mptcp: support userspace path management
Subsystems (3)
v2 cover mptcp: support userspace path management
This patch series brings together all the required changes to flexibly manage paths/subflows over MPTCP connections from path manager implementations running in userspace. Path management decisions may be made on either end of MPTCP connections based on state captured through MPTCP netlink events.
The series starts with a set of general fixes and enhancements in the related kernel code. It is followed with base functionality and new netlink APIs for handling userspace path management. Further, it extends the MPTCP self-test framework with the new netlink APIs, along with the ability to capture MPTCP netlink events to aid in functional/behavioral validations. Lastly, it adds a new self-test script with a suite of test cases covering all the userspace path management capabilities.
v1 -> v2:
fixed formatting, check for 3rd ACK retransmission only on passive side of the MPJ handshake, fixed build error when IPv6 is not enabled
Florian Westphal (2):
mptcp: netlink: split mptcp_pm_parse_addr into two functions
mptcp: netlink: allow userspace-driven subflow establishment
Kishen Maloor (19):
mptcp: do not restrict subflows with non-kernel PMs
mptcp: store remote id from MP_JOIN SYN/ACK in local ctx
mptcp: reflect remote port (not 0) in ANNOUNCED events
mptcp: establish subflows from either end of connection
mptcp: netlink: store per namespace list of refcounted listen socks
mptcp: netlink: store lsk ref in mptcp_pm_addr_entry
mptcp: netlink: process IPv6 addrs in creating listening sockets
mptcp: attempt to add listening sockets for announced addrs
mptcp: allow ADD_ADDR reissuance by userspace PMs
mptcp: handle local addrs announced by userspace PMs
mptcp: read attributes of addr entries managed by userspace PMs
mptcp: netlink: Add MPTCP_PM_CMD_ANNOUNCE
mptcp: selftests: support MPTCP_PM_CMD_ANNOUNCE
mptcp: netlink: Add MPTCP_PM_CMD_REMOVE
mptcp: selftests: support MPTCP_PM_CMD_REMOVE
mptcp: selftests: support MPTCP_PM_CMD_SUBFLOW_CREATE
mptcp: selftests: support MPTCP_PM_CMD_SUBFLOW_DESTROY
mptcp: selftests: capture netlink events
selftests: mptcp: functional tests for the userspace PM type
include/uapi/linux/mptcp.h | 7 + net/mptcp/options.c | 6 +- net/mptcp/pm.c | 12 +- net/mptcp/pm_netlink.c | 844 ++++++++++++++++-- net/mptcp/protocol.c | 7 +- net/mptcp/protocol.h | 12 +- net/mptcp/subflow.c | 6 +- tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 587 +++++++++++- .../selftests/net/mptcp/userspace_pm.sh | 561 ++++++++++++ 9 files changed, 1938 insertions(+), 104 deletions(-) create mode 100755 tools/testing/selftests/net/mptcp/userspace_pm.sh
base-commit: f81a8b95bfe9cae8ff02739e3e263d9310422af7
Signature · 1 line
2.31.1
PATCH v2 1/21 mptcp: do not restrict subflows with non-kernel PMs
All numerical restrictions on # of addresses/subflows currently apply only to in-kernel PM managed connections. Thus this change removes limitations on adding new subflows by non-kernel (e.g. userspace) PMs.
--- net/mptcp/pm.c | 7 +++++++ net/mptcp/subflow.c | 3 ++- 2 files changed, 9 insertions(+), 1 deletion(-)
net/mptcp/pm.c +7 -0
@@ -87,6 +87,13 @@ bool mptcp_pm_allow_new_subflow(struct mptcp_sock *msk)unsigned int subflows_max;int ret =0 ; +if (READ_ONCE (pm->pm_type) !=MPTCP_PM_TYPE_KERNEL ) { +spin_lock_bh (&msk->pm.lock); + ++pm->subflows; +spin_unlock_bh (&msk->pm.lock); +return true; + } + subflows_max =mptcp_pm_get_subflows_max (msk); pr_debug("msk=%p subflows=%d max=%d allow=%d", msk, pm->subflows,
net/mptcp/subflow.c +2 -1
@@ -62,7 +62,8 @@ static void subflow_generate_hmac(u64 key1, u64 key2, u32 nonce1, u32 nonce2,static bool mptcp_can_accept_new_subflow (const struct mptcp_sock *msk) {return mptcp_is_fully_established ((void *)msk) && -READ_ONCE (msk->pm.accept_subflow); + ((READ_ONCE (msk->pm.pm_type) !=MPTCP_PM_TYPE_KERNEL ) || +READ_ONCE (msk->pm.accept_subflow)); }/* validate received token and create truncated hmac and nonce for SYN-ACK */
Signature · 1 line
2.31.1
PATCH v2 2/21 mptcp: store remote id from MP_JOIN SYN/ACK in local ctx
In reply to Kishen MaloorThis change reads the addr id assigned to the remote endpoint of a subflow from the MP_JOIN SYN/ACK message and stores it in the related subflow context. The remote id was not being captured prior to this change, and will now provide a consistent view of remote endpoints and their ids as seen through netlink events.
--- net/mptcp/subflow.c | 1 + 1 file changed, 1 insertion(+)
net/mptcp/subflow.c +1 -0
@@ -444,6 +444,7 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb) subflow->backup = mp_opt.backup; subflow->thmac = mp_opt.thmac; subflow->remote_nonce = mp_opt.nonce; + subflow->remote_id = mp_opt.join_id;pr_debug ("subflow=%p, thmac=%llu, remote_nonce=%u backup=%d" , subflow, subflow->thmac, subflow->remote_nonce, subflow->backup);
Signature · 1 line
2.31.1
PATCH v2 3/21 mptcp: reflect remote port (not 0) in ANNOUNCED events
In reply to Kishen MaloorPer RFC 8684, if no port is specified in an ADD_ADDR message, MPTCP SHOULD attempt to connect to the specified address on the same port as the port that is already in use by the subflow on which the ADD_ADDR signal was sent.
To facilitate that, this change reflects the specific remote port in use by that subflow in MPTCP_EVENT_ANNOUNCED events.
--- net/mptcp/options.c | 2 +- net/mptcp/pm.c | 5 +++-- net/mptcp/pm_netlink.c | 8 ++++++-- net/mptcp/protocol.h | 6 ++++-- 4 files changed, 14 insertions(+), 7 deletions(-)
net/mptcp/options.c +1 -1
@@ -1123,7 +1123,7 @@ bool mptcp_incoming_options(struct sock *sk, struct sk_buff *skb)if ((mp_opt.suboptions &OPTION_MPTCP_ADD_ADDR ) &&add_addr_hmac_valid (msk, &mp_opt)) {if (!mp_opt.echo) { -mptcp_pm_add_addr_received (msk, &mp_opt.addr); +mptcp_pm_add_addr_received (msk, &mp_opt.addr, sk);MPTCP_INC_STATS (sock_net (sk),MPTCP_MIB_ADDADDR ); }else {mptcp_pm_add_addr_echoed (msk, &mp_opt.addr);
net/mptcp/pm.c +3 -2
@@ -204,14 +204,15 @@ void mptcp_pm_subflow_check_next(struct mptcp_sock *msk, const struct sock *ssk, }void mptcp_pm_add_addr_received (struct mptcp_sock *msk, -const struct mptcp_addr_info *addr) +const struct mptcp_addr_info *addr, +const struct sock *ssk) {struct mptcp_pm_data *pm = &msk->pm;pr_debug ("msk=%p remote_id=%d accept=%d" , msk, addr->id,READ_ONCE (pm->accept_addr)); -mptcp_event_addr_announced (msk, addr); +mptcp_event_addr_announced (msk, addr, ssk);spin_lock_bh (&pm->lock);
net/mptcp/pm_netlink.c +6 -2
@@ -1933,7 +1933,8 @@ void mptcp_event_addr_removed(const struct mptcp_sock *msk, uint8_t id) }void mptcp_event_addr_announced (const struct mptcp_sock *msk, -const struct mptcp_addr_info *info) +const struct mptcp_addr_info *info, +const struct sock *ssk) {struct net *net =sock_net ((const struct sock *)msk);struct nlmsghdr *nlh;
@@ -1957,7 +1958,10 @@ void mptcp_event_addr_announced(const struct mptcp_sock *msk,if (nla_put_u8 (skb,MPTCP_ATTR_REM_ID , info->id)) goto nla_put_failure; -if (nla_put_be16 (skb,MPTCP_ATTR_DPORT , info->port)) +if (nla_put_be16 (skb,MPTCP_ATTR_DPORT , + info->port ==0 ? + ((struct inet_sock *)inet_sk (ssk))->inet_dport : + info->port)) goto nla_put_failure;switch (info->family) {
net/mptcp/protocol.h +4 -2
@@ -752,7 +752,8 @@ bool mptcp_pm_nl_check_work_pending(struct mptcp_sock *msk);void mptcp_pm_subflow_check_next (struct mptcp_sock *msk,const struct sock *ssk,const struct mptcp_subflow_context *subflow);void mptcp_pm_add_addr_received (struct mptcp_sock *msk, -const struct mptcp_addr_info *addr); +const struct mptcp_addr_info *addr, +const struct sock *ssk);void mptcp_pm_add_addr_echoed (struct mptcp_sock *msk,struct mptcp_addr_info *addr);void mptcp_pm_add_addr_send_ack (struct mptcp_sock *msk);
@@ -780,7 +781,8 @@ int mptcp_pm_remove_subflow(struct mptcp_sock *msk, const struct mptcp_rm_list *void mptcp_event (enum mptcp_event_type type,const struct mptcp_sock *msk,const struct sock *ssk,gfp_t gfp); -void mptcp_event_addr_announced (const struct mptcp_sock *msk,const struct mptcp_addr_info *info); +void mptcp_event_addr_announced (const struct mptcp_sock *msk,const struct mptcp_addr_info *info, +const struct sock *ssk);void mptcp_event_addr_removed (const struct mptcp_sock *msk,u8 id);static inline bool mptcp_pm_should_add_signal (struct mptcp_sock *msk)
Signature · 1 line
2.31.1
PATCH v2 4/21 mptcp: establish subflows from either end of connection
In reply to Kishen MaloorThis change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
The essence of this change lies in not relying on the "server_side" flag (which continues to be available if needed).
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
--- net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)
net/mptcp/options.c +1 -1
@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }
net/mptcp/protocol.c +1 -4
@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it
net/mptcp/protocol.h +0 -2
@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }
Signature · 1 line
2.31.1
Re: [PATCH mptcp-next v2 04/21] mptcp: establish subflows from either end of connection
Quoted 9 lines · Kishen Maloor
This change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
The essence of this change lies in not relying on the "
server_side" flag (which continues to be available if needed).
After this patch, server_side is unused except for some debug output. If it's really no longer referenced (see below first), may be better to just remove it.
Quoted 49 lines
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)net/mptcp/options.c +1
-1@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }net/mptcp/protocol.c +1
-4@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it net/mptcp/protocol.h +0
-0
net/mptcp/protocol.h static inline bool mptcp_check_infinite_map(struct sk_buff *skb) · line 914
Quoted 9 lines
net/mptcp/protocol.h +0
-2@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }
Are changes in this function needed? This code was added to address
https://github.com/multipath-tcp/mptcp_net-next/issues/35
which was a weird case encountered with syzkaller where a socket tried to connect to itself, triggering the "simultaneous open" code path. It doesn't seem to be applicable to MP_JOINs.
--
Mat Martineau
Intel
Paolo Abeni
maintainerRe: [PATCH mptcp-next v2 04/21] mptcp: establish subflows from either end of connection
Quoted 70 lines · Mat Martineau
Quoted 9 lines · Kishen Maloor
This change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
The essence of this change lies in not relying on the "
server_side" flag (which continues to be available if needed).After this patch,
server_sideis unused except for some debug output. If it's really no longer referenced (see below first), may be better to just remove it.Quoted 54 lines
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)net/mptcp/options.c +1
-1@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }net/mptcp/protocol.c +1
-4@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it net/mptcp/protocol.h +0
-1@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;
Quoted 10 lines
net/mptcp/protocol.h static inline bool mptcp_check_infinite_map(struct sk_buff *skb) · line 914
Quoted 4 lines
net/mptcp/protocol.h +0
-1return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }Are changes in this function needed? This code was added to address
https://github.com/multipath-tcp/mptcp_net-next/issues/35
which was a weird case encountered with syzkaller where a socket tried to connect to itself, triggering the "simultaneous open" code path. It doesn't seem to be applicable toMP_JOINs.
I *think* we can reach here even with MPJ, e.g. if we have port-based endpoint matching the src address/port - possibly syzkaller could be able to reach such scenario.
Additionally I fear dropping the '!mptcp_sk(parent)->pm.server_side &&' check could re-introduce the initial issue, at least for the MPC handshake.
/P
Re: [PATCH mptcp-next v2 04/21] mptcp: establish subflows from either end of connection
Quoted 9 lines · Paolo Abeni
Quoted 8 lines · Mat Martineau
Quoted 6 lines · Kishen Maloor
This change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
The essence of this change lies in not relying on the "
server_side" flag (which continues to be available if needed).After this patch,
server_sideis unused except for some debug output. If it's really no longer referenced (see below first), may be better to just remove it.
I think we'll need it.
One possible use is to inform the PM daemon of the type of application (client/server) associated with a connection.
As we're discussing the idea of creating listening sockets only upon request, the knowledge that we are on the client end of a connection is something a PM daemon could use to request creation of listening sockets whenever it advertises an address.
Quoted 66 lines
Quoted 55 lines
Quoted 55 lines
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)net/mptcp/options.c +1
-1@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }net/mptcp/protocol.c +1
-4@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it net/mptcp/protocol.h +0
-1@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;Quoted 10 lines
net/mptcp/protocol.h static inline bool mptcp_check_infinite_map(struct sk_buff *skb) · line 914
Quoted 4 lines
net/mptcp/protocol.h +0
-1return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }Are changes in this function needed? This code was added to address
https://github.com/multipath-tcp/mptcp_net-next/issues/35
which was a weird case encountered with syzkaller where a socket tried to connect to itself, triggering the "simultaneous open" code path. It doesn't seem to be applicable toMP_JOINs.
Quoted 7 lines
I *think* we can reach here even with MPJ, e.g. if we have port-based endpoint matching the src address/port - possibly syzkaller could be able to reach such scenario.
Additionally I fear dropping the '!mptcp_sk(parent)->pm.server_side &&' check could re-introduce the initial issue, at least for the MPC handshake.
The purpose of this commit is to permit the MPJ sequence to happen at either end of the connection. The change in mptcp_finish_join() had the most direct impact for enabling this, so I looked at other server_side checks mostly in terms of establishing that bi-directional symmetry.
If the above condition holds only with MPCs, i.e. for the initial connection, then I think it would be fine to preserve the server_side check. But if it also holds for MPJs, then we might have to account for both cases. Perhaps replace:
!mptcp_sk(parent)->pm.server_side)
with something like this?
((subflow->request_mptcp && !mptcp_sk(parent)->pm.server_side) ||
subflow->request_join)
Quoted 3 lines
/P
Paolo Abeni
maintainerRe: [PATCH mptcp-next v2 04/21] mptcp: establish subflows from either end of connection
Quoted 110 lines · Kishen Maloor
Quoted 9 lines · Paolo Abeni
Quoted 8 lines · Mat Martineau
Quoted 6 lines · Kishen Maloor
This change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
The essence of this change lies in not relying on the "
server_side" flag (which continues to be available if needed).After this patch,
server_sideis unused except for some debug output. If it's really no longer referenced (see below first), may be better to just remove it.I think we'll need it.
One possible use is to inform the PM daemon of the type of application (client/server) associated with a connection.
As we're discussing the idea of creating listening sockets only upon request, the knowledge that we are on the client end of a connection is something a PM daemon could use to request creation of listening sockets whenever it advertises an address.
Quoted 66 lines
Quoted 55 lines
Quoted 55 lines
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)net/mptcp/options.c +1
-1@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }net/mptcp/protocol.c +1
-4@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it net/mptcp/protocol.h +0
-1@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;Quoted 10 lines
net/mptcp/protocol.h static inline bool mptcp_check_infinite_map(struct sk_buff *skb) · line 914
Quoted 4 lines
net/mptcp/protocol.h +0
-1return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }Are changes in this function needed? This code was added to address
https://github.com/multipath-tcp/mptcp_net-next/issues/35
which was a weird case encountered with syzkaller where a socket tried to connect to itself, triggering the "simultaneous open" code path. It doesn't seem to be applicable toMP_JOINs.Quoted 7 lines
I *think* we can reach here even with MPJ, e.g. if we have port-based endpoint matching the src address/port - possibly syzkaller could be able to reach such scenario.
Additionally I fear dropping the '!mptcp_sk(parent)->pm.server_side &&' check could re-introduce the initial issue, at least for the MPC handshake.
The purpose of this commit is to permit the MPJ sequence to happen at either end of the connection. The change in
mptcp_finish_join()had the most direct impact for enabling this, so I looked at otherserver_sidechecks mostly in terms of establishing that bi-directional symmetry.If the above condition holds only with MPCs, i.e. for the initial connection, then I think it would be fine to preserve the
server_sidecheck. But if it also holds for MPJs, then we might have to account for both cases. Perhaps replace:
Quoted 6 lines
!mptcp_sk(parent)->pm.server_side)
with something like this?
((
subflow->request_mptcp&& !mptcp_sk(parent)->pm.server_side) ||subflow->request_join)
It looks like 'request_mptcp' is 0 on passive sockets, so:
subflow->request_mptcp || subflow->request_join
should suffice.
Possibly factoring out an helper for the above condition would make the code more readable.
/P
Re: [PATCH mptcp-next v2 04/21] mptcp: establish subflows from either end of connection
Quoted 115 lines · Paolo Abeni
Quoted 114 lines · Kishen Maloor
Quoted 9 lines · Paolo Abeni
Quoted 8 lines · Mat Martineau
Quoted 6 lines · Kishen Maloor
This change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
The essence of this change lies in not relying on the "
server_side" flag (which continues to be available if needed).After this patch,
server_sideis unused except for some debug output. If it's really no longer referenced (see below first), may be better to just remove it.I think we'll need it.
One possible use is to inform the PM daemon of the type of application (client/server) associated with a connection.
As we're discussing the idea of creating listening sockets only upon request, the knowledge that we are on the client end of a connection is something a PM daemon could use to request creation of listening sockets whenever it advertises an address.
Quoted 66 lines
Quoted 55 lines
Quoted 55 lines
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)net/mptcp/options.c +1
-1@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }net/mptcp/protocol.c +1
-4@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it net/mptcp/protocol.h +0
-1@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;Quoted 10 lines
net/mptcp/protocol.h static inline bool mptcp_check_infinite_map(struct sk_buff *skb) · line 914
Quoted 4 lines
net/mptcp/protocol.h +0
-1return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }Are changes in this function needed? This code was added to address
https://github.com/multipath-tcp/mptcp_net-next/issues/35
which was a weird case encountered with syzkaller where a socket tried to connect to itself, triggering the "simultaneous open" code path. It doesn't seem to be applicable toMP_JOINs.Quoted 7 lines
I *think* we can reach here even with MPJ, e.g. if we have port-based endpoint matching the src address/port - possibly syzkaller could be able to reach such scenario.
Additionally I fear dropping the '!mptcp_sk(parent)->pm.server_side &&' check could re-introduce the initial issue, at least for the MPC handshake.
The purpose of this commit is to permit the MPJ sequence to happen at either end of the connection. The change in
mptcp_finish_join()had the most direct impact for enabling this, so I looked at otherserver_sidechecks mostly in terms of establishing that bi-directional symmetry.If the above condition holds only with MPCs, i.e. for the initial connection, then I think it would be fine to preserve the
server_sidecheck. But if it also holds for MPJs, then we might have to account for both cases. Perhaps replace:!mptcp_sk(parent)->pm.server_side)
with something like this?
Quoted 8 lines
Quoted 2 lines
((
subflow->request_mptcp&& !mptcp_sk(parent)->pm.server_side) ||subflow->request_join)It looks like '
request_mptcp' is 0 on passive sockets, so:
subflow->request_mptcp||subflow->request_joinshould suffice.
That's right. Makes sense.
Quoted 3 lines
Possibly factoring out an helper for the above condition would make the code more readable.
Thanks! Will do so.
Quoted 3 lines
/P
Re: [PATCH mptcp-next v2 04/21] mptcp: establish subflows from either end of connection
Quoted 22 lines · Kishen Maloor
Quoted 21 lines · Paolo Abeni
Quoted 20 lines · Kishen Maloor
Quoted 9 lines · Paolo Abeni
Quoted 8 lines · Mat Martineau
Quoted 6 lines · Kishen Maloor
This change updates internal logic to permit subflows to be established from either the client or server ends of MPTCP connections. This symmetry and added flexibility may be harnessed by PM implementations running on either end in creating new subflows.
Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
The essence of this change lies in not relying on the "
server_side" flag (which continues to be available if needed).After this patch,
server_sideis unused except for some debug output. If it's really no longer referenced (see below first), may be better to just remove it.I think we'll need it.
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
One possible use is to inform the PM daemon of the type of application (client/server) associated with a connection.
