* [GIT PULL nf-next] IPVS optimisations for v3.10
@ 2013-03-29 4:11 Simon Horman
2013-03-29 4:11 ` [PATCH 01/34] net: add skb_dst_set_noref_force Simon Horman
` (34 more replies)
0 siblings, 35 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang, Julian Anastasov
Hi Pablo,
The following changes since commit e5c5d22e8dcf7c2d430336cbf8e180bd38e8daf1:
net: add ETH_P_802_3_MIN (2013-03-28 01:20:42 -0400)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/horms/ipvs-next.git tags/ipvs2-for-v3.10
for you to fetch changes up to e0566a7cef2f1afca4aa98244224fb9ca6ded08e:
ipvs: do not disable bh for long time (2013-03-29 12:59:02 +0900)
----------------------------------------------------------------
IPVS optimisations for v3.10 by Julian Anastasov
----------------------------------------------------------------
Julian Anastasov (34):
net: add skb_dst_set_noref_force
ipvs: avoid routing by TOS for real server
ipvs: prefer NETDEV_DOWN event to free cached dsts
ipvs: convert the IP_VS_XMIT macros to functions
ipvs: rename functions related to dst_cache reset
ipvs: no need to reroute anymore on DNAT over loopback
ipvs: do not use skb_share_check
ipvs: consolidate all dst checks on transmit in one place
ipvs: optimize dst usage for real server
ipvs: convert app locks
ipvs: remove rs_lock by using RCU
ipvs: convert locks used in persistence engines
ipvs: convert connection locking
ipvs: reorder keys in connection structure
ipvs: avoid kmem_cache_zalloc in ip_vs_conn_new
ipvs: change ip_vs_sched_lock to mutex
ipvs: preparations for using rcu in schedulers
ipvs: add ip_vs_dest_hold and ip_vs_dest_put
ipvs: convert dh scheduler to rcu
ipvs: convert lblc scheduler to rcu
ipvs: convert lblcr scheduler to rcu
ipvs: convert lc scheduler to rcu
ipvs: convert nq scheduler to rcu
ipvs: convert rr scheduler to rcu
ipvs: convert sed scheduler to rcu
ipvs: convert sh scheduler to rcu
ipvs: convert wlc scheduler to rcu
ipvs: convert wrr scheduler to rcu
ipvs: reorganize dest trash
ipvs: do not expect result from done_service
ipvs: convert sched_lock to spin lock
ipvs: convert dests to rcu
ipvs: convert services to rcu
ipvs: do not disable bh for long time
include/linux/skbuff.h | 35 +-
include/net/ip_vs.h | 130 ++--
net/core/dst.c | 9 +-
net/netfilter/ipvs/ip_vs_app.c | 31 +-
net/netfilter/ipvs/ip_vs_conn.c | 306 +++++-----
net/netfilter/ipvs/ip_vs_core.c | 73 +--
net/netfilter/ipvs/ip_vs_ctl.c | 641 ++++++++++----------
net/netfilter/ipvs/ip_vs_dh.c | 86 +--
net/netfilter/ipvs/ip_vs_ftp.c | 4 +
net/netfilter/ipvs/ip_vs_lblc.c | 115 ++--
net/netfilter/ipvs/ip_vs_lblcr.c | 190 +++---
net/netfilter/ipvs/ip_vs_lc.c | 3 +-
net/netfilter/ipvs/ip_vs_nq.c | 3 +-
net/netfilter/ipvs/ip_vs_pe.c | 55 +-
net/netfilter/ipvs/ip_vs_pe_sip.c | 1 +
net/netfilter/ipvs/ip_vs_proto_sctp.c | 36 +-
net/netfilter/ipvs/ip_vs_proto_tcp.c | 40 +-
net/netfilter/ipvs/ip_vs_proto_udp.c | 33 +-
net/netfilter/ipvs/ip_vs_rr.c | 64 +-
net/netfilter/ipvs/ip_vs_sched.c | 63 +-
net/netfilter/ipvs/ip_vs_sed.c | 5 +-
net/netfilter/ipvs/ip_vs_sh.c | 86 +--
net/netfilter/ipvs/ip_vs_sync.c | 35 +-
net/netfilter/ipvs/ip_vs_wlc.c | 5 +-
net/netfilter/ipvs/ip_vs_wrr.c | 176 +++---
net/netfilter/ipvs/ip_vs_xmit.c | 1050 ++++++++++++++-------------------
26 files changed, 1611 insertions(+), 1664 deletions(-)
^ permalink raw reply [flat|nested] 42+ messages in thread
* [PATCH 01/34] net: add skb_dst_set_noref_force
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-04-01 12:06 ` Pablo Neira Ayuso
2013-03-29 4:11 ` [PATCH 02/34] ipvs: avoid routing by TOS for real server Simon Horman
` (33 subsequent siblings)
34 siblings, 1 reply; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Rename skb_dst_set_noref to __skb_dst_set_noref and
add force flag as suggested by David Miller. The new wrapper
skb_dst_set_noref_force will force dst entries that are not
cached to be attached as skb dst without taking reference
as long as provided dst is reclaimed after RCU grace period.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/linux/skbuff.h | 35 ++++++++++++++++++++++++++++++++++-
net/core/dst.c | 9 +++++----
2 files changed, 39 insertions(+), 5 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 878e0ee..364e244 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -575,7 +575,40 @@ static inline void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst)
skb->_skb_refdst = (unsigned long)dst;
}
-extern void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst);
+extern void __skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst,
+ bool force);
+
+/**
+ * skb_dst_set_noref - sets skb dst, hopefully, without taking reference
+ * @skb: buffer
+ * @dst: dst entry
+ *
+ * Sets skb dst, assuming a reference was not taken on dst.
+ * If dst entry is cached, we do not take reference and dst_release
+ * will be avoided by refdst_drop. If dst entry is not cached, we take
+ * reference, so that last dst_release can destroy the dst immediately.
+ */
+static inline void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst)
+{
+ __skb_dst_set_noref(skb, dst, false);
+}
+
+/**
+ * skb_dst_set_noref_force - sets skb dst, without taking reference
+ * @skb: buffer
+ * @dst: dst entry
+ *
+ * Sets skb dst, assuming a reference was not taken on dst.
+ * No reference is taken and no dst_release will be called. While for
+ * cached dsts deferred reclaim is a basic feature, for entries that are
+ * not cached it is caller's job to guarantee that last dst_release for
+ * provided dst happens when nobody uses it, eg. after a RCU grace period.
+ */
+static inline void skb_dst_set_noref_force(struct sk_buff *skb,
+ struct dst_entry *dst)
+{
+ __skb_dst_set_noref(skb, dst, true);
+}
/**
* skb_dst_is_noref - Test if skb dst isn't refcounted
diff --git a/net/core/dst.c b/net/core/dst.c
index 35fd12f..df9cc81 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -320,27 +320,28 @@ void __dst_destroy_metrics_generic(struct dst_entry *dst, unsigned long old)
EXPORT_SYMBOL(__dst_destroy_metrics_generic);
/**
- * skb_dst_set_noref - sets skb dst, without a reference
+ * __skb_dst_set_noref - sets skb dst, without a reference
* @skb: buffer
* @dst: dst entry
+ * @force: if force is set, use noref version even for DST_NOCACHE entries
*
* Sets skb dst, assuming a reference was not taken on dst
* skb_dst_drop() should not dst_release() this dst
*/
-void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst)
+void __skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst, bool force)
{
WARN_ON(!rcu_read_lock_held() && !rcu_read_lock_bh_held());
/* If dst not in cache, we must take a reference, because
* dst_release() will destroy dst as soon as its refcount becomes zero
*/
- if (unlikely(dst->flags & DST_NOCACHE)) {
+ if (unlikely((dst->flags & DST_NOCACHE) && !force)) {
dst_hold(dst);
skb_dst_set(skb, dst);
} else {
skb->_skb_refdst = (unsigned long)dst | SKB_DST_NOREF;
}
}
-EXPORT_SYMBOL(skb_dst_set_noref);
+EXPORT_SYMBOL(__skb_dst_set_noref);
/* Dirty hack. We did it in 2.2 (in __dst_free),
* we have _very_ good reasons not to repeat
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 02/34] ipvs: avoid routing by TOS for real server
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
2013-03-29 4:11 ` [PATCH 01/34] net: add skb_dst_set_noref_force Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 03/34] ipvs: prefer NETDEV_DOWN event to free cached dsts Simon Horman
` (32 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Avoid replacing the cached route for real server
on every packet with different TOS. I doubt that routing
by TOS for real server is used at all, so we should be
better with such optimization.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 1 -
net/netfilter/ipvs/ip_vs_xmit.c | 58 +++++++++++++++++----------------------
2 files changed, 25 insertions(+), 34 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index bee87ba..64db117 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -753,7 +753,6 @@ struct ip_vs_dest {
/* for destination cache */
spinlock_t dst_lock; /* lock of dst_cache */
struct dst_entry *dst_cache; /* destination cache entry */
- u32 dst_rtos; /* RT_TOS(tos) for dst */
u32 dst_cookie;
union nf_inet_addr dst_saddr;
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index ee6b7a9..4b0bd15 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -57,27 +57,24 @@ enum {
* Destination cache to speed up outgoing route lookup
*/
static inline void
-__ip_vs_dst_set(struct ip_vs_dest *dest, u32 rtos, struct dst_entry *dst,
- u32 dst_cookie)
+__ip_vs_dst_set(struct ip_vs_dest *dest, struct dst_entry *dst, u32 dst_cookie)
{
struct dst_entry *old_dst;
old_dst = dest->dst_cache;
dest->dst_cache = dst;
- dest->dst_rtos = rtos;
dest->dst_cookie = dst_cookie;
dst_release(old_dst);
}
static inline struct dst_entry *
-__ip_vs_dst_check(struct ip_vs_dest *dest, u32 rtos)
+__ip_vs_dst_check(struct ip_vs_dest *dest)
{
struct dst_entry *dst = dest->dst_cache;
if (!dst)
return NULL;
- if ((dst->obsolete || rtos != dest->dst_rtos) &&
- dst->ops->check(dst, dest->dst_cookie) == NULL) {
+ if (dst->obsolete && dst->ops->check(dst, dest->dst_cookie) == NULL) {
dest->dst_cache = NULL;
dst_release(dst);
return NULL;
@@ -104,7 +101,7 @@ __mtu_check_toobig_v6(const struct sk_buff *skb, u32 mtu)
/* Get route to daddr, update *saddr, optionally bind route to saddr */
static struct rtable *do_output_route4(struct net *net, __be32 daddr,
- u32 rtos, int rt_mode, __be32 *saddr)
+ int rt_mode, __be32 *saddr)
{
struct flowi4 fl4;
struct rtable *rt;
@@ -113,7 +110,6 @@ static struct rtable *do_output_route4(struct net *net, __be32 daddr,
memset(&fl4, 0, sizeof(fl4));
fl4.daddr = daddr;
fl4.saddr = (rt_mode & IP_VS_RT_MODE_CONNECT) ? *saddr : 0;
- fl4.flowi4_tos = rtos;
fl4.flowi4_flags = (rt_mode & IP_VS_RT_MODE_KNOWN_NH) ?
FLOWI_FLAG_KNOWN_NH : 0;
@@ -124,7 +120,7 @@ retry:
if (PTR_ERR(rt) == -EINVAL && *saddr &&
rt_mode & IP_VS_RT_MODE_CONNECT && !loop) {
*saddr = 0;
- flowi4_update_output(&fl4, 0, rtos, daddr, 0);
+ flowi4_update_output(&fl4, 0, 0, daddr, 0);
goto retry;
}
IP_VS_DBG_RL("ip_route_output error, dest: %pI4\n", &daddr);
@@ -132,7 +128,7 @@ retry:
} else if (!*saddr && rt_mode & IP_VS_RT_MODE_CONNECT && fl4.saddr) {
ip_rt_put(rt);
*saddr = fl4.saddr;
- flowi4_update_output(&fl4, 0, rtos, daddr, fl4.saddr);
+ flowi4_update_output(&fl4, 0, 0, daddr, fl4.saddr);
loop++;
goto retry;
}
@@ -143,7 +139,7 @@ retry:
/* Get route to destination or remote server */
static struct rtable *
__ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
- __be32 daddr, u32 rtos, int rt_mode, __be32 *ret_saddr)
+ __be32 daddr, int rt_mode, __be32 *ret_saddr)
{
struct net *net = dev_net(skb_dst(skb)->dev);
struct rtable *rt; /* Route to the other host */
@@ -152,19 +148,18 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
if (dest) {
spin_lock(&dest->dst_lock);
- if (!(rt = (struct rtable *)
- __ip_vs_dst_check(dest, rtos))) {
- rt = do_output_route4(net, dest->addr.ip, rtos,
- rt_mode, &dest->dst_saddr.ip);
+ rt = (struct rtable *) __ip_vs_dst_check(dest);
+ if (!rt) {
+ rt = do_output_route4(net, dest->addr.ip, rt_mode,
+ &dest->dst_saddr.ip);
if (!rt) {
spin_unlock(&dest->dst_lock);
return NULL;
}
- __ip_vs_dst_set(dest, rtos, dst_clone(&rt->dst), 0);
- IP_VS_DBG(10, "new dst %pI4, src %pI4, refcnt=%d, "
- "rtos=%X\n",
+ __ip_vs_dst_set(dest, dst_clone(&rt->dst), 0);
+ IP_VS_DBG(10, "new dst %pI4, src %pI4, refcnt=%d\n",
&dest->addr.ip, &dest->dst_saddr.ip,
- atomic_read(&rt->dst.__refcnt), rtos);
+ atomic_read(&rt->dst.__refcnt));
}
daddr = dest->addr.ip;
if (ret_saddr)
@@ -177,7 +172,7 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
* for performance reasons because we do not remember saddr
*/
rt_mode &= ~IP_VS_RT_MODE_CONNECT;
- rt = do_output_route4(net, daddr, rtos, rt_mode, &saddr);
+ rt = do_output_route4(net, daddr, rt_mode, &saddr);
if (!rt)
return NULL;
if (ret_saddr)
@@ -307,7 +302,7 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
if (dest) {
spin_lock(&dest->dst_lock);
- rt = (struct rt6_info *)__ip_vs_dst_check(dest, 0);
+ rt = (struct rt6_info *)__ip_vs_dst_check(dest);
if (!rt) {
u32 cookie;
@@ -320,7 +315,7 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
}
rt = (struct rt6_info *) dst;
cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
- __ip_vs_dst_set(dest, 0, dst_clone(&rt->dst), cookie);
+ __ip_vs_dst_set(dest, dst_clone(&rt->dst), cookie);
IP_VS_DBG(10, "new dst %pI6, src %pI6, refcnt=%d\n",
&dest->addr.in6, &dest->dst_saddr.in6,
atomic_read(&rt->dst.__refcnt));
@@ -449,8 +444,9 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
- if (!(rt = __ip_vs_get_out_rt(skb, NULL, iph->daddr, RT_TOS(iph->tos),
- IP_VS_RT_MODE_NON_LOCAL, NULL)))
+ rt = __ip_vs_get_out_rt(skb, NULL, iph->daddr, IP_VS_RT_MODE_NON_LOCAL,
+ NULL);
+ if (!rt)
goto tx_error_icmp;
/* MTU checking */
@@ -581,10 +577,9 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
}
if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- RT_TOS(iph->tos),
IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL |
- IP_VS_RT_MODE_RDR, NULL)))
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_RDR, NULL)))
goto tx_error_icmp;
local = rt->rt_flags & RTCF_LOCAL;
/*
@@ -832,10 +827,9 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- RT_TOS(tos), IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL |
- IP_VS_RT_MODE_CONNECT,
- &saddr)))
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_CONNECT, &saddr)))
goto tx_error_icmp;
if (rt->rt_flags & RTCF_LOCAL) {
ip_rt_put(rt);
@@ -1067,7 +1061,6 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- RT_TOS(iph->tos),
IP_VS_RT_MODE_LOCAL |
IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_KNOWN_NH, NULL)))
@@ -1223,7 +1216,6 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_RT_MODE_LOCAL | IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_RDR : IP_VS_RT_MODE_NON_LOCAL;
if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- RT_TOS(ip_hdr(skb)->tos),
rt_mode, NULL)))
goto tx_error_icmp;
local = rt->rt_flags & RTCF_LOCAL;
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 03/34] ipvs: prefer NETDEV_DOWN event to free cached dsts
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
2013-03-29 4:11 ` [PATCH 01/34] net: add skb_dst_set_noref_force Simon Horman
2013-03-29 4:11 ` [PATCH 02/34] ipvs: avoid routing by TOS for real server Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 04/34] ipvs: convert the IP_VS_XMIT macros to functions Simon Horman
` (31 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The real server becomes unreachable on down event,
no need to wait device unregistration. Should help in
releasing dsts early before dst->dev is replaced with lo.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_ctl.c | 8 +++-----
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 8104120..6b55ba6 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -1514,10 +1514,8 @@ __ip_vs_dev_reset(struct ip_vs_dest *dest, struct net_device *dev)
spin_unlock_bh(&dest->dst_lock);
}
-/*
- * Netdev event receiver
- * Currently only NETDEV_UNREGISTER is handled, i.e. if we hold a reference to
- * a device that is "unregister" it must be released.
+/* Netdev event receiver
+ * Currently only NETDEV_DOWN is handled to release refs to cached dsts
*/
static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
void *ptr)
@@ -1529,7 +1527,7 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
struct ip_vs_dest *dest;
unsigned int idx;
- if (event != NETDEV_UNREGISTER || !ipvs)
+ if (event != NETDEV_DOWN || !ipvs)
return NOTIFY_DONE;
IP_VS_DBG(3, "%s() dev=%s\n", __func__, dev->name);
EnterFunction(2);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 04/34] ipvs: convert the IP_VS_XMIT macros to functions
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (2 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 03/34] ipvs: prefer NETDEV_DOWN event to free cached dsts Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 05/34] ipvs: rename functions related to dst_cache reset Simon Horman
` (30 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
It was a bad idea to hide return statements in macros.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_xmit.c | 134 +++++++++++++++++++++------------------
1 file changed, 72 insertions(+), 62 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 4b0bd15..7cd7c61 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -376,45 +376,59 @@ ip_vs_dst_reset(struct ip_vs_dest *dest)
dest->dst_saddr.ip = 0;
}
-#define IP_VS_XMIT_TUNNEL(skb, cp) \
-({ \
- int __ret = NF_ACCEPT; \
- \
- (skb)->ipvs_property = 1; \
- if (unlikely((cp)->flags & IP_VS_CONN_F_NFCT)) \
- __ret = ip_vs_confirm_conntrack(skb); \
- if (__ret == NF_ACCEPT) { \
- nf_reset(skb); \
- skb_forward_csum(skb); \
- } \
- __ret; \
-})
-
-#define IP_VS_XMIT_NAT(pf, skb, cp, local) \
-do { \
- (skb)->ipvs_property = 1; \
- if (likely(!((cp)->flags & IP_VS_CONN_F_NFCT))) \
- ip_vs_notrack(skb); \
- else \
- ip_vs_update_conntrack(skb, cp, 1); \
- if (local) \
- return NF_ACCEPT; \
- skb_forward_csum(skb); \
- NF_HOOK(pf, NF_INET_LOCAL_OUT, (skb), NULL, \
- skb_dst(skb)->dev, dst_output); \
-} while (0)
-
-#define IP_VS_XMIT(pf, skb, cp, local) \
-do { \
- (skb)->ipvs_property = 1; \
- if (likely(!((cp)->flags & IP_VS_CONN_F_NFCT))) \
- ip_vs_notrack(skb); \
- if (local) \
- return NF_ACCEPT; \
- skb_forward_csum(skb); \
- NF_HOOK(pf, NF_INET_LOCAL_OUT, (skb), NULL, \
- skb_dst(skb)->dev, dst_output); \
-} while (0)
+/* return NF_ACCEPT to allow forwarding or other NF_xxx on error */
+static inline int ip_vs_tunnel_xmit_prepare(struct sk_buff *skb,
+ struct ip_vs_conn *cp)
+{
+ int ret = NF_ACCEPT;
+
+ skb->ipvs_property = 1;
+ if (unlikely(cp->flags & IP_VS_CONN_F_NFCT))
+ ret = ip_vs_confirm_conntrack(skb);
+ if (ret == NF_ACCEPT) {
+ nf_reset(skb);
+ skb_forward_csum(skb);
+ }
+ return ret;
+}
+
+/* return NF_STOLEN (sent) or NF_ACCEPT if local=1 (not sent) */
+static inline int ip_vs_nat_send_or_cont(int pf, struct sk_buff *skb,
+ struct ip_vs_conn *cp, int local)
+{
+ int ret = NF_STOLEN;
+
+ skb->ipvs_property = 1;
+ if (likely(!(cp->flags & IP_VS_CONN_F_NFCT)))
+ ip_vs_notrack(skb);
+ else
+ ip_vs_update_conntrack(skb, cp, 1);
+ if (!local) {
+ skb_forward_csum(skb);
+ NF_HOOK(pf, NF_INET_LOCAL_OUT, skb, NULL, skb_dst(skb)->dev,
+ dst_output);
+ } else
+ ret = NF_ACCEPT;
+ return ret;
+}
+
+/* return NF_STOLEN (sent) or NF_ACCEPT if local=1 (not sent) */
+static inline int ip_vs_send_or_cont(int pf, struct sk_buff *skb,
+ struct ip_vs_conn *cp, int local)
+{
+ int ret = NF_STOLEN;
+
+ skb->ipvs_property = 1;
+ if (likely(!(cp->flags & IP_VS_CONN_F_NFCT)))
+ ip_vs_notrack(skb);
+ if (!local) {
+ skb_forward_csum(skb);
+ NF_HOOK(pf, NF_INET_LOCAL_OUT, skb, NULL, skb_dst(skb)->dev,
+ dst_output);
+ } else
+ ret = NF_ACCEPT;
+ return ret;
+}
/*
@@ -425,7 +439,7 @@ ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
/* we do not touch skb and do not need pskb ptr */
- IP_VS_XMIT(NFPROTO_IPV4, skb, cp, 1);
+ return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
}
@@ -476,7 +490,7 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT(NFPROTO_IPV4, skb, cp, 0);
+ ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 0);
LeaveFunction(10);
return NF_STOLEN;
@@ -537,7 +551,7 @@ ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT(NFPROTO_IPV6, skb, cp, 0);
+ ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 0);
LeaveFunction(10);
return NF_STOLEN;
@@ -562,7 +576,7 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct rtable *rt; /* Route to the other host */
int mtu;
struct iphdr *iph = ip_hdr(skb);
- int local;
+ int local, rc;
EnterFunction(10);
@@ -655,10 +669,10 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT_NAT(NFPROTO_IPV4, skb, cp, local);
+ rc = ip_vs_nat_send_or_cont(NFPROTO_IPV4, skb, cp, local);
LeaveFunction(10);
- return NF_STOLEN;
+ return rc;
tx_error_icmp:
dst_link_failure(skb);
@@ -678,7 +692,7 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
{
struct rt6_info *rt; /* Route to the other host */
int mtu;
- int local;
+ int local, rc;
EnterFunction(10);
@@ -771,10 +785,10 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT_NAT(NFPROTO_IPV6, skb, cp, local);
+ rc = ip_vs_nat_send_or_cont(NFPROTO_IPV6, skb, cp, local);
LeaveFunction(10);
- return NF_STOLEN;
+ return rc;
tx_error_icmp:
dst_link_failure(skb);
@@ -833,7 +847,7 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error_icmp;
if (rt->rt_flags & RTCF_LOCAL) {
ip_rt_put(rt);
- IP_VS_XMIT(NFPROTO_IPV4, skb, cp, 1);
+ return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
}
tdev = rt->dst.dev;
@@ -905,7 +919,7 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- ret = IP_VS_XMIT_TUNNEL(skb, cp);
+ ret = ip_vs_tunnel_xmit_prepare(skb, cp);
if (ret == NF_ACCEPT)
ip_local_out(skb);
else if (ret == NF_DROP)
@@ -948,7 +962,7 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error_icmp;
if (__ip_vs_is_local_route6(rt)) {
dst_release(&rt->dst);
- IP_VS_XMIT(NFPROTO_IPV6, skb, cp, 1);
+ return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
}
tdev = rt->dst.dev;
@@ -1023,7 +1037,7 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- ret = IP_VS_XMIT_TUNNEL(skb, cp);
+ ret = ip_vs_tunnel_xmit_prepare(skb, cp);
if (ret == NF_ACCEPT)
ip6_local_out(skb);
else if (ret == NF_DROP)
@@ -1067,7 +1081,7 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error_icmp;
if (rt->rt_flags & RTCF_LOCAL) {
ip_rt_put(rt);
- IP_VS_XMIT(NFPROTO_IPV4, skb, cp, 1);
+ return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
}
/* MTU checking */
@@ -1097,7 +1111,7 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT(NFPROTO_IPV4, skb, cp, 0);
+ ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 0);
LeaveFunction(10);
return NF_STOLEN;
@@ -1126,7 +1140,7 @@ ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error_icmp;
if (__ip_vs_is_local_route6(rt)) {
dst_release(&rt->dst);
- IP_VS_XMIT(NFPROTO_IPV6, skb, cp, 1);
+ return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
}
/* MTU checking */
@@ -1162,7 +1176,7 @@ ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT(NFPROTO_IPV6, skb, cp, 0);
+ ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 0);
LeaveFunction(10);
return NF_STOLEN;
@@ -1283,9 +1297,7 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT_NAT(NFPROTO_IPV4, skb, cp, local);
-
- rc = NF_STOLEN;
+ rc = ip_vs_nat_send_or_cont(NFPROTO_IPV4, skb, cp, local);
goto out;
tx_error_icmp:
@@ -1404,9 +1416,7 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
- IP_VS_XMIT_NAT(NFPROTO_IPV6, skb, cp, local);
-
- rc = NF_STOLEN;
+ rc = ip_vs_nat_send_or_cont(NFPROTO_IPV6, skb, cp, local);
goto out;
tx_error_icmp:
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 05/34] ipvs: rename functions related to dst_cache reset
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (3 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 04/34] ipvs: convert the IP_VS_XMIT macros to functions Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback Simon Horman
` (29 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Move and give better names to two functions:
- ip_vs_dst_reset to __ip_vs_dst_cache_reset
- __ip_vs_dev_reset to ip_vs_forget_dev
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 1 -
net/netfilter/ipvs/ip_vs_ctl.c | 34 ++++++++++++++++++++++------------
net/netfilter/ipvs/ip_vs_xmit.c | 14 --------------
3 files changed, 22 insertions(+), 27 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 64db117..8ad73a8 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -1427,7 +1427,6 @@ extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, int offset,
unsigned int hooknum, struct ip_vs_iphdr *iph);
-extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
#ifdef CONFIG_IP_VS_IPV6
extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 6b55ba6..5265eaa 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -641,6 +641,17 @@ struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
return dest;
}
+/* Release dst_cache for dest in user context */
+static void __ip_vs_dst_cache_reset(struct ip_vs_dest *dest)
+{
+ struct dst_entry *old_dst;
+
+ old_dst = dest->dst_cache;
+ dest->dst_cache = NULL;
+ dst_release(old_dst);
+ dest->dst_saddr.ip = 0;
+}
+
/*
* Lookup dest by {svc,addr,port} in the destination trash.
* The destination trash is used to hold the destinations that are removed
@@ -690,7 +701,7 @@ ip_vs_trash_get_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
IP_VS_DBG_ADDR(svc->af, &dest->addr),
ntohs(dest->port));
list_del(&dest->n_list);
- ip_vs_dst_reset(dest);
+ __ip_vs_dst_cache_reset(dest);
__ip_vs_unbind_svc(dest);
free_percpu(dest->stats.cpustats);
kfree(dest);
@@ -717,7 +728,7 @@ static void ip_vs_trash_cleanup(struct net *net)
list_for_each_entry_safe(dest, nxt, &ipvs->dest_trash, n_list) {
list_del(&dest->n_list);
- ip_vs_dst_reset(dest);
+ __ip_vs_dst_cache_reset(dest);
__ip_vs_unbind_svc(dest);
free_percpu(dest->stats.cpustats);
kfree(dest);
@@ -811,7 +822,7 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
dest->l_threshold = udest->l_threshold;
spin_lock_bh(&dest->dst_lock);
- ip_vs_dst_reset(dest);
+ __ip_vs_dst_cache_reset(dest);
spin_unlock_bh(&dest->dst_lock);
if (add)
@@ -1037,7 +1048,7 @@ static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest)
dest->vfwmark,
IP_VS_DBG_ADDR(dest->af, &dest->addr),
ntohs(dest->port));
- ip_vs_dst_reset(dest);
+ __ip_vs_dst_cache_reset(dest);
/* simply decrease svc->refcnt here, let the caller check
and release the service if nobody refers to it.
Only user context can release destination and service,
@@ -1496,11 +1507,10 @@ void ip_vs_service_net_cleanup(struct net *net)
mutex_unlock(&__ip_vs_mutex);
LeaveFunction(2);
}
-/*
- * Release dst hold by dst_cache
- */
+
+/* Put all references for device (dst_cache) */
static inline void
-__ip_vs_dev_reset(struct ip_vs_dest *dest, struct net_device *dev)
+ip_vs_forget_dev(struct ip_vs_dest *dest, struct net_device *dev)
{
spin_lock_bh(&dest->dst_lock);
if (dest->dst_cache && dest->dst_cache->dev == dev) {
@@ -1509,7 +1519,7 @@ __ip_vs_dev_reset(struct ip_vs_dest *dest, struct net_device *dev)
IP_VS_DBG_ADDR(dest->af, &dest->addr),
ntohs(dest->port),
atomic_read(&dest->refcnt));
- ip_vs_dst_reset(dest);
+ __ip_vs_dst_cache_reset(dest);
}
spin_unlock_bh(&dest->dst_lock);
@@ -1537,7 +1547,7 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
if (net_eq(svc->net, net)) {
list_for_each_entry(dest, &svc->destinations,
n_list) {
- __ip_vs_dev_reset(dest, dev);
+ ip_vs_forget_dev(dest, dev);
}
}
}
@@ -1546,7 +1556,7 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
if (net_eq(svc->net, net)) {
list_for_each_entry(dest, &svc->destinations,
n_list) {
- __ip_vs_dev_reset(dest, dev);
+ ip_vs_forget_dev(dest, dev);
}
}
@@ -1554,7 +1564,7 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
}
list_for_each_entry(dest, &ipvs->dest_trash, n_list) {
- __ip_vs_dev_reset(dest, dev);
+ ip_vs_forget_dev(dest, dev);
}
mutex_unlock(&__ip_vs_mutex);
LeaveFunction(2);
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 7cd7c61..6448a2e 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -362,20 +362,6 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
#endif
-/*
- * Release dest->dst_cache before a dest is removed
- */
-void
-ip_vs_dst_reset(struct ip_vs_dest *dest)
-{
- struct dst_entry *old_dst;
-
- old_dst = dest->dst_cache;
- dest->dst_cache = NULL;
- dst_release(old_dst);
- dest->dst_saddr.ip = 0;
-}
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (4 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 05/34] ipvs: rename functions related to dst_cache reset Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 14:44 ` Sergei Shtylyov
2013-03-29 4:11 ` [PATCH 07/34] ipvs: do not use skb_share_check Simon Horman
` (28 subsequent siblings)
34 siblings, 1 reply; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
After commit 70e7341673 (ipv4: Show that ip_send_reply()
is purely unicast routine.) we do not need to reroute DNAT-ed
traffic over loopback because reply uses iph daddr and not
rt_spec_dst.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_xmit.c | 58 ++-------------------------------------
1 file changed, 2 insertions(+), 56 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 6448a2e..c942d36 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -207,44 +207,6 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
return rt;
}
-/* Reroute packet to local IPv4 stack after DNAT */
-static int
-__ip_vs_reroute_locally(struct sk_buff *skb)
-{
- struct rtable *rt = skb_rtable(skb);
- struct net_device *dev = rt->dst.dev;
- struct net *net = dev_net(dev);
- struct iphdr *iph = ip_hdr(skb);
-
- if (rt_is_input_route(rt)) {
- unsigned long orefdst = skb->_skb_refdst;
-
- if (ip_route_input(skb, iph->daddr, iph->saddr,
- iph->tos, skb->dev))
- return 0;
- refdst_drop(orefdst);
- } else {
- struct flowi4 fl4 = {
- .daddr = iph->daddr,
- .saddr = iph->saddr,
- .flowi4_tos = RT_TOS(iph->tos),
- .flowi4_mark = skb->mark,
- };
-
- rt = ip_route_output_key(net, &fl4);
- if (IS_ERR(rt))
- return 0;
- if (!(rt->rt_flags & RTCF_LOCAL)) {
- ip_rt_put(rt);
- return 0;
- }
- /* Drop old route. */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
- }
- return 1;
-}
-
#ifdef CONFIG_IP_VS_IPV6
static inline int __ip_vs_is_local_route6(struct rt6_info *rt)
@@ -635,16 +597,8 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* drop old route */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
- } else {
+ } else
ip_rt_put(rt);
- /*
- * Some IPv4 replies get local address from routes,
- * not from iph, so while we DNAT after routing
- * we need this second input/output route.
- */
- if (!__ip_vs_reroute_locally(skb))
- goto tx_error;
- }
IP_VS_DBG_PKT(10, AF_INET, pp, skb, 0, "After DNAT");
@@ -1269,16 +1223,8 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/* drop the old route when skb is not shared */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
- } else {
+ } else
ip_rt_put(rt);
- /*
- * Some IPv4 replies get local address from routes,
- * not from iph, so while we DNAT after routing
- * we need this second input/output route.
- */
- if (!__ip_vs_reroute_locally(skb))
- goto tx_error;
- }
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 07/34] ipvs: do not use skb_share_check
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (5 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 08/34] ipvs: consolidate all dst checks on transmit in one place Simon Horman
` (27 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
We run in contexts like ip_rcv, ipv6_rcv, br_handle_frame,
do not expect shared skbs.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_xmit.c | 45 ++-------------------------------------
1 file changed, 2 insertions(+), 43 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index c942d36..e6e2d3f 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -421,14 +421,6 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error;
}
- /*
- * Call ip_send_check because we are not sure it is called
- * after ip_defrag. Is copy-on-write needed?
- */
- if (unlikely((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL)) {
- ip_rt_put(rt);
- return NF_STOLEN;
- }
ip_send_check(ip_hdr(skb));
/* drop old route */
@@ -482,16 +474,6 @@ ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error;
}
- /*
- * Call ip_send_check because we are not sure it is called
- * after ip_defrag. Is copy-on-write needed?
- */
- skb = skb_share_check(skb, GFP_ATOMIC);
- if (unlikely(skb == NULL)) {
- dst_release(&rt->dst);
- return NF_STOLEN;
- }
-
/* drop old route */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
@@ -708,7 +690,6 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
ipv6_hdr(skb)->daddr = cp->daddr.in6;
if (!local || !skb->dev) {
- /* drop the old route when skb is not shared */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
} else {
@@ -814,8 +795,7 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
*/
max_headroom = LL_RESERVED_SPACE(tdev) + sizeof(struct iphdr);
- if (skb_headroom(skb) < max_headroom
- || skb_cloned(skb) || skb_shared(skb)) {
+ if (skb_headroom(skb) < max_headroom || skb_cloned(skb)) {
struct sk_buff *new_skb =
skb_realloc_headroom(skb, max_headroom);
if (!new_skb) {
@@ -935,8 +915,7 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
*/
max_headroom = LL_RESERVED_SPACE(tdev) + sizeof(struct ipv6hdr);
- if (skb_headroom(skb) < max_headroom
- || skb_cloned(skb) || skb_shared(skb)) {
+ if (skb_headroom(skb) < max_headroom || skb_cloned(skb)) {
struct sk_buff *new_skb =
skb_realloc_headroom(skb, max_headroom);
if (!new_skb) {
@@ -1034,14 +1013,6 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error;
}
- /*
- * Call ip_send_check because we are not sure it is called
- * after ip_defrag. Is copy-on-write needed?
- */
- if (unlikely((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL)) {
- ip_rt_put(rt);
- return NF_STOLEN;
- }
ip_send_check(ip_hdr(skb));
/* drop old route */
@@ -1099,16 +1070,6 @@ ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
goto tx_error;
}
- /*
- * Call ip_send_check because we are not sure it is called
- * after ip_defrag. Is copy-on-write needed?
- */
- skb = skb_share_check(skb, GFP_ATOMIC);
- if (unlikely(skb == NULL)) {
- dst_release(&rt->dst);
- return NF_STOLEN;
- }
-
/* drop old route */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
@@ -1220,7 +1181,6 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
ip_vs_nat_icmp(skb, pp, cp, 0);
if (!local) {
- /* drop the old route when skb is not shared */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
} else
@@ -1337,7 +1297,6 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
ip_vs_nat_icmp_v6(skb, pp, cp, 0);
if (!local || !skb->dev) {
- /* drop the old route when skb is not shared */
skb_dst_drop(skb);
skb_dst_set(skb, &rt->dst);
} else {
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 08/34] ipvs: consolidate all dst checks on transmit in one place
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (6 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 07/34] ipvs: do not use skb_share_check Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 09/34] ipvs: optimize dst usage for real server Simon Horman
` (26 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Consolidate the PMTU checks, ICMP sending and
skb_dst modification in __ip_vs_get_out_rt and
__ip_vs_get_out_rt_v6. Now skb_dst is changed early
to simplify the transmitters.
Make sure update_pmtu is called only for local clients.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_xmit.c | 627 +++++++++++++++------------------------
1 file changed, 235 insertions(+), 392 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index e6e2d3f..603eb8a 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -51,6 +51,7 @@ enum {
*/
IP_VS_RT_MODE_CONNECT = 8, /* Always bind route to saddr */
IP_VS_RT_MODE_KNOWN_NH = 16,/* Route via remote addr */
+ IP_VS_RT_MODE_TUNNEL = 32,/* Tunnel mode */
};
/*
@@ -137,13 +138,17 @@ retry:
}
/* Get route to destination or remote server */
-static struct rtable *
+static int
__ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
__be32 daddr, int rt_mode, __be32 *ret_saddr)
{
struct net *net = dev_net(skb_dst(skb)->dev);
+ struct netns_ipvs *ipvs = net_ipvs(net);
struct rtable *rt; /* Route to the other host */
struct rtable *ort; /* Original route */
+ struct iphdr *iph;
+ __be16 df;
+ int mtu;
int local;
if (dest) {
@@ -154,7 +159,7 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
&dest->dst_saddr.ip);
if (!rt) {
spin_unlock(&dest->dst_lock);
- return NULL;
+ goto err_unreach;
}
__ip_vs_dst_set(dest, dst_clone(&rt->dst), 0);
IP_VS_DBG(10, "new dst %pI4, src %pI4, refcnt=%d\n",
@@ -174,37 +179,78 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
rt_mode &= ~IP_VS_RT_MODE_CONNECT;
rt = do_output_route4(net, daddr, rt_mode, &saddr);
if (!rt)
- return NULL;
+ goto err_unreach;
if (ret_saddr)
*ret_saddr = saddr;
}
- local = rt->rt_flags & RTCF_LOCAL;
+ local = (rt->rt_flags & RTCF_LOCAL) ? 1 : 0;
if (!((local ? IP_VS_RT_MODE_LOCAL : IP_VS_RT_MODE_NON_LOCAL) &
rt_mode)) {
IP_VS_DBG_RL("Stopping traffic to %s address, dest: %pI4\n",
(rt->rt_flags & RTCF_LOCAL) ?
