net/smc: Introduce BPF injection capability
Subsystems (8)
D. Wythe
maintainerv2 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.
D. Wythe (2):
net/smc: Introduce BPF injection capability for SMC bpf/selftests: Test for SMC protocol negotiate
include/linux/btf_ids.h | 15 ++ include/net/smc.h | 254 ++++++++++++++++++ kernel/bpf/bpf_struct_ops_types.h | 4 + net/Makefile | 5 + net/smc/af_smc.c | 10 +- net/smc/bpf_smc_struct_ops.c | 146 +++++++++++ net/smc/smc.h | 220 ---------------- tools/testing/selftests/bpf/prog_tests/bpf_smc.c | 39 +++ tools/testing/selftests/bpf/progs/bpf_smc.c | 315 +++++++++++++++++++++++ 9 files changed, 787 insertions(+), 221 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
D. Wythe
maintainerPATCH v2 1/2 net/smc: Introduce BPF injection capability for SMC
From: "D. Wythe" <[email protected]>
This PATCH attempts to introduce BPF injection capability for SMC. 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 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 | 15 +++ include/net/smc.h | 254 ++++++++++++++++++++++++++++++++++++++ kernel/bpf/bpf_struct_ops_types.h | 4 + net/Makefile | 5 + net/smc/af_smc.c | 10 +- net/smc/bpf_smc_struct_ops.c | 146 ++++++++++++++++++++++ net/smc/smc.h | 220 --------------------------------- 7 files changed, 433 insertions(+), 221 deletions(-) create mode 100644 net/smc/bpf_smc_struct_ops.c
include/linux/btf_ids.h +15 -0
@@ -264,6 +264,21 @@ enum { MAX_BTF_TRACING_TYPE, }; +#if IS_ENABLED (CONFIG_SMC ) +#define BTF_SMC_TYPE_xxx \ + 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) + +enum { +#define BTF_SMC_TYPE(name, type) name, +BTF_SMC_TYPE_xxx +#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 +254 -0
271 lines not on this page · raw message
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 ) +#include <net/smc.h> +BPF_STRUCT_OPS_TYPE(smc_sock_negotiator_ops) +#endif #endif
net/Makefile +5 -0
@@ -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/af_smc.c +9 -1
@@ -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 (smc,SMC_SOCK_CLOSED_TIMING ); + old_state = sk->sk_state;/* cleanup for a dangling non-blocking connect */
@@ -1627,7 +1633,9 @@ 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 ; +tcp_sk (smc->clcsock->sk)->syn_smc = (smc_sock_should_select_smc (smc) ==SK_PASS ) ? +1 :0 ; +if (smc->connect_nonblock) { rc = -EALREADY ; goto out;
net/smc/bpf_smc_struct_ops.c new file +146 -0
146 lines not on this page · raw message
net/smc/smc.h +0 -220
232 lines not on this page · raw message
Signature · 1 line
1.8.3.1
Martin KaFai Lau
reviewerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 18 lines · D. Wythe
From: "D. Wythe" <[email protected]>
This PATCH attempts to introduce BPF injection capability for SMC. 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.
Quoted 6 lines
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.
I have never used smc. I can only comment at its high level usage and details on the bpf side.
Quoted 24 lines
Signed-off-by D. Wythe <[email protected]>--- include/linux/btf_ids.h | 15 +++ include/net/smc.h | 254 ++++++++++++++++++++++++++++++++++++++ kernel/bpf/bpf_struct_ops_types.h | 4 + net/Makefile | 5 + net/smc/af_smc.c | 10 +- net/smc/bpf_smc_struct_ops.c | 146 ++++++++++++++++++++++ net/smc/smc.h | 220 --------------------------------- 7 files changed, 433 insertions(+), 221 deletions(-)create mode 100644
net/smc/bpf_smc_struct_ops.cinclude/linux/btf_ids.h +3
-0@@ -264,6 +264,21 @@ enum { MAX_BTF_TRACING_TYPE, }; +#if IS_ENABLED (CONFIG_SMC ) +#define BTF_SMC_TYPE_xxx \ + BTF_SMC_TYPE(BTF_SMC_TYPE_SOCK, smc_sock) \
include/linux/btf_ids.h enum { · line 279
Quoted 10 lines
include/linux/btf_ids.h +10
-0+ BTF_SMC_TYPE(BTF_SMC_TYPE_CONNECTION, smc_connection) \ + BTF_SMC_TYPE(BTF_SMC_TYPE_HOST_CURSOR, smc_host_cursor) + +enum { +#define BTF_SMC_TYPE (name, type) name, +BTF_SMC_TYPE_xxx +#undef BTF_SMC_TYPE +MAX_BTF_SMC_TYPE, +}; +extern u32 btf_smc_ids[];
Do all these need to be in btf_ids.h?
include/net/smc.h
Quoted 9 lines
include/linux/btf_ids.h +2
-0+#endif +extern u32 btf_tracing_ids[];extern u32 bpf_cgroup_btf_id[];extern u32 bpf_local_storage_map_btf_id[];include/net/smc.h +0
-0
It is not obvious to me why the header moving is needed (from net/smc/smc.h to include/net/smc.h ?). This can use some comment in the commit message and please break it out to another patch.
[ ... ]
net/Makefile obj-$(CONFIG_TIPC) += tipc/ · line 57
Quoted 9 lines
net/Makefile +3
-0@@ -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
This will ensure bpf_smc_struct_ops.c compiled as builtin even when smc is compiled as module?
Quoted 10 lines
net/smc/bpf_smc_struct_ops.c new file +4
-0@@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include <linux/kernel.h> +#include <linux/bpf_verifier.h>
net/smc/bpf_smc_struct_ops.c line 14
Quoted 10 lines
net/smc/bpf_smc_struct_ops.c new file +10
-0+#include <linux/btf_ids.h> +#include <linux/bpf.h> +#include <linux/btf.h> +#include <net/sock.h> +#include <net/smc.h> + +extern struct bpf_struct_ops smc_sock_negotiator_ops; + +DEFINE_RWLOCK (smc_sock_negotiator_ops_rwlock); +struct smc_sock_negotiator_ops *negotiator;
Is it sure one global negotiator (policy) will work for all smc_sock? or each sk should have its own negotiator and the negotiator is selected by setsockopt.
net/smc/bpf_smc_struct_ops.c line 26
Quoted 12 lines
net/smc/bpf_smc_struct_ops.c new file +12
-0+ +/* convert sk to smc_sock */ +static inline struct smc_sock *smc_sk (const struct sock *sk) +{ +return (struct smc_sock *)sk; +} + +/* register ops */ +static inline void smc_reg_passive_sk_ops (struct smc_sock_negotiator_ops *ops) +{ +write_lock_bh (&smc_sock_negotiator_ops_rwlock); + negotiator = ops;
What happens to the existing negotiator?
