* [PATCH v4 RFC 4/8] netlink: Add NLM_F_NOREPLY flags.
From: David Miller @ 2011-03-13 0:36 UTC (permalink / raw)
To: netdev
It is used to bypass the response generation for "GET" type
operations.
This mechanism can then be used to allow a query operation to be used
in a "benchmarking" role, since the actual overhead of the netlink
communications will be muted.
The first query to support this is inet_rtm_getroute().
Signed-off-by: David S. Miller <davem@davemloft.net>
---
include/linux/netlink.h | 2 ++
net/ipv4/route.c | 3 +++
2 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/include/linux/netlink.h b/include/linux/netlink.h
index 4c4ac3f..a63c0e5 100644
--- a/include/linux/netlink.h
+++ b/include/linux/netlink.h
@@ -61,6 +61,8 @@ struct nlmsghdr {
#define NLM_F_CREATE 0x400 /* Create, if it does not exist */
#define NLM_F_APPEND 0x800 /* Add to end of list */
+#define NLM_F_NOREPLY 0x1000 /* Do not generate reply */
+
/*
4.4BSD ADD NLM_F_CREATE|NLM_F_EXCL
4.4BSD CHANGE NLM_F_REPLACE
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 6036476..96dd498 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -2130,6 +2130,9 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
goto errout_free;
skb_dst_set(skb, &rt->dst);
+ if (nlh->nlmsg_flags & NLM_F_NOREPLY)
+ goto errout_free;
+
if (rtm->rtm_flags & RTM_F_NOTIFY)
rt->rt_flags |= RTCF_NOTIFY;
--
1.7.4.1
^ permalink raw reply related
* [PATCH v4 RFC 5/8] net: Make dst_alloc() take more explicit initializations.
From: David Miller @ 2011-03-13 0:37 UTC (permalink / raw)
To: netdev
Now the dst->dev, dev->obsolete, and dst->flags values can
be specified as well.
Signed-off-by: David S. Miller <davem@davemloft.net>
---
include/net/dst.h | 3 ++-
net/core/dst.c | 18 +++++++++++++-----
net/decnet/dn_route.c | 13 ++-----------
net/ipv4/route.c | 48 +++++++++++++++++++-----------------------------
net/ipv6/route.c | 29 +++++++++++------------------
net/xfrm/xfrm_policy.c | 2 +-
6 files changed, 48 insertions(+), 65 deletions(-)
diff --git a/include/net/dst.h b/include/net/dst.h
index 2a46cba..d9b4e78 100644
--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -352,7 +352,8 @@ static inline struct dst_entry *skb_dst_pop(struct sk_buff *skb)
}
extern int dst_discard(struct sk_buff *skb);
-extern void *dst_alloc(struct dst_ops * ops, int initial_ref);
+extern void *dst_alloc(struct dst_ops * ops, struct net_device *dev,
+ int initial_ref, int initial_obsolete, int flags);
extern void __dst_free(struct dst_entry * dst);
extern struct dst_entry *dst_destroy(struct dst_entry * dst);
diff --git a/net/core/dst.c b/net/core/dst.c
index 91104d3..9505778 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -166,7 +166,8 @@ EXPORT_SYMBOL(dst_discard);
const u32 dst_default_metrics[RTAX_MAX];
-void *dst_alloc(struct dst_ops *ops, int initial_ref)
+void *dst_alloc(struct dst_ops *ops, struct net_device *dev,
+ int initial_ref, int initial_obsolete, int flags)
{
struct dst_entry *dst;
@@ -177,12 +178,19 @@ void *dst_alloc(struct dst_ops *ops, int initial_ref)
dst = kmem_cache_zalloc(ops->kmem_cachep, GFP_ATOMIC);
if (!dst)
return NULL;
- atomic_set(&dst->__refcnt, initial_ref);
dst->ops = ops;
- dst->lastuse = jiffies;
- dst->path = dst;
- dst->input = dst->output = dst_discard;
+ dst->dev = dev;
+ if (dev)
+ dev_hold(dev);
dst_init_metrics(dst, dst_default_metrics, true);
+ dst->path = dst;
+ dst->input = dst_discard;
+ dst->output = dst_discard;
+
+ dst->obsolete = initial_obsolete;
+ atomic_set(&dst->__refcnt, initial_ref);
+ dst->lastuse = jiffies;
+ dst->flags = flags;
#if RT_CACHE_DEBUG >= 2
atomic_inc(&dst_total);
#endif
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index 9f09d4f..f489b08 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -1125,13 +1125,10 @@ make_route:
if (dev_out->flags & IFF_LOOPBACK)
flags |= RTCF_LOCAL;
- rt = dst_alloc(&dn_dst_ops, 0);
+ rt = dst_alloc(&dn_dst_ops, dev_out, 1, 0, DST_HOST);
if (rt == NULL)
goto e_nobufs;
- atomic_set(&rt->dst.__refcnt, 1);
- rt->dst.flags = DST_HOST;
-
rt->fld.saddr = oldflp->saddr;
rt->fld.daddr = oldflp->daddr;
rt->fld.flowidn_oif = oldflp->flowidn_oif;
@@ -1146,8 +1143,6 @@ make_route:
rt->rt_dst_map = fld.daddr;
rt->rt_src_map = fld.saddr;
- rt->dst.dev = dev_out;
- dev_hold(dev_out);
rt->dst.neighbour = neigh;
neigh = NULL;
@@ -1399,7 +1394,7 @@ static int dn_route_input_slow(struct sk_buff *skb)
}
make_route:
- rt = dst_alloc(&dn_dst_ops, 0);
+ rt = dst_alloc(&dn_dst_ops, out_dev, 0, 0, DST_HOST);
if (rt == NULL)
goto e_nobufs;
@@ -1419,9 +1414,7 @@ make_route:
rt->fld.flowidn_iif = in_dev->ifindex;
rt->fld.flowidn_mark = fld.flowidn_mark;
- rt->dst.flags = DST_HOST;
rt->dst.neighbour = neigh;
- rt->dst.dev = out_dev;
rt->dst.lastuse = jiffies;
rt->dst.output = dn_rt_bug;
switch(res.type) {
@@ -1440,8 +1433,6 @@ make_route:
rt->dst.input = dst_discard;
}
rt->rt_flags = flags;
- if (rt->dst.dev)
- dev_hold(rt->dst.dev);
err = dn_rt_set_next_hop(rt, &res);
if (err)
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 96dd498..d0c78ec 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1103,21 +1103,17 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *oldflp4,
rt->rt_type = type;
}
-static struct rtable *rt_dst_alloc(bool nopolicy, bool noxfrm)
+static struct rtable *rt_dst_alloc(struct net_device *dev,
+ bool nopolicy, bool noxfrm)
{
- struct rtable *rt = dst_alloc(&ipv4_dst_ops, 1);
- if (rt) {
- rt->dst.obsolete = -1;
-
- /* To avoid expensive rcu stuff for this uncached dst, we set
- * DST_NOCACHE so that dst_release() can free dst without
- * waiting a grace period.
- */
- rt->dst.flags = DST_NOCACHE | DST_HOST |
- (nopolicy ? DST_NOPOLICY : 0) |
- (noxfrm ? DST_NOXFRM : 0);
- }
- return rt;
+ /* To avoid expensive rcu stuff for this uncached dst, we set
+ * DST_NOCACHE so that dst_release() can free dst without
+ * waiting a grace period.
