* [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops
@ 2024-05-01 7:43 Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 2/3] Allow to write tp->snd_cwnd_stamp in bpf_tcp_ca Miao Xu
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Miao Xu @ 2024-05-01 7:43 UTC (permalink / raw
To: Eric Dumazet, David S . Miller, Jakub Kicinski, Paolo Abeni,
David Ahern, Martin Lau
Cc: netdev, bpf, Miao Xu
This patch adds two new arguments for cong_control of struct
tcp_congestion_ops:
- ack
- flag
These two arguments are inherited from the caller tcp_cong_control in
tcp_intput.c. One use case of them is to update cwnd and pacing rate
inside cong_control based on the info they provide. For example, the
flag can be used to decide if it is the right time to raise or reduce a
sender's cwnd.
Reviewed-by: Eric Dumazet <edumazet@google.com>
--
Changes in v2:
* Split the v1 patch into 2 separate patches. In particular, spin out
bpf_tcp_ca.c as a separate patch because it is bpf specific.
Signed-off-by: Miao Xu <miaxu@meta.com>
---
include/net/tcp.h | 2 +-
net/ipv4/bpf_tcp_ca.c | 3 ++-
net/ipv4/tcp_bbr.c | 2 +-
net/ipv4/tcp_input.c | 2 +-
4 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index fe98fb01879b..7294da8fb780 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1172,7 +1172,7 @@ struct tcp_congestion_ops {
/* call when packets are delivered to update cwnd and pacing rate,
* after all the ca_state processing. (optional)
*/
- void (*cong_control)(struct sock *sk, const struct rate_sample *rs);
+ void (*cong_control)(struct sock *sk, u32 ack, int flag, const struct rate_sample *rs);
/* new value of cwnd after loss (required) */
diff --git a/net/ipv4/bpf_tcp_ca.c b/net/ipv4/bpf_tcp_ca.c
index 7f518ea5f4ac..6bd7f8db189a 100644
--- a/net/ipv4/bpf_tcp_ca.c
+++ b/net/ipv4/bpf_tcp_ca.c
@@ -307,7 +307,8 @@ static u32 bpf_tcp_ca_min_tso_segs(struct sock *sk)
return 0;
}
-static void bpf_tcp_ca_cong_control(struct sock *sk, const struct rate_sample *rs)
+static void bpf_tcp_ca_cong_control(struct sock *sk, u32 ack, int flag,
+ const struct rate_sample *rs)
{
}
diff --git a/net/ipv4/tcp_bbr.c b/net/ipv4/tcp_bbr.c
index 7e52ab24e40a..760941e55153 100644
--- a/net/ipv4/tcp_bbr.c
+++ b/net/ipv4/tcp_bbr.c
@@ -1024,7 +1024,7 @@ static void bbr_update_model(struct sock *sk, const struct rate_sample *rs)
bbr_update_gains(sk);
}
-__bpf_kfunc static void bbr_main(struct sock *sk, const struct rate_sample *rs)
+__bpf_kfunc static void bbr_main(struct sock *sk, u32 ack, int flag, const struct rate_sample *rs)
{
struct bbr *bbr = inet_csk_ca(sk);
u32 bw;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 53e1150f706f..23ccfc7b1d3c 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3541,7 +3541,7 @@ static void tcp_cong_control(struct sock *sk, u32 ack, u32 acked_sacked,
const struct inet_connection_sock *icsk = inet_csk(sk);
if (icsk->icsk_ca_ops->cong_control) {
- icsk->icsk_ca_ops->cong_control(sk, rs);
+ icsk->icsk_ca_ops->cong_control(sk, ack, flag, rs);
return;
}
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH net-next v2 2/3] Allow to write tp->snd_cwnd_stamp in bpf_tcp_ca
2024-05-01 7:43 [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Miao Xu
@ 2024-05-01 7:43 ` Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 3/3] Add test for the use of new args in cong_control Miao Xu
2024-05-01 13:26 ` [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Jakub Kicinski
2 siblings, 0 replies; 6+ messages in thread
From: Miao Xu @ 2024-05-01 7:43 UTC (permalink / raw
To: Eric Dumazet, David S . Miller, Jakub Kicinski, Paolo Abeni,
David Ahern, Martin Lau
Cc: netdev, bpf, Miao Xu
This patch allows the write of tp->snd_cwnd_stamp in a bpf tcp
ca program. An use case of writing this field is to keep track
of the time whenever tp->snd_cwnd is raised or reduced inside
the `cong_control` callback.
