Merge branch 'qed-Doorbell-overflow-recovery'

Ariel Elior says:

====================
qed*: Doorbell overflow recovery

Doorbell Overflow
If sufficient CPU cores will send doorbells at a sufficiently high rate, they
can cause an overflow in the doorbell queue block message fifo. When fill level
reaches maximum, the device stops accepting all doorbells from that PF until a
recovery procedure has taken place.

Doorbell Overflow Recovery
The recovery procedure basically means resending the last doorbell for every
doorbelling entity. A doorbelling entity is anything which may send doorbells:
L2 tx ring, rdma sq/rq/cq, light l2, vf l2 tx ring, spq, etc. This relies on
the design assumption that all doorbells are aggregative, so last doorbell
carries the information of all previous doorbells.

APIs
All doorbelling entities need to register with the mechanism before sending
doorbells. The registration entails providing the doorbell address the entity
would be using, and a virtual address where last doorbell data can be found.
Typically fastpath structures already have this construct.

Executing the recovery procedure
Handling the attentions, iterating over all the registered entities and
resending their doorbells, is all handled within qed core module.

Relevance
All doorbelling entities in all protocols need to register with the mechanism,
via the new APIs. Technically this is quite simple (just call the API). Some
protocol fastpath implementation may not have the doorbell data stored anywhere
(compute it from scratch every time) and will have to add such a place.
This is rare and is also better practice (save some cycles on the fastpath).

Performance Penalty
No performance penalty should incur as a result of this feature. If anything
performance can improve by avoiding recalcualtion of doorbell data everytime
doorbell is sent (in some flows).

Add the database used to register doorbelling entities, and APIs for adding
and deleting entries, and logic for traversing the database and doorbelling
once on behalf of all entities.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/qlogic/qed/qed.h

@@ -296,6 +296,12 @@ enum qed_wol_support {
 	QED_WOL_SUPPORT_PME,
 };
 
+enum qed_db_rec_exec {
+	DB_REC_DRY_RUN,
+	DB_REC_REAL_DEAL,
+	DB_REC_ONCE,
+};
+
 struct qed_hw_info {
 	/* PCI personality */
 	enum qed_pci_personality personality;
@@ -425,6 +431,14 @@ struct qed_qm_info {
 	u8 num_pf_rls;
 };
 
+struct qed_db_recovery_info {
+	struct list_head list;
+
+	/* Lock to protect the doorbell recovery mechanism list */
+	spinlock_t lock;
+	u32 db_recovery_counter;
+};
+
 struct storm_stats {
 	u32     address;
 	u32     len;
@@ -522,6 +536,7 @@ struct qed_simd_fp_handler {
 
 enum qed_slowpath_wq_flag {
 	QED_SLOWPATH_MFW_TLV_REQ,
+	QED_SLOWPATH_PERIODIC_DB_REC,
 };
 
 struct qed_hwfn {
@@ -640,6 +655,9 @@ struct qed_hwfn {
 	/* L2-related */
 	struct qed_l2_info *p_l2_info;
 
+	/* Mechanism for recovering from doorbell drop */
+	struct qed_db_recovery_info db_recovery_info;
+
 	/* Nvm images number and attributes */
 	struct qed_nvm_image_info nvm_info;
 
@@ -652,11 +670,12 @@ struct qed_hwfn {
 	struct delayed_work iov_task;
 	unsigned long iov_task_flags;
 #endif
-
-	struct z_stream_s		*stream;
+	struct z_stream_s *stream;
+	bool slowpath_wq_active;
 	struct workqueue_struct *slowpath_wq;
 	struct delayed_work slowpath_task;
 	unsigned long slowpath_task_flags;
+	u32 periodic_db_rec_count;
 };
 
 struct pci_params {
@@ -897,6 +916,12 @@ u16 qed_get_cm_pq_idx_llt_mtc(struct qed_hwfn *p_hwfn, u8 tc);
 
 #define QED_LEADING_HWFN(dev)   (&dev->hwfns[0])
 
+/* doorbell recovery mechanism */
+void qed_db_recovery_dp(struct qed_hwfn *p_hwfn);
+void qed_db_recovery_execute(struct qed_hwfn *p_hwfn,
+			     enum qed_db_rec_exec db_exec);
+bool qed_edpm_enabled(struct qed_hwfn *p_hwfn);
+
 /* Other Linux specific common definitions */
 #define DP_NAME(cdev) ((cdev)->name)
 
@@ -931,4 +956,6 @@ int qed_mfw_fill_tlv_data(struct qed_hwfn *hwfn,
 			  union qed_mfw_tlv_data *tlv_data);
 
 void qed_hw_info_set_offload_tc(struct qed_hw_info *p_info, u8 tc);
+
+void qed_periodic_db_rec_start(struct qed_hwfn *p_hwfn);
 #endif /* _QED_H */

drivers/net/ethernet/qlogic/qed/qed_dev_api.h

@@ -472,6 +472,34 @@ int qed_get_queue_coalesce(struct qed_hwfn *p_hwfn, u16 *coal, void *handle);
 int
 qed_set_queue_coalesce(u16 rx_coal, u16 tx_coal, void *p_handle);
 
