[PATCH net-next 0/6] inet: no longer use RTNL to protect inet_dump_ifaddr()

[PATCH net-next 0/6] inet: no longer use RTNL to protect inet_dump_ifaddr()

[Eric Dumazet]

This series convert inet so that a dump of addresses (ip -4 addr)
no longer requires RTNL.

Eric Dumazet (6):
  inet: annotate data-races around ifa->ifa_tstamp and ifa->ifa_cstamp
  inet: annotate data-races around ifa->ifa_valid_lft
  inet: annotate data-races around ifa->ifa_preferred_lft
  inet: annotate data-races around ifa->ifa_flags
  inet: prepare inet_base_seq() to run without RTNL
  inet: use xa_array iterator to implement inet_dump_ifaddr()

 net/core/dev.c     |   5 +-
 net/ipv4/devinet.c | 166 +++++++++++++++++++++------------------------
 2 files changed, 79 insertions(+), 92 deletions(-)

-- 
2.44.0.278.ge034bb2e1d-goog

[PATCH net-next 1/6] inet: annotate data-races around ifa->ifa_tstamp and ifa->ifa_cstamp

[Eric Dumazet]

ifa->ifa_tstamp can be read locklessly.

Add appropriate READ_ONCE()/WRITE_ONCE() annotations.

Do the same for ifa->ifa_cstamp to prepare upcoming changes.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/devinet.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index af741af61830aeb695e7e75608515547dade8f39..1316046d5f28955376091d9e02ab4594e19fbd09 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -713,13 +713,14 @@ static void check_lifetime(struct work_struct *work)
 
 		rcu_read_lock();
 		hlist_for_each_entry_rcu(ifa, &inet_addr_lst[i], hash) {
-			unsigned long age;
+			unsigned long age, tstamp;
 
 			if (ifa->ifa_flags & IFA_F_PERMANENT)
 				continue;
 
+			tstamp = READ_ONCE(ifa->ifa_tstamp);
 			/* We try to batch several events at once. */
-			age = (now - ifa->ifa_tstamp +
+			age = (now - tstamp +
 			       ADDRCONF_TIMER_FUZZ_MINUS) / HZ;
 
 			if (ifa->ifa_valid_lft != INFINITY_LIFE_TIME &&
@@ -729,17 +730,17 @@ static void check_lifetime(struct work_struct *work)
 				   INFINITY_LIFE_TIME) {
 				continue;
 			} else if (age >= ifa->ifa_preferred_lft) {
-				if (time_before(ifa->ifa_tstamp +
+				if (time_before(tstamp +
 						ifa->ifa_valid_lft * HZ, next))
-					next = ifa->ifa_tstamp +
+					next = tstamp +
 					       ifa->ifa_valid_lft * HZ;
 
 				if (!(ifa->ifa_flags & IFA_F_DEPRECATED))
 					change_needed = true;
-			} else if (time_before(ifa->ifa_tstamp +
+			} else if (time_before(tstamp +
 					       ifa->ifa_preferred_lft * HZ,
 					       next)) {
-				next = ifa->ifa_tstamp +
+				next = tstamp +
 				       ifa->ifa_preferred_lft * HZ;
 			}
 		}
@@ -819,9 +820,9 @@ static void set_ifa_lifetime(struct in_ifaddr *ifa, __u32 valid_lft,
 			ifa->ifa_flags |= IFA_F_DEPRECATED;
 		ifa->ifa_preferred_lft = timeout;
 	}
-	ifa->ifa_tstamp = jiffies;
+	WRITE_ONCE(ifa->ifa_tstamp, jiffies);
 	if (!ifa->ifa_cstamp)
-		ifa->ifa_cstamp = ifa->ifa_tstamp;
+		WRITE_ONCE(ifa->ifa_cstamp, ifa->ifa_tstamp);
 }
 
 static struct in_ifaddr *rtm_to_ifaddr(struct net *net, struct nlmsghdr *nlh,
@@ -1676,6 +1677,7 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 {
 	struct ifaddrmsg *ifm;
 	struct nlmsghdr  *nlh;
+	unsigned long tstamp;
 	u32 preferred, valid;
 
 	nlh = nlmsg_put(skb, args->portid, args->seq, args->event, sizeof(*ifm),
@@ -1694,11 +1696,12 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 	    nla_put_s32(skb, IFA_TARGET_NETNSID, args->netnsid))
 		goto nla_put_failure;
 
