On Wed, 21 Feb 2018 11:33:28 +0000
Stefan Hajnoczi <stefa...@redhat.com> wrote:

> On Wed, Feb 21, 2018 at 12:11:35AM +0100, Tomáš Golembiovský wrote:
> > On Tue, 20 Feb 2018 17:15:44 +0000
> > Stefan Hajnoczi <stefa...@redhat.com> wrote:
> > 
> > > On Mon, Feb 19, 2018 at 02:10:29PM +0100, Tomáš Golembiovský wrote:
> > > > +\item[VIRTIO_BALLOON_S_CACHES (7)] The amount of memory currently used 
> > > > for
> > > > +  caching files and other data from disk (in bytes).
> > > >  \end{description}  
> > > 
> > > This one is somewhat vague.  I'm not sure if it means guest page cache
> > > specifically or something more general.
> > > 
> > > On the other hand, the numbers tend to be OS specific...
> > 
> > We should come up with a definition that is as definite as possible
> > while leaving room for all systems to substitute a value that is
> > natively available. So if anyone has a better formulation then please
> > advise. The original intent was not to create a Linux specific field.
> > 
> > On Linux the value corresponds to Buffers + Cached + SwapCached memory
> > stats. Correct me if I'm wrong but that should be all clean pages that
> > have a counterpart on disk and can be quickly reclaimed and used for
> > something else (without additional IO).
> 
> Thanks for clarifying.  Based on what you posted, how about:
> 
> "The amount of memory, in bytes, that can be quickly reclaimed without
> additional I/O.  Typically these pages are used for caching files from
> disk."
> 
> ?

Sounds reasonable, thanks

    Tomas


> 
> > On Windows this roughly corresponds to the standby list. Or maybe
> > standby list plus something else. I suppose there may be some
> > alternative on BSD too.
> > 
> >     Tomas
> > 
> > 
> > -- 
> > Tomáš Golembiovský <tgole...@redhat.com>


-- 
Tomáš Golembiovský <tgole...@redhat.com>

---------------------------------------------------------------------
To unsubscribe, e-mail: virtio-dev-unsubscr...@lists.oasis-open.org
For additional commands, e-mail: virtio-dev-h...@lists.oasis-open.org

Reply via email to