+/**
+ * @brief db_recovery_add - add doorbell information to the doorbell
+ * recovery mechanism.
+ *
+ * @param cdev
+ * @param db_addr - doorbell address
+ * @param db_data - address of where db_data is stored
+ * @param db_width - doorbell is 32b pr 64b
+ * @param db_space - doorbell recovery addresses are user or kernel space
+ */
+int qed_db_recovery_add(struct qed_dev *cdev,
+			void __iomem *db_addr,
+			void *db_data,
+			enum qed_db_rec_width db_width,
+			enum qed_db_rec_space db_space);
+
+/**
+ * @brief db_recovery_del - remove doorbell information from the doorbell
+ * recovery mechanism. db_data serves as key (db_addr is not unique).
+ *
+ * @param cdev
+ * @param db_addr - doorbell address
+ * @param db_data - address where db_data is stored. Serves as key for the
+ *                  entry to delete.
+ */
+int qed_db_recovery_del(struct qed_dev *cdev,
+			void __iomem *db_addr, void *db_data);
+
 
 const char *qed_hw_get_resc_name(enum qed_resources res_id);
 #endif

drivers/net/ethernet/qlogic/qed/qed_int.c

@@ -361,29 +361,147 @@ static int qed_pglub_rbc_attn_cb(struct qed_hwfn *p_hwfn)
 	return 0;
 }
 
-#define QED_DORQ_ATTENTION_REASON_MASK	(0xfffff)
-#define QED_DORQ_ATTENTION_OPAQUE_MASK (0xffff)
-#define QED_DORQ_ATTENTION_SIZE_MASK	(0x7f)
-#define QED_DORQ_ATTENTION_SIZE_SHIFT	(16)
+#define QED_DORQ_ATTENTION_REASON_MASK  (0xfffff)
+#define QED_DORQ_ATTENTION_OPAQUE_MASK  (0xffff)
+#define QED_DORQ_ATTENTION_OPAQUE_SHIFT (0x0)
+#define QED_DORQ_ATTENTION_SIZE_MASK            (0x7f)
+#define QED_DORQ_ATTENTION_SIZE_SHIFT           (16)
+
+#define QED_DB_REC_COUNT                        1000
+#define QED_DB_REC_INTERVAL                     100
+
+static int qed_db_rec_flush_queue(struct qed_hwfn *p_hwfn,
+				  struct qed_ptt *p_ptt)
+{
+	u32 count = QED_DB_REC_COUNT;
+	u32 usage = 1;
+
+	/* wait for usage to zero or count to run out. This is necessary since
+	 * EDPM doorbell transactions can take multiple 64b cycles, and as such
+	 * can "split" over the pci. Possibly, the doorbell drop can happen with
+	 * half an EDPM in the queue and other half dropped. Another EDPM
+	 * doorbell to the same address (from doorbell recovery mechanism or
+	 * from the doorbelling entity) could have first half dropped and second
+	 * half interpreted as continuation of the first. To prevent such
+	 * malformed doorbells from reaching the device, flush the queue before
+	 * releasing the overflow sticky indication.
+	 */
+	while (count-- && usage) {
+		usage = qed_rd(p_hwfn, p_ptt, DORQ_REG_PF_USAGE_CNT);
+		udelay(QED_DB_REC_INTERVAL);
+	}
+
+	/* should have been depleted by now */
+	if (usage) {
+		DP_NOTICE(p_hwfn->cdev,
+			  "DB recovery: doorbell usage failed to zero after %d usec. usage was %x\n",
+			  QED_DB_REC_INTERVAL * QED_DB_REC_COUNT, usage);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+int qed_db_rec_handler(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
+{
+	u32 overflow;
+	int rc;
+
+	overflow = qed_rd(p_hwfn, p_ptt, DORQ_REG_PF_OVFL_STICKY);
+	DP_NOTICE(p_hwfn, "PF Overflow sticky 0x%x\n", overflow);
+	if (!overflow) {
+		qed_db_recovery_execute(p_hwfn, DB_REC_ONCE);
+		return 0;
+	}
+
+	if (qed_edpm_enabled(p_hwfn)) {
+		rc = qed_db_rec_flush_queue(p_hwfn, p_ptt);
+		if (rc)
+			return rc;
+	}
+
+	/* Flush any pending (e)dpm as they may never arrive */
+	qed_wr(p_hwfn, p_ptt, DORQ_REG_DPM_FORCE_ABORT, 0x1);
+
+	/* Release overflow sticky indication (stop silently dropping everything) */
+	qed_wr(p_hwfn, p_ptt, DORQ_REG_PF_OVFL_STICKY, 0x0);
+
+	/* Repeat all last doorbells (doorbell drop recovery) */
+	qed_db_recovery_execute(p_hwfn, DB_REC_REAL_DEAL);
+
+	return 0;
+}
+
 static int qed_dorq_attn_cb(struct qed_hwfn *p_hwfn)
 {
-	u32 reason;
+	u32 int_sts, first_drop_reason, details, address, all_drops_reason;
+	struct qed_ptt *p_ptt = p_hwfn->p_dpc_ptt;
+	int rc;
 