+	tstamp = READ_ONCE(ifa->ifa_tstamp);
 	if (!(ifm->ifa_flags & IFA_F_PERMANENT)) {
 		preferred = ifa->ifa_preferred_lft;
 		valid = ifa->ifa_valid_lft;
 		if (preferred != INFINITY_LIFE_TIME) {
-			long tval = (jiffies - ifa->ifa_tstamp) / HZ;
+			long tval = (jiffies - tstamp) / HZ;
 
 			if (preferred > tval)
 				preferred -= tval;
@@ -1728,7 +1731,7 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 	    nla_put_u32(skb, IFA_FLAGS, ifa->ifa_flags) ||
 	    (ifa->ifa_rt_priority &&
 	     nla_put_u32(skb, IFA_RT_PRIORITY, ifa->ifa_rt_priority)) ||
-	    put_cacheinfo(skb, ifa->ifa_cstamp, ifa->ifa_tstamp,
+	    put_cacheinfo(skb, READ_ONCE(ifa->ifa_cstamp), tstamp,
 			  preferred, valid))
 		goto nla_put_failure;
 
-- 
2.44.0.278.ge034bb2e1d-goog

[PATCH net-next 2/6] inet: annotate data-races around ifa->ifa_valid_lft

[Eric Dumazet]

ifa->ifa_valid_lft can be read locklessly.

Add appropriate READ_ONCE()/WRITE_ONCE() annotations.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/devinet.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 1316046d5f28955376091d9e02ab4594e19fbd09..99f3e7c57d36ff028edadd4efd66d996ddc5d9b4 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -714,26 +714,26 @@ static void check_lifetime(struct work_struct *work)
 		rcu_read_lock();
 		hlist_for_each_entry_rcu(ifa, &inet_addr_lst[i], hash) {
 			unsigned long age, tstamp;
+			u32 valid_lft;
 
 			if (ifa->ifa_flags & IFA_F_PERMANENT)
 				continue;
 
+			valid_lft = READ_ONCE(ifa->ifa_valid_lft);
 			tstamp = READ_ONCE(ifa->ifa_tstamp);
 			/* We try to batch several events at once. */
 			age = (now - tstamp +
 			       ADDRCONF_TIMER_FUZZ_MINUS) / HZ;
 
-			if (ifa->ifa_valid_lft != INFINITY_LIFE_TIME &&
-			    age >= ifa->ifa_valid_lft) {
+			if (valid_lft != INFINITY_LIFE_TIME &&
+			    age >= valid_lft) {
 				change_needed = true;
 			} else if (ifa->ifa_preferred_lft ==
 				   INFINITY_LIFE_TIME) {
 				continue;
 			} else if (age >= ifa->ifa_preferred_lft) {
-				if (time_before(tstamp +
-						ifa->ifa_valid_lft * HZ, next))
-					next = tstamp +
-					       ifa->ifa_valid_lft * HZ;
+				if (time_before(tstamp + valid_lft * HZ, next))
+					next = tstamp + valid_lft * HZ;
 
 				if (!(ifa->ifa_flags & IFA_F_DEPRECATED))
 					change_needed = true;
@@ -810,7 +810,7 @@ static void set_ifa_lifetime(struct in_ifaddr *ifa, __u32 valid_lft,
 
 	timeout = addrconf_timeout_fixup(valid_lft, HZ);
 	if (addrconf_finite_timeout(timeout))
-		ifa->ifa_valid_lft = timeout;
+		WRITE_ONCE(ifa->ifa_valid_lft, timeout);
 	else
 		ifa->ifa_flags |= IFA_F_PERMANENT;
 
@@ -1699,7 +1699,7 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 	tstamp = READ_ONCE(ifa->ifa_tstamp);
 	if (!(ifm->ifa_flags & IFA_F_PERMANENT)) {
 		preferred = ifa->ifa_preferred_lft;
-		valid = ifa->ifa_valid_lft;
+		valid = READ_ONCE(ifa->ifa_valid_lft);
 		if (preferred != INFINITY_LIFE_TIME) {
 			long tval = (jiffies - tstamp) / HZ;
 
-- 
2.44.0.278.ge034bb2e1d-goog

[PATCH net-next 3/6] inet: annotate data-races around ifa->ifa_preferred_lft

[Eric Dumazet]

ifa->ifa_preferred_lft can be read locklessly.

