Re: [dpdk-dev] [PATCH v12 2/3] ethdev: extend RSS offload types

[Andrew Rybchenko <arybchenko@solarflare.com>]

On 10/15/19 1:55 PM, Su, Simei wrote:
> Hi, Andrew
>
>> -----Original Message-----
>> From: Andrew Rybchenko [mailto:arybchenko@solarflare.com]
>> Sent: Tuesday, October 15, 2019 5:08 PM
>> To: Su, Simei <simei.su@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>; Ye,
>> Xiaolong <xiaolong.ye@intel.com>; Yigit, Ferruh <ferruh.yigit@intel.com>
>> Cc: dev@dpdk.org
>> Subject: Re: [dpdk-dev] [PATCH v12 2/3] ethdev: extend RSS offload types
>>
>> On 10/14/19 4:36 PM, Simei Su wrote:
>>> This patch reserves several bits as input set selection from the high
>>> end of the 64 bits. It is combined with exisiting ETH_RSS_* to
>>> represent RSS types. This patch also checks the simultaneous use of
>>> SRC_ONLY and DST_ONLY of the same level.
>>>
>>> Signed-off-by: Simei Su <simei.su@intel.com>
>>> Reviewed-by: Qi Zhang <qi.z.zhang@intel.com>
>>> Acked-by: Ori Kam <orika@mellanox.com>
>>> ---
>>>    lib/librte_ethdev/rte_ethdev.c |  5 +++++
>>>    lib/librte_ethdev/rte_ethdev.h | 25 +++++++++++++++++++++++++
>>>    2 files changed, 30 insertions(+)
>>>
>>> diff --git a/lib/librte_ethdev/rte_ethdev.c
>>> b/lib/librte_ethdev/rte_ethdev.c index 7caaa0b..69a3d71 100644
>>> --- a/lib/librte_ethdev/rte_ethdev.c
>>> +++ b/lib/librte_ethdev/rte_ethdev.c
>>> @@ -1269,6 +1269,9 @@ struct rte_eth_dev *
>>>    		goto rollback;
>>>    	}
>>>
>>> +
>> 	RTE_ETH_RSS_HF_REFINE(dev->data->dev_conf.rx_adv_conf.rss_conf.rss_
>> hf,
>>> +				dev_conf->rx_adv_conf.rss_conf.rss_hf);
>>> +
>>>    	/* Check that device supports requested rss hash functions. */
>>>    	if ((dev_info.flow_type_rss_offloads |
>>>    	     dev_conf->rx_adv_conf.rss_conf.rss_hf) != @@ -3112,6 +3115,8
>>> @@ struct rte_eth_dev *
>>>    	if (ret != 0)
>>>    		return ret;
>>>
>>> +	RTE_ETH_RSS_HF_REFINE(rss_conf->rss_hf, rss_conf->rss_hf);
>>> +
>>>    	dev = &rte_eth_devices[port_id];
>>>    	if ((dev_info.flow_type_rss_offloads | rss_conf->rss_hf) !=
>>>    	    dev_info.flow_type_rss_offloads) { diff --git
>>> a/lib/librte_ethdev/rte_ethdev.h b/lib/librte_ethdev/rte_ethdev.h
>>> index 40aa6a3..389a3e6 100644
>>> --- a/lib/librte_ethdev/rte_ethdev.h
>>> +++ b/lib/librte_ethdev/rte_ethdev.h
>>> @@ -165,6 +165,17 @@
>>>    #define RTE_ETHDEV_LOG(level, ...) \
>>>    	rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
>>>
>>> +#define RTE_ETH_RSS_HF_REFINE(output, input) do {         \
>>> +	if ((input & ETH_RSS_L3_SRC_ONLY) &&              \
>>> +		(input & ETH_RSS_L3_DST_ONLY))            \
>>> +		output = input & (~(ETH_RSS_L3_SRC_ONLY | \
>>> +			ETH_RSS_L3_DST_ONLY));            \
>>> +	if ((input & ETH_RSS_L4_SRC_ONLY) &&              \
>>> +		(input & ETH_RSS_L4_DST_ONLY))            \
>>> +		output = input & (~(ETH_RSS_L4_SRC_ONLY | \
>>> +			ETH_RSS_L4_DST_ONLY));            \
>>> +} while (0)
>>> +
>>>    struct rte_mbuf;
>>>
>>>    /**
>> Sorry, I missed why macro is used. Inline function is better and easier to read
>> since macros which modify one of its parameters are very error-prone.
> The history is
> [Qi] The consideration is when handle rte_flow_action_rss, we still need to strip it out since this route will bypass the dev_configure or rss_update So there are two options 1, strip out at rte_flow_create , this relief all the PMDs, but code looks a little bit strange.
> 2. handled by PMD themselves
> Anyway both of the cases need this helper function be exposed by rte_ethdev.h, maybe we can define a macro named RTE_ETH_RSS_HF_REFINE?
>
> So I try to use a macro to replace the function of strip_out_src_dst_only. Now, there are two options:
> (1)define a macro in rte_ethdev.h, just like above patch.
> (2)define an inline function "rte_eth_rss_strip_out_src_dst_only" in rte_ethdev.h.
> What do you think? Thanks!

May be inline function rte_eth_rss_hf_refine()?