As we're discussing the idea of creating listening sockets only upon request, the knowledge that we are on the client end of a connection is something a PM daemon could use to request creation of listening sockets whenever it advertises an address.
Ok, makes sense.
Quoted 90 lines
Quoted 90 lines
Quoted 89 lines
Quoted 66 lines
Quoted 55 lines
Quoted 55 lines
v2: check for 3rd ACK retransmission only on passive side of the MPJ handshake
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/options.c | 2 +- net/mptcp/protocol.c | 5 +---- net/mptcp/protocol.h | 2 -- 3 files changed, 2 insertions(+), 7 deletions(-)net/mptcp/options.c +1
-1@@ -921,7 +921,7 @@ static bool check_fully_established(struct mptcp_sock *msk, struct sock *ssk,if (TCP_SKB_CB (skb)->seq == subflow->ssn_offset +1 &&TCP_SKB_CB (skb)->end_seq ==TCP_SKB_CB (skb)->seq && subflow->mp_join && (mp_opt->suboptions &OPTIONS_MPTCP_MPJ ) && -READ_ONCE (msk->pm.server_side)) + !subflow->request_join)tcp_send_ack (ssk); goto fully_established; }net/mptcp/protocol.c +1
-4@@ -3248,15 +3248,12 @@ bool mptcp_finish_join(struct sock *ssk)return false; } -if (!msk->pm.server_side) +if (!list_empty (&subflow->node)) goto out;if (!mptcp_pm_allow_new_subflow (msk)) goto err_prohibited; -if (WARN_ON_ONCE (!list_empty (&subflow->node))) - goto err_prohibited; -/* active connections are already on conn_list. * If we can't acquire msk socket lock here, let the release callback * handle it net/mptcp/protocol.h +0
-1@@ -908,10 +908,8 @@ static inline bool mptcp_check_infinite_map(struct sk_buff *skb)static inline bool subflow_simultaneous_connect (struct sock *sk) {struct mptcp_subflow_context *subflow =mptcp_subflow_ctx (sk); -struct sock *parent = subflow->conn;Quoted 10 lines
net/mptcp/protocol.h static inline bool mptcp_check_infinite_map(struct sk_buff *skb) · line 914
Quoted 4 lines
net/mptcp/protocol.h +0
-1return sk->sk_state ==TCP_ESTABLISHED && - !mptcp_sk (parent)->pm.server_side && !subflow->conn_finished; }Are changes in this function needed? This code was added to address
https://github.com/multipath-tcp/mptcp_net-next/issues/35
which was a weird case encountered with syzkaller where a socket tried to connect to itself, triggering the "simultaneous open" code path. It doesn't seem to be applicable toMP_JOINs.Quoted 7 lines
I *think* we can reach here even with MPJ, e.g. if we have port-based endpoint matching the src address/port - possibly syzkaller could be able to reach such scenario.
Additionally I fear dropping the '!mptcp_sk(parent)->pm.server_side &&' check could re-introduce the initial issue, at least for the MPC handshake.
The purpose of this commit is to permit the MPJ sequence to happen at either end of the connection. The change in
mptcp_finish_join()had the most direct impact for enabling this, so I looked at otherserver_sidechecks mostly in terms of establishing that bi-directional symmetry.If the above condition holds only with MPCs, i.e. for the initial connection, then I think it would be fine to preserve the
server_sidecheck. But if it also holds for MPJs, then we might have to account for both cases. Perhaps replace:!mptcp_sk(parent)->pm.server_side)
with something like this?
((
subflow->request_mptcp&& !mptcp_sk(parent)->pm.server_side) ||subflow->request_join)
Quoted 8 lines
Quoted 5 lines
It looks like '
request_mptcp' is 0 on passive sockets, so:
subflow->request_mptcp||subflow->request_joinshould suffice.
That's right. Makes sense.
Yeah, good to simplify this even if 'pm.server_side' is still around.
Quoted 6 lines
Quoted 3 lines
Possibly factoring out an helper for the above condition would make the code more readable.
Thanks! Will do so.
--
Mat Martineau
Intel
PATCH v2 5/21 mptcp: netlink: store per namespace list of refcounted listen socks
In reply to Kishen MaloorThe kernel maintains listening sockets bound to announced addresses via the ADD_ADDR option to be able to receive MP_JOIN requests. Path managers may choose to advertise the same address over multiple MPTCP connections. So this change provides a simple framework to manage a list of all distinct listning sockets created in a namespace by encapsulating it in a structure that is ref counted and can be shared across multiple connections. The sockets (and their enclosing structure) are released when there are no more references.
v2: fixed formatting
--- net/mptcp/pm_netlink.c | 76 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 76 insertions(+)
net/mptcp/pm_netlink.c +76 -0
@@ -22,6 +22,14 @@ static struct genl_family mptcp_genl_family;static int pm_nl_pernet_id; +struct mptcp_local_lsk { +struct list_head list; +struct mptcp_addr_info addr; +struct socket *lsk; +struct rcu_head rcu; +refcount_t refcount; +}; +struct mptcp_pm_addr_entry {struct list_head list;struct mptcp_addr_info addr;
@@ -41,7 +49,10 @@ struct mptcp_pm_add_entry {struct pm_nl_pernet {/* protects pernet updates */ spinlock_t lock; +/* protects access to pernet lsk list */ +spinlock_t lsk_list_lock;struct list_head local_addr_list; +struct list_head lsk_list;unsigned int addrs;unsigned int stale_loss_cnt;unsigned int add_addr_signal_max;
@@ -83,6 +94,69 @@ static bool addresses_equal(const struct mptcp_addr_info *a,return a->port == b->port; } +static struct mptcp_local_lsk *lsk_list_find (struct pm_nl_pernet *pernet, +struct mptcp_addr_info *addr) +{ +struct mptcp_local_lsk *lsk_ref =NULL ; +struct mptcp_local_lsk *i; + +rcu_read_lock (); + +list_for_each_entry_rcu (i, &pernet->lsk_list, list) { +if (addresses_equal (&i->addr, addr, true)) { +if (refcount_inc_not_zero (&i->refcount)) { + lsk_ref = i; +break ; + } + } + } + +rcu_read_unlock (); + +return lsk_ref; +} + +static void lsk_list_add_ref (struct mptcp_local_lsk *lsk_ref) +{ +refcount_inc (&lsk_ref->refcount); +} + +static struct mptcp_local_lsk *lsk_list_add (struct pm_nl_pernet *pernet, +struct mptcp_addr_info *addr, +struct socket *lsk) +{ +struct mptcp_local_lsk *lsk_ref; + + lsk_ref =kmalloc (sizeof (*lsk_ref),GFP_ATOMIC ); + +if (!lsk_ref) +return NULL ; + + lsk_ref->lsk = lsk; +memcpy (&lsk_ref->addr, addr,sizeof (struct mptcp_addr_info )); +refcount_set (&lsk_ref->refcount,1 ); + +spin_lock_bh (&pernet->lsk_list_lock); +list_add_rcu (&lsk_ref->list, &pernet->lsk_list); +spin_unlock_bh (&pernet->lsk_list_lock); + +return lsk_ref; +} + +static void lsk_list_release (struct pm_nl_pernet *pernet, +struct mptcp_local_lsk *lsk_ref) +{ +if (lsk_ref &&refcount_dec_and_test (&lsk_ref->refcount)) { +sock_release (lsk_ref->lsk); + +spin_lock_bh (&pernet->lsk_list_lock); +list_del_rcu (&lsk_ref->list); +spin_unlock_bh (&pernet->lsk_list_lock); + +kfree_rcu (lsk_ref, rcu); + } +} +static bool address_zero (const struct mptcp_addr_info *addr) {struct mptcp_addr_info zero;
@@ -2098,12 +2172,14 @@ static int __net_init pm_nl_init_net(struct net *net)struct pm_nl_pernet *pernet =net_generic (net, pm_nl_pernet_id);INIT_LIST_HEAD_RCU (&pernet->local_addr_list); +INIT_LIST_HEAD_RCU (&pernet->lsk_list);/* Cit. 2 subflows ought to be enough for anybody. */ pernet->subflows_max =2 ; pernet->next_id =1 ; pernet->stale_loss_cnt =4 ;spin_lock_init (&pernet->lock); +spin_lock_init (&pernet->lsk_list_lock);/* No need to initialize other pernet fields, the struct is zeroed at * allocation time.
Signature · 1 line
2.31.1
PATCH v2 6/21 mptcp: netlink: store lsk ref in mptcp_pm_addr_entry
In reply to Kishen MaloorThis change updates struct mptcp_pm_addr_entry to store a listening socket (lsk) reference, i.e. a pointer to a reference counted structure containing the lsk (struct socket *) instead of the lsk itself. Code blocks that directly operated on the lsk in struct mptcp_pm_addr_entry have been updated to work with the lsk ref instead, utilizing the new helper functions that operate on lsk refs.
v2: fixed formatting
--- net/mptcp/pm_netlink.c | 62 ++++++++++++++++++++++++++++-------------- 1 file changed, 41 insertions(+), 21 deletions(-)
net/mptcp/pm_netlink.c +41 -21
@@ -35,7 +35,7 @@ struct mptcp_pm_addr_entry {struct mptcp_addr_info addr;u8 flags;int ifindex; -struct socket *lsk; +struct mptcp_local_lsk *lsk_ref; };struct mptcp_pm_add_entry {
@@ -983,7 +983,8 @@ static int mptcp_pm_nl_append_new_local_addr(struct pm_nl_pernet *pernet, }static int mptcp_pm_nl_create_listen_socket (struct sock *sk, -struct mptcp_pm_addr_entry *entry) +struct mptcp_pm_addr_entry *entry, +struct socket **lsk) {struct sockaddr_storage addr;struct mptcp_sock *msk;
@@ -992,11 +993,11 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk,int err; err =sock_create_kern (sock_net (sk), entry->addr.family, -SOCK_STREAM ,IPPROTO_MPTCP , &entry->lsk); +SOCK_STREAM ,IPPROTO_MPTCP , lsk);if (err)return err; - msk =mptcp_sk (entry->lsk->sk); + msk =mptcp_sk ((*lsk)->sk);if (!msk) { err = -EINVAL ; goto out;
@@ -1025,7 +1026,8 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk,return 0 ; out: -sock_release (entry->lsk); +sock_release (*lsk); + *lsk =NULL ;return err; }
@@ -1074,7 +1076,7 @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct sock_common *skc) entry->addr.port =0 ; entry->ifindex =0 ; entry->flags =0 ; - entry->lsk =NULL ; + entry->lsk_ref =NULL ; ret =mptcp_pm_nl_append_new_local_addr (pernet, entry);if (ret <0 )kfree (entry);
@@ -1270,6 +1272,7 @@ static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info)struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR ];struct pm_nl_pernet *pernet =genl_info_pm_nl (info);struct mptcp_pm_addr_entry addr, *entry; +struct socket *lsk;int ret; ret =mptcp_pm_parse_addr (attr, info, true, &addr);
@@ -1284,18 +1287,34 @@ static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info) *entry = addr;if (entry->addr.port) { - ret =mptcp_pm_nl_create_listen_socket (skb->sk, entry); -if (ret) { -GENL_SET_ERR_MSG (info,"create listen socket error" ); -kfree (entry); -return ret; + entry->lsk_ref =lsk_list_find (pernet, &entry->addr); + +if (!entry->lsk_ref) { + ret =mptcp_pm_nl_create_listen_socket (skb->sk, entry, &lsk); + +if (ret) { +GENL_SET_ERR_MSG (info,"create listen socket error" ); +kfree (entry); +return ret; + } + + entry->lsk_ref =lsk_list_add (pernet, &entry->addr, lsk); + +if (!entry->lsk_ref) { +GENL_SET_ERR_MSG (info,"can't allocate lsk ref" ); +sock_release (lsk); +kfree (entry); +return -ENOMEM ; + } } } + ret =mptcp_pm_nl_append_new_local_addr (pernet, entry); +if (ret <0 ) {GENL_SET_ERR_MSG (info,"too many addresses or duplicate one" ); -if (entry->lsk) -sock_release (entry->lsk); +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref);kfree (entry);return ret; }
@@ -1398,10 +1417,11 @@ static int mptcp_nl_remove_subflow_and_signal_addr(struct net *net, }/* caller must ensure the RCU grace period is already elapsed */ -static void __mptcp_pm_release_addr_entry (struct mptcp_pm_addr_entry *entry) +static void __mptcp_pm_release_addr_entry (struct pm_nl_pernet *pernet, +struct mptcp_pm_addr_entry *entry) { -if (entry->lsk) -sock_release (entry->lsk); +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref);kfree (entry); }
@@ -1483,7 +1503,7 @@ static int mptcp_nl_cmd_del_addr(struct sk_buff *skb, struct genl_info *info)mptcp_nl_remove_subflow_and_signal_addr (sock_net (skb->sk), &entry->addr);synchronize_rcu (); -__mptcp_pm_release_addr_entry (entry); +__mptcp_pm_release_addr_entry (pernet, entry);return ret; }
@@ -1539,7 +1559,7 @@ static void mptcp_nl_remove_addrs_list(struct net *net, }/* caller must ensure the RCU grace period is already elapsed */ -static void __flush_addrs (struct list_head *list) +static void __flush_addrs (struct pm_nl_pernet *pernet,struct list_head *list) {while (!list_empty (list)) {struct mptcp_pm_addr_entry *cur;
@@ -1547,7 +1567,7 @@ static void __flush_addrs(struct list_head *list) cur =list_entry (list->next, struct mptcp_pm_addr_entry, list);list_del_rcu (&cur->list); -__mptcp_pm_release_addr_entry (cur); +__mptcp_pm_release_addr_entry (pernet, cur); } }
@@ -1572,7 +1592,7 @@ static int mptcp_nl_cmd_flush_addrs(struct sk_buff *skb, struct genl_info *info)spin_unlock_bh (&pernet->lock);mptcp_nl_remove_addrs_list (sock_net (skb->sk), &free_list);synchronize_rcu (); -__flush_addrs (&free_list); +__flush_addrs (pernet, &free_list);return 0 ; }
@@ -2199,7 +2219,7 @@ static void __net_exit pm_nl_exit_net(struct list_head *net_list)* other modifiers, also netns core already waited for a * RCU grace period. */ -__flush_addrs (&pernet->local_addr_list); +__flush_addrs (pernet, &pernet->local_addr_list); } }
Signature · 1 line
2.31.1
PATCH v2 7/21 mptcp: netlink: process IPv6 addrs in creating listening sockets
In reply to Kishen MaloorThis change updates mptcp_pm_nl_create_listen_socket() to create listening sockets bound to IPv6 addresses (where IPv6 is supported).
--- net/mptcp/pm_netlink.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
net/mptcp/pm_netlink.c +6 -2
@@ -986,6 +986,7 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk,struct mptcp_pm_addr_entry *entry, struct socket **lsk) { + int addrlen = sizeof(struct sockaddr_in);struct sockaddr_storage addr;struct mptcp_sock *msk;struct socket *ssock;
@@ -1010,8 +1011,11 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, }mptcp_info2sockaddr (&entry->addr, &addr, entry->addr.family); - err =kernel_bind (ssock, (struct sockaddr *)&addr, -sizeof (struct sockaddr_in )); +#if IS_ENABLED (CONFIG_MPTCP_IPV6 ) +if (entry->addr.family ==AF_INET6 ) + addrlen =sizeof (struct sockaddr_in6 ); +#endif + err =kernel_bind (ssock, (struct sockaddr *)&addr, addrlen);if (err) {pr_warn ("kernel_bind error, err=%d" , err); goto out;
Signature · 1 line
2.31.1
Paolo Abeni
maintainerRe: [PATCH mptcp-next v2 07/21] mptcp: netlink: process IPv6 addrs in creating listening sockets
Hello,
Quoted 24 lines · Kishen Maloor
This change updates
mptcp_pm_nl_create_listen_socket()to create listening sockets bound to IPv6 addresses (where IPv6 is supported).Signed-off-by Kishen Maloor <[email protected]>--- net/mptcp/pm_netlink.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)net/mptcp/pm_netlink.c +1
-0@@ -986,6 +986,7 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk,struct mptcp_pm_addr_entry *entry, struct socket **lsk) { + int addrlen = sizeof(struct sockaddr_in);struct sockaddr_storage addr;struct mptcp_sock *msk;struct socket *ssock;@@ -1010,8 +1011,11 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, }mptcp_info2sockaddr (&entry->addr, &addr, entry->addr.family);
net/mptcp/pm_netlink.c static int mptcp_pm_nl_create_listen_socket(struct sock *sk, · line 1021
Quoted 10 lines
net/mptcp/pm_netlink.c +5
-2- err =kernel_bind (ssock, (struct sockaddr *)&addr, -sizeof (struct sockaddr_in )); +#if IS_ENABLED (CONFIG_MPTCP_IPV6 ) +if (entry->addr.family ==AF_INET6 ) + addrlen =sizeof (struct sockaddr_in6 ); +#endif + err =kernel_bind (ssock, (struct sockaddr *)&addr, addrlen);if (err) {pr_warn ("kernel_bind error, err=%d" , err); goto out;
This looks a bugfix for -net to me?
Possibly worthy additional an additional mp_join self-test for the ipv6 case.
Thanks!
/P
Re: [PATCH mptcp-next v2 07/21] mptcp: netlink: process IPv6 addrs in creating listening sockets
Quoted 6 lines · Paolo Abeni
Hello,
Quoted 3 lines · Kishen Maloor
This change updates
mptcp_pm_nl_create_listen_socket()to create listening sockets bound to IPv6 addresses (where IPv6 is supported).
This tag -
would help with backporting to stable.
Quoted 34 lines
Quoted 21 lines
Signed-off-by Kishen Maloor <[email protected]>--- net/mptcp/pm_netlink.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)net/mptcp/pm_netlink.c +1
-0@@ -986,6 +986,7 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk,struct mptcp_pm_addr_entry *entry, struct socket **lsk) { + int addrlen = sizeof(struct sockaddr_in);struct sockaddr_storage addr;struct mptcp_sock *msk;struct socket *ssock;@@ -1010,8 +1011,11 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, }mptcp_info2sockaddr (&entry->addr, &addr, entry->addr.family);net/mptcp/pm_netlink.c static int mptcp_pm_nl_create_listen_socket(struct sock *sk, · line 1021
Quoted 10 lines
net/mptcp/pm_netlink.c +5
-2- err =kernel_bind (ssock, (struct sockaddr *)&addr, -sizeof (struct sockaddr_in )); +#if IS_ENABLED (CONFIG_MPTCP_IPV6 ) +if (entry->addr.family ==AF_INET6 ) + addrlen =sizeof (struct sockaddr_in6 ); +#endif + err =kernel_bind (ssock, (struct sockaddr *)&addr, addrlen);if (err) {pr_warn ("kernel_bind error, err=%d" , err); goto out;This looks a bugfix for -net to me?
Quoted 3 lines
Possibly worthy additional an additional
mp_joinself-test for the ipv6 case.
I agree, this would be good for -net. Kishen can you add the suggested selftest too and repost separately for mptcp-net?
--
Mat Martineau
Intel
Re: [PATCH mptcp-next v2 07/21] mptcp: netlink: process IPv6 addrs in creating listening sockets
Quoted 9 lines · Mat Martineau
Quoted 6 lines · Paolo Abeni
Hello,
Quoted 3 lines · Kishen Maloor
This change updates
mptcp_pm_nl_create_listen_socket()to create listening sockets bound to IPv6 addresses (where IPv6 is supported).
Quoted 39 lines
This tag -
Fixes 1729cf186d8a ("mptcp: create the listening socket for new port")would help with backporting to stable.
Quoted 33 lines
Quoted 21 lines
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/pm_netlink.c | 8 ++++++--1 file changed, 6 insertions(+), 2 deletions(-)net/mptcp/pm_netlink.c +1
-0@@ -986,6 +986,7 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, struct mptcp_pm_addr_entry *entry, struct socket **lsk) { + int addrlen = sizeof(struct sockaddr_in); struct sockaddr_storage addr; struct mptcp_sock *msk; struct socket *ssock;@@ -1010,8 +1011,11 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, } mptcp_info2sockaddr(&entry->addr, &addr, entry->addr.family);net/mptcp/pm_netlink.c static int mptcp_pm_nl_create_listen_socket(struct sock *sk, · line 1021
Quoted 10 lines
net/mptcp/pm_netlink.c +5
-2- err = kernel_bind(ssock, (struct sockaddr *)&addr, - sizeof(struct sockaddr_in)); +#if IS_ENABLED (CONFIG_MPTCP_IPV6 ) + if (entry->addr.family == AF_INET6) + addrlen = sizeof(struct sockaddr_in6); +#endif + err = kernel_bind(ssock, (struct sockaddr *)&addr, addrlen); if (err) { pr_warn("kernel_bind error, err=%d", err); goto out;This looks a bugfix for -net to me?
Sure, we can record a bug and the fact that this commit fixes it.
Quoted 7 lines
Quoted 4 lines
Possibly worthy additional an additional
mp_joinself-test for the ipv6 case.I agree, this would be good for -net. Kishen can you add the suggested selftest too and repost separately for mptcp-net?
Actually, this path is currently exercised by self-tests in userspace_pm.sh through address advertisements from the namespace containing the MPTCP client, for which a listening socket is created to subsequently receive MPJs.
