From: Edwin Peer <edwin.peer@broadcom.com>
Date: Fri, 29 Oct 2021 03:47:46 -0400
Subject: bnxt_en: improve fw diagnose devlink health messages
Patch-mainline: v5.16-rc1
Git-commit: 8cc95ceb7087d6910050286301d05f4824a0bf59
References: jsc#SLE-18978

Add firmware event counters as well as health state severity. In
the unhealthy state, recommend a remedy and inform the user as to
its impact.

Readability of the devlink tool's output is negatively impacted by
adding these fields to the diagnosis. The single line of text, as
rendered by devlink health diagnose, benefits from more terse
descriptions, which can be substituted without loss of clarity, even
in pretty printed JSON mode.

Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c         |   19 ++-
 drivers/net/ethernet/broadcom/bnxt/bnxt.h         |   25 ++++
 drivers/net/ethernet/broadcom/bnxt/bnxt_devlink.c |  131 +++++++++++++++++-----
 3 files changed, 148 insertions(+), 27 deletions(-)

--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -2138,10 +2138,12 @@ static int bnxt_async_event_process(stru
 			set_bit(BNXT_STATE_FW_ACTIVATE_RESET, &bp->state);
 		} else if (EVENT_DATA1_RESET_NOTIFY_FATAL(data1)) {
 			type_str = "Fatal";
+			bp->fw_health->fatalities++;
 			set_bit(BNXT_STATE_FW_FATAL_COND, &bp->state);
 		} else if (data2 && BNXT_FW_STATUS_HEALTHY !=
 			   EVENT_DATA2_RESET_NOTIFY_FW_STATUS_CODE(data2)) {
 			type_str = "Non-fatal";
+			bp->fw_health->survivals++;
 			set_bit(BNXT_STATE_FW_NON_FATAL_COND, &bp->state);
 		}
 		netif_warn(bp, hw, bp->dev,
@@ -7604,6 +7606,7 @@ static int __bnxt_alloc_fw_health(struct
 	if (!bp->fw_health)
 		return -ENOMEM;
 
+	mutex_init(&bp->fw_health->lock);
 	return 0;
 }
 
@@ -7650,12 +7653,16 @@ static void bnxt_inv_fw_health_reg(struc
 	struct bnxt_fw_health *fw_health = bp->fw_health;
 	u32 reg_type;
 
-	if (!fw_health || !fw_health->status_reliable)
+	if (!fw_health)
 		return;
 
 	reg_type = BNXT_FW_HEALTH_REG_TYPE(fw_health->regs[BNXT_FW_HEALTH_REG]);
 	if (reg_type == BNXT_FW_HEALTH_REG_TYPE_GRC)
 		fw_health->status_reliable = false;
+
+	reg_type = BNXT_FW_HEALTH_REG_TYPE(fw_health->regs[BNXT_FW_RESET_CNT_REG]);
+	if (reg_type == BNXT_FW_HEALTH_REG_TYPE_GRC)
+		fw_health->resets_reliable = false;
 }
 
 static void bnxt_try_map_fw_health_reg(struct bnxt *bp)
@@ -7712,6 +7719,7 @@ static int bnxt_map_fw_health_regs(struc
 	int i;
 
 	bp->fw_health->status_reliable = false;
+	bp->fw_health->resets_reliable = false;
 	/* Only pre-map the monitoring GRC registers using window 3 */
 	for (i = 0; i < 4; i++) {
 		u32 reg = fw_health->regs[i];
@@ -7725,6 +7733,7 @@ static int bnxt_map_fw_health_regs(struc
 		fw_health->mapped_regs[i] = BNXT_FW_HEALTH_WIN_OFF(reg);
 	}
 	bp->fw_health->status_reliable = true;
+	bp->fw_health->resets_reliable = true;
 	if (reg_base == 0xffffffff)
 		return 0;
 
@@ -11264,14 +11273,18 @@ static void bnxt_fw_health_check(struct
 	}
 
 	val = bnxt_fw_health_readl(bp, BNXT_FW_HEARTBEAT_REG);
-	if (val == fw_health->last_fw_heartbeat)
+	if (val == fw_health->last_fw_heartbeat) {
+		fw_health->arrests++;
 		goto fw_reset;
+	}
 
 	fw_health->last_fw_heartbeat = val;
 
 	val = bnxt_fw_health_readl(bp, BNXT_FW_RESET_CNT_REG);
-	if (val != fw_health->last_fw_reset_cnt)
+	if (val != fw_health->last_fw_reset_cnt) {
+		fw_health->discoveries++;
 		goto fw_reset;
+	}
 
 	fw_health->tmr_counter = fw_health->tmr_multiplier;
 	return;
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
@@ -1523,6 +1523,21 @@ struct bnxt_ctx_mem_info {
 	struct bnxt_mem_init	mem_init[BNXT_CTX_MEM_INIT_MAX];
 };
 
