git: 8513b4f15362 - stable/12 - nfscl: Fix NFSv4.1/4.2 mount recovery from an expired lease

[Rick Macklem]

The branch stable/12 has been updated by rmacklem:

URL: https://cgit.FreeBSD.org/src/commit/?id=8513b4f15362161c38c18a4262997468951d3ea0

commit 8513b4f15362161c38c18a4262997468951d3ea0
Author:     Rick Macklem <rmacklem at FreeBSD.org>
AuthorDate: 2021-05-19 21:52:56 +0000
Commit:     Rick Macklem <rmacklem at FreeBSD.org>
CommitDate: 2021-06-04 03:19:04 +0000

    nfscl: Fix NFSv4.1/4.2 mount recovery from an expired lease
    
    The most difficult NFSv4 client recovery case happens when the
    lease has expired on the server.  For NFSv4.0, the client will
    receive a NFSERR_EXPIRED reply from the server to indicate this
    has happened.
    For NFSv4.1/4.2, most RPCs have a Sequence operation and, as such,
    the client will receive a NFSERR_BADSESSION reply when the lease
    has expired for these RPCs.  The client will then call nfscl_recover()
    to handle the NFSERR_BADSESSION reply.  However, for the expired lease
    case, the first reclaim Open will fail with NFSERR_NOGRACE.
    
    This patch recognizes this case and calls nfscl_expireclient()
    to handle the recovery from an expired lease.
    
    This patch only affects NFSv4.1/4.2 mounts when the lease
    expires on the server, due to a network partitioning that
    exceeds the lease duration or similar.
    
    (cherry picked from commit c28cb257ddfe3339756f6fd659fa4a2efa4de2cb)
---
 sys/fs/nfsclient/nfs_clstate.c | 37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)

diff --git a/sys/fs/nfsclient/nfs_clstate.c b/sys/fs/nfsclient/nfs_clstate.c
index e705af31185b..9965b7ca7e4f 100644
--- a/sys/fs/nfsclient/nfs_clstate.c
+++ b/sys/fs/nfsclient/nfs_clstate.c
@@ -1977,6 +1977,7 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 	u_int32_t delegtype = NFSV4OPEN_DELEGATEWRITE, mode;
 	int i, igotlock = 0, error, trycnt, firstlock;
 	struct nfscllayout *lyp, *nlyp;
+	bool recovered_one;
 
 	/*
 	 * First, lock the client structure, so everyone else will
@@ -2050,6 +2051,7 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 	 * Now traverse the state lists, doing Open and Lock Reclaims.
 	 */
 	tcred = newnfs_getcred();
+	recovered_one = false;
 	owp = LIST_FIRST(&clp->nfsc_owner);
 	while (owp != NULL) {
 	    nowp = LIST_NEXT(owp, nfsow_list);
@@ -2083,6 +2085,7 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 			op->nfso_mode, op, NULL, 0, &ndp, 1, delegtype,
 			tcred, p);
 		    if (!error) {
+			recovered_one = true;
 			/* Handle any replied delegation */
 			if (ndp != NULL && ((ndp->nfsdl_flags & NFSCLDL_WRITE)
 			    || NFSMNT_RDONLY(nmp->nm_mountp))) {
@@ -2141,6 +2144,21 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 				nfscl_freelockowner(lp, 0);
 			    lp = nlp;
 			}
+		    } else if (error == NFSERR_NOGRACE && !recovered_one &&
+			NFSHASNFSV4N(nmp)) {
+			/*
+			 * For NFSv4.1/4.2, the NFSERR_EXPIRED case will
+			 * actually end up here, since the client will do
+			 * a recovery for NFSERR_BADSESSION, but will get
+			 * an NFSERR_NOGRACE reply for the first "reclaim"
+			 * attempt.
+			 * So, call nfscl_expireclient() to recover the
+			 * opens as best we can and then do a reclaim
+			 * complete and return.
+			 */
+			nfsrpc_reclaimcomplete(nmp, cred, p);
+			nfscl_expireclient(clp, nmp, tcred, p);
+			goto out;
 		    }
 		}
 		if (error != 0 && error != NFSERR_BADSESSION)
@@ -2227,6 +2245,23 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 		if (error) {
 		    if (nop != NULL)
 			free(nop, M_NFSCLOPEN);
+		    if (error == NFSERR_NOGRACE && !recovered_one &&
+			NFSHASNFSV4N(nmp)) {
+			/*
+			 * For NFSv4.1/4.2, the NFSERR_EXPIRED case will
+			 * actually end up here, since the client will do
+			 * a recovery for NFSERR_BADSESSION, but will get
+			 * an NFSERR_NOGRACE reply for the first "reclaim"
+			 * attempt.
+			 * So, call nfscl_expireclient() to recover the
+			 * opens as best we can and then do a reclaim
+			 * complete and return.
+			 */
+			nfsrpc_reclaimcomplete(nmp, cred, p);
+			nfscl_expireclient(clp, nmp, tcred, p);
+			free(nowp, M_NFSCLOWNER);
+			goto out;
+		    }
 		    /*
 		     * Couldn't reclaim it, so throw the state
 		     * away. Ouch!!
@@ -2234,6 +2269,7 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 		    nfscl_cleandeleg(dp);
 		    nfscl_freedeleg(&clp->nfsc_deleg, dp, true);
 		} else {
+		    recovered_one = true;
 		    LIST_INSERT_HEAD(&extra_open, nop, nfso_list);
 		}
 	    }
@@ -2271,6 +2307,7 @@ nfscl_recover(struct nfsclclient *clp, struct ucred *cred, NFSPROC_T *p)
 	if (NFSHASNFSV4N(nmp))
 		(void)nfsrpc_reclaimcomplete(nmp, cred, p);
 
+out:
 	NFSLOCKCLSTATE();
 	clp->nfsc_flags &= ~NFSCLFLAGS_RECVRINPROG;
 	wakeup(&clp->nfsc_flags);