Re: Terminate process that fails on a constrained allocation V3

[Christoph Lameter]

On Thu, 9 Feb 2006, Andi Kleen wrote:

> > - Do not return NULL and therefore do not change the return values of
> >   __alloc_pages.
> 
> Hmm, to make this work well i guess mmap() would need to be changed
> to take the policy into account when doing the !MAP_NORESERVE checking
> when the kernel runs in strict no overcommit mode.
> Otherwise there is no fool proof way an application can prevent getting killed.

There is no fool proof way right now for the MPOL_BIND case. 

> And mbind() would need to recompute it and fail if the new policy's possible
> allocation are not guaranteed to work.

Well we need to fix that but the patch does not introduce the problem.

> Doing this all properly would probably get quite messy.

I'd say making overcommit working nicely with MPOL_BIND is yet another 
fundamental issue for the policy layer but it does not matter for this 
patch.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/