lpfc: Add Lancer Temperature Event support to the lpfc driver

PATCH In review James Smart 2 participants 4 messages Feb 5, 2015 – Feb 17, 2015

Subsystems (2)

SCSI Emulex/Broadcom lpfc FC/FCoE SCSI driver

lore

Share & tools

lore

James Smart

author

2015-02-05 19:23 UTC #0 lore raw

PATCH 2/21 lpfc: Add Lancer Temperature Event support to the lpfc driver

---
 drivers/scsi/lpfc/lpfc_hw4.h  |   1 +
 drivers/scsi/lpfc/lpfc_init.c | 179 +++++++++++++++++++++++++++++-------------
 2 files changed, 125 insertions(+), 55 deletions(-)

drivers/scsi/lpfc/lpfc_hw4.h +1 -0

index f432ec1..3121ec4 100644
@@ -3244,6 +3244,7 @@ struct lpfc_acqe_sli {
 #define LPFC_SLI_EVENT_TYPE_NVLOG_POST		0x4
 #define LPFC_SLI_EVENT_TYPE_DIAG_DUMP		0x5
 #define LPFC_SLI_EVENT_TYPE_MISCONFIGURED	0x9
+#define LPFC_SLI_EVENT_TYPE_REMOTE_DPORT	0xA
 };
 
 /*

drivers/scsi/lpfc/lpfc_init.c +124 ~~-55~~

261 lines not on this page · raw message

Signature · 4 lines

1.7.11.7

Signed-off-by: James Smart <[email protected]>
Signed-off-by: Dick Kennedy <[email protected]>

Tomas Henzl

2015-02-06 12:16 UTC #1 lore raw

Quoted 22 lines · James Smart

---
 drivers/scsi/lpfc/lpfc_hw4.h  |   1 +
 drivers/scsi/lpfc/lpfc_init.c | 179 +++++++++++++++++++++++++++++-------------
 2 files changed, 125 insertions(+), 55 deletions(-)

drivers/scsi/lpfc/lpfc_hw4.h +1 -0

index f432ec1..3121ec4 100644
@@ -3244,6 +3244,7 @@ struct lpfc_acqe_sli {
 #define LPFC_SLI_EVENT_TYPE_NVLOG_POST		0x4
 #define LPFC_SLI_EVENT_TYPE_DIAG_DUMP		0x5
 #define LPFC_SLI_EVENT_TYPE_MISCONFIGURED	0x9
+#define LPFC_SLI_EVENT_TYPE_REMOTE_DPORT	0xA
 };
 
 /*

drivers/scsi/lpfc/lpfc_init.c +0 -0

index 2b5b910..4ba91af 100644

drivers/scsi/lpfc/lpfc_init.c lpfc_offline_eratt(struct lpfc_hba *phba) · line 1335

Quoted 6 lines

drivers/scsi/lpfc/lpfc_init.c +3 -0

@@ -1330,13 +1330,14 @@ lpfc_offline_eratt(struct lpfc_hba *phba)
 void
 lpfc_sli4_offline_eratt(struct lpfc_hba *phba)
 {
+	spin_lock_irq(&phba->hbalock);
+	phba->link_state = LPFC_HBA_ERROR;
+	spin_unlock_irq(&phba->hbalock);

Hi James,
please explain why is the spinlock^ needed?
There seems to be lot of other places where link_state is not protected, for example it is evaluated in lpfc_sli4_handle_received_buffer with no lock.

Could you please also add some more description to the of the body your mails, sometimes a subject only is not enough.

Thanks,
Tomas

Quoted 263 lines

drivers/scsi/lpfc/lpfc_init.c +121 ~~-55~~

+
 	lpfc_offline_prep(phba, LPFC_MBX_NO_WAIT);
 	lpfc_offline(phba);
-	lpfc_sli4_brdreset(phba);
 	lpfc_hba_down_post(phba);
-	lpfc_sli4_post_status_check(phba);
 	lpfc_unblock_mgmt_io(phba);
-	phba->link_state = LPFC_HBA_ERROR;
 }
 
 /**

@@ -1629,6 +1630,7 @@ lpfc_handle_eratt_s4(struct lpfc_hba *phba)
 	uint32_t uerrlo_reg, uemasklo_reg;
 	uint32_t pci_rd_rc1, pci_rd_rc2;
 	bool en_rn_msg = true;
+	struct temp_event temp_event_data;
 	int rc;
 
