qca-wifi: Move protocol and flow tagging code out of common host components

Move protocol and flow tagging code out of common host driver components

CRs-Fixed: 2501573

Change-Id: Ia3c9d036216c2fa914f6649a92abe8ca45578274

dp/wifi3.0/dp_rx_fst.c

@@ -0,0 +1,723 @@
+/*
+ * Copyright (c) 2019 The Linux Foundation. All rights reserved.
+ *
+ * Permission to use, copy, modify, and/or distribute this software for
+ * any purpose with or without fee is hereby granted, provided that the
+ * above copyright notice and this permission notice appear in all
+ * copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
+ * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
+ * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
+ * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
+ * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
+ * PERFORMANCE OF THIS SOFTWARE.
+ */
+
+#ifdef WLAN_SUPPORT_RX_FLOW_TAG
+#include "dp_types.h"
+#include "qdf_mem.h"
+#include "qdf_nbuf.h"
+#include "cfg_dp.h"
+#include "wlan_cfg.h"
+#include "dp_types.h"
+#include "hal_rx_flow.h"
+#include "dp_htt.h"
+
+/**
+ * In Hawkeye, a hardware bug disallows SW to only clear a single flow entry
+ * when added/deleted by upper layer. Workaround is to clear entire cache,
+ * which can have a performance impact. Flow additions/deletions
+ * are bundled together over 100ms to save HW cycles if upper layer
+ * adds/deletes multiple flows together. Use a longer timeout during setup
+ * stage since no flows are anticipated at this time.
+ */
+#define HW_RX_FSE_CACHE_INVALIDATE_BUNDLE_PERIOD_MS (100)
+#define HW_RX_FSE_CACHE_INVALIDATE_DELAYED_FST_SETUP_MS (5000)
+
+/**
+ * dp_rx_flow_get_fse() - Obtain flow search entry from flow hash
+ * @fst: Rx FST Handle
+ * @flow_hash: Computed hash value of flow
+ *
+ * Return: Handle to flow search table entry
+ */
+static inline struct dp_rx_fse *
+dp_rx_flow_get_fse(struct dp_rx_fst *fst, uint32_t flow_hash)
+{
+	struct dp_rx_fse *fse;
+	uint32_t idx = hal_rx_get_hal_hash(fst->hal_rx_fst, flow_hash);
+
+	fse = (struct dp_rx_fse *)((uint8_t *)fst->base + (idx *
+					sizeof(struct dp_rx_fse)));
+
+	return fse;
+}
+
+/**
+ * dp_rx_flow_dump_flow_entry() - Print flow search entry from 5-tuple
+ * @fst: Rx FST Handle
+ * @flow_info: Flow 5-tuple
+ *
+ * Return: None
+ */
+void dp_rx_flow_dump_flow_entry(struct dp_rx_fst *fst,
+				struct cdp_rx_flow_info *flow_info)
+{
+	dp_info("Dest IP address %x:%x:%x:%x",
+		flow_info->flow_tuple_info.dest_ip_127_96,
+		flow_info->flow_tuple_info.dest_ip_95_64,
+		flow_info->flow_tuple_info.dest_ip_63_32,
+		flow_info->flow_tuple_info.dest_ip_31_0);
+	dp_info("Source IP address %x:%x:%x:%x",
+		flow_info->flow_tuple_info.src_ip_127_96,
+		flow_info->flow_tuple_info.src_ip_95_64,
+		flow_info->flow_tuple_info.src_ip_63_32,
+		flow_info->flow_tuple_info.src_ip_31_0);
+	dp_info("Dest port %u, Src Port %u, Protocol %u",
+		flow_info->flow_tuple_info.dest_port,
+		flow_info->flow_tuple_info.src_port,
+		flow_info->flow_tuple_info.l4_protocol);
+}
+
+/**
+ * dp_rx_flow_compute_flow_hash() - Print flow search entry from 5-tuple
+ * @fst: Rx FST Handle
+ * @rx_flow_info: DP Rx Flow 5-tuple programmed by upper layer
+ * @flow: HAL (HW) flow entry
+ *
+ * Return: Computed Toeplitz hash
+ */
+uint32_t dp_rx_flow_compute_flow_hash(struct dp_rx_fst *fst,
+				      struct cdp_rx_flow_info *rx_flow_info,
+				      struct hal_rx_flow *flow)
+{
+	flow->tuple_info.dest_ip_127_96 =
+			rx_flow_info->flow_tuple_info.dest_ip_127_96;
+	flow->tuple_info.dest_ip_95_64 =
+			rx_flow_info->flow_tuple_info.dest_ip_95_64;
+	flow->tuple_info.dest_ip_63_32 =
+			rx_flow_info->flow_tuple_info.dest_ip_63_32;
+	flow->tuple_info.dest_ip_31_0 =
+			rx_flow_info->flow_tuple_info.dest_ip_31_0;
+	flow->tuple_info.src_ip_127_96 =
+			rx_flow_info->flow_tuple_info.src_ip_127_96;
+	flow->tuple_info.src_ip_95_64 =
+			rx_flow_info->flow_tuple_info.src_ip_95_64;
+	flow->tuple_info.src_ip_63_32 =
+			rx_flow_info->flow_tuple_info.src_ip_63_32;
+	flow->tuple_info.src_ip_31_0 =
+			rx_flow_info->flow_tuple_info.src_ip_31_0;
+	flow->tuple_info.dest_port =
+			rx_flow_info->flow_tuple_info.dest_port;
+	flow->tuple_info.src_port =
+			rx_flow_info->flow_tuple_info.src_port;
+	flow->tuple_info.l4_protocol =
+			rx_flow_info->flow_tuple_info.l4_protocol;
+
+	return hal_flow_toeplitz_hash(fst->hal_rx_fst, flow);
+}
+
+/**
+ * dp_rx_flow_alloc_entry() - Create DP and HAL flow entries in FST
+ * @fst: Rx FST Handle
+ * @rx_flow_info: DP Rx Flow 5-tuple to be added to DP FST
+ * @flow: HAL (HW) flow entry that is created
+ *
+ * Return: Computed Toeplitz hash
+ */
+struct dp_rx_fse *dp_rx_flow_alloc_entry(struct dp_rx_fst *fst,
+					 struct cdp_rx_flow_info *rx_flow_info,
+					 struct hal_rx_flow *flow)
+{
+	struct dp_rx_fse *fse = NULL;
+	uint32_t flow_hash;
