Messages in this thread | | | Subject | Re: [RFC][PATCHSET] mremap/mmap mess | From | Peter Zijlstra <> | Date | Wed, 09 Dec 2009 14:39:18 +0100 |
| |
On Wed, 2009-12-09 at 13:24 +0000, Al Viro wrote:
> > That is, IIRC the problem is that you need to have the argc/env in place > > for the binfmt engine thing, and need to have ran the binfmt engine > > thing before you know the personality. > > > > As to your idea, if that were feasible we could do without the copy and > > simply steal the pages directly from the old mm. > > *raised brows* > Old mm may bloody well be still shared, so I'd be vary of trying that - > there's a lot of nasty surprises we could get on that way.
Good point, but I think we might be able to do the copy from old to new in flush_old_exec() if we can get the binfmt things to pass the right address along (or set it in the brpm struct). It looks like it has clue about personality at that point.
| |