net/smc: Introduce BPF injection capability
Subsystems (8)
D. Wythe
maintainerv3 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 format errors in subject.
2. Remove unnecessary conditions in Kconfig.
v2 -> v1:
1. Fix complie error if CONFIG_BPF_SYSCALL set while CONFIG_SMC_BPF not.
2. Fix potential reference leaks, smc_destruct may be prematurely retired due to pre conditions.
D. Wythe (5):
net/smc: move smc_sock related structure definition net/smc: allow smc to negotiate protocols on policies net/smc: allow set or get smc negotiator by sockopt
bpf: add smc negotiator support in BPF struct_ops bpf/selftests: add selftest for SMC bpf capability
include/net/smc.h | 268 +++++++++++++++++ include/uapi/linux/smc.h | 1 + kernel/bpf/bpf_struct_ops_types.h | 4 + net/Makefile | 1 + net/smc/Kconfig | 11 + net/smc/af_smc.c | 203 ++++++++++--- net/smc/bpf_smc.c | 360 +++++++++++++++++++++++ net/smc/smc.h | 224 -------------- tools/testing/selftests/bpf/prog_tests/bpf_smc.c | 107 +++++++ tools/testing/selftests/bpf/progs/bpf_smc.c | 265 +++++++++++++++++ 10 files changed, 1185 insertions(+), 259 deletions(-) create mode 100644 net/smc/bpf_smc.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
D. Wythe
maintainerRFC PATCH v3 1/5 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
@@ -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 */
@@ -91,4 +96,224 @@ 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
D. Wythe
maintainerRFC PATCH v3 2/5 net/smc: allow smc to negotiate protocols on policies
In reply to D. WytheFrom: "D. Wythe" <[email protected]>
As we all know, 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.
Just like the congestion control implementation in TCP, this patch provides a generic negotiator implementation. If necessary, we can provide different protocol negotiation strategies for apps based on this implementation.
But most importantly, this patch provides the possibility of eBPF injection, allowing users to implement their own protocol negotiation policy in userspace.
--- include/net/smc.h | 43 ++++++++++++ net/Makefile | 1 + net/smc/Kconfig | 11 +++ net/smc/af_smc.c | 68 +++++++++++++++++- net/smc/bpf_smc.c | 201 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 5 files changed, 323 insertions(+), 1 deletion(-) create mode 100644 net/smc/bpf_smc.c
include/net/smc.h +43 -0
@@ -296,6 +296,8 @@ struct smc_sock { /* smc sock container */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; +/* protocol negotiator ops */ +const struct smc_sock_negotiator_ops *negotiator_ops;/* original af ops */ int sockopt_defer_accept;/* sockopt TCP_DEFER_ACCEPT
@@ -316,4 +318,45 @@ struct smc_sock { /* smc sock container */*/ }; +#ifdef CONFIG_SMC_BPF + +#define SMC_NEGOTIATOR_NAME_MAX (16) +#define SMC_SOCK_CLOSED_TIMING (0) + +/* BPF struct ops for smc protocol negotiator */ +struct smc_sock_negotiator_ops { + +struct list_head list; + +/* ops name */ +char name[16 ]; +/* key for name */ +u32 key; + +/* init with sk */ +void (*init)(struct sock *sk); + +/* release with sk */ +void (*release)(struct sock *sk); + +/* advice for negotiate */ +int (*negotiate)(struct sock *sk); + +/* info gathering timing */ +void (*collect_info)(struct sock *sk,int timing); + +/* module owner */ +struct module *owner; +}; + +int smc_sock_register_negotiator_ops (struct smc_sock_negotiator_ops *ops); +int smc_sock_update_negotiator_ops (struct smc_sock_negotiator_ops *ops, +struct smc_sock_negotiator_ops *old_ops); +void smc_sock_unregister_negotiator_ops (struct smc_sock_negotiator_ops *ops); +int smc_sock_assign_negotiator_ops (struct smc_sock *smc,const char *name); +void smc_sock_cleanup_negotiator_ops (struct smc_sock *smc,int in_release); +void smc_sock_clone_negotiator_ops (struct sock *parent,struct sock *child); + +#endif +#endif /* _SMC_H */
net/Makefile +1 -0
@@ -52,6 +52,7 @@ obj-$(CONFIG_TIPC) += tipc/ obj-$(CONFIG_NETLABEL) += netlabel/ obj-$(CONFIG_IUCV) += iucv/ obj-$(CONFIG_SMC) += smc/ +obj-$(CONFIG_SMC_BPF) += smc/bpf_smc.o obj-$(CONFIG_RFKILL) += rfkill/ obj-$(CONFIG_NET_9P) += 9p/ obj-$(CONFIG_CAIF) += caif/
net/smc/Kconfig +11 -0
@@ -19,3 +19,14 @@ config SMC_DIAG smcss. if unsure, say Y. + +config SMC_BPF + bool "SMC: support eBPF" if SMC + depends on BPF_SYSCALL + default n + help + Supports eBPF to allows user mode participation in SMC's protocol process + via ebpf programs. Alternatively, obtain information about the SMC socks + through the ebpf program. + + If unsure, say N.
