Folks,
The following bit of code in net/sctp/output.c
sctp_packet_append_data
/* Update our view of the receiver's rwnd. Include sk_buff overhead
* while updating peer.rwnd so that it reduces the chances of a
* receiver running out of receive buffer space even when receive
* window is still open. This can happen when a sender is sending
* sending small messages.
*/
datasize += sizeof(struct sk_buff);
if (datasize < rwnd)
rwnd -= datasize;
else
rwnd = 0;
adds a whole sk_buff overhead for each data chunk in a packet. On
x86_64 sk_buff is 232 bytes, on an interface with 9K MTU sender which
streams 12 byte packets to a receiver with "default" 64K window is
going to run ouf of "space" before it can fill a single MTU-worth of
bytes. It seems (to me) the overhead calculation is too
aggressive. Does it really assume that each data chunk will get its
own sk_buff on the receive side?
max
--
To unsubscribe from this list: send the line "unsubscribe linux-sctp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