+enum bnxt_health_severity {
+	SEVERITY_NORMAL = 0,
+	SEVERITY_WARNING,
+	SEVERITY_RECOVERABLE,
+	SEVERITY_FATAL,
+};
+
+enum bnxt_health_remedy {
+	REMEDY_DEVLINK_RECOVER,
+	REMEDY_POWER_CYCLE_DEVICE,
+	REMEDY_POWER_CYCLE_HOST,
+	REMEDY_FW_UPDATE,
+	REMEDY_HW_REPLACE,
+};
+
 struct bnxt_fw_health {
 	u32 flags;
 	u32 polling_dsecs;
@@ -1542,6 +1557,7 @@ struct bnxt_fw_health {
 	u8 enabled:1;
 	u8 primary:1;
 	u8 status_reliable:1;
+	u8 resets_reliable:1;
 	u8 tmr_multiplier;
 	u8 tmr_counter;
 	u8 fw_reset_seq_cnt;
@@ -1551,6 +1567,15 @@ struct bnxt_fw_health {
 	u32 echo_req_data1;
 	u32 echo_req_data2;
 	struct devlink_health_reporter	*fw_reporter;
+	/* Protects severity and remedy */
+	struct mutex lock;
+	enum bnxt_health_severity severity;
+	enum bnxt_health_remedy remedy;
+	u32 arrests;
+	u32 discoveries;
+	u32 survivals;
+	u32 fatalities;
+	u32 diagnoses;
 };
 
 #define BNXT_FW_HEALTH_REG_TYPE_MASK	3
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_devlink.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_devlink.c
@@ -71,43 +71,110 @@ static int bnxt_hwrm_remote_dev_reset_se
 	return hwrm_req_send(bp, req);
 }
 
+static char *bnxt_health_severity_str(enum bnxt_health_severity severity)
+{
+	switch (severity) {
+	case SEVERITY_NORMAL: return "normal";
+	case SEVERITY_WARNING: return "warning";
+	case SEVERITY_RECOVERABLE: return "recoverable";
+	case SEVERITY_FATAL: return "fatal";
+	default: return "unknown";
+	}
+}
+
+static char *bnxt_health_remedy_str(enum bnxt_health_remedy remedy)
+{
+	switch (remedy) {
+	case REMEDY_DEVLINK_RECOVER: return "devlink recover";
+	case REMEDY_POWER_CYCLE_DEVICE: return "device power cycle";
+	case REMEDY_POWER_CYCLE_HOST: return "host power cycle";
+	case REMEDY_FW_UPDATE: return "update firmware";
+	case REMEDY_HW_REPLACE: return "replace hardware";
+	default: return "unknown";
+	}
+}
+
 static int bnxt_fw_diagnose(struct devlink_health_reporter *reporter,
 			    struct devlink_fmsg *fmsg,
 			    struct netlink_ext_ack *extack)
 {
 	struct bnxt *bp = devlink_health_reporter_priv(reporter);
-	u32 val;
+	struct bnxt_fw_health *h = bp->fw_health;
+	u32 fw_status, fw_resets;
 	int rc;
 
 	if (test_bit(BNXT_STATE_IN_FW_RESET, &bp->state))
-		return 0;
+		return devlink_fmsg_string_pair_put(fmsg, "Status", "recovering");
 
-	val = bnxt_fw_health_readl(bp, BNXT_FW_HEALTH_REG);
+	if (!h->status_reliable)
+		return devlink_fmsg_string_pair_put(fmsg, "Status", "unknown");
 
-	if (BNXT_FW_IS_BOOTING(val)) {
-		rc = devlink_fmsg_string_pair_put(fmsg, "Description",
-						  "Not yet completed initialization");
+	mutex_lock(&h->lock);
+	fw_status = bnxt_fw_health_readl(bp, BNXT_FW_HEALTH_REG);
+	if (BNXT_FW_IS_BOOTING(fw_status)) {
+		rc = devlink_fmsg_string_pair_put(fmsg, "Status", "initializing");
 		if (rc)
-			return rc;
-	} else if (BNXT_FW_IS_ERR(val)) {
-		rc = devlink_fmsg_string_pair_put(fmsg, "Description",
-						  "Encountered fatal error and cannot recover");
+			goto unlock;
+	} else if (h->severity || fw_status != BNXT_FW_STATUS_HEALTHY) {
+		if (!h->severity) {
+			h->severity = SEVERITY_FATAL;
+			h->remedy = REMEDY_POWER_CYCLE_DEVICE;
+			h->diagnoses++;
+			devlink_health_report(h->fw_reporter,
+					      "FW error diagnosed", h);
+		}
+		rc = devlink_fmsg_string_pair_put(fmsg, "Status", "error");
 		if (rc)
-			return rc;
+			goto unlock;
+		rc = devlink_fmsg_u32_pair_put(fmsg, "Syndrome", fw_status);
+		if (rc)
+			goto unlock;
+	} else {
+		rc = devlink_fmsg_string_pair_put(fmsg, "Status", "healthy");
+		if (rc)
+			goto unlock;
 	}
 
