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.\" $DragonFly: src/share/man/man4/netgraph.4,v 1.2 2003/06/17 04:36:59 dillon Exp $ 38.\" $Whistle: netgraph.4,v 1.7 1999/01/28 23:54:52 julian Exp $ 39.\" 40.Dd January 19, 1999 41.Dt NETGRAPH 4 42.Os 43.Sh NAME 44.Nm netgraph 45.Nd graph based kernel networking subsystem 46.Sh DESCRIPTION 47The 48.Nm 49system provides a uniform and modular system for the implementation 50of kernel objects which perform various networking functions. The objects, 51known as 52.Em nodes , 53can be arranged into arbitrarily complicated graphs. Nodes 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. It provides: 62.Pp 63.Bl -bullet -compact -offset 2n 64.It 65A flexible way of combining protocol and link level drivers 66.It 67A modular way to implement new protocols 68.It 69A common framework for kernel entities to inter-communicate 70.It 71A reasonably fast, kernel-based implementation 72.El 73.Sh Nodes and Types 74The most fundamental concept in 75.Nm 76is that of a 77.Em node . 78All nodes implement a number of predefined methods which allow them 79to interact with other nodes in a well defined manner. 80.Pp 81Each node has a 82.Em type , 83which is a static property of the node determined at node creation time. 84A node's type is described by a unique 85.Tn ASCII 86type name. 87The type implies what the node does and how it may be connected 88to other nodes. 89.Pp 90In object-oriented language, types are classes and nodes are instances 91of their respective class. All node types are subclasses of the generic node 92type, and hence inherit certain common functionality and capabilities 93(e.g., the ability to have an 94.Tn ASCII 95name). 96.Pp 97Nodes may be assigned a globally unique 98.Tn ASCII 99name which can be 100used to refer to the node. 101The name must not contain the characters 102.Dq .\& 103or 104.Dq \&: 105and is limited to 106.Dv "NG_NODELEN + 1" 107characters (including NUL byte). 108.Pp 109Each node instance has a unique 110.Em ID number 111which is expressed as a 32-bit hex value. This value may be used to 112refer to a node when there is no 113.Tn ASCII 114name assigned to it. 115.Sh Hooks 116Nodes are connected to other nodes by connecting a pair of 117.Em hooks , 118one from each node. Data flows bidirectionally between nodes along 119connected pairs of hooks. A node may have as many hooks as it 120needs, and may assign whatever meaning it wants to a hook. 121.Pp 122Hooks have these properties: 123.Pp 124.Bl -bullet -compact -offset 2n 125.It 126A hook has an 127.Tn ASCII 128name which is unique among all hooks 129on that node (other hooks on other nodes may have the same name). 130The name must not contain a 131.Dq .\& 132or a 133.Dq \&: 134and is 135limited to 136.Dv "NG_HOOKLEN + 1" 137characters (including NUL byte). 138.It 139A hook is always connected to another hook. That is, hooks are 140created at the time they are connected, and breaking an edge by 141removing either hook destroys both hooks. 142.El 143.Pp 144A node may decide to assign special meaning to some hooks. 145For example, connecting to the hook named 146.Dq debug 147might trigger 148the node to start sending debugging information to that hook. 149.Sh Data Flow 150Two types of information flow between nodes: data messages and 151control messages. Data messages are passed in mbuf chains along the edges 152in the graph, one edge at a time. The first mbuf in a chain must have the 153.Dv M_PKTHDR 154flag set. Each node decides how to handle data coming in on its hooks. 155.Pp 156Control messages are type-specific C structures sent from one node 157directly to some arbitrary other node. Control messages have a common 158header format, followed by type-specific data, and are binary structures 159for efficiency. However, node types also may support conversion of the 160type specific data between binary and 161.Tn ASCII 162for debugging and human interface purposes (see the 163.Dv NGM_ASCII2BINARY 164and 165.Dv NGM_BINARY2ASCII 166generic control messages below). Nodes are not required to support 167these conversions. 168.Pp 169There are two ways to address a control message. If 170there is a sequence of edges connecting the two nodes, the message 171may be 172.Dq source routed 173by specifying the corresponding sequence 174of hooks as the destination address for the message (relative 175addressing). Otherwise, the recipient node global 176.Tn ASCII 177name 178(or equivalent ID based name) is used as the destination address 179for the message (absolute addressing). The two types of addressing 180may be combined, by specifying an absolute start node and a sequence 181of hooks. 