Skip to contentSkip to conversation

net/smc: Introduce BPF injection capability

PATCH Superseded D. Wythe 4 participants 12 messages Feb 28, 2023 – Mar 1, 2023

Subsystems (8)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

D. Wythe

author maintainer
#0 lore raw

v3 cover net/smc: Introduce BPF injection capability

From: "D. Wythe" <[email protected]>

This patches attempt to introduce BPF injection capability for SMC, and add selftest to ensure code stability.

As we all know that the SMC protocol is not suitable for all scenarios, especially for short-lived. However, for most applications, they cannot guarantee that there are no such scenarios at all. Therefore, apps may need some specific strategies to decide shall we need to use SMC or not, for example, apps can limit the scope of the SMC to a specific IP address or port.

Based on the consideration of transparent replacement, we hope that apps can remain transparent even if they need to formulate some specific strategies for SMC using. That is, do not need to recompile their code.

On the other hand, we need to ensure the scalability of strategies implementation. Although it is simple to use socket options or sysctl, it will bring more complexity to subsequent expansion.

Fortunately, BPF can solve these concerns very well, users can write thire own strategies in eBPF to choose whether to use SMC or not. And it's quite easy for them to modify their strategies in the future.

This patches implement injection capability for SMC via struct_ops. In that way, we can add new injection scenarios in the future.

v3 -> v2:
1. fix checkpatch error and warning.
2. split patch for better review.
3. enhance selftest to cover more scenarios.

v2 -> v1:
1. fix compile error and warning.

D. Wythe (4):
net/smc: move smc_sock related structure definition
bpf: add SMC support in BPF struct_ops net/smc: add BPF injection on smc negotiation bpf/selftests: add selftest for SMC bpf capability

 include/linux/btf_ids.h                          |  12 +
 include/net/smc.h                                | 248 ++++++++++++++++++
 kernel/bpf/bpf_struct_ops_types.h                |   4 +
 net/Makefile                                     |   5 +
 net/smc/af_smc.c                                 |  15 +-
 net/smc/bpf_smc_struct_ops.c                     | 148 +++++++++++
 net/smc/smc.h                                    | 224 ----------------
 tools/testing/selftests/bpf/prog_tests/bpf_smc.c |  37 +++
 tools/testing/selftests/bpf/progs/bpf_smc.c      | 320 +++++++++++++++++++++++
 9 files changed, 788 insertions(+), 225 deletions(-)
 create mode 100644 net/smc/bpf_smc_struct_ops.c
 create mode 100644 tools/testing/selftests/bpf/prog_tests/bpf_smc.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_smc.c
Signature · 1 line
1.8.3.1
author maintainer
#1 lore raw

PATCH v3 1/4 net/smc: move smc_sock related structure definition

From: "D. Wythe" <[email protected]>

This patch only try to move the definition of smc_sock and its related structure, from et/smc/smc.h to include/net/smc/smc.h. In that way can ebpf generate the BTF ID corresponding to our structure.

Of course, we can also choose to hide the structure and only to expose an intermediate structure, but it requires an additional transformation. If we need to obtain some information frequently, this may cause some performance problems.

---
 include/net/smc.h | 225 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 net/smc/smc.h     | 224 -----------------------------------------------------
 2 files changed, 225 insertions(+), 224 deletions(-)
include/net/smc.h +225 -0
index 597cb93..eccbd37 100644
@@ -11,12 +11,17 @@
 #ifndef _SMC_H
 #define _SMC_H
 
+#include <net/inet_connection_sock.h>
 #include <linux/device.h>
 #include <linux/spinlock.h>
 #include <linux/types.h>
 #include <linux/wait.h>
 #include "linux/ism.h"
 
+#ifdef ATOMIC64_INIT
+#define KERNEL_HAS_ATOMIC64
+#endif
+
 struct sock;
 
 #define SMC_MAX_PNETID_LEN	16	/* Max. length of PNET id */
@@ -90,4 +95,224 @@ struct smcd_dev {
 	u8 going_away : 1;
 };
 
