From: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
To: netfilter-devel@vger.kernel.org
Cc: Pablo Neira Ayuso <pablo@netfilter.org>
Subject: [PATCH 2/8] netfilter: ipset: Prepare ipset core for RCU locking instead of rwlock per set
Date: Fri, 12 Dec 2014 22:58:03 +0100 [thread overview]
Message-ID: <1418421489-17411-3-git-send-email-kadlec@blackhole.kfki.hu> (raw)
In-Reply-To: <1418421489-17411-1-git-send-email-kadlec@blackhole.kfki.hu>
Signed-off-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
---
include/linux/netfilter/ipset/ip_set.h | 2 +-
include/linux/netfilter/ipset/ip_set_timeout.h | 27 +++++------
net/netfilter/ipset/ip_set_core.c | 66 ++++++++++++++------------
3 files changed, 47 insertions(+), 48 deletions(-)
diff --git a/include/linux/netfilter/ipset/ip_set.h b/include/linux/netfilter/ipset/ip_set.h
index f1606fa..b3fe9f9 100644
--- a/include/linux/netfilter/ipset/ip_set.h
+++ b/include/linux/netfilter/ipset/ip_set.h
@@ -223,7 +223,7 @@ struct ip_set {
/* The name of the set */
char name[IPSET_MAXNAMELEN];
/* Lock protecting the set data */
- rwlock_t lock;
+ spinlock_t lock;
/* References to the set */
u32 ref;
/* The core set type */
diff --git a/include/linux/netfilter/ipset/ip_set_timeout.h b/include/linux/netfilter/ipset/ip_set_timeout.h
index 83c2f9e..1d6a935 100644
--- a/include/linux/netfilter/ipset/ip_set_timeout.h
+++ b/include/linux/netfilter/ipset/ip_set_timeout.h
@@ -40,38 +40,33 @@ ip_set_timeout_uget(struct nlattr *tb)
}
static inline bool
-ip_set_timeout_test(unsigned long timeout)
+ip_set_timeout_expired(unsigned long *t)
{
- return timeout == IPSET_ELEM_PERMANENT ||
- time_is_after_jiffies(timeout);
-}
-
-static inline bool
-ip_set_timeout_expired(unsigned long *timeout)
-{
- return *timeout != IPSET_ELEM_PERMANENT &&
- time_is_before_jiffies(*timeout);
+ return *t != IPSET_ELEM_PERMANENT && time_is_before_jiffies(*t);
}
static inline void
-ip_set_timeout_set(unsigned long *timeout, u32 t)
+ip_set_timeout_set(unsigned long *timeout, u32 value)
{
- if (!t) {
+ unsigned long t;
+
+ if (!value) {
*timeout = IPSET_ELEM_PERMANENT;
return;
}
- *timeout = msecs_to_jiffies(t * 1000) + jiffies;
- if (*timeout == IPSET_ELEM_PERMANENT)
+ t = msecs_to_jiffies(value * MSEC_PER_SEC) + jiffies;
+ if (t == IPSET_ELEM_PERMANENT)
/* Bingo! :-) */
- (*timeout)--;
+ t--;
+ *timeout = t;
}
static inline u32
ip_set_timeout_get(unsigned long *timeout)
{
return *timeout == IPSET_ELEM_PERMANENT ? 0 :
- jiffies_to_msecs(*timeout - jiffies)/1000;
+ jiffies_to_msecs(*timeout - jiffies)/MSEC_PER_SEC;
}
#endif /* __KERNEL__ */
diff --git a/net/netfilter/ipset/ip_set_core.c b/net/netfilter/ipset/ip_set_core.c
index d259da3..ef24cdb 100644
--- a/net/netfilter/ipset/ip_set_core.c
+++ b/net/netfilter/ipset/ip_set_core.c
@@ -42,7 +42,7 @@ static inline struct ip_set_net *ip_set_pernet(struct net *net)
}
#define IP_SET_INC 64
-#define STREQ(a, b) (strncmp(a, b, IPSET_MAXNAMELEN) == 0)
+#define STRNCMP(a, b) (strncmp(a, b, IPSET_MAXNAMELEN) == 0)
static unsigned int max_sets;
@@ -85,7 +85,7 @@ find_set_type(const char *name, u8 family, u8 revision)
struct ip_set_type *type;
list_for_each_entry_rcu(type, &ip_set_type_list, list)
- if (STREQ(type->name, name) &&
+ if (STRNCMP(type->name, name) &&
(type->family == family ||
type->family == NFPROTO_UNSPEC) &&
revision >= type->revision_min &&
@@ -132,7 +132,7 @@ __find_set_type_get(const char *name, u8 family, u8 revision,
/* Make sure the type is already loaded
* but we don't support the revision */
list_for_each_entry_rcu(type, &ip_set_type_list, list)
- if (STREQ(type->name, name)) {
+ if (STRNCMP(type->name, name)) {
err = -IPSET_ERR_FIND_TYPE;
goto unlock;
}
@@ -166,7 +166,7 @@ __find_set_type_minmax(const char *name, u8 family, u8 *min, u8 *max,
*min = 255; *max = 0;
