Merge branch 'sfc-TC-offload-counters'

Edward Cree says:

====================
sfc: TC offload counters

EF100 hardware supports attaching counters to action-sets in the MAE.
Use these counters to implement stats for TC flower offload.

The counters are delivered to the host over a special hardware RX queue
 which should only ever receive counter update messages, not 'real'
 network packets.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/sfc/Makefile

@@ -9,7 +9,7 @@ sfc-y			+= efx.o efx_common.o efx_channels.o nic.o \
 			   ef100_ethtool.o ef100_rx.o ef100_tx.o
 sfc-$(CONFIG_SFC_MTD)	+= mtd.o
 sfc-$(CONFIG_SFC_SRIOV)	+= sriov.o ef10_sriov.o ef100_sriov.o ef100_rep.o \
-                           mae.o tc.o tc_bindings.o
+                           mae.o tc.o tc_bindings.o tc_counters.o
 
 obj-$(CONFIG_SFC)	+= sfc.o
 

drivers/net/ethernet/sfc/ef100_rx.c

@@ -21,7 +21,7 @@
 /* Get the value of a field in the RX prefix */
 #define PREFIX_OFFSET_W(_f)	(ESF_GZ_RX_PREFIX_ ## _f ## _LBN / 32)
 #define PREFIX_OFFSET_B(_f)	(ESF_GZ_RX_PREFIX_ ## _f ## _LBN % 32)
-#define PREFIX_WIDTH_MASK(_f)	((1UL << ESF_GZ_RX_PREFIX_ ## _f ## _WIDTH) - 1)
+#define PREFIX_WIDTH_MASK(_f)	((1ULL << ESF_GZ_RX_PREFIX_ ## _f ## _WIDTH) - 1)
 #define PREFIX_WORD(_p, _f)	le32_to_cpu((__force __le32)(_p)[PREFIX_OFFSET_W(_f)])
 #define PREFIX_FIELD(_p, _f)	((PREFIX_WORD(_p, _f) >> PREFIX_OFFSET_B(_f)) & \
 				 PREFIX_WIDTH_MASK(_f))
@@ -67,6 +67,13 @@ void __ef100_rx_packet(struct efx_channel *channel)
 
 	prefix = (u32 *)(eh - ESE_GZ_RX_PKT_PREFIX_LEN);
 
+	if (channel->type->receive_raw) {
+		u32 mark = PREFIX_FIELD(prefix, USER_MARK);
+
+		if (channel->type->receive_raw(rx_queue, mark))
+			return; /* packet was consumed */
+	}
+
 	if (ef100_has_fcs_error(channel, prefix) &&
 	    unlikely(!(efx->net_dev->features & NETIF_F_RXALL)))
 		goto out;
@@ -183,24 +190,32 @@ void efx_ef100_ev_rx(struct efx_channel *channel, const efx_qword_t *p_event)
 
 void ef100_rx_write(struct efx_rx_queue *rx_queue)
 {
+	unsigned int notified_count = rx_queue->notified_count;
 	struct efx_rx_buffer *rx_buf;
 	unsigned int idx;
 	efx_qword_t *rxd;
 	efx_dword_t rxdb;
 
-	while (rx_queue->notified_count != rx_queue->added_count) {
-		idx = rx_queue->notified_count & rx_queue->ptr_mask;
+	while (notified_count != rx_queue->added_count) {
+		idx = notified_count & rx_queue->ptr_mask;
 		rx_buf = efx_rx_buffer(rx_queue, idx);
 		rxd = efx_rx_desc(rx_queue, idx);
 
 		EFX_POPULATE_QWORD_1(*rxd, ESF_GZ_RX_BUF_ADDR, rx_buf->dma_addr);
 
-		++rx_queue->notified_count;
+		++notified_count;
 	}
+	if (notified_count == rx_queue->notified_count)
+		return;
 
 	wmb();
 	EFX_POPULATE_DWORD_1(rxdb, ERF_GZ_RX_RING_PIDX,
 			     rx_queue->added_count & rx_queue->ptr_mask);
 	efx_writed_page(rx_queue->efx, &rxdb,
 			ER_GZ_RX_RING_DOORBELL, efx_rx_queue_index(rx_queue));
+	if (rx_queue->grant_credits)
+		wmb();
+	rx_queue->notified_count = notified_count;
+	if (rx_queue->grant_credits)
+		schedule_work(&rx_queue->grant_work);
 }

drivers/net/ethernet/sfc/efx_channels.c

@@ -1119,6 +1119,8 @@ void efx_start_channels(struct efx_nic *efx)
 	struct efx_channel *channel;
 
 	efx_for_each_channel_rev(channel, efx) {
+		if (channel->type->start)
+			channel->type->start(channel);
 		efx_for_each_channel_tx_queue(tx_queue, channel) {
 			efx_init_tx_queue(tx_queue);
 			atomic_inc(&efx->active_queues);
@@ -1143,8 +1145,13 @@ void efx_stop_channels(struct efx_nic *efx)
 	struct efx_channel *channel;
 	int rc = 0;
 
-	/* Stop RX refill */
+	/* Stop special channels and RX refill.
+	 * The channel's stop has to be called first, since it might wait
+	 * for a sentinel RX to indicate the channel has fully drained.
+	 */
 	efx_for_each_channel(channel, efx) {
+		if (channel->type->stop)
+			channel->type->stop(channel);
 		efx_for_each_channel_rx_queue(rx_queue, channel)
 			rx_queue->refill_enabled = false;
 	}

drivers/net/ethernet/sfc/mae.c

@@ -112,6 +112,117 @@ int efx_mae_lookup_mport(struct efx_nic *efx, u32 selector, u32 *id)
 	return 0;
 }
 