+struct smc_wr_rx_hdr {	/* common prefix part of LLC and CDC to demultiplex */
+	union {
+		u8 type;
+#if defined(__BIG_ENDIAN_BITFIELD)
+		struct {
+			u8 llc_version:4,
+			   llc_type:4;
+		};
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+		struct {
+			u8 llc_type:4,
+			   llc_version:4;
+		};
+#endif
+	};
+} __aligned(1);
+
+struct smc_cdc_conn_state_flags {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u8	peer_done_writing : 1;	/* Sending done indicator */
+	u8	peer_conn_closed : 1;	/* Peer connection closed indicator */
+	u8	peer_conn_abort : 1;	/* Abnormal close indicator */
+	u8	reserved : 5;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u8	reserved : 5;
+	u8	peer_conn_abort : 1;
+	u8	peer_conn_closed : 1;
+	u8	peer_done_writing : 1;
+#endif
+};
+
+struct smc_cdc_producer_flags {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u8	write_blocked : 1;	/* Writing Blocked, no rx buf space */
+	u8	urg_data_pending : 1;	/* Urgent Data Pending */
+	u8	urg_data_present : 1;	/* Urgent Data Present */
+	u8	cons_curs_upd_req : 1;	/* cursor update requested */
+	u8	failover_validation : 1;/* message replay due to failover */
+	u8	reserved : 3;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u8	reserved : 3;
+	u8	failover_validation : 1;
+	u8	cons_curs_upd_req : 1;
+	u8	urg_data_present : 1;
+	u8	urg_data_pending : 1;
+	u8	write_blocked : 1;
+#endif
+};
+
+/* in host byte order */
+union smc_host_cursor {	/* SMC cursor - an offset in an RMBE */
+	struct {
+		u16	reserved;
+		u16	wrap;		/* window wrap sequence number */
+		u32	count;		/* cursor (= offset) part */
+	};
+#ifdef KERNEL_HAS_ATOMIC64
+	atomic64_t		acurs;	/* for atomic processing */
+#else
+	u64			acurs;	/* for atomic processing */
+#endif
+} __aligned(8);
+
+/* in host byte order, except for flag bitfields in network byte order */
+struct smc_host_cdc_msg {		/* Connection Data Control message */
+	struct smc_wr_rx_hdr		common; /* .type = 0xFE */
+	u8				len;	/* length = 44 */
+	u16				seqno;	/* connection seq # */
+	u32				token;	/* alert_token */
+	union smc_host_cursor		prod;		/* producer cursor */
+	union smc_host_cursor		cons;		/* consumer cursor,
+							 * piggy backed "ack"
+							 */
+	struct smc_cdc_producer_flags	prod_flags;	/* conn. tx/rx status */
+	struct smc_cdc_conn_state_flags	conn_state_flags; /* peer conn. status*/
+	u8				reserved[18];
+} __aligned(8);
+
+enum smc_urg_state {
+	SMC_URG_VALID	= 1,			/* data present */
+	SMC_URG_NOTYET	= 2,			/* data pending */
+	SMC_URG_READ	= 3,			/* data was already read */
+};
+
+struct smc_connection {
+	struct rb_node		alert_node;
+	struct smc_link_group	*lgr;		/* link group of connection */
+	struct smc_link		*lnk;		/* assigned SMC-R link */
+	u32			alert_token_local; /* unique conn. id */
+	u8			peer_rmbe_idx;	/* from tcp handshake */
+	int			peer_rmbe_size;	/* size of peer rx buffer */
+	atomic_t		peer_rmbe_space;/* remaining free bytes in peer
+						 * rmbe
+						 */
+	int			rtoken_idx;	/* idx to peer RMB rkey/addr */
+
+	struct smc_buf_desc	*sndbuf_desc;	/* send buffer descriptor */
+	struct smc_buf_desc	*rmb_desc;	/* RMBE descriptor */
+	int			rmbe_size_short;/* compressed notation */
+	int			rmbe_update_limit;
+						/* lower limit for consumer
+						 * cursor update
+						 */
+
+	struct smc_host_cdc_msg	local_tx_ctrl;	/* host byte order staging
+						 * buffer for CDC msg send
+						 * .prod cf. TCP snd_nxt
+						 * .cons cf. TCP sends ack
+						 */
+	union smc_host_cursor	local_tx_ctrl_fin;
+						/* prod crsr - confirmed by peer
+						 */
+	union smc_host_cursor	tx_curs_prep;	/* tx - prepared data
+						 * snd_max..wmem_alloc
+						 */
+	union smc_host_cursor	tx_curs_sent;	/* tx - sent data
+						 * snd_nxt ?
+						 */
+	union smc_host_cursor	tx_curs_fin;	/* tx - confirmed by peer
+						 * snd-wnd-begin ?
+						 */
+	atomic_t		sndbuf_space;	/* remaining space in sndbuf */
+	u16			tx_cdc_seq;	/* sequence # for CDC send */
+	u16			tx_cdc_seq_fin;	/* sequence # - tx completed */
+	spinlock_t		send_lock;	/* protect wr_sends */
+	atomic_t		cdc_pend_tx_wr; /* number of pending tx CDC wqe
+						 * - inc when post wqe,
+						 * - dec on polled tx cqe
+						 */
+	wait_queue_head_t	cdc_pend_tx_wq; /* wakeup on no cdc_pend_tx_wr*/
+	atomic_t		tx_pushing;     /* nr_threads trying tx push */
+	struct delayed_work	tx_work;	/* retry of smc_cdc_msg_send */
+	u32			tx_off;		/* base offset in peer rmb */
+
+	struct smc_host_cdc_msg	local_rx_ctrl;	/* filled during event_handl.
+						 * .prod cf. TCP rcv_nxt
+						 * .cons cf. TCP snd_una
+						 */
+	union smc_host_cursor	rx_curs_confirmed; /* confirmed to peer
+						    * source of snd_una ?
+						    */
+	union smc_host_cursor	urg_curs;	/* points at urgent byte */
+	enum smc_urg_state	urg_state;
+	bool			urg_tx_pend;	/* urgent data staged */
+	bool			urg_rx_skip_pend;
+						/* indicate urgent oob data
+						 * read, but previous regular
+						 * data still pending
+						 */
+	char			urg_rx_byte;	/* urgent byte */
+	bool			tx_in_release_sock;
+						/* flush pending tx data in
+						 * sock release_cb()
+						 */
+	atomic_t		bytes_to_rcv;	/* arrived data,
+						 * not yet received
+						 */
+	atomic_t		splice_pending;	/* number of spliced bytes
+						 * pending processing
+						 */
+#ifndef KERNEL_HAS_ATOMIC64
+	spinlock_t		acurs_lock;	/* protect cursors */
+#endif
+	struct work_struct	close_work;	/* peer sent some closing */
+	struct work_struct	abort_work;	/* abort the connection */
+	struct tasklet_struct	rx_tsklet;	/* Receiver tasklet for SMC-D */
+	u8			rx_off;		/* receive offset:
+						 * 0 for SMC-R, 32 for SMC-D
+						 */
+	u64			peer_token;	/* SMC-D token of peer */
+	u8			killed : 1;	/* abnormal termination */
+	u8			freed : 1;	/* normal termiation */
+	u8			out_of_sync : 1; /* out of sync with peer */
+};
+
+struct smc_sock {				/* smc sock container */
+	struct sock		sk;
+	struct socket		*clcsock;	/* internal tcp socket */
+	void			(*clcsk_state_change)(struct sock *sk);
+						/* original stat_change fct. */
+	void			(*clcsk_data_ready)(struct sock *sk);
+						/* original data_ready fct. */
+	void			(*clcsk_write_space)(struct sock *sk);
+						/* original write_space fct. */
+	void			(*clcsk_error_report)(struct sock *sk);
+						/* original error_report fct. */
+	struct smc_connection	conn;		/* smc connection */
+	struct smc_sock		*listen_smc;	/* listen parent */
+	struct work_struct	connect_work;	/* handle non-blocking connect*/
+	struct work_struct	tcp_listen_work;/* handle tcp socket accepts */
+	struct work_struct	smc_listen_work;/* prepare new accept socket */
+	struct list_head	accept_q;	/* sockets to be accepted */
+	spinlock_t		accept_q_lock;	/* protects accept_q */
+	bool			limit_smc_hs;	/* put constraint on handshake */
+	bool			use_fallback;	/* fallback to tcp */
+	int			fallback_rsn;	/* reason for fallback */
+	u32			peer_diagnosis; /* decline reason from peer */
+	atomic_t                queued_smc_hs;  /* queued smc handshakes */
+	struct inet_connection_sock_af_ops		af_ops;
+	const struct inet_connection_sock_af_ops	*ori_af_ops;
+						/* original af ops */
+	int			sockopt_defer_accept;
+						/* sockopt TCP_DEFER_ACCEPT
+						 * value
+						 */
+	u8			wait_close_tx_prepared : 1;
+						/* shutdown wr or close
+						 * started, waiting for unsent
+						 * data to be sent
+						 */
+	u8			connect_nonblock : 1;
+						/* non-blocking connect in
+						 * flight
+						 */
+	struct mutex            clcsock_release_lock;
+						/* protects clcsock of a listen
+						 * socket
+						 */
+};
+
 #endif	/* _SMC_H */
net/smc/smc.h +0 -224

242 lines not on this page · raw message

Signature · 1 line
1.8.3.1
author maintainer
#2 lore raw

PATCH v3 2/4 bpf: add SMC support in BPF struct_ops

In reply to D. Wythe

From: "D. Wythe" <[email protected]>

This PATCH attempts to introduce BPF injection capability for SMC. Considering that the SMC protocol is not suitable for all scenarios, especially for short-lived. However, for most applications, they cannot guarantee that there are no such scenarios at all. Therefore, apps may need some specific strategies to decide shall we need to use SMC or not, for example, apps can limit the scope of the SMC to a specific IP address or port.

