scsi: cxlflash: Support multiple hardware queues

Introduce multiple hardware queues to improve legacy I/O path performance.
Each hardware queue is comprised of a master context and associated I/O
resources. The hardware queues are initially implemented as a static array
embedded in the AFU. This will be transitioned to a dynamic allocation in a
later series to improve the memory footprint of the driver.
Signed-off-by: NUma Krishnan <ukrishn@linux.vnet.ibm.com>
Acked-by: NMatthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

drivers/scsi/cxlflash/common.h

@@ -60,6 +60,9 @@ extern const struct file_operations cxlflash_cxl_fops;
 /* SQ for master issued cmds */
 #define NUM_SQ_ENTRY			CXLFLASH_MAX_CMDS
 
+#define CXLFLASH_NUM_HWQS		1
+#define PRIMARY_HWQ			0
+
 
 static inline void check_sizes(void)
 {
@@ -98,7 +101,6 @@ enum cxlflash_state {
 
 struct cxlflash_cfg {
 	struct afu *afu;
-	struct cxl_context *mcctx;
 
 	struct pci_dev *dev;
 	struct pci_device_id *dev_id;
@@ -144,6 +146,7 @@ struct afu_cmd {
 	struct list_head queue;
 
 	u8 cmd_tmf:1;
+	u32 hwq_index;
 
 	/* As per the SISLITE spec the IOARCB EA has to be 16-byte aligned.
 	 * However for performance reasons the IOARCB/IOASA should be
@@ -164,7 +167,7 @@ static inline struct afu_cmd *sc_to_afucz(struct scsi_cmnd *sc)
 	return afuc;
 }
 
-struct afu {
+struct hwq {
 	/* Stuff requiring alignment go first. */
 	struct sisl_ioarcb sq[NUM_SQ_ENTRY];		/* 16K SQ */
 	u64 rrq_entry[NUM_RRQ_ENTRY];			/* 2K RRQ */
@@ -172,17 +175,13 @@ struct afu {
 	/* Beware of alignment till here. Preferably introduce new
 	 * fields after this point
 	 */
-
-	int (*send_cmd)(struct afu *, struct afu_cmd *);
-	void (*context_reset)(struct afu_cmd *);
-
-	/* AFU HW */
+	struct afu *afu;
+	struct cxl_context *ctx;
 	struct cxl_ioctl_start_work work;
-	struct cxlflash_afu_map __iomem *afu_map;	/* entire MMIO map */
 	struct sisl_host_map __iomem *host_map;		/* MC host map */
 	struct sisl_ctrl_map __iomem *ctrl_map;		/* MC control map */
-
 	ctx_hndl_t ctx_hndl;	/* master's context handle */
+	u32 index;		/* Index of this hwq */
 
 	atomic_t hsq_credits;
 	spinlock_t hsq_slock;
@@ -194,9 +193,22 @@ struct afu {
 	u64 *hrrq_end;
 	u64 *hrrq_curr;
 	bool toggle;
-	atomic_t cmds_active;	/* Number of currently active AFU commands */
+
 	s64 room;
 	spinlock_t rrin_slock; /* Lock to rrin queuing and cmd_room updates */
+
+	struct irq_poll irqpoll;
+} __aligned(cache_line_size());
+
+struct afu {
+	struct hwq hwqs[CXLFLASH_NUM_HWQS];
+	int (*send_cmd)(struct afu *, struct afu_cmd *);
+	void (*context_reset)(struct afu_cmd *);
+
+	/* AFU HW */
+	struct cxlflash_afu_map __iomem *afu_map;	/* entire MMIO map */
+
+	atomic_t cmds_active;	/* Number of currently active AFU commands */
 	u64 hb;
 	u32 internal_lun;	/* User-desired LUN mode for this AFU */
 
@@ -204,11 +216,16 @@ struct afu {
 	u64 interface_version;
 
 	u32 irqpoll_weight;
-	struct irq_poll irqpoll;
 	struct cxlflash_cfg *parent; /* Pointer back to parent cxlflash_cfg */
-
 };
 
+static inline struct hwq *get_hwq(struct afu *afu, u32 index)
+{
+	WARN_ON(index >= CXLFLASH_NUM_HWQS);
+
+	return &afu->hwqs[index];
+}
+
 static inline bool afu_is_irqpoll_enabled(struct afu *afu)
 {
 	return !!afu->irqpoll_weight;

drivers/scsi/cxlflash/superpipe.c

@@ -254,6 +254,7 @@ static int afu_attach(struct cxlflash_cfg *cfg, struct ctx_info *ctxi)
 	struct afu *afu = cfg->afu;
 	struct sisl_ctrl_map __iomem *ctrl_map = ctxi->ctrl_map;
 	int rc = 0;
+	struct hwq *hwq = get_hwq(afu, PRIMARY_HWQ);
 	u64 val;
 
 	/* Unlock cap and restrict user to read/write cmds in translated mode */
@@ -270,7 +271,7 @@ static int afu_attach(struct cxlflash_cfg *cfg, struct ctx_info *ctxi)
 
 	/* Set up MMIO registers pointing to the RHT */
 	writeq_be((u64)ctxi->rht_start, &ctrl_map->rht_start);
-	val = SISL_RHT_CNT_ID((u64)MAX_RHT_PER_CONTEXT, (u64)(afu->ctx_hndl));
+	val = SISL_RHT_CNT_ID((u64)MAX_RHT_PER_CONTEXT, (u64)(hwq->ctx_hndl));
 	writeq_be(val, &ctrl_map->rht_cnt_id);
 out:
 	dev_dbg(dev, "%s: returning rc=%d\n", __func__, rc);
@@ -1626,6 +1627,7 @@ static int cxlflash_afu_recover(struct scsi_device *sdev,
 	struct afu *afu = cfg->afu;
 	struct ctx_info *ctxi = NULL;
 	struct mutex *mutex = &cfg->ctx_recovery_mutex;
+	struct hwq *hwq = get_hwq(afu, PRIMARY_HWQ);
 	u64 flags;
 	u64 ctxid = DECODE_CTXID(recover->context_id),
 	    rctxid = recover->context_id;
@@ -1696,7 +1698,7 @@ static int cxlflash_afu_recover(struct scsi_device *sdev,
 	}
 
 	/* Test if in error state */
-	reg = readq_be(&afu->ctrl_map->mbox_r);
+	reg = readq_be(&hwq->ctrl_map->mbox_r);
 	if (reg == -1) {
 		dev_dbg(dev, "%s: MMIO fail, wait for recovery.\n", __func__);