182.Pp 183Messages often represent commands that are followed by a reply message 184in the reverse direction. To facilitate this, the recipient of a 185control message is supplied with a 186.Dq return address 187that is suitable for addressing a reply. 188.Pp 189Each control message contains a 32 bit value called a 190.Em typecookie 191indicating the type of the message, i.e., how to interpret it. 192Typically each type defines a unique typecookie for the messages 193that it understands. However, a node may choose to recognize and 194implement more than one type of message. 195.Sh Netgraph is Functional 196In order to minimize latency, most 197.Nm 198operations are functional. 199That is, data and control messages are delivered by making function 200calls rather than by using queues and mailboxes. For example, if node 201A wishes to send a data mbuf to neighboring node B, it calls the 202generic 203.Nm 204data delivery function. This function in turn locates 205node B and calls B's 206.Dq receive data 207method. While this mode of operation 208results in good performance, it has a few implications for node 209developers: 210.Pp 211.Bl -bullet -compact -offset 2n 212.It 213Whenever a node delivers a data or control message, the node 214may need to allow for the possibility of receiving a returning 215message before the original delivery function call returns. 216.It 217Netgraph nodes and support routines generally run at 218.Fn splnet . 219However, some nodes may want to send data and control messages 220from a different priority level. Netgraph supplies queueing routines which 221utilize the NETISR system to move message delivery to 222.Fn splnet . 223Note that messages are always received at 224.Fn splnet . 225.It 226It's possible for an infinite loop to occur if the graph contains cycles. 227.El 228.Pp 229So far, these issues have not proven problematical in practice. 230.Sh Interaction With Other Parts of the Kernel 231A node may have a hidden interaction with other components of the 232kernel outside of the 233.Nm 234subsystem, such as device hardware, 235kernel protocol stacks, etc. In fact, one of the benefits of 236.Nm 237is the ability to join disparate kernel networking entities together in a 238consistent communication framework. 239.Pp 240An example is the node type 241.Em socket 242which is both a netgraph node and a 243.Xr socket 2 244.Bx 245socket in the protocol family 246.Dv PF_NETGRAPH . 247Socket nodes allow user processes to participate in 248.Nm . 249Other nodes communicate with socket nodes using the usual methods, and the 250node hides the fact that it is also passing information to and from a 251cooperating user process. 252.Pp 253Another example is a device driver that presents 254a node interface to the hardware. 255.Sh Node Methods 256Nodes are notified of the following actions via function calls 257to the following node methods (all at 258.Fn splnet ) 259and may accept or reject that action (by returning the appropriate 260error code): 261.Bl -tag -width xxx 262.It Creation of a new node 263The constructor for the type is called. If creation of a new node is 264allowed, the constructor must call the generic node creation 265function (in object-oriented terms, the superclass constructor) 266and then allocate any special resources it needs. For nodes that 267correspond to hardware, this is typically done during the device 268attach routine. Often a global 269.Tn ASCII 270name corresponding to the 271device name is assigned here as well. 272.It Creation of a new hook 273The hook is created and tentatively 274linked to the node, and the node is told about the name that will be 275used to describe this hook. The node sets up any special data structures 276it needs, or may reject the connection, based on the name of the hook. 277.It Successful connection of two hooks 278After both ends have accepted their 279hooks, and the links have been made, the nodes get a chance to 280find out who their peer is across the link and can then decide to reject 281the connection. Tear-down is automatic. 