Commit f2bb4bed authored by David S. Miller's avatar David S. Miller

ipv4: Cache output routes in fib_info nexthops.

If we have an output route that lacks nexthop exceptions, we can cache
it in the FIB info nexthop.

Such routes will have DST_HOST cleared because such routes refer to a
family of destinations, rather than just one.

The sequence of the handling of exceptions during route lookup is
adjusted to make the logic work properly.

Before we allocate the route, we lookup the exception.

Then we know if we will cache this route or not, and therefore whether
DST_HOST should be set on the allocated route.

Then we use DST_HOST to key off whether we should store the resulting
route, during rt_set_nexthop(), in the FIB nexthop cache.

With help from Eric Dumazet.
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent ceb33206
...@@ -46,6 +46,7 @@ struct fib_config { ...@@ -46,6 +46,7 @@ struct fib_config {
}; };
struct fib_info; struct fib_info;
struct rtable;
struct fib_nh_exception { struct fib_nh_exception {
struct fib_nh_exception __rcu *fnhe_next; struct fib_nh_exception __rcu *fnhe_next;
...@@ -80,6 +81,7 @@ struct fib_nh { ...@@ -80,6 +81,7 @@ struct fib_nh {
__be32 nh_gw; __be32 nh_gw;
__be32 nh_saddr; __be32 nh_saddr;
int nh_saddr_genid; int nh_saddr_genid;
struct rtable *nh_rth_output;
struct fnhe_hash_bucket *nh_exceptions; struct fnhe_hash_bucket *nh_exceptions;
}; };
......
...@@ -171,6 +171,8 @@ static void free_fib_info_rcu(struct rcu_head *head) ...@@ -171,6 +171,8 @@ static void free_fib_info_rcu(struct rcu_head *head)
dev_put(nexthop_nh->nh_dev); dev_put(nexthop_nh->nh_dev);
if (nexthop_nh->nh_exceptions) if (nexthop_nh->nh_exceptions)
free_nh_exceptions(nexthop_nh); free_nh_exceptions(nexthop_nh);
if (nexthop_nh->nh_rth_output)
dst_release(&nexthop_nh->nh_rth_output->dst);
} endfor_nexthops(fi); } endfor_nexthops(fi);
release_net(fi->fib_net); release_net(fi->fib_net);
......
...@@ -1158,8 +1158,7 @@ static unsigned int ipv4_mtu(const struct dst_entry *dst) ...@@ -1158,8 +1158,7 @@ static unsigned int ipv4_mtu(const struct dst_entry *dst)
return mtu; return mtu;
} }
static void rt_init_metrics(struct rtable *rt, const struct flowi4 *fl4, static void rt_init_metrics(struct rtable *rt, struct fib_info *fi)
struct fib_info *fi)
{ {
if (fi->fib_metrics != (u32 *) dst_default_metrics) { if (fi->fib_metrics != (u32 *) dst_default_metrics) {
rt->fi = fi; rt->fi = fi;
...@@ -1168,22 +1167,34 @@ static void rt_init_metrics(struct rtable *rt, const struct flowi4 *fl4, ...@@ -1168,22 +1167,34 @@ static void rt_init_metrics(struct rtable *rt, const struct flowi4 *fl4,
dst_init_metrics(&rt->dst, fi->fib_metrics, true); dst_init_metrics(&rt->dst, fi->fib_metrics, true);
} }
static void rt_bind_exception(struct rtable *rt, struct fib_nh *nh, __be32 daddr) static struct fib_nh_exception *find_exception(struct fib_nh *nh, __be32 daddr)
{ {
struct fnhe_hash_bucket *hash = nh->nh_exceptions; struct fnhe_hash_bucket *hash = nh->nh_exceptions;
struct fib_nh_exception *fnhe; struct fib_nh_exception *fnhe;
u32 hval; u32 hval;
if (!hash)
return NULL;
hval = fnhe_hashfun(daddr); hval = fnhe_hashfun(daddr);
restart:
for (fnhe = rcu_dereference(hash[hval].chain); fnhe; for (fnhe = rcu_dereference(hash[hval].chain); fnhe;
fnhe = rcu_dereference(fnhe->fnhe_next)) { fnhe = rcu_dereference(fnhe->fnhe_next)) {
if (fnhe->fnhe_daddr == daddr)
return fnhe;
}
return NULL;
}
static void rt_bind_exception(struct rtable *rt, struct fib_nh_exception *fnhe,
__be32 daddr)
{