+int efx_mae_start_counters(struct efx_nic *efx, struct efx_rx_queue *rx_queue)
+{
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_MAE_COUNTERS_STREAM_START_V2_IN_LEN);
+	MCDI_DECLARE_BUF(outbuf, MC_CMD_MAE_COUNTERS_STREAM_START_OUT_LEN);
+	u32 out_flags;
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_WORD(inbuf, MAE_COUNTERS_STREAM_START_V2_IN_QID,
+		      efx_rx_queue_index(rx_queue));
+	MCDI_SET_WORD(inbuf, MAE_COUNTERS_STREAM_START_V2_IN_PACKET_SIZE,
+		      efx->net_dev->mtu);
+	MCDI_SET_DWORD(inbuf, MAE_COUNTERS_STREAM_START_V2_IN_COUNTER_TYPES_MASK,
+		       BIT(MAE_COUNTER_TYPE_AR) | BIT(MAE_COUNTER_TYPE_CT) |
+		       BIT(MAE_COUNTER_TYPE_OR));
+	rc = efx_mcdi_rpc(efx, MC_CMD_MAE_COUNTERS_STREAM_START,
+			  inbuf, sizeof(inbuf), outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		return rc;
+	if (outlen < sizeof(outbuf))
+		return -EIO;
+	out_flags = MCDI_DWORD(outbuf, MAE_COUNTERS_STREAM_START_OUT_FLAGS);
+	if (out_flags & BIT(MC_CMD_MAE_COUNTERS_STREAM_START_OUT_USES_CREDITS_OFST)) {
+		netif_dbg(efx, drv, efx->net_dev,
+			  "MAE counter stream uses credits\n");
+		rx_queue->grant_credits = true;
+		out_flags &= ~BIT(MC_CMD_MAE_COUNTERS_STREAM_START_OUT_USES_CREDITS_OFST);
+	}
+	if (out_flags) {
+		netif_err(efx, drv, efx->net_dev,
+			  "MAE counter stream start: unrecognised flags %x\n",
+			  out_flags);
+		goto out_stop;
+	}
+	return 0;
+out_stop:
+	efx_mae_stop_counters(efx, rx_queue);
+	return -EOPNOTSUPP;
+}
+
+static bool efx_mae_counters_flushed(u32 *flush_gen, u32 *seen_gen)
+{
+	int i;
+
+	for (i = 0; i < EFX_TC_COUNTER_TYPE_MAX; i++)
+		if ((s32)(flush_gen[i] - seen_gen[i]) > 0)
+			return false;
+	return true;
+}
+
+int efx_mae_stop_counters(struct efx_nic *efx, struct efx_rx_queue *rx_queue)
+{
+	MCDI_DECLARE_BUF(outbuf, MC_CMD_MAE_COUNTERS_STREAM_STOP_V2_OUT_LENMAX);
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_MAE_COUNTERS_STREAM_STOP_IN_LEN);
+	size_t outlen;
+	int rc, i;
+
+	MCDI_SET_WORD(inbuf, MAE_COUNTERS_STREAM_STOP_IN_QID,
+		      efx_rx_queue_index(rx_queue));
+	rc = efx_mcdi_rpc(efx, MC_CMD_MAE_COUNTERS_STREAM_STOP,
+			  inbuf, sizeof(inbuf), outbuf, sizeof(outbuf), &outlen);
+
+	if (rc)
+		return rc;
+
+	netif_dbg(efx, drv, efx->net_dev, "Draining counters:\n");
+	/* Only process received generation counts */
+	for (i = 0; (i < (outlen / 4)) && (i < EFX_TC_COUNTER_TYPE_MAX); i++) {
+		efx->tc->flush_gen[i] = MCDI_ARRAY_DWORD(outbuf,
+							 MAE_COUNTERS_STREAM_STOP_V2_OUT_GENERATION_COUNT,
+							 i);
+		netif_dbg(efx, drv, efx->net_dev,
+			  "\ttype %u, awaiting gen %u\n", i,
+			  efx->tc->flush_gen[i]);
+	}
+
+	efx->tc->flush_counters = true;
+
+	/* Drain can take up to 2 seconds owing to FWRIVERHD-2884; whatever
+	 * timeout we use, that delay is added to unload on nonresponsive
+	 * hardware, so 2500ms seems like a reasonable compromise.
+	 */
+	if (!wait_event_timeout(efx->tc->flush_wq,
+				efx_mae_counters_flushed(efx->tc->flush_gen,
+							 efx->tc->seen_gen),
+				msecs_to_jiffies(2500)))
+		netif_warn(efx, drv, efx->net_dev,
+			   "Failed to drain counters RXQ, FW may be unhappy\n");
+
+	efx->tc->flush_counters = false;
+
+	return rc;
+}
+
+void efx_mae_counters_grant_credits(struct work_struct *work)
+{
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_MAE_COUNTERS_STREAM_GIVE_CREDITS_IN_LEN);
+	struct efx_rx_queue *rx_queue = container_of(work, struct efx_rx_queue,
+						     grant_work);
+	struct efx_nic *efx = rx_queue->efx;
+	unsigned int credits;
+
+	BUILD_BUG_ON(MC_CMD_MAE_COUNTERS_STREAM_GIVE_CREDITS_OUT_LEN);
+	credits = READ_ONCE(rx_queue->notified_count) - rx_queue->granted_count;
+	MCDI_SET_DWORD(inbuf, MAE_COUNTERS_STREAM_GIVE_CREDITS_IN_NUM_CREDITS,
+		       credits);
+	if (!efx_mcdi_rpc(efx, MC_CMD_MAE_COUNTERS_STREAM_GIVE_CREDITS,
+			  inbuf, sizeof(inbuf), NULL, 0, NULL))
+		rx_queue->granted_count += credits;
+}
+
 static int efx_mae_get_basic_caps(struct efx_nic *efx, struct mae_caps *caps)
 {
 	MCDI_DECLARE_BUF(outbuf, MC_CMD_MAE_GET_CAPS_OUT_LEN);
@@ -323,6 +434,57 @@ int efx_mae_match_check_caps(struct efx_nic *efx,
 #undef CHECK_BIT
 #undef CHECK
 
+int efx_mae_allocate_counter(struct efx_nic *efx, struct efx_tc_counter *cnt)
+{
+	MCDI_DECLARE_BUF(outbuf, MC_CMD_MAE_COUNTER_ALLOC_OUT_LEN(1));
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_MAE_COUNTER_ALLOC_V2_IN_LEN);
+	size_t outlen;
+	int rc;
+
+	if (!cnt)
+		return -EINVAL;
+
+	MCDI_SET_DWORD(inbuf, MAE_COUNTER_ALLOC_V2_IN_REQUESTED_COUNT, 1);
+	MCDI_SET_DWORD(inbuf, MAE_COUNTER_ALLOC_V2_IN_COUNTER_TYPE, cnt->type);
+	rc = efx_mcdi_rpc(efx, MC_CMD_MAE_COUNTER_ALLOC, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		return rc;
+	/* pcol says this can't happen, since count is 1 */
+	if (outlen < sizeof(outbuf))
+		return -EIO;
+	cnt->fw_id = MCDI_DWORD(outbuf, MAE_COUNTER_ALLOC_OUT_COUNTER_ID);
+	cnt->gen = MCDI_DWORD(outbuf, MAE_COUNTER_ALLOC_OUT_GENERATION_COUNT);
+	return 0;
+}
+
+int efx_mae_free_counter(struct efx_nic *efx, struct efx_tc_counter *cnt)
+{
+	MCDI_DECLARE_BUF(outbuf, MC_CMD_MAE_COUNTER_FREE_OUT_LEN(1));
+	MCDI_DECLARE_BUF(inbuf, MC_CMD_MAE_COUNTER_FREE_V2_IN_LEN);
+	size_t outlen;
+	int rc;
+
+	MCDI_SET_DWORD(inbuf, MAE_COUNTER_FREE_V2_IN_COUNTER_ID_COUNT, 1);
+	MCDI_SET_DWORD(inbuf, MAE_COUNTER_FREE_V2_IN_FREE_COUNTER_ID, cnt->fw_id);
+	MCDI_SET_DWORD(inbuf, MAE_COUNTER_FREE_V2_IN_COUNTER_TYPE, cnt->type);
+	rc = efx_mcdi_rpc(efx, MC_CMD_MAE_COUNTER_FREE, inbuf, sizeof(inbuf),
+			  outbuf, sizeof(outbuf), &outlen);
+	if (rc)
+		return rc;
+	/* pcol says this can't happen, since count is 1 */
+	if (outlen < sizeof(outbuf))
+		return -EIO;
+	/* FW freed a different ID than we asked for, should also never happen.
+	 * Warn because it means we've now got a different idea to the FW of
+	 * what counters exist, which could cause mayhem later.
+	 */
+	if (WARN_ON(MCDI_DWORD(outbuf, MAE_COUNTER_FREE_OUT_FREED_COUNTER_ID) !=
+		    cnt->fw_id))
+		return -EIO;
+	return 0;
+}
+
 static bool efx_mae_asl_id(u32 id)
 {
 	return !!(id & BIT(31));
@@ -339,6 +501,10 @@ int efx_mae_alloc_action_set(struct efx_nic *efx, struct efx_tc_action_set *act)
 		       MC_CMD_MAE_MAC_ADDR_ALLOC_OUT_MAC_ID_NULL);
 	MCDI_SET_DWORD(inbuf, MAE_ACTION_SET_ALLOC_IN_DST_MAC_ID,
 		       MC_CMD_MAE_MAC_ADDR_ALLOC_OUT_MAC_ID_NULL);
+	if (act->count && !WARN_ON(!act->count->cnt))
+		MCDI_SET_DWORD(inbuf, MAE_ACTION_SET_ALLOC_IN_COUNTER_ID,
+			       act->count->cnt->fw_id);
+	else
 		MCDI_SET_DWORD(inbuf, MAE_ACTION_SET_ALLOC_IN_COUNTER_ID,
 			       MC_CMD_MAE_COUNTER_ALLOC_OUT_COUNTER_ID_NULL);
 	MCDI_SET_DWORD(inbuf, MAE_ACTION_SET_ALLOC_IN_COUNTER_LIST_ID,

drivers/net/ethernet/sfc/mae.h

@@ -27,6 +27,10 @@ void efx_mae_mport_mport(struct efx_nic *efx, u32 mport_id, u32 *out);
 
 int efx_mae_lookup_mport(struct efx_nic *efx, u32 selector, u32 *id);
 
