1 files changed, 228 insertions, 28 deletions

diff --git a/fs/nfs/nfs4filelayoutdev.c b/fs/nfs/nfs4filelayoutdev.c
index f5c9b125e8cc..68143c162e3b 100644
--- a/fs/nfs/nfs4filelayoutdev.c
+++ b/fs/nfs/nfs4filelayoutdev.c
@@ -37,6 +37,30 @@
 #define NFSDBG_FACILITY		NFSDBG_PNFS_LD
 
 /*
+ * Device ID RCU cache. A device ID is unique per client ID and layout type.
+ */
+#define NFS4_FL_DEVICE_ID_HASH_BITS	5
+#define NFS4_FL_DEVICE_ID_HASH_SIZE	(1 << NFS4_FL_DEVICE_ID_HASH_BITS)
+#define NFS4_FL_DEVICE_ID_HASH_MASK	(NFS4_FL_DEVICE_ID_HASH_SIZE - 1)
+
+static inline u32
+nfs4_fl_deviceid_hash(struct nfs4_deviceid *id)
+{
+	unsigned char *cptr = (unsigned char *)id->data;
+	unsigned int nbytes = NFS4_DEVICEID4_SIZE;
+	u32 x = 0;
+
+	while (nbytes--) {
+		x *= 37;
+		x += *cptr++;
+	}
+	return x & NFS4_FL_DEVICE_ID_HASH_MASK;
+}
+
+static struct hlist_head filelayout_deviceid_cache[NFS4_FL_DEVICE_ID_HASH_SIZE];
+static DEFINE_SPINLOCK(filelayout_deviceid_lock);
+
+/*
  * Data server cache
  *
  * Data servers can be mapped to different device ids.
@@ -104,6 +128,67 @@ _data_server_lookup_locked(u32 ip_addr, u32 port)
 	return NULL;
 }
 
+/*
+ * Create an rpc connection to the nfs4_pnfs_ds data server
+ * Currently only support IPv4
+ */
+static int
+nfs4_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds)
+{
+	struct nfs_client *clp;
+	struct sockaddr_in sin;
+	int status = 0;
+
+	dprintk("--> %s ip:port %x:%hu au_flavor %d\n", __func__,
+		ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
+		mds_srv->nfs_client->cl_rpcclient->cl_auth->au_flavor);
+
+	sin.sin_family = AF_INET;
+	sin.sin_addr.s_addr = ds->ds_ip_addr;
+	sin.sin_port = ds->ds_port;
+
+	clp = nfs4_set_ds_client(mds_srv->nfs_client, (struct sockaddr *)&sin,
+				 sizeof(sin), IPPROTO_TCP);
+	if (IS_ERR(clp)) {
+		status = PTR_ERR(clp);
+		goto out;
+	}
+
+	if ((clp->cl_exchange_flags & EXCHGID4_FLAG_MASK_PNFS) != 0) {
+		if (!is_ds_client(clp)) {
+			status = -ENODEV;
+			goto out_put;
+		}
+		ds->ds_clp = clp;
+		dprintk("%s [existing] ip=%x, port=%hu\n", __func__,
+			ntohl(ds->ds_ip_addr), ntohs(ds->ds_port));
+		goto out;
+	}
+
+	/*
+	 * Do not set NFS_CS_CHECK_LEASE_TIME instead set the DS lease to
+	 * be equal to the MDS lease. Renewal is scheduled in create_session.
+	 */
+	spin_lock(&mds_srv->nfs_client->cl_lock);
+	clp->cl_lease_time = mds_srv->nfs_client->cl_lease_time;
+	spin_unlock(&mds_srv->nfs_client->cl_lock);
+	clp->cl_last_renewal = jiffies;
+
+	/* New nfs_client */
+	status = nfs4_init_ds_session(clp);
+	if (status)
+		goto out_put;
+
+	ds->ds_clp = clp;
+	dprintk("%s [new] ip=%x, port=%hu\n", __func__, ntohl(ds->ds_ip_addr),
+		ntohs(ds->ds_port));
+out:
+	return status;
+out_put:
+	nfs_put_client(clp);
+	goto out;
+}
+
 static void
 destroy_ds(struct nfs4_pnfs_ds *ds)
 {
@@ -122,7 +207,7 @@ nfs4_fl_free_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
 	struct nfs4_pnfs_ds *ds;
 	int i;
 
