nfsd: fix startup order in nfsd_reply_cache_init
authorJeff Layton <jlayton@redhat.com>
Fri, 15 Mar 2013 13:16:29 +0000 (09:16 -0400)
committerJ. Bruce Fields <bfields@redhat.com>
Mon, 18 Mar 2013 21:21:30 +0000 (17:21 -0400)
commitac534ff2d5508bdff1358a55d88053da729ff46b
treedb11b9341feb5b90305fa3594549a86e0ea89448
parenta517b608fa3d9b65930ef53ffe4a2f9800e10f7d
nfsd: fix startup order in nfsd_reply_cache_init

If we end up doing "goto out_nomem" in this function, we'll call
nfsd_reply_cache_shutdown. That will attempt to walk the LRU list and
free entries, but that list may not be initialized yet if the server is
starting up for the first time. It's also possible for the shrinker to
kick in before we've initialized the LRU list.

Rearrange the initialization so that the LRU list_head and cache size
are initialized before doing any of the allocations that might fail.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
fs/nfsd/nfscache.c