Re: slcanpty

[Oliver Hartkopp <socketcan@hartkopp.net>]

On 18.06.2013 10:24, j.uzycki@elproma.com.pl wrote:

>> BTW:
>> I've learnt about HW filters in SJA1000. I compared the datasheet
>> with acceptance filter explanation here
>> http://www.can232.com/?page_id=72
>> According to datasheet's figures filter output bit acceptance is
>> (AMRn | (input ^ ACCn)) and according to the tables in explanation
>> acceptance is (AMRn | !(input ^ ACCn)). Do you have experience in
>> practive about it?
>> I want to convert SJA1000 filter to SocetCAN. SJA1000 has several
>> difficult features: dual filter (problem with overlapped filters -
>> additional checking required), data filter for 1st and/or 2nd data
>> byte (BCM mode supports, RAW mode requires userland filtering but ok).
>> Finally I consider it is better to use socketCAN filters or implement
>> the whole well defined filter (not universal like socketcan filter) in
>> userland. What could be more CPU effective?
> 
> I've just implemented SJA1000 filter emulator. I didn't test it yet. What do
> you think about the idea?

Pretty funny.

Your referenced docs (and your patch) are very informative, but I just wonder
if the filters have really ever been used by anybody.

They look *very* special.

If there's a need to implement it, it's ok.
But IMHO adding the filters just for the completeness of an emulation can be
omitted when nobody is ever using it.

Best regards,
Oliver