Re: [linux-pm] [PATCH] hibernation should work ok with memory hotplug

[Rafael J. Wysocki]

On Tuesday, 4 of November 2008, Dave Hansen wrote:
> On Tue, 2008-11-04 at 09:54 +0100, Rafael J. Wysocki wrote:
> > To handle this, I need to know two things:
> > 1) what changes of the zones are possible due to memory hotplugging
> > (i.e.    can they grow, shring, change boundaries etc.)
> 
> All of the above. 

OK

If I allocate a page frame corresponding to specific pfn, is it guaranteed to
be associated with the same pfn in future?

> > 2) what kind of locking is needed to prevent zones from changing.
> 
> The amount of locking is pretty minimal.  We depend on some locking in
> sysfs to keep two attempts to online from stepping on the other.
> 
> There is the zone_span_seq*() set of functions.  These are used pretty
> sparsely, but we do use them in page_outside_zone_boundaries() to notice
> when a zone is resized.
> 
> There are also the pgdat_resize*() locks.  Those are more for internal
> use guarding the sparsemem structures and so forth.
> 
> Could you describe a little more why you need to lock down zone
> resizing?  Do you *really* mean zones, or do you mean "the set of memory
> on the system"?

The latter, but our internal data structures are designed with zones in mind.

> Why walk zones instead of pgdats? 

This is a historical thing rather than anything else.  I think we could switch
to pgdats, but that would require a code rewrite that's likely to introduce
bugs, while our image-creating code is really well tested and doesn't change
very often.

Thanks,
Rafael
--
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/