1.\" Copyright (c) 1996-1999 Whistle Communications, Inc. 2.\" All rights reserved. 3.\" 4.\" Subject to the following obligations and disclaimer of warranty, use and 5.\" redistribution of this software, in source or object code forms, with or 6.\" without modifications are expressly permitted by Whistle Communications; 7.\" provided, however, that: 8.\" 1. Any and all reproductions of the source or object code must include the 9.\" copyright notice above and the following disclaimer of warranties; and 10.\" 2. No rights are granted, in any manner or form, to use Whistle 11.\" Communications, Inc. trademarks, including the mark "WHISTLE 12.\" COMMUNICATIONS" on advertising, endorsements, or otherwise except as 13.\" such appears in the above copyright notice or in the software. 14.\" 15.\" THIS SOFTWARE IS BEING PROVIDED BY WHISTLE COMMUNICATIONS "AS IS", AND 16.\" TO THE MAXIMUM EXTENT PERMITTED BY LAW, WHISTLE COMMUNICATIONS MAKES NO 17.\" REPRESENTATIONS OR WARRANTIES, EXPRESS OR IMPLIED, REGARDING THIS SOFTWARE, 18.\" INCLUDING WITHOUT LIMITATION, ANY AND ALL IMPLIED WARRANTIES OF 19.\" MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT. 20.\" WHISTLE COMMUNICATIONS DOES NOT WARRANT, GUARANTEE, OR MAKE ANY 21.\" REPRESENTATIONS REGARDING THE USE OF, OR THE RESULTS OF THE USE OF THIS 22.\" SOFTWARE IN TERMS OF ITS CORRECTNESS, ACCURACY, RELIABILITY OR OTHERWISE. 23.\" IN NO EVENT SHALL WHISTLE COMMUNICATIONS BE LIABLE FOR ANY DAMAGES 24.\" RESULTING FROM OR ARISING OUT OF ANY USE OF THIS SOFTWARE, INCLUDING 25.\" WITHOUT LIMITATION, ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, 26.\" PUNITIVE, OR CONSEQUENTIAL DAMAGES, PROCUREMENT OF SUBSTITUTE GOODS OR 27.\" SERVICES, LOSS OF USE, DATA OR PROFITS, HOWEVER CAUSED AND UNDER ANY 28.\" THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 29.\" (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 30.\" THIS SOFTWARE, EVEN IF WHISTLE COMMUNICATIONS IS ADVISED OF THE POSSIBILITY 31.\" OF SUCH DAMAGE. 32.\" 33.\" Authors: Julian Elischer <julian@FreeBSD.org> 34.\" Archie Cobbs <archie@FreeBSD.org> 35.\" 36.\" $FreeBSD: src/share/man/man4/netgraph.4,v 1.39.2.1 2001/12/21 09:00:50 ru Exp $ 37.\" $Whistle: netgraph.4,v 1.7 1999/01/28 23:54:52 julian Exp $ 38.\" 39.Dd September 2, 2008 40.Dt NETGRAPH 4 41.Os 42.Sh NAME 43.Nm netgraph 44.Nd graph based kernel networking subsystem 45.Sh DESCRIPTION 46The 47.Nm 48system provides a uniform and modular system for the implementation 49of kernel objects which perform various networking functions. 50The objects, known as 51.Em nodes , 52can be arranged into arbitrarily complicated graphs. 53Nodes have 54.Em hooks 55which are used to connect two nodes together, forming the edges in the graph. 56Nodes communicate along the edges to process data, implement protocols, etc. 57.Pp 58The aim of 59.Nm 60is to supplement rather than replace the existing kernel networking 61infrastructure. 62It provides: 63.Pp 64.Bl -bullet -compact -offset 2n 65.It 66A flexible way of combining protocol and link level drivers 67.It 68A modular way to implement new protocols 69.It 70A common framework for kernel entities to inter-communicate 71.It 72A reasonably fast, kernel-based implementation 73.El 74.Sh Nodes and Types 75The most fundamental concept in 76.Nm 77is that of a 78.Em node . 79All nodes implement a number of predefined methods which allow them 80to interact with other nodes in a well defined manner. 81.Pp 82Each node has a 83.Em type , 84which is a static property of the node determined at node creation time. 85A node's type is described by a unique 86.Tn ASCII 87type name. 88The type implies what the node does and how it may be connected 89to other nodes. 90.Pp 91In object-oriented language, types are classes and nodes are instances 92of their respective class. 93All node types are subclasses of the generic node 94type, and hence inherit certain common functionality and capabilities 95(e.g., the ability to have an 96.Tn ASCII 97name). 98.Pp 99Nodes may be assigned a globally unique 100.Tn ASCII 101name which can be 102used to refer to the node. 103The name must not contain the characters 104.Dq .\& 105or 106.Dq \&: 107and is limited to 108.Dv "NG_NODESIZ" 109characters (including NUL byte). 110.Pp 111Each node instance has a unique 112.Em ID number 113which is expressed as a 32-bit hex value. 114This value may be used to refer to a node when there is no 115.Tn ASCII 116name assigned to it. 117.Sh Hooks 118Nodes are connected to other nodes by connecting a pair of 119.Em hooks , 120one from each node. 121Data flows bidirectionally between nodes along connected pairs of hooks. 