Problems with em0 failing to initialize on stable?

[George V. Neville-Neil]

At Thu, 17 Apr 2008 10:32:33 -0700,
Jack Vogel wrote:
> 
> On Thu, Apr 17, 2008 at 8:28 AM, Ben Stuyts <ben at altesco.nl> wrote:
> >
> >
> >  On 17 apr 2008, at 16:11, Jeremy Chadwick wrote:
> >
> >
> > > On Thu, Apr 17, 2008 at 02:02:51PM +0200, Ben Stuyts wrote:
> > >
> > > > On 17 apr 2008, at 13:20, Jeremy Chadwick wrote:
> > > >
> > > > > On Thu, Apr 17, 2008 at 06:32:32PM +0900, gnn at freebsd.org wrote:
> > > > >
> > > > > > I am running 7-STABLE with machines that are net booted.  On
> > occasion,
> > > > > > that is not with any level of predictability, this happens:
> > > > > >
> > > > > > em0: Hardware Initialization Failed
> > > > > > em0: Unable to initialize the hardware
> > > > > >
> > > > > > which of course stops the machine in its tracks.  A normal dmesg is
> > > > > > also included.
> > > > > >
> > > > > > Any steps I should take to help debug this?
> > > > > >
> > > > >
> > > > > George, can you provide the following?
> > > > >
> > > > > * Motherboard type and model (a URL to the board would be good)
> > > > > * kenv | grep smbios
> > > > > * pciconf -lv
> > > > >
> > > >
> > > > I have seen this too. A couple of times on a new server here. Maybe one
> > in
> > > > two or three reboots. So far, it only happens after a reboot, but not
> > after
> > > > a hard reset or power cycle. This board has 3 em network if's. Two on
> > the
> > > > board itself, and one on a daughter card on the ipmi add-on card. It is
> > > > always em0 that is giving trouble. The board is a Supermicro X7DBP-i:
> > > >
> > <http://www.supermicro.com/products/motherboard/xeon1333/5000P/X7DBP-i.cfm>
> > > >
> > >
> > > I'm wondering if the IPMI add-on is what's doing it.  However, the IPMI
> > > cards compatible with the X7DBP all have a dedicated NIC (versus
> > > piggybacking on top of the existing mainboards' NICs, which almost
> > > always causes problems of the mysterious sort).  I'm not sure, but I
> > > don't think em0 will be that NIC.
> > >
> >
> >  Correct, em0 and em1 are on the mainboard. em2 is on the
> > daughter-daughtercard and also single use for the OS. And then there's a 4th
> > nic which is the ipmi interface. It is completely separate. (I have the
> > SIM1U-3B & SIM1U-3D daughter cards:
> > <http://www.supermicro.com/products/accessories/addon/SIM.cfm>
> >
> >
> >
> > > Both you and George have boards that use the 82563EB.  jv@ will have to
> > > help with this one.  I wonder if it's a BIOS bug of some kind, where
> > > something on the NIC isn't getting reset by the BIOS on a soft boot...
> > >
> >
> >  Let me know if there's anything I can do to help. I can test patches during
> > evening hours when this server is mostly idle.
> >
> >  Ben
> >
> >
> 
> There is a fix in the shared code that is checked into CURRENT that
> addresses this, however, yesterday evening a problem cropped up that
> might still be unfixed with that code, I'm looking into that
> today...

Once that's fixed can it be MFC'd?

Best,
George