Quoted 3 lines
Signature · 2 lines
Mat Martineau Intel
Re: [PATCH mptcp-next v2 07/21] mptcp: netlink: process IPv6 addrs in creating listening sockets
Quoted 57 lines · Kishen Maloor
Quoted 9 lines · Mat Martineau
Quoted 6 lines · Paolo Abeni
Hello,
Quoted 3 lines · Kishen Maloor
This change updates
mptcp_pm_nl_create_listen_socket()to create listening sockets bound to IPv6 addresses (where IPv6 is supported).Quoted 39 lines
This tag -
Fixes 1729cf186d8a ("mptcp: create the listening socket for new port")would help with backporting to stable.
Quoted 33 lines
Quoted 21 lines
Signed-off-by Kishen Maloor <[email protected]>---
net/mptcp/pm_netlink.c | 8 ++++++--1 file changed, 6 insertions(+), 2 deletions(-)net/mptcp/pm_netlink.c +1
-0@@ -986,6 +986,7 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, struct mptcp_pm_addr_entry *entry, struct socket **lsk) { + int addrlen = sizeof(struct sockaddr_in); struct sockaddr_storage addr; struct mptcp_sock *msk; struct socket *ssock;@@ -1010,8 +1011,11 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, } mptcp_info2sockaddr(&entry->addr, &addr, entry->addr.family);net/mptcp/pm_netlink.c static int mptcp_pm_nl_create_listen_socket(struct sock *sk, · line 1021
Quoted 10 lines
net/mptcp/pm_netlink.c +5
-2- err = kernel_bind(ssock, (struct sockaddr *)&addr, - sizeof(struct sockaddr_in)); +#if IS_ENABLED (CONFIG_MPTCP_IPV6 ) + if (entry->addr.family == AF_INET6) + addrlen = sizeof(struct sockaddr_in6); +#endif + err = kernel_bind(ssock, (struct sockaddr *)&addr, addrlen); if (err) { pr_warn("kernel_bind error, err=%d", err); goto out;This looks a bugfix for -net to me?
Sure, we can record a bug and the fact that this commit fixes it.
Quoted 5 lines
Quoted 4 lines
Possibly worthy additional an additional
mp_joinself-test for the ipv6 case.
Quoted 6 lines
Quoted 2 lines
I agree, this would be good for -net. Kishen can you add the suggested selftest too and repost separately for mptcp-net?
Actually, this path is currently exercised by self-tests in
userspace_pm.shthrough address advertisements from the namespace containing the MPTCP client, for which a listening socket is created to subsequently receive MPJs.
Ok, it's good that it's covered going forward. Re-reading Paolo's comment, maybe that was the main purpose of the requested test (rather than adding to self test coverage in -stable). Should be enough to have the kernel fix itself for the -net patch.
--
Mat Martineau
Intel
PATCH v2 8/21 mptcp: attempt to add listening sockets for announced addrs
In reply to Kishen MaloorWhen ADD_ADDR announcements use the port associated with an active subflow, this change ensures that a listening socket is bound to the announced address and port for subsequently receiving MP_JOINs from the remote end. In case there's a recorded lsk bound to that address+port, it is reused. But if a listening socket for this address is already held by the application then no further action is taken.
When a listening socket is created, it is stored in struct mptcp_pm_add_entry and released accordingly.
v2: fixed formatting
--- net/mptcp/pm_netlink.c | 56 ++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 54 insertions(+), 2 deletions(-)
net/mptcp/pm_netlink.c +54 -2
@@ -43,6 +43,7 @@ struct mptcp_pm_add_entry {struct mptcp_addr_info addr;struct timer_list add_timer;struct mptcp_sock *sock; +struct mptcp_local_lsk *lsk_ref;u8 retrans_times; };
@@ -66,6 +67,10 @@ struct pm_nl_pernet {#define MPTCP_PM_ADDR_MAX 8#define ADD_ADDR_RETRANS_MAX 3 +static int mptcp_pm_nl_create_listen_socket (struct sock *sk, +struct mptcp_pm_addr_entry *entry, +struct socket **lsk); +static bool addresses_equal (const struct mptcp_addr_info *a,const struct mptcp_addr_info *b,bool use_port) {
@@ -438,7 +443,8 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk, }static bool mptcp_pm_alloc_anno_list (struct mptcp_sock *msk, -struct mptcp_pm_addr_entry *entry) +struct mptcp_pm_addr_entry *entry, +struct mptcp_local_lsk *lsk_ref) {struct mptcp_pm_add_entry *add_entry =NULL ;struct sock *sk = (struct sock *)msk;
@@ -458,6 +464,10 @@ static bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, add_entry->addr = entry->addr; add_entry->sock = msk; add_entry->retrans_times =0 ; + add_entry->lsk_ref = lsk_ref; + +if (lsk_ref) +lsk_list_add_ref (lsk_ref);timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); sk_reset_timer(sk, &add_entry->add_timer,
@@ -470,8 +480,11 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk) {struct mptcp_pm_add_entry *entry, *tmp;struct sock *sk = (struct sock *)msk; +struct pm_nl_pernet *pernet;LIST_HEAD (free_list); + pernet =net_generic (sock_net (sk), pm_nl_pernet_id); +pr_debug ("msk=%p" , msk);spin_lock_bh (&msk->pm.lock);
@@ -480,6 +493,8 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk)list_for_each_entry_safe (entry, tmp, &free_list, list) {sk_stop_timer_sync (sk, &entry->add_timer); +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref);kfree (entry); } }
@@ -570,13 +585,16 @@ lookup_id_by_addr(struct pm_nl_pernet *pernet, const struct mptcp_addr_info *add }static void mptcp_pm_create_subflow_or_signal_addr (struct mptcp_sock *msk) + __must_hold(&msk->pm.lock) { +struct mptcp_local_lsk *lsk_ref =NULL ;struct sock *sk = (struct sock *)msk;struct mptcp_pm_addr_entry *local;unsigned int add_addr_signal_max;unsigned int local_addr_max;struct pm_nl_pernet *pernet;unsigned int subflows_max; +struct socket *lsk; pernet =net_generic (sock_net (sk), pm_nl_pernet_id);
@@ -607,12 +625,39 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) local =select_signal_address (pernet, msk);if (local) { -if (mptcp_pm_alloc_anno_list (msk, local)) { +if (!local->addr.port) { + local->addr.port = + ((struct inet_sock *)inet_sk + ((struct sock *)msk))->inet_sport; + + lsk_ref =lsk_list_find (pernet, &local->addr); + +if (!lsk_ref) { +spin_unlock_bh (&msk->pm.lock); + +mptcp_pm_nl_create_listen_socket (sk, local, &lsk); + +spin_lock_bh (&msk->pm.lock); + +if (lsk) + lsk_ref =lsk_list_add (pernet, &local->addr, lsk); + +if (lsk && !lsk_ref) +sock_release (lsk); + } + + local->addr.port =0 ; + } + +if (mptcp_pm_alloc_anno_list (msk, local, lsk_ref)) {__clear_bit (local->addr.id, msk->pm.id_avail_bitmap); msk->pm.add_addr_signaled++;mptcp_pm_announce_addr (msk, &local->addr, false);mptcp_pm_nl_addr_send_ack (msk); } + +if (lsk_ref) +lsk_list_release (pernet, lsk_ref); } }
@@ -704,6 +749,7 @@ static unsigned int fill_local_addresses_vec(struct mptcp_sock *msk, }static void mptcp_pm_nl_add_addr_received (struct mptcp_sock *msk) + __must_hold(&msk->pm.lock) {struct mptcp_addr_info addrs[MPTCP_PM_ADDR_MAX ];struct sock *sk = (struct sock *)msk;
@@ -1352,11 +1398,17 @@ int mptcp_pm_get_flags_and_ifindex_by_id(struct net *net, unsigned int id,static bool remove_anno_list_by_saddr (struct mptcp_sock *msk,struct mptcp_addr_info *addr) { +struct sock *sk = (struct sock *)msk;struct mptcp_pm_add_entry *entry; +struct pm_nl_pernet *pernet; + + pernet =net_generic (sock_net (sk), pm_nl_pernet_id); entry =mptcp_pm_del_add_timer (msk, addr, false);if (entry) {list_del (&entry->list); +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref);kfree (entry);return true; }
Signature · 1 line
2.31.1
Paolo Abeni
maintainerRe: [PATCH mptcp-next v2 08/21] mptcp: attempt to add listening sockets for announced addrs
Hello,
Quoted 8 lines · Kishen Maloor
When
ADD_ADDRannouncements use the port associated with an active subflow, this change ensures that a listening socket is bound to the announced address and port for subsequently receivingMP_JOINsfrom the remote end. In case there's a recorded lsk bound to that address+port, it is reused. But if a listening socket for this address is already held by the application then no further action is taken.
Quoted 8 lines
When a listening socket is created, it is stored in struct
mptcp_pm_add_entryand released accordingly.v2: fixed formatting
Signed-off-by Kishen Maloor <[email protected]>
should be either:
"""
<changelog>
<tags>
"""
or:
"""
<tags>
---
<changelog>
"""
we usually keep the changelog outside the commit message for development history before landing on netdev, that is:
"""
---
v2: fixed formatting
"""
Quoted 100 lines
--- net/mptcp/pm_netlink.c | 56 ++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 54 insertions(+), 2 deletions(-)net/mptcp/pm_netlink.c +28
-2@@ -43,6 +43,7 @@ struct mptcp_pm_add_entry {struct mptcp_addr_info addr;struct timer_list add_timer;struct mptcp_sock *sock; +struct mptcp_local_lsk *lsk_ref;u8 retrans_times; };@@ -66,6 +67,10 @@ struct pm_nl_pernet {#define MPTCP_PM_ADDR_MAX 8#define ADD_ADDR_RETRANS_MAX 3 +static int mptcp_pm_nl_create_listen_socket (struct sock *sk, +struct mptcp_pm_addr_entry *entry, +struct socket **lsk); +static bool addresses_equal (const struct mptcp_addr_info *a,const struct mptcp_addr_info *b,bool use_port) {@@ -438,7 +443,8 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk, }static bool mptcp_pm_alloc_anno_list (struct mptcp_sock *msk, -struct mptcp_pm_addr_entry *entry) +struct mptcp_pm_addr_entry *entry, +struct mptcp_local_lsk *lsk_ref) {struct mptcp_pm_add_entry *add_entry =NULL ;struct sock *sk = (struct sock *)msk;@@ -458,6 +464,10 @@ static bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, add_entry->addr = entry->addr; add_entry->sock = msk; add_entry->retrans_times =0 ; + add_entry->lsk_ref = lsk_ref; + +if (lsk_ref) +lsk_list_add_ref (lsk_ref);timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); sk_reset_timer(sk, &add_entry->add_timer,@@ -470,8 +480,11 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk) {struct mptcp_pm_add_entry *entry, *tmp;struct sock *sk = (struct sock *)msk; +struct pm_nl_pernet *pernet;LIST_HEAD (free_list); + pernet =net_generic (sock_net (sk), pm_nl_pernet_id); +pr_debug ("msk=%p" , msk);spin_lock_bh (&msk->pm.lock);@@ -480,6 +493,8 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk)list_for_each_entry_safe (entry, tmp, &free_list, list) {sk_stop_timer_sync (sk, &entry->add_timer); +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref);kfree (entry); } }@@ -570,13 +585,16 @@ lookup_id_by_addr(struct pm_nl_pernet *pernet, const struct mptcp_addr_info *add }static void mptcp_pm_create_subflow_or_signal_addr (struct mptcp_sock *msk) + __must_hold(&msk->pm.lock) { +struct mptcp_local_lsk *lsk_ref =NULL ;struct sock *sk = (struct sock *)msk;struct mptcp_pm_addr_entry *local;unsigned int add_addr_signal_max;unsigned int local_addr_max;struct pm_nl_pernet *pernet;unsigned int subflows_max; +struct socket *lsk; pernet =net_generic (sock_net (sk), pm_nl_pernet_id);@@ -607,12 +625,39 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) local =select_signal_address (pernet, msk);if (local) { -if (mptcp_pm_alloc_anno_list (msk, local)) { +if (!local->addr.port) { + local->addr.port = + ((struct inet_sock *)inet_sk + ((struct sock *)msk))->inet_sport; + + lsk_ref =lsk_list_find (pernet, &local->addr); + +if (!lsk_ref) { +spin_unlock_bh (&msk->pm.lock);
net/mptcp/pm_netlink.c static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) · line 646
Quoted 10 lines
net/mptcp/pm_netlink.c +10
-0+ +mptcp_pm_nl_create_listen_socket (sk, local, &lsk); + +spin_lock_bh (&msk->pm.lock); + +if (lsk) + lsk_ref =lsk_list_add (pernet, &local->addr, lsk); + +if (lsk && !lsk_ref) +sock_release (lsk);
Let's suppose an user-space application listen on 2 different address (A, B) and does:
"""
s1 = socket()bind(s1, A)
listen(s1)
// at this point incoming MPTCP connection can be established on s1 // and ADD_ADDR sub-options could be sent back
s2 = socket()bind(s2, B)
listen(s2)
"""
If there is a signal endpoint on B, the above listen can race with the mptcp_pm_nl_create_listen_socket() call, leading to hard to track startup issues for user-space application.
I really think we want at list a configuration option, off by default, for this feature. Some specific self-test would be a plus.
It will help reviewing, splitting this series in at least 2 chunks:
* pre-reqs up to ~this patch
* user-space PM specific stuff
Side note: it would be nice reducing the level of intentation, e.g. factoring-out part of the inner code in some helper.
Quoted 4 lines
net/mptcp/pm_netlink.c +4
-0+ } + + local->addr.port =0 ; + }
net/mptcp/pm_netlink.c static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) · line 660
Quoted 10 lines
net/mptcp/pm_netlink.c +5
-0+ +if (mptcp_pm_alloc_anno_list (msk, local, lsk_ref)) {__clear_bit (local->addr.id, msk->pm.id_avail_bitmap); msk->pm.add_addr_signaled++;mptcp_pm_announce_addr (msk, &local->addr, false);mptcp_pm_nl_addr_send_ack (msk); } + +if (lsk_ref) +lsk_list_release (pernet, lsk_ref);
Probaly not very relevant, but something alike:
rcu_read_lock() lsk_ref = __lsk_list_find(); if (lst_ref) if (mptcp_pm_alloc_anno_list(...) rcu_read_unlock()
would save a pair of possibly contended atomic operations in the common case.
Thanks!
Paolo
Re: [PATCH mptcp-next v2 08/21] mptcp: attempt to add listening sockets for announced addrs
Hi Kishen, Paolo,
Quoted 6 lines · Paolo Abeni
I really think we want at list a configuration option, off by default, for this feature. Some specific self-test would be a plus.
It will help reviewing, splitting this series in at least 2 chunks:
* pre-reqs up to ~this patch
* user-space PM specific stuff
Good idea, I think it will help indeed. And it will be required when we will send these patches to netdev.
For me, it makes sense to have more features only for the userspace PM and of course more flexibility. But we still need to make sure the in-kernel PM can be controlled to avoid some behaviours. For example, when sending the command to emit an ADD_ADDR, a parameter could be used to (try to) create a listening socket as well.
I think an in-kernel PM makes sense for a typical server mainly having to send ADD_ADDRs and accept subflows. Having a userspace PM daemon can be costly for such "basic task".
Also a server could also send an ADD_ADDR for an IP that it doesn't directly "own" and we might save some resources by not creating or trying to create listening sockets. We could also have a new flag with 'ip mptcp' to create listening sockets for the IP we want to "signal".
I mean: even if it is now easier to create listening sockets, we might not want to do it all the time.
Cheers,
Matt
Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions www.tessares.net
Re: [PATCH mptcp-next v2 08/21] mptcp: attempt to add listening sockets for announced addrs
In reply to Paolo AbeniOn Fri, 14 Jan 2022, Paolo Abeni wrote:
...
Quoted 4 lines
It will help reviewing, splitting this series in at least 2 chunks:
* pre-reqs up to ~this patch
* user-space PM specific stuff
Seems like the way to go here is to split patches 1-6 into another series, and have patch 7 as a standalone (or separate series w/ selftest) for mptcp-net.
--
Mat Martineau
Intel
Re: [PATCH mptcp-next v2 08/21] mptcp: attempt to add listening sockets for announced addrs
In reply to Paolo AbeniHi Paolo, Matthieu,
Quoted 30 lines · Paolo Abeni
Hello,
Quoted 8 lines · Kishen Maloor
When
ADD_ADDRannouncements use the port associated with an active subflow, this change ensures that a listening socket is bound to the announced address and port for subsequently receivingMP_JOINsfrom the remote end. In case there's a recorded lsk bound to that address+port, it is reused. But if a listening socket for this address is already held by the application then no further action is taken.Quoted 8 lines
When a listening socket is created, it is stored in struct
mptcp_pm_add_entryand released accordingly.v2: fixed formatting
Signed-off-by Kishen Maloor <[email protected]>should be either:
"""
<changelog><tags>
"""or:
Quoted 8 lines
"""
<tags>
---
<changelog>
"""we usually keep the changelog outside the commit message for development history before landing on netdev, that is:
Thanks! I shall reflect this change in the related commit messages.
Quoted 134 lines
"""
Signed-off-by Kishen Maloor <[email protected]>---
v2: fixed formatting
"""Quoted 100 lines
--- net/mptcp/pm_netlink.c | 56 ++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 54 insertions(+), 2 deletions(-)net/mptcp/pm_netlink.c +28
-2@@ -43,6 +43,7 @@ struct mptcp_pm_add_entry {struct mptcp_addr_info addr;struct timer_list add_timer;struct mptcp_sock *sock; +struct mptcp_local_lsk *lsk_ref;u8 retrans_times; };@@ -66,6 +67,10 @@ struct pm_nl_pernet {#define MPTCP_PM_ADDR_MAX 8#define ADD_ADDR_RETRANS_MAX 3 +static int mptcp_pm_nl_create_listen_socket (struct sock *sk, +struct mptcp_pm_addr_entry *entry, +struct socket **lsk); +static bool addresses_equal (const struct mptcp_addr_info *a,const struct mptcp_addr_info *b,bool use_port) {@@ -438,7 +443,8 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk, }static bool mptcp_pm_alloc_anno_list (struct mptcp_sock *msk, -struct mptcp_pm_addr_entry *entry) +struct mptcp_pm_addr_entry *entry, +struct mptcp_local_lsk *lsk_ref) {struct mptcp_pm_add_entry *add_entry =NULL ;struct sock *sk = (struct sock *)msk;@@ -458,6 +464,10 @@ static bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, add_entry->addr = entry->addr; add_entry->sock = msk; add_entry->retrans_times =0 ; + add_entry->lsk_ref = lsk_ref; + +if (lsk_ref) +lsk_list_add_ref (lsk_ref);timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); sk_reset_timer(sk, &add_entry->add_timer,@@ -470,8 +480,11 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk) {struct mptcp_pm_add_entry *entry, *tmp;struct sock *sk = (struct sock *)msk; +struct pm_nl_pernet *pernet;LIST_HEAD (free_list); + pernet =net_generic (sock_net (sk), pm_nl_pernet_id); +pr_debug ("msk=%p" , msk);spin_lock_bh (&msk->pm.lock);@@ -480,6 +493,8 @@ void mptcp_pm_free_anno_list(struct mptcp_sock *msk)list_for_each_entry_safe (entry, tmp, &free_list, list) {sk_stop_timer_sync (sk, &entry->add_timer); +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref);kfree (entry); } }@@ -570,13 +585,16 @@ lookup_id_by_addr(struct pm_nl_pernet *pernet, const struct mptcp_addr_info *add }static void mptcp_pm_create_subflow_or_signal_addr (struct mptcp_sock *msk) + __must_hold(&msk->pm.lock) { +struct mptcp_local_lsk *lsk_ref =NULL ;struct sock *sk = (struct sock *)msk;struct mptcp_pm_addr_entry *local;unsigned int add_addr_signal_max;unsigned int local_addr_max;struct pm_nl_pernet *pernet;unsigned int subflows_max; +struct socket *lsk; pernet =net_generic (sock_net (sk), pm_nl_pernet_id);@@ -607,12 +625,39 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) local =select_signal_address (pernet, msk);if (local) { -if (mptcp_pm_alloc_anno_list (msk, local)) { +if (!local->addr.port) { + local->addr.port = + ((struct inet_sock *)inet_sk + ((struct sock *)msk))->inet_sport; + + lsk_ref =lsk_list_find (pernet, &local->addr); + +if (!lsk_ref) { +spin_unlock_bh (&msk->pm.lock);net/mptcp/pm_netlink.c static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) · line 646
Quoted 10 lines
net/mptcp/pm_netlink.c +10
-0+ +mptcp_pm_nl_create_listen_socket (sk, local, &lsk); + +spin_lock_bh (&msk->pm.lock); + +if (lsk) + lsk_ref =lsk_list_add (pernet, &local->addr, lsk); + +if (lsk && !lsk_ref) +sock_release (lsk);Let's suppose an user-space application listen on 2 different address (A, B) and does:
"""
s1 =socket()bind(s1, A)
listen(s1)
// at this point incoming MPTCP connection can be established on s1 // andADD_ADDRsub-options could be sent backs2 =
socket()bind(s2, B)
listen(s2)
"""
Quoted 6 lines
If there is a signal endpoint on B, the above listen can race with the
mptcp_pm_nl_create_listen_socket()call, leading to hard to track startup issues for user-space application.I really think we want at list a configuration option, off by default, for this feature. Some specific self-test would be a plus.
