> However, though I can well believe that your patch works well for you,
> I don't think it's general enough: there is no guarantee that the tests
> in can_do_hugetlb_shm() will give the same answer to the user who ends
> up calling shm_destroy() as it did once upon a time to the user who
> called hugetlb_file_setup().
> 
> So, please could you try this alternative patch below, to see if it
> passes your testing too, and let us know the result?  I'm sure we'd
> like to get a fix into 2.6.31, and into 2.6.30-stable.


Yes, your observation is right and your modified patch works good for
me.

So long
Stefan