122A node may have as many hooks as it needs, 123and may assign whatever meaning it wants to a hook. 124.Pp 125Hooks have these properties: 126.Pp 127.Bl -bullet -compact -offset 2n 128.It 129A hook has an 130.Tn ASCII 131name which is unique among all hooks 132on that node (other hooks on other nodes may have the same name). 133The name must not contain a 134.Dq .\& 135or a 136.Dq \&: 137and is 138limited to 139.Dv "NG_HOOKSIZ" 140characters (including NUL byte). 141.It 142A hook is always connected to another hook. 143That is, hooks are created at the time they are connected, 144and breaking an edge by removing either hook destroys both hooks. 145.El 146.Pp 147A node may decide to assign special meaning to some hooks. 148For example, connecting to the hook named 149.Dq debug 150might trigger 151the node to start sending debugging information to that hook. 152.Sh Data Flow 153Two types of information flow between nodes: data messages and 154control messages. 155Data messages are passed in mbuf chains along the edges 156in the graph, one edge at a time. 157The first mbuf in a chain must have the 158.Dv M_PKTHDR 159flag set. 160Each node decides how to handle data coming in on its hooks. 161.Pp 162Control messages are type-specific C structures sent from one node 163directly to some arbitrary other node. 164Control messages have a common header format, 165followed by type-specific data, and are binary structures 166for efficiency. 167However, node types also may support conversion of the 168type specific data between binary and 169.Tn ASCII 170for debugging and human interface purposes (see the 171.Dv NGM_ASCII2BINARY 172and 173.Dv NGM_BINARY2ASCII 174generic control messages below). 175Nodes are not required to support these conversions. 176.Pp 177There are two ways to address a control message. 178If there is a sequence of edges connecting the two nodes, 179the message may be 180.Dq source routed 181by specifying the corresponding sequence 182of hooks as the destination address for the message (relative 183addressing). 184Otherwise, the recipient node global 185.Tn ASCII 186name 187(or equivalent ID based name) is used as the destination address 188for the message (absolute addressing). 189The two types of addressing may be combined, 190by specifying an absolute start node and a sequence of hooks. 191.Pp 192Messages often represent commands that are followed by a reply message 193in the reverse direction. 194To facilitate this, the recipient of a 195control message is supplied with a 196.Dq return address 197that is suitable for addressing a reply. 198.Pp 199Each control message contains a 32 bit value called a 200.Em typecookie 201indicating the type of the message, i.e. how to interpret it. 202Typically each type defines a unique typecookie for the messages 203that it understands. 204However, a node may choose to recognize and 205implement more than one type of message. 206.Sh Netgraph is Functional 207In order to minimize latency, most 208.Nm 209operations are functional. 210That is, data and control messages are delivered by making function 211calls rather than by using queues and mailboxes. 212For example, if node A wishes to send a data mbuf to neighboring node B, 213it calls the generic 214.Nm 215data delivery function. 216This function in turn locates node B and calls B's 217.Dq receive data 218method. 219While this mode of operation results in good performance, 220it has a few implications for node developers: 221.Pp 222.Bl -bullet -compact -offset 2n 223.It 224Whenever a node delivers a data or control message, the node 225may need to allow for the possibility of receiving a returning 226message before the original delivery function call returns. 227.It 228Netgraph nodes and support routines generally run inside critical 229sections. 230However, some nodes may want to send data and control messages 231from a different priority level. 232Netgraph supplies queueing routines which utilize the NETISR system to 233move message delivery inside a critical section. 234Note that messages are always received from inside a critical section. 235.It 236It's possible for an infinite loop to occur if the graph contains cycles. 237.El 238.Pp 239So far, these issues have not proven problematical in practice. 240.Sh Interaction With Other Parts of the Kernel 241A node may have a hidden interaction with other components of the 242kernel outside of the 243.Nm 244subsystem, such as device hardware, kernel protocol stacks, etc. 245In fact, one of the benefits of 246.Nm 247is the ability to join disparate kernel networking entities together in a 248consistent communication framework. 249.Pp 250An example is the node type 251.Em socket 252which is both a netgraph node and a 253.Xr socket 2 254.Bx 255socket in the protocol family 256.Dv PF_NETGRAPH . 