[Bug 276002] nfscl: data corruption using copy_file_range

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=276002

--- Comment #2 from Rick Macklem <rmacklem@FreeBSD.org> ---
Do we also need to add something like...

   if ((obj = invp->v_object) != NULL &&
       vm_object_mightbedirty(obj)) {
          VM_OBJECT_WLOCK(obj);
          vm_object_page_clean(obj, 0, 0, OBJPC_SYNC);
          VM_OBJECT_WUNLOCK(obj);
    }
in front of the
    error = ncl_flush(invp, MNT_WAIT, curthread, 1, 0);

If this is needed, the vnode locking code will need to be
changed to LK_EXCLUSIVE lock the invp.

Man, I hate mmap'd I/O.;-)

-- 
You are receiving this mail because:
You are the assignee for the bug.