Re: Terminate process that fails on a constrained allocation

[Andi Kleen]

On Thursday 09 February 2006 00:54, Christoph Lameter wrote:
> On Wed, 8 Feb 2006, Andrew Morton wrote:
> 
> > > If a caller cannot handle NULL then __GFP_NOFAIL has to be set, right?
> > 
> > That would assume non-buggy code.  I'm talking about the exercising of
> > hitherto-unused codepaths.  We've fixed many, many pieces of code which
> > simply assumed that kmalloc(GFP_KERNEL) succeeds.  I doubt if many such
> > simple bugs still exist, but there will be more subtle ones in there.
> 
> We could add __GFP_NOFAIL to kmem_getpages in slab.c to insure that 
> kmalloc waits rather than return NULL. Also a too drastic measure right?

Definitely too drastic. I bet that would cause deadlocks in quite some loads.

-Andi

-
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/