"local":"non-local", &daddr);
- ip_rt_put(rt);
- return NULL;
+ goto err_put;
}
- if (local && !(rt_mode & IP_VS_RT_MODE_RDR) &&
- !((ort = skb_rtable(skb)) && ort->rt_flags & RTCF_LOCAL)) {
- IP_VS_DBG_RL("Redirect from non-local address %pI4 to local "
- "requires NAT method, dest: %pI4\n",
- &ip_hdr(skb)->daddr, &daddr);
+ iph = ip_hdr(skb);
+ if (likely(!local)) {
+ if (unlikely(ipv4_is_loopback(iph->saddr))) {
+ IP_VS_DBG_RL("Stopping traffic from loopback address "
+ "%pI4 to non-local address, dest: %pI4\n",
+ &iph->saddr, &daddr);
+ goto err_put;
+ }
+ } else {
+ ort = skb_rtable(skb);
+ if (!(rt_mode & IP_VS_RT_MODE_RDR) &&
+ !(ort->rt_flags & RTCF_LOCAL)) {
+ IP_VS_DBG_RL("Redirect from non-local address %pI4 to "
+ "local requires NAT method, dest: %pI4\n",
+ &iph->daddr, &daddr);
+ goto err_put;
+ }
+ /* skb to local stack, preserve old route */
ip_rt_put(rt);
- return NULL;
+ return local;
}
- if (unlikely(!local && ipv4_is_loopback(ip_hdr(skb)->saddr))) {
- IP_VS_DBG_RL("Stopping traffic from loopback address %pI4 "
- "to non-local address, dest: %pI4\n",
- &ip_hdr(skb)->saddr, &daddr);
- ip_rt_put(rt);
- return NULL;
+
+ if (likely(!(rt_mode & IP_VS_RT_MODE_TUNNEL))) {
+ mtu = dst_mtu(&rt->dst);
+ df = iph->frag_off & htons(IP_DF);
+ } else {
+ struct sock *sk = skb->sk;
+
+ mtu = dst_mtu(&rt->dst) - sizeof(struct iphdr);
+ if (mtu < 68) {
+ IP_VS_DBG_RL("%s(): mtu less than 68\n", __func__);
+ goto err_put;
+ }
+ ort = skb_rtable(skb);
+ if (!skb->dev && sk && sk->sk_state != TCP_TIME_WAIT)
+ ort->dst.ops->update_pmtu(&ort->dst, sk, NULL, mtu);
+ /* MTU check allowed? */
+ df = sysctl_pmtu_disc(ipvs) ? iph->frag_off & htons(IP_DF) : 0;
}
- return rt;
+ /* MTU checking */
+ if (unlikely(df && skb->len > mtu && !skb_is_gso(skb))) {
+ icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, htonl(mtu));
+ IP_VS_DBG(1, "frag needed for %pI4\n", &iph->saddr);
+ goto err_put;
+ }
+
+ skb_dst_drop(skb);
+ skb_dst_set(skb, &rt->dst);
+
+ return local;
+
+err_put:
+ ip_rt_put(rt);
+ return -1;
+
+err_unreach:
+ dst_link_failure(skb);
+ return -1;
}
#ifdef CONFIG_IP_VS_IPV6
@@ -251,15 +297,16 @@ out_err:
/*
* Get route to destination or remote server
*/
-static struct rt6_info *
+static int
__ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
struct in6_addr *daddr, struct in6_addr *ret_saddr,
- int do_xfrm, int rt_mode)
+ struct ip_vs_iphdr *ipvsh, int do_xfrm, int rt_mode)
{
struct net *net = dev_net(skb_dst(skb)->dev);
struct rt6_info *rt; /* Route to the other host */
struct rt6_info *ort; /* Original route */
struct dst_entry *dst;
+ int mtu;
int local;
if (dest) {
@@ -273,7 +320,7 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
do_xfrm);
if (!dst) {
spin_unlock(&dest->dst_lock);
- return NULL;
+ goto err_unreach;
}
rt = (struct rt6_info *) dst;
cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
@@ -288,7 +335,7 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
} else {
dst = __ip_vs_route_output_v6(net, daddr, ret_saddr, do_xfrm);
if (!dst)
- return NULL;
+ goto err_unreach;
rt = (struct rt6_info *) dst;
}
@@ -297,29 +344,72 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
rt_mode)) {
IP_VS_DBG_RL("Stopping traffic to %s address, dest: %pI6c\n",
local ? "local":"non-local", daddr);
- dst_release(&rt->dst);
- return NULL;
+ goto err_put;
}
- if (local && !(rt_mode & IP_VS_RT_MODE_RDR) &&
- !((ort = (struct rt6_info *) skb_dst(skb)) &&
- __ip_vs_is_local_route6(ort))) {
- IP_VS_DBG_RL("Redirect from non-local address %pI6c to local "
- "requires NAT method, dest: %pI6c\n",
- &ipv6_hdr(skb)->daddr, daddr);
+ if (likely(!local)) {
+ if (unlikely((!skb->dev || skb->dev->flags & IFF_LOOPBACK) &&
+ ipv6_addr_type(&ipv6_hdr(skb)->saddr) &
+ IPV6_ADDR_LOOPBACK)) {
+ IP_VS_DBG_RL("Stopping traffic from loopback address "
+ "%pI6c to non-local address, "
+ "dest: %pI6c\n",
+ &ipv6_hdr(skb)->saddr, daddr);
+ goto err_put;
+ }
+ } else {
+ ort = (struct rt6_info *) skb_dst(skb);
+ if (!(rt_mode & IP_VS_RT_MODE_RDR) &&
+ !__ip_vs_is_local_route6(ort)) {
+ IP_VS_DBG_RL("Redirect from non-local address %pI6c "
+ "to local requires NAT method, "
+ "dest: %pI6c\n",
+ &ipv6_hdr(skb)->daddr, daddr);
+ goto err_put;
+ }
+ /* skb to local stack, preserve old route */
dst_release(&rt->dst);
- return NULL;
+ return local;
}
- if (unlikely(!local && (!skb->dev || skb->dev->flags & IFF_LOOPBACK) &&
- ipv6_addr_type(&ipv6_hdr(skb)->saddr) &
- IPV6_ADDR_LOOPBACK)) {
- IP_VS_DBG_RL("Stopping traffic from loopback address %pI6c "
- "to non-local address, dest: %pI6c\n",
- &ipv6_hdr(skb)->saddr, daddr);
- dst_release(&rt->dst);
- return NULL;
+
+ /* MTU checking */
+ if (likely(!(rt_mode & IP_VS_RT_MODE_TUNNEL)))
+ mtu = dst_mtu(&rt->dst);
+ else {
+ struct sock *sk = skb->sk;
+
+ mtu = dst_mtu(&rt->dst) - sizeof(struct ipv6hdr);
+ if (mtu < IPV6_MIN_MTU) {
+ IP_VS_DBG_RL("%s(): mtu less than %d\n", __func__,
+ IPV6_MIN_MTU);
+ goto err_put;
+ }
+ ort = (struct rt6_info *) skb_dst(skb);
+ if (!skb->dev && sk && sk->sk_state != TCP_TIME_WAIT)
+ ort->dst.ops->update_pmtu(&ort->dst, sk, NULL, mtu);
}
- return rt;
+ if (unlikely(__mtu_check_toobig_v6(skb, mtu))) {
+ if (!skb->dev)
+ skb->dev = net->loopback_dev;
+ /* only send ICMP too big on first fragment */
+ if (!ipvsh->fragoffs)
+ icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
+ IP_VS_DBG(1, "frag needed for %pI6c\n", &ipv6_hdr(skb)->saddr);
+ goto err_put;
+ }
+
+ skb_dst_drop(skb);
+ skb_dst_set(skb, &rt->dst);
+
+ return local;
+
+err_put:
+ dst_release(&rt->dst);
+ return -1;
+
+err_unreach:
+ dst_link_failure(skb);
+ return -1;
}
#endif
@@ -400,32 +490,15 @@ int
ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
- struct rtable *rt; /* Route to the other host */
struct iphdr *iph = ip_hdr(skb);
- int mtu;
EnterFunction(10);
- rt = __ip_vs_get_out_rt(skb, NULL, iph->daddr, IP_VS_RT_MODE_NON_LOCAL,
- NULL);
- if (!rt)
- goto tx_error_icmp;
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if ((skb->len > mtu) && (iph->frag_off & htons(IP_DF)) &&
- !skb_is_gso(skb)) {
- ip_rt_put(rt);
- icmp_send(skb, ICMP_DEST_UNREACH,ICMP_FRAG_NEEDED, htonl(mtu));
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
+ if (__ip_vs_get_out_rt(skb, NULL, iph->daddr, IP_VS_RT_MODE_NON_LOCAL,
+ NULL) < 0)
goto tx_error;
- }
- ip_send_check(ip_hdr(skb));
-
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
+ ip_send_check(iph);
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
@@ -435,8 +508,6 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
LeaveFunction(10);
return NF_STOLEN;
- tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
@@ -446,37 +517,13 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
#ifdef CONFIG_IP_VS_IPV6
int
ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
- struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph)
+ struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
- struct rt6_info *rt; /* Route to the other host */
- int mtu;
-
EnterFunction(10);
- rt = __ip_vs_get_out_rt_v6(skb, NULL, &iph->daddr.in6, NULL, 0,
- IP_VS_RT_MODE_NON_LOCAL);
- if (!rt)
- goto tx_error_icmp;
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if (__mtu_check_toobig_v6(skb, mtu)) {
- if (!skb->dev) {
- struct net *net = dev_net(skb_dst(skb)->dev);
-
- skb->dev = net->loopback_dev;
- }
- /* only send ICMP too big on first fragment */
- if (!iph->fragoffs)
- icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
- dst_release(&rt->dst);
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
+ if (__ip_vs_get_out_rt_v6(skb, NULL, &ipvsh->daddr.in6, NULL,
+ ipvsh, 0, IP_VS_RT_MODE_NON_LOCAL) < 0)
goto tx_error;
- }
-
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
@@ -486,8 +533,6 @@ ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
LeaveFunction(10);
return NF_STOLEN;
- tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
@@ -504,28 +549,29 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
struct rtable *rt; /* Route to the other host */
- int mtu;
- struct iphdr *iph = ip_hdr(skb);
- int local, rc;
+ int local, rc, was_input;
EnterFunction(10);
/* check if it is a connection of no-client-port */
if (unlikely(cp->flags & IP_VS_CONN_F_NO_CPORT)) {
__be16 _pt, *p;
- p = skb_header_pointer(skb, iph->ihl*4, sizeof(_pt), &_pt);
+
+ p = skb_header_pointer(skb, ipvsh->len, sizeof(_pt), &_pt);
if (p == NULL)
goto tx_error;
ip_vs_conn_fill_cport(cp, *p);
IP_VS_DBG(10, "filled cport=%d\n", ntohs(*p));
}
- if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL |
- IP_VS_RT_MODE_RDR, NULL)))
- goto tx_error_icmp;
- local = rt->rt_flags & RTCF_LOCAL;
+ was_input = rt_is_input_route(skb_rtable(skb));
+ local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_RDR, NULL);
+ if (local < 0)
+ goto tx_error;
+ rt = skb_rtable(skb);
/*
* Avoid duplicate tuple in reply direction for NAT traffic
* to local address when connection is sync-ed
@@ -539,49 +585,31 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_DBG_RL_PKT(10, AF_INET, pp, skb, 0,
"ip_vs_nat_xmit(): "
"stopping DNAT to local address");
- goto tx_error_put;
+ goto tx_error;
}
}
#endif
/* From world but DNAT to loopback address? */
- if (local && ipv4_is_loopback(cp->daddr.ip) &&
- rt_is_input_route(skb_rtable(skb))) {
+ if (local && ipv4_is_loopback(cp->daddr.ip) && was_input) {
IP_VS_DBG_RL_PKT(1, AF_INET, pp, skb, 0, "ip_vs_nat_xmit(): "
"stopping DNAT to loopback address");
- goto tx_error_put;
- }
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if ((skb->len > mtu) && (iph->frag_off & htons(IP_DF)) &&
- !skb_is_gso(skb)) {
- icmp_send(skb, ICMP_DEST_UNREACH,ICMP_FRAG_NEEDED, htonl(mtu));
- IP_VS_DBG_RL_PKT(0, AF_INET, pp, skb, 0,
- "ip_vs_nat_xmit(): frag needed for");
- goto tx_error_put;
+ goto tx_error;
}
/* copy-on-write the packet before mangling it */
if (!skb_make_writable(skb, sizeof(struct iphdr)))
- goto tx_error_put;
+ goto tx_error;
if (skb_cow(skb, rt->dst.dev->hard_header_len))
- goto tx_error_put;
+ goto tx_error;
/* mangle the packet */
if (pp->dnat_handler && !pp->dnat_handler(skb, pp, cp, ipvsh))
- goto tx_error_put;
+ goto tx_error;
ip_hdr(skb)->daddr = cp->daddr.ip;
ip_send_check(ip_hdr(skb));
- if (!local) {
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
- } else
- ip_rt_put(rt);
-
IP_VS_DBG_PKT(10, AF_INET, pp, skb, 0, "After DNAT");
/* FIXME: when application helper enlarges the packet and the length
@@ -596,44 +624,40 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
LeaveFunction(10);
return rc;
- tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
return NF_STOLEN;
- tx_error_put:
- ip_rt_put(rt);
- goto tx_error;
}
#ifdef CONFIG_IP_VS_IPV6
int
ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
- struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph)
+ struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
struct rt6_info *rt; /* Route to the other host */
- int mtu;
int local, rc;
EnterFunction(10);
/* check if it is a connection of no-client-port */
- if (unlikely(cp->flags & IP_VS_CONN_F_NO_CPORT && !iph->fragoffs)) {
+ if (unlikely(cp->flags & IP_VS_CONN_F_NO_CPORT && !ipvsh->fragoffs)) {
__be16 _pt, *p;
- p = skb_header_pointer(skb, iph->len, sizeof(_pt), &_pt);
+ p = skb_header_pointer(skb, ipvsh->len, sizeof(_pt), &_pt);
if (p == NULL)
goto tx_error;
ip_vs_conn_fill_cport(cp, *p);
IP_VS_DBG(10, "filled cport=%d\n", ntohs(*p));
}
- if (!(rt = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
- 0, (IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL |
- IP_VS_RT_MODE_RDR))))
- goto tx_error_icmp;
- local = __ip_vs_is_local_route6(rt);
+ local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
+ ipvsh, 0,
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_RDR);
+ if (local < 0)
+ goto tx_error;
+ rt = (struct rt6_info *) skb_dst(skb);
/*
* Avoid duplicate tuple in reply direction for NAT traffic
* to local address when connection is sync-ed
@@ -647,7 +671,7 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_DBG_RL_PKT(10, AF_INET6, pp, skb, 0,
"ip_vs_nat_xmit_v6(): "
"stopping DNAT to local address");
- goto tx_error_put;
+ goto tx_error;
}
}
#endif
@@ -658,45 +682,21 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_DBG_RL_PKT(1, AF_INET6, pp, skb, 0,
"ip_vs_nat_xmit_v6(): "
"stopping DNAT to loopback address");
- goto tx_error_put;
- }
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if (__mtu_check_toobig_v6(skb, mtu)) {
- if (!skb->dev) {
- struct net *net = dev_net(skb_dst(skb)->dev);
-
- skb->dev = net->loopback_dev;
- }
- /* only send ICMP too big on first fragment */
- if (!iph->fragoffs)
- icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
- IP_VS_DBG_RL_PKT(0, AF_INET6, pp, skb, 0,
- "ip_vs_nat_xmit_v6(): frag needed for");
- goto tx_error_put;
+ goto tx_error;
}
/* copy-on-write the packet before mangling it */
if (!skb_make_writable(skb, sizeof(struct ipv6hdr)))
- goto tx_error_put;
+ goto tx_error;
if (skb_cow(skb, rt->dst.dev->hard_header_len))
- goto tx_error_put;
+ goto tx_error;
/* mangle the packet */
- if (pp->dnat_handler && !pp->dnat_handler(skb, pp, cp, iph))
+ if (pp->dnat_handler && !pp->dnat_handler(skb, pp, cp, ipvsh))
goto tx_error;
ipv6_hdr(skb)->daddr = cp->daddr.in6;
- if (!local || !skb->dev) {
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
- } else {
- /* destined to loopback, do we need to change route? */
- dst_release(&rt->dst);
- }
-
IP_VS_DBG_PKT(10, AF_INET6, pp, skb, 0, "After DNAT");
/* FIXME: when application helper enlarges the packet and the length
@@ -711,15 +711,10 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
LeaveFunction(10);
return rc;
-tx_error_icmp:
- dst_link_failure(skb);
tx_error:
LeaveFunction(10);
kfree_skb(skb);
return NF_STOLEN;
-tx_error_put:
- dst_release(&rt->dst);
- goto tx_error;
}
#endif
@@ -756,40 +751,26 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
__be16 df;
struct iphdr *iph; /* Our new IP header */
unsigned int max_headroom; /* The extra header space needed */
- int mtu;
- int ret;
+ int ret, local;
EnterFunction(10);
- if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL |
- IP_VS_RT_MODE_CONNECT, &saddr)))
- goto tx_error_icmp;
- if (rt->rt_flags & RTCF_LOCAL) {
- ip_rt_put(rt);
+ local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_CONNECT |
+ IP_VS_RT_MODE_TUNNEL, &saddr);
+ if (local < 0)
+ goto tx_error;
+ if (local)
return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
- }
+ rt = skb_rtable(skb);
tdev = rt->dst.dev;
- mtu = dst_mtu(&rt->dst) - sizeof(struct iphdr);
- if (mtu < 68) {
- IP_VS_DBG_RL("%s(): mtu less than 68\n", __func__);
- goto tx_error_put;
- }
- if (rt_is_output_route(skb_rtable(skb)))
- skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
-
/* Copy DF, reset fragment offset and MF */
df = sysctl_pmtu_disc(ipvs) ? old_iph->frag_off & htons(IP_DF) : 0;
- if (df && mtu < ntohs(old_iph->tot_len) && !skb_is_gso(skb)) {
- icmp_send(skb, ICMP_DEST_UNREACH,ICMP_FRAG_NEEDED, htonl(mtu));
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
- goto tx_error_put;
- }
-
/*
* Okay, now see if we can stuff it in the buffer as-is.
*/
@@ -798,12 +779,9 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
if (skb_headroom(skb) < max_headroom || skb_cloned(skb)) {
struct sk_buff *new_skb =
skb_realloc_headroom(skb, max_headroom);
- if (!new_skb) {
- ip_rt_put(rt);
- kfree_skb(skb);
- IP_VS_ERR_RL("%s(): no memory\n", __func__);
- return NF_STOLEN;
- }
+
+ if (!new_skb)
+ goto tx_error;
consume_skb(skb);
skb = new_skb;
old_iph = ip_hdr(skb);
@@ -818,10 +796,6 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
skb_reset_network_header(skb);
memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
-
/*
* Push down and install the IPIP header.
*/
@@ -849,15 +823,10 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
return NF_STOLEN;
- tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
return NF_STOLEN;
-tx_error_put:
- ip_rt_put(rt);
- goto tx_error;
}
#ifdef CONFIG_IP_VS_IPV6
@@ -871,45 +840,23 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ipv6hdr *old_iph = ipv6_hdr(skb);
struct ipv6hdr *iph; /* Our new IP header */
unsigned int max_headroom; /* The extra header space needed */
- int mtu;
- int ret;
+ int ret, local;
EnterFunction(10);
- if (!(rt = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6,
- &saddr, 1, (IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL))))
- goto tx_error_icmp;
- if (__ip_vs_is_local_route6(rt)) {
- dst_release(&rt->dst);
+ local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6,
+ &saddr, ipvsh, 1,
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_TUNNEL);
+ if (local < 0)
+ goto tx_error;
+ if (local)
return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
- }
+ rt = (struct rt6_info *) skb_dst(skb);
tdev = rt->dst.dev;
- mtu = dst_mtu(&rt->dst) - sizeof(struct ipv6hdr);
- if (mtu < IPV6_MIN_MTU) {
- IP_VS_DBG_RL("%s(): mtu less than %d\n", __func__,
- IPV6_MIN_MTU);
- goto tx_error_put;
- }
- if (skb_dst(skb))
- skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
-
- /* MTU checking: Notice that 'mtu' have been adjusted before hand */
- if (__mtu_check_toobig_v6(skb, mtu)) {
- if (!skb->dev) {
- struct net *net = dev_net(skb_dst(skb)->dev);
-
- skb->dev = net->loopback_dev;
- }
- /* only send ICMP too big on first fragment */
- if (!ipvsh->fragoffs)
- icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
- goto tx_error_put;
- }
-
/*
* Okay, now see if we can stuff it in the buffer as-is.
*/
@@ -918,12 +865,9 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
if (skb_headroom(skb) < max_headroom || skb_cloned(skb)) {
struct sk_buff *new_skb =
skb_realloc_headroom(skb, max_headroom);
- if (!new_skb) {
- dst_release(&rt->dst);
- kfree_skb(skb);
- IP_VS_ERR_RL("%s(): no memory\n", __func__);
- return NF_STOLEN;
- }
+
+ if (!new_skb)
+ goto tx_error;
consume_skb(skb);
skb = new_skb;
old_iph = ipv6_hdr(skb);
@@ -935,10 +879,6 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
skb_reset_network_header(skb);
memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
-
/*
* Push down and install the IPIP header.
*/
@@ -966,15 +906,10 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
return NF_STOLEN;
-tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
return NF_STOLEN;
-tx_error_put:
- dst_release(&rt->dst);
- goto tx_error;
}
#endif
@@ -987,38 +922,21 @@ int
ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
- struct rtable *rt; /* Route to the other host */
- struct iphdr *iph = ip_hdr(skb);
- int mtu;
+ int local;
EnterFunction(10);
- if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL |
- IP_VS_RT_MODE_KNOWN_NH, NULL)))
- goto tx_error_icmp;
- if (rt->rt_flags & RTCF_LOCAL) {
- ip_rt_put(rt);
- return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
- }
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if ((iph->frag_off & htons(IP_DF)) && skb->len > mtu &&
- !skb_is_gso(skb)) {
- icmp_send(skb, ICMP_DEST_UNREACH,ICMP_FRAG_NEEDED, htonl(mtu));
- ip_rt_put(rt);
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
+ local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL |
+ IP_VS_RT_MODE_KNOWN_NH, NULL);
+ if (local < 0)
goto tx_error;
- }
+ if (local)
+ return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
ip_send_check(ip_hdr(skb));
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
-
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
@@ -1027,8 +945,6 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
LeaveFunction(10);
return NF_STOLEN;
- tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
@@ -1038,41 +954,20 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
#ifdef CONFIG_IP_VS_IPV6
int
ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
- struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph)
+ struct ip_vs_protocol *pp, struct ip_vs_iphdr *ipvsh)
{
- struct rt6_info *rt; /* Route to the other host */
- int mtu;
+ int local;
EnterFunction(10);
- if (!(rt = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
- 0, (IP_VS_RT_MODE_LOCAL |
- IP_VS_RT_MODE_NON_LOCAL))))
- goto tx_error_icmp;
- if (__ip_vs_is_local_route6(rt)) {
- dst_release(&rt->dst);
- return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
- }
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if (__mtu_check_toobig_v6(skb, mtu)) {
- if (!skb->dev) {
- struct net *net = dev_net(skb_dst(skb)->dev);
-
- skb->dev = net->loopback_dev;
- }
- /* only send ICMP too big on first fragment */
- if (!iph->fragoffs)
- icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
- dst_release(&rt->dst);
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
+ local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
+ ipvsh, 0,
+ IP_VS_RT_MODE_LOCAL |
+ IP_VS_RT_MODE_NON_LOCAL);
+ if (local < 0)
goto tx_error;
- }
-
- /* drop old route */
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
+ if (local)
+ return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
@@ -1082,8 +977,6 @@ ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
LeaveFunction(10);
return NF_STOLEN;
-tx_error_icmp:
- dst_link_failure(skb);
tx_error:
kfree_skb(skb);
LeaveFunction(10);
@@ -1102,10 +995,9 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_iphdr *iph)
{
struct rtable *rt; /* Route to the other host */
- int mtu;
int rc;
int local;
- int rt_mode;
+ int rt_mode, was_input;
EnterFunction(10);
@@ -1125,15 +1017,16 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
/*
* mangle and send the packet here (only for VS/NAT)
*/
+ was_input = rt_is_input_route(skb_rtable(skb));
/* LOCALNODE from FORWARD hook is not supported */
rt_mode = (hooknum != NF_INET_FORWARD) ?
IP_VS_RT_MODE_LOCAL | IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_RDR : IP_VS_RT_MODE_NON_LOCAL;
- if (!(rt = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
- rt_mode, NULL)))
- goto tx_error_icmp;
- local = rt->rt_flags & RTCF_LOCAL;
+ local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip, rt_mode, NULL);
+ if (local < 0)
+ goto tx_error;
+ rt = skb_rtable(skb);
/*
* Avoid duplicate tuple in reply direction for NAT traffic
@@ -1148,71 +1041,49 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_DBG(10, "%s(): "
"stopping DNAT to local address %pI4\n",
__func__, &cp->daddr.ip);
- goto tx_error_put;
+ goto tx_error;
}
}
#endif
/* From world but DNAT to loopback address? */
- if (local && ipv4_is_loopback(cp->daddr.ip) &&
- rt_is_input_route(skb_rtable(skb))) {
+ if (local && ipv4_is_loopback(cp->daddr.ip) && was_input) {
IP_VS_DBG(1, "%s(): "
"stopping DNAT to loopback %pI4\n",
__func__, &cp->daddr.ip);
- goto tx_error_put;
- }
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if ((skb->len > mtu) && (ip_hdr(skb)->frag_off & htons(IP_DF)) &&
- !skb_is_gso(skb)) {
- icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, htonl(mtu));
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
- goto tx_error_put;
+ goto tx_error;
}
/* copy-on-write the packet before mangling it */
if (!skb_make_writable(skb, offset))
- goto tx_error_put;
+ goto tx_error;
if (skb_cow(skb, rt->dst.dev->hard_header_len))
- goto tx_error_put;
+ goto tx_error;
ip_vs_nat_icmp(skb, pp, cp, 0);
- if (!local) {
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
- } else
- ip_rt_put(rt);
-
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
rc = ip_vs_nat_send_or_cont(NFPROTO_IPV4, skb, cp, local);
goto out;
- tx_error_icmp:
- dst_link_failure(skb);
tx_error:
dev_kfree_skb(skb);
rc = NF_STOLEN;
out:
LeaveFunction(10);
return rc;
- tx_error_put:
- ip_rt_put(rt);
- goto tx_error;
}
#ifdef CONFIG_IP_VS_IPV6
int
ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, int offset, unsigned int hooknum,
- struct ip_vs_iphdr *iph)
+ struct ip_vs_iphdr *ipvsh)
{
struct rt6_info *rt; /* Route to the other host */
- int mtu;
int rc;
int local;
int rt_mode;
@@ -1224,7 +1095,7 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
translate address/port back */
if (IP_VS_FWD_METHOD(cp) != IP_VS_CONN_F_MASQ) {
if (cp->packet_xmit)
- rc = cp->packet_xmit(skb, cp, pp, iph);
+ rc = cp->packet_xmit(skb, cp, pp, ipvsh);
else
rc = NF_ACCEPT;
/* do not touch skb anymore */
@@ -1240,11 +1111,11 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
rt_mode = (hooknum != NF_INET_FORWARD) ?
IP_VS_RT_MODE_LOCAL | IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_RDR : IP_VS_RT_MODE_NON_LOCAL;
- if (!(rt = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
- 0, rt_mode)))
- goto tx_error_icmp;
-
- local = __ip_vs_is_local_route6(rt);
+ local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
+ ipvsh, 0, rt_mode);
+ if (local < 0)
+ goto tx_error;
+ rt = (struct rt6_info *) skb_dst(skb);
/*
* Avoid duplicate tuple in reply direction for NAT traffic
* to local address when connection is sync-ed
@@ -1258,7 +1129,7 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_DBG(10, "%s(): "
"stopping DNAT to local address %pI6\n",
__func__, &cp->daddr.in6);
- goto tx_error_put;
+ goto tx_error;
}
}
#endif
@@ -1269,57 +1140,29 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_DBG(1, "%s(): "
"stopping DNAT to loopback %pI6\n",
__func__, &cp->daddr.in6);
- goto tx_error_put;
- }
-
- /* MTU checking */
- mtu = dst_mtu(&rt->dst);
- if (__mtu_check_toobig_v6(skb, mtu)) {
- if (!skb->dev) {
- struct net *net = dev_net(skb_dst(skb)->dev);
-
- skb->dev = net->loopback_dev;
- }
- /* only send ICMP too big on first fragment */
- if (!iph->fragoffs)
- icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
- IP_VS_DBG_RL("%s(): frag needed\n", __func__);
- goto tx_error_put;
+ goto tx_error;
}
/* copy-on-write the packet before mangling it */
if (!skb_make_writable(skb, offset))
- goto tx_error_put;
+ goto tx_error;
if (skb_cow(skb, rt->dst.dev->hard_header_len))
- goto tx_error_put;
+ goto tx_error;
ip_vs_nat_icmp_v6(skb, pp, cp, 0);
- if (!local || !skb->dev) {
- skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
- } else {
- /* destined to loopback, do we need to change route? */
- dst_release(&rt->dst);
- }
-
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
rc = ip_vs_nat_send_or_cont(NFPROTO_IPV6, skb, cp, local);
goto out;
-tx_error_icmp:
- dst_link_failure(skb);
tx_error:
dev_kfree_skb(skb);
rc = NF_STOLEN;
out:
LeaveFunction(10);
return rc;
-tx_error_put:
- dst_release(&rt->dst);
- goto tx_error;
}
#endif
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 09/34] ipvs: optimize dst usage for real server
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (7 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 08/34] ipvs: consolidate all dst checks on transmit in one place Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 10/34] ipvs: convert app locks Simon Horman
` (25 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Currently when forwarding requests to real servers
we use dst_lock and atomic operations when cloning the
dst_cache value. As the dst_cache value does not change
most of the time it is better to use RCU and to lock
dst_lock only when we need to replace the obsoleted dst.
For this to work we keep dst_cache in new structure protected
by RCU. For packets to remote real servers we will use noref
version of dst_cache, it will be valid while we are in RCU
read-side critical section because now dst_release for replaced
dsts will be invoked after the grace period. Packets to
local real servers that are passed to local stack with
NF_ACCEPT need a dst clone.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 12 ++-
net/netfilter/ipvs/ip_vs_core.c | 11 ++-
net/netfilter/ipvs/ip_vs_ctl.c | 25 ++++--
net/netfilter/ipvs/ip_vs_xmit.c | 188 +++++++++++++++++++++++++++++----------
4 files changed, 177 insertions(+), 59 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 8ad73a8..a150ff5 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -724,6 +724,13 @@ struct ip_vs_service {
struct ip_vs_pe *pe;
};
+/* Information for cached dst */
+struct ip_vs_dest_dst {
+ struct dst_entry *dst_cache; /* destination cache entry */
+ u32 dst_cookie;
+ union nf_inet_addr dst_saddr;
+ struct rcu_head rcu_head;
+};
/*
* The real server destination forwarding entry
@@ -752,9 +759,7 @@ struct ip_vs_dest {
/* for destination cache */
spinlock_t dst_lock; /* lock of dst_cache */
- struct dst_entry *dst_cache; /* destination cache entry */
- u32 dst_cookie;
- union nf_inet_addr dst_saddr;
+ struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
/* for virtual service */
struct ip_vs_service *svc; /* service it belongs to */
@@ -1427,6 +1432,7 @@ extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
struct ip_vs_protocol *pp, int offset,
unsigned int hooknum, struct ip_vs_iphdr *iph);
+extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
#ifdef CONFIG_IP_VS_IPV6
extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c
index 2aef23e..6ad24e7 100644
--- a/net/netfilter/ipvs/ip_vs_core.c
+++ b/net/netfilter/ipvs/ip_vs_core.c
@@ -1395,10 +1395,13 @@ ip_vs_in_icmp(struct sk_buff *skb, int *related, unsigned int hooknum)
goto ignore_ipip;
/* Prefer the resulting PMTU */
if (dest) {
- spin_lock(&dest->dst_lock);
- if (dest->dst_cache)
- mtu = dst_mtu(dest->dst_cache);
- spin_unlock(&dest->dst_lock);
+ struct ip_vs_dest_dst *dest_dst;
+
+ rcu_read_lock();
+ dest_dst = rcu_dereference(dest->dest_dst);
+ if (dest_dst)
+ mtu = dst_mtu(dest_dst->dst_cache);
+ rcu_read_unlock();
}
if (mtu > 68 + sizeof(struct iphdr))
mtu -= sizeof(struct iphdr);
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 5265eaa..ef48cc5 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -641,15 +641,26 @@ struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
return dest;
}
-/* Release dst_cache for dest in user context */
+void ip_vs_dest_dst_rcu_free(struct rcu_head *head)
+{
+ struct ip_vs_dest_dst *dest_dst = container_of(head,
+ struct ip_vs_dest_dst,
+ rcu_head);
+
+ dst_release(dest_dst->dst_cache);
+ kfree(dest_dst);
+}
+
+/* Release dest_dst and dst_cache for dest in user context */
static void __ip_vs_dst_cache_reset(struct ip_vs_dest *dest)
{
- struct dst_entry *old_dst;
+ struct ip_vs_dest_dst *old;
- old_dst = dest->dst_cache;
- dest->dst_cache = NULL;
- dst_release(old_dst);
- dest->dst_saddr.ip = 0;
+ old = rcu_dereference_protected(dest->dest_dst, 1);
+ if (old) {
+ RCU_INIT_POINTER(dest->dest_dst, NULL);
+ call_rcu(&old->rcu_head, ip_vs_dest_dst_rcu_free);
+ }
}
/*
@@ -1513,7 +1524,7 @@ static inline void
ip_vs_forget_dev(struct ip_vs_dest *dest, struct net_device *dev)
{
spin_lock_bh(&dest->dst_lock);
- if (dest->dst_cache && dest->dst_cache->dev == dev) {
+ if (dest->dest_dst && dest->dest_dst->dst_cache->dev == dev) {
IP_VS_DBG_BUF(3, "Reset dev:%s dest %s:%u ,dest->refcnt=%d\n",
dev->name,
IP_VS_DBG_ADDR(dest->af, &dest->addr),
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 603eb8a..3db7889 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -17,6 +17,8 @@
* - not all connections have destination server, for example,
* connections in backup server when fwmark is used
* - bypass connections use daddr from packet
+ * - we can use dst without ref while sending in RCU section, we use
+ * ref when returning NF_ACCEPT for NAT-ed packet via loopback
* LOCAL_OUT rules:
* - skb->dev is NULL, skb->protocol is not set (both are set in POST_ROUTING)
* - skb->pkt_type is not set yet
@@ -54,34 +56,51 @@ enum {
IP_VS_RT_MODE_TUNNEL = 32,/* Tunnel mode */
};
+static inline struct ip_vs_dest_dst *ip_vs_dest_dst_alloc(void)
+{
+ return kmalloc(sizeof(struct ip_vs_dest_dst), GFP_ATOMIC);
+}
+
+static inline void ip_vs_dest_dst_free(struct ip_vs_dest_dst *dest_dst)
+{
+ kfree(dest_dst);
+}
+
/*
* Destination cache to speed up outgoing route lookup
*/
static inline void
-__ip_vs_dst_set(struct ip_vs_dest *dest, struct dst_entry *dst, u32 dst_cookie)
+__ip_vs_dst_set(struct ip_vs_dest *dest, struct ip_vs_dest_dst *dest_dst,
+ struct dst_entry *dst, u32 dst_cookie)
{
- struct dst_entry *old_dst;
+ struct ip_vs_dest_dst *old;
+
+ old = rcu_dereference_protected(dest->dest_dst,
+ lockdep_is_held(&dest->dst_lock));
+
+ if (dest_dst) {
+ dest_dst->dst_cache = dst;
+ dest_dst->dst_cookie = dst_cookie;
+ }
+ rcu_assign_pointer(dest->dest_dst, dest_dst);
- old_dst = dest->dst_cache;
- dest->dst_cache = dst;
- dest->dst_cookie = dst_cookie;
- dst_release(old_dst);
+ if (old)
+ call_rcu(&old->rcu_head, ip_vs_dest_dst_rcu_free);
}
-static inline struct dst_entry *
+static inline struct ip_vs_dest_dst *
__ip_vs_dst_check(struct ip_vs_dest *dest)
{
- struct dst_entry *dst = dest->dst_cache;
+ struct ip_vs_dest_dst *dest_dst = rcu_dereference(dest->dest_dst);
+ struct dst_entry *dst;
- if (!dst)
+ if (!dest_dst)
return NULL;
- if (dst->obsolete && dst->ops->check(dst, dest->dst_cookie) == NULL) {
- dest->dst_cache = NULL;
- dst_release(dst);
+ dst = dest_dst->dst_cache;
+ if (dst->obsolete &&
+ dst->ops->check(dst, dest_dst->dst_cookie) == NULL)
return NULL;
- }
- dst_hold(dst);
- return dst;
+ return dest_dst;
}
static inline bool
@@ -144,35 +163,48 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
{
struct net *net = dev_net(skb_dst(skb)->dev);
struct netns_ipvs *ipvs = net_ipvs(net);
+ struct ip_vs_dest_dst *dest_dst;
struct rtable *rt; /* Route to the other host */
struct rtable *ort; /* Original route */
struct iphdr *iph;
__be16 df;
int mtu;
- int local;
+ int local, noref = 1;
if (dest) {
- spin_lock(&dest->dst_lock);
- rt = (struct rtable *) __ip_vs_dst_check(dest);
- if (!rt) {
+ dest_dst = __ip_vs_dst_check(dest);
+ if (likely(dest_dst))
+ rt = (struct rtable *) dest_dst->dst_cache;
+ else {
+ dest_dst = ip_vs_dest_dst_alloc();
+ spin_lock(&dest->dst_lock);
+ if (!dest_dst) {
+ __ip_vs_dst_set(dest, NULL, NULL, 0);
+ spin_unlock(&dest->dst_lock);
+ goto err_unreach;
+ }
rt = do_output_route4(net, dest->addr.ip, rt_mode,
- &dest->dst_saddr.ip);
+ &dest_dst->dst_saddr.ip);
if (!rt) {
+ __ip_vs_dst_set(dest, NULL, NULL, 0);
spin_unlock(&dest->dst_lock);
+ ip_vs_dest_dst_free(dest_dst);
goto err_unreach;
}
- __ip_vs_dst_set(dest, dst_clone(&rt->dst), 0);
+ __ip_vs_dst_set(dest, dest_dst, &rt->dst, 0);
+ spin_unlock(&dest->dst_lock);
IP_VS_DBG(10, "new dst %pI4, src %pI4, refcnt=%d\n",
- &dest->addr.ip, &dest->dst_saddr.ip,
+ &dest->addr.ip, &dest_dst->dst_saddr.ip,
atomic_read(&rt->dst.__refcnt));
}
daddr = dest->addr.ip;
if (ret_saddr)
- *ret_saddr = dest->dst_saddr.ip;
- spin_unlock(&dest->dst_lock);
+ *ret_saddr = dest_dst->dst_saddr.ip;
} else {
__be32 saddr = htonl(INADDR_ANY);
+ noref = 0;
+
/* For such unconfigured boxes avoid many route lookups
* for performance reasons because we do not remember saddr
*/
@@ -210,7 +242,8 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
goto err_put;
}
/* skb to local stack, preserve old route */
- ip_rt_put(rt);
+ if (!noref)
+ ip_rt_put(rt);
return local;
}
@@ -240,12 +273,19 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
}
skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
+ if (noref) {
+ if (!local)
+ skb_dst_set_noref_force(skb, &rt->dst);
+ else
+ skb_dst_set(skb, dst_clone(&rt->dst));
+ } else
+ skb_dst_set(skb, &rt->dst);
return local;
err_put:
- ip_rt_put(rt);
+ if (!noref)
+ ip_rt_put(rt);
return -1;
err_unreach:
@@ -303,36 +343,48 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
struct ip_vs_iphdr *ipvsh, int do_xfrm, int rt_mode)
{
struct net *net = dev_net(skb_dst(skb)->dev);
+ struct ip_vs_dest_dst *dest_dst;
struct rt6_info *rt; /* Route to the other host */
struct rt6_info *ort; /* Original route */
struct dst_entry *dst;
int mtu;
- int local;
+ int local, noref = 1;
if (dest) {
- spin_lock(&dest->dst_lock);
- rt = (struct rt6_info *)__ip_vs_dst_check(dest);
- if (!rt) {
+ dest_dst = __ip_vs_dst_check(dest);
+ if (likely(dest_dst))
+ rt = (struct rt6_info *) dest_dst->dst_cache;
+ else {
u32 cookie;
+ dest_dst = ip_vs_dest_dst_alloc();
+ spin_lock(&dest->dst_lock);
+ if (!dest_dst) {
+ __ip_vs_dst_set(dest, NULL, NULL, 0);
+ spin_unlock(&dest->dst_lock);
+ goto err_unreach;
+ }
dst = __ip_vs_route_output_v6(net, &dest->addr.in6,
- &dest->dst_saddr.in6,
+ &dest_dst->dst_saddr.in6,
do_xfrm);
if (!dst) {
+ __ip_vs_dst_set(dest, NULL, NULL, 0);
spin_unlock(&dest->dst_lock);
+ ip_vs_dest_dst_free(dest_dst);
goto err_unreach;
}
rt = (struct rt6_info *) dst;
cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
- __ip_vs_dst_set(dest, dst_clone(&rt->dst), cookie);
+ __ip_vs_dst_set(dest, dest_dst, &rt->dst, cookie);
+ spin_unlock(&dest->dst_lock);
IP_VS_DBG(10, "new dst %pI6, src %pI6, refcnt=%d\n",
- &dest->addr.in6, &dest->dst_saddr.in6,
+ &dest->addr.in6, &dest_dst->dst_saddr.in6,
atomic_read(&rt->dst.__refcnt));
}
if (ret_saddr)
- *ret_saddr = dest->dst_saddr.in6;
- spin_unlock(&dest->dst_lock);
+ *ret_saddr = dest_dst->dst_saddr.in6;
} else {
+ noref = 0;
dst = __ip_vs_route_output_v6(net, daddr, ret_saddr, do_xfrm);
if (!dst)
goto err_unreach;
@@ -367,7 +419,8 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
goto err_put;
}
/* skb to local stack, preserve old route */
- dst_release(&rt->dst);
+ if (!noref)
+ dst_release(&rt->dst);
return local;
}
@@ -399,12 +452,19 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
}
skb_dst_drop(skb);
- skb_dst_set(skb, &rt->dst);
+ if (noref) {
+ if (!local)
+ skb_dst_set_noref_force(skb, &rt->dst);
+ else
+ skb_dst_set(skb, dst_clone(&rt->dst));
+ } else
+ skb_dst_set(skb, &rt->dst);
return local;
err_put:
- dst_release(&rt->dst);
+ if (!noref)
+ dst_release(&rt->dst);
return -1;
err_unreach:
@@ -494,6 +554,7 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
if (__ip_vs_get_out_rt(skb, NULL, iph->daddr, IP_VS_RT_MODE_NON_LOCAL,
NULL) < 0)
goto tx_error;
@@ -504,12 +565,14 @@ ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 0);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -521,6 +584,7 @@ ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
{
EnterFunction(10);
+ rcu_read_lock();
if (__ip_vs_get_out_rt_v6(skb, NULL, &ipvsh->daddr.in6, NULL,
ipvsh, 0, IP_VS_RT_MODE_NON_LOCAL) < 0)
goto tx_error;
@@ -529,12 +593,14 @@ ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 0);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -553,6 +619,7 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
/* check if it is a connection of no-client-port */
if (unlikely(cp->flags & IP_VS_CONN_F_NO_CPORT)) {
__be16 _pt, *p;
@@ -620,12 +687,14 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
rc = ip_vs_nat_send_or_cont(NFPROTO_IPV4, skb, cp, local);
+ rcu_read_unlock();
LeaveFunction(10);
return rc;
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -640,6 +709,7 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
/* check if it is a connection of no-client-port */
if (unlikely(cp->flags & IP_VS_CONN_F_NO_CPORT && !ipvsh->fragoffs)) {
__be16 _pt, *p;
@@ -707,6 +777,7 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
rc = ip_vs_nat_send_or_cont(NFPROTO_IPV6, skb, cp, local);
+ rcu_read_unlock();
LeaveFunction(10);
return rc;
@@ -714,6 +785,7 @@ ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
tx_error:
LeaveFunction(10);
kfree_skb(skb);
+ rcu_read_unlock();
return NF_STOLEN;
}
#endif
@@ -755,6 +827,7 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
IP_VS_RT_MODE_LOCAL |
IP_VS_RT_MODE_NON_LOCAL |
@@ -762,8 +835,10 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_RT_MODE_TUNNEL, &saddr);
if (local < 0)
goto tx_error;
- if (local)
+ if (local) {
+ rcu_read_unlock();
return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
+ }
rt = skb_rtable(skb);
tdev = rt->dst.dev;
@@ -818,6 +893,7 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
ip_local_out(skb);
else if (ret == NF_DROP)
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
@@ -825,6 +901,7 @@ ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -844,6 +921,7 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6,
&saddr, ipvsh, 1,
IP_VS_RT_MODE_LOCAL |
@@ -851,8 +929,10 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
IP_VS_RT_MODE_TUNNEL);
if (local < 0)
goto tx_error;
- if (local)
+ if (local) {
+ rcu_read_unlock();
return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
+ }
rt = (struct rt6_info *) skb_dst(skb);
tdev = rt->dst.dev;
@@ -901,6 +981,7 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
ip6_local_out(skb);
else if (ret == NF_DROP)
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
@@ -908,6 +989,7 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -926,14 +1008,17 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip,
IP_VS_RT_MODE_LOCAL |
IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_KNOWN_NH, NULL);
if (local < 0)
goto tx_error;
- if (local)
+ if (local) {
+ rcu_read_unlock();
return ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 1);
+ }
ip_send_check(ip_hdr(skb));
@@ -941,12 +1026,14 @@ ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
ip_vs_send_or_cont(NFPROTO_IPV4, skb, cp, 0);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -960,25 +1047,30 @@ ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
EnterFunction(10);
+ rcu_read_lock();
local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
ipvsh, 0,
IP_VS_RT_MODE_LOCAL |
IP_VS_RT_MODE_NON_LOCAL);
if (local < 0)
goto tx_error;
- if (local)
+ if (local) {
+ rcu_read_unlock();
return ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 1);
+ }
/* Another hack: avoid icmp_send in ip_fragment */
skb->local_df = 1;
ip_vs_send_or_cont(NFPROTO_IPV6, skb, cp, 0);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
tx_error:
kfree_skb(skb);
+ rcu_read_unlock();
LeaveFunction(10);
return NF_STOLEN;
}
@@ -1023,6 +1115,7 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
rt_mode = (hooknum != NF_INET_FORWARD) ?
