From: Gal Pressman <gal@nvidia.com>

Report RX hardware GRO statistics via the netdev queue stats API by
mapping the existing gro_packets and gro_bytes counters to the
hw_gro_packets and hw_gro_bytes fields.

Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 0e955568c2f4..774a2e32d5f9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5532,6 +5532,10 @@ static void mlx5e_get_queue_stats_rx(struct net_device *dev, int i,
 	stats->bytes = rq_stats->bytes + xskrq_stats->bytes;
 	stats->alloc_fail = rq_stats->buff_alloc_err +
 			    xskrq_stats->buff_alloc_err;
+
+	stats->hw_gro_packets =
+		rq_stats->gro_packets + xskrq_stats->gro_packets;
+	stats->hw_gro_bytes = rq_stats->gro_bytes + xskrq_stats->gro_bytes;
 }
 
 static void mlx5e_get_queue_stats_tx(struct net_device *dev, int i,
@@ -5568,6 +5572,8 @@ static void mlx5e_get_base_stats(struct net_device *dev,
 		rx->packets = 0;
 		rx->bytes = 0;
 		rx->alloc_fail = 0;
+		rx->hw_gro_packets = 0;
+		rx->hw_gro_bytes = 0;
 
 		for (i = priv->channels.params.num_channels; i < priv->stats_nch; i++) {
 			struct netdev_queue_stats_rx rx_i = {0};
@@ -5577,6 +5583,8 @@ static void mlx5e_get_base_stats(struct net_device *dev,
 			rx->packets += rx_i.packets;
 			rx->bytes += rx_i.bytes;
 			rx->alloc_fail += rx_i.alloc_fail;
+			rx->hw_gro_packets += rx_i.hw_gro_packets;
+			rx->hw_gro_bytes += rx_i.hw_gro_bytes;
 		}
 
 		/* always report PTP RX stats from base as there is no
@@ -5588,6 +5596,8 @@ static void mlx5e_get_base_stats(struct net_device *dev,
 
 			rx->packets += rq_stats->packets;
 			rx->bytes += rq_stats->bytes;
+			rx->hw_gro_packets += rq_stats->gro_packets;
+			rx->hw_gro_bytes += rq_stats->gro_bytes;
 		}
 	}
 
-- 
2.44.0

On Wed, 4 Feb 2026 21:33:12 +0200 Tariq Toukan wrote:
> +	stats->hw_gro_packets =
> +		rq_stats->gro_packets + xskrq_stats->gro_packets;
> +	stats->hw_gro_bytes = rq_stats->gro_bytes + xskrq_stats->gro_bytes;

Doesn't look right.. 

mlx5e_shampo_flush_skb(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe, bool match) 
{                                                                               
        struct sk_buff *skb = rq->hw_gro_data->skb;                             
        struct mlx5e_rq_stats *stats = rq->stats;                               
        u16 gro_count = NAPI_GRO_CB(skb)->count;                                
                                                                                
        if (likely(skb_shinfo(skb)->nr_frags))                                  
                mlx5e_shampo_align_fragment(skb, rq->mpwqe.log_stride_sz);      
        if (gro_count > 1) {                                                    
                stats->gro_skbs++;                                              
                stats->gro_packets += gro_count;       

And:
      -
        name: rx-hw-gro-packets
        doc: |
          Number of packets that were coalesced from smaller packets by the
          device. Counts only packets coalesced with the HW-GRO netdevice
          feature, LRO-coalesced packets are not counted.

      -
        name: rx-hw-gro-wire-packets
        doc: |
          Number of packets that were coalesced to bigger packetss with the
          HW-GRO netdevice feature. LRO-coalesced packets are not counted.
        type: uint

Your gro_packets are "gro-wire-packets" and "gro-packets" are your
gro_skbs.

I really wish the AI was clever enough to catch uAPI mis-reading :(
-- 
pw-bot: cr

On 05/02/2026 7:23, Jakub Kicinski wrote:
> On Wed, 4 Feb 2026 21:33:12 +0200 Tariq Toukan wrote:
>> +	stats->hw_gro_packets =
>> +		rq_stats->gro_packets + xskrq_stats->gro_packets;
>> +	stats->hw_gro_bytes = rq_stats->gro_bytes + xskrq_stats->gro_bytes;
> 
> Doesn't look right.. 
> 
> mlx5e_shampo_flush_skb(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe, bool match) 
> {                                                                               
>         struct sk_buff *skb = rq->hw_gro_data->skb;                             
>         struct mlx5e_rq_stats *stats = rq->stats;                               
>         u16 gro_count = NAPI_GRO_CB(skb)->count;                                
>                                                                                 
>         if (likely(skb_shinfo(skb)->nr_frags))                                  
>                 mlx5e_shampo_align_fragment(skb, rq->mpwqe.log_stride_sz);      
>         if (gro_count > 1) {                                                    
>                 stats->gro_skbs++;                                              
>                 stats->gro_packets += gro_count;       
> 
> And:
>       -
>         name: rx-hw-gro-packets
>         doc: |
>           Number of packets that were coalesced from smaller packets by the
>           device. Counts only packets coalesced with the HW-GRO netdevice
>           feature, LRO-coalesced packets are not counted.
> 
>       -
>         name: rx-hw-gro-wire-packets
>         doc: |
>           Number of packets that were coalesced to bigger packetss with the
>           HW-GRO netdevice feature. LRO-coalesced packets are not counted.
>         type: uint
> 
> Your gro_packets are "gro-wire-packets" and "gro-packets" are your
> gro_skbs.

You're absolutely right, thanks Jakub!

> 
> I really wish the AI was clever enough to catch uAPI mis-reading :(