Do not limit cb_flags when creating child sk from listen sk
Subsystems (6)
cover Do not limit cb_flags when creating child sk from listen sk
This set fixes an issue that the bpf_skops_init_child() unnecessarily limited the child sk from inheriting all bpf_sock_ops_cb_flags of the listen sk. It also adds a test to check that.
Martin KaFai Lau (2):
bpf: tcp: Do not limit cb_flags when creating child sk from listen sk
bpf: selftest: Ensure the child sk inherited all bpf_sock_ops_cb_flags
include/net/tcp.h | 33 ------------------- net/ipv4/tcp_minisocks.c | 1 - .../bpf/prog_tests/tcp_hdr_options.c | 12 +++++++ .../bpf/progs/test_misc_tcp_hdr_options.c | 4 +-- .../bpf/progs/test_tcp_hdr_options.c | 7 ++-- .../selftests/bpf/test_tcp_hdr_options.h | 5 +-- 6 files changed, 22 insertions(+), 40 deletions(-)
Signature · 1 line
2.24.1
PATCH 1/2 bpf: tcp: Do not limit cb_flags when creating child sk from listen sk
The commit 0813a841566f ("bpf: tcp: Allow bpf prog to write and parse TCP header option") unnecessarily introduced bpf_skops_init_child() which limited the child
sk from inheriting all bpf_sock_ops_cb_flags of the listen sk. That
breaks existing user expectation.
This patch removes the bpf_skops_init_child() and just allows sock_copy() to do its job to copy everything from listen sk to the child sk.
--- include/net/tcp.h | 33 --------------------------------- net/ipv4/tcp_minisocks.c | 1 - 2 files changed, 34 deletions(-)
include/net/tcp.h +0 -33
@@ -2228,34 +2228,6 @@ int __tcp_bpf_recvmsg(struct sock *sk, struct sk_psock *psock,#endif /* CONFIG_NET_SOCK_MSG */ #ifdef CONFIG_CGROUP_BPF -/* Copy the listen sk's HDR_OPT_CB flags to its child. -* -* During 3-Way-HandShake, the synack is usually sent from -* the listen sk with the HDR_OPT_CB flags set so that -* bpf-prog will be called to write the BPF hdr option. -* -* In fastopen, the child sk is used to send synack instead -* of the listen sk. Thus, inheriting the HDR_OPT_CB flags -* from the listen sk gives the bpf-prog a chance to write -* BPF hdr option in the synack pkt during fastopen. -* -* Both fastopen and non-fastopen child will inherit the -* HDR_OPT_CB flags to keep the bpf-prog having a consistent -* behavior when deciding to clear this cb flags (or not) -* during the PASSIVE_ESTABLISHED_CB. -* -* In the future, other cb flags could be inherited here also. -*/ -static inline void bpf_skops_init_child (const struct sock *sk, -struct sock *child) -{ -tcp_sk (child)->bpf_sock_ops_cb_flags = -tcp_sk (sk)->bpf_sock_ops_cb_flags & - (BPF_SOCK_OPS_PARSE_ALL_HDR_OPT_CB_FLAG | -BPF_SOCK_OPS_PARSE_UNKNOWN_HDR_OPT_CB_FLAG | -BPF_SOCK_OPS_WRITE_HDR_OPT_CB_FLAG ); -} -static inline void bpf_skops_init_skb (struct bpf_sock_ops_kern *skops,struct sk_buff *skb,unsigned int end_offset)
@@ -2264,11 +2236,6 @@ static inline void bpf_skops_init_skb(struct bpf_sock_ops_kern *skops, skops->skb_data_end = skb->data + end_offset; }#else -static inline void bpf_skops_init_child (const struct sock *sk, -struct sock *child) -{ -} -static inline void bpf_skops_init_skb (struct bpf_sock_ops_kern *skops,struct sk_buff *skb,unsigned int end_offset)
net/ipv4/tcp_minisocks.c +0 -1
@@ -548,7 +548,6 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, newtp->fastopen_req =NULL ;RCU_INIT_POINTER (newtp->fastopen_rsk,NULL ); -bpf_skops_init_child (sk, newsk);tcp_bpf_clone (sk, newsk);__TCP_INC_STATS (sock_net (sk),TCP_MIB_PASSIVEOPENS );
Signature · 1 line
2.24.1
PATCH 2/2 bpf: selftest: Ensure the child sk inherited all bpf_sock_ops_cb_flags
In reply to Martin KaFai LauThis patch adds a test to ensure the child sk inherited everything from the bpf_sock_ops_cb_flags of the listen sk:
1. Sets one more cb_flags (BPF_SOCK_OPS_STATE_CB_FLAG) to the listen sk in test_tcp_hdr_options.c
2. Saves the skops->bpf_sock_ops_cb_flags when handling the newly established passive connection
3. CHECK() it is the same as the listen sk
This also covers the fastopen case as the existing test_tcp_hdr_options.c does.