-	if (val >> 16) {
-		rc = devlink_fmsg_u32_pair_put(fmsg, "Error code", val >> 16);
+	rc = devlink_fmsg_string_pair_put(fmsg, "Severity",
+					  bnxt_health_severity_str(h->severity));
+	if (rc)
+		goto unlock;
+
+	if (h->severity) {
+		rc = devlink_fmsg_string_pair_put(fmsg, "Remedy",
+						  bnxt_health_remedy_str(h->remedy));
 		if (rc)
-			return rc;
+			goto unlock;
+		if (h->remedy == REMEDY_DEVLINK_RECOVER) {
+			rc = devlink_fmsg_string_pair_put(fmsg, "Impact",
+							  "traffic+ntuple_cfg");
+			if (rc)
+				goto unlock;
+		}
 	}
 
-	val = bnxt_fw_health_readl(bp, BNXT_FW_RESET_CNT_REG);
-	rc = devlink_fmsg_u32_pair_put(fmsg, "Reset count", val);
-	if (rc)
+unlock:
+	mutex_unlock(&h->lock);
+	if (rc || !h->resets_reliable)
 		return rc;
 
-	return 0;
+	fw_resets = bnxt_fw_health_readl(bp, BNXT_FW_RESET_CNT_REG);
+	rc = devlink_fmsg_u32_pair_put(fmsg, "Resets", fw_resets);
+	if (rc)
+		return rc;
+	rc = devlink_fmsg_u32_pair_put(fmsg, "Arrests", h->arrests);
+	if (rc)
+		return rc;
+	rc = devlink_fmsg_u32_pair_put(fmsg, "Survivals", h->survivals);
+	if (rc)
+		return rc;
+	rc = devlink_fmsg_u32_pair_put(fmsg, "Discoveries", h->discoveries);
+	if (rc)
+		return rc;
+	rc = devlink_fmsg_u32_pair_put(fmsg, "Fatalities", h->fatalities);
+	if (rc)
+		return rc;
+	return devlink_fmsg_u32_pair_put(fmsg, "Diagnoses", h->diagnoses);
 }
 
 static int bnxt_fw_recover(struct devlink_health_reporter *reporter,
@@ -116,6 +183,9 @@ static int bnxt_fw_recover(struct devlin
 {
 	struct bnxt *bp = devlink_health_reporter_priv(reporter);
 
+	if (bp->fw_health->severity == SEVERITY_FATAL)
+		return -ENODEV;
+
 	set_bit(BNXT_STATE_RECOVER, &bp->state);
 	__bnxt_fw_recover(bp);
 
@@ -165,6 +235,7 @@ void bnxt_dl_fw_reporters_destroy(struct
 void bnxt_devlink_health_fw_report(struct bnxt *bp)
 {
 	struct bnxt_fw_health *fw_health = bp->fw_health;
+	int rc;
 
 	if (!fw_health)
 		return;
@@ -174,20 +245,32 @@ void bnxt_devlink_health_fw_report(struc
 		return;
 	}
 
-	devlink_health_report(fw_health->fw_reporter, "FW error reported", NULL);
+	mutex_lock(&fw_health->lock);
+	fw_health->severity = SEVERITY_RECOVERABLE;
+	fw_health->remedy = REMEDY_DEVLINK_RECOVER;
+	mutex_unlock(&fw_health->lock);
+	rc = devlink_health_report(fw_health->fw_reporter, "FW error reported",
+				   fw_health);
+	if (rc == -ECANCELED)
+		__bnxt_fw_recover(bp);
 }
 
 void bnxt_dl_health_fw_status_update(struct bnxt *bp, bool healthy)
 {
-	struct bnxt_fw_health *health = bp->fw_health;
+	struct bnxt_fw_health *fw_health = bp->fw_health;
 	u8 state;
 
-	if (healthy)
+	mutex_lock(&fw_health->lock);
+	if (healthy) {
+		fw_health->severity = SEVERITY_NORMAL;
 		state = DEVLINK_HEALTH_REPORTER_STATE_HEALTHY;
-	else
+	} else {
+		fw_health->severity = SEVERITY_FATAL;
+		fw_health->remedy = REMEDY_POWER_CYCLE_DEVICE;
 		state = DEVLINK_HEALTH_REPORTER_STATE_ERROR;
-
-	devlink_health_reporter_state_update(health->fw_reporter, state);
+	}
+	mutex_unlock(&fw_health->lock);
+	devlink_health_reporter_state_update(fw_health->fw_reporter, state);
 }
 
 void bnxt_dl_health_fw_recovery_done(struct bnxt *bp)