+ */
+ return dst_alloc(&ipv4_dst_ops, dev, 1, -1,
+ DST_NOCACHE | DST_HOST |
+ (nopolicy ? DST_NOPOLICY : 0) |
+ (noxfrm ? DST_NOXFRM : 0));
}
/* called in rcu_read_lock() section */
@@ -1149,7 +1145,8 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
if (err < 0)
goto e_err;
}
- rth = rt_dst_alloc(IN_DEV_CONF_GET(in_dev, NOPOLICY), false);
+ rth = rt_dst_alloc(init_net.loopback_dev,
+ IN_DEV_CONF_GET(in_dev, NOPOLICY), false);
if (!rth)
goto e_nobufs;
@@ -1165,8 +1162,6 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
rth->dst.tclassid = itag;
#endif
rth->rt_iif = dev->ifindex;
- rth->dst.dev = init_net.loopback_dev;
- dev_hold(rth->dst.dev);
rth->rt_oif = 0;
rth->rt_gateway = daddr;
rth->rt_spec_dst= spec_dst;
@@ -1284,7 +1279,8 @@ static int __mkroute_input(struct sk_buff *skb,
}
}
- rth = rt_dst_alloc(IN_DEV_CONF_GET(in_dev, NOPOLICY),
+ rth = rt_dst_alloc(out_dev->dev,
+ IN_DEV_CONF_GET(in_dev, NOPOLICY),
IN_DEV_CONF_GET(out_dev, NOXFRM));
if (!rth) {
err = -ENOBUFS;
@@ -1299,8 +1295,6 @@ static int __mkroute_input(struct sk_buff *skb,
rth->rt_src = saddr;
rth->rt_gateway = daddr;
rth->rt_iif = in_dev->dev->ifindex;
- rth->dst.dev = (out_dev)->dev;
- dev_hold(rth->dst.dev);
rth->rt_oif = 0;
rth->rt_spec_dst= spec_dst;
@@ -1453,7 +1447,8 @@ brd_input:
RT_CACHE_STAT_INC(in_brd);
local_input:
- rth = rt_dst_alloc(IN_DEV_CONF_GET(in_dev, NOPOLICY), false);
+ rth = rt_dst_alloc(net->loopback_dev,
+ IN_DEV_CONF_GET(in_dev, NOPOLICY), false);
if (!rth)
goto e_nobufs;
@@ -1470,8 +1465,6 @@ local_input:
rth->dst.tclassid = itag;
#endif
rth->rt_iif = dev->ifindex;
- rth->dst.dev = net->loopback_dev;
- dev_hold(rth->dst.dev);
rth->rt_gateway = daddr;
rth->rt_spec_dst= spec_dst;
rth->dst.input= ip_local_deliver;
@@ -1618,7 +1611,8 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
fi = NULL;
}
- rth = rt_dst_alloc(IN_DEV_CONF_GET(in_dev, NOPOLICY),
+ rth = rt_dst_alloc(dev_out,
+ IN_DEV_CONF_GET(in_dev, NOPOLICY),
IN_DEV_CONF_GET(in_dev, NOXFRM));
if (!rth)
return ERR_PTR(-ENOBUFS);
@@ -1631,10 +1625,6 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
rth->rt_dst = fl4->daddr;
rth->rt_src = fl4->saddr;
rth->rt_iif = 0;
- /* get references to the devices that are to be hold by the routing
- cache entry */
- rth->dst.dev = dev_out;
- dev_hold(dev_out);
rth->rt_gateway = fl4->daddr;
rth->rt_spec_dst= fl4->saddr;
@@ -1885,7 +1875,7 @@ static struct dst_ops ipv4_dst_blackhole_ops = {
struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig)
{
- struct rtable *rt = dst_alloc(&ipv4_dst_blackhole_ops, 1);
+ struct rtable *rt = dst_alloc(&ipv4_dst_blackhole_ops, NULL, 1, 0, 0);
struct rtable *ort = (struct rtable *) dst_orig;
if (rt) {
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 6814c87..018db5d 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -220,9 +220,10 @@ static struct rt6_info ip6_blk_hole_entry_template = {
#endif
/* allocate dst with ip6_dst_ops */
-static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops)
+static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops,
+ struct net_device *dev)
{
- return (struct rt6_info *)dst_alloc(ops, 0);
+ return (struct rt6_info *)dst_alloc(ops, dev, 0, 0, 0);
}
static void ip6_dst_destroy(struct dst_entry *dst)
@@ -874,10 +875,10 @@ EXPORT_SYMBOL(ip6_route_output);
struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
{
- struct rt6_info *rt = dst_alloc(&ip6_dst_blackhole_ops, 1);
- struct rt6_info *ort = (struct rt6_info *) dst_orig;
+ struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
struct dst_entry *new = NULL;
+ rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, 0, 0);
if (rt) {
new = &rt->dst;
@@ -886,9 +887,6 @@ struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_ori
new->output = dst_discard;
dst_copy_metrics(new, &ort->dst);
- new->dev = ort->dst.dev;
- if (new->dev)
- dev_hold(new->dev);
rt->rt6i_idev = ort->rt6i_idev;
if (rt->rt6i_idev)
in6_dev_hold(rt->rt6i_idev);
@@ -1031,13 +1029,12 @@ struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
if (unlikely(idev == NULL))
return NULL;
- rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
+ rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops, dev);
if (unlikely(rt == NULL)) {
in6_dev_put(idev);
goto out;
}
- dev_hold(dev);
if (neigh)
neigh_hold(neigh);
else {
@@ -1046,7 +1043,6 @@ struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
neigh = NULL;
}
- rt->rt6i_dev = dev;
rt->rt6i_idev = idev;
rt->rt6i_nexthop = neigh;
atomic_set(&rt->dst.__refcnt, 1);
@@ -1205,7 +1201,7 @@ int ip6_route_add(struct fib6_config *cfg)
goto out;
}
- rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
+ rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops, NULL);
if (rt == NULL) {
err = -ENOMEM;
@@ -1714,7 +1710,8 @@ void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
{
struct net *net = dev_net(ort->rt6i_dev);
- struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
+ struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops,
+ ort->dst.dev);
if (rt) {
rt->dst.input = ort->dst.input;
@@ -1722,9 +1719,6 @@ static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
dst_copy_metrics(&rt->dst, &ort->dst);
rt->dst.error = ort->dst.error;
- rt->dst.dev = ort->dst.dev;
- if (rt->dst.dev)
- dev_hold(rt->dst.dev);
rt->rt6i_idev = ort->rt6i_idev;
if (rt->rt6i_idev)
in6_dev_hold(rt->rt6i_idev);
@@ -1994,7 +1988,8 @@ struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
int anycast)
{
struct net *net = dev_net(idev->dev);
- struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
+ struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops,
+ net->loopback_dev);
struct neighbour *neigh;
if (rt == NULL) {
@@ -2004,13 +1999,11 @@ struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
return ERR_PTR(-ENOMEM);
}
- dev_hold(net->loopback_dev);
in6_dev_hold(idev);
rt->dst.flags = DST_HOST;
rt->dst.input = ip6_input;
rt->dst.output = ip6_output;
- rt->rt6i_dev = net->loopback_dev;
rt->rt6i_idev = idev;
dst_metric_set(&rt->dst, RTAX_HOPLIMIT, -1);
rt->dst.obsolete = -1;
diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
index 1ba0258..2404b53 100644
--- a/net/xfrm/xfrm_policy.c
+++ b/net/xfrm/xfrm_policy.c
@@ -1348,7 +1348,7 @@ static inline struct xfrm_dst *xfrm_alloc_dst(struct net *net, int family)
default:
BUG();
}
- xdst = dst_alloc(dst_ops, 0);
+ xdst = dst_alloc(dst_ops, NULL, 0, 0, 0);
xfrm_policy_put_afinfo(afinfo);
if (likely(xdst))
--
1.7.4.1
^ permalink raw reply related
* [PATCH v4 RFC 6/8] net: Use non-zero allocations in dst_alloc().
From: David Miller @ 2011-03-13 0:37 UTC (permalink / raw)
To: netdev
Make dst_alloc() and it's users explicitly initialize the entire
entry.
The zero'ing done by kmem_cache_zalloc() was almost entirely
redundant.
Signed-off-by: David S. Miller <davem@davemloft.net>
---
net/core/dst.c | 20 ++++++++++--
net/decnet/dn_route.c | 2 +
net/ipv4/route.c | 81 ++++++++++++++++++++++++++++--------------------
net/ipv6/route.c | 8 ++++-
net/xfrm/xfrm_policy.c | 1 +
5 files changed, 74 insertions(+), 38 deletions(-)
diff --git a/net/core/dst.c b/net/core/dst.c
index 9505778..30f0093 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -175,22 +175,36 @@ void *dst_alloc(struct dst_ops *ops, struct net_device *dev,
if (ops->gc(ops))
return NULL;
}
- dst = kmem_cache_zalloc(ops->kmem_cachep, GFP_ATOMIC);
+ dst = kmem_cache_alloc(ops->kmem_cachep, GFP_ATOMIC);
if (!dst)
return NULL;
- dst->ops = ops;
+ dst->child = NULL;
dst->dev = dev;
if (dev)
dev_hold(dev);
+ dst->ops = ops;
dst_init_metrics(dst, dst_default_metrics, true);
+ dst->expires = 0UL;
dst->path = dst;
+ dst->neighbour = NULL;
+ dst->hh = NULL;
+#ifdef CONFIG_XFRM
+ dst->xfrm = NULL;
+#endif
dst->input = dst_discard;
dst->output = dst_discard;
-
+ dst->error = 0;
dst->obsolete = initial_obsolete;
+ dst->header_len = 0;
+ dst->trailer_len = 0;
+#ifdef CONFIG_IP_ROUTE_CLASSID
+ dst->tclassid = 0;
+#endif
atomic_set(&dst->__refcnt, initial_ref);
+ dst->__use = 0;
dst->lastuse = jiffies;
dst->flags = flags;
+ dst->next = NULL;
#if RT_CACHE_DEBUG >= 2
atomic_inc(&dst_total);
#endif
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index f489b08..74544bc 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -1129,6 +1129,7 @@ make_route:
if (rt == NULL)
goto e_nobufs;
+ memset(&rt->fld, 0, sizeof(rt->fld));
rt->fld.saddr = oldflp->saddr;
rt->fld.daddr = oldflp->daddr;
rt->fld.flowidn_oif = oldflp->flowidn_oif;
@@ -1398,6 +1399,7 @@ make_route:
if (rt == NULL)
goto e_nobufs;
+ memset(&rt->fld, 0, sizeof(rt->fld));
rt->rt_saddr = fld.saddr;
rt->rt_daddr = fld.daddr;
rt->rt_gateway = fld.daddr;
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index d0c78ec..0b02eb3 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1100,7 +1100,6 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *oldflp4,
#endif
set_class_tag(rt, itag);
#endif
- rt->rt_type = type;
}
static struct rtable *rt_dst_alloc(struct net_device *dev,
@@ -1150,24 +1149,27 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
if (!rth)
goto e_nobufs;
+#ifdef CONFIG_IP_ROUTE_CLASSID
+ rth->dst.tclassid = itag;
+#endif
rth->dst.output = ip_rt_bug;
rth->rt_key_dst = daddr;
- rth->rt_dst = daddr;
- rth->rt_tos = tos;
- rth->rt_mark = skb->mark;
rth->rt_key_src = saddr;
+ rth->rt_genid = rt_genid(dev_net(dev));
+ rth->rt_flags = RTCF_MULTICAST;
+ rth->rt_type = RTN_MULTICAST;
+ rth->rt_tos = tos;
+ rth->rt_dst = daddr;
rth->rt_src = saddr;
-#ifdef CONFIG_IP_ROUTE_CLASSID
- rth->dst.tclassid = itag;
-#endif
rth->rt_iif = dev->ifindex;
rth->rt_oif = 0;