Reviewed-by: Eric Dumazet <edumazet@google.com>
--
Changes in v2:
* None. It is a spinout from the original 1st patch.
Signed-off-by: Miao Xu <miaxu@meta.com>
---
net/ipv4/bpf_tcp_ca.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/net/ipv4/bpf_tcp_ca.c b/net/ipv4/bpf_tcp_ca.c
index 6bd7f8db189a..18227757ec0c 100644
--- a/net/ipv4/bpf_tcp_ca.c
+++ b/net/ipv4/bpf_tcp_ca.c
@@ -107,6 +107,9 @@ static int bpf_tcp_ca_btf_struct_access(struct bpf_verifier_log *log,
case offsetof(struct tcp_sock, snd_cwnd_cnt):
end = offsetofend(struct tcp_sock, snd_cwnd_cnt);
break;
+ case offsetof(struct tcp_sock, snd_cwnd_stamp):
+ end = offsetofend(struct tcp_sock, snd_cwnd_stamp);
+ break;
case offsetof(struct tcp_sock, snd_ssthresh):
end = offsetofend(struct tcp_sock, snd_ssthresh);
break;
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH net-next v2 3/3] Add test for the use of new args in cong_control
2024-05-01 7:43 [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 2/3] Allow to write tp->snd_cwnd_stamp in bpf_tcp_ca Miao Xu
@ 2024-05-01 7:43 ` Miao Xu
2024-05-01 20:19 ` Martin KaFai Lau
2024-05-01 13:26 ` [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Jakub Kicinski
2 siblings, 1 reply; 6+ messages in thread
From: Miao Xu @ 2024-05-01 7:43 UTC (permalink / raw
To: Eric Dumazet, David S . Miller, Jakub Kicinski, Paolo Abeni,
David Ahern, Martin Lau
Cc: netdev, bpf, Miao Xu
This patch adds a selftest to show the usage of the new arguments in
cong_control. For simplicity's sake, the testing example reuses cubic's
kernel functions.
--
Changes in v2:
* Added highlights to explain major differences between the bpf program
and tcp_cubic.c.
* bpf_tcp_helpers.h should not be further extended, so remove the
dependency on this file. Use vmlinux.h instead.
* Minor changes such as indentation.
Signed-off-by: Miao Xu <miaxu@meta.com>
---
.../bpf/progs/bpf_cubic_cong_control.c | 207 ++++++++++++++++++
.../selftests/bpf/progs/bpf_tracing_net.h | 10 +
2 files changed, 217 insertions(+)
create mode 100644 tools/testing/selftests/bpf/progs/bpf_cubic_cong_control.c
diff --git a/tools/testing/selftests/bpf/progs/bpf_cubic_cong_control.c b/tools/testing/selftests/bpf/progs/bpf_cubic_cong_control.c
new file mode 100644
index 000000000000..7ec9da0356c3
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/bpf_cubic_cong_control.c
@@ -0,0 +1,207 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+/* Highlights:
+ * 1. The major difference between this bpf program and tcp_cubic.c
+ * is that this bpf program relies on `cong_control` rather than
+ * `cong_avoid` in the struct tcp_congestion_ops.
+ * 2. Logic such as tcp_cwnd_reduction, tcp_cong_avoid, and
+ * tcp_update_pacing_rate is bypassed when `cong_control` is
+ * defined, so moving these logic to `cong_control`.
+ * 3. WARNING: This bpf program is NOT the same as tcp_cubic.c.
+ * The main purpose is to show use cases of the arguments in
+ * `cong_control`. For simplicity's sake, it reuses tcp cubic's
+ * kernel functions.