Looking at your example above, assuming both A and B are bound to the same port then yes, a race such as you suggest could occur.
But if you consider bug #203, it arose only because there was no listener in the application (and unexpectedly so). So if the path manager creates a listener (at the time of the address advertisement) to facilitate MPJs then that would have the usual side effects of creating listeners (in general).
For e.g,. I think this clash could also occur with the existing code in the kernel PM and using a port-based endpoint when the app happens to bind a socket to that specific addr+port.
The other scenario where the path manager needs to always establish a listener is when running alongside an MPTCP client.
We could certainly add an "attempt to create lsk" option to the ADD_ADDR netlink commands, as I believe you've both suggested, but perhaps we should think further about the guidance regarding usage of this option.
For e.g., if creating lsks is not the default behavior, then bug #203 would persist unless the entity that issues the ADD_ADDR command exercises this option.
Quoted 38 lines
It will help reviewing, splitting this series in at least 2 chunks:
* pre-reqs up to ~this patch
* user-space PM specific stuffSide note: it would be nice reducing the level of intentation, e.g. factoring-out part of the inner code in some helper.
Quoted 4 lines
net/mptcp/pm_netlink.c +4
-0+ } + + local->addr.port =0 ; + }net/mptcp/pm_netlink.c static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk) · line 660
Quoted 10 lines
net/mptcp/pm_netlink.c +5
-0+ +if (mptcp_pm_alloc_anno_list (msk, local, lsk_ref)) {__clear_bit (local->addr.id, msk->pm.id_avail_bitmap); msk->pm.add_addr_signaled++;mptcp_pm_announce_addr (msk, &local->addr, false);mptcp_pm_nl_addr_send_ack (msk); } + +if (lsk_ref) +lsk_list_release (pernet, lsk_ref);Probaly not very relevant, but something alike:
rcu_read_lock() lsk_ref = __lsk_list_find(); if (lst_ref) if (mptcp_pm_alloc_anno_list(...) rcu_read_unlock()would save a pair of possibly contended atomic operations in the common case.
Thanks!
Paolo
PATCH v2 9/21 mptcp: allow ADD_ADDR reissuance by userspace PMs
In reply to Kishen MaloorThis change allows userspace PM implementations to reissue ADD_ADDR announcements (if necessary) based on their chosen policy.
--- net/mptcp/pm_netlink.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-)
net/mptcp/pm_netlink.c +10 -2
@@ -452,8 +452,16 @@ static bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,lockdep_assert_held (&msk->pm.lock); -if (mptcp_lookup_anno_list_by_saddr (msk, &entry->addr)) -return false; + add_entry =mptcp_lookup_anno_list_by_saddr (msk, &entry->addr); + +if (add_entry) { +if (READ_ONCE (msk->pm.pm_type) ==MPTCP_PM_TYPE_KERNEL ) +return false; + +sk_reset_timer (sk, &add_entry->add_timer, + jiffies +mptcp_get_add_addr_timeout (net)); +return true; + } add_entry =kmalloc (sizeof (*add_entry),GFP_ATOMIC );if (!add_entry)
Signature · 1 line
2.31.1
PATCH v2 10/21 mptcp: handle local addrs announced by userspace PMs
In reply to Kishen MaloorThis change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1) ADD_ADDR announcements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance of ADD_ADDRs when there's a successful match of addr/id.
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the REMOVE_ADDR flow is freed while the sock is destructed.
--- net/mptcp/pm_netlink.c | 79 ++++++++++++++++++++++++++++++++++++++++++ net/mptcp/protocol.c | 2 ++ net/mptcp/protocol.h | 2 ++ 3 files changed, 83 insertions(+)
net/mptcp/pm_netlink.c +79 -0
@@ -484,6 +484,31 @@ static bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,return true; } +void mptcp_free_local_addr_list (struct mptcp_sock *msk) +{ +struct mptcp_pm_addr_entry *entry, *tmp; +struct sock *sk = (struct sock *)msk; +struct pm_nl_pernet *pernet; +LIST_HEAD (free_list); + +if (READ_ONCE (msk->pm.pm_type) ==MPTCP_PM_TYPE_KERNEL ) +return ; + + pernet =net_generic (sock_net (sk), pm_nl_pernet_id); + +pr_debug ("msk=%p" , msk); + +mptcp_data_lock (sk); +list_splice_init (&msk->local_addr_list, &free_list); +mptcp_data_unlock (sk); + +list_for_each_entry_safe (entry, tmp, &free_list, list) { +if (entry->lsk_ref) +lsk_list_release (pernet, entry->lsk_ref); +kfree (entry); + } +} +void mptcp_pm_free_anno_list (struct mptcp_sock *msk) {struct mptcp_pm_add_entry *entry, *tmp;
@@ -972,6 +997,60 @@ static bool address_use_port(struct mptcp_pm_addr_entry *entry)MPTCP_PM_ADDR_FLAG_SIGNAL ; } +static int mptcp_userspace_pm_append_new_local_addr (struct mptcp_sock *msk, +struct mptcp_pm_addr_entry *entry) +{ +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; +bool id_match = false; +int ret = -EINVAL ; + +bitmap_zero (id_bitmap,MPTCP_PM_MAX_ADDR_ID +1 ); + +mptcp_data_lock (sk); +list_for_each_entry (e, &msk->local_addr_list, list) { + addr_match =addresses_equal (&e->addr, &entry->addr, true); +if (addr_match && entry->addr.id ==0 ) + entry->addr.id = e->addr.id; + id_match = (e->addr.id == entry->addr.id); +if (addr_match && id_match) { + match = e; +break ; + }else if (addr_match || id_match) { +break ; + } +__set_bit (e->addr.id, id_bitmap); + } + +if (!match && !addr_match && !id_match) { + e =kmalloc (sizeof (*e),GFP_ATOMIC ); +if (!e) { +mptcp_data_unlock (sk); +return -ENOMEM ; + } + + *e = *entry; +if (!e->addr.id) + e->addr.id =find_next_zero_bit (id_bitmap, +MPTCP_PM_MAX_ADDR_ID +1 , +1 ); +list_add_tail_rcu (&e->list, &msk->local_addr_list); + ret = e->addr.id; + +if (e->lsk_ref && e->addr.port) +lsk_list_add_ref (e->lsk_ref); + }else if (match) { + ret = entry->addr.id; + } + +mptcp_data_unlock (sk); + +return ret; +} +static int mptcp_pm_nl_append_new_local_addr (struct pm_nl_pernet *pernet,struct mptcp_pm_addr_entry *entry) {
net/mptcp/protocol.c +2 -0
@@ -2531,6 +2531,7 @@ static int __mptcp_init_sock(struct sock *sk)INIT_LIST_HEAD (&msk->conn_list);INIT_LIST_HEAD (&msk->join_list);INIT_LIST_HEAD (&msk->rtx_queue); +INIT_LIST_HEAD (&msk->local_addr_list);INIT_WORK (&msk->work, mptcp_worker);__skb_queue_head_init (&msk->receive_queue); msk->out_of_order_queue =RB_ROOT ;
@@ -3027,6 +3028,7 @@ void mptcp_destroy_common(struct mptcp_sock *msk) msk->rmem_fwd_alloc =0 ;mptcp_token_destroy (msk);mptcp_pm_free_anno_list (msk); +mptcp_free_local_addr_list (msk); }static void mptcp_destroy (struct sock *sk)
net/mptcp/protocol.h +2 -0
@@ -281,6 +281,7 @@ struct mptcp_sock {struct sk_buff_head receive_queue;struct list_head conn_list;struct list_head rtx_queue; +struct list_head local_addr_list;struct mptcp_data_frag *first_pending;struct list_head join_list;struct socket *subflow;/* outgoing connect/listener/!mp_capable */
@@ -733,6 +734,7 @@ struct mptcp_sock *mptcp_token_get_sock(struct net *net, u32 token);struct mptcp_sock *mptcp_token_iter_next (const struct net *net,long *s_slot,long *s_num);void mptcp_token_destroy (struct mptcp_sock *msk); +void mptcp_free_local_addr_list (struct mptcp_sock *msk);void mptcp_crypto_key_sha (u64 key,u32 *token,u64 *idsn);
Signature · 1 line
2.31.1
Paolo Abeni
maintainerRe: [PATCH mptcp-next v2 10/21] mptcp: handle local addrs announced by userspace PMs
Quoted 9 lines · Kishen Maloor
This change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1)ADD_ADDRannouncements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance ofADD_ADDRswhen there's a successful match of addr/id.
Quoted 3 lines
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the
REMOVE_ADDRflow is freed while the sock is destructed.
It feels strange to me that we need to maintain an additional addresses list inside the kernel for the user-space PM - which should take care of all the status information.
Why anno_list is not enough? Why isn't cheapest to extend it?
Being the list unlimited a malicius (or buggy) user-space could consume all the kernel memory. I think we need some limits, or at least some accounting.
/P
Re: [PATCH mptcp-next v2 10/21] mptcp: handle local addrs announced by userspace PMs
Quoted 10 lines · Paolo Abeni
Quoted 9 lines · Kishen Maloor
This change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1)ADD_ADDRannouncements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance ofADD_ADDRswhen there's a successful match of addr/id.
Quoted 7 lines
Quoted 3 lines
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the
REMOVE_ADDRflow is freed while the sock is destructed.It feels strange to me that we need to maintain an additional addresses list inside the kernel for the user-space PM - which should take care of all the status information.
The PM daemon will have the complete picture, but the protocol needs to know the local ids assigned to addresses so as such the kernel has to store addresses (with their ids) regardless of PM.
Quoted 2 lines
Why
anno_listis not enough? Why isn't cheapest to extend it?
The anno_list is the list of addresses that were announced via the ADD_ADDR command, to be used specifically for doing re-transmissions.
However, the implementation can also accept connections at local addresses that were not explicitly announced (hence not in anno_list), and in this case the kernel records the address and assigns it a local id unique in the scope of its connection.
So basically msk->local_addr_list is the list of all known local addresses, both announced and not so their ids can be later retrieved.
To give you more context, in my last iteration of the code before I posted the series, I was storing local addrs in pernet->local_addr_list just as its done for the kernel PM, but later moved it to
a per-msk list to eliminate contention (in accessing that list) with other userspace or kernel PM managed connections.
Quoted 4 lines
Being the list unlimited a malicius (or buggy) user-space could consume all the kernel memory. I think we need some limits, or at least some accounting.
At this point we're taking the PM daemon as a trusted entity which can issue these netlink commands for path management. So there is currently no configurable ceiling in the kernel on the size of the PM's kernel stored context.
Quoted 3 lines
/P
Re: [PATCH mptcp-next v2 10/21] mptcp: handle local addrs announced by userspace PMs
Quoted 42 lines · Kishen Maloor
Quoted 10 lines · Paolo Abeni
Quoted 9 lines · Kishen Maloor
This change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1)ADD_ADDRannouncements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance ofADD_ADDRswhen there's a successful match of addr/id.Quoted 7 lines
Quoted 3 lines
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the
REMOVE_ADDRflow is freed while the sock is destructed.It feels strange to me that we need to maintain an additional addresses list inside the kernel for the user-space PM - which should take care of all the status information.
The PM daemon will have the complete picture, but the protocol needs to know the local ids assigned to addresses so as such the kernel has to store addresses (with their ids) regardless of PM.
Quoted 2 lines
Why
anno_listis not enough? Why isn't cheapest to extend it?The
anno_listis the list of addresses that were announced via theADD_ADDRcommand, to be used specifically for doing re-transmissions.However, the implementation can also accept connections at local addresses that were not explicitly announced (hence not in
anno_list), and in this case the kernel records the address and assigns it a local id unique in the scope of its connection.So basically
msk->local_addr_listis the list of all known local addresses, both announced and not so their ids can be later retrieved.To give you more context, in my last iteration of the code before I posted the series, I was storing local addrs in
pernet->local_addr_listjust as its done for the kernel PM, but later moved it to
a per-msk list to eliminate contention (in accessing that list) with other userspace or kernel PM managed connections.Quoted 1 line
Quoted 8 lines
Quoted 3 lines
Being the list unlimited a malicius (or buggy) user-space could consume all the kernel memory. I think we need some limits, or at least some accounting.
At this point we're taking the PM daemon as a trusted entity which can issue these netlink commands for path management. So there is currently no configurable ceiling in the kernel on the size of the PM's kernel stored context.
I think it's still worthwhile to have some limits/accounting as Paolo suggests - part of the point of pushing PM code to userspace is so bugs or other vulnerabilities don't take down the whole machine.
--
Mat Martineau
Intel
Re: [PATCH mptcp-next v2 10/21] mptcp: handle local addrs announced by userspace PMs
Quoted 48 lines · Mat Martineau
Quoted 46 lines · Kishen Maloor
Quoted 10 lines · Paolo Abeni
Quoted 9 lines · Kishen Maloor
This change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1)ADD_ADDRannouncements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance ofADD_ADDRswhen there's a successful match of addr/id.Quoted 7 lines
Quoted 3 lines
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the
REMOVE_ADDRflow is freed while the sock is destructed.It feels strange to me that we need to maintain an additional addresses list inside the kernel for the user-space PM - which should take care of all the status information.
The PM daemon will have the complete picture, but the protocol needs to know the local ids assigned to addresses so as such the kernel has to store addresses (with their ids) regardless of PM.
Quoted 2 lines
Why
anno_listis not enough? Why isn't cheapest to extend it?The
anno_listis the list of addresses that were announced via theADD_ADDRcommand, to be used specifically for doing re-transmissions.However, the implementation can also accept connections at local addresses that were not explicitly announced (hence not in
anno_list), and in this case the kernel records the address and assigns it a local id unique in the scope of its connection.So basically
msk->local_addr_listis the list of all known local addresses, both announced and not so their ids can be later retrieved.To give you more context, in my last iteration of the code before I posted the series, I was storing local addrs in
pernet->local_addr_listjust as its done for the kernel PM, but later moved it to
a per-msk list to eliminate contention (in accessing that list) with other userspace or kernel PM managed connections.Quoted 4 lines
Being the list unlimited a malicius (or buggy) user-space could consume all the kernel memory. I think we need some limits, or at least some accounting.
Quoted 7 lines
Quoted 4 lines
At this point we're taking the PM daemon as a trusted entity which can issue these netlink commands for path management. So there is currently no configurable ceiling in the kernel on the size of the PM's kernel stored context.
I think it's still worthwhile to have some limits/accounting as Paolo suggests - part of the point of pushing PM code to userspace is so bugs or other vulnerabilities don't take down the whole machine.
Agreed, but do we put circuit breakers inside the kernel or the PM daemon? (for e.g., if a PM plug-in can see limitations imposed by the daemon?)
If we consider userspace path management as providing "more flexibility", then conceivably the kernel would expose hooks to the PM daemon to bump up any kernel enforced limits when necessary. So a malicious or buggy PM could also raise those limits.
Since interactions over the netlink interface are to be carried out by a privileged entity, I've been assuming that the PM daemon is to be trusted.
On the other hand, if we're talking about having fixed upper bounds in the kernel wrt
# of addrs/subflows (which cannot be changed by the PM daemon), then that could make sense.
Quoted 3 lines
Signature · 2 lines
Mat Martineau Intel
Re: [PATCH mptcp-next v2 10/21] mptcp: handle local addrs announced by userspace PMs
Quoted 54 lines · Kishen Maloor
Quoted 53 lines · Mat Martineau
Quoted 50 lines · Kishen Maloor
Quoted 10 lines · Paolo Abeni
Quoted 9 lines · Kishen Maloor
This change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1)ADD_ADDRannouncements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance ofADD_ADDRswhen there's a successful match of addr/id.Quoted 7 lines
Quoted 3 lines
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the
REMOVE_ADDRflow is freed while the sock is destructed.It feels strange to me that we need to maintain an additional addresses list inside the kernel for the user-space PM - which should take care of all the status information.
The PM daemon will have the complete picture, but the protocol needs to know the local ids assigned to addresses so as such the kernel has to store addresses (with their ids) regardless of PM.
Quoted 2 lines
Why
anno_listis not enough? Why isn't cheapest to extend it?The
anno_listis the list of addresses that were announced via theADD_ADDRcommand, to be used specifically for doing re-transmissions.However, the implementation can also accept connections at local addresses that were not explicitly announced (hence not in
anno_list), and in this case the kernel records the address and assigns it a local id unique in the scope of its connection.So basically
msk->local_addr_listis the list of all known local addresses, both announced and not so their ids can be later retrieved.To give you more context, in my last iteration of the code before I posted the series, I was storing local addrs in
pernet->local_addr_listjust as its done for the kernel PM, but later moved it to
a per-msk list to eliminate contention (in accessing that list) with other userspace or kernel PM managed connections.Quoted 4 lines
Being the list unlimited a malicius (or buggy) user-space could consume all the kernel memory. I think we need some limits, or at least some accounting.
At this point we're taking the PM daemon as a trusted entity which can issue these netlink commands for path management. So there is currently no configurable ceiling in the kernel on the size of the PM's kernel stored context.
Quoted 6 lines
Quoted 2 lines
I think it's still worthwhile to have some limits/accounting as Paolo suggests - part of the point of pushing PM code to userspace is so bugs or other vulnerabilities don't take down the whole machine.
Agreed, but do we put circuit breakers inside the kernel or the PM daemon? (for e.g., if a PM plug-in can see limitations imposed by the daemon?)
Kernel, I'd say.
Quoted 5 lines
If we consider userspace path management as providing "more flexibility", then conceivably the kernel would expose hooks to the PM daemon to bump up any kernel enforced limits when necessary. So a malicious or buggy PM could also raise those limits.
True, an admin (or bad actor) can get themselves into trouble by setting nonsensical limits. But the limits remain useful for stability and managing resources.
Quoted 3 lines
Since interactions over the netlink interface are to be carried out by a privileged entity, I've been assuming that the PM daemon is to be trusted.
We're trusting the PM daemon with certain (granular) privileges/capabilities using CAP_NET_ADMIN, not the old "root-controls-all" model.
Quoted 3 lines
On the other hand, if we're talking about having fixed upper bounds in the kernel wrt
# of addrs/subflows (which cannot be changed by the PM daemon), then that could make sense.
That's pretty much what I'm talking about, maybe tunable (within a range) by a sysctl.
--
Mat Martineau
Intel
Re: [PATCH mptcp-next v2 10/21] mptcp: handle local addrs announced by userspace PMs
Quoted 77 lines · Mat Martineau
Quoted 54 lines · Kishen Maloor
Quoted 53 lines · Mat Martineau
Quoted 50 lines · Kishen Maloor
Quoted 10 lines · Paolo Abeni
Quoted 9 lines · Kishen Maloor
This change adds a new internal function to store/retrieve local addrs announced by userspace PM implementations from the kernel context. The function does not stipulate any limitation on the # of addrs, and handles the requirements of three scenarios:
1)ADD_ADDRannouncements (which require that a local id be provided), 2) retrieving the local id associated with an address, also where one may need to be assigned, and 3) reissuance ofADD_ADDRswhen there's a successful match of addr/id.Quoted 7 lines
Quoted 3 lines
The list of all stored local addr entries is held under the MPTCP sock structure. This list, if not released by the
REMOVE_ADDRflow is freed while the sock is destructed.It feels strange to me that we need to maintain an additional addresses list inside the kernel for the user-space PM - which should take care of all the status information.
The PM daemon will have the complete picture, but the protocol needs to know the local ids assigned to addresses so as such the kernel has to store addresses (with their ids) regardless of PM.
Quoted 2 lines
Why
anno_listis not enough? Why isn't cheapest to extend it?The
anno_listis the list of addresses that were announced via theADD_ADDRcommand, to be used specifically for doing re-transmissions.However, the implementation can also accept connections at local addresses that were not explicitly announced (hence not in
anno_list), and in this case the kernel records the address and assigns it a local id unique in the scope of its connection.So basically
msk->local_addr_listis the list of all known local addresses, both announced and not so their ids can be later retrieved.To give you more context, in my last iteration of the code before I posted the series, I was storing local addrs in
pernet->local_addr_listjust as its done for the kernel PM, but later moved it to
a per-msk list to eliminate contention (in accessing that list) with other userspace or kernel PM managed connections.Quoted 4 lines
Being the list unlimited a malicius (or buggy) user-space could consume all the kernel memory. I think we need some limits, or at least some accounting.
At this point we're taking the PM daemon as a trusted entity which can issue these netlink commands for path management. So there is currently no configurable ceiling in the kernel on the size of the PM's kernel stored context.
Quoted 6 lines
Quoted 2 lines
I think it's still worthwhile to have some limits/accounting as Paolo suggests - part of the point of pushing PM code to userspace is so bugs or other vulnerabilities don't take down the whole machine.
Agreed, but do we put circuit breakers inside the kernel or the PM daemon? (for e.g., if a PM plug-in can see limitations imposed by the daemon?)
Kernel, I'd say.
