Skip to contentSkip to conversation

BPF path manager, part 3

PATCH In review Geliang Tang 3 participants 11 messages Jan 24, 2025

Subsystems (2)
lore
Share & tools
lore

Geliang Tang

author reviewer
#0 lore raw

v5 cover BPF path manager, part 3

From: Geliang Tang <[email protected]>

v5:
- drop "change is_backup interfaces as get_flags"
- move "add struct mptcp_pm_addr_id_bitmap" and "hold msk lock before
removing id 0 address" out of this set.
- a new patch "use ipv6_addr_equal in addresses_equal"
- patch 4, drop inet6_sk() in mptcp_event_add_subflow() and mptcp_event_pm_listener() too.

v4:
- add a new patch, "add a build check for userspace_pm_dump_addr" as Matt suggested.
- patch 4, fix a warning here reported by CI:
WARNING: line length of 84 exceeds 80 columns
#27: FILE: net/mptcp/pm_userspace.c:468:

v3:
- drop "mptcp: update address type of get_local_id".

v2:
- a new patch "hold msk lock before removing id 0 address"

Drop the patch "mptcp: reuse sending nlmsg code in dump_addr", add a new helper mptcp_pm_genl_fill_addr() to save the duplicated code.

Define struct mptcp_pm_addr_id_bitmap in protocol.h, instead of defining mptcp_pm_addr_id_bitmap_t in include/net/mptcp.h, since get_addr() and dump_addr() interfaces of BPF userspace pm is dropped as Matt suggested.

Geliang Tang (5):
mptcp: add a build check for userspace_pm_dump_addr
mptcp: add mptcp_pm_genl_fill_addr helper
mptcp: drop match in userspace_pm_append_new_local_addr
mptcp: drop inet6_sk after inet_sk
mptcp: use ipv6_addr_equal in addresses_equal

 net/mptcp/pm.c           | 21 +++++++++++++++++++++
 net/mptcp/pm_netlink.c   | 22 ++++------------------
 net/mptcp/pm_userspace.c | 29 +++++++----------------------
 net/mptcp/protocol.h     |  3 +++
 4 files changed, 35 insertions(+), 40 deletions(-)
Signature · 1 line
2.43.0
author reviewer
#1 lore raw

PATCH v5 1/5 mptcp: add a build check for userspace_pm_dump_addr

From: Geliang Tang <[email protected]>

This patch adds a build check for mptcp_userspace_pm_dump_addr() to make sure there is enough space in 'cb->ctx' to store an address id bitmap.

---
 net/mptcp/pm_userspace.c | 2 ++
 1 file changed, 2 insertions(+)
net/mptcp/pm_userspace.c +2 -0
index 277cf092a870..b69fb5b18130 100644
@@ -643,6 +643,8 @@ int mptcp_userspace_pm_dump_addr(struct sk_buff *msg,
 	struct sock *sk;
 	void *hdr;
 
+	BUILD_BUG_ON(sizeof(struct id_bitmap) > sizeof(cb->ctx));
+
 	bitmap = (struct id_bitmap *)cb->ctx;
 
 	msk = mptcp_userspace_pm_get_sock(info);
Signature · 1 line
2.43.0
author reviewer
#2 lore raw

PATCH v5 2/5 mptcp: add mptcp_pm_genl_fill_addr helper

In reply to Geliang Tang

From: Geliang Tang <[email protected]>

To save some redundant code in dump_addr() interfaces of both the netlink PM and userspace PM, the code that calls netlink message helpers (genlmsg_put/cancel/end) and mptcp_nl_fill_addr() is wrapped into a new helper mptcp_pm_genl_fill_addr().

