Merge branch 'rhashtable-New-features-in-walk-and-bucket'

Tom Herbert says:

====================
rhashtable: New features in walk and bucket

This patch contains some changes to related rhashtable:

- Above allow rhashtable_walk_start to return void
- Add a functon to peek at the next entry during a walk
- Abstract out function to compute a has for a table
- A library function to alloc a spinlocks bucket array
- Call the above function for rhashtable locks allocation

Tested: Exercised using various operations on an ILA xlat
table.

v2:
 - Apply feedback from Herbert. Don't change semantics of resize
   event reporting and -EAGAIN, just simplify API for callers that
   ignore those.
 - Add end_of_table in iter to reliably tell when the iterator has
   reached to the eno.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/broadcom/bnxt/bnxt_tc.c

@@ -1412,11 +1412,7 @@ bnxt_tc_flow_stats_batch_prep(struct bnxt *bp,
 	void *flow_node;
 	int rc, i;
 
-	rc = rhashtable_walk_start(iter);
-	if (rc && rc != -EAGAIN) {
-		i = 0;
-		goto done;
-	}
+	rhashtable_walk_start(iter);
 
 	rc = 0;
 	for (i = 0; i < BNXT_FLOW_STATS_BATCH_MAX; i++) {

drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c

@@ -763,9 +763,7 @@ static void ch_flower_stats_handler(struct work_struct *work)
 
 	rhashtable_walk_enter(&adap->flower_tbl, &iter);
 	do {
-		flower_entry = ERR_PTR(rhashtable_walk_start(&iter));
-		if (IS_ERR(flower_entry))
-			goto walk_stop;
+		rhashtable_walk_start(&iter);
 
 		while ((flower_entry = rhashtable_walk_next(&iter)) &&
 		       !IS_ERR(flower_entry)) {
@@ -784,8 +782,9 @@ static void ch_flower_stats_handler(struct work_struct *work)
 				spin_unlock(&flower_entry->lock);
 			}
 		}
-walk_stop:
+
 		rhashtable_walk_stop(&iter);
+
 	} while (flower_entry == ERR_PTR(-EAGAIN));
 	rhashtable_walk_exit(&iter);
 	mod_timer(&adap->flower_stats_timer, jiffies + STATS_CHECK_PERIOD);

fs/gfs2/glock.c

@@ -1549,16 +1549,13 @@ static void glock_hash_walk(glock_examiner examiner, const struct gfs2_sbd *sdp)
 	rhashtable_walk_enter(&gl_hash_table, &iter);
 
 	do {
-		gl = ERR_PTR(rhashtable_walk_start(&iter));
-		if (IS_ERR(gl))
-			goto walk_stop;
+		rhashtable_walk_start(&iter);
 
 		while ((gl = rhashtable_walk_next(&iter)) && !IS_ERR(gl))
 			if (gl->gl_name.ln_sbd == sdp &&
 			    lockref_get_not_dead(&gl->gl_lockref))
 				examiner(gl);
 
-walk_stop:
 		rhashtable_walk_stop(&iter);
 	} while (cond_resched(), gl == ERR_PTR(-EAGAIN));
 
@@ -1947,7 +1944,7 @@ static void *gfs2_glock_seq_start(struct seq_file *seq, loff_t *pos)
 	loff_t n = *pos;
 
 	rhashtable_walk_enter(&gl_hash_table, &gi->hti);
-	if (rhashtable_walk_start(&gi->hti) != 0)
+	if (rhashtable_walk_start_check(&gi->hti) != 0)
 		return NULL;
 
 	do {

include/linux/rhashtable.h

@@ -207,6 +207,7 @@ struct rhashtable_iter {
 	struct rhashtable_walker walker;
 	unsigned int slot;
 	unsigned int skip;
+	bool end_of_table;
 };
 
 static inline unsigned long rht_marker(const struct rhashtable *ht, u32 hash)
@@ -239,34 +240,42 @@ static inline unsigned int rht_bucket_index(const struct bucket_table *tbl,
 	return (hash >> RHT_HASH_RESERVED_SPACE) & (tbl->size - 1);
 }
 
-static inline unsigned int rht_key_hashfn(
-	struct rhashtable *ht, const struct bucket_table *tbl,
-	const void *key, const struct rhashtable_params params)
+static inline unsigned int rht_key_get_hash(struct rhashtable *ht,
+	const void *key, const struct rhashtable_params params,
+	unsigned int hash_rnd)
 {
 	unsigned int hash;
 
 	/* params must be equal to ht->p if it isn't constant. */
 	if (!__builtin_constant_p(params.key_len))
-		hash = ht->p.hashfn(key, ht->key_len, tbl->hash_rnd);
+		hash = ht->p.hashfn(key, ht->key_len, hash_rnd);
 	else if (params.key_len) {
 		unsigned int key_len = params.key_len;
 
 		if (params.hashfn)
-			hash = params.hashfn(key, key_len, tbl->hash_rnd);
+			hash = params.hashfn(key, key_len, hash_rnd);
 		else if (key_len & (sizeof(u32) - 1))
-			hash = jhash(key, key_len, tbl->hash_rnd);
+			hash = jhash(key, key_len, hash_rnd);
 		else
-			hash = jhash2(key, key_len / sizeof(u32),
-				      tbl->hash_rnd);
+			hash = jhash2(key, key_len / sizeof(u32), hash_rnd);
 	} else {
 		unsigned int key_len = ht->p.key_len;
 
 		if (params.hashfn)
-			hash = params.hashfn(key, key_len, tbl->hash_rnd);
+			hash = params.hashfn(key, key_len, hash_rnd);
 		else
-			hash = jhash(key, key_len, tbl->hash_rnd);
+			hash = jhash(key, key_len, hash_rnd);
 	}
 
