scsi: qla2xxx: move fields from qla_hw_data to qla_qpair

- Move chip_reset, enable_class_2 fields from qla_hw_data to qla_qpair
  to reduce cache thrash for target MQ.
- Optimizations to reduce unnecessary memory load for good path io.
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

drivers/scsi/qla2xxx/qla_attr.c

@@ -2289,7 +2289,7 @@ qla2x00_init_host_attr(scsi_qla_host_t *vha)
 	fc_host_dev_loss_tmo(vha->host) = ha->port_down_retry_count;
 	fc_host_node_name(vha->host) = wwn_to_u64(vha->node_name);
 	fc_host_port_name(vha->host) = wwn_to_u64(vha->port_name);
-	fc_host_supported_classes(vha->host) = ha->tgt.enable_class_2 ?
+	fc_host_supported_classes(vha->host) = ha->base_qpair->enable_class_2 ?
 			(FC_COS_CLASS2|FC_COS_CLASS3) : FC_COS_CLASS3;
 	fc_host_max_npiv_vports(vha->host) = ha->max_npiv_vports;
 	fc_host_npiv_vports_inuse(vha->host) = ha->cur_vport_count;

drivers/scsi/qla2xxx/qla_def.h

@@ -3252,6 +3252,7 @@ struct qla_qpair {
 	 */
 	spinlock_t *qp_lock_ptr;
 	struct scsi_qla_host *vha;
+	u32 chip_reset;
 
 	/* distill these fields down to 'online=0/1'
 	 * ha->flags.eeh_busy
@@ -3263,6 +3264,8 @@ struct qla_qpair {
 	uint32_t difdix_supported:1;
 	uint32_t delete_in_progress:1;
 	uint32_t fw_started:1;
+	uint32_t enable_class_2:1;
+	uint32_t enable_explicit_conf:1;
 
 	uint16_t id;			/* qp number used with FW */
 	uint16_t vp_idx;		/* vport ID */
@@ -3296,8 +3299,6 @@ struct scsi_qlt_host {
 
 struct qlt_hw_data {
 	/* Protected by hw lock */
-	uint32_t enable_class_2:1;
-	uint32_t enable_explicit_conf:1;
 	uint32_t node_name_set:1;
 
 	dma_addr_t atio_dma;	/* Physical address. */
@@ -3954,7 +3955,6 @@ struct qla_hw_data {
 	struct work_struct board_disable;
 
 	struct mr_data_fx00 mr;
-	uint32_t chip_reset;
 
 	struct qlt_hw_data tgt;
 	int	allow_cna_fw_dump;
@@ -4247,6 +4247,25 @@ struct qla2_sgx {
 #define QLA_QPAIR_MARK_NOT_BUSY(__qpair)		\
 	atomic_dec(&__qpair->ref_count);		\
 
+
+#define QLA_ENA_CONF(_ha) {\
+    int i;\
+    _ha->base_qpair->enable_explicit_conf = 1;	\
+    for (i = 0; i < _ha->max_qpairs; i++) {	\
+	if (_ha->queue_pair_map[i])		\
+	    _ha->queue_pair_map[i]->enable_explicit_conf = 1; \
+    }						\
+}
+
+#define QLA_DIS_CONF(_ha) {\
+    int i;\
+    _ha->base_qpair->enable_explicit_conf = 0;	\
+    for (i = 0; i < _ha->max_qpairs; i++) {	\
+	if (_ha->queue_pair_map[i])		\
+	    _ha->queue_pair_map[i]->enable_explicit_conf = 0; \
+    }						\
+}
+
 /*
  * qla2x00 local function return status codes
  */

drivers/scsi/qla2xxx/qla_init.c

@@ -1322,7 +1322,7 @@ qla24xx_handle_plogi_done_event(struct scsi_qla_host *vha, struct event_arg *ea)
 		ql_dbg(ql_dbg_disc, vha, 0x20ea,
 		    "%s %d %8phC post gpdb\n",
 		    __func__, __LINE__, ea->fcport->port_name);
-		ea->fcport->chip_reset = vha->hw->chip_reset;
+		ea->fcport->chip_reset = vha->hw->base_qpair->chip_reset;
 		ea->fcport->logout_on_delete = 1;
 		qla24xx_post_gpdb_work(vha, ea->fcport, 0);
 		break;
@@ -5524,6 +5524,7 @@ qla2x00_abort_isp_cleanup(scsi_qla_host_t *vha)
 	struct scsi_qla_host *vp;
 	unsigned long flags;
 	fc_port_t *fcport;
+	u16 i;
 
