Re: sbefifo userspace api

[Stewart Smith <stewart@linux.vnet.ibm.com>]

Brad Bishop <bradleyb@fuzziesquirrel.com> writes:
>> On Feb 21, 2017, at 4:49 PM, Patrick Williams <patrick@stwcx.xyz> wrote:
>> 
>> On Tue, Feb 21, 2017 at 02:34:18AM +1100, Alistair Popple wrote:
>>> I would also be interested in the answer to the inverse - why put them
>>> in the kernel? There's the OCC hwmon driver - what chip-ops does that
>>> need? Just get/putscom? If that's the case I thought we were already
>>> exposing scom chip-op operations via an OpenFSI master?
>> 
>> When the P9 chip has security enabled, you cannot even issue SCOM
>> operations without going through the SBE FIFO.  The SBE provides a
>> white-list filtering system that prevents the majority of the SCOM space
>> from being accessed.  We will need a "SCOM over FIFO" driver on top of
>> FSI.
>
> Well, we need a way to do scoms over sbe.  Not sure that equals being
> a device driver.  As it stands if it were a driver, it would just be
> very thin wrapper around the sbefifo driver.

it'd be nice if getscom/putscom userspace could be the same on BMC and
host though.

-- 
Stewart Smith
OPAL Architect, IBM.