nfsd4: use idr for stateid's
The idr system is designed exactly for generating id and looking up
integer id's. Thanks to Trond for pointing it out.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index cb36c9a..a9e71cd 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -32,6 +32,7 @@
*
*/
+#include <linux/idr.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/slab.h>
@@ -49,7 +50,6 @@
time_t nfsd4_lease = 90; /* default lease time */
time_t nfsd4_grace = 90;
static time_t boot_time;
-static u32 current_stateid = 1;
static stateid_t zerostateid; /* bits all 0 */
static stateid_t onestateid; /* bits all 1 */
static u64 current_sessionid = 1;
@@ -149,10 +149,7 @@
#define FILE_HASH_BITS 8
#define FILE_HASH_SIZE (1 << FILE_HASH_BITS)
-/* hash table for (open)nfs4_ol_stateid */
-#define STATEID_HASH_BITS 10
-#define STATEID_HASH_SIZE (1 << STATEID_HASH_BITS)
-#define STATEID_HASH_MASK (STATEID_HASH_SIZE - 1)
+struct idr stateids;
static unsigned int file_hashval(struct inode *ino)
{
@@ -160,13 +157,7 @@
return hash_ptr(ino, FILE_HASH_BITS);
}
-static unsigned int stateid_hashval(stateid_t *s)
-{
- return opaque_hashval(&s->si_opaque, sizeof(stateid_opaque_t)) & STATEID_HASH_MASK;
-}
-
static struct list_head file_hashtbl[FILE_HASH_SIZE];
-static struct list_head stateid_hashtbl[STATEID_HASH_SIZE];
static void __nfs4_file_get_access(struct nfs4_file *fp, int oflag)
{
@@ -215,26 +206,52 @@
__nfs4_file_put_access(fp, oflag);
}
-static inline void hash_stid(struct nfs4_stid *stid)
+static inline int get_new_stid(struct nfs4_stid *stid)
{
- stateid_t *s = &stid->sc_stateid;
- unsigned int hashval;
+ static int min_stateid = 0;
+ int new_stid;
+ int error;
- hashval = stateid_hashval(s);
- list_add(&stid->sc_hash, &stateid_hashtbl[hashval]);
+ if (!idr_pre_get(&stateids, GFP_KERNEL))
+ return -ENOMEM;
+
+ error = idr_get_new_above(&stateids, stid, min_stateid, &new_stid);
+ /*
+ * All this code is currently serialized; the preallocation
+ * above should still be ours:
+ */
+ BUG_ON(error);
+ /*
+ * It shouldn't be a problem to reuse an opaque stateid value.
+ * I don't think it is for 4.1. But with 4.0 I worry that, for
+ * example, a stray write retransmission could be accepted by
+ * the server when it should have been rejected. Therefore,
+ * adopt a trick from the sctp code to attempt to maximize the
+ * amount of time until an id is reused, by ensuring they always
+ * "increase" (mod INT_MAX):
+ */
+
+ min_stateid = new_stid+1;
+ if (min_stateid == INT_MAX)
+ min_stateid = 0;
+ return new_stid;
}
-static void init_stid(struct nfs4_stid *stid, struct nfs4_client *cl, unsigned char type)
+static inline __be32 init_stid(struct nfs4_stid *stid, struct nfs4_client *cl, unsigned char type)
{
stateid_t *s = &stid->sc_stateid;
+ int new_id;
stid->sc_type = type;
stid->sc_client = cl;
s->si_opaque.so_clid = cl->cl_clientid;
- s->si_opaque.so_id = current_stateid++;
+ new_id = get_new_stid(stid);
+ if (new_id < 0)
+ return nfserr_jukebox;
+ s->si_opaque.so_id = (u32)new_id;
/* Will be incremented before return to client: */
s->si_generation = 0;
- hash_stid(stid);
+ return 0;
}
static struct nfs4_delegation *
@@ -242,6 +259,7 @@
{
struct nfs4_delegation *dp;
struct nfs4_file *fp = stp->st_file;
+ __be32 status;
dprintk("NFSD alloc_init_deleg\n");
/*
@@ -258,11 +276,15 @@
dp = kmem_cache_alloc(deleg_slab, GFP_KERNEL);
if (dp == NULL)
return dp;
- init_stid(&dp->dl_stid, clp, NFS4_DELEG_STID);
+ status = init_stid(&dp->dl_stid, clp, NFS4_DELEG_STID);
+ if (status) {
+ kmem_cache_free(deleg_slab, dp);
+ return NULL;
+ }
/*
* delegation seqid's are never incremented. The 4.1 special
- * meaning of seqid 0 isn't really meaningful, really, but let's
- * avoid 0 anyway just for consistency and use 1:
+ * meaning of seqid 0 isn't meaningful, really, but let's avoid
+ * 0 anyway just for consistency and use 1:
*/
dp->dl_stid.sc_stateid.si_generation = 1;
num_delegations++;
@@ -300,11 +322,16 @@
}
}
+static void unhash_stid(struct nfs4_stid *s)
