From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jiri Pirko Subject: Re: [PATCH net-next v5 3/5] netlink: allow sending extended ACK with cookie on success Date: Wed, 12 Apr 2017 15:47:09 +0200 Message-ID: <20170412134709.GF1952@nanopsycho> References: <20170412123408.22012-1-johannes@sipsolutions.net> <20170412123408.22012-4-johannes@sipsolutions.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-wireless-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, pablo-Cap9r6Oaw4JrovVCs/uTlw@public.gmane.org, Jamal Hadi Salim , Jiri Benc , David Ahern , Johannes Berg To: Johannes Berg Return-path: Content-Disposition: inline In-Reply-To: <20170412123408.22012-4-johannes-cdvu00un1VgdHxzADdlk8Q@public.gmane.org> Sender: linux-wireless-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: netdev.vger.kernel.org Wed, Apr 12, 2017 at 02:34:06PM CEST, johannes-cdvu00un1VgdHxzADdlk8Q@public.gmane.org wrote: >From: Johannes Berg > >Now that we have extended error reporting and a new message format >for netlink ACK messages, also extend this to be able to return >arbitrary cookie data on success. > >This will allow, for example, nl80211 to not send an extra message >for cookies identifying newly created objects, but return those >directly in the ACK message. > >The cookie data size is currently limited to 20 bytes (since Jamal >talked about using SHA1 for identifiers.) > >Thanks to Jamal Hadi Salim for bringing up this idea during the >discussions. > >Signed-off-by: Johannes Berg Reviewed-by: Jiri Pirko [...] >diff --git a/include/uapi/linux/netlink.h b/include/uapi/linux/netlink.h >index 7df88770e029..f86127a46cfc 100644 >--- a/include/uapi/linux/netlink.h >+++ b/include/uapi/linux/netlink.h >@@ -122,6 +122,9 @@ struct nlmsgerr { > * @NLMSGERR_ATTR_MSG: error message string (string) > * @NLMSGERR_ATTR_OFFS: offset of the invalid attribute in the original > * message, counting from the beginning of the header (u32) >+ * @NLMSGERR_ATTR_COOKIE: arbitrary subsystem specific cookie to >+ * be used - in the success case - to identify a created >+ * object or operation or similar (binary) > * @__NLMSGERR_ATTR_MAX: number of attributes > * @NLMSGERR_ATTR_MAX: highest attribute number > */ >@@ -129,6 +132,7 @@ enum nlmsgerr_attrs { > NLMSGERR_ATTR_UNUSED, > NLMSGERR_ATTR_MSG, > NLMSGERR_ATTR_OFFS, >+ NLMSGERR_ATTR_COOKIE, > > __NLMSGERR_ATTR_MAX, > NLMSGERR_ATTR_MAX = __NLMSGERR_ATTR_MAX - 1 >diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c >index c1564768000e..ee841f00a6ec 100644 >--- a/net/netlink/af_netlink.c >+++ b/net/netlink/af_netlink.c >@@ -2311,6 +2311,10 @@ void netlink_ack(struct sk_buff *in_skb, struct nlmsghdr *nlh, int err, > } > } else { > flags |= NLM_F_CAPPED; >+ >+ if (nlk->flags & NETLINK_F_EXT_ACK && >+ extack && extack->cookie_len) >+ tlvlen += nla_total_size(extack->cookie_len); > } > Now I see why you have "if (tlvlen)" where you have it :) > if (tlvlen)