-	print_deviceid(&dsaddr->deviceid.de_id);
+	print_deviceid(&dsaddr->deviceid);
 
 	for (i = 0; i < dsaddr->ds_num; i++) {
 		ds = dsaddr->ds_list[i];
@@ -139,15 +224,6 @@ nfs4_fl_free_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
 	kfree(dsaddr);
 }
 
-void
-nfs4_fl_free_deviceid_callback(struct pnfs_deviceid_node *device)
-{
-	struct nfs4_file_layout_dsaddr *dsaddr =
-		container_of(device, struct nfs4_file_layout_dsaddr, deviceid);
-
-	nfs4_fl_free_deviceid(dsaddr);
-}
-
 static struct nfs4_pnfs_ds *
 nfs4_pnfs_ds_add(struct inode *inode, u32 ip_addr, u32 port)
 {
@@ -219,6 +295,10 @@ decode_and_add_ds(__be32 **pp, struct inode *inode)
 		goto out_err;
 	}
 	buf = kmalloc(rlen + 1, GFP_KERNEL);
+	if (!buf) {
+		dprintk("%s: Not enough memory\n", __func__);
+		goto out_err;
+	}
 	buf[rlen] = '\0';
 	memcpy(buf, r_addr, rlen);
 
@@ -296,7 +376,7 @@ decode_device(struct inode *ino, struct pnfs_device *pdev)
 	dsaddr->stripe_count = cnt;
 	dsaddr->ds_num = num;
 
-	memcpy(&dsaddr->deviceid.de_id, &pdev->dev_id, sizeof(pdev->dev_id));
+	memcpy(&dsaddr->deviceid, &pdev->dev_id, sizeof(pdev->dev_id));
 
 	/* Go back an read stripe indices */
 	p = indicesp;
@@ -346,28 +426,37 @@ out_err:
 }
 
 /*
- * Decode the opaque device specified in 'dev'
- * and add it to the list of available devices.
- * If the deviceid is already cached, nfs4_add_deviceid will return
- * a pointer to the cached struct and throw away the new.
+ * Decode the opaque device specified in 'dev' and add it to the cache of
+ * available devices.
  */
-static struct nfs4_file_layout_dsaddr*
+static struct nfs4_file_layout_dsaddr *
 decode_and_add_device(struct inode *inode, struct pnfs_device *dev)
 {
-	struct nfs4_file_layout_dsaddr *dsaddr;
-	struct pnfs_deviceid_node *d;
+	struct nfs4_file_layout_dsaddr *d, *new;
+	long hash;
 
-	dsaddr = decode_device(inode, dev);
-	if (!dsaddr) {
+	new = decode_device(inode, dev);
+	if (!new) {
 		printk(KERN_WARNING "%s: Could not decode or add device\n",
 			__func__);
 		return NULL;
 	}
 
-	d = pnfs_add_deviceid(NFS_SERVER(inode)->nfs_client->cl_devid_cache,
-			      &dsaddr->deviceid);
+	spin_lock(&filelayout_deviceid_lock);
+	d = nfs4_fl_find_get_deviceid(&new->deviceid);
+	if (d) {
+		spin_unlock(&filelayout_deviceid_lock);
+		nfs4_fl_free_deviceid(new);
+		return d;
+	}
+
+	INIT_HLIST_NODE(&new->node);
+	atomic_set(&new->ref, 1);
+	hash = nfs4_fl_deviceid_hash(&new->deviceid);
+	hlist_add_head_rcu(&new->node, &filelayout_deviceid_cache[hash]);
+	spin_unlock(&filelayout_deviceid_lock);
 
-	return container_of(d, struct nfs4_file_layout_dsaddr, deviceid);
+	return new;
 }
 