Based on the consideration of transparent replacement, we hope that apps can remain transparent even if they need to formulate some specific strategies for SMC using. That is, do not need to recompile their code.

On the other hand, we need to ensure the scalability of strategies implementation. Although it is simple to use socket options or sysctl, it will bring more complexity to subsequent expansion.

Fortunately, BPF can solve these concerns very well, users can write thire own strategies in eBPF to choose whether to use SMC or not. And it's quite easy for them to modify their strategies in the future.

This PATCH implement injection capability for SMC via struct_ops. In that way, we can add new injection scenarios in the future.

---
 include/linux/btf_ids.h           |  12 ++++
 include/net/smc.h                 |  23 ++++++
 kernel/bpf/bpf_struct_ops_types.h |   4 ++
 net/Makefile                      |   5 ++
 net/smc/bpf_smc_struct_ops.c      | 148 ++++++++++++++++++++++++++++++++++++++
 5 files changed, 192 insertions(+)
 create mode 100644 net/smc/bpf_smc_struct_ops.c
include/linux/btf_ids.h +12 -0
index 3a4f7cd..d4cf0a9 100644
@@ -264,6 +264,18 @@ enum {
 MAX_BTF_TRACING_TYPE,
 };
 
+#if IS_ENABLED(CONFIG_SMC)
+enum {
+#define BTF_SMC_TYPE(name, type) name,
+BTF_SMC_TYPE(BTF_SMC_TYPE_SOCK, smc_sock)
+BTF_SMC_TYPE(BTF_SMC_TYPE_CONNECTION, smc_connection)
+BTF_SMC_TYPE(BTF_SMC_TYPE_HOST_CURSOR, smc_host_cursor)
+#undef BTF_SMC_TYPE
+MAX_BTF_SMC_TYPE,
+};
+extern u32 btf_smc_ids[];
+#endif
+
 extern u32 btf_tracing_ids[];
 extern u32 bpf_cgroup_btf_id[];
 extern u32 bpf_local_storage_map_btf_id[];
include/net/smc.h +23 -0
index eccbd37..c612fec 100644
@@ -16,6 +16,7 @@
 #include <linux/spinlock.h>
 #include <linux/types.h>
 #include <linux/wait.h>
+#include <linux/bpf.h>
 #include "linux/ism.h"
 
 #ifdef ATOMIC64_INIT
@@ -315,4 +316,26 @@ struct smc_sock {				/* smc sock container */
 						 */
 };
 
+#define SMC_SOCK_CLOSED_TIMING	(0)
+
+/* BPF struct ops for smc protocol negotiator */
+struct smc_sock_negotiator_ops {
+	/* ret for negotiate */
+	int (*negotiate)(struct smc_sock *sk);
+
+	/* info gathering timing */
+	void (*collect_info)(struct sock *sk, int timing);
+};
+
+/* Query if current sock should go with SMC protocol
+ * SK_PASS for yes, otherwise for no.
+ */
+int smc_sock_should_select_smc(const struct smc_sock *smc);
+
+/* At some specific points in time,
+ * let negotiator can perform info gathering
+ * on target sock.
+ */
+void smc_sock_perform_collecting_info(const struct sock *sk, int timing);
+
 #endif	/* _SMC_H */
kernel/bpf/bpf_struct_ops_types.h +4 -0
index 5678a9d..35cdd15 100644
@@ -9,4 +9,8 @@
 #include <net/tcp.h>
 BPF_STRUCT_OPS_TYPE(tcp_congestion_ops)
 #endif
+#if IS_ENABLED(CONFIG_SMC)
+#include <net/smc.h>
+BPF_STRUCT_OPS_TYPE(smc_sock_negotiator_ops)
+#endif
 #endif
net/Makefile +5 -0
index 0914bea..47a4c00 100644
@@ -52,6 +52,11 @@ obj-$(CONFIG_TIPC)		+= tipc/
 obj-$(CONFIG_NETLABEL)		+= netlabel/
 obj-$(CONFIG_IUCV)		+= iucv/
 obj-$(CONFIG_SMC)		+= smc/
+ifneq ($(CONFIG_SMC),)
+ifeq ($(CONFIG_BPF_SYSCALL),y)
+obj-y				+= smc/bpf_smc_struct_ops.o
+endif
+endif
 obj-$(CONFIG_RFKILL)		+= rfkill/
 obj-$(CONFIG_NET_9P)		+= 9p/
 obj-$(CONFIG_CAIF)		+= caif/
net/smc/bpf_smc_struct_ops.c new file +148 -0

148 lines not on this page · raw message

Signature · 1 line
1.8.3.1
author maintainer
#3 lore raw

PATCH v3 3/4 net/smc: add BPF injection on smc negotiation

In reply to D. Wythe

From: "D. Wythe" <[email protected]>

This patch try add BPF injection on smc negotiation, so that the application can decided whether to use smc or not through eBPF progs.

In particular, some applications may need global dynamic information to make decision. Therefore, we also inject a information collect point into smc_release.

Note that, in order to make negotiation can be decided by application, sockets must have SMC_LIMIT_HS set.

---
 net/smc/af_smc.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)
net/smc/af_smc.c +14 -1
index a4cccdf..7ebe5e8 100644
@@ -166,6 +166,9 @@ static bool smc_hs_congested(const struct sock *sk)
 	if (workqueue_congested(WORK_CPU_UNBOUND, smc_hs_wq))
 		return true;
 
+	if (!smc_sock_should_select_smc(smc))
+		return true;
+
 	return false;
 }
 
@@ -320,6 +323,9 @@ static int smc_release(struct socket *sock)
 	sock_hold(sk); /* sock_put below */
 	smc = smc_sk(sk);
 
+	/* trigger info gathering if needed.*/
+	smc_sock_perform_collecting_info(sk, SMC_SOCK_CLOSED_TIMING);
+
 	old_state = sk->sk_state;
 
 	/* cleanup for a dangling non-blocking connect */
@@ -1627,7 +1633,14 @@ static int smc_connect(struct socket *sock, struct sockaddr *addr,
 	}
 
 	smc_copy_sock_settings_to_clc(smc);
-	tcp_sk(smc->clcsock->sk)->syn_smc = 1;
+	/* accept out connection as SMC connection */
+	if (smc_sock_should_select_smc(smc) == SK_PASS) {
+		tcp_sk(smc->clcsock->sk)->syn_smc = 1;
+	} else {
+		tcp_sk(smc->clcsock->sk)->syn_smc = 0;
+		smc_switch_to_fallback(smc, /* just a chooice */ 0);
+	}
+
 	if (smc->connect_nonblock) {
 		rc = -EALREADY;
 		goto out;
Signature · 1 line
1.8.3.1
#4 lore raw

