[PATCH net] amd-xgbe: avoid misleading per-packet error log

[PATCH net] amd-xgbe: avoid misleading per-packet error log

[Raju Rangoju]

On the receive path, packet can be damaged because of buffer
overflow in Rx FIFO. Avoid misleading per-packet error log when
packet->errors is set, which could spam logs and confuse users.
Instead, rely on the stats.

Fixes: c5aa9e3b8156 ("amd-xgbe: Initial AMD 10GbE platform driver")
Signed-off-by: Raju Rangoju <Raju.Rangoju@amd.com>
---
 drivers/net/ethernet/amd/xgbe/xgbe-common.h  | 3 +++
 drivers/net/ethernet/amd/xgbe/xgbe-dev.c     | 5 +++++
 drivers/net/ethernet/amd/xgbe/xgbe-drv.c     | 6 +++---
 drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c | 2 ++
 drivers/net/ethernet/amd/xgbe/xgbe.h         | 2 ++
 5 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-common.h b/drivers/net/ethernet/amd/xgbe/xgbe-common.h
index 62b01de93db4..4a0548e0cbaf 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-common.h
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-common.h
@@ -1134,6 +1134,9 @@
 #define RX_NORMAL_DESC3_RSV_INDEX		26
 #define RX_NORMAL_DESC3_RSV_WIDTH		1
 
+/* RX normal DESC3 values */
+#define RX_NORMAL_DESC3_ETLT_FIFO_OVERFLOW	0x7
+
 #define RX_DESC3_L34T_IPV4_TCP			1
 #define RX_DESC3_L34T_IPV4_UDP			2
 #define RX_DESC3_L34T_IPV4_ICMP			3
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-dev.c b/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
index b646ae575e6a..9cb57e0b26bc 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
@@ -1968,9 +1968,14 @@ static int xgbe_dev_read(struct xgbe_channel *channel)
 			XGMAC_SET_BITS(packet->attributes, RX_PACKET_ATTRIBUTES,
 				       TNPCSUM_DONE, 0);
 			pdata->ext_stats.rx_vxlan_csum_errors++;
+		} else if (etlt == RX_NORMAL_DESC3_ETLT_FIFO_OVERFLOW) {
+			XGMAC_SET_BITS(packet->errors, RX_PACKET_ERRORS,
+				       OVERRUN, 1);
+			pdata->ext_stats.rx_buffer_overflow++;
 		} else {
 			XGMAC_SET_BITS(packet->errors, RX_PACKET_ERRORS,
 				       FRAME, 1);
+			pdata->ext_stats.rx_pkt_errors++;
 		}
 	}
 
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
index 3ddd896d6987..132a2e1368ed 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
@@ -2292,9 +2292,9 @@ static int xgbe_rx_poll(struct xgbe_channel *channel, int budget)
 			goto read_again;
 
 		if (error || packet->errors) {
-			if (packet->errors)
-				netif_err(pdata, rx_err, netdev,
-					  "error in received packet\n");
+			/* packet->errors may indicate RX FIFO overflow;
+			 * drop without per-packet log
+			 */
 			dev_kfree_skb(skb);
 			goto next_packet;
 		}
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c b/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
index 0d19b09497a0..c8797a338542 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
@@ -75,6 +75,8 @@ static const struct xgbe_stats xgbe_gstring_stats[] = {
 	XGMAC_MMC_STAT("rx_pause_frames", rxpauseframes),
 	XGMAC_EXT_STAT("rx_split_header_packets", rx_split_header_packets),
 	XGMAC_EXT_STAT("rx_buffer_unavailable", rx_buffer_unavailable),
+	XGMAC_EXT_STAT("rx_buffer_overflow", rx_buffer_overflow),
+	XGMAC_EXT_STAT("rx_pkt_errors", rx_pkt_errors),
 };
 
 #define XGBE_STATS_COUNT	ARRAY_SIZE(xgbe_gstring_stats)
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe.h b/drivers/net/ethernet/amd/xgbe/xgbe.h
index 03ef0f548483..735c757e1603 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe.h
+++ b/drivers/net/ethernet/amd/xgbe/xgbe.h
@@ -675,6 +675,8 @@ struct xgbe_ext_stats {
 	u64 rx_vxlan_packets;
 	u64 rx_csum_errors;
 	u64 rx_vxlan_csum_errors;
+	u64 rx_buffer_overflow;
+	u64 rx_pkt_errors;
 };
 
 struct xgbe_pps_config {
-- 
2.34.1

Re: [PATCH net] amd-xgbe: avoid misleading per-packet error log

[Jakub Kicinski]

On Sat, 10 Jan 2026 22:34:22 +0530 Raju Rangoju wrote:
> diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c b/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
> index 0d19b09497a0..c8797a338542 100644
> --- a/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
> +++ b/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
> @@ -75,6 +75,8 @@ static const struct xgbe_stats xgbe_gstring_stats[] = {
>  	XGMAC_MMC_STAT("rx_pause_frames", rxpauseframes),
>  	XGMAC_EXT_STAT("rx_split_header_packets", rx_split_header_packets),
>  	XGMAC_EXT_STAT("rx_buffer_unavailable", rx_buffer_unavailable),
> +	XGMAC_EXT_STAT("rx_buffer_overflow", rx_buffer_overflow),

"rx_buffer_overflow" sounds like something that should be categorized
to one of the struct rtnl_link_stats64 stats. See the kdoc in if_link.h

> +	XGMAC_EXT_STAT("rx_pkt_errors", rx_pkt_errors),

Are these counted to standard rx_errors somewhere?
-- 
pw-bot: cr

Re: [PATCH net] amd-xgbe: avoid misleading per-packet error log

[Rangoju, Raju]

On 1/13/2026 8:58 AM, Jakub Kicinski wrote:
> Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
> 
> 
> On Sat, 10 Jan 2026 22:34:22 +0530 Raju Rangoju wrote:
>> diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c b/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
>> index 0d19b09497a0..c8797a338542 100644
>> --- a/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
>> +++ b/drivers/net/ethernet/amd/xgbe/xgbe-ethtool.c
>> @@ -75,6 +75,8 @@ static const struct xgbe_stats xgbe_gstring_stats[] = {
>>        XGMAC_MMC_STAT("rx_pause_frames", rxpauseframes),
>>        XGMAC_EXT_STAT("rx_split_header_packets", rx_split_header_packets),
>>        XGMAC_EXT_STAT("rx_buffer_unavailable", rx_buffer_unavailable),
>> +     XGMAC_EXT_STAT("rx_buffer_overflow", rx_buffer_overflow),
> 
> "rx_buffer_overflow" sounds like something that should be categorized
> to one of the struct rtnl_link_stats64 stats. See the kdoc in if_link.h
> 
>> +     XGMAC_EXT_STAT("rx_pkt_errors", rx_pkt_errors),
> 
> Are these counted to standard rx_errors somewhere?

Yes, I believe this and the above stats are counted as part of standard 
rtnl_link_stats64. Thanks for reviewing this Jakub.

I'll respin v2 fixing these.

> --
> pw-bot: cr