Re: [PATCH] [RFC] fadvise: Add _VOLATILE,_ISVOLATILE, and _NONVOLATILEflags

[Dave Hansen]

On 11/22/2011 02:45 AM, Rik van Riel wrote:
> 4) Virtualization.  Marking an object (and its pages)
>     _VOLATILE inside a guest will not be visible on the
>     host side, which means a virtual system may continue
>     to suffer the performance penalty anyway.

Yeah, I guess we still have to communicate it _somehow_.

I guess we could theoretically pass the calls up to the
hypervisor and it could even make its own VOLATILE calls
to the host kernel.  We'd also have to pass back down the
"was this evicted" information during a re-pin. That seems
messy to me.

Is it really any different of a problem than page cache?
The guest has data sitting in RAM that it probably doesn't
need.  If we passed up just the amount of unpinned data back
up to the hypervisor, it would have a decent idea how much
it could balloon the guest, for instance.  That would fit
in well with some of the existing schemes that folks have
and be *much* nicer than what they've got at the moment.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/