---
 net/mptcp/pm.c           | 21 +++++++++++++++++++++
 net/mptcp/pm_netlink.c   | 12 +-----------
 net/mptcp/pm_userspace.c | 12 +-----------
 net/mptcp/protocol.h     |  3 +++
 4 files changed, 26 insertions(+), 22 deletions(-)
net/mptcp/pm.c +21 -0
index b1f36dc1a091..16cacce6c10f 100644
@@ -489,6 +489,27 @@ int mptcp_pm_nl_get_addr_doit(struct sk_buff *skb, struct genl_info *info)
 	return ret;
 }
 
+int mptcp_pm_genl_fill_addr(struct sk_buff *msg,
+			    struct netlink_callback *cb,
+			    struct mptcp_pm_addr_entry *entry)
+{
+	void *hdr;
+
+	hdr = genlmsg_put(msg, NETLINK_CB(cb->skb).portid,
+			  cb->nlh->nlmsg_seq, &mptcp_genl_family,
+			  NLM_F_MULTI, MPTCP_PM_CMD_GET_ADDR);
+	if (!hdr)
+		return -EINVAL;
+
+	if (mptcp_nl_fill_addr(msg, entry) < 0) {
+		genlmsg_cancel(msg, hdr);
+		return -EINVAL;
+	}
+
+	genlmsg_end(msg, hdr);
+	return 0;
+}
+
 static int mptcp_pm_dump_addr(struct sk_buff *msg, struct netlink_callback *cb)
 {
 	const struct genl_info *info = genl_info_dump(cb);
net/mptcp/pm_netlink.c +1 -11
index 99705a9c2238..02e2945911df 100644
@@ -1798,7 +1798,6 @@ int mptcp_pm_nl_dump_addr(struct sk_buff *msg,
 	struct mptcp_pm_addr_entry *entry;
 	struct pm_nl_pernet *pernet;
 	int id = cb->args[0];
-	void *hdr;
 	int i;
 
 	pernet = pm_nl_get_pernet(net);
@@ -1813,19 +1812,10 @@ int mptcp_pm_nl_dump_addr(struct sk_buff *msg,
 			if (entry->addr.id <= id)
 				continue;
 
-			hdr = genlmsg_put(msg, NETLINK_CB(cb->skb).portid,
-					  cb->nlh->nlmsg_seq, &mptcp_genl_family,
-					  NLM_F_MULTI, MPTCP_PM_CMD_GET_ADDR);
-			if (!hdr)
+			if (mptcp_pm_genl_fill_addr(msg, cb, entry) < 0)
 				break;
 
-			if (mptcp_nl_fill_addr(msg, entry) < 0) {
-				genlmsg_cancel(msg, hdr);
-				break;
-			}
-
 			id = entry->addr.id;
-			genlmsg_end(msg, hdr);
 		}
 	}
 	rcu_read_unlock();
net/mptcp/pm_userspace.c +1 -11
index b69fb5b18130..bedd6f9ebc8b 100644
@@ -641,7 +641,6 @@ int mptcp_userspace_pm_dump_addr(struct sk_buff *msg,
 	struct mptcp_sock *msk;
 	int ret = -EINVAL;
 	struct sock *sk;
-	void *hdr;
 
 	BUILD_BUG_ON(sizeof(struct id_bitmap) > sizeof(cb->ctx));
 
@@ -659,19 +658,10 @@ int mptcp_userspace_pm_dump_addr(struct sk_buff *msg,
 		if (test_bit(entry->addr.id, bitmap->map))
 			continue;
 
-		hdr = genlmsg_put(msg, NETLINK_CB(cb->skb).portid,
-				  cb->nlh->nlmsg_seq, &mptcp_genl_family,
-				  NLM_F_MULTI, MPTCP_PM_CMD_GET_ADDR);
-		if (!hdr)
+		if (mptcp_pm_genl_fill_addr(msg, cb, entry) < 0)
 			break;
 
