[PATCH 0/3] smartpqi updates

[PATCH 0/3] smartpqi updates

[Don Brace]

These patches are based on Martin Petersen's 6.8/scsi-queue tree
  https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git
  6.8/scsi-queue

The only functional change to smartpqi is correction of a race condition
during a scan/rescan operation. The device->rescan flag can be updated
by multiple threads causing issues.

The other two patches add PCI-IDs for new controllers and change the
driver version.

This set of changes consists of:
* smartpqi-Add-new-controller-PCI-IDs
  No functional changes. Just adding in more device support.
* smartpqi: fix logical volume rescan race condition
  The driver has several threads that can contend for accessing
  a controller instance wide 'rescan' flag. In some rare cases
  a race condition can occur that bypasses a call to rescan for
  device changes. Some of these changes are; device removal, logical
  volume changes, and device adds. We basically added a spin-lock
  to protect the 'rescan' flag.
* smartpqi-change-driver-version-to-2.1.26-030
  No functional changes.

---

David Strahan (1):
  smartpqi: Add new controller PCI IDs

Don Brace (1):
  smartpqi: bump driver version to 2.1.26-030

Mahesh Rajashekhara (1):
  smartpqi: fix logical volume rescan race condition

 drivers/scsi/smartpqi/smartpqi.h      |  1 -
 drivers/scsi/smartpqi/smartpqi_init.c | 89 ++++++++++++++++++++++++---
 2 files changed, 79 insertions(+), 11 deletions(-)

-- 
2.43.0.76.g1a87c842ec

[PATCH 1/3] smartpqi: Add new controller PCI IDs

[Don Brace]

From: David Strahan <david.strahan@microchip.com>

All PCI ID entries in Hex.

    Add PCI IDs for Cisco controllers:
                                                VID  / DID  / SVID / SDID
                                                ----   ----   ----   ----
        Cisco 24G TriMode M1 RAID 4GB FBWC 32D  9005 / 028f / 1137 / 02f8
        Cisco 24G TriMode M1 RAID 4GB FBWC 16D  9005 / 028f / 1137 / 02f9
        Cisco 24G TriMode M1 HBA 16D            9005 / 028f / 1137 / 02fa

    Add PCI IDs for CloudNine controllers:
                                                VID  / DID  / SVID / SDID
                                                ----   ----   ----   ----
        SmartRAID P7604N-16i                    9005 / 028f / 1f51 / 100e
        SmartRAID P7604N-8i                     9005 / 028f / 1f51 / 100f
        SmartRAID P7504N-16i                    9005 / 028f / 1f51 / 1010
        SmartRAID P7504N-8i                     9005 / 028f / 1f51 / 1011
        SmartRAID P7504N-8i                     9005 / 028f / 1f51 / 1043
        SmartHBA  P6500-8i                      9005 / 028f / 1f51 / 1044
        SmartRAID P7504-8i                      9005 / 028f / 1f51 / 1045

Reviewed-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: David Strahan <david.strahan@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
---
 drivers/scsi/smartpqi/smartpqi_init.c | 40 +++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/drivers/scsi/smartpqi/smartpqi_init.c b/drivers/scsi/smartpqi/smartpqi_init.c
index 9a58df9312fa..d56201120087 100644
--- a/drivers/scsi/smartpqi/smartpqi_init.c
+++ b/drivers/scsi/smartpqi/smartpqi_init.c
@@ -10142,6 +10142,18 @@ static const struct pci_device_id pqi_pci_id_table[] = {
 		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
 				0x1014, 0x0718)
 	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1137, 0x02f8)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1137, 0x02f9)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1137, 0x02fa)
+	},
 	{
 		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
 				0x1e93, 0x1000)
@@ -10198,6 +10210,34 @@ static const struct pci_device_id pqi_pci_id_table[] = {
 		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
 				0x1f51, 0x100a)
 	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x100e)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x100f)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x1010)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x1011)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x1043)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x1044)
+	},
+	{
+		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
+			       0x1f51, 0x1045)
+	},
 	{
 		PCI_DEVICE_SUB(PCI_VENDOR_ID_ADAPTEC2, 0x028f,
 			       PCI_ANY_ID, PCI_ANY_ID)
-- 
2.43.0.76.g1a87c842ec

[PATCH 2/3] smartpqi: fix logical volume rescan race condition

[Don Brace]

From: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>

Correct rescan flag race condition.