net/smc/af_smc.c +67 -1
104 lines not on this page · raw message
net/smc/bpf_smc.c new file +201 -0
201 lines not on this page · raw message
Signature · 1 line
1.8.3.1
D. Wythe
maintainerRFC PATCH v3 3/5 net/smc: allow set or get smc negotiator by sockopt
In reply to D. WytheFrom: "D. Wythe" <[email protected]>
Allow applications to set specific protocol negotiation rules for SMC Socks. Typically, applications need to know the name of the negotiator and then set it through the syscall setsockopt, for examples:
const char name[] = "apps";setsockopt(fd, SOL_SMC, SMC_NEGOTIATOR, name, sizeof(name) - 1);
Noted that there is no default negotiator in SMC implementation, the application needs to inject the specific implementation through eBPF before setting it up. Although no default negotiator implementation is provided,
Note that SMC does not provide a default negotiator in SMC implementation,
and the application needs to inject the specific implementation through eBPF before setting it up. Although no default negotiator implementation is provided, logically it can be seen as an implementation that always return SK_PASS.
--- include/uapi/linux/smc.h | 1 + net/smc/af_smc.c | 135 +++++++++++++++++++++++++++++++++++------------ 2 files changed, 102 insertions(+), 34 deletions(-)
include/uapi/linux/smc.h +1 -0
@@ -299,5 +299,6 @@ enum {/* SMC socket options */ #define SMC_LIMIT_HS 1/* constraint on smc handshake */ +#define SMC_NEGOTIATOR 2/* SMC protocol negotiator */ #endif /* _UAPI_LINUX_SMC_H */
net/smc/af_smc.c +101 -34
184 lines not on this page · raw message
Signature · 1 line
1.8.3.1
D. Wythe
maintainerRFC PATCH v3 4/5 bpf: add smc negotiator support in BPF struct_ops
In reply to D. WytheFrom: "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.
--- kernel/bpf/bpf_struct_ops_types.h | 4 + net/smc/bpf_smc.c | 159 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 163 insertions(+)
kernel/bpf/bpf_struct_ops_types.h +4 -0
@@ -9,4 +9,8 @@#include <net/tcp.h> BPF_STRUCT_OPS_TYPE(tcp_congestion_ops)#endif +#if IS_ENABLED (CONFIG_SMC_BPF ) +#include <net/smc.h> +BPF_STRUCT_OPS_TYPE(smc_sock_negotiator_ops) +#endif #endif
net/smc/bpf_smc.c +159 -0
176 lines not on this page · raw message
Signature · 1 line
1.8.3.1
D. Wythe
maintainerRFC PATCH v3 5/5 bpf/selftests: add selftest for SMC bpf capability
In reply to D. WytheFrom: "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 smc
Results shows:
18/1 bpf_smc/load:OK 18/2 bpf_smc/update:OK 18/3 bpf_smc/ref:OK 18 bpf_smc:OK
Summary: 1/3 PASSED, 0 SKIPPED, 0 FAILED
--- tools/testing/selftests/bpf/prog_tests/bpf_smc.c | 107 +++++++++ tools/testing/selftests/bpf/progs/bpf_smc.c | 265 +++++++++++++++++++++++ 2 files changed, 372 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 +107 -0
@@ -0,0 +1,107 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include <linux/err.h> +#include <netinet/tcp.h> +#include <test_progs.h> +#include "network_helpers.h" +#include "bpf_smc.skel.h" + +#define SOL_SMC 286 +#define SMC_NEGOTIATOR 2 +static const char name[] ="apps" ; + +void run_smc (void ) +{ +int fd, err; + + fd =socket (AF_SMC ,SOCK_STREAM ,0 ); +ASSERT_GT (fd,0 ,"create smc socket" ); + + err =setsockopt (fd,SOL_SMC ,SMC_NEGOTIATOR , name,sizeof (name) /sizeof (char )); +ASSERT_EQ (err,0 ,"setsockopt" ); + +close (fd); +} + +void test_load (void ) +{ +struct bpf_smc *smc_skel; +struct bpf_link *link; + + smc_skel =bpf_smc__open_and_load (); +if (!ASSERT_OK_PTR (smc_skel,"skel_open" )) +return ; + + 