+	uint32_t flow_idx;
+	QDF_STATUS status;
+
+	flow_hash = dp_rx_flow_compute_flow_hash(fst, rx_flow_info, flow);
+
+	status = hal_rx_insert_flow_entry(fst->hal_rx_fst,
+					  flow_hash,
+					  &rx_flow_info->flow_tuple_info,
+					  &flow_idx);
+	if (status != QDF_STATUS_SUCCESS) {
+		dp_err("Add entry failed with status %d for tuple with hash %u",
+		       status, flow_hash);
+		return NULL;
+	}
+
+	fse = dp_rx_flow_get_fse(fst, flow_idx);
+	fse->is_ipv4_addr_entry = rx_flow_info->is_addr_ipv4;
+	fse->flow_hash = flow_hash;
+	fse->flow_id = flow_idx;
+	fse->stats.msdu_count = 0;
+	fse->is_valid = true;
+
+	return fse;
+}
+
+/**
+ * dp_rx_flow_find_entry_by_tuple() - Find the DP FSE matching a given 5-tuple
+ * @fst: Rx FST Handle
+ * @rx_flow_info: DP Rx Flow 5-tuple
+ * @flow: Pointer to the  HAL (HW) flow entry
+ *
+ * Return: Pointer to the DP FSE entry
+ */
+struct dp_rx_fse *
+dp_rx_flow_find_entry_by_tuple(struct dp_rx_fst *fst,
+			       struct cdp_rx_flow_info *rx_flow_info,
+			       struct hal_rx_flow *flow)
+{
+	uint32_t flow_hash;
+	uint32_t flow_idx;
+	QDF_STATUS status;
+
+	flow_hash = dp_rx_flow_compute_flow_hash(fst, rx_flow_info, flow);
+
+	status = hal_rx_find_flow_from_tuple(fst->hal_rx_fst,
+					     flow_hash,
+					     &rx_flow_info->flow_tuple_info,
+					     &flow_idx);
+	if (status != QDF_STATUS_SUCCESS) {
+		dp_err("Could not find tuple with hash %u", flow_hash);
+		dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+		return NULL;
+	}
+	return dp_rx_flow_get_fse(fst, flow_idx);
+}
+
+/**
+ * dp_rx_flow_find_entry_by_flowid() - Find DP FSE matching a given flow index
+ * @fst: Rx FST Handle
+ * @flow_id: Flow index of the requested flow
+ *
+ * Return: Pointer to the DP FSE entry
+ */
+struct dp_rx_fse *
+dp_rx_flow_find_entry_by_flowid(struct dp_rx_fst *fst,
+				uint32_t flow_id)
+{
+	struct dp_rx_fse *fse = NULL;
+
+	fse = dp_rx_flow_get_fse(fst, flow_id);
+	if (!fse->is_valid)
+		return NULL;
+
+	dp_info("flow_idx= %d, flow_addr = %pK", flow_id, fse);
+	qdf_assert_always(fse->flow_id == flow_id);
+
+	return fse;
+}
+
+/**
+ * dp_rx_flow_send_htt_operation_cmd() - Send HTT FSE command to FW for flow
+ *					 addition/removal
+ * @pdev: Pdev instance
+ * @op: Add/delete operation
+ * @info: DP Flow parameters of the flow added/deleted
+ *
+ * Return: Success on sending HTT command to FW, error on failure
+ */
+QDF_STATUS dp_rx_flow_send_htt_operation_cmd(struct dp_pdev *pdev,
+					     enum dp_htt_flow_fst_operation op,
+					     struct cdp_rx_flow_info *info)
+{
+	struct dp_htt_rx_flow_fst_operation fst_op;
+	struct wlan_cfg_dp_soc_ctxt *cfg = pdev->soc->wlan_cfg_ctx;
+
+	qdf_mem_set(&fst_op, 0, sizeof(struct dp_htt_rx_flow_fst_operation));
+
+	if (qdf_unlikely(wlan_cfg_is_rx_flow_search_table_per_pdev(cfg))) {
+		/* Firmware pdev ID starts from 1 */
+		fst_op.pdev_id = DP_SW2HW_MACID(pdev->pdev_id);
+	} else {
+		fst_op.pdev_id = 0;
+	}
+
+	fst_op.op_code = op;
+	fst_op.rx_flow = info;
+
+	return dp_htt_rx_flow_fse_operation(pdev, &fst_op);
+}
+
+/**
+ * dp_rx_flow_add_entry() - Add a flow entry to flow search table
+ * @pdev: DP pdev instance
+ * @rx_flow_info: DP flow paramaters
+ *
+ * Return: Success when flow is added, no-memory or already exists on error
+ */
+QDF_STATUS dp_rx_flow_add_entry(struct dp_pdev *pdev,
+				struct cdp_rx_flow_info *rx_flow_info)
+{
+	struct hal_rx_flow flow = { 0 };
+	struct dp_rx_fse *fse;
+	struct dp_soc *soc = pdev->soc;
+	struct dp_rx_fst *fst;
+
+	fst = pdev->rx_fst;
+
+	/* Initialize unused bits in IPv6 address for IPv4 address */
+	if (rx_flow_info->is_addr_ipv4) {
+		rx_flow_info->flow_tuple_info.dest_ip_63_32 = 0;
+		rx_flow_info->flow_tuple_info.dest_ip_95_64 = 0;
+		rx_flow_info->flow_tuple_info.dest_ip_127_96 =
+			HAL_IP_DA_SA_PREFIX_IPV4_COMPATIBLE_IPV6;
+
+		rx_flow_info->flow_tuple_info.src_ip_63_32 = 0;
+		rx_flow_info->flow_tuple_info.src_ip_95_64 = 0;
+		rx_flow_info->flow_tuple_info.src_ip_127_96 =
+			HAL_IP_DA_SA_PREFIX_IPV4_COMPATIBLE_IPV6;
+	}
+
+	/* Allocate entry in DP FST */
+	fse = dp_rx_flow_alloc_entry(fst, rx_flow_info, &flow);
+	if (NULL == fse) {
+		dp_err("RX FSE alloc failed");
+		dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+		return QDF_STATUS_E_NOMEM;
+	}
+	dp_info("flow_addr = %pK, flow_id = %u, valid = %d, v4 = %d\n",
+		fse, fse->flow_id, fse->is_valid, fse->is_ipv4_addr_entry);
+
+	/* Initialize other parameters for HW flow & populate HW FSE entry */
+	flow.reo_destination_indication = (fse->flow_hash &
+				HAL_REO_DEST_IND_HASH_MASK);
+
+	/**
+	 * Reo destination of each flow is mapped to match the same used
+	 * by RX Hash algorithm. If RX Hash is disabled, then the REO
+	 * destination below is directly got from pdev, rather than using
+	 * dp_peer_setup_get_reo_hash since we do not have vdev handle here.
