On Tue, Dec 3, 2019 at 4:43 AM Matthew Wilcox [off-list ref] wrote:

On Tue, Dec 03, 2019 at 03:01:17PM +1100, Alastair D'Silva wrote:

quoted

On Mon, 2019-12-02 at 19:50 -0800, Matthew Wilcox wrote:

quoted

On Tue, Dec 03, 2019 at 02:46:28PM +1100, Alastair D'Silva wrote:

quoted

This series adds support for OpenCAPI SCM devices, exposing

Could we _not_ introduce yet another term for persistent memory?

"Storage Class Memory" is an industry wide term, and is used repeatedly
in the device specifications. It's not something that has been pulled
out of thin air.

"Somebody else also wrote down Storage Class Memory".  Don't care.
Google gets 750k hits for Persistent Memory and 150k hits for
Storage Class Memory.  This term lost.

quoted

The term is also already in use within the 'papr_scm' driver.

The acronym "SCM" is already in use.  Socket Control Messages go back
to early Unix (SCM_RIGHTS, SCM_CREDENTIALS, etc).  Really, you're just
making the case that IBM already uses the term SCM.  You should stop.

I tend to agree. The naming of things under
arch/powerpc/platforms/pseries/ is not under my purview, but
drivers/nvdimm/ is. Since this driver is colocated with the "pmem"
driver let's not proliferate the "scm" vs "pmem" confusion.