Quoted 25 lines
net/smc/bpf_smc_struct_ops.c new file +25
-0+write_unlock_bh (&smc_sock_negotiator_ops_rwlock); +} + +/* unregister ops */ +static inline void smc_unreg_passive_sk_ops (struct smc_sock_negotiator_ops *ops) +{ +write_lock_bh (&smc_sock_negotiator_ops_rwlock); +if (negotiator == ops) + negotiator =NULL ; +write_unlock_bh (&smc_sock_negotiator_ops_rwlock); +} + +int smc_sock_should_select_smc (const struct smc_sock *smc) +{ +int ret =SK_PASS ; + +read_lock_bh (&smc_sock_negotiator_ops_rwlock); +if (negotiator && negotiator->negotiate) + ret = negotiator->negotiate ((struct smc_sock *)smc); +read_unlock_bh (&smc_sock_negotiator_ops_rwlock); +return ret; +} +EXPORT_SYMBOL_GPL (smc_sock_should_select_smc); + +void smc_sock_perform_collecting_info (const struct smc_sock *smc,int timing)
net/smc/bpf_smc_struct_ops.c line 61
Quoted 10 lines
net/smc/bpf_smc_struct_ops.c new file +10
-0+{ +read_lock_bh (&smc_sock_negotiator_ops_rwlock); +if (negotiator && negotiator->collect_info) + negotiator->collect_info ((struct smc_sock *)smc, timing); +read_unlock_bh (&smc_sock_negotiator_ops_rwlock); +} +EXPORT_SYMBOL_GPL (smc_sock_perform_collecting_info); + +/* define global smc ID for smc_struct_ops */ +BTF_ID_LIST_GLOBAL (btf_smc_ids,MAX_BTF_SMC_TYPE )
How is btf_smc_ids used?
Quoted 4 lines
net/smc/bpf_smc_struct_ops.c new file +4
-0+#define BTF_SMC_TYPE (name, type) BTF_ID(struct, type) +BTF_SMC_TYPE_xxx +#undef BTF_SMC_TYPE +
D. Wythe
maintainerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 19 lines · Martin KaFai Lau
Quoted 18 lines · D. Wythe
From: "D. Wythe" <[email protected]>
This PATCH attempts to introduce BPF injection capability for SMC. 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.
Quoted 8 lines
Quoted 6 lines
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.I have never used smc. I can only comment at its high level usage and details on the bpf side.
Hi Martin,
Thank you very much for your comments and I'm very sorry for my mistakes.
Quoted 30 lines
Quoted 30 lines
Signed-off-by D. Wythe <[email protected]>---
include/linux/btf_ids.h | 15 +++ include/net/smc.h | 254
++++++++++++++++++++++++++++++++++++++
kernel/bpf/bpf_struct_ops_types.h| 4 + net/Makefile | 5 + net/smc/af_smc.c | 10 +- net/smc/bpf_smc_struct_ops.c | 146 ++++++++++++++++++++++ net/smc/smc.h | 220
---------------------------------7 files changed, 433 insertions(+), 221 deletions(-)create mode 100644
net/smc/bpf_smc_struct_ops.cinclude/linux/btf_ids.h +6
-0@@ -264,6 +264,21 @@ enum { MAX_BTF_TRACING_TYPE, }; +#if IS_ENABLED(CONFIG_SMC) +#define BTF_SMC_TYPE_xxx \ + 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) + +enum {
Quoted 8 lines
include/linux/btf_ids.h enum { · line 279
Quoted 6 lines
include/linux/btf_ids.h +6
-0+#define BTF_SMC_TYPE (name, type) name, +BTF_SMC_TYPE_xxx +#undef BTF_SMC_TYPE +MAX_BTF_SMC_TYPE, +}; +extern u32 btf_smc_ids[];Do all these need to be in
btf_ids.h?
My original intention is to do some security checks via btf_smc_ids,
but since it is not implemented at present, so it is not necessary here.
Quoted 12 lines
include/net/smc.h
Quoted 10 lines
include/linux/btf_ids.h +2
-0+#endif + extern u32 btf_tracing_ids[]; extern u32 bpf_cgroup_btf_id[]; extern u32 bpf_local_storage_map_btf_id[];include/net/smc.h +0
-0It is not obvious to me why the header moving is needed (from
net/smc/smc.htoinclude/net/smc.h?). This can use some comment in the commit message and please break it out to another patch.
Got it, , I have finished the splitting.
Quoted 14 lines
[ ... ]
net/Makefile obj-$(CONFIG_TIPC) += tipc/ · line 57
Quoted 9 lines
net/Makefile +0
-0@@ -52,6 +52,11 @@ obj-$(CONFIG_TIPC) += tipc/obj-$(CONFIG_NETLABEL) += netlabel/
obj-$(CONFIG_IUCV) += iucv/
obj-$(CONFIG_SMC) += smc/net/Makefile +3
-0+ifneq ($(CONFIG_SMC),) +ifeq ($(CONFIG_BPF_SYSCALL),y) +obj-y += smc/bpf_smc_struct_ops.oThis will ensure
bpf_smc_struct_ops.ccompiled as builtin even when smc is compiled as module?
Yes, smc allow compiled as module.
We are also struggling here. If you have a better way, please let me know. 😁
Quoted 15 lines
Quoted 15 lines
net/smc/bpf_smc_struct_ops.c new file +8
-0@@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include <linux/kernel.h> +#include <linux/bpf_verifier.h> +#include <linux/btf_ids.h> +#include <linux/bpf.h> +#include <linux/btf.h> +#include <net/sock.h>
Quoted 9 lines
net/smc/bpf_smc_struct_ops.c line 14
Quoted 6 lines
net/smc/bpf_smc_struct_ops.c new file +6
-0+#include <net/smc.h> + +extern struct bpf_struct_ops smc_sock_negotiator_ops; + +DEFINE_RWLOCK (smc_sock_negotiator_ops_rwlock); +struct smc_sock_negotiator_ops *negotiator;Is it sure one global negotiator (policy) will work for all
smc_sock? or each sk should have its own negotiator and the negotiator is selected by setsockopt.
This is really a good question, we can really consider adding an independent negotiator for each sock.
But just like the TCP congestion control , the global negotiator can be used for sock without
special requirements.