+int efx_mae_start_counters(struct efx_nic *efx, struct efx_rx_queue *rx_queue);
+int efx_mae_stop_counters(struct efx_nic *efx, struct efx_rx_queue *rx_queue);
+void efx_mae_counters_grant_credits(struct work_struct *work);
+
 #define MAE_NUM_FIELDS	(MAE_FIELD_ENC_VNET_ID + 1)
 
 struct mae_caps {
@@ -41,6 +45,9 @@ int efx_mae_match_check_caps(struct efx_nic *efx,
 			     const struct efx_tc_match_fields *mask,
 			     struct netlink_ext_ack *extack);
 
+int efx_mae_allocate_counter(struct efx_nic *efx, struct efx_tc_counter *cnt);
+int efx_mae_free_counter(struct efx_nic *efx, struct efx_tc_counter *cnt);
+
 int efx_mae_alloc_action_set(struct efx_nic *efx, struct efx_tc_action_set *act);
 int efx_mae_free_action_set(struct efx_nic *efx, u32 fw_id);
 

drivers/net/ethernet/sfc/mae_counter_format.h

+/* SPDX-License-Identifier: GPL-2.0-only */
+/****************************************************************************
+ * Driver for Solarflare network controllers and boards
+ * Copyright 2020 Xilinx, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+/* Format of counter packets (version 2) from the ef100 Match-Action Engine */
+
+#ifndef EFX_MAE_COUNTER_FORMAT_H
+#define EFX_MAE_COUNTER_FORMAT_H
+
+
+/*------------------------------------------------------------*/
+/*
+ * ER_RX_SL_PACKETISER_HEADER_WORD(160bit):
+ * 
+ */
+#define ER_RX_SL_PACKETISER_HEADER_WORD_SIZE 20
+#define ER_RX_SL_PACKETISER_HEADER_WORD_WIDTH 160
+
+#define ERF_SC_PACKETISER_HEADER_VERSION_LBN 0
+#define ERF_SC_PACKETISER_HEADER_VERSION_WIDTH 8
+#define ERF_SC_PACKETISER_HEADER_VERSION_VALUE 2
+#define ERF_SC_PACKETISER_HEADER_IDENTIFIER_LBN 8
+#define ERF_SC_PACKETISER_HEADER_IDENTIFIER_WIDTH 8
+#define ERF_SC_PACKETISER_HEADER_IDENTIFIER_AR 0
+#define ERF_SC_PACKETISER_HEADER_IDENTIFIER_CT 1
+#define ERF_SC_PACKETISER_HEADER_IDENTIFIER_OR 2
+#define ERF_SC_PACKETISER_HEADER_HEADER_OFFSET_LBN 16
+#define ERF_SC_PACKETISER_HEADER_HEADER_OFFSET_WIDTH 8
+#define ERF_SC_PACKETISER_HEADER_HEADER_OFFSET_DEFAULT 0x4
+#define ERF_SC_PACKETISER_HEADER_PAYLOAD_OFFSET_LBN 24
+#define ERF_SC_PACKETISER_HEADER_PAYLOAD_OFFSET_WIDTH 8
+#define ERF_SC_PACKETISER_HEADER_PAYLOAD_OFFSET_DEFAULT 0x14
+#define ERF_SC_PACKETISER_HEADER_INDEX_LBN 32
+#define ERF_SC_PACKETISER_HEADER_INDEX_WIDTH 16
+#define ERF_SC_PACKETISER_HEADER_COUNT_LBN 48
+#define ERF_SC_PACKETISER_HEADER_COUNT_WIDTH 16
+#define ERF_SC_PACKETISER_HEADER_RESERVED_0_LBN 64
+#define ERF_SC_PACKETISER_HEADER_RESERVED_0_WIDTH 32
+#define ERF_SC_PACKETISER_HEADER_RESERVED_1_LBN 96
+#define ERF_SC_PACKETISER_HEADER_RESERVED_1_WIDTH 32
+#define ERF_SC_PACKETISER_HEADER_RESERVED_2_LBN 128
+#define ERF_SC_PACKETISER_HEADER_RESERVED_2_WIDTH 32
+
+
+/*------------------------------------------------------------*/
+/*
+ * ER_RX_SL_PACKETISER_PAYLOAD_WORD(128bit):
+ * 
+ */
+#define ER_RX_SL_PACKETISER_PAYLOAD_WORD_SIZE 16
+#define ER_RX_SL_PACKETISER_PAYLOAD_WORD_WIDTH 128
+
+#define ERF_SC_PACKETISER_PAYLOAD_COUNTER_INDEX_LBN 0
+#define ERF_SC_PACKETISER_PAYLOAD_COUNTER_INDEX_WIDTH 24
+#define ERF_SC_PACKETISER_PAYLOAD_RESERVED_LBN 24
+#define ERF_SC_PACKETISER_PAYLOAD_RESERVED_WIDTH 8
+#define ERF_SC_PACKETISER_PAYLOAD_PACKET_COUNT_OFST 4
+#define ERF_SC_PACKETISER_PAYLOAD_PACKET_COUNT_SIZE 6
+#define ERF_SC_PACKETISER_PAYLOAD_PACKET_COUNT_LBN 32
+#define ERF_SC_PACKETISER_PAYLOAD_PACKET_COUNT_WIDTH 48
+#define ERF_SC_PACKETISER_PAYLOAD_BYTE_COUNT_OFST 10
+#define ERF_SC_PACKETISER_PAYLOAD_BYTE_COUNT_SIZE 6
+#define ERF_SC_PACKETISER_PAYLOAD_BYTE_COUNT_LBN 80
+#define ERF_SC_PACKETISER_PAYLOAD_BYTE_COUNT_WIDTH 48
+
+
+#endif /* EFX_MAE_COUNTER_FORMAT_H */

drivers/net/ethernet/sfc/mcdi.h

@@ -221,6 +221,11 @@ void efx_mcdi_sensor_event(struct efx_nic *efx, efx_qword_t *ev);
 #define MCDI_BYTE(_buf, _field)						\
 	((void)BUILD_BUG_ON_ZERO(MC_CMD_ ## _field ## _LEN != 1),	\
 	 *MCDI_PTR(_buf, _field))
+#define MCDI_SET_WORD(_buf, _field, _value) do {			\
+	BUILD_BUG_ON(MC_CMD_ ## _field ## _LEN != 2);			\
+	BUILD_BUG_ON(MC_CMD_ ## _field ## _OFST & 1);			\
+	*(__force __le16 *)MCDI_PTR(_buf, _field) = cpu_to_le16(_value);\
+	} while (0)
 #define MCDI_WORD(_buf, _field)						\
 	((u16)BUILD_BUG_ON_ZERO(MC_CMD_ ## _field ## _LEN != 2) +	\
 	 le16_to_cpu(*(__force const __le16 *)MCDI_PTR(_buf, _field)))

drivers/net/ethernet/sfc/net_driver.h

@@ -56,7 +56,8 @@
 #define EFX_MAX_RX_QUEUES EFX_MAX_CHANNELS
 #define EFX_EXTRA_CHANNEL_IOV	0
 #define EFX_EXTRA_CHANNEL_PTP	1
-#define EFX_MAX_EXTRA_CHANNELS	2U
+#define EFX_EXTRA_CHANNEL_TC	2
+#define EFX_MAX_EXTRA_CHANNELS	3U
 