+	 */
+	if (wlan_cfg_is_rx_hash_enabled(soc->wlan_cfg_ctx)) {
+		flow.reo_destination_indication |=
+			HAL_REO_DEST_IND_START_OFFSET;
+	} else {
+		flow.reo_destination_indication = pdev->reo_dest;
+	}
+
+	flow.reo_destination_handler = HAL_RX_FSE_REO_DEST_FT;
+	flow.fse_metadata = rx_flow_info->fse_metadata;
+	fse->hal_rx_fse = hal_rx_flow_setup_fse(fst->hal_rx_fst,
+						fse->flow_id, &flow);
+	if (qdf_unlikely(!fse->hal_rx_fse)) {
+		dp_err("Unable to alloc FSE entry");
+		dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+		/* Free up the FSE entry as returning failure */
+		fse->is_valid = false;
+		return QDF_STATUS_E_EXISTS;
+	}
+
+	/* Increment number of valid entries in table */
+	fst->num_entries++;
+	dp_info("FST num_entries = %d, reo_dest_ind = %d, reo_dest_hand = %u",
+		fst->num_entries, flow.reo_destination_indication,
+		flow.reo_destination_handler);
+
+	if (soc->is_rx_fse_full_cache_invalidate_war_enabled) {
+		qdf_atomic_set(&fst->is_cache_update_pending, 1);
+	} else {
+		QDF_STATUS status;
+		/**
+		 * Send HTT cache invalidation command to firmware to
+		 * reflect the added flow
+		 */
+		status = dp_rx_flow_send_htt_operation_cmd(
+					pdev,
+					DP_HTT_FST_CACHE_INVALIDATE_ENTRY,
+					rx_flow_info);
+
+		if (QDF_STATUS_SUCCESS != status) {
+			dp_err("Send cache invalidate entry to fw failed: %u",
+			       status);
+			dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+			/* Free DP FSE and HAL FSE */
+			hal_rx_flow_delete_entry(fst->hal_rx_fst,
+						 fse->hal_rx_fse);
+			fse->is_valid = false;
+			return status;
+		}
+	}
+
+	return QDF_STATUS_SUCCESS;
+}
+
+/**
+ * dp_rx_flow_delete_entry() - Delete a flow entry from flow search table
+ * @pdev: pdev handle
+ * @rx_flow_info: DP flow parameters
+ *
+ * Return: Success when flow is deleted, error on failure
+ */
+QDF_STATUS dp_rx_flow_delete_entry(struct dp_pdev *pdev,
+				   struct cdp_rx_flow_info *rx_flow_info)
+{
+	struct hal_rx_flow flow = { 0 };
+	struct dp_rx_fse *fse;
+	struct dp_soc *soc = pdev->soc;
+	struct dp_rx_fst *fst;
+	QDF_STATUS status;
+
+	fst = pdev->rx_fst;
+
+	/* Find the given flow entry DP FST */
+	fse = dp_rx_flow_find_entry_by_tuple(fst, rx_flow_info, &flow);
+	if (!fse) {
+		dp_err("RX flow delete entry failed");
+		dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+		return QDF_STATUS_E_INVAL;
+	}
+
+	/* Delete the FSE in HW FST */
+	status = hal_rx_flow_delete_entry(fst->hal_rx_fst, fse->hal_rx_fse);
+	qdf_assert_always(status == QDF_STATUS_SUCCESS);
+
+	/* Free the FSE in DP FST */
+	fse->is_valid = false;
+
+	/* Decrement number of valid entries in table */
+	fst->num_entries--;
+
+	if (soc->is_rx_fse_full_cache_invalidate_war_enabled) {
+		qdf_atomic_set(&fst->is_cache_update_pending, 1);
+	} else {
+		/**
+		 * Send HTT cache invalidation command to firmware
+		 * to reflect the deleted flow
+		 */
+		status = dp_rx_flow_send_htt_operation_cmd(
+					pdev,
+					DP_HTT_FST_CACHE_INVALIDATE_ENTRY,
+					rx_flow_info);
+
+		if (QDF_STATUS_SUCCESS != status) {
+			dp_err("Send cache invalidate entry to fw failed: %u",
+			       status);
+			dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+			/* Do not add entry back in DP FSE and HAL FSE */
+			return status;
+		}
+	}
+
+	return QDF_STATUS_SUCCESS;
+}
+
+/* dp_rx_flow_update_fse_stats() - Update a flow's statistics
+ * @pdev: pdev handle
+ * @flow_id: flow index (truncated hash) in the Rx FST
+ *
+ * Return: Success when flow statistcs is updated, error on failure
+ */
+QDF_STATUS dp_rx_flow_update_fse_stats(struct dp_pdev *pdev, uint32_t flow_id)
+{
+	struct dp_rx_fse *fse;
+
+	fse = dp_rx_flow_find_entry_by_flowid(pdev->rx_fst, flow_id);
+
+	if (NULL == fse) {
+		dp_err("Flow not found, flow ID %u", flow_id);
+		return QDF_STATUS_E_NOENT;
+	}
+
+	fse->stats.msdu_count += 1;
+	return QDF_STATUS_SUCCESS;
+}
+
+/**
+ * dp_rx_flow_get_fse_stats() - Fetch a flow's stats based on DP flow parameter
+ * @pdev: pdev handle
+ * @rx_flow_info: Pointer to the DP flow struct of the requested flow
+ * @stats: Matching flow's stats returned to caller
+ *
+ * Return: Success when flow statistcs is updated, error on failure
+ */
+QDF_STATUS dp_rx_flow_get_fse_stats(struct dp_pdev *pdev,
+				    struct cdp_rx_flow_info *rx_flow_info,
+				    struct cdp_flow_stats *stats)
+{
+	struct dp_rx_fst *fst;
+	struct dp_rx_fse *fse;
+	struct hal_rx_flow flow;
+
+	fst = pdev->rx_fst;
+
+	/* Find the given flow entry DP FST */
+	fse = dp_rx_flow_find_entry_by_tuple(fst, rx_flow_info, &flow);
+	if (!fse) {
+		dp_err("RX flow entry search failed");
+		dp_rx_flow_dump_flow_entry(fst, rx_flow_info);
+		return QDF_STATUS_E_INVAL;
+	}
+
+	stats->msdu_count = fse->stats.msdu_count;
+	return QDF_STATUS_SUCCESS;
+}
+
+/**
+ * dp_rx_flow_cache_invalidate_timer_handler() - Timer handler used for bundling
+ * flows before invalidating entire cache
+ * @ctx: Pdev handle
+ *
+ * Return: None
+ */
+void dp_rx_flow_cache_invalidate_timer_handler(void *ctx)
+{
+	struct dp_pdev *pdev = (struct dp_pdev *)ctx;
+	struct dp_rx_fst *fst;
+	bool is_update_pending;
+	QDF_STATUS status;
+
+	fst = pdev->rx_fst;
+
+	qdf_assert_always(fst);
+	is_update_pending = qdf_atomic_read(&fst->is_cache_update_pending);
+	qdf_atomic_set(&fst->is_cache_update_pending, 0);
+
+	if (is_update_pending) {
+		/* Send full cache invalidate command to firmware */
+		status = dp_rx_flow_send_htt_operation_cmd(
+				pdev,
+				DP_HTT_FST_CACHE_INVALIDATE_FULL,
+				NULL);
+
+		if (QDF_STATUS_SUCCESS != status)
+			dp_err("Send full cache inv to fw failed: %u", status);
+	}
+
+	qdf_timer_start(&fst->cache_invalidate_timer,
+			HW_RX_FSE_CACHE_INVALIDATE_BUNDLE_PERIOD_MS);
+}
+
+/**
+ * dp_rx_fst_attach() - Initialize Rx FST and setup necessary parameters
+ * @soc: SoC handle
+ * @pdev: Pdev handle
+ *
+ * Return: Handle to flow search table entry
+ */
+QDF_STATUS dp_rx_fst_attach(struct dp_soc *soc, struct dp_pdev *pdev)
+{
+	struct dp_rx_fst *fst;
+	uint8_t *hash_key;
+	struct wlan_cfg_dp_soc_ctxt *cfg = soc->wlan_cfg_ctx;
+	bool is_rx_flow_search_table_per_pdev =
+		wlan_cfg_is_rx_flow_search_table_per_pdev(cfg);
+
+	if (qdf_unlikely(!wlan_cfg_is_rx_flow_tag_enabled(cfg))) {
+		dp_err("RX Flow tag feature disabled");
+		return QDF_STATUS_E_NOSUPPORT;
+	}
+
+	if (!wlan_psoc_nif_fw_ext_cap_get((void *)pdev->ctrl_pdev,
+					  WLAN_SOC_CEXT_RX_FSE_SUPPORT)) {
+		QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+			  "rx fse disabled in FW\n");
+		wlan_cfg_set_rx_flow_tag_enabled(cfg, false);
+		return QDF_STATUS_E_NOSUPPORT;
+	}
+	/**
+	 * Func. is called for every pdev. If FST is per SOC, then return
+	 * if it was already called once.
+	 */
+	if (!is_rx_flow_search_table_per_pdev && soc->rx_fst) {
+		pdev->rx_fst = soc->rx_fst;
+		QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+			  "RX FST for SoC is already initialized");
+		return QDF_STATUS_SUCCESS;
+	}
+
+	/**
+	 * Func. is called for this pdev already. This is an error.