Add appropriate READ_ONCE()/WRITE_ONCE() annotations.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/devinet.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 99f3e7c57d36ff028edadd4efd66d996ddc5d9b4..368fb56e1f1b2e3b7888f611a3f113f3bd5fc2af 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -714,11 +714,13 @@ static void check_lifetime(struct work_struct *work)
 		rcu_read_lock();
 		hlist_for_each_entry_rcu(ifa, &inet_addr_lst[i], hash) {
 			unsigned long age, tstamp;
+			u32 preferred_lft;
 			u32 valid_lft;
 
 			if (ifa->ifa_flags & IFA_F_PERMANENT)
 				continue;
 
+			preferred_lft = READ_ONCE(ifa->ifa_preferred_lft);
 			valid_lft = READ_ONCE(ifa->ifa_valid_lft);
 			tstamp = READ_ONCE(ifa->ifa_tstamp);
 			/* We try to batch several events at once. */
@@ -728,20 +730,18 @@ static void check_lifetime(struct work_struct *work)
 			if (valid_lft != INFINITY_LIFE_TIME &&
 			    age >= valid_lft) {
 				change_needed = true;
-			} else if (ifa->ifa_preferred_lft ==
+			} else if (preferred_lft ==
 				   INFINITY_LIFE_TIME) {
 				continue;
-			} else if (age >= ifa->ifa_preferred_lft) {
+			} else if (age >= preferred_lft) {
 				if (time_before(tstamp + valid_lft * HZ, next))
 					next = tstamp + valid_lft * HZ;
 
 				if (!(ifa->ifa_flags & IFA_F_DEPRECATED))
 					change_needed = true;
-			} else if (time_before(tstamp +
-					       ifa->ifa_preferred_lft * HZ,
+			} else if (time_before(tstamp + preferred_lft * HZ,
 					       next)) {
-				next = tstamp +
-				       ifa->ifa_preferred_lft * HZ;
+				next = tstamp + preferred_lft * HZ;
 			}
 		}
 		rcu_read_unlock();
@@ -818,7 +818,7 @@ static void set_ifa_lifetime(struct in_ifaddr *ifa, __u32 valid_lft,
 	if (addrconf_finite_timeout(timeout)) {
 		if (timeout == 0)
 			ifa->ifa_flags |= IFA_F_DEPRECATED;
-		ifa->ifa_preferred_lft = timeout;
+		WRITE_ONCE(ifa->ifa_preferred_lft, timeout);
 	}
 	WRITE_ONCE(ifa->ifa_tstamp, jiffies);
 	if (!ifa->ifa_cstamp)
@@ -1698,7 +1698,7 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 
 	tstamp = READ_ONCE(ifa->ifa_tstamp);
 	if (!(ifm->ifa_flags & IFA_F_PERMANENT)) {
-		preferred = ifa->ifa_preferred_lft;
+		preferred = READ_ONCE(ifa->ifa_preferred_lft);
 		valid = READ_ONCE(ifa->ifa_valid_lft);
 		if (preferred != INFINITY_LIFE_TIME) {
 			long tval = (jiffies - tstamp) / HZ;
-- 
2.44.0.278.ge034bb2e1d-goog

[PATCH net-next 4/6] inet: annotate data-races around ifa->ifa_flags

[Eric Dumazet]

ifa->ifa_flags can be read locklessly.

Add appropriate READ_ONCE()/WRITE_ONCE() annotations.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/devinet.c | 22 +++++++++++++---------
 1 file changed, 13 insertions(+), 9 deletions(-)

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 368fb56e1f1b2e3b7888f611a3f113f3bd5fc2af..550b775cbbf3c140c66e224c69996df7051b3d36 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -716,8 +716,10 @@ static void check_lifetime(struct work_struct *work)
 			unsigned long age, tstamp;
 			u32 preferred_lft;
 			u32 valid_lft;
+			u32 flags;
 
-			if (ifa->ifa_flags & IFA_F_PERMANENT)
+			flags = READ_ONCE(ifa->ifa_flags);
+			if (flags & IFA_F_PERMANENT)
 				continue;
 
 			preferred_lft = READ_ONCE(ifa->ifa_preferred_lft);
@@ -737,7 +739,7 @@ static void check_lifetime(struct work_struct *work)
 				if (time_before(tstamp + valid_lft * HZ, next))
 					next = tstamp + valid_lft * HZ;
 
