Merge branch 'rhashtable-fixes'
Thomas Graf says: ==================== rhashtable rehashing fixes Some rhashtable rehashing bugs found while testing with the next rhashtable self-test queued up for the next devel cycle: https://github.com/tgraf/net-next/commits/rht v2: - Moved schedule_work() call into rhashtable_insert_rehash() ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
commit
a2c3935aaa
@ -405,13 +405,18 @@ int rhashtable_insert_rehash(struct rhashtable *ht)
|
||||
|
||||
if (rht_grow_above_75(ht, tbl))
|
||||
size *= 2;
|
||||
/* More than two rehashes (not resizes) detected. */
|
||||
else if (WARN_ON(old_tbl != tbl && old_tbl->size == size))
|
||||
/* Do not schedule more than one rehash */
|
||||
else if (old_tbl != tbl)
|
||||
return -EBUSY;
|
||||
|
||||
new_tbl = bucket_table_alloc(ht, size, GFP_ATOMIC);
|
||||
if (new_tbl == NULL)
|
||||
if (new_tbl == NULL) {
|
||||
/* Schedule async resize/rehash to try allocation
|
||||
* non-atomic context.
|
||||
*/
|
||||
schedule_work(&ht->run_work);
|
||||
return -ENOMEM;
|
||||
}
|
||||
|
||||
err = rhashtable_rehash_attach(ht, tbl, new_tbl);
|
||||
if (err) {
|
||||
|
Loading…
x
Reference in New Issue
Block a user