Re: [PATCH bpf-next v3 3/4] net/smc: add BPF injection on smc negotiation

Thank you for the patch! Yet something to improve:

Hi Wythe,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/D-Wythe/net-smc-move-smc_sock-related-structure-definition/20230228-173007
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/1677576294-33411-4-git-send-email-alibuda%40linux.alibaba.com
patch subject: [PATCH bpf-next v3 3/4] net/smc: add BPF injection on smc negotiation
config: x86_64-randconfig-a015-20230227 (https://download.01.org/0day-ci/archive/20230228/[email protected]/config)
compiler: gcc-11 (Debian 11.3.0-8) 11.3.0
reproduce (this is a W=1 build):
        # https://github.com/intel-lab-lkp/linux/commit/aa482ab82f4bf9b99d490f8ba5d88e1491156ccf
        git remote add linux-review https://github.com/intel-lab-lkp/linux
        git fetch --no-tags linux-review D-Wythe/net-smc-move-smc_sock-related-structure-definition/20230228-173007
        git checkout aa482ab82f4bf9b99d490f8ba5d88e1491156ccf
        # save the config file
        mkdir build_dir && cp config build_dir/.config
        make W=1 O=build_dir ARCH=x86_64 olddefconfig
        make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash

If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <[email protected]>
| Link: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

ld: net/smc/af_smc.o: in function `smc_hs_congested':

>> net/smc/af_smc.c:169: undefined reference to `smc_sock_should_select_smc'
   ld: net/smc/af_smc.o: in function `smc_release':
>> net/smc/af_smc.c:327: undefined reference to `smc_sock_perform_collecting_info'

ld: net/smc/af_smc.o: in function `smc_connect':
net/smc/af_smc.c:1637: undefined reference to `smc_sock_should_select_smc'

vim +169 net/smc/af_smc.c

156
157 static bool smc_hs_congested(const struct sock *sk)
158 {
159 const struct smc_sock *smc;
160
161 smc = smc_clcsock_user_data(sk);
162
163 if (!smc)
164 return true;
165
166 if (workqueue_congested(WORK_CPU_UNBOUND, smc_hs_wq))
167 return true;
168

 > 169		if (!smc_sock_should_select_smc(smc))
   170			return true;
   171	
   172		return false;
   173	}
   174	
   175	static struct smc_hashinfo smc_v4_hashinfo = {
   176		.lock = __RW_LOCK_UNLOCKED(smc_v4_hashinfo.lock),
   177	};
   178	
   179	static struct smc_hashinfo smc_v6_hashinfo = {
   180		.lock = __RW_LOCK_UNLOCKED(smc_v6_hashinfo.lock),
   181	};
   182	
   183	int smc_hash_sk(struct sock *sk)
   184	{
   185		struct smc_hashinfo *h = sk->sk_prot->h.smc_hash;
   186		struct hlist_head *head;
   187	
   188		head = &h->ht;
   189	
   190		write_lock_bh(&h->lock);
   191		sk_add_node(sk, head);
   192		write_unlock_bh(&h->lock);
   193		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
   194	
   195		return 0;
   196	}
   197	EXPORT_SYMBOL_GPL(smc_hash_sk);
   198	
   199	void smc_unhash_sk(struct sock *sk)
   200	{
   201		struct smc_hashinfo *h = sk->sk_prot->h.smc_hash;
   202	
   203		write_lock_bh(&h->lock);
   204		if (sk_del_node_init(sk))
   205			sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
   206		write_unlock_bh(&h->lock);
   207	}
   208	EXPORT_SYMBOL_GPL(smc_unhash_sk);
   209	
   210	/* This will be called before user really release sock_lock. So do the
   211	 * work which we didn't do because of user hold the sock_lock in the
   212	 * BH context
   213	 */
   214	static void smc_release_cb(struct sock *sk)
   215	{
   216		struct smc_sock *smc = smc_sk(sk);
   217	
   218		if (smc->conn.tx_in_release_sock) {
   219			smc_tx_pending(&smc->conn);
   220			smc->conn.tx_in_release_sock = false;
   221		}
   222	}
   223	
   224	struct proto smc_proto = {
   225		.name		= "SMC",
   226		.owner		= THIS_MODULE,
   227		.keepalive	= smc_set_keepalive,
   228		.hash		= smc_hash_sk,
   229		.unhash		= smc_unhash_sk,
   230		.release_cb	= smc_release_cb,
   231		.obj_size	= sizeof(struct smc_sock),
   232		.h.smc_hash	= &smc_v4_hashinfo,
   233		.slab_flags	= SLAB_TYPESAFE_BY_RCU,
   234	};
   235	EXPORT_SYMBOL_GPL(smc_proto);
   236	
   237	struct proto smc_proto6 = {
   238		.name		= "SMC6",
   239		.owner		= THIS_MODULE,
   240		.keepalive	= smc_set_keepalive,
   241		.hash		= smc_hash_sk,
   242		.unhash		= smc_unhash_sk,
   243		.release_cb	= smc_release_cb,
   244		.obj_size	= sizeof(struct smc_sock),
   245		.h.smc_hash	= &smc_v6_hashinfo,
   246		.slab_flags	= SLAB_TYPESAFE_BY_RCU,
   247	};
   248	EXPORT_SYMBOL_GPL(smc_proto6);
   249	
   250	static void smc_fback_restore_callbacks(struct smc_sock *smc)
   251	{
   252		struct sock *clcsk = smc->clcsock->sk;
   253	
   254		write_lock_bh(&clcsk->sk_callback_lock);
   255		clcsk->sk_user_data = NULL;
   256	
   257		smc_clcsock_restore_cb(&clcsk->sk_state_change, &smc->clcsk_state_change);
   258		smc_clcsock_restore_cb(&clcsk->sk_data_ready, &smc->clcsk_data_ready);
   259		smc_clcsock_restore_cb(&clcsk->sk_write_space, &smc->clcsk_write_space);
   260		smc_clcsock_restore_cb(&clcsk->sk_error_report, &smc->clcsk_error_report);
   261	
   262		write_unlock_bh(&clcsk->sk_callback_lock);
   263	}
   264	
   265	static void smc_restore_fallback_changes(struct smc_sock *smc)
   266	{
   267		if (smc->clcsock->file) { /* non-accepted sockets have no file yet */
   268			smc->clcsock->file->private_data = smc->sk.sk_socket;
   269			smc->clcsock->file = NULL;
   270			smc_fback_restore_callbacks(smc);
   271		}
   272	}
   273	
   274	static int __smc_release(struct smc_sock *smc)
   275	{
   276		struct sock *sk = &smc->sk;
   277		int rc = 0;
   278	
   279		if (!smc->use_fallback) {
   280			rc = smc_close_active(smc);
   281			sock_set_flag(sk, SOCK_DEAD);
   282			sk->sk_shutdown |= SHUTDOWN_MASK;
   283		} else {
   284			if (sk->sk_state != SMC_CLOSED) {
   285				if (sk->sk_state != SMC_LISTEN &&
   286				    sk->sk_state != SMC_INIT)
   287					sock_put(sk); /* passive closing */
   288				if (sk->sk_state == SMC_LISTEN) {
   289					/* wake up clcsock accept */
   290					rc = kernel_sock_shutdown(smc->clcsock,
   291								  SHUT_RDWR);
   292				}
   293				sk->sk_state = SMC_CLOSED;
   294				sk->sk_state_change(sk);
   295			}
   296			smc_restore_fallback_changes(smc);
   297		}
   298	
   299		sk->sk_prot->unhash(sk);
   300	
   301		if (sk->sk_state == SMC_CLOSED) {
   302			if (smc->clcsock) {
   303				release_sock(sk);
   304				smc_clcsock_release(smc);
   305				lock_sock(sk);
   306			}
   307			if (!smc->use_fallback)
   308				smc_conn_free(&smc->conn);
   309		}
   310	
   311		return rc;
   312	}
   313	
   314	static int smc_release(struct socket *sock)
   315	{
   316		struct sock *sk = sock->sk;
   317		struct smc_sock *smc;
   318		int old_state, rc = 0;
   319	
   320		if (!sk)
   321			goto out;
   322	
   323		sock_hold(sk); /* sock_put below */
   324		smc = smc_sk(sk);
   325	
   326		/* trigger info gathering if needed.*/
 > 327		smc_sock_perform_collecting_info(sk, SMC_SOCK_CLOSED_TIMING);