-				if (!(ifa->ifa_flags & IFA_F_DEPRECATED))
+				if (!(flags & IFA_F_DEPRECATED))
 					change_needed = true;
 			} else if (time_before(tstamp + preferred_lft * HZ,
 					       next)) {
@@ -805,21 +807,23 @@ static void set_ifa_lifetime(struct in_ifaddr *ifa, __u32 valid_lft,
 			     __u32 prefered_lft)
 {
 	unsigned long timeout;
+	u32 flags;
 
-	ifa->ifa_flags &= ~(IFA_F_PERMANENT | IFA_F_DEPRECATED);
+	flags = ifa->ifa_flags & ~(IFA_F_PERMANENT | IFA_F_DEPRECATED);
 
 	timeout = addrconf_timeout_fixup(valid_lft, HZ);
 	if (addrconf_finite_timeout(timeout))
 		WRITE_ONCE(ifa->ifa_valid_lft, timeout);
 	else
-		ifa->ifa_flags |= IFA_F_PERMANENT;
+		flags |= IFA_F_PERMANENT;
 
 	timeout = addrconf_timeout_fixup(prefered_lft, HZ);
 	if (addrconf_finite_timeout(timeout)) {
 		if (timeout == 0)
-			ifa->ifa_flags |= IFA_F_DEPRECATED;
+			flags |= IFA_F_DEPRECATED;
 		WRITE_ONCE(ifa->ifa_preferred_lft, timeout);
 	}
+	WRITE_ONCE(ifa->ifa_flags, flags);
 	WRITE_ONCE(ifa->ifa_tstamp, jiffies);
 	if (!ifa->ifa_cstamp)
 		WRITE_ONCE(ifa->ifa_cstamp, ifa->ifa_tstamp);
@@ -1313,7 +1317,7 @@ static __be32 in_dev_select_addr(const struct in_device *in_dev,
 	const struct in_ifaddr *ifa;
 
 	in_dev_for_each_ifa_rcu(ifa, in_dev) {
-		if (ifa->ifa_flags & IFA_F_SECONDARY)
+		if (READ_ONCE(ifa->ifa_flags) & IFA_F_SECONDARY)
 			continue;
 		if (ifa->ifa_scope != RT_SCOPE_LINK &&
 		    ifa->ifa_scope <= scope)
@@ -1341,7 +1345,7 @@ __be32 inet_select_addr(const struct net_device *dev, __be32 dst, int scope)
 		localnet_scope = RT_SCOPE_LINK;
 
 	in_dev_for_each_ifa_rcu(ifa, in_dev) {
-		if (ifa->ifa_flags & IFA_F_SECONDARY)
+		if (READ_ONCE(ifa->ifa_flags) & IFA_F_SECONDARY)
 			continue;
 		if (min(ifa->ifa_scope, localnet_scope) > scope)
 			continue;
@@ -1688,7 +1692,7 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 	ifm = nlmsg_data(nlh);
 	ifm->ifa_family = AF_INET;
 	ifm->ifa_prefixlen = ifa->ifa_prefixlen;
-	ifm->ifa_flags = ifa->ifa_flags;
+	ifm->ifa_flags = READ_ONCE(ifa->ifa_flags);
 	ifm->ifa_scope = ifa->ifa_scope;
 	ifm->ifa_index = ifa->ifa_dev->dev->ifindex;
 
@@ -1728,7 +1732,7 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
 	     nla_put_string(skb, IFA_LABEL, ifa->ifa_label)) ||
 	    (ifa->ifa_proto &&
 	     nla_put_u8(skb, IFA_PROTO, ifa->ifa_proto)) ||
-	    nla_put_u32(skb, IFA_FLAGS, ifa->ifa_flags) ||
+	    nla_put_u32(skb, IFA_FLAGS, ifm->ifa_flags) ||
 	    (ifa->ifa_rt_priority &&
 	     nla_put_u32(skb, IFA_RT_PRIORITY, ifa->ifa_rt_priority)) ||
 	    put_cacheinfo(skb, READ_ONCE(ifa->ifa_cstamp), tstamp,
-- 
2.44.0.278.ge034bb2e1d-goog

[PATCH net-next 5/6] inet: prepare inet_base_seq() to run without RTNL

[Eric Dumazet]

In the following patch, inet_base_seq() will no longer be called
with RTNL held.

