[dpdk-dev] [PATCH v3 0/6] Enable VF RSS for Niantic

[Vlad Zolotarov]

On 01/06/15 03:11, Ouyang, Changchun wrote:
>
>> -----Original Message-----
>> From: Vlad Zolotarov [mailto:vladz at cloudius-systems.com]
>> Sent: Monday, January 5, 2015 9:02 PM
>> To: Richardson, Bruce; Ouyang, Changchun
>> Cc: dev at dpdk.org
>> Subject: Re: [dpdk-dev] [PATCH v3 0/6] Enable VF RSS for Niantic
>>
>>
>> On 01/05/15 12:38, Bruce Richardson wrote:
>>> On Thu, Dec 25, 2014 at 01:46:54AM +0000, Ouyang, Changchun wrote:
>>>> Hi,
>>>>
>>>>> -----Original Message-----
>>>>> From: Vlad Zolotarov [mailto:vladz at cloudius-systems.com]
>>>>> Sent: Wednesday, December 24, 2014 5:59 PM
>>>>> To: Ouyang, Changchun; dev at dpdk.org
>>>>> Subject: Re: [dpdk-dev] [PATCH v3 0/6] Enable VF RSS for Niantic
>>>>>
>>>    >
>>>>> On the contrary - it's a very good idea! We use DPDK on Amazon's
>>>>> guests with enhanced networking and we have no access to the PF. We
>>>>> still need to know the RSS redirection rules for our VF pool. From
>>>>> the 82599 spec, chapter
>>>>> 4.6.10.1.1: "redirection table is common to all the pools and only
>>>>> indicates the queue inside the pool to use once the pool is chosen".
>>>>> In that case we need to get the whole 128 entries of the RETA. Is
>>>>> there a reason why we can't have it?
>>>>>
>>>> Due to hardware limitation, VF could not query its own reta table,
>>>> because there is not its own reta, The reta table shared by pf and all vfs.
>>>> If you need know it, query them on pf is feasible way to do it.
>>>>
>>> It's not feasible if you only have access to a guest. :-) IMHO since
>>> the guest is seeing the results of the RSS redirection table, it
>>> should be able to query the table, if it wants. It should not,
>>> however, be able to modify the table, as it is owned by the PF.
>> This is exactly what I meant! ;)
>> The problem at the moment is that upstream PF driver has no VF-PF
>> command for that and I'm in the process of pushing the patch for it.
>> Then it's accepted (and pushed into the Amazon's HV ;)) then DPDK's VF
>> driver may proceed with what u and me are suggesting.
> Besides lack of command between pf and vf, another issue, pf also need know which entries from the whole 128 entries in reta table are assigned
> To a specified vf.

First of all PF knows since it configures it for a VF in a x550 and for 
older devices the (whole) RETA is shared between the PF and VF. There is 
a per-pool RTYPE[n].RQPL
register that defines the number of lsb's from the redirection table to 
consider (see my patch series "ixgbevf: Allow querying VFs RSS 
indirection table and key" in the netdev list).

>
>> Not related question to Intel guys: I can't find a x550 spec in the net.
>> Can anybody tell me where it may be found? ;)
> AFAIK, not yet

Well, too bad... ;) I'll have to hope I guessed right during the driver 
reverse engineering... ;)

>
>>> Regards,
>>> /Bruce
>>>