328
329 old_state = sk->sk_state;
330
331 /* cleanup for a dangling non-blocking connect */
332 if (smc->connect_nonblock && old_state == SMC_INIT)
333 tcp_abort(smc->clcsock->sk, ECONNABORTED);
334
335 if (cancel_work_sync(&smc->connect_work))
336 sock_put(&smc->sk); /* sock_hold in smc_connect for passive closing */
337
338 if (sk->sk_state == SMC_LISTEN)
339 /* smc_close_non_accepted() is called and acquires
340 * sock lock for child sockets again
341 */
342 lock_sock_nested(sk, SINGLE_DEPTH_NESTING);
343 else
344 lock_sock(sk);
345
346 if (old_state == SMC_INIT && sk->sk_state == SMC_ACTIVE &&
347 !smc->use_fallback)
348 smc_close_active_abort(smc);
349
350 rc = __smc_release(smc);
351
352 /* detach socket */
353 sock_orphan(sk);
354 sock->sk = NULL;
355 release_sock(sk);
356
357 sock_put(sk); /* sock_hold above */
358 sock_put(sk); /* final sock_put */
359 out:
360 return rc;
361 }
362

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests
author maintainer
#5 lore raw

Re: [PATCH bpf-next v3 3/4] net/smc: add BPF injection on smc negotiation

Hi robot,

It's seems impossible. I have compiled and tested this patch locally for many times. Maybe you compiled this patch separately. However, this patch depends on the previous one.

Thanks
D. Wythe

Quoted 246 lines · kernel test robot

Hi Wythe,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/D-Wythe/net-smc-move-smc_sock-related-structure-definition/20230228-173007
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/1677576294-33411-4-git-send-email-alibuda%40linux.alibaba.com
patch subject: [PATCH bpf-next v3 3/4] net/smc: add BPF injection on smc negotiation
config: x86_64-randconfig-a015-20230227 (https://download.01.org/0day-ci/archive/20230228/[email protected]/config)
compiler: gcc-11 (Debian 11.3.0-8) 11.3.0
reproduce (this is a W=1 build):
         # https://github.com/intel-lab-lkp/linux/commit/aa482ab82f4bf9b99d490f8ba5d88e1491156ccf
         git remote add linux-review https://github.com/intel-lab-lkp/linux
         git fetch --no-tags linux-review D-Wythe/net-smc-move-smc_sock-related-structure-definition/20230228-173007
         git checkout aa482ab82f4bf9b99d490f8ba5d88e1491156ccf
         # save the config file
         mkdir build_dir && cp config build_dir/.config
         make W=1 O=build_dir ARCH=x86_64 olddefconfig
         make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash

If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <[email protected]>
| Link: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

ld: net/smc/af_smc.o: in function `smc_hs_congested':

Quoted 1 line
Quoted 1 line

net/smc/af_smc.c:169: undefined reference to `smc_sock_should_select_smc'

ld: net/smc/af_smc.o: in function `smc_release':

Quoted 1 line
Quoted 1 line

net/smc/af_smc.c:327: undefined reference to `smc_sock_perform_collecting_info'

ld: net/smc/af_smc.o: in function `smc_connect':
net/smc/af_smc.c:1637: undefined reference to `smc_sock_should_select_smc'

vim +169 net/smc/af_smc.c

    156	
    157	static bool smc_hs_congested(const struct sock *sk)
    158	{
    159		const struct smc_sock *smc;
    160	
    161		smc = smc_clcsock_user_data(sk);
    162	
    163		if (!smc)
    164			return true;
    165	
    166		if (workqueue_congested(WORK_CPU_UNBOUND, smc_hs_wq))
    167			return true;
    168	
Quoted 1 line

169 if (!smc_sock_should_select_smc(smc))