IP_VS_RT_MODE_LOCAL | IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_RDR : IP_VS_RT_MODE_NON_LOCAL;
+ rcu_read_lock();
local = __ip_vs_get_out_rt(skb, cp->dest, cp->daddr.ip, rt_mode, NULL);
if (local < 0)
goto tx_error;
@@ -1067,10 +1160,12 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
rc = ip_vs_nat_send_or_cont(NFPROTO_IPV4, skb, cp, local);
+ rcu_read_unlock();
goto out;
tx_error:
- dev_kfree_skb(skb);
+ kfree_skb(skb);
+ rcu_read_unlock();
rc = NF_STOLEN;
out:
LeaveFunction(10);
@@ -1111,6 +1206,7 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
rt_mode = (hooknum != NF_INET_FORWARD) ?
IP_VS_RT_MODE_LOCAL | IP_VS_RT_MODE_NON_LOCAL |
IP_VS_RT_MODE_RDR : IP_VS_RT_MODE_NON_LOCAL;
+ rcu_read_lock();
local = __ip_vs_get_out_rt_v6(skb, cp->dest, &cp->daddr.in6, NULL,
ipvsh, 0, rt_mode);
if (local < 0)
@@ -1156,10 +1252,12 @@ ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
skb->local_df = 1;
rc = ip_vs_nat_send_or_cont(NFPROTO_IPV6, skb, cp, local);
+ rcu_read_unlock();
goto out;
tx_error:
- dev_kfree_skb(skb);
+ kfree_skb(skb);
+ rcu_read_unlock();
rc = NF_STOLEN;
out:
LeaveFunction(10);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 10/34] ipvs: convert app locks
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (8 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 09/34] ipvs: optimize dst usage for real server Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 11/34] ipvs: remove rs_lock by using RCU Simon Horman
` (24 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
We use locks like tcp_app_lock, udp_app_lock,
sctp_app_lock to protect access to the protocol hash tables
from readers in packet context while the application
instances (inc) are [un]registered under global mutex.
As the hash tables are mostly read when conns are
created and bound to app, use RCU for readers and reclaim
app instance after grace period.
Simplify ip_vs_app_inc_get because we use usecnt
only for statistics and rely on module refcounting.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 4 +---
net/netfilter/ipvs/ip_vs_app.c | 27 +++++++++++++++++++--------
net/netfilter/ipvs/ip_vs_ftp.c | 2 ++
net/netfilter/ipvs/ip_vs_proto_sctp.c | 18 ++++++------------
net/netfilter/ipvs/ip_vs_proto_tcp.c | 18 ++++++------------
net/netfilter/ipvs/ip_vs_proto_udp.c | 19 ++++++-------------
6 files changed, 40 insertions(+), 48 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index a150ff5..84ca171 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -823,6 +823,7 @@ struct ip_vs_app {
struct ip_vs_app *app; /* its real application */
__be16 port; /* port number in net order */
atomic_t usecnt; /* usage counter */
+ struct rcu_head rcu_head;
/*
* output hook: Process packet in inout direction, diff set for TCP.
@@ -908,7 +909,6 @@ struct netns_ipvs {
#define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
#define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
struct list_head tcp_apps[TCP_APP_TAB_SIZE];
- spinlock_t tcp_app_lock;
#endif
/* ip_vs_proto_udp */
#ifdef CONFIG_IP_VS_PROTO_UDP
@@ -916,7 +916,6 @@ struct netns_ipvs {
#define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
#define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
struct list_head udp_apps[UDP_APP_TAB_SIZE];
- spinlock_t udp_app_lock;
#endif
/* ip_vs_proto_sctp */
#ifdef CONFIG_IP_VS_PROTO_SCTP
@@ -925,7 +924,6 @@ struct netns_ipvs {
#define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
/* Hash table for SCTP application incarnations */
struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
- spinlock_t sctp_app_lock;
#endif
/* ip_vs_conn */
atomic_t conn_count; /* connection counter */
diff --git a/net/netfilter/ipvs/ip_vs_app.c b/net/netfilter/ipvs/ip_vs_app.c
index 0b779d7..a956030 100644
--- a/net/netfilter/ipvs/ip_vs_app.c
+++ b/net/netfilter/ipvs/ip_vs_app.c
@@ -58,6 +58,18 @@ static inline void ip_vs_app_put(struct ip_vs_app *app)
module_put(app->module);
}
+static void ip_vs_app_inc_destroy(struct ip_vs_app *inc)
+{
+ kfree(inc->timeout_table);
+ kfree(inc);
+}
+
+static void ip_vs_app_inc_rcu_free(struct rcu_head *head)
+{
+ struct ip_vs_app *inc = container_of(head, struct ip_vs_app, rcu_head);
+
+ ip_vs_app_inc_destroy(inc);
+}
/*
* Allocate/initialize app incarnation and register it in proto apps.
@@ -106,8 +118,7 @@ ip_vs_app_inc_new(struct net *net, struct ip_vs_app *app, __u16 proto,
return 0;
out:
- kfree(inc->timeout_table);
- kfree(inc);
+ ip_vs_app_inc_destroy(inc);
return ret;
}
@@ -131,8 +142,7 @@ ip_vs_app_inc_release(struct net *net, struct ip_vs_app *inc)
list_del(&inc->a_list);
- kfree(inc->timeout_table);
- kfree(inc);
+ call_rcu(&inc->rcu_head, ip_vs_app_inc_rcu_free);
}
@@ -144,9 +154,9 @@ int ip_vs_app_inc_get(struct ip_vs_app *inc)
{
int result;
- atomic_inc(&inc->usecnt);
- if (unlikely((result = ip_vs_app_get(inc->app)) != 1))
- atomic_dec(&inc->usecnt);
+ result = ip_vs_app_get(inc->app);
+ if (result)
+ atomic_inc(&inc->usecnt);
return result;
}
@@ -156,8 +166,8 @@ int ip_vs_app_inc_get(struct ip_vs_app *inc)
*/
void ip_vs_app_inc_put(struct ip_vs_app *inc)
{
- ip_vs_app_put(inc->app);
atomic_dec(&inc->usecnt);
+ ip_vs_app_put(inc->app);
}
@@ -218,6 +228,7 @@ out_unlock:
/*
* ip_vs_app unregistration routine
* We are sure there are no app incarnations attached to services
+ * Caller should use synchronize_rcu() or rcu_barrier()
*/
void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app)
{
diff --git a/net/netfilter/ipvs/ip_vs_ftp.c b/net/netfilter/ipvs/ip_vs_ftp.c
index 4f53a5f..7f90825 100644
--- a/net/netfilter/ipvs/ip_vs_ftp.c
+++ b/net/netfilter/ipvs/ip_vs_ftp.c
@@ -480,6 +480,7 @@ static int __init ip_vs_ftp_init(void)
int rv;
rv = register_pernet_subsys(&ip_vs_ftp_ops);
+ /* rcu_barrier() is called by netns on error */
return rv;
}
@@ -489,6 +490,7 @@ static int __init ip_vs_ftp_init(void)
static void __exit ip_vs_ftp_exit(void)
{
unregister_pernet_subsys(&ip_vs_ftp_ops);
+ /* rcu_barrier() is called by netns */
}
diff --git a/net/netfilter/ipvs/ip_vs_proto_sctp.c b/net/netfilter/ipvs/ip_vs_proto_sctp.c
index cd1d729..f7190cd 100644
--- a/net/netfilter/ipvs/ip_vs_proto_sctp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_sctp.c
@@ -1016,30 +1016,25 @@ static int sctp_register_app(struct net *net, struct ip_vs_app *inc)
hash = sctp_app_hashkey(port);
- spin_lock_bh(&ipvs->sctp_app_lock);
list_for_each_entry(i, &ipvs->sctp_apps[hash], p_list) {
if (i->port == port) {
ret = -EEXIST;
goto out;
}
}
- list_add(&inc->p_list, &ipvs->sctp_apps[hash]);
+ list_add_rcu(&inc->p_list, &ipvs->sctp_apps[hash]);
atomic_inc(&pd->appcnt);
out:
- spin_unlock_bh(&ipvs->sctp_app_lock);
return ret;
}
static void sctp_unregister_app(struct net *net, struct ip_vs_app *inc)
{
- struct netns_ipvs *ipvs = net_ipvs(net);
struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_SCTP);
- spin_lock_bh(&ipvs->sctp_app_lock);
atomic_dec(&pd->appcnt);
- list_del(&inc->p_list);
- spin_unlock_bh(&ipvs->sctp_app_lock);
+ list_del_rcu(&inc->p_list);
}
static int sctp_app_conn_bind(struct ip_vs_conn *cp)
@@ -1055,12 +1050,12 @@ static int sctp_app_conn_bind(struct ip_vs_conn *cp)
/* Lookup application incarnations and bind the right one */
hash = sctp_app_hashkey(cp->vport);
- spin_lock(&ipvs->sctp_app_lock);
- list_for_each_entry(inc, &ipvs->sctp_apps[hash], p_list) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(inc, &ipvs->sctp_apps[hash], p_list) {
if (inc->port == cp->vport) {
if (unlikely(!ip_vs_app_inc_get(inc)))
break;
- spin_unlock(&ipvs->sctp_app_lock);
+ rcu_read_unlock();
IP_VS_DBG_BUF(9, "%s: Binding conn %s:%u->"
"%s:%u to app %s on port %u\n",
@@ -1076,7 +1071,7 @@ static int sctp_app_conn_bind(struct ip_vs_conn *cp)
goto out;
}
}
- spin_unlock(&ipvs->sctp_app_lock);
+ rcu_read_unlock();
out:
return result;
}
@@ -1090,7 +1085,6 @@ static int __ip_vs_sctp_init(struct net *net, struct ip_vs_proto_data *pd)
struct netns_ipvs *ipvs = net_ipvs(net);
ip_vs_init_hash_table(ipvs->sctp_apps, SCTP_APP_TAB_SIZE);
- spin_lock_init(&ipvs->sctp_app_lock);
pd->timeout_table = ip_vs_create_timeout_table((int *)sctp_timeouts,
sizeof(sctp_timeouts));
if (!pd->timeout_table)
diff --git a/net/netfilter/ipvs/ip_vs_proto_tcp.c b/net/netfilter/ipvs/ip_vs_proto_tcp.c
index 9af653a..0bbc3fe 100644
--- a/net/netfilter/ipvs/ip_vs_proto_tcp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_tcp.c
@@ -580,18 +580,16 @@ static int tcp_register_app(struct net *net, struct ip_vs_app *inc)
hash = tcp_app_hashkey(port);
- spin_lock_bh(&ipvs->tcp_app_lock);
list_for_each_entry(i, &ipvs->tcp_apps[hash], p_list) {
if (i->port == port) {
ret = -EEXIST;
goto out;
}
}
- list_add(&inc->p_list, &ipvs->tcp_apps[hash]);
+ list_add_rcu(&inc->p_list, &ipvs->tcp_apps[hash]);
atomic_inc(&pd->appcnt);
out:
- spin_unlock_bh(&ipvs->tcp_app_lock);
return ret;
}
@@ -599,13 +597,10 @@ static int tcp_register_app(struct net *net, struct ip_vs_app *inc)
static void
tcp_unregister_app(struct net *net, struct ip_vs_app *inc)
{
- struct netns_ipvs *ipvs = net_ipvs(net);
struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_TCP);
- spin_lock_bh(&ipvs->tcp_app_lock);
atomic_dec(&pd->appcnt);
- list_del(&inc->p_list);
- spin_unlock_bh(&ipvs->tcp_app_lock);
+ list_del_rcu(&inc->p_list);
}
@@ -624,12 +619,12 @@ tcp_app_conn_bind(struct ip_vs_conn *cp)
/* Lookup application incarnations and bind the right one */
hash = tcp_app_hashkey(cp->vport);
- spin_lock(&ipvs->tcp_app_lock);
- list_for_each_entry(inc, &ipvs->tcp_apps[hash], p_list) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(inc, &ipvs->tcp_apps[hash], p_list) {
if (inc->port == cp->vport) {
if (unlikely(!ip_vs_app_inc_get(inc)))
break;
- spin_unlock(&ipvs->tcp_app_lock);
+ rcu_read_unlock();
IP_VS_DBG_BUF(9, "%s(): Binding conn %s:%u->"
"%s:%u to app %s on port %u\n",
@@ -646,7 +641,7 @@ tcp_app_conn_bind(struct ip_vs_conn *cp)
goto out;
}
}
- spin_unlock(&ipvs->tcp_app_lock);
+ rcu_read_unlock();
out:
return result;
@@ -676,7 +671,6 @@ static int __ip_vs_tcp_init(struct net *net, struct ip_vs_proto_data *pd)
struct netns_ipvs *ipvs = net_ipvs(net);
ip_vs_init_hash_table(ipvs->tcp_apps, TCP_APP_TAB_SIZE);
- spin_lock_init(&ipvs->tcp_app_lock);
pd->timeout_table = ip_vs_create_timeout_table((int *)tcp_timeouts,
sizeof(tcp_timeouts));
if (!pd->timeout_table)
diff --git a/net/netfilter/ipvs/ip_vs_proto_udp.c b/net/netfilter/ipvs/ip_vs_proto_udp.c
index 503a842..1a03e2d 100644
--- a/net/netfilter/ipvs/ip_vs_proto_udp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_udp.c
@@ -359,19 +359,16 @@ static int udp_register_app(struct net *net, struct ip_vs_app *inc)
hash = udp_app_hashkey(port);
-
- spin_lock_bh(&ipvs->udp_app_lock);
list_for_each_entry(i, &ipvs->udp_apps[hash], p_list) {
if (i->port == port) {
ret = -EEXIST;
goto out;
}
}
- list_add(&inc->p_list, &ipvs->udp_apps[hash]);
+ list_add_rcu(&inc->p_list, &ipvs->udp_apps[hash]);
atomic_inc(&pd->appcnt);
out:
- spin_unlock_bh(&ipvs->udp_app_lock);
return ret;
}
@@ -380,12 +377,9 @@ static void
udp_unregister_app(struct net *net, struct ip_vs_app *inc)
{
struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_UDP);
- struct netns_ipvs *ipvs = net_ipvs(net);
- spin_lock_bh(&ipvs->udp_app_lock);
atomic_dec(&pd->appcnt);
- list_del(&inc->p_list);
- spin_unlock_bh(&ipvs->udp_app_lock);
+ list_del_rcu(&inc->p_list);
}
@@ -403,12 +397,12 @@ static int udp_app_conn_bind(struct ip_vs_conn *cp)
/* Lookup application incarnations and bind the right one */
hash = udp_app_hashkey(cp->vport);
- spin_lock(&ipvs->udp_app_lock);
- list_for_each_entry(inc, &ipvs->udp_apps[hash], p_list) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(inc, &ipvs->udp_apps[hash], p_list) {
if (inc->port == cp->vport) {
if (unlikely(!ip_vs_app_inc_get(inc)))
break;
- spin_unlock(&ipvs->udp_app_lock);
+ rcu_read_unlock();
IP_VS_DBG_BUF(9, "%s(): Binding conn %s:%u->"
"%s:%u to app %s on port %u\n",
@@ -425,7 +419,7 @@ static int udp_app_conn_bind(struct ip_vs_conn *cp)
goto out;
}
}
- spin_unlock(&ipvs->udp_app_lock);
+ rcu_read_unlock();
out:
return result;
@@ -467,7 +461,6 @@ static int __udp_init(struct net *net, struct ip_vs_proto_data *pd)
struct netns_ipvs *ipvs = net_ipvs(net);
ip_vs_init_hash_table(ipvs->udp_apps, UDP_APP_TAB_SIZE);
- spin_lock_init(&ipvs->udp_app_lock);
pd->timeout_table = ip_vs_create_timeout_table((int *)udp_timeouts,
sizeof(udp_timeouts));
if (!pd->timeout_table)
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 11/34] ipvs: remove rs_lock by using RCU
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (9 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 10/34] ipvs: convert app locks Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 12/34] ipvs: convert locks used in persistence engines Simon Horman
` (23 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
rs_lock was used to protect rs_table (hash table)
from updaters (under global mutex) and readers (packet handlers).
We can remove rs_lock by using RCU lock for readers. Reclaiming
dest only with kfree_rcu is enough because the readers access
only fields from the ip_vs_dest structure.
Use hlist for rs_table.
As we are now using hlist_del_rcu, introduce in_rs_table
flag as replacement for the list_empty checks which do not
work with RCU. It is needed because only NAT dests are in
the rs_table.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 14 ++++---
net/netfilter/ipvs/ip_vs_core.c | 5 +--
net/netfilter/ipvs/ip_vs_ctl.c | 88 ++++++++++++++-------------------------
3 files changed, 41 insertions(+), 66 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 84ca171..b06aa6c 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -738,7 +738,7 @@ struct ip_vs_dest_dst {
*/
struct ip_vs_dest {
struct list_head n_list; /* for the dests in the service */
- struct list_head d_list; /* for table with all the dests */
+ struct hlist_node d_list; /* for table with all the dests */
u16 af; /* address family */
__be16 port; /* port number of the server */
@@ -767,6 +767,9 @@ struct ip_vs_dest {
__be16 vport; /* virtual port number */
union nf_inet_addr vaddr; /* virtual IP address */
__u32 vfwmark; /* firewall mark of service */
+
+ struct rcu_head rcu_head;
+ unsigned int in_rs_table:1; /* we are in rs_table */
};
@@ -897,7 +900,7 @@ struct netns_ipvs {
#define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
#define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
- struct list_head rs_table[IP_VS_RTAB_SIZE];
+ struct hlist_head rs_table[IP_VS_RTAB_SIZE];
/* ip_vs_app */
struct list_head app_list;
/* ip_vs_proto */
@@ -933,7 +936,6 @@ struct netns_ipvs {
int num_services; /* no of virtual services */
- rwlock_t rs_lock; /* real services table */
/* Trash for destinations */
struct list_head dest_trash;
/* Service counters */
@@ -1376,9 +1378,9 @@ static inline void ip_vs_service_put(struct ip_vs_service *svc)
atomic_dec(&svc->usecnt);
}
-extern struct ip_vs_dest *
-ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
- const union nf_inet_addr *daddr, __be16 dport);
+extern bool
+ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
+ const union nf_inet_addr *daddr, __be16 dport);
extern int ip_vs_use_count_inc(void);
extern void ip_vs_use_count_dec(void);
diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c
index 6ad24e7..4fc749c 100644
--- a/net/netfilter/ipvs/ip_vs_core.c
+++ b/net/netfilter/ipvs/ip_vs_core.c
@@ -1161,9 +1161,8 @@ ip_vs_out(unsigned int hooknum, struct sk_buff *skb, int af)
sizeof(_ports), _ports, &iph);
if (pptr == NULL)
return NF_ACCEPT; /* Not for me */
- if (ip_vs_lookup_real_service(net, af, iph.protocol,
- &iph.saddr,
- pptr[0])) {
+ if (ip_vs_has_real_service(net, af, iph.protocol, &iph.saddr,
+ pptr[0])) {
/*
* Notify the real server: there is no
* existing entry if it is not RST
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index ef48cc5..182d958 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -508,17 +508,13 @@ static inline unsigned int ip_vs_rs_hashkey(int af,
& IP_VS_RTAB_MASK;
}
-/*
- * Hashes ip_vs_dest in rs_table by <proto,addr,port>.
- * should be called with locked tables.
- */
-static int ip_vs_rs_hash(struct netns_ipvs *ipvs, struct ip_vs_dest *dest)
+/* Hash ip_vs_dest in rs_table by <proto,addr,port>. */
+static void ip_vs_rs_hash(struct netns_ipvs *ipvs, struct ip_vs_dest *dest)
{
unsigned int hash;
- if (!list_empty(&dest->d_list)) {
- return 0;
- }
+ if (dest->in_rs_table)
+ return;
/*
* Hash by proto,addr,port,
@@ -526,60 +522,47 @@ static int ip_vs_rs_hash(struct netns_ipvs *ipvs, struct ip_vs_dest *dest)
*/
hash = ip_vs_rs_hashkey(dest->af, &dest->addr, dest->port);
- list_add(&dest->d_list, &ipvs->rs_table[hash]);
-
- return 1;
+ hlist_add_head_rcu(&dest->d_list, &ipvs->rs_table[hash]);
+ dest->in_rs_table = 1;
}
-/*
- * UNhashes ip_vs_dest from rs_table.
- * should be called with locked tables.
- */
-static int ip_vs_rs_unhash(struct ip_vs_dest *dest)
+/* Unhash ip_vs_dest from rs_table. */
+static void ip_vs_rs_unhash(struct ip_vs_dest *dest)
{
/*
* Remove it from the rs_table table.
*/
- if (!list_empty(&dest->d_list)) {
- list_del_init(&dest->d_list);
+ if (dest->in_rs_table) {
+ hlist_del_rcu(&dest->d_list);
+ dest->in_rs_table = 0;
}
-
- return 1;
}
-/*
- * Lookup real service by <proto,addr,port> in the real service table.
- */
-struct ip_vs_dest *
-ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
- const union nf_inet_addr *daddr,
- __be16 dport)
+/* Check if real service by <proto,addr,port> is present */
+bool ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
+ const union nf_inet_addr *daddr, __be16 dport)
{
struct netns_ipvs *ipvs = net_ipvs(net);
unsigned int hash;
struct ip_vs_dest *dest;
- /*
- * Check for "full" addressed entries
- * Return the first found entry
- */
+ /* Check for "full" addressed entries */
hash = ip_vs_rs_hashkey(af, daddr, dport);
- read_lock(&ipvs->rs_lock);
- list_for_each_entry(dest, &ipvs->rs_table[hash], d_list) {
- if ((dest->af == af)
- && ip_vs_addr_equal(af, &dest->addr, daddr)
- && (dest->port == dport)
- && ((dest->protocol == protocol) ||
- dest->vfwmark)) {
+ rcu_read_lock();
+ hlist_for_each_entry_rcu(dest, &ipvs->rs_table[hash], d_list) {
+ if (dest->port == dport &&
+ dest->af == af &&
+ ip_vs_addr_equal(af, &dest->addr, daddr) &&
+ (dest->protocol == protocol || dest->vfwmark)) {
/* HIT */
- read_unlock(&ipvs->rs_lock);
- return dest;
+ rcu_read_unlock();
+ return true;
}
}
- read_unlock(&ipvs->rs_lock);
+ rcu_read_unlock();
- return NULL;
+ return false;
}
/*
@@ -612,9 +595,6 @@ ip_vs_lookup_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
* the backup synchronization daemon. It finds the
* destination to be bound to the received connection
* on the backup.
- *
- * ip_vs_lookup_real_service() looked promissing, but
- * seems not working as expected.
*/
struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
const union nf_inet_addr *daddr,
@@ -715,7 +695,7 @@ ip_vs_trash_get_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
__ip_vs_dst_cache_reset(dest);
__ip_vs_unbind_svc(dest);
free_percpu(dest->stats.cpustats);
- kfree(dest);
+ kfree_rcu(dest, rcu_head);
}
}
@@ -742,7 +722,7 @@ static void ip_vs_trash_cleanup(struct net *net)
__ip_vs_dst_cache_reset(dest);
__ip_vs_unbind_svc(dest);
free_percpu(dest->stats.cpustats);
- kfree(dest);
+ kfree_rcu(dest, rcu_head);
}
}
@@ -807,9 +787,7 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
* Put the real service in rs_table if not present.
* For now only for NAT!
*/
- write_lock_bh(&ipvs->rs_lock);
ip_vs_rs_hash(ipvs, dest);
- write_unlock_bh(&ipvs->rs_lock);
}
atomic_set(&dest->conn_flags, conn_flags);
@@ -905,7 +883,7 @@ ip_vs_new_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest,
atomic_set(&dest->persistconns, 0);
atomic_set(&dest->refcnt, 1);
- INIT_LIST_HEAD(&dest->d_list);
+ INIT_HLIST_NODE(&dest->d_list);
spin_lock_init(&dest->dst_lock);
spin_lock_init(&dest->stats.lock);
__ip_vs_update_dest(svc, dest, udest, 1);
@@ -1045,9 +1023,7 @@ static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest)
/*
* Remove it from the d-linked list with the real services.
*/
- write_lock_bh(&ipvs->rs_lock);
ip_vs_rs_unhash(dest);
- write_unlock_bh(&ipvs->rs_lock);
/*
* Decrease the refcnt of the dest, and free the dest
@@ -1067,7 +1043,7 @@ static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest)
time, so the operation here is OK */
atomic_dec(&dest->svc->refcnt);
free_percpu(dest->stats.cpustats);
- kfree(dest);
+ kfree_rcu(dest, rcu_head);
} else {
IP_VS_DBG_BUF(3, "Moving dest %s:%u into trash, "
"dest->refcnt=%d\n",
@@ -3811,11 +3787,9 @@ int __net_init ip_vs_control_net_init(struct net *net)
int idx;
struct netns_ipvs *ipvs = net_ipvs(net);
- rwlock_init(&ipvs->rs_lock);
-
/* Initialize rs_table */
for (idx = 0; idx < IP_VS_RTAB_SIZE; idx++)
- INIT_LIST_HEAD(&ipvs->rs_table[idx]);
+ INIT_HLIST_HEAD(&ipvs->rs_table[idx]);
INIT_LIST_HEAD(&ipvs->dest_trash);
atomic_set(&ipvs->ftpsvc_counter, 0);
@@ -3892,7 +3866,7 @@ int __init ip_vs_control_init(void)
EnterFunction(2);
- /* Initialize svc_table, ip_vs_svc_fwm_table, rs_table */
+ /* Initialize svc_table, ip_vs_svc_fwm_table */
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
INIT_LIST_HEAD(&ip_vs_svc_table[idx]);
INIT_LIST_HEAD(&ip_vs_svc_fwm_table[idx]);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 12/34] ipvs: convert locks used in persistence engines
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (10 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 11/34] ipvs: remove rs_lock by using RCU Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 13/34] ipvs: convert connection locking Simon Horman
` (22 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Allow the readers to use RCU lock and for
PE module registrations use global mutex instead of
spinlock. All PE modules need to use synchronize_rcu
in their module exit handler.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_pe.c | 43 +++++++++++--------------------------
net/netfilter/ipvs/ip_vs_pe_sip.c | 1 +
2 files changed, 14 insertions(+), 30 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_pe.c b/net/netfilter/ipvs/ip_vs_pe.c
index 5cf859c..5d9774c 100644
--- a/net/netfilter/ipvs/ip_vs_pe.c
+++ b/net/netfilter/ipvs/ip_vs_pe.c
@@ -13,8 +13,8 @@
/* IPVS pe list */
static LIST_HEAD(ip_vs_pe);
-/* lock for service table */
-static DEFINE_SPINLOCK(ip_vs_pe_lock);
+/* semaphore for IPVS PEs. */
+static DEFINE_MUTEX(ip_vs_pe_mutex);
/* Bind a service with a pe */
void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe)
@@ -36,9 +36,8 @@ struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name)
IP_VS_DBG(10, "%s(): pe_name \"%s\"\n", __func__,
pe_name);
- spin_lock_bh(&ip_vs_pe_lock);
-
- list_for_each_entry(pe, &ip_vs_pe, n_list) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(pe, &ip_vs_pe, n_list) {
/* Test and get the modules atomically */
if (pe->module &&
!try_module_get(pe->module)) {
@@ -47,14 +46,14 @@ struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name)
}
if (strcmp(pe_name, pe->name)==0) {
/* HIT */
- spin_unlock_bh(&ip_vs_pe_lock);
+ rcu_read_unlock();
return pe;
}
if (pe->module)
module_put(pe->module);
}
+ rcu_read_unlock();
- spin_unlock_bh(&ip_vs_pe_lock);
return NULL;
}
@@ -83,22 +82,13 @@ int register_ip_vs_pe(struct ip_vs_pe *pe)
/* increase the module use count */
ip_vs_use_count_inc();
- spin_lock_bh(&ip_vs_pe_lock);
-
- if (!list_empty(&pe->n_list)) {
- spin_unlock_bh(&ip_vs_pe_lock);
- ip_vs_use_count_dec();
- pr_err("%s(): [%s] pe already linked\n",
- __func__, pe->name);
- return -EINVAL;
- }
-
+ mutex_lock(&ip_vs_pe_mutex);
/* Make sure that the pe with this name doesn't exist
* in the pe list.
*/
list_for_each_entry(tmp, &ip_vs_pe, n_list) {
if (strcmp(tmp->name, pe->name) == 0) {
- spin_unlock_bh(&ip_vs_pe_lock);
+ mutex_unlock(&ip_vs_pe_mutex);
ip_vs_use_count_dec();
pr_err("%s(): [%s] pe already existed "
"in the system\n", __func__, pe->name);
@@ -106,8 +96,8 @@ int register_ip_vs_pe(struct ip_vs_pe *pe)
}
}
/* Add it into the d-linked pe list */
- list_add(&pe->n_list, &ip_vs_pe);
- spin_unlock_bh(&ip_vs_pe_lock);
+ list_add_rcu(&pe->n_list, &ip_vs_pe);
+ mutex_unlock(&ip_vs_pe_mutex);
pr_info("[%s] pe registered.\n", pe->name);
@@ -118,17 +108,10 @@ EXPORT_SYMBOL_GPL(register_ip_vs_pe);
/* Unregister a pe from the pe list */
int unregister_ip_vs_pe(struct ip_vs_pe *pe)
{
- spin_lock_bh(&ip_vs_pe_lock);
- if (list_empty(&pe->n_list)) {
- spin_unlock_bh(&ip_vs_pe_lock);
- pr_err("%s(): [%s] pe is not in the list. failed\n",
- __func__, pe->name);
- return -EINVAL;
- }
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 13/34] ipvs: convert connection locking
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (11 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 12/34] ipvs: convert locks used in persistence engines Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 14/34] ipvs: reorder keys in connection structure Simon Horman
` (21 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Convert __ip_vs_conntbl_lock_array as follows:
- readers that do not modify conn lists will use RCU lock
- updaters that modify lists will use spinlock_t
Now for conn lookups we will use RCU read-side
critical section. Without using __ip_vs_conn_get such
places have access to connection fields and can
dereference some pointers like pe and pe_data plus
the ability to update timer expiration. If full access
is required we contend for reference.
We add barrier in __ip_vs_conn_put, so that
other CPUs see the refcnt operation after other writes.
With the introduction of ip_vs_conn_unlink()
we try to reorganize ip_vs_conn_expire(), so that
unhashing of connections that should stay more time is
avoided, even if it is for very short time.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 12 ++
net/netfilter/ipvs/ip_vs_conn.c | 230 +++++++++++++++++++++------------------
2 files changed, 134 insertions(+), 108 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index b06aa6c..5700b07 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -620,6 +620,8 @@ struct ip_vs_conn {
const struct ip_vs_pe *pe;
char *pe_data;
__u8 pe_data_len;
+
+ struct rcu_head rcu_head;
};
/*
@@ -1185,9 +1187,19 @@ struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
const struct ip_vs_iphdr *iph,
int inverse);
+/* Get reference to gain full access to conn.
+ * By default, RCU read-side critical sections have access only to
+ * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
+ */
+static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
+{
+ return atomic_inc_not_zero(&cp->refcnt);
+}
+
/* put back the conn without restarting its timer */
static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
{
+ smp_mb__before_atomic_dec();
atomic_dec(&cp->refcnt);
}
extern void ip_vs_conn_put(struct ip_vs_conn *cp);
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index 704e514..b0cd2be 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -79,51 +79,21 @@ static unsigned int ip_vs_conn_rnd __read_mostly;
struct ip_vs_aligned_lock
{
- rwlock_t l;
+ spinlock_t l;
} __attribute__((__aligned__(SMP_CACHE_BYTES)));
/* lock array for conn table */
static struct ip_vs_aligned_lock
__ip_vs_conntbl_lock_array[CT_LOCKARRAY_SIZE] __cacheline_aligned;
-static inline void ct_read_lock(unsigned int key)
-{
- read_lock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
-}
-
-static inline void ct_read_unlock(unsigned int key)
-{
- read_unlock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
-}
-
static inline void ct_write_lock(unsigned int key)
{
- write_lock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
+ spin_lock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
}
static inline void ct_write_unlock(unsigned int key)
{
- write_unlock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
-}
-
-static inline void ct_read_lock_bh(unsigned int key)
-{
- read_lock_bh(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
-}
-
-static inline void ct_read_unlock_bh(unsigned int key)
-{
- read_unlock_bh(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
-}
-
-static inline void ct_write_lock_bh(unsigned int key)
-{
- write_lock_bh(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
-}
-
-static inline void ct_write_unlock_bh(unsigned int key)
-{
- write_unlock_bh(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
+ spin_unlock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
}
@@ -201,9 +171,9 @@ static inline int ip_vs_conn_hash(struct ip_vs_conn *cp)
spin_lock(&cp->lock);
if (!(cp->flags & IP_VS_CONN_F_HASHED)) {
- hlist_add_head(&cp->c_list, &ip_vs_conn_tab[hash]);
cp->flags |= IP_VS_CONN_F_HASHED;
atomic_inc(&cp->refcnt);
+ hlist_add_head_rcu(&cp->c_list, &ip_vs_conn_tab[hash]);
ret = 1;
} else {
pr_err("%s(): request for already hashed, called from %pF\n",
@@ -220,7 +190,7 @@ static inline int ip_vs_conn_hash(struct ip_vs_conn *cp)
/*
* UNhashes ip_vs_conn from ip_vs_conn_tab.
- * returns bool success.
+ * returns bool success. Caller should hold conn reference.
*/
static inline int ip_vs_conn_unhash(struct ip_vs_conn *cp)
{
@@ -234,7 +204,7 @@ static inline int ip_vs_conn_unhash(struct ip_vs_conn *cp)
spin_lock(&cp->lock);
if (cp->flags & IP_VS_CONN_F_HASHED) {
- hlist_del(&cp->c_list);
+ hlist_del_rcu(&cp->c_list);
cp->flags &= ~IP_VS_CONN_F_HASHED;
atomic_dec(&cp->refcnt);
ret = 1;
@@ -247,6 +217,36 @@ static inline int ip_vs_conn_unhash(struct ip_vs_conn *cp)
return ret;
}
+/* Try to unlink ip_vs_conn from ip_vs_conn_tab.
+ * returns bool success.
+ */
+static inline bool ip_vs_conn_unlink(struct ip_vs_conn *cp)
+{
+ unsigned int hash;
+ bool ret;
+
+ hash = ip_vs_conn_hashkey_conn(cp);
+
+ ct_write_lock(hash);
+ spin_lock(&cp->lock);
+
+ if (cp->flags & IP_VS_CONN_F_HASHED) {
+ ret = false;
+ /* Decrease refcnt and unlink conn only if we are last user */
+ if (atomic_cmpxchg(&cp->refcnt, 1, 0) == 1) {
+ hlist_del_rcu(&cp->c_list);
+ cp->flags &= ~IP_VS_CONN_F_HASHED;
+ ret = true;
+ }
+ } else
+ ret = atomic_read(&cp->refcnt) ? false : true;
+
+ spin_unlock(&cp->lock);
+ ct_write_unlock(hash);
+
+ return ret;
+}
+
/*
* Gets ip_vs_conn associated with supplied parameters in the ip_vs_conn_tab.
@@ -262,9 +262,9 @@ __ip_vs_conn_in_get(const struct ip_vs_conn_param *p)
hash = ip_vs_conn_hashkey_param(p, false);
- ct_read_lock(hash);
+ rcu_read_lock();
- hlist_for_each_entry(cp, &ip_vs_conn_tab[hash], c_list) {
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
if (cp->af == p->af &&
p->cport == cp->cport && p->vport == cp->vport &&
ip_vs_addr_equal(p->af, p->caddr, &cp->caddr) &&
@@ -272,14 +272,15 @@ __ip_vs_conn_in_get(const struct ip_vs_conn_param *p)
((!p->cport) ^ (!(cp->flags & IP_VS_CONN_F_NO_CPORT))) &&
p->protocol == cp->protocol &&
ip_vs_conn_net_eq(cp, p->net)) {
+ if (!__ip_vs_conn_get(cp))
+ continue;
/* HIT */
- atomic_inc(&cp->refcnt);
- ct_read_unlock(hash);
+ rcu_read_unlock();
return cp;
}
}
- ct_read_unlock(hash);
+ rcu_read_unlock();
return NULL;
}
@@ -346,14 +347,16 @@ struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p)
hash = ip_vs_conn_hashkey_param(p, false);
- ct_read_lock(hash);
+ rcu_read_lock();
- hlist_for_each_entry(cp, &ip_vs_conn_tab[hash], c_list) {
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
if (!ip_vs_conn_net_eq(cp, p->net))
continue;
if (p->pe_data && p->pe->ct_match) {
- if (p->pe == cp->pe && p->pe->ct_match(p, cp))
- goto out;
+ if (p->pe == cp->pe && p->pe->ct_match(p, cp)) {
+ if (__ip_vs_conn_get(cp))
+ goto out;
+ }
continue;
}
@@ -365,15 +368,15 @@ struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p)
p->af, p->vaddr, &cp->vaddr) &&
p->cport == cp->cport && p->vport == cp->vport &&
cp->flags & IP_VS_CONN_F_TEMPLATE &&
- p->protocol == cp->protocol)
- goto out;
+ p->protocol == cp->protocol) {
+ if (__ip_vs_conn_get(cp))
+ goto out;
+ }
}
cp = NULL;
out:
- if (cp)
- atomic_inc(&cp->refcnt);
- ct_read_unlock(hash);
+ rcu_read_unlock();
IP_VS_DBG_BUF(9, "template lookup/in %s %s:%d->%s:%d %s\n",
ip_vs_proto_name(p->protocol),
@@ -398,23 +401,24 @@ struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p)
*/
hash = ip_vs_conn_hashkey_param(p, true);
- ct_read_lock(hash);
+ rcu_read_lock();
- hlist_for_each_entry(cp, &ip_vs_conn_tab[hash], c_list) {
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
if (cp->af == p->af &&
p->vport == cp->cport && p->cport == cp->dport &&
ip_vs_addr_equal(p->af, p->vaddr, &cp->caddr) &&
ip_vs_addr_equal(p->af, p->caddr, &cp->daddr) &&
p->protocol == cp->protocol &&
ip_vs_conn_net_eq(cp, p->net)) {
+ if (!__ip_vs_conn_get(cp))
+ continue;
/* HIT */
- atomic_inc(&cp->refcnt);
ret = cp;
break;
}
}
- ct_read_unlock(hash);
+ rcu_read_unlock();
IP_VS_DBG_BUF(9, "lookup/out %s %s:%d->%s:%d %s\n",
ip_vs_proto_name(p->protocol),
@@ -757,41 +761,36 @@ int ip_vs_check_template(struct ip_vs_conn *ct)
* Simply decrease the refcnt of the template,
* don't restart its timer.
*/
- atomic_dec(&ct->refcnt);
+ __ip_vs_conn_put(ct);
return 0;
}
return 1;
}
+static void ip_vs_conn_rcu_free(struct rcu_head *head)
+{
+ struct ip_vs_conn *cp = container_of(head, struct ip_vs_conn,
+ rcu_head);
+
+ ip_vs_pe_put(cp->pe);
+ kfree(cp->pe_data);
+ kmem_cache_free(ip_vs_conn_cachep, cp);
+}
+
static void ip_vs_conn_expire(unsigned long data)
{
struct ip_vs_conn *cp = (struct ip_vs_conn *)data;
struct net *net = ip_vs_conn_net(cp);
struct netns_ipvs *ipvs = net_ipvs(net);
- cp->timeout = 60*HZ;
-
- /*
- * hey, I'm using it
- */
- atomic_inc(&cp->refcnt);
-
/*
* do I control anybody?
*/
if (atomic_read(&cp->n_control))
goto expire_later;
- /*
- * unhash it if it is hashed in the conn table
- */
- if (!ip_vs_conn_unhash(cp) && !(cp->flags & IP_VS_CONN_F_ONE_PACKET))
- goto expire_later;
-
- /*
- * refcnt==1 implies I'm the only one referrer
- */
- if (likely(atomic_read(&cp->refcnt) == 1)) {
+ /* Unlink conn if not referenced anymore */
+ if (likely(ip_vs_conn_unlink(cp))) {
/* delete the timer if it is activated by other users */
del_timer(&cp->timer);
@@ -810,38 +809,41 @@ static void ip_vs_conn_expire(unsigned long data)
ip_vs_conn_drop_conntrack(cp);
}
- ip_vs_pe_put(cp->pe);
- kfree(cp->pe_data);
if (unlikely(cp->app != NULL))
ip_vs_unbind_app(cp);
ip_vs_unbind_dest(cp);
if (cp->flags & IP_VS_CONN_F_NO_CPORT)
atomic_dec(&ip_vs_conn_no_cport_cnt);
+ call_rcu(&cp->rcu_head, ip_vs_conn_rcu_free);
atomic_dec(&ipvs->conn_count);
-
- kmem_cache_free(ip_vs_conn_cachep, cp);
return;
}
- /* hash it back to the table */
- ip_vs_conn_hash(cp);
-
expire_later:
- IP_VS_DBG(7, "delayed: conn->refcnt-1=%d conn->n_control=%d\n",
- atomic_read(&cp->refcnt)-1,
+ IP_VS_DBG(7, "delayed: conn->refcnt=%d conn->n_control=%d\n",
+ atomic_read(&cp->refcnt),
atomic_read(&cp->n_control));
+ atomic_inc(&cp->refcnt);
+ cp->timeout = 60*HZ;
+
if (ipvs->sync_state & IP_VS_STATE_MASTER)
ip_vs_sync_conn(net, cp, sysctl_sync_threshold(ipvs));
ip_vs_conn_put(cp);
}
-
+/* Modify timer, so that it expires as soon as possible.
+ * Can be called without reference only if under RCU lock.
+ */
void ip_vs_conn_expire_now(struct ip_vs_conn *cp)
{
- if (del_timer(&cp->timer))
- mod_timer(&cp->timer, jiffies);
+ /* Using mod_timer_pending will ensure the timer is not
+ * modified after the final del_timer in ip_vs_conn_expire.