Quoted 5 lines
If we consider userspace path management as providing "more flexibility", then conceivably the kernel would expose hooks to the PM daemon to bump up any kernel enforced limits when necessary. So a malicious or buggy PM could also raise those limits.
True, an admin (or bad actor) can get themselves into trouble by setting nonsensical limits. But the limits remain useful for stability and managing resources.
Quoted 3 lines
Since interactions over the netlink interface are to be carried out by a privileged entity, I've been assuming that the PM daemon is to be trusted.
Quoted 7 lines
We're trusting the PM daemon with certain (granular) privileges/capabilities using
CAP_NET_ADMIN, not the old "root-controls-all" model.Quoted 3 lines
On the other hand, if we're talking about having fixed upper bounds in the kernel wrt
# of addrs/subflows (which cannot be changed by the PM daemon), then that could make sense.That's pretty much what I'm talking about, maybe tunable (within a range) by a sysctl.
Sounds good, I will look into adding a couple new params, configurable via sysctl for regulating activities by userspace PMs. Note that these will be unrelated to the equivalent limits that we have in place for the kernel PM.
Quoted 4 lines
Signature · 3 lines
-- Mat Martineau Intel
PATCH v2 11/21 mptcp: read attributes of addr entries managed by userspace PMs
In reply to Kishen MaloorThis change introduces a parallel path in the kernel for retrieving the local id, flags, if_index for an addr entry in the context of an MPTCP connection that's being managed by a userspace PM. The userspace and in-kernel PM modes deviate in their procedures for obtaining this information.
--- net/mptcp/pm_netlink.c | 101 ++++++++++++++++++++++++++++------------- net/mptcp/protocol.h | 2 +- net/mptcp/subflow.c | 2 +- 3 files changed, 71 insertions(+), 34 deletions(-)
net/mptcp/pm_netlink.c +69 -32
@@ -1175,6 +1175,7 @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct sock_common *skc)struct mptcp_addr_info msk_local;struct pm_nl_pernet *pernet;int ret =-1 ; +int pm_type;if (WARN_ON_ONCE (!msk))return -1 ;
@@ -1192,31 +1193,50 @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct sock_common *skc) pernet =net_generic (sock_net ((struct sock *)msk), pm_nl_pernet_id); -rcu_read_lock (); -list_for_each_entry_rcu (entry, &pernet->local_addr_list, list) { -if (addresses_equal (&entry->addr, &skc_local, entry->addr.port)) { - ret = entry->addr.id; -break ; - } - } -rcu_read_unlock (); -if (ret >=0 ) -return ret; -/* address not found, add to local list */ - entry =kmalloc (sizeof (*entry),GFP_ATOMIC ); -if (!entry) -return -ENOMEM ; - - entry->addr = skc_local; - entry->addr.id =0 ; - entry->addr.port =0 ; - entry->ifindex =0 ; - entry->flags =0 ; - entry->lsk_ref =NULL ; - ret =mptcp_pm_nl_append_new_local_addr (pernet, entry); -if (ret <0 ) -kfree (entry); + + pm_type =READ_ONCE (msk->pm.pm_type); + +if (pm_type ==MPTCP_PM_TYPE_KERNEL ) { +rcu_read_lock (); +list_for_each_entry_rcu (entry, &pernet->local_addr_list, list) { +if (addresses_equal (&entry->addr, &skc_local, entry->addr.port)) { + ret = entry->addr.id; +break ; + } + } +rcu_read_unlock (); + +if (ret >=0 ) +return ret; + + entry =kmalloc (sizeof (*entry),GFP_ATOMIC ); +if (!entry) +return -ENOMEM ; + + entry->addr = skc_local; + entry->addr.id =0 ; + entry->addr.port =0 ; + entry->ifindex =0 ; + entry->flags =0 ; + entry->lsk_ref =NULL ; + ret =mptcp_pm_nl_append_new_local_addr (pernet, entry); +if (ret <0 ) +kfree (entry); + }else if (pm_type ==MPTCP_PM_TYPE_USERSPACE ) { +struct mptcp_pm_addr_entry new_entry; +__be16 msk_sport = ((struct inet_sock *) +inet_sk ((struct sock *)msk))->inet_sport; + +memset (&new_entry,0 ,sizeof (struct mptcp_pm_addr_entry )); + new_entry.addr = skc_local; + new_entry.addr.id =0 ; + +if (new_entry.addr.port == msk_sport) + new_entry.addr.port =0 ; + + ret =mptcp_userspace_pm_append_new_local_addr (msk, &new_entry); + }return ret; }
@@ -1461,22 +1481,39 @@ static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info)return 0 ; } -int mptcp_pm_get_flags_and_ifindex_by_id (struct net *net,unsigned int id, +int mptcp_pm_get_flags_and_ifindex_by_id (struct mptcp_sock *msk,unsigned int id,u8 *flags,int *ifindex) { -struct mptcp_pm_addr_entry *entry; +struct mptcp_pm_addr_entry *entry, *match =NULL ; +struct sock *sk = (struct sock *)msk; +struct net *net =sock_net (sk); *flags =0 ; *ifindex =0 ;if (id) { -rcu_read_lock (); - entry =__lookup_addr_by_id (net_generic (net, pm_nl_pernet_id), id); -if (entry) { - *flags = entry->flags; - *ifindex = entry->ifindex; +if (READ_ONCE (msk->pm.pm_type) ==MPTCP_PM_TYPE_KERNEL ) { +rcu_read_lock (); + entry =__lookup_addr_by_id (net_generic (net, pm_nl_pernet_id), id); +if (entry) { + *flags = entry->flags; + *ifindex = entry->ifindex; + } +rcu_read_unlock (); + }else { +mptcp_data_lock (sk); +list_for_each_entry (entry, &msk->local_addr_list, list) { +if (id == entry->addr.id) { + match = entry; +break ; + } + } +mptcp_data_unlock (sk); +if (match) { + *flags = match->flags; + *ifindex = match->ifindex; + } } -rcu_read_unlock (); }return 0 ;
net/mptcp/protocol.h +1 -1
@@ -772,7 +772,7 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk,struct mptcp_pm_add_entry *mptcp_lookup_anno_list_by_saddr (struct mptcp_sock *msk,struct mptcp_addr_info *addr); -int mptcp_pm_get_flags_and_ifindex_by_id (struct net *net,unsigned int id, +int mptcp_pm_get_flags_and_ifindex_by_id (struct mptcp_sock *msk,unsigned int id,u8 *flags,int *ifindex); int mptcp_pm_announce_addr(struct mptcp_sock *msk,
net/mptcp/subflow.c +1 -1
@@ -1419,7 +1419,7 @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc, local_id = err; } -mptcp_pm_get_flags_and_ifindex_by_id (sock_net (sk), local_id, +mptcp_pm_get_flags_and_ifindex_by_id (msk, local_id, &flags, &ifindex); subflow->remote_key = msk->remote_key; subflow->local_key = msk->local_key;
Signature · 1 line
2.31.1
PATCH v2 12/21 mptcp: netlink: split mptcp_pm_parse_addr into two functions
In reply to Kishen MaloorFrom: Florian Westphal <[email protected]>
Next patch will need to parse MPTCP_PM_ATTR_ADDR attributes and fill an mptcp_addr_info structure from a different genl command callback.
To avoid copy-paste, split the existing function to a helper that does the common part and then call the helper from the (renamed)mptcp_pm_parse_entry function.
--- net/mptcp/pm_netlink.c | 60 +++++++++++++++++++++++++++--------------- 1 file changed, 39 insertions(+), 21 deletions(-)
net/mptcp/pm_netlink.c +39 -21
@@ -1316,11 +1316,12 @@ static int mptcp_pm_family_to_addr(int family)return MPTCP_PM_ADDR_ATTR_ADDR4 ; } -static int mptcp_pm_parse_addr (struct nlattr *attr,struct genl_info *info, -bool require_family, -struct mptcp_pm_addr_entry *entry) +static int mptcp_pm_parse_pm_addr_attr (struct nlattr *tb[], +const struct nlattr *attr, +struct genl_info *info, +struct mptcp_addr_info *addr, +bool require_family) { -struct nlattr *tb[MPTCP_PM_ADDR_ATTR_MAX +1 ];int err, addr_addr;if (!attr) {
@@ -1334,27 +1335,29 @@ static int mptcp_pm_parse_addr(struct nlattr *attr, struct genl_info *info,if (err)return err; -memset (entry,0 ,sizeof (*entry)); +if (tb[MPTCP_PM_ADDR_ATTR_ID ]) + addr->id =nla_get_u8 (tb[MPTCP_PM_ADDR_ATTR_ID ]); +if (!tb[MPTCP_PM_ADDR_ATTR_FAMILY ]) {if (!require_family) - goto skip_family; +return err;NL_SET_ERR_MSG_ATTR (info->extack, attr,"missing family" );return -EINVAL ; } - entry->addr.family =nla_get_u16 (tb[MPTCP_PM_ADDR_ATTR_FAMILY ]); - if (entry->addr.family != AF_INET + addr->family =nla_get_u16 (tb[MPTCP_PM_ADDR_ATTR_FAMILY ]); + if (addr->family != AF_INET #if IS_ENABLED(CONFIG_MPTCP_IPV6) - && entry->addr.family !=AF_INET6 + && addr->family !=AF_INET6 #endif ) {NL_SET_ERR_MSG_ATTR (info->extack, attr,"unknown address family" );return -EINVAL ; } - addr_addr =mptcp_pm_family_to_addr (entry->addr.family); + addr_addr =mptcp_pm_family_to_addr (addr->family);if (!tb[addr_addr]) {NL_SET_ERR_MSG_ATTR (info->extack, attr,"missing address data" );
@@ -1362,22 +1365,37 @@ static int mptcp_pm_parse_addr(struct nlattr *attr, struct genl_info *info, }#if IS_ENABLED (CONFIG_MPTCP_IPV6 ) -if (entry->addr.family ==AF_INET6 ) - entry->addr.addr6 =nla_get_in6_addr (tb[addr_addr]); +if (addr->family ==AF_INET6 ) + addr->addr6 =nla_get_in6_addr (tb[addr_addr]); else#endif - entry->addr.addr.s_addr =nla_get_in_addr (tb[addr_addr]); + addr->addr.s_addr =nla_get_in_addr (tb[addr_addr]); + +if (tb[MPTCP_PM_ADDR_ATTR_PORT ]) + addr->port =htons (nla_get_u16 (tb[MPTCP_PM_ADDR_ATTR_PORT ])); + +return err; +} + +static int mptcp_pm_parse_entry (struct nlattr *attr,struct genl_info *info, +bool require_family, +struct mptcp_pm_addr_entry *entry) +{ +struct nlattr *tb[MPTCP_PM_ADDR_ATTR_MAX +1 ]; +int err; + +memset (entry,0 ,sizeof (*entry)); + + err =mptcp_pm_parse_pm_addr_attr (tb, attr, info, &entry->addr, require_family); +if (err) +return err; -skip_family:if (tb[MPTCP_PM_ADDR_ATTR_IF_IDX ]) {u32 val =nla_get_s32 (tb[MPTCP_PM_ADDR_ATTR_IF_IDX ]); entry->ifindex = val; } -if (tb[MPTCP_PM_ADDR_ATTR_ID ]) - entry->addr.id =nla_get_u8 (tb[MPTCP_PM_ADDR_ATTR_ID ]); -if (tb[MPTCP_PM_ADDR_ATTR_FLAGS ]) entry->flags =nla_get_u32 (tb[MPTCP_PM_ADDR_ATTR_FLAGS ]);
@@ -1432,7 +1450,7 @@ static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info)struct socket *lsk;int ret; - ret =mptcp_pm_parse_addr (attr, info, true, &addr); + ret =mptcp_pm_parse_entry (attr, info, true, &addr);if (ret <0 )return ret;
@@ -1648,7 +1666,7 @@ static int mptcp_nl_cmd_del_addr(struct sk_buff *skb, struct genl_info *info)unsigned int addr_max;int ret; - ret =mptcp_pm_parse_addr (attr, info, false, &addr); + ret =mptcp_pm_parse_entry (attr, info, false, &addr);if (ret <0 )return ret;
@@ -1824,7 +1842,7 @@ static int mptcp_nl_cmd_get_addr(struct sk_buff *skb, struct genl_info *info)void *reply;int ret; - ret =mptcp_pm_parse_addr (attr, info, false, &addr); + ret =mptcp_pm_parse_entry (attr, info, false, &addr);if (ret <0 )return ret;
@@ -2019,7 +2037,7 @@ static int mptcp_nl_cmd_set_flags(struct sk_buff *skb, struct genl_info *info)u8 bkup =0 , lookup_by_id =0 ;int ret; - ret =mptcp_pm_parse_addr (attr, info, false, &addr); + ret =mptcp_pm_parse_entry (attr, info, false, &addr);if (ret <0 )return ret;
Signature · 1 line
2.31.1
PATCH v2 13/21 mptcp: netlink: Add MPTCP_PM_CMD_ANNOUNCE
In reply to Kishen MaloorThis change adds a MPTCP netlink interface for issuing ADD_ADDR advertisements over the chosen MPTCP connection from a userspace path manager.
The command requires the following parameters:
{ token, { loc_id, family, daddr4 | daddr6 [, dport] } [, if_idx], flags/signal }.
--- include/uapi/linux/mptcp.h | 2 + net/mptcp/pm_netlink.c | 111 +++++++++++++++++++++++++++++++++++++ 2 files changed, 113 insertions(+)
include/uapi/linux/mptcp.h +2 -0
@@ -55,6 +55,7 @@ enum {MPTCP_PM_ATTR_ADDR ,/* nested address */ MPTCP_PM_ATTR_RCV_ADD_ADDRS ,/* u32 */ MPTCP_PM_ATTR_SUBFLOWS ,/* u32 */ +MPTCP_PM_ATTR_TOKEN ,/* u32 */ __MPTCP_PM_ATTR_MAX };
@@ -92,6 +93,7 @@ enum {MPTCP_PM_CMD_SET_LIMITS ,MPTCP_PM_CMD_GET_LIMITS ,MPTCP_PM_CMD_SET_FLAGS , +MPTCP_PM_CMD_ANNOUNCE , __MPTCP_PM_CMD_AFTER_LAST };
net/mptcp/pm_netlink.c +111 -0
129 lines not on this page · raw message
Signature · 1 line
2.31.1
Paolo Abeni
maintainerRe: [PATCH mptcp-next v2 13/21] mptcp: netlink: Add MPTCP_PM_CMD_ANNOUNCE
Quoted 108 lines · Kishen Maloor
This change adds a MPTCP netlink interface for issuing
ADD_ADDRadvertisements over the chosen MPTCP connection from a userspace path manager.The command requires the following parameters:
{ token, {loc_id, family, daddr4 | daddr6 [, dport] } [,if_idx], flags/signal }.Signed-off-by Kishen Maloor <[email protected]>--- include/uapi/linux/mptcp.h | 2 + net/mptcp/pm_netlink.c | 111 +++++++++++++++++++++++++++++++++++++ 2 files changed, 113 insertions(+)include/uapi/linux/mptcp.h +2
-0@@ -55,6 +55,7 @@ enum {MPTCP_PM_ATTR_ADDR ,/* nested address */ MPTCP_PM_ATTR_RCV_ADD_ADDRS ,/* u32 */ MPTCP_PM_ATTR_SUBFLOWS ,/* u32 */ +MPTCP_PM_ATTR_TOKEN ,/* u32 */ __MPTCP_PM_ATTR_MAX };@@ -92,6 +93,7 @@ enum {MPTCP_PM_CMD_SET_LIMITS ,MPTCP_PM_CMD_GET_LIMITS ,MPTCP_PM_CMD_SET_FLAGS , +MPTCP_PM_CMD_ANNOUNCE , __MPTCP_PM_CMD_AFTER_LAST };net/mptcp/pm_netlink.c +59
-0@@ -1268,6 +1268,7 @@ static const struct nla_policy mptcp_pm_policy[MPTCP_PM_ATTR_MAX + 1] = { NLA_POLICY_NESTED(mptcp_pm_addr_policy), [MPTCP_PM_ATTR_RCV_ADD_ADDRS] = { .type = NLA_U32, }, [MPTCP_PM_ATTR_SUBFLOWS] = { .type = NLA_U32, }, + [MPTCP_PM_ATTR_TOKEN] = { .type = NLA_U32, }, };void mptcp_pm_nl_subflow_chk_stale (const struct mptcp_sock *msk,struct sock *ssk)@@ -2028,6 +2029,111 @@ static int mptcp_nl_addr_backup(struct net *net,return ret; } +static int mptcp_nl_cmd_announce (struct sk_buff *skb,struct genl_info *info) +{ +struct nlattr *token = info->attrs[MPTCP_PM_ATTR_TOKEN ]; +struct nlattr *addr = info->attrs[MPTCP_PM_ATTR_ADDR ]; +struct pm_nl_pernet *pernet =genl_info_pm_nl (info); +struct mptcp_pm_addr_entry addr_val; +struct mptcp_local_lsk *lsk_ref; +bool reuse_port = false; +struct mptcp_sock *msk; +struct socket *lsk; +u32 token_val; +int err; + +if (!addr || !token) { +GENL_SET_ERR_MSG (info,"missing required inputs" ); +return -EINVAL ; + } + + token_val =nla_get_u32 (token); + + msk =mptcp_token_get_sock (sock_net (skb->sk), token_val); +if (!msk) { +NL_SET_ERR_MSG_ATTR (info->extack, token,"invalid token" ); +return -EINVAL ; + } + +if (READ_ONCE (msk->pm.pm_type) !=MPTCP_PM_TYPE_USERSPACE ) { +GENL_SET_ERR_MSG (info,"invalid request; userspace PM not selected" ); +return -EINVAL ; + } + + err =mptcp_pm_parse_entry (addr, info, true, &addr_val); +if (err <0 ) { +GENL_SET_ERR_MSG (info,"error parsing local address" ); +return err; + } + +if (addr_val.addr.id ==0 || !(addr_val.flags &MPTCP_PM_ADDR_FLAG_SIGNAL )) { +GENL_SET_ERR_MSG (info,"invalid addr id or flags" ); +return -EINVAL ; + } + +if (!addr_val.addr.port) { + addr_val.addr.port = + ((struct inet_sock *)inet_sk + ((struct sock *)msk))->inet_sport; + + reuse_port = true; + } + + lsk_ref =lsk_list_find (pernet, &addr_val.addr); + +if (!lsk_ref) { + err =mptcp_pm_nl_create_listen_socket (skb->sk, &addr_val, &lsk); +if ((err && !reuse_port) || (err && (err != -EADDRINUSE ) && reuse_port)) { +GENL_SET_ERR_MSG (info,"error creating listen socket" ); +return err; + }
net/mptcp/pm_netlink.c static int mptcp_nl_addr_backup(struct net *net, · line 2099
Quoted 10 lines
net/mptcp/pm_netlink.c +10
-0+ +if (lsk) { + lsk_ref =lsk_list_add (pernet, &addr_val.addr, lsk); +if (!lsk_ref) { +GENL_SET_ERR_MSG (info,"can't allocate lsk ref" ); +sock_release (lsk); +return -ENOMEM ; + } + } + }
It looks like that the above chunk of code is sort of recurring. it's likely worthy to factor it in some helper alike "lsk_list_find_or_create()"
Cheers,
Paolo
Re: [PATCH mptcp-next v2 13/21] mptcp: netlink: Add MPTCP_PM_CMD_ANNOUNCE
Quoted 115 lines · Paolo Abeni
Quoted 114 lines · Kishen Maloor
This change adds a MPTCP netlink interface for issuing
ADD_ADDRadvertisements over the chosen MPTCP connection from a userspace path manager.The command requires the following parameters:
{ token, {loc_id, family, daddr4 | daddr6 [, dport] } [,if_idx], flags/signal }.Signed-off-by Kishen Maloor <[email protected]>--- include/uapi/linux/mptcp.h | 2 + net/mptcp/pm_netlink.c | 111 +++++++++++++++++++++++++++++++++++++ 2 files changed, 113 insertions(+)include/uapi/linux/mptcp.h +2
-0@@ -55,6 +55,7 @@ enum {MPTCP_PM_ATTR_ADDR ,/* nested address */ MPTCP_PM_ATTR_RCV_ADD_ADDRS ,/* u32 */ MPTCP_PM_ATTR_SUBFLOWS ,/* u32 */ +MPTCP_PM_ATTR_TOKEN ,/* u32 */ __MPTCP_PM_ATTR_MAX };@@ -92,6 +93,7 @@ enum {MPTCP_PM_CMD_SET_LIMITS ,MPTCP_PM_CMD_GET_LIMITS ,MPTCP_PM_CMD_SET_FLAGS , +MPTCP_PM_CMD_ANNOUNCE , __MPTCP_PM_CMD_AFTER_LAST };net/mptcp/pm_netlink.c +65
-0@@ -1268,6 +1268,7 @@ static const struct nla_policy mptcp_pm_policy[MPTCP_PM_ATTR_MAX + 1] = { NLA_POLICY_NESTED(mptcp_pm_addr_policy), [MPTCP_PM_ATTR_RCV_ADD_ADDRS] = { .type = NLA_U32, }, [MPTCP_PM_ATTR_SUBFLOWS] = { .type = NLA_U32, }, + [MPTCP_PM_ATTR_TOKEN] = { .type = NLA_U32, }, };void mptcp_pm_nl_subflow_chk_stale (const struct mptcp_sock *msk,struct sock *ssk)@@ -2028,6 +2029,111 @@ static int mptcp_nl_addr_backup(struct net *net,return ret; } +static int mptcp_nl_cmd_announce (struct sk_buff *skb,struct genl_info *info) +{ +struct nlattr *token = info->attrs[MPTCP_PM_ATTR_TOKEN ]; +struct nlattr *addr = info->attrs[MPTCP_PM_ATTR_ADDR ]; +struct pm_nl_pernet *pernet =genl_info_pm_nl (info); +struct mptcp_pm_addr_entry addr_val; +struct mptcp_local_lsk *lsk_ref; +bool reuse_port = false; +struct mptcp_sock *msk; +struct socket *lsk; +u32 token_val; +int err; + +if (!addr || !token) { +GENL_SET_ERR_MSG (info,"missing required inputs" ); +return -EINVAL ; + } + + token_val =nla_get_u32 (token); + + msk =mptcp_token_get_sock (sock_net (skb->sk), token_val); +if (!msk) { +NL_SET_ERR_MSG_ATTR (info->extack, token,"invalid token" ); +return -EINVAL ; + } + +if (READ_ONCE (msk->pm.pm_type) !=MPTCP_PM_TYPE_USERSPACE ) { +GENL_SET_ERR_MSG (info,"invalid request; userspace PM not selected" ); +return -EINVAL ; + } + + err =mptcp_pm_parse_entry (addr, info, true, &addr_val); +if (err <0 ) { +GENL_SET_ERR_MSG (info,"error parsing local address" ); +return err; + } + +if (addr_val.addr.id ==0 || !(addr_val.flags &MPTCP_PM_ADDR_FLAG_SIGNAL )) { +GENL_SET_ERR_MSG (info,"invalid addr id or flags" ); +return -EINVAL ; + } + +if (!addr_val.addr.port) { + addr_val.addr.port = + ((struct inet_sock *)inet_sk + ((struct sock *)msk))->inet_sport; + + reuse_port = true; + } + + lsk_ref =lsk_list_find (pernet, &addr_val.addr); + +if (!lsk_ref) { + err =mptcp_pm_nl_create_listen_socket (skb->sk, &addr_val, &lsk); +if ((err && !reuse_port) || (err && (err != -EADDRINUSE ) && reuse_port)) { +GENL_SET_ERR_MSG (info,"error creating listen socket" ); +return err; + } + +if (lsk) { + lsk_ref =lsk_list_add (pernet, &addr_val.addr, lsk); +if (!lsk_ref) { +GENL_SET_ERR_MSG (info,"can't allocate lsk ref" ); +sock_release (lsk);
Quoted 8 lines
net/mptcp/pm_netlink.c static int mptcp_nl_addr_backup(struct net *net, · line 2099
Quoted 4 lines
net/mptcp/pm_netlink.c +4
-0+return -ENOMEM ; + } + } + }It looks like that the above chunk of code is sort of recurring. it's likely worthy to factor it in some helper alike "
lsk_list_find_or_create()"
The find->create_listen_sock sequence is run in 3 places, but there are deviations in the logic that follows a failed create_listen_sock. Hence, I did not over optimize. But I will check if it makes sense to refactor.
