Re: RFC/POC Make Page Tables Relocatable

[Dave Hansen <haveblue@us.ibm.com>]

On Thu, 2007-10-25 at 16:10 -0400, Ross Biro wrote:
> On 10/25/07, Dave Hansen <haveblue@us.ibm.com> wrote:
> > How would it get freed?
> 
> The process exists or ummaps the range of memory.  The relocation code
> is likely called on a different cpu in the node and currently has no
> way to pin the data in memory.  Perhaps finding a way to pin the page
> would help the other locking issues, so it might solve lots of
> problems.

Taking a simple reference count on the page will keep it from getting
freed.  It won't keep it from getting _unused_, but it will against
getting actually freed back to the allocator.

But, if you get to this point and you have a page and the only person
with a reference to it is you, it _should_ be completely empty of pte
entries.  They were all cleared at zap_pte_range() time.

You need other mechanisms in place, anyway, to keep ptes from being
instantiated or shot down while you're doing the copy itself.

-- Dave

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>