Re: [Qemu-devel] [PATCH 3/3] nvdimm: platform capabilities command line option

[Ross Zwisler <ross.zwisler@linux.intel.com>]

On Thu, May 10, 2018 at 03:28:48PM +0200, Igor Mammedov wrote:
> On Fri, 27 Apr 2018 15:53:14 -0600
> Ross Zwisler <ross.zwisler@linux.intel.com> wrote:
> 
> > Add a device command line option to allow the user to control the Platform
> > Capabilities Structure in the virtualized NFIT.
> > 
> > Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
> > ---
> >  docs/nvdimm.txt         | 22 ++++++++++++++++++++++
> >  hw/acpi/nvdimm.c        | 29 +++++++++++++++++++++++++----
> >  hw/mem/nvdimm.c         | 28 ++++++++++++++++++++++++++++
> >  include/hw/mem/nvdimm.h |  6 ++++++
> >  4 files changed, 81 insertions(+), 4 deletions(-)
> > 
> > diff --git a/docs/nvdimm.txt b/docs/nvdimm.txt
> > index e903d8bb09..13a2c15b70 100644
> > --- a/docs/nvdimm.txt
> > +++ b/docs/nvdimm.txt
> > @@ -153,3 +153,25 @@ guest NVDIMM region mapping structure.  This unarmed flag indicates
> >  guest software that this vNVDIMM device contains a region that cannot
> >  accept persistent writes. In result, for example, the guest Linux
> >  NVDIMM driver, marks such vNVDIMM device as read-only.
> > +
> > +Platform Capabilities
> > +---------------------
> > +
> > +ACPI 6.2 Errata A added support for a new Platform Capabilities Structure
> > +which allows the platform to communicate what features it supports related to
> > +NVDIMM data durability.  Users can provide a capabilities value to a guest via
> > +the optional "cap" device command line option:
> > +
> > + -device nvdimm,id=nvdimm1,memdev=mem1,cap=3
> > +
> > +As of ACPI 6.2 Errata A, the following values are valid for the bottom two
> > +bits:
> > +
> > +2 - Memory Controller Flush to NVDIMM Durability on Power Loss Capable.
> > +3 - CPU Cache Flush to NVDIMM Durability on Power Loss Capable.
> > +
> > +For a complete list of the flags available please consult the ACPI spec.
> > +
> > +These platform capabilities apply to the entire virtual platform, so it is
> > +recommended that only one "cap" device command option be given per virtual
> > +machine.  This value will apply to all NVDIMMs in the virtual platform.
> This looks like it should be machine property instead of per device one,
> you can get rid of static variable and mismatch check and a weird nvdimm CLI
> option that implies that the option is per device.

Yep, that's much better.  I have this implemented and ready to go.

> Also an extra patch to for make check that will test setting 'cap'
> would be nice (an extra testcase in tests/bios-tables-test.c)

Hmm...I've been looking at this, and it doesn't look like there is any
verification around a lot of the ACPI tables (NFIT, SRAT, etc).

I've verified my patch by interacting with a guest with various settings - is
this good enough, or do you really want me to test the value (which I think
would just be "do I get out what I put in at the command line") via the unit
test infrastructure?

Thank you for the review.