 /* Checksum generation is a per-queue option in hardware, so each
  * queue visible to the networking core is backed by two hardware TX
@@ -363,8 +364,12 @@ struct efx_rx_page_state {
  * @refill_enabled: Enable refill whenever fill level is low
  * @flush_pending: Set when a RX flush is pending. Has the same lifetime as
  *	@rxq_flush_pending.
+ * @grant_credits: Posted RX descriptors need to be granted to the MAE with
+ *	%MC_CMD_MAE_COUNTERS_STREAM_GIVE_CREDITS.  For %EFX_EXTRA_CHANNEL_TC,
+ *	and only supported on EF100.
  * @added_count: Number of buffers added to the receive queue.
  * @notified_count: Number of buffers given to NIC (<= @added_count).
+ * @granted_count: Number of buffers granted to the MAE (<= @notified_count).
  * @removed_count: Number of buffers removed from the receive queue.
  * @scatter_n: Used by NIC specific receive code.
  * @scatter_len: Used by NIC specific receive code.
@@ -385,6 +390,7 @@ struct efx_rx_page_state {
  *	refill was triggered.
  * @recycle_count: RX buffer recycle counter.
  * @slow_fill: Timer used to defer efx_nic_generate_fill_event().
+ * @grant_work: workitem used to grant credits to the MAE if @grant_credits
  * @xdp_rxq_info: XDP specific RX queue information.
  * @xdp_rxq_info_valid: Is xdp_rxq_info valid data?.
  */
@@ -396,9 +402,11 @@ struct efx_rx_queue {
 	unsigned int ptr_mask;
 	bool refill_enabled;
 	bool flush_pending;
+	bool grant_credits;
 
 	unsigned int added_count;
 	unsigned int notified_count;
+	unsigned int granted_count;
 	unsigned int removed_count;
 	unsigned int scatter_n;
 	unsigned int scatter_len;
@@ -416,6 +424,7 @@ struct efx_rx_queue {
 	unsigned int recycle_count;
 	struct timer_list slow_fill;
 	unsigned int slow_fill_count;
+	struct work_struct grant_work;
 	/* Statistics to supplement MAC stats */
 	unsigned long rx_packets;
 	struct xdp_rxq_info xdp_rxq_info;
@@ -577,12 +586,15 @@ struct efx_msi_context {
  * struct efx_channel_type - distinguishes traffic and extra channels
  * @handle_no_channel: Handle failure to allocate an extra channel
  * @pre_probe: Set up extra state prior to initialisation
+ * @start: called early in efx_start_channels()
+ * @stop: called early in efx_stop_channels()
  * @post_remove: Tear down extra state after finalisation, if allocated.
  *	May be called on channels that have not been probed.
  * @get_name: Generate the channel's name (used for its IRQ handler)
  * @copy: Copy the channel state prior to reallocation.  May be %NULL if
  *	reallocation is not supported.
  * @receive_skb: Handle an skb ready to be passed to netif_receive_skb()
+ * @receive_raw: Handle an RX buffer ready to be passed to __efx_rx_packet()
  * @want_txqs: Determine whether this channel should have TX queues
  *	created.  If %NULL, TX queues are not created.
  * @keep_eventq: Flag for whether event queue should be kept initialised
@@ -593,10 +605,13 @@ struct efx_msi_context {
 struct efx_channel_type {
 	void (*handle_no_channel)(struct efx_nic *);
 	int (*pre_probe)(struct efx_channel *);
+	int (*start)(struct efx_channel *);
+	void (*stop)(struct efx_channel *);
 	void (*post_remove)(struct efx_channel *);
 	void (*get_name)(struct efx_channel *, char *buf, size_t len);
 	struct efx_channel *(*copy)(const struct efx_channel *);
 	bool (*receive_skb)(struct efx_channel *, struct sk_buff *);
+	bool (*receive_raw)(struct efx_rx_queue *, u32);
 	bool (*want_txqs)(struct efx_channel *);
 	bool keep_eventq;
 	bool want_pio;

drivers/net/ethernet/sfc/rx_common.c

@@ -229,6 +229,7 @@ void efx_init_rx_queue(struct efx_rx_queue *rx_queue)
 	/* Initialise ptr fields */
 	rx_queue->added_count = 0;
 	rx_queue->notified_count = 0;
+	rx_queue->granted_count = 0;
 	rx_queue->removed_count = 0;
 	rx_queue->min_fill = -1U;
 	efx_init_rx_recycle_ring(rx_queue);
@@ -281,6 +282,8 @@ void efx_fini_rx_queue(struct efx_rx_queue *rx_queue)
 		  "shutting down RX queue %d\n", efx_rx_queue_index(rx_queue));
 
 	del_timer_sync(&rx_queue->slow_fill);
+	if (rx_queue->grant_credits)
+		flush_work(&rx_queue->grant_work);
 
 	/* Release RX buffers from the current read ptr to the write ptr */
 	if (rx_queue->buffer) {

drivers/net/ethernet/sfc/tc.c

@@ -77,6 +77,8 @@ static void efx_tc_free_action_set(struct efx_nic *efx,
 		 */
 		list_del(&act->list);
 	}
+	if (act->count)
+		efx_tc_flower_put_counter_index(efx, act->count);
 	kfree(act);
 }
 
@@ -282,6 +284,29 @@ static int efx_tc_flower_parse_match(struct efx_nic *efx,
 	return 0;
 }
 
+/* For details of action order constraints refer to SF-123102-TC-1§12.6.1 */
+enum efx_tc_action_order {
+	EFX_TC_AO_COUNT,
+	EFX_TC_AO_DELIVER
+};
+/* Determine whether we can add @new action without violating order */
+static bool efx_tc_flower_action_order_ok(const struct efx_tc_action_set *act,
+					  enum efx_tc_action_order new)
+{
+	switch (new) {
+	case EFX_TC_AO_COUNT:
+		if (act->count)
+			return false;
+		fallthrough;
+	case EFX_TC_AO_DELIVER:
+		return !act->deliver;
+	default:
+		/* Bad caller.  Whatever they wanted to do, say they can't. */
+		WARN_ON_ONCE(1);
+		return false;
+	}
+}
+
 static int efx_tc_flower_replace(struct efx_nic *efx,
 				 struct net_device *net_dev,
 				 struct flow_cls_offload *tc,
@@ -376,6 +401,47 @@ static int efx_tc_flower_replace(struct efx_nic *efx,
 			goto release;
 		}
 
+		if ((fa->id == FLOW_ACTION_REDIRECT ||
+		     fa->id == FLOW_ACTION_MIRRED ||
+		     fa->id == FLOW_ACTION_DROP) && fa->hw_stats) {
+			struct efx_tc_counter_index *ctr;
+
+			/* Currently the only actions that want stats are
+			 * mirred and gact (ok, shot, trap, goto-chain), which
+			 * means we want stats just before delivery.  Also,
+			 * note that tunnel_key set shouldn't change the length
+			 * — it's only the subsequent mirred that does that,
+			 * and the stats are taken _before_ the mirred action
+			 * happens.
+			 */
+			if (!efx_tc_flower_action_order_ok(act, EFX_TC_AO_COUNT)) {
+				/* All supported actions that count either steal
+				 * (gact shot, mirred redirect) or clone act
+				 * (mirred mirror), so we should never get two
+				 * count actions on one action_set.
+				 */
+				NL_SET_ERR_MSG_MOD(extack, "Count-action conflict (can't happen)");
+				rc = -EOPNOTSUPP;
+				goto release;
+			}
+
+			if (!(fa->hw_stats & FLOW_ACTION_HW_STATS_DELAYED)) {
+				NL_SET_ERR_MSG_FMT_MOD(extack, "hw_stats_type %u not supported (only 'delayed')",
+						       fa->hw_stats);
+				rc = -EOPNOTSUPP;
+				goto release;
+			}
+
+			ctr = efx_tc_flower_get_counter_index(efx, tc->cookie,
+							      EFX_TC_COUNTER_TYPE_AR);
+			if (IS_ERR(ctr)) {
+				rc = PTR_ERR(ctr);
+				NL_SET_ERR_MSG_MOD(extack, "Failed to obtain a counter");
+				goto release;
+			}
+			act->count = ctr;
+		}
+
 		switch (fa->id) {
 		case FLOW_ACTION_DROP:
 			rc = efx_mae_alloc_action_set(efx, act);
@@ -389,6 +455,14 @@ static int efx_tc_flower_replace(struct efx_nic *efx,
 		case FLOW_ACTION_REDIRECT:
 		case FLOW_ACTION_MIRRED:
 			save = *act;
+
+			if (!efx_tc_flower_action_order_ok(act, EFX_TC_AO_DELIVER)) {
+				/* can't happen */
+				rc = -EOPNOTSUPP;
+				NL_SET_ERR_MSG_MOD(extack, "Deliver action violates action order (can't happen)");
+				goto release;
+			}
+
 			to_efv = efx_tc_flower_lookup_efv(efx, fa->dev);
 			if (IS_ERR(to_efv)) {
 				NL_SET_ERR_MSG_MOD(extack, "Mirred egress device not on switch");
@@ -412,6 +486,7 @@ static int efx_tc_flower_replace(struct efx_nic *efx,
 			if (fa->id == FLOW_ACTION_REDIRECT)
 				break; /* end of the line */
 			/* Mirror, so continue on with saved act */
+			save.count = NULL;
 			act = kzalloc(sizeof(*act), GFP_USER);
 			if (!act) {
 				rc = -ENOMEM;
@@ -520,6 +595,42 @@ static int efx_tc_flower_destroy(struct efx_nic *efx,
 	return 0;
 }
 