-	reason = qed_rd(p_hwfn, p_hwfn->p_dpc_ptt, DORQ_REG_DB_DROP_REASON) &
-			QED_DORQ_ATTENTION_REASON_MASK;
-	if (reason) {
-		u32 details = qed_rd(p_hwfn, p_hwfn->p_dpc_ptt,
-				     DORQ_REG_DB_DROP_DETAILS);
+	int_sts = qed_rd(p_hwfn, p_ptt, DORQ_REG_INT_STS);
+	DP_NOTICE(p_hwfn->cdev, "DORQ attention. int_sts was %x\n", int_sts);
 
-		DP_INFO(p_hwfn->cdev,
-			"DORQ db_drop: address 0x%08x Opaque FID 0x%04x Size [bytes] 0x%08x Reason: 0x%08x\n",
-			qed_rd(p_hwfn, p_hwfn->p_dpc_ptt,
-			       DORQ_REG_DB_DROP_DETAILS_ADDRESS),
-			(u16)(details & QED_DORQ_ATTENTION_OPAQUE_MASK),
-			GET_FIELD(details, QED_DORQ_ATTENTION_SIZE) * 4,
-			reason);
+	/* int_sts may be zero since all PFs were interrupted for doorbell
+	 * overflow but another one already handled it. Can abort here. If
+	 * This PF also requires overflow recovery we will be interrupted again.
+	 * The masked almost full indication may also be set. Ignoring.
+	 */
+	if (!(int_sts & ~DORQ_REG_INT_STS_DORQ_FIFO_AFULL))
+		return 0;
+
+	/* check if db_drop or overflow happened */
+	if (int_sts & (DORQ_REG_INT_STS_DB_DROP |
+		       DORQ_REG_INT_STS_DORQ_FIFO_OVFL_ERR)) {
+		/* Obtain data about db drop/overflow */
+		first_drop_reason = qed_rd(p_hwfn, p_ptt,
+					   DORQ_REG_DB_DROP_REASON) &
+		    QED_DORQ_ATTENTION_REASON_MASK;
+		details = qed_rd(p_hwfn, p_ptt, DORQ_REG_DB_DROP_DETAILS);
+		address = qed_rd(p_hwfn, p_ptt,
+				 DORQ_REG_DB_DROP_DETAILS_ADDRESS);
+		all_drops_reason = qed_rd(p_hwfn, p_ptt,
+					  DORQ_REG_DB_DROP_DETAILS_REASON);
+
+		/* Log info */
+		DP_NOTICE(p_hwfn->cdev,
+			  "Doorbell drop occurred\n"
+			  "Address\t\t0x%08x\t(second BAR address)\n"
+			  "FID\t\t0x%04x\t\t(Opaque FID)\n"
+			  "Size\t\t0x%04x\t\t(in bytes)\n"
+			  "1st drop reason\t0x%08x\t(details on first drop since last handling)\n"
+			  "Sticky reasons\t0x%08x\t(all drop reasons since last handling)\n",
+			  address,
+			  GET_FIELD(details, QED_DORQ_ATTENTION_OPAQUE),
+			  GET_FIELD(details, QED_DORQ_ATTENTION_SIZE) * 4,
+			  first_drop_reason, all_drops_reason);
+
+		rc = qed_db_rec_handler(p_hwfn, p_ptt);
+		qed_periodic_db_rec_start(p_hwfn);
+		if (rc)
+			return rc;
+
+		/* Clear the doorbell drop details and prepare for next drop */
+		qed_wr(p_hwfn, p_ptt, DORQ_REG_DB_DROP_DETAILS_REL, 0);
+
+		/* Mark interrupt as handled (note: even if drop was due to a different
+		 * reason than overflow we mark as handled)
+		 */
+		qed_wr(p_hwfn,
+		       p_ptt,
+		       DORQ_REG_INT_STS_WR,
+		       DORQ_REG_INT_STS_DB_DROP |
+		       DORQ_REG_INT_STS_DORQ_FIFO_OVFL_ERR);
+
+		/* If there are no indications other than drop indications, success */
+		if ((int_sts & ~(DORQ_REG_INT_STS_DB_DROP |
+				 DORQ_REG_INT_STS_DORQ_FIFO_OVFL_ERR |
+				 DORQ_REG_INT_STS_DORQ_FIFO_AFULL)) == 0)
+			return 0;
 	}
 