Add READ_ONCE()/WRITE_ONCE() annotations in dev_base_seq_inc()
and inet_base_seq().

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/core/dev.c     | 5 +++--
 net/ipv4/devinet.c | 2 +-
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 053fac78305c7322b894ceb07a925f7e64ed70aa..873e095e141db3e631e51763435ddafbf0d280af 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -180,8 +180,9 @@ static DECLARE_RWSEM(devnet_rename_sem);
 
 static inline void dev_base_seq_inc(struct net *net)
 {
-	while (++net->dev_base_seq == 0)
-		;
+	unsigned int val = net->dev_base_seq + 1;
+
+	WRITE_ONCE(net->dev_base_seq, val ?: 1);
 }
 
 static inline struct hlist_head *dev_name_hash(struct net *net, const char *name)
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 550b775cbbf3c140c66e224c69996df7051b3d36..2afe78dfc3c2f6c0394925f1c35532a2dfd26d71 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1837,7 +1837,7 @@ static int in_dev_dump_addr(struct in_device *in_dev, struct sk_buff *skb,
 static u32 inet_base_seq(const struct net *net)
 {
 	u32 res = atomic_read(&net->ipv4.dev_addr_genid) +
-		  net->dev_base_seq;
+		  READ_ONCE(net->dev_base_seq);
 
 	/* Must not return 0 (see nl_dump_check_consistent()).
 	 * Chose a value far away from 0.
-- 
2.44.0.278.ge034bb2e1d-goog

[PATCH net-next 6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()

[Eric Dumazet]

1) inet_dump_ifaddr() can can run under RCU protection
   instead of RTNL.

2) properly return 0 at the end of a dump, avoiding an
   an extra recvmsg() system call.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/devinet.c | 95 ++++++++++++++++++----------------------------
 1 file changed, 37 insertions(+), 58 deletions(-)

diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 2afe78dfc3c2f6c0394925f1c35532a2dfd26d71..4daa8124f247c256c4f8c1ff29ac621570af0755 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1676,7 +1676,7 @@ static int put_cacheinfo(struct sk_buff *skb, unsigned long cstamp,
 	return nla_put(skb, IFA_CACHEINFO, sizeof(ci), &ci);
 }
 
-static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
+static int inet_fill_ifaddr(struct sk_buff *skb, const struct in_ifaddr *ifa,
 			    struct inet_fill_args *args)
 {
 	struct ifaddrmsg *ifm;
@@ -1805,15 +1805,15 @@ static int inet_valid_dump_ifaddr_req(const struct nlmsghdr *nlh,
 }
 
 static int in_dev_dump_addr(struct in_device *in_dev, struct sk_buff *skb,
-			    struct netlink_callback *cb, int s_ip_idx,
+			    struct netlink_callback *cb, int *s_ip_idx,
 			    struct inet_fill_args *fillargs)
 {
 	struct in_ifaddr *ifa;
 	int ip_idx = 0;
 	int err;
 
-	in_dev_for_each_ifa_rtnl(ifa, in_dev) {
-		if (ip_idx < s_ip_idx) {
+	in_dev_for_each_ifa_rcu(ifa, in_dev) {
+		if (ip_idx < *s_ip_idx) {
 			ip_idx++;
 			continue;
 		}
@@ -1825,9 +1825,9 @@ static int in_dev_dump_addr(struct in_device *in_dev, struct sk_buff *skb,
 		ip_idx++;
 	}
 	err = 0;
-
+	ip_idx = 0;
 done:
-	cb->args[2] = ip_idx;
+	*s_ip_idx = ip_idx;
 
 	return err;
 }
@@ -1859,75 +1859,53 @@ static int inet_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb)
 	};
 	struct net *net = sock_net(skb->sk);
 	struct net *tgt_net = net;
-	int h, s_h;
-	int idx, s_idx;
-	int s_ip_idx;
-	struct net_device *dev;
+	struct {
+		unsigned long ifindex;
+		int ip_idx;
+	} *ctx = (void *)cb->ctx;
 	struct in_device *in_dev;
-	struct hlist_head *head;
+	struct net_device *dev;
 	int err = 0;
 