+ rth->rt_mark = skb->mark;
rth->rt_gateway = daddr;
rth->rt_spec_dst= spec_dst;
- rth->rt_genid = rt_genid(dev_net(dev));
- rth->rt_flags = RTCF_MULTICAST;
- rth->rt_type = RTN_MULTICAST;
+ rth->rt_peer_genid = 0;
+ rth->peer = NULL;
+ rth->fi = NULL;
if (our) {
rth->dst.input= ip_local_deliver;
rth->rt_flags |= RTCF_LOCAL;
@@ -1288,24 +1290,27 @@ static int __mkroute_input(struct sk_buff *skb,
}
rth->rt_key_dst = daddr;
- rth->rt_dst = daddr;
- rth->rt_tos = tos;
- rth->rt_mark = skb->mark;
rth->rt_key_src = saddr;
+ rth->rt_genid = rt_genid(dev_net(rth->dst.dev));
+ rth->rt_flags = flags;
+ rth->rt_type = res->type;
+ rth->rt_tos = tos;
+ rth->rt_dst = daddr;
rth->rt_src = saddr;
- rth->rt_gateway = daddr;
rth->rt_iif = in_dev->dev->ifindex;
rth->rt_oif = 0;
+ rth->rt_mark = skb->mark;
+ rth->rt_gateway = daddr;
rth->rt_spec_dst= spec_dst;
+ rth->rt_peer_genid = 0;
+ rth->peer = NULL;
+ rth->fi = NULL;
rth->dst.input = ip_forward;
rth->dst.output = ip_output;
- rth->rt_genid = rt_genid(dev_net(rth->dst.dev));
rt_set_nexthop(rth, NULL, res, res->fi, res->type, itag);
- rth->rt_flags = flags;
-
*result = rth;
err = 0;
cleanup:
@@ -1452,29 +1457,33 @@ local_input:
if (!rth)
goto e_nobufs;
+ rth->dst.input= ip_local_deliver;
rth->dst.output= ip_rt_bug;
- rth->rt_genid = rt_genid(net);
+#ifdef CONFIG_IP_ROUTE_CLASSID
+ rth->dst.tclassid = itag;
+#endif
rth->rt_key_dst = daddr;
- rth->rt_dst = daddr;
- rth->rt_tos = tos;
- rth->rt_mark = skb->mark;
rth->rt_key_src = saddr;
+ rth->rt_genid = rt_genid(net);
+ rth->rt_flags = flags|RTCF_LOCAL;
+ rth->rt_type = res.type;
+ rth->rt_tos = tos;
+ rth->rt_dst = daddr;
rth->rt_src = saddr;
-#ifdef CONFIG_IP_ROUTE_CLASSID
- rth->dst.tclassid = itag;
-#endif
rth->rt_iif = dev->ifindex;
+ rth->rt_oif = 0;
+ rth->rt_mark = skb->mark;
rth->rt_gateway = daddr;
rth->rt_spec_dst= spec_dst;
- rth->dst.input= ip_local_deliver;
- rth->rt_flags = flags|RTCF_LOCAL;
+ rth->rt_peer_genid = 0;
+ rth->peer = NULL;
+ rth->fi = NULL;
if (res.type == RTN_UNREACHABLE) {
rth->dst.input= ip_error;
rth->dst.error= -err;
rth->rt_flags &= ~RTCF_LOCAL;
}
- rth->rt_type = res.type;
rth = rt_finalize(rth, skb);
err = 0;
if (IS_ERR(rth))
@@ -1617,19 +1626,24 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
if (!rth)
return ERR_PTR(-ENOBUFS);
+ rth->dst.output = ip_output;
+
rth->rt_key_dst = oldflp4->daddr;
- rth->rt_tos = tos;
rth->rt_key_src = oldflp4->saddr;
- rth->rt_oif = oldflp4->flowi4_oif;
- rth->rt_mark = oldflp4->flowi4_mark;
+ rth->rt_genid = rt_genid(dev_net(dev_out));
+ rth->rt_flags = flags;
+ rth->rt_type = type;
+ rth->rt_tos = tos;
rth->rt_dst = fl4->daddr;
rth->rt_src = fl4->saddr;
rth->rt_iif = 0;
+ rth->rt_oif = oldflp4->flowi4_oif;
+ rth->rt_mark = oldflp4->flowi4_mark;
rth->rt_gateway = fl4->daddr;
rth->rt_spec_dst= fl4->saddr;
-
- rth->dst.output=ip_output;
- rth->rt_genid = rt_genid(dev_net(dev_out));
+ rth->rt_peer_genid = 0;
+ rth->peer = NULL;
+ rth->fi = NULL;
RT_CACHE_STAT_INC(out_slow_tot);
@@ -1657,7 +1671,6 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
rt_set_nexthop(rth, oldflp4, res, fi, type, 0);
- rth->rt_flags = flags;
return rth;
}
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 018db5d..d434cd5 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -223,7 +223,11 @@ static struct rt6_info ip6_blk_hole_entry_template = {
static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops,
struct net_device *dev)
{
- return (struct rt6_info *)dst_alloc(ops, dev, 0, 0, 0);
+ struct rt6_info *rt = dst_alloc(ops, dev, 0, 0, 0);
+
+ memset(&rt->rt6i_table, 0, sizeof(*rt) - sizeof(struct dst_entry));
+
+ return rt;
}
static void ip6_dst_destroy(struct dst_entry *dst)
@@ -880,6 +884,8 @@ struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_ori
rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, 0, 0);
if (rt) {
+ memset(&rt->rt6i_table, 0, sizeof(*rt) - sizeof(struct dst_entry));
+
new = &rt->dst;
new->__use = 1;
diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
index 2404b53..7a028ad 100644
--- a/net/xfrm/xfrm_policy.c
+++ b/net/xfrm/xfrm_policy.c
@@ -1349,6 +1349,7 @@ static inline struct xfrm_dst *xfrm_alloc_dst(struct net *net, int family)
BUG();
}
xdst = dst_alloc(dst_ops, NULL, 0, 0, 0);
+ memset(&xdst->u.rt6.rt6i_table, 0, sizeof(*xdst) - sizeof(struct dst_entry));
xfrm_policy_put_afinfo(afinfo);
if (likely(xdst))
--
1.7.4.1
^ permalink raw reply related
* [PATCH v4 RFC 7/8] ipv4: Kill rt_key_{src,dst} from struct rtable.
From: David Miller @ 2011-03-13 0:37 UTC (permalink / raw)
To: netdev
They are always used in contexts where they can be reconstituted,
or where the finally resolved rt->rt_{src,dst} is semantically
equivalent.
Signed-off-by: David S. Miller <davem@davemloft.net>
---
include/net/route.h | 8 ++------
net/ipv4/ipmr.c | 4 ++--
net/ipv4/route.c | 24 +++++++-----------------
net/ipv4/xfrm4_policy.c | 2 --
4 files changed, 11 insertions(+), 27 deletions(-)
diff --git a/include/net/route.h b/include/net/route.h
index 78e8ce1..154e589 100644
--- a/include/net/route.h
+++ b/include/net/route.h
@@ -53,10 +53,6 @@ struct fib_info;
struct rtable {
struct dst_entry dst;
- /* Lookup key. */
- __be32 rt_key_dst;
- __be32 rt_key_src;
-
int rt_genid;
unsigned rt_flags;
__u16 rt_type;
@@ -256,8 +252,8 @@ static inline struct rtable *ip_route_newports(struct rtable *rt,
struct flowi4 fl4 = {
.flowi4_oif = rt->rt_oif,
.flowi4_mark = rt->rt_mark,
- .daddr = rt->rt_key_dst,
- .saddr = rt->rt_key_src,
+ .daddr = rt->rt_dst,
+ .saddr = rt->rt_src,
.flowi4_tos = rt->rt_tos,
.flowi4_proto = protocol,
.fl4_sport = sport,
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index 1f62eae..0441b26 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -1791,8 +1791,8 @@ dont_forward:
static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct rtable *rt)
{
struct flowi4 fl4 = {
- .daddr = rt->rt_key_dst,
- .saddr = rt->rt_key_src,
+ .daddr = rt->rt_dst,
+ .saddr = rt->rt_src,
.flowi4_tos = rt->rt_tos,
.flowi4_oif = rt->rt_oif,
.flowi4_iif = rt->rt_iif,
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 0b02eb3..21fb343 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -978,8 +978,8 @@ void ip_rt_get_source(u8 *addr, struct rtable *rt)
src = rt->rt_src;
else {
struct flowi4 fl4 = {
- .daddr = rt->rt_key_dst,
- .saddr = rt->rt_key_src,
+ .daddr = rt->rt_dst,
+ .saddr = rt->rt_src,
.flowi4_tos = rt->rt_tos,
.flowi4_oif = rt->rt_oif,
.flowi4_iif = rt->rt_iif,
@@ -1154,8 +1154,6 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
#endif
rth->dst.output = ip_rt_bug;
- rth->rt_key_dst = daddr;
- rth->rt_key_src = saddr;
rth->rt_genid = rt_genid(dev_net(dev));
rth->rt_flags = RTCF_MULTICAST;
rth->rt_type = RTN_MULTICAST;
@@ -1289,8 +1287,6 @@ static int __mkroute_input(struct sk_buff *skb,
goto cleanup;
}
- rth->rt_key_dst = daddr;
- rth->rt_key_src = saddr;
rth->rt_genid = rt_genid(dev_net(rth->dst.dev));
rth->rt_flags = flags;
rth->rt_type = res->type;
@@ -1463,8 +1459,6 @@ local_input:
rth->dst.tclassid = itag;
#endif
- rth->rt_key_dst = daddr;
- rth->rt_key_src = saddr;
rth->rt_genid = rt_genid(net);
rth->rt_flags = flags|RTCF_LOCAL;
rth->rt_type = res.type;
@@ -1628,8 +1622,6 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
rth->dst.output = ip_output;
- rth->rt_key_dst = oldflp4->daddr;
- rth->rt_key_src = oldflp4->saddr;
rth->rt_genid = rt_genid(dev_net(dev_out));
rth->rt_flags = flags;
rth->rt_type = type;
@@ -1903,8 +1895,6 @@ struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_or
if (new->dev)
dev_hold(new->dev);
- rt->rt_key_dst = ort->rt_key_dst;
- rt->rt_key_src = ort->rt_key_src;
rt->rt_tos = ort->rt_tos;
rt->rt_iif = ort->rt_iif;
rt->rt_oif = ort->rt_oif;
@@ -1955,7 +1945,7 @@ struct rtable *ip_route_output_flow(struct net *net, struct flowi4 *flp4,
}
EXPORT_SYMBOL_GPL(ip_route_output_flow);
-static int rt_fill_info(struct net *net,
+static int rt_fill_info(struct net *net, __be32 src,
struct sk_buff *skb, u32 pid, u32 seq, int event,
int nowait, unsigned int flags)
{
@@ -1985,9 +1975,9 @@ static int rt_fill_info(struct net *net,
NLA_PUT_BE32(skb, RTA_DST, rt->rt_dst);
- if (rt->rt_key_src) {
+ if (src) {
r->rtm_src_len = 32;
- NLA_PUT_BE32(skb, RTA_SRC, rt->rt_key_src);
+ NLA_PUT_BE32(skb, RTA_SRC, src);
}
if (rt->dst.dev)
NLA_PUT_U32(skb, RTA_OIF, rt->dst.dev->ifindex);
@@ -1997,7 +1987,7 @@ static int rt_fill_info(struct net *net,
#endif
if (rt_is_input_route(rt))
NLA_PUT_BE32(skb, RTA_PREFSRC, rt->rt_spec_dst);
- else if (rt->rt_src != rt->rt_key_src)
+ else if (rt->rt_src != src)
NLA_PUT_BE32(skb, RTA_PREFSRC, rt->rt_src);
if (rt->rt_dst != rt->rt_gateway)
@@ -2139,7 +2129,7 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
if (rtm->rtm_flags & RTM_F_NOTIFY)
rt->rt_flags |= RTCF_NOTIFY;
- err = rt_fill_info(net, skb, NETLINK_CB(in_skb).pid, nlh->nlmsg_seq,
+ err = rt_fill_info(net, src, skb, NETLINK_CB(in_skb).pid, nlh->nlmsg_seq,
RTM_NEWROUTE, 0, 0);
if (err <= 0)
goto errout_free;
diff --git a/net/ipv4/xfrm4_policy.c b/net/ipv4/xfrm4_policy.c
index 13e0e7f..813fb34 100644
--- a/net/ipv4/xfrm4_policy.c
+++ b/net/ipv4/xfrm4_policy.c
@@ -71,8 +71,6 @@ static int xfrm4_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
struct rtable *rt = (struct rtable *)xdst->route;
const struct flowi4 *fl4 = &fl->u.ip4;
- rt->rt_key_dst = fl4->daddr;
- rt->rt_key_src = fl4->saddr;
rt->rt_tos = fl4->flowi4_tos;
rt->rt_iif = fl4->flowi4_iif;
rt->rt_oif = fl4->flowi4_oif;
--
1.7.4.1
^ permalink raw reply related
* [PATCH v4 RFC 8/8] ipv4: Use caller's on-stack flowi as-is in output route lookups.