+	/* Some other indication was present - non recoverable */
+	DP_INFO(p_hwfn, "DORQ fatal attention\n");
+
 	return -EINVAL;
 }
 

drivers/net/ethernet/qlogic/qed/qed_int.h

@@ -190,6 +190,16 @@ void qed_int_get_num_sbs(struct qed_hwfn	*p_hwfn,
  */
 void qed_int_disable_post_isr_release(struct qed_dev *cdev);
 
+/**
+ * @brief - Doorbell Recovery handler.
+ *          Run DB_REAL_DEAL doorbell recovery in case of PF overflow
+ *          (and flush DORQ if needed), otherwise run DB_REC_ONCE.
+ *
+ * @param p_hwfn
+ * @param p_ptt
+ */
+int qed_db_rec_handler(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt);
+
 #define QED_CAU_DEF_RX_TIMER_RES 0
 #define QED_CAU_DEF_TX_TIMER_RES 0
 

drivers/net/ethernet/qlogic/qed/qed_ll2.c

@@ -1085,7 +1085,14 @@ static int qed_sp_ll2_tx_queue_start(struct qed_hwfn *p_hwfn,
 
 	p_ramrod->gsi_offload_flag = p_ll2_conn->input.gsi_enable;
 
-	return qed_spq_post(p_hwfn, p_ent, NULL);
+	rc = qed_spq_post(p_hwfn, p_ent, NULL);
+	if (rc)
+		return rc;
+
+	rc = qed_db_recovery_add(p_hwfn->cdev, p_tx->doorbell_addr,
+				 &p_tx->db_msg, DB_REC_WIDTH_32B,
+				 DB_REC_KERNEL);
+	return rc;
 }
 
 static int qed_sp_ll2_rx_queue_stop(struct qed_hwfn *p_hwfn,
@@ -1119,9 +1126,11 @@ static int qed_sp_ll2_rx_queue_stop(struct qed_hwfn *p_hwfn,
 static int qed_sp_ll2_tx_queue_stop(struct qed_hwfn *p_hwfn,
 				    struct qed_ll2_info *p_ll2_conn)
 {
+	struct qed_ll2_tx_queue *p_tx = &p_ll2_conn->tx_queue;
 	struct qed_spq_entry *p_ent = NULL;
 	struct qed_sp_init_data init_data;
 	int rc = -EINVAL;
+	qed_db_recovery_del(p_hwfn->cdev, p_tx->doorbell_addr, &p_tx->db_msg);
 
 	/* Get SPQ entry */
 	memset(&init_data, 0, sizeof(init_data));
@@ -1542,6 +1551,13 @@ int qed_ll2_establish_connection(void *cxt, u8 connection_handle)
 	p_tx->doorbell_addr = (u8 __iomem *)p_hwfn->doorbells +
 					    qed_db_addr(p_ll2_conn->cid,
 							DQ_DEMS_LEGACY);
+	/* prepare db data */
+	SET_FIELD(p_tx->db_msg.params, CORE_DB_DATA_DEST, DB_DEST_XCM);
+	SET_FIELD(p_tx->db_msg.params, CORE_DB_DATA_AGG_CMD, DB_AGG_CMD_SET);
+	SET_FIELD(p_tx->db_msg.params, CORE_DB_DATA_AGG_VAL_SEL,
+		  DQ_XCM_CORE_TX_BD_PROD_CMD);
+	p_tx->db_msg.agg_flags = DQ_XCM_CORE_DQ_CF_CMD;
+
 
 	rc = qed_ll2_establish_connection_rx(p_hwfn, p_ll2_conn);
 	if (rc)
@@ -1780,7 +1796,6 @@ static void qed_ll2_tx_packet_notify(struct qed_hwfn *p_hwfn,
 	bool b_notify = p_ll2_conn->tx_queue.cur_send_packet->notify_fw;
 	struct qed_ll2_tx_queue *p_tx = &p_ll2_conn->tx_queue;
 	struct qed_ll2_tx_packet *p_pkt = NULL;
-	struct core_db_data db_msg = { 0, 0, 0 };
 	u16 bd_prod;
 
 	/* If there are missing BDs, don't do anything now */
@@ -1809,24 +1824,19 @@ static void qed_ll2_tx_packet_notify(struct qed_hwfn *p_hwfn,
 		list_move_tail(&p_pkt->list_entry, &p_tx->active_descq);
 	}
 
-	SET_FIELD(db_msg.params, CORE_DB_DATA_DEST, DB_DEST_XCM);
-	SET_FIELD(db_msg.params, CORE_DB_DATA_AGG_CMD, DB_AGG_CMD_SET);
-	SET_FIELD(db_msg.params, CORE_DB_DATA_AGG_VAL_SEL,
-		  DQ_XCM_CORE_TX_BD_PROD_CMD);
-	db_msg.agg_flags = DQ_XCM_CORE_DQ_CF_CMD;
-	db_msg.spq_prod = cpu_to_le16(bd_prod);
+	p_tx->db_msg.spq_prod = cpu_to_le16(bd_prod);
 