257Socket nodes allow user processes to participate in 258.Nm . 259Other nodes communicate with socket nodes using the usual methods, and the 260node hides the fact that it is also passing information to and from a 261cooperating user process. 262.Pp 263Another example is a device driver that presents 264a node interface to the hardware. 265.Sh Node Methods 266Nodes are notified of the following actions via function calls 267to the following node methods (all from inside critical sections) 268and may accept or reject that action (by returning the appropriate 269error code): 270.Bl -tag -width xxx 271.It Creation of a new node 272The constructor for the type is called. 273If creation of a new node is allowed, 274the constructor must call the generic node creation 275function (in object-oriented terms, the superclass constructor) 276and then allocate any special resources it needs. 277For nodes that correspond to hardware, this is typically done during 278the device attach routine. 279Often a global 280.Tn ASCII 281name corresponding to the 282device name is assigned here as well. 283.It Creation of a new hook 284The hook is created and tentatively 285linked to the node, and the node is told about the name that will be 286used to describe this hook. 287The node sets up any special data structures it needs, 288or may reject the connection, based on the name of the hook. 289.It Successful connection of two hooks 290After both ends have accepted their 291hooks, and the links have been made, the nodes get a chance to 292find out who their peer is across the link and can then decide to reject 293the connection. 294Tear-down is automatic. 295.It Destruction of a hook 296The node is notified of a broken connection. 297The node may consider some hooks to be critical to operation and others 298to be expendable: the disconnection of one hook may be an acceptable 299event while for another it may affect a total shutdown for the node. 300.It Shutdown of a node 301This method allows a node to clean up 302and to ensure that any actions that need to be performed 303at this time are taken. 304The method must call the generic (i.e. superclass) 305node destructor to get rid of the generic components of the node. 306Some nodes (usually associated with a piece of hardware) may be 307.Em persistent 308in that a shutdown breaks all edges and resets the node, 309but doesn't remove it, in which case the generic destructor is not called. 310.El 311.Sh Sending and Receiving Data 312Three other methods are also supported by all nodes: 313.Bl -tag -width xxx 314.It Receive data message 315An mbuf chain is passed to the node. 316The node is notified on which hook the data arrived, 317and can use this information in its processing decision. 318The node must always 319.Fn m_freem 320the mbuf chain on completion or error, or pass it on to another node 321(or kernel module) which will then be responsible for freeing it. 322.Pp 323In addition to the mbuf chain itself there is also a pointer to a 324structure describing meta-data about the message 325(e.g. priority information). 326This pointer may be 327.Dv NULL 328if there is no additional information. 329The format for this information is described in 330.In netgraph/netgraph.h . 331The memory for meta-data must allocated via 332.Fn malloc 333with type 334.Dv M_NETGRAPH . 335As with the data itself, it is the receiver's responsibility to 336.Fn free 337the meta-data. 338If the mbuf chain is freed the meta-data must be freed at the same time. 339If the meta-data is freed but the real data on is passed on, then a 340.Dv NULL 341pointer must be substituted. 342.Pp 343The receiving node may decide to defer the data by queueing it in the 344.Nm 345NETISR system (see below). 346.Pp 347The structure and use of meta-data is still experimental, but is 348presently used in frame-relay to indicate that management packets 349should be queued for transmission 350at a higher priority than data packets. 351This is required for conformance with Frame Relay standards. 352.It Receive queued data message 353Usually this will be the same function as 354.Em Receive data message. 355This is the entry point called when a data message is being handed to 356the node after having been queued in the NETISR system. 357This allows a node to decide in the 358.Em Receive data message 359method that a message should be deferred and queued, 360and be sure that when it is processed from the queue, 361it will not be queued again. 362.It Receive control message 363This method is called when a control message is addressed to the node. 364A return address is always supplied, giving the address of the node 365that originated the message so a reply message can be sent anytime later. 