On Oct 4, 2005, at 7:00 AM, Ferdinand Goldmann wrote:
Ok, to followup on this issue:
Today I installed the driver which I had downloaded from the Intel
website (em-3.2.15.tar.gz). This driver does not solve the issue
with Ierrs rising rapidly when polling is enabled (I tried HZ=1000
and burst_max 300 and 600).
Neither throughput nor interrupt load were any better than with the
old driver; in fact I'd say they were worse. :-( So we ended up
rebooting with the old kernel.
This is pretty bad, because we did assume that Intel cards are
among the ones well supported by FreeBSD.
Can anyone give recommendations on copper Gbit cards that work
_well_ with recent FreeBSD 5-STABLE?
This is pretty odd. We've got dozens of servers using various
versions of 5.x, and many different em cards, and have no problem,
even when shoving near line rate speeds out of them.
Add in cards like:
em0: <Intel(R) PRO/1000 Network Connection, Version - 1.7.35> port
0xccc0-0xccff mem 0xfcd20000-0xfcd3ffff irq 24 at device 6.0 on pci2
and built in to newer Dell servers:
em0: <Intel(R) PRO/1000 Network Connection, Version - 1.7.35> port
0xecc0-0xecff mem 0xdfae0000-0xdfafffff irq 64 at device 7.0 on pci6
After you experience your problems, can you do "sysctl -w
hw.em0.stats=1" and "sysctl -w hw.em0.debug_info=1" and post what
gets dumped to your syslog/dmesg output?
We're using polling on nearly all the servers, and don't see ierrs at
all. Have you tried contacting Intel directly about this?
[EMAIL PROTECTED] has been pretty helpful with em specific
problems in the past.
_______________________________________________
freebsd-net@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