+ */
+
+#include "vmlinux.h"
+
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+#include "bpf_tracing_net.h"
+
+#define BPF_STRUCT_OPS(name, args...) \
+SEC("struct_ops/"#name) \
+BPF_PROG(name, args)
+
+
+#define min(a, b) ((a) < (b) ? (a) : (b))
+#define max(a, b) ((a) > (b) ? (a) : (b))
+
+static __always_inline struct inet_connection_sock *inet_csk(const struct sock *sk)
+{
+ return (struct inet_connection_sock *)sk;
+}
+
+static __always_inline struct tcp_sock *tcp_sk(const struct sock *sk)
+{
+ return (struct tcp_sock *)sk;
+}
+
+static __always_inline bool before(__u32 seq1, __u32 seq2)
+{
+ return (__s32)(seq1-seq2) < 0;
+}
+#define after(seq2, seq1) before(seq1, seq2)
+
+char _license[] SEC("license") = "GPL";
+
+extern void cubictcp_init(struct sock *sk) __ksym;
+extern void cubictcp_cwnd_event(struct sock *sk, enum tcp_ca_event event) __ksym;
+extern __u32 cubictcp_recalc_ssthresh(struct sock *sk) __ksym;
+extern void cubictcp_state(struct sock *sk, __u8 new_state) __ksym;
+extern __u32 tcp_reno_undo_cwnd(struct sock *sk) __ksym;
+extern void cubictcp_acked(struct sock *sk, const struct ack_sample *sample) __ksym;
+extern void cubictcp_cong_avoid(struct sock *sk, __u32 ack, __u32 acked) __ksym;
+
+
+void BPF_STRUCT_OPS(bpf_cubic_init, struct sock *sk)
+{
+ cubictcp_init(sk);
+}
+
+void BPF_STRUCT_OPS(bpf_cubic_cwnd_event, struct sock *sk, enum tcp_ca_event event)
+{
+ cubictcp_cwnd_event(sk, event);
+}
+
+#define USEC_PER_SEC 1000000UL
+#define TCP_PACING_SS_RATIO (200)
+#define TCP_PACING_CA_RATIO (120)
+#define TCP_REORDERING (12)
+#define likely(x) (__builtin_expect(!!(x), 1))
+
+static __always_inline __u64 div64_u64(__u64 dividend, __u64 divisor)
+{
+ return dividend / divisor;
+}
+
+static __always_inline void tcp_update_pacing_rate(struct sock *sk)
+{
+ const struct tcp_sock *tp = tcp_sk(sk);
+ __u64 rate;
+
+ /* set sk_pacing_rate to 200 % of current rate (mss * cwnd / srtt) */
+ rate = (__u64)tp->mss_cache * ((USEC_PER_SEC / 100) << 3);
+
+ /* current rate is (cwnd * mss) / srtt
+ * In Slow Start [1], set sk_pacing_rate to 200 % the current rate.
+ * In Congestion Avoidance phase, set it to 120 % the current rate.
+ *
+ * [1] : Normal Slow Start condition is (tp->snd_cwnd < tp->snd_ssthresh)
+ * If snd_cwnd >= (tp->snd_ssthresh / 2), we are approaching
+ * end of slow start and should slow down.
+ */
+ if (tp->snd_cwnd < tp->snd_ssthresh / 2)
+ rate *= TCP_PACING_SS_RATIO;
+ else
+ rate *= TCP_PACING_CA_RATIO;
+
+ rate *= max(tp->snd_cwnd, tp->packets_out);
+
+ if (likely(tp->srtt_us))
+ rate = div64_u64(rate, (__u64)tp->srtt_us);
+
+ sk->sk_pacing_rate = min(rate, sk->sk_max_pacing_rate);
+}
+
+static __always_inline void tcp_cwnd_reduction(
+ struct sock *sk,
+ int newly_acked_sacked,
+ int newly_lost,
+ int flag) {
+ struct tcp_sock *tp = tcp_sk(sk);
+ int sndcnt = 0;
+ __u32 pkts_in_flight = tp->packets_out - (tp->sacked_out + tp->lost_out) + tp->retrans_out;
+ int delta = tp->snd_ssthresh - pkts_in_flight;
+
+ if (newly_acked_sacked <= 0 || !tp->prior_cwnd)
+ return;
+
+ __u32 prr_delivered = tp->prr_delivered + newly_acked_sacked;
+
+ if (delta < 0) {
+ __u64 dividend =
+ (__u64)tp->snd_ssthresh * prr_delivered + tp->prior_cwnd - 1;