From: David Miller @ 2011-03-13 0:38 UTC (permalink / raw)
To: netdev
Signed-off-by: David S. Miller <davem@davemloft.net>
---
include/net/route.h | 2 +-
net/ipv4/route.c | 134 ++++++++++++++++++++++++--------------------------
2 files changed, 65 insertions(+), 71 deletions(-)
diff --git a/include/net/route.h b/include/net/route.h
index 154e589..304cd6c 100644
--- a/include/net/route.h
+++ b/include/net/route.h
@@ -117,7 +117,7 @@ extern int ip_rt_init(void);
extern void ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw,
__be32 src, struct net_device *dev);
extern void rt_cache_flush(struct net *net, int how);
-extern struct rtable *__ip_route_output_key(struct net *, const struct flowi4 *flp);
+extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
struct sock *sk);
extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 21fb343..ab0053b 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1037,7 +1037,7 @@ static unsigned int ipv4_default_mtu(const struct dst_entry *dst)
return mtu;
}
-static void rt_init_metrics(struct rtable *rt, const struct flowi4 *oldflp4,
+static void rt_init_metrics(struct rtable *rt, const struct flowi4 *fl4,
struct fib_info *fi)
{
struct inet_peer *peer;
@@ -1046,7 +1046,7 @@ static void rt_init_metrics(struct rtable *rt, const struct flowi4 *oldflp4,
/* If a peer entry exists for this destination, we must hook
* it up in order to get at cached metrics.
*/
- if (oldflp4 && (oldflp4->flowi4_flags & FLOWI_FLAG_PRECOW_METRICS))
+ if (fl4 && (fl4->flowi4_flags & FLOWI_FLAG_PRECOW_METRICS))
create = 1;
rt->peer = peer = inet_getpeer_v4(rt->rt_dst, create);
@@ -1073,7 +1073,7 @@ static void rt_init_metrics(struct rtable *rt, const struct flowi4 *oldflp4,
}
}
-static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *oldflp4,
+static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *fl4,
const struct fib_result *res,
struct fib_info *fi, u16 type, u32 itag)
{
@@ -1083,7 +1083,7 @@ static void rt_set_nexthop(struct rtable *rt, const struct flowi4 *oldflp4,
if (FIB_RES_GW(*res) &&
FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK)
rt->rt_gateway = FIB_RES_GW(*res);
- rt_init_metrics(rt, oldflp4, fi);
+ rt_init_metrics(rt, fl4, fi);
#ifdef CONFIG_IP_ROUTE_CLASSID
dst->tclassid = FIB_RES_NH(*res).nh_tclassid;
#endif
@@ -1571,12 +1571,11 @@ EXPORT_SYMBOL(ip_route_input);
/* called with rcu_read_lock() */
static struct rtable *__mkroute_output(const struct fib_result *res,
const struct flowi4 *fl4,
- const struct flowi4 *oldflp4,
struct net_device *dev_out,
unsigned int flags)
{
struct fib_info *fi = res->fi;
- u32 tos = RT_FL_TOS(oldflp4);
+ u32 tos = RT_FL_TOS(fl4);
struct in_device *in_dev;
u16 type = res->type;
struct rtable *rth;
@@ -1603,8 +1602,8 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
fi = NULL;
} else if (type == RTN_MULTICAST) {
flags |= RTCF_MULTICAST | RTCF_LOCAL;
- if (!ip_check_mc_rcu(in_dev, oldflp4->daddr, oldflp4->saddr,
- oldflp4->flowi4_proto))
+ if (!ip_check_mc_rcu(in_dev, fl4->daddr, fl4->saddr,
+ fl4->flowi4_proto))
flags &= ~RTCF_LOCAL;
/* If multicast route do not exist use
* default one, but do not gateway in this case.
@@ -1629,8 +1628,8 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
rth->rt_dst = fl4->daddr;
rth->rt_src = fl4->saddr;
rth->rt_iif = 0;
- rth->rt_oif = oldflp4->flowi4_oif;
- rth->rt_mark = oldflp4->flowi4_mark;
+ rth->rt_oif = fl4->flowi4_oif;
+ rth->rt_mark = fl4->flowi4_mark;
rth->rt_gateway = fl4->daddr;
rth->rt_spec_dst= fl4->saddr;
rth->rt_peer_genid = 0;
@@ -1653,7 +1652,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
#ifdef CONFIG_IP_MROUTE
if (type == RTN_MULTICAST) {
if (IN_DEV_MFORWARD(in_dev) &&
- !ipv4_is_local_multicast(oldflp4->daddr)) {
+ !ipv4_is_local_multicast(fl4->daddr)) {
rth->dst.input = ip_mr_input;
rth->dst.output = ip_mc_output;
}
@@ -1661,7 +1660,7 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
#endif
}
- rt_set_nexthop(rth, oldflp4, res, fi, type, 0);
+ rt_set_nexthop(rth, fl4, res, fi, type, 0);
return rth;
}
@@ -1670,13 +1669,12 @@ static struct rtable *__mkroute_output(const struct fib_result *res,
* Major route resolver routine.
*/
-struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldflp4)
+struct rtable *__ip_route_output_key(struct net *net, struct flowi4 *fl4)
{
- u32 tos = RT_FL_TOS(oldflp4);
- struct flowi4 fl4;
- struct fib_result res;
- unsigned int flags = 0;
struct net_device *dev_out = NULL;
+ u32 tos = RT_FL_TOS(fl4);
+ unsigned int flags = 0;
+ struct fib_result res;
struct rtable *rth;
res.fi = NULL;
@@ -1684,21 +1682,17 @@ struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldfl
res.r = NULL;
#endif
- fl4.flowi4_oif = oldflp4->flowi4_oif;
- fl4.flowi4_iif = net->loopback_dev->ifindex;
- fl4.flowi4_mark = oldflp4->flowi4_mark;
- fl4.daddr = oldflp4->daddr;
- fl4.saddr = oldflp4->saddr;
- fl4.flowi4_tos = tos & IPTOS_RT_MASK;
- fl4.flowi4_scope = ((tos & RTO_ONLINK) ?
- RT_SCOPE_LINK : RT_SCOPE_UNIVERSE);
+ fl4->flowi4_iif = net->loopback_dev->ifindex;
+ fl4->flowi4_tos = tos & IPTOS_RT_MASK;
+ fl4->flowi4_scope = ((tos & RTO_ONLINK) ?
+ RT_SCOPE_LINK : RT_SCOPE_UNIVERSE);
rcu_read_lock();
- if (oldflp4->saddr) {
+ if (fl4->saddr) {
rth = ERR_PTR(-EINVAL);
- if (ipv4_is_multicast(oldflp4->saddr) ||
- ipv4_is_lbcast(oldflp4->saddr) ||
- ipv4_is_zeronet(oldflp4->saddr))
+ if (ipv4_is_multicast(fl4->saddr) ||
+ ipv4_is_lbcast(fl4->saddr) ||
+ ipv4_is_zeronet(fl4->saddr))
goto out;
/* I removed check for oif == dev_out->oif here.