__be32 fnhe_daddr, gw; __be32 fnhe_daddr, gw;
unsigned long expires; unsigned long expires;
unsigned int seq; unsigned int seq;
u32 pmtu; u32 pmtu;
restart:
seq = read_seqbegin(&fnhe_seqlock); seq = read_seqbegin(&fnhe_seqlock);
fnhe_daddr = fnhe->fnhe_daddr; fnhe_daddr = fnhe->fnhe_daddr;
gw = fnhe->fnhe_gw; gw = fnhe->fnhe_gw;
...@@ -1191,8 +1202,10 @@ static void rt_bind_exception(struct rtable *rt, struct fib_nh *nh, __be32 daddr ...@@ -1191,8 +1202,10 @@ static void rt_bind_exception(struct rtable *rt, struct fib_nh *nh, __be32 daddr
expires = fnhe->fnhe_expires; expires = fnhe->fnhe_expires;
if (read_seqretry(&fnhe_seqlock, seq)) if (read_seqretry(&fnhe_seqlock, seq))
goto restart; goto restart;
if (daddr != fnhe_daddr) if (daddr != fnhe_daddr)
continue; return;
if (pmtu) { if (pmtu) {
unsigned long diff = expires - jiffies; unsigned long diff = expires - jiffies;
...@@ -1206,12 +1219,31 @@ static void rt_bind_exception(struct rtable *rt, struct fib_nh *nh, __be32 daddr ...@@ -1206,12 +1219,31 @@ static void rt_bind_exception(struct rtable *rt, struct fib_nh *nh, __be32 daddr
rt->rt_gateway = gw; rt->rt_gateway = gw;
} }
fnhe->fnhe_stamp = jiffies; fnhe->fnhe_stamp = jiffies;
break; }
static inline void rt_release_rcu(struct rcu_head *head)
{
struct dst_entry *dst = container_of(head, struct dst_entry, rcu_head);
dst_release(dst);
}
static void rt_cache_route(struct fib_nh *nh, struct rtable *rt)
{
struct rtable *orig, *prev, **p = &nh->nh_rth_output;
orig = *p;
prev = cmpxchg(p, orig, rt);
if (prev == orig) {
dst_clone(&rt->dst);
if (orig)
call_rcu_bh(&orig->dst.rcu_head, rt_release_rcu);
} }
} }
static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4, static void rt_set_nexthop(struct rtable *rt, __be32 daddr,
const struct fib_result *res, const struct fib_result *res,
struct fib_nh_exception *fnhe,
struct fib_info *fi, u16 type, u32 itag) struct fib_info *fi, u16 type, u32 itag)
{ {
if (fi) { if (fi) {
...@@ -1219,12 +1251,15 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4, ...@@ -1219,12 +1251,15 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4,
if (nh->nh_gw && nh->nh_scope == RT_SCOPE_LINK) if (nh->nh_gw && nh->nh_scope == RT_SCOPE_LINK)
rt->rt_gateway = nh->nh_gw; rt->rt_gateway = nh->nh_gw;
if (unlikely(nh->nh_exceptions)) if (unlikely(fnhe))
rt_bind_exception(rt, nh, fl4->daddr); rt_bind_exception(rt, fnhe, daddr);
rt_init_metrics(rt, fl4, fi); rt_init_metrics(rt, fi);
#ifdef CONFIG_IP_ROUTE_CLASSID #ifdef CONFIG_IP_ROUTE_CLASSID
rt->dst.tclassid = FIB_RES_NH(*res).nh_tclassid; rt->dst.tclassid = nh->nh_tclassid;
#endif #endif
if (!(rt->dst.flags & DST_HOST) &&
rt_is_output_route(rt))
rt_cache_route(nh, rt);
} }
#ifdef CONFIG_IP_ROUTE_CLASSID #ifdef CONFIG_IP_ROUTE_CLASSID
...@@ -1236,10 +1271,10 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4, ...@@ -1236,10 +1271,10 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4,
} }
static struct rtable *rt_dst_alloc(struct net_device *dev, static struct rtable *rt_dst_alloc(struct net_device *dev,
bool nopolicy, bool noxfrm) bool nopolicy, bool noxfrm, bool will_cache)
{ {
return dst_alloc(&ipv4_dst_ops, dev, 1, DST_OBSOLETE_FORCE_CHK, return dst_alloc(&ipv4_dst_ops, dev, 1, DST_OBSOLETE_FORCE_CHK,
DST_HOST | DST_NOCACHE | (will_cache ? 0 : DST_HOST) | DST_NOCACHE |
(nopolicy ? DST_NOPOLICY : 0) | (nopolicy ? DST_NOPOLICY : 0) |
(noxfrm ? DST_NOXFRM : 0)); (noxfrm ? DST_NOXFRM : 0));
} }
...@@ -1276,7 +1311,7 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr, ...@@ -1276,7 +1311,7 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
goto e_err; goto e_err;
