| Message ID | 20240212161615.161935-2-stanislaw.gruszka@linux.intel.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 3de21a8990d3c2cc507e9cc4ed00f36358d5b93e |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | thermal/netlink/intel_hfi: Enable HFI feature only when required | expand |
On Mon, 12 Feb 2024 17:16:13 +0100 Stanislaw Gruszka wrote: > Add genetlink family bind()/unbind() callbacks when adding/removing > multicast group to/from netlink client socket via setsockopt() or > bind() syscall. > > They can be used to track if consumers of netlink multicast messages > emerge or disappear. Thus, a client implementing callbacks, can now > send events only when there are active consumers, preventing unnecessary > work when none exist. > > Suggested-by: Jakub Kicinski <kuba@kernel.org> > Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com> LGTM! genetlink code is a bit hot lately, to avoid any conflicts can I put the first patch (or all of them) on a shared branch for both netdev and PM to pull in? Once the other two patches are reviewed, obviously.
On Mon, 2024-02-12 at 17:07 -0800, Jakub Kicinski wrote: > On Mon, 12 Feb 2024 17:16:13 +0100 Stanislaw Gruszka wrote: > > Add genetlink family bind()/unbind() callbacks when adding/removing > > multicast group to/from netlink client socket via setsockopt() or > > bind() syscall. > > > > They can be used to track if consumers of netlink multicast > > messages > > emerge or disappear. Thus, a client implementing callbacks, can now > > send events only when there are active consumers, preventing > > unnecessary > > work when none exist. > > > > Suggested-by: Jakub Kicinski <kuba@kernel.org> > > Signed-off-by: Stanislaw Gruszka > > <stanislaw.gruszka@linux.intel.com> > > LGTM! genetlink code is a bit hot lately, to avoid any conflicts can > I put the first patch (or all of them) on a shared branch for both > netdev and PM to pull in? Once the other two patches are reviewed, > obviously. > If netlink maintainers are happy with the 1/3, you can take 1/3. Once merged, the PM patches can go separately as they need 1/3. Hi Daniel, Please look at 2/3 and also 3/3 if you can. Thanks, Srinivas
Mon, Feb 12, 2024 at 05:16:13PM CET, stanislaw.gruszka@linux.intel.com wrote: >Add genetlink family bind()/unbind() callbacks when adding/removing >multicast group to/from netlink client socket via setsockopt() or >bind() syscall. > >They can be used to track if consumers of netlink multicast messages >emerge or disappear. Thus, a client implementing callbacks, can now >send events only when there are active consumers, preventing unnecessary >work when none exist. > >Suggested-by: Jakub Kicinski <kuba@kernel.org> >Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com> Reviewed-by: Jiri Pirko <jiri@nvidia.com>
diff --git a/include/net/genetlink.h b/include/net/genetlink.h index e61469129402..ecadba836ae5 100644 --- a/include/net/genetlink.h +++ b/include/net/genetlink.h @@ -41,6 +41,8 @@ struct genl_info; * do additional, common, filtering and return an error * @post_doit: called after an operation's doit callback, it may * undo operations done by pre_doit, for example release locks + * @bind: called when family multicast group is added to a netlink socket + * @unbind: called when family multicast group is removed from a netlink socket * @module: pointer to the owning module (set to THIS_MODULE) * @mcgrps: multicast groups used by this family * @n_mcgrps: number of multicast groups @@ -84,6 +86,8 @@ struct genl_family { void (*post_doit)(const struct genl_split_ops *ops, struct sk_buff *skb, struct genl_info *info); + int (*bind)(int mcgrp); + void (*unbind)(int mcgrp); const struct genl_ops * ops; const struct genl_small_ops *small_ops; const struct genl_split_ops *split_ops; diff --git a/net/netlink/genetlink.c b/net/netlink/genetlink.c index 8c7af02f8454..50ec599a5cff 100644 --- a/net/netlink/genetlink.c +++ b/net/netlink/genetlink.c @@ -1836,6 +1836,9 @@ static int genl_bind(struct net *net, int group) !ns_capable(net->user_ns, CAP_SYS_ADMIN)) ret = -EPERM; + if (family->bind) + family->bind(i); + break; } @@ -1843,12 +1846,39 @@ static int genl_bind(struct net *net, int group) return ret; } +static void genl_unbind(struct net *net, int group) +{ + const struct genl_family *family; + unsigned int id; + + down_read(&cb_lock); + + idr_for_each_entry(&genl_fam_idr, family, id) { + int i; + + if (family->n_mcgrps == 0) + continue; + + i = group - family->mcgrp_offset; + if (i < 0 || i >= family->n_mcgrps) + continue; + + if (family->unbind) + family->unbind(i); + + break; + } + + up_read(&cb_lock); +} + static int __net_init genl_pernet_init(struct net *net) { struct netlink_kernel_cfg cfg = { .input = genl_rcv, .flags = NL_CFG_F_NONROOT_RECV, .bind = genl_bind, + .unbind = genl_unbind, .release = genl_release, };
Add genetlink family bind()/unbind() callbacks when adding/removing multicast group to/from netlink client socket via setsockopt() or bind() syscall. They can be used to track if consumers of netlink multicast messages emerge or disappear. Thus, a client implementing callbacks, can now send events only when there are active consumers, preventing unnecessary work when none exist. Suggested-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com> --- include/net/genetlink.h | 4 ++++ net/netlink/genetlink.c | 30 ++++++++++++++++++++++++++++++ 2 files changed, 34 insertions(+)