Quoted 5 lines
Cheers,
Paolo
PATCH v2 14/21 mptcp: selftests: support MPTCP_PM_CMD_ANNOUNCE
In reply to Kishen MaloorThis change updates the "pm_nl_ctl" testing sample with an "ann" (announce) option to support the newly added netlink interface command MPTCP_PM_CMD_ANNOUNCE to issue ADD_ADDR advertisements over the chosen MPTCP connection.
E.g. ./pm_nl_ctl ann 192.168.122.75 token 823274047 id 25 dev enp1s0
--- tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 130 ++++++++++++++++++ 1 file changed, 130 insertions(+)
tools/testing/selftests/net/mptcp/pm_nl_ctl.c +130 -0
@@ -6,6 +6,7 @@#include <stdlib.h> #include <string.h> #include <unistd.h> +#include <limits.h> #include <sys/socket.h> #include <sys/types.h>
@@ -26,6 +27,7 @@ static void syntax(char *argv[]) {fprintf (stderr,"%s add|get|set|del|flush|dump|accept [<args>]\n" , argv[0 ]);fprintf (stderr,"\tadd [flags signal|subflow|backup|fullmesh] [id <nr>] [dev <name>] <ip>\n" ); +fprintf (stderr,"\tann <local-ip> id <local-id> token <token> [port <local-port>] [dev <name>]\n" );fprintf (stderr,"\tdel <id> [<ip>]\n" );fprintf (stderr,"\tget <id>\n" );fprintf (stderr,"\tset <ip> [flags backup|nobackup]\n" );
@@ -170,6 +172,132 @@ static int resolve_mptcp_pm_netlink(int fd)return genl_parse_getfamily ((void *)data); } +int announce_addr (int fd,int pm_family,int argc,char *argv[]) +{ +char data[NLMSG_ALIGN (sizeof (struct nlmsghdr )) + +NLMSG_ALIGN (sizeof (struct genlmsghdr )) + +1024 ]; +u_int32_t flags =MPTCP_PM_ADDR_FLAG_SIGNAL ; +u_int32_t token =UINT_MAX ; +struct rtattr *rta, *addr; +u_int32_t id =UINT_MAX ; +struct nlmsghdr *nh; +u_int16_t family; +int addr_start; +int off =0 ; +int arg; + +memset (data,0 ,sizeof (data)); + nh = (void *)data; + off =init_genl_req (data, pm_family,MPTCP_PM_CMD_ANNOUNCE , +MPTCP_PM_VER ); + +if (argc <7 ) +syntax (argv); + +/* local-ip header */ + addr_start = off; + addr = (void *)(data + off); + addr->rta_type =NLA_F_NESTED |MPTCP_PM_ATTR_ADDR ; + addr->rta_len =RTA_LENGTH (0 ); + off +=NLMSG_ALIGN (addr->rta_len); + +/* local-ip data */ +/* record addr type */ + rta = (void *)(data + off); +if (inet_pton (AF_INET , argv[2 ],RTA_DATA (rta))) { + family =AF_INET ; + rta->rta_type =MPTCP_PM_ADDR_ATTR_ADDR4 ; + rta->rta_len =RTA_LENGTH (4 ); + }else if (inet_pton (AF_INET6 , argv[2 ],RTA_DATA (rta))) { + family =AF_INET6 ; + rta->rta_type =MPTCP_PM_ADDR_ATTR_ADDR6 ; + rta->rta_len =RTA_LENGTH (16 ); + }else +error (1 , errno,"can't parse ip %s" , argv[2 ]); + off +=NLMSG_ALIGN (rta->rta_len); + +/* addr family */ + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_FAMILY ; + rta->rta_len =RTA_LENGTH (2 ); +memcpy (RTA_DATA (rta), &family,2 ); + off +=NLMSG_ALIGN (rta->rta_len); + +/* addr flags (..SIGNAL) */ + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_FLAGS ; + rta->rta_len =RTA_LENGTH (4 ); +memcpy (RTA_DATA (rta), &flags,4 ); + off +=NLMSG_ALIGN (rta->rta_len); + +for (arg =3 ; arg < argc; arg++) { +if (!strcmp (argv[arg],"id" )) { +/* local-id */ +if (++arg >= argc) +error (1 ,0 ," missing id value" ); + + id =atoi (argv[arg]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_ID ; + rta->rta_len =RTA_LENGTH (1 ); +memcpy (RTA_DATA (rta), &id,1 ); + off +=NLMSG_ALIGN (rta->rta_len); + }else if (!strcmp (argv[arg],"dev" )) { +/* for the if_index */ +int32_t ifindex; + +if (++arg >= argc) +error (1 ,0 ," missing dev name" ); + + ifindex =if_nametoindex (argv[arg]); +if (!ifindex) +error (1 , errno,"unknown device %s" , argv[arg]); + + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_IF_IDX ; + rta->rta_len =RTA_LENGTH (4 ); +memcpy (RTA_DATA (rta), &ifindex,4 ); + off +=NLMSG_ALIGN (rta->rta_len); + }else if (!strcmp (argv[arg],"port" )) { +/* local-port (optional) */ +u_int16_t port; + +if (++arg >= argc) +error (1 ,0 ," missing port value" ); + + port =atoi (argv[arg]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_PORT ; + rta->rta_len =RTA_LENGTH (2 ); +memcpy (RTA_DATA (rta), &port,2 ); + off +=NLMSG_ALIGN (rta->rta_len); + }else if (!strcmp (argv[arg],"token" )) { +/* MPTCP connection token */ +if (++arg >= argc) +error (1 ,0 ," missing token value" ); + + token =atoi (argv[arg]); + }else +error (1 ,0 ,"unknown keyword %s" , argv[arg]); + } + addr->rta_len = off - addr_start; + +if (id ==UINT_MAX || token ==UINT_MAX ) +error (1 ,0 ," missing mandatory inputs" ); + +/* token */ + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ATTR_TOKEN ; + rta->rta_len =RTA_LENGTH (4 ); +memcpy (RTA_DATA (rta), &token,4 ); + off +=NLMSG_ALIGN (rta->rta_len); + +do_nl_req (fd, nh, off,0 ); + +return 0 ; +} +int add_addr (int fd,int pm_family,int argc,char *argv[]) { char data[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
@@ -744,6 +872,8 @@ int main(int argc, char *argv[])if (!strcmp (argv[1 ],"add" ))return add_addr (fd, pm_family, argc, argv); +else if (!strcmp (argv[1 ],"ann" )) +return announce_addr (fd, pm_family, argc, argv);else if (!strcmp (argv[1 ],"del" ))return del_addr (fd, pm_family, argc, argv);else if (!strcmp (argv[1 ],"flush" ))
Signature · 1 line
2.31.1
PATCH v2 15/21 mptcp: netlink: Add MPTCP_PM_CMD_REMOVE
In reply to Kishen MaloorThis change adds a MPTCP netlink command for issuing REMOVE_ADDR signals for a specific address over the chosen MPTCP connection from a userspace path manager.
The command requires the following parameters: {token, loc_id}.
--- include/uapi/linux/mptcp.h | 2 ++ net/mptcp/pm_netlink.c | 67 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 69 insertions(+)
include/uapi/linux/mptcp.h +2 -0
@@ -56,6 +56,7 @@ enum {MPTCP_PM_ATTR_RCV_ADD_ADDRS ,/* u32 */ MPTCP_PM_ATTR_SUBFLOWS ,/* u32 */ MPTCP_PM_ATTR_TOKEN ,/* u32 */ +MPTCP_PM_ATTR_LOC_ID ,/* u8 */ __MPTCP_PM_ATTR_MAX };
@@ -94,6 +95,7 @@ enum {MPTCP_PM_CMD_GET_LIMITS ,MPTCP_PM_CMD_SET_FLAGS ,MPTCP_PM_CMD_ANNOUNCE , +MPTCP_PM_CMD_REMOVE , __MPTCP_PM_CMD_AFTER_LAST };
net/mptcp/pm_netlink.c +67 -0
91 lines not on this page · raw message
Signature · 1 line
2.31.1
PATCH v2 16/21 mptcp: selftests: support MPTCP_PM_CMD_REMOVE
In reply to Kishen MaloorThis change updates the "pm_nl_ctl" testing sample with a "rem" (remove) option to support the newly added netlink interface command MPTCP_PM_CMD_REMOVE to issue REMOVE_ADDR signals over the chosen MPTCP connection.
E.g. ./pm_nl_ctl rem token 823274047 id 23
--- tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 52 +++++++++++++++++++ 1 file changed, 52 insertions(+)
tools/testing/selftests/net/mptcp/pm_nl_ctl.c +52 -0
@@ -28,6 +28,7 @@ static void syntax(char *argv[])fprintf (stderr,"%s add|get|set|del|flush|dump|accept [<args>]\n" , argv[0 ]);fprintf (stderr,"\tadd [flags signal|subflow|backup|fullmesh] [id <nr>] [dev <name>] <ip>\n" );fprintf (stderr,"\tann <local-ip> id <local-id> token <token> [port <local-port>] [dev <name>]\n" ); +fprintf (stderr,"\trem id <local-id> token <token>\n" );fprintf (stderr,"\tdel <id> [<ip>]\n" );fprintf (stderr,"\tget <id>\n" );fprintf (stderr,"\tset <ip> [flags backup|nobackup]\n" );
@@ -172,6 +173,55 @@ static int resolve_mptcp_pm_netlink(int fd)return genl_parse_getfamily ((void *)data); } +int remove_addr (int fd,int pm_family,int argc,char *argv[]) +{ +char data[NLMSG_ALIGN (sizeof (struct nlmsghdr )) + +NLMSG_ALIGN (sizeof (struct genlmsghdr )) + +1024 ]; +struct nlmsghdr *nh; +struct rtattr *rta; +u_int32_t token; +u_int8_t id; +int off =0 ; +int arg; + +memset (data,0 ,sizeof (data)); + nh = (void *)data; + off =init_genl_req (data, pm_family,MPTCP_PM_CMD_REMOVE , +MPTCP_PM_VER ); + +if (argc <6 ) +syntax (argv); + +for (arg =2 ; arg < argc; arg++) { +if (!strcmp (argv[arg],"id" )) { +if (++arg >= argc) +error (1 ,0 ," missing id value" ); + + id =atoi (argv[arg]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ATTR_LOC_ID ; + rta->rta_len =RTA_LENGTH (1 ); +memcpy (RTA_DATA (rta), &id,1 ); + off +=NLMSG_ALIGN (rta->rta_len); + }else if (!strcmp (argv[arg],"token" )) { +if (++arg >= argc) +error (1 ,0 ," missing token value" ); + + token =atoi (argv[arg]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ATTR_TOKEN ; + rta->rta_len =RTA_LENGTH (4 ); +memcpy (RTA_DATA (rta), &token,4 ); + off +=NLMSG_ALIGN (rta->rta_len); + }else +error (1 ,0 ,"unknown keyword %s" , argv[arg]); + } + +do_nl_req (fd, nh, off,0 ); +return 0 ; +} +int announce_addr (int fd,int pm_family,int argc,char *argv[]) { char data[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
@@ -874,6 +924,8 @@ int main(int argc, char *argv[])return add_addr (fd, pm_family, argc, argv); else if (!strcmp(argv[1], "ann")) return announce_addr(fd, pm_family, argc, argv); + else if (!strcmp(argv[1], "rem")) + return remove_addr(fd, pm_family, argc, argv); else if (!strcmp(argv[1], "del")) return del_addr(fd, pm_family, argc, argv); else if (!strcmp(argv[1], "flush"))
Signature · 1 line
2.31.1
PATCH v2 17/21 mptcp: netlink: allow userspace-driven subflow establishment
In reply to Kishen MaloorFrom: Florian Westphal <[email protected]>
This allows userspace to tell kernel to add a new subflow to an existing mptcp connection.
Userspace provides the token to identify the mptcp-level connection that needs a change in active subflows and the local and remote addresses of the new or the to-be-removed subflow.
MPTCP_PM_CMD_SUBFLOW_CREATE requires the following parameters:
{ token, { loc_id, family, loc_addr4 | loc_addr6 }, { family, rem_addr4 | rem_addr6, rem_port }
MPTCP_PM_CMD_SUBFLOW_DESTROY requires the following parameters:
{ token, { family, loc_addr4 | loc_addr6, loc_port }, { family, rem_addr4 | rem_addr6, rem_port }
v2: fix build error when IPv6 is not enabled
--- include/uapi/linux/mptcp.h | 3 + net/mptcp/pm_netlink.c | 206 +++++++++++++++++++++++++++++++++++++ 2 files changed, 209 insertions(+)
include/uapi/linux/mptcp.h +3 -0
@@ -57,6 +57,7 @@ enum {MPTCP_PM_ATTR_SUBFLOWS ,/* u32 */ MPTCP_PM_ATTR_TOKEN ,/* u32 */ MPTCP_PM_ATTR_LOC_ID ,/* u8 */ +MPTCP_PM_ATTR_ADDR_REMOTE ,/* nested address */ __MPTCP_PM_ATTR_MAX };
@@ -96,6 +97,8 @@ enum {MPTCP_PM_CMD_SET_FLAGS ,MPTCP_PM_CMD_ANNOUNCE ,MPTCP_PM_CMD_REMOVE , +MPTCP_PM_CMD_SUBFLOW_CREATE , +MPTCP_PM_CMD_SUBFLOW_DESTROY , __MPTCP_PM_CMD_AFTER_LAST };
net/mptcp/pm_netlink.c +206 -0
230 lines not on this page · raw message
Signature · 1 line
2.31.1
PATCH v2 18/21 mptcp: selftests: support MPTCP_PM_CMD_SUBFLOW_CREATE
In reply to Kishen MaloorThis change updates the "pm_nl_ctl" testing sample with a "csf" (create subflow) option to support the newly added netlink interface command MPTCP_PM_CMD_SUBFLOW_CREATE over the chosen MPTCP connection.
E.g. ./pm_nl_ctl csf lip 10.0.2.1 lid 23 rip 10.0.2.2 rport 56789 token 823274047
--- tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 129 ++++++++++++++++++ 1 file changed, 129 insertions(+)
tools/testing/selftests/net/mptcp/pm_nl_ctl.c +129 -0
@@ -29,6 +29,7 @@ static void syntax(char *argv[])fprintf (stderr,"\tadd [flags signal|subflow|backup|fullmesh] [id <nr>] [dev <name>] <ip>\n" );fprintf (stderr,"\tann <local-ip> id <local-id> token <token> [port <local-port>] [dev <name>]\n" );fprintf (stderr,"\trem id <local-id> token <token>\n" ); +fprintf (stderr,"\tcsf lip <local-ip> lid <local-id> rip <remote-ip> rport <remote-port> token <token>\n" );fprintf (stderr,"\tdel <id> [<ip>]\n" );fprintf (stderr,"\tget <id>\n" );fprintf (stderr,"\tset <ip> [flags backup|nobackup]\n" );
@@ -173,6 +174,132 @@ static int resolve_mptcp_pm_netlink(int fd)return genl_parse_getfamily ((void *)data); } +int csf (int fd,int pm_family,int argc,char *argv[]) +{ +char data[NLMSG_ALIGN (sizeof (struct nlmsghdr )) + +NLMSG_ALIGN (sizeof (struct genlmsghdr )) + +1024 ]; +const char *params[5 ]; +struct nlmsghdr *nh; +struct rtattr *addr; +struct rtattr *rta; +u_int16_t family; +u_int32_t token; +u_int16_t port; +int addr_start; +u_int8_t id; +int off =0 ; +int arg; + +memset (params,0 ,5 *sizeof (const char *)); + +memset (data,0 ,sizeof (data)); + nh = (void *)data; + off =init_genl_req (data, pm_family,MPTCP_PM_CMD_SUBFLOW_CREATE , +MPTCP_PM_VER ); + +if (argc <12 ) +syntax (argv); + +/* Params recorded in this order: +* <local-ip>, <local-id>, <remote-ip>, <remote-port>, <token> +*/ +for (arg =2 ; arg < argc; arg++) { +if (!strcmp (argv[arg],"lip" )) { +if (++arg >= argc) +error (1 ,0 ," missing local IP" ); + + params[0 ] = argv[arg]; + }else if (!strcmp (argv[arg],"lid" )) { +if (++arg >= argc) +error (1 ,0 ," missing local id" ); + + params[1 ] = argv[arg]; + }else if (!strcmp (argv[arg],"rip" )) { +if (++arg >= argc) +error (1 ,0 ," missing remote ip" ); + + params[2 ] = argv[arg]; + }else if (!strcmp (argv[arg],"rport" )) { +if (++arg >= argc) +error (1 ,0 ," missing remote port" ); + + params[3 ] = argv[arg]; + }else if (!strcmp (argv[arg],"token" )) { +if (++arg >= argc) +error (1 ,0 ," missing token" ); + + params[4 ] = argv[arg]; + }else +error (1 ,0 ,"unknown param %s" , argv[arg]); + } + +for (arg =0 ; arg <4 ; arg = arg +2 ) { +/* addr header */ + addr_start = off; + addr = (void *)(data + off); + addr->rta_type =NLA_F_NESTED | + ((arg ==0 ) ?MPTCP_PM_ATTR_ADDR :MPTCP_PM_ATTR_ADDR_REMOTE ); + addr->rta_len =RTA_LENGTH (0 ); + off +=NLMSG_ALIGN (addr->rta_len); + +/* addr data */ + rta = (void *)(data + off); +if (inet_pton (AF_INET , params[arg],RTA_DATA (rta))) { + family =AF_INET ; + rta->rta_type =MPTCP_PM_ADDR_ATTR_ADDR4 ; + rta->rta_len =RTA_LENGTH (4 ); + }else if (inet_pton (AF_INET6 , params[arg],RTA_DATA (rta))) { + family =AF_INET6 ; + rta->rta_type =MPTCP_PM_ADDR_ATTR_ADDR6 ; + rta->rta_len =RTA_LENGTH (16 ); + }else +error (1 , errno,"can't parse ip %s" , params[arg]); + off +=NLMSG_ALIGN (rta->rta_len); + +/* family */ + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_FAMILY ; + rta->rta_len =RTA_LENGTH (2 ); +memcpy (RTA_DATA (rta), &family,2 ); + off +=NLMSG_ALIGN (rta->rta_len); + +if (arg ==2 ) { +/* port */ + port =atoi (params[arg +1 ]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_PORT ; + rta->rta_len =RTA_LENGTH (2 ); +memcpy (RTA_DATA (rta), &port,2 ); + off +=NLMSG_ALIGN (rta->rta_len); + } + +if (arg ==0 ) { +/* id */ + id =atoi (params[arg +1 ]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_ID ; + rta->rta_len =RTA_LENGTH (1 ); +memcpy (RTA_DATA (rta), &id,1 ); + off +=NLMSG_ALIGN (rta->rta_len); + } + + addr->rta_len = off - addr_start; + } + +/* token */ + token =atoi (params[4 ]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ATTR_TOKEN ; + rta->rta_len =RTA_LENGTH (4 ); +memcpy (RTA_DATA (rta), &token,4 ); + off +=NLMSG_ALIGN (rta->rta_len); + +do_nl_req (fd, nh, off,0 ); + +return 0 ; +} +int remove_addr (int fd,int pm_family,int argc,char *argv[]) { char data[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
@@ -926,6 +1053,8 @@ int main(int argc, char *argv[])return announce_addr (fd, pm_family, argc, argv); else if (!strcmp(argv[1], "rem")) return remove_addr(fd, pm_family, argc, argv); + else if (!strcmp(argv[1], "csf")) + return csf(fd, pm_family, argc, argv); else if (!strcmp(argv[1], "del")) return del_addr(fd, pm_family, argc, argv); else if (!strcmp(argv[1], "flush"))
Signature · 1 line
2.31.1
PATCH v2 19/21 mptcp: selftests: support MPTCP_PM_CMD_SUBFLOW_DESTROY
In reply to Kishen MaloorThis change updates the "pm_nl_ctl" testing sample with a "dsf" (destroy subflow) option to support the newly added netlink interface command MPTCP_PM_CMD_SUBFLOW_DESTROY over the chosen MPTCP connection.