 	/* Make sure the BDs data is updated before ringing the doorbell */
 	wmb();
 
-	DIRECT_REG_WR(p_tx->doorbell_addr, *((u32 *)&db_msg));
+	DIRECT_REG_WR(p_tx->doorbell_addr, *((u32 *)&p_tx->db_msg));
 
 	DP_VERBOSE(p_hwfn,
 		   (NETIF_MSG_TX_QUEUED | QED_MSG_LL2),
 		   "LL2 [q 0x%02x cid 0x%08x type 0x%08x] Doorbelled [producer 0x%04x]\n",
 		   p_ll2_conn->queue_id,
 		   p_ll2_conn->cid,
-		   p_ll2_conn->input.conn_type, db_msg.spq_prod);
+		   p_ll2_conn->input.conn_type, p_tx->db_msg.spq_prod);
 }
 
 int qed_ll2_prepare_tx_packet(void *cxt,

drivers/net/ethernet/qlogic/qed/qed_ll2.h

@@ -103,6 +103,7 @@ struct qed_ll2_tx_queue {
 	struct qed_ll2_tx_packet cur_completing_packet;
 	u16 cur_completing_bd_idx;
 	void __iomem *doorbell_addr;
+	struct core_db_data db_msg;
 	u16 bds_idx;
 	u16 cur_send_frag_num;
 	u16 cur_completing_frag_num;

drivers/net/ethernet/qlogic/qed/qed_main.c

@@ -966,9 +966,47 @@ static void qed_update_pf_params(struct qed_dev *cdev,
 	}
 }
 
+#define QED_PERIODIC_DB_REC_COUNT		100
+#define QED_PERIODIC_DB_REC_INTERVAL_MS		100
+#define QED_PERIODIC_DB_REC_INTERVAL \
+	msecs_to_jiffies(QED_PERIODIC_DB_REC_INTERVAL_MS)
+#define QED_PERIODIC_DB_REC_WAIT_COUNT		10
+#define QED_PERIODIC_DB_REC_WAIT_INTERVAL \
+	(QED_PERIODIC_DB_REC_INTERVAL_MS / QED_PERIODIC_DB_REC_WAIT_COUNT)
+
+static int qed_slowpath_delayed_work(struct qed_hwfn *hwfn,
+				     enum qed_slowpath_wq_flag wq_flag,
+				     unsigned long delay)
+{
+	if (!hwfn->slowpath_wq_active)
+		return -EINVAL;
+
+	/* Memory barrier for setting atomic bit */
+	smp_mb__before_atomic();
+	set_bit(wq_flag, &hwfn->slowpath_task_flags);
+	smp_mb__after_atomic();
+	queue_delayed_work(hwfn->slowpath_wq, &hwfn->slowpath_task, delay);
+
+	return 0;
+}
+
+void qed_periodic_db_rec_start(struct qed_hwfn *p_hwfn)
+{
+	/* Reset periodic Doorbell Recovery counter */
+	p_hwfn->periodic_db_rec_count = QED_PERIODIC_DB_REC_COUNT;
+
+	/* Don't schedule periodic Doorbell Recovery if already scheduled */
+	if (test_bit(QED_SLOWPATH_PERIODIC_DB_REC,
+		     &p_hwfn->slowpath_task_flags))
+		return;
+
+	qed_slowpath_delayed_work(p_hwfn, QED_SLOWPATH_PERIODIC_DB_REC,
+				  QED_PERIODIC_DB_REC_INTERVAL);
+}
+
 static void qed_slowpath_wq_stop(struct qed_dev *cdev)
 {
-	int i;
+	int i, sleep_count = QED_PERIODIC_DB_REC_WAIT_COUNT;
 
 	if (IS_VF(cdev))
 		return;
@@ -977,6 +1015,15 @@ static void qed_slowpath_wq_stop(struct qed_dev *cdev)
 		if (!cdev->hwfns[i].slowpath_wq)
 			continue;
 
+		/* Stop queuing new delayed works */
+		cdev->hwfns[i].slowpath_wq_active = false;
+
+		/* Wait until the last periodic doorbell recovery is executed */
+		while (test_bit(QED_SLOWPATH_PERIODIC_DB_REC,
+				&cdev->hwfns[i].slowpath_task_flags) &&
+		       sleep_count--)
+			msleep(QED_PERIODIC_DB_REC_WAIT_INTERVAL);
+
 		flush_workqueue(cdev->hwfns[i].slowpath_wq);
 		destroy_workqueue(cdev->hwfns[i].slowpath_wq);
 	}
@@ -989,7 +1036,10 @@ static void qed_slowpath_task(struct work_struct *work)
 	struct qed_ptt *ptt = qed_ptt_acquire(hwfn);
 
 	if (!ptt) {
-		queue_delayed_work(hwfn->slowpath_wq, &hwfn->slowpath_task, 0);
+		if (hwfn->slowpath_wq_active)
+			queue_delayed_work(hwfn->slowpath_wq,
+					   &hwfn->slowpath_task, 0);
+
 		return;
 	}
 
