in_cksum(9) [netbsd man page]

IN_CKSUM(9)						   BSD Kernel Developer's Manual					       IN_CKSUM(9)

NAME

     in_cksum, in4_cksum, in6_cksum -- compute Internet checksum

SYNOPSIS

     uint16_t
     in_cksum(struct mbuf *m, int len);

     uint16_t
     in4_cksum(struct mbuf *m, uint8_t nxt, int off, int len);

     uint16_t
     in6_cksum(struct mbuf *m, uint8_t nxt, int off, int len);

DESCRIPTION

     These functions are used to compute the ones-complement checksum required by IP and IPv6.	The in4_cksum() function is used to compute the
     transport-layer checksum required by tcp(4) and udp(4) over a range of bytes starting at off and continuing on for len bytes within the mbuf
     m.

     If the nxt parameter is non-zero, it is assumed to be an IP protocol number.  It is also assumed that the data within m starts with an IP
     header, and the transport-layer header starts at off; a pseudo-header is constructed as specified in RFC768 and RFC793, and the pseudo-header
     is prepended to the data covered by the checksum.

     The in6_cksum() function is similar; if nxt is non-zero, it is assumed that m starts with an IPv6 header, and that the transport-layer header
     starts after off bytes.

     The in_cksum() function is equivalent to in4_cksum(m, 0, 0, len).

     These functions are always performance critical and should be reimplemented in assembler or optimized C for each platform; when available,
     use of repeated full-width add-with-carry followed by reduction of the sum to a 16 bit width usually leads to best results.  See RFC's 1071,
     1141, 1624, and 1936 for more information about efficient computation of the internet checksum.

RETURN VALUES

     All three functions return the computed checksum value.

SEE ALSO

     inet(4), inet6(4), tcp(4), udp(4), protocols(5), mbuf(9)

STANDARDS

     These functions implement the Internet transport-layer checksum as specified in RFC768, RFC793, and RFC2460.

BUGS

     The in6_cksum() function currently requires special handling of link-local addresses in the pseudo-header due to the use of embedded scope-
     id's within link-local addresses.

BSD
								   May 22, 2001 							       BSD

M_TAG(9)						   BSD Kernel Developer's Manual						  M_TAG(9)

NAME

     m_tag, m_tag_get, m_tag_free, m_tag_prepend, m_tag_unlink, m_tag_delete, m_tag_delete_chain, m_tag_delete_nonpersistent, m_tag_find,
     m_tag_copy, m_tag_copy_chain, m_tag_init, m_tag_first, m_tag_next -- mbuf tagging interfaces

SYNOPSIS

     #include <sys/mbuf.h>

     struct m_tag *
     m_tag_get(int type, int len, int wait);

     void
     m_tag_free(struct m_tag *t);

     void
     m_tag_prepend(struct mbuf *m, struct m_tag *t);

     void
     m_tag_unlink(struct mbuf *m, struct m_tag *t);

     void
     m_tag_delete(struct mbuf *m, struct m_tag *t);

     void
     m_tag_delete_chain(struct mbuf *m, struct m_tag *t);

     void
     m_tag_delete_nonpersistent(struct mbuf *);

     struct m_tag *
     m_tag_find(struct mbuf *m, int type, struct m_tag *t);

     struct m_tag *
     m_tag_copy(struct m_tag *m);

     int
     m_tag_copy_chain(struct mbuf *to, struct mbuf *from);

     void
     m_tag_init(struct mbuf *m);

     struct m_tag *
     m_tag_first(struct mbuf *m);

     struct m_tag *
     m_tag_next(struct mbuf *m, struct m_tag *t);

DESCRIPTION

     The m_tag interface is used to ``tag'' mbufs.

FUNCTIONS

     m_tag_get(type, len, wait)
	      Allocate an mbuf tag.  type is one of the PACKET_TAG_ macros.  len is the size of the data associated with the tag, in bytes.  wait
	      is either M_WAITOK or M_NOWAIT.

     m_tag_free(t)
	      Free the mbuf tag t.

     m_tag_prepend(m, t)
	      Prepend the mbuf tag t to the mbuf m.  t will become the first tag of the mbuf m.  When m is freed, t will also be freed.

     m_tag_unlink(m, t)
	      Unlink the mbuf tag t from the mbuf m.

     m_tag_delete(m, t)
	      The same as m_tag_unlink() followed by m_tag_free().

     m_tag_delete_chain(m, t)
	      Unlink and free mbuf tags beginning with the mbuf tag t from the mbuf m.	If t is NULL, m_tag_delete_chain() unlinks and frees all
	      mbuf tags associated with the mbuf m.

     m_tag_delete_nonpersistent(m)
	      Unlink and free all non persistent tags associated with the mbuf m.

     m_tag_find(m, type, t)
	      Find an mbuf tag with type type after the mbuf tag t in the tag chain associated with the mbuf m.  If t is NULL, search from the
	      first mbuf tag.  If an mbuf tag is found, return a pointer to it.  Otherwise return NULL.

     m_tag_copy(t)
	      Copy an mbuf tag t.  Return a new mbuf tag on success.  Otherwise return NULL.

     m_tag_copy_chain(to, from)
	      Copy all mbuf tags associated with the mbuf from to the mbuf to.	If to already has any mbuf tags, they will be unlinked and freed
	      beforehand.  Return 1 on success.  Otherwise return 0.

     m_tag_init(m)
	      Initialize mbuf tag chain of the mbuf m.

     m_tag_first(m)
	      Return the first mbuf tag associated with the mbuf m.  Return NULL if no mbuf tags are found.

     m_tag_next(m, t)
	      Return the next mbuf tag after t associated with the mbuf m.  Return NULL if t is the last tag in the chain.

CODE REFERENCES

     The mbuf tagging interfaces are implemented within the file sys/kern/uipc_mbuf2.c.

     The PACKET_TAG_ macros are defined in the file sys/sys/mbuf.h.

SEE ALSO

     intro(9), malloc(9), mbuf(9)

BUGS

     The semantics of the term "persistent tag" are vague.

BSD
								 September 7, 2004							       BSD