E.g. ./pm_nl_ctl dsf lip 10.0.2.1 lport 44567 rip 10.0.2.2 rport 56789 token 823274047
--- tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 115 ++++++++++++++++++ 1 file changed, 115 insertions(+)
tools/testing/selftests/net/mptcp/pm_nl_ctl.c +115 -0
@@ -30,6 +30,7 @@ static void syntax(char *argv[])fprintf (stderr,"\tann <local-ip> id <local-id> token <token> [port <local-port>] [dev <name>]\n" );fprintf (stderr,"\trem id <local-id> token <token>\n" );fprintf (stderr,"\tcsf lip <local-ip> lid <local-id> rip <remote-ip> rport <remote-port> token <token>\n" ); +fprintf (stderr,"\tdsf lip <local-ip> lport <local-port> rip <remote-ip> rport <remote-port> token <token>\n" );fprintf (stderr,"\tdel <id> [<ip>]\n" );fprintf (stderr,"\tget <id>\n" );fprintf (stderr,"\tset <ip> [flags backup|nobackup]\n" );
@@ -174,6 +175,118 @@ static int resolve_mptcp_pm_netlink(int fd)return genl_parse_getfamily ((void *)data); } +int dsf (int fd,int pm_family,int argc,char *argv[]) +{ +char data[NLMSG_ALIGN (sizeof (struct nlmsghdr )) + +NLMSG_ALIGN (sizeof (struct genlmsghdr )) + +1024 ]; +struct rtattr *rta, *addr; +u_int16_t family, port; +struct nlmsghdr *nh; +u_int32_t token; +int addr_start; +int off =0 ; +int arg; + +const char *params[5 ]; + +memset (params,0 ,5 *sizeof (const char *)); + +memset (data,0 ,sizeof (data)); + nh = (void *)data; + off =init_genl_req (data, pm_family,MPTCP_PM_CMD_SUBFLOW_DESTROY , +MPTCP_PM_VER ); + +if (argc <12 ) +syntax (argv); + +/* Params recorded in this order: +* <local-ip>, <local-port>, <remote-ip>, <remote-port>, <token> +*/ +for (arg =2 ; arg < argc; arg++) { +if (!strcmp (argv[arg],"lip" )) { +if (++arg >= argc) +error (1 ,0 ," missing local IP" ); + + params[0 ] = argv[arg]; + }else if (!strcmp (argv[arg],"lport" )) { +if (++arg >= argc) +error (1 ,0 ," missing local port" ); + + params[1 ] = argv[arg]; + }else if (!strcmp (argv[arg],"rip" )) { +if (++arg >= argc) +error (1 ,0 ," missing remote IP" ); + + params[2 ] = argv[arg]; + }else if (!strcmp (argv[arg],"rport" )) { +if (++arg >= argc) +error (1 ,0 ," missing remote port" ); + + params[3 ] = argv[arg]; + }else if (!strcmp (argv[arg],"token" )) { +if (++arg >= argc) +error (1 ,0 ," missing token" ); + + params[4 ] = argv[arg]; + }else +error (1 ,0 ,"unknown keyword %s" , argv[arg]); + } + +for (arg =0 ; arg <4 ; arg = arg +2 ) { +/* addr header */ + addr_start = off; + addr = (void *)(data + off); + addr->rta_type =NLA_F_NESTED | + ((arg ==0 ) ?MPTCP_PM_ATTR_ADDR :MPTCP_PM_ATTR_ADDR_REMOTE ); + addr->rta_len =RTA_LENGTH (0 ); + off +=NLMSG_ALIGN (addr->rta_len); + +/* addr data */ + rta = (void *)(data + off); +if (inet_pton (AF_INET , params[arg],RTA_DATA (rta))) { + family =AF_INET ; + rta->rta_type =MPTCP_PM_ADDR_ATTR_ADDR4 ; + rta->rta_len =RTA_LENGTH (4 ); + }else if (inet_pton (AF_INET6 , params[arg],RTA_DATA (rta))) { + family =AF_INET6 ; + rta->rta_type =MPTCP_PM_ADDR_ATTR_ADDR6 ; + rta->rta_len =RTA_LENGTH (16 ); + }else +error (1 , errno,"can't parse ip %s" , params[arg]); + off +=NLMSG_ALIGN (rta->rta_len); + +/* family */ + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_FAMILY ; + rta->rta_len =RTA_LENGTH (2 ); +memcpy (RTA_DATA (rta), &family,2 ); + off +=NLMSG_ALIGN (rta->rta_len); + +/* port */ + port =atoi (params[arg +1 ]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ADDR_ATTR_PORT ; + rta->rta_len =RTA_LENGTH (2 ); +memcpy (RTA_DATA (rta), &port,2 ); + off +=NLMSG_ALIGN (rta->rta_len); + + addr->rta_len = off - addr_start; + } + +/* token */ + token =atoi (params[4 ]); + rta = (void *)(data + off); + rta->rta_type =MPTCP_PM_ATTR_TOKEN ; + rta->rta_len =RTA_LENGTH (4 ); +memcpy (RTA_DATA (rta), &token,4 ); + off +=NLMSG_ALIGN (rta->rta_len); + +do_nl_req (fd, nh, off,0 ); + +return 0 ; +} +int csf (int fd,int pm_family,int argc,char *argv[]) { char data[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
@@ -1055,6 +1168,8 @@ int main(int argc, char *argv[])return remove_addr (fd, pm_family, argc, argv); else if (!strcmp(argv[1], "csf")) return csf(fd, pm_family, argc, argv); + else if (!strcmp(argv[1], "dsf")) + return dsf(fd, pm_family, argc, argv); else if (!strcmp(argv[1], "del")) return del_addr(fd, pm_family, argc, argv); else if (!strcmp(argv[1], "flush"))
Signature · 1 line
2.31.1
PATCH v2 20/21 mptcp: selftests: capture netlink events
In reply to Kishen MaloorThis change adds to self-testing support for the MPTCP netlink interface by capturing various MPTCP netlink events (and all their metadata) associated with connections, subflows and local address announcements. It can be incorporated into self-test scripts that exercise the MPTCP netlink commands to then precisely validate those operations through the dispatched MPTCP netlink events in response to those commands.
--- tools/testing/selftests/net/mptcp/pm_nl_ctl.c | 161 +++++++++++++++++- 1 file changed, 154 insertions(+), 7 deletions(-)
tools/testing/selftests/net/mptcp/pm_nl_ctl.c +154 -7
@@ -22,6 +22,9 @@#ifndef MPTCP_PM_NAME #define MPTCP_PM_NAME "mptcp_pm"#endif +#ifndef MPTCP_PM_EVENTS +#define MPTCP_PM_EVENTS "mptcp_pm_events" +#endif static void syntax (char *argv[]) {
@@ -37,6 +40,7 @@ static void syntax(char *argv[])fprintf (stderr,"\tflush\n" );fprintf (stderr,"\tdump\n" );fprintf (stderr,"\tlimits [<rcv addr max> <subflow max>]\n" ); +fprintf (stderr,"\tevents\n" );exit (0 ); }
@@ -88,6 +92,105 @@ static void nl_error(struct nlmsghdr *nh) } } +static int capture_events (int fd,int event_group) +{ +u_int8_t buffer[NLMSG_ALIGN (sizeof (struct nlmsghdr )) + +NLMSG_ALIGN (sizeof (struct genlmsghdr )) +1024 ]; +struct genlmsghdr *ghdr; +struct rtattr *attrs; +struct nlmsghdr *nh; +int ret =0 ; +int res_len; +int msg_len; +fd_set rfds; + +if (setsockopt (fd,SOL_NETLINK ,NETLINK_ADD_MEMBERSHIP , + &event_group,sizeof (event_group)) <0 ) +error (1 , errno,"could not join the " MPTCP_PM_EVENTS " mcast group" ); + +do { +FD_ZERO (&rfds); +FD_SET (fd, &rfds); + res_len =NLMSG_ALIGN (sizeof (struct nlmsghdr )) + +NLMSG_ALIGN (sizeof (struct genlmsghdr )) +1024 ; + + ret =select (FD_SETSIZE , &rfds,NULL ,NULL ,NULL ); + +if (ret <0 ) +error (1 , ret,"error in select() on NL socket" ); + + res_len =recv (fd, buffer, res_len,0 ); +if (res_len <0 ) +error (1 , res_len,"error on recv() from NL socket" ); + + nh = (struct nlmsghdr *)buffer; + +for (;NLMSG_OK (nh, res_len); nh =NLMSG_NEXT (nh, res_len)) { +if (nh->nlmsg_type ==NLMSG_ERROR ) +error (1 ,NLMSG_ERROR ,"received invalid NL message" ); + + ghdr = (struct genlmsghdr *)NLMSG_DATA (nh); + +if (ghdr->cmd ==0 ) +continue ; + +fprintf (stderr,"type:%d" , ghdr->cmd); + + msg_len = nh->nlmsg_len -NLMSG_LENGTH (GENL_HDRLEN ); + + attrs = (struct rtattr *) ((char *) ghdr +GENL_HDRLEN ); +while (RTA_OK (attrs, msg_len)) { +if (attrs->rta_type ==MPTCP_ATTR_TOKEN ) +fprintf (stderr,",token:%u" , *(__u32 *)RTA_DATA (attrs)); +else if (attrs->rta_type ==MPTCP_ATTR_FAMILY ) +fprintf (stderr,",family:%u" , *(__u16 *)RTA_DATA (attrs)); +else if (attrs->rta_type ==MPTCP_ATTR_LOC_ID ) +fprintf (stderr,",loc_id:%u" , *(__u8 *)RTA_DATA (attrs)); +else if (attrs->rta_type ==MPTCP_ATTR_REM_ID ) +fprintf (stderr,",rem_id:%u" , *(__u8 *)RTA_DATA (attrs)); +else if (attrs->rta_type ==MPTCP_ATTR_SADDR4 ) { +u_int32_t saddr4 =ntohl (*(__u32 *)RTA_DATA (attrs)); + +fprintf (stderr,",saddr4:%u.%u.%u.%u" , saddr4 >>24 , + (saddr4 >>16 ) &0xFF , (saddr4 >>8 ) &0xFF , + (saddr4 &0xFF )); + }else if (attrs->rta_type ==MPTCP_ATTR_SADDR6 ) { +char buf[INET6_ADDRSTRLEN ]; + +if (inet_ntop (AF_INET6 ,RTA_DATA (attrs), buf, +sizeof (buf)) !=NULL ) +fprintf (stderr,",saddr6:%s" , buf); + }else if (attrs->rta_type ==MPTCP_ATTR_DADDR4 ) { +u_int32_t daddr4 =ntohl (*(__u32 *)RTA_DATA (attrs)); + +fprintf (stderr,",daddr4:%u.%u.%u.%u" , daddr4 >>24 , + (daddr4 >>16 ) &0xFF , (daddr4 >>8 ) &0xFF , + (daddr4 &0xFF )); + }else if (attrs->rta_type ==MPTCP_ATTR_DADDR6 ) { +char buf[INET6_ADDRSTRLEN ]; + +if (inet_ntop (AF_INET6 ,RTA_DATA (attrs), buf, +sizeof (buf)) !=NULL ) +fprintf (stderr,",daddr6:%s" , buf); + }else if (attrs->rta_type ==MPTCP_ATTR_SPORT ) +fprintf (stderr,",sport:%u" , +ntohs (*(__u16 *)RTA_DATA (attrs))); +else if (attrs->rta_type ==MPTCP_ATTR_DPORT ) +fprintf (stderr,",dport:%u" , +ntohs (*(__u16 *)RTA_DATA (attrs))); +else if (attrs->rta_type ==MPTCP_ATTR_BACKUP ) +fprintf (stderr,",backup:%u" , *(__u8 *)RTA_DATA (attrs)); +else if (attrs->rta_type ==MPTCP_ATTR_ERROR ) +fprintf (stderr,",error:%u" , *(__u8 *)RTA_DATA (attrs)); + attrs =RTA_NEXT (attrs, msg_len); + } + } +fprintf (stderr,"\n" ); + }while (1 ); + +return 0 ; +} +/* do a netlink command and, if max > 0, fetch the reply */ static int do_nl_req (int fd,struct nlmsghdr *nh,int len,int max) {
@@ -121,11 +224,18 @@ static int do_nl_req(int fd, struct nlmsghdr *nh, int len, int max)return ret; } -static int genl_parse_getfamily (struct nlmsghdr *nlh) +static int genl_parse_getfamily (struct nlmsghdr *nlh,int *pm_family, +int *events_mcast_grp) {struct genlmsghdr *ghdr =NLMSG_DATA (nlh);int len = nlh->nlmsg_len;struct rtattr *attrs; +struct rtattr *grps; +struct rtattr *grp; +int got_events_grp; +int got_family; +int grps_len; +int grp_len;if (nlh->nlmsg_type !=GENL_ID_CTRL ) error(1, errno, "Not a controller message, len=%d type=0x%x\n",
@@ -140,9 +250,42 @@ static int genl_parse_getfamily(struct nlmsghdr *nlh)error (1 , errno,"Unknown controller command %d\n" , ghdr->cmd); attrs = (struct rtattr *) ((char *) ghdr +GENL_HDRLEN ); + got_family =0 ; + got_events_grp =0 ; +while (RTA_OK (attrs, len)) { -if (attrs->rta_type ==CTRL_ATTR_FAMILY_ID ) -return *(__u16 *)RTA_DATA (attrs); +if (attrs->rta_type ==CTRL_ATTR_FAMILY_ID ) { + *pm_family = *(__u16 *)RTA_DATA (attrs); + got_family =1 ; + }else if (attrs->rta_type ==CTRL_ATTR_MCAST_GROUPS ) { + grps =RTA_DATA (attrs); + grps_len =RTA_PAYLOAD (attrs); + +while (RTA_OK (grps, grps_len)) { + grp =RTA_DATA (grps); + grp_len =RTA_PAYLOAD (grps); + got_events_grp =0 ; + +while (RTA_OK (grp, grp_len)) { +if (grp->rta_type ==CTRL_ATTR_MCAST_GRP_ID ) + *events_mcast_grp = *(__u32 *)RTA_DATA (grp); +else if (grp->rta_type ==CTRL_ATTR_MCAST_GRP_NAME && + !strcmp (RTA_DATA (grp),MPTCP_PM_EVENTS )) + got_events_grp =1 ; + + grp =RTA_NEXT (grp, grp_len); + } + +if (got_events_grp) +break ; + + grps =RTA_NEXT (grps, grps_len); + } + } + +if (got_family && got_events_grp) +return 0 ; + attrs =RTA_NEXT (attrs, len); }
@@ -150,7 +293,7 @@ static int genl_parse_getfamily(struct nlmsghdr *nlh)return -1 ; } -static int resolve_mptcp_pm_netlink (int fd) +static int resolve_mptcp_pm_netlink (int fd,int *pm_family,int *events_mcast_grp) { char data[NLMSG_ALIGN(sizeof(struct nlmsghdr)) + NLMSG_ALIGN(sizeof(struct genlmsghdr)) +
@@ -172,7 +315,7 @@ static int resolve_mptcp_pm_netlink(int fd) off +=NLMSG_ALIGN (rta->rta_len);do_nl_req (fd, nh, off,sizeof (data)); -return genl_parse_getfamily ((void *)data); +return genl_parse_getfamily ((void *)data, pm_family, events_mcast_grp); }int dsf (int fd,int pm_family,int argc,char *argv[])
@@ -1149,7 +1292,9 @@ int set_flags(int fd, int pm_family, int argc, char *argv[])int main (int argc,char *argv[]) { -int fd, pm_family; +int events_mcast_grp; +int pm_family; +int fd;if (argc <2 )syntax (argv);
@@ -1158,7 +1303,7 @@ int main(int argc, char *argv[])if (fd ==-1 )error (1 , errno,"socket netlink" ); - pm_family =resolve_mptcp_pm_netlink (fd); +resolve_mptcp_pm_netlink (fd, &pm_family, &events_mcast_grp);if (!strcmp (argv[1 ],"add" ))return add_addr (fd, pm_family, argc, argv);
@@ -1182,6 +1327,8 @@ int main(int argc, char *argv[])return get_set_limits (fd, pm_family, argc, argv); else if (!strcmp(argv[1], "set")) return set_flags(fd, pm_family, argc, argv); + else if (!strcmp(argv[1], "events")) + return capture_events(fd, events_mcast_grp);fprintf (stderr,"unknown sub-command: %s" , argv[1 ]);syntax (argv);
Signature · 1 line
2.31.1
PATCH v2 21/21 selftests: mptcp: functional tests for the userspace PM type
In reply to Kishen MaloorThis change adds a selftest script that performs a comprehensive behavioral/functional test of all userspace PM capabilities by exercising all the newly added APIs and changes to support said capabilities.