rcu_read_lock();
list_for_each_entry_rcu(type, &ip_set_type_list, list)
- if (STREQ(type->name, name) &&
+ if (STRNCMP(type->name, name) &&
(type->family == family ||
type->family == NFPROTO_UNSPEC)) {
found = true;
@@ -208,15 +208,15 @@ ip_set_type_register(struct ip_set_type *type)
pr_warn("ip_set type %s, family %s with revision min %u already registered!\n",
type->name, family_name(type->family),
type->revision_min);
- ret = -EINVAL;
- goto unlock;
+ ip_set_type_unlock();
+ return -EINVAL;
}
list_add_rcu(&type->list, &ip_set_type_list);
pr_debug("type %s, family %s, revision %u:%u registered.\n",
type->name, family_name(type->family),
type->revision_min, type->revision_max);
-unlock:
ip_set_type_unlock();
+ synchronize_rcu();
return ret;
}
EXPORT_SYMBOL_GPL(ip_set_type_register);
@@ -230,12 +230,12 @@ ip_set_type_unregister(struct ip_set_type *type)
pr_warn("ip_set type %s, family %s with revision min %u not registered\n",
type->name, family_name(type->family),
type->revision_min);
- goto unlock;
+ ip_set_type_unlock();
+ return;
}
list_del_rcu(&type->list);
pr_debug("type %s, family %s with revision min %u unregistered.\n",
type->name, family_name(type->family), type->revision_min);
-unlock:
ip_set_type_unlock();
synchronize_rcu();
@@ -496,16 +496,16 @@ ip_set_test(ip_set_id_t index, const struct sk_buff *skb,
!(opt->family == set->family || set->family == NFPROTO_UNSPEC))
return 0;
- read_lock_bh(&set->lock);
+ rcu_read_lock_bh();
ret = set->variant->kadt(set, skb, par, IPSET_TEST, opt);
- read_unlock_bh(&set->lock);
+ rcu_read_unlock_bh();
if (ret == -EAGAIN) {
/* Type requests element to be completed */
pr_debug("element must be completed, ADD is triggered\n");
- write_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
set->variant->kadt(set, skb, par, IPSET_ADD, opt);
- write_unlock_bh(&set->lock);
+ spin_unlock_bh(&set->lock);
ret = 1;
} else {
/* --return-nomatch: invert matched element */
@@ -535,9 +535,9 @@ ip_set_add(ip_set_id_t index, const struct sk_buff *skb,
!(opt->family == set->family || set->family == NFPROTO_UNSPEC))
return -IPSET_ERR_TYPE_MISMATCH;
- write_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
ret = set->variant->kadt(set, skb, par, IPSET_ADD, opt);
- write_unlock_bh(&set->lock);
+ spin_unlock_bh(&set->lock);
return ret;
}
@@ -558,9 +558,9 @@ ip_set_del(ip_set_id_t index, const struct sk_buff *skb,
!(opt->family == set->family || set->family == NFPROTO_UNSPEC))
return -IPSET_ERR_TYPE_MISMATCH;
- write_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
ret = set->variant->kadt(set, skb, par, IPSET_DEL, opt);
- write_unlock_bh(&set->lock);
+ spin_unlock_bh(&set->lock);
return ret;
}
@@ -581,7 +581,7 @@ ip_set_get_byname(struct net *net, const char *name, struct ip_set **set)
rcu_read_lock();
for (i = 0; i < inst->ip_set_max; i++) {
s = rcu_dereference(inst->ip_set_list)[i];
- if (s != NULL && STREQ(s->name, name)) {
+ if (s != NULL && STRNCMP(s->name, name)) {
__ip_set_get(s);
index = i;
*set = s;
@@ -758,7 +758,7 @@ find_set_and_id(struct ip_set_net *inst, const char *name, ip_set_id_t *id)
*id = IPSET_INVALID_ID;
for (i = 0; i < inst->ip_set_max; i++) {
set = ip_set(inst, i);
- if (set != NULL && STREQ(set->name, name)) {
+ if (set != NULL && STRNCMP(set->name, name)) {
*id = i;
break;
}
@@ -787,7 +787,7 @@ find_free_id(struct ip_set_net *inst, const char *name, ip_set_id_t *index,
if (s == NULL) {
if (*index == IPSET_INVALID_ID)
*index = i;
- } else if (STREQ(name, s->name)) {
+ } else if (STRNCMP(name, s->name)) {
/* Name clash */
*set = s;
return -EEXIST;
@@ -845,7 +845,7 @@ ip_set_create(struct sock *ctnl, struct sk_buff *skb,
set = kzalloc(sizeof(struct ip_set), GFP_KERNEL);
if (!set)
return -ENOMEM;
- rwlock_init(&set->lock);
+ spin_lock_init(&set->lock);