+{
+ idr_remove(&stateids, s->sc_stateid.si_opaque.so_id);
+}
+
/* Called under the state lock. */
static void
unhash_delegation(struct nfs4_delegation *dp)
{
- list_del_init(&dp->dl_stid.sc_hash);
+ unhash_stid(&dp->dl_stid);
list_del_init(&dp->dl_perclnt);
spin_lock(&recall_lock);
list_del_init(&dp->dl_perfile);
@@ -457,7 +484,7 @@
struct file *file;
unhash_generic_stateid(stp);
- list_del(&stp->st_stid.sc_hash);
+ unhash_stid(&stp->st_stid);
file = find_any_file(stp->st_file);
if (file)
locks_remove_posix(file, (fl_owner_t)lockowner(stp->st_stateowner));
@@ -506,7 +533,7 @@
static void release_open_stateid(struct nfs4_ol_stateid *stp)
{
unhash_open_stateid(stp);
- list_del(&stp->st_stid.sc_hash);
+ unhash_stid(&stp->st_stid);
free_generic_stateid(stp);
}
@@ -528,7 +555,7 @@
struct nfs4_ol_stateid *s = oo->oo_last_closed_stid;
if (s) {
- list_del_init(&s->st_stid.sc_hash);
+ unhash_stid(&s->st_stid);
free_generic_stateid(s);
oo->oo_last_closed_stid = NULL;
}
@@ -1099,23 +1126,9 @@
*p++ = i++;
}
-static int
-same_stateid(stateid_t *id_one, stateid_t *id_two)
-{
- return 0 == memcmp(&id_one->si_opaque, &id_two->si_opaque,
- sizeof(stateid_opaque_t));
-}
-
static struct nfs4_stid *find_stateid(stateid_t *t)
{
- struct nfs4_stid *s;
- unsigned int hashval;
-
- hashval = stateid_hashval(t);
- list_for_each_entry(s, &stateid_hashtbl[hashval], sc_hash)
- if (same_stateid(&s->sc_stateid, t))
- return s;
- return NULL;
+ return idr_find(&stateids, t->si_opaque.so_id);
}
static struct nfs4_stid *find_stateid_by_type(stateid_t *t, char typemask)
@@ -2342,12 +2355,14 @@
return oo;
}
-static inline void
-init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
+static inline __be32 init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp, struct nfsd4_open *open) {
struct nfs4_openowner *oo = open->op_openowner;
struct nfs4_client *clp = oo->oo_owner.so_client;
+ __be32 status;
- init_stid(&stp->st_stid, clp, NFS4_OPEN_STID);
+ status = init_stid(&stp->st_stid, clp, NFS4_OPEN_STID);
+ if (status)
+ return status;
INIT_LIST_HEAD(&stp->st_lockowners);
list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
list_add(&stp->st_perfile, &fp->fi_stateids);
@@ -2360,6 +2375,7 @@
&stp->st_access_bmap);
__set_bit(open->op_share_deny, &stp->st_deny_bmap);
stp->st_openstp = NULL;
+ return nfs_ok;
}
static void
@@ -2949,7 +2965,11 @@
status = nfs4_new_open(rqstp, &stp, fp, current_fh, open);
if (status)
goto out;
- init_open_stateid(stp, fp, open);
+ status = init_open_stateid(stp, fp, open);
+ if (status) {
+ release_open_stateid(stp);
+ goto out;
+ }
status = nfsd4_truncate(rqstp, current_fh, open);
if (status) {
release_open_stateid(stp);
@@ -3824,11 +3844,16 @@
{
struct nfs4_ol_stateid *stp;
struct nfs4_client *clp = lo->lo_owner.so_client;
+ __be32 status;
stp = nfs4_alloc_stateid();
if (stp == NULL)
- goto out;
- init_stid(&stp->st_stid, clp, NFS4_LOCK_STID);
+ return NULL;
+ status = init_stid(&stp->st_stid, clp, NFS4_LOCK_STID);
+ if (status) {
+ free_generic_stateid(stp);
+ return NULL;
+ }
list_add(&stp->st_perfile, &fp->fi_stateids);
list_add(&stp->st_perstateowner, &lo->lo_owner.so_stateids);
stp->st_stateowner = &lo->lo_owner;
@@ -3837,8 +3862,6 @@
stp->st_access_bmap = 0;
stp->st_deny_bmap = open_stp->st_deny_bmap;
stp->st_openstp = open_stp;
-
-out:
return stp;
}
@@ -4386,8 +4409,7 @@
for (i = 0; i < OPEN_OWNER_HASH_SIZE; i++) {
INIT_LIST_HEAD(&open_ownerstr_hashtbl[i]);
}
- for (i = 0; i < STATEID_HASH_SIZE; i++)
- INIT_LIST_HEAD(&stateid_hashtbl[i]);
+ idr_init(&stateids);
for (i = 0; i < LOCK_HASH_SIZE; i++) {
INIT_LIST_HEAD(&lock_ownerstr_hashtbl[i]);
}
diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
index 70062b7..3ed5f99 100644
--- a/fs/nfsd/state.h
+++ b/fs/nfsd/state.h
@@ -79,7 +79,6 @@
/* For an open stateid kept around *only* to process close replays: */
#define NFS4_CLOSED_STID 8
unsigned char sc_type;
- struct list_head sc_hash;
stateid_t sc_stateid;
struct nfs4_client *sc_client;
};