--- .../selftests/net/mptcp/userspace_pm.sh | 561 ++++++++++++++++++ 1 file changed, 561 insertions(+) create mode 100755 tools/testing/selftests/net/mptcp/userspace_pm.sh
tools/testing/selftests/net/mptcp/userspace_pm.sh new file +561 -0
@@ -0,0 +1,561 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0 + +ip -Version > /dev/null 2>&1 +if [ $? -ne 0 ];then + echo "SKIP: Cannot not run test without ip tool" + exit 1 +fi + +ANNOUNCED=6 # MPTCP_EVENT_ANNOUNCED +REMOVED=7 # MPTCP_EVENT_REMOVED +SUB_ESTABLISHED=10 # MPTCP_EVENT_SUB_ESTABLISHED +SUB_CLOSED=11 # MPTCP_EVENT_SUB_CLOSED + +AF_INET=2 +AF_INET6=10 + +evts_pid=0 +client4_pid=0 +server4_pid=0 +client6_pid=0 +server6_pid=0 +client4_token="" +server4_token="" +client6_token="" +server6_token="" +client4_port=0; +client6_port=0; +app4_port=50002 +new4_port=50003 +app6_port=50004 +client_addr_id=${RANDOM:0:2} +server_addr_id=${RANDOM:0:2} + +sec=$(date +%s) +rndh=$(printf %x $sec)-$(mktemp -u XXXXXX) +ns1="ns1-$rndh" +ns2="ns2-$rndh" + +cleanup() +{ + echo "cleanup" + + # Terminate the MPTCP connection and related processes + kill -SIGUSR1 $client4_pid > /dev/null 2>&1 + kill $server4_pid > /dev/null 2>&1 + kill -SIGUSR1 $client6_pid > /dev/null 2>&1 + kill $server6_pid > /dev/null 2>&1 + + kill $evts_pid > /dev/null 2>&1 + + local netns + for netns in "$ns1" "$ns2" ;do + ip netns del $netns + done +} + +trap cleanup EXIT + +for i in "$ns1" "$ns2" ;do + ip netns add $i || exit 1 + ip -net $i link set lo up + ip netns exec $i sysctl -q net.mptcp.enabled=1 + ip netns exec $i sysctl -q net.mptcp.pm_type=1 +done + +# "$ns1" ns2 +# ns1eth2 ns2eth1 + +ip link add ns1eth2 netns "$ns1" type veth peer name ns2eth1 netns "$ns2" + +ip -net "$ns1" addr add 10.0.1.1/24 dev ns1eth2 +ip -net "$ns1" addr add 10.0.2.1/24 dev ns1eth2 +ip -net "$ns1" addr add dead:beef:1::1/64 dev ns1eth2 nodad +ip -net "$ns1" addr add dead:beef:2::1/64 dev ns1eth2 nodad +ip -net "$ns1" link set ns1eth2 up + +ip -net "$ns2" addr add 10.0.1.2/24 dev ns2eth1 +ip -net "$ns2" addr add 10.0.2.2/24 dev ns2eth1 +ip -net "$ns2" addr add dead:beef:1::2/64 dev ns2eth1 nodad +ip -net "$ns2" addr add dead:beef:2::2/64 dev ns2eth1 nodad +ip -net "$ns2" link set ns2eth1 up + +printf "Created network namespaces ns1, ns2 \t\t\t[OK]\n" + +make_file() +{ + local name=$1 + local who=$2 + local ksize=1 + + dd if=/dev/urandom of="$name" bs=1024 count=$ksize 2> /dev/null + echo -e "\nMPTCP_TEST_FILE_END_MARKER" >> "$name" +} + +make_connection() +{ + local file=$(mktemp) + make_file "$file" "client" + + local is_v6=$1 + local app_port=$app4_port + local connect_addr="10.0.1.1" + local listen_addr="0.0.0.0" + if [ "$is_v6" = "v6" ] + then + connect_addr="dead:beef:1::1" + listen_addr="::" + app_port=$app6_port + else + is_v6="v4" + fi + + local client_evts=$(mktemp) + :>"$client_evts" + ip netns exec $ns2 ./pm_nl_ctl events >> "$client_evts" 2>&1 & + local client_evts_pid=$! + local server_evts=$(mktemp) + :>"$server_evts" + ip netns exec $ns1 ./pm_nl_ctl events >> "$server_evts" 2>&1 & + local server_evts_pid=$! + sleep 0.1 + + # Run the server + ip netns exec $ns1 \ + ./mptcp_connect -s MPTCP -w 300 -p $app_port -l $listen_addr 2>&1 > /dev/null & + local server_pid=$! + sleep 0.1 + + # Run the client, transfer $file and stay connected to the server + # to conduct tests + ip netns exec $ns2 \ + ./mptcp_connect -s MPTCP -w 300 -m sendfile -p $app_port $connect_addr 2>&1 > /dev/null < $file & + local client_pid=$! + sleep 0.1 + + kill $client_evts_pid + local client_token=$(sed -n 's/.*\(token:\)\([[:digit:]]*\).*$/\2/p;q' "$client_evts") + local client_port=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$client_evts") + + kill $server_evts_pid + local server_token=$(sed -n 's/.*\(token:\)\([[:digit:]]*\).*$/\2/p;q' "$server_evts") + rm -f "$client_evts" "$server_evts" "$file" + + if [ $client_token != "" ] && [ $server_token != "" ] + then + printf "Established IP%s MPTCP Connection ns2 => ns1 \t\t[OK]\n" $is_v6 + else + exit 1 + fi + + if [ "$is_v6" = "v6" ] + then + client6_token=$client_token + server6_token=$server_token + client6_port=$client_port + client6_pid=$client_pid + server6_pid=$server_pid + else + client4_token=$client_token + server4_token=$server_token + client4_port=$client_port + client4_pid=$client_pid + server4_pid=$server_pid + fi +} + +verify_announce_event() +{ + local evt=$1 + local e_type=$2 + local e_token=$3 + local e_addr=$4 + local e_id=$5 + local e_dport=$6 + local e_af=$7 + + local type=$(sed -n 's/.*\(type:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local token=$(sed -n 's/.*\(token:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local addr="" + if [ "$e_af" = "v6" ] + then + addr=$(sed -n 's/.*\(daddr6:\)\([0-9a-f:.]*\).*$/\2/p;q' "$evt") + else + addr=$(sed -n 's/.*\(daddr4:\)\([0-9.]*\).*$/\2/p;q' "$evt") + fi + local dport=$(sed -n 's/.*\(dport:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local id=$(sed -n 's/.*\(rem_id:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + if [ "$type" = "$e_type" ] && [ "$token" = "$e_token" ] && [ "$addr" = "$e_addr" ] && [ "$dport" = "$e_dport" ] && [ "$id" = "$e_id" ] + then + printf "[OK]\n" + return 0 + fi + printf "[FAIL]\n" + exit 1 +} + +test_announce() +{ + local evts=$(mktemp) + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl events >> "$evts" 2>&1 & + evts_pid=$! + sleep 0.1 + + local invalid_token=$(( $client4_token - 1)) + ip netns exec $ns2 ./pm_nl_ctl ann 10.0.2.2 token $invalid_token id $client_addr_id dev ns2eth1 2>&1 > /dev/null + local type=$(sed -n 's/.*\(type:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + printf "ADD_ADDR 10.0.2.2 (ns2) => ns1, invalid token \t\t" + if [ "$type" = "" ] + then + printf "[OK]\n" + else + printf "[FAIL]\n" + exit 1 + fi + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl ann 10.0.2.2 token $client4_token id $client_addr_id dev ns2eth1 2>&1 > /dev/null + printf "ADD_ADDR id:%d 10.0.2.2 (ns2) => ns1, reuse port \t\t" $client_addr_id + sleep 0.1 + verify_announce_event "$evts" "$ANNOUNCED" "$server4_token" "10.0.2.2" "$client_addr_id" "$client4_port" + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl ann dead:beef:2::2 token $client6_token id $client_addr_id dev ns2eth1 2>&1 > /dev/null + printf "ADD_ADDR6 id:%d dead:beef:2::2 (ns2) => ns1, reuse port\t\t" $client_addr_id + sleep 0.1 + verify_announce_event "$evts" "$ANNOUNCED" "$server6_token" "dead:beef:2::2" "$client_addr_id" "$client6_port" "v6" + + :>"$evts" + client_addr_id=$((client_addr_id+1)) + ip netns exec $ns2 ./pm_nl_ctl ann 10.0.2.2 token $client4_token id $client_addr_id dev ns2eth1 port $new4_port 2>&1 > /dev/null + printf "ADD_ADDR id:%d 10.0.2.2 (ns2) => ns1, new port \t\t\t" $client_addr_id + sleep 0.1 + verify_announce_event "$evts" "$ANNOUNCED" "$server4_token" "10.0.2.2" "$client_addr_id" "$new4_port" + + kill $evts_pid + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl events >> "$evts" 2>&1 & + evts_pid=$! + sleep 0.1 + + ip netns exec $ns1 ./pm_nl_ctl ann 10.0.2.1 token $server4_token id $server_addr_id dev ns1eth2 2>&1 > /dev/null + printf "ADD_ADDR id:%d 10.0.2.1 (ns1) => ns2, reuse port \t\t" $server_addr_id + sleep 0.1 + verify_announce_event "$evts" "$ANNOUNCED" "$client4_token" "10.0.2.1" "$server_addr_id" "$app4_port" + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl ann dead:beef:2::1 token $server6_token id $server_addr_id dev ns1eth2 2>&1 > /dev/null + printf "ADD_ADDR6 id:%d dead:beef:2::1 (ns1) => ns2, reuse port\t\t" $server_addr_id + sleep 0.1 + verify_announce_event "$evts" "$ANNOUNCED" "$client6_token" "dead:beef:2::1" "$server_addr_id" "$app6_port" "v6" + + :>"$evts" + server_addr_id=$((server_addr_id+1)) + ip netns exec $ns1 ./pm_nl_ctl ann 10.0.2.1 token $server4_token id $server_addr_id dev ns1eth2 port $new4_port 2>&1 > /dev/null + printf "ADD_ADDR id:%d 10.0.2.1 (ns1) => ns2, new port \t\t\t" $server_addr_id + sleep 0.1 + verify_announce_event "$evts" "$ANNOUNCED" "$client4_token" "10.0.2.1" "$server_addr_id" "$new4_port" + + kill $evts_pid + rm -f "$evts" +} + +verify_remove_event() +{ + local evt=$1 + local e_type=$2 + local e_token=$3 + local e_id=$4 + + local type=$(sed -n 's/.*\(type:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local token=$(sed -n 's/.*\(token:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local id=$(sed -n 's/.*\(rem_id:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + if [ "$type" = "$e_type" ] && [ "$token" = "$e_token" ] && [ "$id" = "$e_id" ] + then + printf "[OK]\n" + return 0 + fi + printf "[FAIL]\n" + exit 1 +} + +test_remove() +{ + local evts=$(mktemp) + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl events >> "$evts" 2>&1 & + evts_pid=$! + sleep 0.1 + + local invalid_token=$(( $client4_token - 1 )) + ip netns exec $ns2 ./pm_nl_ctl rem token $invalid_token id $client_addr_id 2>&1 > /dev/null + printf "RM_ADDR id:%d ns2 => ns1, invalid token \t" $client_addr_id + local type=$(sed -n 's/.*\(type:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + if [ "$type" = "" ] + then + printf "[OK]\n" + else + printf "[FAIL]\n" + fi + + local invalid_id=$(( $client_addr_id + 1 )) + ip netns exec $ns2 ./pm_nl_ctl rem token $client4_token id $invalid_id 2>&1 > /dev/null + printf "RM_ADDR id:%d ns2 => ns1, invalid id \t" $invalid_id + type=$(sed -n 's/.*\(type:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + if [ "$type" = "" ] + then + printf "[OK]\n" + else + printf "[FAIL]\n" + fi + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl rem token $client4_token id $client_addr_id 2>&1 > /dev/null + printf "RM_ADDR id:%d ns2 => ns1 \t" $client_addr_id + sleep 0.1 + verify_remove_event "$evts" "$REMOVED" "$server4_token" "$client_addr_id" + + :>"$evts" + client_addr_id=$(( $client_addr_id - 1 )) + ip netns exec $ns2 ./pm_nl_ctl rem token $client4_token id $client_addr_id 2>&1 > /dev/null + printf "RM_ADDR id:%d ns2 => ns1 \t" $client_addr_id + sleep 0.1 + verify_remove_event "$evts" "$REMOVED" "$server4_token" "$client_addr_id" + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl rem token $client6_token id $client_addr_id 2>&1 > /dev/null + printf "RM_ADDR6 id:%d ns2 => ns1 \t" $client_addr_id + sleep 0.1 + verify_remove_event "$evts" "$REMOVED" "$server6_token" "$client_addr_id" + + kill $evts_pid + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl events >> "$evts" 2>&1 & + evts_pid=$! + sleep 0.1 + + ip netns exec $ns1 ./pm_nl_ctl rem token $server4_token id $server_addr_id 2>&1 > /dev/null + printf "RM_ADDR id:%d ns1 => ns2 \t" $server_addr_id + sleep 0.1 + verify_remove_event "$evts" "$REMOVED" "$client4_token" "$server_addr_id" + + :>"$evts" + server_addr_id=$(( $server_addr_id - 1 )) + ip netns exec $ns1 ./pm_nl_ctl rem token $server4_token id $server_addr_id 2>&1 > /dev/null + printf "RM_ADDR id:%d ns1 => ns2 \t" $server_addr_id + sleep 0.1 + verify_remove_event "$evts" "$REMOVED" "$client4_token" "$server_addr_id" + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl rem token $server6_token id $server_addr_id 2>&1 > /dev/null + printf "RM_ADDR6 id:%d ns1 => ns2 \t" $server_addr_id + sleep 0.1 + verify_remove_event "$evts" "$REMOVED" "$client6_token" "$server_addr_id" + + kill $evts_pid + rm -f "$evts" +} + +verify_subflow_events() +{ + local evt=$1 + local e_type=$2 + local e_token=$3 + local e_family=$4 + local e_saddr=$5 + local e_daddr=$6 + local e_dport=$7 + local e_locid=$8 + local e_remid=$9 + shift 2 + local e_from=$8 + local e_to=$9 + + if [ "$e_type" = "$SUB_ESTABLISHED" ] + then + if [ "$e_family" = "$AF_INET6" ] + then + printf "CREATE_SUBFLOW6 %s (%s) => %s (%s) " $e_saddr $e_from $e_daddr $e_to + else + printf "CREATE_SUBFLOW %s (%s) => %s (%s) \t" $e_saddr $e_from $e_daddr $e_to + fi + else + if [ "$e_family" = "$AF_INET6" ] + then + printf "DESTROY_SUBFLOW6 %s (%s) => %s (%s) " $e_saddr $e_from $e_daddr $e_to + else + printf "DESTROY_SUBFLOW %s (%s) => %s (%s) \t" $e_saddr $e_from $e_daddr $e_to + fi + fi + + local type=$(sed -n 's/.*\(type:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local token=$(sed -n 's/.*\(token:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local family=$(sed -n 's/.*\(family:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local dport=$(sed -n 's/.*\(dport:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local locid=$(sed -n 's/.*\(loc_id:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local remid=$(sed -n 's/.*\(rem_id:\)\([[:digit:]]*\).*$/\2/p;q' "$evt") + local saddr="" + local daddr="" + if [ "$family" = "$AF_INET6" ] + then + saddr=$(sed -n 's/.*\(saddr6:\)\([0-9a-f:.]*\).*$/\2/p;q' "$evt") + daddr=$(sed -n 's/.*\(daddr6:\)\([0-9a-f:.]*\).*$/\2/p;q' "$evt") + else + saddr=$(sed -n 's/.*\(saddr4:\)\([0-9.]*\).*$/\2/p;q' "$evt") + daddr=$(sed -n 's/.*\(daddr4:\)\([0-9.]*\).*$/\2/p;q' "$evt") + fi + + if [ "$type" = "$e_type" ] && [ "$token" = "$e_token" ] && [ "$daddr" = "$e_daddr" ] && [ "$e_dport" = "$dport" ] && [ "$family" = "$e_family" ] && [ "$saddr" = "$e_saddr" ] && [ "$e_locid" = "$locid" ] && [ "$e_remid" = "$remid" ] + then + printf "[OK]\n" + return 0 + fi + printf "[FAIL]\n" + exit 1 +} + +test_subflows() +{ + local evts=$(mktemp) + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl events >> "$evts" 2>&1 & + evts_pid=$! + sleep 0.1 + + ip netns exec $ns2 ./pm_nl_ctl ann 10.0.2.2 token $client4_token id $client_addr_id 2>&1 > /dev/null + sleep 0.1 + printf "ADD_ADDR 10.0.2.2 (ns2) => ns1, reuse port \t[OK]\n" + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl csf lip 10.0.2.1 lid 23 rip 10.0.2.2 rport $client4_port token $server4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_ESTABLISHED" "$server4_token" "$AF_INET" "10.0.2.1" "10.0.2.2" "$client4_port" "23" "$client_addr_id" "ns1" "ns2" + + local sport=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl dsf lip 10.0.2.1 lport $sport rip 10.0.2.2 rport $client4_port token $server4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_CLOSED" "$server4_token" "$AF_INET" "10.0.2.1" "10.0.2.2" "$client4_port" "23" "$client_addr_id" "ns1" "ns2" + + ip netns exec $ns2 ./pm_nl_ctl rem id $client_addr_id token $client4_token 2>&1 > /dev/null + sleep 0.1 + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl ann dead:beef:2::2 token $client6_token id $client_addr_id 2>&1 > /dev/null + sleep 0.1 + printf "ADD_ADDR6 dead:beef:2::2 (ns2) => ns1, reuse port \t[OK]\n" + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl csf lip dead:beef:2::1 lid 23 rip dead:beef:2::2 rport $client6_port token $server6_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_ESTABLISHED" "$server6_token" "$AF_INET6" "dead:beef:2::1" "dead:beef:2::2" "$client6_port" "23" "$client_addr_id" "ns1" "ns2" + + local sport=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl dsf lip dead:beef:2::1 lport $sport rip dead:beef:2::2 rport $client6_port token $server6_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_CLOSED" "$server6_token" "$AF_INET6" "dead:beef:2::1" "dead:beef:2::2" "$client6_port" "23" "$client_addr_id" "ns1" "ns2" + + ip netns exec $ns2 ./pm_nl_ctl rem id $client_addr_id token $client6_token 2>&1 > /dev/null + sleep 0.1 + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl ann 10.0.2.2 token $client4_token id $client_addr_id port $new4_port 2>&1 > /dev/null + sleep 0.1 + printf "ADD_ADDR 10.0.2.2 (ns2) => ns1, new port \t[OK]\n" + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl csf lip 10.0.2.1 lid 23 rip 10.0.2.2 rport $new4_port token $server4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_ESTABLISHED" "$server4_token" "$AF_INET" "10.0.2.1" "10.0.2.2" "$new4_port" "23" "$client_addr_id" "ns1" "ns2" + + sport=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl dsf lip 10.0.2.1 lport $sport rip 10.0.2.2 rport $new4_port token $server4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_CLOSED" "$server4_token" "$AF_INET" "10.0.2.1" "10.0.2.2" "$new4_port" "23" "$client_addr_id" "ns1" "ns2" + + ip netns exec $ns2 ./pm_nl_ctl rem id $client_addr_id token $client4_token 2>&1 > /dev/null + + kill $evts_pid + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl events >> "$evts" 2>&1 & + evts_pid=$! + sleep 0.1 + + ip netns exec $ns1 ./pm_nl_ctl ann 10.0.2.1 token $server4_token id $server_addr_id 2>&1 > /dev/null + sleep 0.1 + printf "ADD_ADDR 10.0.2.1 (ns1) => ns2, reuse port \t[OK]\n" + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl csf lip 10.0.2.2 lid 23 rip 10.0.2.1 rport $app4_port token $client4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_ESTABLISHED" "$client4_token" "$AF_INET" "10.0.2.2" "10.0.2.1" "$app4_port" "23" "$server_addr_id" "ns2" "ns1" + + sport=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl dsf lip 10.0.2.2 lport $sport rip 10.0.2.1 rport $app4_port token $client4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_CLOSED" "$client4_token" "$AF_INET" "10.0.2.2" "10.0.2.1" "$app4_port" "23" "$server_addr_id" "ns2" "ns1" + + ip netns exec $ns1 ./pm_nl_ctl rem id $server_addr_id token $server4_token 2>&1 > /dev/null + sleep 0.1 + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl ann dead:beef:2::1 token $server6_token id $server_addr_id 2>&1 > /dev/null + sleep 0.1 + printf "ADD_ADDR6 dead:beef:2::1 (ns1) => ns2, reuse port \t[OK]\n" + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl csf lip dead:beef:2::2 lid 23 rip dead:beef:2::1 rport $app6_port token $client6_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_ESTABLISHED" "$client6_token" "$AF_INET6" "dead:beef:2::2" "dead:beef:2::1" "$app6_port" "23" "$server_addr_id" "ns2" "ns1" + + local sport=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl dsf lip dead:beef:2::2 lport $sport rip dead:beef:2::1 rport $app6_port token $client6_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_CLOSED" "$client6_token" "$AF_INET6" "dead:beef:2::2" "dead:beef:2::1" "$app6_port" "23" "$server_addr_id" "ns2" "ns1" + + ip netns exec $ns1 ./pm_nl_ctl rem id $server_addr_id token $server6_token 2>&1 > /dev/null + sleep 0.1 + + :>"$evts" + ip netns exec $ns1 ./pm_nl_ctl ann 10.0.2.1 token $server4_token id $server_addr_id port $new4_port 2>&1 > /dev/null + sleep 0.1 + printf "ADD_ADDR 10.0.2.1 (ns1) => ns2, new port \t[OK]\n" + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl csf lip 10.0.2.2 lid 23 rip 10.0.2.1 rport $new4_port token $client4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_ESTABLISHED" "$client4_token" "$AF_INET" "10.0.2.2" "10.0.2.1" "$new4_port" "23" "$server_addr_id" "ns2" "ns1" + + sport=$(sed -n 's/.*\(sport:\)\([[:digit:]]*\).*$/\2/p;q' "$evts") + + :>"$evts" + ip netns exec $ns2 ./pm_nl_ctl dsf lip 10.0.2.2 lport $sport rip 10.0.2.1 rport $new4_port token $client4_token 2>&1 > /dev/null + sleep 0.1 + verify_subflow_events "$evts" "$SUB_CLOSED" "$client4_token" "$AF_INET" "10.0.2.2" "10.0.2.1" "$new4_port" "23" "$server_addr_id" "ns2" "ns1" + + ip netns exec $ns1 ./pm_nl_ctl rem id $server_addr_id token $server4_token 2>&1 > /dev/null + + kill $evts_pid + rm -f "$evts" +} + +make_connection +make_connection "v6" +test_announce +test_remove +test_subflows +exit 0
Signature · 1 line
2.31.1
Re: selftests: mptcp: functional tests for the userspace PM type: Build Failure
Hi Kishen,
Thank you for your modifications, that's great!
But sadly, our CI spotted some issues with it when trying to build it.
You can find more details there:
https://patchwork.kernel.org/project/mptcp/patch/[email protected]/
https://github.com/multipath-tcp/mptcp_net-next/actions/runs/1689981079
Status: failure
Initiator: MPTCPimporter
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/bce46a55a41c
Feel free to reply to this email if you cannot access logs, if you need some support to fix the error, if this doesn't seem to be caused by your modifications or if the error is a false positive one.
Cheers,
MPTCP GH Action bot