/etc/rc.d/netif & em interfaces
bjmccann at gmail.com
Fri Dec 3 13:18:31 UTC 2010
On Wed, Dec 1, 2010 at 11:15 AM, Brian McCann <bjmccann at gmail.com> wrote:
> On Wed, Dec 1, 2010 at 10:51 AM, Brian McCann <bjmccann at gmail.com> wrote:
>> This isn't a Dell PoweEdge, but at least this is two platforms it's
>> happening on now. Strangely, my VMWare virtual machines that use the
>> em driver do not have this problem.
>> TSO is disabled.
>> I just tried adding "up" to the $ifconfig_emX lines in rc.conf, and it
>> didn't help.
>> What I discovered (and I'm still playing with a bit), is that it has
>> to do with checksum offloading. If you run "ifconfig emX -rxcsum",
>> the interface comes back to life. I just added "-rxcsum" to the
>> $ifconfig_emX lines in rc.conf, and I can run "/etc/rc.d/netif
>> restart" now, and the NIC continues to work fine.
>> I tried emailing the folks @ Intel that are mentioned in the man page
>> for the em driver, but the mail bounced...
> Brian (and everyone else reading the thread),
> I just found this...the patch was submitted some time ago, but somehow
> didn't make it into 8.1.
> http://www.freebsd.org/cgi/query-pr.cgi?pr=141843 . I'm trying it
> now. I'll keep you posted.
So, the patch didn't work against 8.1-RELEASE...it wouldn't compile
since there were some new files in the driver package. So, I checked
out the e1000 driver from HEAD and put it in my src tree (in addition
to some supporting files that were necessary), and compiled that. I
went through the SVN notes for the e1000 files, found the SVN version
number, and updated all the files listed in the commit. Here's the
Once I installed the new kernel module, all was well...no more
problems. Hopefully this will be included in 8.2-RELEASE.
Thanks, and good luck to all!
"I don't have to take this abuse from you -- I've got hundreds of
people waiting to abuse me."
-- Bill Murray, "Ghostbusters"
More information about the freebsd-questions