+static int efx_tc_flower_stats(struct efx_nic *efx, struct net_device *net_dev,
+			       struct flow_cls_offload *tc)
+{
+	struct netlink_ext_ack *extack = tc->common.extack;
+	struct efx_tc_counter_index *ctr;
+	struct efx_tc_counter *cnt;
+	u64 packets, bytes;
+
+	ctr = efx_tc_flower_find_counter_index(efx, tc->cookie);
+	if (!ctr) {
+		/* See comment in efx_tc_flower_destroy() */
+		if (!IS_ERR(efx_tc_flower_lookup_efv(efx, net_dev)))
+			if (net_ratelimit())
+				netif_warn(efx, drv, efx->net_dev,
+					   "Filter %lx not found for stats\n",
+					   tc->cookie);
+		NL_SET_ERR_MSG_MOD(extack, "Flow cookie not found in offloaded rules");
+		return -ENOENT;
+	}
+	if (WARN_ON(!ctr->cnt)) /* can't happen */
+		return -EIO;
+	cnt = ctr->cnt;
+
+	spin_lock_bh(&cnt->lock);
+	/* Report only new pkts/bytes since last time TC asked */
+	packets = cnt->packets;
+	bytes = cnt->bytes;
+	flow_stats_update(&tc->stats, bytes - cnt->old_bytes,
+			  packets - cnt->old_packets, 0, cnt->touched,
+			  FLOW_ACTION_HW_STATS_DELAYED);
+	cnt->old_packets = packets;
+	cnt->old_bytes = bytes;
+	spin_unlock_bh(&cnt->lock);
+	return 0;
+}
+
 int efx_tc_flower(struct efx_nic *efx, struct net_device *net_dev,
 		  struct flow_cls_offload *tc, struct efx_rep *efv)
 {
@@ -536,6 +647,9 @@ int efx_tc_flower(struct efx_nic *efx, struct net_device *net_dev,
 	case FLOW_CLS_DESTROY:
 		rc = efx_tc_flower_destroy(efx, net_dev, tc);
 		break;
+	case FLOW_CLS_STATS:
+		rc = efx_tc_flower_stats(efx, net_dev, tc);
+		break;
 	default:
 		rc = -EOPNOTSUPP;
 		break;
@@ -751,6 +865,10 @@ int efx_init_struct_tc(struct efx_nic *efx)
 	INIT_LIST_HEAD(&efx->tc->block_list);
 
 	mutex_init(&efx->tc->mutex);
+	init_waitqueue_head(&efx->tc->flush_wq);
+	rc = efx_tc_init_counters(efx);
+	if (rc < 0)
+		goto fail_counters;
 	rc = rhashtable_init(&efx->tc->match_action_ht, &efx_tc_match_action_ht_params);
 	if (rc < 0)
 		goto fail_match_action_ht;
@@ -760,8 +878,11 @@ int efx_init_struct_tc(struct efx_nic *efx)
 	efx->tc->dflt.pf.fw_id = MC_CMD_MAE_ACTION_RULE_INSERT_OUT_ACTION_RULE_ID_NULL;
 	INIT_LIST_HEAD(&efx->tc->dflt.wire.acts.list);
 	efx->tc->dflt.wire.fw_id = MC_CMD_MAE_ACTION_RULE_INSERT_OUT_ACTION_RULE_ID_NULL;
+	efx->extra_channel_type[EFX_EXTRA_CHANNEL_TC] = &efx_tc_channel_type;
 	return 0;
 fail_match_action_ht:
+	efx_tc_destroy_counters(efx);
+fail_counters:
 	mutex_destroy(&efx->tc->mutex);
 	kfree(efx->tc->caps);
 fail_alloc_caps:
@@ -782,6 +903,7 @@ void efx_fini_struct_tc(struct efx_nic *efx)
 			     MC_CMD_MAE_ACTION_RULE_INSERT_OUT_ACTION_RULE_ID_NULL);
 	rhashtable_free_and_destroy(&efx->tc->match_action_ht, efx_tc_flow_free,
 				    efx);
+	efx_tc_fini_counters(efx);
 	mutex_unlock(&efx->tc->mutex);
 	mutex_destroy(&efx->tc->mutex);
 	kfree(efx->tc->caps);

drivers/net/ethernet/sfc/tc.h

@@ -14,11 +14,13 @@
 #include <net/flow_offload.h>
 #include <linux/rhashtable.h>
 #include "net_driver.h"
+#include "tc_counters.h"
 
 #define IS_ALL_ONES(v)	(!(typeof (v))~(v))
 
 struct efx_tc_action_set {
 	u16 deliver:1;
+	struct efx_tc_counter_index *count;
 	u32 dest_mport;
 	u32 fw_id; /* index of this entry in firmware actions table */
 	struct list_head list;
@@ -74,11 +76,19 @@ enum efx_tc_rule_prios {
  * @caps: MAE capabilities reported by MCDI
  * @block_list: List of &struct efx_tc_block_binding
  * @mutex: Used to serialise operations on TC hashtables
+ * @counter_ht: Hashtable of TC counters (FW IDs and counter values)
+ * @counter_id_ht: Hashtable mapping TC counter cookies to counters
  * @match_action_ht: Hashtable of TC match-action rules
  * @reps_mport_id: MAE port allocated for representor RX
  * @reps_filter_uc: VNIC filter for representor unicast RX (promisc)
  * @reps_filter_mc: VNIC filter for representor multicast RX (allmulti)
  * @reps_mport_vport_id: vport_id for representor RX filters
+ * @flush_counters: counters have been stopped, waiting for drain
+ * @flush_gen: final generation count per type array as reported by
+ *             MC_CMD_MAE_COUNTERS_STREAM_STOP
+ * @seen_gen: most recent generation count per type as seen by efx_tc_rx()
+ * @flush_wq: wait queue used by efx_mae_stop_counters() to wait for
+ *	MAE counters RXQ to finish draining
  * @dflt: Match-action rules for default switching; at priority
  *	%EFX_TC_PRIO_DFLT.  Named by *ingress* port
  * @dflt.pf: rule for traffic ingressing from PF (egresses to wire)
@@ -89,9 +99,15 @@ struct efx_tc_state {
 	struct mae_caps *caps;
 	struct list_head block_list;
 	struct mutex mutex;
+	struct rhashtable counter_ht;
+	struct rhashtable counter_id_ht;
 	struct rhashtable match_action_ht;
 	u32 reps_mport_id, reps_mport_vport_id;
 	s32 reps_filter_uc, reps_filter_mc;
+	bool flush_counters;
+	u32 flush_gen[EFX_TC_COUNTER_TYPE_MAX];
+	u32 seen_gen[EFX_TC_COUNTER_TYPE_MAX];
+	wait_queue_head_t flush_wq;
 	struct {
 		struct efx_tc_flow_rule pf;
 		struct efx_tc_flow_rule wire;

