From: Xiang Chen <chenxiang66@hisilicon.com>
Date: Mon, 24 Sep 2018 23:06:33 +0800
Subject: scsi: hisi_sas: Use block layer tag instead for IPTT
Patch-mainline: v4.20-rc1
Git-commit: 784b46b7cba0ae914dd293f23848c5057c6ba017
References: FATE#326253

Currently we use the IPTT defined in LLDD to identify IOs. Actually for
IOs which are from the block layer, they have tags to identify them. So
for those IOs, use tag of the block layer directly, and for IOs which is
not from the block layer (such as internal IOs from libsas/LLDD), reserve
96 IPTTs for them.

Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Johannes Thumshirn <jthumshirn@suse.de>
---
 drivers/scsi/hisi_sas/hisi_sas.h       |    3 -
 drivers/scsi/hisi_sas/hisi_sas_main.c  |   89 +++++++++++++++++++++------------
 drivers/scsi/hisi_sas/hisi_sas_v1_hw.c |    1 
 drivers/scsi/hisi_sas/hisi_sas_v2_hw.c |    9 +--
 drivers/scsi/hisi_sas/hisi_sas_v3_hw.c |    8 +-
 5 files changed, 70 insertions(+), 40 deletions(-)

--- a/drivers/scsi/hisi_sas/hisi_sas.h
+++ b/drivers/scsi/hisi_sas/hisi_sas.h
@@ -34,6 +34,7 @@
 #define HISI_SAS_MAX_DEVICES HISI_SAS_MAX_ITCT_ENTRIES
 #define HISI_SAS_RESET_BIT	0
 #define HISI_SAS_REJECT_CMD_BIT	1
+#define HISI_SAS_RESERVED_IPTT_CNT  96
 
 #define HISI_SAS_STATUS_BUF_SZ (sizeof(struct hisi_sas_status_buffer))
 #define HISI_SAS_COMMAND_TABLE_SZ (sizeof(union hisi_sas_command_table))
@@ -218,7 +219,7 @@ struct hisi_sas_hw {
 	int (*hw_init)(struct hisi_hba *hisi_hba);
 	void (*setup_itct)(struct hisi_hba *hisi_hba,
 			   struct hisi_sas_device *device);
-	int (*slot_index_alloc)(struct hisi_hba *hisi_hba, int *slot_idx,
+	int (*slot_index_alloc)(struct hisi_hba *hisi_hba,
 				struct domain_device *device);
 	struct hisi_sas_device *(*alloc_dev)(struct domain_device *device);
 	void (*sl_notify)(struct hisi_hba *hisi_hba, int phy_no);
--- a/drivers/scsi/hisi_sas/hisi_sas_main.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_main.c
@@ -183,7 +183,14 @@ static void hisi_sas_slot_index_clear(st
 
 static void hisi_sas_slot_index_free(struct hisi_hba *hisi_hba, int slot_idx)
 {
-	hisi_sas_slot_index_clear(hisi_hba, slot_idx);
+	unsigned long flags;
+
+	if (hisi_hba->hw->slot_index_alloc || (slot_idx >=
+	    hisi_hba->hw->max_command_entries - HISI_SAS_RESERVED_IPTT_CNT)) {
+		spin_lock_irqsave(&hisi_hba->lock, flags);
+		hisi_sas_slot_index_clear(hisi_hba, slot_idx);
+		spin_unlock_irqrestore(&hisi_hba->lock, flags);
+	}
 }
 
 static void hisi_sas_slot_index_set(struct hisi_hba *hisi_hba, int slot_idx)
@@ -193,24 +200,34 @@ static void hisi_sas_slot_index_set(stru
 	set_bit(slot_idx, bitmap);
 }
 
-static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba, int *slot_idx)
+static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
+				     struct scsi_cmnd *scsi_cmnd)
 {
-	unsigned int index;
+	int index;
 	void *bitmap = hisi_hba->slot_index_tags;
+	unsigned long flags;
 
+	if (scsi_cmnd)
+		return scsi_cmnd->request->tag;
+
+	spin_lock_irqsave(&hisi_hba->lock, flags);
 	index = find_next_zero_bit(bitmap, hisi_hba->slot_index_count,
-			hisi_hba->last_slot_index + 1);
+				   hisi_hba->last_slot_index + 1);
 	if (index >= hisi_hba->slot_index_count) {
-		index = find_next_zero_bit(bitmap, hisi_hba->slot_index_count,
-					   0);
-		if (index >= hisi_hba->slot_index_count)
+		index = find_next_zero_bit(bitmap,
+				hisi_hba->slot_index_count,
+				hisi_hba->hw->max_command_entries -
+				HISI_SAS_RESERVED_IPTT_CNT);
+		if (index >= hisi_hba->slot_index_count) {
+			spin_unlock_irqrestore(&hisi_hba->lock, flags);
 			return -SAS_QUEUE_FULL;
+		}
 	}
 	hisi_sas_slot_index_set(hisi_hba, index);
-	*slot_idx = index;
 	hisi_hba->last_slot_index = index;
+	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
-	return 0;
+	return index;
 }
 
 static void hisi_sas_slot_index_init(struct hisi_hba *hisi_hba)
@@ -249,9 +266,7 @@ void hisi_sas_slot_task_free(struct hisi
 
 	memset(slot, 0, offsetof(struct hisi_sas_slot, buf));
 
