I recently committed to 9-CURRENT a fix for the sis(4) driver to improve performance when the host can't keep up with the processing requirements of the network. Previously, when an overflow occurred the chip was reset and a sleep of 100ms was introduced while programming a short cable fix for the NatSemi FA311 chip. Now, the reset is only done once when a media change occurs and the delay was reduced to 100us.
However there's some doubt as to whether the change may cause problems on SiS based controllers as found on some motherboards - they may require the chip to be reset when an overflow occurs. If anyone has a machine with a SiS based controller I'd welcome any testing that could be done to check that I've not broken overflow handling. Overflows are indicated by the Ierrs column in the output of "netstat -s". -- Bruce Cran _______________________________________________ [email protected] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "[email protected]"