-		if (mptcp_nl_fill_addr(msg, entry) < 0) {
-			genlmsg_cancel(msg, hdr);
-			break;
-		}
-
 		__set_bit(entry->addr.id, bitmap->map);
-		genlmsg_end(msg, hdr);
 	}
 	spin_unlock_bh(&msk->pm.lock);
 	release_sock(sk);
net/mptcp/protocol.h +3 -0
index 2c1232bfd722..bac5c925a72f 100644
@@ -1061,6 +1061,9 @@ void mptcp_fastopen_subflow_synack_set_params(struct mptcp_subflow_context *subf
 					      struct request_sock *req);
 int mptcp_nl_fill_addr(struct sk_buff *skb,
 		       struct mptcp_pm_addr_entry *entry);
+int mptcp_pm_genl_fill_addr(struct sk_buff *msg,
+			    struct netlink_callback *cb,
+			    struct mptcp_pm_addr_entry *entry);
 
 static inline bool mptcp_pm_should_add_signal(struct mptcp_sock *msk)
 {
Signature · 1 line
2.43.0
maintainer Reviewed-by on v5 2/5
#3 lore raw

Re: [PATCH mptcp-next v5 2/5] mptcp: add mptcp_pm_genl_fill_addr helper

Hi Geliang,

Quoted 6 lines · Geliang Tang

From: Geliang Tang <[email protected]>

To save some redundant code in dump_addr() interfaces of both the netlink PM and userspace PM, the code that calls netlink message helpers (genlmsg_put/cancel/end) and mptcp_nl_fill_addr() is wrapped into a new helper mptcp_pm_genl_fill_addr().

Thank you for also having added '< 0' when looking for errors from mptcp_pm_genl_fill_addr().

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
author reviewer
#4 lore raw

PATCH v5 3/5 mptcp: drop match in userspace_pm_append_new_local_addr

In reply to Geliang Tang

From: Geliang Tang <[email protected]>

The variable 'match' in mptcp_userspace_pm_append_new_local_addr() is a redundant one, and this patch drops it.

No need to define 'match' as 'struct mptcp_pm_addr_entry *' type. In this function, it's only used to check whether it's NULL. It can be defined as a Boolean one.

Also other variables 'addr_match' and 'id_match' make 'match' a redundant one, which can be replaced by directly checking 'addr_match && id_match'.

---
 net/mptcp/pm_userspace.c | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)
net/mptcp/pm_userspace.c +3 -8
index bedd6f9ebc8b..a16e2fb45a6c 100644
@@ -48,7 +48,6 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 						    bool needs_id)
 {
 	DECLARE_BITMAP(id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
-	struct mptcp_pm_addr_entry *match = NULL;
 	struct sock *sk = (struct sock *)msk;
 	struct mptcp_pm_addr_entry *e;
 	bool addr_match = false;
@@ -63,16 +62,12 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 		if (addr_match && entry->addr.id == 0 && needs_id)
 			entry->addr.id = e->addr.id;
 		id_match = (e->addr.id == entry->addr.id);
-		if (addr_match && id_match) {
-			match = e;
+		if (addr_match || id_match)
 			break;
-		} else if (addr_match || id_match) {
-			break;
-		}
 		__set_bit(e->addr.id, id_bitmap);
 	}
 
-	if (!match && !addr_match && !id_match) {
+	if (!addr_match && !id_match) {
 		/* Memory for the entry is allocated from the
 		 * sock option buffer.
 		 */
@@ -90,7 +85,7 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 		list_add_tail_rcu(&e->list, &msk->pm.userspace_pm_local_addr_list);
 		msk->pm.local_addr_used++;
 		ret = e->addr.id;
-	} else if (match) {
+	} else if (addr_match && id_match) {
 		ret = entry->addr.id;
 	}
 
Signature · 1 line
2.43.0
author reviewer
#5 lore raw

PATCH v5 4/5 mptcp: drop inet6_sk after inet_sk

In reply to Geliang Tang

From: Geliang Tang <[email protected]>

In mptcp_event_add_subflow(), mptcp_event_pm_listener() and mptcp_nl_find_ssk(), 'issk' has already been got through inet_sk(). No need to use inet6_sk() to get 'ipv6_pinfo' again, just use issk->pinet6 instead. This patch also drops these 'ipv6_pinfo' variables.

---
v2:
- drop inet6_sk() in mptcp_event_add_subflow() and mptcp_event_pm_listener() too.