Quoted 7 lines
Quoted 7 lines
net/smc/bpf_smc_struct_ops.c new file +7
-0+ +/* convert sk to smc_sock */ +static inline struct smc_sock *smc_sk (const struct sock *sk) +{ + return (struct smc_sock *)sk; +} +
Quoted 8 lines
Quoted 6 lines
net/smc/bpf_smc_struct_ops.c new file +2
-0+/* register ops */ +static inline void smc_reg_passive_sk_ops(struct
smc_sock_negotiator_ops*ops)
+{
+write_lock_bh(&smc_sock_negotiator_ops_rwlock); + negotiator = ops;What happens to the existing negotiator?
What if we return a failure when the negotiator already exists ?
Quoted 32 lines
Quoted 32 lines
+
write_unlock_bh(&smc_sock_negotiator_ops_rwlock);net/smc/bpf_smc_struct_ops.c new file +4
-0+} + +/* unregister ops */ +static inline void smc_unreg_passive_sk_ops(struct
smc_sock_negotiator_ops*ops)net/smc/bpf_smc_struct_ops.c new file +20
-0+{ + write_lock_bh(&smc_sock_negotiator_ops_rwlock); + if (negotiator == ops) + negotiator = NULL; + write_unlock_bh(&smc_sock_negotiator_ops_rwlock); +} + +int smc_sock_should_select_smc (const struct smc_sock *smc) +{ + int ret = SK_PASS; + + read_lock_bh(&smc_sock_negotiator_ops_rwlock); + if (negotiator && negotiator->negotiate) + ret = negotiator->negotiate((struct smc_sock *)smc); + read_unlock_bh(&smc_sock_negotiator_ops_rwlock); + return ret; +} +EXPORT_SYMBOL_GPL (smc_sock_should_select_smc); + +void smc_sock_perform_collecting_info(const struct smc_sock *smc,int timing)
+{
+read_lock_bh(&smc_sock_negotiator_ops_rwlock);
+ if (negotiator &&negotiator->collect_info)
+negotiator->collect_info((struct smc_sock *)smc, timing);
Quoted 8 lines
net/smc/bpf_smc_struct_ops.c line 61
Quoted 6 lines
+
read_unlock_bh(&smc_sock_negotiator_ops_rwlock);net/smc/bpf_smc_struct_ops.c new file +5
-0+} +EXPORT_SYMBOL_GPL (smc_sock_perform_collecting_info); + +/* define global smc ID for smc_struct_ops */ +BTF_ID_LIST_GLOBAL (btf_smc_ids,MAX_BTF_SMC_TYPE )How is
btf_smc_idsused?
Yes, it is useless here for the time being. I will remove them in the new version.
Quoted 6 lines
Quoted 5 lines
net/smc/bpf_smc_struct_ops.c new file +4
-0+#define BTF_SMC_TYPE (name, type) BTF_ID(struct, type) +BTF_SMC_TYPE_xxx +#undef BTF_SMC_TYPE +
Martin KaFai Lau
reviewerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 20 lines · D. Wythe
Quoted 11 lines
Quoted 11 lines
net/smc/bpf_smc_struct_ops.c new file +8
-0@@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include <linux/kernel.h> +#include <linux/bpf_verifier.h> +#include <linux/btf_ids.h> +#include <linux/bpf.h> +#include <linux/btf.h> +#include <net/sock.h> Quoted 9 lines
net/smc/bpf_smc_struct_ops.c line 14
Quoted 6 lines
net/smc/bpf_smc_struct_ops.c new file +6
-0+#include <net/smc.h> + +extern struct bpf_struct_ops smc_sock_negotiator_ops; + +DEFINE_RWLOCK (smc_sock_negotiator_ops_rwlock); +struct smc_sock_negotiator_ops *negotiator;Is it sure one global negotiator (policy) will work for all
smc_sock? or each sk should have its own negotiator and the negotiator is selected by setsockopt.
Quoted 5 lines
This is really a good question, we can really consider adding an independent negotiator for each sock.
But just like the TCP congestion control , the global negotiator can be used for sock without
special requirements.
It is different from TCP congestion control (CC). TCP CC has a global default but each sk can select what tcp-cc to use and there can be multiple tcp-cc registered under different names.
It sounds like smc using tcp_sock should be able to have different negotiator also (eg. based on dst IP or some other tcp connection characteristic). The tcp-cc registration, per-sock selection and the rcu_read_lock+refcnt are well understood and there are other bpf infrastructure to support the per sock tcp-cc selection (like bpf_setsockopt).
For the network stack, there is little reason other af_* should not follow at the beginning considering the infrastructure has already been built. The one single global negotiator and reader/writer lock in this patch reads like an effort wanted to give it a try and see if it will be useful before implementing the whole thing. It is better to keep it off the tree for now until it is more ready.
D. Wythe
maintainerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 25 lines · Martin KaFai Lau
Quoted 24 lines · D. Wythe
Quoted 11 lines
Quoted 11 lines
net/smc/bpf_smc_struct_ops.c new file +8
-0@@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include <linux/kernel.h> +#include <linux/bpf_verifier.h> +#include <linux/btf_ids.h> +#include <linux/bpf.h> +#include <linux/btf.h> +#include <net/sock.h> Quoted 9 lines
net/smc/bpf_smc_struct_ops.c line 14
Quoted 6 lines
net/smc/bpf_smc_struct_ops.c new file +6
-0+#include <net/smc.h> + +extern struct bpf_struct_ops smc_sock_negotiator_ops; + +DEFINE_RWLOCK (smc_sock_negotiator_ops_rwlock); +struct smc_sock_negotiator_ops *negotiator;Is it sure one global negotiator (policy) will work for all
smc_sock? or each sk should have its own negotiator and the negotiator is selected by setsockopt.This is really a good question, we can really consider adding an independent negotiator for each sock.
But just like the TCP congestion control , the global negotiator can be used for sock without
Quoted 7 lines
Quoted 1 line
special requirements.
It is different from TCP congestion control (CC). TCP CC has a global default but each sk can select what tcp-cc to use and there can be multiple tcp-cc registered under different names.
It sounds like smc using
tcp_sockshould be able to have different negotiator also (eg. based on dst IP or some other tcp connection characteristic). The tcp-cc registration, per-sock selection and the rcu_read_lock+refcnt are well understood and there are other bpf infrastructure to support the per sock tcp-cc selection (likebpf_setsockopt).For the network stack, there is little reason other af_* should not follow at the beginning considering the infrastructure has already been built. The one single global negotiator and reader/writer lock in this patch reads like an effort wanted to give it a try and see if it will be useful before implementing the whole thing. It is better to keep it off the tree for now until it is more ready.
Hi Martin,
Thank you very much for your comments. I have indeed removed global negotiator from my latest implementation.
The latest design is that users can register a negotiator implementation indexed by name, smc_sock can use bpf_setsockopt to specify whether a specific negotiation implementation is required via name. If there are no settings, there will be no negotiators.
What do you think?