+ sndcnt = (__u32)div64_u64(dividend, (__u64)tp->prior_cwnd) - tp->prr_out;
+ } else {
+ sndcnt = max(prr_delivered - tp->prr_out, newly_acked_sacked);
+ if (flag & FLAG_SND_UNA_ADVANCED && !newly_lost)
+ sndcnt++;
+ sndcnt = min(delta, sndcnt);
+ }
+ /* Force a fast retransmit upon entering fast recovery */
+ sndcnt = max(sndcnt, (tp->prr_out ? 0 : 1));
+ tp->snd_cwnd = pkts_in_flight + sndcnt;
+}
+
+/* Decide wheather to run the increase function of congestion control. */
+static __always_inline bool tcp_may_raise_cwnd(
+ const struct sock *sk,
+ const int flag) {
+ if (tcp_sk(sk)->reordering > TCP_REORDERING)
+ return flag & FLAG_FORWARD_PROGRESS;
+
+ return flag & FLAG_DATA_ACKED;
+}
+
+void BPF_STRUCT_OPS(bpf_cubic_cong_control, struct sock *sk, __u32 ack, int flag,
+ const struct rate_sample *rs)
+{
+ struct tcp_sock *tp = tcp_sk(sk);
+
+ if (((1<<TCP_CA_CWR) | (1<<TCP_CA_Recovery)) &
+ (1 << inet_csk(sk)->icsk_ca_state)) {
+ /* Reduce cwnd if state mandates */
+ tcp_cwnd_reduction(sk, rs->acked_sacked, rs->losses, flag);
+
+ if (!before(tp->snd_una, tp->high_seq)) {
+ /* Reset cwnd to ssthresh in CWR or Recovery (unless it's undone) */
+ if (tp->snd_ssthresh < TCP_INFINITE_SSTHRESH &&
+ inet_csk(sk)->icsk_ca_state == TCP_CA_CWR) {
+ tp->snd_cwnd = tp->snd_ssthresh;
+ tp->snd_cwnd_stamp = tcp_jiffies32;
+ }
+ // __cwnd_event(sk, CA_EVENT_COMPLETE_CWR);
+ }
+ } else if (tcp_may_raise_cwnd(sk, flag)) {
+ /* Advance cwnd if state allows */
+ cubictcp_cong_avoid(sk, ack, rs->acked_sacked);
+ tp->snd_cwnd_stamp = tcp_jiffies32;
+ }
+
+ tcp_update_pacing_rate(sk);
+}
+
+__u32 BPF_STRUCT_OPS(bpf_cubic_recalc_ssthresh, struct sock *sk)
+{
+ return cubictcp_recalc_ssthresh(sk);
+}
+
+void BPF_STRUCT_OPS(bpf_cubic_state, struct sock *sk, __u8 new_state)
+{
+ cubictcp_state(sk, new_state);
+}
+
+void BPF_STRUCT_OPS(bpf_cubic_acked, struct sock *sk,
+ const struct ack_sample *sample)
+{
+ cubictcp_acked(sk, sample);
+}
+
+__u32 BPF_STRUCT_OPS(bpf_cubic_undo_cwnd, struct sock *sk)
+{
+ return tcp_reno_undo_cwnd(sk);
+}
+
+
+SEC(".struct_ops")
+struct tcp_congestion_ops cubic = {
+ .init = (void *)bpf_cubic_init,
+ .ssthresh = (void *)bpf_cubic_recalc_ssthresh,
+ .cong_control = (void *)bpf_cubic_cong_control,
+ .set_state = (void *)bpf_cubic_state,
+ .undo_cwnd = (void *)bpf_cubic_undo_cwnd,
+ .cwnd_event = (void *)bpf_cubic_cwnd_event,
+ .pkts_acked = (void *)bpf_cubic_acked,
+ .name = "bpf_cubic",
+};
diff --git a/tools/testing/selftests/bpf/progs/bpf_tracing_net.h b/tools/testing/selftests/bpf/progs/bpf_tracing_net.h
index 7001965d1cc3..f9ec630dfcd5 100644
--- a/tools/testing/selftests/bpf/progs/bpf_tracing_net.h
+++ b/tools/testing/selftests/bpf/progs/bpf_tracing_net.h
@@ -80,6 +80,14 @@
#define TCP_INFINITE_SSTHRESH 0x7fffffff
#define TCP_PINGPONG_THRESH 3
+#define FLAG_DATA_ACKED 0x04 /* This ACK acknowledged new data. */
+#define FLAG_SYN_ACKED 0x10 /* This ACK acknowledged SYN. */
+#define FLAG_DATA_SACKED 0x20 /* New SACK. */
+#define FLAG_SND_UNA_ADVANCED \
+ 0x400 /* Snd_una was changed (!= FLAG_DATA_ACKED) */
+#define FLAG_ACKED (FLAG_DATA_ACKED | FLAG_SYN_ACKED)
+#define FLAG_FORWARD_PROGRESS (FLAG_ACKED | FLAG_DATA_SACKED)