 	/* If the pci channel is offline, ignore possible errors, since

@@ -1636,9 +1638,6 @@ lpfc_handle_eratt_s4(struct lpfc_hba *phba)
 	 */
 	if (pci_channel_offline(phba->pcidev))
 		return;
-	/* If resets are disabled then leave the HBA alone and return */
-	if (!phba->cfg_enable_hba_reset)
-		return;
 
 	if_type = bf_get(lpfc_sli_intf_if_type, &phba->sli4_hba.sli_intf);
 	switch (if_type) {

@@ -1654,6 +1653,7 @@ lpfc_handle_eratt_s4(struct lpfc_hba *phba)
 			return;
 		lpfc_sli4_offline_eratt(phba);
 		break;
+
 	case LPFC_SLI_INTF_IF_TYPE_2:
 		pci_rd_rc1 = lpfc_readl(
 				phba->sli4_hba.u.if_type2.STATUSregaddr,

@@ -1668,15 +1668,27 @@ lpfc_handle_eratt_s4(struct lpfc_hba *phba)
 		reg_err1 = readl(phba->sli4_hba.u.if_type2.ERR1regaddr);
 		reg_err2 = readl(phba->sli4_hba.u.if_type2.ERR2regaddr);
 		if (bf_get(lpfc_sliport_status_oti, &portstat_reg)) {
-			/* TODO: Register for Overtemp async events. */
 			lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
 				"2889 Port Overtemperature event, "
-				"taking port offline\n");
+				"taking port offline Data: x%x x%x\n",
+				reg_err1, reg_err2);
+
+			temp_event_data.event_type = FC_REG_TEMPERATURE_EVENT;
+			temp_event_data.event_code = LPFC_CRIT_TEMP;
+			temp_event_data.data = 0xFFFFFFFF;
+
+			shost = lpfc_shost_from_vport(phba->pport);
+			fc_host_post_vendor_event(shost, fc_get_event_number(),
+						  sizeof(temp_event_data),
+						  (char *)&temp_event_data,
+						  SCSI_NL_VID_TYPE_PCI
+						  | PCI_VENDOR_ID_EMULEX);
+
 			spin_lock_irq(&phba->hbalock);
 			phba->over_temp_state = HBA_OVER_TEMP;
 			spin_unlock_irq(&phba->hbalock);
 			lpfc_sli4_offline_eratt(phba);
-			break;
+			return;
 		}
 		if (reg_err1 == SLIPORT_ERR1_REG_ERR_CODE_2 &&
 		    reg_err2 == SLIPORT_ERR2_REG_FW_RESTART) {

@@ -1693,6 +1705,10 @@ lpfc_handle_eratt_s4(struct lpfc_hba *phba)
 			lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
 					"3145 Port Down: Provisioning\n");
 
+		/* If resets are disabled then leave the HBA alone and return */
+		if (!phba->cfg_enable_hba_reset)
+			return;
+
 		/* Check port status register for function reset */
 		rc = lpfc_sli4_port_sta_fn_reset(phba, LPFC_MBX_NO_WAIT,
 				en_rn_msg);

@@ -4044,18 +4060,21 @@ lpfc_sli4_async_sli_evt(struct lpfc_hba *phba, struct lpfc_acqe_sli *acqe_sli)
 	char port_name;
 	char message[128];
 	uint8_t status;
+	uint8_t evt_type;
+	struct temp_event temp_event_data;
 	struct lpfc_acqe_misconfigured_event *misconfigured;
+	struct Scsi_Host  *shost;
+
+	evt_type = bf_get(lpfc_trailer_type, acqe_sli);
 
-	/* special case misconfigured event as it contains data for all ports */
-	if ((bf_get(lpfc_sli_intf_if_type, &phba->sli4_hba.sli_intf) !=
-		 LPFC_SLI_INTF_IF_TYPE_2) ||
-		(bf_get(lpfc_trailer_type, acqe_sli) !=
-			LPFC_SLI_EVENT_TYPE_MISCONFIGURED)) {
+	/* Special case Lancer */
+	if (bf_get(lpfc_sli_intf_if_type, &phba->sli4_hba.sli_intf) !=
+		 LPFC_SLI_INTF_IF_TYPE_2) {
 		lpfc_printf_log(phba, KERN_INFO, LOG_SLI,
 				"2901 Async SLI event - Event Data1:x%08x Event Data2:"
 				"x%08x SLI Event Type:%d\n",
 				acqe_sli->event_data1, acqe_sli->event_data2,
-				bf_get(lpfc_trailer_type, acqe_sli));
+				evt_type);
 		return;
 	}