366.Pp 367It is possible for a synchronous reply to be made, and in fact this 368is more common in practice. 369This is done by setting a pointer (supplied as an extra function parameter) 370to point to the reply. 371Then when the control message delivery function returns, 372the caller can check if this pointer has been made non-NULL, 373and if so then it points to the reply message allocated via 374.Fn malloc 375and containing the synchronous response. 376In both directions, (request and response) it is up to the 377receiver of that message to 378.Fn free 379the control message buffer. 380All control messages and replies are allocated with 381.Fn malloc 382type 383.Dv M_NETGRAPH . 384.El 385.Pp 386Much use has been made of reference counts, so that nodes being 387free'd of all references are automatically freed, and this behaviour 388has been tested and debugged to present a consistent and trustworthy 389framework for the 390.Dq type module 391writer to use. 392.Sh Addressing 393The 394.Nm 395framework provides an unambiguous and simple to use method of specifically 396addressing any single node in the graph. 397The naming of a node is independent of its type, in that another node, 398or external component need not know anything about the node's type in 399order to address it so as to send it a generic message type. 400Node and hook names should be chosen so as to make addresses meaningful. 401.Pp 402Addresses are either absolute or relative. 403An absolute address begins with a node name (or ID), followed by a colon, 404followed by a sequence of hook names separated by periods. 405This addresses the node reached by starting 406at the named node and following the specified sequence of hooks. 407A relative address includes only the sequence of hook names, implicitly 408starting hook traversal at the local node. 409.Pp 410There are a couple of special possibilities for the node name. 411The name 412.Dq .\& 413(referred to as 414.Dq \&.: ) 415always refers to the local node. 416Also, nodes that have no global name may be addressed by their ID numbers, 417by enclosing the hex representation of the ID number within square brackets. 418Here are some examples of valid netgraph addresses: 419.Bd -literal -offset 4n -compact 420 421 .: 422 foo: 423 .:hook1 424 foo:hook1.hook2 425 [f057cd80]:hook1 426.Ed 427.Pp 428Consider the following set of nodes might be created for a site with 429a single physical frame relay line having two active logical DLCI channels, 430with RFC 1490 frames on DLCI 16 and PPP frames over DLCI 20: 431.Bd -literal 432[type SYNC ] [type FRAME] [type RFC1490] 433[ "Frame1" ](uplink)<-->(data)[<un-named>](dlci16)<-->(mux)[<un-named> ] 434[ A ] [ B ](dlci20)<---+ [ C ] 435 | 436 | [ type PPP ] 437 +>(mux)[<un-named>] 438 [ D ] 439.Ed 440.Pp 441One could always send a control message to node C from anywhere 442by using the name 443.Em "Frame1:uplink.dlci16" . 444Similarly, 445.Em "Frame1:uplink.dlci20" 446could reliably be used to reach node D, and node A could refer 447to node B as 448.Em ".:uplink" , 449or simply 450.Em "uplink" . 451Conversely, B can refer to A as 452.Em "data" . 453The address 454.Em "mux.data" 455could be used by both nodes C and D to address a message to node A. 456.Pp 457Note that this is only for 458.Em control messages . 459Data messages are routed one hop at a time, by specifying the departing 460hook, with each node making the next routing decision. 461So when B receives a frame on hook 462.Em data 463it decodes the frame relay header to determine the DLCI, 464and then forwards the unwrapped frame to either C or D. 465.Sh Netgraph Structures 466Interesting members of the node and hook structures are shown below: 467.Bd -literal 468struct ng_node { 469 char *name; /* Optional globally unique name */ 470 void *private; /* Node implementation private info */ 471 struct ng_type *type; /* The type of this node */ 472 int refs; /* Number of references to this struct */ 473 int numhooks; /* Number of connected hooks */ 474 hook_p hooks; /* Linked list of (connected) hooks */ 475}; 476typedef struct ng_node *node_p; 477 478struct ng_hook { 479 char *name; /* This node's name for this hook */ 480 void *private; /* Node implementation private info */ 481 int refs; /* Number of references to this struct */ 482 struct ng_node *node; /* The node this hook is attached to */ 483 struct ng_hook *peer; /* The other hook in this connected pair */ 484 struct ng_hook *next; /* Next in list of hooks for this node */ 485}; 486typedef struct ng_hook *hook_p; 487.Ed 488.Pp 489The maintenance of the name pointers, reference counts, and linked list 490of hooks for each node is handled automatically by the 491.Nm 492subsystem. 