In addition, I am very sorry that I have not issued my implementation for such a long time, and I have encountered some problems with the implementation because the SMC needs to be built as kernel module, I have struggled with the bpf_setsockopt implementation, and there are some new self-testes that need to be written.
However, I believe that I can send a new version as soon as possible.
Best wishes
D. Wythe
Martin KaFai Lau
reviewerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 5 lines · D. Wythe
The latest design is that users can register a negotiator implementation indexed by name,
smc_sockcan usebpf_setsockoptto specify whether a specific negotiation implementation is required via name. If there are no settings, there will be no negotiators.What do you think?
tbh, bpf_setsockopt is many steps away. It needs to begin with a syscall setsockopt first. There is little reason it can only be done with a bpf prog. and how does the user know which negotiator a smc sock is using? Currently, ss can learn the tcp-cc of a sk.
~~~~~~~~
If this effort is serious, the code quality has to be much improved. The obvious bug and unused variables make this set at most a RFC.
From the bpf perspective, it is ok-ish to start with a global negotiator first and skip the setsockopt details for now. However, it needs to be have a name. The new link_update (https://lore.kernel.org/bpf/[email protected]/) has to work also. The struct_ops is rcu reader safe, so leverage it whenever it can instead of the read/write lock. It is how struct_ops work for tcp, so try to stay consistent as much as possible in the networking stack.
Quoted 4 lines
In addition, I am very sorry that I have not issued my implementation for such a long time, and I have encountered some problems with the implementation because the SMC needs to be built as kernel module, I have struggled with the
bpf_setsockoptimplementation, and there are some new self-testes that need to be written.
Regarding compiling as module,
+ifneq ($(CONFIG_SMC),)
+ifeq ($(CONFIG_BPF_SYSCALL),y)
+obj-y += smc/bpf_smc_struct_ops.o
+endif
struct_ops does not support module now. It is on the todo list. The bpf_smc_struct_ops.o above can only be used when CONFIG_SMC=y. Otherwise, the bpf_smc_struct_ops is always built in while most users will never load the smc module.
D. Wythe
maintainerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Hi Martin,
Sorry to have been responding so late, I've been working on the link_update you mentioned in last week, I have completed the support and testing of the related functions of it. and it is expected to be released in the next few days.
As you mentioned, I do have much experience in kernel network development, so I plan to resend the PATCH in the form of RFC. I really hope to receive your suggestions in next serials. Thank you.😉
Best wishes.
D. Wythe
Quoted 28 lines · Martin KaFai Lau
Quoted 5 lines · D. Wythe
The latest design is that users can register a negotiator implementation indexed by name,
smc_sockcan usebpf_setsockoptto specify whether a specific negotiation implementation is required via name. If there are no settings, there will be no negotiators.What do you think?
tbh,
bpf_setsockoptis many steps away. It needs to begin with a syscall setsockopt first. There is little reason it can only be done with a bpf prog. and how does the user know which negotiator a smc sock is using? Currently, ss can learn the tcp-cc of a sk.~~~~~~~~
If this effort is serious, the code quality has to be much improved. The obvious bug and unused variables make this set at most a RFC.
From the bpf perspective, it is ok-ish to start with a global negotiator first and skip the setsockopt details for now. However, it needs to be have a name. The new
link_update(https://lore.kernel.org/bpf/[email protected]/) has to work also. Thestruct_opsis rcu reader safe, so leverage it whenever it can instead of the read/write lock. It is howstruct_opswork for tcp, so try to stay consistent as much as possible in the networking stack.Quoted 4 lines
In addition, I am very sorry that I have not issued my implementation for such a long time, and I have encountered some problems with the implementation because the SMC needs to be built as kernel module, I have struggled with the
bpf_setsockoptimplementation, and there are some new self-testes that need to be written.Regarding compiling as module,
net/Makefile +4
-0+ifneq ($(CONFIG_SMC),) +ifeq ($(CONFIG_BPF_SYSCALL),y) +obj-y += smc/bpf_smc_struct_ops.o +endif
struct_opsdoes not support module now. It is on the todo list. The bpf_smc_struct_ops.o above can only be used whenCONFIG_SMC=y. Otherwise, thebpf_smc_struct_opsis always built in while most users will never load the smc module.
Re: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
In reply to D. WytheQuoted 498 lines · D. Wythe
From: "D. Wythe" <[email protected]>
This PATCH attempts to introduce BPF injection capability for SMC. 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 PATCH implement injection capability for SMC via
struct_ops. In that way, we can add new injection scenarios in the future.Signed-off-by D. Wythe <[email protected]>--- include/linux/btf_ids.h | 15 +++ include/net/smc.h | 254 ++++++++++++++++++++++++++++++++++++++ kernel/bpf/bpf_struct_ops_types.h | 4 + net/Makefile | 5 + net/smc/af_smc.c | 10 +- net/smc/bpf_smc_struct_ops.c | 146 ++++++++++++++++++++++ net/smc/smc.h | 220 --------------------------------- 7 files changed, 433 insertions(+), 221 deletions(-)create mode 100644
net/smc/bpf_smc_struct_ops.cinclude/linux/btf_ids.h +15
-0@@ -264,6 +264,21 @@ enum { MAX_BTF_TRACING_TYPE, }; +#if IS_ENABLED (CONFIG_SMC ) +#define BTF_SMC_TYPE_xxx \ + 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) + +enum { +#define BTF_SMC_TYPE(name, type) name, +BTF_SMC_TYPE_xxx +#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 +254
