[dpdk-dev] [PATCH 1/2] lib/security: add support for saving app cookie

[Radu Nicolau]

Hi


On 11/20/2017 3:32 PM, Anoob wrote:
> Hi,
>
> Having something like "get_pkt_metadata" should be fine for inline 
> protocol usage. But we will still need a "get cookie" call to get the 
> cookie, as the application would need something it can interpret.
Why can't you have a get_pkt_metadata that returns the "cookie" - which 
I would call it userdata or similar? What I'm trying to say is that we 
should try to keep it as generic as possible. For example, I wouldn't 
assume that the cookie is stored in pkt->udata64 in the application.

> And, even though it seems both are symmetric operations(get & set pkt 
> metadata), there are some minor differences in what they would do. Set 
> metadata would be setting metadata(udata64 member) in mbuf, while get 
> metadata is not exactly returning metadata. We use the actual metadata 
> to get something else(security session in the proposed patch). That 
> was the primary motive for adding "session_get" API.
What they do exactly is left to the PMD implementation. From the user's 
perspective, it does not matter.
There is no requirement that set pkt metadata will set the udata64 member.
>
> Thanks,
> Anoob
>
> On 11/20/2017 05:42 PM, Radu Nicolau wrote:
>> Hi,
>>
>>
>> Why not have something similar to rte_security_set_pkt_metadata, for 
>> example:
>>
>> void *
>> rte_security_get_pkt_metadata(struct rte_security_ctx *instance,
>>                   struct rte_mbuf *mb);
>>
>> and keep internally in the PMD all the required references. The 
>> returned value will be device-specific, so it's flexible enough to 
>> include anything required (just as void *params is in the 
>> set_pkt_metadata).
>>
>> I think it will make a cleaner and more consistent implementation.
>>
>>
>> Regards,
>>
>> Radu
>>
>>
>>
>> On 11/20/2017 10:31 AM, Anoob Joseph wrote:
>>> In case of inline protocol processed ingress traffic, the packet may 
>>> not
>>> have enough information to determine the security parameters with which
>>> the packet was processed. In such cases, the application could register
>>> a cookie, which will be saved in the the security session.
>>>
>>> As the ingress packets are received in the application, it will have
>>> some metadata set in the mbuf. Application can pass this metadata to
>>> "rte_security_session_get" API to retrieve the security session. Once
>>> the security session is determined, another driver call with the
>>> security session will give the application the cookie it had 
>>> registered.
>>>
>>> The cookie will be registered while creating the security session.
>>> Without the cookie, the selector check (SP-SA check) for the incoming
>>> IPsec traffic won't be possible in the application.
>>>
>>> Application can choose what it should register as the cookie. It can
>>> register SPI or a pointer to SA.
>>>
>>> Signed-off-by: Anoob Joseph <anoob.joseph at caviumnetworks.com>
>>> ---
>>>   lib/librte_security/rte_security.c        | 26 
>>> +++++++++++++++++++++++
>>>   lib/librte_security/rte_security.h        | 30 
>>> +++++++++++++++++++++++++++
>>>   lib/librte_security/rte_security_driver.h | 34 
>>> +++++++++++++++++++++++++++++++
>>>   3 files changed, 90 insertions(+)
>>> <snip>
>