Skip to contentSkip to conversation

nfsd: containerise NFSv4 recovery state

PATCH In review Stanislav Kinsbursky 2 participants 4 messages Nov 26, 2012 – Nov 28, 2012

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover nfsd: containerise NFSv4 recovery state

This patch set makes rec_file and in_grace global static variables allocated per network namespace.
Both of them are a part of client tracking engine, which is containerised already.

The following series implements...

---

Stanislav Kinsbursky (2):
nfsd: recovery - make rec_file per net
nfsd: recovery - make in_grace per net

 fs/nfsd/netns.h       |    3 ++
 fs/nfsd/nfs4recover.c |   79 ++++++++++++++++++++++++-------------------------
 2 files changed, 42 insertions(+), 40 deletions(-)
#1 lore raw

PATCH 1/2 nfsd: recovery - make rec_file per net

Opening and closing of this file is done in client tracking init and exit operations.
Client tracking is done in network namespace context already. So let's make this file opened and closed per network context - this will simlify it's management.

---
 fs/nfsd/netns.h       |    2 +
 fs/nfsd/nfs4recover.c |   70 +++++++++++++++++++++++++------------------------
 2 files changed, 37 insertions(+), 35 deletions(-)
fs/nfsd/netns.h +2 -0
index 08d5fa1..1305632 100644
@@ -84,6 +84,8 @@ struct nfsd_net {
 
 	/* client_lock protects the client lru list and session hash table */
 	spinlock_t client_lock;
+
+	struct file *rec_file;
 };
 
 extern int nfsd_net_id;
fs/nfsd/nfs4recover.c +35 -35

225 lines not on this page · raw message

#2 lore raw

PATCH 2/2 nfsd: recovery - make in_grace per net

In reply to Stanislav Kinsbursky

Flag in_grace is a part of client tracking state, which is network namesapce aware. So let'a replace global static variable with per-net one.

---
 fs/nfsd/netns.h       |    1 +
 fs/nfsd/nfs4recover.c |    9 ++++-----
 2 files changed, 5 insertions(+), 5 deletions(-)
fs/nfsd/netns.h +1 -0
index 1305632..9047706 100644
@@ -86,6 +86,7 @@ struct nfsd_net {
 	spinlock_t client_lock;
 
 	struct file *rec_file;
+	bool in_grace;
 };
 
 extern int nfsd_net_id;
fs/nfsd/nfs4recover.c +4 -5
index 3e76d28..359793f 100644
@@ -64,7 +64,6 @@ struct nfsd4_client_tracking_ops {
 /* Globals */
 static char user_recovery_dirname[PATH_MAX] = "/var/lib/nfs/v4recovery";
 static struct nfsd4_client_tracking_ops *client_tracking_ops;
-static bool in_grace;
 
 static int
 nfs4_save_creds(const struct cred **original_creds)
@@ -221,7 +220,7 @@ out_put:
 out_unlock:
 	mutex_unlock(&dir->d_inode->i_mutex);
 	if (status == 0) {
-		if (in_grace) {
+		if (nn->in_grace) {
 			crp = nfs4_client_to_reclaim(dname, nn);
 			if (crp)
 				crp->cr_clp = clp;
@@ -358,7 +357,7 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
 	nfs4_reset_creds(original_cred);
 	if (status == 0) {
 		vfs_fsync(nn->rec_file, 0);
-		if (in_grace) {
+		if (nn->in_grace) {
 			/* remove reclaim record */
 			crp = nfsd4_find_reclaim_client(dname, nn);
 			if (crp)
@@ -394,7 +393,7 @@ nfsd4_recdir_purge_old(struct nfsd_net *nn, time_t boot_time)
 {
 	int status;
 
-	in_grace = false;
+	nn->in_grace = false;
 	if (!nn->rec_file)
 		return;
 	status = mnt_want_write_file(nn->rec_file);
@@ -473,7 +472,7 @@ nfsd4_init_recdir(struct net *net)
 
 	nfs4_reset_creds(original_cred);
 	if (!status)
-		in_grace = true;
+		nn->in_grace = true;
 	return status;
 }
 
#3 lore raw
In reply to Stanislav Kinsbursky
Quoted 4 lines · Stanislav Kinsbursky

This patch set makes rec_file and in_grace global static variables allocated per network namespace.
Both of them are a part of client tracking engine, which is containerised already.

Thanks, applying for 3.8.--b.

Quoted 14 lines

The following series implements...

---

Stanislav Kinsbursky (2):
nfsd: recovery - make rec_file per net
nfsd: recovery - make in_grace per net

 fs/nfsd/netns.h       |    3 ++
 fs/nfsd/nfs4recover.c |   79 ++++++++++++++++++++++++-------------------------
 2 files changed, 42 insertions(+), 40 deletions(-)