Re: [PATCH 1/3] virtio: Basic implementation of virtio pstore driver

[Cornelia Huck <cornelia.huck@de.ibm.com>]

On Mon, 18 Jul 2016 17:29:55 +0900
Namhyung Kim <namhyung@kernel.org> wrote:

> On Mon, Jul 18, 2016 at 09:54:39AM +0200, Cornelia Huck wrote:
> > On Mon, 18 Jul 2016 13:37:39 +0900
> > Namhyung Kim <namhyung@kernel.org> wrote:

> > > +#define VIRT_PSTORE_ORDER    2
> > > +#define VIRT_PSTORE_BUFSIZE  (4096 << VIRT_PSTORE_ORDER)
> > 
> > It may make sense to make the size of the buffer configurable through
> > the config space.
> 
> Right.  I'm considering it too, but it needs a buffer larger than
> kmsg_bytes (= 10K) to work properly in the current implementation.  As
> this version is just to verify the idea is sane and useful, I used a
> fixed size buffer.  Will change in the next version.

Sure, that makes sense for a prototype. We can guard any config space
entry with a feature bit, but this one makes sense to add from the
beginning.

> > Speaking of the standard: I think it makes sense to at least reserve a
> > device id for pstore, as the idea is sound. Maybe prepare a patch to
> > the standard as well if you have time?
> 
> I'd love to.  As I mentioned earlier, I don't have enough knowledge in
> this area.  Could you please provide some links about how can I do that?

See the virtio page at OASIS
(https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=virtio)
for a link to our subversion (yes...) repository. Just do two patches:
one to reserve a device id, and one that specifies how device and
driver work. (For examples, look at the proposed device types that have
been posted to the virtualization lists, e.g. virtio-crypto or
virtio-sdm). You just need to be patient, we're currently a bit
stalled...