+	return hash;
+}
+
+static inline unsigned int rht_key_hashfn(
+	struct rhashtable *ht, const struct bucket_table *tbl,
+	const void *key, const struct rhashtable_params params)
+{
+	unsigned int hash = rht_key_get_hash(ht, key, params, tbl->hash_rnd);
+
 	return rht_bucket_index(tbl, hash);
 }
 
@@ -378,8 +387,15 @@ void *rhashtable_insert_slow(struct rhashtable *ht, const void *key,
 void rhashtable_walk_enter(struct rhashtable *ht,
 			   struct rhashtable_iter *iter);
 void rhashtable_walk_exit(struct rhashtable_iter *iter);
-int rhashtable_walk_start(struct rhashtable_iter *iter) __acquires(RCU);
+int rhashtable_walk_start_check(struct rhashtable_iter *iter) __acquires(RCU);
+
+static inline void rhashtable_walk_start(struct rhashtable_iter *iter)
+{
+	(void)rhashtable_walk_start_check(iter);
+}
+
 void *rhashtable_walk_next(struct rhashtable_iter *iter);
+void *rhashtable_walk_peek(struct rhashtable_iter *iter);
 void rhashtable_walk_stop(struct rhashtable_iter *iter) __releases(RCU);
 
 void rhashtable_free_and_destroy(struct rhashtable *ht,

include/linux/spinlock.h

@@ -414,4 +414,10 @@ extern int _atomic_dec_and_lock(atomic_t *atomic, spinlock_t *lock);
 #define atomic_dec_and_lock(atomic, lock) \
 		__cond_lock(lock, _atomic_dec_and_lock(atomic, lock))
 
+int alloc_bucket_spinlocks(spinlock_t **locks, unsigned int *lock_mask,
+			   size_t max_size, unsigned int cpu_mult,
+			   gfp_t gfp);
+
+void free_bucket_spinlocks(spinlock_t *locks);
+
 #endif /* __LINUX_SPINLOCK_H */

include/net/sctp/sctp.h

@@ -116,7 +116,7 @@ extern struct percpu_counter sctp_sockets_allocated;
 int sctp_asconf_mgmt(struct sctp_sock *, struct sctp_sockaddr_entry *);
 struct sk_buff *sctp_skb_recv_datagram(struct sock *, int, int, int *);
 
-int sctp_transport_walk_start(struct rhashtable_iter *iter);
+void sctp_transport_walk_start(struct rhashtable_iter *iter);
 void sctp_transport_walk_stop(struct rhashtable_iter *iter);
 struct sctp_transport *sctp_transport_get_next(struct net *net,
 			struct rhashtable_iter *iter);

lib/Makefile

@@ -39,7 +39,7 @@ obj-y += bcd.o div64.o sort.o parser.o debug_locks.o random32.o \
 	 gcd.o lcm.o list_sort.o uuid.o flex_array.o iov_iter.o clz_ctz.o \
 	 bsearch.o find_bit.o llist.o memweight.o kfifo.o \
 	 percpu-refcount.o percpu_ida.o rhashtable.o reciprocal_div.o \
-	 once.o refcount.o usercopy.o errseq.o
+	 once.o refcount.o usercopy.o errseq.o bucket_locks.o
 obj-$(CONFIG_STRING_SELFTEST) += test_string.o
 obj-y += string_helpers.o
 obj-$(CONFIG_TEST_STRING_HELPERS) += test-string_helpers.o

lib/bucket_locks.c

@@ -0,0 +1,54 @@
+#include <linux/export.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+
+/* Allocate an array of spinlocks to be accessed by a hash. Two arguments
+ * indicate the number of elements to allocate in the array. max_size
+ * gives the maximum number of elements to allocate. cpu_mult gives
+ * the number of locks per CPU to allocate. The size is rounded up
+ * to a power of 2 to be suitable as a hash table.
+ */
+
+int alloc_bucket_spinlocks(spinlock_t **locks, unsigned int *locks_mask,
+			   size_t max_size, unsigned int cpu_mult, gfp_t gfp)
+{
+	spinlock_t *tlocks = NULL;
+	unsigned int i, size;
+#if defined(CONFIG_PROVE_LOCKING)
+	unsigned int nr_pcpus = 2;
+#else
+	unsigned int nr_pcpus = num_possible_cpus();
+#endif
+
+	if (cpu_mult) {
+		nr_pcpus = min_t(unsigned int, nr_pcpus, 64UL);
+		size = min_t(unsigned int, nr_pcpus * cpu_mult, max_size);
+	} else {
+		size = max_size;
+	}
+
+	if (sizeof(spinlock_t) != 0) {
+		if (gfpflags_allow_blocking(gfp))
+			tlocks = kvmalloc(size * sizeof(spinlock_t), gfp);
+		else
+			tlocks = kmalloc_array(size, sizeof(spinlock_t), gfp);
+		if (!tlocks)
+			return -ENOMEM;
+		for (i = 0; i < size; i++)
+			spin_lock_init(&tlocks[i]);
+	}
+
+	*locks = tlocks;
+	*locks_mask = size - 1;
+
+	return 0;
+}
+EXPORT_SYMBOL(alloc_bucket_spinlocks);
+
+void free_bucket_spinlocks(spinlock_t *locks)
+{
+	kvfree(locks);
+}
+EXPORT_SYMBOL(free_bucket_spinlocks);