493Typically a node's private info contains a back-pointer to the node or hook 494structure, which counts as a new reference that must be registered by 495incrementing 496.Dv "node->refs" . 497.Pp 498From a hook you can obtain the corresponding node, and from 499a node the list of all active hooks. 500.Pp 501Node types are described by these structures: 502.Bd -literal 503/** How to convert a control message from binary <-> ASCII */ 504struct ng_cmdlist { 505 u_int32_t cookie; /* typecookie */ 506 int cmd; /* command number */ 507 const char *name; /* command name */ 508 const struct ng_parse_type *mesgType; /* args if !NGF_RESP */ 509 const struct ng_parse_type *respType; /* args if NGF_RESP */ 510}; 511 512struct ng_type { 513 u_int32_t version; /* Must equal NG_VERSION */ 514 const char *name; /* Unique type name */ 515 516 /* Module event handler */ 517 modeventhand_t mod_event; /* Handle load/unload (optional) */ 518 519 /* Constructor */ 520 int (*constructor)(node_p *node); /* Create a new node */ 521 522 /** Methods using the node **/ 523 int (*rcvmsg)(node_p node, /* Receive control message */ 524 struct ng_mesg *msg, /* The message */ 525 const char *retaddr, /* Return address */ 526 struct ng_mesg **resp); /* Synchronous response */ 527 int (*shutdown)(node_p node); /* Shutdown this node */ 528 int (*newhook)(node_p node, /* create a new hook */ 529 hook_p hook, /* Pre-allocated struct */ 530 const char *name); /* Name for new hook */ 531 532 /** Methods using the hook **/ 533 int (*connect)(hook_p hook); /* Confirm new hook attachment */ 534 int (*rcvdata)(hook_p hook, /* Receive data on a hook */ 535 struct mbuf *m, /* The data in an mbuf */ 536 meta_p meta); /* Meta-data, if any */ 537 int (*disconnect)(hook_p hook); /* Notify disconnection of hook */ 538 539 /** How to convert control messages binary <-> ASCII */ 540 const struct ng_cmdlist *cmdlist; /* Optional; may be NULL */ 541}; 542.Ed 543.Pp 544Control messages have the following structure: 545.Bd -literal 546#define NG_CMDSTRSIZ 16 /* Max command string (including null) */ 547 548struct ng_mesg { 549 struct ng_msghdr { 550 u_char version; /* Must equal NG_VERSION */ 551 u_char spare; /* Pad to 2 bytes */ 552 u_short arglen; /* Length of cmd/resp data */ 553 u_long flags; /* Message status flags */ 554 u_long token; /* Reply should have the same token */ 555 u_long typecookie; /* Node type understanding this message */ 556 u_long cmd; /* Command identifier */ 557 u_char cmdstr[NG_CMDSTRSIZ]; /* Cmd string (for debug) */ 558 } header; 559 char data[0]; /* Start of cmd/resp data */ 560}; 561 562#define NG_VERSION 1 /* Netgraph version */ 563#define NGF_ORIG 0x0000 /* Command */ 564#define NGF_RESP 0x0001 /* Response */ 565.Ed 566.Pp 567Control messages have the fixed header shown above, followed by a 568variable length data section which depends on the type cookie 569and the command. 570Each field is explained below: 571.Bl -tag -width xxx 572.It Dv version 573Indicates the version of netgraph itself. 574The current version is 575.Dv NG_VERSION . 576.It Dv arglen 577This is the length of any extra arguments, which begin at 578.Dv data . 579.It Dv flags 580Indicates whether this is a command or a response control message. 581.It Dv token 582The 583.Dv token 584is a means by which a sender can match a reply message to the 585corresponding command message; the reply always has the same token. 586.It Dv typecookie 587The corresponding node type's unique 32-bit value. 588If a node doesn't recognize the type cookie it must reject the message 589by returning 590.Er EINVAL . 591.Pp 592Each type should have an include file that defines the commands, 593argument format, and cookie for its own messages. 594The typecookie 595insures that the same header file was included by both sender and 596receiver; when an incompatible change in the header file is made, 597the typecookie 598.Em must 599be changed. 600The de facto method for generating unique type cookies is to take the 601seconds from the epoch at the time the header file is written 602(i.e. the output of 603.Dv "date -u +'%s'" ) . 604.Pp 605There is a predefined typecookie 606.Dv NGM_GENERIC_COOKIE 607for the 608.Dq generic 609node type, and 610a corresponding set of generic messages which all nodes understand. 611The handling of these messages is automatic. 612.It Dv command 613The identifier for the message command. 614This is type specific, 615and is defined in the same header file as the typecookie. 616.It Dv cmdstr 617Room for a short human readable version of 618.Dq command 619(for debugging purposes only). 620.El 621.Pp 622Some modules may choose to implement messages from more than one 623of the header files and thus recognize more than one type cookie. 624.Sh Control Message ASCII Form 625Control messages are in binary format for efficiency. 626However, for debugging and human interface purposes, 627and if the node type supports it, 628control messages may be converted to and from an equivalent 629.Tn ASCII 630form. 631The 632.Tn ASCII 633form is similar to the binary form, with two exceptions: 634.Pp 635.Bl -tag -compact -width xxx 636.It o 637The 638.Dv cmdstr 639header field must contain the 640.Tn ASCII 641name of the command, corresponding to the 642.Dv cmd 643header field. 644.It o 645The 646.Dv args 647field contains a NUL-terminated 648.Tn ASCII 649string version of the message arguments. 650.El 651.Pp 652In general, the arguments field of a control message can be any 653arbitrary C data type. 654Netgraph includes parsing routines to support 655some pre-defined datatypes in 656.Tn ASCII 657with this simple syntax: 658.Pp 659.Bl -tag -compact -width xxx 660.It o 661Integer types are represented by base 8, 10, or 16 numbers. 662.It o 663Strings are enclosed in double quotes and respect the normal 664C language backslash escapes. 665.It o 666IP addresses have the obvious form. 667.It o 668Arrays are enclosed in square brackets, with the elements listed 669consecutively starting at index zero. 670An element may have an optional index and equals sign preceding it. 671Whenever an element does not have an explicit index, the index is 672implicitly the previous element's index plus one. 673.It o 674Structures are enclosed in curly braces, and each field is specified 675in the form 676.Dq fieldname=value . 677.It o 678Any array element or structure field whose value is equal to its 679.Dq default value 680may be omitted. 681For integer types, the default value is usually zero; 682for string types, the empty string. 683.It o 684Array elements and structure fields may be specified in any order. 685.El 686.Pp 687Each node type may define its own arbitrary types by providing 688the necessary routines to parse and unparse. 689.Tn ASCII 690forms defined 691for a specific node type are documented in the documentation for 692that node type. 693.Sh Generic Control Messages 694There are a number of standard predefined messages that will work 695for any node, as they are supported directly by the framework itself. 696These are defined in 697.In netgraph/ng_message.h 698along with the basic layout of messages and other similar information. 699.Bl -tag -width xxx 700.It Dv NGM_CONNECT 701Connect to another node, using the supplied hook names on either end. 702.It Dv NGM_MKPEER 703Construct a node of the given type and then connect to it using the 704supplied hook names. 705.It Dv NGM_SHUTDOWN 706The target node should disconnect from all its neighbours and shut down. 707Persistent nodes such as those representing physical hardware 708might not disappear from the node namespace, but only reset themselves. 709The node must disconnect all of its hooks. 710This may result in neighbors shutting themselves down, and possibly a 711cascading shutdown of the entire connected graph. 712.It Dv NGM_NAME 713Assign a name to a node. 714Nodes can exist without having a name, and this 715is the default for nodes created using the 716.Dv NGM_MKPEER 717method. 718Such nodes can only be addressed relatively or by their ID number. 719.It Dv NGM_RMHOOK 720Ask the node to break a hook connection to one of its neighbours. 721Both nodes will have their 722.Dq disconnect 723method invoked. 724Either node may elect to totally shut down as a result. 725.It Dv NGM_NODEINFO 726Asks the target node to describe itself. 727The four returned fields are the node name (if named), the node type, 728the node ID and the number of hooks attached. 729The ID is an internal number unique to that node. 730.It Dv NGM_LISTHOOKS 731This returns the information given by 732.Dv NGM_NODEINFO , 733but in addition 734includes an array of fields describing each link, and the description for 735the node at the far end of that link. 736.It Dv NGM_LISTNAMES 737This returns an array of node descriptions (as for 738.Dv NGM_NODEINFO ")" 739where each entry of the array describes a named node. 740All named nodes will be described. 741.It Dv NGM_LISTNODES 742This is the same as 743.Dv NGM_LISTNAMES 744except that all nodes are listed regardless of whether they have a name or not. 745.It Dv NGM_LISTTYPES 746This returns a list of all currently installed netgraph types. 747.It Dv NGM_TEXT_STATUS 748The node may return a text formatted status message. 749The status information is determined entirely by the node type. 750It is the only "generic" message 751that requires any support within the node itself and as such the node may 752elect to not support this message. 753The text response must be less than 754.Dv NG_TEXTRESPONSE 755bytes in length (presently 1024). 