 	/* For ISP82XX, driver waits for completion of the commands.
 	 * online flag should be set.
@@ -5549,7 +5550,12 @@ qla2x00_abort_isp_cleanup(scsi_qla_host_t *vha)
 	ha->current_topology = 0;
 	ha->flags.fw_started = 0;
 	ha->flags.fw_init_done = 0;
-	ha->chip_reset++;
+	ha->base_qpair->chip_reset++;
+	for (i = 0; i < ha->max_qpairs; i++) {
+		if (ha->queue_pair_map[i])
+			ha->queue_pair_map[i]->chip_reset =
+				ha->base_qpair->chip_reset;
+	}
 
 	atomic_set(&vha->loop_down_timer, LOOP_DOWN_TIME);
 	if (atomic_read(&vha->loop_state) != LOOP_DOWN) {
@@ -7624,6 +7630,10 @@ struct qla_qpair *qla2xxx_create_qpair(struct scsi_qla_host *vha, int qos,
 		qpair->id = qpair_id;
 		qpair->vp_idx = vp_idx;
 		INIT_LIST_HEAD(&qpair->hints_list);
+		qpair->chip_reset = ha->base_qpair->chip_reset;
+		qpair->enable_class_2 = ha->base_qpair->enable_class_2;
+		qpair->enable_explicit_conf =
+		    ha->base_qpair->enable_explicit_conf;
 
 		for (i = 0; i < ha->msix_count; i++) {
 			msix = &ha->msix_entries[i];

drivers/scsi/qla2xxx/qla_os.c

@@ -387,6 +387,7 @@ static int qla2x00_alloc_queues(struct qla_hw_data *ha, struct req_que *req,
 	/* init qpair to this cpu. Will adjust at run time. */
 	ha->base_qpair->msix = &ha->msix_entries[QLA_MSIX_RSP_Q];
 	INIT_LIST_HEAD(&ha->base_qpair->hints_list);
+	ha->base_qpair->enable_class_2 = ql2xenableclass2;
 	qla_cpu_update(rsp->qpair, smp_processor_id());
 
 	if (ql2xmqsupport && ha->max_qpairs) {
@@ -2708,7 +2709,6 @@ qla2x00_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	ql_dbg_pci(ql_dbg_init, pdev, 0x000a,
 	    "Memory allocated for ha=%p.\n", ha);
 	ha->pdev = pdev;
-	ha->tgt.enable_class_2 = ql2xenableclass2;
 	INIT_LIST_HEAD(&ha->tgt.q_full_list);
 	spin_lock_init(&ha->tgt.q_full_lock);
 	spin_lock_init(&ha->tgt.sess_lock);

drivers/scsi/qla2xxx/qla_target.h

@@ -881,6 +881,7 @@ struct qla_tgt_cmd {
 	struct se_cmd se_cmd;
 	struct fc_port *sess;
 	struct qla_qpair *qpair;
+	uint32_t reset_count;
 	int state;
 	struct work_struct work;
 	/* Sense buffer that will be mapped into outgoing status */
@@ -906,7 +907,6 @@ struct qla_tgt_cmd {
 	int offset;
 	u64 unpacked_lun;
 	enum dma_data_direction dma_data_direction;
-	uint32_t reset_count;
 
 	uint16_t loop_id;	/* to save extra sess dereferences */
 	struct qla_tgt *tgt;	/* to save extra sess dereferences */
@@ -980,7 +980,6 @@ struct qla_tgt_prm {
 	int seg_cnt;
 	int req_cnt;
 	uint16_t rq_result;
-	uint16_t scsi_status;
 	int sense_buffer_len;
 	int residual;
 	int add_status_pkt;