diff mbox series

[net-next,3/3] netlink: specs: use exact-len for IPv6 addr

Message ID 20231215035009.498049-4-liuhangbin@gmail.com (mailing list archive)
State Changes Requested
Delegated to: Netdev Maintainers
Headers show
Series ynl-gen: update check format | expand

Checks

Context Check Description
netdev/tree_selection success Clearly marked for net-next
netdev/apply fail Patch does not apply to net-next

Commit Message

Hangbin Liu Dec. 15, 2023, 3:50 a.m. UTC
We should use the exact-len instead of min-len for IPv6 address.

Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
---
 Documentation/netlink/specs/fou.yaml   | 4 ++--
 Documentation/netlink/specs/mptcp.yaml | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

Comments

Jakub Kicinski Dec. 16, 2023, 2:09 a.m. UTC | #1
On Fri, 15 Dec 2023 11:50:09 +0800 Hangbin Liu wrote:
> We should use the exact-len instead of min-len for IPv6 address.

It does make sense, but these families historically used min-len..
Not sure if it's worth changing this now or we risk regressions.
Hangbin Liu Dec. 16, 2023, 8:48 a.m. UTC | #2
On Fri, Dec 15, 2023 at 06:09:11PM -0800, Jakub Kicinski wrote:
> On Fri, 15 Dec 2023 11:50:09 +0800 Hangbin Liu wrote:
> > We should use the exact-len instead of min-len for IPv6 address.
> 
> It does make sense, but these families historically used min-len..
> Not sure if it's worth changing this now or we risk regressions.

The addr6 in mptcp.yaml also use exact-len. I don't think the IPv6 address
could be larger than 16 bytes. So the min-len check looks incorrect.

Thanks
Hangbin
Davide Caratti Dec. 18, 2023, 11:40 a.m. UTC | #3
hello Hangbin!

for the mptcp part: I didn't convert mptcp events, here transition
from min-len to exact-len should not generate issues at all.

Acked-by: Davide Caratti <dcaratti@redhat.com>

On Fri, Dec 15, 2023 at 4:50 AM Hangbin Liu <liuhangbin@gmail.com> wrote:
>
> We should use the exact-len instead of min-len for IPv6 address.
>
> Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
> ---
>  Documentation/netlink/specs/fou.yaml   | 4 ++--
>  Documentation/netlink/specs/mptcp.yaml | 4 ++--
>  2 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/Documentation/netlink/specs/fou.yaml b/Documentation/netlink/specs/fou.yaml
> index 0af5ab842c04..d472fd5055bd 100644
> --- a/Documentation/netlink/specs/fou.yaml
> +++ b/Documentation/netlink/specs/fou.yaml
> @@ -52,7 +52,7 @@ attribute-sets:
>          name: local_v6
>          type: binary
>          checks:
> -          min-len: 16
> +          exact-len: 16
>        -
>          name: peer_v4
>          type: u32
> @@ -60,7 +60,7 @@ attribute-sets:
>          name: peer_v6
>          type: binary
>          checks:
> -          min-len: 16
> +          exact-len: 16
>        -
>          name: peer_port
>          type: u16
> diff --git a/Documentation/netlink/specs/mptcp.yaml b/Documentation/netlink/specs/mptcp.yaml
> index 49f90cfb4698..2f694b79c3a7 100644
> --- a/Documentation/netlink/specs/mptcp.yaml
> +++ b/Documentation/netlink/specs/mptcp.yaml
> @@ -223,7 +223,7 @@ attribute-sets:
>          name: saddr6
>          type: binary
>          checks:
> -          min-len: 16
> +          exact-len: 16
>        -
>          name: daddr4
>          type: u32
> @@ -232,7 +232,7 @@ attribute-sets:
>          name: daddr6
>          type: binary
>          checks:
> -          min-len: 16
> +          exact-len: 16
>        -
>          name: sport
>          type: u16
> --
> 2.43.0
>
>
Jakub Kicinski Dec. 18, 2023, 10:24 p.m. UTC | #4
On Sat, 16 Dec 2023 16:48:01 +0800 Hangbin Liu wrote:
> On Fri, Dec 15, 2023 at 06:09:11PM -0800, Jakub Kicinski wrote:
> > On Fri, 15 Dec 2023 11:50:09 +0800 Hangbin Liu wrote:  
> > > We should use the exact-len instead of min-len for IPv6 address.  
> > 
> > It does make sense, but these families historically used min-len..
> > Not sure if it's worth changing this now or we risk regressions.  
> 
> The addr6 in mptcp.yaml also use exact-len. I don't think the IPv6 address
> could be larger than 16 bytes. So the min-len check looks incorrect.

I understand, what I'm saying is that the nla_policy before we started
using specs was buggy, so we kept the bugginess in the conversion.
But it should be fine, so we can change it now and hope for the best..
diff mbox series

Patch

diff --git a/Documentation/netlink/specs/fou.yaml b/Documentation/netlink/specs/fou.yaml
index 0af5ab842c04..d472fd5055bd 100644
--- a/Documentation/netlink/specs/fou.yaml
+++ b/Documentation/netlink/specs/fou.yaml
@@ -52,7 +52,7 @@  attribute-sets:
         name: local_v6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
       -
         name: peer_v4
         type: u32
@@ -60,7 +60,7 @@  attribute-sets:
         name: peer_v6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
       -
         name: peer_port
         type: u16
diff --git a/Documentation/netlink/specs/mptcp.yaml b/Documentation/netlink/specs/mptcp.yaml
index 49f90cfb4698..2f694b79c3a7 100644
--- a/Documentation/netlink/specs/mptcp.yaml
+++ b/Documentation/netlink/specs/mptcp.yaml
@@ -223,7 +223,7 @@  attribute-sets:
         name: saddr6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
       -
         name: daddr4
         type: u32
@@ -232,7 +232,7 @@  attribute-sets:
         name: daddr6
         type: binary
         checks:
-          min-len: 16
+          exact-len: 16
       -
         name: sport
         type: u16