Re: Issues around packet capture when secondary process is doing rx/tx

[Ferruh Yigit <ferruh.yigit@amd.com>]

On 1/9/2024 11:07 PM, Stephen Hemminger wrote:
> On Tue, 9 Jan 2024 15:06:47 -0800
> Stephen Hemminger <stephen@networkplumber.org> wrote:
> 
>> On Mon, 8 Jan 2024 15:13:25 +0000
>> Konstantin Ananyev <konstantin.ananyev@huawei.com> wrote:
>>
>>>> I have been looking at a problem reported by Sandesh
>>>> where packet capture does not work if rx/tx burst is done in secondary process.
>>>>
>>>> The root cause is that existing rx/tx callback model just doesn't work
>>>> unless the process doing the rx/tx burst calls is the same one that
>>>> registered the callbacks.
>>>>
>>>> An example sequence would be:
>>>> 	1. dumpcap (or pdump) as secondary tells pdump in primary to register callback
>>>> 	2. secondary process calls rx_burst.
>>>> 	3. rx_burst sees the callback but it has pointer pdump_rx which is not necessarily
>>>> 	   at same location in primary and secondary process.
>>>> 	4. indirect function call in secondary to bad location likely causes crash.    
>>>
>>> As I remember, RX/TX callbacks were never intended to work over multiple processes.
>>> Right now RX/TX callbacks are private for the process, different process simply should not
>>> see/execute them.
>>> I.E. it callbacks list is part of 'struct rte_eth_dev' itself, not the rte_eth_dev.data that is shared
>>> between processes.
>>> It should be normal, wehn for the same port/queue you will end-up with different list of callbacks
>>> for different processes.  
>>> So, unless I am missing something, I don't see how we can end-up with 3) and 4) from above:
>>> From my understanding secondary process will never see/call primary's callbacks.
>>>
>>> About pdump itself, it was a while when I looked at it last time, but as I remember to start it to work,
>>> server process has to call rte_pdump_init() which in terns register PDUMP_MP handler.
>>> I suppose for the secondary process to act as a 'pdump server' it needs to call rte_pdump_init() itself,
>>> though I am not sure such option is supported right now. 
>>>    
>>
>> Did some more tests with modified testpmd, and reached some conclusions:
>>
>> The logical interface would be to allow rte_pdump_init() to be called by
>>    the process that would be using rx/tx burst API's.
>>
>>   This doesn't work as it should because the multi-process socket API
>>   assumes that the it only runs the server in primary.  The secondary
>>   can start its own MP thread, but it won't work:
>>
>>   Primary EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
>>   Secondary: EAL: Multi-process socket /var/run/dpdk/rte/mp_socket_6057_1ccd4157fd5
>>
>>   The problem is when client (pdump or dumpcap) tries to run, it uses the mp_socket
>>   in the primary which causes: EAL: Cannot find action: mp_pdump
>>
>>   Looks like the whole MP socket mechanism is just not up to this.
>>
>> Maybe pdump needs to have its own socket and control thread?
>> Or MP socket needs to have some multicast fanout to all secondaries?
>>

I replied to old email but you seem already figured out the root cause.

So when a secondary sends an MP message, the registered MP handler in
another secondary is not called.

As you suggested fan-out to all secondaries with a flag in the message
can be an option.


And one of the reasons MP socket added was, when a device hotplugged in
secondary, this new device populated both in primary and secondary.
And as far as I know if there are multiple secondaries, device populated
to all, not via secondary to secondary communication, but via primary to
all secondaries communication.
So some kind of fan out to all secondaries should be happening for
hotplugging device usecase.