* [PATCH net-next v2 1/2] tcp_metrics: add UAPI to the header guard
2024-06-27 21:35 [PATCH net-next v2 0/2] tcp_metrics: add netlink protocol spec in YAML Jakub Kicinski
@ 2024-06-27 21:35 ` Jakub Kicinski
2024-06-28 10:23 ` Donald Hunter
2024-06-27 21:35 ` [PATCH net-next v2 2/2] tcp_metrics: add netlink protocol spec in YAML Jakub Kicinski
2024-07-01 8:50 ` [PATCH net-next v2 0/2] " patchwork-bot+netdevbpf
2 siblings, 1 reply; 6+ messages in thread
From: Jakub Kicinski @ 2024-06-27 21:35 UTC (permalink / raw)
To: davem; +Cc: netdev, edumazet, pabeni, donald.hunter, Jakub Kicinski
tcp_metrics' header lacks the customary _UAPI in the header guard.
This makes YNL build rules work less seamlessly.
We can easily fix that on YNL side, but this could also be
problematic if we ever needed to create a kernel-only tcp_metrics.h.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
include/uapi/linux/tcp_metrics.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/include/uapi/linux/tcp_metrics.h b/include/uapi/linux/tcp_metrics.h
index 7cb4a172feed..c48841076998 100644
--- a/include/uapi/linux/tcp_metrics.h
+++ b/include/uapi/linux/tcp_metrics.h
@@ -1,8 +1,8 @@
/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
/* tcp_metrics.h - TCP Metrics Interface */
-#ifndef _LINUX_TCP_METRICS_H
-#define _LINUX_TCP_METRICS_H
+#ifndef _UAPI_LINUX_TCP_METRICS_H
+#define _UAPI_LINUX_TCP_METRICS_H
#include <linux/types.h>
@@ -58,4 +58,4 @@ enum {
#define TCP_METRICS_CMD_MAX (__TCP_METRICS_CMD_MAX - 1)
-#endif /* _LINUX_TCP_METRICS_H */
+#endif /* _UAPI_LINUX_TCP_METRICS_H */
--
2.45.2
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH net-next v2 2/2] tcp_metrics: add netlink protocol spec in YAML
2024-06-27 21:35 [PATCH net-next v2 0/2] tcp_metrics: add netlink protocol spec in YAML Jakub Kicinski
2024-06-27 21:35 ` [PATCH net-next v2 1/2] tcp_metrics: add UAPI to the header guard Jakub Kicinski
@ 2024-06-27 21:35 ` Jakub Kicinski
2024-06-28 10:24 ` Donald Hunter
2024-07-01 8:50 ` [PATCH net-next v2 0/2] " patchwork-bot+netdevbpf
2 siblings, 1 reply; 6+ messages in thread
From: Jakub Kicinski @ 2024-06-27 21:35 UTC (permalink / raw)
To: davem; +Cc: netdev, edumazet, pabeni, donald.hunter, Jakub Kicinski
Add a protocol spec for tcp_metrics, so that it's accessible via YNL.
Useful at the very least for testing fixes.
In this episode of "10,000 ways to complicate netlink" the metric
nest has defines which are off by 1. iproute2 does:
struct rtattr *m[TCP_METRIC_MAX + 1 + 1];
parse_rtattr_nested(m, TCP_METRIC_MAX + 1, a);
for (i = 0; i < TCP_METRIC_MAX + 1; i++) {
// ...
attr = m[i + 1];
This is too weird to support in YNL, add a new set of defines
with _correct_ values to the official kernel header.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
CC: donald.hunter@gmail.com
v2:
- remove unused unspec op and attr (TCP_METRICS_ATTR_UNSPEC and
TCP_METRICS_CMD_UNSPEC exist, but we can't generate uAPI, anyway)
- fix typos
- s/_/-/ on metric names
- correct doc on rttvar-us (like rttvar there are 2 fractional bits)
---
Documentation/netlink/specs/tcp_metrics.yaml | 169 +++++++++++++++++++
include/uapi/linux/tcp_metrics.h | 16 ++
tools/net/ynl/Makefile.deps | 1 +
3 files changed, 186 insertions(+)
create mode 100644 Documentation/netlink/specs/tcp_metrics.yaml
diff --git a/Documentation/netlink/specs/tcp_metrics.yaml b/Documentation/netlink/specs/tcp_metrics.yaml
new file mode 100644
index 000000000000..1bd94f43e526
--- /dev/null
+++ b/Documentation/netlink/specs/tcp_metrics.yaml
@@ -0,0 +1,169 @@
+# SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
+
+name: tcp_metrics
+
+protocol: genetlink-legacy
+
+doc: |
+ Management interface for TCP metrics.