+
#define fib_nh_dev nh_common.nhc_dev
#define fib_nh_gw_family nh_common.nhc_gw_family
#define fib_nh_gw6 nh_common.nhc_gw.ipv6
@@ -119,4 +127,6 @@
#define tw_v6_daddr __tw_common.skc_v6_daddr
#define tw_v6_rcv_saddr __tw_common.skc_v6_rcv_saddr
+#define tcp_jiffies32 ((__u32)bpf_jiffies64())
+
#endif
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops
2024-05-01 7:43 [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 2/3] Allow to write tp->snd_cwnd_stamp in bpf_tcp_ca Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 3/3] Add test for the use of new args in cong_control Miao Xu
@ 2024-05-01 13:26 ` Jakub Kicinski
2 siblings, 0 replies; 6+ messages in thread
From: Jakub Kicinski @ 2024-05-01 13:26 UTC (permalink / raw
To: Miao Xu
Cc: Eric Dumazet, David S . Miller, Paolo Abeni, David Ahern,
Martin Lau, netdev, bpf
On Wed, 1 May 2024 00:43:36 -0700 Miao Xu wrote:
> This patch adds two new arguments for cong_control of struct
> tcp_congestion_ops:
> - ack
> - flag
> These two arguments are inherited from the caller tcp_cong_control in
> tcp_intput.c. One use case of them is to update cwnd and pacing rate
> inside cong_control based on the info they provide. For example, the
> flag can be used to decide if it is the right time to raise or reduce a
> sender's cwnd.
>
> Reviewed-by: Eric Dumazet <edumazet@google.com>
> --
three dashes here ---
> Changes in v2:
> * Split the v1 patch into 2 separate patches. In particular, spin out
> bpf_tcp_ca.c as a separate patch because it is bpf specific.
>
> Signed-off-by: Miao Xu <miaxu@meta.com>
This goes after Eric's review tag.
Looks like you need to adjust one of the BPF selftests:
Error: #29/14 bpf_tcp_ca/tcp_ca_kfunc
Error: #29/14 bpf_tcp_ca/tcp_ca_kfunc
libbpf: extern (func ksym) 'bbr_main': func_proto [213] incompatible with vmlinux [48152]
libbpf: failed to load object 'tcp_ca_kfunc'
libbpf: failed to load BPF skeleton 'tcp_ca_kfunc': -22
test_tcp_ca_kfunc:FAIL:tcp_ca_kfunc__open_and_load unexpected error: -22
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net-next v2 3/3] Add test for the use of new args in cong_control
2024-05-01 7:43 ` [PATCH net-next v2 3/3] Add test for the use of new args in cong_control Miao Xu
@ 2024-05-01 20:19 ` Martin KaFai Lau
2024-05-02 0:39 ` Jakub Kicinski
0 siblings, 1 reply; 6+ messages in thread
From: Martin KaFai Lau @ 2024-05-01 20:19 UTC (permalink / raw
To: Miao Xu
Cc: Eric Dumazet, David S . Miller, Jakub Kicinski, Paolo Abeni,
David Ahern, Martin Lau, netdev, bpf
On 5/1/24 12:43 AM, Miao Xu wrote:
> This patch adds a selftest to show the usage of the new arguments in
> cong_control. For simplicity's sake, the testing example reuses cubic's
> kernel functions.
Jakub, is it ok to target the set for the bpf-next?
The bpf_tcp_ca test failed (Jakub also mentioned). The progs/tcp_ca_kfunc.c
requires changes. The func signature of bbr_main and the BPF_PROG(cong_control,
...) has to be adjusted.
Since it needs a respin, a few nits.
Please add "selftests/bpf:" to the subject line of this patch 3. I think Patch 1
can use "tcp:" and patch 2 can use "bpf: tcp:" also.
Please also add a cover letter, git format-patch --cover-letter ...