strlcpy(set->name, name, IPSET_MAXNAMELEN);
set->family = family;
set->revision = revision;
@@ -887,7 +887,7 @@ ip_set_create(struct sock *ctnl, struct sk_buff *skb,
if (ret == -EEXIST) {
/* If this is the same set and requested, ignore error */
if ((flags & IPSET_FLAG_EXIST) &&
- STREQ(set->type->name, clash->type->name) &&
+ STRNCMP(set->type->name, clash->type->name) &&
set->type->family == clash->type->family &&
set->type->revision_min == clash->type->revision_min &&
set->type->revision_max == clash->type->revision_max &&
@@ -1024,9 +1024,9 @@ ip_set_flush_set(struct ip_set *set)
{
pr_debug("set: %s\n", set->name);
- write_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
set->variant->flush(set);
- write_unlock_bh(&set->lock);
+ spin_unlock_bh(&set->lock);
}
static int
@@ -1098,7 +1098,7 @@ ip_set_rename(struct sock *ctnl, struct sk_buff *skb,
name2 = nla_data(attr[IPSET_ATTR_SETNAME2]);
for (i = 0; i < inst->ip_set_max; i++) {
s = ip_set(inst, i);
- if (s != NULL && STREQ(s->name, name2)) {
+ if (s != NULL && STRNCMP(s->name, name2)) {
ret = -IPSET_ERR_EXIST_SETNAME2;
goto out;
}
@@ -1327,9 +1327,9 @@ dump_last:
goto next_set;
/* Fall through and add elements */
default:
- read_lock_bh(&set->lock);
+ rcu_read_lock_bh();
ret = set->variant->list(set, skb, cb);
- read_unlock_bh(&set->lock);
+ rcu_read_unlock_bh();
if (!cb->args[IPSET_CB_ARG0])
/* Set is done, proceed with next one */
goto next_set;
@@ -1407,9 +1407,13 @@ call_ad(struct sock *ctnl, struct sk_buff *skb, struct ip_set *set,
bool eexist = flags & IPSET_FLAG_EXIST, retried = false;
do {
- write_lock_bh(&set->lock);
+ spin_lock_bh(&set->lock);
ret = set->variant->uadt(set, tb, adt, &lineno, flags, retried);
- write_unlock_bh(&set->lock);
+ spin_unlock_bh(&set->lock);
+ if (ret == -EINPROGRESS) {
+ synchronize_rcu_bh();
+ ret = 0;
+ }
retried = true;
} while (ret == -EAGAIN &&
set->variant->resize &&
@@ -1589,9 +1593,9 @@ ip_set_utest(struct sock *ctnl, struct sk_buff *skb,
set->type->adt_policy))
return -IPSET_ERR_PROTOCOL;
- read_lock_bh(&set->lock);
+ rcu_read_lock_bh();
ret = set->variant->uadt(set, tb, IPSET_TEST, NULL, 0, 0);
- read_unlock_bh(&set->lock);
+ rcu_read_unlock_bh();
/* Userspace can't trigger element to be re-added */
if (ret == -EAGAIN)
ret = 1;
--
1.8.5.1
next prev parent reply other threads:[~2014-12-12 21:57 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-12-12 21:58 [PATCH 0/8] ipset patches for nf-next, v3 Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 1/8] netfilter: ipset: Remove rbtree from hash:net,iface in order to run under RCU Jozsef Kadlecsik
2014-12-12 21:58 ` Jozsef Kadlecsik [this message]
2014-12-18 18:38 ` [PATCH 2/8] netfilter: ipset: Prepare ipset core for RCU locking instead of rwlock per set Pablo Neira Ayuso
2014-12-21 11:52 ` Jozsef Kadlecsik
2014-12-21 12:23 ` Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 3/8] netfilter: ipset: Introduce RCU locking in the bitmap types Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 4/8] netfilter: ipset: Introduce RCU locking in the list type Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 5/8] netfilter: ipset: Introduce RCU locking in the hash types Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 6/8] netfilter: ipset: styles warned by checkpatch.pl fixed Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 7/8] netfilter: ipset: Fix parallel resizing and listing of the same set Jozsef Kadlecsik
2014-12-12 21:58 ` [PATCH 8/8] netfilter: ipset: Fix sparse warning Jozsef Kadlecsik
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1418421489-17411-3-git-send-email-kadlec@blackhole.kfki.hu \
--to=kadlec@blackhole.kfki.hu \
--cc=netfilter-devel@vger.kernel.org \
--cc=pablo@netfilter.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).