-0@@ -11,13 +11,16 @@#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/bpf.h> #include "linux/ism.h" struct sock ; +struct smc_diag_conninfo ;#define SMC_MAX_PNETID_LEN 16/* Max. length of PNET id */ @@ -90,4 +93,255 @@ struct smcd_dev { u8 going_away : 1; }; +#if IS_ENABLED (CONFIG_SMC ) + +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 +}; + +enum smc_urg_state { +SMC_URG_VALID =1 ,/* data present */ +SMC_URG_NOTYET =2 ,/* data pending */ +SMC_URG_READ =3 ,/* data was already read */ +}; + +/* 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 ATOMIC64_INIT +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) ; + +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 +*/ +}; + +#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 smc_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 smc_sock *smc,int timing); + +#else +struct smc_sock {}; +struct smc_connection {}; +struct smc_sock_negotiator_ops {}; +union smc_host_cursor {}; +#endif /* CONFIG_SMC */ +#endif /* _SMC_H */ 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 ) +#include <net/smc.h> +BPF_STRUCT_OPS_TYPE(smc_sock_negotiator_ops) +#endif #endif net/Makefile +5
-0@@ -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/af_smc.c +9
-1@@ -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 (smc,SMC_SOCK_CLOSED_TIMING ); + old_state = sk->sk_state;/* cleanup for a dangling non-blocking connect */ @@ -1627,7 +1633,9 @@ 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 ; +tcp_sk (smc->clcsock->sk)->syn_smc = (smc_sock_should_select_smc (smc) ==SK_PASS ) ? +1 :0 ; +if (smc->connect_nonblock) { rc = -EALREADY ; goto out;net/smc/bpf_smc_struct_ops.c new file +88
-0@@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include <linux/kernel.h> +#include <linux/bpf_verifier.h> +#include <linux/btf_ids.h> +#include <linux/bpf.h> +#include <linux/btf.h> +#include <net/sock.h> +#include <net/smc.h> + +extern struct bpf_struct_ops smc_sock_negotiator_ops; + +DEFINE_RWLOCK (smc_sock_negotiator_ops_rwlock); +struct smc_sock_negotiator_ops *negotiator; + +/* convert sk to smc_sock */ +static inline struct smc_sock *smc_sk (const struct sock *sk) +{ +return (struct smc_sock *)sk; +} + +/* register ops */ +static inline void smc_reg_passive_sk_ops (struct smc_sock_negotiator_ops *ops) +{ +write_lock_bh (&smc_sock_negotiator_ops_rwlock); + negotiator = ops; +write_unlock_bh (&smc_sock_negotiator_ops_rwlock); +} + +/* unregister ops */ +static inline void smc_unreg_passive_sk_ops (struct smc_sock_negotiator_ops *ops) +{ +write_lock_bh (&smc_sock_negotiator_ops_rwlock); +if (negotiator == ops) + negotiator =NULL ; +write_unlock_bh (&smc_sock_negotiator_ops_rwlock); +} + +int smc_sock_should_select_smc (const struct smc_sock *smc) +{ +int ret =SK_PASS ; + +read_lock_bh (&smc_sock_negotiator_ops_rwlock); +if (negotiator && negotiator->negotiate) + ret = negotiator->negotiate ((struct smc_sock *)smc); +read_unlock_bh (&smc_sock_negotiator_ops_rwlock); +return ret; +} +EXPORT_SYMBOL_GPL (smc_sock_should_select_smc); + +void smc_sock_perform_collecting_info (const struct smc_sock *smc,int timing) +{ +read_lock_bh (&smc_sock_negotiator_ops_rwlock); +if (negotiator && negotiator->collect_info) + negotiator->collect_info ((struct smc_sock *)smc, timing); +read_unlock_bh (&smc_sock_negotiator_ops_rwlock); +} +EXPORT_SYMBOL_GPL (smc_sock_perform_collecting_info); + +/* define global smc ID for smc_struct_ops */ +BTF_ID_LIST_GLOBAL(btf_smc_ids, MAX_BTF_SMC_TYPE) +#define BTF_SMC_TYPE (name, type) BTF_ID(struct, type) +BTF_SMC_TYPE_xxx +#undef BTF_SMC_TYPE + +static int bpf_smc_passive_sk_init (struct btf *btf) +{ +return 0 ; +} + +/* register ops by BPF */ +static int bpf_smc_passive_sk_ops_reg (void *kdata) +{ +struct smc_sock_negotiator_ops *ops = kdata; + +/* at least one ops need implement */ +if (!ops->negotiate || !ops->collect_info) { +pr_err ("At least one ops need implement.\n" ); +return -EINVAL ; + } + +smc_reg_passive_sk_ops (ops); +/* always success now */ +return 0 ; +} + +/* unregister ops by BPF */ +static void bpf_smc_passive_sk_ops_unreg (void *kdata)
net/smc/bpf_smc_struct_ops.c line 98
Quoted 10 lines
net/smc/bpf_smc_struct_ops.c new file +10
-0+{ +smc_unreg_passive_sk_ops (kdata); +} + +static int bpf_smc_passive_sk_ops_check_member (const struct btf_type *t, +const struct btf_member *member, +const struct bpf_prog *prog) +{ +return 0 ; +}
Please check the right pointer type of check_member:
int (*check_member)(const struct btf_type *t,
const struct btf_member *member);
Quoted 285 lines
net/smc/bpf_smc_struct_ops.c new file +48
-0+ +static int bpf_smc_passive_sk_ops_init_member (const struct btf_type *t, +const struct btf_member *member, +void *kdata,const void *udata) +{ +return 0 ; +} + +static const struct bpf_func_proto * +smc_passive_sk_prog_func_proto (enum bpf_func_id func_id,const struct bpf_prog *prog) +{ +return bpf_base_func_proto (func_id); +} + +static bool smc_passive_sk_ops_prog_is_valid_access (int off,int size,enum bpf_access_type type, +const struct bpf_prog *prog, +struct bpf_insn_access_aux *info) +{ +return bpf_tracing_btf_ctx_access (off, size, type, prog, info); +} + +static int smc_passive_sk_ops_prog_struct_access (struct bpf_verifier_log *log, +const struct bpf_reg_state *reg, +int off,int size,enum bpf_access_type atype, +u32 *next_btf_id,enum bpf_type_flag *flag) +{ +/* only allow read now*/ +if (atype ==BPF_READ ) +return btf_struct_access (log, reg, off, size, atype, next_btf_id, flag); + +return -EACCES ; +} + +static const struct bpf_verifier_ops bpf_smc_passive_sk_verifier_ops = { + .get_func_proto = smc_passive_sk_prog_func_proto, + .is_valid_access = smc_passive_sk_ops_prog_is_valid_access, + .btf_struct_access = smc_passive_sk_ops_prog_struct_access +}; + +struct bpf_struct_ops bpf_smc_sock_negotiator_ops = { + .verifier_ops = &bpf_smc_passive_sk_verifier_ops, + .init = bpf_smc_passive_sk_init, + .check_member = bpf_smc_passive_sk_ops_check_member, + .init_member = bpf_smc_passive_sk_ops_init_member, + .reg = bpf_smc_passive_sk_ops_reg, + .unreg = bpf_smc_passive_sk_ops_unreg, + .name ="smc_sock_negotiator_ops" , +};net/smc/smc.h +0
-220@@ -57,232 +57,12 @@ enum smc_state { /* possible states of an SMC socket */struct smc_link_group ; -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_mark_woken {bool woken;void *key;wait_queue_entry_t wait_entry; }; -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 -* */ -}; -static inline struct smc_sock *smc_sk (const struct sock *sk) {return (struct smc_sock *)sk;
D. Wythe
maintainerRe: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 12 lines · Wenjia Zhang
Quoted 9 lines · D. Wythe
From: "D. Wythe" <[email protected]>
This PATCH attempts to introduce BPF injection capability for SMC. 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.