+	 * Return failure
+	 */
+	if (is_rx_flow_search_table_per_pdev && pdev->rx_fst) {
+		QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+			  "RX FST for PDEV %u is already initialized",
+			  pdev->pdev_id);
+		return QDF_STATUS_E_EXISTS;
+	}
+
+	fst = qdf_mem_malloc(sizeof(struct dp_rx_fst));
+	if (!fst) {
+		QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+			  "RX FST allocation failed\n");
+		return QDF_STATUS_E_NOMEM;
+	}
+
+	qdf_mem_set(fst, 0, sizeof(struct dp_rx_fst));
+
+	fst->max_skid_length = wlan_cfg_rx_fst_get_max_search(cfg);
+	fst->max_entries = wlan_cfg_get_rx_flow_search_table_size(cfg);
+	hash_key = wlan_cfg_rx_fst_get_hash_key(cfg);
+
+	if (!(fst->max_entries &&
+	      (!(fst->max_entries & (fst->max_entries - 1))))) {
+		uint32_t next_power_of_2 = fst->max_entries - 1;
+
+		next_power_of_2 |= (next_power_of_2 >> 1);
+		next_power_of_2 |= (next_power_of_2 >> 2);
+		next_power_of_2 |= (next_power_of_2 >> 4);
+		next_power_of_2 |= (next_power_of_2 >> 8);
+		next_power_of_2 |= (next_power_of_2 >> 16);
+		next_power_of_2++;
+		if (next_power_of_2 > WLAN_CFG_RX_FLOW_SEARCH_TABLE_SIZE_MAX)
+			next_power_of_2 =
+				 WLAN_CFG_RX_FLOW_SEARCH_TABLE_SIZE_MAX;
+		dp_info("Num entries in cfg is not a ^2:%u, using next ^2:%u",
+			fst->max_entries, next_power_of_2);
+		fst->max_entries = next_power_of_2;
+	}
+	fst->hash_mask = fst->max_entries - 1;
+	fst->num_entries = 0;
+
+	fst->base = (uint8_t *) qdf_mem_malloc(sizeof(struct dp_rx_fse) *
+					       fst->max_entries);
+
+	if (!fst->base) {
+		QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+			  "Rx fst->base allocation failed, #entries:%d\n",
+			  fst->max_entries);
+
+		qdf_mem_free(fst);
+		return QDF_STATUS_E_NOMEM;
+	}
+
+	qdf_mem_set((uint8_t *)fst->base, 0,
+		    (sizeof(struct dp_rx_fse) * fst->max_entries));
+
+	fst->hal_rx_fst = hal_rx_fst_attach(
+				soc->osdev,
+				&fst->hal_rx_fst_base_paddr,
+				fst->max_entries,
+				fst->max_skid_length,
+				hash_key);
+
+	if (qdf_unlikely(!fst->hal_rx_fst)) {
+		QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+			  "Rx Hal fst allocation failed, #entries:%d\n",
+			  fst->max_entries);
+		qdf_mem_free(fst->base);
+		qdf_mem_free(fst);
+		return QDF_STATUS_E_NOMEM;
+	}
+	if (!is_rx_flow_search_table_per_pdev)
+		soc->rx_fst = fst;
+
+	pdev->rx_fst = fst;
+
+	if (soc->is_rx_fse_full_cache_invalidate_war_enabled) {
+		QDF_STATUS status;
+
+		status = qdf_timer_init(
+				soc->osdev,
+				&fst->cache_invalidate_timer,
+				dp_rx_flow_cache_invalidate_timer_handler,
+				(void *)pdev,
+				QDF_TIMER_TYPE_SW);
+
+		qdf_assert_always(status == QDF_STATUS_SUCCESS);
+
+		/* Start the timer */
+		qdf_timer_start(
+			&fst->cache_invalidate_timer,
+			HW_RX_FSE_CACHE_INVALIDATE_DELAYED_FST_SETUP_MS);
+
+		qdf_atomic_set(&fst->is_cache_update_pending, false);
+	}
+
+	QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_INFO,
+		  "Rx FST attach successful, #entries:%d\n",
+		  fst->max_entries);
+	return QDF_STATUS_SUCCESS;
+}
+
+/**
+ * dp_rx_fst_detach() - De-initialize Rx FST
+ * @soc: SoC handle
+ * @pdev: Pdev handle
+ *
+ * Return: None
+ */
+void dp_rx_fst_detach(struct dp_soc *soc, struct dp_pdev *pdev)
+{
+	struct dp_rx_fst *dp_fst;
+	struct wlan_cfg_dp_soc_ctxt *cfg = soc->wlan_cfg_ctx;
+
+	if (qdf_unlikely(wlan_cfg_is_rx_flow_search_table_per_pdev(cfg))) {
+		dp_fst = pdev->rx_fst;
+		pdev->rx_fst = NULL;
+	} else {
+		dp_fst = soc->rx_fst;
+		soc->rx_fst = NULL;
+	}
+
+	if (qdf_likely(dp_fst)) {
+		hal_rx_fst_detach(dp_fst->hal_rx_fst, soc->osdev);
+		if (soc->is_rx_fse_full_cache_invalidate_war_enabled) {
+			qdf_timer_sync_cancel(&dp_fst->cache_invalidate_timer);
+			qdf_timer_stop(&dp_fst->cache_invalidate_timer);
+			qdf_timer_free(&dp_fst->cache_invalidate_timer);
+		}
+		qdf_mem_free(dp_fst->base);
+		qdf_mem_free(dp_fst);
+	}
+	QDF_TRACE(QDF_MODULE_ID_DP, QDF_TRACE_LEVEL_DEBUG,
+		  "Rx FST detached for pdev %u\n", pdev->pdev_id);
+}
+
+/**
+ * dp_rx_flow_send_fst_fw_setup() - Program FST parameters in FW/HW post-attach
+ * @soc: SoC handle
+ * @pdev: Pdev handle
+ *
+ * Return: Success when fst parameters are programmed in FW, error otherwise
+ */
+QDF_STATUS dp_rx_flow_send_fst_fw_setup(struct dp_soc *soc,
+					struct dp_pdev *pdev)
+{
+	struct dp_htt_rx_flow_fst_setup fst_setup;
+	struct dp_rx_fst *fst;
+	QDF_STATUS status;
+	struct wlan_cfg_dp_soc_ctxt *cfg = soc->wlan_cfg_ctx;
+
+	if (qdf_unlikely(!wlan_cfg_is_rx_flow_tag_enabled(cfg)))
+		return QDF_STATUS_SUCCESS;
+
+	qdf_mem_set(&fst_setup, 0, sizeof(struct dp_htt_rx_flow_fst_setup));
+
+	if (qdf_unlikely(wlan_cfg_is_rx_flow_search_table_per_pdev(cfg))) {
+		/* Firmware pdev ID starts from 1 */
+		fst_setup.pdev_id = DP_SW2HW_MACID(pdev->pdev_id);
+		fst = pdev->rx_fst;
+	} else {
+		fst_setup.pdev_id = 0;
+		fst = soc->rx_fst;
+	}
+	fst_setup.max_entries = fst->max_entries;
+	fst_setup.max_search = fst->max_skid_length;
+	fst_setup.base_addr_lo = (uint32_t)fst->hal_rx_fst_base_paddr;
+	fst_setup.base_addr_hi =
+		(uint32_t)((uint64_t)fst->hal_rx_fst_base_paddr >> 32);
+	fst_setup.ip_da_sa_prefix =
+		HAL_FST_IP_DA_SA_PFX_TYPE_IPV4_COMPATIBLE_IPV6;
+	fst_setup.hash_key =  wlan_cfg_rx_fst_get_hash_key(cfg);
+	fst_setup.hash_key_len = HAL_FST_HASH_KEY_SIZE_BYTES;
+
+	status = dp_htt_rx_flow_fst_setup(pdev, &fst_setup);
+	if (status == QDF_STATUS_SUCCESS) {
+		fst->fse_setup_done = true;
+		return status;
+	}
+	QDF_TRACE(QDF_MODULE_ID_ANY, QDF_TRACE_LEVEL_ERROR,
+		  "Failed to send Rx FSE Setup pdev%d status %d\n",
+		  pdev->pdev_id, status);
+	/* Free all the memory allocations and data structures */
+	dp_rx_fst_detach(pdev->soc, pdev);
+	return status;
+}
+#endif /* WLAN_SUPPORT_RX_FLOW_TAG */

dp/wifi3.0/dp_rx_tag.c

@@ -0,0 +1,520 @@
+/*
+ * Copyright (c) 2016-2019 The Linux Foundation. All rights reserved.