282.It Destruction of a hook 283The node is notified of a broken connection. The node may consider some hooks 284to be critical to operation and others to be expendable: the disconnection 285of one hook may be an acceptable event while for another it 286may affect a total shutdown for the node. 287.It Shutdown of a node 288This method allows a node to clean up 289and to ensure that any actions that need to be performed 290at this time are taken. The method must call the generic (i.e., superclass) 291node destructor to get rid of the generic components of the node. 292Some nodes (usually associated with a piece of hardware) may be 293.Em persistent 294in that a shutdown breaks all edges and resets the node, 295but doesn't remove it, in which case the generic destructor is not called. 296.El 297.Sh Sending and Receiving Data 298Three other methods are also supported by all nodes: 299.Bl -tag -width xxx 300.It Receive data message 301An mbuf chain is passed to the node. 302The node is notified on which hook the data arrived, 303and can use this information in its processing decision. 304The node must must always 305.Fn m_freem 306the mbuf chain on completion or error, or pass it on to another node 307(or kernel module) which will then be responsible for freeing it. 308.Pp 309In addition to the mbuf chain itself there is also a pointer to a 310structure describing meta-data about the message 311(e.g. priority information). This pointer may be 312.Dv NULL 313if there is no additional information. The format for this information is 314described in 315.Pa sys/netgraph/netgraph.h . 316The memory for meta-data must allocated via 317.Fn malloc 318with type 319.Dv M_NETGRAPH . 320As with the data itself, it is the receiver's responsibility to 321.Fn free 322the meta-data. If the mbuf chain is freed the meta-data must 323be freed at the same time. If the meta-data is freed but the 324real data on is passed on, then a 325.Dv NULL 326pointer must be substituted. 327.Pp 328The receiving node may decide to defer the data by queueing it in the 329.Nm 330NETISR system (see below). 331.Pp 332The structure and use of meta-data is still experimental, but is 333presently used in frame-relay to indicate that management packets 334should be queued for transmission 335at a higher priority than data packets. This is required for 336conformance with Frame Relay standards. 337.Pp 338.It Receive queued data message 339Usually this will be the same function as 340.Em Receive data message. 341This is the entry point called when a data message is being handed to 342the node after having been queued in the NETISR system. 343This allows a node to decide in the 344.Em Receive data message 345method that a message should be deferred and queued, 346and be sure that when it is processed from the queue, 347it will not be queued again. 348.It Receive control message 349This method is called when a control message is addressed to the node. 350A return address is always supplied, giving the address of the node 351that originated the message so a reply message can be sent anytime later. 352.Pp 353It is possible for a synchronous reply to be made, and in fact this 354is more common in practice. 355This is done by setting a pointer (supplied as an extra function parameter) 356to point to the reply. 357Then when the control message delivery function returns, 358the caller can check if this pointer has been made non-NULL, 359and if so then it points to the reply message allocated via 360.Fn malloc 361and containing the synchronous response. In both directions, 362(request and response) it is up to the 363receiver of that message to 364.Fn free 365the control message buffer. All control messages and replies are 366allocated with 367.Fn malloc 368type 369.Dv M_NETGRAPH . 370.El 371.Pp 372Much use has been made of reference counts, so that nodes being 373free'd of all references are automatically freed, and this behaviour 374has been tested and debugged to present a consistent and trustworthy 375framework for the 376.Dq type module 377writer to use. 378.Sh Addressing 379The 380.Nm 381framework provides an unambiguous and simple to use method of specifically 382addressing any single node in the graph. The naming of a node is 383independent of its type, in that another node, or external component 384need not know anything about the node's type in order to address it so as 385to send it a generic message type. Node and hook names should be 386chosen so as to make addresses meaningful. 387.Pp 388Addresses are either absolute or relative. An absolute address begins 389with a node name, (or ID), followed by a colon, followed by a sequence of hook 390names separated by periods. This addresses the node reached by starting 391at the named node and following the specified sequence of hooks. 392A relative address includes only the sequence of hook names, implicitly 393starting hook traversal at the local node. 394.Pp 395There are a couple of special possibilities for the node name. 396The name 397.Dq .