+ */
+ if (timer_pending(&cp->timer) &&
+ time_after(cp->timer.expires, jiffies))
+ mod_timer_pending(&cp->timer, jiffies);
}
@@ -952,14 +954,17 @@ static void *ip_vs_conn_array(struct seq_file *seq, loff_t pos)
struct ip_vs_iter_state *iter = seq->private;
for (idx = 0; idx < ip_vs_conn_tab_size; idx++) {
- ct_read_lock_bh(idx);
- hlist_for_each_entry(cp, &ip_vs_conn_tab[idx], c_list) {
+ rcu_read_lock();
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[idx], c_list) {
+ /* __ip_vs_conn_get() is not needed by
+ * ip_vs_conn_seq_show and ip_vs_conn_sync_seq_show
+ */
if (pos-- == 0) {
iter->l = &ip_vs_conn_tab[idx];
return cp;
}
}
- ct_read_unlock_bh(idx);
+ rcu_read_unlock();
}
return NULL;
@@ -977,6 +982,7 @@ static void *ip_vs_conn_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct ip_vs_conn *cp = v;
struct ip_vs_iter_state *iter = seq->private;
+ struct hlist_node *e;
struct hlist_head *l = iter->l;
int idx;
@@ -985,19 +991,19 @@ static void *ip_vs_conn_seq_next(struct seq_file *seq, void *v, loff_t *pos)
return ip_vs_conn_array(seq, 0);
/* more on same hash chain? */
- if (cp->c_list.next)
- return hlist_entry(cp->c_list.next, struct ip_vs_conn, c_list);
+ e = rcu_dereference(hlist_next_rcu(&cp->c_list));
+ if (e)
+ return hlist_entry(e, struct ip_vs_conn, c_list);
+ rcu_read_unlock();
idx = l - ip_vs_conn_tab;
- ct_read_unlock_bh(idx);
-
while (++idx < ip_vs_conn_tab_size) {
- ct_read_lock_bh(idx);
- hlist_for_each_entry(cp, &ip_vs_conn_tab[idx], c_list) {
+ rcu_read_lock();
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[idx], c_list) {
iter->l = &ip_vs_conn_tab[idx];
return cp;
}
- ct_read_unlock_bh(idx);
+ rcu_read_unlock();
}
iter->l = NULL;
return NULL;
@@ -1009,7 +1015,7 @@ static void ip_vs_conn_seq_stop(struct seq_file *seq, void *v)
struct hlist_head *l = iter->l;
if (l)
- ct_read_unlock_bh(l - ip_vs_conn_tab);
+ rcu_read_unlock();
}
static int ip_vs_conn_seq_show(struct seq_file *seq, void *v)
@@ -1188,7 +1194,7 @@ static inline int todrop_entry(struct ip_vs_conn *cp)
void ip_vs_random_dropentry(struct net *net)
{
int idx;
- struct ip_vs_conn *cp;
+ struct ip_vs_conn *cp, *cp_c;
/*
* Randomly scan 1/32 of the whole table every second
@@ -1199,9 +1205,9 @@ void ip_vs_random_dropentry(struct net *net)
/*
* Lock is actually needed in this loop.
*/
- ct_write_lock_bh(hash);
+ rcu_read_lock();
- hlist_for_each_entry(cp, &ip_vs_conn_tab[hash], c_list) {
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
if (cp->flags & IP_VS_CONN_F_TEMPLATE)
/* connection template */
continue;
@@ -1228,12 +1234,15 @@ void ip_vs_random_dropentry(struct net *net)
IP_VS_DBG(4, "del connection\n");
ip_vs_conn_expire_now(cp);
- if (cp->control) {
+ cp_c = cp->control;
+ /* cp->control is valid only with reference to cp */
+ if (cp_c && __ip_vs_conn_get(cp)) {
IP_VS_DBG(4, "del conn template\n");
- ip_vs_conn_expire_now(cp->control);
+ ip_vs_conn_expire_now(cp_c);
+ __ip_vs_conn_put(cp);
}
}
- ct_write_unlock_bh(hash);
+ rcu_read_unlock();
}
}
@@ -1244,7 +1253,7 @@ void ip_vs_random_dropentry(struct net *net)
static void ip_vs_conn_flush(struct net *net)
{
int idx;
- struct ip_vs_conn *cp;
+ struct ip_vs_conn *cp, *cp_c;
struct netns_ipvs *ipvs = net_ipvs(net);
flush_again:
@@ -1252,19 +1261,22 @@ flush_again:
/*
* Lock is actually needed in this loop.
*/
- ct_write_lock_bh(idx);
+ rcu_read_lock();
- hlist_for_each_entry(cp, &ip_vs_conn_tab[idx], c_list) {
+ hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[idx], c_list) {
if (!ip_vs_conn_net_eq(cp, net))
continue;
IP_VS_DBG(4, "del connection\n");
ip_vs_conn_expire_now(cp);
- if (cp->control) {
+ cp_c = cp->control;
+ /* cp->control is valid only with reference to cp */
+ if (cp_c && __ip_vs_conn_get(cp)) {
IP_VS_DBG(4, "del conn template\n");
- ip_vs_conn_expire_now(cp->control);
+ ip_vs_conn_expire_now(cp_c);
+ __ip_vs_conn_put(cp);
}
}
- ct_write_unlock_bh(idx);
+ rcu_read_unlock();
}
/* the counter may be not NULL, because maybe some conn entries
@@ -1331,7 +1343,7 @@ int __init ip_vs_conn_init(void)
INIT_HLIST_HEAD(&ip_vs_conn_tab[idx]);
for (idx = 0; idx < CT_LOCKARRAY_SIZE; idx++) {
- rwlock_init(&__ip_vs_conntbl_lock_array[idx].l);
+ spin_lock_init(&__ip_vs_conntbl_lock_array[idx].l);
}
/* calculate the random value for connection hash */
@@ -1342,6 +1354,8 @@ int __init ip_vs_conn_init(void)
void ip_vs_conn_cleanup(void)
{
+ /* Wait all ip_vs_conn_rcu_free() callbacks to complete */
+ rcu_barrier();
/* Release the empty cache */
kmem_cache_destroy(ip_vs_conn_cachep);
vfree(ip_vs_conn_tab);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 14/34] ipvs: reorder keys in connection structure
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (12 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 13/34] ipvs: convert connection locking Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 15/34] ipvs: avoid kmem_cache_zalloc in ip_vs_conn_new Simon Horman
` (20 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
__ip_vs_conn_in_get and ip_vs_conn_out_get are
hot places. Optimize them, so that ports are matched first.
By moving net and fwmark below, on 32-bit arch we can fit
caddr in 32-byte cache line and all addresses in 64-byte
cache line.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 12 ++++++------
net/netfilter/ipvs/ip_vs_conn.c | 19 ++++++++++---------
2 files changed, 16 insertions(+), 15 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 5700b07..929e04c 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -566,20 +566,19 @@ struct ip_vs_conn_param {
*/
struct ip_vs_conn {
struct hlist_node c_list; /* hashed list heads */
-#ifdef CONFIG_NET_NS
- struct net *net; /* Name space */
-#endif
/* Protocol, addresses and port numbers */
- u16 af; /* address family */
__be16 cport;
- __be16 vport;
__be16 dport;
- __u32 fwmark; /* Fire wall mark from skb */
+ __be16 vport;
+ u16 af; /* address family */
union nf_inet_addr caddr; /* client address */
union nf_inet_addr vaddr; /* virtual address */
union nf_inet_addr daddr; /* destination address */
volatile __u32 flags; /* status flags */
__u16 protocol; /* Which protocol (TCP/UDP) */
+#ifdef CONFIG_NET_NS
+ struct net *net; /* Name space */
+#endif
/* counter and timer */
atomic_t refcnt; /* reference count */
@@ -593,6 +592,7 @@ struct ip_vs_conn {
* state transition triggerd
* synchronization
*/
+ __u32 fwmark; /* Fire wall mark from skb */
unsigned long sync_endtime; /* jiffies + sent_retries */
/* Control members */
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index b0cd2be..416015b 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -265,8 +265,8 @@ __ip_vs_conn_in_get(const struct ip_vs_conn_param *p)
rcu_read_lock();
hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
- if (cp->af == p->af &&
- p->cport == cp->cport && p->vport == cp->vport &&
+ if (p->cport == cp->cport && p->vport == cp->vport &&
+ cp->af == p->af &&
ip_vs_addr_equal(p->af, p->caddr, &cp->caddr) &&
ip_vs_addr_equal(p->af, p->vaddr, &cp->vaddr) &&
((!p->cport) ^ (!(cp->flags & IP_VS_CONN_F_NO_CPORT))) &&
@@ -350,9 +350,9 @@ struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p)
rcu_read_lock();
hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
- if (!ip_vs_conn_net_eq(cp, p->net))
- continue;
- if (p->pe_data && p->pe->ct_match) {
+ if (unlikely(p->pe_data && p->pe->ct_match)) {
+ if (!ip_vs_conn_net_eq(cp, p->net))
+ continue;
if (p->pe == cp->pe && p->pe->ct_match(p, cp)) {
if (__ip_vs_conn_get(cp))
goto out;
@@ -366,9 +366,10 @@ struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p)
* p->vaddr is a fwmark */
ip_vs_addr_equal(p->protocol == IPPROTO_IP ? AF_UNSPEC :
p->af, p->vaddr, &cp->vaddr) &&
- p->cport == cp->cport && p->vport == cp->vport &&
+ p->vport == cp->vport && p->cport == cp->cport &&
cp->flags & IP_VS_CONN_F_TEMPLATE &&
- p->protocol == cp->protocol) {
+ p->protocol == cp->protocol &&
+ ip_vs_conn_net_eq(cp, p->net)) {
if (__ip_vs_conn_get(cp))
goto out;
}
@@ -404,8 +405,8 @@ struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p)
rcu_read_lock();
hlist_for_each_entry_rcu(cp, &ip_vs_conn_tab[hash], c_list) {
- if (cp->af == p->af &&
- p->vport == cp->cport && p->cport == cp->dport &&
+ if (p->vport == cp->cport && p->cport == cp->dport &&
+ cp->af == p->af &&
ip_vs_addr_equal(p->af, p->vaddr, &cp->caddr) &&
ip_vs_addr_equal(p->af, p->caddr, &cp->daddr) &&
p->protocol == cp->protocol &&
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 15/34] ipvs: avoid kmem_cache_zalloc in ip_vs_conn_new
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (13 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 14/34] ipvs: reorder keys in connection structure Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 16/34] ipvs: change ip_vs_sched_lock to mutex Simon Horman
` (19 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
We have many fields to set and few to reset,
use kmem_cache_alloc instead to save some cycles.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 15 +++++++++++++++
net/netfilter/ipvs/ip_vs_conn.c | 24 +++++++++++++++++++-----
2 files changed, 34 insertions(+), 5 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 929e04c..43886bb 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -233,6 +233,21 @@ static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
dst->ip = src->ip;
}
+static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
+ const union nf_inet_addr *src)
+{
+#ifdef CONFIG_IP_VS_IPV6
+ if (af == AF_INET6) {
+ dst->in6 = src->in6;
+ return;
+ }
+#endif
+ dst->ip = src->ip;
+ dst->all[1] = 0;
+ dst->all[2] = 0;
+ dst->all[3] = 0;
+}
+
static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
const union nf_inet_addr *b)
{
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index 416015b..e3e2b4d 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -861,7 +861,7 @@ ip_vs_conn_new(const struct ip_vs_conn_param *p,
struct ip_vs_proto_data *pd = ip_vs_proto_data_get(p->net,
p->protocol);
- cp = kmem_cache_zalloc(ip_vs_conn_cachep, GFP_ATOMIC);
+ cp = kmem_cache_alloc(ip_vs_conn_cachep, GFP_ATOMIC);
if (cp == NULL) {
IP_VS_ERR_RL("%s(): no memory\n", __func__);
return NULL;
@@ -872,13 +872,13 @@ ip_vs_conn_new(const struct ip_vs_conn_param *p,
ip_vs_conn_net_set(cp, p->net);
cp->af = p->af;
cp->protocol = p->protocol;
- ip_vs_addr_copy(p->af, &cp->caddr, p->caddr);
+ ip_vs_addr_set(p->af, &cp->caddr, p->caddr);
cp->cport = p->cport;
- ip_vs_addr_copy(p->af, &cp->vaddr, p->vaddr);
+ ip_vs_addr_set(p->af, &cp->vaddr, p->vaddr);
cp->vport = p->vport;
/* proto should only be IPPROTO_IP if d_addr is a fwmark */
- ip_vs_addr_copy(p->protocol == IPPROTO_IP ? AF_UNSPEC : p->af,
- &cp->daddr, daddr);
+ ip_vs_addr_set(p->protocol == IPPROTO_IP ? AF_UNSPEC : p->af,
+ &cp->daddr, daddr);
cp->dport = dport;
cp->flags = flags;
cp->fwmark = fwmark;
@@ -887,6 +887,10 @@ ip_vs_conn_new(const struct ip_vs_conn_param *p,
cp->pe = p->pe;
cp->pe_data = p->pe_data;
cp->pe_data_len = p->pe_data_len;
+ } else {
+ cp->pe = NULL;
+ cp->pe_data = NULL;
+ cp->pe_data_len = 0;
}
spin_lock_init(&cp->lock);
@@ -897,18 +901,28 @@ ip_vs_conn_new(const struct ip_vs_conn_param *p,
*/
atomic_set(&cp->refcnt, 1);
+ cp->control = NULL;
atomic_set(&cp->n_control, 0);
atomic_set(&cp->in_pkts, 0);
+ cp->packet_xmit = NULL;
+ cp->app = NULL;
+ cp->app_data = NULL;
+ /* reset struct ip_vs_seq */
+ cp->in_seq.delta = 0;
+ cp->out_seq.delta = 0;
+
atomic_inc(&ipvs->conn_count);
if (flags & IP_VS_CONN_F_NO_CPORT)
atomic_inc(&ip_vs_conn_no_cport_cnt);
/* Bind the connection with a destination server */
+ cp->dest = NULL;
ip_vs_bind_dest(cp, dest);
/* Set its state and timeout */
cp->state = 0;
+ cp->old_state = 0;
cp->timeout = 3*HZ;
cp->sync_endtime = jiffies & ~3UL;
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 16/34] ipvs: change ip_vs_sched_lock to mutex
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (14 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 15/34] ipvs: avoid kmem_cache_zalloc in ip_vs_conn_new Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 17/34] ipvs: preparations for using rcu in schedulers Simon Horman
` (18 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The global list with schedulers ip_vs_schedulers
is accessed only from user context - configuration and
scheduler module [un]registration. Use ip_vs_sched_mutex
instead.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_sched.c | 24 ++++++++++++------------
1 file changed, 12 insertions(+), 12 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_sched.c b/net/netfilter/ipvs/ip_vs_sched.c
index d6bf20d..7f11d3d 100644
--- a/net/netfilter/ipvs/ip_vs_sched.c
+++ b/net/netfilter/ipvs/ip_vs_sched.c
@@ -35,8 +35,8 @@ EXPORT_SYMBOL(ip_vs_scheduler_err);
*/
static LIST_HEAD(ip_vs_schedulers);
-/* lock for service table */
-static DEFINE_SPINLOCK(ip_vs_sched_lock);
+/* semaphore for schedulers */
+static DEFINE_MUTEX(ip_vs_sched_mutex);
/*
@@ -92,7 +92,7 @@ static struct ip_vs_scheduler *ip_vs_sched_getbyname(const char *sched_name)
IP_VS_DBG(2, "%s(): sched_name \"%s\"\n", __func__, sched_name);
- spin_lock_bh(&ip_vs_sched_lock);
+ mutex_lock(&ip_vs_sched_mutex);
list_for_each_entry(sched, &ip_vs_schedulers, n_list) {
/*
@@ -106,14 +106,14 @@ static struct ip_vs_scheduler *ip_vs_sched_getbyname(const char *sched_name)
}
if (strcmp(sched_name, sched->name)==0) {
/* HIT */
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
return sched;
}
if (sched->module)
module_put(sched->module);
}
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
return NULL;
}
@@ -192,10 +192,10 @@ int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler)
/* increase the module use count */
ip_vs_use_count_inc();
- spin_lock_bh(&ip_vs_sched_lock);
+ mutex_lock(&ip_vs_sched_mutex);
if (!list_empty(&scheduler->n_list)) {
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
ip_vs_use_count_dec();
pr_err("%s(): [%s] scheduler already linked\n",
__func__, scheduler->name);
@@ -208,7 +208,7 @@ int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler)
*/
list_for_each_entry(sched, &ip_vs_schedulers, n_list) {
if (strcmp(scheduler->name, sched->name) == 0) {
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
ip_vs_use_count_dec();
pr_err("%s(): [%s] scheduler already existed "
"in the system\n", __func__, scheduler->name);
@@ -219,7 +219,7 @@ int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler)
* Add it into the d-linked scheduler list
*/
list_add(&scheduler->n_list, &ip_vs_schedulers);
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
pr_info("[%s] scheduler registered.\n", scheduler->name);
@@ -237,9 +237,9 @@ int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler)
return -EINVAL;
}
- spin_lock_bh(&ip_vs_sched_lock);
+ mutex_lock(&ip_vs_sched_mutex);
if (list_empty(&scheduler->n_list)) {
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
pr_err("%s(): [%s] scheduler is not in the list. failed\n",
__func__, scheduler->name);
return -EINVAL;
@@ -249,7 +249,7 @@ int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler)
* Remove it from the d-linked scheduler list
*/
list_del(&scheduler->n_list);
- spin_unlock_bh(&ip_vs_sched_lock);
+ mutex_unlock(&ip_vs_sched_mutex);
/* decrease the module use count */
ip_vs_use_count_dec();
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 17/34] ipvs: preparations for using rcu in schedulers
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (15 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 16/34] ipvs: change ip_vs_sched_lock to mutex Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 18/34] ipvs: add ip_vs_dest_hold and ip_vs_dest_put Simon Horman
` (17 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Allow schedulers to use rcu_dereference when
returning destination on lookup. The RCU read-side critical
section will allow ip_vs_bind_dest to get dest refcnt as
preparation for the step where destinations will be
deleted without an IP_VS_WAIT_WHILE guard that holds the
packet processing during update.
Add new optional scheduler methods add_dest,
del_dest and upd_dest. For now the methods are called
together with update_service but update_service will be
removed in a following change.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 6 ++++++
net/netfilter/ipvs/ip_vs_core.c | 6 ++++++
net/netfilter/ipvs/ip_vs_ctl.c | 8 ++++++++
3 files changed, 20 insertions(+)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 43886bb..d91385c 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -805,6 +805,12 @@ struct ip_vs_scheduler {
int (*done_service)(struct ip_vs_service *svc);
/* scheduler updating service */
int (*update_service)(struct ip_vs_service *svc);
+ /* dest is linked */
+ int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
+ /* dest is unlinked */
+ int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
+ /* dest is updated */
+ int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
/* selecting a server from the given service */
struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c
index 4fc749c..939ad11 100644
--- a/net/netfilter/ipvs/ip_vs_core.c
+++ b/net/netfilter/ipvs/ip_vs_core.c
@@ -301,8 +301,10 @@ ip_vs_sched_persist(struct ip_vs_service *svc,
* template is not available.
* return *ignored=0 i.e. ICMP and NF_DROP
*/
+ rcu_read_lock();
dest = svc->scheduler->schedule(svc, skb);
if (!dest) {
+ rcu_read_unlock();
IP_VS_DBG(1, "p-schedule: no dest found.\n");
kfree(param.pe_data);
*ignored = 0;
@@ -318,6 +320,7 @@ ip_vs_sched_persist(struct ip_vs_service *svc,
* when the template expires */
ct = ip_vs_conn_new(¶m, &dest->addr, dport,
IP_VS_CONN_F_TEMPLATE, dest, skb->mark);
+ rcu_read_unlock();
if (ct == NULL) {
kfree(param.pe_data);
*ignored = -1;
@@ -446,8 +449,10 @@ ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
return NULL;
}
+ rcu_read_lock();
dest = svc->scheduler->schedule(svc, skb);
if (dest == NULL) {
+ rcu_read_unlock();
IP_VS_DBG(1, "Schedule: no dest found.\n");
return NULL;
}
@@ -468,6 +473,7 @@ ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
cp = ip_vs_conn_new(&p, &dest->addr,
dest->port ? dest->port : pptr[1],
flags, dest, skb->mark);
+ rcu_read_unlock();
if (!cp) {
*ignored = -1;
return NULL;
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 182d958..d64f800 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -825,6 +825,11 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
if (add) {
list_add(&dest->n_list, &svc->destinations);
svc->num_dests++;
+ if (svc->scheduler->add_dest)
+ svc->scheduler->add_dest(svc, dest);
+ } else {
+ if (svc->scheduler->upd_dest)
+ svc->scheduler->upd_dest(svc, dest);
}
/* call the update_service, because server weight may be changed */
@@ -1071,6 +1076,9 @@ static void __ip_vs_unlink_dest(struct ip_vs_service *svc,
list_del(&dest->n_list);
svc->num_dests--;
+ if (svcupd && svc->scheduler->del_dest)
+ svc->scheduler->del_dest(svc, dest);
+
/*
* Call the update_service function of its scheduler
*/
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 18/34] ipvs: add ip_vs_dest_hold and ip_vs_dest_put
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (16 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 17/34] ipvs: preparations for using rcu in schedulers Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 19/34] ipvs: convert dh scheduler to rcu Simon Horman
` (16 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
ip_vs_dest_hold will be used under RCU lock
while ip_vs_dest_put can be called even after dest
is removed from service, as it happens for conns and
some schedulers.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 10 ++++++++++
net/netfilter/ipvs/ip_vs_conn.c | 9 ++-------
net/netfilter/ipvs/ip_vs_ctl.c | 4 ++--
net/netfilter/ipvs/ip_vs_sync.c | 4 ++--
4 files changed, 16 insertions(+), 11 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index d91385c..7d3027f 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -1427,6 +1427,16 @@ ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
__u16 protocol, __u32 fwmark, __u32 flags);
extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
+static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
+{
+ atomic_inc(&dest->refcnt);
+}
+
+static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
+{
+ smp_mb__before_atomic_dec();
+ atomic_dec(&dest->refcnt);
+}
/*
* IPVS sync daemon data and function prototypes
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index e3e2b4d..1b29e4a 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -554,7 +554,7 @@ ip_vs_bind_dest(struct ip_vs_conn *cp, struct ip_vs_dest *dest)
return;
/* Increase the refcnt counter of the dest */
- atomic_inc(&dest->refcnt);
+ ip_vs_dest_hold(dest);
conn_flags = atomic_read(&dest->conn_flags);
if (cp->protocol != IPPROTO_UDP)
@@ -700,12 +700,7 @@ static inline void ip_vs_unbind_dest(struct ip_vs_conn *cp)
dest->flags &= ~IP_VS_DEST_F_OVERLOAD;
}
- /*
- * Simply decrease the refcnt of the dest, because the
- * dest will be either in service's destination list
- * or in the trash.
- */
- atomic_dec(&dest->refcnt);
+ ip_vs_dest_put(dest);
}
static int expire_quiescent_template(struct netns_ipvs *ipvs,
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index d64f800..a4f6388 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -616,7 +616,7 @@ struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
if (!dest)
dest = ip_vs_lookup_dest(svc, daddr, port ^ dport);
if (dest)
- atomic_inc(&dest->refcnt);
+ ip_vs_dest_hold(dest);
ip_vs_service_put(svc);
return dest;
}
@@ -1056,7 +1056,7 @@ static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest)
ntohs(dest->port),
atomic_read(&dest->refcnt));
list_add(&dest->n_list, &ipvs->dest_trash);
- atomic_inc(&dest->refcnt);
+ ip_vs_dest_hold(dest);
}
}
diff --git a/net/netfilter/ipvs/ip_vs_sync.c b/net/netfilter/ipvs/ip_vs_sync.c
index 44fd10c..6cc3e52 100644
--- a/net/netfilter/ipvs/ip_vs_sync.c
+++ b/net/netfilter/ipvs/ip_vs_sync.c
@@ -861,7 +861,7 @@ static void ip_vs_proc_conn(struct net *net, struct ip_vs_conn_param *param,
if (!dest) {
dest = ip_vs_try_bind_dest(cp);
if (dest)
- atomic_dec(&dest->refcnt);
+ ip_vs_dest_put(dest);
}
} else {
/*
@@ -874,7 +874,7 @@ static void ip_vs_proc_conn(struct net *net, struct ip_vs_conn_param *param,
cp = ip_vs_conn_new(param, daddr, dport, flags, dest, fwmark);
if (dest)
- atomic_dec(&dest->refcnt);
+ ip_vs_dest_put(dest);
if (!cp) {
if (param->pe_data)
kfree(param->pe_data);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 19/34] ipvs: convert dh scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (17 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 18/34] ipvs: add ip_vs_dest_hold and ip_vs_dest_put Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 20/34] ipvs: convert lblc " Simon Horman
` (15 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Use the new add_dest and del_dest methods
to reassign dests.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_dh.c | 81 +++++++++++++++++++++++------------------
1 file changed, 45 insertions(+), 36 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_dh.c b/net/netfilter/ipvs/ip_vs_dh.c
index 7f3b0cc..ebe80f4 100644
--- a/net/netfilter/ipvs/ip_vs_dh.c
+++ b/net/netfilter/ipvs/ip_vs_dh.c
@@ -51,7 +51,7 @@
* IPVS DH bucket
*/
struct ip_vs_dh_bucket {
- struct ip_vs_dest *dest; /* real server (cache) */
+ struct ip_vs_dest __rcu *dest; /* real server (cache) */
};
/*
@@ -64,6 +64,10 @@ struct ip_vs_dh_bucket {
#define IP_VS_DH_TAB_SIZE (1 << IP_VS_DH_TAB_BITS)
#define IP_VS_DH_TAB_MASK (IP_VS_DH_TAB_SIZE - 1)
+struct ip_vs_dh_state {
+ struct ip_vs_dh_bucket buckets[IP_VS_DH_TAB_SIZE];
+ struct rcu_head rcu_head;
+};
/*
* Returns hash value for IPVS DH entry
@@ -85,10 +89,9 @@ static inline unsigned int ip_vs_dh_hashkey(int af, const union nf_inet_addr *ad
* Get ip_vs_dest associated with supplied parameters.
*/
static inline struct ip_vs_dest *
-ip_vs_dh_get(int af, struct ip_vs_dh_bucket *tbl,
- const union nf_inet_addr *addr)
+ip_vs_dh_get(int af, struct ip_vs_dh_state *s, const union nf_inet_addr *addr)
{
- return (tbl[ip_vs_dh_hashkey(af, addr)]).dest;
+ return rcu_dereference(s->buckets[ip_vs_dh_hashkey(af, addr)].dest);
}
@@ -96,25 +99,30 @@ ip_vs_dh_get(int af, struct ip_vs_dh_bucket *tbl,
* Assign all the hash buckets of the specified table with the service.
*/
static int
-ip_vs_dh_assign(struct ip_vs_dh_bucket *tbl, struct ip_vs_service *svc)
+ip_vs_dh_reassign(struct ip_vs_dh_state *s, struct ip_vs_service *svc)
{
int i;
struct ip_vs_dh_bucket *b;
struct list_head *p;
struct ip_vs_dest *dest;
+ bool empty;
- b = tbl;
+ b = &s->buckets[0];
p = &svc->destinations;
+ empty = list_empty(p);
for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
- if (list_empty(p)) {
- b->dest = NULL;
- } else {
+ dest = rcu_dereference_protected(b->dest, 1);
+ if (dest)
+ ip_vs_dest_put(dest);
+ if (empty)
+ RCU_INIT_POINTER(b->dest, NULL);
+ else {
if (p == &svc->destinations)
p = p->next;
dest = list_entry(p, struct ip_vs_dest, n_list);
- atomic_inc(&dest->refcnt);
- b->dest = dest;
+ ip_vs_dest_hold(dest);
+ RCU_INIT_POINTER(b->dest, dest);
p = p->next;
}
@@ -127,16 +135,18 @@ ip_vs_dh_assign(struct ip_vs_dh_bucket *tbl, struct ip_vs_service *svc)
/*
* Flush all the hash buckets of the specified table.
*/
-static void ip_vs_dh_flush(struct ip_vs_dh_bucket *tbl)
+static void ip_vs_dh_flush(struct ip_vs_dh_state *s)
{
int i;
struct ip_vs_dh_bucket *b;
+ struct ip_vs_dest *dest;
- b = tbl;
+ b = &s->buckets[0];
for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
- if (b->dest) {
- atomic_dec(&b->dest->refcnt);
- b->dest = NULL;
+ dest = rcu_dereference_protected(b->dest, 1);
+ if (dest) {
+ ip_vs_dest_put(dest);
+ RCU_INIT_POINTER(b->dest, NULL);
}
b++;
}
@@ -145,21 +155,20 @@ static void ip_vs_dh_flush(struct ip_vs_dh_bucket *tbl)
static int ip_vs_dh_init_svc(struct ip_vs_service *svc)
{
- struct ip_vs_dh_bucket *tbl;
+ struct ip_vs_dh_state *s;
/* allocate the DH table for this service */
- tbl = kmalloc(sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE,
- GFP_KERNEL);
- if (tbl == NULL)
+ s = kzalloc(sizeof(struct ip_vs_dh_state), GFP_KERNEL);
+ if (s == NULL)
return -ENOMEM;
- svc->sched_data = tbl;
+ svc->sched_data = s;
IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) allocated for "
"current service\n",
sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
- /* assign the hash buckets with the updated service */
- ip_vs_dh_assign(tbl, svc);
+ /* assign the hash buckets with current dests */
+ ip_vs_dh_reassign(s, svc);
return 0;
}
@@ -167,13 +176,13 @@ static int ip_vs_dh_init_svc(struct ip_vs_service *svc)
static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
{
- struct ip_vs_dh_bucket *tbl = svc->sched_data;
+ struct ip_vs_dh_state *s = svc->sched_data;
/* got to clean up hash buckets here */
- ip_vs_dh_flush(tbl);
+ ip_vs_dh_flush(s);
/* release the table itself */
- kfree(svc->sched_data);
+ kfree_rcu(s, rcu_head);
IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) released\n",
sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
@@ -181,15 +190,13 @@ static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
}
-static int ip_vs_dh_update_svc(struct ip_vs_service *svc)
+static int ip_vs_dh_dest_changed(struct ip_vs_service *svc,
+ struct ip_vs_dest *dest)
{
- struct ip_vs_dh_bucket *tbl = svc->sched_data;
-
- /* got to clean up hash buckets here */
- ip_vs_dh_flush(tbl);
+ struct ip_vs_dh_state *s = svc->sched_data;
/* assign the hash buckets with the updated service */
- ip_vs_dh_assign(tbl, svc);
+ ip_vs_dh_reassign(s, svc);
return 0;
}
@@ -212,19 +219,20 @@ static struct ip_vs_dest *
ip_vs_dh_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
{
struct ip_vs_dest *dest;
- struct ip_vs_dh_bucket *tbl;
+ struct ip_vs_dh_state *s;
struct ip_vs_iphdr iph;
ip_vs_fill_iph_addr_only(svc->af, skb, &iph);
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
- tbl = (struct ip_vs_dh_bucket *)svc->sched_data;
- dest = ip_vs_dh_get(svc->af, tbl, &iph.daddr);
+ s = (struct ip_vs_dh_state *) svc->sched_data;
+ dest = ip_vs_dh_get(svc->af, s, &iph.daddr);
if (!dest
|| !(dest->flags & IP_VS_DEST_F_AVAILABLE)
|| atomic_read(&dest->weight) <= 0
|| is_overloaded(dest)) {
+ ip_vs_scheduler_err(svc, "no destination available");
return NULL;
}
@@ -248,7 +256,8 @@ static struct ip_vs_scheduler ip_vs_dh_scheduler =
.n_list = LIST_HEAD_INIT(ip_vs_dh_scheduler.n_list),
.init_service = ip_vs_dh_init_svc,
.done_service = ip_vs_dh_done_svc,
- .update_service = ip_vs_dh_update_svc,
+ .add_dest = ip_vs_dh_dest_changed,
+ .del_dest = ip_vs_dh_dest_changed,
.schedule = ip_vs_dh_schedule,
};
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 20/34] ipvs: convert lblc scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (18 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 19/34] ipvs: convert dh scheduler to rcu Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 21/34] ipvs: convert lblcr " Simon Horman
` (14 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations. The read_lock for sched_lock is
removed. Use a dead flag to prevent new entries to be created
while scheduler is reclaimed. Use hlist for the hash table.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_lblc.c | 96 ++++++++++++++++++++++-----------------
1 file changed, 55 insertions(+), 41 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c
index fdd89b9..b873e17 100644
--- a/net/netfilter/ipvs/ip_vs_lblc.c
+++ b/net/netfilter/ipvs/ip_vs_lblc.c
@@ -90,11 +90,12 @@
* IP address and its destination server
*/
struct ip_vs_lblc_entry {
- struct list_head list;
+ struct hlist_node list;
int af; /* address family */
union nf_inet_addr addr; /* destination IP address */
- struct ip_vs_dest *dest; /* real server (cache) */
+ struct ip_vs_dest __rcu *dest; /* real server (cache) */
unsigned long lastuse; /* last used time */
+ struct rcu_head rcu_head;
};
@@ -102,12 +103,14 @@ struct ip_vs_lblc_entry {
* IPVS lblc hash table
*/
struct ip_vs_lblc_table {
- struct list_head bucket[IP_VS_LBLC_TAB_SIZE]; /* hash bucket */
+ struct rcu_head rcu_head;
+ struct hlist_head __rcu bucket[IP_VS_LBLC_TAB_SIZE]; /* hash bucket */
+ struct timer_list periodic_timer; /* collect stale entries */
atomic_t entries; /* number of entries */
int max_size; /* maximum size of entries */
- struct timer_list periodic_timer; /* collect stale entries */
int rover; /* rover for expire check */
int counter; /* counter for no expire */
+ bool dead;
};
@@ -129,13 +132,16 @@ static ctl_table vs_vars_table[] = {
static inline void ip_vs_lblc_free(struct ip_vs_lblc_entry *en)
{
- list_del(&en->list);
+ struct ip_vs_dest *dest;
+
+ hlist_del_rcu(&en->list);
/*
* We don't kfree dest because it is referred either by its service
* or the trash dest list.
*/
- atomic_dec(&en->dest->refcnt);
- kfree(en);
+ dest = rcu_dereference_protected(en->dest, 1);
+ ip_vs_dest_put(dest);
+ kfree_rcu(en, rcu_head);
}
@@ -165,15 +171,12 @@ ip_vs_lblc_hash(struct ip_vs_lblc_table *tbl, struct ip_vs_lblc_entry *en)
{
unsigned int hash = ip_vs_lblc_hashkey(en->af, &en->addr);
- list_add(&en->list, &tbl->bucket[hash]);
+ hlist_add_head_rcu(&en->list, &tbl->bucket[hash]);
atomic_inc(&tbl->entries);
}
-/*
- * Get ip_vs_lblc_entry associated with supplied parameters. Called under read
- * lock
- */
+/* Get ip_vs_lblc_entry associated with supplied parameters. */
static inline struct ip_vs_lblc_entry *
ip_vs_lblc_get(int af, struct ip_vs_lblc_table *tbl,
const union nf_inet_addr *addr)
@@ -181,7 +184,7 @@ ip_vs_lblc_get(int af, struct ip_vs_lblc_table *tbl,
unsigned int hash = ip_vs_lblc_hashkey(af, addr);
struct ip_vs_lblc_entry *en;
- list_for_each_entry(en, &tbl->bucket[hash], list)
+ hlist_for_each_entry_rcu(en, &tbl->bucket[hash], list)
if (ip_vs_addr_equal(af, &en->addr, addr))
return en;
@@ -209,14 +212,20 @@ ip_vs_lblc_new(struct ip_vs_lblc_table *tbl, const union nf_inet_addr *daddr,
ip_vs_addr_copy(dest->af, &en->addr, daddr);
en->lastuse = jiffies;
- atomic_inc(&dest->refcnt);
- en->dest = dest;
+ ip_vs_dest_hold(dest);
+ RCU_INIT_POINTER(en->dest, dest);
ip_vs_lblc_hash(tbl, en);
- } else if (en->dest != dest) {
- atomic_dec(&en->dest->refcnt);
- atomic_inc(&dest->refcnt);
- en->dest = dest;
+ } else {
+ struct ip_vs_dest *old_dest;
+
+ old_dest = rcu_dereference_protected(en->dest, 1);
+ if (old_dest != dest) {
+ ip_vs_dest_put(old_dest);
+ ip_vs_dest_hold(dest);
+ /* No ordering constraints for refcnt */
+ RCU_INIT_POINTER(en->dest, dest);
+ }
}
return en;
@@ -226,17 +235,22 @@ ip_vs_lblc_new(struct ip_vs_lblc_table *tbl, const union nf_inet_addr *daddr,
/*
* Flush all the entries of the specified table.
*/
-static void ip_vs_lblc_flush(struct ip_vs_lblc_table *tbl)
+static void ip_vs_lblc_flush(struct ip_vs_service *svc)
{
- struct ip_vs_lblc_entry *en, *nxt;
+ struct ip_vs_lblc_table *tbl = svc->sched_data;
+ struct ip_vs_lblc_entry *en;
+ struct hlist_node *next;
int i;
+ write_lock_bh(&svc->sched_lock);
+ tbl->dead = 1;
for (i=0; i<IP_VS_LBLC_TAB_SIZE; i++) {
- list_for_each_entry_safe(en, nxt, &tbl->bucket[i], list) {
+ hlist_for_each_entry_safe(en, next, &tbl->bucket[i], list) {
ip_vs_lblc_free(en);
atomic_dec(&tbl->entries);
}
}
+ write_unlock_bh(&svc->sched_lock);
}
static int sysctl_lblc_expiration(struct ip_vs_service *svc)
@@ -252,7 +266,8 @@ static int sysctl_lblc_expiration(struct ip_vs_service *svc)
static inline void ip_vs_lblc_full_check(struct ip_vs_service *svc)
{
struct ip_vs_lblc_table *tbl = svc->sched_data;
- struct ip_vs_lblc_entry *en, *nxt;
+ struct ip_vs_lblc_entry *en;
+ struct hlist_node *next;
unsigned long now = jiffies;
int i, j;
@@ -260,7 +275,7 @@ static inline void ip_vs_lblc_full_check(struct ip_vs_service *svc)
j = (j + 1) & IP_VS_LBLC_TAB_MASK;
write_lock(&svc->sched_lock);
- list_for_each_entry_safe(en, nxt, &tbl->bucket[j], list) {
+ hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_before(now,
en->lastuse +
sysctl_lblc_expiration(svc)))
@@ -293,7 +308,8 @@ static void ip_vs_lblc_check_expire(unsigned long data)
unsigned long now = jiffies;
int goal;
int i, j;
- struct ip_vs_lblc_entry *en, *nxt;
+ struct ip_vs_lblc_entry *en;
+ struct hlist_node *next;
if ((tbl->counter % COUNT_FOR_FULL_EXPIRATION) == 0) {
/* do full expiration check */
@@ -315,7 +331,7 @@ static void ip_vs_lblc_check_expire(unsigned long data)
j = (j + 1) & IP_VS_LBLC_TAB_MASK;
write_lock(&svc->sched_lock);
- list_for_each_entry_safe(en, nxt, &tbl->bucket[j], list) {
+ hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_before(now, en->lastuse + ENTRY_TIMEOUT))
continue;
@@ -354,11 +370,12 @@ static int ip_vs_lblc_init_svc(struct ip_vs_service *svc)
* Initialize the hash buckets
*/
for (i=0; i<IP_VS_LBLC_TAB_SIZE; i++) {
- INIT_LIST_HEAD(&tbl->bucket[i]);
+ INIT_HLIST_HEAD(&tbl->bucket[i]);
}
tbl->max_size = IP_VS_LBLC_TAB_SIZE*16;
tbl->rover = 0;
tbl->counter = 1;
+ tbl->dead = 0;
/*
* Hook periodic timer for garbage collection
@@ -379,10 +396,10 @@ static int ip_vs_lblc_done_svc(struct ip_vs_service *svc)
del_timer_sync(&tbl->periodic_timer);
/* got to clean up table entries here */
- ip_vs_lblc_flush(tbl);
+ ip_vs_lblc_flush(svc);
/* release the table itself */
- kfree(tbl);
+ kfree_rcu(tbl, rcu_head);
IP_VS_DBG(6, "LBLC hash table (memory=%Zdbytes) released\n",
sizeof(*tbl));
@@ -408,7 +425,7 @@ __ip_vs_lblc_schedule(struct ip_vs_service *svc)
* The server with weight=0 is quiesced and will not receive any
* new connection.
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
if (atomic_read(&dest->weight) > 0) {
@@ -423,7 +440,7 @@ __ip_vs_lblc_schedule(struct ip_vs_service *svc)
* Find the destination with the least load.
*/
nextstage:
- list_for_each_entry_continue(dest, &svc->destinations, n_list) {
+ list_for_each_entry_continue_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
@@ -457,7 +474,7 @@ is_overloaded(struct ip_vs_dest *dest, struct ip_vs_service *svc)
if (atomic_read(&dest->activeconns) > atomic_read(&dest->weight)) {
struct ip_vs_dest *d;
- list_for_each_entry(d, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(d, &svc->destinations, n_list) {
if (atomic_read(&d->activeconns)*2
< atomic_read(&d->weight)) {
return 1;
@@ -484,7 +501,6 @@ ip_vs_lblc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
/* First look in our cache */
- read_lock(&svc->sched_lock);
en = ip_vs_lblc_get(svc->af, tbl, &iph.daddr);
if (en) {
/* We only hold a read lock, but this is atomic */
@@ -499,14 +515,11 @@ ip_vs_lblc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* free up entries from the trash at any time.
*/
- if (en->dest->flags & IP_VS_DEST_F_AVAILABLE)
- dest = en->dest;
+ dest = rcu_dereference(en->dest);
+ if ((dest->flags & IP_VS_DEST_F_AVAILABLE) &&
+ atomic_read(&dest->weight) > 0 && !is_overloaded(dest, svc))
+ goto out;
}
- read_unlock(&svc->sched_lock);
-
- /* If the destination has a weight and is not overloaded, use it */
- if (dest && atomic_read(&dest->weight) > 0 && !is_overloaded(dest, svc))
- goto out;
/* No cache entry or it is invalid, time to schedule */
dest = __ip_vs_lblc_schedule(svc);
@@ -517,7 +530,8 @@ ip_vs_lblc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
/* If we fail to create a cache entry, we'll just use the valid dest */
write_lock(&svc->sched_lock);
- ip_vs_lblc_new(tbl, &iph.daddr, dest);
+ if (!tbl->dead)
+ ip_vs_lblc_new(tbl, &iph.daddr, dest);
write_unlock(&svc->sched_lock);
out:
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 21/34] ipvs: convert lblcr scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (19 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 20/34] ipvs: convert lblc " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 22/34] ipvs: convert lc " Simon Horman
` (13 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations. The read_lock for sched_lock is
removed. The set.lock is removed because now it is used in
rare cases, mostly under sched_lock.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_lblcr.c | 171 ++++++++++++++++++++------------------
1 file changed, 90 insertions(+), 81 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_lblcr.c b/net/netfilter/ipvs/ip_vs_lblcr.c
index c03b6a3..c22f173 100644
--- a/net/netfilter/ipvs/ip_vs_lblcr.c
+++ b/net/netfilter/ipvs/ip_vs_lblcr.c
@@ -89,40 +89,44 @@
*/
struct ip_vs_dest_set_elem {
struct list_head list; /* list link */
- struct ip_vs_dest *dest; /* destination server */
+ struct ip_vs_dest __rcu *dest; /* destination server */
+ struct rcu_head rcu_head;
};
struct ip_vs_dest_set {
atomic_t size; /* set size */
unsigned long lastmod; /* last modified time */
struct list_head list; /* destination list */
- rwlock_t lock; /* lock for this list */
};
-static struct ip_vs_dest_set_elem *
-ip_vs_dest_set_insert(struct ip_vs_dest_set *set, struct ip_vs_dest *dest)
+static void ip_vs_dest_set_insert(struct ip_vs_dest_set *set,
+ struct ip_vs_dest *dest, bool check)
{
struct ip_vs_dest_set_elem *e;
- list_for_each_entry(e, &set->list, list) {
- if (e->dest == dest)
- /* already existed */
- return NULL;
+ if (check) {
+ list_for_each_entry(e, &set->list, list) {
+ struct ip_vs_dest *d;
+
+ d = rcu_dereference_protected(e->dest, 1);
+ if (d == dest)
+ /* already existed */
+ return;
+ }
}
e = kmalloc(sizeof(*e), GFP_ATOMIC);
if (e == NULL)
- return NULL;
+ return;
- atomic_inc(&dest->refcnt);
- e->dest = dest;
+ ip_vs_dest_hold(dest);
+ RCU_INIT_POINTER(e->dest, dest);
- list_add(&e->list, &set->list);
+ list_add_rcu(&e->list, &set->list);
atomic_inc(&set->size);
set->lastmod = jiffies;
- return e;
}
static void
@@ -131,13 +135,16 @@ ip_vs_dest_set_erase(struct ip_vs_dest_set *set, struct ip_vs_dest *dest)
struct ip_vs_dest_set_elem *e;
list_for_each_entry(e, &set->list, list) {
- if (e->dest == dest) {
+ struct ip_vs_dest *d;
+
+ d = rcu_dereference_protected(e->dest, 1);
+ if (d == dest) {
/* HIT */
atomic_dec(&set->size);
set->lastmod = jiffies;
- atomic_dec(&e->dest->refcnt);
- list_del(&e->list);
- kfree(e);
+ ip_vs_dest_put(dest);
+ list_del_rcu(&e->list);
+ kfree_rcu(e, rcu_head);
break;
}
}
@@ -147,17 +154,18 @@ static void ip_vs_dest_set_eraseall(struct ip_vs_dest_set *set)
{
struct ip_vs_dest_set_elem *e, *ep;
- write_lock(&set->lock);
list_for_each_entry_safe(e, ep, &set->list, list) {
+ struct ip_vs_dest *d;
+
+ d = rcu_dereference_protected(e->dest, 1);
/*
* We don't kfree dest because it is referred either
* by its service or by the trash dest list.
*/
- atomic_dec(&e->dest->refcnt);
- list_del(&e->list);
- kfree(e);
+ ip_vs_dest_put(d);
+ list_del_rcu(&e->list);
+ kfree_rcu(e, rcu_head);
}
- write_unlock(&set->lock);
}
/* get weighted least-connection node in the destination set */
@@ -171,8 +179,8 @@ static inline struct ip_vs_dest *ip_vs_dest_set_min(struct ip_vs_dest_set *set)
return NULL;
/* select the first destination server, whose weight > 0 */
- list_for_each_entry(e, &set->list, list) {
- least = e->dest;
+ list_for_each_entry_rcu(e, &set->list, list) {
+ least = rcu_dereference(e->dest);
if (least->flags & IP_VS_DEST_F_OVERLOAD)
continue;
@@ -186,8 +194,8 @@ static inline struct ip_vs_dest *ip_vs_dest_set_min(struct ip_vs_dest_set *set)
/* find the destination with the weighted least load */
nextstage:
- list_for_each_entry(e, &set->list, list) {
- dest = e->dest;
+ list_for_each_entry_continue_rcu(e, &set->list, list) {
+ dest = rcu_dereference(e->dest);
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
@@ -224,7 +232,7 @@ static inline struct ip_vs_dest *ip_vs_dest_set_max(struct ip_vs_dest_set *set)
/* select the first destination server, whose weight > 0 */
list_for_each_entry(e, &set->list, list) {
- most = e->dest;
+ most = rcu_dereference_protected(e->dest, 1);
if (atomic_read(&most->weight) > 0) {
moh = ip_vs_dest_conn_overhead(most);
goto nextstage;
@@ -234,8 +242,8 @@ static inline struct ip_vs_dest *ip_vs_dest_set_max(struct ip_vs_dest_set *set)
/* find the destination with the weighted most load */
nextstage:
- list_for_each_entry(e, &set->list, list) {
- dest = e->dest;
+ list_for_each_entry_continue(e, &set->list, list) {
+ dest = rcu_dereference_protected(e->dest, 1);
doh = ip_vs_dest_conn_overhead(dest);
/* moh/mw < doh/dw ==> moh*dw < doh*mw, where mw,dw>0 */
if ((moh * atomic_read(&dest->weight) <
@@ -262,11 +270,12 @@ static inline struct ip_vs_dest *ip_vs_dest_set_max(struct ip_vs_dest_set *set)
* IP address and its destination server set
*/
struct ip_vs_lblcr_entry {
- struct list_head list;
+ struct hlist_node list;
int af; /* address family */
union nf_inet_addr addr; /* destination IP address */
struct ip_vs_dest_set set; /* destination server set */
unsigned long lastuse; /* last used time */
+ struct rcu_head rcu_head;
};
@@ -274,12 +283,14 @@ struct ip_vs_lblcr_entry {
* IPVS lblcr hash table
*/
struct ip_vs_lblcr_table {
- struct list_head bucket[IP_VS_LBLCR_TAB_SIZE]; /* hash bucket */
+ struct rcu_head rcu_head;
+ struct hlist_head __rcu bucket[IP_VS_LBLCR_TAB_SIZE]; /* hash bucket */
atomic_t entries; /* number of entries */
int max_size; /* maximum size of entries */
struct timer_list periodic_timer; /* collect stale entries */
int rover; /* rover for expire check */
int counter; /* counter for no expire */
+ bool dead;
};
@@ -302,9 +313,9 @@ static ctl_table vs_vars_table[] = {
static inline void ip_vs_lblcr_free(struct ip_vs_lblcr_entry *en)
{
- list_del(&en->list);
+ hlist_del_rcu(&en->list);
ip_vs_dest_set_eraseall(&en->set);
- kfree(en);
+ kfree_rcu(en, rcu_head);
}
@@ -334,15 +345,12 @@ ip_vs_lblcr_hash(struct ip_vs_lblcr_table *tbl, struct ip_vs_lblcr_entry *en)
{
unsigned int hash = ip_vs_lblcr_hashkey(en->af, &en->addr);
- list_add(&en->list, &tbl->bucket[hash]);
+ hlist_add_head_rcu(&en->list, &tbl->bucket[hash]);
atomic_inc(&tbl->entries);
}
-/*
- * Get ip_vs_lblcr_entry associated with supplied parameters. Called under
- * read lock.
- */
+/* Get ip_vs_lblcr_entry associated with supplied parameters. */
static inline struct ip_vs_lblcr_entry *
ip_vs_lblcr_get(int af, struct ip_vs_lblcr_table *tbl,
const union nf_inet_addr *addr)
@@ -350,7 +358,7 @@ ip_vs_lblcr_get(int af, struct ip_vs_lblcr_table *tbl,
unsigned int hash = ip_vs_lblcr_hashkey(af, addr);
struct ip_vs_lblcr_entry *en;
- list_for_each_entry(en, &tbl->bucket[hash], list)
+ hlist_for_each_entry_rcu(en, &tbl->bucket[hash], list)
if (ip_vs_addr_equal(af, &en->addr, addr))
return en;
@@ -381,14 +389,14 @@ ip_vs_lblcr_new(struct ip_vs_lblcr_table *tbl, const union nf_inet_addr *daddr,
/* initialize its dest set */
atomic_set(&(en->set.size), 0);
INIT_LIST_HEAD(&en->set.list);
- rwlock_init(&en->set.lock);
+
+ ip_vs_dest_set_insert(&en->set, dest, false);
ip_vs_lblcr_hash(tbl, en);
+ return en;
}
- write_lock(&en->set.lock);
- ip_vs_dest_set_insert(&en->set, dest);
- write_unlock(&en->set.lock);
+ ip_vs_dest_set_insert(&en->set, dest, true);
return en;
}
@@ -397,17 +405,21 @@ ip_vs_lblcr_new(struct ip_vs_lblcr_table *tbl, const union nf_inet_addr *daddr,
/*
* Flush all the entries of the specified table.
*/
-static void ip_vs_lblcr_flush(struct ip_vs_lblcr_table *tbl)
+static void ip_vs_lblcr_flush(struct ip_vs_service *svc)
{
+ struct ip_vs_lblcr_table *tbl = svc->sched_data;
int i;
- struct ip_vs_lblcr_entry *en, *nxt;
+ struct ip_vs_lblcr_entry *en;
+ struct hlist_node *next;
- /* No locking required, only called during cleanup. */
+ write_lock_bh(&svc->sched_lock);
+ tbl->dead = 1;
for (i=0; i<IP_VS_LBLCR_TAB_SIZE; i++) {
- list_for_each_entry_safe(en, nxt, &tbl->bucket[i], list) {
+ hlist_for_each_entry_safe(en, next, &tbl->bucket[i], list) {
ip_vs_lblcr_free(en);
}
}
+ write_unlock_bh(&svc->sched_lock);
}
static int sysctl_lblcr_expiration(struct ip_vs_service *svc)
@@ -425,13 +437,14 @@ static inline void ip_vs_lblcr_full_check(struct ip_vs_service *svc)
struct ip_vs_lblcr_table *tbl = svc->sched_data;
unsigned long now = jiffies;
int i, j;
- struct ip_vs_lblcr_entry *en, *nxt;
+ struct ip_vs_lblcr_entry *en;
+ struct hlist_node *next;
for (i=0, j=tbl->rover; i<IP_VS_LBLCR_TAB_SIZE; i++) {
j = (j + 1) & IP_VS_LBLCR_TAB_MASK;
write_lock(&svc->sched_lock);
- list_for_each_entry_safe(en, nxt, &tbl->bucket[j], list) {
+ hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_after(en->lastuse +
sysctl_lblcr_expiration(svc), now))
continue;
@@ -463,7 +476,8 @@ static void ip_vs_lblcr_check_expire(unsigned long data)
unsigned long now = jiffies;
int goal;
int i, j;
- struct ip_vs_lblcr_entry *en, *nxt;
+ struct ip_vs_lblcr_entry *en;
+ struct hlist_node *next;
if ((tbl->counter % COUNT_FOR_FULL_EXPIRATION) == 0) {
/* do full expiration check */
@@ -485,7 +499,7 @@ static void ip_vs_lblcr_check_expire(unsigned long data)
j = (j + 1) & IP_VS_LBLCR_TAB_MASK;
write_lock(&svc->sched_lock);
- list_for_each_entry_safe(en, nxt, &tbl->bucket[j], list) {
+ hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_before(now, en->lastuse+ENTRY_TIMEOUT))
continue;
@@ -523,11 +537,12 @@ static int ip_vs_lblcr_init_svc(struct ip_vs_service *svc)
* Initialize the hash buckets
*/
for (i=0; i<IP_VS_LBLCR_TAB_SIZE; i++) {
- INIT_LIST_HEAD(&tbl->bucket[i]);
+ INIT_HLIST_HEAD(&tbl->bucket[i]);
}
tbl->max_size = IP_VS_LBLCR_TAB_SIZE*16;
tbl->rover = 0;
tbl->counter = 1;
+ tbl->dead = 0;
/*
* Hook periodic timer for garbage collection
@@ -548,10 +563,10 @@ static int ip_vs_lblcr_done_svc(struct ip_vs_service *svc)
del_timer_sync(&tbl->periodic_timer);
/* got to clean up table entries here */
- ip_vs_lblcr_flush(tbl);
+ ip_vs_lblcr_flush(svc);
/* release the table itself */
- kfree(tbl);
+ kfree_rcu(tbl, rcu_head);
IP_VS_DBG(6, "LBLCR hash table (memory=%Zdbytes) released\n",
sizeof(*tbl));
@@ -577,7 +592,7 @@ __ip_vs_lblcr_schedule(struct ip_vs_service *svc)
* The server with weight=0 is quiesced and will not receive any
* new connection.
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
@@ -593,7 +608,7 @@ __ip_vs_lblcr_schedule(struct ip_vs_service *svc)
* Find the destination with the least load.
*/
nextstage:
- list_for_each_entry_continue(dest, &svc->destinations, n_list) {
+ list_for_each_entry_continue_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
@@ -627,7 +642,7 @@ is_overloaded(struct ip_vs_dest *dest, struct ip_vs_service *svc)
if (atomic_read(&dest->activeconns) > atomic_read(&dest->weight)) {
struct ip_vs_dest *d;
- list_for_each_entry(d, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(d, &svc->destinations, n_list) {
if (atomic_read(&d->activeconns)*2
< atomic_read(&d->weight)) {
return 1;
@@ -646,7 +661,7 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
{
struct ip_vs_lblcr_table *tbl = svc->sched_data;
struct ip_vs_iphdr iph;
- struct ip_vs_dest *dest = NULL;
+ struct ip_vs_dest *dest;
struct ip_vs_lblcr_entry *en;
ip_vs_fill_iph_addr_only(svc->af, skb, &iph);
@@ -654,53 +669,46 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
/* First look in our cache */
- read_lock(&svc->sched_lock);
en = ip_vs_lblcr_get(svc->af, tbl, &iph.daddr);
if (en) {
- /* We only hold a read lock, but this is atomic */
en->lastuse = jiffies;
/* Get the least loaded destination */
- read_lock(&en->set.lock);
dest = ip_vs_dest_set_min(&en->set);
- read_unlock(&en->set.lock);
/* More than one destination + enough time passed by, cleanup */
if (atomic_read(&en->set.size) > 1 &&
- time_after(jiffies, en->set.lastmod +
+ time_after(jiffies, en->set.lastmod +
sysctl_lblcr_expiration(svc))) {
- struct ip_vs_dest *m;
+ write_lock(&svc->sched_lock);
+ if (atomic_read(&en->set.size) > 1) {
+ struct ip_vs_dest *m;
- write_lock(&en->set.lock);
- m = ip_vs_dest_set_max(&en->set);
- if (m)
- ip_vs_dest_set_erase(&en->set, m);
- write_unlock(&en->set.lock);
+ m = ip_vs_dest_set_max(&en->set);
+ if (m)
+ ip_vs_dest_set_erase(&en->set, m);
+ }
+ write_unlock(&svc->sched_lock);
}
/* If the destination is not overloaded, use it */
- if (dest && !is_overloaded(dest, svc)) {
- read_unlock(&svc->sched_lock);
+ if (dest && !is_overloaded(dest, svc))
goto out;
- }
/* The cache entry is invalid, time to schedule */
dest = __ip_vs_lblcr_schedule(svc);
if (!dest) {
ip_vs_scheduler_err(svc, "no destination available");
- read_unlock(&svc->sched_lock);
return NULL;
}
/* Update our cache entry */
- write_lock(&en->set.lock);
- ip_vs_dest_set_insert(&en->set, dest);
- write_unlock(&en->set.lock);
- }
- read_unlock(&svc->sched_lock);
-
- if (dest)
+ write_lock(&svc->sched_lock);
+ if (!tbl->dead)
+ ip_vs_dest_set_insert(&en->set, dest, true);
+ write_unlock(&svc->sched_lock);
goto out;
+ }
/* No cache entry, time to schedule */
dest = __ip_vs_lblcr_schedule(svc);
@@ -711,7 +719,8 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
/* If we fail to create a cache entry, we'll just use the valid dest */
write_lock(&svc->sched_lock);
- ip_vs_lblcr_new(tbl, &iph.daddr, dest);
+ if (!tbl->dead)
+ ip_vs_lblcr_new(tbl, &iph.daddr, dest);
write_unlock(&svc->sched_lock);
out:
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 22/34] ipvs: convert lc scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (20 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 21/34] ipvs: convert lblcr " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 23/34] ipvs: convert nq " Simon Horman
` (12 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_lc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/netfilter/ipvs/ip_vs_lc.c b/net/netfilter/ipvs/ip_vs_lc.c
index f391819..0cabf78 100644
--- a/net/netfilter/ipvs/ip_vs_lc.c
+++ b/net/netfilter/ipvs/ip_vs_lc.c
@@ -42,7 +42,7 @@ ip_vs_lc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* served, but no new connection is assigned to the server.
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if ((dest->flags & IP_VS_DEST_F_OVERLOAD) ||
atomic_read(&dest->weight) == 0)
continue;
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 23/34] ipvs: convert nq scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (21 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 22/34] ipvs: convert lc " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 24/34] ipvs: convert rr " Simon Horman
` (11 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_nq.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/netfilter/ipvs/ip_vs_nq.c b/net/netfilter/ipvs/ip_vs_nq.c
index 984d9c1..51dc0cf 100644
--- a/net/netfilter/ipvs/ip_vs_nq.c
+++ b/net/netfilter/ipvs/ip_vs_nq.c
@@ -75,7 +75,7 @@ ip_vs_nq_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* new connections.
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD ||
!atomic_read(&dest->weight))
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 24/34] ipvs: convert rr scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (22 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 23/34] ipvs: convert nq " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 25/34] ipvs: convert sed " Simon Horman
` (10 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations. As the previous entry
could be unlinked, limit the list traversals to 2 when
lookup started from previous entry.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_rr.c | 55 ++++++++++++++++++++++++++---------------
1 file changed, 35 insertions(+), 20 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_rr.c b/net/netfilter/ipvs/ip_vs_rr.c
index c49b388..3942890 100644
--- a/net/netfilter/ipvs/ip_vs_rr.c
+++ b/net/netfilter/ipvs/ip_vs_rr.c
@@ -35,9 +35,18 @@ static int ip_vs_rr_init_svc(struct ip_vs_service *svc)
}
-static int ip_vs_rr_update_svc(struct ip_vs_service *svc)
+static int ip_vs_rr_del_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest)
{
- svc->sched_data = &svc->destinations;
+ struct list_head *p;
+
+ write_lock_bh(&svc->sched_lock);
+ p = (struct list_head *) svc->sched_data;
+ /* dest is already unlinked, so p->prev is not valid but
+ * p->next is valid, use it to reach previous entry.
+ */
+ if (p == &dest->n_list)
+ svc->sched_data = p->next->prev;
+ write_unlock_bh(&svc->sched_lock);
return 0;
}
@@ -48,35 +57,40 @@ static int ip_vs_rr_update_svc(struct ip_vs_service *svc)
static struct ip_vs_dest *
ip_vs_rr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
{
- struct list_head *p, *q;
- struct ip_vs_dest *dest;
+ struct list_head *p;
+ struct ip_vs_dest *dest, *last;
+ int pass = 0;
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
write_lock(&svc->sched_lock);
- p = (struct list_head *)svc->sched_data;
- p = p->next;
- q = p;
+ p = (struct list_head *) svc->sched_data;
+ last = dest = list_entry(p, struct ip_vs_dest, n_list);
+
do {
- /* skip list head */
- if (q == &svc->destinations) {
- q = q->next;
- continue;
+ list_for_each_entry_continue_rcu(dest,
+ &svc->destinations,
+ n_list) {
+ if (!(dest->flags & IP_VS_DEST_F_OVERLOAD) &&
+ atomic_read(&dest->weight) > 0)
+ /* HIT */
+ goto out;
+ if (dest == last)
+ goto stop;
}
+ pass++;
+ /* Previous dest could be unlinked, do not loop forever.
+ * If we stay at head there is no need for 2nd pass.
+ */
+ } while (pass < 2 && p != &svc->destinations);
- dest = list_entry(q, struct ip_vs_dest, n_list);
- if (!(dest->flags & IP_VS_DEST_F_OVERLOAD) &&
- atomic_read(&dest->weight) > 0)
- /* HIT */
- goto out;
- q = q->next;
- } while (q != p);
+stop:
write_unlock(&svc->sched_lock);
ip_vs_scheduler_err(svc, "no destination available");
return NULL;
out:
- svc->sched_data = q;
+ svc->sched_data = &dest->n_list;
write_unlock(&svc->sched_lock);
IP_VS_DBG_BUF(6, "RR: server %s:%u "
"activeconns %d refcnt %d weight %d\n",
@@ -94,7 +108,8 @@ static struct ip_vs_scheduler ip_vs_rr_scheduler = {
.module = THIS_MODULE,
.n_list = LIST_HEAD_INIT(ip_vs_rr_scheduler.n_list),
.init_service = ip_vs_rr_init_svc,
- .update_service = ip_vs_rr_update_svc,
+ .add_dest = NULL,
+ .del_dest = ip_vs_rr_del_dest,
.schedule = ip_vs_rr_schedule,
};
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 25/34] ipvs: convert sed scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (23 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 24/34] ipvs: convert rr " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 26/34] ipvs: convert sh " Simon Horman
` (9 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_sed.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_sed.c b/net/netfilter/ipvs/ip_vs_sed.c
index 89ead246..d011870 100644
--- a/net/netfilter/ipvs/ip_vs_sed.c
+++ b/net/netfilter/ipvs/ip_vs_sed.c
@@ -79,7 +79,7 @@ ip_vs_sed_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* new connections.
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if (!(dest->flags & IP_VS_DEST_F_OVERLOAD) &&
atomic_read(&dest->weight) > 0) {
least = dest;
@@ -94,7 +94,7 @@ ip_vs_sed_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* Find the destination with the least load.
*/
nextstage:
- list_for_each_entry_continue(dest, &svc->destinations, n_list) {
+ list_for_each_entry_continue_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
doh = ip_vs_sed_dest_overhead(dest);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 26/34] ipvs: convert sh scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (24 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 25/34] ipvs: convert sed " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 27/34] ipvs: convert wlc " Simon Horman
` (8 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
Use the 3 new methods to reassign dests.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_sh.c | 81 +++++++++++++++++++++++------------------
1 file changed, 45 insertions(+), 36 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_sh.c b/net/netfilter/ipvs/ip_vs_sh.c
index e331269..55e76d8 100644
--- a/net/netfilter/ipvs/ip_vs_sh.c
+++ b/net/netfilter/ipvs/ip_vs_sh.c
@@ -53,7 +53,7 @@
* IPVS SH bucket
*/
struct ip_vs_sh_bucket {
- struct ip_vs_dest *dest; /* real server (cache) */
+ struct ip_vs_dest __rcu *dest; /* real server (cache) */
};
/*
@@ -66,6 +66,10 @@ struct ip_vs_sh_bucket {
#define IP_VS_SH_TAB_SIZE (1 << IP_VS_SH_TAB_BITS)
#define IP_VS_SH_TAB_MASK (IP_VS_SH_TAB_SIZE - 1)
+struct ip_vs_sh_state {
+ struct ip_vs_sh_bucket buckets[IP_VS_SH_TAB_SIZE];
+ struct rcu_head rcu_head;
+};
/*
* Returns hash value for IPVS SH entry
@@ -87,10 +91,9 @@ static inline unsigned int ip_vs_sh_hashkey(int af, const union nf_inet_addr *ad
* Get ip_vs_dest associated with supplied parameters.
*/
static inline struct ip_vs_dest *
-ip_vs_sh_get(int af, struct ip_vs_sh_bucket *tbl,
- const union nf_inet_addr *addr)
+ip_vs_sh_get(int af, struct ip_vs_sh_state *s, const union nf_inet_addr *addr)
{
- return (tbl[ip_vs_sh_hashkey(af, addr)]).dest;
+ return rcu_dereference(s->buckets[ip_vs_sh_hashkey(af, addr)].dest);
}
@@ -98,27 +101,32 @@ ip_vs_sh_get(int af, struct ip_vs_sh_bucket *tbl,
* Assign all the hash buckets of the specified table with the service.
*/
static int
-ip_vs_sh_assign(struct ip_vs_sh_bucket *tbl, struct ip_vs_service *svc)
+ip_vs_sh_reassign(struct ip_vs_sh_state *s, struct ip_vs_service *svc)
{
int i;
struct ip_vs_sh_bucket *b;
struct list_head *p;
struct ip_vs_dest *dest;
int d_count;
+ bool empty;
- b = tbl;
+ b = &s->buckets[0];
p = &svc->destinations;
+ empty = list_empty(p);
d_count = 0;
for (i=0; i<IP_VS_SH_TAB_SIZE; i++) {
- if (list_empty(p)) {
- b->dest = NULL;
- } else {
+ dest = rcu_dereference_protected(b->dest, 1);
+ if (dest)
+ ip_vs_dest_put(dest);
+ if (empty)
+ RCU_INIT_POINTER(b->dest, NULL);
+ else {
if (p == &svc->destinations)
p = p->next;
dest = list_entry(p, struct ip_vs_dest, n_list);
- atomic_inc(&dest->refcnt);
- b->dest = dest;
+ ip_vs_dest_hold(dest);
+ RCU_INIT_POINTER(b->dest, dest);
IP_VS_DBG_BUF(6, "assigned i: %d dest: %s weight: %d\n",
i, IP_VS_DBG_ADDR(svc->af, &dest->addr),
@@ -140,16 +148,18 @@ ip_vs_sh_assign(struct ip_vs_sh_bucket *tbl, struct ip_vs_service *svc)
/*
* Flush all the hash buckets of the specified table.
*/
-static void ip_vs_sh_flush(struct ip_vs_sh_bucket *tbl)
+static void ip_vs_sh_flush(struct ip_vs_sh_state *s)
{
int i;
struct ip_vs_sh_bucket *b;
+ struct ip_vs_dest *dest;
- b = tbl;
+ b = &s->buckets[0];
for (i=0; i<IP_VS_SH_TAB_SIZE; i++) {
- if (b->dest) {
- atomic_dec(&b->dest->refcnt);
- b->dest = NULL;
+ dest = rcu_dereference_protected(b->dest, 1);
+ if (dest) {
+ ip_vs_dest_put(dest);
+ RCU_INIT_POINTER(b->dest, NULL);
}
b++;
}
@@ -158,21 +168,20 @@ static void ip_vs_sh_flush(struct ip_vs_sh_bucket *tbl)
static int ip_vs_sh_init_svc(struct ip_vs_service *svc)
{
- struct ip_vs_sh_bucket *tbl;
+ struct ip_vs_sh_state *s;
/* allocate the SH table for this service */
- tbl = kmalloc(sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE,
- GFP_KERNEL);
- if (tbl == NULL)
+ s = kzalloc(sizeof(struct ip_vs_sh_state), GFP_KERNEL);
+ if (s == NULL)
return -ENOMEM;
- svc->sched_data = tbl;
+ svc->sched_data = s;
IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) allocated for "
"current service\n",
sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);
- /* assign the hash buckets with the updated service */
- ip_vs_sh_assign(tbl, svc);
+ /* assign the hash buckets with current dests */
+ ip_vs_sh_reassign(s, svc);
return 0;
}
@@ -180,13 +189,13 @@ static int ip_vs_sh_init_svc(struct ip_vs_service *svc)
static int ip_vs_sh_done_svc(struct ip_vs_service *svc)
{
- struct ip_vs_sh_bucket *tbl = svc->sched_data;
+ struct ip_vs_sh_state *s = svc->sched_data;
/* got to clean up hash buckets here */
- ip_vs_sh_flush(tbl);
+ ip_vs_sh_flush(s);
/* release the table itself */
- kfree(svc->sched_data);
+ kfree_rcu(s, rcu_head);
IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) released\n",
sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);
@@ -194,15 +203,13 @@ static int ip_vs_sh_done_svc(struct ip_vs_service *svc)
}
-static int ip_vs_sh_update_svc(struct ip_vs_service *svc)
+static int ip_vs_sh_dest_changed(struct ip_vs_service *svc,
+ struct ip_vs_dest *dest)
{
- struct ip_vs_sh_bucket *tbl = svc->sched_data;
-
- /* got to clean up hash buckets here */
- ip_vs_sh_flush(tbl);
+ struct ip_vs_sh_state *s = svc->sched_data;
/* assign the hash buckets with the updated service */
- ip_vs_sh_assign(tbl, svc);
+ ip_vs_sh_reassign(s, svc);
return 0;
}
@@ -225,15 +232,15 @@ static struct ip_vs_dest *
ip_vs_sh_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
{
struct ip_vs_dest *dest;
- struct ip_vs_sh_bucket *tbl;
+ struct ip_vs_sh_state *s;
struct ip_vs_iphdr iph;
ip_vs_fill_iph_addr_only(svc->af, skb, &iph);
IP_VS_DBG(6, "ip_vs_sh_schedule(): Scheduling...\n");
- tbl = (struct ip_vs_sh_bucket *)svc->sched_data;
- dest = ip_vs_sh_get(svc->af, tbl, &iph.saddr);
+ s = (struct ip_vs_sh_state *) svc->sched_data;
+ dest = ip_vs_sh_get(svc->af, s, &iph.saddr);
if (!dest
|| !(dest->flags & IP_VS_DEST_F_AVAILABLE)
|| atomic_read(&dest->weight) <= 0
@@ -262,7 +269,9 @@ static struct ip_vs_scheduler ip_vs_sh_scheduler =
.n_list = LIST_HEAD_INIT(ip_vs_sh_scheduler.n_list),
.init_service = ip_vs_sh_init_svc,
.done_service = ip_vs_sh_done_svc,
- .update_service = ip_vs_sh_update_svc,
+ .add_dest = ip_vs_sh_dest_changed,
+ .del_dest = ip_vs_sh_dest_changed,
+ .upd_dest = ip_vs_sh_dest_changed,
.schedule = ip_vs_sh_schedule,
};
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 27/34] ipvs: convert wlc scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (25 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 26/34] ipvs: convert sh " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 28/34] ipvs: convert wrr " Simon Horman
` (7 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_wlc.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_wlc.c b/net/netfilter/ipvs/ip_vs_wlc.c
index bc1bfc4..dafae88 100644
--- a/net/netfilter/ipvs/ip_vs_wlc.c
+++ b/net/netfilter/ipvs/ip_vs_wlc.c
@@ -51,7 +51,7 @@ ip_vs_wlc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* new connections.
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if (!(dest->flags & IP_VS_DEST_F_OVERLOAD) &&
atomic_read(&dest->weight) > 0) {
least = dest;
@@ -66,7 +66,7 @@ ip_vs_wlc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
* Find the destination with the least load.
*/
nextstage:
- list_for_each_entry_continue(dest, &svc->destinations, n_list) {
+ list_for_each_entry_continue_rcu(dest, &svc->destinations, n_list) {
if (dest->flags & IP_VS_DEST_F_OVERLOAD)
continue;
doh = ip_vs_dest_conn_overhead(dest);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 28/34] ipvs: convert wrr scheduler to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (26 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 27/34] ipvs: convert wlc " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 29/34] ipvs: reorganize dest trash Simon Horman
` (6 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
The schedule method now needs _rcu list-traversal
primitive for svc->destinations. As the weight for some
dest can be reduced during dest selection, change the
algorithm to check weights by using minimum weights in the
1 .. max_weight-(di-1) range, with the same step (di). By this
way we ensure that there will be always a weight >= 1 check
before claiming that all destinations are overloaded.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_wrr.c | 167 +++++++++++++++++++++++++---------------
1 file changed, 103 insertions(+), 64 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_wrr.c b/net/netfilter/ipvs/ip_vs_wrr.c
index 231be7d..98cb05e 100644
--- a/net/netfilter/ipvs/ip_vs_wrr.c
+++ b/net/netfilter/ipvs/ip_vs_wrr.c
@@ -29,14 +29,45 @@
#include <net/ip_vs.h>
+/* The WRR algorithm depends on some caclulations:
+ * - mw: maximum weight
+ * - di: weight step, greatest common divisor from all weights
+ * - cw: current required weight
+ * As result, all weights are in the [di..mw] range with a step=di.
+ *
+ * First, we start with cw = mw and select dests with weight >= cw.
+ * Then cw is reduced with di and all dests are checked again.
+ * Last pass should be with cw = di. We have mw/di passes in total:
+ *
+ * pass 1: cw = max weight
+ * pass 2: cw = max weight - di
+ * pass 3: cw = max weight - 2 * di
+ * ...
+ * last pass: cw = di
+ *
+ * Weights are supposed to be >= di but we run in parallel with
+ * weight changes, it is possible some dest weight to be reduced
+ * below di, bad if it is the only available dest.
+ *
+ * So, we modify how mw is calculated, now it is reduced with (di - 1),
+ * so that last cw is 1 to catch such dests with weight below di:
+ * pass 1: cw = max weight - (di - 1)
+ * pass 2: cw = max weight - di - (di - 1)
+ * pass 3: cw = max weight - 2 * di - (di - 1)
+ * ...
+ * last pass: cw = 1
+ *
+ */
+
/*
* current destination pointer for weighted round-robin scheduling
*/
struct ip_vs_wrr_mark {
- struct list_head *cl; /* current list head */
+ struct ip_vs_dest *cl; /* current dest or head */
int cw; /* current weight */
int mw; /* maximum weight */
int di; /* decreasing interval */
+ struct rcu_head rcu_head;
};
@@ -88,10 +119,10 @@ static int ip_vs_wrr_init_svc(struct ip_vs_service *svc)
if (mark == NULL)
return -ENOMEM;
- mark->cl = &svc->destinations;
- mark->cw = 0;
- mark->mw = ip_vs_wrr_max_weight(svc);
+ mark->cl = list_entry(&svc->destinations, struct ip_vs_dest, n_list);
mark->di = ip_vs_wrr_gcd_weight(svc);
+ mark->mw = ip_vs_wrr_max_weight(svc) - (mark->di - 1);
+ mark->cw = mark->mw;
svc->sched_data = mark;
return 0;
@@ -100,24 +131,31 @@ static int ip_vs_wrr_init_svc(struct ip_vs_service *svc)
static int ip_vs_wrr_done_svc(struct ip_vs_service *svc)
{
+ struct ip_vs_wrr_mark *mark = svc->sched_data;
+
/*
* Release the mark variable
*/
- kfree(svc->sched_data);
+ kfree_rcu(mark, rcu_head);
return 0;
}
-static int ip_vs_wrr_update_svc(struct ip_vs_service *svc)
+static int ip_vs_wrr_dest_changed(struct ip_vs_service *svc,
+ struct ip_vs_dest *dest)
{
struct ip_vs_wrr_mark *mark = svc->sched_data;
- mark->cl = &svc->destinations;
- mark->mw = ip_vs_wrr_max_weight(svc);
+ write_lock_bh(&svc->sched_lock);
+ mark->cl = list_entry(&svc->destinations, struct ip_vs_dest, n_list);
mark->di = ip_vs_wrr_gcd_weight(svc);
- if (mark->cw > mark->mw)
- mark->cw = 0;
+ mark->mw = ip_vs_wrr_max_weight(svc) - (mark->di - 1);
+ if (mark->cw > mark->mw || !mark->cw)
+ mark->cw = mark->mw;
+ else if (mark->di > 1)
+ mark->cw = (mark->cw / mark->di) * mark->di + 1;
+ write_unlock_bh(&svc->sched_lock);
return 0;
}
@@ -128,80 +166,79 @@ static int ip_vs_wrr_update_svc(struct ip_vs_service *svc)
static struct ip_vs_dest *
ip_vs_wrr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
{
- struct ip_vs_dest *dest;
+ struct ip_vs_dest *dest, *last, *stop = NULL;
struct ip_vs_wrr_mark *mark = svc->sched_data;
- struct list_head *p;
+ bool last_pass = false, restarted = false;
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
- /*
- * This loop will always terminate, because mark->cw in (0, max_weight]
- * and at least one server has its weight equal to max_weight.
- */
write_lock(&svc->sched_lock);
- p = mark->cl;
+ dest = mark->cl;
+ /* No available dests? */
+ if (mark->mw == 0)
+ goto err_noavail;
+ last = dest;
+ /* Stop only after all dests were checked for weight >= 1 (last pass) */
while (1) {
- if (mark->cl == &svc->destinations) {
- /* it is at the head of the destination list */
-
- if (mark->cl == mark->cl->next) {
- /* no dest entry */
- ip_vs_scheduler_err(svc,
- "no destination available: "
- "no destinations present");
- dest = NULL;
- goto out;
- }
-
- mark->cl = svc->destinations.next;
- mark->cw -= mark->di;
- if (mark->cw <= 0) {
- mark->cw = mark->mw;
- /*
- * Still zero, which means no available servers.
- */
- if (mark->cw == 0) {
- mark->cl = &svc->destinations;
- ip_vs_scheduler_err(svc,
- "no destination available");
- dest = NULL;
- goto out;
- }
- }
- } else
- mark->cl = mark->cl->next;
-
- if (mark->cl != &svc->destinations) {
- /* not at the head of the list */
- dest = list_entry(mark->cl, struct ip_vs_dest, n_list);
+ list_for_each_entry_continue_rcu(dest,
+ &svc->destinations,
+ n_list) {
if (!(dest->flags & IP_VS_DEST_F_OVERLOAD) &&
- atomic_read(&dest->weight) >= mark->cw) {
- /* got it */
- break;
- }
+ atomic_read(&dest->weight) >= mark->cw)
+ goto found;
+ if (dest == stop)
+ goto err_over;
}
-
- if (mark->cl == p && mark->cw == mark->di) {
- /* back to the start, and no dest is found.
- It is only possible when all dests are OVERLOADED */
- dest = NULL;
- ip_vs_scheduler_err(svc,
- "no destination available: "
- "all destinations are overloaded");
- goto out;
+ mark->cw -= mark->di;
+ if (mark->cw <= 0) {
+ mark->cw = mark->mw;
+ /* Stop if we tried last pass from first dest:
+ * 1. last_pass: we started checks when cw > di but
+ * then all dests were checked for w >= 1
+ * 2. last was head: the first and only traversal
+ * was for weight >= 1, for all dests.
+ */
+ if (last_pass ||
+ &last->n_list == &svc->destinations)
+ goto err_over;
+ restarted = true;
+ }
+ last_pass = mark->cw <= mark->di;
+ if (last_pass && restarted &&
+ &last->n_list != &svc->destinations) {
+ /* First traversal was for w >= 1 but only
+ * for dests after 'last', now do the same
+ * for all dests up to 'last'.
+ */
+ stop = last;
}
}
+found:
IP_VS_DBG_BUF(6, "WRR: server %s:%u "
"activeconns %d refcnt %d weight %d\n",
IP_VS_DBG_ADDR(svc->af, &dest->addr), ntohs(dest->port),
atomic_read(&dest->activeconns),
atomic_read(&dest->refcnt),
atomic_read(&dest->weight));
+ mark->cl = dest;
out:
write_unlock(&svc->sched_lock);
return dest;
+
+err_noavail:
+ mark->cl = dest;
+ dest = NULL;
+ ip_vs_scheduler_err(svc, "no destination available");
+ goto out;
+
+err_over:
+ mark->cl = dest;
+ dest = NULL;
+ ip_vs_scheduler_err(svc, "no destination available: "
+ "all destinations are overloaded");
+ goto out;
}
@@ -212,7 +249,9 @@ static struct ip_vs_scheduler ip_vs_wrr_scheduler = {
.n_list = LIST_HEAD_INIT(ip_vs_wrr_scheduler.n_list),
.init_service = ip_vs_wrr_init_svc,
.done_service = ip_vs_wrr_done_svc,
- .update_service = ip_vs_wrr_update_svc,
+ .add_dest = ip_vs_wrr_dest_changed,
+ .del_dest = ip_vs_wrr_dest_changed,
+ .upd_dest = ip_vs_wrr_dest_changed,
.schedule = ip_vs_wrr_schedule,
};
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 29/34] ipvs: reorganize dest trash
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (27 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 28/34] ipvs: convert wrr " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 30/34] ipvs: do not expect result from done_service Simon Horman
` (5 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
All dests will go to trash, no exceptions.
But we have to use new list node t_list for this, due
to RCU changes in following patches. Dests will wait there
initial grace period and later all conns and schedulers to
put their reference. The dests don't get reference for
staying in dest trash as before.
As result, we do not load ip_vs_dest_put with
extra checks for last refcnt and the schedulers do not
need to play games with atomic_inc_not_zero while
selecting best destination.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 9 ++
net/netfilter/ipvs/ip_vs_ctl.c | 178 +++++++++++++++++++++++-----------------
2 files changed, 113 insertions(+), 74 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 7d3027f..18aeb85 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -749,6 +749,8 @@ struct ip_vs_dest_dst {
struct rcu_head rcu_head;
};
+/* In grace period after removing */
+#define IP_VS_DEST_STATE_REMOVING 0x01
/*
* The real server destination forwarding entry
* with ip address, port number, and so on.
@@ -766,6 +768,7 @@ struct ip_vs_dest {
atomic_t refcnt; /* reference counter */
struct ip_vs_stats stats; /* statistics */
+ unsigned long state; /* state flags */
/* connection counters and thresholds */
atomic_t activeconns; /* active connections */
@@ -785,6 +788,7 @@ struct ip_vs_dest {
union nf_inet_addr vaddr; /* virtual IP address */
__u32 vfwmark; /* firewall mark of service */
+ struct list_head t_list; /* in dest_trash */
struct rcu_head rcu_head;
unsigned int in_rs_table:1; /* we are in rs_table */
};
@@ -912,6 +916,9 @@ struct ipvs_master_sync_state {
struct netns_ipvs *ipvs;
};
+/* How much time to keep dests in trash */
+#define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
+
/* IPVS in network namespace */
struct netns_ipvs {
int gen; /* Generation */
@@ -961,6 +968,8 @@ struct netns_ipvs {
/* Trash for destinations */
struct list_head dest_trash;
+ spinlock_t dest_trash_lock;
+ struct timer_list dest_trash_timer; /* expiration timer */
/* Service counters */
atomic_t ftpsvc_counter;
atomic_t nullsvc_counter;
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index a4f6388..80d366b 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -71,7 +71,7 @@ int ip_vs_get_debug_level(void)
/* Protos */
-static void __ip_vs_del_service(struct ip_vs_service *svc);
+static void __ip_vs_del_service(struct ip_vs_service *svc, bool cleanup);
#ifdef CONFIG_IP_VS_IPV6
@@ -657,19 +657,25 @@ static struct ip_vs_dest *
ip_vs_trash_get_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
__be16 dport)
{
- struct ip_vs_dest *dest, *nxt;
+ struct ip_vs_dest *dest;
struct netns_ipvs *ipvs = net_ipvs(svc->net);
/*
* Find the destination in trash
*/
- list_for_each_entry_safe(dest, nxt, &ipvs->dest_trash, n_list) {
+ spin_lock_bh(&ipvs->dest_trash_lock);
+ list_for_each_entry(dest, &ipvs->dest_trash, t_list) {
IP_VS_DBG_BUF(3, "Destination %u/%s:%u still in trash, "
"dest->refcnt=%d\n",
dest->vfwmark,
IP_VS_DBG_ADDR(svc->af, &dest->addr),
ntohs(dest->port),
atomic_read(&dest->refcnt));
+ /* We can not reuse dest while in grace period
+ * because conns still can use dest->svc
+ */
+ if (test_bit(IP_VS_DEST_STATE_REMOVING, &dest->state))
+ continue;
if (dest->af == svc->af &&
ip_vs_addr_equal(svc->af, &dest->addr, daddr) &&
dest->port == dport &&
@@ -679,29 +685,27 @@ ip_vs_trash_get_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
(ip_vs_addr_equal(svc->af, &dest->vaddr, &svc->addr) &&
dest->vport == svc->port))) {
/* HIT */
- return dest;
- }
-
- /*
- * Try to purge the destination from trash if not referenced
- */
- if (atomic_read(&dest->refcnt) == 1) {
- IP_VS_DBG_BUF(3, "Removing destination %u/%s:%u "
- "from trash\n",
- dest->vfwmark,
- IP_VS_DBG_ADDR(svc->af, &dest->addr),
- ntohs(dest->port));
- list_del(&dest->n_list);
- __ip_vs_dst_cache_reset(dest);
- __ip_vs_unbind_svc(dest);
- free_percpu(dest->stats.cpustats);
- kfree_rcu(dest, rcu_head);
+ list_del(&dest->t_list);
+ ip_vs_dest_hold(dest);
+ goto out;
}
}
- return NULL;
+ dest = NULL;
+
+out:
+ spin_unlock_bh(&ipvs->dest_trash_lock);
+
+ return dest;
}
+static void ip_vs_dest_free(struct ip_vs_dest *dest)
+{
+ __ip_vs_dst_cache_reset(dest);
+ __ip_vs_unbind_svc(dest);
+ free_percpu(dest->stats.cpustats);
+ kfree(dest);
+}
/*
* Clean up all the destinations in the trash
@@ -710,19 +714,18 @@ ip_vs_trash_get_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
* When the ip_vs_control_clearup is activated by ipvs module exit,
* the service tables must have been flushed and all the connections
* are expired, and the refcnt of each destination in the trash must
- * be 1, so we simply release them here.
+ * be 0, so we simply release them here.
*/
static void ip_vs_trash_cleanup(struct net *net)
{
struct ip_vs_dest *dest, *nxt;
struct netns_ipvs *ipvs = net_ipvs(net);
- list_for_each_entry_safe(dest, nxt, &ipvs->dest_trash, n_list) {
- list_del(&dest->n_list);
- __ip_vs_dst_cache_reset(dest);
- __ip_vs_unbind_svc(dest);
- free_percpu(dest->stats.cpustats);
- kfree_rcu(dest, rcu_head);
+ del_timer_sync(&ipvs->dest_trash_timer);
+ /* No need to use dest_trash_lock */
+ list_for_each_entry_safe(dest, nxt, &ipvs->dest_trash, t_list) {
+ list_del(&dest->t_list);
+ ip_vs_dest_free(dest);
}
}
@@ -955,11 +958,6 @@ ip_vs_add_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
IP_VS_DBG_ADDR(svc->af, &dest->vaddr),
ntohs(dest->vport));
- /*
- * Get the destination from the trash
- */
- list_del(&dest->n_list);
-
__ip_vs_update_dest(svc, dest, udest, 1);
ret = 0;
} else {
@@ -1015,11 +1013,21 @@ ip_vs_edit_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
return 0;
}
+static void ip_vs_dest_wait_readers(struct rcu_head *head)
+{
+ struct ip_vs_dest *dest = container_of(head, struct ip_vs_dest,
+ rcu_head);
+
+ /* End of grace period after unlinking */
+ clear_bit(IP_VS_DEST_STATE_REMOVING, &dest->state);
+}
+
/*
* Delete a destination (must be already unlinked from the service)
*/
-static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest)
+static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest,
+ bool cleanup)
{
struct netns_ipvs *ipvs = net_ipvs(net);
@@ -1030,34 +1038,22 @@ static void __ip_vs_del_dest(struct net *net, struct ip_vs_dest *dest)
*/
ip_vs_rs_unhash(dest);
- /*
- * Decrease the refcnt of the dest, and free the dest
- * if nobody refers to it (refcnt=0). Otherwise, throw
- * the destination into the trash.
- */
- if (atomic_dec_and_test(&dest->refcnt)) {
- IP_VS_DBG_BUF(3, "Removing destination %u/%s:%u\n",
- dest->vfwmark,
- IP_VS_DBG_ADDR(dest->af, &dest->addr),
- ntohs(dest->port));
- __ip_vs_dst_cache_reset(dest);
- /* simply decrease svc->refcnt here, let the caller check
- and release the service if nobody refers to it.
- Only user context can release destination and service,
- and only one user context can update virtual service at a
- time, so the operation here is OK */
- atomic_dec(&dest->svc->refcnt);
- free_percpu(dest->stats.cpustats);
- kfree_rcu(dest, rcu_head);
- } else {
- IP_VS_DBG_BUF(3, "Moving dest %s:%u into trash, "
- "dest->refcnt=%d\n",
- IP_VS_DBG_ADDR(dest->af, &dest->addr),
- ntohs(dest->port),
- atomic_read(&dest->refcnt));
- list_add(&dest->n_list, &ipvs->dest_trash);
- ip_vs_dest_hold(dest);
+ if (!cleanup) {
+ set_bit(IP_VS_DEST_STATE_REMOVING, &dest->state);
+ call_rcu(&dest->rcu_head, ip_vs_dest_wait_readers);
}
+
+ spin_lock_bh(&ipvs->dest_trash_lock);
+ IP_VS_DBG_BUF(3, "Moving dest %s:%u into trash, dest->refcnt=%d\n",
+ IP_VS_DBG_ADDR(dest->af, &dest->addr), ntohs(dest->port),
+ atomic_read(&dest->refcnt));
+ if (list_empty(&ipvs->dest_trash) && !cleanup)
+ mod_timer(&ipvs->dest_trash_timer,
+ jiffies + IP_VS_DEST_TRASH_PERIOD);
+ /* dest lives in trash without reference */
+ list_add(&dest->t_list, &ipvs->dest_trash);
+ spin_unlock_bh(&ipvs->dest_trash_lock);
+ ip_vs_dest_put(dest);
}
@@ -1122,13 +1118,38 @@ ip_vs_del_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
/*
* Delete the destination
*/
- __ip_vs_del_dest(svc->net, dest);
+ __ip_vs_del_dest(svc->net, dest, false);
LeaveFunction(2);
return 0;
}
+static void ip_vs_dest_trash_expire(unsigned long data)
+{
+ struct net *net = (struct net *) data;
+ struct netns_ipvs *ipvs = net_ipvs(net);
+ struct ip_vs_dest *dest, *next;
+
+ spin_lock(&ipvs->dest_trash_lock);
+ list_for_each_entry_safe(dest, next, &ipvs->dest_trash, t_list) {
+ /* Skip if dest is in grace period */
+ if (test_bit(IP_VS_DEST_STATE_REMOVING, &dest->state))
+ continue;
+ if (atomic_read(&dest->refcnt) > 0)
+ continue;
+ IP_VS_DBG_BUF(3, "Removing destination %u/%s:%u from trash\n",
+ dest->vfwmark,
+ IP_VS_DBG_ADDR(dest->svc->af, &dest->addr),
+ ntohs(dest->port));
+ list_del(&dest->t_list);
+ ip_vs_dest_free(dest);
+ }
+ if (!list_empty(&ipvs->dest_trash))
+ mod_timer(&ipvs->dest_trash_timer,
+ jiffies + IP_VS_DEST_TRASH_PERIOD);
+ spin_unlock(&ipvs->dest_trash_lock);
+}
/*
* Add a service into the service hash table
@@ -1358,7 +1379,7 @@ out:
* - The service must be unlinked, unlocked and not referenced!
* - We are called under _bh lock
*/
-static void __ip_vs_del_service(struct ip_vs_service *svc)
+static void __ip_vs_del_service(struct ip_vs_service *svc, bool cleanup)
{
struct ip_vs_dest *dest, *nxt;
struct ip_vs_scheduler *old_sched;
@@ -1394,7 +1415,7 @@ static void __ip_vs_del_service(struct ip_vs_service *svc)
*/
list_for_each_entry_safe(dest, nxt, &svc->destinations, n_list) {
__ip_vs_unlink_dest(svc, dest, 0);
- __ip_vs_del_dest(svc->net, dest);
+ __ip_vs_del_dest(svc->net, dest, cleanup);
}
/*
@@ -1424,7 +1445,7 @@ static void __ip_vs_del_service(struct ip_vs_service *svc)
/*
* Unlink a service from list and try to delete it if its refcnt reached 0
*/
-static void ip_vs_unlink_service(struct ip_vs_service *svc)
+static void ip_vs_unlink_service(struct ip_vs_service *svc, bool cleanup)
{
/*
* Unhash it from the service table
@@ -1438,7 +1459,7 @@ static void ip_vs_unlink_service(struct ip_vs_service *svc)
*/
IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
- __ip_vs_del_service(svc);
+ __ip_vs_del_service(svc, cleanup);
write_unlock_bh(&__ip_vs_svc_lock);
}
@@ -1450,7 +1471,7 @@ static int ip_vs_del_service(struct ip_vs_service *svc)
{
if (svc == NULL)
return -EEXIST;
- ip_vs_unlink_service(svc);
+ ip_vs_unlink_service(svc, false);
return 0;
}
@@ -1459,7 +1480,7 @@ static int ip_vs_del_service(struct ip_vs_service *svc)
/*
* Flush all the virtual services
*/
-static int ip_vs_flush(struct net *net)
+static int ip_vs_flush(struct net *net, bool cleanup)
{
int idx;
struct ip_vs_service *svc, *nxt;
@@ -1471,7 +1492,7 @@ static int ip_vs_flush(struct net *net)
list_for_each_entry_safe(svc, nxt, &ip_vs_svc_table[idx],
s_list) {
if (net_eq(svc->net, net))
- ip_vs_unlink_service(svc);
+ ip_vs_unlink_service(svc, cleanup);
}
}
@@ -1482,7 +1503,7 @@ static int ip_vs_flush(struct net *net)
list_for_each_entry_safe(svc, nxt,
&ip_vs_svc_fwm_table[idx], f_list) {
if (net_eq(svc->net, net))
- ip_vs_unlink_service(svc);
+ ip_vs_unlink_service(svc, cleanup);
}
}
@@ -1498,7 +1519,7 @@ void ip_vs_service_net_cleanup(struct net *net)
EnterFunction(2);
/* Check for "full" addressed entries */
mutex_lock(&__ip_vs_mutex);
- ip_vs_flush(net);
+ ip_vs_flush(net, true);
mutex_unlock(&__ip_vs_mutex);
LeaveFunction(2);
}
@@ -1558,9 +1579,11 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
}
}
- list_for_each_entry(dest, &ipvs->dest_trash, n_list) {
+ spin_lock_bh(&ipvs->dest_trash_lock);
+ list_for_each_entry(dest, &ipvs->dest_trash, t_list) {
ip_vs_forget_dev(dest, dev);
}
+ spin_unlock_bh(&ipvs->dest_trash_lock);
mutex_unlock(&__ip_vs_mutex);
LeaveFunction(2);
return NOTIFY_DONE;
@@ -2394,7 +2417,7 @@ do_ip_vs_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
if (cmd == IP_VS_SO_SET_FLUSH) {
/* Flush the virtual service */
- ret = ip_vs_flush(net);
+ ret = ip_vs_flush(net, false);
goto out_unlock;
} else if (cmd == IP_VS_SO_SET_TIMEOUT) {
/* Set timeout values for (tcp tcpfin udp) */
@@ -3403,7 +3426,7 @@ static int ip_vs_genl_set_cmd(struct sk_buff *skb, struct genl_info *info)
mutex_lock(&__ip_vs_mutex);
if (cmd == IPVS_CMD_FLUSH) {
- ret = ip_vs_flush(net);
+ ret = ip_vs_flush(net, false);
goto out;
} else if (cmd == IPVS_CMD_SET_CONFIG) {
ret = ip_vs_genl_set_config(net, info->attrs);
@@ -3800,6 +3823,9 @@ int __net_init ip_vs_control_net_init(struct net *net)
INIT_HLIST_HEAD(&ipvs->rs_table[idx]);
INIT_LIST_HEAD(&ipvs->dest_trash);
+ spin_lock_init(&ipvs->dest_trash_lock);
+ setup_timer(&ipvs->dest_trash_timer, ip_vs_dest_trash_expire,
+ (unsigned long) net);
atomic_set(&ipvs->ftpsvc_counter, 0);
atomic_set(&ipvs->nullsvc_counter, 0);
@@ -3829,6 +3855,10 @@ void __net_exit ip_vs_control_net_cleanup(struct net *net)
{
struct netns_ipvs *ipvs = net_ipvs(net);
+ /* Some dest can be in grace period even before cleanup, we have to
+ * defer ip_vs_trash_cleanup until ip_vs_dest_wait_readers is called.
+ */
+ rcu_barrier();
ip_vs_trash_cleanup(net);
ip_vs_stop_estimator(net, &ipvs->tot_stats);
ip_vs_control_net_cleanup_sysctl(net);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 30/34] ipvs: do not expect result from done_service
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (28 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 29/34] ipvs: reorganize dest trash Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 31/34] ipvs: convert sched_lock to spin lock Simon Horman
` (4 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
This method releases the scheduler state,
it can not fail. Such change will help to properly
replace the scheduler in following patch.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 4 ++--
net/netfilter/ipvs/ip_vs_ctl.c | 5 +----
net/netfilter/ipvs/ip_vs_dh.c | 4 +---
net/netfilter/ipvs/ip_vs_lblc.c | 4 +---
net/netfilter/ipvs/ip_vs_lblcr.c | 4 +---
net/netfilter/ipvs/ip_vs_sched.c | 13 ++++---------
net/netfilter/ipvs/ip_vs_sh.c | 4 +---
net/netfilter/ipvs/ip_vs_wrr.c | 4 +---
8 files changed, 12 insertions(+), 30 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 18aeb85..4990de6 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -806,7 +806,7 @@ struct ip_vs_scheduler {
/* scheduler initializing service */
int (*init_service)(struct ip_vs_service *svc);
/* scheduling service finish */
- int (*done_service)(struct ip_vs_service *svc);
+ void (*done_service)(struct ip_vs_service *svc);
/* scheduler updating service */
int (*update_service)(struct ip_vs_service *svc);
/* dest is linked */
@@ -1392,7 +1392,7 @@ extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
struct ip_vs_scheduler *scheduler);
-extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
+extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc);
extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
extern struct ip_vs_conn *
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 80d366b..d022726 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -1334,10 +1334,7 @@ ip_vs_edit_service(struct ip_vs_service *svc, struct ip_vs_service_user_kern *u)
/*
* Unbind the old scheduler
*/
- if ((ret = ip_vs_unbind_scheduler(svc))) {
- old_sched = sched;
- goto out_unlock;
- }
+ ip_vs_unbind_scheduler(svc);
/*
* Bind the new scheduler
diff --git a/net/netfilter/ipvs/ip_vs_dh.c b/net/netfilter/ipvs/ip_vs_dh.c
index ebe80f4..89c2723 100644
--- a/net/netfilter/ipvs/ip_vs_dh.c
+++ b/net/netfilter/ipvs/ip_vs_dh.c
@@ -174,7 +174,7 @@ static int ip_vs_dh_init_svc(struct ip_vs_service *svc)
}
-static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
+static void ip_vs_dh_done_svc(struct ip_vs_service *svc)
{
struct ip_vs_dh_state *s = svc->sched_data;
@@ -185,8 +185,6 @@ static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
kfree_rcu(s, rcu_head);
IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) released\n",
sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
-
- return 0;
}
diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c
index b873e17..c7ff978 100644
--- a/net/netfilter/ipvs/ip_vs_lblc.c
+++ b/net/netfilter/ipvs/ip_vs_lblc.c
@@ -388,7 +388,7 @@ static int ip_vs_lblc_init_svc(struct ip_vs_service *svc)
}
-static int ip_vs_lblc_done_svc(struct ip_vs_service *svc)
+static void ip_vs_lblc_done_svc(struct ip_vs_service *svc)
{
struct ip_vs_lblc_table *tbl = svc->sched_data;
@@ -402,8 +402,6 @@ static int ip_vs_lblc_done_svc(struct ip_vs_service *svc)
kfree_rcu(tbl, rcu_head);
IP_VS_DBG(6, "LBLC hash table (memory=%Zdbytes) released\n",
sizeof(*tbl));
-
- return 0;
}
diff --git a/net/netfilter/ipvs/ip_vs_lblcr.c b/net/netfilter/ipvs/ip_vs_lblcr.c
index c22f173..6049b85 100644
--- a/net/netfilter/ipvs/ip_vs_lblcr.c
+++ b/net/netfilter/ipvs/ip_vs_lblcr.c
@@ -555,7 +555,7 @@ static int ip_vs_lblcr_init_svc(struct ip_vs_service *svc)
}
-static int ip_vs_lblcr_done_svc(struct ip_vs_service *svc)
+static void ip_vs_lblcr_done_svc(struct ip_vs_service *svc)
{
struct ip_vs_lblcr_table *tbl = svc->sched_data;
@@ -569,8 +569,6 @@ static int ip_vs_lblcr_done_svc(struct ip_vs_service *svc)
kfree_rcu(tbl, rcu_head);
IP_VS_DBG(6, "LBLCR hash table (memory=%Zdbytes) released\n",
sizeof(*tbl));
-
- return 0;
}
diff --git a/net/netfilter/ipvs/ip_vs_sched.c b/net/netfilter/ipvs/ip_vs_sched.c
index 7f11d3d..1b715d0 100644
--- a/net/netfilter/ipvs/ip_vs_sched.c
+++ b/net/netfilter/ipvs/ip_vs_sched.c
@@ -64,22 +64,17 @@ int ip_vs_bind_scheduler(struct ip_vs_service *svc,
/*
* Unbind a service with its scheduler
*/
-int ip_vs_unbind_scheduler(struct ip_vs_service *svc)
+void ip_vs_unbind_scheduler(struct ip_vs_service *svc)
{
struct ip_vs_scheduler *sched = svc->scheduler;
if (!sched)
- return 0;
+ return;
- if (sched->done_service) {
- if (sched->done_service(svc) != 0) {
- pr_err("%s(): done error\n", __func__);
- return -EINVAL;
- }
- }
+ if (sched->done_service)
+ sched->done_service(svc);
svc->scheduler = NULL;
- return 0;
}
diff --git a/net/netfilter/ipvs/ip_vs_sh.c b/net/netfilter/ipvs/ip_vs_sh.c
index 55e76d8..81c1a10 100644
--- a/net/netfilter/ipvs/ip_vs_sh.c
+++ b/net/netfilter/ipvs/ip_vs_sh.c
@@ -187,7 +187,7 @@ static int ip_vs_sh_init_svc(struct ip_vs_service *svc)
}
-static int ip_vs_sh_done_svc(struct ip_vs_service *svc)
+static void ip_vs_sh_done_svc(struct ip_vs_service *svc)
{
struct ip_vs_sh_state *s = svc->sched_data;
@@ -198,8 +198,6 @@ static int ip_vs_sh_done_svc(struct ip_vs_service *svc)
kfree_rcu(s, rcu_head);
IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) released\n",
sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);
-
- return 0;
}
diff --git a/net/netfilter/ipvs/ip_vs_wrr.c b/net/netfilter/ipvs/ip_vs_wrr.c
index 98cb05e..a74fd9b 100644
--- a/net/netfilter/ipvs/ip_vs_wrr.c
+++ b/net/netfilter/ipvs/ip_vs_wrr.c
@@ -129,7 +129,7 @@ static int ip_vs_wrr_init_svc(struct ip_vs_service *svc)
}
-static int ip_vs_wrr_done_svc(struct ip_vs_service *svc)
+static void ip_vs_wrr_done_svc(struct ip_vs_service *svc)
{
struct ip_vs_wrr_mark *mark = svc->sched_data;
@@ -137,8 +137,6 @@ static int ip_vs_wrr_done_svc(struct ip_vs_service *svc)
* Release the mark variable
*/
kfree_rcu(mark, rcu_head);
-
- return 0;
}
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 31/34] ipvs: convert sched_lock to spin lock
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (29 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 30/34] ipvs: do not expect result from done_service Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 32/34] ipvs: convert dests to rcu Simon Horman
` (3 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
As all read_locks are gone spin lock is preferred.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 2 +-
net/netfilter/ipvs/ip_vs_ctl.c | 2 +-
net/netfilter/ipvs/ip_vs_lblc.c | 18 +++++++++---------
net/netfilter/ipvs/ip_vs_lblcr.c | 26 +++++++++++++-------------
net/netfilter/ipvs/ip_vs_rr.c | 10 +++++-----
net/netfilter/ipvs/ip_vs_wrr.c | 8 ++++----
6 files changed, 33 insertions(+), 33 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 4990de6..4a7bc63 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -734,7 +734,7 @@ struct ip_vs_service {
/* for scheduling */
struct ip_vs_scheduler *scheduler; /* bound scheduler object */
- rwlock_t sched_lock; /* lock sched_data */
+ spinlock_t sched_lock; /* lock sched_data */
void *sched_data; /* scheduler application data */
/* alternate persistence engine */
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index d022726..2bfd807 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -1219,7 +1219,7 @@ ip_vs_add_service(struct net *net, struct ip_vs_service_user_kern *u,
svc->net = net;
INIT_LIST_HEAD(&svc->destinations);
- rwlock_init(&svc->sched_lock);
+ spin_lock_init(&svc->sched_lock);
spin_lock_init(&svc->stats.lock);
/* Bind the scheduler */
diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c
index c7ff978..ffef8a1 100644
--- a/net/netfilter/ipvs/ip_vs_lblc.c
+++ b/net/netfilter/ipvs/ip_vs_lblc.c
@@ -194,7 +194,7 @@ ip_vs_lblc_get(int af, struct ip_vs_lblc_table *tbl,
/*
* Create or update an ip_vs_lblc_entry, which is a mapping of a destination IP
- * address to a server. Called under write lock.
+ * address to a server. Called under spin lock.
*/
static inline struct ip_vs_lblc_entry *
ip_vs_lblc_new(struct ip_vs_lblc_table *tbl, const union nf_inet_addr *daddr,
@@ -242,7 +242,7 @@ static void ip_vs_lblc_flush(struct ip_vs_service *svc)
struct hlist_node *next;
int i;
- write_lock_bh(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
tbl->dead = 1;
for (i=0; i<IP_VS_LBLC_TAB_SIZE; i++) {
hlist_for_each_entry_safe(en, next, &tbl->bucket[i], list) {
@@ -250,7 +250,7 @@ static void ip_vs_lblc_flush(struct ip_vs_service *svc)
atomic_dec(&tbl->entries);
}
}
- write_unlock_bh(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
}
static int sysctl_lblc_expiration(struct ip_vs_service *svc)
@@ -274,7 +274,7 @@ static inline void ip_vs_lblc_full_check(struct ip_vs_service *svc)
for (i=0, j=tbl->rover; i<IP_VS_LBLC_TAB_SIZE; i++) {
j = (j + 1) & IP_VS_LBLC_TAB_MASK;
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_before(now,
en->lastuse +
@@ -284,7 +284,7 @@ static inline void ip_vs_lblc_full_check(struct ip_vs_service *svc)
ip_vs_lblc_free(en);
atomic_dec(&tbl->entries);
}
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
}
tbl->rover = j;
}
@@ -330,7 +330,7 @@ static void ip_vs_lblc_check_expire(unsigned long data)
for (i=0, j=tbl->rover; i<IP_VS_LBLC_TAB_SIZE; i++) {
j = (j + 1) & IP_VS_LBLC_TAB_MASK;
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_before(now, en->lastuse + ENTRY_TIMEOUT))
continue;
@@ -339,7 +339,7 @@ static void ip_vs_lblc_check_expire(unsigned long data)
atomic_dec(&tbl->entries);
goal--;
}
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
if (goal <= 0)
break;
}
@@ -527,10 +527,10 @@ ip_vs_lblc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
}
/* If we fail to create a cache entry, we'll just use the valid dest */
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
if (!tbl->dead)
ip_vs_lblc_new(tbl, &iph.daddr, dest);
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
out:
IP_VS_DBG_BUF(6, "LBLC: destination IP address %s --> server %s:%d\n",
diff --git a/net/netfilter/ipvs/ip_vs_lblcr.c b/net/netfilter/ipvs/ip_vs_lblcr.c
index 6049b85..cdfe6a9 100644
--- a/net/netfilter/ipvs/ip_vs_lblcr.c
+++ b/net/netfilter/ipvs/ip_vs_lblcr.c
@@ -368,7 +368,7 @@ ip_vs_lblcr_get(int af, struct ip_vs_lblcr_table *tbl,
/*
* Create or update an ip_vs_lblcr_entry, which is a mapping of a destination
- * IP address to a server. Called under write lock.
+ * IP address to a server. Called under spin lock.
*/
static inline struct ip_vs_lblcr_entry *
ip_vs_lblcr_new(struct ip_vs_lblcr_table *tbl, const union nf_inet_addr *daddr,
@@ -412,14 +412,14 @@ static void ip_vs_lblcr_flush(struct ip_vs_service *svc)
struct ip_vs_lblcr_entry *en;
struct hlist_node *next;
- write_lock_bh(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
tbl->dead = 1;
for (i=0; i<IP_VS_LBLCR_TAB_SIZE; i++) {
hlist_for_each_entry_safe(en, next, &tbl->bucket[i], list) {
ip_vs_lblcr_free(en);
}
}
- write_unlock_bh(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
}
static int sysctl_lblcr_expiration(struct ip_vs_service *svc)
@@ -443,7 +443,7 @@ static inline void ip_vs_lblcr_full_check(struct ip_vs_service *svc)
for (i=0, j=tbl->rover; i<IP_VS_LBLCR_TAB_SIZE; i++) {
j = (j + 1) & IP_VS_LBLCR_TAB_MASK;
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_after(en->lastuse +
sysctl_lblcr_expiration(svc), now))
@@ -452,7 +452,7 @@ static inline void ip_vs_lblcr_full_check(struct ip_vs_service *svc)
ip_vs_lblcr_free(en);
atomic_dec(&tbl->entries);
}
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
}
tbl->rover = j;
}
@@ -498,7 +498,7 @@ static void ip_vs_lblcr_check_expire(unsigned long data)
for (i=0, j=tbl->rover; i<IP_VS_LBLCR_TAB_SIZE; i++) {
j = (j + 1) & IP_VS_LBLCR_TAB_MASK;
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
hlist_for_each_entry_safe(en, next, &tbl->bucket[j], list) {
if (time_before(now, en->lastuse+ENTRY_TIMEOUT))
continue;
@@ -507,7 +507,7 @@ static void ip_vs_lblcr_check_expire(unsigned long data)
atomic_dec(&tbl->entries);
goal--;
}
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
if (goal <= 0)
break;
}
@@ -678,7 +678,7 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
if (atomic_read(&en->set.size) > 1 &&
time_after(jiffies, en->set.lastmod +
sysctl_lblcr_expiration(svc))) {
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
if (atomic_read(&en->set.size) > 1) {
struct ip_vs_dest *m;
@@ -686,7 +686,7 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
if (m)
ip_vs_dest_set_erase(&en->set, m);
}
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
}
/* If the destination is not overloaded, use it */
@@ -701,10 +701,10 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
}
/* Update our cache entry */
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
if (!tbl->dead)
ip_vs_dest_set_insert(&en->set, dest, true);
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
goto out;
}
@@ -716,10 +716,10 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
}
/* If we fail to create a cache entry, we'll just use the valid dest */
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
if (!tbl->dead)
ip_vs_lblcr_new(tbl, &iph.daddr, dest);
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
out:
IP_VS_DBG_BUF(6, "LBLCR: destination IP address %s --> server %s:%d\n",
diff --git a/net/netfilter/ipvs/ip_vs_rr.c b/net/netfilter/ipvs/ip_vs_rr.c
index 3942890..aa4601f 100644
--- a/net/netfilter/ipvs/ip_vs_rr.c
+++ b/net/netfilter/ipvs/ip_vs_rr.c
@@ -39,14 +39,14 @@ static int ip_vs_rr_del_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest)
{
struct list_head *p;
- write_lock_bh(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
p = (struct list_head *) svc->sched_data;
/* dest is already unlinked, so p->prev is not valid but
* p->next is valid, use it to reach previous entry.
*/
if (p == &dest->n_list)
svc->sched_data = p->next->prev;
- write_unlock_bh(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
return 0;
}
@@ -63,7 +63,7 @@ ip_vs_rr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
p = (struct list_head *) svc->sched_data;
last = dest = list_entry(p, struct ip_vs_dest, n_list);
@@ -85,13 +85,13 @@ ip_vs_rr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
} while (pass < 2 && p != &svc->destinations);
stop:
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
ip_vs_scheduler_err(svc, "no destination available");
return NULL;
out:
svc->sched_data = &dest->n_list;
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
IP_VS_DBG_BUF(6, "RR: server %s:%u "
"activeconns %d refcnt %d weight %d\n",
IP_VS_DBG_ADDR(svc->af, &dest->addr), ntohs(dest->port),
diff --git a/net/netfilter/ipvs/ip_vs_wrr.c b/net/netfilter/ipvs/ip_vs_wrr.c
index a74fd9b..b173ef9 100644
--- a/net/netfilter/ipvs/ip_vs_wrr.c
+++ b/net/netfilter/ipvs/ip_vs_wrr.c
@@ -145,7 +145,7 @@ static int ip_vs_wrr_dest_changed(struct ip_vs_service *svc,
{
struct ip_vs_wrr_mark *mark = svc->sched_data;
- write_lock_bh(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
mark->cl = list_entry(&svc->destinations, struct ip_vs_dest, n_list);
mark->di = ip_vs_wrr_gcd_weight(svc);
mark->mw = ip_vs_wrr_max_weight(svc) - (mark->di - 1);
@@ -153,7 +153,7 @@ static int ip_vs_wrr_dest_changed(struct ip_vs_service *svc,
mark->cw = mark->mw;
else if (mark->di > 1)
mark->cw = (mark->cw / mark->di) * mark->di + 1;
- write_unlock_bh(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
return 0;
}
@@ -170,7 +170,7 @@ ip_vs_wrr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
- write_lock(&svc->sched_lock);
+ spin_lock(&svc->sched_lock);
dest = mark->cl;
/* No available dests? */
if (mark->mw == 0)
@@ -222,7 +222,7 @@ found:
mark->cl = dest;
out:
- write_unlock(&svc->sched_lock);
+ spin_unlock(&svc->sched_lock);
return dest;
err_noavail:
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 32/34] ipvs: convert dests to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (30 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 31/34] ipvs: convert sched_lock to spin lock Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 33/34] ipvs: convert services " Simon Horman
` (2 subsequent siblings)
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
In previous commits the schedulers started to access
svc->destinations with _rcu list traversal primitives
because the IP_VS_WAIT_WHILE macro still plays the role of
grace period. Now it is time to finish the updating part,
i.e. adding and deleting of dests with _rcu suffix before
removing the IP_VS_WAIT_WHILE in next commit.
We use the same rule for conns as for the
schedulers: dests can be searched in RCU read-side critical
section where ip_vs_dest_hold can be called by ip_vs_bind_dest.
Some things are not perfect, for example, calling
functions like ip_vs_lookup_dest from updating code under
RCU, just because we use some function both from reader
and from updater.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 2 +-
net/netfilter/ipvs/ip_vs_conn.c | 8 +++++---
net/netfilter/ipvs/ip_vs_ctl.c | 32 +++++++++++++++++---------------
net/netfilter/ipvs/ip_vs_sync.c | 11 ++++-------
4 files changed, 27 insertions(+), 26 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 4a7bc63..78a6634 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -1434,7 +1434,7 @@ extern struct ip_vs_dest *
ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
__be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
__u16 protocol, __u32 fwmark, __u32 flags);
-extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
+extern void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
{
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index 1b29e4a..54de340 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -611,10 +611,11 @@ ip_vs_bind_dest(struct ip_vs_conn *cp, struct ip_vs_dest *dest)
* Check if there is a destination for the connection, if so
* bind the connection to the destination.
*/
-struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp)
+void ip_vs_try_bind_dest(struct ip_vs_conn *cp)
{
struct ip_vs_dest *dest;
+ rcu_read_lock();
dest = ip_vs_find_dest(ip_vs_conn_net(cp), cp->af, &cp->daddr,
cp->dport, &cp->vaddr, cp->vport,
cp->protocol, cp->fwmark, cp->flags);
@@ -624,7 +625,8 @@ struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp)
spin_lock(&cp->lock);
if (cp->dest) {
spin_unlock(&cp->lock);
- return dest;
+ rcu_read_unlock();
+ return;
}
/* Applications work depending on the forwarding method
@@ -648,7 +650,7 @@ struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp)
if (pd && atomic_read(&pd->appcnt))
ip_vs_bind_app(cp, pd->pp);
}
- return dest;
+ rcu_read_unlock();
}
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 2bfd807..0763cc6 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -565,8 +565,8 @@ bool ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
return false;
}
-/*
- * Lookup destination by {addr,port} in the given service
+/* Lookup destination by {addr,port} in the given service
+ * Called under RCU lock.
*/
static struct ip_vs_dest *
ip_vs_lookup_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
@@ -577,7 +577,7 @@ ip_vs_lookup_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
/*
* Find the destination for the given service
*/
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
if ((dest->af == svc->af)
&& ip_vs_addr_equal(svc->af, &dest->addr, daddr)
&& (dest->port == dport)) {
@@ -591,10 +591,11 @@ ip_vs_lookup_dest(struct ip_vs_service *svc, const union nf_inet_addr *daddr,
/*
* Find destination by {daddr,dport,vaddr,protocol}
- * Cretaed to be used in ip_vs_process_message() in
+ * Created to be used in ip_vs_process_message() in
* the backup synchronization daemon. It finds the
* destination to be bound to the received connection
* on the backup.
+ * Called under RCU lock, no refcnt is returned.
*/
struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
const union nf_inet_addr *daddr,
@@ -615,8 +616,6 @@ struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
dest = ip_vs_lookup_dest(svc, daddr, port);
if (!dest)
dest = ip_vs_lookup_dest(svc, daddr, port ^ dport);
- if (dest)
- ip_vs_dest_hold(dest);
ip_vs_service_put(svc);
return dest;
}
@@ -826,7 +825,7 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
if (add) {
- list_add(&dest->n_list, &svc->destinations);
+ list_add_rcu(&dest->n_list, &svc->destinations);
svc->num_dests++;
if (svc->scheduler->add_dest)
svc->scheduler->add_dest(svc, dest);
@@ -933,10 +932,10 @@ ip_vs_add_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
ip_vs_addr_copy(svc->af, &daddr, &udest->addr);
- /*
- * Check if the dest already exists in the list
- */
+ /* We use function that requires RCU lock */
+ rcu_read_lock();
dest = ip_vs_lookup_dest(svc, &daddr, dport);
+ rcu_read_unlock();
if (dest != NULL) {
IP_VS_DBG(1, "%s(): dest already exists\n", __func__);
@@ -997,10 +996,10 @@ ip_vs_edit_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
ip_vs_addr_copy(svc->af, &daddr, &udest->addr);
- /*
- * Lookup the destination list
- */
+ /* We use function that requires RCU lock */
+ rcu_read_lock();
dest = ip_vs_lookup_dest(svc, &daddr, dport);
+ rcu_read_unlock();
if (dest == NULL) {
IP_VS_DBG(1, "%s(): dest doesn't exist\n", __func__);
@@ -1069,7 +1068,7 @@ static void __ip_vs_unlink_dest(struct ip_vs_service *svc,
/*
* Remove it from the d-linked destination list.
*/
- list_del(&dest->n_list);
+ list_del_rcu(&dest->n_list);
svc->num_dests--;
if (svcupd && svc->scheduler->del_dest)
@@ -1094,7 +1093,10 @@ ip_vs_del_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
EnterFunction(2);
+ /* We use function that requires RCU lock */
+ rcu_read_lock();
dest = ip_vs_lookup_dest(svc, &udest->addr, dport);
+ rcu_read_unlock();
if (dest == NULL) {
IP_VS_DBG(1, "%s(): destination not found!\n", __func__);
@@ -2104,7 +2106,7 @@ static int ip_vs_info_seq_show(struct seq_file *seq, void *v)
else
seq_putc(seq, '\n');
- list_for_each_entry(dest, &svc->destinations, n_list) {
+ list_for_each_entry_rcu(dest, &svc->destinations, n_list) {
#ifdef CONFIG_IP_VS_IPV6
if (dest->af == AF_INET6)
seq_printf(seq,
diff --git a/net/netfilter/ipvs/ip_vs_sync.c b/net/netfilter/ipvs/ip_vs_sync.c
index 6cc3e52..9724174 100644
--- a/net/netfilter/ipvs/ip_vs_sync.c
+++ b/net/netfilter/ipvs/ip_vs_sync.c
@@ -858,23 +858,20 @@ static void ip_vs_proc_conn(struct net *net, struct ip_vs_conn_param *param,
flags |= cp->flags & ~IP_VS_CONN_F_BACKUP_UPD_MASK;
cp->flags = flags;
spin_unlock(&cp->lock);
- if (!dest) {
- dest = ip_vs_try_bind_dest(cp);
- if (dest)
- ip_vs_dest_put(dest);
- }
+ if (!dest)
+ ip_vs_try_bind_dest(cp);
} else {
/*
* Find the appropriate destination for the connection.
* If it is not found the connection will remain unbound
* but still handled.
*/
+ rcu_read_lock();
dest = ip_vs_find_dest(net, type, daddr, dport, param->vaddr,
param->vport, protocol, fwmark, flags);
cp = ip_vs_conn_new(param, daddr, dport, flags, dest, fwmark);
- if (dest)
- ip_vs_dest_put(dest);
+ rcu_read_unlock();
if (!cp) {
if (param->pe_data)
kfree(param->pe_data);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 33/34] ipvs: convert services to rcu
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (31 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 32/34] ipvs: convert dests to rcu Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 34/34] ipvs: do not disable bh for long time Simon Horman
2013-04-01 22:41 ` [GIT PULL nf-next] IPVS optimisations for v3.10 Pablo Neira Ayuso
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
This is the final step in RCU conversion.
Things that are removed:
- svc->usecnt: now svc is accessed under RCU read lock
- svc->inc: and some unused code
- ip_vs_bind_pe and ip_vs_unbind_pe: no ability to replace PE
- __ip_vs_svc_lock: replaced with RCU
- IP_VS_WAIT_WHILE: now readers lookup svcs and dests under
RCU and work in parallel with configuration
Other changes:
- before now, a RCU read-side critical section included the
calling of the schedule method, now it is extended to include
service lookup
- ip_vs_svc_table and ip_vs_svc_fwm_table are now using hlist
- svc->pe and svc->scheduler remain to the end (of grace period),
the schedulers are prepared for such RCU readers
even after done_service is called but they need
to use synchronize_rcu because last ip_vs_scheduler_put
can happen while RCU read-side critical sections
use an outdated svc->scheduler pointer
- as planned, update_service is removed
- empty services can be freed immediately after grace period.
If dests were present, the services are freed from
the dest trash code
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
include/net/ip_vs.h | 28 +--
net/netfilter/ipvs/ip_vs_core.c | 26 +--
net/netfilter/ipvs/ip_vs_ctl.c | 299 ++++++++++++++-------------------
net/netfilter/ipvs/ip_vs_dh.c | 1 +
net/netfilter/ipvs/ip_vs_lblc.c | 1 +
net/netfilter/ipvs/ip_vs_lblcr.c | 1 +
net/netfilter/ipvs/ip_vs_lc.c | 1 +
net/netfilter/ipvs/ip_vs_nq.c | 1 +
net/netfilter/ipvs/ip_vs_pe.c | 12 --
net/netfilter/ipvs/ip_vs_proto_sctp.c | 14 +-
net/netfilter/ipvs/ip_vs_proto_tcp.c | 14 +-
net/netfilter/ipvs/ip_vs_proto_udp.c | 14 +-
net/netfilter/ipvs/ip_vs_rr.c | 1 +
net/netfilter/ipvs/ip_vs_sched.c | 28 +--
net/netfilter/ipvs/ip_vs_sed.c | 1 +
net/netfilter/ipvs/ip_vs_sh.c | 1 +
net/netfilter/ipvs/ip_vs_wlc.c | 1 +
net/netfilter/ipvs/ip_vs_wrr.c | 1 +
18 files changed, 186 insertions(+), 259 deletions(-)
diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index 78a6634..f9f5b05 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -359,8 +359,6 @@ static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
#define LeaveFunction(level) do {} while (0)
#endif
-#define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
-
/*
* The port number of FTP service (in network order).
@@ -712,10 +710,9 @@ struct ip_vs_dest_user_kern {
* and the forwarding entries
*/
struct ip_vs_service {
- struct list_head s_list; /* for normal service table */
- struct list_head f_list; /* for fwmark-based service table */
+ struct hlist_node s_list; /* for normal service table */
+ struct hlist_node f_list; /* for fwmark-based service table */
atomic_t refcnt; /* reference counter */
- atomic_t usecnt; /* use counter */
u16 af; /* address family */
__u16 protocol; /* which protocol (TCP/UDP) */
@@ -730,15 +727,16 @@ struct ip_vs_service {
struct list_head destinations; /* real server d-linked list */
__u32 num_dests; /* number of servers */
struct ip_vs_stats stats; /* statistics for the service */
- struct ip_vs_app *inc; /* bind conns to this app inc */
/* for scheduling */
- struct ip_vs_scheduler *scheduler; /* bound scheduler object */
+ struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
spinlock_t sched_lock; /* lock sched_data */
void *sched_data; /* scheduler application data */
/* alternate persistence engine */
- struct ip_vs_pe *pe;
+ struct ip_vs_pe __rcu *pe;
+
+ struct rcu_head rcu_head;
};
/* Information for cached dst */
@@ -807,8 +805,6 @@ struct ip_vs_scheduler {
int (*init_service)(struct ip_vs_service *svc);
/* scheduling service finish */
void (*done_service)(struct ip_vs_service *svc);
- /* scheduler updating service */
- int (*update_service)(struct ip_vs_service *svc);
/* dest is linked */
int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
/* dest is unlinked */
@@ -1344,8 +1340,6 @@ extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
-void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
-void ip_vs_unbind_pe(struct ip_vs_service *svc);
int register_ip_vs_pe(struct ip_vs_pe *pe);
int unregister_ip_vs_pe(struct ip_vs_pe *pe);
struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
@@ -1392,7 +1386,8 @@ extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
struct ip_vs_scheduler *scheduler);
-extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc);
+extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
+ struct ip_vs_scheduler *sched);
extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
extern struct ip_vs_conn *
@@ -1412,14 +1407,9 @@ extern struct ip_vs_stats ip_vs_stats;
extern int sysctl_ip_vs_sync_ver;
extern struct ip_vs_service *
-ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
+ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
const union nf_inet_addr *vaddr, __be16 vport);
-static inline void ip_vs_service_put(struct ip_vs_service *svc)
-{
- atomic_dec(&svc->usecnt);
-}
-
extern bool
ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
const union nf_inet_addr *daddr, __be16 dport);
diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c
index 939ad11..79df3c6 100644
--- a/net/netfilter/ipvs/ip_vs_core.c
+++ b/net/netfilter/ipvs/ip_vs_core.c
@@ -203,7 +203,7 @@ ip_vs_conn_fill_param_persist(const struct ip_vs_service *svc,
{
ip_vs_conn_fill_param(svc->net, svc->af, protocol, caddr, cport, vaddr,
vport, p);
- p->pe = svc->pe;
+ p->pe = rcu_dereference(svc->pe);
if (p->pe && p->pe->fill_param)
return p->pe->fill_param(p, skb);
@@ -296,15 +296,16 @@ ip_vs_sched_persist(struct ip_vs_service *svc,
/* Check if a template already exists */
ct = ip_vs_ct_in_get(¶m);
if (!ct || !ip_vs_check_template(ct)) {
+ struct ip_vs_scheduler *sched;
+
/*
* No template found or the dest of the connection
* template is not available.
* return *ignored=0 i.e. ICMP and NF_DROP
*/
- rcu_read_lock();
- dest = svc->scheduler->schedule(svc, skb);
+ sched = rcu_dereference(svc->scheduler);
+ dest = sched->schedule(svc, skb);
if (!dest) {
- rcu_read_unlock();
IP_VS_DBG(1, "p-schedule: no dest found.\n");
kfree(param.pe_data);
*ignored = 0;
@@ -320,7 +321,6 @@ ip_vs_sched_persist(struct ip_vs_service *svc,
* when the template expires */
ct = ip_vs_conn_new(¶m, &dest->addr, dport,
IP_VS_CONN_F_TEMPLATE, dest, skb->mark);
- rcu_read_unlock();
if (ct == NULL) {
kfree(param.pe_data);
*ignored = -1;
@@ -394,6 +394,7 @@ ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
{
struct ip_vs_protocol *pp = pd->pp;
struct ip_vs_conn *cp = NULL;
+ struct ip_vs_scheduler *sched;
struct ip_vs_dest *dest;
__be16 _ports[2], *pptr;
unsigned int flags;
@@ -449,10 +450,9 @@ ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
return NULL;
}
- rcu_read_lock();
- dest = svc->scheduler->schedule(svc, skb);
+ sched = rcu_dereference(svc->scheduler);
+ dest = sched->schedule(svc, skb);
if (dest == NULL) {
- rcu_read_unlock();
IP_VS_DBG(1, "Schedule: no dest found.\n");
return NULL;
}
@@ -473,7 +473,6 @@ ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
cp = ip_vs_conn_new(&p, &dest->addr,
dest->port ? dest->port : pptr[1],
flags, dest, skb->mark);
- rcu_read_unlock();
if (!cp) {
*ignored = -1;
return NULL;
@@ -510,7 +509,6 @@ int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
pptr = frag_safe_skb_hp(skb, iph->len, sizeof(_ports), _ports, iph);
if (pptr == NULL) {
- ip_vs_service_put(svc);
return NF_DROP;
}
@@ -536,8 +534,6 @@ int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
IP_VS_CONN_F_ONE_PACKET : 0;
union nf_inet_addr daddr = { .all = { 0, 0, 0, 0 } };
- ip_vs_service_put(svc);
-
/* create a new connection entry */
IP_VS_DBG(6, "%s(): create a cache_bypass entry\n", __func__);
{
@@ -574,12 +570,8 @@ int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
* listed in the ipvs table), pass the packets, because it is
* not ipvs job to decide to drop the packets.
*/
- if ((svc->port == FTPPORT) && (pptr[1] != FTPPORT)) {
- ip_vs_service_put(svc);
+ if ((svc->port == FTPPORT) && (pptr[1] != FTPPORT))
return NF_ACCEPT;
- }
-
- ip_vs_service_put(svc);
/*
* Notify the client that the destination is unreachable, and
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 0763cc6..9e4074c 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -55,9 +55,6 @@
/* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
static DEFINE_MUTEX(__ip_vs_mutex);
-/* lock for service table */
-static DEFINE_RWLOCK(__ip_vs_svc_lock);
-
/* sysctl variables */
#ifdef CONFIG_IP_VS_DEBUG
@@ -257,9 +254,9 @@ ip_vs_use_count_dec(void)
#define IP_VS_SVC_TAB_MASK (IP_VS_SVC_TAB_SIZE - 1)
/* the service table hashed by <protocol, addr, port> */
-static struct list_head ip_vs_svc_table[IP_VS_SVC_TAB_SIZE];
+static struct hlist_head ip_vs_svc_table[IP_VS_SVC_TAB_SIZE];
/* the service table hashed by fwmark */
-static struct list_head ip_vs_svc_fwm_table[IP_VS_SVC_TAB_SIZE];
+static struct hlist_head ip_vs_svc_fwm_table[IP_VS_SVC_TAB_SIZE];
/*
@@ -314,13 +311,13 @@ static int ip_vs_svc_hash(struct ip_vs_service *svc)
*/
hash = ip_vs_svc_hashkey(svc->net, svc->af, svc->protocol,
&svc->addr, svc->port);
- list_add(&svc->s_list, &ip_vs_svc_table[hash]);
+ hlist_add_head_rcu(&svc->s_list, &ip_vs_svc_table[hash]);
} else {
/*
* Hash it by fwmark in svc_fwm_table
*/
hash = ip_vs_svc_fwm_hashkey(svc->net, svc->fwmark);
- list_add(&svc->f_list, &ip_vs_svc_fwm_table[hash]);
+ hlist_add_head_rcu(&svc->f_list, &ip_vs_svc_fwm_table[hash]);
}
svc->flags |= IP_VS_SVC_F_HASHED;
@@ -344,10 +341,10 @@ static int ip_vs_svc_unhash(struct ip_vs_service *svc)
if (svc->fwmark == 0) {
/* Remove it from the svc_table table */
- list_del(&svc->s_list);
+ hlist_del_rcu(&svc->s_list);
} else {
/* Remove it from the svc_fwm_table table */
- list_del(&svc->f_list);
+ hlist_del_rcu(&svc->f_list);
}
svc->flags &= ~IP_VS_SVC_F_HASHED;
@@ -369,7 +366,7 @@ __ip_vs_service_find(struct net *net, int af, __u16 protocol,
/* Check for "full" addressed entries */
hash = ip_vs_svc_hashkey(net, af, protocol, vaddr, vport);
- list_for_each_entry(svc, &ip_vs_svc_table[hash], s_list){
+ hlist_for_each_entry_rcu(svc, &ip_vs_svc_table[hash], s_list) {
if ((svc->af == af)
&& ip_vs_addr_equal(af, &svc->addr, vaddr)
&& (svc->port == vport)
@@ -396,7 +393,7 @@ __ip_vs_svc_fwm_find(struct net *net, int af, __u32 fwmark)
/* Check for fwmark addressed entries */
hash = ip_vs_svc_fwm_hashkey(net, fwmark);
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[hash], f_list) {
+ hlist_for_each_entry_rcu(svc, &ip_vs_svc_fwm_table[hash], f_list) {
if (svc->fwmark == fwmark && svc->af == af
&& net_eq(svc->net, net)) {
/* HIT */
@@ -407,15 +404,14 @@ __ip_vs_svc_fwm_find(struct net *net, int af, __u32 fwmark)
return NULL;
}
+/* Find service, called under RCU lock */
struct ip_vs_service *
-ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
- const union nf_inet_addr *vaddr, __be16 vport)
+ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
+ const union nf_inet_addr *vaddr, __be16 vport)
{
struct ip_vs_service *svc;
struct netns_ipvs *ipvs = net_ipvs(net);
- read_lock(&__ip_vs_svc_lock);
-
/*
* Check the table hashed by fwmark first
*/
@@ -451,10 +447,6 @@ ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
}
out:
- if (svc)
- atomic_inc(&svc->usecnt);
- read_unlock(&__ip_vs_svc_lock);
-
IP_VS_DBG_BUF(9, "lookup service: fwm %u %s %s:%u %s\n",
fwmark, ip_vs_proto_name(protocol),
IP_VS_DBG_ADDR(af, vaddr), ntohs(vport),
@@ -471,6 +463,13 @@ __ip_vs_bind_svc(struct ip_vs_dest *dest, struct ip_vs_service *svc)
dest->svc = svc;
}
+static void ip_vs_service_free(struct ip_vs_service *svc)
+{
+ if (svc->stats.cpustats)
+ free_percpu(svc->stats.cpustats);
+ kfree(svc);
+}
+
static void
__ip_vs_unbind_svc(struct ip_vs_dest *dest)
{
@@ -478,12 +477,11 @@ __ip_vs_unbind_svc(struct ip_vs_dest *dest)
dest->svc = NULL;
if (atomic_dec_and_test(&svc->refcnt)) {
- IP_VS_DBG_BUF(3, "Removing service %u/%s:%u usecnt=%d\n",
+ IP_VS_DBG_BUF(3, "Removing service %u/%s:%u\n",
svc->fwmark,
IP_VS_DBG_ADDR(svc->af, &svc->addr),
- ntohs(svc->port), atomic_read(&svc->usecnt));
- free_percpu(svc->stats.cpustats);
- kfree(svc);
+ ntohs(svc->port));
+ ip_vs_service_free(svc);
}
}
@@ -608,7 +606,7 @@ struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
struct ip_vs_service *svc;
__be16 port = dport;
- svc = ip_vs_service_get(net, af, fwmark, protocol, vaddr, vport);
+ svc = ip_vs_service_find(net, af, fwmark, protocol, vaddr, vport);
if (!svc)
return NULL;
if (fwmark && (flags & IP_VS_CONN_F_FWD_MASK) != IP_VS_CONN_F_MASQ)
@@ -616,7 +614,6 @@ struct ip_vs_dest *ip_vs_find_dest(struct net *net, int af,
dest = ip_vs_lookup_dest(svc, daddr, port);
if (!dest)
dest = ip_vs_lookup_dest(svc, daddr, port ^ dport);
- ip_vs_service_put(svc);
return dest;
}
@@ -774,6 +771,7 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
struct ip_vs_dest_user_kern *udest, int add)
{
struct netns_ipvs *ipvs = net_ipvs(svc->net);
+ struct ip_vs_scheduler *sched;
int conn_flags;
/* set the weight and the flags */
@@ -816,29 +814,17 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
__ip_vs_dst_cache_reset(dest);
spin_unlock_bh(&dest->dst_lock);
- if (add)
- ip_vs_start_estimator(svc->net, &dest->stats);
-
- write_lock_bh(&__ip_vs_svc_lock);
-
- /* Wait until all other svc users go away */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
-
+ sched = rcu_dereference_protected(svc->scheduler, 1);
if (add) {
+ ip_vs_start_estimator(svc->net, &dest->stats);
list_add_rcu(&dest->n_list, &svc->destinations);
svc->num_dests++;
- if (svc->scheduler->add_dest)
- svc->scheduler->add_dest(svc, dest);
+ if (sched->add_dest)
+ sched->add_dest(svc, dest);
} else {
- if (svc->scheduler->upd_dest)
- svc->scheduler->upd_dest(svc, dest);
+ if (sched->upd_dest)
+ sched->upd_dest(svc, dest);
}
-
- /* call the update_service, because server weight may be changed */
- if (svc->scheduler->update_service)
- svc->scheduler->update_service(svc);
-
- write_unlock_bh(&__ip_vs_svc_lock);
}
@@ -1071,14 +1057,13 @@ static void __ip_vs_unlink_dest(struct ip_vs_service *svc,
list_del_rcu(&dest->n_list);
svc->num_dests--;
- if (svcupd && svc->scheduler->del_dest)
- svc->scheduler->del_dest(svc, dest);
+ if (svcupd) {
+ struct ip_vs_scheduler *sched;
- /*
- * Call the update_service function of its scheduler
- */
- if (svcupd && svc->scheduler->update_service)
- svc->scheduler->update_service(svc);
+ sched = rcu_dereference_protected(svc->scheduler, 1);
+ if (sched->del_dest)
+ sched->del_dest(svc, dest);
+ }
}
@@ -1103,20 +1088,11 @@ ip_vs_del_dest(struct ip_vs_service *svc, struct ip_vs_dest_user_kern *udest)
return -ENOENT;
}
- write_lock_bh(&__ip_vs_svc_lock);
-
- /*
- * Wait until all other svc users go away.
- */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
-
/*
* Unlink dest from the service
*/
__ip_vs_unlink_dest(svc, dest, 1);
- write_unlock_bh(&__ip_vs_svc_lock);
-
/*
* Delete the destination
*/
@@ -1207,7 +1183,6 @@ ip_vs_add_service(struct net *net, struct ip_vs_service_user_kern *u,
}
/* I'm the first user of the service */
- atomic_set(&svc->usecnt, 0);
atomic_set(&svc->refcnt, 0);
svc->af = u->af;
@@ -1231,7 +1206,7 @@ ip_vs_add_service(struct net *net, struct ip_vs_service_user_kern *u,
sched = NULL;
/* Bind the ct retriever */
- ip_vs_bind_pe(svc, pe);
+ RCU_INIT_POINTER(svc->pe, pe);
pe = NULL;
/* Update the virtual service counters */
@@ -1247,9 +1222,7 @@ ip_vs_add_service(struct net *net, struct ip_vs_service_user_kern *u,
ipvs->num_services++;
/* Hash the service into the service table */
- write_lock_bh(&__ip_vs_svc_lock);
ip_vs_svc_hash(svc);
- write_unlock_bh(&__ip_vs_svc_lock);
*svc_p = svc;
/* Now there is a service - full throttle */
@@ -1259,15 +1232,8 @@ ip_vs_add_service(struct net *net, struct ip_vs_service_user_kern *u,
out_err:
if (svc != NULL) {
- ip_vs_unbind_scheduler(svc);
- if (svc->inc) {
- local_bh_disable();
- ip_vs_app_inc_put(svc->inc);
- local_bh_enable();
- }
- if (svc->stats.cpustats)
- free_percpu(svc->stats.cpustats);
- kfree(svc);
+ ip_vs_unbind_scheduler(svc, sched);
+ ip_vs_service_free(svc);
}
ip_vs_scheduler_put(sched);
ip_vs_pe_put(pe);
@@ -1317,12 +1283,17 @@ ip_vs_edit_service(struct ip_vs_service *svc, struct ip_vs_service_user_kern *u)
}
#endif
- write_lock_bh(&__ip_vs_svc_lock);
-
- /*
- * Wait until all other svc users go away.
- */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
+ old_sched = rcu_dereference_protected(svc->scheduler, 1);
+ if (sched != old_sched) {
+ /* Bind the new scheduler */
+ ret = ip_vs_bind_scheduler(svc, sched);
+ if (ret) {
+ old_sched = sched;
+ goto out;
+ }
+ /* Unbind the old scheduler on success */
+ ip_vs_unbind_scheduler(svc, old_sched);
+ }
/*
* Set the flags and timeout value
@@ -1331,47 +1302,23 @@ ip_vs_edit_service(struct ip_vs_service *svc, struct ip_vs_service_user_kern *u)
svc->timeout = u->timeout * HZ;
svc->netmask = u->netmask;
- old_sched = svc->scheduler;
- if (sched != old_sched) {
- /*
- * Unbind the old scheduler
- */
- ip_vs_unbind_scheduler(svc);
+ old_pe = rcu_dereference_protected(svc->pe, 1);
+ if (pe != old_pe)
+ rcu_assign_pointer(svc->pe, pe);
- /*
- * Bind the new scheduler
- */
- if ((ret = ip_vs_bind_scheduler(svc, sched))) {
- /*
- * If ip_vs_bind_scheduler fails, restore the old
- * scheduler.
- * The main reason of failure is out of memory.
- *
- * The question is if the old scheduler can be
- * restored all the time. TODO: if it cannot be
- * restored some time, we must delete the service,
- * otherwise the system may crash.
- */
- ip_vs_bind_scheduler(svc, old_sched);
- old_sched = sched;
- goto out_unlock;
- }
- }
-
- old_pe = svc->pe;
- if (pe != old_pe) {
- ip_vs_unbind_pe(svc);
- ip_vs_bind_pe(svc, pe);
- }
-
-out_unlock:
- write_unlock_bh(&__ip_vs_svc_lock);
out:
ip_vs_scheduler_put(old_sched);
ip_vs_pe_put(old_pe);
return ret;
}
+static void ip_vs_service_rcu_free(struct rcu_head *head)
+{
+ struct ip_vs_service *svc;
+
+ svc = container_of(head, struct ip_vs_service, rcu_head);
+ ip_vs_service_free(svc);
+}
/*
* Delete a service from the service list
@@ -1394,21 +1341,14 @@ static void __ip_vs_del_service(struct ip_vs_service *svc, bool cleanup)
ip_vs_stop_estimator(svc->net, &svc->stats);
/* Unbind scheduler */
- old_sched = svc->scheduler;
- ip_vs_unbind_scheduler(svc);
+ old_sched = rcu_dereference_protected(svc->scheduler, 1);
+ ip_vs_unbind_scheduler(svc, old_sched);
ip_vs_scheduler_put(old_sched);
- /* Unbind persistence engine */
- old_pe = svc->pe;
- ip_vs_unbind_pe(svc);
+ /* Unbind persistence engine, keep svc->pe */
+ old_pe = rcu_dereference_protected(svc->pe, 1);
ip_vs_pe_put(old_pe);
- /* Unbind app inc */
- if (svc->inc) {
- ip_vs_app_inc_put(svc->inc);
- svc->inc = NULL;
- }
-
/*
* Unlink the whole destination list
*/
@@ -1428,13 +1368,12 @@ static void __ip_vs_del_service(struct ip_vs_service *svc, bool cleanup)
/*
* Free the service if nobody refers to it
*/
- if (atomic_read(&svc->refcnt) == 0) {
- IP_VS_DBG_BUF(3, "Removing service %u/%s:%u usecnt=%d\n",
+ if (atomic_dec_and_test(&svc->refcnt)) {
+ IP_VS_DBG_BUF(3, "Removing service %u/%s:%u\n",
svc->fwmark,
IP_VS_DBG_ADDR(svc->af, &svc->addr),
- ntohs(svc->port), atomic_read(&svc->usecnt));
- free_percpu(svc->stats.cpustats);
- kfree(svc);
+ ntohs(svc->port));
+ call_rcu(&svc->rcu_head, ip_vs_service_rcu_free);
}
/* decrease the module use count */
@@ -1446,21 +1385,14 @@ static void __ip_vs_del_service(struct ip_vs_service *svc, bool cleanup)
*/
static void ip_vs_unlink_service(struct ip_vs_service *svc, bool cleanup)
{
+ /* Hold svc to avoid double release from dest_trash */
+ atomic_inc(&svc->refcnt);
/*
* Unhash it from the service table
*/
- write_lock_bh(&__ip_vs_svc_lock);
-
ip_vs_svc_unhash(svc);
- /*
- * Wait until all the svc users go away.
- */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
-
__ip_vs_del_service(svc, cleanup);
-
- write_unlock_bh(&__ip_vs_svc_lock);
}
/*
@@ -1482,14 +1414,15 @@ static int ip_vs_del_service(struct ip_vs_service *svc)
static int ip_vs_flush(struct net *net, bool cleanup)
{
int idx;
- struct ip_vs_service *svc, *nxt;
+ struct ip_vs_service *svc;
+ struct hlist_node *n;
/*
* Flush the service table hashed by <netns,protocol,addr,port>
*/
for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry_safe(svc, nxt, &ip_vs_svc_table[idx],
- s_list) {
+ hlist_for_each_entry_safe(svc, n, &ip_vs_svc_table[idx],
+ s_list) {
if (net_eq(svc->net, net))
ip_vs_unlink_service(svc, cleanup);
}
@@ -1499,8 +1432,8 @@ static int ip_vs_flush(struct net *net, bool cleanup)
* Flush the service table hashed by fwmark
*/
for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry_safe(svc, nxt,
- &ip_vs_svc_fwm_table[idx], f_list) {
+ hlist_for_each_entry_safe(svc, n, &ip_vs_svc_fwm_table[idx],
+ f_list) {
if (net_eq(svc->net, net))
ip_vs_unlink_service(svc, cleanup);
}
@@ -1558,7 +1491,7 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
EnterFunction(2);
mutex_lock(&__ip_vs_mutex);
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
if (net_eq(svc->net, net)) {
list_for_each_entry(dest, &svc->destinations,
n_list) {
@@ -1567,7 +1500,7 @@ static int ip_vs_dst_event(struct notifier_block *this, unsigned long event,
}
}
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
if (net_eq(svc->net, net)) {
list_for_each_entry(dest, &svc->destinations,
n_list) {
@@ -1595,12 +1528,10 @@ static int ip_vs_zero_service(struct ip_vs_service *svc)
{
struct ip_vs_dest *dest;
- write_lock_bh(&__ip_vs_svc_lock);
list_for_each_entry(dest, &svc->destinations, n_list) {
ip_vs_zero_stats(&dest->stats);
}
ip_vs_zero_stats(&svc->stats);
- write_unlock_bh(&__ip_vs_svc_lock);
return 0;
}
@@ -1610,14 +1541,14 @@ static int ip_vs_zero_all(struct net *net)
struct ip_vs_service *svc;
for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
if (net_eq(svc->net, net))
ip_vs_zero_service(svc);
}
}
for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
if (net_eq(svc->net, net))
ip_vs_zero_service(svc);
}
@@ -1945,7 +1876,7 @@ static struct ctl_table vs_vars[] = {
struct ip_vs_iter {
struct seq_net_private p; /* Do not move this, netns depends upon it*/
- struct list_head *table;
+ struct hlist_head *table;
int bucket;
};
@@ -1978,7 +1909,7 @@ static struct ip_vs_service *ip_vs_info_array(struct seq_file *seq, loff_t pos)
/* look in hash by protocol */
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
+ hlist_for_each_entry_rcu(svc, &ip_vs_svc_table[idx], s_list) {
if (net_eq(svc->net, net) && pos-- == 0) {
iter->table = ip_vs_svc_table;
iter->bucket = idx;
@@ -1989,7 +1920,8 @@ static struct ip_vs_service *ip_vs_info_array(struct seq_file *seq, loff_t pos)
/* keep looking in fwmark */
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
+ hlist_for_each_entry_rcu(svc, &ip_vs_svc_fwm_table[idx],
+ f_list) {
if (net_eq(svc->net, net) && pos-- == 0) {
iter->table = ip_vs_svc_fwm_table;
iter->bucket = idx;
@@ -2002,17 +1934,16 @@ static struct ip_vs_service *ip_vs_info_array(struct seq_file *seq, loff_t pos)
}
static void *ip_vs_info_seq_start(struct seq_file *seq, loff_t *pos)
-__acquires(__ip_vs_svc_lock)
{
- read_lock_bh(&__ip_vs_svc_lock);
+ rcu_read_lock();
return *pos ? ip_vs_info_array(seq, *pos - 1) : SEQ_START_TOKEN;
}
static void *ip_vs_info_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
- struct list_head *e;
+ struct hlist_node *e;
struct ip_vs_iter *iter;
struct ip_vs_service *svc;
@@ -2025,13 +1956,14 @@ static void *ip_vs_info_seq_next(struct seq_file *seq, void *v, loff_t *pos)
if (iter->table == ip_vs_svc_table) {
/* next service in table hashed by protocol */
- if ((e = svc->s_list.next) != &ip_vs_svc_table[iter->bucket])
- return list_entry(e, struct ip_vs_service, s_list);
-
+ e = rcu_dereference(hlist_next_rcu(&svc->s_list));
+ if (e)
+ return hlist_entry(e, struct ip_vs_service, s_list);
while (++iter->bucket < IP_VS_SVC_TAB_SIZE) {
- list_for_each_entry(svc,&ip_vs_svc_table[iter->bucket],
- s_list) {
+ hlist_for_each_entry_rcu(svc,
+ &ip_vs_svc_table[iter->bucket],
+ s_list) {
return svc;
}
}
@@ -2042,13 +1974,15 @@ static void *ip_vs_info_seq_next(struct seq_file *seq, void *v, loff_t *pos)
}
/* next service in hashed by fwmark */
- if ((e = svc->f_list.next) != &ip_vs_svc_fwm_table[iter->bucket])
- return list_entry(e, struct ip_vs_service, f_list);
+ e = rcu_dereference(hlist_next_rcu(&svc->f_list));
+ if (e)
+ return hlist_entry(e, struct ip_vs_service, f_list);
scan_fwmark:
while (++iter->bucket < IP_VS_SVC_TAB_SIZE) {
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[iter->bucket],
- f_list)
+ hlist_for_each_entry_rcu(svc,
+ &ip_vs_svc_fwm_table[iter->bucket],
+ f_list)
return svc;
}
@@ -2056,9 +1990,8 @@ static void *ip_vs_info_seq_next(struct seq_file *seq, void *v, loff_t *pos)
}
static void ip_vs_info_seq_stop(struct seq_file *seq, void *v)
-__releases(__ip_vs_svc_lock)
{
- read_unlock_bh(&__ip_vs_svc_lock);
+ rcu_read_unlock();
}
@@ -2076,6 +2009,7 @@ static int ip_vs_info_seq_show(struct seq_file *seq, void *v)
const struct ip_vs_service *svc = v;
const struct ip_vs_iter *iter = seq->private;
const struct ip_vs_dest *dest;
+ struct ip_vs_scheduler *sched = rcu_dereference(svc->scheduler);
if (iter->table == ip_vs_svc_table) {
#ifdef CONFIG_IP_VS_IPV6
@@ -2084,18 +2018,18 @@ static int ip_vs_info_seq_show(struct seq_file *seq, void *v)
ip_vs_proto_name(svc->protocol),
&svc->addr.in6,
ntohs(svc->port),
- svc->scheduler->name);
+ sched->name);
else
#endif
seq_printf(seq, "%s %08X:%04X %s %s ",
ip_vs_proto_name(svc->protocol),
ntohl(svc->addr.ip),
ntohs(svc->port),
- svc->scheduler->name,
+ sched->name,
(svc->flags & IP_VS_SVC_F_ONEPACKET)?"ops ":"");
} else {
seq_printf(seq, "FWM %08X %s %s",
- svc->fwmark, svc->scheduler->name,
+ svc->fwmark, sched->name,
(svc->flags & IP_VS_SVC_F_ONEPACKET)?"ops ":"");
}
@@ -2451,11 +2385,13 @@ do_ip_vs_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
}
/* Lookup the exact service by <protocol, addr, port> or fwmark */
+ rcu_read_lock();
if (usvc.fwmark == 0)
svc = __ip_vs_service_find(net, usvc.af, usvc.protocol,
&usvc.addr, usvc.port);
else
svc = __ip_vs_svc_fwm_find(net, usvc.af, usvc.fwmark);
+ rcu_read_unlock();
if (cmd != IP_VS_SO_SET_ADD
&& (svc == NULL || svc->protocol != usvc.protocol)) {
@@ -2507,11 +2443,14 @@ do_ip_vs_set_ctl(struct sock *sk, int cmd, void __user *user, unsigned int len)
static void
ip_vs_copy_service(struct ip_vs_service_entry *dst, struct ip_vs_service *src)
{
+ struct ip_vs_scheduler *sched;
+
+ sched = rcu_dereference_protected(src->scheduler, 1);
dst->protocol = src->protocol;
dst->addr = src->addr.ip;
dst->port = src->port;
dst->fwmark = src->fwmark;
- strlcpy(dst->sched_name, src->scheduler->name, sizeof(dst->sched_name));
+ strlcpy(dst->sched_name, sched->name, sizeof(dst->sched_name));
dst->flags = src->flags;
dst->timeout = src->timeout / HZ;
dst->netmask = src->netmask;
@@ -2530,7 +2469,7 @@ __ip_vs_get_service_entries(struct net *net,
int ret = 0;
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_table[idx], s_list) {
/* Only expose IPv4 entries to old interface */
if (svc->af != AF_INET || !net_eq(svc->net, net))
continue;
@@ -2549,7 +2488,7 @@ __ip_vs_get_service_entries(struct net *net,
}
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_fwm_table[idx], f_list) {
/* Only expose IPv4 entries to old interface */
if (svc->af != AF_INET || !net_eq(svc->net, net))
continue;
@@ -2578,11 +2517,13 @@ __ip_vs_get_dest_entries(struct net *net, const struct ip_vs_get_dests *get,
union nf_inet_addr addr = { .ip = get->addr };
int ret = 0;
+ rcu_read_lock();
if (get->fwmark)
svc = __ip_vs_svc_fwm_find(net, AF_INET, get->fwmark);
else
svc = __ip_vs_service_find(net, AF_INET, get->protocol, &addr,
get->port);
+ rcu_read_unlock();
if (svc) {
int count = 0;
@@ -2765,12 +2706,14 @@ do_ip_vs_get_ctl(struct sock *sk, int cmd, void __user *user, int *len)
entry = (struct ip_vs_service_entry *)arg;
addr.ip = entry->addr;
+ rcu_read_lock();
if (entry->fwmark)
svc = __ip_vs_svc_fwm_find(net, AF_INET, entry->fwmark);
else
svc = __ip_vs_service_find(net, AF_INET,
entry->protocol, &addr,
entry->port);
+ rcu_read_unlock();
if (svc) {
ip_vs_copy_service(entry, svc);
if (copy_to_user(user, entry, sizeof(*entry)) != 0)
@@ -2927,6 +2870,7 @@ nla_put_failure:
static int ip_vs_genl_fill_service(struct sk_buff *skb,
struct ip_vs_service *svc)
{
+ struct ip_vs_scheduler *sched;
struct nlattr *nl_service;
struct ip_vs_flags flags = { .flags = svc->flags,
.mask = ~0 };
@@ -2947,7 +2891,8 @@ static int ip_vs_genl_fill_service(struct sk_buff *skb,
goto nla_put_failure;
}
- if (nla_put_string(skb, IPVS_SVC_ATTR_SCHED_NAME, svc->scheduler->name) ||
+ sched = rcu_dereference_protected(svc->scheduler, 1);
+ if (nla_put_string(skb, IPVS_SVC_ATTR_SCHED_NAME, sched->name) ||
(svc->pe &&
nla_put_string(skb, IPVS_SVC_ATTR_PE_NAME, svc->pe->name)) ||
nla_put(skb, IPVS_SVC_ATTR_FLAGS, sizeof(flags), &flags) ||
@@ -2998,7 +2943,7 @@ static int ip_vs_genl_dump_services(struct sk_buff *skb,
mutex_lock(&__ip_vs_mutex);
for (i = 0; i < IP_VS_SVC_TAB_SIZE; i++) {
- list_for_each_entry(svc, &ip_vs_svc_table[i], s_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_table[i], s_list) {
if (++idx <= start || !net_eq(svc->net, net))
continue;
if (ip_vs_genl_dump_service(skb, svc, cb) < 0) {
@@ -3009,7 +2954,7 @@ static int ip_vs_genl_dump_services(struct sk_buff *skb,
}
for (i = 0; i < IP_VS_SVC_TAB_SIZE; i++) {
- list_for_each_entry(svc, &ip_vs_svc_fwm_table[i], f_list) {
+ hlist_for_each_entry(svc, &ip_vs_svc_fwm_table[i], f_list) {
if (++idx <= start || !net_eq(svc->net, net))
continue;
if (ip_vs_genl_dump_service(skb, svc, cb) < 0) {
@@ -3069,11 +3014,13 @@ static int ip_vs_genl_parse_service(struct net *net,
usvc->fwmark = 0;
}
+ rcu_read_lock();
if (usvc->fwmark)
svc = __ip_vs_svc_fwm_find(net, usvc->af, usvc->fwmark);
else
svc = __ip_vs_service_find(net, usvc->af, usvc->protocol,
&usvc->addr, usvc->port);
+ rcu_read_unlock();
*ret_svc = svc;
/* If a full entry was requested, check for the additional fields */
@@ -3905,8 +3852,8 @@ int __init ip_vs_control_init(void)
/* Initialize svc_table, ip_vs_svc_fwm_table */
for (idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
- INIT_LIST_HEAD(&ip_vs_svc_table[idx]);
- INIT_LIST_HEAD(&ip_vs_svc_fwm_table[idx]);
+ INIT_HLIST_HEAD(&ip_vs_svc_table[idx]);
+ INIT_HLIST_HEAD(&ip_vs_svc_fwm_table[idx]);
}
smp_wmb(); /* Do we really need it now ? */
diff --git a/net/netfilter/ipvs/ip_vs_dh.c b/net/netfilter/ipvs/ip_vs_dh.c
index 89c2723..ccab120 100644
--- a/net/netfilter/ipvs/ip_vs_dh.c
+++ b/net/netfilter/ipvs/ip_vs_dh.c
@@ -269,6 +269,7 @@ static int __init ip_vs_dh_init(void)
static void __exit ip_vs_dh_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_dh_scheduler);
+ synchronize_rcu();
}
diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c
index ffef8a1..d8e5238 100644
--- a/net/netfilter/ipvs/ip_vs_lblc.c
+++ b/net/netfilter/ipvs/ip_vs_lblc.c
@@ -633,6 +633,7 @@ static void __exit ip_vs_lblc_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_lblc_scheduler);
unregister_pernet_subsys(&ip_vs_lblc_ops);
+ synchronize_rcu();
}
diff --git a/net/netfilter/ipvs/ip_vs_lblcr.c b/net/netfilter/ipvs/ip_vs_lblcr.c
index cdfe6a9..041b7cc 100644
--- a/net/netfilter/ipvs/ip_vs_lblcr.c
+++ b/net/netfilter/ipvs/ip_vs_lblcr.c
@@ -821,6 +821,7 @@ static void __exit ip_vs_lblcr_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_lblcr_scheduler);
unregister_pernet_subsys(&ip_vs_lblcr_ops);
+ synchronize_rcu();
}
diff --git a/net/netfilter/ipvs/ip_vs_lc.c b/net/netfilter/ipvs/ip_vs_lc.c
index 0cabf78..5128e33 100644
--- a/net/netfilter/ipvs/ip_vs_lc.c
+++ b/net/netfilter/ipvs/ip_vs_lc.c
@@ -84,6 +84,7 @@ static int __init ip_vs_lc_init(void)
static void __exit ip_vs_lc_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_lc_scheduler);
+ synchronize_rcu();
}
module_init(ip_vs_lc_init);
diff --git a/net/netfilter/ipvs/ip_vs_nq.c b/net/netfilter/ipvs/ip_vs_nq.c
index 51dc0cf..646cfd4 100644
--- a/net/netfilter/ipvs/ip_vs_nq.c
+++ b/net/netfilter/ipvs/ip_vs_nq.c
@@ -133,6 +133,7 @@ static int __init ip_vs_nq_init(void)
static void __exit ip_vs_nq_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_nq_scheduler);
+ synchronize_rcu();
}
module_init(ip_vs_nq_init);
diff --git a/net/netfilter/ipvs/ip_vs_pe.c b/net/netfilter/ipvs/ip_vs_pe.c
index 5d9774c..1a82b29 100644
--- a/net/netfilter/ipvs/ip_vs_pe.c
+++ b/net/netfilter/ipvs/ip_vs_pe.c
@@ -16,18 +16,6 @@ static LIST_HEAD(ip_vs_pe);
/* semaphore for IPVS PEs. */
static DEFINE_MUTEX(ip_vs_pe_mutex);
-/* Bind a service with a pe */
-void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe)
-{
- svc->pe = pe;
-}
-
-/* Unbind a service from its pe */
-void ip_vs_unbind_pe(struct ip_vs_service *svc)
-{
- svc->pe = NULL;
-}
-
/* Get pe in the pe list by name */
struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name)
{
diff --git a/net/netfilter/ipvs/ip_vs_proto_sctp.c b/net/netfilter/ipvs/ip_vs_proto_sctp.c
index f7190cd..4de5176 100644
--- a/net/netfilter/ipvs/ip_vs_proto_sctp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_sctp.c
@@ -27,9 +27,10 @@ sctp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
if (sch == NULL)
return 0;
net = skb_net(skb);
+ rcu_read_lock();
if ((sch->type == SCTP_CID_INIT) &&
- (svc = ip_vs_service_get(net, af, skb->mark, iph->protocol,
- &iph->daddr, sh->dest))) {
+ (svc = ip_vs_service_find(net, af, skb->mark, iph->protocol,
+ &iph->daddr, sh->dest))) {
int ignored;
if (ip_vs_todrop(net_ipvs(net))) {
@@ -37,7 +38,7 @@ sctp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
* It seems that we are very loaded.
* We have to drop this packet :(
*/
- ip_vs_service_put(svc);
+ rcu_read_unlock();
*verdict = NF_DROP;
return 0;
}
@@ -49,14 +50,13 @@ sctp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
if (!*cpp && ignored <= 0) {
if (!ignored)
*verdict = ip_vs_leave(svc, skb, pd, iph);
- else {
- ip_vs_service_put(svc);
+ else
*verdict = NF_DROP;
- }
+ rcu_read_unlock();
return 0;
}
- ip_vs_service_put(svc);
}
+ rcu_read_unlock();
/* NF_ACCEPT */
return 1;
}
diff --git a/net/netfilter/ipvs/ip_vs_proto_tcp.c b/net/netfilter/ipvs/ip_vs_proto_tcp.c
index 0bbc3fe..7de3342 100644
--- a/net/netfilter/ipvs/ip_vs_proto_tcp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_tcp.c
@@ -47,9 +47,10 @@ tcp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
}
net = skb_net(skb);
/* No !th->ack check to allow scheduling on SYN+ACK for Active FTP */
+ rcu_read_lock();
if (th->syn &&
- (svc = ip_vs_service_get(net, af, skb->mark, iph->protocol,
- &iph->daddr, th->dest))) {
+ (svc = ip_vs_service_find(net, af, skb->mark, iph->protocol,
+ &iph->daddr, th->dest))) {
int ignored;
if (ip_vs_todrop(net_ipvs(net))) {
@@ -57,7 +58,7 @@ tcp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
* It seems that we are very loaded.
* We have to drop this packet :(
*/
- ip_vs_service_put(svc);
+ rcu_read_unlock();
*verdict = NF_DROP;
return 0;
}
@@ -70,14 +71,13 @@ tcp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
if (!*cpp && ignored <= 0) {
if (!ignored)
*verdict = ip_vs_leave(svc, skb, pd, iph);
- else {
- ip_vs_service_put(svc);
+ else
*verdict = NF_DROP;
- }
+ rcu_read_unlock();
return 0;
}
- ip_vs_service_put(svc);
}
+ rcu_read_unlock();
/* NF_ACCEPT */
return 1;
}
diff --git a/net/netfilter/ipvs/ip_vs_proto_udp.c b/net/netfilter/ipvs/ip_vs_proto_udp.c
index 1a03e2d..b62a3c0 100644
--- a/net/netfilter/ipvs/ip_vs_proto_udp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_udp.c
@@ -44,8 +44,9 @@ udp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
return 0;
}
net = skb_net(skb);
- svc = ip_vs_service_get(net, af, skb->mark, iph->protocol,
- &iph->daddr, uh->dest);
+ rcu_read_lock();
+ svc = ip_vs_service_find(net, af, skb->mark, iph->protocol,
+ &iph->daddr, uh->dest);
if (svc) {
int ignored;
@@ -54,7 +55,7 @@ udp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
* It seems that we are very loaded.
* We have to drop this packet :(
*/
- ip_vs_service_put(svc);
+ rcu_read_unlock();
*verdict = NF_DROP;
return 0;
}
@@ -67,14 +68,13 @@ udp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
if (!*cpp && ignored <= 0) {
if (!ignored)
*verdict = ip_vs_leave(svc, skb, pd, iph);
- else {
- ip_vs_service_put(svc);
+ else
*verdict = NF_DROP;
- }
+ rcu_read_unlock();
return 0;
}
- ip_vs_service_put(svc);
}
+ rcu_read_unlock();
/* NF_ACCEPT */
return 1;
}
diff --git a/net/netfilter/ipvs/ip_vs_rr.c b/net/netfilter/ipvs/ip_vs_rr.c
index aa4601f..749c98a 100644
--- a/net/netfilter/ipvs/ip_vs_rr.c
+++ b/net/netfilter/ipvs/ip_vs_rr.c
@@ -121,6 +121,7 @@ static int __init ip_vs_rr_init(void)
static void __exit ip_vs_rr_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_rr_scheduler);
+ synchronize_rcu();
}
module_init(ip_vs_rr_init);
diff --git a/net/netfilter/ipvs/ip_vs_sched.c b/net/netfilter/ipvs/ip_vs_sched.c
index 1b715d0..4dbcda6 100644
--- a/net/netfilter/ipvs/ip_vs_sched.c
+++ b/net/netfilter/ipvs/ip_vs_sched.c
@@ -47,8 +47,6 @@ int ip_vs_bind_scheduler(struct ip_vs_service *svc,
{
int ret;
- svc->scheduler = scheduler;
-
if (scheduler->init_service) {
ret = scheduler->init_service(svc);
if (ret) {
@@ -56,7 +54,7 @@ int ip_vs_bind_scheduler(struct ip_vs_service *svc,
return ret;
}
}
-
+ rcu_assign_pointer(svc->scheduler, scheduler);
return 0;
}
@@ -64,17 +62,19 @@ int ip_vs_bind_scheduler(struct ip_vs_service *svc,
/*
* Unbind a service with its scheduler
*/
-void ip_vs_unbind_scheduler(struct ip_vs_service *svc)
+void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
+ struct ip_vs_scheduler *sched)
{
- struct ip_vs_scheduler *sched = svc->scheduler;
+ struct ip_vs_scheduler *cur_sched;
- if (!sched)
+ cur_sched = rcu_dereference_protected(svc->scheduler, 1);
+ /* This check proves that old 'sched' was installed */
+ if (!cur_sched)
return;
if (sched->done_service)
sched->done_service(svc);
-
- svc->scheduler = NULL;
+ /* svc->scheduler can not be set to NULL */
}
@@ -148,21 +148,21 @@ void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler)
void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg)
{
+ struct ip_vs_scheduler *sched;
+
+ sched = rcu_dereference(svc->scheduler);
if (svc->fwmark) {
IP_VS_ERR_RL("%s: FWM %u 0x%08X - %s\n",
- svc->scheduler->name, svc->fwmark,
- svc->fwmark, msg);
+ sched->name, svc->fwmark, svc->fwmark, msg);
#ifdef CONFIG_IP_VS_IPV6
} else if (svc->af == AF_INET6) {
IP_VS_ERR_RL("%s: %s [%pI6c]:%d - %s\n",
- svc->scheduler->name,
- ip_vs_proto_name(svc->protocol),
+ sched->name, ip_vs_proto_name(svc->protocol),
&svc->addr.in6, ntohs(svc->port), msg);
#endif
} else {
IP_VS_ERR_RL("%s: %s %pI4:%d - %s\n",
- svc->scheduler->name,
- ip_vs_proto_name(svc->protocol),
+ sched->name, ip_vs_proto_name(svc->protocol),
&svc->addr.ip, ntohs(svc->port), msg);
}
}
diff --git a/net/netfilter/ipvs/ip_vs_sed.c b/net/netfilter/ipvs/ip_vs_sed.c
index d011870..f320592 100644
--- a/net/netfilter/ipvs/ip_vs_sed.c
+++ b/net/netfilter/ipvs/ip_vs_sed.c
@@ -134,6 +134,7 @@ static int __init ip_vs_sed_init(void)
static void __exit ip_vs_sed_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_sed_scheduler);
+ synchronize_rcu();
}
module_init(ip_vs_sed_init);
diff --git a/net/netfilter/ipvs/ip_vs_sh.c b/net/netfilter/ipvs/ip_vs_sh.c
index 81c1a10..0df269d 100644
--- a/net/netfilter/ipvs/ip_vs_sh.c
+++ b/net/netfilter/ipvs/ip_vs_sh.c
@@ -283,6 +283,7 @@ static int __init ip_vs_sh_init(void)
static void __exit ip_vs_sh_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_sh_scheduler);
+ synchronize_rcu();
}
diff --git a/net/netfilter/ipvs/ip_vs_wlc.c b/net/netfilter/ipvs/ip_vs_wlc.c
index dafae88..c60a81c 100644
--- a/net/netfilter/ipvs/ip_vs_wlc.c
+++ b/net/netfilter/ipvs/ip_vs_wlc.c
@@ -106,6 +106,7 @@ static int __init ip_vs_wlc_init(void)
static void __exit ip_vs_wlc_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_wlc_scheduler);
+ synchronize_rcu();
}
module_init(ip_vs_wlc_init);
diff --git a/net/netfilter/ipvs/ip_vs_wrr.c b/net/netfilter/ipvs/ip_vs_wrr.c
index b173ef9..32c646e 100644
--- a/net/netfilter/ipvs/ip_vs_wrr.c
+++ b/net/netfilter/ipvs/ip_vs_wrr.c
@@ -261,6 +261,7 @@ static int __init ip_vs_wrr_init(void)
static void __exit ip_vs_wrr_cleanup(void)
{
unregister_ip_vs_scheduler(&ip_vs_wrr_scheduler);
+ synchronize_rcu();
}
module_init(ip_vs_wrr_init);
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* [PATCH 34/34] ipvs: do not disable bh for long time
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (32 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 33/34] ipvs: convert services " Simon Horman
@ 2013-03-29 4:11 ` Simon Horman
2013-04-01 22:41 ` [GIT PULL nf-next] IPVS optimisations for v3.10 Pablo Neira Ayuso
34 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-03-29 4:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
From: Julian Anastasov <ja@ssi.bg>
We used a global BH disable in LOCAL_OUT hook.