 /*
@@ -442,12 +531,123 @@ out_free:
 	return dsaddr;
 }
 
+void
+nfs4_fl_put_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
+{
+	if (atomic_dec_and_lock(&dsaddr->ref, &filelayout_deviceid_lock)) {
+		hlist_del_rcu(&dsaddr->node);
+		spin_unlock(&filelayout_deviceid_lock);
+
+		synchronize_rcu();
+		nfs4_fl_free_deviceid(dsaddr);
+	}
+}
+
 struct nfs4_file_layout_dsaddr *
-nfs4_fl_find_get_deviceid(struct nfs_client *clp, struct nfs4_deviceid *id)
+nfs4_fl_find_get_deviceid(struct nfs4_deviceid *id)
+{
+	struct nfs4_file_layout_dsaddr *d;
+	struct hlist_node *n;
+	long hash = nfs4_fl_deviceid_hash(id);
+
+
+	rcu_read_lock();
+	hlist_for_each_entry_rcu(d, n, &filelayout_deviceid_cache[hash], node) {
+		if (!memcmp(&d->deviceid, id, sizeof(*id))) {
+			if (!atomic_inc_not_zero(&d->ref))
+				goto fail;
+			rcu_read_unlock();
+			return d;
+		}
+	}
+fail:
+	rcu_read_unlock();
+	return NULL;
+}
+
+/*
+ * Want res = (offset - layout->pattern_offset)/ layout->stripe_unit
+ * Then: ((res + fsi) % dsaddr->stripe_count)
+ */
+u32
+nfs4_fl_calc_j_index(struct pnfs_layout_segment *lseg, loff_t offset)
+{
+	struct nfs4_filelayout_segment *flseg = FILELAYOUT_LSEG(lseg);
+	u64 tmp;
+
+	tmp = offset - flseg->pattern_offset;
+	do_div(tmp, flseg->stripe_unit);
+	tmp += flseg->first_stripe_index;
+	return do_div(tmp, flseg->dsaddr->stripe_count);
+}
+
+u32
+nfs4_fl_calc_ds_index(struct pnfs_layout_segment *lseg, u32 j)
 {
-	struct pnfs_deviceid_node *d;
+	return FILELAYOUT_LSEG(lseg)->dsaddr->stripe_indices[j];
+}
 
-	d = pnfs_find_get_deviceid(clp->cl_devid_cache, id);
-	return (d == NULL) ? NULL :
-		container_of(d, struct nfs4_file_layout_dsaddr, deviceid);
+struct nfs_fh *
+nfs4_fl_select_ds_fh(struct pnfs_layout_segment *lseg, u32 j)
+{
+	struct nfs4_filelayout_segment *flseg = FILELAYOUT_LSEG(lseg);
+	u32 i;
+
+	if (flseg->stripe_type == STRIPE_SPARSE) {
+		if (flseg->num_fh == 1)
+			i = 0;
+		else if (flseg->num_fh == 0)
+			/* Use the MDS OPEN fh set in nfs_read_rpcsetup */
+			return NULL;
+		else
+			i = nfs4_fl_calc_ds_index(lseg, j);
+	} else
+		i = j;
+	return flseg->fh_array[i];
+}
+
+static void
+filelayout_mark_devid_negative(struct nfs4_file_layout_dsaddr *dsaddr,
+			       int err, u32 ds_addr)
+{
+	u32 *p = (u32 *)&dsaddr->deviceid;
+
+	printk(KERN_ERR "NFS: data server %x connection error %d."
+		" Deviceid [%x%x%x%x] marked out of use.\n",
+		ds_addr, err, p[0], p[1], p[2], p[3]);
+
+	spin_lock(&filelayout_deviceid_lock);
+	dsaddr->flags |= NFS4_DEVICE_ID_NEG_ENTRY;
+	spin_unlock(&filelayout_deviceid_lock);
+}
+
+struct nfs4_pnfs_ds *
+nfs4_fl_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx)
+{
+	struct nfs4_file_layout_dsaddr *dsaddr = FILELAYOUT_LSEG(lseg)->dsaddr;
+	struct nfs4_pnfs_ds *ds = dsaddr->ds_list[ds_idx];
+
+	if (ds == NULL) {
+		printk(KERN_ERR "%s: No data server for offset index %d\n",
+			__func__, ds_idx);
+		return NULL;
+	}
+
+	if (!ds->ds_clp) {
+		struct nfs_server *s = NFS_SERVER(lseg->pls_layout->plh_inode);
+		int err;
+
+		if (dsaddr->flags & NFS4_DEVICE_ID_NEG_ENTRY) {
+			/* Already tried to connect, don't try again */
+			dprintk("%s Deviceid marked out of use\n", __func__);
+			return NULL;
+		}
+		err = nfs4_ds_connect(s, ds);
+		if (err) {
+			filelayout_mark_devid_negative(dsaddr, err,
+						       ntohl(ds->ds_ip_addr));
+			return NULL;
+		}
+	}
+	return ds;
 }