\& 398(referred to as 399.Dq \&.: ) 400always refers to the local node. 401Also, nodes that have no global name may be addressed by their ID numbers, 402by enclosing the hex representation of the ID number within square brackets. 403Here are some examples of valid netgraph addresses: 404.Bd -literal -offset 4n -compact 405 406 .: 407 foo: 408 .:hook1 409 foo:hook1.hook2 410 [f057cd80]:hook1 411.Ed 412.Pp 413Consider the following set of nodes might be created for a site with 414a single physical frame relay line having two active logical DLCI channels, 415with RFC-1490 frames on DLCI 16 and PPP frames over DLCI 20: 416.Pp 417.Bd -literal 418[type SYNC ] [type FRAME] [type RFC1490] 419[ "Frame1" ](uplink)<-->(data)[<un-named>](dlci16)<-->(mux)[<un-named> ] 420[ A ] [ B ](dlci20)<---+ [ C ] 421 | 422 | [ type PPP ] 423 +>(mux)[<un-named>] 424 [ D ] 425.Ed 426.Pp 427One could always send a control message to node C from anywhere 428by using the name 429.Em "Frame1:uplink.dlci16" . 430Similarly, 431.Em "Frame1:uplink.dlci20" 432could reliably be used to reach node D, and node A could refer 433to node B as 434.Em ".:uplink" , 435or simply 436.Em "uplink" . 437Conversely, B can refer to A as 438.Em "data" . 439The address 440.Em "mux.data" 441could be used by both nodes C and D to address a message to node A. 442.Pp 443Note that this is only for 444.Em control messages . 445Data messages are routed one hop at a time, by specifying the departing 446hook, with each node making the next routing decision. So when B 447receives a frame on hook 448.Em data 449it decodes the frame relay header to determine the DLCI, 450and then forwards the unwrapped frame to either C or D. 451.Pp 452A similar graph might be used to represent multi-link PPP running 453over an ISDN line: 454.Pp 455.Bd -literal 456[ type BRI ](B1)<--->(link1)[ type MPP ] 457[ "ISDN1" ](B2)<--->(link2)[ (no name) ] 458[ ](D) <-+ 459 | 460 +----------------+ 461 | 462 +->(switch)[ type Q.921 ](term1)<---->(datalink)[ type Q.931 ] 463 [ (no name) ] [ (no name) ] 464.Ed 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_CMDSTRLEN 15 /* Max command string (16 with 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_CMDSTRLEN+1]; /* 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. Each field is explained below: 570.Bl -tag -width xxx 571.It Dv version 572Indicates the version of netgraph itself. The current version is 573.Dv NG_VERSION . 574.It Dv arglen 575This is the length of any extra arguments, which begin at 576.Dv data . 577.It Dv flags 578Indicates whether this is a command or a response control message. 579.It Dv token 580The 581.Dv token 582is a means by which a sender can match a reply message to the 583corresponding command message; the reply always has the same token. 584.Pp 585.It Dv typecookie 586The corresponding node type's unique 32-bit value. 587If a node doesn't recognize the type cookie it must reject the message 588by returning 589.Er EINVAL . 590.Pp 591Each type should have an include file that defines the commands, 592argument format, and cookie for its own messages. 593The typecookie 594insures that the same header file was included by both sender and 595receiver; when an incompatible change in the header file is made, 596the typecookie 597.Em must 598be changed. 599The de facto method for generating unique type cookies is to take the 600seconds from the epoch at the time the header file is written 601(i.e., the output of 602.Dv "date -u +'%s'" ) . 603.Pp 604There is a predefined typecookie 605.Dv NGM_GENERIC_COOKIE 606for the 607.Dq generic 608node type, and 609a corresponding set of generic messages which all nodes understand. 610The handling of these messages is automatic. 611.It Dv command 612The identifier for the message command. This is type specific, 613and is defined in the same header file as the typecookie. 614.It Dv cmdstr 615Room for a short human readable version of 616.Dq command 617(for debugging purposes only). 618.El 619.Pp 620Some modules may choose to implement messages from more than one 621of the header files and thus recognize more than one type cookie. 622.Sh Control Message ASCII Form 623Control messages are in binary format for efficiency. However, for 624debugging and human interface purposes, and if the node type supports 625it, control messages may be converted to and from an equivalent 626.Tn ASCII 627form. The 628.Tn ASCII 629form is similar to the binary form, with two exceptions: 630.Pp 631.Bl -tag -compact -width xxx 632.It o 633The 634.Dv cmdstr 635header field must contain the 636.Tn ASCII 637name of the command, corresponding to the 638.Dv cmd 639header field. 