On Sat, Mar 24, 2018 at 12:47 AM, Duncan Roe <duncan_roe@xxxxxxxxxxxxxxx> wrote:
> On Wed, Mar 14, 2018 at 10:00:35PM +0100, Laura Garcia Liebana wrote:
>> Create the new type u32_integer with a fixed size in order to
>> be used as a key in maps and sets. The type integer cannot be
>> used as a key cause is a dynamic size type and is used as a
>> base type of some subtypes.
>>
>> Without this patch we obtain the following error:
>>
>> Error: unqualified key type integer specified in map definition
>> add map nftlb mapa { type integer : ipv4_addr; }
>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>
>> After this patch, we can use an u32 integer as a key for sets
>> and maps:
>>
>> table ip nftlb {
>> map mapa {
>> type u32_integer : ipv4_addr
>> }
>>
>> set conjunto {
>> type u32_integer
>> }
>> }
>>
>> Signed-off-by: Laura Garcia Liebana <nevola@xxxxxxxxx>
>> ---
>> This is the v2 of ("fix integer type size to be used as a key
>> for sets and maps"), due to this approach fits better with the
>> current design of nft types than the previous one and avoids
>> possible side effects.
>>
>> include/datatype.h | 3 +++
>> src/datatype.c | 10 ++++++++++
>> 2 files changed, 13 insertions(+)
>>
>> diff --git a/include/datatype.h b/include/datatype.h
>> index 3f612e5..7f106cd 100644
>> --- a/include/datatype.h
>> +++ b/include/datatype.h
>> @@ -42,6 +42,7 @@
>> * @TYPE_DEVGROUP: devgroup code (integer subtype)
>> * @TYPE_DSCP: Differentiated Services Code Point (integer subtype)
>> * @TYPE_IFNAME: interface name (string subtype)
>> + * @TYPE_U32_INTEGER: unsigned 32 bits integer (integer subtype)
>> */
>> enum datatypes {
>> TYPE_INVALID,
>> @@ -86,6 +87,7 @@ enum datatypes {
>> TYPE_BOOLEAN,
>> TYPE_CT_EVENTBIT,
>> TYPE_IFNAME,
>> + TYPE_U32_INTEGER,
>> __TYPE_MAX
>> };
>> #define TYPE_MAX (__TYPE_MAX - 1)
>> @@ -240,6 +242,7 @@ extern const struct datatype icmpv6_code_type;
>> extern const struct datatype icmpx_code_type;
>> extern const struct datatype time_type;
>> extern const struct datatype boolean_type;
>> +extern const struct datatype u32_integer_type;
>>
>> extern const struct datatype *concat_type_alloc(uint32_t type);
>> extern void concat_type_destroy(const struct datatype *dtype);
>> diff --git a/src/datatype.c b/src/datatype.c
>> index 324ac80..f2d1d2b 100644
>> --- a/src/datatype.c
>> +++ b/src/datatype.c
>> @@ -69,6 +69,7 @@ static const struct datatype *datatypes[TYPE_MAX + 1] = {
>> [TYPE_FIB_ADDR] = &fib_addr_type,
>> [TYPE_BOOLEAN] = &boolean_type,
>> [TYPE_IFNAME] = &ifname_type,
>> + [TYPE_U32_INTEGER] = &u32_integer_type,
>> };
>>
>> const struct datatype *datatype_lookup(enum datatypes type)
>> @@ -1144,3 +1145,12 @@ const struct datatype boolean_type = {
>> .basetype = &integer_type,
>> .sym_tbl = &boolean_tbl,
>> };
>> +
>> +const struct datatype u32_integer_type = {
>> + .type = TYPE_U32_INTEGER,
>> + .name = "u32_integer",
>> + .desc = "32 bits integer",
>> + .size = 4 * BITS_PER_BYTE,
>> + .byteorder = BYTEORDER_HOST_ENDIAN,
>> + .basetype = &integer_type,
>> +};
>> --
>> 2.11.0
>>
> Why do we need an invented type when there is already uint32_t in
> /usr/include/stdint.h?
Hi Duncan,
Cause the nft parser doesn't understand uint32_t and the way
that nft sends the types to the kernel requires to specify
the base type, size, byteorder, etc.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