-	s_h = cb->args[0];
-	s_idx = idx = cb->args[1];
-	s_ip_idx = cb->args[2];
-
+	rcu_read_lock();
 	if (cb->strict_check) {
 		err = inet_valid_dump_ifaddr_req(nlh, &fillargs, &tgt_net,
 						 skb->sk, cb);
 		if (err < 0)
-			goto put_tgt_net;
+			goto done;
 
-		err = 0;
 		if (fillargs.ifindex) {
-			dev = __dev_get_by_index(tgt_net, fillargs.ifindex);
-			if (!dev) {
-				err = -ENODEV;
-				goto put_tgt_net;
-			}
-
-			in_dev = __in_dev_get_rtnl(dev);
-			if (in_dev) {
-				err = in_dev_dump_addr(in_dev, skb, cb, s_ip_idx,
-						       &fillargs);
-			}
-			goto put_tgt_net;
-		}
-	}
-
-	for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) {
-		idx = 0;
-		head = &tgt_net->dev_index_head[h];
-		rcu_read_lock();
-		cb->seq = inet_base_seq(tgt_net);
-		hlist_for_each_entry_rcu(dev, head, index_hlist) {
-			if (idx < s_idx)
-				goto cont;
-			if (h > s_h || idx > s_idx)
-				s_ip_idx = 0;
+			err = -ENODEV;
+			dev = dev_get_by_index_rcu(tgt_net, fillargs.ifindex);
+			if (!dev)
+				goto done;
 			in_dev = __in_dev_get_rcu(dev);
 			if (!in_dev)
-				goto cont;
-
-			err = in_dev_dump_addr(in_dev, skb, cb, s_ip_idx,
-					       &fillargs);
-			if (err < 0) {
-				rcu_read_unlock();
 				goto done;
-			}
-cont:
-			idx++;
+			err = in_dev_dump_addr(in_dev, skb, cb, &ctx->ip_idx,
+					       &fillargs);
+			goto done;
 		}
-		rcu_read_unlock();
 	}
 
+	cb->seq = inet_base_seq(tgt_net);
+
+	for_each_netdev_dump(net, dev, ctx->ifindex) {
+		in_dev = __in_dev_get_rcu(dev);
+		if (!in_dev)
+			continue;
+		err = in_dev_dump_addr(in_dev, skb, cb, &ctx->ip_idx,
+				       &fillargs);
+		if (err < 0)
+			goto done;
+	}
 done:
-	cb->args[0] = h;
-	cb->args[1] = idx;
-put_tgt_net:
+	if (err < 0 && likely(skb->len))
+		err = skb->len;
 	if (fillargs.netnsid >= 0)
 		put_net(tgt_net);
-
-	return skb->len ? : err;
+	rcu_read_unlock();
+	return err;
 }
 
 static void rtmsg_ifa(int event, struct in_ifaddr *ifa, struct nlmsghdr *nlh,
@@ -2818,7 +2796,8 @@ void __init devinet_init(void)
 
 	rtnl_register(PF_INET, RTM_NEWADDR, inet_rtm_newaddr, NULL, 0);
 	rtnl_register(PF_INET, RTM_DELADDR, inet_rtm_deladdr, NULL, 0);
-	rtnl_register(PF_INET, RTM_GETADDR, NULL, inet_dump_ifaddr, 0);
+	rtnl_register(PF_INET, RTM_GETADDR, NULL, inet_dump_ifaddr,
+		      RTNL_FLAG_DUMP_UNLOCKED);
 	rtnl_register(PF_INET, RTM_GETNETCONF, inet_netconf_get_devconf,
 		      inet_netconf_dump_devconf,
 		      RTNL_FLAG_DOIT_UNLOCKED | RTNL_FLAG_DUMP_UNLOCKED);
-- 
2.44.0.278.ge034bb2e1d-goog

Re: [PATCH net-next 0/6] inet: no longer use RTNL to protect inet_dump_ifaddr()

[Jiri Pirko]

Thu, Feb 29, 2024 at 12:40:10PM CET, edumazet@google.com wrote:
>This series convert inet so that a dump of addresses (ip -4 addr)
>no longer requires RTNL.
>
>Eric Dumazet (6):
>  inet: annotate data-races around ifa->ifa_tstamp and ifa->ifa_cstamp
>  inet: annotate data-races around ifa->ifa_valid_lft
>  inet: annotate data-races around ifa->ifa_preferred_lft
>  inet: annotate data-races around ifa->ifa_flags
>  inet: prepare inet_base_seq() to run without RTNL
>  inet: use xa_array iterator to implement inet_dump_ifaddr()
>
> net/core/dev.c     |   5 +-
> net/ipv4/devinet.c | 166 +++++++++++++++++++++------------------------
> 2 files changed, 79 insertions(+), 92 deletions(-)