    170			return true;
    171	
    172		return false;
    173	}
    174	
    175	static struct smc_hashinfo smc_v4_hashinfo = {
    176		.lock = __RW_LOCK_UNLOCKED(smc_v4_hashinfo.lock),
    177	};
    178	
    179	static struct smc_hashinfo smc_v6_hashinfo = {
    180		.lock = __RW_LOCK_UNLOCKED(smc_v6_hashinfo.lock),
    181	};
    182	
    183	int smc_hash_sk(struct sock *sk)
    184	{
    185		struct smc_hashinfo *h = sk->sk_prot->h.smc_hash;
    186		struct hlist_head *head;
    187	
    188		head = &h->ht;
    189	
    190		write_lock_bh(&h->lock);
    191		sk_add_node(sk, head);
    192		write_unlock_bh(&h->lock);
    193		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
    194	
    195		return 0;
    196	}
    197	EXPORT_SYMBOL_GPL(smc_hash_sk);
    198	
    199	void smc_unhash_sk(struct sock *sk)
    200	{
    201		struct smc_hashinfo *h = sk->sk_prot->h.smc_hash;
    202	
    203		write_lock_bh(&h->lock);
    204		if (sk_del_node_init(sk))
    205			sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
    206		write_unlock_bh(&h->lock);
    207	}
    208	EXPORT_SYMBOL_GPL(smc_unhash_sk);
    209	
    210	/* This will be called before user really release sock_lock. So do the
    211	 * work which we didn't do because of user hold the sock_lock in the
    212	 * BH context
    213	 */
    214	static void smc_release_cb(struct sock *sk)
    215	{
    216		struct smc_sock *smc = smc_sk(sk);
    217	
    218		if (smc->conn.tx_in_release_sock) {
    219			smc_tx_pending(&smc->conn);
    220			smc->conn.tx_in_release_sock = false;
    221		}
    222	}
    223	
    224	struct proto smc_proto = {
    225		.name		= "SMC",
    226		.owner		= THIS_MODULE,
    227		.keepalive	= smc_set_keepalive,
    228		.hash		= smc_hash_sk,
    229		.unhash		= smc_unhash_sk,
    230		.release_cb	= smc_release_cb,
    231		.obj_size	= sizeof(struct smc_sock),
    232		.h.smc_hash	= &smc_v4_hashinfo,
    233		.slab_flags	= SLAB_TYPESAFE_BY_RCU,
    234	};
    235	EXPORT_SYMBOL_GPL(smc_proto);
    236	
    237	struct proto smc_proto6 = {
    238		.name		= "SMC6",
    239		.owner		= THIS_MODULE,
    240		.keepalive	= smc_set_keepalive,
    241		.hash		= smc_hash_sk,
    242		.unhash		= smc_unhash_sk,
    243		.release_cb	= smc_release_cb,
    244		.obj_size	= sizeof(struct smc_sock),
    245		.h.smc_hash	= &smc_v6_hashinfo,
    246		.slab_flags	= SLAB_TYPESAFE_BY_RCU,
    247	};
    248	EXPORT_SYMBOL_GPL(smc_proto6);
    249	
    250	static void smc_fback_restore_callbacks(struct smc_sock *smc)
    251	{
    252		struct sock *clcsk = smc->clcsock->sk;
    253	
    254		write_lock_bh(&clcsk->sk_callback_lock);
    255		clcsk->sk_user_data = NULL;
    256	
    257		smc_clcsock_restore_cb(&clcsk->sk_state_change, &smc->clcsk_state_change);
    258		smc_clcsock_restore_cb(&clcsk->sk_data_ready, &smc->clcsk_data_ready);
    259		smc_clcsock_restore_cb(&clcsk->sk_write_space, &smc->clcsk_write_space);
    260		smc_clcsock_restore_cb(&clcsk->sk_error_report, &smc->clcsk_error_report);
    261	
    262		write_unlock_bh(&clcsk->sk_callback_lock);
    263	}
    264	
    265	static void smc_restore_fallback_changes(struct smc_sock *smc)
    266	{
    267		if (smc->clcsock->file) { /* non-accepted sockets have no file yet */
    268			smc->clcsock->file->private_data = smc->sk.sk_socket;
    269			smc->clcsock->file = NULL;
    270			smc_fback_restore_callbacks(smc);
    271		}
    272	}
    273	
    274	static int __smc_release(struct smc_sock *smc)
    275	{
    276		struct sock *sk = &smc->sk;
    277		int rc = 0;
    278	
    279		if (!smc->use_fallback) {
    280			rc = smc_close_active(smc);
    281			sock_set_flag(sk, SOCK_DEAD);
    282			sk->sk_shutdown |= SHUTDOWN_MASK;
    283		} else {
    284			if (sk->sk_state != SMC_CLOSED) {
    285				if (sk->sk_state != SMC_LISTEN &&
    286				    sk->sk_state != SMC_INIT)
    287					sock_put(sk); /* passive closing */
    288				if (sk->sk_state == SMC_LISTEN) {
    289					/* wake up clcsock accept */
    290					rc = kernel_sock_shutdown(smc->clcsock,
    291								  SHUT_RDWR);
    292				}
    293				sk->sk_state = SMC_CLOSED;
    294				sk->sk_state_change(sk);
    295			}
    296			smc_restore_fallback_changes(smc);
    297		}
    298	
    299		sk->sk_prot->unhash(sk);
    300	
    301		if (sk->sk_state == SMC_CLOSED) {
    302			if (smc->clcsock) {
    303				release_sock(sk);
    304				smc_clcsock_release(smc);
    305				lock_sock(sk);
    306			}
    307			if (!smc->use_fallback)
    308				smc_conn_free(&smc->conn);
    309		}
    310	
    311		return rc;
    312	}
    313	
    314	static int smc_release(struct socket *sock)
    315	{
    316		struct sock *sk = sock->sk;
    317		struct smc_sock *smc;
    318		int old_state, rc = 0;
    319	
    320		if (!sk)
    321			goto out;
    322	
    323		sock_hold(sk); /* sock_put below */
    324		smc = smc_sk(sk);
    325	
    326		/* trigger info gathering if needed.*/
Quoted 1 line

327 smc_sock_perform_collecting_info(sk, SMC_SOCK_CLOSED_TIMING);

    328	
    329		old_state = sk->sk_state;
    330	
    331		/* cleanup for a dangling non-blocking connect */
    332		if (smc->connect_nonblock && old_state == SMC_INIT)
    333			tcp_abort(smc->clcsock->sk, ECONNABORTED);
    334	
    335		if (cancel_work_sync(&smc->connect_work))
    336			sock_put(&smc->sk); /* sock_hold in smc_connect for passive closing */
    337	
    338		if (sk->sk_state == SMC_LISTEN)
    339			/* smc_close_non_accepted() is called and acquires
    340			 * sock lock for child sockets again
    341			 */
    342			lock_sock_nested(sk, SINGLE_DEPTH_NESTING);
    343		else
    344			lock_sock(sk);
    345	
    346		if (old_state == SMC_INIT && sk->sk_state == SMC_ACTIVE &&
    347		    !smc->use_fallback)
    348			smc_close_active_abort(smc);
    349	
    350		rc = __smc_release(smc);
    351	
    352		/* detach socket */
    353		sock_orphan(sk);
    354		sock->sk = NULL;
    355		release_sock(sk);
    356	
    357		sock_put(sk); /* sock_hold above */
    358		sock_put(sk); /* final sock_put */
    359	out:
    360		return rc;
    361	}
    362	
author maintainer
#6 lore raw

Re: [PATCH bpf-next v3 3/4] net/smc: add BPF injection on smc negotiation

In reply to kernel test robot

You are right. Thanks for you test!!
I will send new version to fix this.!

