svn commit: r299201 - head/sys/fs/nfsserver

[Rick Macklem]

Author: rmacklem
Date: Fri May  6 23:26:17 2016
New Revision: 299201
URL: https://svnweb.freebsd.org/changeset/base/299201

Log:
  Give mountd -S priority over outstanding RPC requests when suspending the nfsd.
  
  It was reported via email that under certain heavy RPC loads
  long delays before the exports would be updated was observed
  when using "mountd -S". This patch reverses the priority between
  the exclusive lock request to suspend the nfsd threads and the
  shared lock request for performing RPCs.
  As such, when mountd attempts to suspend the nfsd threads, it
  gets priority over outstanding RPC requests to do this.
  I suspect that the case reported was an artificial test load,
  but this patch did fix the problem for the reporter.
  
  Reported and Tested by:	josephlai at qnap.com
  MFC after:	2 weeks

Modified:
  head/sys/fs/nfsserver/nfs_nfsdkrpc.c

Modified: head/sys/fs/nfsserver/nfs_nfsdkrpc.c
==============================================================================
--- head/sys/fs/nfsserver/nfs_nfsdkrpc.c	Fri May  6 22:54:56 2016	(r299200)
+++ head/sys/fs/nfsserver/nfs_nfsdkrpc.c	Fri May  6 23:26:17 2016	(r299201)
@@ -231,10 +231,16 @@ nfssvc_program(struct svc_req *rqst, SVC
 		 * Get a refcnt (shared lock) on nfsd_suspend_lock.
 		 * NFSSVC_SUSPENDNFSD will take an exclusive lock on
 		 * nfsd_suspend_lock to suspend these threads.
+		 * The call to nfsv4_lock() that preceeds nfsv4_getref()
+		 * ensures that the acquisition of the exclusive lock
+		 * takes priority over acquisition of the shared lock by
+		 * waiting for any exclusive lock request to complete.
 		 * This must be done here, before the check of
 		 * nfsv4root exports by nfsvno_v4rootexport().
 		 */
 		NFSLOCKV4ROOTMUTEX();
+		nfsv4_lock(&nfsd_suspend_lock, 0, NULL, NFSV4ROOTLOCKMUTEXPTR,
+		    NULL);
 		nfsv4_getref(&nfsd_suspend_lock, NULL, NFSV4ROOTLOCKMUTEXPTR,
 		    NULL);
 		NFSUNLOCKV4ROOTMUTEX();