@@ -1709,11 +1703,11 @@ struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldfl
of another iface. --ANK
*/
- if (oldflp4->flowi4_oif == 0 &&
- (ipv4_is_multicast(oldflp4->daddr) ||
- ipv4_is_lbcast(oldflp4->daddr))) {
+ if (fl4->flowi4_oif == 0 &&
+ (ipv4_is_multicast(fl4->daddr) ||
+ ipv4_is_lbcast(fl4->daddr))) {
/* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */
- dev_out = __ip_dev_find(net, oldflp4->saddr, false);
+ dev_out = __ip_dev_find(net, fl4->saddr, false);
if (dev_out == NULL)
goto out;
@@ -1732,20 +1726,20 @@ struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldfl
Luckily, this hack is good workaround.
*/
- fl4.flowi4_oif = dev_out->ifindex;
+ fl4->flowi4_oif = dev_out->ifindex;
goto make_route;
}
- if (!(oldflp4->flowi4_flags & FLOWI_FLAG_ANYSRC)) {
+ if (!(fl4->flowi4_flags & FLOWI_FLAG_ANYSRC)) {
/* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */
- if (!__ip_dev_find(net, oldflp4->saddr, false))
+ if (!__ip_dev_find(net, fl4->saddr, false))
goto out;
}
}
- if (oldflp4->flowi4_oif) {
- dev_out = dev_get_by_index_rcu(net, oldflp4->flowi4_oif);
+ if (fl4->flowi4_oif) {
+ dev_out = dev_get_by_index_rcu(net, fl4->flowi4_oif);
rth = ERR_PTR(-ENODEV);
if (dev_out == NULL)
goto out;
@@ -1755,37 +1749,37 @@ struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldfl
rth = ERR_PTR(-ENETUNREACH);
goto out;
}
- if (ipv4_is_local_multicast(oldflp4->daddr) ||
- ipv4_is_lbcast(oldflp4->daddr)) {
- if (!fl4.saddr)
- fl4.saddr = inet_select_addr(dev_out, 0,
- RT_SCOPE_LINK);
+ if (ipv4_is_local_multicast(fl4->daddr) ||
+ ipv4_is_lbcast(fl4->daddr)) {
+ if (!fl4->saddr)
+ fl4->saddr = inet_select_addr(dev_out, 0,
+ RT_SCOPE_LINK);
goto make_route;
}
- if (!fl4.saddr) {
- if (ipv4_is_multicast(oldflp4->daddr))
- fl4.saddr = inet_select_addr(dev_out, 0,
- fl4.flowi4_scope);
- else if (!oldflp4->daddr)
- fl4.saddr = inet_select_addr(dev_out, 0,
- RT_SCOPE_HOST);
+ if (fl4->saddr) {
+ if (ipv4_is_multicast(fl4->daddr))
+ fl4->saddr = inet_select_addr(dev_out, 0,
+ fl4->flowi4_scope);
+ else if (!fl4->daddr)
+ fl4->saddr = inet_select_addr(dev_out, 0,
+ RT_SCOPE_HOST);
}
}
- if (!fl4.daddr) {
- fl4.daddr = fl4.saddr;
- if (!fl4.daddr)
- fl4.daddr = fl4.saddr = htonl(INADDR_LOOPBACK);
+ if (!fl4->daddr) {
+ fl4->daddr = fl4->saddr;
+ if (!fl4->daddr)
+ fl4->daddr = fl4->saddr = htonl(INADDR_LOOPBACK);
dev_out = net->loopback_dev;
- fl4.flowi4_oif = net->loopback_dev->ifindex;
+ fl4->flowi4_oif = net->loopback_dev->ifindex;
res.type = RTN_LOCAL;
flags |= RTCF_LOCAL;
goto make_route;
}
- if (fib_lookup(net, &fl4, &res)) {
+ if (fib_lookup(net, fl4, &res)) {
res.fi = NULL;
- if (oldflp4->flowi4_oif) {
+ if (fl4->flowi4_oif) {
/* Apparently, routing tables are wrong. Assume,
that the destination is on link.
@@ -1804,9 +1798,9 @@ struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldfl
likely IPv6, but we do not.
*/
- if (fl4.saddr == 0)
- fl4.saddr = inet_select_addr(dev_out, 0,
- RT_SCOPE_LINK);
+ if (fl4->saddr == 0)
+ fl4->saddr = inet_select_addr(dev_out, 0,
+ RT_SCOPE_LINK);
res.type = RTN_UNICAST;
goto make_route;
}
@@ -1815,36 +1809,36 @@ struct rtable *__ip_route_output_key(struct net *net, const struct flowi4 *oldfl
}
if (res.type == RTN_LOCAL) {
- if (!fl4.saddr) {
+ if (!fl4->saddr) {
if (res.fi->fib_prefsrc)
- fl4.saddr = res.fi->fib_prefsrc;
+ fl4->saddr = res.fi->fib_prefsrc;
else
- fl4.saddr = fl4.daddr;
+ fl4->saddr = fl4->daddr;
}
dev_out = net->loopback_dev;
- fl4.flowi4_oif = dev_out->ifindex;
+ fl4->flowi4_oif = dev_out->ifindex;
res.fi = NULL;
flags |= RTCF_LOCAL;
goto make_route;
}
#ifdef CONFIG_IP_ROUTE_MULTIPATH
- if (res.fi->fib_nhs > 1 && fl4.flowi4_oif == 0)
+ if (res.fi->fib_nhs > 1 && fl4->flowi4_oif == 0)
fib_select_multipath(&res);
else
#endif
- if (!res.prefixlen && res.type == RTN_UNICAST && !fl4.flowi4_oif)
+ if (!res.prefixlen && res.type == RTN_UNICAST && !fl4->flowi4_oif)
fib_select_default(&res);
- if (!fl4.saddr)
- fl4.saddr = FIB_RES_PREFSRC(res);
+ if (!fl4->saddr)
+ fl4->saddr = FIB_RES_PREFSRC(res);
dev_out = FIB_RES_DEV(res);
- fl4.flowi4_oif = dev_out->ifindex;
+ fl4->flowi4_oif = dev_out->ifindex;
make_route:
- rth = __mkroute_output(&res, &fl4, oldflp4, dev_out, flags);
+ rth = __mkroute_output(&res, fl4, dev_out, flags);
if (!IS_ERR(rth))
rth = rt_finalize(rth, NULL);
--
1.7.4.1
^ permalink raw reply related
* Re: [RFC v2 PATCH 1/9] crypto: authencesn - Add algorithm to handle IPsec extended sequence numbers
From: Herbert Xu @ 2011-03-13 6:30 UTC (permalink / raw)
To: Steffen Klassert
Cc: David Miller, Alex Badea, Andreas Gruenbacher, netdev,
linux-crypto
In-Reply-To: <20110308100458.GC31402@secunet.com>
On Tue, Mar 08, 2011 at 11:04:58AM +0100, Steffen Klassert wrote:
>
> + return crypto_authenc_esn_genicv(req, iv, CRYPTO_TFM_REQ_MAY_SLEEP);
BTW, we should get rid of these MAY_SLEEP flags. Originally I
included them for the IPsec encryption path because that happens
to be sleepable. We should instead be getting it from the request
itself.
Also, as soon as we enter the async path it becomes unsleepable.
Of course this affects the existing authenc file as well.
Cheers,
--
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* Re: [RFC v2 PATCH 2/9] xfrm: Add basic infrastructure to support IPsec extended sequence numbers
From: Herbert Xu @ 2011-03-13 6:44 UTC (permalink / raw)
To: Steffen Klassert
Cc: David Miller, Alex Badea, Andreas Gruenbacher, netdev,
linux-crypto
In-Reply-To: <20110308100543.GD31402@secunet.com>
On Tue, Mar 08, 2011 at 11:05:43AM +0100, Steffen Klassert wrote:
> This patch adds the struct xfrm_replay_state_esn which will be
> used to support IPsec extended sequence numbers and anti replay windows
> bigger than 32 packets. Also we add a function that returns the actual
> size of the xfrm_replay_state_esn, a xfrm netlink atribute and a xfrm state
> flag for the use of extended sequence numbers.
>
> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
All the remaining eight patches look good to me.
Dave, how would you like to handle this patch-set? As Steffen's
first crypto patch is completely self-contained, perhaps you could
just include it along with the rest of the series in your tree?
Thanks,
--
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* Re: [RFC v2 PATCH 2/9] xfrm: Add basic infrastructure to support IPsec extended sequence numbers
From: David Miller @ 2011-03-13 6:53 UTC (permalink / raw)
To: herbert; +Cc: steffen.klassert, abadea, agruen, netdev, linux-crypto
In-Reply-To: <20110313064457.GB10849@gondor.apana.org.au>
From: Herbert Xu <herbert@gondor.apana.org.au>
Date: Sun, 13 Mar 2011 14:44:57 +0800
> On Tue, Mar 08, 2011 at 11:05:43AM +0100, Steffen Klassert wrote:
>> This patch adds the struct xfrm_replay_state_esn which will be
>> used to support IPsec extended sequence numbers and anti replay windows
>> bigger than 32 packets. Also we add a function that returns the actual
>> size of the xfrm_replay_state_esn, a xfrm netlink atribute and a xfrm state
>> flag for the use of extended sequence numbers.
>>
>> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
>
> All the remaining eight patches look good to me.
>
> Dave, how would you like to handle this patch-set? As Steffen's
> first crypto patch is completely self-contained, perhaps you could
> just include it along with the rest of the series in your tree?
Sure, I can do that.
Can I add your ACK to all of the patches?
^ permalink raw reply
* Re: [RFC v2 PATCH 2/9] xfrm: Add basic infrastructure to support IPsec extended sequence numbers
From: Herbert Xu @ 2011-03-13 6:58 UTC (permalink / raw)
To: David Miller; +Cc: steffen.klassert, abadea, agruen, netdev, linux-crypto
In-Reply-To: <20110312.225333.193705435.davem@davemloft.net>
On Sat, Mar 12, 2011 at 10:53:33PM -0800, David Miller wrote:
>
> Sure, I can do that.