drivers/net/ethernet/sfc/tc_counters.c

+// SPDX-License-Identifier: GPL-2.0-only
+/****************************************************************************
+ * Driver for Solarflare network controllers and boards
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#include "tc_counters.h"
+#include "mae_counter_format.h"
+#include "mae.h"
+#include "rx_common.h"
+
+/* Counter-management hashtables */
+
+static const struct rhashtable_params efx_tc_counter_id_ht_params = {
+	.key_len	= offsetof(struct efx_tc_counter_index, linkage),
+	.key_offset	= 0,
+	.head_offset	= offsetof(struct efx_tc_counter_index, linkage),
+};
+
+static const struct rhashtable_params efx_tc_counter_ht_params = {
+	.key_len	= offsetof(struct efx_tc_counter, linkage),
+	.key_offset	= 0,
+	.head_offset	= offsetof(struct efx_tc_counter, linkage),
+};
+
+static void efx_tc_counter_free(void *ptr, void *__unused)
+{
+	struct efx_tc_counter *cnt = ptr;
+
+	kfree(cnt);
+}
+
+static void efx_tc_counter_id_free(void *ptr, void *__unused)
+{
+	struct efx_tc_counter_index *ctr = ptr;
+
+	WARN_ON(refcount_read(&ctr->ref));
+	kfree(ctr);
+}
+
+int efx_tc_init_counters(struct efx_nic *efx)
+{
+	int rc;
+
+	rc = rhashtable_init(&efx->tc->counter_id_ht, &efx_tc_counter_id_ht_params);
+	if (rc < 0)
+		goto fail_counter_id_ht;
+	rc = rhashtable_init(&efx->tc->counter_ht, &efx_tc_counter_ht_params);
+	if (rc < 0)
+		goto fail_counter_ht;
+	return 0;
+fail_counter_ht:
+	rhashtable_destroy(&efx->tc->counter_id_ht);
+fail_counter_id_ht:
+	return rc;
+}
+
+/* Only call this in init failure teardown.
+ * Normal exit should fini instead as there may be entries in the table.
+ */
+void efx_tc_destroy_counters(struct efx_nic *efx)
+{
+	rhashtable_destroy(&efx->tc->counter_ht);
+	rhashtable_destroy(&efx->tc->counter_id_ht);
+}
+
+void efx_tc_fini_counters(struct efx_nic *efx)
+{
+	rhashtable_free_and_destroy(&efx->tc->counter_id_ht, efx_tc_counter_id_free, NULL);
+	rhashtable_free_and_destroy(&efx->tc->counter_ht, efx_tc_counter_free, NULL);
+}
+
+/* Counter allocation */
+
+static struct efx_tc_counter *efx_tc_flower_allocate_counter(struct efx_nic *efx,
+							     int type)
+{
+	struct efx_tc_counter *cnt;
+	int rc, rc2;
+
+	cnt = kzalloc(sizeof(*cnt), GFP_USER);
+	if (!cnt)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&cnt->lock);
+	cnt->touched = jiffies;
+	cnt->type = type;
+
+	rc = efx_mae_allocate_counter(efx, cnt);
+	if (rc)
+		goto fail1;
+	rc = rhashtable_insert_fast(&efx->tc->counter_ht, &cnt->linkage,
+				    efx_tc_counter_ht_params);
+	if (rc)
+		goto fail2;
+	return cnt;
+fail2:
+	/* If we get here, it implies that we couldn't insert into the table,
+	 * which in turn probably means that the fw_id was already taken.
+	 * In that case, it's unclear whether we really 'own' the fw_id; but
+	 * the firmware seemed to think we did, so it's proper to free it.
+	 */
+	rc2 = efx_mae_free_counter(efx, cnt);
+	if (rc2)
+		netif_warn(efx, hw, efx->net_dev,
+			   "Failed to free MAE counter %u, rc %d\n",
+			   cnt->fw_id, rc2);
+fail1:
+	kfree(cnt);
+	return ERR_PTR(rc > 0 ? -EIO : rc);
+}
+
+static void efx_tc_flower_release_counter(struct efx_nic *efx,
+					  struct efx_tc_counter *cnt)
+{
+	int rc;
+
+	rhashtable_remove_fast(&efx->tc->counter_ht, &cnt->linkage,
+			       efx_tc_counter_ht_params);
+	rc = efx_mae_free_counter(efx, cnt);
+	if (rc)
+		netif_warn(efx, hw, efx->net_dev,
+			   "Failed to free MAE counter %u, rc %d\n",
+			   cnt->fw_id, rc);
+	/* This doesn't protect counter updates coming in arbitrarily long
+	 * after we deleted the counter.  The RCU just ensures that we won't
+	 * free the counter while another thread has a pointer to it.
+	 * Ensuring we don't update the wrong counter if the ID gets re-used
+	 * is handled by the generation count.
+	 */
+	synchronize_rcu();
+	EFX_WARN_ON_PARANOID(spin_is_locked(&cnt->lock));
+	kfree(cnt);
+}
+
+static struct efx_tc_counter *efx_tc_flower_find_counter_by_fw_id(
+				struct efx_nic *efx, int type, u32 fw_id)
+{
+	struct efx_tc_counter key = {};
+
+	key.fw_id = fw_id;
+	key.type = type;
+
+	return rhashtable_lookup_fast(&efx->tc->counter_ht, &key,
+				      efx_tc_counter_ht_params);
+}
+
+/* TC cookie to counter mapping */
+
+void efx_tc_flower_put_counter_index(struct efx_nic *efx,
+				     struct efx_tc_counter_index *ctr)
+{
+	if (!refcount_dec_and_test(&ctr->ref))
+		return; /* still in use */
+	rhashtable_remove_fast(&efx->tc->counter_id_ht, &ctr->linkage,
+			       efx_tc_counter_id_ht_params);
+	efx_tc_flower_release_counter(efx, ctr->cnt);
+	kfree(ctr);
+}
+
+struct efx_tc_counter_index *efx_tc_flower_get_counter_index(
+				struct efx_nic *efx, unsigned long cookie,
+				enum efx_tc_counter_type type)
+{
+	struct efx_tc_counter_index *ctr, *old;
+	struct efx_tc_counter *cnt;
+
+	ctr = kzalloc(sizeof(*ctr), GFP_USER);
+	if (!ctr)
+		return ERR_PTR(-ENOMEM);
+	ctr->cookie = cookie;
+	old = rhashtable_lookup_get_insert_fast(&efx->tc->counter_id_ht,
+						&ctr->linkage,
+						efx_tc_counter_id_ht_params);
+	if (old) {
+		/* don't need our new entry */
+		kfree(ctr);
+		if (!refcount_inc_not_zero(&old->ref))
+			return ERR_PTR(-EAGAIN);
+		/* existing entry found */
+		ctr = old;
+	} else {
+		cnt = efx_tc_flower_allocate_counter(efx, type);
+		if (IS_ERR(cnt)) {
+			rhashtable_remove_fast(&efx->tc->counter_id_ht,
+					       &ctr->linkage,
+					       efx_tc_counter_id_ht_params);
+			kfree(ctr);
+			return (void *)cnt; /* it's an ERR_PTR */
+		}
+		ctr->cnt = cnt;
+		refcount_set(&ctr->ref, 1);
+	}
+	return ctr;
+}
+
+struct efx_tc_counter_index *efx_tc_flower_find_counter_index(
+				struct efx_nic *efx, unsigned long cookie)
+{
+	struct efx_tc_counter_index key = {};
+
+	key.cookie = cookie;
+	return rhashtable_lookup_fast(&efx->tc->counter_id_ht, &key,
+				      efx_tc_counter_id_ht_params);
+}
+
+/* TC Channel.  Counter updates are delivered on this channel's RXQ. */
+
+static void efx_tc_handle_no_channel(struct efx_nic *efx)
+{
+	netif_warn(efx, drv, efx->net_dev,
+		   "MAE counters require MSI-X and 1 additional interrupt vector.\n");
+}
+
+static int efx_tc_probe_channel(struct efx_channel *channel)
+{
+	struct efx_rx_queue *rx_queue = &channel->rx_queue;
+
+	channel->irq_moderation_us = 0;
+	rx_queue->core_index = 0;
+
+	INIT_WORK(&rx_queue->grant_work, efx_mae_counters_grant_credits);
+
+	return 0;
+}
+
+static int efx_tc_start_channel(struct efx_channel *channel)
+{
+	struct efx_rx_queue *rx_queue = efx_channel_get_rx_queue(channel);
+	struct efx_nic *efx = channel->efx;
+
+	return efx_mae_start_counters(efx, rx_queue);
+}
+
+static void efx_tc_stop_channel(struct efx_channel *channel)
+{
+	struct efx_rx_queue *rx_queue = efx_channel_get_rx_queue(channel);
+	struct efx_nic *efx = channel->efx;
+	int rc;
+
+	rc = efx_mae_stop_counters(efx, rx_queue);
+	if (rc)
+		netif_warn(efx, drv, efx->net_dev,
+			   "Failed to stop MAE counters streaming, rc=%d.\n",
+			   rc);
+	rx_queue->grant_credits = false;
+	flush_work(&rx_queue->grant_work);
+}
+
+static void efx_tc_remove_channel(struct efx_channel *channel)
+{
+}
+
+static void efx_tc_get_channel_name(struct efx_channel *channel,
+				    char *buf, size_t len)
+{
+	snprintf(buf, len, "%s-mae", channel->efx->name);
+}
+
+static void efx_tc_counter_update(struct efx_nic *efx,
+				  enum efx_tc_counter_type counter_type,
+				  u32 counter_idx, u64 packets, u64 bytes,
+				  u32 mark)
+{
+	struct efx_tc_counter *cnt;
+
+	rcu_read_lock(); /* Protect against deletion of 'cnt' */
+	cnt = efx_tc_flower_find_counter_by_fw_id(efx, counter_type, counter_idx);
+	if (!cnt) {
+		/* This can legitimately happen when a counter is removed,
+		 * with updates for the counter still in-flight; however this
+		 * should be an infrequent occurrence.
+		 */
+		if (net_ratelimit())
+			netif_dbg(efx, drv, efx->net_dev,
