libtlv

Simple tlv library supporting variable length encoding, padding, alignment and etc.

Read (get) and write (put) functions operate on tlv buffer (empty or not) of a given size with 0 as end or empty, and 0xff as padding.

Reading or get

int libtlv_get(unsigned int opt, void *buf, size_t size, unsigned int *t, unsigned int *l, void *v);

caller needs to check *t to see if it matches tvl ends with single 0 (no length or value)

• opt: options
• buf: tlv buffer
• size: buffer size
• t: [INOUT] null to find end, -> 0 to find next, or -> type to match
• l: [INOUT] null to ignore length, -> 0 for get ptr in v, otherwise -> max of v, return actual length
• v: [OUT] null to ignore value, ptr to value if l -> 0, otherwise copy value (max *l)
• Return: negative error or offset at end or after found tlv

Note this function can be used to

• find particular type t, copy value into v (no more than *l) and return actual length into l
• find last tlv entry matching t
• find end of tlv (zero or end of buffer)
• return pointer to value into v
• get next tlv entry for looping

Writing or put

int libtlv_put(unsigned int opt, void *buf, size_t size, unsigned int  t, unsigned int  l, void *v);

• opt: options
• buf: tlv buffer
• size: buffer size
• t: type
• l: length
• v: value
• Return: negative error or offset at end or after put tlv

Note this function can be used to

• skip existing tlv and put at the end
• optionally return failure if type t already in buffer

Encoding

Both t and l are encoded in network order (big-endian) of 1 or 2 bytes in fixed encoding:

• 1 byte: max 0xFF
• 2 bytes: max 0xFFFF

For variable sized t and l, maximum 4-bytes of 7-bit encoding is used as as in VLQ.

• 1 byte: max 0x7F
• 2 bytes: max 0x3FFF
• 3 bytes: max 0x1FFFFF
• 4 bytes: max 0xFFFFFFF

Padding and alignment

Choosing 0xFF as padding makes alignment (of t, l or v) possible, but limit range of t:

• 1 byte: max 0xFE
• 2 bytes: max 0xFEFF
• var 4 bytes: max 0xFDFFFFF

Note for variable encoding, to avoid 0xFF, some values start with 0x80.

Recursive tlv

When outputing recursive tlv, space of t and l can be reserved (initialized to zero) and fill up later:

ret = libtlv_put(opt | LIBTLV_OPT_PUT_INIT, buf, size, t, 0, NULL); /* space of t,l set to 0 */
l = fill_sub_tlv(buf + ret, size-ret, ...);
ret = libtlv_put(opt | LIBTLV_OPT_PUT_FINI, buf, size, t, l, NULL); /* space of t,l filled up */