[ovs-dev] [PATCH] dpcls: Avoid one 8-byte chunk in subtable mask.

[Jarno Rajahalme]

> On Jan 4, 2017, at 6:41 AM, antonio.fischetti at intel.com wrote:
> 
> This patch allows to skip the chunk comprising of dp_hash and in_port
> in the subtable mask when the packet is not recirculated.  This will
> slightly speed up the hash computation as one expensive function call
> to hash_add64() can be skipped.
> For each new netdev flow we wildcard in_port in the mask, so in the
> physical case where dp_hash is also wildcarded, the resulting 8-byte
> chunk will not be part of the subtable mask.
> 
> The idea behind is that all the packets within a given dpcls will
> have the same in_port value and typically dp_hash == 0.  So they will
> have the same 8-bytes chunk in their {dp_hash, in_port} portion of the
> miniflow. This doesn't contribute effectively in spreading hash values
> and avoiding collisions.
> 
> Signed-off-by: Antonio Fischetti <antonio.fischetti at intel.com>
> Signed-off-by: Bhanuprakash Bodireddy <bhanuprakash.bodireddy at intel.com>
> Co-authored-by: Bhanuprakash Bodireddy <bhanuprakash.bodireddy at intel.com>
> Signed-off-by: Jarno Rajahalme <jarno at ovn.org>
> Co-authored-by: Jarno Rajahalme <jarno at ovn.org>
> ---
> lib/dpif-netdev.c | 14 +++++++++++++-
> 1 file changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
> index 1a47a45..a808f4b 100644
> --- a/lib/dpif-netdev.c
> +++ b/lib/dpif-netdev.c
> @@ -2277,8 +2277,20 @@ dp_netdev_flow_add(struct dp_netdev_pmd_thread *pmd,
>     struct netdev_flow_key mask;
>     struct dpcls *cls;
>     odp_port_t in_port = match->flow.in_port.odp_port;
> -
> +    uint32_t old_odp_port;
> +
> +    /* As we select the dpcls based on the port number, each netdev flow
> +     * belonging to the same dpcls will have the same odp_port value.
> +     * For performance reasons here we wildcard odp_port in the mask.  In the
> +     * physical case where dp_hash is also wildcarded, the resulting 8-byte
> +     * chunk {dp_hash, in_port} will be ignored by netdev_flow_mask_init() and
> +     * will not be part of the subtable mask.
> +     * This will speed up the hash computation during dpcls_lookup() because
> +     * one call to hash_add64() will be skipped. */
> +    old_odp_port = match->wc.masks.in_port.odp_port;


Please move this line further down in the function here and do not save the old value:

    ovs_assert(match->wc.masks.in_port.odp_port == ODPP_NONE);

> +    match->wc.masks.in_port.odp_port = 0;
>     netdev_flow_mask_init(&mask, match);
> +    match->wc.masks.in_port.odp_port = old_odp_port;

And do this instead:

match->wc.masks.in_port.odp_port = ODPP_NONE;

>     /* Make sure wc does not have metadata. */
>     ovs_assert(!FLOWMAP_HAS_FIELD(&mask.mf.map, metadata)
>                && !FLOWMAP_HAS_FIELD(&mask.mf.map, regs));
> -- 
> 2.4.11
>