Re: [RFC] mlock/stack guard interaction fixup

[Peter Zijlstra <peterz@infradead.org>]

On Mon, 2010-08-23 at 11:50 -0700, Jeremy Fitzhardinge wrote:
> On 08/23/2010 10:34 AM, Linus Torvalds wrote:
> > Quite frankly, I personally believe that people who play games with
> > mlock are misguided. The _one_ special case is for protecting keys or
> > private data that you do not want to hit the disk in some unencrypted
> > mode, and quite frankly, you should strive to handle those way more
> > specially than just putting them in some random place ("on the stack"
> > or "in some malloc()'ed area"). The sane model for doing that is
> > generally to explicitly mmap() and mlock the area, so that you get a
> > very controlled access pattern, and never have to worry about things
> > like COW etc.
> 
> Is that guaranteed to work (in Linux or in general)?  mlock has always
> meant "won't generate disk IO to fault in/evicted", but does it prevent
> dirty pages from being written out so long as they also remain
> resident?  Or does it depend on the precise type of page you're
> mlocking?  For example, what does mlock on a shared writeable mapping mean?

mlock() simply avoids major faults, nothing more.

I think both page migration and page-out for shared pages where some
maps are !mlocked can cause unmaps and thus minor faults.

mlock and dirty do not interact, they will still be cleaned/written out
as normal.