756This can be used to return general 757status information in human readable form. 758.It Dv NGM_BINARY2ASCII 759This message converts a binary control message to its 760.Tn ASCII 761form. 762The entire control message to be converted is contained within the 763arguments field of the 764.Dv NGM_BINARY2ASCII 765message itself. 766If successful, the reply will contain the same control message in 767.Tn ASCII 768form. 769A node will typically only know how to translate messages that it 770itself understands, so the target node of the 771.Dv NGM_BINARY2ASCII 772is often the same node that would actually receive that message. 773.It Dv NGM_ASCII2BINARY 774The opposite of 775.Dv NGM_BINARY2ASCII . 776The entire control message to be converted, in 777.Tn ASCII 778form, is contained 779in the arguments section of the 780.Dv NGM_ASCII2BINARY 781and need only have the 782.Dv flags , 783.Dv cmdstr , 784and 785.Dv arglen 786header fields filled in, plus the NUL-terminated string version of 787the arguments in the arguments field. 788If successful, the reply 789contains the binary version of the control message. 790.El 791.Sh Metadata 792Data moving through the 793.Nm 794system can be accompanied by meta-data that describes some 795aspect of that data. 796The form of the meta-data is a fixed header, 797which contains enough information for most uses, and can optionally 798be supplemented by trailing 799.Em option 800structures, which contain a 801.Em cookie 802(see the section on control messages), an identifier, a length and optional 803data. 804If a node does not recognize the cookie associated with an option, 805it should ignore that option. 806.Pp 807Meta data might include such things as priority, discard eligibility, 808or special processing requirements. 809It might also mark a packet for debug status, etc. 810The use of meta-data is still experimental. 811.Sh INITIALIZATION 812The base 813.Nm 814code may either be statically compiled 815into the kernel or else loaded dynamically as a KLD via 816.Xr kldload 8 . 817In the former case, include 818.Pp 819.D1 Cd options NETGRAPH 820.Pp 821in your kernel configuration file. 822You may also include selected 823node types in the kernel compilation, for example: 824.Bd -unfilled -offset indent 825.Cd options NETGRAPH 826.Cd options NETGRAPH_SOCKET 827.Cd options NETGRAPH_ECHO 828.Ed 829.Pp 830Once the 831.Nm 832subsystem is loaded, individual node types may be loaded at any time 833as KLD modules via 834.Xr kldload 8 . 835Moreover, 836.Nm 837knows how to automatically do this; when a request to create a new 838node of unknown type 839.Em type 840is made, 841.Nm 842will attempt to load the KLD module 843.Pa ng_type.ko . 844.Pp 845Types can also be installed at boot time, as certain device drivers 846may want to export each instance of the device as a netgraph node. 847.Pp 848In general, new types can be installed at any time from within the 849kernel by calling 850.Fn ng_newtype , 851supplying a pointer to the type's 852.Dv struct ng_type 853structure. 854.Pp 855The 856.Fn NETGRAPH_INIT 857macro automates this process by using a linker set. 858.Sh EXISTING NODE TYPES 859Several node types currently exist. 860Each is fully documented in its own man page: 861.Bl -tag -width xxx 862.It SOCKET 863The socket type implements two new sockets in the new protocol domain 864.Dv PF_NETGRAPH . 865The new sockets protocols are 866.Dv NG_DATA 867and 868.Dv NG_CONTROL , 869both of type 870.Dv SOCK_DGRAM . 871Typically one of each is associated with a socket node. 872When both sockets have closed, the node will shut down. 873The 874.Dv NG_DATA 875socket is used for sending and receiving data, while the 876.Dv NG_CONTROL 877socket is used for sending and receiving control messages. 878Data and control messages are passed using the 879.Xr sendto 2 880and 881.Xr recvfrom 2 882calls, using a 883.Dv struct sockaddr_ng 884socket address. 885.It HOLE 886Responds only to generic messages and is a 887.Dq black hole 888for data, Useful for testing. 889Always accepts new hooks. 890.It ECHO 891Responds only to generic messages and always echoes data back through the 892hook from which it arrived. 893Returns any non generic messages as their own response. 894Useful for testing. 895Always accepts new hooks. 896.It TEE 897This node is useful for 898.Dq snooping . 899It has 4 hooks: 900.Dv left , 901.Dv right , 902.Dv left2right , 903and 904.Dv right2left . 905Data entering from the right is passed to the left and duplicated on 906.Dv right2left , 907and data entering from the left is passed to the right and 908duplicated on 909.Dv left2right . 