Thanks!
> Can I add your ACK to all of the patches?
Of course:
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Cheers,
--
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* RE: how to utilize multi tx queue to sent packets
From: Jon Zhou @ 2011-03-13 9:06 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: netdev@vger.kernel.org
In-Reply-To: <20110311090636.6e393f2b@nehalam>
> -----Original Message-----
> From: Stephen Hemminger [mailto:shemminger@vyatta.com]
> Sent: Saturday, March 12, 2011 1:07 AM
> To: Jon Zhou
> Cc: netdev@vger.kernel.org
> Subject: Re: how to utilize multi tx queue to sent packets
>
> On Thu, 10 Mar 2011 22:52:30 -0800
> Jon Zhou <Jon.Zhou@jdsu.com> wrote:
>
> > hi
> >
> > I am doing some test according to the
> > website:http://wiki.ipxwarzone.com/index.php5?title=Linux_packet_mmap
> >
> > use packet_mmap tx_ring to send packet.
> >
> > I modified the sample code "packetmmap.c" to make it send packets
> have different outer ip.
> > so that with the help of RSS, I can achieve higher throughput.
> >
> > but one thing I saw at the tx side, is that all the packets are sent
> via the same tx_queue,
> > which is conflict with what I saw at the rx side.
> > any idea to make it sent packets via different tx_queues? (spread
> across the tx_queues)
>
> You need to have multiple threads to get Tx scaling.
> In you case that also means multiple AF_PACKET sockets and separate
> rings.
>
> Or just run multiple copies of the same test each with a different IP
How does kernel know which tx_queue to use?
Will dev_queue_xmit determine tx_queue(sw queue) to send packets?
Or just let NIC select a tx_queue(hardware) ?
^ permalink raw reply
* Re: [PATCH] inetpeer: Don't disable BH for initial fast RCU lookup.
From: Eric Dumazet @ 2011-03-13 10:04 UTC (permalink / raw)
To: David Miller; +Cc: netdev
In-Reply-To: <20110308.145954.59682618.davem@davemloft.net>
Le mardi 08 mars 2011 à 14:59 -0800, David Miller a écrit :
> If modifications on other cpus are ok, then modifications to
> the tree during lookup done by the local cpu are ok too.
>
> Signed-off-by: David S. Miller <davem@davemloft.net>
> ---
> net/ipv4/inetpeer.c | 18 +++++++++---------
> 1 files changed, 9 insertions(+), 9 deletions(-)
>
> diff --git a/net/ipv4/inetpeer.c b/net/ipv4/inetpeer.c
> index f604ffd..6442c35 100644
> --- a/net/ipv4/inetpeer.c
> +++ b/net/ipv4/inetpeer.c
> @@ -206,16 +206,16 @@ static int addr_compare(const struct inetpeer_addr *a,
> })
>
> /*
> - * Called with rcu_read_lock_bh()
> + * Called with rcu_read_lock()
> * Because we hold no lock against a writer, its quite possible we fall
> * in an endless loop.
> * But every pointer we follow is guaranteed to be valid thanks to RCU.
> * We exit from this function if number of links exceeds PEER_MAXDEPTH
> */
> -static struct inet_peer *lookup_rcu_bh(const struct inetpeer_addr *daddr,
> - struct inet_peer_base *base)
> +static struct inet_peer *lookup_rcu(const struct inetpeer_addr *daddr,
> + struct inet_peer_base *base)
> {
> - struct inet_peer *u = rcu_dereference_bh(base->root);
> + struct inet_peer *u = rcu_dereference(base->root);
> int count = 0;
>
> while (u != peer_avl_empty) {
> @@ -231,9 +231,9 @@ static struct inet_peer *lookup_rcu_bh(const struct inetpeer_addr *daddr,
> return u;
> }
> if (cmp == -1)
> - u = rcu_dereference_bh(u->avl_left);
> + u = rcu_dereference(u->avl_left);
> else
> - u = rcu_dereference_bh(u->avl_right);
> + u = rcu_dereference(u->avl_right);
> if (unlikely(++count == PEER_MAXDEPTH))
> break;
> }
> @@ -470,11 +470,11 @@ struct inet_peer *inet_getpeer(struct inetpeer_addr *daddr, int create)
> /* Look up for the address quickly, lockless.
> * Because of a concurrent writer, we might not find an existing entry.
> */
> - rcu_read_lock_bh();
> + rcu_read_lock();
> sequence = read_seqbegin(&base->lock);
> - p = lookup_rcu_bh(daddr, base);
> + p = lookup_rcu(daddr, base);
> invalidated = read_seqretry(&base->lock, sequence);
> - rcu_read_unlock_bh();
> + rcu_read_unlock();
>
> if (p) {
> /* The existing node has been found.
David, I am not sure this is safe, since we use call_rcu_bh() when
freeing one item. One cpu could decide to kfree() one item while another
cpu could still use it.
rcu_read_lock_bh() was signalling to others cpu we were in a softirq
section, so we were delaying a possible kfree().
^ permalink raw reply
* Re: ANNOUNCE: debloat-testing kernel git tree
From: Pavel Machek @ 2011-03-13 10:27 UTC (permalink / raw)
To: Rick Jones
Cc: Tianji Li, Dave T?ht, sedat.dilek, netdev, linux-wireless,
bloat-devel, linux-kernel
In-Reply-To: <1299610096.2087.8.camel@tardy>
Hi!
> > Bandwidth varies greatly with time. On cellphone, you can go from
> > 5KB/sec GPRS to 300KB/sec HSDPA and back...
>
> How quickly/often and does that trigger an event in the host that would
> allow adjustment of the queue?
I'd expect it to get fast when you leave the building, and back to
slow when you re-enter. It all depends how good 3G coverage is, but
seeing it change more than once a minute is not that uncommon.
Event... not sure. Linux can ask for current signal strength, and I
assume connection type works in similar way...
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
^ permalink raw reply
* Re: [PATCH 03/18] ipvs: zero percpu stats
From: Eric Dumazet @ 2011-03-13 10:57 UTC (permalink / raw)
To: Julian Anastasov
Cc: Simon Horman, netdev, netfilter-devel, netfilter, lvs-devel,
Hans Schillstrom
In-Reply-To: <alpine.LFD.2.00.1103061218470.1879@ja.ssi.bg>
Le dimanche 06 mars 2011 à 14:18 +0200, Julian Anastasov a écrit :
> Hello,
>
> On Sun, 6 Mar 2011, Eric Dumazet wrote:
>
> >> Zero the new percpu stats because we copy from there.
> >>
> >> Signed-off-by: Julian Anastasov <ja@ssi.bg>
> >> Signed-off-by: Simon Horman <horms@verge.net.au>
> >> ---
> >> net/netfilter/ipvs/ip_vs_ctl.c | 17 +++++++++++++++++
> >> 1 files changed, 17 insertions(+), 0 deletions(-)
> >>
> >> diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
> >> index a2a67ad..fd74527 100644
> >> --- a/net/netfilter/ipvs/ip_vs_ctl.c
> >> +++ b/net/netfilter/ipvs/ip_vs_ctl.c
> >> @@ -715,8 +715,25 @@ static void ip_vs_trash_cleanup(struct net *net)
> >> static void
> >> ip_vs_zero_stats(struct ip_vs_stats *stats)
> >> {
> >> + struct ip_vs_cpu_stats *cpustats = stats->cpustats;
> >> + int i;
> >> +
> >> spin_lock_bh(&stats->lock);
> >>
> >> + for_each_possible_cpu(i) {
> >> + struct ip_vs_cpu_stats *u = per_cpu_ptr(cpustats, i);
> >> + unsigned int start;
> >> +
> >> + /* Do not pretend to be writer, it is enough to
> >> + * sync with writers that modify the u64 counters
> >> + * because under stats->lock we are the only reader.
> >> + */
> >> + do {
> >> + start = u64_stats_fetch_begin(&u->syncp);
> >> + memset(&u->ustats, 0, sizeof(u->ustats));
> >> + } while (u64_stats_fetch_retry(&u->syncp, start));
> >
> >
> > Sorry this makes no sense to me.
>
> Hm, yes, the comment is a little bit misleading.
> I fixed it below...
>
> > This code _is_ a writer, and hardly a hot path.
>
> Yes, the picture is as follows:
>
> - in 2.6.38-rc we remove the global spin lock (stats->lock)
> from packet processing which is a hot path, adding percpu
> counters instead
>
> - we need protection for percpu counters and for the sum
>
> - the chain is: interrupts increment percpu counters, the
> estimation timer reads them and creates sum every 2 seconds,
> then user context can read the sum or even to show the percpu
> counters, not to forget the zeroing of sum and counters
>
> The players in detail:
>
> - packet processing:
> - softirq context, hot path
> - increments counters by using u64_stats_update_begin and
> u64_stats_update_end, does not wait readers or zeroing
> - sum not touched, stats->lock usage removed in 2.6.38-rc
>
> - 2-second estimation timer:
> - funcs: estimation_timer()
> - timer context, softirq
> - reads percpu counters with u64_stats_fetch_begin and
> u64_stats_fetch_retry to sync with counter incrementing
> - uses spin_lock (stats->lock) to protect the written sum
> which is later read by user context: provides
> at least u64 atomicity but additionally the relation
> between packets and bytes
>
> - sum readers:
> - funcs: ip_vs_stats_show(), ip_vs_stats_percpu_show(),
> ip_vs_copy_stats(), ip_vs_genl_fill_stats()
> - user context, not a hot path
> - uses spin_lock_bh (stats->lock) for atomic reading of
> the sum created by estimation_timer()
>
> - show percpu counters:
> - funcs: ip_vs_stats_percpu_show()
> - user context, not a hot path
> - uses u64_stats_fetch_begin_bh and u64_stats_fetch_retry_bh
> to synchronize with counter incrementing
> - still missing: should use spin_lock_bh (stats->lock)
> to synchronize with ip_vs_zero_stats() that modifies
> percpu counters.
