
> would still prefer to retain the condition (dbt->bitrate > 1000000).
> 
> Do you have any other suggestions?

Using TDC on low bitrates is indeed unstable and can lead to bus errors.
But the condition which you suggest

  (dbt->bitrate > 1000000)

is indirectly linked with the brp values. That's why the brp check is
sufficient (and more precise than the empirical 1 Mbit/s threshold value).

Are you able to trigger any bus errors when using TDC with a brp value
of 1 or 2 and a bitrate below 1 Mbit/sec? If no, there is no reason to
add this extra check and we can continue to rely on the framework test.


Yours sincerely,
Vincent Mailhol