Looks fine to me.

set-
Reviewed-by: Jiri Pirko <jiri@nvidia.com>

Re: [PATCH net-next 6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()

[Jakub Kicinski]

On Thu, 29 Feb 2024 11:40:16 +0000 Eric Dumazet wrote:
> +	if (err < 0 && likely(skb->len))
> +		err = skb->len;

I think Ido may have commented on one of your early series, but if we
set err to skb->len we'll have to do an extra empty message to terminate
the dump.

You basically only want to return skb->len when message has
overflown, so the somewhat idiomatic way to do this is:

	err = (err == -EMSGSIZE) ? skb->len : err;

Assuming err can't be set to some weird positive value.

IDK if you want to do this in future patches or it's risky, but I have
the itch to tell you every time I see a conversion which doesn't follow
this pattern :)

Re: [PATCH net-next 6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()

[Eric Dumazet]

On Thu, Feb 29, 2024 at 4:37 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Thu, 29 Feb 2024 11:40:16 +0000 Eric Dumazet wrote:
> > +     if (err < 0 && likely(skb->len))
> > +             err = skb->len;
>
> I think Ido may have commented on one of your early series, but if we
> set err to skb->len we'll have to do an extra empty message to terminate
> the dump.
>
> You basically only want to return skb->len when message has
> overflown, so the somewhat idiomatic way to do this is:
>
>         err = (err == -EMSGSIZE) ? skb->len : err;
>
> Assuming err can't be set to some weird positive value.
>
> IDK if you want to do this in future patches or it's risky, but I have
> the itch to tell you every time I see a conversion which doesn't follow
> this pattern :)

This totally makes sense.

I will send a followup patch to fix all these in one go, if this is ok
with you ?

Thanks.

Re: [PATCH net-next 6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()

[Eric Dumazet]

On Thu, Feb 29, 2024 at 4:45 PM Eric Dumazet <edumazet@google.com> wrote:
>
> On Thu, Feb 29, 2024 at 4:37 PM Jakub Kicinski <kuba@kernel.org> wrote:
> >
> > On Thu, 29 Feb 2024 11:40:16 +0000 Eric Dumazet wrote:
> > > +     if (err < 0 && likely(skb->len))
> > > +             err = skb->len;
> >
> > I think Ido may have commented on one of your early series, but if we
> > set err to skb->len we'll have to do an extra empty message to terminate
> > the dump.
> >
> > You basically only want to return skb->len when message has
> > overflown, so the somewhat idiomatic way to do this is:
> >
> >         err = (err == -EMSGSIZE) ? skb->len : err;

This would set err to zero if skb is empty at this point.

I guess a more correct action would be:

if (err == -EMSGSIZE && likely(skb->len))
    err = skb->len;


> >
> > Assuming err can't be set to some weird positive value.
> >
> > IDK if you want to do this in future patches or it's risky, but I have
> > the itch to tell you every time I see a conversion which doesn't follow
> > this pattern :)
>
> This totally makes sense.
>
> I will send a followup patch to fix all these in one go, if this is ok
> with you ?
>
> Thanks.

Re: [PATCH net-next 6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()

[Jakub Kicinski]

On Thu, 29 Feb 2024 16:50:45 +0100 Eric Dumazet wrote:
> > > You basically only want to return skb->len when message has
> > > overflown, so the somewhat idiomatic way to do this is:
> > >
> > >         err = (err == -EMSGSIZE) ? skb->len : err;  
> 
> This would set err to zero if skb is empty at this point.
> 
> I guess a more correct action would be:
> 
> if (err == -EMSGSIZE && likely(skb->len))
>     err = skb->len;

Ugh, fair point.
We should probably move the EMSGSIZE handling to the core, this is
getting too complicated for average humans.. Like this?

diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 765921cf1194..ce27003b90a8 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -2264,6 +2264,8 @@ static int netlink_dump(struct sock *sk, bool lock_taken)
 		if (extra_mutex)
 			mutex_lock(extra_mutex);
 		nlk->dump_done_errno = cb->dump(skb, cb);
+		if (nlk->dump_done_errno == -EMSGSIZE && skb->len)
+			nlk->dump_done_errno = skb->len;
 		if (extra_mutex)
 			mutex_unlock(extra_mutex);

> > >
> > > Assuming err can't be set to some weird positive value.
> > >
> > > IDK if you want to do this in future patches or it's risky, but I have
> > > the itch to tell you every time I see a conversion which doesn't follow
> > > this pattern :)  
> >
> > This totally makes sense.
> >
> > I will send a followup patch to fix all these in one go, if this is ok
> > with you ?

