[PATCH v12 4/5] cachefiles: Don't rely on backing fs storage map for most use cases

[Petr Chmelar]

David Howells <dhowells at redhat.com> wrote:

> Note that these patches are now upstream with this change incorporated.

Thanks, David, and sorry for the slow reply to your question about
the fix - I was on vacation.

I've now tested mainline (72a16bcff2) with the same reproducers as in
my earlier mail, and it fixes the problem.  All 4096 blocks read back
correctly on both full reads, and the reads were served from the
cache:

 - 9p, cache=fscache: 0 wrong blocks (936 on 13e340e397).
 - NFSv4.2 loopback with fsc: 0 wrong blocks (817 on 9f8b64f66),
   also after a server-side rewrite of half the file and after
   truncating and regrowing it through the mount.

Thanks again for the quick fix.

Petr