...
Quoted 9 lines
net/smc/bpf_smc_struct_ops.c line 98
Quoted 6 lines
net/smc/bpf_smc_struct_ops.c new file +6
-0+static int bpf_smc_passive_sk_ops_check_member (const struct btf_type *t, + const struct btf_member *member, + const struct bpf_prog *prog) +{ + return 0; +}Please check the right pointer type of
check_member:
Quoted 3 lines
int (*check_member)(const struct
btf_type*t,
const structbtf_member*member);
Hi Wenjia,
That's weird. the prototype of check_member on
latested net-next and bpf-next is:
struct bpf_struct_ops {
const struct bpf_verifier_ops *verifier_ops;
int (*init)(struct btf *btf);
int (*check_member)(const struct btf_type *t,
const struct btf_member *member,
const struct bpf_prog *prog);
int (*init_member)(const struct btf_type *t,
const struct btf_member *member,
void *kdata, const void *udata);
int (*reg)(void *kdata);
void (*unreg)(void *kdata);
const struct btf_type *type;
const struct btf_type *value_type;
const char *name;
struct btf_func_model func_models[BPF_STRUCT_OPS_MAX_NR_MEMBERS];
u32 type_id;
u32 value_id;
};I wonder if there is any code out of sync?
And also I found that this patch is too complex and mixed with the code of two modules (smc & bpf). I will split them out for easier review today.
Best wishes
D. Wythe
Re: [PATCH bpf-next v2 1/2] net/smc: Introduce BPF injection capability for SMC
Quoted 56 lines · D. Wythe
Quoted 12 lines · Wenjia Zhang
Quoted 9 lines · D. Wythe
From: "D. Wythe" <[email protected]>
This PATCH attempts to introduce BPF injection capability for SMC. 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.
...
Quoted 10 lines
net/smc/bpf_smc_struct_ops.c line 69
Quoted 7 lines
+static int bpf_smc_passive_sk_ops_check_member(const struct
btf_type*t,
+ const structbtf_member*member, + const structbpf_prog*prog)net/smc/bpf_smc_struct_ops.c new file +3
-0+{ + return 0; +}Please check the right pointer type of
check_member:Quoted 3 lines
int (*check_member)(const struct
btf_type*t,
const structbtf_member*member);Hi Wenjia,
That's weird. the prototype of
check_memberon
latested net-next and bpf-next is:struct
bpf_struct_ops{
const structbpf_verifier_ops*verifier_ops;
int (*init)(struct btf *btf);
int (*check_member)(const structbtf_type*t,
const structbtf_member*member,
const structbpf_prog*prog);
int (*init_member)(const structbtf_type*t,
const structbtf_member*member,
void *kdata, const void *udata);
int (*reg)(void *kdata);
void (*unreg)(void *kdata);
const structbtf_type*type;
const structbtf_type*value_type;
const char *name;
structbtf_func_modelfunc_models[BPF_STRUCT_OPS_MAX_NR_MEMBERS]; u32type_id;
u32value_id;
};
Quoted 8 lines
I wonder if there is any code out of sync?
And also I found that this patch is too complex and mixed with the code of two modules (smc & bpf). I will split them out for easier review today.
Best wishes
D. Wythe
Good question, the base I used is the current torvalds tree, maybe some code there is still not up-to-date.
But it would be great if you can split them out for better review.
Thanks
Wenjia
D. Wythe
maintainerPATCH v2 2/2 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 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 | 39 +++ tools/testing/selftests/bpf/progs/bpf_smc.c | 315 +++++++++++++++++++++++ 2 files changed, 354 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 +39 -0
@@ -0,0 +1,39 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2019 Facebook */ + +#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 +315 -0
315 lines not on this page · raw message
Signature · 1 line
1.8.3.1
Martin KaFai Lau
reviewerRe: [PATCH bpf-next v2 2/2] bpf/selftests: add selftest for SMC bpf capability
Quoted 30 lines · 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
cdtools/testing/selftests/bpf
sudo ./test_progs -tbpf_smcResults shows:
18 bpf_smc:OK
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILEDSigned-off-by D. Wythe <[email protected]>--- tools/testing/selftests/bpf/prog_tests/bpf_smc.c | 39 +++ tools/testing/selftests/bpf/progs/bpf_smc.c | 315 +++++++++++++++++++++++ 2 files changed, 354 insertions(+)create mode 100644
tools/testing/selftests/bpf/prog_tests/bpf_smc.ccreate mode 100644tools/testing/selftests/bpf/progs/bpf_smc.ctools/testing/selftests/bpf/prog_tests/bpf_smc.c new file +0
-0
tools/testing/selftests/bpf/prog_tests/bpf_smc.c line 2
Quoted 2 lines
tools/testing/selftests/bpf/prog_tests/bpf_smc.c new file +2
-0@@ -0,0 +1,39 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2019 Facebook */
copy-and-paste left-over...
Quoted 56 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +50
-0@@ -0,0 +1,315 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include <linux/bpf.h> +#include <linux/stddef.h> +#include <linux/smc.h> +#include <stdbool.h> +#include <linux/types.h> +#include <bpf/bpf_helpers.h> +#include <bpf/bpf_core_read.h> +#include <bpf/bpf_tracing.h> + +#define BPF_STRUCT_OPS(name, args...) \ + SEC("struct_ops/"#name) \ + BPF_PROG(name, args) + +#define SMC_LISTEN (10) +#define SMC_SOCK_CLOSED_TIMING (0) +extern unsigned long CONFIG_HZ __kconfig; +#define HZ CONFIG_HZ + +char _license[] SEC("license") = "GPL"; +#define max (a, b) ((a) > (b) ? (a) : (b)) + +struct sock_common { +unsigned char skc_state; +__u16 skc_num; +} __attribute__((preserve_access_index)); + +struct sock { +struct sock_common __sk_common; +int sk_sndbuf; +} __attribute__((preserve_access_index)); + +struct inet_sock { +struct sock sk; +} __attribute__((preserve_access_index)); + +struct inet_connection_sock { +struct inet_sock icsk_inet; +} __attribute__((preserve_access_index)); + +struct tcp_sock { +struct inet_connection_sock inet_conn; +__u32 rcv_nxt; +__u32 snd_nxt; +__u32 snd_una; +__u32 delivered; +__u8 syn_data:1 ,/* SYN includes data */ + syn_fastopen:1 ,/* SYN includes Fast Open option */ + syn_fastopen_exp:1,/* SYN includes Fast Open exp. option */
tools/testing/selftests/bpf/progs/bpf_smc.c line 60
Quoted 10 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +10
-0+ syn_fastopen_ch:1, /* Active TFO re-enabling probe */ + syn_data_acked:1,/* data in SYN is acked by SYN-ACK */ + save_syn:1, /* Save headers of SYN packet */ + is_cwnd_limited:1,/* forward progress limited by snd_cwnd? */ + syn_smc:1; /* SYN includes SMC */ +} __attribute__((preserve_access_index)); + +struct socket { +struct sock *sk; +} __attribute__((preserve_access_index));
All these tcp_sock, socket, inet_sock definitions can go away if it includes "vmlinux.h". tcp_ca_write_sk_pacing.c is a better example to follow. Try to define the "common" (eg. tcp, tc...etc) missing macros in bpf_tracing_net.h. The smc specific macros can stay in this file.