+				  "Got update for unwanted MAE counter %u type %u\n",
+				  counter_idx, counter_type);
+		goto out;
+	}
+
+	spin_lock_bh(&cnt->lock);
+	if ((s32)mark - (s32)cnt->gen < 0) {
+		/* This counter update packet is from before the counter was
+		 * allocated; thus it must be for a previous counter with
+		 * the same ID that has since been freed, and it should be
+		 * ignored.
+		 */
+	} else {
+		/* Update latest seen generation count.  This ensures that
+		 * even a long-lived counter won't start getting ignored if
+		 * the generation count wraps around, unless it somehow
+		 * manages to go 1<<31 generations without an update.
+		 */
+		cnt->gen = mark;
+		/* update counter values */
+		cnt->packets += packets;
+		cnt->bytes += bytes;
+		cnt->touched = jiffies;
+	}
+	spin_unlock_bh(&cnt->lock);
+out:
+	rcu_read_unlock();
+}
+
+static void efx_tc_rx_version_1(struct efx_nic *efx, const u8 *data, u32 mark)
+{
+	u16 n_counters, i;
+
+	/* Header format:
+	 * + |   0    |   1    |   2    |   3    |
+	 * 0 |version |         reserved         |
+	 * 4 |    seq_index    |   n_counters    |
+	 */
+
+	n_counters = le16_to_cpu(*(const __le16 *)(data + 6));
+
+	/* Counter update entry format:
+	 * | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | a | b | c | d | e | f |
+	 * |  counter_idx  |     packet_count      |      byte_count       |
+	 */
+	for (i = 0; i < n_counters; i++) {
+		const void *entry = data + 8 + 16 * i;
+		u64 packet_count, byte_count;
+		u32 counter_idx;
+
+		counter_idx = le32_to_cpu(*(const __le32 *)entry);
+		packet_count = le32_to_cpu(*(const __le32 *)(entry + 4)) |
+			       ((u64)le16_to_cpu(*(const __le16 *)(entry + 8)) << 32);
+		byte_count = le16_to_cpu(*(const __le16 *)(entry + 10)) |
+			     ((u64)le32_to_cpu(*(const __le32 *)(entry + 12)) << 16);
+		efx_tc_counter_update(efx, EFX_TC_COUNTER_TYPE_AR, counter_idx,
+				      packet_count, byte_count, mark);
+	}
+}
+
+#define TCV2_HDR_PTR(pkt, field)						\
+	((void)BUILD_BUG_ON_ZERO(ERF_SC_PACKETISER_HEADER_##field##_LBN & 7),	\
+	 (pkt) + ERF_SC_PACKETISER_HEADER_##field##_LBN / 8)
+#define TCV2_HDR_BYTE(pkt, field)						\
+	((void)BUILD_BUG_ON_ZERO(ERF_SC_PACKETISER_HEADER_##field##_WIDTH != 8),\
+	 *TCV2_HDR_PTR(pkt, field))
+#define TCV2_HDR_WORD(pkt, field)						\
+	((void)BUILD_BUG_ON_ZERO(ERF_SC_PACKETISER_HEADER_##field##_WIDTH != 16),\
+	 (void)BUILD_BUG_ON_ZERO(ERF_SC_PACKETISER_HEADER_##field##_LBN & 15),	\
+	 *(__force const __le16 *)TCV2_HDR_PTR(pkt, field))
+#define TCV2_PKT_PTR(pkt, poff, i, field)					\
+	((void)BUILD_BUG_ON_ZERO(ERF_SC_PACKETISER_PAYLOAD_##field##_LBN & 7),	\
+	 (pkt) + ERF_SC_PACKETISER_PAYLOAD_##field##_LBN/8 + poff +		\
+	 i * ER_RX_SL_PACKETISER_PAYLOAD_WORD_SIZE)
+
+/* Read a little-endian 48-bit field with 16-bit alignment */
+static u64 efx_tc_read48(const __le16 *field)
+{
+	u64 out = 0;
+	int i;
+
+	for (i = 0; i < 3; i++)
+		out |= (u64)le16_to_cpu(field[i]) << (i * 16);
+	return out;
+}
+
+static enum efx_tc_counter_type efx_tc_rx_version_2(struct efx_nic *efx,
+						    const u8 *data, u32 mark)
+{
+	u8 payload_offset, header_offset, ident;
+	enum efx_tc_counter_type type;
+	u16 n_counters, i;
+
+	ident = TCV2_HDR_BYTE(data, IDENTIFIER);
+	switch (ident) {
+	case ERF_SC_PACKETISER_HEADER_IDENTIFIER_AR:
+		type = EFX_TC_COUNTER_TYPE_AR;
+		break;
+	case ERF_SC_PACKETISER_HEADER_IDENTIFIER_CT:
+		type = EFX_TC_COUNTER_TYPE_CT;
+		break;
+	case ERF_SC_PACKETISER_HEADER_IDENTIFIER_OR:
+		type = EFX_TC_COUNTER_TYPE_OR;
+		break;
+	default:
+		if (net_ratelimit())
+			netif_err(efx, drv, efx->net_dev,
+				  "ignored v2 MAE counter packet (bad identifier %u"
+				  "), counters may be inaccurate\n", ident);
+		return EFX_TC_COUNTER_TYPE_MAX;
+	}
+	header_offset = TCV2_HDR_BYTE(data, HEADER_OFFSET);
+	/* mae_counter_format.h implies that this offset is fixed, since it
+	 * carries on with SOP-based LBNs for the fields in this header
+	 */
+	if (header_offset != ERF_SC_PACKETISER_HEADER_HEADER_OFFSET_DEFAULT) {
+		if (net_ratelimit())
+			netif_err(efx, drv, efx->net_dev,
+				  "choked on v2 MAE counter packet (bad header_offset %u"
+				  "), counters may be inaccurate\n", header_offset);
+		return EFX_TC_COUNTER_TYPE_MAX;
+	}
+	payload_offset = TCV2_HDR_BYTE(data, PAYLOAD_OFFSET);
+	n_counters = le16_to_cpu(TCV2_HDR_WORD(data, COUNT));
+
+	for (i = 0; i < n_counters; i++) {
+		const void *counter_idx_p, *packet_count_p, *byte_count_p;
+		u64 packet_count, byte_count;
+		u32 counter_idx;
+
+		/* 24-bit field with 32-bit alignment */
+		counter_idx_p = TCV2_PKT_PTR(data, payload_offset, i, COUNTER_INDEX);
+		BUILD_BUG_ON(ERF_SC_PACKETISER_PAYLOAD_COUNTER_INDEX_WIDTH != 24);
+		BUILD_BUG_ON(ERF_SC_PACKETISER_PAYLOAD_COUNTER_INDEX_LBN & 31);
+		counter_idx = le32_to_cpu(*(const __le32 *)counter_idx_p) & 0xffffff;
+		/* 48-bit field with 16-bit alignment */
+		packet_count_p = TCV2_PKT_PTR(data, payload_offset, i, PACKET_COUNT);
+		BUILD_BUG_ON(ERF_SC_PACKETISER_PAYLOAD_PACKET_COUNT_WIDTH != 48);
+		BUILD_BUG_ON(ERF_SC_PACKETISER_PAYLOAD_PACKET_COUNT_LBN & 15);
+		packet_count = efx_tc_read48((const __le16 *)packet_count_p);
+		/* 48-bit field with 16-bit alignment */
+		byte_count_p = TCV2_PKT_PTR(data, payload_offset, i, BYTE_COUNT);
+		BUILD_BUG_ON(ERF_SC_PACKETISER_PAYLOAD_BYTE_COUNT_WIDTH != 48);
+		BUILD_BUG_ON(ERF_SC_PACKETISER_PAYLOAD_BYTE_COUNT_LBN & 15);
+		byte_count = efx_tc_read48((const __le16 *)byte_count_p);
+
+		if (type == EFX_TC_COUNTER_TYPE_CT) {
+			/* CT counters are 1-bit saturating counters to update
+			 * the lastuse time in CT stats. A received CT counter
+			 * should have packet counter to 0 and only LSB bit on
+			 * in byte counter.
+			 */
+			if (packet_count || byte_count != 1)
+				netdev_warn_once(efx->net_dev,
+						 "CT counter with inconsistent state (%llu, %llu)\n",
+						 packet_count, byte_count);
+			/* Do not increment the driver's byte counter */
+			byte_count = 0;
+		}
+
+		efx_tc_counter_update(efx, type, counter_idx, packet_count,
+				      byte_count, mark);
+	}
+	return type;
+}
+
+/* We always swallow the packet, whether successful or not, since it's not
+ * a network packet and shouldn't ever be forwarded to the stack.
+ * @mark is the generation count for counter allocations.
+ */
+static bool efx_tc_rx(struct efx_rx_queue *rx_queue, u32 mark)
+{
+	struct efx_channel *channel = efx_rx_queue_channel(rx_queue);
+	struct efx_rx_buffer *rx_buf = efx_rx_buffer(rx_queue,
+						     channel->rx_pkt_index);
+	const u8 *data = efx_rx_buf_va(rx_buf);
+	struct efx_nic *efx = rx_queue->efx;
+	enum efx_tc_counter_type type;
+	u8 version;
+
+	/* version is always first byte of packet */
+	version = *data;
+	switch (version) {
+	case 1:
+		type = EFX_TC_COUNTER_TYPE_AR;
+		efx_tc_rx_version_1(efx, data, mark);
+		break;
+	case ERF_SC_PACKETISER_HEADER_VERSION_VALUE: // 2
+		type = efx_tc_rx_version_2(efx, data, mark);
+		break;
+	default:
+		if (net_ratelimit())
+			netif_err(efx, drv, efx->net_dev,
+				  "choked on MAE counter packet (bad version %u"
+				  "); counters may be inaccurate\n",
+				  version);
+		goto out;
+	}
+
+	/* Update seen_gen unconditionally, to avoid a missed wakeup if
+	 * we race with efx_mae_stop_counters().
+	 */
+	efx->tc->seen_gen[type] = mark;
+	if (efx->tc->flush_counters &&
+	    (s32)(efx->tc->flush_gen[type] - mark) <= 0)
+		wake_up(&efx->tc->flush_wq);
+out:
+	efx_free_rx_buffers(rx_queue, rx_buf, 1);
+	channel->rx_pkt_n_frags = 0;
+	return true;
+}
+
+const struct efx_channel_type efx_tc_channel_type = {
+	.handle_no_channel	= efx_tc_handle_no_channel,
+	.pre_probe		= efx_tc_probe_channel,
+	.start			= efx_tc_start_channel,
+	.stop			= efx_tc_stop_channel,
+	.post_remove		= efx_tc_remove_channel,
+	.get_name		= efx_tc_get_channel_name,
+	.receive_raw		= efx_tc_rx,
+	.keep_eventq		= true,
+};