@@ -997,6 +1047,15 @@ static void qed_slowpath_task(struct work_struct *work)
 			       &hwfn->slowpath_task_flags))
 		qed_mfw_process_tlv_req(hwfn, ptt);
 
+	if (test_and_clear_bit(QED_SLOWPATH_PERIODIC_DB_REC,
+			       &hwfn->slowpath_task_flags)) {
+		qed_db_rec_handler(hwfn, ptt);
+		if (hwfn->periodic_db_rec_count--)
+			qed_slowpath_delayed_work(hwfn,
+						  QED_SLOWPATH_PERIODIC_DB_REC,
+						  QED_PERIODIC_DB_REC_INTERVAL);
+	}
+
 	qed_ptt_release(hwfn, ptt);
 }
 
@@ -1023,6 +1082,7 @@ static int qed_slowpath_wq_start(struct qed_dev *cdev)
 		}
 
 		INIT_DELAYED_WORK(&hwfn->slowpath_task, qed_slowpath_task);
+		hwfn->slowpath_wq_active = true;
 	}
 
 	return 0;
@@ -2324,6 +2384,8 @@ const struct qed_common_ops qed_common_ops_pass = {
 	.update_mac = &qed_update_mac,
 	.update_mtu = &qed_update_mtu,
 	.update_wol = &qed_update_wol,
+	.db_recovery_add = &qed_db_recovery_add,
+	.db_recovery_del = &qed_db_recovery_del,
 	.read_module_eeprom = &qed_read_module_eeprom,
 };
 

drivers/net/ethernet/qlogic/qed/qed_reg_addr.h

@@ -1243,6 +1243,56 @@
 	0x1701534UL
 #define TSEM_REG_DBG_FORCE_FRAME \
 	0x1701538UL
+#define DORQ_REG_PF_USAGE_CNT \
+	0x1009c0UL
+#define DORQ_REG_PF_OVFL_STICKY	\
+	0x1009d0UL
+#define DORQ_REG_DPM_FORCE_ABORT \
+	0x1009d8UL
+#define DORQ_REG_INT_STS \
+	0x100180UL
+#define DORQ_REG_INT_STS_ADDRESS_ERROR \
+	(0x1UL << 0)
+#define DORQ_REG_INT_STS_WR \
+	0x100188UL
+#define DORQ_REG_DB_DROP_DETAILS_REL \
+	0x100a28UL
+#define DORQ_REG_INT_STS_ADDRESS_ERROR_SHIFT \
+	0
+#define DORQ_REG_INT_STS_DB_DROP \
+		(0x1UL << 1)
+#define DORQ_REG_INT_STS_DB_DROP_SHIFT \
+	1
+#define DORQ_REG_INT_STS_DORQ_FIFO_OVFL_ERR \
+		(0x1UL << 2)
+#define DORQ_REG_INT_STS_DORQ_FIFO_OVFL_ERR_SHIFT \
+	2
+#define DORQ_REG_INT_STS_DORQ_FIFO_AFULL\
+		(0x1UL << 3)
+#define DORQ_REG_INT_STS_DORQ_FIFO_AFULL_SHIFT \
+	3
+#define DORQ_REG_INT_STS_CFC_BYP_VALIDATION_ERR \
+		(0x1UL << 4)
+#define DORQ_REG_INT_STS_CFC_BYP_VALIDATION_ERR_SHIFT \
+	4
+#define DORQ_REG_INT_STS_CFC_LD_RESP_ERR \
+		(0x1UL << 5)
+#define DORQ_REG_INT_STS_CFC_LD_RESP_ERR_SHIFT \
+	5
+#define DORQ_REG_INT_STS_XCM_DONE_CNT_ERR \
+		(0x1UL << 6)
+#define DORQ_REG_INT_STS_XCM_DONE_CNT_ERR_SHIFT	\
+	6
+#define DORQ_REG_INT_STS_CFC_LD_REQ_FIFO_OVFL_ERR \
+		(0x1UL << 7)
+#define DORQ_REG_INT_STS_CFC_LD_REQ_FIFO_OVFL_ERR_SHIFT	\
+	7
+#define DORQ_REG_INT_STS_CFC_LD_REQ_FIFO_UNDER_ERR \
+		(0x1UL << 8)
+#define DORQ_REG_INT_STS_CFC_LD_REQ_FIFO_UNDER_ERR_SHIFT \
+	8
+#define DORQ_REG_DB_DROP_DETAILS_REASON	\
+	0x100a20UL
 #define MSEM_REG_DBG_SELECT \
 	0x1801528UL
 #define MSEM_REG_DBG_DWORD_ENABLE \

drivers/net/ethernet/qlogic/qed/qed_sp.h

@@ -227,7 +227,9 @@ struct qed_spq {
 	u32			comp_count;
 
 	u32			cid;
-	qed_spq_async_comp_cb async_comp_cb[MAX_PROTOCOL_TYPE];
+	u32			db_addr_offset;
+	struct core_db_data	db_data;
+	qed_spq_async_comp_cb	async_comp_cb[MAX_PROTOCOL_TYPE];
 };
 