@@ -4063,58 +4082,107 @@ lpfc_sli4_async_sli_evt(struct lpfc_hba *phba, struct lpfc_acqe_sli *acqe_sli)
 	if (port_name == 0x00)
 		port_name = '?'; /* get port name is empty */
 
-	misconfigured = (struct lpfc_acqe_misconfigured_event *)
+	switch (evt_type) {
+	case LPFC_SLI_EVENT_TYPE_OVER_TEMP:
+		temp_event_data.event_type = FC_REG_TEMPERATURE_EVENT;
+		temp_event_data.event_code = LPFC_THRESHOLD_TEMP;
+		temp_event_data.data = (uint32_t)acqe_sli->event_data1;
+
+		lpfc_printf_log(phba, KERN_WARNING, LOG_SLI,
+				"3190 Over Temperature:%d Celsius- Port Name %c\n",
+				acqe_sli->event_data1, port_name);
+
+		shost = lpfc_shost_from_vport(phba->pport);
+		fc_host_post_vendor_event(shost, fc_get_event_number(),
+					  sizeof(temp_event_data),
+					  (char *)&temp_event_data,
+					  SCSI_NL_VID_TYPE_PCI
+					  | PCI_VENDOR_ID_EMULEX);
+		break;
+	case LPFC_SLI_EVENT_TYPE_NORM_TEMP:
+		temp_event_data.event_type = FC_REG_TEMPERATURE_EVENT;
+		temp_event_data.event_code = LPFC_NORMAL_TEMP;
+		temp_event_data.data = (uint32_t)acqe_sli->event_data1;
+
+		lpfc_printf_log(phba, KERN_INFO, LOG_SLI,
+				"3191 Normal Temperature:%d Celsius - Port Name %c\n",
+				acqe_sli->event_data1, port_name);
+
+		shost = lpfc_shost_from_vport(phba->pport);
+		fc_host_post_vendor_event(shost, fc_get_event_number(),
+					  sizeof(temp_event_data),
+					  (char *)&temp_event_data,
+					  SCSI_NL_VID_TYPE_PCI
+					  | PCI_VENDOR_ID_EMULEX);
+		break;
+	case LPFC_SLI_EVENT_TYPE_MISCONFIGURED:
+		misconfigured = (struct lpfc_acqe_misconfigured_event *)
 					&acqe_sli->event_data1;
 
-	/* fetch the status for this port */
-	switch (phba->sli4_hba.lnk_info.lnk_no) {
-	case LPFC_LINK_NUMBER_0:
-		status = bf_get(lpfc_sli_misconfigured_port0,
+		/* fetch the status for this port */
+		switch (phba->sli4_hba.lnk_info.lnk_no) {
+		case LPFC_LINK_NUMBER_0:
+			status = bf_get(lpfc_sli_misconfigured_port0,
 					&misconfigured->theEvent);
-		break;
-	case LPFC_LINK_NUMBER_1:
-		status = bf_get(lpfc_sli_misconfigured_port1,
+			break;
+		case LPFC_LINK_NUMBER_1:
+			status = bf_get(lpfc_sli_misconfigured_port1,
 					&misconfigured->theEvent);
-		break;
-	case LPFC_LINK_NUMBER_2:
-		status = bf_get(lpfc_sli_misconfigured_port2,
+			break;
+		case LPFC_LINK_NUMBER_2:
+			status = bf_get(lpfc_sli_misconfigured_port2,
 					&misconfigured->theEvent);
-		break;
-	case LPFC_LINK_NUMBER_3:
-		status = bf_get(lpfc_sli_misconfigured_port3,
+			break;
+		case LPFC_LINK_NUMBER_3:
+			status = bf_get(lpfc_sli_misconfigured_port3,
 					&misconfigured->theEvent);
-		break;
-	default:
-		status = ~LPFC_SLI_EVENT_STATUS_VALID;
-		break;
-	}
+			break;
+		default:
+			status = ~LPFC_SLI_EVENT_STATUS_VALID;
+			break;
+		}
 