drivers/net/ethernet/sfc/tc_counters.h

+/* SPDX-License-Identifier: GPL-2.0-only */
+/****************************************************************************
+ * Driver for Solarflare network controllers and boards
+ * Copyright 2022 Advanced Micro Devices, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation, incorporated herein by reference.
+ */
+
+#ifndef EFX_TC_COUNTERS_H
+#define EFX_TC_COUNTERS_H
+#include <linux/refcount.h>
+#include "net_driver.h"
+
+#include "mcdi_pcol.h" /* for MAE_COUNTER_TYPE_* */
+
+enum efx_tc_counter_type {
+	EFX_TC_COUNTER_TYPE_AR = MAE_COUNTER_TYPE_AR,
+	EFX_TC_COUNTER_TYPE_CT = MAE_COUNTER_TYPE_CT,
+	EFX_TC_COUNTER_TYPE_OR = MAE_COUNTER_TYPE_OR,
+	EFX_TC_COUNTER_TYPE_MAX
+};
+
+struct efx_tc_counter {
+	u32 fw_id; /* index in firmware counter table */
+	enum efx_tc_counter_type type;
+	struct rhash_head linkage; /* efx->tc->counter_ht */
+	spinlock_t lock; /* Serialises updates to counter values */
+	u32 gen; /* Generation count at which this counter is current */
+	u64 packets, bytes;
+	u64 old_packets, old_bytes; /* Values last time passed to userspace */
+	/* jiffies of the last time we saw packets increase */
+	unsigned long touched;
+};
+
+struct efx_tc_counter_index {
+	unsigned long cookie;
+	struct rhash_head linkage; /* efx->tc->counter_id_ht */
+	refcount_t ref;
+	struct efx_tc_counter *cnt;
+};
+
+/* create/uncreate/teardown hashtables */
+int efx_tc_init_counters(struct efx_nic *efx);
+void efx_tc_destroy_counters(struct efx_nic *efx);
+void efx_tc_fini_counters(struct efx_nic *efx);
+
+struct efx_tc_counter_index *efx_tc_flower_get_counter_index(
+				struct efx_nic *efx, unsigned long cookie,
+				enum efx_tc_counter_type type);
+void efx_tc_flower_put_counter_index(struct efx_nic *efx,
+				     struct efx_tc_counter_index *ctr);
+struct efx_tc_counter_index *efx_tc_flower_find_counter_index(
+				struct efx_nic *efx, unsigned long cookie);
+
+extern const struct efx_channel_type efx_tc_channel_type;
+
+#endif /* EFX_TC_COUNTERS_H */