+
+c-family-name: tcp-metrics-genl-name
+c-version-name: tcp-metrics-genl-version
+max-by-define: true
+kernel-policy: global
+
+definitions:
+ -
+ name: tcp-fastopen-cookie-max
+ type: const
+ value: 16
+
+attribute-sets:
+ -
+ name: tcp-metrics
+ name-prefix: tcp-metrics-attr-
+ attributes:
+ -
+ name: addr-ipv4
+ type: u32
+ byte-order: big-endian
+ display-hint: ipv4
+ -
+ name: addr-ipv6
+ type: binary
+ checks:
+ min-len: 16
+ byte-order: big-endian
+ display-hint: ipv6
+ -
+ name: age
+ type: u64
+ -
+ name: tw-tsval
+ type: u32
+ doc: unused
+ -
+ name: tw-ts-stamp
+ type: s32
+ doc: unused
+ -
+ name: vals
+ type: nest
+ nested-attributes: metrics
+ -
+ name: fopen-mss
+ type: u16
+ -
+ name: fopen-syn-drops
+ type: u16
+ -
+ name: fopen-syn-drop-ts
+ type: u64
+ -
+ name: fopen-cookie
+ type: binary
+ checks:
+ min-len: tcp-fastopen-cookie-max
+ -
+ name: saddr-ipv4
+ type: u32
+ byte-order: big-endian
+ display-hint: ipv4
+ -
+ name: saddr-ipv6
+ type: binary
+ checks:
+ min-len: 16
+ byte-order: big-endian
+ display-hint: ipv6
+ -
+ name: pad
+ type: pad
+
+ -
+ name: metrics
+ # Intentionally don't define the name-prefix, see below.
+ doc: |
+ Attributes with metrics. Note that the values here do not match
+ the TCP_METRIC_* defines in the kernel, because kernel defines
+ are off-by one (e.g. rtt is defined as enum 0, while netlink carries
+ attribute type 1).
+ attributes:
+ -
+ name: rtt
+ type: u32
+ doc: |
+ Round Trip Time (RTT), in msecs with 3 bits fractional
+ (left-shift by 3 to get the msec value).
+ -
+ name: rttvar
+ type: u32
+ doc: |
+ Round Trip Time VARiance (RTT), in msecs with 2 bits fractional
+ (left-shift by 2 to get the msec value).
+ -
+ name: ssthresh
+ type: u32
+ doc: Slow Start THRESHold.
+ -
+ name: cwnd
+ type: u32
+ doc: Congestion Window.
+ -
+ name: reodering
+ type: u32
+ doc: Reodering metric.
+ -
+ name: rtt-us
+ type: u32
+ doc: |
+ Round Trip Time (RTT), in usecs, with 3 bits fractional
+ (left-shift by 3 to get the msec value).
+ -
+ name: rttvar-us
+ type: u32
+ doc: |
+ Round Trip Time (RTT), in usecs, with 2 bits fractional
+ (left-shift by 3 to get the msec value).
+
+operations:
+ list:
+ -
+ name: get
+ doc: Retrieve metrics.
+ attribute-set: tcp-metrics
+
+ dont-validate: [ strict, dump ]
+
+ do:
+ request: &sel_attrs
+ attributes:
+ - addr-ipv4
+ - addr-ipv6
+ - saddr-ipv4
+ - saddr-ipv6
+ reply: &all_attrs
+ attributes:
+ - addr-ipv4
+ - addr-ipv6
+ - saddr-ipv4
+ - saddr-ipv6
+ - age
+ - vals
+ - fopen-mss
+ - fopen-syn-drops
+ - fopen-syn-drop-ts
+ - fopen-cookie
+ dump:
+ reply: *all_attrs
+
+ -
+ name: del
+ doc: Delete metrics.
+ attribute-set: tcp-metrics
+
+ dont-validate: [ strict, dump ]
+ flags: [ admin-perm ]
+
+ do:
+ request: *sel_attrs
diff --git a/include/uapi/linux/tcp_metrics.h b/include/uapi/linux/tcp_metrics.h
index c48841076998..927c735a5b0e 100644
--- a/include/uapi/linux/tcp_metrics.h
+++ b/include/uapi/linux/tcp_metrics.h
@@ -27,6 +27,22 @@ enum tcp_metric_index {
#define TCP_METRIC_MAX (__TCP_METRIC_MAX - 1)
+/* Re-define enum tcp_metric_index, again, using the values carried
+ * as netlink attribute types.
+ */
+enum {
+ TCP_METRICS_A_METRICS_RTT = 1,
+ TCP_METRICS_A_METRICS_RTTVAR,
+ TCP_METRICS_A_METRICS_SSTHRESH,
+ TCP_METRICS_A_METRICS_CWND,
+ TCP_METRICS_A_METRICS_REODERING,
+ TCP_METRICS_A_METRICS_RTT_US,
+ TCP_METRICS_A_METRICS_RTTVAR_US,
+
+ __TCP_METRICS_A_METRICS_MAX
+};
+#define TCP_METRICS_A_METRICS_MAX (__TCP_METRICS_A_METRICS_MAX - 1)
+
enum {
TCP_METRICS_ATTR_UNSPEC,
TCP_METRICS_ATTR_ADDR_IPV4, /* u32 */
diff --git a/tools/net/ynl/Makefile.deps b/tools/net/ynl/Makefile.deps
index f4e8eb79c1b8..33399fbe9851 100644
--- a/tools/net/ynl/Makefile.deps
+++ b/tools/net/ynl/Makefile.deps
@@ -25,3 +25,4 @@ CFLAGS_nfsd:=$(call get_hdr_inc,_LINUX_NFSD_NETLINK_H,nfsd_netlink.h)
CFLAGS_ovs_datapath:=$(call get_hdr_inc,__LINUX_OPENVSWITCH_H,openvswitch.h)
CFLAGS_ovs_flow:=$(call get_hdr_inc,__LINUX_OPENVSWITCH_H,openvswitch.h)
CFLAGS_ovs_vport:=$(call get_hdr_inc,__LINUX_OPENVSWITCH_H,openvswitch.h)
+CFLAGS_tcp_metrics:=$(call get_hdr_inc,_LINUX_TCP_METRICS_H,tcp_metrics.h)
--
2.45.2
^ permalink raw reply related [flat|nested] 6+ messages in thread