+ *
+ * Permission to use, copy, modify, and/or distribute this software for
+ * any purpose with or without fee is hereby granted, provided that the
+ * above copyright notice and this permission notice appear in all
+ * copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
+ * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
+ * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
+ * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
+ * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
+ * PERFORMANCE OF THIS SOFTWARE.
+ */
+
+#include "hal_hw_headers.h"
+#include "hal_api.h"
+#include "hal_rx.h"
+#include "qdf_trace.h"
+#include "dp_tx.h"
+#include "dp_peer.h"
+#include "dp_internal.h"
+#include "dp_rx_tag.h"
+
+#if defined(WLAN_SUPPORT_RX_TAG_STATISTICS) && \
+	defined(WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG)
+/**
+ * dp_rx_update_rx_protocol_tag_stats() - Increments the protocol tag stats
+ *                                        for the given protocol type
+ * @soc: core txrx main context
+ * @pdev: TXRX pdev context for which stats should be incremented
+ * @protocol_index: Protocol index for which the stats should be incremented
+ * @ring_index: REO ring number from which this tag was received.
+ *
+ * Since HKv2 is a SMP, two or more cores may simultaneously receive packets
+ * of same type, and hence attempt to increment counters for the same protocol
+ * type at the same time. This creates the possibility of missing stats.
+ *
+ * For example,  when two or more CPUs have each read the old tag value, V,
+ * for protocol type, P and each increment the value to V+1. Instead, the
+ * operations should have been  sequenced to achieve a final value of V+2.
+ *
+ * In order to avoid this scenario,  we can either use locks or store stats
+ * on a per-CPU basis. Since tagging happens in the core data path, locks
+ * are not preferred. Instead, we use a per-ring counter, since each CPU
+ * operates on a REO ring.
+ *
+ * Return: void
+ */
+void dp_rx_update_rx_protocol_tag_stats(struct dp_pdev *pdev,
+					uint16_t protocol_index,
+					uint16_t ring_index)
+{
+	if (ring_index >= MAX_REO_DEST_RINGS)
+		return;
+
+	pdev->reo_proto_tag_stats[ring_index][protocol_index].tag_ctr++;
+}
+#endif /* WLAN_SUPPORT_RX_TAG_STATISTICS */
+
+#if defined(WLAN_SUPPORT_RX_TAG_STATISTICS) && \
+	defined(WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG)
+/**
+ * dp_rx_update_rx_err_protocol_tag_stats() - Increments the protocol tag stats
+ *                                        for the given protocol type
+ *                                        received from exception ring
+ * @soc: core txrx main context
+ * @pdev: TXRX pdev context for which stats should be incremented
+ * @protocol_index: Protocol index for which the stats should be incremented
+ *
+ * In HKv2, all exception packets are received on Ring-0 (along with normal
+ * Rx). Hence tags are maintained separately for exception ring as well.
+ *
+ * Return: void
+ */
+void dp_rx_update_rx_err_protocol_tag_stats(struct dp_pdev *pdev,
+					    uint16_t protocol_index)
+{
+	pdev->rx_err_proto_tag_stats[protocol_index].tag_ctr++;
+}
+#endif /* WLAN_SUPPORT_RX_TAG_STATISTICS */
+
+/**
+ * dp_rx_update_protocol_tag() - Reads CCE metadata from the RX MSDU end TLV
+ *                              and set the corresponding tag in QDF packet
+ * @soc: core txrx main context
+ * @vdev: vdev on which the packet is received
+ * @nbuf: QDF pkt buffer on which the protocol tag should be set
+ * @rx_tlv_hdr: rBbase address where the RX TLVs starts
+ * @ring_index: REO ring number, not used for error & monitor ring
+ * @is_reo_exception: flag to indicate if rx from REO ring or exception ring
+ * @is_update_stats: flag to indicate whether to update stats or not
+ * Return: void
+ */
+#ifdef WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG
+void
+dp_rx_update_protocol_tag(struct dp_soc *soc, struct dp_vdev *vdev,
+			  qdf_nbuf_t nbuf, uint8_t *rx_tlv_hdr,
+			  uint16_t ring_index,
+			  bool is_reo_exception, bool is_update_stats)
+{
+	uint16_t cce_metadata = RX_PROTOCOL_TAG_START_OFFSET;
+	bool     cce_match = false;
+	struct   dp_pdev *pdev;
+	uint16_t protocol_tag = 0;
+
+	if (qdf_unlikely(!vdev))
+		return;
+
+	pdev = vdev->pdev;
+
+	if (qdf_likely(!pdev->is_rx_protocol_tagging_enabled))
+		return;
+
+	/*
+	 * In case of raw frames, rx_attention and rx_msdu_end tlv
+	 * may be stale or invalid. Do not tag such frames.
+	 * Default decap_type is set to ethernet for monitor vdev,
+	 * therefore, cannot check decap_type for monitor mode.
+	 * We will call this only for eth frames from dp_rx_mon_dest.c.
+	 */
+	if (qdf_likely(!(pdev->monitor_vdev && pdev->monitor_vdev == vdev) &&
+		       (vdev->rx_decap_type !=  htt_cmn_pkt_type_ethernet)))
+		return;
+
+	/*
+	 * Check whether HW has filled in the CCE metadata in
+	 * this packet, if not filled, just return
+	 */
+	if (qdf_likely(!hal_rx_msdu_cce_match_get(rx_tlv_hdr)))
+		return;
+
+	cce_match = true;
+	/* Get the cce_metadata from RX MSDU TLV */
+	cce_metadata = (hal_rx_msdu_cce_metadata_get(rx_tlv_hdr) &
+			RX_MSDU_END_16_CCE_METADATA_MASK);
+	/*
+	 * Received CCE metadata should be within the
+	 * valid limits
+	 */
+	qdf_assert_always((cce_metadata >= RX_PROTOCOL_TAG_START_OFFSET) &&
+			  (cce_metadata < (RX_PROTOCOL_TAG_START_OFFSET +
+			   RX_PROTOCOL_TAG_MAX)));
+
+	/*
+	 * The CCE metadata received is just the
+	 * packet_type + RX_PROTOCOL_TAG_START_OFFSET
+	 */
+	cce_metadata -= RX_PROTOCOL_TAG_START_OFFSET;
+
+	/*
+	 * Update the QDF packet with the user-specified
+	 * tag/metadata by looking up tag value for
+	 * received protocol type.
