Re: [PATCH] mm,migration: Avoid race between shift_arg_pages() and rmap_walk() during migration by not migrating temporary stacks

[Mel Gorman <mel@csn.ul.ie>]

On Wed, May 12, 2010 at 12:54:27PM -0700, Andrew Morton wrote:
> On Wed, 12 May 2010 09:23:44 +0900 (JST)
> KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> wrote:
> 
> > > diff --git a/fs/exec.c b/fs/exec.c
> > > index 725d7ef..13f8e7f 100644
> > > --- a/fs/exec.c
> > > +++ b/fs/exec.c
> > > @@ -242,9 +242,10 @@ static int __bprm_mm_init(struct linux_binprm *bprm)
> > >  	 * use STACK_TOP because that can depend on attributes which aren't
> > >  	 * configured yet.
> > >  	 */
> > > +	BUG_ON(VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP);
> > 
> > Can we use BUILD_BUG_ON()? 
> 
> That's vastly preferable - I made that change.
> 

I will be surprised if it works. On x86, that is

#define VM_STACK_FLAGS  (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
#define VM_DATA_DEFAULT_FLAGS \
        (((current->personality & READ_IMPLIES_EXEC) ? VM_EXEC : 0 ) | \
         VM_READ | VM_WRITE | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)

so VM_STACK_FLAGS is depending on the value of current->personality
which we don't know at build time.

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab