* [PATCH net-next 1/9] inet: frag: constify match, hashfn and constructor arguments
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 2/9] inet: frag: remove hash size assumptions from callers Nikolay Aleksandrov
` (9 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
---
include/net/inet_frag.h | 11 ++++++-----
include/net/ipv6.h | 4 ++--
net/ieee802154/reassembly.c | 14 +++++++-------
net/ipv4/ip_fragment.c | 14 +++++++-------
net/ipv6/netfilter/nf_conntrack_reasm.c | 2 +-
net/ipv6/reassembly.c | 14 +++++++-------
6 files changed, 30 insertions(+), 29 deletions(-)
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 6f59de98dabd..15033057d44e 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -71,10 +71,11 @@ struct inet_frags {
u32 rnd;
int qsize;
- unsigned int (*hashfn)(struct inet_frag_queue *);
- bool (*match)(struct inet_frag_queue *q, void *arg);
+ unsigned int (*hashfn)(const struct inet_frag_queue *);
+ bool (*match)(const struct inet_frag_queue *q,
+ const void *arg);
void (*constructor)(struct inet_frag_queue *q,
- void *arg);
+ const void *arg);
void (*destructor)(struct inet_frag_queue *);
void (*skb_free)(struct sk_buff *);
void (*frag_expire)(unsigned long data);
@@ -131,9 +132,9 @@ static inline void init_frag_mem_limit(struct netns_frags *nf)
percpu_counter_init(&nf->mem, 0);
}
-static inline int sum_frag_mem_limit(struct netns_frags *nf)
+static inline unsigned int sum_frag_mem_limit(struct netns_frags *nf)
{
- int res;
+ unsigned int res;
local_bh_disable();
res = percpu_counter_sum_positive(&nf->mem);
diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index a25017247457..25c2170e1298 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -496,8 +496,8 @@ struct ip6_create_arg {
u8 ecn;
};
-void ip6_frag_init(struct inet_frag_queue *q, void *a);
-bool ip6_frag_match(struct inet_frag_queue *q, void *a);
+void ip6_frag_init(struct inet_frag_queue *q, const void *a);
+bool ip6_frag_match(const struct inet_frag_queue *q, const void *a);
/*
* Equivalent of ipv4 struct ip
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index b85bd3f7048e..a5588be18024 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -61,18 +61,18 @@ static unsigned int lowpan_hash_frag(__be16 tag, u16 d_size,
return c & (INETFRAGS_HASHSZ - 1);
}
-static unsigned int lowpan_hashfn(struct inet_frag_queue *q)
+static unsigned int lowpan_hashfn(const struct inet_frag_queue *q)
{
- struct lowpan_frag_queue *fq;
+ const struct lowpan_frag_queue *fq;
fq = container_of(q, struct lowpan_frag_queue, q);
return lowpan_hash_frag(fq->tag, fq->d_size, &fq->saddr, &fq->daddr);
}
-static bool lowpan_frag_match(struct inet_frag_queue *q, void *a)
+static bool lowpan_frag_match(const struct inet_frag_queue *q, const void *a)
{
- struct lowpan_frag_queue *fq;
- struct lowpan_create_arg *arg = a;
+ const struct lowpan_frag_queue *fq;
+ const struct lowpan_create_arg *arg = a;
fq = container_of(q, struct lowpan_frag_queue, q);
return fq->tag == arg->tag && fq->d_size == arg->d_size &&
@@ -80,10 +80,10 @@ static bool lowpan_frag_match(struct inet_frag_queue *q, void *a)
ieee802154_addr_equal(&fq->daddr, arg->dst);
}
-static void lowpan_frag_init(struct inet_frag_queue *q, void *a)
+static void lowpan_frag_init(struct inet_frag_queue *q, const void *a)
{
+ const struct lowpan_create_arg *arg = a;
struct lowpan_frag_queue *fq;
- struct lowpan_create_arg *arg = a;
fq = container_of(q, struct lowpan_frag_queue, q);
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index ed32313e307c..586e6aaf9e6e 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -112,18 +112,18 @@ static unsigned int ipqhashfn(__be16 id, __be32 saddr, __be32 daddr, u8 prot)
ip4_frags.rnd) & (INETFRAGS_HASHSZ - 1);
}
-static unsigned int ip4_hashfn(struct inet_frag_queue *q)
+static unsigned int ip4_hashfn(const struct inet_frag_queue *q)
{
- struct ipq *ipq;
+ const struct ipq *ipq;
ipq = container_of(q, struct ipq, q);
return ipqhashfn(ipq->id, ipq->saddr, ipq->daddr, ipq->protocol);
}
-static bool ip4_frag_match(struct inet_frag_queue *q, void *a)
+static bool ip4_frag_match(const struct inet_frag_queue *q, const void *a)
{
- struct ipq *qp;
- struct ip4_create_arg *arg = a;
+ const struct ipq *qp;
+ const struct ip4_create_arg *arg = a;
qp = container_of(q, struct ipq, q);
return qp->id == arg->iph->id &&
@@ -133,14 +133,14 @@ static bool ip4_frag_match(struct inet_frag_queue *q, void *a)
qp->user == arg->user;
}
-static void ip4_frag_init(struct inet_frag_queue *q, void *a)
+static void ip4_frag_init(struct inet_frag_queue *q, const void *a)
{
struct ipq *qp = container_of(q, struct ipq, q);
struct netns_ipv4 *ipv4 = container_of(q->net, struct netns_ipv4,
frags);
struct net *net = container_of(ipv4, struct net, ipv4);
- struct ip4_create_arg *arg = a;
+ const struct ip4_create_arg *arg = a;
qp->protocol = arg->iph->protocol;
qp->id = arg->iph->id;
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 0d5279fd852a..c2c3f2116bc5 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -156,7 +156,7 @@ static unsigned int nf_hash_frag(__be32 id, const struct in6_addr *saddr,
}
-static unsigned int nf_hashfn(struct inet_frag_queue *q)
+static unsigned int nf_hashfn(const struct inet_frag_queue *q)
{
const struct frag_queue *nq;
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index cc85a9ba5010..0c6932cc08cb 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -94,18 +94,18 @@ static unsigned int inet6_hash_frag(__be32 id, const struct in6_addr *saddr,
return c & (INETFRAGS_HASHSZ - 1);
}
-static unsigned int ip6_hashfn(struct inet_frag_queue *q)
+static unsigned int ip6_hashfn(const struct inet_frag_queue *q)
{
- struct frag_queue *fq;
+ const struct frag_queue *fq;
fq = container_of(q, struct frag_queue, q);
return inet6_hash_frag(fq->id, &fq->saddr, &fq->daddr);
}
-bool ip6_frag_match(struct inet_frag_queue *q, void *a)
+bool ip6_frag_match(const struct inet_frag_queue *q, const void *a)
{
- struct frag_queue *fq;
- struct ip6_create_arg *arg = a;
+ const struct frag_queue *fq;
+ const struct ip6_create_arg *arg = a;
fq = container_of(q, struct frag_queue, q);
return fq->id == arg->id &&
@@ -115,10 +115,10 @@ bool ip6_frag_match(struct inet_frag_queue *q, void *a)
}
EXPORT_SYMBOL(ip6_frag_match);
-void ip6_frag_init(struct inet_frag_queue *q, void *a)
+void ip6_frag_init(struct inet_frag_queue *q, const void *a)
{
struct frag_queue *fq = container_of(q, struct frag_queue, q);
- struct ip6_create_arg *arg = a;
+ const struct ip6_create_arg *arg = a;
fq->id = arg->id;
fq->user = arg->user;
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 2/9] inet: frag: remove hash size assumptions from callers
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 1/9] inet: frag: constify match, hashfn and constructor arguments Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 3/9] inet: frag: move evictor calls into frag_find function Nikolay Aleksandrov
` (8 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
hide actual hash size from individual users: The _find
function will now fold the given hash value into the required range.
Signed-off-by: Florian Westphal <fw@strlen.de>
---
net/ieee802154/reassembly.c | 12 ++++--------
net/ipv4/inet_fragment.c | 13 ++++++++++---
net/ipv4/ip_fragment.c | 2 +-
net/ipv6/netfilter/nf_conntrack_reasm.c | 7 ++-----
net/ipv6/reassembly.c | 8 ++------
5 files changed, 19 insertions(+), 23 deletions(-)
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index a5588be18024..a707995fd4d7 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -50,15 +50,11 @@ static unsigned int lowpan_hash_frag(__be16 tag, u16 d_size,
const struct ieee802154_addr *saddr,
const struct ieee802154_addr *daddr)
{
- u32 c;
-
net_get_random_once(&lowpan_frags.rnd, sizeof(lowpan_frags.rnd));
- c = jhash_3words(ieee802154_addr_hash(saddr),
- ieee802154_addr_hash(daddr),
- (__force u32)(tag + (d_size << 16)),
- lowpan_frags.rnd);
-
- return c & (INETFRAGS_HASHSZ - 1);
+ return jhash_3words(ieee802154_addr_hash(saddr),
+ ieee802154_addr_hash(daddr),
+ (__force u32)(tag + (d_size << 16)),
+ lowpan_frags.rnd);
}
static unsigned int lowpan_hashfn(const struct inet_frag_queue *q)
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 3b01959bf4bb..930d23870811 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -46,6 +46,12 @@ const u8 ip_frag_ecn_table[16] = {
};
EXPORT_SYMBOL(ip_frag_ecn_table);
+static unsigned int
+inet_frag_hashfn(const struct inet_frags *f, const struct inet_frag_queue *q)
+{
+ return f->hashfn(q) & (INETFRAGS_HASHSZ - 1);
+}
+
static void inet_frag_secret_rebuild(unsigned long dummy)
{
struct inet_frags *f = (struct inet_frags *)dummy;
@@ -63,7 +69,7 @@ static void inet_frag_secret_rebuild(unsigned long dummy)
hb = &f->hash[i];
hlist_for_each_entry_safe(q, n, &hb->chain, list) {
- unsigned int hval = f->hashfn(q);
+ unsigned int hval = inet_frag_hashfn(f, q);
if (hval != i) {
struct inet_frag_bucket *hb_dest;
@@ -133,7 +139,7 @@ static inline void fq_unlink(struct inet_frag_queue *fq, struct inet_frags *f)
unsigned int hash;
read_lock(&f->lock);
- hash = f->hashfn(fq);
+ hash = inet_frag_hashfn(f, fq);
hb = &f->hash[hash];
spin_lock(&hb->chain_lock);
@@ -252,7 +258,7 @@ static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
* the rnd seed, so we need to re-calculate the hash
* chain. Fortunatelly the qp_in can be used to get one.
*/
- hash = f->hashfn(qp_in);
+ hash = inet_frag_hashfn(f, qp_in);
hb = &f->hash[hash];
spin_lock(&hb->chain_lock);
@@ -326,6 +332,7 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
struct inet_frag_queue *q;
int depth = 0;
+ hash &= (INETFRAGS_HASHSZ - 1);
hb = &f->hash[hash];
spin_lock(&hb->chain_lock);
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 586e6aaf9e6e..b769eb6c83c0 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -109,7 +109,7 @@ static unsigned int ipqhashfn(__be16 id, __be32 saddr, __be32 daddr, u8 prot)
net_get_random_once(&ip4_frags.rnd, sizeof(ip4_frags.rnd));
return jhash_3words((__force u32)id << 16 | prot,
(__force u32)saddr, (__force u32)daddr,
- ip4_frags.rnd) & (INETFRAGS_HASHSZ - 1);
+ ip4_frags.rnd);
}
static unsigned int ip4_hashfn(const struct inet_frag_queue *q)
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index c2c3f2116bc5..607e4a94ef41 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -147,12 +147,9 @@ static inline u8 ip6_frag_ecn(const struct ipv6hdr *ipv6h)
static unsigned int nf_hash_frag(__be32 id, const struct in6_addr *saddr,
const struct in6_addr *daddr)
{
- u32 c;
-
net_get_random_once(&nf_frags.rnd, sizeof(nf_frags.rnd));
- c = jhash_3words(ipv6_addr_hash(saddr), ipv6_addr_hash(daddr),
- (__force u32)id, nf_frags.rnd);
- return c & (INETFRAGS_HASHSZ - 1);
+ return jhash_3words(ipv6_addr_hash(saddr), ipv6_addr_hash(daddr),
+ (__force u32)id, nf_frags.rnd);
}
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 0c6932cc08cb..2b76549a1016 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -85,13 +85,9 @@ static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
static unsigned int inet6_hash_frag(__be32 id, const struct in6_addr *saddr,
const struct in6_addr *daddr)
{
- u32 c;
-
net_get_random_once(&ip6_frags.rnd, sizeof(ip6_frags.rnd));
- c = jhash_3words(ipv6_addr_hash(saddr), ipv6_addr_hash(daddr),
- (__force u32)id, ip6_frags.rnd);
-
- return c & (INETFRAGS_HASHSZ - 1);
+ return jhash_3words(ipv6_addr_hash(saddr), ipv6_addr_hash(daddr),
+ (__force u32)id, ip6_frags.rnd);
}
static unsigned int ip6_hashfn(const struct inet_frag_queue *q)
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 3/9] inet: frag: move evictor calls into frag_find function
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 1/9] inet: frag: constify match, hashfn and constructor arguments Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 2/9] inet: frag: remove hash size assumptions from callers Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 4/9] inet: frag: move eviction of queues to work queue Nikolay Aleksandrov
` (7 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
First step to move eviction handling into a work queue.
We lose two spots that accounted evicted fragments in MIB counters.
Accounting will be restored since the upcoming work-queue evictor
invokes the frag queue timer callbacks instead.
Signed-off-by: Florian Westphal <fw@strlen.de>
---
include/net/inet_frag.h | 1 -
net/ieee802154/reassembly.c | 2 --
net/ipv4/inet_fragment.c | 16 +++++++++-------
net/ipv4/ip_fragment.c | 15 ---------------
net/ipv6/netfilter/nf_conntrack_reasm.c | 4 ----
net/ipv6/reassembly.c | 6 ------
6 files changed, 9 insertions(+), 35 deletions(-)
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 15033057d44e..9fe644d1a26e 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -90,7 +90,6 @@ void inet_frags_exit_net(struct netns_frags *nf, struct inet_frags *f);
void inet_frag_kill(struct inet_frag_queue *q, struct inet_frags *f);
void inet_frag_destroy(struct inet_frag_queue *q,
struct inet_frags *f, int *work);
-int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force);
struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
struct inet_frags *f, void *key, unsigned int hash)
__releases(&f->lock);
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index a707995fd4d7..9503a48556f7 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -369,8 +369,6 @@ int lowpan_frag_rcv(struct sk_buff *skb, const u8 frag_type)
if (frag_info->d_size > ieee802154_lowpan->max_dsize)
goto err;
- inet_frag_evictor(&ieee802154_lowpan->frags, &lowpan_frags, false);
-
fq = fq_find(net, frag_info, &source, &dest);
if (fq != NULL) {
int ret;
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 930d23870811..535636017534 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -46,6 +46,8 @@ const u8 ip_frag_ecn_table[16] = {
};
EXPORT_SYMBOL(ip_frag_ecn_table);
+static int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force);
+
static unsigned int
inet_frag_hashfn(const struct inet_frags *f, const struct inet_frag_queue *q)
{
@@ -203,16 +205,11 @@ void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f,
}
EXPORT_SYMBOL(inet_frag_destroy);
-int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force)
+static int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force)
{
struct inet_frag_queue *q;
int work, evicted = 0;
- if (!force) {
- if (frag_mem_limit(nf) <= nf->high_thresh)
- return 0;
- }
-
work = frag_mem_limit(nf) - nf->low_thresh;
while (work > 0 || force) {
spin_lock(&nf->lru_lock);
@@ -242,7 +239,6 @@ int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force)
return evicted;
}
-EXPORT_SYMBOL(inet_frag_evictor);
static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
struct inet_frag_queue *qp_in, struct inet_frags *f,
@@ -296,6 +292,9 @@ static struct inet_frag_queue *inet_frag_alloc(struct netns_frags *nf,
{
struct inet_frag_queue *q;
+ if (frag_mem_limit(nf) > nf->high_thresh)
+ return NULL;
+
q = kzalloc(f->qsize, GFP_ATOMIC);
if (q == NULL)
return NULL;
@@ -332,6 +331,9 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
struct inet_frag_queue *q;
int depth = 0;
+ if (frag_mem_limit(nf) > nf->high_thresh)
+ inet_frag_evictor(nf, f, false);
+
hash &= (INETFRAGS_HASHSZ - 1);
hb = &f->hash[hash];
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index b769eb6c83c0..54988672d00d 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -177,18 +177,6 @@ static void ipq_kill(struct ipq *ipq)
inet_frag_kill(&ipq->q, &ip4_frags);
}
-/* Memory limiting on fragments. Evictor trashes the oldest
- * fragment queue until we are back under the threshold.
- */
-static void ip_evictor(struct net *net)
-{
- int evicted;
-
- evicted = inet_frag_evictor(&net->ipv4.frags, &ip4_frags, false);
- if (evicted)
- IP_ADD_STATS_BH(net, IPSTATS_MIB_REASMFAILS, evicted);
-}
-
/*
* Oops, a fragment queue timed out. Kill it and send an ICMP reply.
*/
@@ -655,9 +643,6 @@ int ip_defrag(struct sk_buff *skb, u32 user)
net = skb->dev ? dev_net(skb->dev) : dev_net(skb_dst(skb)->dev);
IP_INC_STATS_BH(net, IPSTATS_MIB_REASMREQDS);
- /* Start by cleaning up the memory. */
- ip_evictor(net);
-
/* Lookup (or create) queue header */
if ((qp = ip_find(net, ip_hdr(skb), user)) != NULL) {
int ret;
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 607e4a94ef41..58e32cf91c95 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -594,10 +594,6 @@ struct sk_buff *nf_ct_frag6_gather(struct sk_buff *skb, u32 user)
hdr = ipv6_hdr(clone);
fhdr = (struct frag_hdr *)skb_transport_header(clone);
- local_bh_disable();
- inet_frag_evictor(&net->nf_frag.frags, &nf_frags, false);
- local_bh_enable();
-
fq = fq_find(net, fhdr->identification, user, &hdr->saddr, &hdr->daddr,
ip6_frag_ecn(hdr));
if (fq == NULL) {
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 2b76549a1016..97acbc490d9e 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -519,7 +519,6 @@ static int ipv6_frag_rcv(struct sk_buff *skb)
struct frag_queue *fq;
const struct ipv6hdr *hdr = ipv6_hdr(skb);
struct net *net = dev_net(skb_dst(skb)->dev);
- int evicted;
if (IP6CB(skb)->flags & IP6SKB_FRAGMENTED)
goto fail_hdr;
@@ -548,11 +547,6 @@ static int ipv6_frag_rcv(struct sk_buff *skb)
return 1;
}
- evicted = inet_frag_evictor(&net->ipv6.frags, &ip6_frags, false);
- if (evicted)
- IP6_ADD_STATS_BH(net, ip6_dst_idev(skb_dst(skb)),
- IPSTATS_MIB_REASMFAILS, evicted);
-
fq = fq_find(net, fhdr->identification, &hdr->saddr, &hdr->daddr,
ip6_frag_ecn(hdr));
if (fq != NULL) {
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 4/9] inet: frag: move eviction of queues to work queue
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (2 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 3/9] inet: frag: move evictor calls into frag_find function Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 5/9] inet: frag: don't account number of fragment queues Nikolay Aleksandrov
` (6 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
When the high_thresh limit is reached we try to toss the 'oldest'
incomplete fragment queues until memory limits are below the low_thresh
value. This happens in softirq/packet processing context.
This has two drawbacks:
1) processors might evict a queue that was about to be completed
by another cpu, because they will compete wrt. resource usage and
resource reclaim.
2) LRU list maintenance is expensive.
But when constantly overloaded, even the 'least recently used' element is
recent, so removing 'lru' queue first is not 'fairer' than removing any
other fragment queue.
This moves eviction out of the fast path:
When the low threshold is reached, a work queue is scheduled
which then iterates over the table and removes the queues that exceed
the memory limits of the namespace. It sets a new flag called
INET_FRAG_EVICTED on the evicted queues so the proper counters will get
incremented when the queue is forcefully expired.
When the high threshold is reached, no more fragment queues are
created until we're below the limit again.
The LRU list is now unused and will be removed in a followup patch.
Joint work with Nikolay Aleksandrov.
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
---
Documentation/networking/ip-sysctl.txt | 4 +-
include/net/inet_frag.h | 6 +-
net/ipv4/inet_fragment.c | 142 +++++++++++++++++++++++----------
net/ipv4/ip_fragment.c | 3 +-
net/ipv6/reassembly.c | 4 +-
5 files changed, 112 insertions(+), 47 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index f35bfe43bf7a..625c8dda4be7 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -104,7 +104,9 @@ ipfrag_high_thresh - INTEGER
is reached.
ipfrag_low_thresh - INTEGER
- See ipfrag_high_thresh
+ Maximum memory used to reassemble IP fragments before the kernel
+ begins to remove incomplete fragment queues to free up resources.
+ The kernel still accepts new fragments for defragmentation.
ipfrag_time - INTEGER
Time in seconds to keep an IP fragment in memory.
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 9fe644d1a26e..e975032ea11b 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -32,6 +32,7 @@ struct inet_frag_queue {
int meat;
__u8 last_in; /* first/last segment arrived? */
+#define INET_FRAG_EVICTED 8
#define INET_FRAG_COMPLETE 4
#define INET_FRAG_FIRST_IN 2
#define INET_FRAG_LAST_IN 1
@@ -48,7 +49,7 @@ struct inet_frag_queue {
* rounded up (SKB_TRUELEN(0) + sizeof(struct ipq or
* struct frag_queue))
*/
-#define INETFRAGS_MAXDEPTH 128
+#define INETFRAGS_MAXDEPTH 128
struct inet_frag_bucket {
struct hlist_head chain;
@@ -65,6 +66,9 @@ struct inet_frags {
int secret_interval;
struct timer_list secret_timer;
+ struct work_struct frags_work;
+ unsigned int next_bucket;
+
/* The first call to hashfn is responsible to initialize
* rnd. This is best done with net_get_random_once.
*/
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 535636017534..43315ecb9400 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -25,6 +25,9 @@
#include <net/inet_frag.h>
#include <net/inet_ecn.h>
+#define INETFRAGS_EVICT_BUCKETS 128
+#define INETFRAGS_EVICT_MAX 512
+
/* Given the OR values of all fragments, apply RFC 3168 5.3 requirements
* Value : 0xff if frame should be dropped.
* 0 or INET_ECN_CE value, to be ORed in to final iph->tos field
@@ -46,8 +49,6 @@ const u8 ip_frag_ecn_table[16] = {
};
EXPORT_SYMBOL(ip_frag_ecn_table);
-static int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force);
-
static unsigned int
inet_frag_hashfn(const struct inet_frags *f, const struct inet_frag_queue *q)
{
@@ -89,10 +90,92 @@ static void inet_frag_secret_rebuild(unsigned long dummy)
mod_timer(&f->secret_timer, now + f->secret_interval);
}
+static bool inet_fragq_should_evict(const struct inet_frag_queue *q)
+{
+ return q->net->low_thresh == 0 ||
+ frag_mem_limit(q->net) >= q->net->low_thresh;
+}
+
+static unsigned int
+inet_evict_bucket(struct inet_frags *f, struct inet_frag_bucket *hb)
+{
+ struct inet_frag_queue *fq;
+ struct hlist_node *n;
+ unsigned int evicted = 0;
+ HLIST_HEAD(expired);
+
+evict_again:
+ spin_lock(&hb->chain_lock);
+
+ hlist_for_each_entry_safe(fq, n, &hb->chain, list) {
+ if (!inet_fragq_should_evict(fq))
+ continue;
+
+ if (!del_timer(&fq->timer)) {
+ /* q expiring right now thus increment its refcount so
+ * it won't be freed under us and wait until the timer
+ * has finished executing then destroy it
+ */
+ atomic_inc(&fq->refcnt);
+ spin_unlock(&hb->chain_lock);
+ del_timer_sync(&fq->timer);
+ WARN_ON(atomic_read(&fq->refcnt) != 1);
+ inet_frag_put(fq, f);
+ goto evict_again;
+ }
+
+ /* suppress xmit of (icmp) error packet */
+ fq->last_in &= ~INET_FRAG_FIRST_IN;
+ fq->last_in |= INET_FRAG_EVICTED;
+ hlist_del(&fq->list);
+ hlist_add_head(&fq->list, &expired);
+ ++evicted;
+ }
+
+ spin_unlock(&hb->chain_lock);
+
+ hlist_for_each_entry_safe(fq, n, &expired, list)
+ f->frag_expire((unsigned long) fq);
+
+ return evicted;
+}
+
+static void inet_frag_worker(struct work_struct *work)
+{
+ unsigned int budget = INETFRAGS_EVICT_BUCKETS;
+ unsigned int i, evicted = 0;
+ struct inet_frags *f;
+
+ f = container_of(work, struct inet_frags, frags_work);
+
+ BUILD_BUG_ON(INETFRAGS_EVICT_BUCKETS >= INETFRAGS_HASHSZ);
+
+ read_lock_bh(&f->lock);
+
+ for (i = ACCESS_ONCE(f->next_bucket); budget; --budget) {
+ evicted += inet_evict_bucket(f, &f->hash[i]);
+ i = (i + 1) & (INETFRAGS_HASHSZ - 1);
+ if (evicted > INETFRAGS_EVICT_MAX)
+ break;
+ }
+
+ f->next_bucket = i;
+
+ read_unlock_bh(&f->lock);
+}
+
+static void inet_frag_schedule_worker(struct inet_frags *f)
+{
+ if (unlikely(!work_pending(&f->frags_work)))
+ schedule_work(&f->frags_work);
+}
+
void inet_frags_init(struct inet_frags *f)
{
int i;
+ INIT_WORK(&f->frags_work, inet_frag_worker);
+
for (i = 0; i < INETFRAGS_HASHSZ; i++) {
struct inet_frag_bucket *hb = &f->hash[i];
@@ -120,16 +203,22 @@ EXPORT_SYMBOL(inet_frags_init_net);
void inet_frags_fini(struct inet_frags *f)
{
del_timer(&f->secret_timer);
+ cancel_work_sync(&f->frags_work);
}
EXPORT_SYMBOL(inet_frags_fini);
void inet_frags_exit_net(struct netns_frags *nf, struct inet_frags *f)
{
+ int i;
+
nf->low_thresh = 0;
- local_bh_disable();
- inet_frag_evictor(nf, f, true);
- local_bh_enable();
+ read_lock_bh(&f->lock);
+
+ for (i = 0; i < INETFRAGS_HASHSZ ; i++)
+ inet_evict_bucket(f, &f->hash[i]);
+
+ read_unlock_bh(&f->lock);
percpu_counter_destroy(&nf->mem);
}
@@ -205,41 +294,6 @@ void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f,
}
EXPORT_SYMBOL(inet_frag_destroy);
-static int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f, bool force)
-{
- struct inet_frag_queue *q;
- int work, evicted = 0;
-
- work = frag_mem_limit(nf) - nf->low_thresh;
- while (work > 0 || force) {
- spin_lock(&nf->lru_lock);
-
- if (list_empty(&nf->lru_list)) {
- spin_unlock(&nf->lru_lock);
- break;
- }
-
- q = list_first_entry(&nf->lru_list,
- struct inet_frag_queue, lru_list);
- atomic_inc(&q->refcnt);
- /* Remove q from list to avoid several CPUs grabbing it */
- list_del_init(&q->lru_list);
-
- spin_unlock(&nf->lru_lock);
-
- spin_lock(&q->lock);
- if (!(q->last_in & INET_FRAG_COMPLETE))
- inet_frag_kill(q, f);
- spin_unlock(&q->lock);
-
- if (atomic_dec_and_test(&q->refcnt))
- inet_frag_destroy(q, f, &work);
- evicted++;
- }
-
- return evicted;
-}
-
static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
struct inet_frag_queue *qp_in, struct inet_frags *f,
void *arg)
@@ -292,8 +346,10 @@ static struct inet_frag_queue *inet_frag_alloc(struct netns_frags *nf,
{
struct inet_frag_queue *q;
- if (frag_mem_limit(nf) > nf->high_thresh)
+ if (frag_mem_limit(nf) > nf->high_thresh) {
+ inet_frag_schedule_worker(f);
return NULL;
+ }
q = kzalloc(f->qsize, GFP_ATOMIC);
if (q == NULL)
@@ -331,8 +387,8 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
struct inet_frag_queue *q;
int depth = 0;
- if (frag_mem_limit(nf) > nf->high_thresh)
- inet_frag_evictor(nf, f, false);
+ if (frag_mem_limit(nf) > nf->low_thresh)
+ inet_frag_schedule_worker(f);
hash &= (INETFRAGS_HASHSZ - 1);
hb = &f->hash[hash];
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 54988672d00d..54bd170c5eb4 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -195,7 +195,8 @@ static void ip_expire(unsigned long arg)
ipq_kill(qp);
- IP_INC_STATS_BH(net, IPSTATS_MIB_REASMTIMEOUT);
+ if (!(qp->q.last_in & INET_FRAG_EVICTED))
+ IP_INC_STATS_BH(net, IPSTATS_MIB_REASMTIMEOUT);
IP_INC_STATS_BH(net, IPSTATS_MIB_REASMFAILS);
if ((qp->q.last_in & INET_FRAG_FIRST_IN) && qp->q.fragments != NULL) {
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 97acbc490d9e..b3924b10dff3 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -141,7 +141,9 @@ void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq,
if (!dev)
goto out_rcu_unlock;
- IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMTIMEOUT);
+ if (!(fq->q.last_in & INET_FRAG_EVICTED))
+ IP6_INC_STATS_BH(net, __in6_dev_get(dev),
+ IPSTATS_MIB_REASMTIMEOUT);
IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMFAILS);
/* Don't send error if the first segment did not arrive. */
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 5/9] inet: frag: don't account number of fragment queues
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (3 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 4/9] inet: frag: move eviction of queues to work queue Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-25 6:15 ` David Miller
2014-07-24 14:50 ` [PATCH net-next 6/9] inet: frag: remove lru list Nikolay Aleksandrov
` (5 subsequent siblings)
10 siblings, 1 reply; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
The 'nqueues' counter is protected by the lru list lock,
once thats removed this needs to be converted to atomic
counter. Given this isn't used for anything except for
reporting it to userspace via /proc, just remove it.
We still report the memory currently used by fragment
reassembly queues.
Signed-off-by: Florian Westphal <fw@strlen.de>
---
include/net/inet_frag.h | 3 ---
include/net/ip.h | 1 -
include/net/ipv6.h | 5 -----
net/ipv4/inet_fragment.c | 1 -
net/ipv4/ip_fragment.c | 5 -----
net/ipv4/proc.c | 5 +++--
net/ipv6/proc.c | 4 ++--
7 files changed, 5 insertions(+), 19 deletions(-)
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index e975032ea11b..68de33765705 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -4,7 +4,6 @@
#include <linux/percpu_counter.h>
struct netns_frags {
- int nqueues;
struct list_head lru_list;
spinlock_t lru_lock;
@@ -158,7 +157,6 @@ static inline void inet_frag_lru_del(struct inet_frag_queue *q)
{
spin_lock(&q->net->lru_lock);
list_del_init(&q->lru_list);
- q->net->nqueues--;
spin_unlock(&q->net->lru_lock);
}
@@ -167,7 +165,6 @@ static inline void inet_frag_lru_add(struct netns_frags *nf,
{
spin_lock(&nf->lru_lock);
list_add_tail(&q->lru_list, &nf->lru_list);
- q->net->nqueues++;
spin_unlock(&nf->lru_lock);
}
diff --git a/include/net/ip.h b/include/net/ip.h
index 2e8f055989c3..ca14799545fd 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -495,7 +495,6 @@ static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user)
}
#endif
int ip_frag_mem(struct net *net);
-int ip_frag_nqueues(struct net *net);
/*
* Functions provided by ip_forward.c
diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index 25c2170e1298..a2db816e8461 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -299,11 +299,6 @@ static inline bool ipv6_accept_ra(struct inet6_dev *idev)
}
#if IS_ENABLED(CONFIG_IPV6)
-static inline int ip6_frag_nqueues(struct net *net)
-{
- return net->ipv6.frags.nqueues;
-}
-
static inline int ip6_frag_mem(struct net *net)
{
return sum_frag_mem_limit(&net->ipv6.frags);
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 43315ecb9400..231ca0b40811 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -193,7 +193,6 @@ EXPORT_SYMBOL(inet_frags_init);
void inet_frags_init_net(struct netns_frags *nf)
{
- nf->nqueues = 0;
init_frag_mem_limit(nf);
INIT_LIST_HEAD(&nf->lru_list);
spin_lock_init(&nf->lru_lock);
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 54bd170c5eb4..1f42c2e3966b 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -86,11 +86,6 @@ static inline u8 ip4_frag_ecn(u8 tos)
static struct inet_frags ip4_frags;
-int ip_frag_nqueues(struct net *net)
-{
- return net->ipv4.frags.nqueues;
-}
-
int ip_frag_mem(struct net *net)
{
return sum_frag_mem_limit(&net->ipv4.frags);
diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c
index ae0af9386f7c..8e3eb39f84e7 100644
--- a/net/ipv4/proc.c
+++ b/net/ipv4/proc.c
@@ -52,6 +52,7 @@
static int sockstat_seq_show(struct seq_file *seq, void *v)
{
struct net *net = seq->private;
+ unsigned int frag_mem;
int orphans, sockets;
local_bh_disable();
@@ -71,8 +72,8 @@ static int sockstat_seq_show(struct seq_file *seq, void *v)
sock_prot_inuse_get(net, &udplite_prot));
seq_printf(seq, "RAW: inuse %d\n",
sock_prot_inuse_get(net, &raw_prot));
- seq_printf(seq, "FRAG: inuse %d memory %d\n",
- ip_frag_nqueues(net), ip_frag_mem(net));
+ frag_mem = ip_frag_mem(net);
+ seq_printf(seq, "FRAG: inuse %u memory %u\n", !!frag_mem, frag_mem);
return 0;
}
diff --git a/net/ipv6/proc.c b/net/ipv6/proc.c
index 3317440ea341..2d6f860e5c1e 100644
--- a/net/ipv6/proc.c
+++ b/net/ipv6/proc.c
@@ -33,6 +33,7 @@
static int sockstat6_seq_show(struct seq_file *seq, void *v)
{
struct net *net = seq->private;
+ unsigned int frag_mem = ip6_frag_mem(net);
seq_printf(seq, "TCP6: inuse %d\n",
sock_prot_inuse_get(net, &tcpv6_prot));
@@ -42,8 +43,7 @@ static int sockstat6_seq_show(struct seq_file *seq, void *v)
sock_prot_inuse_get(net, &udplitev6_prot));
seq_printf(seq, "RAW6: inuse %d\n",
sock_prot_inuse_get(net, &rawv6_prot));
- seq_printf(seq, "FRAG6: inuse %d memory %d\n",
- ip6_frag_nqueues(net), ip6_frag_mem(net));
+ seq_printf(seq, "FRAG6: inuse %u memory %u\n", !!frag_mem, frag_mem);
return 0;
}
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* Re: [PATCH net-next 5/9] inet: frag: don't account number of fragment queues
2014-07-24 14:50 ` [PATCH net-next 5/9] inet: frag: don't account number of fragment queues Nikolay Aleksandrov
@ 2014-07-25 6:15 ` David Miller
2014-07-25 8:32 ` Florian Westphal
0 siblings, 1 reply; 15+ messages in thread
From: David Miller @ 2014-07-25 6:15 UTC (permalink / raw)
To: nikolay
Cc: netdev, fw, kuznet, jmorris, yoshfuji, kaber, alex.aring,
eric.dumazet
From: Nikolay Aleksandrov <nikolay@redhat.com>
Date: Thu, 24 Jul 2014 16:50:33 +0200
> From: Florian Westphal <fw@strlen.de>
>
> The 'nqueues' counter is protected by the lru list lock,
> once thats removed this needs to be converted to atomic
> counter. Given this isn't used for anything except for
> reporting it to userspace via /proc, just remove it.
>
> We still report the memory currently used by fragment
> reassembly queues.
>
> Signed-off-by: Florian Westphal <fw@strlen.de>
It's a user visible value, you can't just change it's semantics.
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH net-next 5/9] inet: frag: don't account number of fragment queues
2014-07-25 6:15 ` David Miller
@ 2014-07-25 8:32 ` Florian Westphal
0 siblings, 0 replies; 15+ messages in thread
From: Florian Westphal @ 2014-07-25 8:32 UTC (permalink / raw)
To: David Miller
Cc: nikolay, netdev, fw, kuznet, jmorris, yoshfuji, kaber, alex.aring,
eric.dumazet
David Miller <davem@davemloft.net> wrote:
> From: Nikolay Aleksandrov <nikolay@redhat.com>
> Date: Thu, 24 Jul 2014 16:50:33 +0200
>
> > From: Florian Westphal <fw@strlen.de>
> >
> > The 'nqueues' counter is protected by the lru list lock,
> > once thats removed this needs to be converted to atomic
> > counter. Given this isn't used for anything except for
> > reporting it to userspace via /proc, just remove it.
> >
> > We still report the memory currently used by fragment
> > reassembly queues.
> >
> > Signed-off-by: Florian Westphal <fw@strlen.de>
>
> It's a user visible value, you can't just change it's semantics.
Really? It won't break scripts.
In my tests this never reported any useful information to me.
If you insist, we can keep this but it seems silly to me to add atomic
ops just for this one counter which - IMO - does not provide
any useful information.
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH net-next 6/9] inet: frag: remove lru list
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (4 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 5/9] inet: frag: don't account number of fragment queues Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 7/9] inet: frag: remove periodic secret rebuild timer Nikolay Aleksandrov
` (4 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
no longer used.
Signed-off-by: Florian Westphal <fw@strlen.de>
---
include/net/inet_frag.h | 32 ++------------------------------
net/ieee802154/reassembly.c | 1 -
net/ipv4/inet_fragment.c | 12 ++----------
net/ipv4/ip_fragment.c | 1 -
net/ipv6/netfilter/nf_conntrack_reasm.c | 1 -
net/ipv6/reassembly.c | 1 -
6 files changed, 4 insertions(+), 44 deletions(-)
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 68de33765705..90d21ea62c59 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -4,9 +4,6 @@
#include <linux/percpu_counter.h>
struct netns_frags {
- struct list_head lru_list;
- spinlock_t lru_lock;
-
/* The percpu_counter "mem" need to be cacheline aligned.
* mem.count must not share cacheline with other writers
*/
@@ -21,7 +18,6 @@ struct netns_frags {
struct inet_frag_queue {
spinlock_t lock;
struct timer_list timer; /* when will this queue expire? */
- struct list_head lru_list; /* lru list member */
struct hlist_node list;
atomic_t refcnt;
struct sk_buff *fragments; /* list of received fragments */
@@ -91,8 +87,7 @@ void inet_frags_init_net(struct netns_frags *nf);
void inet_frags_exit_net(struct netns_frags *nf, struct inet_frags *f);
void inet_frag_kill(struct inet_frag_queue *q, struct inet_frags *f);
-void inet_frag_destroy(struct inet_frag_queue *q,
- struct inet_frags *f, int *work);
+void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f);
struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
struct inet_frags *f, void *key, unsigned int hash)
__releases(&f->lock);
@@ -102,7 +97,7 @@ void inet_frag_maybe_warn_overflow(struct inet_frag_queue *q,
static inline void inet_frag_put(struct inet_frag_queue *q, struct inet_frags *f)
{
if (atomic_dec_and_test(&q->refcnt))
- inet_frag_destroy(q, f, NULL);
+ inet_frag_destroy(q, f);
}
/* Memory Tracking Functions. */
@@ -145,29 +140,6 @@ static inline unsigned int sum_frag_mem_limit(struct netns_frags *nf)
return res;
}
-static inline void inet_frag_lru_move(struct inet_frag_queue *q)
-{
- spin_lock(&q->net->lru_lock);
- if (!list_empty(&q->lru_list))
- list_move_tail(&q->lru_list, &q->net->lru_list);
- spin_unlock(&q->net->lru_lock);
-}
-
-static inline void inet_frag_lru_del(struct inet_frag_queue *q)
-{
- spin_lock(&q->net->lru_lock);
- list_del_init(&q->lru_list);
- spin_unlock(&q->net->lru_lock);
-}
-
-static inline void inet_frag_lru_add(struct netns_frags *nf,
- struct inet_frag_queue *q)
-{
- spin_lock(&nf->lru_lock);
- list_add_tail(&q->lru_list, &nf->lru_list);
- spin_unlock(&nf->lru_lock);
-}
-
/* RFC 3168 support :
* We want to check ECN values of all fragments, do detect invalid combinations.
* In ipq->ecn, we store the OR value of each ip4_frag_ecn() fragment value.
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index 9503a48556f7..b4bc7a50eccf 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -219,7 +219,6 @@ found:
return res;
}
- inet_frag_lru_move(&fq->q);
return -1;
err:
kfree_skb(skb);
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 231ca0b40811..198a5ed7a815 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -194,8 +194,6 @@ EXPORT_SYMBOL(inet_frags_init);
void inet_frags_init_net(struct netns_frags *nf)
{
init_frag_mem_limit(nf);
- INIT_LIST_HEAD(&nf->lru_list);
- spin_lock_init(&nf->lru_lock);
}
EXPORT_SYMBOL(inet_frags_init_net);
@@ -237,7 +235,6 @@ static inline void fq_unlink(struct inet_frag_queue *fq, struct inet_frags *f)
spin_unlock(&hb->chain_lock);
read_unlock(&f->lock);
- inet_frag_lru_del(fq);
}
void inet_frag_kill(struct inet_frag_queue *fq, struct inet_frags *f)
@@ -261,8 +258,7 @@ static inline void frag_kfree_skb(struct netns_frags *nf, struct inet_frags *f,
kfree_skb(skb);
}
-void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f,
- int *work)
+void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f)
{
struct sk_buff *fp;
struct netns_frags *nf;
@@ -282,14 +278,11 @@ void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f,
fp = xp;
}
sum = sum_truesize + f->qsize;
- if (work)
- *work -= sum;
sub_frag_mem_limit(q, sum);
if (f->destructor)
f->destructor(q);
kfree(q);
-
}
EXPORT_SYMBOL(inet_frag_destroy);
@@ -333,7 +326,7 @@ static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
atomic_inc(&qp->refcnt);
hlist_add_head(&qp->list, &hb->chain);
- inet_frag_lru_add(nf, qp);
+
spin_unlock(&hb->chain_lock);
read_unlock(&f->lock);
@@ -361,7 +354,6 @@ static struct inet_frag_queue *inet_frag_alloc(struct netns_frags *nf,
setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
spin_lock_init(&q->lock);
atomic_set(&q->refcnt, 1);
- INIT_LIST_HEAD(&q->lru_list);
return q;
}
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 1f42c2e3966b..8fbeee495037 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -489,7 +489,6 @@ found:
}
skb_dst_drop(skb);
- inet_frag_lru_move(&qp->q);
return -EINPROGRESS;
err:
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 58e32cf91c95..fb0f72a0ff31 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -349,7 +349,6 @@ found:
fq->q.last_in |= INET_FRAG_FIRST_IN;
}
- inet_frag_lru_move(&fq->q);
return 0;
discard_fq:
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index b3924b10dff3..af85551682c2 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -351,7 +351,6 @@ found:
}
skb_dst_drop(skb);
- inet_frag_lru_move(&fq->q);
return -1;
discard_fq:
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 7/9] inet: frag: remove periodic secret rebuild timer
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (5 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 6/9] inet: frag: remove lru list Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 8/9] inet: frag: use seqlock for hash rebuild Nikolay Aleksandrov
` (3 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
merge functionality into the eviction workqueue.
Instead of rebuilding every n seconds, take advantage of the upper
hash chain length limit.
If we hit it, mark table for rebuild and schedule workqueue.
To prevent frequent rebuilds when we're completely overloaded,
don't rebuild more than once every 5 seconds.
ipfrag_secret_interval sysctl is now obsolete and has been marked as
deprecated, it still can be changed so scripts won't be broken but it
won't have any effect. A comment is left above each unused secret_timer
variable to avoid confusion.
Joint work with Nikolay Aleksandrov.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
---
Documentation/networking/ip-sysctl.txt | 10 --------
include/net/inet_frag.h | 4 +--
net/ieee802154/reassembly.c | 5 ++--
net/ipv4/inet_fragment.c | 43 ++++++++++++++++++++++-----------
net/ipv4/ip_fragment.c | 5 ++--
net/ipv6/netfilter/nf_conntrack_reasm.c | 1 -
net/ipv6/reassembly.c | 5 ++--
7 files changed, 40 insertions(+), 33 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 625c8dda4be7..e8c304e37831 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -111,11 +111,6 @@ ipfrag_low_thresh - INTEGER
ipfrag_time - INTEGER
Time in seconds to keep an IP fragment in memory.
-ipfrag_secret_interval - INTEGER
- Regeneration interval (in seconds) of the hash secret (or lifetime
- for the hash secret) for IP fragments.
- Default: 600
-
ipfrag_max_dist - INTEGER
ipfrag_max_dist is a non-negative integer value which defines the
maximum "disorder" which is allowed among fragments which share a
@@ -1164,11 +1159,6 @@ ip6frag_low_thresh - INTEGER
ip6frag_time - INTEGER
Time in seconds to keep an IPv6 fragment in memory.
-ip6frag_secret_interval - INTEGER
- Regeneration interval (in seconds) of the hash secret (or lifetime
- for the hash secret) for IPv6 fragments.
- Default: 600
-
conf/default/*:
Change the interface-specific default settings.
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 90d21ea62c59..d9cc5bb64854 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -58,11 +58,11 @@ struct inet_frags {
* Its primarily a rebuild protection rwlock.
*/
rwlock_t lock ____cacheline_aligned_in_smp;
- int secret_interval;
- struct timer_list secret_timer;
struct work_struct frags_work;
unsigned int next_bucket;
+ unsigned long last_rebuild_jiffies;
+ bool rebuild;
/* The first call to hashfn is responsible to initialize
* rnd. This is best done with net_get_random_once.
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index b4bc7a50eccf..20d219682d84 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -419,10 +419,12 @@ static struct ctl_table lowpan_frags_ns_ctl_table[] = {
{ }
};
+/* secret interval has been deprecated */
+static int lowpan_frags_secret_interval_unused;
static struct ctl_table lowpan_frags_ctl_table[] = {
{
.procname = "6lowpanfrag_secret_interval",
- .data = &lowpan_frags.secret_interval,
+ .data = &lowpan_frags_secret_interval_unused,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
@@ -562,7 +564,6 @@ int __init lowpan_net_frag_init(void)
lowpan_frags.qsize = sizeof(struct frag_queue);
lowpan_frags.match = lowpan_frag_match;
lowpan_frags.frag_expire = lowpan_frag_expire;
- lowpan_frags.secret_interval = 10 * 60 * HZ;
inet_frags_init(&lowpan_frags);
return ret;
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 198a5ed7a815..58d4c38534f6 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -28,6 +28,9 @@
#define INETFRAGS_EVICT_BUCKETS 128
#define INETFRAGS_EVICT_MAX 512
+/* don't rebuild inetfrag table with new secret more often than this */
+#define INETFRAGS_MIN_REBUILD_INTERVAL (5 * HZ)
+
/* Given the OR values of all fragments, apply RFC 3168 5.3 requirements
* Value : 0xff if frame should be dropped.
* 0 or INET_ECN_CE value, to be ORed in to final iph->tos field
@@ -55,16 +58,24 @@ inet_frag_hashfn(const struct inet_frags *f, const struct inet_frag_queue *q)
return f->hashfn(q) & (INETFRAGS_HASHSZ - 1);
}
-static void inet_frag_secret_rebuild(unsigned long dummy)
+static bool inet_frag_may_rebuild(struct inet_frags *f)
+{
+ return time_after(jiffies,
+ f->last_rebuild_jiffies + INETFRAGS_MIN_REBUILD_INTERVAL);
+}
+
+static void inet_frag_secret_rebuild(struct inet_frags *f)
{
- struct inet_frags *f = (struct inet_frags *)dummy;
- unsigned long now = jiffies;
int i;
/* Per bucket lock NOT needed here, due to write lock protection */
- write_lock(&f->lock);
+ write_lock_bh(&f->lock);
+
+ if (!inet_frag_may_rebuild(f))
+ goto out;
get_random_bytes(&f->rnd, sizeof(u32));
+
for (i = 0; i < INETFRAGS_HASHSZ; i++) {
struct inet_frag_bucket *hb;
struct inet_frag_queue *q;
@@ -85,9 +96,11 @@ static void inet_frag_secret_rebuild(unsigned long dummy)
}
}
}
- write_unlock(&f->lock);
- mod_timer(&f->secret_timer, now + f->secret_interval);
+ f->rebuild = false;
+ f->last_rebuild_jiffies = jiffies;
+out:
+ write_unlock_bh(&f->lock);
}
static bool inet_fragq_should_evict(const struct inet_frag_queue *q)
@@ -162,6 +175,8 @@ static void inet_frag_worker(struct work_struct *work)
f->next_bucket = i;
read_unlock_bh(&f->lock);
+ if (f->rebuild && inet_frag_may_rebuild(f))
+ inet_frag_secret_rebuild(f);
}
static void inet_frag_schedule_worker(struct inet_frags *f)
@@ -183,11 +198,7 @@ void inet_frags_init(struct inet_frags *f)
INIT_HLIST_HEAD(&hb->chain);
}
rwlock_init(&f->lock);
-
- setup_timer(&f->secret_timer, inet_frag_secret_rebuild,
- (unsigned long)f);
- f->secret_timer.expires = jiffies + f->secret_interval;
- add_timer(&f->secret_timer);
+ f->last_rebuild_jiffies = 0;
}
EXPORT_SYMBOL(inet_frags_init);
@@ -199,7 +210,6 @@ EXPORT_SYMBOL(inet_frags_init_net);
void inet_frags_fini(struct inet_frags *f)
{
- del_timer(&f->secret_timer);
cancel_work_sync(&f->frags_work);
}
EXPORT_SYMBOL(inet_frags_fini);
@@ -399,8 +409,13 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
if (depth <= INETFRAGS_MAXDEPTH)
return inet_frag_create(nf, f, key);
- else
- return ERR_PTR(-ENOBUFS);
+
+ if (inet_frag_may_rebuild(f)) {
+ f->rebuild = true;
+ inet_frag_schedule_worker(f);
+ }
+
+ return ERR_PTR(-ENOBUFS);
}
EXPORT_SYMBOL(inet_frag_find);
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 8fbeee495037..44e591a7e03f 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -720,10 +720,12 @@ static struct ctl_table ip4_frags_ns_ctl_table[] = {
{ }
};
+/* secret interval has been deprecated */
+static int ip4_frags_secret_interval_unused;
static struct ctl_table ip4_frags_ctl_table[] = {
{
.procname = "ipfrag_secret_interval",
- .data = &ip4_frags.secret_interval,
+ .data = &ip4_frags_secret_interval_unused,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
@@ -853,6 +855,5 @@ void __init ipfrag_init(void)
ip4_frags.qsize = sizeof(struct ipq);
ip4_frags.match = ip4_frag_match;
ip4_frags.frag_expire = ip_expire;
- ip4_frags.secret_interval = 10 * 60 * HZ;
inet_frags_init(&ip4_frags);
}
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index fb0f72a0ff31..3b3ef9774cc2 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -669,7 +669,6 @@ int nf_ct_frag6_init(void)
nf_frags.qsize = sizeof(struct frag_queue);
nf_frags.match = ip6_frag_match;
nf_frags.frag_expire = nf_ct_frag6_expire;
- nf_frags.secret_interval = 10 * 60 * HZ;
inet_frags_init(&nf_frags);
ret = register_pernet_subsys(&nf_ct_net_ops);
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index af85551682c2..987fea46b915 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -604,10 +604,12 @@ static struct ctl_table ip6_frags_ns_ctl_table[] = {
{ }
};
+/* secret interval has been deprecated */
+static int ip6_frags_secret_interval_unused;
static struct ctl_table ip6_frags_ctl_table[] = {
{
.procname = "ip6frag_secret_interval",
- .data = &ip6_frags.secret_interval,
+ .data = &ip6_frags_secret_interval_unused,
.maxlen = sizeof(int),
.mode = 0644,
.proc_handler = proc_dointvec_jiffies,
@@ -737,7 +739,6 @@ int __init ipv6_frag_init(void)
ip6_frags.qsize = sizeof(struct frag_queue);
ip6_frags.match = ip6_frag_match;
ip6_frags.frag_expire = ip6_frag_expire;
- ip6_frags.secret_interval = 10 * 60 * HZ;
inet_frags_init(&ip6_frags);
out:
return ret;
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 8/9] inet: frag: use seqlock for hash rebuild
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (6 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 7/9] inet: frag: remove periodic secret rebuild timer Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-24 14:50 ` [PATCH net-next 9/9] inet: frag: set limits and make init_net's high_thresh limit global Nikolay Aleksandrov
` (2 subsequent siblings)
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
From: Florian Westphal <fw@strlen.de>
rehash is rare operation, don't force readers to take
the read-side rwlock.
Instead, we only have to detect the (rare) case where
the secret was altered while we are trying to insert
a new inetfrag queue into the table.
If it was changed, drop the bucket lock and recompute
the hash to get the 'new' chain bucket that we have to
insert into.
Joint work with Nikolay Aleksandrov.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
---
include/net/inet_frag.h | 13 +++--
net/ieee802154/reassembly.c | 1 -
net/ipv4/inet_fragment.c | 88 ++++++++++++++++++++-------------
net/ipv4/ip_fragment.c | 1 -
net/ipv6/netfilter/nf_conntrack_reasm.c | 2 +-
net/ipv6/reassembly.c | 1 -
6 files changed, 62 insertions(+), 44 deletions(-)
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index d9cc5bb64854..6f4930a0b660 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -53,11 +53,6 @@ struct inet_frag_bucket {
struct inet_frags {
struct inet_frag_bucket hash[INETFRAGS_HASHSZ];
- /* This rwlock is a global lock (seperate per IPv4, IPv6 and
- * netfilter). Important to keep this on a seperate cacheline.
- * Its primarily a rebuild protection rwlock.
- */
- rwlock_t lock ____cacheline_aligned_in_smp;
struct work_struct frags_work;
unsigned int next_bucket;
@@ -66,8 +61,12 @@ struct inet_frags {
/* The first call to hashfn is responsible to initialize
* rnd. This is best done with net_get_random_once.
+ *
+ * rnd_seqlock is used to let hash insertion detect
+ * when it needs to re-lookup the hash chain to use.
*/
u32 rnd;
+ seqlock_t rnd_seqlock;
int qsize;
unsigned int (*hashfn)(const struct inet_frag_queue *);
@@ -89,8 +88,8 @@ void inet_frags_exit_net(struct netns_frags *nf, struct inet_frags *f);
void inet_frag_kill(struct inet_frag_queue *q, struct inet_frags *f);
void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f);
struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
- struct inet_frags *f, void *key, unsigned int hash)
- __releases(&f->lock);
+ struct inet_frags *f, void *key, unsigned int hash);
+
void inet_frag_maybe_warn_overflow(struct inet_frag_queue *q,
const char *prefix);
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index 20d219682d84..8da635d92a58 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -124,7 +124,6 @@ fq_find(struct net *net, const struct lowpan_frag_info *frag_info,
arg.src = src;
arg.dst = dst;
- read_lock(&lowpan_frags.lock);
hash = lowpan_hash_frag(frag_info->d_tag, frag_info->d_size, src, dst);
q = inet_frag_find(&ieee802154_lowpan->frags,
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 58d4c38534f6..62b1f73749dc 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -68,8 +68,7 @@ static void inet_frag_secret_rebuild(struct inet_frags *f)
{
int i;
- /* Per bucket lock NOT needed here, due to write lock protection */
- write_lock_bh(&f->lock);
+ write_seqlock_bh(&f->rnd_seqlock);
if (!inet_frag_may_rebuild(f))
goto out;
@@ -82,6 +81,8 @@ static void inet_frag_secret_rebuild(struct inet_frags *f)
struct hlist_node *n;
hb = &f->hash[i];
+ spin_lock(&hb->chain_lock);
+
hlist_for_each_entry_safe(q, n, &hb->chain, list) {
unsigned int hval = inet_frag_hashfn(f, q);
@@ -92,15 +93,28 @@ static void inet_frag_secret_rebuild(struct inet_frags *f)
/* Relink to new hash chain. */
hb_dest = &f->hash[hval];
+
+ /* This is the only place where we take
+ * another chain_lock while already holding
+ * one. As this will not run concurrently,
+ * we cannot deadlock on hb_dest lock below, if its
+ * already locked it will be released soon since
+ * other caller cannot be waiting for hb lock
+ * that we've taken above.
+ */
+ spin_lock_nested(&hb_dest->chain_lock,
+ SINGLE_DEPTH_NESTING);
hlist_add_head(&q->list, &hb_dest->chain);
+ spin_unlock(&hb_dest->chain_lock);
}
}
+ spin_unlock(&hb->chain_lock);
}
f->rebuild = false;
f->last_rebuild_jiffies = jiffies;
out:
- write_unlock_bh(&f->lock);
+ write_sequnlock_bh(&f->rnd_seqlock);
}
static bool inet_fragq_should_evict(const struct inet_frag_queue *q)
@@ -163,7 +177,7 @@ static void inet_frag_worker(struct work_struct *work)
BUILD_BUG_ON(INETFRAGS_EVICT_BUCKETS >= INETFRAGS_HASHSZ);
- read_lock_bh(&f->lock);
+ local_bh_disable();
for (i = ACCESS_ONCE(f->next_bucket); budget; --budget) {
evicted += inet_evict_bucket(f, &f->hash[i]);
@@ -174,7 +188,8 @@ static void inet_frag_worker(struct work_struct *work)
f->next_bucket = i;
- read_unlock_bh(&f->lock);
+ local_bh_enable();
+
if (f->rebuild && inet_frag_may_rebuild(f))
inet_frag_secret_rebuild(f);
}
@@ -197,7 +212,8 @@ void inet_frags_init(struct inet_frags *f)
spin_lock_init(&hb->chain_lock);
INIT_HLIST_HEAD(&hb->chain);
}
- rwlock_init(&f->lock);
+
+ seqlock_init(&f->rnd_seqlock);
f->last_rebuild_jiffies = 0;
}
EXPORT_SYMBOL(inet_frags_init);
@@ -216,35 +232,56 @@ EXPORT_SYMBOL(inet_frags_fini);
void inet_frags_exit_net(struct netns_frags *nf, struct inet_frags *f)
{
+ unsigned int seq;
int i;
nf->low_thresh = 0;
+ local_bh_disable();
- read_lock_bh(&f->lock);
+evict_again:
+ seq = read_seqbegin(&f->rnd_seqlock);
for (i = 0; i < INETFRAGS_HASHSZ ; i++)
inet_evict_bucket(f, &f->hash[i]);
- read_unlock_bh(&f->lock);
+ if (read_seqretry(&f->rnd_seqlock, seq))
+ goto evict_again;
+
+ local_bh_enable();
percpu_counter_destroy(&nf->mem);
}
EXPORT_SYMBOL(inet_frags_exit_net);
-static inline void fq_unlink(struct inet_frag_queue *fq, struct inet_frags *f)
+static struct inet_frag_bucket *
+get_frag_bucket_locked(struct inet_frag_queue *fq, struct inet_frags *f)
+__acquires(hb->chain_lock)
{
struct inet_frag_bucket *hb;
- unsigned int hash;
+ unsigned int seq, hash;
+
+ restart:
+ seq = read_seqbegin(&f->rnd_seqlock);
- read_lock(&f->lock);
hash = inet_frag_hashfn(f, fq);
hb = &f->hash[hash];
spin_lock(&hb->chain_lock);
+ if (read_seqretry(&f->rnd_seqlock, seq)) {
+ spin_unlock(&hb->chain_lock);
+ goto restart;
+ }
+
+ return hb;
+}
+
+static inline void fq_unlink(struct inet_frag_queue *fq, struct inet_frags *f)
+{
+ struct inet_frag_bucket *hb;
+
+ hb = get_frag_bucket_locked(fq, f);
hlist_del(&fq->list);
spin_unlock(&hb->chain_lock);
-
- read_unlock(&f->lock);
}
void inet_frag_kill(struct inet_frag_queue *fq, struct inet_frags *f)
@@ -300,30 +337,18 @@ static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
struct inet_frag_queue *qp_in, struct inet_frags *f,
void *arg)
{
- struct inet_frag_bucket *hb;
+ struct inet_frag_bucket *hb = get_frag_bucket_locked(qp_in, f);
struct inet_frag_queue *qp;
- unsigned int hash;
-
- read_lock(&f->lock); /* Protects against hash rebuild */
- /*
- * While we stayed w/o the lock other CPU could update
- * the rnd seed, so we need to re-calculate the hash
- * chain. Fortunatelly the qp_in can be used to get one.
- */
- hash = inet_frag_hashfn(f, qp_in);
- hb = &f->hash[hash];
- spin_lock(&hb->chain_lock);
#ifdef CONFIG_SMP
/* With SMP race we have to recheck hash table, because
- * such entry could be created on other cpu, while we
- * released the hash bucket lock.
+ * such entry could have been created on other cpu before
+ * we acquired hash bucket lock.
*/
hlist_for_each_entry(qp, &hb->chain, list) {
if (qp->net == nf && f->match(qp, arg)) {
atomic_inc(&qp->refcnt);
spin_unlock(&hb->chain_lock);
- read_unlock(&f->lock);
qp_in->last_in |= INET_FRAG_COMPLETE;
inet_frag_put(qp_in, f);
return qp;
@@ -338,7 +363,6 @@ static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
hlist_add_head(&qp->list, &hb->chain);
spin_unlock(&hb->chain_lock);
- read_unlock(&f->lock);
return qp;
}
@@ -382,7 +406,6 @@ static struct inet_frag_queue *inet_frag_create(struct netns_frags *nf,
struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
struct inet_frags *f, void *key, unsigned int hash)
- __releases(&f->lock)
{
struct inet_frag_bucket *hb;
struct inet_frag_queue *q;
@@ -399,19 +422,18 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
if (q->net == nf && f->match(q, key)) {
atomic_inc(&q->refcnt);
spin_unlock(&hb->chain_lock);
- read_unlock(&f->lock);
return q;
}
depth++;
}
spin_unlock(&hb->chain_lock);
- read_unlock(&f->lock);
if (depth <= INETFRAGS_MAXDEPTH)
return inet_frag_create(nf, f, key);
if (inet_frag_may_rebuild(f)) {
- f->rebuild = true;
+ if (!f->rebuild)
+ f->rebuild = true;
inet_frag_schedule_worker(f);
}
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 44e591a7e03f..ccee68dffd6e 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -244,7 +244,6 @@ static inline struct ipq *ip_find(struct net *net, struct iphdr *iph, u32 user)
arg.iph = iph;
arg.user = user;
- read_lock(&ip4_frags.lock);
hash = ipqhashfn(iph->id, iph->saddr, iph->daddr, iph->protocol);
q = inet_frag_find(&net->ipv4.frags, &ip4_frags, &arg, hash);
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 3b3ef9774cc2..4d9da1e35f8c 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -193,7 +193,7 @@ static inline struct frag_queue *fq_find(struct net *net, __be32 id,
arg.dst = dst;
arg.ecn = ecn;
- read_lock_bh(&nf_frags.lock);
+ local_bh_disable();
hash = nf_hash_frag(id, src, dst);
q = inet_frag_find(&net->nf_frag.frags, &nf_frags, &arg, hash);
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 987fea46b915..57a9707b2032 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -190,7 +190,6 @@ fq_find(struct net *net, __be32 id, const struct in6_addr *src,
arg.dst = dst;
arg.ecn = ecn;
- read_lock(&ip6_frags.lock);
hash = inet6_hash_frag(id, src, dst);
q = inet_frag_find(&net->ipv6.frags, &ip6_frags, &arg, hash);
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* [PATCH net-next 9/9] inet: frag: set limits and make init_net's high_thresh limit global
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (7 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 8/9] inet: frag: use seqlock for hash rebuild Nikolay Aleksandrov
@ 2014-07-24 14:50 ` Nikolay Aleksandrov
2014-07-28 5:37 ` [PATCH net-next 0/9] inet: frag: cleanup and update David Miller
2014-07-28 6:43 ` Alexander Aring
10 siblings, 0 replies; 15+ messages in thread
From: Nikolay Aleksandrov @ 2014-07-24 14:50 UTC (permalink / raw)
To: netdev
Cc: Florian Westphal, Nikolay Aleksandrov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy, Alexander Aring, Eric Dumazet
This patch makes init_net's high_thresh limit to be the maximum for all
namespaces, thus introducing a global memory limit threshold equal to the
sum of the individual high_thresh limits which are capped.
It also introduces some sane minimums for low_thresh as it shouldn't be
able to drop below 0 (or > high_thresh in the unsigned case), and
overall low_thresh should not ever be above high_thresh, so we make the
following relations for a namespace:
init_net:
high_thresh - max(not capped), min(init_net low_thresh)
low_thresh - max(init_net high_thresh), min (0)
all other namespaces:
high_thresh = max(init_net high_thresh), min(namespace's low_thresh)
low_thresh = max(namespace's high_thresh), min(0)
The major issue with having low_thresh > high_thresh is that we'll
schedule eviction but never evict anything and thus rely only on the
timers.
Signed-off-by: Nikolay Aleksandrov <nikolay@redhat.com>
---
Documentation/networking/ip-sysctl.txt | 3 ++-
net/ieee802154/reassembly.c | 12 ++++++++++--
net/ipv4/ip_fragment.c | 10 ++++++++--
net/ipv6/netfilter/nf_conntrack_reasm.c | 12 ++++++++++--
net/ipv6/reassembly.c | 12 ++++++++++--
5 files changed, 40 insertions(+), 9 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index e8c304e37831..29a93518bf18 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -101,7 +101,8 @@ ipfrag_high_thresh - INTEGER
Maximum memory used to reassemble IP fragments. When
ipfrag_high_thresh bytes of memory is allocated for this purpose,
the fragment handler will toss packets until ipfrag_low_thresh
- is reached.
+ is reached. This also serves as a maximum limit to namespaces
+ different from the initial one.
ipfrag_low_thresh - INTEGER
Maximum memory used to reassemble IP fragments before the kernel
diff --git a/net/ieee802154/reassembly.c b/net/ieee802154/reassembly.c
index 8da635d92a58..f13d4f32e207 100644
--- a/net/ieee802154/reassembly.c
+++ b/net/ieee802154/reassembly.c
@@ -386,20 +386,25 @@ err:
EXPORT_SYMBOL(lowpan_frag_rcv);
#ifdef CONFIG_SYSCTL
+static int zero;
+
static struct ctl_table lowpan_frags_ns_ctl_table[] = {
{
.procname = "6lowpanfrag_high_thresh",
.data = &init_net.ieee802154_lowpan.frags.high_thresh,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = proc_dointvec
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &init_net.ieee802154_lowpan.frags.low_thresh
},
{
.procname = "6lowpanfrag_low_thresh",
.data = &init_net.ieee802154_lowpan.frags.low_thresh,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = proc_dointvec
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &init_net.ieee802154_lowpan.frags.high_thresh
},
{
.procname = "6lowpanfrag_time",
@@ -446,7 +451,10 @@ static int __net_init lowpan_frags_ns_sysctl_register(struct net *net)
goto err_alloc;
table[0].data = &ieee802154_lowpan->frags.high_thresh;
+ table[0].extra1 = &ieee802154_lowpan->frags.low_thresh;
+ table[0].extra2 = &init_net.ieee802154_lowpan.frags.high_thresh;
table[1].data = &ieee802154_lowpan->frags.low_thresh;
+ table[1].extra2 = &ieee802154_lowpan->frags.high_thresh;
table[2].data = &ieee802154_lowpan->frags.timeout;
table[3].data = &ieee802154_lowpan->max_dsize;
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index ccee68dffd6e..634fc31aa243 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -700,14 +700,17 @@ static struct ctl_table ip4_frags_ns_ctl_table[] = {
.data = &init_net.ipv4.frags.high_thresh,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = proc_dointvec
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &init_net.ipv4.frags.low_thresh
},
{
.procname = "ipfrag_low_thresh",
.data = &init_net.ipv4.frags.low_thresh,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = proc_dointvec
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &init_net.ipv4.frags.high_thresh
},
{
.procname = "ipfrag_time",
@@ -752,7 +755,10 @@ static int __net_init ip4_frags_ns_ctl_register(struct net *net)
goto err_alloc;
table[0].data = &net->ipv4.frags.high_thresh;
+ table[0].extra1 = &net->ipv4.frags.low_thresh;
+ table[0].extra2 = &init_net.ipv4.frags.high_thresh;
table[1].data = &net->ipv4.frags.low_thresh;
+ table[1].extra2 = &net->ipv4.frags.high_thresh;
table[2].data = &net->ipv4.frags.timeout;
/* Don't export sysctls to unprivileged users */
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 4d9da1e35f8c..3d4bccf6d67d 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -63,6 +63,8 @@ struct nf_ct_frag6_skb_cb
static struct inet_frags nf_frags;
#ifdef CONFIG_SYSCTL
+static int zero;
+
static struct ctl_table nf_ct_frag6_sysctl_table[] = {
{
.procname = "nf_conntrack_frag6_timeout",
@@ -76,14 +78,17 @@ static struct ctl_table nf_ct_frag6_sysctl_table[] = {
.data = &init_net.nf_frag.frags.low_thresh,
.maxlen = sizeof(unsigned int),
.mode = 0644,
- .proc_handler = proc_dointvec,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &init_net.nf_frag.frags.high_thresh
},
{
.procname = "nf_conntrack_frag6_high_thresh",
.data = &init_net.nf_frag.frags.high_thresh,
.maxlen = sizeof(unsigned int),
.mode = 0644,
- .proc_handler = proc_dointvec,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &init_net.nf_frag.frags.low_thresh
},
{ }
};
@@ -102,7 +107,10 @@ static int nf_ct_frag6_sysctl_register(struct net *net)
table[0].data = &net->nf_frag.frags.timeout;
table[1].data = &net->nf_frag.frags.low_thresh;
+ table[1].extra2 = &net->nf_frag.frags.high_thresh;
table[2].data = &net->nf_frag.frags.high_thresh;
+ table[2].extra1 = &net->nf_frag.frags.low_thresh;
+ table[2].extra2 = &init_net.nf_frag.frags.high_thresh;
}
hdr = register_net_sysctl(net, "net/netfilter", table);
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 57a9707b2032..f1709c4a289a 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -578,20 +578,25 @@ static const struct inet6_protocol frag_protocol =
};
#ifdef CONFIG_SYSCTL
+static int zero;
+
static struct ctl_table ip6_frags_ns_ctl_table[] = {
{
.procname = "ip6frag_high_thresh",
.data = &init_net.ipv6.frags.high_thresh,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = proc_dointvec
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &init_net.ipv6.frags.low_thresh
},
{
.procname = "ip6frag_low_thresh",
.data = &init_net.ipv6.frags.low_thresh,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = proc_dointvec
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &init_net.ipv6.frags.high_thresh
},
{
.procname = "ip6frag_time",
@@ -628,7 +633,10 @@ static int __net_init ip6_frags_ns_sysctl_register(struct net *net)
goto err_alloc;
table[0].data = &net->ipv6.frags.high_thresh;
+ table[0].extra1 = &net->ipv6.frags.low_thresh;
+ table[0].extra2 = &init_net.ipv6.frags.high_thresh;
table[1].data = &net->ipv6.frags.low_thresh;
+ table[1].extra2 = &net->ipv6.frags.high_thresh;
table[2].data = &net->ipv6.frags.timeout;
/* Don't export sysctls to unprivileged users */
--
1.9.3
^ permalink raw reply related [flat|nested] 15+ messages in thread* Re: [PATCH net-next 0/9] inet: frag: cleanup and update
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (8 preceding siblings ...)
2014-07-24 14:50 ` [PATCH net-next 9/9] inet: frag: set limits and make init_net's high_thresh limit global Nikolay Aleksandrov
@ 2014-07-28 5:37 ` David Miller
2014-07-28 8:00 ` Florian Westphal
2014-07-28 6:43 ` Alexander Aring
10 siblings, 1 reply; 15+ messages in thread
From: David Miller @ 2014-07-28 5:37 UTC (permalink / raw)
To: nikolay
Cc: netdev, fw, kuznet, jmorris, yoshfuji, kaber, alex.aring,
eric.dumazet
From: Nikolay Aleksandrov <nikolay@redhat.com>
Date: Thu, 24 Jul 2014 16:50:28 +0200
> The end goal of this patchset is to remove the LRU list and to move the
> frag eviction to a work queue. It also does a couple of necessary cleanups
> and fixes. Brief patch descriptions:
Ok I decided to apply this series.
Honestly I thought the frag counter would be useful for diagnosing problems
("are there a lot of frag queues bunched up on my machine?"). But whatever.
You could make it accurate again by just scanning the hash table really
quickly and counting how many entries you see, if you wanted to.
^ permalink raw reply [flat|nested] 15+ messages in thread* Re: [PATCH net-next 0/9] inet: frag: cleanup and update
2014-07-28 5:37 ` [PATCH net-next 0/9] inet: frag: cleanup and update David Miller
@ 2014-07-28 8:00 ` Florian Westphal
0 siblings, 0 replies; 15+ messages in thread
From: Florian Westphal @ 2014-07-28 8:00 UTC (permalink / raw)
To: David Miller
Cc: nikolay, netdev, fw, kuznet, jmorris, yoshfuji, kaber, alex.aring,
eric.dumazet
David Miller <davem@davemloft.net> wrote:
> From: Nikolay Aleksandrov <nikolay@redhat.com>
> Date: Thu, 24 Jul 2014 16:50:28 +0200
>
> > The end goal of this patchset is to remove the LRU list and to move the
> > frag eviction to a work queue. It also does a couple of necessary cleanups
> > and fixes. Brief patch descriptions:
>
> Ok I decided to apply this series.
>
> Honestly I thought the frag counter would be useful for diagnosing problems
> ("are there a lot of frag queues bunched up on my machine?"). But whatever.
>
> You could make it accurate again by just scanning the hash table really
> quickly and counting how many entries you see, if you wanted to.
Right. In case someone reports a bug vs. "bogus frag counter" either
myself or Nik will submit a patch to do this.
Thanks David.
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH net-next 0/9] inet: frag: cleanup and update
2014-07-24 14:50 [PATCH net-next 0/9] inet: frag: cleanup and update Nikolay Aleksandrov
` (9 preceding siblings ...)
2014-07-28 5:37 ` [PATCH net-next 0/9] inet: frag: cleanup and update David Miller
@ 2014-07-28 6:43 ` Alexander Aring
10 siblings, 0 replies; 15+ messages in thread
From: Alexander Aring @ 2014-07-28 6:43 UTC (permalink / raw)
To: Nikolay Aleksandrov
Cc: netdev, Florian Westphal, David S. Miller, Alexey Kuznetsov,
James Morris, Hideaki YOSHIFUJI, Patrick McHardy, Eric Dumazet
On Thu, Jul 24, 2014 at 04:50:28PM +0200, Nikolay Aleksandrov wrote:
> Hello,
> The end goal of this patchset is to remove the LRU list and to move the
> frag eviction to a work queue. It also does a couple of necessary cleanups
> and fixes.
For net/ieee802154/reassembly.c:
Acked-by: Alexander Aring <alex.aring@gmail.com>
^ permalink raw reply [flat|nested] 15+ messages in thread