[dpdk-dev] [PATCH 1/5] net/mlx5: fix protocol size for raw encap judgement

[Ori Kam]

> -----Original Message-----
> From: Matan Azrad <matan at nvidia.com>
> Sent: Sunday, November 22, 2020 6:05 PM
> Subject: RE: [dpdk-dev] [PATCH 1/5] net/mlx5: fix protocol size for raw encap
> judgement
> 
> 
> 
> From: Thomas Monjalon
> 
> > 16/11/2020 08:55, Xiaoyu Min:
> > > From: Xiaoyu Min <jackmin at nvidia.com>
> > >
> > > The rte_flow_item_eth and rte_flow_item_vlan items are refined.
> > > The structs do not exactly represent the packet bits captured on the
> > > wire anymore.
> > > Should use real header instead of the whole struct.
> > >
> > > Replace the rte_flow_item_* with the existing corresponding rte_*_hdr.
> > >
> > > Fixes: 09315fc83861 ("ethdev: add VLAN attributes to ethernet and VLAN
> > > items")
> > >
> > > Signed-off-by: Xiaoyu Min <jackmin at nvidia.com>
> > > ---
> > >  drivers/net/mlx5/mlx5_flow.c | 12 ++++++------
> > > drivers/net/mlx5/mlx5_flow.h |  4 ++--
> > >  2 files changed, 8 insertions(+), 8 deletions(-)
> > >
> > > diff --git a/drivers/net/mlx5/mlx5_flow.c
> > > b/drivers/net/mlx5/mlx5_flow.c index 324349ed19..34b7a2f137 100644
> > > --- a/drivers/net/mlx5/mlx5_flow.c
> > > +++ b/drivers/net/mlx5/mlx5_flow.c
> > > @@ -3655,8 +3655,8 @@ flow_check_hairpin_split(struct rte_eth_dev
> *dev,
> > >               case RTE_FLOW_ACTION_TYPE_RAW_ENCAP:
> > >                       raw_encap = actions->conf;
> > >                       if (raw_encap->size >
> > > -                         (sizeof(struct rte_flow_item_eth) +
> > > -                          sizeof(struct rte_flow_item_ipv4)))
> > > +                         (sizeof(struct rte_ether_hdr) +
> > > +                          sizeof(struct rte_ipv4_hdr)))
> > >                               split++;
> > >                       action_n++;
> > >                       break;
> > > @@ -4092,8 +4092,8 @@ flow_hairpin_split(struct rte_eth_dev *dev,
> > >               case RTE_FLOW_ACTION_TYPE_RAW_ENCAP:
> > >                       raw_encap = actions->conf;
> > >                       if (raw_encap->size >
> > > -                         (sizeof(struct rte_flow_item_eth) +
> > > -                          sizeof(struct rte_flow_item_ipv4))) {
> > > +                         (sizeof(struct rte_ether_hdr) +
> > > +                          sizeof(struct rte_ipv4_hdr))) {
> > >                               memcpy(actions_tx, actions,
> > >                                      sizeof(struct rte_flow_action));
> > >                               actions_tx++; @@ -4107,8 +4107,8 @@
> > > flow_hairpin_split(struct rte_eth_dev *dev,
> > >               case RTE_FLOW_ACTION_TYPE_RAW_DECAP:
> > >                       raw_decap = actions->conf;
> > >                       if (raw_decap->size <
> > > -                         (sizeof(struct rte_flow_item_eth) +
> > > -                          sizeof(struct rte_flow_item_ipv4))) {
> > > +                         (sizeof(struct rte_ether_hdr) +
> > > +                          sizeof(struct rte_ipv4_hdr))) {
> > >                               memcpy(actions_tx, actions,
> > >                                      sizeof(struct rte_flow_action));
> > >                               actions_tx++; diff --git
> > > a/drivers/net/mlx5/mlx5_flow.h b/drivers/net/mlx5/mlx5_flow.h index
> > > 5fac8672fc..3b28c48a1d 100644
> > > --- a/drivers/net/mlx5/mlx5_flow.h
> > > +++ b/drivers/net/mlx5/mlx5_flow.h
> > > @@ -345,8 +345,8 @@ enum mlx5_feature_name {  #define
> > > MLX5_GENEVE_OPT_LEN_0 14  #define MLX5_GENEVE_OPT_LEN_1 63
> > >
> > > -#define MLX5_ENCAPSULATION_DECISION_SIZE (sizeof(struct
> > rte_flow_item_eth) + \
> > > -                                       sizeof(struct rte_flow_item_ipv4))
> > > +#define MLX5_ENCAPSULATION_DECISION_SIZE (sizeof(struct
> > rte_ether_hdr) + \
> > > +                                       sizeof(struct rte_ipv4_hdr))
> >
> > This constant should be used above in hairpin functions.
> > Will change while merging.
> 
> +1
> 

Why above the hairpin functions? This is correct for all encap decisions.

Best,
Ori