} }
rth = rt_dst_alloc(dev_net(dev)->loopback_dev, rth = rt_dst_alloc(dev_net(dev)->loopback_dev,
IN_DEV_CONF_GET(in_dev, NOPOLICY), false); IN_DEV_CONF_GET(in_dev, NOPOLICY), false, false);
if (!rth) if (!rth)
goto e_nobufs; goto e_nobufs;
...@@ -1349,6 +1384,7 @@ static int __mkroute_input(struct sk_buff *skb, ...@@ -1349,6 +1384,7 @@ static int __mkroute_input(struct sk_buff *skb,
__be32 daddr, __be32 saddr, u32 tos, __be32 daddr, __be32 saddr, u32 tos,
struct rtable **result) struct rtable **result)
{ {
struct fib_nh_exception *fnhe;
struct rtable *rth; struct rtable *rth;
int err; int err;
struct in_device *out_dev; struct in_device *out_dev;
...@@ -1395,9 +1431,13 @@ static int __mkroute_input(struct sk_buff *skb, ...@@ -1395,9 +1431,13 @@ static int __mkroute_input(struct sk_buff *skb,
} }
} }
fnhe = NULL;
if (res->fi)
fnhe = find_exception(&FIB_RES_NH(*res), daddr);
rth = rt_dst_alloc(out_dev->dev, rth = rt_dst_alloc(out_dev->dev,
IN_DEV_CONF_GET(in_dev, NOPOLICY), IN_DEV_CONF_GET(in_dev, NOPOLICY),
IN_DEV_CONF_GET(out_dev, NOXFRM)); IN_DEV_CONF_GET(out_dev, NOXFRM), false);
if (!rth) { if (!rth) {
err = -ENOBUFS; err = -ENOBUFS;
goto cleanup; goto cleanup;
...@@ -1416,7 +1456,7 @@ static int __mkroute_input(struct sk_buff *skb, ...@@ -1416,7 +1456,7 @@ static int __mkroute_input(struct sk_buff *skb,
rth->dst.input = ip_forward; rth->dst.input = ip_forward;
rth->dst.output = ip_output; rth->dst.output = ip_output;
rt_set_nexthop(rth, NULL, res, res->fi, res->type, itag); rt_set_nexthop(rth, daddr, res, fnhe, res->fi, res->type, itag);
*result = rth; *result = rth;
err = 0; err = 0;
...@@ -1558,7 +1598,7 @@ out: return err; ...@@ -1558,7 +1598,7 @@ out: return err;
local_input: local_input:
rth = rt_dst_alloc(net->loopback_dev, rth = rt_dst_alloc(net->loopback_dev,
IN_DEV_CONF_GET(in_dev, NOPOLICY), false); IN_DEV_CONF_GET(in_dev, NOPOLICY), false, false);
if (!rth) if (!rth)
goto e_nobufs; goto e_nobufs;
...@@ -1672,6 +1712,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res, ...@@ -1672,6 +1712,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
unsigned int flags) unsigned int flags)
{ {
struct fib_info *fi = res->fi; struct fib_info *fi = res->fi;
struct fib_nh_exception *fnhe;
struct in_device *in_dev; struct in_device *in_dev;
u16 type = res->type; u16 type = res->type;
struct rtable *rth; struct rtable *rth;
...@@ -1710,9 +1751,22 @@ static struct rtable *__mkroute_output(const struct fib_result *res, ...@@ -1710,9 +1751,22 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
fi = NULL; fi = NULL;
} }
fnhe = NULL;
if (fi) {
fnhe = find_exception(&FIB_RES_NH(*res), fl4->daddr);
if (!fnhe) {
rth = FIB_RES_NH(*res).nh_rth_output;
if (rth &&
rth->dst.obsolete == DST_OBSOLETE_FORCE_CHK) {
dst_use(&rth->dst, jiffies);
return rth;
}
}
}
rth = rt_dst_alloc(dev_out, rth = rt_dst_alloc(dev_out,
IN_DEV_CONF_GET(in_dev, NOPOLICY), IN_DEV_CONF_GET(in_dev, NOPOLICY),
IN_DEV_CONF_GET(in_dev, NOXFRM)); IN_DEV_CONF_GET(in_dev, NOXFRM),
fi && !fnhe);
if (!rth) if (!rth)
return ERR_PTR(-ENOBUFS); return ERR_PTR(-ENOBUFS);
...@@ -1749,7 +1803,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res, ...@@ -1749,7 +1803,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
#endif #endif
} }
rt_set_nexthop(rth, fl4, res, fi, type, 0); rt_set_nexthop(rth, fl4->daddr, res, fnhe, fi, type, 0);
if (fl4->flowi4_flags & FLOWI_FLAG_RT_NOCACHE) if (fl4->flowi4_flags & FLOWI_FLAG_RT_NOCACHE)
rth->dst.flags |= DST_NOCACHE; rth->dst.flags |= DST_NOCACHE;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment