diff mbox series

[net-next,12/12] bnxt_en: Event handler for Thermal event

Message ID 20230815045658.80494-13-michael.chan@broadcom.com (mailing list archive)
State Changes Requested
Headers show
Series None | expand

Commit Message

Michael Chan Aug. 15, 2023, 4:56 a.m. UTC 
From: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>

Newer FW will send a new async event when it detects that
the chip's temperature has crossed the configured threshold value.
The driver will now notify hwmon and will log a warning message.

Cc: Jean Delvare <jdelvare@suse.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: linux-hwmon@vger.kernel.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 44 +++++++++++++++++++
 .../net/ethernet/broadcom/bnxt/bnxt_hwmon.c   |  8 ++++
 .../net/ethernet/broadcom/bnxt/bnxt_hwmon.h   |  5 +++
 3 files changed, 57 insertions(+)

Comments

Guenter Roeck Aug. 15, 2023, 3:16 p.m. UTC | #1 
On Mon, Aug 14, 2023 at 09:56:58PM -0700, Michael Chan wrote:
> From: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
> 
> Newer FW will send a new async event when it detects that
> the chip's temperature has crossed the configured threshold value.
> The driver will now notify hwmon and will log a warning message.
> 
> Cc: Jean Delvare <jdelvare@suse.com>
> Cc: Guenter Roeck <linux@roeck-us.net>
> Cc: linux-hwmon@vger.kernel.org
> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
> Signed-off-by: Michael Chan <michael.chan@broadcom.com>
> ---
>  drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 44 +++++++++++++++++++
>  .../net/ethernet/broadcom/bnxt/bnxt_hwmon.c   |  8 ++++
>  .../net/ethernet/broadcom/bnxt/bnxt_hwmon.h   |  5 +++
>  3 files changed, 57 insertions(+)
> 
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> index 5e97a3d93e87..c8e04c9501ee 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
> @@ -2130,6 +2130,17 @@ static u16 bnxt_agg_ring_id_to_grp_idx(struct bnxt *bp, u16 ring_id)
>  	return INVALID_HW_RING_ID;
>  }
>  
> +#define BNXT_EVENT_THERMAL_CURRENT_TEMP(data2)				\
> +	((data2) & ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA2_CURRENT_TEMP_MASK)
> +
> +#define BNXT_EVENT_THERMAL_THRESHOLD_TEMP(data2)					\
> +	(((data2) &									\
> +	  ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA2_THRESHOLD_TEMP_MASK) >>	\
> +	 ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA2_THRESHOLD_TEMP_SFT)
> +
> +#define EVENT_DATA1_THERMAL_THRESHOLD_TYPE(data1)			\
> +	((data1) & ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_MASK)
> +
>  static void bnxt_event_error_report(struct bnxt *bp, u32 data1, u32 data2)
>  {
>  	u32 err_type = BNXT_EVENT_ERROR_REPORT_TYPE(data1);
> @@ -2145,6 +2156,39 @@ static void bnxt_event_error_report(struct bnxt *bp, u32 data1, u32 data2)
>  	case ASYNC_EVENT_CMPL_ERROR_REPORT_BASE_EVENT_DATA1_ERROR_TYPE_DOORBELL_DROP_THRESHOLD:
>  		netdev_warn(bp->dev, "One or more MMIO doorbells dropped by the device!\n");
>  		break;
> +	case ASYNC_EVENT_CMPL_ERROR_REPORT_BASE_EVENT_DATA1_ERROR_TYPE_THERMAL_THRESHOLD: {
> +		char *threshold_type;
> +		u32 attr;
> +
> +		switch (EVENT_DATA1_THERMAL_THRESHOLD_TYPE(data1)) {
> +		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_WARN:
> +			attr = hwmon_temp_lcrit_alarm;

As with previous patch, wrong attribute

> +			threshold_type = "warning";
> +			break;
> +		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_CRITICAL:
> +			attr = hwmon_temp_crit_alarm;
> +			threshold_type = "critical";
> +			break;
> +		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_FATAL:
> +			attr = hwmon_temp_emergency_alarm;
> +			threshold_type = "fatal";
> +			break;
> +		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_SHUTDOWN:
> +			attr = hwmon_temp_max_alarm;

Same here.

Overall it seems to me it would be better to keep hwmon internals out
of this file and just pass the threshold type to the hwmon code.

> +			threshold_type = "shutdown";
> +			break;
> +		default:
> +			netdev_err(bp->dev, "Unknown Thermal threshold type event\n");
> +			return;
> +		}
> +		netdev_warn(bp->dev, "Chip temperature has crossed the %s thermal threshold!\n",
> +			    threshold_type);
> +		netdev_warn(bp->dev, "Temperature (In Celsius), Current: %lu, threshold: %lu\n",
> +			    BNXT_EVENT_THERMAL_CURRENT_TEMP(data2),
> +			    BNXT_EVENT_THERMAL_THRESHOLD_TEMP(data2));
> +		bnxt_hwmon_notify_event(bp, attr);
> +		break;
> +	}
>  	default:
>  		netdev_err(bp->dev, "FW reported unknown error type %u\n",
>  			   err_type);
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c
> index f5affac1169a..483571264276 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c
> @@ -18,6 +18,14 @@
>  #include "bnxt_hwrm.h"
>  #include "bnxt_hwmon.h"
>  
> +void bnxt_hwmon_notify_event(struct bnxt *bp, u32 attr)
> +{
> +	if (!bp->hwmon_dev)
> +		return;
> +
> +	hwmon_notify_event(&bp->pdev->dev, hwmon_temp, attr, 0);
> +}
> +
>  static int bnxt_hwrm_temp_query(struct bnxt *bp, u8 *temp)
>  {
>  	struct hwrm_temp_monitor_query_output *resp;
> diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h
> index af310066687c..5cf127702764 100644
> --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h
> +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h
> @@ -11,9 +11,14 @@
>  #define BNXT_HWMON_H
>  
>  #ifdef CONFIG_BNXT_HWMON
> +void bnxt_hwmon_notify_event(struct bnxt *bp, u32 attr);
>  void bnxt_hwmon_uninit(struct bnxt *bp);
>  void bnxt_hwmon_init(struct bnxt *bp);
>  #else
> +static inline void bnxt_hwmon_notify_event(struct bnxt *bp, u32 attr)
> +{
> +}
> +
>  static inline void bnxt_hwmon_uninit(struct bnxt *bp)
>  {
>  }
> -- 
> 2.30.1
>
diff mbox series