640.It o 641The 642.Dv args 643field contains a NUL-terminated 644.Tn ASCII 645string version of the message arguments. 646.El 647.Pp 648In general, the arguments field of a control messgage can be any 649arbitrary C data type. Netgraph includes parsing routines to support 650some pre-defined datatypes in 651.Tn ASCII 652with this simple syntax: 653.Pp 654.Bl -tag -compact -width xxx 655.It o 656Integer types are represented by base 8, 10, or 16 numbers. 657.It o 658Strings are enclosed in double quotes and respect the normal 659C language backslash escapes. 660.It o 661IP addresses have the obvious form. 662.It o 663Arrays are enclosed in square brackets, with the elements listed 664consecutively starting at index zero. An element may have an optional 665index and equals sign preceding it. Whenever an element 666does not have an explicit index, the index is implicitly the previous 667element's index plus one. 668.It o 669Structures are enclosed in curly braces, and each field is specified 670in the form 671.Dq fieldname=value . 672.It o 673Any array element or structure field whose value is equal to its 674.Dq default value 675may be omitted. For integer types, the default value 676is usually zero; for string types, the empty string. 677.It o 678Array elements and structure fields may be specified in any order. 679.El 680.Pp 681Each node type may define its own arbitrary types by providing 682the necessary routines to parse and unparse. 683.Tn ASCII 684forms defined 685for a specific node type are documented in the documentation for 686that node type. 687.Sh Generic Control Messages 688There are a number of standard predefined messages that will work 689for any node, as they are supported directly by the framework itself. 690These are defined in 691.Pa ng_message.h 692along with the basic layout of messages and other similar information. 693.Bl -tag -width xxx 694.It Dv NGM_CONNECT 695Connect to another node, using the supplied hook names on either end. 696.It Dv NGM_MKPEER 697Construct a node of the given type and then connect to it using the 698supplied hook names. 699.It Dv NGM_SHUTDOWN 700The target node should disconnect from all its neighbours and shut down. 701Persistent nodes such as those representing physical hardware 702might not disappear from the node namespace, but only reset themselves. 703The node must disconnect all of its hooks. 704This may result in neighbors shutting themselves down, and possibly a 705cascading shutdown of the entire connected graph. 706.It Dv NGM_NAME 707Assign a name to a node. Nodes can exist without having a name, and this 708is the default for nodes created using the 709.Dv NGM_MKPEER 710method. Such nodes can only be addressed relatively or by their ID number. 711.It Dv NGM_RMHOOK 712Ask the node to break a hook connection to one of its neighbours. 713Both nodes will have their 714.Dq disconnect 715method invoked. 716Either node may elect to totally shut down as a result. 717.It Dv NGM_NODEINFO 718Asks the target node to describe itself. The four returned fields 719are the node name (if named), the node type, the node ID and the 720number of hooks attached. The ID is an internal number unique to that node. 721.It Dv NGM_LISTHOOKS 722This returns the information given by 723.Dv NGM_NODEINFO , 724but in addition 725includes an array of fields describing each link, and the description for 726the node at the far end of that link. 727.It Dv NGM_LISTNAMES 728This returns an array of node descriptions (as for 729.Dv NGM_NODEINFO ")" 730where each entry of the array describes a named node. 731All named nodes will be described. 732.It Dv NGM_LISTNODES 733This is the same as 734.Dv NGM_LISTNAMES 735except that all nodes are listed regardless of whether they have a name or not. 736.It Dv NGM_LISTTYPES 737This returns a list of all currently installed netgraph types. 738.It Dv NGM_TEXT_STATUS 739The node may return a text formatted status message. 740The status information is determined entirely by the node type. 741It is the only "generic" message 742that requires any support within the node itself and as such the node may 743elect to not support this message. The text response must be less than 744.Dv NG_TEXTRESPONSE 745bytes in length (presently 1024). This can be used to return general 746status information in human readable form. 747.It Dv NGM_BINARY2ASCII 748This message converts a binary control message to its 749.Tn ASCII 750form. 751The entire control message to be converted is contained within the 752arguments field of the 753.Dv NGM_BINARY2ASCII 754message itself. If successful, the reply will contain the same control 755message in 756.Tn ASCII 757form. 758A node will typically only know how to translate messages that it 759itself understands, so the target node of the 760.Dv NGM_BINARY2ASCII 761is often the same node that would actually receive that message. 