--- .../selftests/bpf/prog_tests/tcp_hdr_options.c | 12 ++++++++++++ .../selftests/bpf/progs/test_misc_tcp_hdr_options.c | 4 ++-- .../selftests/bpf/progs/test_tcp_hdr_options.c | 7 +++++-- tools/testing/selftests/bpf/test_tcp_hdr_options.h | 5 +++-- 4 files changed, 22 insertions(+), 6 deletions(-)
tools/testing/selftests/bpf/prog_tests/tcp_hdr_options.c +12 -0
@@ -264,9 +264,19 @@ static int check_error_linum(const struct sk_fds *sk_fds)static void check_hdr_and_close_fds (struct sk_fds *sk_fds) { +const __u32 expected_inherit_cb_flags = +BPF_SOCK_OPS_PARSE_UNKNOWN_HDR_OPT_CB_FLAG | +BPF_SOCK_OPS_WRITE_HDR_OPT_CB_FLAG | +BPF_SOCK_OPS_STATE_CB_FLAG ; +if (sk_fds_shutdown (sk_fds)) goto check_linum; +if (CHECK (expected_inherit_cb_flags != skel->bss->inherit_cb_flags, +"Unexpected inherit_cb_flags" ,"0x%x != 0x%x\n" , + skel->bss->inherit_cb_flags, expected_inherit_cb_flags)) + goto check_linum; +if (check_hdr_stg (&exp_passive_hdr_stg, sk_fds->passive_fd,"passive_hdr_stg" )) goto check_linum;
@@ -321,6 +331,8 @@ static void reset_test(void)memset (&skel->bss->active_estab_in,0 , optsize);memset (&skel->bss->active_fin_in,0 , optsize); + skel->bss->inherit_cb_flags =0 ; + skel->data->test_kind =TCPOPT_EXP ; skel->data->test_magic =0xeB9F ;
tools/testing/selftests/bpf/progs/test_misc_tcp_hdr_options.c +2 -2
@@ -304,10 +304,10 @@ int misc_estab(struct bpf_sock_ops *skops) passive_lport_n =__bpf_htons (passive_lport_h);bpf_setsockopt (skops,SOL_TCP ,TCP_SAVE_SYN , &true_val,sizeof (true_val)); -set_hdr_cb_flags (skops); +set_hdr_cb_flags (skops,0 );break ;case BPF_SOCK_OPS_TCP_CONNECT_CB : -set_hdr_cb_flags (skops); +set_hdr_cb_flags (skops,0 );break ;case BPF_SOCK_OPS_PARSE_HDR_OPT_CB :return handle_parse_hdr (skops);
tools/testing/selftests/bpf/progs/test_tcp_hdr_options.c +5 -2
@@ -21,6 +21,7 @@__u8 test_kind =TCPOPT_EXP ;__u16 test_magic =0xeB9F ; +__u32 inherit_cb_flags =0 ;struct bpf_test_option passive_synack_out = {};struct bpf_test_option passive_fin_out = {};
@@ -467,6 +468,8 @@ static int handle_passive_estab(struct bpf_sock_ops *skops)struct tcphdr *th;int err; + inherit_cb_flags = skops->bpf_sock_ops_cb_flags; + err =load_option (skops, &passive_estab_in, true);if (err == -ENOENT ) {/* saved_syn is not found. It was in syncookie mode.
@@ -600,10 +603,10 @@ int estab(struct bpf_sock_ops *skops)case BPF_SOCK_OPS_TCP_LISTEN_CB :bpf_setsockopt (skops,SOL_TCP ,TCP_SAVE_SYN , &true_val,sizeof (true_val)); -set_hdr_cb_flags (skops); +set_hdr_cb_flags (skops,BPF_SOCK_OPS_STATE_CB_FLAG );break ;case BPF_SOCK_OPS_TCP_CONNECT_CB : -set_hdr_cb_flags (skops); +set_hdr_cb_flags (skops,0 );break ;case BPF_SOCK_OPS_PARSE_HDR_OPT_CB :return handle_parse_hdr (skops);
tools/testing/selftests/bpf/test_tcp_hdr_options.h +3 -2
@@ -110,12 +110,13 @@ static inline void clear_hdr_cb_flags(struct bpf_sock_ops *skops) BPF_SOCK_OPS_WRITE_HDR_OPT_CB_FLAG)); } -static inline void set_hdr_cb_flags (struct bpf_sock_ops *skops) +static inline void set_hdr_cb_flags (struct bpf_sock_ops *skops,__u32 extra) {bpf_sock_ops_cb_flags_set (skops, skops->bpf_sock_ops_cb_flags |BPF_SOCK_OPS_PARSE_UNKNOWN_HDR_OPT_CB_FLAG | -BPF_SOCK_OPS_WRITE_HDR_OPT_CB_FLAG ); +BPF_SOCK_OPS_WRITE_HDR_OPT_CB_FLAG | + extra); }static inline void clear_parse_all_hdr_cb_flags (struct bpf_sock_ops *skops)
Signature · 1 line
2.24.1
Quoted 3 lines · Martin KaFai Lau
This set fixes an issue that the
bpf_skops_init_child()unnecessarily limited the child sk from inheriting allbpf_sock_ops_cb_flagsof the listen sk. It also adds a test to check that.
Thank you, that fixed it for me!
This series was applied to bpf/bpf-next.git (refs/heads/master):
Hello:
This series was applied to bpf/bpf-next.git (refs/heads/master):
Quoted 9 lines · On Thu, 1 Oct 2020 18:34:42 -0700 you wrote:
This set fixes an issue that the
bpf_skops_init_child()unnecessarily limited the child sk from inheriting allbpf_sock_ops_cb_flagsof the listen sk. It also adds a test to check that.Martin KaFai Lau (2):
bpf: tcp: Do not limitcb_flagswhen creating child sk from listen sk
bpf: selftest: Ensure the child sk inherited allbpf_sock_ops_cb_flags[...]
Here is the summary with links:
- [bpf-next,1/2] bpf: tcp: Do not limit cb_flags when creating child sk from listen sk
https://git.kernel.org/bpf/bpf-next/c/82f45c6c4a70 - [bpf-next,2/2] bpf: selftest: Ensure the child sk inherited all bpf_sock_ops_cb_flags https://git.kernel.org/bpf/bpf-next/c/96d46c508506
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html