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); +} + +void test_update (void ) +{ +struct bpf_smc *smc_skel; +struct bpf_link *link; +int err; + + smc_skel =bpf_smc__open_and_load (); +if (!ASSERT_OK_PTR (smc_skel,"skel_open" )) +return ; + + link =bpf_map__attach_struct_ops (smc_skel->maps.accept); +if (!ASSERT_OK_PTR (link,"bpf_map__attach_struct_ops" )) + goto error; + +run_smc (); +ASSERT_EQ (smc_skel->bss->accept_cnt,1 ,"accept_cnt" ); + + err =bpf_link__update_map (link, smc_skel->maps.drop); +ASSERT_OK (err,"update_map" ); + +run_smc (); +ASSERT_EQ (smc_skel->bss->accept_cnt,1 ,"accept_cnt" ); +ASSERT_EQ (smc_skel->bss->drop_cnt,1 ,"drop_cnt" ); + +bpf_link__destroy (link); +error: +bpf_smc__destroy (smc_skel); +} + +void test_ref (void ) +{ +struct bpf_smc *smc_skel; +struct bpf_link *link; +int fd =0 , err; + + smc_skel =bpf_smc__open_and_load (); +if (!ASSERT_OK_PTR (smc_skel,"skel_open" )) +return ; + + link =bpf_map__attach_struct_ops (smc_skel->maps.accept); +if (!ASSERT_OK_PTR (link,"bpf_map__attach_struct_ops" )) + goto error; + + fd =socket (AF_SMC ,SOCK_STREAM ,0 ); +ASSERT_GT (fd,0 ,"create smc socket" ); + err =setsockopt (fd,SOL_SMC ,SMC_NEGOTIATOR , name,sizeof (name) /sizeof (char )); +ASSERT_EQ (err,0 ,"setsockopt" ); +bpf_link__destroy (link); +if (fd >0 ) +close (fd); +ASSERT_EQ (smc_skel->bss->accept_release_cnt,1 ,"accept_release_cnt" ); +error: +bpf_smc__destroy (smc_skel); +} + +void test_bpf_smc (void ) +{ +if (test__start_subtest ("load" )) +test_load (); +if (test__start_subtest ("update" )) +test_update (); +if (test__start_subtest ("ref" )) +test_ref (); +}
tools/testing/selftests/bpf/progs/bpf_smc.c new file +265 -0
265 lines not on this page · raw message
Signature · 1 line
1.8.3.1
D. Wythe
maintainerHi Martin,
At now, most of the issues you mentioned have been resolved, mainly including
1. Use RCU instead of read write lock
2. Support for update
3. Negotiator requires a name
4. Added a KConfig with a default value of N to avoid affecting users who do not require SMC.
5. Revised some issues with test cases.
Do you have any other suggestions? If any, please let me know. Thank a lot! 😁 If there are no other opinions, I plan to convert the RFC into a formal PATCH. What do you think?
Best wishes
D. Wythe
Quoted 63 lines · D. Wythe
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 format errors in subject.
2. Remove unnecessary conditions inKconfig.v2 -> v1:
1. Fix complie error if
CONFIG_BPF_SYSCALLset whileCONFIG_SMC_BPFnot.Reported-by kernel test robot <[email protected]>2. Fix potential reference leaks,
smc_destructmay be prematurely retired due to pre conditions.D. Wythe (5):
net/smc: movesmc_sockrelated structure definition net/smc: allow smc to negotiate protocols on policies net/smc: allow set or get smc negotiator by sockopt
bpf: add smc negotiator support in BPFstruct_opsbpf/selftests: add selftest for SMC bpf capabilityinclude/net/smc.h | 268 +++++++++++++++++ include/uapi/linux/smc.h | 1 + kernel/bpf/bpf_struct_ops_types.h | 4 + net/Makefile | 1 + net/smc/Kconfig | 11 + net/smc/af_smc.c | 203 ++++++++++--- net/smc/bpf_smc.c | 360 +++++++++++++++++++++++ net/smc/smc.h | 224 -------------- tools/testing/selftests/bpf/prog_tests/bpf_smc.c | 107 +++++++ tools/testing/selftests/bpf/progs/bpf_smc.c | 265 +++++++++++++++++ 10 files changed, 1185 insertions(+), 259 deletions(-)create mode 100644
net/smc/bpf_smc.c
create mode 100644tools/testing/selftests/bpf/prog_tests/bpf_smc.ccreate mode 100644tools/testing/selftests/bpf/progs/bpf_smc.c