Quoted 246 lines · kernel test robot

Hi Wythe,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on bpf-next/master]

url:    https://github.com/intel-lab-lkp/linux/commits/D-Wythe/net-smc-move-smc_sock-related-structure-definition/20230228-173007
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git master
patch link:    https://lore.kernel.org/r/1677576294-33411-4-git-send-email-alibuda%40linux.alibaba.com
patch subject: [PATCH bpf-next v3 3/4] net/smc: add BPF injection on smc negotiation
config: x86_64-randconfig-a015-20230227 (https://download.01.org/0day-ci/archive/20230228/[email protected]/config)
compiler: gcc-11 (Debian 11.3.0-8) 11.3.0
reproduce (this is a W=1 build):
         # https://github.com/intel-lab-lkp/linux/commit/aa482ab82f4bf9b99d490f8ba5d88e1491156ccf
         git remote add linux-review https://github.com/intel-lab-lkp/linux
         git fetch --no-tags linux-review D-Wythe/net-smc-move-smc_sock-related-structure-definition/20230228-173007
         git checkout aa482ab82f4bf9b99d490f8ba5d88e1491156ccf
         # save the config file
         mkdir build_dir && cp config build_dir/.config
         make W=1 O=build_dir ARCH=x86_64 olddefconfig
         make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash

If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <[email protected]>
| Link: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

ld: net/smc/af_smc.o: in function `smc_hs_congested':

Quoted 1 line
Quoted 1 line

net/smc/af_smc.c:169: undefined reference to `smc_sock_should_select_smc'

ld: net/smc/af_smc.o: in function `smc_release':

Quoted 1 line
Quoted 1 line

net/smc/af_smc.c:327: undefined reference to `smc_sock_perform_collecting_info'

ld: net/smc/af_smc.o: in function `smc_connect':
net/smc/af_smc.c:1637: undefined reference to `smc_sock_should_select_smc'

vim +169 net/smc/af_smc.c

    156	
    157	static bool smc_hs_congested(const struct sock *sk)
    158	{
    159		const struct smc_sock *smc;
    160	
    161		smc = smc_clcsock_user_data(sk);
    162	
    163		if (!smc)
    164			return true;
    165	
    166		if (workqueue_congested(WORK_CPU_UNBOUND, smc_hs_wq))
    167			return true;
    168	
Quoted 1 line

169 if (!smc_sock_should_select_smc(smc))

    170			return true;
    171	
    172		return false;
    173	}
    174	
    175	static struct smc_hashinfo smc_v4_hashinfo = {
    176		.lock = __RW_LOCK_UNLOCKED(smc_v4_hashinfo.lock),
    177	};
    178	
    179	static struct smc_hashinfo smc_v6_hashinfo = {
    180		.lock = __RW_LOCK_UNLOCKED(smc_v6_hashinfo.lock),
    181	};
    182	
    183	int smc_hash_sk(struct sock *sk)
    184	{
    185		struct smc_hashinfo *h = sk->sk_prot->h.smc_hash;
    186		struct hlist_head *head;
    187	
    188		head = &h->ht;
    189	
    190		write_lock_bh(&h->lock);
    191		sk_add_node(sk, head);
    192		write_unlock_bh(&h->lock);
    193		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
    194	
    195		return 0;
    196	}
    197	EXPORT_SYMBOL_GPL(smc_hash_sk);
    198	
    199	void smc_unhash_sk(struct sock *sk)
    200	{
    201		struct smc_hashinfo *h = sk->sk_prot->h.smc_hash;
    202	
    203		write_lock_bh(&h->lock);
    204		if (sk_del_node_init(sk))
    205			sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
    206		write_unlock_bh(&h->lock);
    207	}
    208	EXPORT_SYMBOL_GPL(smc_unhash_sk);
    209	
    210	/* This will be called before user really release sock_lock. So do the
    211	 * work which we didn't do because of user hold the sock_lock in the
    212	 * BH context
    213	 */
    214	static void smc_release_cb(struct sock *sk)
    215	{
    216		struct smc_sock *smc = smc_sk(sk);
    217	
    218		if (smc->conn.tx_in_release_sock) {
    219			smc_tx_pending(&smc->conn);
    220			smc->conn.tx_in_release_sock = false;
    221		}
    222	}
    223	
    224	struct proto smc_proto = {
    225		.name		= "SMC",
    226		.owner		= THIS_MODULE,
    227		.keepalive	= smc_set_keepalive,
    228		.hash		= smc_hash_sk,
    229		.unhash		= smc_unhash_sk,
    230		.release_cb	= smc_release_cb,
    231		.obj_size	= sizeof(struct smc_sock),
    232		.h.smc_hash	= &smc_v4_hashinfo,
    233		.slab_flags	= SLAB_TYPESAFE_BY_RCU,
    234	};
    235	EXPORT_SYMBOL_GPL(smc_proto);
    236	
    237	struct proto smc_proto6 = {
    238		.name		= "SMC6",
    239		.owner		= THIS_MODULE,
    240		.keepalive	= smc_set_keepalive,
    241		.hash		= smc_hash_sk,
    242		.unhash		= smc_unhash_sk,
    243		.release_cb	= smc_release_cb,
    244		.obj_size	= sizeof(struct smc_sock),
    245		.h.smc_hash	= &smc_v6_hashinfo,
    246		.slab_flags	= SLAB_TYPESAFE_BY_RCU,
    247	};
    248	EXPORT_SYMBOL_GPL(smc_proto6);
    249	
    250	static void smc_fback_restore_callbacks(struct smc_sock *smc)
    251	{
    252		struct sock *clcsk = smc->clcsock->sk;
    253	
    254		write_lock_bh(&clcsk->sk_callback_lock);
    255		clcsk->sk_user_data = NULL;
    256	
    257		smc_clcsock_restore_cb(&clcsk->sk_state_change, &smc->clcsk_state_change);
    258		smc_clcsock_restore_cb(&clcsk->sk_data_ready, &smc->clcsk_data_ready);
    259		smc_clcsock_restore_cb(&clcsk->sk_write_space, &smc->clcsk_write_space);
    260		smc_clcsock_restore_cb(&clcsk->sk_error_report, &smc->clcsk_error_report);
    261	
    262		write_unlock_bh(&clcsk->sk_callback_lock);
    263	}
    264	
    265	static void smc_restore_fallback_changes(struct smc_sock *smc)
    266	{
    267		if (smc->clcsock->file) { /* non-accepted sockets have no file yet */
    268			smc->clcsock->file->private_data = smc->sk.sk_socket;
    269			smc->clcsock->file = NULL;
    270			smc_fback_restore_callbacks(smc);
    271		}
    272	}
    273	
    274	static int __smc_release(struct smc_sock *smc)
    275	{
    276		struct sock *sk = &smc->sk;
    277		int rc = 0;
    278	
    279		if (!smc->use_fallback) {
    280			rc = smc_close_active(smc);
    281			sock_set_flag(sk, SOCK_DEAD);
    282			sk->sk_shutdown |= SHUTDOWN_MASK;
    283		} else {
    284			if (sk->sk_state != SMC_CLOSED) {
    285				if (sk->sk_state != SMC_LISTEN &&
    286				    sk->sk_state != SMC_INIT)
    287					sock_put(sk); /* passive closing */
    288				if (sk->sk_state == SMC_LISTEN) {
    289					/* wake up clcsock accept */
    290					rc = kernel_sock_shutdown(smc->clcsock,
    291								  SHUT_RDWR);
    292				}
    293				sk->sk_state = SMC_CLOSED;
    294				sk->sk_state_change(sk);
    295			}
    296			smc_restore_fallback_changes(smc);
    297		}
    298	
    299		sk->sk_prot->unhash(sk);
    300	
    301		if (sk->sk_state == SMC_CLOSED) {
    302			if (smc->clcsock) {
    303				release_sock(sk);
    304				smc_clcsock_release(smc);
    305				lock_sock(sk);
    306			}
    307			if (!smc->use_fallback)
    308				smc_conn_free(&smc->conn);
    309		}
    310	
    311		return rc;
    312	}
    313	
    314	static int smc_release(struct socket *sock)
    315	{
    316		struct sock *sk = sock->sk;
    317		struct smc_sock *smc;
    318		int old_state, rc = 0;
    319	
    320		if (!sk)
    321			goto out;
    322	
    323		sock_hold(sk); /* sock_put below */
    324		smc = smc_sk(sk);
    325	
    326		/* trigger info gathering if needed.*/
Quoted 1 line