+	 */
+	protocol_tag = pdev->rx_proto_tag_map[cce_metadata].tag;
+	qdf_nbuf_set_rx_protocol_tag(nbuf, protocol_tag);
+	QDF_TRACE(QDF_MODULE_ID_DP, QDF_TRACE_LEVEL_INFO_LOW,
+		  "Seq:%u dcap:%u CCE Match:%u ProtoID:%u Tag:%u stats:%u",
+		  hal_rx_get_rx_sequence(rx_tlv_hdr),
+		  vdev->rx_decap_type, cce_match, cce_metadata,
+		  protocol_tag, is_update_stats);
+
+	if (qdf_likely(!is_update_stats))
+		return;
+
+	if (qdf_unlikely(is_reo_exception)) {
+		dp_rx_update_rx_err_protocol_tag_stats(pdev,
+						       cce_metadata);
+	} else {
+		dp_rx_update_rx_protocol_tag_stats(pdev,
+						   cce_metadata,
+						   ring_index);
+	}
+}
+#endif /* WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG */
+
+/**
+ * dp_rx_update_flow_tag() - Reads FSE metadata from the RX MSDU end TLV
+ *                           and set the corresponding tag in QDF packet
+ * @soc: core txrx main context
+ * @vdev: vdev on which the packet is received
+ * @nbuf: QDF pkt buffer on which the protocol tag should be set
+ * @rx_tlv_hdr: base address where the RX TLVs starts
+ * @is_update_stats: flag to indicate whether to update stats or not
+ *
+ * Return: void
+ */
+#ifdef WLAN_SUPPORT_RX_FLOW_TAG
+void
+dp_rx_update_flow_tag(struct dp_soc *soc, struct dp_vdev *vdev,
+		      qdf_nbuf_t nbuf, uint8_t *rx_tlv_hdr, bool update_stats)
+{
+	bool flow_idx_invalid, flow_idx_timeout;
+	uint32_t flow_idx, fse_metadata;
+	struct dp_pdev *pdev;
+
+	if (qdf_unlikely(!vdev))
+		return;
+
+	pdev = vdev->pdev;
+	if (qdf_likely(!wlan_cfg_is_rx_flow_tag_enabled(soc->wlan_cfg_ctx)))
+		return;
+
+	QDF_TRACE(QDF_MODULE_ID_DP, QDF_TRACE_LEVEL_INFO_LOW,
+		  "Seq:%u dcap:%u invalid:%u timeout:%u flow:%u tag:%u stat:%u",
+		  hal_rx_get_rx_sequence(rx_tlv_hdr),
+		  vdev->rx_decap_type,
+		  hal_rx_msdu_flow_idx_invalid(rx_tlv_hdr),
+		  hal_rx_msdu_flow_idx_timeout(rx_tlv_hdr),
+		  hal_rx_msdu_flow_idx_get(rx_tlv_hdr),
+		  hal_rx_msdu_fse_metadata_get(rx_tlv_hdr),
+		  update_stats);
+
+	/**
+	 * In case of raw frames, rx_msdu_end tlv may be stale or invalid.
+	 * Do not tag such frames in normal REO path.
+	 * Default decap_type is set to ethernet for monitor vdev currently,
+	 * therefore, we will not check decap_type for monitor mode.
+	 * We will call this only for eth frames from dp_rx_mon_dest.c.
+	 */
+	if (qdf_likely((vdev->rx_decap_type !=  htt_cmn_pkt_type_ethernet)))
+		return;
+
+	flow_idx_invalid = hal_rx_msdu_flow_idx_invalid(rx_tlv_hdr);
+	hal_rx_msdu_get_flow_params(rx_tlv_hdr, &flow_idx_invalid,
+				    &flow_idx_timeout, &flow_idx);
+	if (qdf_unlikely(flow_idx_invalid))
+		return;
+
+	if (qdf_unlikely(flow_idx_timeout))
+		return;
+
+	/**
+	 * Limit FSE metadata to 16 bit as we have allocated only
+	 * 16 bits for flow_tag field in skb->cb
+	 */
+	fse_metadata = hal_rx_msdu_fse_metadata_get(rx_tlv_hdr) & 0xFFFF;
+
+	/* update the skb->cb with the user-specified tag/metadata */
+	qdf_nbuf_set_rx_flow_tag(nbuf, fse_metadata);
+
+	QDF_TRACE(QDF_MODULE_ID_DP, QDF_TRACE_LEVEL_INFO_LOW,
+		  "Seq:%u dcap:%u invalid:%u timeout:%u flow:%u tag:%u stat:%u",
+		  hal_rx_get_rx_sequence(rx_tlv_hdr),
+		  vdev->rx_decap_type, flow_idx_invalid, flow_idx_timeout,
+		  flow_idx, fse_metadata, update_stats);
+
+	if (qdf_likely(update_stats))
+		dp_rx_update_rx_flow_tag_stats(pdev, flow_idx);
+}
+#endif /* WLAN_SUPPORT_RX_FLOW_TAG */
+
+#if defined(WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG) ||\
+	defined(WLAN_SUPPORT_RX_FLOW_TAG)
+/**
+ * dp_rx_mon_update_protocol_flow_tag() - Performs necessary checks for monitor
+ *                                       mode and then tags appropriate packets
+ * @soc: core txrx main context
+ * @vdev: pdev on which packet is received
+ * @msdu: QDF packet buffer on which the protocol tag should be set
+ * @rx_desc: base address where the RX TLVs start
+ * Return: void
+ */
+void dp_rx_mon_update_protocol_flow_tag(struct dp_soc *soc,
+					struct dp_pdev *dp_pdev,
+					qdf_nbuf_t msdu, void *rx_desc)
+{
+	uint32_t msdu_ppdu_id = 0;
+	struct mon_rx_status *mon_recv_status;
+
+	bool is_mon_protocol_flow_tag_enabled =
+		wlan_cfg_is_rx_mon_protocol_flow_tag_enabled(soc->wlan_cfg_ctx);
+
+	if (qdf_likely(!is_mon_protocol_flow_tag_enabled))
+		return;
+
+	if (qdf_likely(!dp_pdev->monitor_vdev))
+		return;
+
+	if (qdf_likely(1 != dp_pdev->ppdu_info.rx_status.rxpcu_filter_pass))
+		return;
+
+	msdu_ppdu_id = HAL_RX_HW_DESC_GET_PPDUID_GET(rx_desc);
+
+	if (msdu_ppdu_id != dp_pdev->ppdu_info.com_info.ppdu_id) {
+		QDF_TRACE(QDF_MODULE_ID_DP,
+			  QDF_TRACE_LEVEL_ERROR,
+			  "msdu_ppdu_id=%x,com_info.ppdu_id=%x",
+			  msdu_ppdu_id,
+			  dp_pdev->ppdu_info.com_info.ppdu_id);
+		return;
+	}
+
+	mon_recv_status = &dp_pdev->ppdu_info.rx_status;
+	if (mon_recv_status->frame_control_info_valid &&
+	    ((mon_recv_status->frame_control & IEEE80211_FC0_TYPE_MASK) ==
+	      IEEE80211_FC0_TYPE_DATA)) {
+		/*
+		 * Update the protocol tag in SKB for packets received on BSS.
+		 * Do not update tag stats since it would double actual
+		 * received count.
+		 */
+		dp_rx_update_protocol_tag(soc,
+					  dp_pdev->monitor_vdev,
+					  msdu, rx_desc,
+					  MAX_REO_DEST_RINGS,
+					  false, false);
+		/* Update the flow tag in SKB based on FSE metadata */
+		dp_rx_update_flow_tag(soc, dp_pdev->monitor_vdev,
+				      msdu, rx_desc, false);
+	}
+}
+#endif /* WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG || WLAN_SUPPORT_RX_FLOW_TAG */
+
+#ifdef WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG
+#ifdef WLAN_SUPPORT_RX_TAG_STATISTICS
+/**
+ * dp_summarize_tag_stats - sums up the given protocol type's counters
+ * across all the rings and dumps the same
+ * @pdev_handle: cdp_pdev handle
+ * @protocol_type: protocol type for which stats should be displayed
+ *
+ * Return: none
+ */
+uint64_t dp_summarize_tag_stats(struct cdp_pdev *pdev_handle,
+				uint16_t protocol_type)
+{
+	struct dp_pdev *pdev = (struct dp_pdev *)pdev_handle;
+	uint8_t ring_idx;
+	uint64_t total_tag_cnt = 0;
+
+	for (ring_idx = 0; ring_idx < MAX_REO_DEST_RINGS; ring_idx++) {
+		total_tag_cnt +=
+		pdev->reo_proto_tag_stats[ring_idx][protocol_type].tag_ctr;
+	}
+	total_tag_cnt += pdev->rx_err_proto_tag_stats[protocol_type].tag_ctr;
+	DP_PRINT_STATS("ProtoID: %d, Tag: %u Tagged MSDU cnt: %llu",
+		       protocol_type,
+		       pdev->rx_proto_tag_map[protocol_type].tag,
+		       total_tag_cnt);
+	return total_tag_cnt;
+}
+
+/**
+ * dp_dump_pdev_rx_protocol_tag_stats - dump the number of packets tagged for
+ * given protocol type (RX_PROTOCOL_TAG_ALL indicates for all protocol)
+ * @pdev_handle: cdp_pdev handle
+ * @protocol_type: protocol type for which stats should be displayed
+ *
+ * Return: none
+ */
+void
+dp_dump_pdev_rx_protocol_tag_stats(struct cdp_pdev *pdev_handle,
+				   uint16_t protocol_type)
+{
+	uint16_t proto_idx;
+
+	if (protocol_type != RX_PROTOCOL_TAG_ALL &&
+	    protocol_type >= RX_PROTOCOL_TAG_MAX) {
+		DP_PRINT_STATS("Invalid protocol type : %u", protocol_type);
+		return;
+	}
+
+	/* protocol_type in [0 ... RX_PROTOCOL_TAG_MAX] */
+	if (protocol_type != RX_PROTOCOL_TAG_ALL) {
+		dp_summarize_tag_stats(pdev_handle, protocol_type);
+		return;
+	}
+
+	/* protocol_type == RX_PROTOCOL_TAG_ALL */
+	for (proto_idx = 0; proto_idx < RX_PROTOCOL_TAG_MAX; proto_idx++)
+		dp_summarize_tag_stats(pdev_handle, proto_idx);
+}
+#endif /* WLAN_SUPPORT_RX_TAG_STATISTICS */
+
+#ifdef WLAN_SUPPORT_RX_TAG_STATISTICS
+/**
+ * dp_reset_pdev_rx_protocol_tag_stats - resets the stats counters for
+ * given protocol type
+ * @pdev_handle: cdp_pdev handle
+ * @protocol_type: protocol type for which stats should be reset
+ *
+ * Return: none
+ */
+void
+dp_reset_pdev_rx_protocol_tag_stats(struct cdp_pdev *pdev_handle,
+				    uint16_t protocol_type)
+{
+	struct dp_pdev *pdev = (struct dp_pdev *)pdev_handle;
+	uint8_t ring_idx;
+
+	for (ring_idx = 0; ring_idx < MAX_REO_DEST_RINGS; ring_idx++)
+		pdev->reo_proto_tag_stats[ring_idx][protocol_type].tag_ctr = 0;
+	pdev->rx_err_proto_tag_stats[protocol_type].tag_ctr = 0;
+}
+#endif /* WLAN_SUPPORT_RX_TAG_STATISTICS */
+
+/**
+ * dp_update_pdev_rx_protocol_tag - Add/remove a protocol tag that should be
+ * applied to the desired protocol type packets
+ * @txrx_pdev_handle: cdp_pdev handle
+ * @enable_rx_protocol_tag - bitmask that indicates what protocol types
+ * are enabled for tagging. zero indicates disable feature, non-zero indicates
+ * enable feature
+ * @protocol_type: new protocol type for which the tag is being added
+ * @tag: user configured tag for the new protocol
+ *
+ * Return: QDF_STATUS
+ */
+QDF_STATUS
+dp_update_pdev_rx_protocol_tag(struct cdp_pdev *pdev_handle,
+			       uint32_t enable_rx_protocol_tag,
+			       uint16_t protocol_type,
+			       uint16_t tag)
+{
+	struct dp_pdev *pdev = (struct dp_pdev *)pdev_handle;
+	/*
+	 * dynamically enable/disable tagging based on enable_rx_protocol_tag
+	 * flag.
+	 */
+	if (enable_rx_protocol_tag) {
+		/* Tagging for one or more protocols has been set by user */
+		pdev->is_rx_protocol_tagging_enabled = true;
+	} else {
+		/*
+		 * No protocols being tagged, disable feature till next add
+		 * operation
+		 */
+		pdev->is_rx_protocol_tagging_enabled = false;
+	}
+
+	/** Reset stats counter across all rings for given protocol */
+	dp_reset_pdev_rx_protocol_tag_stats(pdev_handle, protocol_type);
+
+	pdev->rx_proto_tag_map[protocol_type].tag = tag;
+
+	return QDF_STATUS_SUCCESS;
+}
+#endif /* WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG */
+
+#ifdef WLAN_SUPPORT_RX_FLOW_TAG
+/**
+ * dp_set_rx_flow_tag - add/delete a flow
+ * @pdev_handle: cdp_pdev handle
+ * @flow_info: flow tuple that is to be added to/deleted from flow search table
+ *
+ * Return: 0 for success, nonzero for failure
+ */
+QDF_STATUS
+dp_set_rx_flow_tag(struct cdp_pdev *pdev_handle,
+		   struct cdp_rx_flow_info *flow_info)
+{
+	struct dp_pdev *pdev = (struct dp_pdev *)pdev_handle;
+	struct wlan_cfg_dp_soc_ctxt *cfg = pdev->soc->wlan_cfg_ctx;
+
+	if (qdf_unlikely(!wlan_cfg_is_rx_flow_tag_enabled(cfg))) {
+		dp_err("RX Flow tag feature disabled");
+		return QDF_STATUS_E_NOSUPPORT;
+	}
+
+	if (flow_info->op_code == CDP_FLOW_FST_ENTRY_ADD)
+		return dp_rx_flow_add_entry(pdev, flow_info);
+	if (flow_info->op_code == CDP_FLOW_FST_ENTRY_DEL)
+		return dp_rx_flow_delete_entry(pdev, flow_info);
+
+	return QDF_STATUS_E_INVAL;
+}
+
+/**
+ * dp_dump_rx_flow_tag_stats - dump the number of packets tagged for
+ * given flow 5-tuple
+ * @pdev_handle: cdp_pdev handle
+ * @flow_info: flow 5-tuple for which stats should be displayed
+ *
+ * Return: 0 for success, nonzero for failure
+ */
+QDF_STATUS
+dp_dump_rx_flow_tag_stats(struct cdp_pdev *pdev_handle,
+			  struct cdp_rx_flow_info *flow_info)
+{
+	QDF_STATUS status;
+	struct cdp_flow_stats stats;
+	struct dp_pdev *pdev = (struct dp_pdev *)pdev_handle;
+	struct wlan_cfg_dp_soc_ctxt *cfg = pdev->soc->wlan_cfg_ctx;
+
+	if (qdf_unlikely(!wlan_cfg_is_rx_flow_tag_enabled(cfg))) {
+		dp_err("RX Flow tag feature disabled");
+		return QDF_STATUS_E_NOSUPPORT;
+	}
+
+	status = dp_rx_flow_get_fse_stats(pdev, flow_info, &stats);
+
+	if (status != QDF_STATUS_SUCCESS) {
+		dp_err("Unable to get flow stats, error: %u", status);
+		return status;
+	}
+
+	DP_PRINT_STATS("Dest IP address %x:%x:%x:%x",
+		       flow_info->flow_tuple_info.dest_ip_127_96,
+		       flow_info->flow_tuple_info.dest_ip_95_64,
+		       flow_info->flow_tuple_info.dest_ip_63_32,
+		       flow_info->flow_tuple_info.dest_ip_31_0);
+	DP_PRINT_STATS("Source IP address %x:%x:%x:%x",
+		       flow_info->flow_tuple_info.src_ip_127_96,
+		       flow_info->flow_tuple_info.src_ip_95_64,
+		       flow_info->flow_tuple_info.src_ip_63_32,
+		       flow_info->flow_tuple_info.src_ip_31_0);
+	DP_PRINT_STATS("Dest port %u, Src Port %u, Protocol %u",
+		       flow_info->flow_tuple_info.dest_port,
+		       flow_info->flow_tuple_info.src_port,
+		       flow_info->flow_tuple_info.l4_protocol);
+	DP_PRINT_STATS("MSDU Count: %u", stats.msdu_count);
+
+	return status;
+}
+#endif /* WLAN_SUPPORT_RX_FLOW_TAG */

dp/wifi3.0/dp_rx_tag.h

@@ -0,0 +1,150 @@
+/*
+ * Copyright (c) 2016-2019 The Linux Foundation. All rights reserved.
+ *
+ * Permission to use, copy, modify, and/or distribute this software for
+ * any purpose with or without fee is hereby granted, provided that the
+ * above copyright notice and this permission notice appear in all
+ * copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
+ * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
+ * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
+ * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
+ * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
+ * PERFORMANCE OF THIS SOFTWARE.
+ */
+
+#ifndef _DP_RX_TAG_H_
+#define _DP_RX_TAG_H_
+
+#include "dp_internal.h"
+#include "dp_types.h"
+
+#ifdef WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG
+/**
+ * dp_update_pdev_rx_protocol_tag - Add/remove a protocol tag that should be
+ * applied to the desired protocol type packets
+ * @txrx_pdev_handle: cdp_pdev handle
+ * @enable_rx_protocol_tag - bitmask that indicates what protocol types
+ * are enabled for tagging. zero indicates disable feature, non-zero indicates
+ * enable feature
+ * @protocol_type: new protocol type for which the tag is being added
+ * @tag: user configured tag for the new protocol
+ *
+ * Return: QDF_STATUS
+ */
+QDF_STATUS
+dp_update_pdev_rx_protocol_tag(struct cdp_pdev *pdev_handle,
+			       uint32_t enable_rx_protocol_tag,
+			       uint16_t protocol_type,
+			       uint16_t tag);
+
+/**
+ * dp_rx_update_protocol_tag() - Reads CCE metadata from the RX MSDU end TLV
+ *                              and set the corresponding tag in QDF packet
+ * @soc: core txrx main context
+ * @vdev: vdev on which the packet is received
+ * @nbuf: QDF pkt buffer on which the protocol tag should be set
+ * @rx_tlv_hdr: rBbase address where the RX TLVs starts
+ * @ring_index: REO ring number, not used for error & monitor ring
+ * @is_reo_exception: flag to indicate if rx from REO ring or exception ring
+ * @is_update_stats: flag to indicate whether to update stats or not
+ * Return: void
+ */
+void
+dp_rx_update_protocol_tag(struct dp_soc *soc, struct dp_vdev *vdev,
+			  qdf_nbuf_t nbuf, uint8_t *rx_tlv_hdr,
+			  uint16_t ring_index,
+			  bool is_reo_exception, bool is_update_stats);
+
+#ifdef WLAN_SUPPORT_RX_TAG_STATISTICS
+/**
+ * dp_dump_pdev_rx_protocol_tag_stats - dump the number of packets tagged for
+ * given protocol type (RX_PROTOCOL_TAG_ALL indicates for all protocol)
+ * @pdev_handle: cdp_pdev handle
+ * @protocol_type: protocol type for which stats should be displayed
+ *
+ * Return: none
+ */
+void
+dp_dump_pdev_rx_protocol_tag_stats(struct cdp_pdev *pdev_handle,
+				   uint16_t protocol_type);
+#endif /* WLAN_SUPPORT_RX_TAG_STATISTICS */
+#endif /* WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG */
+
+#ifdef WLAN_SUPPORT_RX_FLOW_TAG
+/**
+ * dp_rx_update_flow_tag() - Reads FSE metadata from the RX MSDU end TLV
+ *                           and set the corresponding tag in QDF packet
+ * @soc: core txrx main context
+ * @vdev: vdev on which the packet is received
+ * @nbuf: QDF pkt buffer on which the protocol tag should be set
+ * @rx_tlv_hdr: base address where the RX TLVs starts
+ * @is_update_stats: flag to indicate whether to update stats or not
+ *
+ * Return: void
+ */
+void
+dp_rx_update_flow_tag(struct dp_soc *soc, struct dp_vdev *vdev,
+		      qdf_nbuf_t nbuf, uint8_t *rx_tlv_hdr, bool update_stats);
+
+/**
+ * dp_set_rx_flow_tag - add/delete a flow
+ * @pdev_handle: cdp_pdev handle
+ * @flow_info: flow tuple that is to be added to/deleted from flow search table
+ *
+ * Return: 0 for success, nonzero for failure
+ */
+QDF_STATUS
+dp_set_rx_flow_tag(struct cdp_pdev *pdev_handle,
+		   struct cdp_rx_flow_info *flow_info);
+
+/**
+ * dp_dump_rx_flow_tag_stats - dump the number of packets tagged for
+ * given flow 5-tuple
+ * @pdev_handle: cdp_pdev handle
+ * @flow_info: flow 5-tuple for which stats should be displayed
+ *
+ * Return: 0 for success, nonzero for failure
+ */
+QDF_STATUS
+dp_dump_rx_flow_tag_stats(struct cdp_pdev *pdev_handle,
+			  struct cdp_rx_flow_info *flow_info);
+
+/**
+ * dp_rx_update_rx_flow_tag_stats() - Update stats for given flow index
+ * @pdev: TXRX pdev context for which stats should be incremented
+ * @flow_index: flow index for which the stats should be incremented
+ *
+ * Return: void
+ */
+extern QDF_STATUS
+dp_rx_flow_update_fse_stats(struct dp_pdev *pdev, uint32_t flow_id);
+
+static inline void
+dp_rx_update_rx_flow_tag_stats(struct dp_pdev *pdev,
+			       uint32_t flow_index)
+{
+	dp_rx_flow_update_fse_stats(pdev, flow_index);
+}
+#endif /* WLAN_SUPPORT_RX_FLOW_TAG */
+
+#if defined(WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG) ||\
+	defined(WLAN_SUPPORT_RX_FLOW_TAG)
+/**
+ * dp_rx_mon_update_protocol_flow_tag() - Performs necessary checks for monitor
+ *                                       mode and then tags appropriate packets
+ * @soc: core txrx main context
+ * @vdev: pdev on which packet is received
+ * @msdu: QDF packet buffer on which the protocol tag should be set
+ * @rx_desc: base address where the RX TLVs start
+ * Return: void
+ */
+void dp_rx_mon_update_protocol_flow_tag(struct dp_soc *soc,
+					struct dp_pdev *dp_pdev,
+					qdf_nbuf_t msdu, void *rx_desc);
+#endif /* WLAN_SUPPORT_RX_PROTOCOL_TYPE_TAG || WLAN_SUPPORT_RX_FLOW_TAG */
+
+#endif /* _DP_RX_TAG_H_ */