Add _bh suffix to all places that need it and remove
the disabling from LOCAL_OUT and sync code.
Functions like ip_defrag need protection from
BH, so add it. As for nf_nat_mangle_tcp_packet, it needs
RCU lock.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_app.c | 4 ++--
net/netfilter/ipvs/ip_vs_conn.c | 30 +++++++++++++-------------
net/netfilter/ipvs/ip_vs_core.c | 37 +++++++--------------------------
net/netfilter/ipvs/ip_vs_ftp.c | 2 ++
net/netfilter/ipvs/ip_vs_lblc.c | 4 ++--
net/netfilter/ipvs/ip_vs_lblcr.c | 12 +++++------
net/netfilter/ipvs/ip_vs_proto_sctp.c | 4 ++--
net/netfilter/ipvs/ip_vs_proto_tcp.c | 8 +++----
net/netfilter/ipvs/ip_vs_rr.c | 6 +++---
net/netfilter/ipvs/ip_vs_sync.c | 24 +++++++++------------
net/netfilter/ipvs/ip_vs_wrr.c | 4 ++--
net/netfilter/ipvs/ip_vs_xmit.c | 16 +++++++-------
12 files changed, 64 insertions(+), 87 deletions(-)
diff --git a/net/netfilter/ipvs/ip_vs_app.c b/net/netfilter/ipvs/ip_vs_app.c
index a956030..dfd7b65 100644
--- a/net/netfilter/ipvs/ip_vs_app.c
+++ b/net/netfilter/ipvs/ip_vs_app.c
@@ -352,14 +352,14 @@ static inline void vs_seq_update(struct ip_vs_conn *cp, struct ip_vs_seq *vseq,
unsigned int flag, __u32 seq, int diff)
{
/* spinlock is to keep updating cp->flags atomic */
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
if (!(cp->flags & flag) || after(seq, vseq->init_seq)) {
vseq->previous_delta = vseq->delta;
vseq->delta += diff;
vseq->init_seq = seq;
cp->flags |= flag;
}
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
}
static inline int app_tcp_pkt_out(struct ip_vs_conn *cp, struct sk_buff *skb,
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index 54de340..de64758 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -86,14 +86,14 @@ struct ip_vs_aligned_lock
static struct ip_vs_aligned_lock
__ip_vs_conntbl_lock_array[CT_LOCKARRAY_SIZE] __cacheline_aligned;
-static inline void ct_write_lock(unsigned int key)
+static inline void ct_write_lock_bh(unsigned int key)
{
- spin_lock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
+ spin_lock_bh(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
}
-static inline void ct_write_unlock(unsigned int key)
+static inline void ct_write_unlock_bh(unsigned int key)
{
- spin_unlock(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
+ spin_unlock_bh(&__ip_vs_conntbl_lock_array[key&CT_LOCKARRAY_MASK].l);
}
@@ -167,7 +167,7 @@ static inline int ip_vs_conn_hash(struct ip_vs_conn *cp)
/* Hash by protocol, client address and port */
hash = ip_vs_conn_hashkey_conn(cp);
- ct_write_lock(hash);
+ ct_write_lock_bh(hash);
spin_lock(&cp->lock);
if (!(cp->flags & IP_VS_CONN_F_HASHED)) {
@@ -182,7 +182,7 @@ static inline int ip_vs_conn_hash(struct ip_vs_conn *cp)
}
spin_unlock(&cp->lock);
- ct_write_unlock(hash);
+ ct_write_unlock_bh(hash);
return ret;
}
@@ -200,7 +200,7 @@ static inline int ip_vs_conn_unhash(struct ip_vs_conn *cp)
/* unhash it and decrease its reference counter */
hash = ip_vs_conn_hashkey_conn(cp);
- ct_write_lock(hash);
+ ct_write_lock_bh(hash);
spin_lock(&cp->lock);
if (cp->flags & IP_VS_CONN_F_HASHED) {
@@ -212,7 +212,7 @@ static inline int ip_vs_conn_unhash(struct ip_vs_conn *cp)
ret = 0;
spin_unlock(&cp->lock);
- ct_write_unlock(hash);
+ ct_write_unlock_bh(hash);
return ret;
}
@@ -227,7 +227,7 @@ static inline bool ip_vs_conn_unlink(struct ip_vs_conn *cp)
hash = ip_vs_conn_hashkey_conn(cp);
- ct_write_lock(hash);
+ ct_write_lock_bh(hash);
spin_lock(&cp->lock);
if (cp->flags & IP_VS_CONN_F_HASHED) {
@@ -242,7 +242,7 @@ static inline bool ip_vs_conn_unlink(struct ip_vs_conn *cp)
ret = atomic_read(&cp->refcnt) ? false : true;
spin_unlock(&cp->lock);
- ct_write_unlock(hash);
+ ct_write_unlock_bh(hash);
return ret;
}
@@ -462,13 +462,13 @@ void ip_vs_conn_put(struct ip_vs_conn *cp)
void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport)
{
if (ip_vs_conn_unhash(cp)) {
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
if (cp->flags & IP_VS_CONN_F_NO_CPORT) {
atomic_dec(&ip_vs_conn_no_cport_cnt);
cp->flags &= ~IP_VS_CONN_F_NO_CPORT;
cp->cport = cport;
}
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
/* hash on new dport */
ip_vs_conn_hash(cp);
@@ -622,9 +622,9 @@ void ip_vs_try_bind_dest(struct ip_vs_conn *cp)
if (dest) {
struct ip_vs_proto_data *pd;
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
if (cp->dest) {
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
rcu_read_unlock();
return;
}
@@ -635,7 +635,7 @@ void ip_vs_try_bind_dest(struct ip_vs_conn *cp)
ip_vs_unbind_app(cp);
ip_vs_bind_dest(cp, dest);
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
/* Update its packet transmitter */
cp->packet_xmit = NULL;
diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c
index 79df3c6..f26fe33 100644
--- a/net/netfilter/ipvs/ip_vs_core.c
+++ b/net/netfilter/ipvs/ip_vs_core.c
@@ -638,8 +638,11 @@ static inline enum ip_defrag_users ip_vs_defrag_user(unsigned int hooknum)
static inline int ip_vs_gather_frags(struct sk_buff *skb, u_int32_t user)
{
- int err = ip_defrag(skb, user);
+ int err;
+ local_bh_disable();
+ err = ip_defrag(skb, user);
+ local_bh_enable();
if (!err)
ip_send_check(ip_hdr(skb));
@@ -1217,13 +1220,7 @@ ip_vs_local_reply4(unsigned int hooknum, struct sk_buff *skb,
const struct net_device *in, const struct net_device *out,
int (*okfn)(struct sk_buff *))
{
- unsigned int verdict;
-
- /* Disable BH in LOCAL_OUT until all places are fixed */
- local_bh_disable();
- verdict = ip_vs_out(hooknum, skb, AF_INET);
- local_bh_enable();
- return verdict;
+ return ip_vs_out(hooknum, skb, AF_INET);
}
#ifdef CONFIG_IP_VS_IPV6
@@ -1250,13 +1247,7 @@ ip_vs_local_reply6(unsigned int hooknum, struct sk_buff *skb,
const struct net_device *in, const struct net_device *out,
int (*okfn)(struct sk_buff *))
{
- unsigned int verdict;
-
- /* Disable BH in LOCAL_OUT until all places are fixed */
- local_bh_disable();
- verdict = ip_vs_out(hooknum, skb, AF_INET6);
- local_bh_enable();
- return verdict;
+ return ip_vs_out(hooknum, skb, AF_INET6);
}
#endif
@@ -1714,13 +1705,7 @@ ip_vs_local_request4(unsigned int hooknum, struct sk_buff *skb,
const struct net_device *in, const struct net_device *out,
int (*okfn)(struct sk_buff *))
{
- unsigned int verdict;
-
- /* Disable BH in LOCAL_OUT until all places are fixed */
- local_bh_disable();
- verdict = ip_vs_in(hooknum, skb, AF_INET);
- local_bh_enable();
- return verdict;
+ return ip_vs_in(hooknum, skb, AF_INET);
}
#ifdef CONFIG_IP_VS_IPV6
@@ -1779,13 +1764,7 @@ ip_vs_local_request6(unsigned int hooknum, struct sk_buff *skb,
const struct net_device *in, const struct net_device *out,
int (*okfn)(struct sk_buff *))
{
- unsigned int verdict;
-
- /* Disable BH in LOCAL_OUT until all places are fixed */
- local_bh_disable();
- verdict = ip_vs_in(hooknum, skb, AF_INET6);
- local_bh_enable();
- return verdict;
+ return ip_vs_in(hooknum, skb, AF_INET6);
}
#endif
diff --git a/net/netfilter/ipvs/ip_vs_ftp.c b/net/netfilter/ipvs/ip_vs_ftp.c
index 7f90825..77c1732 100644
--- a/net/netfilter/ipvs/ip_vs_ftp.c
+++ b/net/netfilter/ipvs/ip_vs_ftp.c
@@ -267,10 +267,12 @@ static int ip_vs_ftp_out(struct ip_vs_app *app, struct ip_vs_conn *cp,
* hopefully it will succeed on the retransmitted
* packet.
*/
+ rcu_read_lock();
ret = nf_nat_mangle_tcp_packet(skb, ct, ctinfo,
iph->ihl * 4,
start-data, end-start,
buf, buf_len);
+ rcu_read_unlock();
if (ret) {
ip_vs_nfct_expect_related(skb, ct, n_cp,
IPPROTO_TCP, 0, 0);
diff --git a/net/netfilter/ipvs/ip_vs_lblc.c b/net/netfilter/ipvs/ip_vs_lblc.c
index d8e5238..b2cc252 100644
--- a/net/netfilter/ipvs/ip_vs_lblc.c
+++ b/net/netfilter/ipvs/ip_vs_lblc.c
@@ -527,10 +527,10 @@ ip_vs_lblc_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
}
/* If we fail to create a cache entry, we'll just use the valid dest */
- spin_lock(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
if (!tbl->dead)
ip_vs_lblc_new(tbl, &iph.daddr, dest);
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
out:
IP_VS_DBG_BUF(6, "LBLC: destination IP address %s --> server %s:%d\n",
diff --git a/net/netfilter/ipvs/ip_vs_lblcr.c b/net/netfilter/ipvs/ip_vs_lblcr.c
index 041b7cc..feb9656 100644
--- a/net/netfilter/ipvs/ip_vs_lblcr.c
+++ b/net/netfilter/ipvs/ip_vs_lblcr.c
@@ -678,7 +678,7 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
if (atomic_read(&en->set.size) > 1 &&
time_after(jiffies, en->set.lastmod +
sysctl_lblcr_expiration(svc))) {
- spin_lock(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
if (atomic_read(&en->set.size) > 1) {
struct ip_vs_dest *m;
@@ -686,7 +686,7 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
if (m)
ip_vs_dest_set_erase(&en->set, m);
}
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
}
/* If the destination is not overloaded, use it */
@@ -701,10 +701,10 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
}
/* Update our cache entry */
- spin_lock(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
if (!tbl->dead)
ip_vs_dest_set_insert(&en->set, dest, true);
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
goto out;
}
@@ -716,10 +716,10 @@ ip_vs_lblcr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
}
/* If we fail to create a cache entry, we'll just use the valid dest */
- spin_lock(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
if (!tbl->dead)
ip_vs_lblcr_new(tbl, &iph.daddr, dest);
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
out:
IP_VS_DBG_BUF(6, "LBLCR: destination IP address %s --> server %s:%d\n",
diff --git a/net/netfilter/ipvs/ip_vs_proto_sctp.c b/net/netfilter/ipvs/ip_vs_proto_sctp.c
index 4de5176..6e14a7b 100644
--- a/net/netfilter/ipvs/ip_vs_proto_sctp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_sctp.c
@@ -994,9 +994,9 @@ static void
sctp_state_transition(struct ip_vs_conn *cp, int direction,
const struct sk_buff *skb, struct ip_vs_proto_data *pd)
{
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
set_sctp_state(pd, cp, direction, skb);
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
}
static inline __u16 sctp_app_hashkey(__be16 port)
diff --git a/net/netfilter/ipvs/ip_vs_proto_tcp.c b/net/netfilter/ipvs/ip_vs_proto_tcp.c
index 7de3342..50a1594 100644
--- a/net/netfilter/ipvs/ip_vs_proto_tcp.c
+++ b/net/netfilter/ipvs/ip_vs_proto_tcp.c
@@ -557,9 +557,9 @@ tcp_state_transition(struct ip_vs_conn *cp, int direction,
if (th == NULL)
return;
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
set_tcp_state(pd, cp, direction, th);
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
}
static inline __u16 tcp_app_hashkey(__be16 port)
@@ -655,11 +655,11 @@ void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp)
{
struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_TCP);
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
cp->state = IP_VS_TCP_S_LISTEN;
cp->timeout = (pd ? pd->timeout_table[IP_VS_TCP_S_LISTEN]
: tcp_timeouts[IP_VS_TCP_S_LISTEN]);
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
}
/* ---------------------------------------------
diff --git a/net/netfilter/ipvs/ip_vs_rr.c b/net/netfilter/ipvs/ip_vs_rr.c
index 749c98a..c35986c 100644
--- a/net/netfilter/ipvs/ip_vs_rr.c
+++ b/net/netfilter/ipvs/ip_vs_rr.c
@@ -63,7 +63,7 @@ ip_vs_rr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
- spin_lock(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
p = (struct list_head *) svc->sched_data;
last = dest = list_entry(p, struct ip_vs_dest, n_list);
@@ -85,13 +85,13 @@ ip_vs_rr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
} while (pass < 2 && p != &svc->destinations);
stop:
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
ip_vs_scheduler_err(svc, "no destination available");
return NULL;
out:
svc->sched_data = &dest->n_list;
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
IP_VS_DBG_BUF(6, "RR: server %s:%u "
"activeconns %d refcnt %d weight %d\n",
IP_VS_DBG_ADDR(svc->af, &dest->addr), ntohs(dest->port),
diff --git a/net/netfilter/ipvs/ip_vs_sync.c b/net/netfilter/ipvs/ip_vs_sync.c
index 9724174..8e57077 100644
--- a/net/netfilter/ipvs/ip_vs_sync.c
+++ b/net/netfilter/ipvs/ip_vs_sync.c
@@ -531,9 +531,9 @@ static void ip_vs_sync_conn_v0(struct net *net, struct ip_vs_conn *cp,
if (!ip_vs_sync_conn_needed(ipvs, cp, pkts))
return;
- spin_lock(&ipvs->sync_buff_lock);
+ spin_lock_bh(&ipvs->sync_buff_lock);
if (!(ipvs->sync_state & IP_VS_STATE_MASTER)) {
- spin_unlock(&ipvs->sync_buff_lock);
+ spin_unlock_bh(&ipvs->sync_buff_lock);
return;
}
@@ -552,7 +552,7 @@ static void ip_vs_sync_conn_v0(struct net *net, struct ip_vs_conn *cp,
if (!buff) {
buff = ip_vs_sync_buff_create_v0(ipvs);
if (!buff) {
- spin_unlock(&ipvs->sync_buff_lock);
+ spin_unlock_bh(&ipvs->sync_buff_lock);
pr_err("ip_vs_sync_buff_create failed.\n");
return;
}
@@ -590,7 +590,7 @@ static void ip_vs_sync_conn_v0(struct net *net, struct ip_vs_conn *cp,
sb_queue_tail(ipvs, ms);
ms->sync_buff = NULL;
}
- spin_unlock(&ipvs->sync_buff_lock);
+ spin_unlock_bh(&ipvs->sync_buff_lock);
/* synchronize its controller if it has */
cp = cp->control;
@@ -641,9 +641,9 @@ sloop:
pe_name_len = strnlen(cp->pe->name, IP_VS_PENAME_MAXLEN);
}
- spin_lock(&ipvs->sync_buff_lock);
+ spin_lock_bh(&ipvs->sync_buff_lock);
if (!(ipvs->sync_state & IP_VS_STATE_MASTER)) {
- spin_unlock(&ipvs->sync_buff_lock);
+ spin_unlock_bh(&ipvs->sync_buff_lock);
return;
}
@@ -683,7 +683,7 @@ sloop:
if (!buff) {
buff = ip_vs_sync_buff_create(ipvs);
if (!buff) {
- spin_unlock(&ipvs->sync_buff_lock);
+ spin_unlock_bh(&ipvs->sync_buff_lock);
pr_err("ip_vs_sync_buff_create failed.\n");
return;
}
@@ -750,7 +750,7 @@ sloop:
}
}
- spin_unlock(&ipvs->sync_buff_lock);
+ spin_unlock_bh(&ipvs->sync_buff_lock);
control:
/* synchronize its controller if it has */
@@ -843,7 +843,7 @@ static void ip_vs_proc_conn(struct net *net, struct ip_vs_conn_param *param,
kfree(param->pe_data);
dest = cp->dest;
- spin_lock(&cp->lock);
+ spin_lock_bh(&cp->lock);
if ((cp->flags ^ flags) & IP_VS_CONN_F_INACTIVE &&
!(flags & IP_VS_CONN_F_TEMPLATE) && dest) {
if (flags & IP_VS_CONN_F_INACTIVE) {
@@ -857,7 +857,7 @@ static void ip_vs_proc_conn(struct net *net, struct ip_vs_conn_param *param,
flags &= IP_VS_CONN_F_BACKUP_UPD_MASK;
flags |= cp->flags & ~IP_VS_CONN_F_BACKUP_UPD_MASK;
cp->flags = flags;
- spin_unlock(&cp->lock);
+ spin_unlock_bh(&cp->lock);
if (!dest)
ip_vs_try_bind_dest(cp);
} else {
@@ -1689,11 +1689,7 @@ static int sync_thread_backup(void *data)
break;
}
- /* disable bottom half, because it accesses the data
- shared by softirq while getting/creating conns */
- local_bh_disable();
ip_vs_process_message(tinfo->net, tinfo->buf, len);
- local_bh_enable();
}
}
diff --git a/net/netfilter/ipvs/ip_vs_wrr.c b/net/netfilter/ipvs/ip_vs_wrr.c
index 32c646e..0e68555 100644
--- a/net/netfilter/ipvs/ip_vs_wrr.c
+++ b/net/netfilter/ipvs/ip_vs_wrr.c
@@ -170,7 +170,7 @@ ip_vs_wrr_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
IP_VS_DBG(6, "%s(): Scheduling...\n", __func__);
- spin_lock(&svc->sched_lock);
+ spin_lock_bh(&svc->sched_lock);
dest = mark->cl;
/* No available dests? */
if (mark->mw == 0)
@@ -222,7 +222,7 @@ found:
mark->cl = dest;
out:
- spin_unlock(&svc->sched_lock);
+ spin_unlock_bh(&svc->sched_lock);
return dest;
err_noavail:
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 3db7889..b75ff64 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -177,22 +177,22 @@ __ip_vs_get_out_rt(struct sk_buff *skb, struct ip_vs_dest *dest,
rt = (struct rtable *) dest_dst->dst_cache;
else {
dest_dst = ip_vs_dest_dst_alloc();
- spin_lock(&dest->dst_lock);
+ spin_lock_bh(&dest->dst_lock);
if (!dest_dst) {
__ip_vs_dst_set(dest, NULL, NULL, 0);
- spin_unlock(&dest->dst_lock);
+ spin_unlock_bh(&dest->dst_lock);
goto err_unreach;
}
rt = do_output_route4(net, dest->addr.ip, rt_mode,
&dest_dst->dst_saddr.ip);
if (!rt) {
__ip_vs_dst_set(dest, NULL, NULL, 0);
- spin_unlock(&dest->dst_lock);
+ spin_unlock_bh(&dest->dst_lock);
ip_vs_dest_dst_free(dest_dst);
goto err_unreach;
}
__ip_vs_dst_set(dest, dest_dst, &rt->dst, 0);
- spin_unlock(&dest->dst_lock);
+ spin_unlock_bh(&dest->dst_lock);
IP_VS_DBG(10, "new dst %pI4, src %pI4, refcnt=%d\n",
&dest->addr.ip, &dest_dst->dst_saddr.ip,
atomic_read(&rt->dst.__refcnt));
@@ -358,10 +358,10 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
u32 cookie;
dest_dst = ip_vs_dest_dst_alloc();
- spin_lock(&dest->dst_lock);
+ spin_lock_bh(&dest->dst_lock);
if (!dest_dst) {
__ip_vs_dst_set(dest, NULL, NULL, 0);
- spin_unlock(&dest->dst_lock);
+ spin_unlock_bh(&dest->dst_lock);
goto err_unreach;
}
dst = __ip_vs_route_output_v6(net, &dest->addr.in6,
@@ -369,14 +369,14 @@ __ip_vs_get_out_rt_v6(struct sk_buff *skb, struct ip_vs_dest *dest,
do_xfrm);
if (!dst) {
__ip_vs_dst_set(dest, NULL, NULL, 0);
- spin_unlock(&dest->dst_lock);
+ spin_unlock_bh(&dest->dst_lock);
ip_vs_dest_dst_free(dest_dst);
goto err_unreach;
}
rt = (struct rt6_info *) dst;
cookie = rt->rt6i_node ? rt->rt6i_node->fn_sernum : 0;
__ip_vs_dst_set(dest, dest_dst, &rt->dst, cookie);
- spin_unlock(&dest->dst_lock);
+ spin_unlock_bh(&dest->dst_lock);
IP_VS_DBG(10, "new dst %pI6, src %pI6, refcnt=%d\n",
&dest->addr.in6, &dest_dst->dst_saddr.in6,
atomic_read(&rt->dst.__refcnt));
--
1.7.10.4
^ permalink raw reply related [flat|nested] 42+ messages in thread
* Re: [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback
2013-03-29 4:11 ` [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback Simon Horman
@ 2013-03-29 14:44 ` Sergei Shtylyov
2013-03-29 21:47 ` Julian Anastasov
0 siblings, 1 reply; 42+ messages in thread
From: Sergei Shtylyov @ 2013-03-29 14:44 UTC (permalink / raw)
To: Simon Horman
Cc: Pablo Neira Ayuso, lvs-devel, netdev, netfilter-devel,
Wensong Zhang, Julian Anastasov
Hello.
On 29-03-2013 8:11, Simon Horman wrote:
> From: Julian Anastasov <ja@ssi.bg>
> After commit 70e7341673 (ipv4: Show that ip_send_reply()
> is purely unicast routine.) we do not need to reroute DNAT-ed
> traffic over loopback because reply uses iph daddr and not
> rt_spec_dst.
> Signed-off-by: Julian Anastasov <ja@ssi.bg>
> Signed-off by: Hans Schillstrom <hans@schillstrom.com>
> Signed-off-by: Simon Horman <horms@verge.net.au>
> ---
> net/netfilter/ipvs/ip_vs_xmit.c | 58 ++-------------------------------------
> 1 file changed, 2 insertions(+), 56 deletions(-)
> diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
> index 6448a2e..c942d36 100644
> --- a/net/netfilter/ipvs/ip_vs_xmit.c
> +++ b/net/netfilter/ipvs/ip_vs_xmit.c
[...]
> @@ -635,16 +597,8 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
> /* drop old route */
> skb_dst_drop(skb);
> skb_dst_set(skb, &rt->dst);
> - } else {
> + } else
{} should be kept after *else*, according to Documentation/CodingStyle,
chapter 3.
> ip_rt_put(rt);
> - /*
> - * Some IPv4 replies get local address from routes,
> - * not from iph, so while we DNAT after routing
> - * we need this second input/output route.
> - */
> - if (!__ip_vs_reroute_locally(skb))
> - goto tx_error;
> - }
>
> IP_VS_DBG_PKT(10, AF_INET, pp, skb, 0, "After DNAT");
>
> @@ -1269,16 +1223,8 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
> /* drop the old route when skb is not shared */
> skb_dst_drop(skb);
> skb_dst_set(skb, &rt->dst);
> - } else {
> + } else
Same here.
> ip_rt_put(rt);
> - /*
> - * Some IPv4 replies get local address from routes,
> - * not from iph, so while we DNAT after routing
> - * we need this second input/output route.
> - */
> - if (!__ip_vs_reroute_locally(skb))
> - goto tx_error;
> - }
>
> /* Another hack: avoid icmp_send in ip_fragment */
> skb->local_df = 1;
>
^ permalink raw reply [flat|nested] 42+ messages in thread
* Re: [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback
2013-03-29 14:44 ` Sergei Shtylyov
@ 2013-03-29 21:47 ` Julian Anastasov
0 siblings, 0 replies; 42+ messages in thread
From: Julian Anastasov @ 2013-03-29 21:47 UTC (permalink / raw)
To: Sergei Shtylyov
Cc: Simon Horman, Pablo Neira Ayuso, lvs-devel, netdev,
netfilter-devel, Wensong Zhang
Hello,
On Fri, 29 Mar 2013, Sergei Shtylyov wrote:
> On 29-03-2013 8:11, Simon Horman wrote:
>
> > From: Julian Anastasov <ja@ssi.bg>
>
> > After commit 70e7341673 (ipv4: Show that ip_send_reply()
> > is purely unicast routine.) we do not need to reroute DNAT-ed
> > traffic over loopback because reply uses iph daddr and not
> > rt_spec_dst.
>
> > Signed-off-by: Julian Anastasov <ja@ssi.bg>
> > Signed-off by: Hans Schillstrom <hans@schillstrom.com>
> > Signed-off-by: Simon Horman <horms@verge.net.au>
> > ---
> > net/netfilter/ipvs/ip_vs_xmit.c | 58
> > ++-------------------------------------
> > 1 file changed, 2 insertions(+), 56 deletions(-)
>
> > diff --git a/net/netfilter/ipvs/ip_vs_xmit.c
> > b/net/netfilter/ipvs/ip_vs_xmit.c
> > index 6448a2e..c942d36 100644
> > --- a/net/netfilter/ipvs/ip_vs_xmit.c
> > +++ b/net/netfilter/ipvs/ip_vs_xmit.c
> [...]
> > @@ -635,16 +597,8 @@ ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn
> > *cp,
> > /* drop old route */
> > skb_dst_drop(skb);
> > skb_dst_set(skb, &rt->dst);
> > - } else {
> > + } else
>
> {} should be kept after *else*, according to Documentation/CodingStyle,
> chapter 3.
Sigh. I missed this rule from Nov 2011.
It seems checkpatch.pl can not stop such crimes to occur :)
> > ip_rt_put(rt);
> > - /*
> > - * Some IPv4 replies get local address from routes,
> > - * not from iph, so while we DNAT after routing
> > - * we need this second input/output route.
> > - */
> > - if (!__ip_vs_reroute_locally(skb))
> > - goto tx_error;
> > - }
> >
> > IP_VS_DBG_PKT(10, AF_INET, pp, skb, 0, "After DNAT");
> >
> > @@ -1269,16 +1223,8 @@ ip_vs_icmp_xmit(struct sk_buff *skb, struct
> > ip_vs_conn *cp,
> > /* drop the old route when skb is not shared */
> > skb_dst_drop(skb);
> > skb_dst_set(skb, &rt->dst);
> > - } else {
> > + } else
>
> Same here.
>
> > ip_rt_put(rt);
> > - /*
> > - * Some IPv4 replies get local address from routes,
> > - * not from iph, so while we DNAT after routing
> > - * we need this second input/output route.
> > - */
> > - if (!__ip_vs_reroute_locally(skb))
> > - goto tx_error;
> > - }
> >
> > /* Another hack: avoid icmp_send in ip_fragment */
> > skb->local_df = 1;
Regards
--
Julian Anastasov <ja@ssi.bg>
^ permalink raw reply [flat|nested] 42+ messages in thread
* Re: [PATCH 01/34] net: add skb_dst_set_noref_force
2013-03-29 4:11 ` [PATCH 01/34] net: add skb_dst_set_noref_force Simon Horman
@ 2013-04-01 12:06 ` Pablo Neira Ayuso
2013-04-01 16:57 ` David Miller
0 siblings, 1 reply; 42+ messages in thread
From: Pablo Neira Ayuso @ 2013-04-01 12:06 UTC (permalink / raw)
To: Simon Horman
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang, Julian Anastasov
Hi Simon,
On Fri, Mar 29, 2013 at 01:11:18PM +0900, Simon Horman wrote:
> From: Julian Anastasov <ja@ssi.bg>
>
> Rename skb_dst_set_noref to __skb_dst_set_noref and
> add force flag as suggested by David Miller. The new wrapper
> skb_dst_set_noref_force will force dst entries that are not
> cached to be attached as skb dst without taking reference
> as long as provided dst is reclaimed after RCU grace period.
>
> Signed-off-by: Julian Anastasov <ja@ssi.bg>
> Signed-off by: Hans Schillstrom <hans@schillstrom.com>
> Signed-off-by: Simon Horman <horms@verge.net.au>
I think was acked by David, right?
> ---
> include/linux/skbuff.h | 35 ++++++++++++++++++++++++++++++++++-
> net/core/dst.c | 9 +++++----
> 2 files changed, 39 insertions(+), 5 deletions(-)
>
> diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
> index 878e0ee..364e244 100644
> --- a/include/linux/skbuff.h
> +++ b/include/linux/skbuff.h
> @@ -575,7 +575,40 @@ static inline void skb_dst_set(struct sk_buff *skb, struct dst_entry *dst)
> skb->_skb_refdst = (unsigned long)dst;
> }
>
> -extern void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst);
> +extern void __skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst,
> + bool force);
> +
> +/**
> + * skb_dst_set_noref - sets skb dst, hopefully, without taking reference
> + * @skb: buffer
> + * @dst: dst entry
> + *
> + * Sets skb dst, assuming a reference was not taken on dst.
> + * If dst entry is cached, we do not take reference and dst_release
> + * will be avoided by refdst_drop. If dst entry is not cached, we take
> + * reference, so that last dst_release can destroy the dst immediately.
> + */
> +static inline void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst)
> +{
> + __skb_dst_set_noref(skb, dst, false);
> +}
> +
> +/**
> + * skb_dst_set_noref_force - sets skb dst, without taking reference
> + * @skb: buffer
> + * @dst: dst entry
> + *
> + * Sets skb dst, assuming a reference was not taken on dst.
> + * No reference is taken and no dst_release will be called. While for
> + * cached dsts deferred reclaim is a basic feature, for entries that are
> + * not cached it is caller's job to guarantee that last dst_release for
> + * provided dst happens when nobody uses it, eg. after a RCU grace period.
> + */
> +static inline void skb_dst_set_noref_force(struct sk_buff *skb,
> + struct dst_entry *dst)
> +{
> + __skb_dst_set_noref(skb, dst, true);
> +}
>
> /**
> * skb_dst_is_noref - Test if skb dst isn't refcounted
> diff --git a/net/core/dst.c b/net/core/dst.c
> index 35fd12f..df9cc81 100644
> --- a/net/core/dst.c
> +++ b/net/core/dst.c
> @@ -320,27 +320,28 @@ void __dst_destroy_metrics_generic(struct dst_entry *dst, unsigned long old)
> EXPORT_SYMBOL(__dst_destroy_metrics_generic);
>
> /**
> - * skb_dst_set_noref - sets skb dst, without a reference
> + * __skb_dst_set_noref - sets skb dst, without a reference
> * @skb: buffer
> * @dst: dst entry
> + * @force: if force is set, use noref version even for DST_NOCACHE entries
> *
> * Sets skb dst, assuming a reference was not taken on dst
> * skb_dst_drop() should not dst_release() this dst
> */
> -void skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst)
> +void __skb_dst_set_noref(struct sk_buff *skb, struct dst_entry *dst, bool force)
> {
> WARN_ON(!rcu_read_lock_held() && !rcu_read_lock_bh_held());
> /* If dst not in cache, we must take a reference, because
> * dst_release() will destroy dst as soon as its refcount becomes zero
> */
> - if (unlikely(dst->flags & DST_NOCACHE)) {
> + if (unlikely((dst->flags & DST_NOCACHE) && !force)) {
> dst_hold(dst);
> skb_dst_set(skb, dst);
> } else {
> skb->_skb_refdst = (unsigned long)dst | SKB_DST_NOREF;
> }
> }
> -EXPORT_SYMBOL(skb_dst_set_noref);
> +EXPORT_SYMBOL(__skb_dst_set_noref);
>
> /* Dirty hack. We did it in 2.2 (in __dst_free),
> * we have _very_ good reasons not to repeat
> --
> 1.7.10.4
>
^ permalink raw reply [flat|nested] 42+ messages in thread
* Re: [PATCH 01/34] net: add skb_dst_set_noref_force
2013-04-01 12:06 ` Pablo Neira Ayuso
@ 2013-04-01 16:57 ` David Miller
2013-04-01 22:42 ` Pablo Neira Ayuso
0 siblings, 1 reply; 42+ messages in thread
From: David Miller @ 2013-04-01 16:57 UTC (permalink / raw)
To: pablo; +Cc: horms, lvs-devel, netdev, netfilter-devel, wensong, ja
From: Pablo Neira Ayuso <pablo@netfilter.org>
Date: Mon, 1 Apr 2013 14:06:44 +0200
> Hi Simon,
>
> On Fri, Mar 29, 2013 at 01:11:18PM +0900, Simon Horman wrote:
>> From: Julian Anastasov <ja@ssi.bg>
>>
>> Rename skb_dst_set_noref to __skb_dst_set_noref and
>> add force flag as suggested by David Miller. The new wrapper
>> skb_dst_set_noref_force will force dst entries that are not
>> cached to be attached as skb dst without taking reference
>> as long as provided dst is reclaimed after RCU grace period.
>>
>> Signed-off-by: Julian Anastasov <ja@ssi.bg>
>> Signed-off by: Hans Schillstrom <hans@schillstrom.com>
>> Signed-off-by: Simon Horman <horms@verge.net.au>
>
> I think was acked by David, right?
It was:
Acked-by: David S. Miller <davem@davemloft.net>
^ permalink raw reply [flat|nested] 42+ messages in thread
* Re: [GIT PULL nf-next] IPVS optimisations for v3.10
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
` (33 preceding siblings ...)
2013-03-29 4:11 ` [PATCH 34/34] ipvs: do not disable bh for long time Simon Horman
@ 2013-04-01 22:41 ` Pablo Neira Ayuso
34 siblings, 0 replies; 42+ messages in thread
From: Pablo Neira Ayuso @ 2013-04-01 22:41 UTC (permalink / raw)
To: Simon Horman
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang, Julian Anastasov
On Fri, Mar 29, 2013 at 01:11:17PM +0900, Simon Horman wrote:
> Hi Pablo,
>
> The following changes since commit e5c5d22e8dcf7c2d430336cbf8e180bd38e8daf1:
>
> net: add ETH_P_802_3_MIN (2013-03-28 01:20:42 -0400)
>
> are available in the git repository at:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/horms/ipvs-next.git tags/ipvs2-for-v3.10
>
> for you to fetch changes up to e0566a7cef2f1afca4aa98244224fb9ca6ded08e:
>
> ipvs: do not disable bh for long time (2013-03-29 12:59:02 +0900)
Pulled, thanks Simon.
^ permalink raw reply [flat|nested] 42+ messages in thread
* Re: [PATCH 01/34] net: add skb_dst_set_noref_force
2013-04-01 16:57 ` David Miller
@ 2013-04-01 22:42 ` Pablo Neira Ayuso
2013-04-02 1:11 ` Simon Horman
0 siblings, 1 reply; 42+ messages in thread
From: Pablo Neira Ayuso @ 2013-04-01 22:42 UTC (permalink / raw)
To: David Miller; +Cc: horms, lvs-devel, netdev, netfilter-devel, wensong, ja
On Mon, Apr 01, 2013 at 12:57:33PM -0400, David Miller wrote:
> From: Pablo Neira Ayuso <pablo@netfilter.org>
> Date: Mon, 1 Apr 2013 14:06:44 +0200
>
> > Hi Simon,
> >
> > On Fri, Mar 29, 2013 at 01:11:18PM +0900, Simon Horman wrote:
> >> From: Julian Anastasov <ja@ssi.bg>
> >>
> >> Rename skb_dst_set_noref to __skb_dst_set_noref and
> >> add force flag as suggested by David Miller. The new wrapper
> >> skb_dst_set_noref_force will force dst entries that are not
> >> cached to be attached as skb dst without taking reference
> >> as long as provided dst is reclaimed after RCU grace period.
> >>
> >> Signed-off-by: Julian Anastasov <ja@ssi.bg>
> >> Signed-off by: Hans Schillstrom <hans@schillstrom.com>
> >> Signed-off-by: Simon Horman <horms@verge.net.au>
> >
> > I think was acked by David, right?
>
> It was:
>
> Acked-by: David S. Miller <davem@davemloft.net>
I have added this missing Acked-by to the patch. Simon, you'll have to
rebase your tree, sorry.
^ permalink raw reply [flat|nested] 42+ messages in thread
* Re: [PATCH 01/34] net: add skb_dst_set_noref_force
2013-04-01 22:42 ` Pablo Neira Ayuso
@ 2013-04-02 1:11 ` Simon Horman
0 siblings, 0 replies; 42+ messages in thread
From: Simon Horman @ 2013-04-02 1:11 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: David Miller, lvs-devel, netdev, netfilter-devel, wensong, ja
On Tue, Apr 02, 2013 at 12:42:30AM +0200, Pablo Neira Ayuso wrote:
> On Mon, Apr 01, 2013 at 12:57:33PM -0400, David Miller wrote:
> > From: Pablo Neira Ayuso <pablo@netfilter.org>
> > Date: Mon, 1 Apr 2013 14:06:44 +0200
> >
> > > Hi Simon,
> > >
> > > On Fri, Mar 29, 2013 at 01:11:18PM +0900, Simon Horman wrote:
> > >> From: Julian Anastasov <ja@ssi.bg>
> > >>
> > >> Rename skb_dst_set_noref to __skb_dst_set_noref and
> > >> add force flag as suggested by David Miller. The new wrapper
> > >> skb_dst_set_noref_force will force dst entries that are not
> > >> cached to be attached as skb dst without taking reference
> > >> as long as provided dst is reclaimed after RCU grace period.
> > >>
> > >> Signed-off-by: Julian Anastasov <ja@ssi.bg>
> > >> Signed-off by: Hans Schillstrom <hans@schillstrom.com>
> > >> Signed-off-by: Simon Horman <horms@verge.net.au>
> > >
> > > I think was acked by David, right?
> >
> > It was:
> >
> > Acked-by: David S. Miller <davem@davemloft.net>
>
> I have added this missing Acked-by to the patch. Simon, you'll have to
> rebase your tree, sorry.
Thanks. Sorry for the mix up.
^ permalink raw reply [flat|nested] 42+ messages in thread
end of thread, other threads:[~2013-04-02 1:11 UTC | newest]
Thread overview: 42+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-03-29 4:11 [GIT PULL nf-next] IPVS optimisations for v3.10 Simon Horman
2013-03-29 4:11 ` [PATCH 01/34] net: add skb_dst_set_noref_force Simon Horman
2013-04-01 12:06 ` Pablo Neira Ayuso
2013-04-01 16:57 ` David Miller
2013-04-01 22:42 ` Pablo Neira Ayuso
2013-04-02 1:11 ` Simon Horman
2013-03-29 4:11 ` [PATCH 02/34] ipvs: avoid routing by TOS for real server Simon Horman
2013-03-29 4:11 ` [PATCH 03/34] ipvs: prefer NETDEV_DOWN event to free cached dsts Simon Horman
2013-03-29 4:11 ` [PATCH 04/34] ipvs: convert the IP_VS_XMIT macros to functions Simon Horman
2013-03-29 4:11 ` [PATCH 05/34] ipvs: rename functions related to dst_cache reset Simon Horman
2013-03-29 4:11 ` [PATCH 06/34] ipvs: no need to reroute anymore on DNAT over loopback Simon Horman
2013-03-29 14:44 ` Sergei Shtylyov
2013-03-29 21:47 ` Julian Anastasov
2013-03-29 4:11 ` [PATCH 07/34] ipvs: do not use skb_share_check Simon Horman
2013-03-29 4:11 ` [PATCH 08/34] ipvs: consolidate all dst checks on transmit in one place Simon Horman
2013-03-29 4:11 ` [PATCH 09/34] ipvs: optimize dst usage for real server Simon Horman
2013-03-29 4:11 ` [PATCH 10/34] ipvs: convert app locks Simon Horman
2013-03-29 4:11 ` [PATCH 11/34] ipvs: remove rs_lock by using RCU Simon Horman
2013-03-29 4:11 ` [PATCH 12/34] ipvs: convert locks used in persistence engines Simon Horman
2013-03-29 4:11 ` [PATCH 13/34] ipvs: convert connection locking Simon Horman
2013-03-29 4:11 ` [PATCH 14/34] ipvs: reorder keys in connection structure Simon Horman
2013-03-29 4:11 ` [PATCH 15/34] ipvs: avoid kmem_cache_zalloc in ip_vs_conn_new Simon Horman
2013-03-29 4:11 ` [PATCH 16/34] ipvs: change ip_vs_sched_lock to mutex Simon Horman
2013-03-29 4:11 ` [PATCH 17/34] ipvs: preparations for using rcu in schedulers Simon Horman
2013-03-29 4:11 ` [PATCH 18/34] ipvs: add ip_vs_dest_hold and ip_vs_dest_put Simon Horman
2013-03-29 4:11 ` [PATCH 19/34] ipvs: convert dh scheduler to rcu Simon Horman
2013-03-29 4:11 ` [PATCH 20/34] ipvs: convert lblc " Simon Horman
2013-03-29 4:11 ` [PATCH 21/34] ipvs: convert lblcr " Simon Horman
2013-03-29 4:11 ` [PATCH 22/34] ipvs: convert lc " Simon Horman
2013-03-29 4:11 ` [PATCH 23/34] ipvs: convert nq " Simon Horman
2013-03-29 4:11 ` [PATCH 24/34] ipvs: convert rr " Simon Horman
2013-03-29 4:11 ` [PATCH 25/34] ipvs: convert sed " Simon Horman
2013-03-29 4:11 ` [PATCH 26/34] ipvs: convert sh " Simon Horman
2013-03-29 4:11 ` [PATCH 27/34] ipvs: convert wlc " Simon Horman
2013-03-29 4:11 ` [PATCH 28/34] ipvs: convert wrr " Simon Horman
2013-03-29 4:11 ` [PATCH 29/34] ipvs: reorganize dest trash Simon Horman
2013-03-29 4:11 ` [PATCH 30/34] ipvs: do not expect result from done_service Simon Horman
2013-03-29 4:11 ` [PATCH 31/34] ipvs: convert sched_lock to spin lock Simon Horman
2013-03-29 4:11 ` [PATCH 32/34] ipvs: convert dests to rcu Simon Horman
2013-03-29 4:11 ` [PATCH 33/34] ipvs: convert services " Simon Horman
2013-03-29 4:11 ` [PATCH 34/34] ipvs: do not disable bh for long time Simon Horman
2013-04-01 22:41 ` [GIT PULL nf-next] IPVS optimisations for v3.10 Pablo Neira Ayuso
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.