sunrpc: Don't schedule an upcall on a replaced cache entry.
authorNeilBrown <neilb@suse.de>
Thu, 13 Jun 2013 02:53:42 +0000 (12:53 +1000)
committerJ. Bruce Fields <bfields@redhat.com>
Mon, 1 Jul 2013 21:53:28 +0000 (17:53 -0400)
When a cache entry is replaced, the "expiry_time" get set to
zero by a call to "cache_fresh_locked(..., 0)" at the end of
"sunrpc_cache_update".

This low expiry time makes cache_check() think that the 'refresh_age'
is negative, so the 'age' is comparatively large and a refresh is
triggered.
However refreshing a replaced entry it pointless, it cannot achieve
anything useful.

So teach cache_check to ignore a low refresh_age when expiry_time
is zero.

Reported-by: Bodo Stroesser <bstroesser@ts.fujitsu.com>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
net/sunrpc/cache.c

index 5478a015ab003f0f6b62ba48ae7ffcac7770fa51..49eb37010aa35da3bba5e237a7cff107075114e0 100644 (file)
@@ -262,7 +262,8 @@ int cache_check(struct cache_detail *detail,
        if (rqstp == NULL) {
                if (rv == -EAGAIN)
                        rv = -ENOENT;
-       } else if (rv == -EAGAIN || age > refresh_age/2) {
+       } else if (rv == -EAGAIN ||
+                  (h->expiry_time != 0 && age > refresh_age/2)) {
                dprintk("RPC:       Want update, refage=%ld, age=%ld\n",
                                refresh_age, age);
                if (!test_and_set_bit(CACHE_PENDING, &h->flags)) {