 /**

drivers/net/ethernet/qlogic/qed/qed_spq.c

@@ -252,9 +252,9 @@ static int qed_spq_hw_post(struct qed_hwfn *p_hwfn,
 			   struct qed_spq *p_spq, struct qed_spq_entry *p_ent)
 {
 	struct qed_chain *p_chain = &p_hwfn->p_spq->chain;
+	struct core_db_data *p_db_data = &p_spq->db_data;
 	u16 echo = qed_chain_get_prod_idx(p_chain);
 	struct slow_path_element	*elem;
-	struct core_db_data		db;
 
 	p_ent->elem.hdr.echo	= cpu_to_le16(echo);
 	elem = qed_chain_produce(p_chain);
@@ -266,27 +266,22 @@ static int qed_spq_hw_post(struct qed_hwfn *p_hwfn,
 	*elem = p_ent->elem; /* struct assignment */
 
 	/* send a doorbell on the slow hwfn session */
-	memset(&db, 0, sizeof(db));
-	SET_FIELD(db.params, CORE_DB_DATA_DEST, DB_DEST_XCM);
-	SET_FIELD(db.params, CORE_DB_DATA_AGG_CMD, DB_AGG_CMD_SET);
-	SET_FIELD(db.params, CORE_DB_DATA_AGG_VAL_SEL,
-		  DQ_XCM_CORE_SPQ_PROD_CMD);
-	db.agg_flags = DQ_XCM_CORE_DQ_CF_CMD;
-	db.spq_prod = cpu_to_le16(qed_chain_get_prod_idx(p_chain));
+	p_db_data->spq_prod = cpu_to_le16(qed_chain_get_prod_idx(p_chain));
 
 	/* make sure the SPQE is updated before the doorbell */
 	wmb();
 
-	DOORBELL(p_hwfn, qed_db_addr(p_spq->cid, DQ_DEMS_LEGACY), *(u32 *)&db);
+	DOORBELL(p_hwfn, p_spq->db_addr_offset, *(u32 *)p_db_data);
 
 	/* make sure doorbell is rang */
 	wmb();
 
 	DP_VERBOSE(p_hwfn, QED_MSG_SPQ,
 		   "Doorbelled [0x%08x, CID 0x%08x] with Flags: %02x agg_params: %02x, prod: %04x\n",
-		   qed_db_addr(p_spq->cid, DQ_DEMS_LEGACY),
-		   p_spq->cid, db.params, db.agg_flags,
-		   qed_chain_get_prod_idx(p_chain));
+		   p_spq->db_addr_offset,
+		   p_spq->cid,
+		   p_db_data->params,
+		   p_db_data->agg_flags, qed_chain_get_prod_idx(p_chain));
 
 	return 0;
 }
@@ -490,8 +485,11 @@ void qed_spq_setup(struct qed_hwfn *p_hwfn)
 {
 	struct qed_spq *p_spq = p_hwfn->p_spq;
 	struct qed_spq_entry *p_virt = NULL;
+	struct core_db_data *p_db_data;
+	void __iomem *db_addr;
 	dma_addr_t p_phys = 0;
 	u32 i, capacity;
+	int rc;
 
 	INIT_LIST_HEAD(&p_spq->pending);
 	INIT_LIST_HEAD(&p_spq->completion_pending);
@@ -528,6 +526,25 @@ void qed_spq_setup(struct qed_hwfn *p_hwfn)
 
 	/* reset the chain itself */
 	qed_chain_reset(&p_spq->chain);
+
+	/* Initialize the address/data of the SPQ doorbell */
+	p_spq->db_addr_offset = qed_db_addr(p_spq->cid, DQ_DEMS_LEGACY);
+	p_db_data = &p_spq->db_data;
+	memset(p_db_data, 0, sizeof(*p_db_data));
+	SET_FIELD(p_db_data->params, CORE_DB_DATA_DEST, DB_DEST_XCM);
+	SET_FIELD(p_db_data->params, CORE_DB_DATA_AGG_CMD, DB_AGG_CMD_MAX);
+	SET_FIELD(p_db_data->params, CORE_DB_DATA_AGG_VAL_SEL,
+		  DQ_XCM_CORE_SPQ_PROD_CMD);
+	p_db_data->agg_flags = DQ_XCM_CORE_DQ_CF_CMD;
+
+	/* Register the SPQ doorbell with the doorbell recovery mechanism */
+	db_addr = (void __iomem *)((u8 __iomem *)p_hwfn->doorbells +
+				   p_spq->db_addr_offset);
+	rc = qed_db_recovery_add(p_hwfn->cdev, db_addr, &p_spq->db_data,
+				 DB_REC_WIDTH_32B, DB_REC_KERNEL);
+	if (rc)
+		DP_INFO(p_hwfn,
+			"Failed to register the SPQ doorbell with the doorbell recovery mechanism\n");
 }
 
 int qed_spq_alloc(struct qed_hwfn *p_hwfn)
@@ -575,11 +592,17 @@ int qed_spq_alloc(struct qed_hwfn *p_hwfn)
 void qed_spq_free(struct qed_hwfn *p_hwfn)
 {
 	struct qed_spq *p_spq = p_hwfn->p_spq;
+	void __iomem *db_addr;
 	u32 capacity;
 
 	if (!p_spq)
 		return;
 
