mpt2sas, mpt3sas: log exceeded temperature thresholds

This patch will log a message when driver receives "Temperature Threshold
exceeded" event from any temperature sensor.

The message will look similar to like:
mpt3sas0: Temperature Threshold flags a b c d exceeded for Sensor: x !!!
mpt3sas0: Current Temp In Celsius: y

where a b c d are threshold flags 0 1 2 3

Change_set:
1. Get the number of sensor count of this IOC by reading IO Unit page 8 at
   driver initialization time.
2. Also unmask the Temperature Threshold Event at driver initialization
   time
3. Whenever a MPI2_EVENT_TEMP_THRESHOLD event is received from the
   firmware, then print the sensor number, the maximum threshold number it
   has exceed and the current temperature of this sensor.
Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@avagotech.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

drivers/scsi/mpt2sas/mpt2sas_base.c

@@ -639,6 +639,9 @@ _base_display_event_data(struct MPT2SAS_ADAPTER *ioc,
 		if (!ioc->hide_ir_msg)
 			desc = "Log Entry Added";
 		break;
+	case MPI2_EVENT_TEMP_THRESHOLD:
+		desc = "Temperature Threshold";
+		break;
 	}
 
 	if (!desc)
@@ -2338,6 +2341,7 @@ _base_static_config_pages(struct MPT2SAS_ADAPTER *ioc)
 	mpt2sas_config_get_ioc_pg8(ioc, &mpi_reply, &ioc->ioc_pg8);
 	mpt2sas_config_get_iounit_pg0(ioc, &mpi_reply, &ioc->iounit_pg0);
 	mpt2sas_config_get_iounit_pg1(ioc, &mpi_reply, &ioc->iounit_pg1);
+	mpt2sas_config_get_iounit_pg8(ioc, &mpi_reply, &ioc->iounit_pg8);
 	_base_display_ioc_capabilities(ioc);
 
 	/*
@@ -2355,6 +2359,8 @@ _base_static_config_pages(struct MPT2SAS_ADAPTER *ioc)
 	ioc->iounit_pg1.Flags = cpu_to_le32(iounit_pg1_flags);
 	mpt2sas_config_set_iounit_pg1(ioc, &mpi_reply, &ioc->iounit_pg1);
 
+	if (ioc->iounit_pg8.NumSensors)
+		ioc->temp_sensors_count = ioc->iounit_pg8.NumSensors;
 }
 
 /**
@@ -4594,6 +4600,7 @@ mpt2sas_base_attach(struct MPT2SAS_ADAPTER *ioc)
 	_base_unmask_events(ioc, MPI2_EVENT_IR_PHYSICAL_DISK);
 	_base_unmask_events(ioc, MPI2_EVENT_IR_OPERATION_STATUS);
 	_base_unmask_events(ioc, MPI2_EVENT_LOG_ENTRY_ADDED);
+	_base_unmask_events(ioc, MPI2_EVENT_TEMP_THRESHOLD);
 	r = _base_make_ioc_operational(ioc, CAN_SLEEP);
 	if (r)
 		goto out_free_resources;

drivers/scsi/mpt2sas/mpt2sas_base.h

@@ -725,6 +725,7 @@ typedef void (*MPT2SAS_FLUSH_RUNNING_CMDS)(struct MPT2SAS_ADAPTER *ioc);
  * @ioc_pg8: static ioc page 8
  * @iounit_pg0: static iounit page 0
  * @iounit_pg1: static iounit page 1
+ * @iounit_pg8: static iounit page 8
  * @sas_hba: sas host object
  * @sas_expander_list: expander object list
  * @sas_node_lock:
@@ -795,6 +796,7 @@ typedef void (*MPT2SAS_FLUSH_RUNNING_CMDS)(struct MPT2SAS_ADAPTER *ioc);
  * @reply_post_host_index: head index in the pool where FW completes IO
  * @delayed_tr_list: target reset link list
  * @delayed_tr_volume_list: volume target reset link list
+ * @@temp_sensors_count: flag to carry the number of temperature sensors
  */
 struct MPT2SAS_ADAPTER {
 	struct list_head list;
@@ -892,6 +894,7 @@ struct MPT2SAS_ADAPTER {
 	Mpi2IOCPage8_t ioc_pg8;
 	Mpi2IOUnitPage0_t iounit_pg0;
 	Mpi2IOUnitPage1_t iounit_pg1;
+	Mpi2IOUnitPage8_t iounit_pg8;
 
 	struct _boot_device req_boot_device;
 	struct _boot_device req_alt_boot_device;
@@ -992,6 +995,7 @@ struct MPT2SAS_ADAPTER {
 
 	struct list_head delayed_tr_list;
 	struct list_head delayed_tr_volume_list;
+	u8		temp_sensors_count;
 
 	/* diag buffer support */
 	u8		*diag_buffer[MPI2_DIAG_BUF_TYPE_COUNT];
@@ -1120,6 +1124,8 @@ int mpt2sas_config_get_iounit_pg1(struct MPT2SAS_ADAPTER *ioc, Mpi2ConfigReply_t
     *mpi_reply, Mpi2IOUnitPage1_t *config_page);
 int mpt2sas_config_set_iounit_pg1(struct MPT2SAS_ADAPTER *ioc, Mpi2ConfigReply_t
     *mpi_reply, Mpi2IOUnitPage1_t *config_page);
+int mpt2sas_config_get_iounit_pg8(struct MPT2SAS_ADAPTER *ioc,
+	Mpi2ConfigReply_t *mpi_reply, Mpi2IOUnitPage8_t *config_page);
 int mpt2sas_config_get_iounit_pg3(struct MPT2SAS_ADAPTER *ioc,
 	Mpi2ConfigReply_t *mpi_reply, Mpi2IOUnitPage3_t *config_page, u16 sz);
 int mpt2sas_config_get_sas_iounit_pg1(struct MPT2SAS_ADAPTER *ioc, Mpi2ConfigReply_t

drivers/scsi/mpt2sas/mpt2sas_config.c

@@ -718,6 +718,42 @@ mpt2sas_config_get_iounit_pg3(struct MPT2SAS_ADAPTER *ioc,
 	return r;
 }
 
+/**
+ * mpt2sas_config_get_iounit_pg8 - obtain iounit page 8
+ * @ioc: per adapter object
+ * @mpi_reply: reply mf payload returned from firmware
+ * @config_page: contents of the config page
+ * Context: sleep.
+ *
+ * Returns 0 for success, non-zero for failure.
+ */
+int
+mpt2sas_config_get_iounit_pg8(struct MPT2SAS_ADAPTER *ioc,
+	Mpi2ConfigReply_t *mpi_reply, Mpi2IOUnitPage8_t *config_page)
+{
+	Mpi2ConfigRequest_t mpi_request;
+	int r;
+
+	memset(&mpi_request, 0, sizeof(Mpi2ConfigRequest_t));
+	mpi_request.Function = MPI2_FUNCTION_CONFIG;
+	mpi_request.Action = MPI2_CONFIG_ACTION_PAGE_HEADER;
+	mpi_request.Header.PageType = MPI2_CONFIG_PAGETYPE_IO_UNIT;
+	mpi_request.Header.PageNumber = 8;
+	mpi_request.Header.PageVersion = MPI2_IOUNITPAGE8_PAGEVERSION;
+	mpt2sas_base_build_zero_len_sge(ioc, &mpi_request.PageBufferSGE);
+	r = _config_request(ioc, &mpi_request, mpi_reply,
+	    MPT2_CONFIG_PAGE_DEFAULT_TIMEOUT, NULL, 0);
+	if (r)
+		goto out;
+
+	mpi_request.Action = MPI2_CONFIG_ACTION_PAGE_READ_CURRENT;
+	r = _config_request(ioc, &mpi_request, mpi_reply,
+	    MPT2_CONFIG_PAGE_DEFAULT_TIMEOUT, config_page,
+	    sizeof(*config_page));
+ out:
+	return r;
+}
+
 /**
  * mpt2sas_config_get_ioc_pg8 - obtain ioc page 8
  * @ioc: per adapter object

drivers/scsi/mpt2sas/mpt2sas_scsih.c

@@ -3646,6 +3646,31 @@ _scsih_check_volume_delete_events(struct MPT2SAS_ADAPTER *ioc,
 		    le16_to_cpu(event_data->VolDevHandle));
 }
 
+/**
+ * _scsih_temp_threshold_events - display temperature threshold exceeded events
+ * @ioc: per adapter object
+ * @event_data: the temp threshold event data
+ * Context: interrupt time.
+ *
+ * Return nothing.
+ */
+static void
+_scsih_temp_threshold_events(struct MPT2SAS_ADAPTER *ioc,
+	Mpi2EventDataTemperature_t *event_data)
+{
+	if (ioc->temp_sensors_count >= event_data->SensorNum) {
+		printk(MPT2SAS_ERR_FMT "Temperature Threshold flags %s%s%s%s"
+		  " exceeded for Sensor: %d !!!\n", ioc->name,
+		  ((le16_to_cpu(event_data->Status) & 0x1) == 1) ? "0 " : " ",
+		  ((le16_to_cpu(event_data->Status) & 0x2) == 2) ? "1 " : " ",
+		  ((le16_to_cpu(event_data->Status) & 0x4) == 4) ? "2 " : " ",
+		  ((le16_to_cpu(event_data->Status) & 0x8) == 8) ? "3 " : " ",
+		  event_data->SensorNum);
+		printk(MPT2SAS_ERR_FMT "Current Temp In Celsius: %d\n",
+			ioc->name, event_data->CurrentTemperature);
+	}
+}
+
 /**
  * _scsih_flush_running_cmds - completing outstanding commands.
  * @ioc: per adapter object
@@ -7557,6 +7582,12 @@ mpt2sas_scsih_event_callback(struct MPT2SAS_ADAPTER *ioc, u8 msix_index,
 	case MPI2_EVENT_IR_PHYSICAL_DISK:
 		break;
 
+	case MPI2_EVENT_TEMP_THRESHOLD:
+		_scsih_temp_threshold_events(ioc,
+			(Mpi2EventDataTemperature_t *)
+			mpi_reply->EventData);
+		break;
+
 	default: /* ignore the rest */
 		return;
 	}

drivers/scsi/mpt3sas/mpt3sas_base.c

@@ -619,6 +619,9 @@ _base_display_event_data(struct MPT3SAS_ADAPTER *ioc,
 	case MPI2_EVENT_LOG_ENTRY_ADDED:
 		desc = "Log Entry Added";
 		break;
+	case MPI2_EVENT_TEMP_THRESHOLD:
+		desc = "Temperature Threshold";
+		break;
 	}
 
 	if (!desc)
@@ -2500,6 +2503,7 @@ _base_static_config_pages(struct MPT3SAS_ADAPTER *ioc)
 	mpt3sas_config_get_ioc_pg8(ioc, &mpi_reply, &ioc->ioc_pg8);
 	mpt3sas_config_get_iounit_pg0(ioc, &mpi_reply, &ioc->iounit_pg0);
 	mpt3sas_config_get_iounit_pg1(ioc, &mpi_reply, &ioc->iounit_pg1);
+	mpt3sas_config_get_iounit_pg8(ioc, &mpi_reply, &ioc->iounit_pg8);
 	_base_display_ioc_capabilities(ioc);
 
 	/*
@@ -2516,6 +2520,9 @@ _base_static_config_pages(struct MPT3SAS_ADAPTER *ioc)
 		    MPI2_IOUNITPAGE1_DISABLE_TASK_SET_FULL_HANDLING;
 	ioc->iounit_pg1.Flags = cpu_to_le32(iounit_pg1_flags);
 	mpt3sas_config_set_iounit_pg1(ioc, &mpi_reply, &ioc->iounit_pg1);
+
+	if (ioc->iounit_pg8.NumSensors)
+		ioc->temp_sensors_count = ioc->iounit_pg8.NumSensors;
 }
 
 /**
@@ -4720,6 +4727,7 @@ mpt3sas_base_attach(struct MPT3SAS_ADAPTER *ioc)
 	_base_unmask_events(ioc, MPI2_EVENT_IR_PHYSICAL_DISK);
 	_base_unmask_events(ioc, MPI2_EVENT_IR_OPERATION_STATUS);
 	_base_unmask_events(ioc, MPI2_EVENT_LOG_ENTRY_ADDED);
+	_base_unmask_events(ioc, MPI2_EVENT_TEMP_THRESHOLD);
 
 	r = _base_make_ioc_operational(ioc, CAN_SLEEP);
 	if (r)

drivers/scsi/mpt3sas/mpt3sas_base.h

@@ -659,6 +659,7 @@ typedef void (*MPT3SAS_FLUSH_RUNNING_CMDS)(struct MPT3SAS_ADAPTER *ioc);
  * @ioc_pg8: static ioc page 8
  * @iounit_pg0: static iounit page 0
  * @iounit_pg1: static iounit page 1
+ * @iounit_pg8: static iounit page 8
  * @sas_hba: sas host object
  * @sas_expander_list: expander object list
  * @sas_node_lock:
@@ -728,6 +729,7 @@ typedef void (*MPT3SAS_FLUSH_RUNNING_CMDS)(struct MPT3SAS_ADAPTER *ioc);
  * @reply_post_host_index: head index in the pool where FW completes IO
  * @delayed_tr_list: target reset link list
  * @delayed_tr_volume_list: volume target reset link list
+ * @@temp_sensors_count: flag to carry the number of temperature sensors
  */
 struct MPT3SAS_ADAPTER {
 	struct list_head list;
@@ -834,6 +836,7 @@ struct MPT3SAS_ADAPTER {
 	Mpi2IOCPage8_t ioc_pg8;
 	Mpi2IOUnitPage0_t iounit_pg0;
 	Mpi2IOUnitPage1_t iounit_pg1;
+	Mpi2IOUnitPage8_t iounit_pg8;
 
 	struct _boot_device req_boot_device;
 	struct _boot_device req_alt_boot_device;
@@ -934,6 +937,7 @@ struct MPT3SAS_ADAPTER {
 
 	struct list_head delayed_tr_list;
 	struct list_head delayed_tr_volume_list;
+	u8		temp_sensors_count;
 
 	/* diag buffer support */
 	u8		*diag_buffer[MPI2_DIAG_BUF_TYPE_COUNT];
@@ -1082,6 +1086,8 @@ int mpt3sas_config_get_iounit_pg1(struct MPT3SAS_ADAPTER *ioc, Mpi2ConfigReply_t
 	*mpi_reply, Mpi2IOUnitPage1_t *config_page);
 int mpt3sas_config_set_iounit_pg1(struct MPT3SAS_ADAPTER *ioc, Mpi2ConfigReply_t
 	*mpi_reply, Mpi2IOUnitPage1_t *config_page);
+int mpt3sas_config_get_iounit_pg8(struct MPT3SAS_ADAPTER *ioc, Mpi2ConfigReply_t
+	*mpi_reply, Mpi2IOUnitPage8_t *config_page);
 int mpt3sas_config_get_sas_iounit_pg1(struct MPT3SAS_ADAPTER *ioc,
 	Mpi2ConfigReply_t *mpi_reply, Mpi2SasIOUnitPage1_t *config_page,
 	u16 sz);

drivers/scsi/mpt3sas/mpt3sas_config.c

@@ -870,6 +870,42 @@ mpt3sas_config_set_iounit_pg1(struct MPT3SAS_ADAPTER *ioc,
 	return r;
 }
 
+/**
+ * mpt3sas_config_get_iounit_pg8 - obtain iounit page 8
+ * @ioc: per adapter object
+ * @mpi_reply: reply mf payload returned from firmware
+ * @config_page: contents of the config page
+ * Context: sleep.
+ *
+ * Returns 0 for success, non-zero for failure.
+ */
+int
+mpt3sas_config_get_iounit_pg8(struct MPT3SAS_ADAPTER *ioc,
+	Mpi2ConfigReply_t *mpi_reply, Mpi2IOUnitPage8_t *config_page)
+{
+	Mpi2ConfigRequest_t mpi_request;
+	int r;
+
+	memset(&mpi_request, 0, sizeof(Mpi2ConfigRequest_t));
+	mpi_request.Function = MPI2_FUNCTION_CONFIG;
+	mpi_request.Action = MPI2_CONFIG_ACTION_PAGE_HEADER;
+	mpi_request.Header.PageType = MPI2_CONFIG_PAGETYPE_IO_UNIT;
+	mpi_request.Header.PageNumber = 8;
+	mpi_request.Header.PageVersion = MPI2_IOUNITPAGE8_PAGEVERSION;
+	ioc->build_zero_len_sge_mpi(ioc, &mpi_request.PageBufferSGE);
+	r = _config_request(ioc, &mpi_request, mpi_reply,
+	    MPT3_CONFIG_PAGE_DEFAULT_TIMEOUT, NULL, 0);
+	if (r)
+		goto out;
+
+	mpi_request.Action = MPI2_CONFIG_ACTION_PAGE_READ_CURRENT;
+	r = _config_request(ioc, &mpi_request, mpi_reply,
+	    MPT3_CONFIG_PAGE_DEFAULT_TIMEOUT, config_page,
+	    sizeof(*config_page));
+ out:
+	return r;
+}
+
 /**
  * mpt3sas_config_get_ioc_pg8 - obtain ioc page 8
  * @ioc: per adapter object

drivers/scsi/mpt3sas/mpt3sas_scsih.c

@@ -3341,6 +3341,31 @@ _scsih_check_volume_delete_events(struct MPT3SAS_ADAPTER *ioc,
 		    le16_to_cpu(event_data->VolDevHandle));
 }
 
+/**
+ * _scsih_temp_threshold_events - display temperature threshold exceeded events
+ * @ioc: per adapter object
+ * @event_data: the temp threshold event data
+ * Context: interrupt time.
+ *
+ * Return nothing.
+ */
+static void
+_scsih_temp_threshold_events(struct MPT3SAS_ADAPTER *ioc,
+	Mpi2EventDataTemperature_t *event_data)
+{
+	if (ioc->temp_sensors_count >= event_data->SensorNum) {
+		pr_err(MPT3SAS_FMT "Temperature Threshold flags %s%s%s%s"
+		  " exceeded for Sensor: %d !!!\n", ioc->name,
+		  ((le16_to_cpu(event_data->Status) & 0x1) == 1) ? "0 " : " ",
+		  ((le16_to_cpu(event_data->Status) & 0x2) == 2) ? "1 " : " ",
+		  ((le16_to_cpu(event_data->Status) & 0x4) == 4) ? "2 " : " ",
+		  ((le16_to_cpu(event_data->Status) & 0x8) == 8) ? "3 " : " ",
+		  event_data->SensorNum);
+		pr_err(MPT3SAS_FMT "Current Temp In Celsius: %d\n",
+			ioc->name, event_data->CurrentTemperature);
+	}
+}
+
 /**
  * _scsih_flush_running_cmds - completing outstanding commands.
  * @ioc: per adapter object
@@ -7194,6 +7219,12 @@ mpt3sas_scsih_event_callback(struct MPT3SAS_ADAPTER *ioc, u8 msix_index,
 	case MPI2_EVENT_IR_PHYSICAL_DISK:
 		break;
 
+	case MPI2_EVENT_TEMP_THRESHOLD:
+		_scsih_temp_threshold_events(ioc,
+			(Mpi2EventDataTemperature_t *)
+			mpi_reply->EventData);
+		break;
+
 	default: /* ignore the rest */
 		return 1;
 	}