On Wed, Sep 23, 2026 at 9:19 AM <[email protected]> wrote: > [Severity: Low] > This isn't a bug in the change itself, but should this carry a Fixes: tag? > > The spec has been shipped with the wrong type, and the generated ynl code and > the Python decoder have been treating the netns ID as an opaque byte blob, > with the request path rejecting an integer for a binary attribute. The > kernel side has always declared it signed 32-bit: > > net/ipv4/devinet.c:ifa_ipv4_policy: > [IFA_TARGET_NETNSID] = { .type = NLA_S32 }, > > with nla_put_s32() on the emit side and nla_get_s32() on the parse side in > net/ipv6/addrconf.c. > > git blame on the pre-patch tree attributes the type: binary line in > Documentation/netlink/specs/rt-addr.yaml to: > > dfb0f7d9d97956 ("doc/netlink: Add spec for rt addr messages") > > so would adding > > Fixes: dfb0f7d9d979 ("doc/netlink: Add spec for rt addr messages") > > help backport tooling and users identify which releases carry the wrong > type? > > -- > Sashiko AI review · > https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260921235951.3214-1-sigefriedhyy%40gmail.com
This issue has been discussed in a previous review comment before, we want to keep this patch target net-next without the Fixes tag.
