ppc64 oops..

[Nick Piggin]

Paul Mackerras wrote:
> Nick Piggin writes:
> 
> 
>>I really don't think we've missed PG_reserved. The ZERO_PAGE accounting
>>thing may be a problem, but that problem didn't come about due to
>>removal of PageReserved, but rather the concurrent removal of ZERO_PAGE
>>special casing we had there - it can be reinstated (and a solution for
>>2.6.15 won't be difficult).
> 
> 
> Not that I'm any sort of a VM expert, but it seems to me that we need
> some sort of way to mark things like the hashed page table on PowerPC
> as being "special" memory that is there, but that the VM system should
> just completely ignore.  That's what I thought PG_reserved was for,
> and IMHO it's useful for that.  For sure we should eliminate the
> abuses of PG_reserved, but I don't see that that means we should
> eliminate PG_reserved itself.
> 

Well, a page is introduced to the VM system in one of two ways
really. Either it gets put into the page allocator as a free page,
or it gets returned from ->nopage or otherwise mapped into a user
mapping.

The first case is arch specific, but sure PG_reserved may come in
handy to track these pages. swsusp for example uses this flag, and
in that case I think its usage is valid.

In the second case, the VM can't completely ignore the page. It is
cleaner to specify treatment of pages through this mapping by using
properties of the mapping rather than pages that might be in it
(ie. VM_RESERVED rather than PG_reserved).

-- 
SUSE Labs, Novell Inc.

Send instant messages to your online friends http://au.messenger.yahoo.com