[dpdk-dev] rte_flow: Different devices have different field indianess?

[dpdk-dev] rte_flow: Different devices have different field indianess?

[Tom Barbette]

Hi all,

It seems rte_flow_item_eth takes its ethertype in host byte order with 
i40e, but in network byte order with mlx5.

Wouldn't it be nice to unify that? Else is there a way to know in which 
byte order the spec should be given? I guess that expands to all fields, 
but I only compared the ethertype field.

Thanks,

Tom

Re: [dpdk-dev] rte_flow: Different devices have different field indianess?

[Andrew Rybchenko]

On 3/27/20 5:29 PM, Tom Barbette wrote:
> Hi all,
> 
> It seems rte_flow_item_eth takes its ethertype in host byte order with
> i40e, but in network byte order with mlx5.

If so, it is definitely bug in i40e, since struct rte_flow_item_eth
defines type as rte_be16_t type.

> Wouldn't it be nice to unify that? Else is there a way to know in which
> byte order the spec should be given? I guess that expands to all fields,
> but I only compared the ethertype field.
> 
> Thanks,
> 
> Tom

Re: [dpdk-dev] rte_flow: Different devices have different field indianess?

[Tom Barbette]

Le 27/03/2020 à 15:44, Andrew Rybchenko a écrit :
> On 3/27/20 5:29 PM, Tom Barbette wrote:
>> Hi all,
>>
>> It seems rte_flow_item_eth takes its ethertype in host byte order with
>> i40e, but in network byte order with mlx5.
> 
> If so, it is definitely bug in i40e, since struct rte_flow_item_eth
> defines type as rte_be16_t type.

Indeed, with testpmd I would expect the first one to be the one that works:

testpmd> flow create 0 ingress pattern eth type is 0x0800 / end actions 
mark id 5  / end
port_flow_complain(): Caught PMD error type 13 (specific pattern item): 
cause: 0x227fd75540, Unsupported ether_type.: Invalid argument
testpmd> flow create 0 ingress pattern eth type is 0x0008 / end actions 
mark id 5  / end
Flow rule #0 created


> 
>> Wouldn't it be nice to unify that? Else is there a way to know in which
>> byte order the spec should be given? I guess that expands to all fields,
>> but I only compared the ethertype field.
>>
>> Thanks,
>>
>> Tom
> 

Re: [dpdk-dev] rte_flow: Different devices have different field indianess?

[Thomas Monjalon]

27/03/2020 15:44, Andrew Rybchenko:
> On 3/27/20 5:29 PM, Tom Barbette wrote:
> > Hi all,
> > 
> > It seems rte_flow_item_eth takes its ethertype in host byte order with
> > i40e, but in network byte order with mlx5.
> 
> If so, it is definitely bug in i40e, since struct rte_flow_item_eth
> defines type as rte_be16_t type.

Someone to open a bugzilla for this i40e bug please?

Re: [dpdk-dev] rte_flow: Different devices have different field indianess?

[Xing, Beilei]

> -----Original Message-----
> From: Tom Barbette <barbette@kth.se>
> Sent: Friday, March 27, 2020 10:51 PM
> To: Andrew Rybchenko <arybchenko@solarflare.com>; dev@dpdk.org
> Cc: orika@mellanox.com; Xing, Beilei <beilei.xing@intel.com>; Zhang, Qi Z
> <qi.z.zhang@intel.com>
> Subject: Re: [dpdk-dev] rte_flow: Different devices have different field
> indianess?
> 
> Le 27/03/2020 à 15:44, Andrew Rybchenko a écrit :
> > On 3/27/20 5:29 PM, Tom Barbette wrote:
> >> Hi all,
> >>
> >> It seems rte_flow_item_eth takes its ethertype in host byte order
> >> with i40e, but in network byte order with mlx5.
> >
> > If so, it is definitely bug in i40e, since struct rte_flow_item_eth
> > defines type as rte_be16_t type.
> 
> Indeed, with testpmd I would expect the first one to be the one that works:
> 
> testpmd> flow create 0 ingress pattern eth type is 0x0800 / end actions
> mark id 5  / end
> port_flow_complain(): Caught PMD error type 13 (specific pattern item):
> cause: 0x227fd75540, Unsupported ether_type.: Invalid argument
> testpmd> flow create 0 ingress pattern eth type is 0x0008 / end actions
> mark id 5  / end
> Flow rule #0 created
> 

It's not a bug, but HW limitation.
Only FDIR supports mark action, but i40e FDIR doesn't support filter with eth type 0x0800.

Please refer to the following 
ether_type = rte_be_to_cpu_16(eth_spec->type);

                                if (next_type == RTE_FLOW_ITEM_TYPE_VLAN ||
                                    ether_type == RTE_ETHER_TYPE_IPV4 ||
                                    ether_type == RTE_ETHER_TYPE_IPV6 ||
                                    ether_type == RTE_ETHER_TYPE_ARP ||
                                    ether_type == outer_tpid) {
                                        rte_flow_error_set(error, EINVAL,
                                                     RTE_FLOW_ERROR_TYPE_ITEM,
                                                     item,
                                                     "Unsupported ether_type.");
                                        return -rte_errno;
                                }

Beilei

> 
> >
> >> Wouldn't it be nice to unify that? Else is there a way to know in
> >> which byte order the spec should be given? I guess that expands to
> >> all fields, but I only compared the ethertype field.
> >>
> >> Thanks,
> >>
> >> Tom
> >