freebsd stable - Jun 2014 - Unusable Emulex "oce" driver version in -STABLE, 9.3 and CURRENT

Hello,

This should be fixed before the release of 9.3, and of course as soon as
possible for the rest of the branches.

At least under 10-STABLE the "oce" driver can cause a panic when
there's a lot of network traffic. I have been able to reproduce it
using "iperf". Sometimes the panic can be triggered within seconds of
starting an iperf test.

There's a new driver available for download on the Emulex site, version
10.0.747.0, and it works perfectly. The source code compiles perfectly
under FreeBSD 10-STABLE (at least recent versions since April) and the kernel
module loads without trouble.

There's a bug report covering this but seems to be forgotten :/

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=183391



Thanks!




Borja.

On Jun 17, 2014, at 6:07 PM, Borja Marcos wrote:
> 
> Hello,
> 
> This should be fixed before the release of 9.3, and of course as soon as
possible for the rest of the branches.
> 
> At least under 10-STABLE the "oce" driver can cause a panic when
there's a lot of network traffic. I have been able to reproduce it
> using "iperf". Sometimes the panic can be triggered within
seconds of starting an iperf test.
> 
> There's a new driver available for download on the Emulex site, version
10.0.747.0, and it works perfectly. The source code compiles perfectly
> under FreeBSD 10-STABLE (at least recent versions since April) and the
kernel module loads without trouble.
> 
> There's a bug report covering this but seems to be forgotten :/
> 
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=183391
In my case, the problem manifests itself running 10. Emulex provides a more
recent driver which compiles perfectly and works perfectly (if we can safely
assume that a weekend saturating a couple of interfaces runnning benchmarks with
no glitches is a good hint).

I am now installing the latest beta of 9.3 to see if I can reproduce the panic.
I will follow-up in a couple of hours if possible.





Borja.