[ ... ]
> +void BPF_STRUCT_OPS(bpf_cubic_cong_control, struct sock *sk, __u32 ack, int flag,
> + const struct rate_sample *rs)
> +{
> + struct tcp_sock *tp = tcp_sk(sk);
> +
> + if (((1<<TCP_CA_CWR) | (1<<TCP_CA_Recovery)) &
> + (1 << inet_csk(sk)->icsk_ca_state)) {
> + /* Reduce cwnd if state mandates */
> + tcp_cwnd_reduction(sk, rs->acked_sacked, rs->losses, flag);
> +
> + if (!before(tp->snd_una, tp->high_seq)) {
> + /* Reset cwnd to ssthresh in CWR or Recovery (unless it's undone) */
> + if (tp->snd_ssthresh < TCP_INFINITE_SSTHRESH &&
> + inet_csk(sk)->icsk_ca_state == TCP_CA_CWR) {
> + tp->snd_cwnd = tp->snd_ssthresh;
> + tp->snd_cwnd_stamp = tcp_jiffies32;
> + }
> + // __cwnd_event(sk, CA_EVENT_COMPLETE_CWR);
Remove the commented out code.
> + }
> + } else if (tcp_may_raise_cwnd(sk, flag)) {
> + /* Advance cwnd if state allows */
> + cubictcp_cong_avoid(sk, ack, rs->acked_sacked);
> + tp->snd_cwnd_stamp = tcp_jiffies32;
> + }
> +
> + tcp_update_pacing_rate(sk);
> +}
> +
> +__u32 BPF_STRUCT_OPS(bpf_cubic_recalc_ssthresh, struct sock *sk)
> +{
> + return cubictcp_recalc_ssthresh(sk);
> +}
> +
> +void BPF_STRUCT_OPS(bpf_cubic_state, struct sock *sk, __u8 new_state)
> +{
> + cubictcp_state(sk, new_state);
> +}
> +
> +void BPF_STRUCT_OPS(bpf_cubic_acked, struct sock *sk,
> + const struct ack_sample *sample)
> +{
> + cubictcp_acked(sk, sample);
> +}
> +
> +__u32 BPF_STRUCT_OPS(bpf_cubic_undo_cwnd, struct sock *sk)
> +{
> + return tcp_reno_undo_cwnd(sk);
> +}
> +
> +
> +SEC(".struct_ops")
> +struct tcp_congestion_ops cubic = {
> + .init = (void *)bpf_cubic_init,
> + .ssthresh = (void *)bpf_cubic_recalc_ssthresh,
> + .cong_control = (void *)bpf_cubic_cong_control,
> + .set_state = (void *)bpf_cubic_state,
> + .undo_cwnd = (void *)bpf_cubic_undo_cwnd,
> + .cwnd_event = (void *)bpf_cubic_cwnd_event,
> + .pkts_acked = (void *)bpf_cubic_acked,
> + .name = "bpf_cubic",
nit. It has the same name as the tcp-cc in bpf_cubic.c. Rename it to
"bpf_cc_cubic" ?
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH net-next v2 3/3] Add test for the use of new args in cong_control
2024-05-01 20:19 ` Martin KaFai Lau
@ 2024-05-02 0:39 ` Jakub Kicinski
0 siblings, 0 replies; 6+ messages in thread
From: Jakub Kicinski @ 2024-05-02 0:39 UTC (permalink / raw
To: Martin KaFai Lau
Cc: Miao Xu, Eric Dumazet, David S . Miller, Paolo Abeni, David Ahern,
Martin Lau, netdev, bpf
On Wed, 1 May 2024 13:19:38 -0700 Martin KaFai Lau wrote:
> On 5/1/24 12:43 AM, Miao Xu wrote:
> > This patch adds a selftest to show the usage of the new arguments in
> > cong_control. For simplicity's sake, the testing example reuses cubic's
> > kernel functions.
>
> Jakub, is it ok to target the set for the bpf-next?
SGTM!
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2024-05-02 0:39 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-05-01 7:43 [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 2/3] Allow to write tp->snd_cwnd_stamp in bpf_tcp_ca Miao Xu
2024-05-01 7:43 ` [PATCH net-next v2 3/3] Add test for the use of new args in cong_control Miao Xu
2024-05-01 20:19 ` Martin KaFai Lau
2024-05-02 0:39 ` Jakub Kicinski
2024-05-01 13:26 ` [PATCH net-next v2 1/3] Add new args for cong_control in tcp_congestion_ops Jakub Kicinski
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).