Hi Cristian,
Currently, the DPDK framework does not make any assumption on the actual
layout of a mbuf.
More precisely, the DPDK does not impose any constraint on the actual
location of additional metadata, if any, or on the actual location and
size of its associated payload data buffer.
This is coherent with the fact that mbuf pools are not created by the
DPDK itself, but by network applications that are free to choose
whatever packet mbuf layout that fits their particular needs.
There is one exception to this basic DPDK rule: the mbuf cloning feature
available through the RTE_MBUF_SCATTER_GATHER configuration option
assumes that the payload data buffer of the mbuf immediately follows the
rte_mbuf data structure (see the macros RTE_MBUF_FROM_BADDR,
RTE_MBUF_TO_BADDR, RTE_MBUF_INDIRECT, and RTE_MBUF_DIRECT in the file
lib/librte_mbuf/rte_mbuf.h).
The cloning feature prevents to build packet mbufs with their metadata
located immediately after the rte_mbuf data structure, which is exactly
what your patch introduces.
At least, a comment that clearly warns the user of this incompatibility
might be worth adding into both the code and your patch log.
Regards,
Ivan
On 05/27/2014 07:09 PM, Cristian Dumitrescu wrote:
> Added zero-size field (offset in data structure) to specify the beginning of
> packet meta-data in the packet buffer just after the mbuf.
>
> The size of the packet meta-data is application specific and the packet
> meta-data is managed by the application.
>
> The packet meta-data should always be accessed through the provided macros.
>
> This is used by the Packet Framework libraries (port, table, pipeline).
>
> There is absolutely no performance impact due to this mbuf field, as it does
> not take any space in the mbuf structure (zero-size field).
>
> Signed-off-by: Cristian Dumitrescu <cristian.dumitrescu at intel.com>
> ---
> lib/librte_mbuf/rte_mbuf.h | 17 +++++++++++++++++
> 1 files changed, 17 insertions(+), 0 deletions(-)
>
> diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> index 4a9ab41..bf09618 100644
> --- a/lib/librte_mbuf/rte_mbuf.h
> +++ b/lib/librte_mbuf/rte_mbuf.h
> @@ -201,8 +201,25 @@ struct rte_mbuf {
> struct rte_ctrlmbuf ctrl;
> struct rte_pktmbuf pkt;
> };
> +
> + union {
> + uint8_t metadata[0];
> + uint16_t metadata16[0];
> + uint32_t metadata32[0];
> + uint64_t metadata64[0];
> + };
> } __rte_cache_aligned;
>
> +#define RTE_MBUF_METADATA_UINT8(mbuf, offset) (mbuf->metadata[offset])
> +#define RTE_MBUF_METADATA_UINT16(mbuf, offset)
> (mbuf->metadata16[offset/sizeof(uint16_t)])
> +#define RTE_MBUF_METADATA_UINT32(mbuf, offset)
> (mbuf->metadata32[offset/sizeof(uint32_t)])
> +#define RTE_MBUF_METADATA_UINT64(mbuf, offset)
> (mbuf->metadata64[offset/sizeof(uint64_t)])
> +
> +#define RTE_MBUF_METADATA_UINT8_PTR(mbuf, offset) (&mbuf->metadata[offset])
> +#define RTE_MBUF_METADATA_UINT16_PTR(mbuf, offset)
> (&mbuf->metadata16[offset/sizeof(uint16_t)])
> +#define RTE_MBUF_METADATA_UINT32_PTR(mbuf, offset)
> (&mbuf->metadata32[offset/sizeof(uint32_t)])
> +#define RTE_MBUF_METADATA_UINT64_PTR(mbuf, offset)
> (&mbuf->metadata64[offset/sizeof(uint64_t)])
> +
> /**
> * Given the buf_addr returns the pointer to corresponding mbuf.
> */
>
--
Ivan Boule
6WIND Development Engineer
