Red Hat errata kernels for test...

Jeff Garzik jgarzik at
Mon Jan 27 12:06:01 CST 2003

On Mon, Jan 27, 2003 at 12:53:59PM -0500, Gabriel Rosenkoetter wrote:
> At Monday, January 27, 2003 12:45 PM, Jeff Garzik wrote:
> > On Mon, Jan 27, 2003 at 12:37:02PM -0500, Gabriel Rosenkoetter wrote:
> > > To be precise, dare we hope that the hw tcp csum errors are actually
> > > gone now?
> > 
> > I do not see any fixes specific to that, so I would not have a ton of
> > hope; however it will be useful to test, as some of the e1000 bugfixes
> > may have a positive effect on that problem nonetheless.
> I'm also interested in the tg3 fixes. Are these the same as in your 
> aragorn kernel which I had an initially bad response to and backed  
> off from on a production machine?
> Should I try this kernel rather than the aragorn one if I'm trying  
> something new to (further) fix tg3 hangs? (The bcm driver's right 
> out; throughput speed is direly important on this system, and I'm
> not especially fond of remote DoSes either.)

The latest errata kernel includes all the known e1000 and tg3 fixes so
far, but also some additional changes which may positive impact things.
So for most this should fix things, I think you are seeing a different
problem which may not be fixed by the latest errata kernel.

The most important thing, I think, that is the latest errata kernel is
(a) official Red Hat kernel [currently in Q/A], and
(b) a baseline from which future problems may be debugged/worked.

> Coincidentally, Barry Roomberg (with whom I work) got Matt Domsch all
> riled up about Broadcom at LinuxWorld last week, and Matt suggested
> that some sense might be beat into Broadcom by way of the anti-Linux
> exec there moving off the scene. Heard anything about that?

hehe, mebbe :)

Can't comment publicly nor privately without an NDA, though...


More information about the Linux-PowerEdge mailing list