Re: Race between vmtruncate and mapped areas?

[William Lee Irwin III]

On Tuesday, May 13, 2003 16:20:38 -0700 William Lee Irwin III <[e-mail]> wrote:
>> The mmap_sem works because then ->i_size can't be sampled by
>> filemap_nopage() before the pagetable wiping operation starts.

On Tue, May 13, 2003 at 06:28:20PM -0500, Dave McCracken wrote:
> So why isn't that the right way to do it?  Waiting for mmap_sem guarantees
> we won't catch a page fault in flight, which is the cause of the problem in
> the first place.

Oh it works fine, it's just the explanation for why it's sufficient I
was looking for.

-- wli