-	spin_lock_irqsave(&hisi_hba->lock, flags);
 	hisi_sas_slot_index_free(hisi_hba, slot->idx);
-	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 }
 EXPORT_SYMBOL_GPL(hisi_sas_slot_task_free);
 
@@ -384,16 +399,27 @@ static int hisi_sas_task_prep(struct sas
 		goto err_out_dma_unmap;
 	}
 
-	spin_lock_irqsave(&hisi_hba->lock, flags);
 	if (hisi_hba->hw->slot_index_alloc)
-		rc = hisi_hba->hw->slot_index_alloc(hisi_hba, &slot_idx,
-						    device);
-	else
-		rc = hisi_sas_slot_index_alloc(hisi_hba, &slot_idx);
-	spin_unlock_irqrestore(&hisi_hba->lock, flags);
-	if (rc)
+		rc = hisi_hba->hw->slot_index_alloc(hisi_hba, device);
+	else {
+		struct scsi_cmnd *scsi_cmnd = NULL;
+
+		if (task->uldd_task) {
+			struct ata_queued_cmd *qc;
+
+			if (dev_is_sata(device)) {
+				qc = task->uldd_task;
+				scsi_cmnd = qc->scsicmd;
+			} else {
+				scsi_cmnd = task->uldd_task;
+			}
+		}
+		rc  = hisi_sas_slot_index_alloc(hisi_hba, scsi_cmnd);
+	}
+	if (rc < 0)
 		goto err_out_dma_unmap;
 
+	slot_idx = rc;
 	slot = &hisi_hba->slot_info[slot_idx];
 
 	spin_lock_irqsave(&dq->lock, flags);
@@ -454,9 +480,7 @@ static int hisi_sas_task_prep(struct sas
 	return 0;
 
 err_out_tag:
-	spin_lock_irqsave(&hisi_hba->lock, flags);
 	hisi_sas_slot_index_free(hisi_hba, slot_idx);
-	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 err_out_dma_unmap:
 	if (!sas_protocol_ata(task->task_proto)) {
 		if (task->num_scatter) {
@@ -1741,14 +1765,11 @@ hisi_sas_internal_abort_task_exec(struct
 	port = to_hisi_sas_port(sas_port);
 
 	/* simply get a slot and send abort command */
-	spin_lock_irqsave(&hisi_hba->lock, flags);
-	rc = hisi_sas_slot_index_alloc(hisi_hba, &slot_idx);
-	if (rc) {
-		spin_unlock_irqrestore(&hisi_hba->lock, flags);
+	rc = hisi_sas_slot_index_alloc(hisi_hba, NULL);
+	if (rc < 0)
 		goto err_out;
-	}
-	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
+	slot_idx = rc;
 	slot = &hisi_hba->slot_info[slot_idx];
 
 	spin_lock_irqsave(&dq->lock, flags_dq);
@@ -1784,7 +1805,6 @@ hisi_sas_internal_abort_task_exec(struct
 	spin_lock_irqsave(&task->task_state_lock, flags);
 	task->task_state_flags |= SAS_TASK_AT_INITIATOR;
 	spin_unlock_irqrestore(&task->task_state_lock, flags);
-
 	WRITE_ONCE(slot->ready, 1);
 	/* send abort command to the chip */
 	spin_lock_irqsave(&dq->lock, flags);
@@ -1795,9 +1815,7 @@ hisi_sas_internal_abort_task_exec(struct
 	return 0;
 
 err_out_tag:
-	spin_lock_irqsave(&hisi_hba->lock, flags);
 	hisi_sas_slot_index_free(hisi_hba, slot_idx);
-	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 err_out:
 	dev_err(dev, "internal abort task prep: failed[%d]!\n", rc);
 
@@ -2160,6 +2178,8 @@ int hisi_sas_alloc(struct hisi_hba *hisi
 	hisi_sas_init_mem(hisi_hba);
 
 	hisi_sas_slot_index_init(hisi_hba);
+	hisi_hba->last_slot_index = hisi_hba->hw->max_command_entries -
+		HISI_SAS_RESERVED_IPTT_CNT;
 