   Multiple conditions are being evaluated before notifying OS
   to do a rescan.

   Driver will skip re-scanning the device if any one of the following
   conditions are met:
   - Devices that have not yet been added to the OS or devices
     that have been removed.
   - Devices which are already marked for removal or in the phase
     of removal.

   Under very rare conditions, after logical volume size expansion, the OS
   still sees the size of the logical volume which was before expansion.

   The rescan flag in the driver is used to signal the need for a logical
   volume rescan. A race condition can occur in the driver, and it leads
   to one thread overwriting the flag inadvertently. As a result, driver
   is not notifying the OS SML to rescan the logical volume.

   Move device->rescan update into new function
   pqi_mark_volumes_for_rescan() and protect with a spin lock.

   Move check for device->rescan into new function
   pqi_volume_rescan_needed() and protect function call with
   a spin_lock.

Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Co-developed-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Signed-off-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Signed-off-by: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
---
 drivers/scsi/smartpqi/smartpqi.h      |  1 -
 drivers/scsi/smartpqi/smartpqi_init.c | 43 ++++++++++++++++++++++-----
 2 files changed, 36 insertions(+), 8 deletions(-)

diff --git a/drivers/scsi/smartpqi/smartpqi.h b/drivers/scsi/smartpqi/smartpqi.h
index 041940183516..cdedc271857a 100644
--- a/drivers/scsi/smartpqi/smartpqi.h
+++ b/drivers/scsi/smartpqi/smartpqi.h
@@ -1347,7 +1347,6 @@ struct pqi_ctrl_info {
 	bool		controller_online;
 	bool		block_requests;
 	bool		scan_blocked;
-	u8		logical_volume_rescan_needed : 1;
 	u8		inbound_spanning_supported : 1;
 	u8		outbound_spanning_supported : 1;
 	u8		pqi_mode_enabled : 1;
diff --git a/drivers/scsi/smartpqi/smartpqi_init.c b/drivers/scsi/smartpqi/smartpqi_init.c
index d56201120087..081bb2c09806 100644
--- a/drivers/scsi/smartpqi/smartpqi_init.c
+++ b/drivers/scsi/smartpqi/smartpqi_init.c
@@ -2093,8 +2093,6 @@ static void pqi_scsi_update_device(struct pqi_ctrl_info *ctrl_info,
 		if (existing_device->devtype == TYPE_DISK) {
 			existing_device->raid_level = new_device->raid_level;
 			existing_device->volume_status = new_device->volume_status;
-			if (ctrl_info->logical_volume_rescan_needed)
-				existing_device->rescan = true;
 			memset(existing_device->next_bypass_group, 0, sizeof(existing_device->next_bypass_group));
 			if (!pqi_raid_maps_equal(existing_device->raid_map, new_device->raid_map)) {
 				kfree(existing_device->raid_map);
@@ -2164,6 +2162,20 @@ static inline void pqi_init_device_tmf_work(struct pqi_scsi_dev *device)
 		INIT_WORK(&tmf_work->work_struct, pqi_tmf_worker);
 }
 
+static inline bool pqi_volume_rescan_needed(struct pqi_scsi_dev *device)
+{
+	if (pqi_device_in_remove(device))
+		return false;
+
+	if (device->sdev == NULL)
+		return false;
+
+	if (!scsi_device_online(device->sdev))
+		return false;
+
+	return device->rescan;
+}
+
 static void pqi_update_device_list(struct pqi_ctrl_info *ctrl_info,
 	struct pqi_scsi_dev *new_device_list[], unsigned int num_new_devices)
 {
@@ -2284,9 +2296,13 @@ static void pqi_update_device_list(struct pqi_ctrl_info *ctrl_info,
 		if (device->sdev && device->queue_depth != device->advertised_queue_depth) {
 			device->advertised_queue_depth = device->queue_depth;
 			scsi_change_queue_depth(device->sdev, device->advertised_queue_depth);
-			if (device->rescan) {
-				scsi_rescan_device(device->sdev);
+			spin_lock_irqsave(&ctrl_info->scsi_device_list_lock, flags);
+			if (pqi_volume_rescan_needed(device)) {
 				device->rescan = false;
+				spin_unlock_irqrestore(&ctrl_info->scsi_device_list_lock, flags);
+				scsi_rescan_device(device->sdev);
+			} else {
+				spin_unlock_irqrestore(&ctrl_info->scsi_device_list_lock, flags);
 			}
 		}
 	}
@@ -2308,8 +2324,6 @@ static void pqi_update_device_list(struct pqi_ctrl_info *ctrl_info,
 		}
 	}
 
-	ctrl_info->logical_volume_rescan_needed = false;
-
 }
 
 static inline bool pqi_is_supported_device(struct pqi_scsi_dev *device)
@@ -3702,6 +3716,21 @@ static bool pqi_ofa_process_event(struct pqi_ctrl_info *ctrl_info,
 	return ack_event;
 }
 
+static void pqi_mark_volumes_for_rescan(struct pqi_ctrl_info *ctrl_info)
+{
+	unsigned long flags;
+	struct pqi_scsi_dev *device;
+
+	spin_lock_irqsave(&ctrl_info->scsi_device_list_lock, flags);
+
+	list_for_each_entry(device, &ctrl_info->scsi_device_list, scsi_device_list_entry) {
+		if (pqi_is_logical_device(device) && device->devtype == TYPE_DISK)
+			device->rescan = true;
+	}
+
+	spin_unlock_irqrestore(&ctrl_info->scsi_device_list_lock, flags);
+}
+
 static void pqi_disable_raid_bypass(struct pqi_ctrl_info *ctrl_info)
 {
 	unsigned long flags;
@@ -3742,7 +3771,7 @@ static void pqi_event_worker(struct work_struct *work)
 				ack_event = true;
 				rescan_needed = true;
 				if (event->event_type == PQI_EVENT_TYPE_LOGICAL_DEVICE)
-					ctrl_info->logical_volume_rescan_needed = true;
+					pqi_mark_volumes_for_rescan(ctrl_info);
 				else if (event->event_type == PQI_EVENT_TYPE_AIO_STATE_CHANGE)
 					pqi_disable_raid_bypass(ctrl_info);
 			}
-- 
2.43.0.76.g1a87c842ec

[PATCH 3/3] smartpqi: bump driver version to 2.1.26-030

[Don Brace]

Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Reviewed-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
---
 drivers/scsi/smartpqi/smartpqi_init.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/smartpqi/smartpqi_init.c b/drivers/scsi/smartpqi/smartpqi_init.c
index 081bb2c09806..ceff1ec13f9e 100644
--- a/drivers/scsi/smartpqi/smartpqi_init.c
+++ b/drivers/scsi/smartpqi/smartpqi_init.c
@@ -33,11 +33,11 @@
 #define BUILD_TIMESTAMP
 #endif
 
-#define DRIVER_VERSION		"2.1.24-046"
+#define DRIVER_VERSION		"2.1.26-030"
 #define DRIVER_MAJOR		2
 #define DRIVER_MINOR		1
-#define DRIVER_RELEASE		24
-#define DRIVER_REVISION		46
+#define DRIVER_RELEASE		26
+#define DRIVER_REVISION		30
 
 #define DRIVER_NAME		"Microchip SmartPQI Driver (v" \
 				DRIVER_VERSION BUILD_TIMESTAMP ")"
-- 
2.43.0.76.g1a87c842ec

Re: [PATCH 0/3] smartpqi updates

[Martin K. Petersen]

Don,

> The only functional change to smartpqi is correction of a race
> condition during a scan/rescan operation. The device->rescan flag can
> be updated by multiple threads causing issues.

Applied to 6.8/scsi-staging, thanks!

-- 
Martin K. Petersen	Oracle Linux Engineering