-	switch (status) {
-	case LPFC_SLI_EVENT_STATUS_VALID:
-		return; /* no message if the sfp is okay */
-	case LPFC_SLI_EVENT_STATUS_NOT_PRESENT:
-		sprintf(message, "Optics faulted/incorrectly installed/not " \
-				"installed - Reseat optics, if issue not "
-				"resolved, replace.");
-		break;
-	case LPFC_SLI_EVENT_STATUS_WRONG_TYPE:
-		sprintf(message,
-			"Optics of two types installed - Remove one optic or " \
-			"install matching pair of optics.");
-		break;
-	case LPFC_SLI_EVENT_STATUS_UNSUPPORTED:
-		sprintf(message, "Incompatible optics - Replace with " \
+		switch (status) {
+		case LPFC_SLI_EVENT_STATUS_VALID:
+			return; /* no message if the sfp is okay */
+		case LPFC_SLI_EVENT_STATUS_NOT_PRESENT:
+			sprintf(message, "Optics faulted/incorrectly "
+				"installed/not installed - Reseat optics, "
+				"if issue not resolved, replace.");
+			break;
+		case LPFC_SLI_EVENT_STATUS_WRONG_TYPE:
+			sprintf(message,
+				"Optics of two types installed - Remove one "
+				"optic or install matching pair of optics.");
+			break;
+		case LPFC_SLI_EVENT_STATUS_UNSUPPORTED:
+			sprintf(message, "Incompatible optics - Replace with "
 				"compatible optics for card to function.");
+			break;
+		default:
+			/* firmware is reporting a status we don't know about */
+			sprintf(message, "Unknown event status x%02x", status);
+			break;
+		}
+
+		lpfc_printf_log(phba, KERN_ERR, LOG_SLI,
+				"3176 Misconfigured Physical Port - "
+				"Port Name %c %s\n", port_name, message);
+		break;
+	case LPFC_SLI_EVENT_TYPE_REMOTE_DPORT:
+		lpfc_printf_log(phba, KERN_INFO, LOG_SLI,
+				"3192 Remote DPort Test Initiated - "
+				"Event Data1:x%08x Event Data2: x%08x\n",
+				acqe_sli->event_data1, acqe_sli->event_data2);
 		break;
 	default:
-		/* firmware is reporting a status we don't know about */
-		sprintf(message, "Unknown event status x%02x", status);
+		lpfc_printf_log(phba, KERN_INFO, LOG_SLI,
+				"3193 Async SLI event - Event Data1:x%08x Event Data2:"
+				"x%08x SLI Event Type:%d\n",
+				acqe_sli->event_data1, acqe_sli->event_data2,
+				evt_type);
 		break;
 	}
-
-	lpfc_printf_log(phba, KERN_ERR, LOG_SLI,
-			"3176 Misconfigured Physical Port - "
-			"Port Name %c %s\n", port_name, message);
 }
 
 /**

@@ -5183,6 +5251,7 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba)
 		rc = lpfc_pci_function_reset(phba);
 		if (unlikely(rc))
 			return -ENODEV;
+		phba->temp_sensor_support = 1;
 	}
 
 	/* Create the bootstrap mailbox command */

James Smart

author

2015-02-16 16:11 UTC #2 lore raw

Quoted 34 lines · Tomas Henzl

Quoted 22 lines · James Smart
---
  drivers/scsi/lpfc/lpfc_hw4.h  |   1 +
  drivers/scsi/lpfc/lpfc_init.c | 179 +++++++++++++++++++++++++++++-------------
  2 files changed, 125 insertions(+), 55 deletions(-)
drivers/scsi/lpfc/lpfc_hw4.h +1 -0
index f432ec1..3121ec4 100644
@@ -3244,6 +3244,7 @@ struct lpfc_acqe_sli {
  #define LPFC_SLI_EVENT_TYPE_NVLOG_POST		0x4
  #define LPFC_SLI_EVENT_TYPE_DIAG_DUMP		0x5
  #define LPFC_SLI_EVENT_TYPE_MISCONFIGURED	0x9
+#define LPFC_SLI_EVENT_TYPE_REMOTE_DPORT	0xA
  };
 