 	hisi_hba->wq = create_singlethread_workqueue(dev_name(dev));
 	if (!hisi_hba->wq) {
@@ -2363,8 +2383,15 @@ int hisi_sas_probe(struct platform_devic
 	shost->max_channel = 1;
 	shost->max_cmd_len = 16;
 	shost->sg_tablesize = min_t(u16, SG_ALL, HISI_SAS_SGE_PAGE_CNT);
-	shost->can_queue = hisi_hba->hw->max_command_entries;
-	shost->cmd_per_lun = hisi_hba->hw->max_command_entries;
+	if (hisi_hba->hw->slot_index_alloc) {
+		shost->can_queue = hisi_hba->hw->max_command_entries;
+		shost->cmd_per_lun = hisi_hba->hw->max_command_entries;
+	} else {
+		shost->can_queue = hisi_hba->hw->max_command_entries -
+			HISI_SAS_RESERVED_IPTT_CNT;
+		shost->cmd_per_lun = hisi_hba->hw->max_command_entries -
+			HISI_SAS_RESERVED_IPTT_CNT;
+	}
 
 	sha->sas_ha_name = DRV_NAME;
 	sha->dev = hisi_hba->dev;
--- a/drivers/scsi/hisi_sas/hisi_sas_v1_hw.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_v1_hw.c
@@ -1809,7 +1809,6 @@ static struct scsi_host_template sht_v1_
 	.scan_start		= hisi_sas_scan_start,
 	.change_queue_depth	= sas_change_queue_depth,
 	.bios_param		= sas_bios_param,
-	.can_queue		= 1,
 	.this_id		= -1,
 	.sg_tablesize		= SG_ALL,
 	.max_sectors		= SCSI_DEFAULT_MAX_SECTORS,
--- a/drivers/scsi/hisi_sas/hisi_sas_v2_hw.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_v2_hw.c
@@ -770,7 +770,7 @@ static u32 hisi_sas_phy_read32(struct hi
 
 /* This function needs to be protected from pre-emption. */
 static int
-slot_index_alloc_quirk_v2_hw(struct hisi_hba *hisi_hba, int *slot_idx,
+slot_index_alloc_quirk_v2_hw(struct hisi_hba *hisi_hba,
 			     struct domain_device *device)
 {
 	int sata_dev = dev_is_sata(device);
@@ -778,6 +778,7 @@ slot_index_alloc_quirk_v2_hw(struct hisi
 	struct hisi_sas_device *sas_dev = device->lldd_dev;
 	int sata_idx = sas_dev->sata_idx;
 	int start, end;
+	unsigned long flags;
 
 	if (!sata_dev) {
 		/*
@@ -801,6 +802,7 @@ slot_index_alloc_quirk_v2_hw(struct hisi
 		end = 64 * (sata_idx + 2);
 	}
 
+	spin_lock_irqsave(&hisi_hba->lock, flags);
 	while (1) {
 		start = find_next_zero_bit(bitmap,
 					hisi_hba->slot_index_count, start);
@@ -815,8 +817,8 @@ slot_index_alloc_quirk_v2_hw(struct hisi
 	}
 
 	set_bit(start, bitmap);
-	*slot_idx = start;
-	return 0;
+	spin_unlock_irqrestore(&hisi_hba->lock, flags);
+	return start;
 }
 
 static bool sata_index_alloc_v2_hw(struct hisi_hba *hisi_hba, int *idx)
@@ -3562,7 +3564,6 @@ static struct scsi_host_template sht_v2_
 	.scan_start		= hisi_sas_scan_start,
 	.change_queue_depth	= sas_change_queue_depth,
 	.bios_param		= sas_bios_param,
-	.can_queue		= 1,
 	.this_id		= -1,
 	.sg_tablesize		= SG_ALL,
 	.max_sectors		= SCSI_DEFAULT_MAX_SECTORS,
--- a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
@@ -2098,7 +2098,6 @@ static struct scsi_host_template sht_v3_
 	.scan_start		= hisi_sas_scan_start,
 	.change_queue_depth	= sas_change_queue_depth,
 	.bios_param		= sas_bios_param,
-	.can_queue		= 1,
 	.this_id		= -1,
 	.sg_tablesize		= SG_ALL,
 	.max_sectors		= SCSI_DEFAULT_MAX_SECTORS,
@@ -2108,6 +2107,7 @@ static struct scsi_host_template sht_v3_
 	.target_destroy		= sas_target_destroy,
 	.ioctl			= sas_ioctl,
 	.shost_attrs		= host_attrs,
+	.tag_alloc_policy	= BLK_TAG_ALLOC_RR,
 };
 
 static const struct hisi_sas_hw hisi_sas_v3_hw = {
@@ -2245,8 +2245,10 @@ hisi_sas_v3_probe(struct pci_dev *pdev,
 	shost->max_channel = 1;
 	shost->max_cmd_len = 16;
 	shost->sg_tablesize = min_t(u16, SG_ALL, HISI_SAS_SGE_PAGE_CNT);
-	shost->can_queue = hisi_hba->hw->max_command_entries;
-	shost->cmd_per_lun = hisi_hba->hw->max_command_entries;
+	shost->can_queue = hisi_hba->hw->max_command_entries -
+		HISI_SAS_RESERVED_IPTT_CNT;
+	shost->cmd_per_lun = hisi_hba->hw->max_command_entries -
+		HISI_SAS_RESERVED_IPTT_CNT;
 
 	sha->sas_ha_name = DRV_NAME;
 	sha->dev = dev;