On 3/5/11 1:49 PM, Mark Kettenis wrote:
Date: Fri, 04 Mar 2011 07:40:59 -0500
From: Daniel Ouellet<[email protected]>
I get pretty much a symetric traffic of ~43.5Mbps each direction using
tcpbench as before.
Then after may be one minute, both side will cut off oppose to before
when only one side was doing that. It will then stay off for a few
seconds and come back up on both side at the same time, just like ti
went down at the same time and EACH time this happen I will get a
watchdog timeout.
I will get it on either network port like this:
# dc0: watchdog timeout
dc0: watchdog timeout
dc1: watchdog timeout
dc0: watchdog timeout
dc0: watchdog timeout
dc0: watchdog timeout
dc0: watchdog timeout
dc0: watchdog timeout
dc0: watchdog timeout
dc1: watchdog timeout
dc0: watchdog timeout
Yes. The chip still craps out. My diff doesn't address that. But at
least the machine won't panic anymore, and the network interface will
recover after the timeout. It's committed now.
I'll see if I can figure out why the chip shits itself. May send you
more stuff to test.
Thanks I saw that commit.
Anytime you need me to test more, I am all over it! (:>
Best,
Daniel