910Data entering from 911.Dv left2right 912is sent to the right and data from 913.Dv right2left 914to left. 915.It RFC1490 MUX 916Encapsulates/de-encapsulates frames encoded according to RFC 1490. 917Has a hook for the encapsulated packets 918.Pq Dq downstream 919and one hook 920for each protocol (i.e. IP, PPP, etc.). 921.It FRAME RELAY MUX 922Encapsulates/de-encapsulates Frame Relay frames. 923Has a hook for the encapsulated packets 924.Pq Dq downstream 925and one hook 926for each DLCI. 927.It FRAME RELAY LMI 928Automatically handles frame relay 929.Dq LMI 930(link management interface) operations and packets. 931Automatically probes and detects which of several LMI standards 932is in use at the exchange. 933.It TTY 934This node is also a line discipline. 935It simply converts between mbuf frames and sequential serial data, 936allowing a tty to appear as a netgraph node. 937It has a programmable 938.Dq hotkey 939character. 940.It ASYNC 941This node encapsulates and de-encapsulates asynchronous frames 942according to RFC 1662. 943This is used in conjunction with the TTY node 944type for supporting PPP links over asynchronous serial lines. 945.It INTERFACE 946This node is also a system networking interface. 947It has hooks representing each protocol family (IP, AppleTalk, etc.) 948and appears in the output of 949.Xr ifconfig 8 . 950The interfaces are named 951.Em ng0 , 952.Em ng1 , 953etc. 954.El 955.Sh NOTES 956Whether a named node exists can be checked by trying to send a control message 957to it (e.g., 958.Dv NGM_NODEINFO ) . 959If it does not exist, 960.Er ENOENT 961will be returned. 962.Pp 963All data messages are mbuf chains with the M_PKTHDR flag set. 964.Pp 965Nodes are responsible for freeing what they allocate. 966There are three exceptions: 967.Bl -tag -width xxxx 968.It 1 969Mbufs sent across a data link are never to be freed by the sender. 970.It 2 971Any meta-data information traveling with the data has the same restriction. 972It might be freed by any node the data passes through, and a 973.Dv NULL 974passed onwards, but the caller will never free it. 975Two macros 976.Fn NG_FREE_META "meta" 977and 978.Fn NG_FREE_DATA "m" "meta" 979should be used if possible to free data and meta data (see 980.In netgraph/netgraph.h ) . 981.It 3 982Messages sent using 983.Fn ng_send_msg 984are freed by the callee. 985As in the case above, the addresses 986associated with the message are freed by whatever allocated them so the 987recipient should copy them if it wants to keep that information. 988.El 989.Sh FILES 990.Bl -tag -width xxxxx -compact 991.It In netgraph/netgraph.h 992Definitions for use solely within the kernel by 993.Nm 994nodes. 995.It In netgraph/ng_message.h 996Definitions needed by any file that needs to deal with 997.Nm 998messages. 999.It In netgraph/socket/ng_socket.h 1000Definitions needed to use 1001.Nm 1002socket type nodes. 1003.It In netgraph/{type}/ng_{type}.h 1004Definitions needed to use 1005.Nm 1006{type} 1007nodes, including the type cookie definition. 1008.It Pa /boot/modules/netgraph.ko 1009Netgraph subsystem loadable KLD module. 1010.It Pa /boot/modules/ng_{type}.ko 1011Loadable KLD module for node type {type}. 1012.El 1013.Sh USER MODE SUPPORT 1014There is a library for supporting user-mode programs that wish 1015to interact with the netgraph system. 1016See 1017.Xr netgraph 3 1018for details. 1019.Pp 1020Two user-mode support programs, 1021.Xr ngctl 8 1022and 1023.Xr nghook 8 , 1024are available to assist manual configuration and debugging. 1025.Pp 1026There are a few useful techniques for debugging new node types. 1027First, implementing new node types in user-mode first 1028makes debugging easier. 1029The 1030.Em tee 1031node type is also useful for debugging, especially in conjunction with 1032.Xr ngctl 8 1033and 1034.Xr nghook 8 . 1035.Sh SEE ALSO 1036.Xr socket 2 , 1037.Xr netgraph 3 , 1038.Xr ng_async 4 , 1039.Xr ng_bpf 4 , 1040.Xr ng_bridge 4 , 1041.Xr ng_cisco 4 , 1042.Xr ng_echo 4 , 1043.Xr ng_eiface 4 , 1044.Xr ng_etf 4 , 1045.Xr ng_ether 4 , 1046.Xr ng_frame_relay 4 , 1047.Xr ng_hole 4 , 1048.Xr ng_iface 4 , 1049.Xr ng_ksocket 4 , 1050.Xr ng_l2tp 4 , 1051.Xr ng_lmi 4 , 1052.Xr ng_mppc 4 , 1053.Xr ng_one2many 4 , 1054.Xr ng_ppp 4 , 1055.Xr ng_pppoe 4 , 1056.Xr ng_rfc1490 4 , 1057.Xr ng_socket 4 , 1058.Xr ng_tee 4 , 1059.Xr ng_tty 4 , 1060.Xr ng_UI 4 , 1061.Xr ng_vjc 4 , 1062.Xr ngctl 8 , 1063.Xr nghook 8 1064.Sh HISTORY 1065The 1066.Nm 1067system was designed and first implemented at Whistle Communications, Inc.\& 1068in a version of 1069.Fx 2.2 1070customized for the Whistle InterJet. 1071It first made its debut in the main tree in 1072.Fx 3.4 . 1073.Sh AUTHORS 1074.An -nosplit 1075.An Julian Elischer Aq Mt julian@FreeBSD.org , 1076with contributions by 1077.An Archie Cobbs Aq Mt archie@FreeBSD.org . 1078