Definitely not a blocker

Re: [PATCH net-next 6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()

[Eric Dumazet]

On Thu, Feb 29, 2024 at 5:21 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Thu, 29 Feb 2024 16:50:45 +0100 Eric Dumazet wrote:
> > > > You basically only want to return skb->len when message has
> > > > overflown, so the somewhat idiomatic way to do this is:
> > > >
> > > >         err = (err == -EMSGSIZE) ? skb->len : err;
> >
> > This would set err to zero if skb is empty at this point.
> >
> > I guess a more correct action would be:
> >
> > if (err == -EMSGSIZE && likely(skb->len))
> >     err = skb->len;
>
> Ugh, fair point.
> We should probably move the EMSGSIZE handling to the core, this is
> getting too complicated for average humans.. Like this?
>

Now you are talking ! Yep, definitely nicer.

> diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
> index 765921cf1194..ce27003b90a8 100644
> --- a/net/netlink/af_netlink.c
> +++ b/net/netlink/af_netlink.c
> @@ -2264,6 +2264,8 @@ static int netlink_dump(struct sock *sk, bool lock_taken)
>                 if (extra_mutex)
>                         mutex_lock(extra_mutex);
>                 nlk->dump_done_errno = cb->dump(skb, cb);
> +               if (nlk->dump_done_errno == -EMSGSIZE && skb->len)
> +                       nlk->dump_done_errno = skb->len;
>                 if (extra_mutex)
>                         mutex_unlock(extra_mutex);
>
> > > >
> > > > Assuming err can't be set to some weird positive value.
> > > >
> > > > IDK if you want to do this in future patches or it's risky, but I have
> > > > the itch to tell you every time I see a conversion which doesn't follow
> > > > this pattern :)
> > >
> > > This totally makes sense.
> > >
> > > I will send a followup patch to fix all these in one go, if this is ok
> > > with you ?
>
> Definitely not a blocker

Re: [PATCH net-next 0/6] inet: no longer use RTNL to protect inet_dump_ifaddr()

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (main)
by David S. Miller <davem@davemloft.net>:

On Thu, 29 Feb 2024 11:40:10 +0000 you wrote:
> This series convert inet so that a dump of addresses (ip -4 addr)
> no longer requires RTNL.
> 
> Eric Dumazet (6):
>   inet: annotate data-races around ifa->ifa_tstamp and ifa->ifa_cstamp
>   inet: annotate data-races around ifa->ifa_valid_lft
>   inet: annotate data-races around ifa->ifa_preferred_lft
>   inet: annotate data-races around ifa->ifa_flags
>   inet: prepare inet_base_seq() to run without RTNL
>   inet: use xa_array iterator to implement inet_dump_ifaddr()
> 
> [...]

Here is the summary with links:
  - [net-next,1/6] inet: annotate data-races around ifa->ifa_tstamp and ifa->ifa_cstamp
    https://git.kernel.org/netdev/net-next/c/3cd3e72ccb3a
  - [net-next,2/6] inet: annotate data-races around ifa->ifa_valid_lft
    https://git.kernel.org/netdev/net-next/c/a5fcf74d80be
  - [net-next,3/6] inet: annotate data-races around ifa->ifa_preferred_lft
    https://git.kernel.org/netdev/net-next/c/9f6fa3c4e722
  - [net-next,4/6] inet: annotate data-races around ifa->ifa_flags
    https://git.kernel.org/netdev/net-next/c/3ddc2231c810
  - [net-next,5/6] inet: prepare inet_base_seq() to run without RTNL
    https://git.kernel.org/netdev/net-next/c/590e92cdc835
  - [net-next,6/6] inet: use xa_array iterator to implement inet_dump_ifaddr()
    https://git.kernel.org/netdev/net-next/c/cdb2f80f1c10

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html