762.It Dv NGM_ASCII2BINARY 763The opposite of 764.Dv NGM_BINARY2ASCII . 765The entire control message to be converted, in 766.Tn ASCII 767form, is contained 768in the arguments section of the 769.Dv NGM_ASCII2BINARY 770and need only have the 771.Dv flags , 772.Dv cmdstr , 773and 774.Dv arglen 775header fields filled in, plus the NUL-terminated string version of 776the arguments in the arguments field. If successful, the reply 777contains the binary version of the control message. 778.El 779.Sh Metadata 780Data moving through the 781.Nm 782system can be accompanied by meta-data that describes some 783aspect of that data. The form of the meta-data is a fixed header, 784which contains enough information for most uses, and can optionally 785be supplemented by trailing 786.Em option 787structures, which contain a 788.Em cookie 789(see the section on control messages), an identifier, a length and optional 790data. If a node does not recognize the cookie associated with an option, 791it should ignore that option. 792.Pp 793Meta data might include such things as priority, discard eligibility, 794or special processing requirements. It might also mark a packet for 795debug status, etc. The use of meta-data is still experimental. 796.Sh INITIALIZATION 797The base 798.Nm 799code may either be statically compiled 800into the kernel or else loaded dynamically as a KLD via 801.Xr kldload 8 . 802In the former case, include 803.Pp 804.Dl options NETGRAPH 805.Pp 806in your kernel configuration file. You may also include selected 807node types in the kernel compilation, for example: 808.Bd -literal -offset indent 809options NETGRAPH 810options NETGRAPH_SOCKET 811options NETGRAPH_ECHO 812.Ed 813.Pp 814Once the 815.Nm 816subsystem is loaded, individual node types may be loaded at any time 817as KLD modules via 818.Xr kldload 8 . 819Moreover, 820.Nm 821knows how to automatically do this; when a request to create a new 822node of unknown type 823.Em type 824is made, 825.Nm 826will attempt to load the KLD module 827.Pa ng_type.ko . 828.Pp 829Types can also be installed at boot time, as certain device drivers 830may want to export each instance of the device as a netgraph node. 831.Pp 832In general, new types can be installed at any time from within the 833kernel by calling 834.Fn ng_newtype , 835supplying a pointer to the type's 836.Dv struct ng_type 837structure. 838.Pp 839The 840.Fn NETGRAPH_INIT 841macro automates this process by using a linker set. 842.Sh EXISTING NODE TYPES 843Several node types currently exist. Each is fully documented 844in its own man page: 845.Bl -tag -width xxx 846.It SOCKET 847The socket type implements two new sockets in the new protocol domain 848.Dv PF_NETGRAPH . 849The new sockets protocols are 850.Dv NG_DATA 851and 852.Dv NG_CONTROL , 853both of type 854.Dv SOCK_DGRAM . 855Typically one of each is associated with a socket node. 856When both sockets have closed, the node will shut down. The 857.Dv NG_DATA 858socket is used for sending and receiving data, while the 859.Dv NG_CONTROL 860socket is used for sending and receiving control messages. 861Data and control messages are passed using the 862.Xr sendto 2 863and 864.Xr recvfrom 2 865calls, using a 866.Dv struct sockaddr_ng 867socket address. 868.Pp 869.It HOLE 870Responds only to generic messages and is a 871.Dq black hole 872for data, Useful for testing. Always accepts new hooks. 873.Pp 874.It ECHO 875Responds only to generic messages and always echoes data back through the 876hook from which it arrived. Returns any non generic messages as their 877own response. Useful for testing. Always accepts new hooks. 878.Pp 879.It TEE 880This node is useful for 881.Dq snooping . 882It has 4 hooks: 883.Dv left , 884.Dv right , 885.Dv left2right , 886and 887.Dv right2left . 888Data entering from the right is passed to the left and duplicated on 889.Dv right2left , 890and data entering from the left is passed to the right and 891duplicated on 892.Dv left2right . 893Data entering from 894.Dv left2right 895is sent to the right and data from 896.Dv right2left 897to left. 898.Pp 899.It RFC1490 MUX 900Encapsulates/de-encapsulates frames encoded according to RFC 1490. 