1/*****************************************/ 2Kernel Connector. 3/*****************************************/ 4 5Kernel connector - new netlink based userspace <-> kernel space easy 6to use communication module. 7 8The Connector driver makes it easy to connect various agents using a 9netlink based network. One must register a callback and an identifier. 10When the driver receives a special netlink message with the appropriate 11identifier, the appropriate callback will be called. 12 13From the userspace point of view it's quite straightforward: 14 15 socket(); 16 bind(); 17 send(); 18 recv(); 19 20But if kernelspace wants to use the full power of such connections, the 21driver writer must create special sockets, must know about struct sk_buff 22handling, etc... The Connector driver allows any kernelspace agents to use 23netlink based networking for inter-process communication in a significantly 24easier way: 25 26int cn_add_callback(struct cb_id *id, char *name, void (*callback) (struct cn_msg *, struct netlink_skb_parms *)); 27void cn_netlink_send_multi(struct cn_msg *msg, u16 len, u32 portid, u32 __group, int gfp_mask); 28void cn_netlink_send(struct cn_msg *msg, u32 portid, u32 __group, int gfp_mask); 29 30struct cb_id 31{ 32 __u32 idx; 33 __u32 val; 34}; 35 36idx and val are unique identifiers which must be registered in the 37connector.h header for in-kernel usage. void (*callback) (void *) is a 38callback function which will be called when a message with above idx.val 39is received by the connector core. The argument for that function must 40be dereferenced to struct cn_msg *. 41 42struct cn_msg 43{ 44 struct cb_id id; 45 46 __u32 seq; 47 __u32 ack; 48 49 __u32 len; /* Length of the following data */ 50 __u8 data[0]; 51}; 52 53/*****************************************/ 54Connector interfaces. 55/*****************************************/ 56 57int cn_add_callback(struct cb_id *id, char *name, void (*callback) (struct cn_msg *, struct netlink_skb_parms *)); 58 59 Registers new callback with connector core. 60 61 struct cb_id *id - unique connector's user identifier. 62 It must be registered in connector.h for legal in-kernel users. 63 char *name - connector's callback symbolic name. 64 void (*callback) (struct cn..) - connector's callback. 65 cn_msg and the sender's credentials 66 67 68void cn_del_callback(struct cb_id *id); 69 70 Unregisters new callback with connector core. 71 72 struct cb_id *id - unique connector's user identifier. 73 74 75int cn_netlink_send_multi(struct cn_msg *msg, u16 len, u32 portid, u32 __groups, int gfp_mask); 76int cn_netlink_send(struct cn_msg *msg, u32 portid, u32 __groups, int gfp_mask); 77 78 Sends message to the specified groups. It can be safely called from 79 softirq context, but may silently fail under strong memory pressure. 80 If there are no listeners for given group -ESRCH can be returned. 81 82 struct cn_msg * - message header(with attached data). 83 u16 len - for *_multi multiple cn_msg messages can be sent 84 u32 port - destination port. 85 If non-zero the message will be sent to the 86 given port, which should be set to the 87 original sender. 88 u32 __group - destination group. 89 If port and __group is zero, then appropriate group will 90 be searched through all registered connector users, 91 and message will be delivered to the group which was 92 created for user with the same ID as in msg. 93 If __group is not zero, then message will be delivered 94 to the specified group. 95 int gfp_mask - GFP mask. 96 97 Note: When registering new callback user, connector core assigns 98 netlink group to the user which is equal to its id.idx. 99 100/*****************************************/ 101Protocol description. 102/*****************************************/ 103 104The current framework offers a transport layer with fixed headers. The 105recommended protocol which uses such a header is as following: 106 107msg->seq and msg->ack are used to determine message genealogy. When 108someone sends a message, they use a locally unique sequence and random 109acknowledge number. The sequence number may be copied into 110nlmsghdr->nlmsg_seq too. 111 112The sequence number is incremented with each message sent. 113 114If you expect a reply to the message, then the sequence number in the 115received message MUST be the same as in the original message, and the 116acknowledge number MUST be the same + 1. 117 118If we receive a message and its sequence number is not equal to one we 119are expecting, then it is a new message. If we receive a message and 120its sequence number is the same as one we are expecting, but its 121acknowledge is not equal to the sequence number in the original 122message + 1, then it is a new message. 123 124Obviously, the protocol header contains the above id. 125 126The connector allows event notification in the following form: kernel 127driver or userspace process can ask connector to notify it when 128selected ids will be turned on or off (registered or unregistered its 129callback). It is done by sending a special command to the connector 130driver (it also registers itself with id={-1, -1}). 131 132As example of this usage can be found in the cn_test.c module which 133uses the connector to request notification and to send messages. 134 135/*****************************************/ 136Reliability. 137/*****************************************/ 138 139Netlink itself is not a reliable protocol. That means that messages can 140be lost due to memory pressure or process' receiving queue overflowed, 141so caller is warned that it must be prepared. That is why the struct 142cn_msg [main connector's message header] contains u32 seq and u32 ack 143fields. 144 145/*****************************************/ 146Userspace usage. 147/*****************************************/ 148 1492.6.14 has a new netlink socket implementation, which by default does not 150allow people to send data to netlink groups other than 1. 151So, if you wish to use a netlink socket (for example using connector) 152with a different group number, the userspace application must subscribe to 153that group first. It can be achieved by the following pseudocode: 154 155s = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_CONNECTOR); 156 157l_local.nl_family = AF_NETLINK; 158l_local.nl_groups = 12345; 159l_local.nl_pid = 0; 160 161if (bind(s, (struct sockaddr *)&l_local, sizeof(struct sockaddr_nl)) == -1) { 162 perror("bind"); 163 close(s); 164 return -1; 165} 166 167{ 168 int on = l_local.nl_groups; 169 setsockopt(s, 270, 1, &on, sizeof(on)); 170} 171 172Where 270 above is SOL_NETLINK, and 1 is a NETLINK_ADD_MEMBERSHIP socket 173option. To drop a multicast subscription, one should call the above socket 174option with the NETLINK_DROP_MEMBERSHIP parameter which is defined as 0. 175 1762.6.14 netlink code only allows to select a group which is less or equal to 177the maximum group number, which is used at netlink_kernel_create() time. 178In case of connector it is CN_NETLINK_USERS + 0xf, so if you want to use 179group number 12345, you must increment CN_NETLINK_USERS to that number. 180Additional 0xf numbers are allocated to be used by non-in-kernel users. 181 182Due to this limitation, group 0xffffffff does not work now, so one can 183not use add/remove connector's group notifications, but as far as I know, 184only cn_test.c test module used it. 185 186Some work in netlink area is still being done, so things can be changed in 1872.6.15 timeframe, if it will happen, documentation will be updated for that 188kernel. 189