Re: [PATCH 1 of 4] mm: add a ptep_modify_prot transaction abstraction

[Ingo Molnar <mingo@elte.hu>]

* Hugh Dickins <hugh@veritas.com> wrote:

> On Mon, 16 Jun 2008, Linus Torvalds wrote:
> > On Mon, 16 Jun 2008, Jeremy Fitzhardinge wrote:
> > >
> > > ptep_modify_prot_start() returns the current pte value, and puts the
> > >   pte entry into a state where either the hardware will not update the
> > >   pte, or if it does, the updates will be preserved on commit.
> > > 
> > > ptep_modify_prot_commit() writes back the updated pte, makes sure that
> > >   any hardware updates made since ptep_modify_prot_start() are
> > >   preserved.
> > 
> > Ok, I'm fine with this now that it's renamed to be clearly about just 
> > protection bits.
> > 
> > So 
> > 
> > 	Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
> 
> And seems very reasonable (and exceptionally well described) to me too.
> 
> 	Acked-by: Hugh Dickins <hugh@veritas.com>

thanks guys, i've added the Acked-by's and added a new -tip topic for 
this.

The dependencies are a bit tricky and the changes contain mm/ and 
include/asm-generic/ bits so lets try a new Git trick here to keep it 
all tidy and disciplined for v2.6.27 merging.

So i've created a new tip/mm/xen topic branch (the 85th -tip topic 
branch ;-), which is COW-ed off the current tip/x86/xen topic branch [on 
which branch these changes have some dependencies], and added these 4 
changes to that. The tip/x86/xen (append-only-) topic will continue to 
advance as usual, and we likely wont have dependencies on the stuff in 
tip/mm/xen. (if there will be such dependencies we can handle that too)

In v2.6.27 we can then offer up the two branches separately for upstream 
merge, and tip/x86/xen will still only have x86 and xen changes, not any 
mm changes. (Obviously tip/mm/xen will be offered after tip/x86/xen has 
gone upstream - so that it will only contain these 4 patches ontop of 
already-upstream changes)

it will all be auto-merged into linux-next so there this internal 
structure is not visible, all the changes are available for wide testing 
of course.

i've added these mm/ changes to auto-core-next (not auto-x86-next), if 
that is fine by Andrew.

	Ingo