mdchain, md_initm, md_done, md_append_record, md_next_record,
md_get_uint8, md_get_uint16, md_get_uint16be, md_get_uint16le,
md_get_uint32, md_get_uint32be, md_get_uint32le, md_get_int64,
md_get_int64be, md_get_int64le, md_get_mem, md_get_mbuf, md_get_uio --
set of functions to dissect an mbuf chain to various data types
options LIBMCHAIN
kldload libmchain
#include <sys/param.h>
#include <sys/uio.h>
#include <sys/mchain.h>
void
md_initm(struct mdchain *mdp, struct mbuf *m);
void
md_done(struct mdchain *mdp);
void
md_append_record(struct mdchain *mdp, struct mbuf *top);
int
md_next_record(struct mdchain *mdp);
int
md_get_uint8(struct mdchain *mdp, u_int8_t *x);
int
md_get_uint16(struct mdchain *mdp, u_int16_t *x);
int
md_get_uint16be(struct mdchain *mdp, u_int16_t *x);
int
md_get_uint16le(struct mdchain *mdp, u_int16_t *x);
int
md_get_uint32(struct mdchain *mdp, u_int32_t *x);
int
md_get_uint32be(struct mdchain *mdp, u_int32_t *x);
int
md_get_uint32le(struct mdchain *mdp, u_int32_t *x);
int
md_get_int64(struct mdchain *mdp, int64_t *x);
int
md_get_int64be(struct mdchain *mdp, int64_t *x);
int
md_get_int64le(struct mdchain *mdp, int64_t *x);
int
md_get_mem(struct mdchain *mdp, caddr_t target, int size, int type);
int
md_get_mbuf(struct mdchain *mdp, int size, struct mbuf **m);
int
md_get_uio(struct mdchain *mdp, struct uio *uiop, int size);
These functions are used to decompose mbuf chains to various data types.
The mdchain structure is used as a working context and should be initialized
through a call of the mb_initm() function. It has the following
fields:
md_top (struct mbuf *) a pointer to the top of the parsed mbuf chain
md_cur (struct mbuf *) a pointer to the currently parsed mbuf
md_pas (int) offset in the current mbuf
The md_done() function disposes of an mbuf chain pointed to by the
mdp->md_top field and sets the field to NULL.
The md_append_record() appends a new mbuf chain using m_nextpkt field to
form a single linked list of mbuf chains. If the mdp->md_top field is
NULL, then this function behaves exactly as the md_initm() function.
The md_next_record() function extracts the next mbuf chain and disposes
the current one, if any. For a new mbuf chain it calls the md_initm()
function. If there is no data left the function returns ENOENT.
All md_get_*() functions perform an actual copy of the data from an mbuf
chain. Functions which have le or be suffixes will perform conversion to
the little- or big-endian data formats.
md_get_mem() function copies size bytes of data specified by the source
argument from an mbuf chain. The type argument specifies the method used
to perform a copy, and can be one of the following:
MB_MSYSTEM Use the bcopy() function.
MB_MUSER Use the copyin(9) function.
MB_MINLINE Use an ``inline'' loop which does not call any function.
If target is NULL, an actual copy is not performed and the function just
skips the given number of bytes.
All int functions return zero if successful, otherwise an error code is
returned.
Note: after failure of any function, an mbuf chain is left in the broken
state and only the md_done() function can safely be called to destroy it.
struct mdchain *mdp;
struct mbuf *m;
u_int16_t length;
u_int8_t byte;
receive(so, &m);
md_initm(mdp, m);
if (md_get_uint8(mdp, &byte) != 0 ||
md_get_uint16le(mdp, &length) != 0)
error = EBADRPC;
mb_done(mdp);
mbchain(9), mbuf(9)
This manual page was written by Boris Popov <[email protected]>.
FreeBSD 5.2.1 February 28, 2001 FreeBSD 5.2.1 [ Back ] |