Oliver Hartkopp wrote:
> Hi Matthias,
> 
> Matthias Fuchs wrote:
>> I played a little with various ways to manipulate tx/rxqueuesizes with 
>> SocketCAN.
>> I ended up with some questions:
>>
>> 1) Does setsockopt(... ,SOL_SOCKET, SO_SNDBUF, ....) have any impact on 
>> CAN_RAW sockets?
>> My result: no!
> 
> Ack.
> 
>> 2) Does setsockopt(... ,SOL_SOCKET, SO_RCVBUF, ....) have any impact on 
>> CAN_RAW sockets?
>> Yes, that's also used by candump to prevent losing rx frames.
>>
>> 1+2 brings me to the conclusion that setsockopt(..., SO_RCVBUF, ...) 
>> configures receive
>> behavior and "ip link set can0 txqueuelen <whatever>" configures tx 
>> behavior. Did I miss 
>> anything?
> 
> No. That's the current behaviour.
> 
>> 3) Do we have a way to make write() on CAN_RAW sockets block when there is
>> no more buffer space? Typical behavior is returning ENOBUFS. I haven't tried 
>> to
>> use select() for sending. Does this work?
> 
> Yes. That should work.
> 
> We had a discussion on this recently.
> 
> http://old.nabble.com/-PATCH-0-2--add-support-for-Janz-MODULbus-devices-tp27662172p27715072.html
> 
> Due to that discussion, i added the poll support to the 'cangen' tool:
> 
> http://svn.berlios.de/wsvn/socketcan/?op=comp&compare[]=%2ftrunk%2fcan-ut...@1149&compare[]=%2ftrunk%2fcan-ut...@1146

I short, it does work but the poll/slect seems not to block as expected.


>> 4) Do we have a way to flush the txqueue and block until all messages are on 
>> the bus?
>> fsync() does not seem to be the right choice.
> 
> AFAIK there's no way from the userspace to flush the txqueue - only set the
> interface state to "down" ...
> 
>> 5) Is there any special reason why read/write can only handle single 
>> messages?
>> Is it just because nobody asked? It should be possible with our
>> fixed size stuct can_frame. It could even speed thing up.
> 
> This is an interesting idea. BUT the problem is that all the drivers and
> especially the echo_skb handling deals with single skb's. For a good reason.
> 
> But i could imagine to implement a multi-frame handling inside the CAN_RAW
> socket. This could be done with a loop that splits it into single skb's in
> raw_sendmsg() quite easily.
> 
> At least this would reduce the number of userspace->kernelspace copies.
> 
> But you would need to check the return code to check how many frames have been
> written (when the txqueue is getting full).
> 
> Interested?

It should be done via sendmsg/recvmsg, I think. But it needs to be
supported by the protocol. Here is an example:

http://lxr.linux.no/#linux+v2.6.33/net/ipv4/tcp.c#L929

Wolfgang.


_______________________________________________
Socketcan-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-users

Reply via email to