---
 net/mptcp/pm_netlink.c   | 8 ++------
 net/mptcp/pm_userspace.c | 4 +---
 2 files changed, 3 insertions(+), 9 deletions(-)
net/mptcp/pm_netlink.c +2 -6
index 02e2945911df..1462dc7e886d 100644
@@ -2022,9 +2022,7 @@ static int mptcp_event_add_subflow(struct sk_buff *skb, const struct sock *ssk)
 		break;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 	case AF_INET6: {
-		const struct ipv6_pinfo *np = inet6_sk(ssk);
-
-		if (nla_put_in6_addr(skb, MPTCP_ATTR_SADDR6, &np->saddr))
+		if (nla_put_in6_addr(skb, MPTCP_ATTR_SADDR6, &issk->pinet6->saddr))
 			return -EMSGSIZE;
 		if (nla_put_in6_addr(skb, MPTCP_ATTR_DADDR6, &ssk->sk_v6_daddr))
 			return -EMSGSIZE;
@@ -2251,9 +2249,7 @@ void mptcp_event_pm_listener(const struct sock *ssk,
 		break;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 	case AF_INET6: {
-		const struct ipv6_pinfo *np = inet6_sk(ssk);
-
-		if (nla_put_in6_addr(skb, MPTCP_ATTR_SADDR6, &np->saddr))
+		if (nla_put_in6_addr(skb, MPTCP_ATTR_SADDR6, &issk->pinet6->saddr))
 			goto nla_put_failure;
 		break;
 	}
net/mptcp/pm_userspace.c +1 -3
index a16e2fb45a6c..6bf6a20ef7f3 100644
@@ -460,9 +460,7 @@ static struct sock *mptcp_nl_find_ssk(struct mptcp_sock *msk,
 			break;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 		case AF_INET6: {
-			const struct ipv6_pinfo *pinfo = inet6_sk(ssk);
-
-			if (!ipv6_addr_equal(&local->addr6, &pinfo->saddr) ||
+			if (!ipv6_addr_equal(&local->addr6, &issk->pinet6->saddr) ||
 			    !ipv6_addr_equal(&remote->addr6, &ssk->sk_v6_daddr))
 				continue;
 			break;
Signature · 1 line
2.43.0
maintainer
#6 lore raw

Re: [PATCH mptcp-next v5 4/5] mptcp: drop inet6_sk after inet_sk

Hi Geliang,

Quoted 8 lines · Geliang Tang

From: Geliang Tang <[email protected]>

In mptcp_event_add_subflow(), mptcp_event_pm_listener() and mptcp_nl_find_ssk(), 'issk' has already been got through inet_sk(). No need to use inet6_sk() to get 'ipv6_pinfo' again, just use issk->pinet6 instead. This patch also drops these 'ipv6_pinfo' variables.

Quoted 6 lines

---
v2:
- drop inet6_sk() in mptcp_event_add_subflow() and mptcp_event_pm_listener() too.

Thank you for this comment, that's helpful!

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
author reviewer
#7 lore raw

PATCH v5 5/5 mptcp: use ipv6_addr_equal in addresses_equal

In reply to Geliang Tang

From: Geliang Tang <[email protected]>

Use ipv6_addr_equal() to check whether two IPv6 addresses are equal in mptcp_addresses_equal() is more appropriate than using ipv6_addr_cmp().

---
 net/mptcp/pm_netlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/mptcp/pm_netlink.c +1 -1
index 1462dc7e886d..6461aa742385 100644
@@ -64,7 +64,7 @@ bool mptcp_addresses_equal(const struct mptcp_addr_info *a,
 			addr_equals = a->addr.s_addr == b->addr.s_addr;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 		else
-			addr_equals = !ipv6_addr_cmp(&a->addr6, &b->addr6);
+			addr_equals = ipv6_addr_equal(&a->addr6, &b->addr6);
 	} else if (a->family == AF_INET) {
 		if (ipv6_addr_v4mapped(&b->addr6))
 			addr_equals = a->addr.s_addr == b->addr6.s6_addr32[3];
Signature · 1 line
2.43.0
maintainer Reviewed-by on v5 5/5
#8 lore raw

Re: [PATCH mptcp-next v5 5/5] mptcp: use ipv6_addr_equal in addresses_equal

Hi Geliang,

Quoted 15 lines · Geliang Tang

From: Geliang Tang <[email protected]>

Use ipv6_addr_equal() to check whether two IPv6 addresses are equal in mptcp_addresses_equal() is more appropriate than using ipv6_addr_cmp().

