[openib-general] EEH: MMIO Failure on Power5

[Thaddeus Ternes]

Well, I moved the card to slot 5 and things seem to be working...

I have another Power5 and Mellanox card available, so I decided to
retest with them to see what the solution was.  I dropped the card
into slot 5 on the second Power 5, and it came right up, even without
the firmware upgrade (though the module did inform me that I had old
firmware).  Apparently it was just an issue with the card not being
right slot.

Is there some place that this is documented and I've just missed it? 
Some folks that I work with (who are more familiar with the Power
series than I am) didn't seem to know much about it either. 
Thankfully, you knew about it, Pradeep.  Still... it might be nice if
this was a little more obvious.

Thanks for helping track this down.

Thaddeus


On 9/26/05, Pradeep Satyanarayana <pradeep at us.ibm.com> wrote:
>
>
> Tried to find out the "default superslotes" for an OpenPower 720. Please try either slot 2 or 5. I delayed my response to make certain that these were indeed the superslots. I am still not a 100% certain -no point waiting beyond a certain stage.
>
>  If you can please go ahead and try these and let us see what happens. Also can you provide the output of "lspci -v" before you load the ib_mthca?
>
>  The firmware I was referring to was the OpenPower firmware, not that of the HCA.
>
>  Pradeep
>  pradeep at us.ibm.com
>
>  Thaddeus Ternes <tternes at gmail.com>
>
>
>
>
>
>
>
> Thaddeus Ternes <tternes at gmail.com>
>
> 09/23/2005 11:23 AM
>
> Please respond to
>  Thaddeus Ternes
>
>
> To
>  Pradeep Satyanarayana/Beaverton/IBM at IBMUS
>
>
> cc
>  openib-general at openib.org, Roland Dreier <rolandd at cisco.com>
>
>
> Subject
>  Re: [openib-general] EEH: MMIO Failure on Power5
>
>
>  I've tried a few things, but still seem to get the same error.  My
>  testing has been on 2.6.13.1, with SVN IB code (as of Monday).  The
>  ib_mthca module reports my HCA FW version to be 3.2.0 (which is
>  admittedly old).  Updating this old firmware will likely be my next
>  step.
>
>  Originally, I had installed the card in slot 1.  I've since poked
>  around in a PDF file I found on IBM's site and concluded that I should
>  have installed the card in slot 3, though I'm still not overly
>  confident about that.  I/O Adapter Large Capacity is also now enabled
>  (it wasn't previously, and changing it while the card was in slot 1
>  didn't seem to affect anything).
>
>  Is somebody aware of a clear way to identify which of the slots in the
>  720 are "superslots," as I've had no luck so far in my hunt in the
>  documentation.  Most likely, I've mistakenly skipped over it.
>
>  Thanks.
>
>  Thaddeus
>
>  On 9/22/05, Pradeep Satyanarayana <pradeep at us.ibm.com> wrote:
>  >
>  >
>  > I have filed a bug against the kernel (for p-series) as a starting point.
>  > Could you please flll me on some of the other specifics a) which kernel were
>  > you using b) firmware level (presumably it is uptodate).
>  >
>  >  One other issue that I failed to mention previously - is the HCA in one of
>  > the superslots (I know on my p570 slots 2 and 6 are superslots by default)
>  > and, is this superslot enabled?
>  >
>  >  Here is a quote of how to enable superslots-
>  >
>  >  One issue with the Mellanox cards in pSeries systems is to ensure that the
>  > card is installed in a superslot, and that the "I/O Adapter Enlarged
>  > Capacity" setting has been enabled for the system. For a p570, slots C6 and
>  > C2 are the available super slots. To enable the "Enlarged Capacity" feature,
>  > go to ASM and select the following screens:
>  >
>  >  System Configuration->I/O Adapter Enlarged Capacity
>  >  Set the setting to Enabled and save it.
>  >
>  >  If this does not help, I have already filed the bug. Please let me know
>  > either way.
>  >
>  >  Pradeep
>  >  pradeep at us.ibm.com
>
>
>
>
>