[PATCH net-next 1/2] netlink: specs: tcp_metrics: fix the attribute length checks

Subsystems: networking [general], the rest, yaml netlink (ynl)

COLD24d

5 messages, 3 authors, 24d ago · open the first message on its own page

[PATCH net-next 1/2] netlink: specs: tcp_metrics: fix the attribute length checks

From: Jakub Kicinski <kuba@kernel.org>
Date: 2026-09-12 23:43:42

TCP_FASTOPEN_COOKIE_MAX is the longest cookie we accept, not the shortest
one. Cookies are even sized, from 4 bytes up, and the ones Linux itself
generates are 8 bytes, so a 16 byte minimum declares all but the longest
cookie invalid. The reference policy kept under #if 0 in tcp_metrics.c
spells it as a maximum, which is what .len means for NLA_BINARY.

The IPv6 addresses err the other way. The policy uses
NLA_POLICY_EXACT_LEN() for both, so a request carrying a longer address is
rejected with -ERANGE, even though the spec advertises 16 bytes as a mere
minimum. Were the policy generated from this spec, as kernel-policy: global
promises, the check would turn into NLA_POLICY_MIN_LEN() and start
accepting over-long addresses, of which nla_get_in6_addr() would take the
first 16 bytes.

Nothing generated changes. fopen-cookie is reply-only so it never gets a
policy entry, and exact-len only feeds the policy (and the fixed size array
form, which needs a sub-type).

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 Documentation/netlink/specs/tcp_metrics.yaml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/Documentation/netlink/specs/tcp_metrics.yaml b/Documentation/netlink/specs/tcp_metrics.yaml
index 13144aeed31a..1d365908084d 100644
--- a/Documentation/netlink/specs/tcp_metrics.yaml
+++ b/Documentation/netlink/specs/tcp_metrics.yaml
@@ -32,7 +32,7 @@ kernel-policy: global
         name: addr-ipv6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
         byte-order: big-endian
         display-hint: ipv6
       -
@@ -63,7 +63,7 @@ kernel-policy: global
         name: fopen-cookie
         type: binary
         checks:
-          min-len: tcp-fastopen-cookie-max
+          max-len: tcp-fastopen-cookie-max
       -
         name: saddr-ipv4
         type: u32
@@ -73,7 +73,7 @@ kernel-policy: global
         name: saddr-ipv6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
         byte-order: big-endian
         display-hint: ipv6
       -
-- 
2.55.0

[PATCH net-next 2/2] netlink: specs: tcp_metrics: drop the RTT shift instructions

From: Jakub Kicinski <kuba@kernel.org>
Date: 2026-09-12 23:43:42

All four RTT attributes tell the reader to left-shift, which inflates the
value by 16 to 64 times, and the two usec ones say the result is in msecs.
The attributes carry srtt_us and mdev_us, which hold 3 and 2 fractional
bits. The shift to get the integer part would be a right shift.

Drop the instructions instead of turning them around. The number of
fractional bits is the part worth documenting, whether to shift, divide or
convert to a double is up to the caller.

While at it fix the acronym on the two variance attributes.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 Documentation/netlink/specs/tcp_metrics.yaml | 12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)
diff --git a/Documentation/netlink/specs/tcp_metrics.yaml b/Documentation/netlink/specs/tcp_metrics.yaml
index 1d365908084d..779e8a22b765 100644
--- a/Documentation/netlink/specs/tcp_metrics.yaml
+++ b/Documentation/netlink/specs/tcp_metrics.yaml
@@ -93,14 +93,12 @@ kernel-policy: global
         name: rtt
         type: u32
         doc: |
-          Round Trip Time (RTT), in msecs with 3 bits fractional
-          (left-shift by 3 to get the msec value).
+          Round Trip Time (RTT), in msecs with 3 bits fractional.
       -
         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).
+          Round Trip Time VARiance (RTTVAR), in msecs with 2 bits fractional.
       -
         name: ssthresh
         type: u32
@@ -117,14 +115,12 @@ kernel-policy: global
         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).
+          Round Trip Time (RTT), in usecs, with 3 bits fractional.
       -
         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).
+          Round Trip Time VARiance (RTTVAR), in usecs, with 2 bits fractional.
 
 operations:
   list:
-- 
2.55.0

Re: [PATCH net-next 1/2] netlink: specs: tcp_metrics: fix the attribute length checks

From: Hangbin Liu <hidden>
Date: 2026-09-14 06:41:36

On Sat, Sep 12, 2026 at 04:43:36PM -0700, Jakub Kicinski wrote:
quoted hunk
TCP_FASTOPEN_COOKIE_MAX is the longest cookie we accept, not the shortest
one. Cookies are even sized, from 4 bytes up, and the ones Linux itself
generates are 8 bytes, so a 16 byte minimum declares all but the longest
cookie invalid. The reference policy kept under #if 0 in tcp_metrics.c
spells it as a maximum, which is what .len means for NLA_BINARY.

