Re: [PATCH net-next V6 07/14] devlink: Add parent dev to devlink API

[Cosmin Ratiu <cratiu@nvidia.com>]

On Tue, 2026-01-27 at 13:49 +0000, Simon Horman wrote:
> On Sun, Jan 25, 2026 at 01:31:56PM +0200, Tariq Toukan wrote:
> > From: Cosmin Ratiu <cratiu@nvidia.com>
> > 
> > Upcoming changes to the rate commands need the parent devlink
> > specified.
> > This change adds a nested 'parent-dev' attribute to the API and
> > helpers
> > to obtain and put a reference to the parent devlink instance in
> > info->user_ptr[1].
> > 
> > To avoid deadlocks, the parent devlink is unlocked before obtaining
> > the
> > main devlink instance that is the target of the request.
> > A reference to the parent is kept until the end of the request to
> > avoid
> > it suddenly disappearing.
> > 
> > This means that this reference is of limited use without additional
> > protection.
> > 
> > Signed-off-by: Cosmin Ratiu <cratiu@nvidia.com>
> > Reviewed-by: Carolina Jubran <cjubran@nvidia.com>
> > Reviewed-by: Jiri Pirko <jiri@nvidia.com>
> > Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> 
> ...
> 
> Hi Cosmin, all,
> 
> The netlink_gen.[ch] and spec changes in this patch do not seem
> consistent.
> 
> $ tools/net/ynl/ynl-regen.sh -f
> $ git diff
> diff --git a/net/devlink/netlink_gen.c b/net/devlink/netlink_gen.c
> index 6b691bdbf037..f4c61c2b4f22 100644
> --- a/net/devlink/netlink_gen.c
> +++ b/net/devlink/netlink_gen.c
> @@ -39,11 +39,6 @@ devlink_attr_param_type_validate(const struct
> nlattr *attr,
>  }
> 
>  /* Common nested types */
> -const struct nla_policy
> devlink_dl_parent_dev_nl_policy[DEVLINK_ATTR_DEV_NAME + 1] = {
> -       [DEVLINK_ATTR_BUS_NAME] = { .type = NLA_NUL_STRING, },
> -       [DEVLINK_ATTR_DEV_NAME] = { .type = NLA_NUL_STRING, },
> -};
> -
>  const struct nla_policy
> devlink_dl_port_function_nl_policy[DEVLINK_PORT_FN_ATTR_CAPS + 1] = {
>         [DEVLINK_PORT_FUNCTION_ATTR_HW_ADDR] = { .type = NLA_BINARY,
> },
>         [DEVLINK_PORT_FN_ATTR_STATE] = NLA_POLICY_MAX(NLA_U8, 1),
> diff --git a/net/devlink/netlink_gen.h b/net/devlink/netlink_gen.h
> index d4db82a00522..2817d53a0eba 100644
> --- a/net/devlink/netlink_gen.h
> +++ b/net/devlink/netlink_gen.h
> @@ -13,7 +13,6 @@
>  #include <uapi/linux/devlink.h>
> 
>  /* Common nested types */
> -extern const struct nla_policy
> devlink_dl_parent_dev_nl_policy[DEVLINK_ATTR_DEV_NAME + 1];
>  extern const struct nla_policy
> devlink_dl_port_function_nl_policy[DEVLINK_PORT_FN_ATTR_CAPS + 1];
>  extern const struct nla_policy
> devlink_dl_rate_tc_bws_nl_policy[DEVLINK_RATE_TC_ATTR_BW + 1];
>  extern const struct nla_policy
> devlink_dl_selftest_id_nl_policy[DEVLINK_ATTR_SELFTEST_ID_FLASH + 1];
> @@ -30,19 +29,12 @@ int devlink_nl_pre_doit_dev_lock(const struct
> genl_split_ops *ops,
>  int devlink_nl_pre_doit_port_optional(const struct genl_split_ops
> *ops,
>                                       struct sk_buff *skb,
>                                       struct genl_info *info);
> -int devlink_nl_pre_doit_parent_dev_optional(const struct
> genl_split_ops *ops,
> -                                           struct sk_buff *skb,
> -                                           struct genl_info *info);
>  void
>  devlink_nl_post_doit(const struct genl_split_ops *ops, struct
> sk_buff *skb,
>                      struct genl_info *info);
>  void
>  devlink_nl_post_doit_dev_lock(const struct genl_split_ops *ops,
>                               struct sk_buff *skb, struct genl_info
> *info);
> -void
> -devlink_nl_post_doit_parent_dev_optional(const struct genl_split_ops
> *ops,
> -                                        struct sk_buff *skb,
> -                                        struct genl_info *info);
> 
>  int devlink_nl_get_doit(struct sk_buff *skb, struct genl_info
> *info);
>  int devlink_nl_get_dumpit(struct sk_buff *skb, struct
> netlink_callback *cb);
> ;

Hi Simon,

We had this conversation during v4, I replied then [1].
But thinking about it a bit more, I think it's indeed slightly cleaner
to move the policy and the new pre/post doit handlers to the next
patch, where they are actually used. The only bit is that the policy is
used from devlink_get_parent_from_attrs_lock from this function, but it
appears safe to use NULL there until next patch (the underlying parse
functions tolerate NULL policies).

So I'll do that in the next submission.

[1]
https://lore.kernel.org/netdev/3ec956ea1d0a1c6e56865b2ded6d83ed773ccd4d.camel@nvidia.com/