>
> - zero stats and percpu counters
> - funcs: ip_vs_zero_stats()
> - user context, not a hot path
> - uses spin_lock_bh (stats->lock) while modifying
> sum but also while zeroing percpu counters because
> we are a hidden writer which does not allow other
> percpu counter readers at the same time but we are
> still synchronized with percpu counter incrementing
> without delaying it
>
> To summarize, I see 2 solutions, in order of preference:
>
> 1. all players except packet processing should use stats->lock
> when reading/writing sum or when reading/zeroing percpu
> counters. Use u64_stats to avoid delays in incrementing.
>
> 2. Use seqlock instead of u64_stats if we want to treat the
> percpu counters zeroing as writer. This returns us before
> 2.6.38-rc where we used global stats->lock even for counter
> incrementing. Except that now we can use percpu seqlock
> just to register the zeroing as writer.
>
> > Why try to pretend its a reader and confuse people ?
> >
> > Either :
> >
> > - Another writer can modify the counters in same time, and we must
> > synchronize with them (we are a writer after all)
>
> Global mutex allows only one zeroing at a time.
> But zeroing runs in parallel with incrementing, so we
> have 2 writers for a per-CPU state. This sounds like
> above solution 2 with percpu seqlock? But it adds extra
> spin_lock in hot path, even if it is percpu. It only
> saves the spin_lock_bh while reading percpu counters in
> ip_vs_stats_percpu_show(). That is why a prefer solution 1.
>
> > - Another reader can read the counters in same time, and we must let
> > them catch we mihjt have cleared half of their values.
>
> Yes, zeroing can run in parallel with /proc reading,
> that is why I now try to serialize all readers with the
> stats spin lock to guarantee u64 atomicity.
>
> > - No reader or writer can access data, no synch is needed, a pure
> > memset() is OK.
>
> Packet processing can damage the counters while we
> do memset, so we need at least u64_stats_fetch_* to sync
> with incrementing.
>
OK I now understand what you wanted to do.
Problem is you do synchronize your memset() with a concurrent writer but
one way only. (You detect a writer did some changes on the counters
while you memset() them), but a writer has no way to detect your writes
(could be partially committed to main memory) : It could read a
corrupted value.
I feel memory barriers are wrong and not really fixable without slowing
down the hot path.
As implied in include/linux/u64_stats_sync.h file, a "writer" should be
alone :)
One other way to handle that (and let hot path packet processing without
extra locking) would be to never memset() this data, but use a separate
"summed" value as a relative point, and substract this sum to the
current one (all this in slow path, so not a problem)
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* (unknown),
From: Wang Lei @ 2011-03-13 8:03 UTC (permalink / raw)
I'm Wang Lei, I have a deal worth 25 Million USD if interested,
please contact
me via my personal email: wanglei99@w.cn
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
^ permalink raw reply
* hi
From: Vincent Appiah @ 2011-03-13 11:35 UTC (permalink / raw)
Attn: Please and a good day to you.
I have an important information for the owner of this email but first of all, I
want to ensure if this email is still valid, I want you to confirm if this email
is correct and is being checked by only you so that I will proceed and let you
know what information we have for you.
I am taking this measure in other not to make any mistake because of the
sensitivity of the information i have for you.
Awaiting your response.
Barr vincent Appiah.
^ permalink raw reply
* Hello
From: Johnson Ato @ 2011-03-13 12:50 UTC (permalink / raw)
--
Dear Friend,
I got your email address on the internet, I want you to assist
me by getting a nice property for purchase within your location
as me and my family wants to relocate to a safe environment as
soon as possible due to the crisis in my country.
Your urgent response is highly appreciated.
Johnson Ato
----------------------------------------------------------------
^ permalink raw reply
* Re: [PATCH] ipv4: netfilter: ipt_CLUSTERIP: fix buffer overflow
From: Changli Gao @ 2011-03-13 14:00 UTC (permalink / raw)
To: Vasiliy Kulikov
Cc: linux-kernel, security, Patrick McHardy, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
netfilter-devel, netfilter, coreteam, netdev
In-Reply-To: <1299780740-32652-1-git-send-email-segoon@openwall.com>
On Fri, Mar 11, 2011 at 2:12 AM, Vasiliy Kulikov <segoon@openwall.com> wrote:
> buffer string is copied from userspace. It is not checked whether it is
> zero terminated. This may lead to overflow inside of simple_strtoul().
>
> It was introduced before the git epoch. Files "ipt_CLUSTERIP/*" are
> root writable only by default, however, on some setups permissions might be
> relaxed to e.g. network admin user.
>
> Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
> ---
> Compile tested.
>
> net/ipv4/netfilter/ipt_CLUSTERIP.c | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c
> index 403ca57..7aabf9a 100644
> --- a/net/ipv4/netfilter/ipt_CLUSTERIP.c
> +++ b/net/ipv4/netfilter/ipt_CLUSTERIP.c
> @@ -666,6 +666,7 @@ static ssize_t clusterip_proc_write(struct file *file, const char __user *input,
>
> if (copy_from_user(buffer, input, PROC_WRITELEN))
I think size should be used instead of PROC_WRITELEN.
if (size > PROC_WRITELEN)
return -EIO;
if (copy_from_user(buffer, input, size))
return -EFAULT;
buffer[size] = '\0';
> return -EFAULT;
> + buffer[sizeof(buffer)-1] = 0;
>
> if (*buffer == '+') {
> nodenum = simple_strtoul(buffer+1, NULL, 10);
--
Regards,
Changli Gao(xiaosuo@gmail.com)
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Hello
From: Johnson Ato @ 2011-03-13 13:36 UTC (permalink / raw)
--
Dear Friend,
I got your email address on the internet, I want you to assist
me by getting a nice property for purchase within your location
as me and my family wants to relocate to a safe environment as
soon as possible due to the crisis in my country.
Your urgent response is highly appreciated.
Johnson Ato
----------------------------------------------------------------
^ permalink raw reply
* Re: [PATCH 3/3] vhost-net: use lock_sock_fast() in peek_head_len()
From: Michael S. Tsirkin @ 2011-03-13 15:06 UTC (permalink / raw)
To: Jason Wang; +Cc: virtualization, netdev, kvm, linux-kernel
In-Reply-To: <20110117081117.18900.48672.stgit@dhcp-91-7.nay.redhat.com.englab.nay.redhat.com>
On Mon, Jan 17, 2011 at 04:11:17PM +0800, Jason Wang wrote:
> We can use lock_sock_fast() instead of lock_sock() in order to get
> speedup in peek_head_len().
>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
> drivers/vhost/net.c | 4 ++--
> 1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
> index c32a2e4..50b622a 100644
> --- a/drivers/vhost/net.c
> +++ b/drivers/vhost/net.c
> @@ -211,12 +211,12 @@ static int peek_head_len(struct sock *sk)
> {
> struct sk_buff *head;
> int len = 0;
> + bool slow = lock_sock_fast(sk);
>
> - lock_sock(sk);
> head = skb_peek(&sk->sk_receive_queue);
> if (head)
> len = head->len;
> - release_sock(sk);
> + unlock_sock_fast(sk, slow);
> return len;
> }
>
Wanted to apply this, but looking at the code I think the lock_sock here
is wrong. What we really need is to handle the case where the skb is
pulled from the receive queue after skb_peek. However this is not the
right lock to use for that, sk_receive_queue.lock is.
So I expect the following is the right way to handle this.
Comments?
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index 0329c41..5720301 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -213,12 +213,13 @@ static int peek_head_len(struct sock *sk)
{
struct sk_buff *head;
int len = 0;
+ unsigned long flags;
- lock_sock(sk);
+ spin_lock_irqsave(&sk->sk_receive_queue.lock, flags);
head = skb_peek(&sk->sk_receive_queue);
- if (head)
+ if (likely(head))
len = head->len;
- release_sock(sk);
+ spin_unlock_irqrestore(&sk->sk_receive_queue.lock, flags);
return len;
}
^ permalink raw reply related
* Re: [PATCH 3/3] vhost-net: use lock_sock_fast() in peek_head_len()
From: Eric Dumazet @ 2011-03-13 15:52 UTC (permalink / raw)
To: Michael S. Tsirkin; +Cc: Jason Wang, virtualization, netdev, kvm, linux-kernel
In-Reply-To: <20110313150646.GA30494@redhat.com>
Le dimanche 13 mars 2011 à 17:06 +0200, Michael S. Tsirkin a écrit :
> On Mon, Jan 17, 2011 at 04:11:17PM +0800, Jason Wang wrote:
> > We can use lock_sock_fast() instead of lock_sock() in order to get
> > speedup in peek_head_len().
> >
> > Signed-off-by: Jason Wang <jasowang@redhat.com>
> > ---
> > drivers/vhost/net.c | 4 ++--
> > 1 files changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
> > index c32a2e4..50b622a 100644
> > --- a/drivers/vhost/net.c
> > +++ b/drivers/vhost/net.c
> > @@ -211,12 +211,12 @@ static int peek_head_len(struct sock *sk)
> > {
> > struct sk_buff *head;
> > int len = 0;
> > + bool slow = lock_sock_fast(sk);
> >
> > - lock_sock(sk);
> > head = skb_peek(&sk->sk_receive_queue);
> > if (head)
> > len = head->len;
> > - release_sock(sk);
> > + unlock_sock_fast(sk, slow);
> > return len;
> > }
> >
>
> Wanted to apply this, but looking at the code I think the lock_sock here
> is wrong. What we really need is to handle the case where the skb is
> pulled from the receive queue after skb_peek. However this is not the
> right lock to use for that, sk_receive_queue.lock is.
> So I expect the following is the right way to handle this.
> Comments?