+	/* Delete the SPQ doorbell from the doorbell recovery mechanism */
+	db_addr = (void __iomem *)((u8 __iomem *)p_hwfn->doorbells +
+				   p_spq->db_addr_offset);
+	qed_db_recovery_del(p_hwfn->cdev, db_addr, &p_spq->db_data);
+
 	if (p_spq->p_virt) {
 		capacity = qed_chain_get_capacity(&p_spq->chain);
 		dma_free_coherent(&p_hwfn->cdev->pdev->dev,

drivers/net/ethernet/qlogic/qede/qede_main.c

@@ -1774,6 +1774,10 @@ static int qede_drain_txq(struct qede_dev *edev,
 static int qede_stop_txq(struct qede_dev *edev,
 			 struct qede_tx_queue *txq, int rss_id)
 {
+	/* delete doorbell from doorbell recovery mechanism */
+	edev->ops->common->db_recovery_del(edev->cdev, txq->doorbell_addr,
+					   &txq->tx_db);
+
 	return edev->ops->q_tx_stop(edev->cdev, rss_id, txq->handle);
 }
 
@@ -1910,6 +1914,11 @@ static int qede_start_txq(struct qede_dev *edev,
 		  DQ_XCM_ETH_TX_BD_PROD_CMD);
 	txq->tx_db.data.agg_flags = DQ_XCM_ETH_DQ_CF_CMD;
 
+	/* register doorbell with doorbell recovery mechanism */
+	rc = edev->ops->common->db_recovery_add(edev->cdev, txq->doorbell_addr,
+						&txq->tx_db, DB_REC_WIDTH_32B,
+						DB_REC_KERNEL);
+
 	return rc;
 }
 

include/linux/qed/qed_if.h

@@ -47,6 +47,7 @@
 #include <linux/slab.h>
 #include <linux/qed/common_hsi.h>
 #include <linux/qed/qed_chain.h>
+#include <linux/io-64-nonatomic-lo-hi.h>
 
 enum dcbx_protocol_type {
 	DCBX_PROTOCOL_ISCSI,
@@ -448,11 +449,24 @@ struct qed_mfw_tlv_iscsi {
 	bool tx_bytes_set;
 };
 
+enum qed_db_rec_width {
+	DB_REC_WIDTH_32B,
+	DB_REC_WIDTH_64B,
+};
+
+enum qed_db_rec_space {
+	DB_REC_KERNEL,
+	DB_REC_USER,
+};
+
 #define DIRECT_REG_WR(reg_addr, val) writel((u32)val, \
 					    (void __iomem *)(reg_addr))
 
 #define DIRECT_REG_RD(reg_addr) readl((void __iomem *)(reg_addr))
 
+#define DIRECT_REG_WR64(reg_addr, val) writeq((u32)val,	\
+					      (void __iomem *)(reg_addr))
+
 #define QED_COALESCE_MAX 0x1FF
 #define QED_DEFAULT_RX_USECS 12
 #define QED_DEFAULT_TX_USECS 48
@@ -1015,6 +1029,33 @@ struct qed_common_ops {
  */
 	int (*set_led)(struct qed_dev *cdev,
 		       enum qed_led_mode mode);
+/**
+ * @brief db_recovery_add - add doorbell information to the doorbell
+ * recovery mechanism.
+ *
+ * @param cdev
+ * @param db_addr - doorbell address
+ * @param db_data - address of where db_data is stored
+ * @param db_is_32b - doorbell is 32b pr 64b
+ * @param db_is_user - doorbell recovery addresses are user or kernel space
+ */
+	int (*db_recovery_add)(struct qed_dev *cdev,
+			       void __iomem *db_addr,
+			       void *db_data,
+			       enum qed_db_rec_width db_width,
+			       enum qed_db_rec_space db_space);
+
+/**
+ * @brief db_recovery_del - remove doorbell information from the doorbell
+ * recovery mechanism. db_data serves as key (db_addr is not unique).
+ *
+ * @param cdev
+ * @param db_addr - doorbell address
+ * @param db_data - address where db_data is stored. Serves as key for the
+ *		    entry to delete.
+ */
+	int (*db_recovery_del)(struct qed_dev *cdev,
+			       void __iomem *db_addr, void *db_data);
 
 /**
  * @brief update_drv_state - API to inform the change in the driver state.