I can tell you that it is not a TCP flow control issue. That is a much higher level activity; the driver doesn't know anything about TCP windowing. The issue is that the transmitter is getting stuck... and the ring is filling up... and when it has refused to accept more packets for a long enough time interval, the kernel tries to force the issue, and the driver complains. Can you post your kernel messages again? I don't have any older stuff from you so when you say things are "still" this way, that doesn't help me. -- Dave |
Messages