The IPv6 addresses err the other way. The policy uses
NLA_POLICY_EXACT_LEN() for both, so a request carrying a longer address is
rejected with -ERANGE, even though the spec advertises 16 bytes as a mere
minimum. Were the policy generated from this spec, as kernel-policy: global
promises, the check would turn into NLA_POLICY_MIN_LEN() and start
accepting over-long addresses, of which nla_get_in6_addr() would take the
first 16 bytes.

Nothing generated changes. fopen-cookie is reply-only so it never gets a
policy entry, and exact-len only feeds the policy (and the fixed size array
form, which needs a sub-type).

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 Documentation/netlink/specs/tcp_metrics.yaml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/Documentation/netlink/specs/tcp_metrics.yaml b/Documentation/netlink/specs/tcp_metrics.yaml
index 13144aeed31a..1d365908084d 100644
--- a/Documentation/netlink/specs/tcp_metrics.yaml
+++ b/Documentation/netlink/specs/tcp_metrics.yaml
@@ -32,7 +32,7 @@ kernel-policy: global
         name: addr-ipv6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
         byte-order: big-endian
         display-hint: ipv6
       -
@@ -63,7 +63,7 @@ kernel-policy: global
         name: fopen-cookie
         type: binary
         checks:
-          min-len: tcp-fastopen-cookie-max
+          max-len: tcp-fastopen-cookie-max
       -
         name: saddr-ipv4
         type: u32
@@ -73,7 +73,7 @@ kernel-policy: global
         name: saddr-ipv6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
         byte-order: big-endian
         display-hint: ipv6
       -
-- 
2.55.0
Reviewed-by: Hangbin Liu <redacted>

Re: [PATCH net-next 2/2] netlink: specs: tcp_metrics: drop the RTT shift instructions

From: Hangbin Liu <hidden>
Date: 2026-09-14 06:49:43

On Sat, Sep 12, 2026 at 04:43:37PM -0700, Jakub Kicinski wrote:
quoted hunk
All four RTT attributes tell the reader to left-shift, which inflates the
value by 16 to 64 times, and the two usec ones say the result is in msecs.
The attributes carry srtt_us and mdev_us, which hold 3 and 2 fractional
bits. The shift to get the integer part would be a right shift.

Drop the instructions instead of turning them around. The number of
fractional bits is the part worth documenting, whether to shift, divide or
convert to a double is up to the caller.

While at it fix the acronym on the two variance attributes.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 Documentation/netlink/specs/tcp_metrics.yaml | 12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)
diff --git a/Documentation/netlink/specs/tcp_metrics.yaml b/Documentation/netlink/specs/tcp_metrics.yaml
index 1d365908084d..779e8a22b765 100644
--- a/Documentation/netlink/specs/tcp_metrics.yaml
+++ b/Documentation/netlink/specs/tcp_metrics.yaml
@@ -93,14 +93,12 @@ kernel-policy: global
         name: rtt
         type: u32
         doc: |
-          Round Trip Time (RTT), in msecs with 3 bits fractional
-          (left-shift by 3 to get the msec value).
+          Round Trip Time (RTT), in msecs with 3 bits fractional.
       -
         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).
+          Round Trip Time VARiance (RTTVAR), in msecs with 2 bits fractional.
       -
         name: ssthresh
         type: u32
@@ -117,14 +115,12 @@ kernel-policy: global
         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).
+          Round Trip Time (RTT), in usecs, with 3 bits fractional.
       -
         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).
+          Round Trip Time VARiance (RTTVAR), in usecs, with 2 bits fractional.
 
 operations:
   list:
-- 
2.55.0
Reviewed-by: Hangbin Liu <redacted>

Re: [PATCH net-next 1/2] netlink: specs: tcp_metrics: fix the attribute length checks

From: patchwork-bot+netdevbpf@kernel.org
Date: 2026-09-15 13:11:10

Hello:

This series was applied to netdev/net-next.git (main)
by Paolo Abeni [off-list ref]:

On Sat, 12 Sep 2026 16:43:36 -0700 you wrote:
TCP_FASTOPEN_COOKIE_MAX is the longest cookie we accept, not the shortest
one. Cookies are even sized, from 4 bytes up, and the ones Linux itself
generates are 8 bytes, so a 16 byte minimum declares all but the longest
cookie invalid. The reference policy kept under #if 0 in tcp_metrics.c
spells it as a maximum, which is what .len means for NLA_BINARY.

The IPv6 addresses err the other way. The policy uses
NLA_POLICY_EXACT_LEN() for both, so a request carrying a longer address is
rejected with -ERANGE, even though the spec advertises 16 bytes as a mere
minimum. Were the policy generated from this spec, as kernel-policy: global
promises, the check would turn into NLA_POLICY_MIN_LEN() and start
accepting over-long addresses, of which nla_get_in6_addr() would take the
first 16 bytes.

[...]
Here is the summary with links:
  - [net-next,1/2] netlink: specs: tcp_metrics: fix the attribute length checks
    https://git.kernel.org/netdev/net-next/c/4358a3e8e7cc
  - [net-next,2/2] netlink: specs: tcp_metrics: drop the RTT shift instructions
    https://git.kernel.org/netdev/net-next/c/1a523f29fe4f

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help