Quoted 55 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +55
-0+static inline struct smc_prediction *smc_prediction_get (const struct smc_sock *smc, +const struct tcp_sock *tp,__u64 tstamp) +{ +struct smc_prediction zero = {}, *smc_predictor; +__u16 key; +__u32 gap; +int err; + + err =bpf_core_read (&key,sizeof (__u16), &tp->inet_conn.icsk_inet.sk.__sk_common.skc_num); +if (err) +return NULL ; + +/* BAD key */ +if (key ==0 ) +return NULL ; + + smc_predictor =bpf_map_lookup_elem (&negotiator_map, &key); +if (!smc_predictor) { + zero.start_tstamp =bpf_jiffies64 (); + zero.pacing_delta =SMC_PREDICTION_MIN_PACING_DELTA ; +bpf_map_update_elem (&negotiator_map, &key, &zero,0 ); + smc_predictor =bpf_map_lookup_elem (&negotiator_map, &key); +if (!smc_predictor) +return NULL ; + } + +if (tstamp) { +bpf_spin_lock (&smc_predictor->lock); + gap = (tstamp - smc_predictor->start_tstamp) / smc_predictor->pacing_delta; +/* new splice */ +if (gap >0 ) { + smc_predictor->start_tstamp = tstamp; + smc_predictor->last_rate_of_lcc = + (smc_prediction_calt_rate (smc_predictor) *7 ) >> (2 + gap); + smc_predictor->closed_long_cc =0 ; + smc_predictor->closed_total_cc =0 ; + smc_predictor->incoming_long_cc =0 ; + } +bpf_spin_unlock (&smc_predictor->lock); + } +return smc_predictor; +} + +/* BPF struct ops for smc protocol negotiator */ +struct smc_sock_negotiator_ops { +/* ret for negotiate */ +int (*negotiate)(struct smc_sock *smc); + +/* info gathering timing */ +void (*collect_info)(struct smc_sock *smc,int timing); +}; + +int BPF_STRUCT_OPS (bpf_smc_negotiate ,struct smc_sock *smc) +{ +struct smc_prediction *smc_predictor;
tools/testing/selftests/bpf/progs/bpf_smc.c line 203
Quoted 10 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +10
-0+struct tcp_sock *tp; +struct sock *clcsk; +int ret =SK_DROP ; +__u32 rate =0 ; + +/* Only make decison during listen */ +if (smc->sk.__sk_common.skc_state !=SMC_LISTEN ) +return SK_PASS ; + + clcsk =BPF_CORE_READ (smc, clcsock, sk);
Instead of using bpf_core_read here, why not directly gets the clcsk like the 'smc->sk.__sk_common.skc_state' above.
tools/testing/selftests/bpf/progs/bpf_smc.c line 207
Quoted 4 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +4
-0+if (!clcsk) + goto error; + + tp =tcp_sk (clcsk);
There is a bpf_skc_to_tcp_sock(). Give it a try after changing the above BPF_CORE_READ.
Quoted 23 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +23
-0+if (!tp) + goto error; + + smc_predictor =smc_prediction_get (smc, tp,bpf_jiffies64 ()); +if (!smc_predictor) +return SK_PASS ; + +bpf_spin_lock (&smc_predictor->lock); + +if (smc_predictor->incoming_long_cc ==0 ) + goto out_locked_pass; + +if (smc_predictor->incoming_long_cc >SMC_PREDICTION_MAX_LONGCC_PER_SPLICE ) { + ret =100 ; + goto out_locked_drop; + } + + rate =smc_prediction_calt_rate (smc_predictor); +if (rate <SMC_PREDICTION_LONGCC_RATE_THRESHOLD ) { + ret =200 ; + goto out_locked_drop; + } +out_locked_pass:
tools/testing/selftests/bpf/progs/bpf_smc.c line 240
Quoted 10 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +10
-0+ smc_predictor->incoming_long_cc++; +bpf_spin_unlock (&smc_predictor->lock); +return SK_PASS ; +out_locked_drop: +bpf_spin_unlock (&smc_predictor->lock); +error: +return SK_DROP ; +} + +void BPF_STRUCT_OPS (bpf_smc_collect_info ,struct smc_sock *smc,int timing)
Try to stay with SEC("struct_ops/...") void BPF_PROG(....)
D. Wythe
maintainerRe: [PATCH bpf-next v2 2/2] bpf/selftests: add selftest for SMC bpf capability
Quoted 26 lines · Martin KaFai Lau
Quoted 25 lines · 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
cdtools/testing/selftests/bpf
sudo ./test_progs -tbpf_smcResults shows:
18 bpf_smc:OK
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILEDSigned-off-by D. Wythe <[email protected]>---
tools/testing/selftests/bpf/prog_tests/bpf_smc.c| 39 +++ tools/testing/selftests/bpf/progs/bpf_smc.c | 315
+++++++++++++++++++++++2 files changed, 354 insertions(+)create mode 100644
tools/testing/selftests/bpf/prog_tests/bpf_smc.ccreate mode 100644tools/testing/selftests/bpf/progs/bpf_smc.c
Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/bpf_smc.c line 2
Quoted 8 lines
tools/testing/selftests/bpf/prog_tests/bpf_smc.c new file +2
-0@@ -0,0 +1,39 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2019 Facebook */ copy-and-paste left-over...