901Has a hook for the encapsulated packets 902.Pq Dq downstream 903and one hook 904for each protocol (i.e., IP, PPP, etc.). 905.Pp 906.It FRAME RELAY MUX 907Encapsulates/de-encapsulates Frame Relay frames. 908Has a hook for the encapsulated packets 909.Pq Dq downstream 910and one hook 911for each DLCI. 912.Pp 913.It FRAME RELAY LMI 914Automatically handles frame relay 915.Dq LMI 916(link management interface) operations and packets. 917Automatically probes and detects which of several LMI standards 918is in use at the exchange. 919.Pp 920.It TTY 921This node is also a line discipline. It simply converts between mbuf 922frames and sequential serial data, allowing a tty to appear as a netgraph 923node. It has a programmable 924.Dq hotkey 925character. 926.Pp 927.It ASYNC 928This node encapsulates and de-encapsulates asynchronous frames 929according to RFC 1662. This is used in conjunction with the TTY node 930type for supporting PPP links over asynchronous serial lines. 931.Pp 932.It INTERFACE 933This node is also a system networking interface. It has hooks representing 934each protocol family (IP, AppleTalk, IPX, etc.) and appears in the output of 935.Xr ifconfig 8 . 936The interfaces are named 937.Em ng0 , 938.Em ng1 , 939etc. 940.El 941.Sh NOTES 942Whether a named node exists can be checked by trying to send a control message 943to it (e.g., 944.Dv NGM_NODEINFO ) . 945If it does not exist, 946.Er ENOENT 947will be returned. 948.Pp 949All data messages are mbuf chains with the M_PKTHDR flag set. 950.Pp 951Nodes are responsible for freeing what they allocate. 952There are three exceptions: 953.Bl -tag -width xxxx 954.It 1 955Mbufs sent across a data link are never to be freed by the sender. 956.It 2 957Any meta-data information traveling with the data has the same restriction. 958It might be freed by any node the data passes through, and a 959.Dv NULL 960passed onwards, but the caller will never free it. 961Two macros 962.Fn NG_FREE_META "meta" 963and 964.Fn NG_FREE_DATA "m" "meta" 965should be used if possible to free data and meta data (see 966.Pa netgraph.h ) . 967.It 3 968Messages sent using 969.Fn ng_send_message 970are freed by the callee. As in the case above, the addresses 971associated with the message are freed by whatever allocated them so the 972recipient should copy them if it wants to keep that information. 973.El 974.Sh FILES 975.Bl -tag -width xxxxx -compact 976.It Pa /sys/netgraph/netgraph.h 977Definitions for use solely within the kernel by 978.Nm 979nodes. 980.It Pa /sys/netgraph/ng_message.h 981Definitions needed by any file that needs to deal with 982.Nm 983messages. 984.It Pa /sys/netgraph/ng_socket.h 985Definitions needed to use 986.Nm 987socket type nodes. 988.It Pa /sys/netgraph/ng_{type}.h 989Definitions needed to use 990.Nm 991{type} 992nodes, including the type cookie definition. 993.It Pa /modules/netgraph.ko 994Netgraph subsystem loadable KLD module. 995.It Pa /modules/ng_{type}.ko 996Loadable KLD module for node type {type}. 997.El 998.Sh USER MODE SUPPORT 999There is a library for supporting user-mode programs that wish 1000to interact with the netgraph system. See 1001.Xr netgraph 3 1002for details. 1003.Pp 1004Two user-mode support programs, 1005.Xr ngctl 8 1006and 1007.Xr nghook 8 , 1008are available to assist manual configuration and debugging. 1009.Pp 1010There are a few useful techniques for debugging new node types. 1011First, implementing new node types in user-mode first 1012makes debugging easier. 1013The 1014.Em tee 1015node type is also useful for debugging, especially in conjunction with 1016.Xr ngctl 8 1017and 1018.Xr nghook 8 . 1019.Sh SEE ALSO 1020.Xr socket 2 , 1021.Xr netgraph 3 , 1022.Xr ng_async 4 , 1023.Xr ng_bpf 4 , 1024.Xr ng_cisco 4 , 1025.Xr ng_echo 4 , 1026.Xr ng_ether 4 , 1027.Xr ng_frame_relay 4 , 1028.Xr ng_hole 4 , 1029.Xr ng_iface 4 , 1030.Xr ng_ksocket 4 , 1031.Xr ng_lmi 4 , 1032.Xr ng_mppc 4 , 1033.Xr ng_ppp 4 , 1034.Xr ng_pppoe 4 , 1035.Xr ng_rfc1490 4 , 1036.Xr ng_socket 4 , 1037.Xr ng_tee 4 , 1038.Xr ng_tty 4 , 1039.Xr ng_UI 4 , 1040.Xr ng_vjc 4 , 1041.Xr ngctl 8 , 1042.Xr nghook 8 1043.Sh HISTORY 1044The 1045.Nm 1046system was designed and first implemented at Whistle Communications, Inc.\& 1047in a version of 1048.Fx 2.2 1049customized for the Whistle InterJet. 1050It first made its debut in the main tree in 1051.Fx 3.4 . 1052.Sh AUTHORS 1053.An -nosplit 1054.An Julian Elischer Aq julian@FreeBSD.org , 1055with contributions by 1056.An Archie Cobbs Aq archie@FreeBSD.org . 1057