327 smc_sock_perform_collecting_info(sk, SMC_SOCK_CLOSED_TIMING);

    328	
    329		old_state = sk->sk_state;
    330	
    331		/* cleanup for a dangling non-blocking connect */
    332		if (smc->connect_nonblock && old_state == SMC_INIT)
    333			tcp_abort(smc->clcsock->sk, ECONNABORTED);
    334	
    335		if (cancel_work_sync(&smc->connect_work))
    336			sock_put(&smc->sk); /* sock_hold in smc_connect for passive closing */
    337	
    338		if (sk->sk_state == SMC_LISTEN)
    339			/* smc_close_non_accepted() is called and acquires
    340			 * sock lock for child sockets again
    341			 */
    342			lock_sock_nested(sk, SINGLE_DEPTH_NESTING);
    343		else
    344			lock_sock(sk);
    345	
    346		if (old_state == SMC_INIT && sk->sk_state == SMC_ACTIVE &&
    347		    !smc->use_fallback)
    348			smc_close_active_abort(smc);
    349	
    350		rc = __smc_release(smc);
    351	
    352		/* detach socket */
    353		sock_orphan(sk);
    354		sock->sk = NULL;
    355		release_sock(sk);
    356	
    357		sock_put(sk); /* sock_hold above */
    358		sock_put(sk); /* final sock_put */
    359	out:
    360		return rc;
    361	}
    362	
author maintainer
#7 lore raw

PATCH v3 4/4 bpf/selftests: add selftest for SMC bpf capability

In reply to D. Wythe

From: "D. Wythe" <[email protected]>

This PATCH adds a tiny selftest for SMC bpf capability, making decisions on whether to use SMC by collecting certain information from kernel smc sock.

Follow the steps below to run this test.

make -C tools/testing/selftests/bpf
cd tools/testing/selftests/bpf
sudo ./test_progs -t bpf_smc

Results shows:
18 bpf_smc:OK
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED

---
 tools/testing/selftests/bpf/prog_tests/bpf_smc.c |  37 +++
 tools/testing/selftests/bpf/progs/bpf_smc.c      | 320 +++++++++++++++++++++++
 2 files changed, 357 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/bpf_smc.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_smc.c
tools/testing/selftests/bpf/prog_tests/bpf_smc.c new file +37 -0
new file mode 100644
index 0000000..b57326c
@@ -0,0 +1,37 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/err.h>
+#include <netinet/tcp.h>
+#include <test_progs.h>
+#include "bpf_smc.skel.h"
+
+void test_bpf_smc(void)
+{
+	struct bpf_smc *smc_skel;
+	struct bpf_link *link;
+	int err;
+
+	smc_skel = bpf_smc__open();
+	if (!ASSERT_OK_PTR(smc_skel, "skel_open"))
+		return;
+
+	err = bpf_map__set_type(smc_skel->maps.negotiator_map, BPF_MAP_TYPE_HASH);
+	if (!ASSERT_OK(err, "bpf_map__set_type"))
+		goto error;
+
+	err = bpf_map__set_max_entries(smc_skel->maps.negotiator_map, 1);
+	if (!ASSERT_OK(err, "bpf_map__set_type"))
+		goto error;
+
+	err =  bpf_smc__load(smc_skel);
+	if (!ASSERT_OK(err, "skel_load"))
+		goto error;
+
+	link = bpf_map__attach_struct_ops(smc_skel->maps.ops);
+	if (!ASSERT_OK_PTR(link, "bpf_map__attach_struct_ops"))
+		goto error;
+
+	bpf_link__destroy(link);
+error:
+	bpf_smc__destroy(smc_skel);
+}
tools/testing/selftests/bpf/progs/bpf_smc.c new file +320 -0

320 lines not on this page · raw message

Signature · 1 line
1.8.3.1
author maintainer
#9 lore raw
Quoted 9 lines · Jakub Kicinski
Quoted 4 lines · On Tue, 28 Feb 2023 17:24:50 +0800 D. Wythe wrote:

From: "D. Wythe" <[email protected]>

This patches attempt to introduce BPF injection capability for SMC, and add selftest to ensure code stability.

What happened to fixing the issues Al pointed out long, long time ago?

https://lore.kernel.org/all/YutBc9aCQOvPPlWN@ZenIV/

Sorry about that, I don't know much details about this issues. But I have already fed back the problem to Tony Lu, he will give a reply soon, thank you.

D. Wythe

reviewer
#10 lore raw
In reply to Jakub Kicinski
Quoted 9 lines · Jakub Kicinski
Quoted 4 lines · On Tue, 28 Feb 2023 17:24:50 +0800 D. Wythe wrote:

From: "D. Wythe" <[email protected]>

This patches attempt to introduce BPF injection capability for SMC, and add selftest to ensure code stability.

What happened to fixing the issues Al pointed out long, long time ago?

https://lore.kernel.org/all/YutBc9aCQOvPPlWN@ZenIV/

Actually, this patch set is going to replace the patch of TCP ULP for SMC. If this patch set is accepted, I am going to revert that patch.

For the reasons, the TCP ULP for SMC doesn't use wildly. It's not possible to know which applications are suitable to be replaced with SMC. But it's easier to detect the behavior of applications and determine whether to replace applications with SMC. And this patch set is going to fallback to TCP by behavior with eBPF.

So this is the _fix_ for that patch.

Thank you,
Tony Lu

maintainer
#11 lore raw
Quoted 10 lines · On Wed, 1 Mar 2023 10:41:57 +0800 Tony Lu wrote:

Actually, this patch set is going to replace the patch of TCP ULP for SMC. If this patch set is accepted, I am going to revert that patch.

For the reasons, the TCP ULP for SMC doesn't use wildly. It's not possible to know which applications are suitable to be replaced with SMC. But it's easier to detect the behavior of applications and determine whether to replace applications with SMC. And this patch set is going to fallback to TCP by behavior with eBPF.

So this is the _fix_ for that patch.

Good to hear, I was worried you'd still want to install the ULP at some point whether the decision is via BPF or user space.