Sorry for that, but it might be more appropriate to delete it here... 😂
Quoted 61 lines
Quoted 61 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +54
-0@@ -0,0 +1,315 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include <linux/bpf.h> +#include <linux/stddef.h> +#include <linux/smc.h> +#include <stdbool.h> +#include <linux/types.h> +#include <bpf/bpf_helpers.h> +#include <bpf/bpf_core_read.h> +#include <bpf/bpf_tracing.h> + +#define BPF_STRUCT_OPS(name, args...) \ + SEC("struct_ops/"#name) \ + BPF_PROG(name, args) + +#define SMC_LISTEN (10) +#define SMC_SOCK_CLOSED_TIMING (0) +extern unsigned long CONFIG_HZ __kconfig; +#define HZ CONFIG_HZ + +char _license[] SEC("license") = "GPL"; +#define max (a, b) ((a) > (b) ? (a) : (b)) + +struct sock_common { + unsigned char skc_state; + __u16 skc_num; +} __attribute__((preserve_access_index)); + +struct sock { + struct sock_common __sk_common; + int sk_sndbuf; +} __attribute__((preserve_access_index)); + +struct inet_sock { + struct sock sk; +} __attribute__((preserve_access_index)); + +struct inet_connection_sock { + struct inet_sock icsk_inet; +} __attribute__((preserve_access_index)); + +struct tcp_sock { + struct inet_connection_sock inet_conn; + __u32 rcv_nxt; + __u32 snd_nxt; + __u32 snd_una; + __u32 delivered; + __u8 syn_data:1, /* SYN includes data */ + syn_fastopen:1, /* SYN includes Fast Open option */ + syn_fastopen_exp:1,/* SYN includes Fast Open exp. option */ + syn_fastopen_ch:1, /* Active TFO re-enabling probe */ + syn_data_acked:1,/* data in SYN is acked by SYN-ACK */ + save_syn:1, /* Save headers of SYN packet */ + is_cwnd_limited:1,/* forward progress limited by snd_cwnd? */
Quoted 8 lines
tools/testing/selftests/bpf/progs/bpf_smc.c line 60
Quoted 6 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +6
-0+ syn_smc:1; /* SYN includes SMC */ +} __attribute__((preserve_access_index)); + +struct socket { + struct sock *sk; +} __attribute__((preserve_access_index));All these
tcp_sock, socket,inet_sockdefinitions can go away if it includes "vmlinux.h".tcp_ca_write_sk_pacing.cis a better example to follow. Try to define the "common" (eg. tcp, tc...etc) missing macros inbpf_tracing_net.h. The smc specific macros can stay in this file.
Got it, i'll fix this.
Quoted 72 lines
Quoted 59 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +1
-0+static inline struct smc_prediction *smc_prediction_get(const struct
smc_sock*smc,
+ const structtcp_sock*tp,__u64tstamp)tools/testing/selftests/bpf/progs/bpf_smc.c new file +7
-0+{ + struct smc_prediction zero = {}, *smc_predictor; + __u16 key; + __u32 gap; + int err; + + err = bpf_core_read(&key, sizeof(__u16),
&tp->inet_conn.icsk_inet.sk.__sk_common.skc_num);
+ if (err)
+ return NULL;tools/testing/selftests/bpf/progs/bpf_smc.c new file +18
-0+ + /* BAD key */ + if (key == 0) + return NULL; + + smc_predictor = bpf_map_lookup_elem(&negotiator_map, &key); + if (!smc_predictor) { + zero.start_tstamp = bpf_jiffies64(); + zero.pacing_delta = SMC_PREDICTION_MIN_PACING_DELTA; + bpf_map_update_elem(&negotiator_map, &key, &zero, 0); + smc_predictor = bpf_map_lookup_elem(&negotiator_map, &key); + if (!smc_predictor) + return NULL; + } + + if (tstamp) { + bpf_spin_lock(&smc_predictor->lock); + gap = (tstamp - smc_predictor->start_tstamp) /
smc_predictor->pacing_delta;
+ /* new splice */
+ if (gap > 0) {
+smc_predictor->start_tstamp= tstamp;
+smc_predictor->last_rate_of_lcc=
+ (smc_prediction_calt_rate(smc_predictor)* 7) >> (2
+ gap);
+smc_predictor->closed_long_cc= 0;
+smc_predictor->closed_total_cc= 0;
+smc_predictor->incoming_long_cc= 0;
+ }
+bpf_spin_unlock(&smc_predictor->lock);
+ }
+ returnsmc_predictor;tools/testing/selftests/bpf/progs/bpf_smc.c new file +14
-0+} + +/* BPF struct ops for smc protocol negotiator */ +struct smc_sock_negotiator_ops { + /* ret for negotiate */ + int (*negotiate)(struct smc_sock *smc); + + /* info gathering timing */ + void (*collect_info)(struct smc_sock *smc, int timing); +}; + +int BPF_STRUCT_OPS (bpf_smc_negotiate ,struct smc_sock *smc) +{ + struct smc_prediction *smc_predictor;tools/testing/selftests/bpf/progs/bpf_smc.c line 203
Quoted 10 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +10
-0+ struct tcp_sock *tp; + struct sock *clcsk; + int ret = SK_DROP; + __u32 rate = 0; + + /* Only make decison during listen */ + if (smc->sk.__sk_common.skc_state != SMC_LISTEN) + return SK_PASS; + + clcsk = BPF_CORE_READ(smc, clcsock, sk);Instead of using
bpf_core_readhere, why not directly gets the clcsk like the 'smc->sk.__sk_common.skc_state' above.
Quoted 6 lines
Quoted 4 lines
+ if (!clcsk)
+ goto error;
+
+ tp =tcp_sk(clcsk);There is a
bpf_skc_to_tcp_sock(). Give it a try after changing the aboveBPF_CORE_READ.
Copy that! thanks.
Quoted 30 lines
Quoted 30 lines
+ if (!tp)
+ goto error;tools/testing/selftests/bpf/progs/bpf_smc.c new file +11
-0+ + smc_predictor = smc_prediction_get(smc, tp, bpf_jiffies64()); + if (!smc_predictor) + return SK_PASS; + + bpf_spin_lock(&smc_predictor->lock); + + if (smc_predictor->incoming_long_cc == 0) + goto out_locked_pass; + + if (smc_predictor->incoming_long_cc >
SMC_PREDICTION_MAX_LONGCC_PER_SPLICE) {
+ ret = 100;
+ gotoout_locked_drop;
+ }tools/testing/selftests/bpf/progs/bpf_smc.c new file +12
-0+ + rate = smc_prediction_calt_rate(smc_predictor); + if (rate < SMC_PREDICTION_LONGCC_RATE_THRESHOLD) { + ret = 200; + goto out_locked_drop; + } +out_locked_pass: + smc_predictor->incoming_long_cc++; + bpf_spin_unlock(&smc_predictor->lock); + return SK_PASS; +out_locked_drop: + bpf_spin_unlock(&smc_predictor->lock);
Quoted 8 lines
Quoted 6 lines
tools/testing/selftests/bpf/progs/bpf_smc.c new file +5
-0+error: + return SK_DROP; +} + +void BPF_STRUCT_OPS(bpf_smc_collect_info, struct smc_sock *smc, inttiming)
Try to stay with
SEC("struct_ops/...")voidBPF_PROG(....)
Got it. I have finished this modification in v4.