Patch

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index 5e97a3d93e87..c8e04c9501ee 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -2130,6 +2130,17 @@  static u16 bnxt_agg_ring_id_to_grp_idx(struct bnxt *bp, u16 ring_id)
 	return INVALID_HW_RING_ID;
 }
 
+#define BNXT_EVENT_THERMAL_CURRENT_TEMP(data2)				\
+	((data2) & ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA2_CURRENT_TEMP_MASK)
+
+#define BNXT_EVENT_THERMAL_THRESHOLD_TEMP(data2)					\
+	(((data2) &									\
+	  ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA2_THRESHOLD_TEMP_MASK) >>	\
+	 ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA2_THRESHOLD_TEMP_SFT)
+
+#define EVENT_DATA1_THERMAL_THRESHOLD_TYPE(data1)			\
+	((data1) & ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_MASK)
+
 static void bnxt_event_error_report(struct bnxt *bp, u32 data1, u32 data2)
 {
 	u32 err_type = BNXT_EVENT_ERROR_REPORT_TYPE(data1);
@@ -2145,6 +2156,39 @@  static void bnxt_event_error_report(struct bnxt *bp, u32 data1, u32 data2)
 	case ASYNC_EVENT_CMPL_ERROR_REPORT_BASE_EVENT_DATA1_ERROR_TYPE_DOORBELL_DROP_THRESHOLD:
 		netdev_warn(bp->dev, "One or more MMIO doorbells dropped by the device!\n");
 		break;
+	case ASYNC_EVENT_CMPL_ERROR_REPORT_BASE_EVENT_DATA1_ERROR_TYPE_THERMAL_THRESHOLD: {
+		char *threshold_type;
+		u32 attr;
+
+		switch (EVENT_DATA1_THERMAL_THRESHOLD_TYPE(data1)) {
+		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_WARN:
+			attr = hwmon_temp_lcrit_alarm;
+			threshold_type = "warning";
+			break;
+		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_CRITICAL:
+			attr = hwmon_temp_crit_alarm;
+			threshold_type = "critical";
+			break;
+		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_FATAL:
+			attr = hwmon_temp_emergency_alarm;
+			threshold_type = "fatal";
+			break;
+		case ASYNC_EVENT_CMPL_ERROR_REPORT_THERMAL_EVENT_DATA1_THRESHOLD_TYPE_SHUTDOWN:
+			attr = hwmon_temp_max_alarm;
+			threshold_type = "shutdown";
+			break;
+		default:
+			netdev_err(bp->dev, "Unknown Thermal threshold type event\n");
+			return;
+		}
+		netdev_warn(bp->dev, "Chip temperature has crossed the %s thermal threshold!\n",
+			    threshold_type);
+		netdev_warn(bp->dev, "Temperature (In Celsius), Current: %lu, threshold: %lu\n",
+			    BNXT_EVENT_THERMAL_CURRENT_TEMP(data2),
+			    BNXT_EVENT_THERMAL_THRESHOLD_TEMP(data2));
+		bnxt_hwmon_notify_event(bp, attr);
+		break;
+	}
 	default:
 		netdev_err(bp->dev, "FW reported unknown error type %u\n",
 			   err_type);
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c
index f5affac1169a..483571264276 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.c
@@ -18,6 +18,14 @@ 
 #include "bnxt_hwrm.h"
 #include "bnxt_hwmon.h"
 
+void bnxt_hwmon_notify_event(struct bnxt *bp, u32 attr)
+{
+	if (!bp->hwmon_dev)
+		return;
+
+	hwmon_notify_event(&bp->pdev->dev, hwmon_temp, attr, 0);
+}
+
 static int bnxt_hwrm_temp_query(struct bnxt *bp, u8 *temp)
 {
 	struct hwrm_temp_monitor_query_output *resp;
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h
index af310066687c..5cf127702764 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_hwmon.h
@@ -11,9 +11,14 @@ 
 #define BNXT_HWMON_H
 
 #ifdef CONFIG_BNXT_HWMON
+void bnxt_hwmon_notify_event(struct bnxt *bp, u32 attr);
 void bnxt_hwmon_uninit(struct bnxt *bp);
 void bnxt_hwmon_init(struct bnxt *bp);
 #else
+static inline void bnxt_hwmon_notify_event(struct bnxt *bp, u32 attr)
+{
+}
+
 static inline void bnxt_hwmon_uninit(struct bnxt *bp)
 {
 }