>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
>
> diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
> index 0329c41..5720301 100644
> --- a/drivers/vhost/net.c
> +++ b/drivers/vhost/net.c
> @@ -213,12 +213,13 @@ static int peek_head_len(struct sock *sk)
> {
> struct sk_buff *head;
> int len = 0;
> + unsigned long flags;
>
> - lock_sock(sk);
> + spin_lock_irqsave(&sk->sk_receive_queue.lock, flags);
> head = skb_peek(&sk->sk_receive_queue);
> - if (head)
> + if (likely(head))
> len = head->len;
> - release_sock(sk);
> + spin_unlock_irqrestore(&sk->sk_receive_queue.lock, flags);
> return len;
> }
>
You may be right, only way to be sure is to check the other side.
If it uses skb_queue_tail(), then yes, your patch is fine.
If other side did not lock socket, then your patch is a bug fix.
^ permalink raw reply
* Re: [PATCH 3/3] vhost-net: use lock_sock_fast() in peek_head_len()
From: Michael S. Tsirkin @ 2011-03-13 16:19 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Jason Wang, virtualization, netdev, kvm, linux-kernel
In-Reply-To: <1300031570.2761.22.camel@edumazet-laptop>
On Sun, Mar 13, 2011 at 04:52:50PM +0100, Eric Dumazet wrote:
> Le dimanche 13 mars 2011 à 17:06 +0200, Michael S. Tsirkin a écrit :
> > On Mon, Jan 17, 2011 at 04:11:17PM +0800, Jason Wang wrote:
> > > We can use lock_sock_fast() instead of lock_sock() in order to get
> > > speedup in peek_head_len().
> > >
> > > Signed-off-by: Jason Wang <jasowang@redhat.com>
> > > ---
> > > drivers/vhost/net.c | 4 ++--
> > > 1 files changed, 2 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
> > > index c32a2e4..50b622a 100644
> > > --- a/drivers/vhost/net.c
> > > +++ b/drivers/vhost/net.c
> > > @@ -211,12 +211,12 @@ static int peek_head_len(struct sock *sk)
> > > {
> > > struct sk_buff *head;
> > > int len = 0;
> > > + bool slow = lock_sock_fast(sk);
> > >
> > > - lock_sock(sk);
> > > head = skb_peek(&sk->sk_receive_queue);
> > > if (head)
> > > len = head->len;
> > > - release_sock(sk);
> > > + unlock_sock_fast(sk, slow);
> > > return len;
> > > }
> > >
> >
> > Wanted to apply this, but looking at the code I think the lock_sock here
> > is wrong. What we really need is to handle the case where the skb is
> > pulled from the receive queue after skb_peek. However this is not the
> > right lock to use for that, sk_receive_queue.lock is.
> > So I expect the following is the right way to handle this.
> > Comments?
> >
> > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> >
> > diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
> > index 0329c41..5720301 100644
> > --- a/drivers/vhost/net.c
> > +++ b/drivers/vhost/net.c
> > @@ -213,12 +213,13 @@ static int peek_head_len(struct sock *sk)
> > {
> > struct sk_buff *head;
> > int len = 0;
> > + unsigned long flags;
> >
> > - lock_sock(sk);
> > + spin_lock_irqsave(&sk->sk_receive_queue.lock, flags);
> > head = skb_peek(&sk->sk_receive_queue);
> > - if (head)
> > + if (likely(head))
> > len = head->len;
> > - release_sock(sk);
> > + spin_unlock_irqrestore(&sk->sk_receive_queue.lock, flags);
> > return len;
> > }
> >
>
> You may be right, only way to be sure is to check the other side.
>
> If it uses skb_queue_tail(), then yes, your patch is fine.
>
> If other side did not lock socket, then your patch is a bug fix.
>
>
Other side is in drivers/net/tun.c and net/packet/af_packet.c
At least wrt tun it seems clear socket is not locked.
Besides queue, dequeue seems to be done without socket locked.
--
MST
^ permalink raw reply
* Re: [PATCH 3/3] vhost-net: use lock_sock_fast() in peek_head_len()
From: Eric Dumazet @ 2011-03-13 16:32 UTC (permalink / raw)
To: Michael S. Tsirkin; +Cc: Jason Wang, virtualization, netdev, kvm, linux-kernel
In-Reply-To: <20110313161915.GB30642@redhat.com>
Le dimanche 13 mars 2011 à 18:19 +0200, Michael S. Tsirkin a écrit :
> Other side is in drivers/net/tun.c and net/packet/af_packet.c
> At least wrt tun it seems clear socket is not locked.
Yes (assuming you refer to tun_net_xmit())
> Besides queue, dequeue seems to be done without socket locked.
>
It seems this code (assuming you speak of drivers/vhost/net.c ?) has
some races indeed.
^ permalink raw reply
* Re: [PATCH 3/3] vhost-net: use lock_sock_fast() in peek_head_len()
From: Michael S. Tsirkin @ 2011-03-13 16:43 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Jason Wang, virtualization, netdev, kvm, linux-kernel
In-Reply-To: <1300033927.2761.26.camel@edumazet-laptop>
On Sun, Mar 13, 2011 at 05:32:07PM +0100, Eric Dumazet wrote:
> Le dimanche 13 mars 2011 à 18:19 +0200, Michael S. Tsirkin a écrit :
>
> > Other side is in drivers/net/tun.c and net/packet/af_packet.c
> > At least wrt tun it seems clear socket is not locked.
>
> Yes (assuming you refer to tun_net_xmit())
>
> > Besides queue, dequeue seems to be done without socket locked.
> >
>
> It seems this code (assuming you speak of drivers/vhost/net.c ?) has
> some races indeed.
>
Hmm. Any more besides the one fixed here?
--
MST
^ permalink raw reply
* [PATCH 1/2] NET: cdc-phonet, fix stop-queue handling
From: Jiri Slaby @ 2011-03-13 16:54 UTC (permalink / raw)
To: davem
Cc: jirislaby, netdev, gregkh, linux-usb, linux-kernel, Jiri Slaby,
Rémi Denis-Courmont
Currently there is a warning emitted by the cdc-phonet driver:
WARNING: at include/linux/netdevice.h:1557 usbpn_probe+0x3bb/0x3f0 [cdc_phonet]()
Modules linked in: ...
Pid: 5877, comm: insmod Not tainted 2.6.37.3-16-desktop #1
Call Trace:
[<ffffffff810059b9>] dump_trace+0x79/0x340
[<ffffffff81520fdc>] dump_stack+0x69/0x6f
[<ffffffff810580eb>] warn_slowpath_common+0x7b/0xc0
[<ffffffffa00254fb>] usbpn_probe+0x3bb/0x3f0 [cdc_phonet]
...
---[ end trace f5d3e02908603ab4 ]---
netif_stop_queue() cannot be called before register_netdev()
So remove netif_stop_queue from the probe funtction to avoid that.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
Cc: David S. Miller <davem@davemloft.net>
---
drivers/net/usb/cdc-phonet.c | 1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/drivers/net/usb/cdc-phonet.c b/drivers/net/usb/cdc-phonet.c
index 109751b..4cf4e36 100644
--- a/drivers/net/usb/cdc-phonet.c
+++ b/drivers/net/usb/cdc-phonet.c
@@ -392,7 +392,6 @@ int usbpn_probe(struct usb_interface *intf, const struct usb_device_id *id)
pnd = netdev_priv(dev);
SET_NETDEV_DEV(dev, &intf->dev);
- netif_stop_queue(dev);
pnd->dev = dev;
pnd->usb = usb_get_dev(usbdev);
--
1.7.4.1
^ permalink raw reply related
* [PATCH 2/2] NET: cdc-phonet, handle empty phonet header
From: Jiri Slaby @ 2011-03-13 16:54 UTC (permalink / raw)
To: davem
Cc: jirislaby, netdev, gregkh, linux-usb, linux-kernel, Jiri Slaby,
Rémi Denis-Courmont
In-Reply-To: <1300035271-8138-1-git-send-email-jslaby@suse.cz>
Currently, for N 5800 XM I get:
cdc_phonet: probe of 1-6:1.10 failed with error -22
It's because phonet_header is empty. Extra altsetting looks like
there:
E 05 24 00 01 10 03 24 ab 05 24 06 0a 0b 04 24 fd .$....$..$....$.
E 00 .
I don't see the header used anywhere so just check if the phonet
descriptor is there, not the structure itself.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
Cc: David S. Miller <davem@davemloft.net>
---
drivers/net/usb/cdc-phonet.c | 9 +++------
1 files changed, 3 insertions(+), 6 deletions(-)
diff --git a/drivers/net/usb/cdc-phonet.c b/drivers/net/usb/cdc-phonet.c
index 4cf4e36..f967913 100644
--- a/drivers/net/usb/cdc-phonet.c
+++ b/drivers/net/usb/cdc-phonet.c
@@ -328,13 +328,13 @@ int usbpn_probe(struct usb_interface *intf, const struct usb_device_id *id)
{
static const char ifname[] = "usbpn%d";
const struct usb_cdc_union_desc *union_header = NULL;
- const struct usb_cdc_header_desc *phonet_header = NULL;
const struct usb_host_interface *data_desc;
struct usb_interface *data_intf;
struct usb_device *usbdev = interface_to_usbdev(intf);
struct net_device *dev;
struct usbpn_dev *pnd;
u8 *data;
+ int phonet = 0;
int len, err;
data = intf->altsetting->extra;
@@ -355,10 +355,7 @@ int usbpn_probe(struct usb_interface *intf, const struct usb_device_id *id)
(struct usb_cdc_union_desc *)data;
break;
case 0xAB:
- if (phonet_header || dlen < 5)
- break;
- phonet_header =
- (struct usb_cdc_header_desc *)data;
+ phonet = 1;
break;
}
}
@@ -366,7 +363,7 @@ int usbpn_probe(struct usb_interface *intf, const struct usb_device_id *id)
len -= dlen;
}
- if (!union_header || !phonet_header)
+ if (!union_header || !phonet)
return -EINVAL;
data_intf = usb_ifnum_to_if(usbdev, union_header->bSlaveInterface0);
--
1.7.4.1
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox