Re: [PATCH] restore protections after forced fault in get_user_pages

[Roland McGrath]

> That should be sufficient, I think: since "handle_mm_fault()" marks the 
> page dirty (but not writable) and will have done all the work to do a COW, 
> we know that once we do the "follow_page()", we'll be getting a private 
> copy. Which is what we wanted.

The only potential hole I can see here is if there is an (exceedingly rare)
race where the page could be ejected after handle_mm_fault, then brought
back in but not marked dirty, before follow_page looks it up and returns a
page to be used for writing without marking it dirty.  Obviously this is a
ridiculously unlikely case.  But what I don't know is whether it is
strictly speaking impossible.  That is, that a page once dirty then later
not present, would ever again be present without also being dirty.


Thanks,
Roland
-
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/