[PATCH] net/mlx5: add hairpin counter per queue

[Dariusz Sosnowski]

> -----Original Message-----
> From: Shani Peretz <shperetz at nvidia.com>
> Sent: Tuesday, September 10, 2024 09:41
> To: dev at dpdk.org
> Cc: Shani Peretz <shperetz at nvidia.com>; Maayan Kashani
> <mkashani at nvidia.com>; Raslan Darawsheh <rasland at nvidia.com>; Dariusz
> Sosnowski <dsosnowski at nvidia.com>; Slava Ovsiienko
> <viacheslavo at nvidia.com>; Bing Zhao <bingz at nvidia.com>; Ori Kam
> <orika at nvidia.com>; Suanming Mou <suanmingm at nvidia.com>; Matan
> Azrad <matan at nvidia.com>; Anatoly Burakov <anatoly.burakov at intel.com>
> Subject: [PATCH] net/mlx5: add hairpin counter per queue
> 
> Currently mlx5 PMD exposes hairpin_out_of_buffer counter that specifically
> tracks packets dropped when hairpin Rx queue is full.
> 
> This patch allows the user to choose whether they want to track packet drops
> at a per-port level or a per-queue level.
> 
> If the user chooses to have a counter per queue, HW objects will be created for
> each individual queue, and each Rx queue will be assigned to its specific
> counter.
> 
> However, if the user has more hairpin Rx queues than available hardware
> counters, we will attach the first hairpin queues to the available counters, and
> the remaining queues will not have their packet drops tracked by these
> counters.
> 
> This approach provides flexibility for users to monitor packet drops at either a
> port-level or a queue-level granularity.
> 
> Signed-off-by: Shani Peretz <shperetz at nvidia.com>

Acked-by: Dariusz Sosnowski <dsosnowski at nvidia.com>

Best regards,
Dariusz Sosnowski