---
 net/mptcp/pm_netlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/mptcp/pm_netlink.c +0 -0
index 1462dc7e886d..6461aa742385 100644

net/mptcp/pm_netlink.c bool mptcp_addresses_equal(const struct mptcp_addr_info *a, · line 67

Quoted 5 lines
net/mptcp/pm_netlink.c +1 -1
@@ -64,7 +64,7 @@ bool mptcp_addresses_equal(const struct mptcp_addr_info *a,
 			addr_equals = a->addr.s_addr == b->addr.s_addr;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
 		else
-			addr_equals = !ipv6_addr_cmp(&a->addr6, &b->addr6);
+			addr_equals = ipv6_addr_equal(&a->addr6, &b->addr6);

Good idea, that looks more logical!

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.
#9 lore raw
In reply to Geliang Tang

Hi Geliang,

Thank you for your modifications, that's great!

Our CI did some validations and here is its report:

- KVM Validation: normal: Success! ✅
- KVM Validation: debug: Critical: Global Timeout ❌
- KVM Validation: btf-normal (only bpftest_all): Success! ✅
- KVM Validation: btf-debug (only bpftest_all): Unstable: 1 failed test(s): bpftest_test_progs-cpuv4_mptcp 🔴
- Task: https://github.com/multipath-tcp/mptcp_net-next/actions/runs/12945359042

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/6e73d43e9da0
Patchwork: https://patchwork.kernel.org/project/mptcp/list/?series=928039

If there are some issues, you can reproduce them using the same environment as the one used by the CI thanks to a docker image, e.g.:

    $ cd [kernel source code]
    $ docker run -v "${PWD}:${PWD}:rw" -w "${PWD}" --privileged --rm -it \
        --pull always mptcp/mptcp-upstream-virtme-docker:latest \
        auto-normal

For more details:

https://github.com/multipath-tcp/mptcp-upstream-virtme-docker

Please note that despite all the efforts that have been already done to have a stable tests suite when executed on a public CI like here, it is possible some reported issues are not due to your modifications. Still, do not hesitate to help us improve that ;-)

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (NGI0 Core)

maintainer
#10 lore raw
In reply to Geliang Tang

Hi Geliang,

Quoted 9 lines · Geliang Tang

From: Geliang Tang <[email protected]>

v5:
- drop "change is_backup interfaces as get_flags"
- move "add struct mptcp_pm_addr_id_bitmap" and "hold msk lock before
removing id 0 address" out of this set.
- a new patch "use ipv6_addr_equal in addresses_equal"
- patch 4, drop inet6_sk() in mptcp_event_add_subflow() and mptcp_event_pm_listener() too.

Thank you for the new version!

Now in our tree (feat. for net-next):

New patches for t/upstream:
- 8b880bfa4362: mptcp: add a build check for userspace_pm_dump_addr
- 5c44b4a4e41f: mptcp: add mptcp_pm_genl_fill_addr helper
- f0d7ae749373: mptcp: drop match in userspace_pm_append_new_local_addr
- be35f8e33917: mptcp: drop inet6_sk after inet_sk
- 6e78ea4d3c94: mptcp: use ipv6_addr_equal in addresses_equal
- Results: e80a31769cf4..ef015705ac0b (export)

Tests are now in progress:

- export:
https://github.com/multipath-tcp/mptcp_net-next/commit/b36f3b98b02191d9346b78a16e8da775eb3d780f/checks

Cheers,
Matt

Signature · 1 line
Sponsored by the NGI0 Core fund.