  /*
drivers/scsi/lpfc/lpfc_init.c +0 -0
index 2b5b910..4ba91af 100644
drivers/scsi/lpfc/lpfc_init.c lpfc_offline_eratt(struct lpfc_hba *phba) · line 1335
Quoted 6 lines
drivers/scsi/lpfc/lpfc_init.c +3 -0
@@ -1330,13 +1330,14 @@ lpfc_offline_eratt(struct lpfc_hba *phba)
  void
  lpfc_sli4_offline_eratt(struct lpfc_hba *phba)
  {
+	spin_lock_irq(&phba->hbalock);
+	phba->link_state = LPFC_HBA_ERROR;
+	spin_unlock_irq(&phba->hbalock);
Hi James,
please explain why is the spinlock^ needed?
There seems to be lot of other places where link_state is not protected, for example it is evaluated in lpfc_sli4_handle_received_buffer with no lock.

Quoted 5 lines

Could you please also add some more description to the of the body your mails, sometimes a subject only is not enough.
Thanks,
Tomas

Well - good question - access should be under lock, and it wouldn't surprise me if there are some cases, especially read checks, where it isn't. In most cases, this bad behavior works, as the variable doesn't change often, and even when it does, there's very long time delays before something would reference the change. Not trying to say it shouldn't be fixed, but rather why it hasn't been an issue. This one will take a little bit to clean up - it reached this point historically over many successive changes.

Yes, I'll add a little meat to future patches.

-- james s

Tomas Henzl

2015-02-17 13:01 UTC #3 lore raw

Quoted 40 lines · James Smart

Quoted 34 lines · Tomas Henzl
Quoted 22 lines · James Smart
---
  drivers/scsi/lpfc/lpfc_hw4.h  |   1 +
  drivers/scsi/lpfc/lpfc_init.c | 179 +++++++++++++++++++++++++++++-------------
  2 files changed, 125 insertions(+), 55 deletions(-)
drivers/scsi/lpfc/lpfc_hw4.h +1 -0
index f432ec1..3121ec4 100644
@@ -3244,6 +3244,7 @@ struct lpfc_acqe_sli {
  #define LPFC_SLI_EVENT_TYPE_NVLOG_POST		0x4
  #define LPFC_SLI_EVENT_TYPE_DIAG_DUMP		0x5
  #define LPFC_SLI_EVENT_TYPE_MISCONFIGURED	0x9
+#define LPFC_SLI_EVENT_TYPE_REMOTE_DPORT	0xA
  };
 
  /*
drivers/scsi/lpfc/lpfc_init.c +0 -0
index 2b5b910..4ba91af 100644
drivers/scsi/lpfc/lpfc_init.c lpfc_offline_eratt(struct lpfc_hba *phba) · line 1335
Quoted 6 lines
drivers/scsi/lpfc/lpfc_init.c +3 -0
@@ -1330,13 +1330,14 @@ lpfc_offline_eratt(struct lpfc_hba *phba)
  void
  lpfc_sli4_offline_eratt(struct lpfc_hba *phba)
  {
+	spin_lock_irq(&phba->hbalock);
+	phba->link_state = LPFC_HBA_ERROR;
+	spin_unlock_irq(&phba->hbalock);
Hi James,
please explain why is the spinlock^ needed?
There seems to be lot of other places where link_state is not protected, for example it is evaluated in lpfc_sli4_handle_received_buffer with no lock.
Quoted 5 lines
Could you please also add some more description to the of the body your mails, sometimes a subject only is not enough.
Thanks,
Tomas

Quoted 8 lines

Well - good question - access should be under lock, and it wouldn't surprise me if there are some cases, especially read checks, where it isn't. In most cases, this bad behavior works, as the variable doesn't change often, and even when it does, there's very long time delays before something would reference the change. Not trying to say it shouldn't be fixed, but rather why it hasn't been an issue. This one will take a little bit to clean up - it reached this point historically over many successive changes.

OK, makes sense, I have found at least one place where is a spinlock used, his is a start for an longer process - I'm fine with the explanation. and also with your response in 7/21+10/21.

Quoted 2 lines

Yes, I'll add a little meat to future patches.

Thanks.

tomash

Quoted 9 lines

-- james s
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html