Re: [net-next 10/15] net/mlx5e: Let channels be SD-aware

[Jakub Kicinski <kuba@kernel.org>]

On Mon, 8 Jan 2024 14:30:54 +0200 Gal Pressman wrote:
> On 05/01/2024 0:50, Jakub Kicinski wrote:
> > On Wed, 20 Dec 2023 16:57:16 -0800 Saeed Mahameed wrote:  
> >> Example for 2 mdevs and 6 channels:
> >> +-------+---------+
> >> | ch ix | mdev ix |
> >> +-------+---------+
> >> |   0   |    0    |
> >> |   1   |    1    |
> >> |   2   |    0    |
> >> |   3   |    1    |
> >> |   4   |    0    |
> >> |   5   |    1    |
> >> +-------+---------+  
> > 
> > Meaning Rx queue 0 goes to PF 0, Rx queue 1 goes to PF 1, etc.?  
> 
> Correct.
> 
> > Is the user then expected to magic pixie dust the XPS or some such
> > to get to the right queue?  
> 
> I'm confused, how are RX queues related to XPS?

Separate sentence, perhaps I should be more verbose..

> XPS shouldn't be affected, we just make sure that whatever queue XPS
> chose will go out through the "right" PF.

But you said "correct" to queue 0 going to PF 0 and queue 1 to PF 1.
The queue IDs in my question refer to the queue mapping form the stacks
perspective. If user wants to send everything to queue 0 will it use
both PFs?

> So for example, XPS will choose a queue according to the CPU, and the
> driver will make sure that packets transmitted from this SQ are going
> out through the PF closer to that NUMA.

Sounds like queue 0 is duplicated in both PFs, then?

> > How is this going to get represented in the recently merged Netlink
> > queue API?  
> 
> Can you share a link please?

commit a90d56049acc45802f67cd7d4c058ac45b1bc26f
 
> All the logic is internal to the driver, so I expect it to be fine, but
> I'd like to double check.

Herm, "internal to the driver" is a bit of a landmine. It will be fine
for iperf testing but real users will want to configure the NIC.