On 20 Oct 2000, Trond Myklebust wrote:
> Under NFS the problem is that pages can (and *should*) be invalidated
> despite there being pending write backs. The server can trigger the
> need for a cache invalidation at any time.
OK, so what should happen if user does mmap() on NFS file, dirties the
page and server tells that page should be invalidated? Should we still see
that page in process' address space? Silently making it anonymous is the
worst thing possible - it's still accessible through process page tables
and you (a) still see the (allegedly) invalidated data and (b) have no way
in hell to get the cache coherency between processes. I.e. you are in for
situations when we do mmap(), fork(), write data in child and attempt to
read from that address in parent and child yield different
results. Moreover, future attempts to modify the data by either of them
will be invisible in another.
IOW, what do you really want from the invalidation? It looks like the
right thing _might_ be "remove from all page tables", but that can become
really interesting with mlock() and friends.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Oct 23 2000 - 21:00:17 EST