Skip to contentSkip to conversation

scsi: Remember if a device is an ATA device

PATCH Applied Damien Le Moal 4 participants 8 messages Jun 11, 2025 – Jun 20, 2025 SCSI

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 scsi: Remember if a device is an ATA device

scsi_add_lun() tests the device vendor string of SCSI devices to detect if a SCSI device is in fact an ATA device, in order to correctly handle SATL power management. The function scsi_cdl_enable() also requires knowing if a SCSI device is an ATA device to control the state of the device CDL feature but this function does that by testing for the presence of the VPD page 89h (ATA INFORMATION page). sd_read_write_same() also has a similar test.

Simplify these different methods by adding the is_ata field to struct scsi_device to remember that a SCSI device is in fact an ATA one based on the device vendor name test. This filed can also allow low level SCSI host adapter drivers to take special actions for ATA devices (e.g. to better handle ATA NCQ errors).

With this, simplify scsi_cdl_enable() and sd_read_write_same().

---
Changes from v1:
- Add simplification of sd_read_write_same()

 drivers/scsi/scsi.c        |  7 +------
 drivers/scsi/scsi_scan.c   |  3 ++-
 drivers/scsi/sd.c          | 13 ++++---------
 include/scsi/scsi_device.h |  5 +++++
 4 files changed, 12 insertions(+), 16 deletions(-)
drivers/scsi/scsi.c +1 -6
index 518a252eb6aa..534310224e8f 100644
@@ -708,20 +708,15 @@ void scsi_cdl_check(struct scsi_device *sdev)
 int scsi_cdl_enable(struct scsi_device *sdev, bool enable)
 {
 	char buf[64];
-	bool is_ata;
 	int ret;
 
 	if (!sdev->cdl_supported)
 		return -EOPNOTSUPP;
 
-	rcu_read_lock();
-	is_ata = rcu_dereference(sdev->vpd_pg89);
-	rcu_read_unlock();
-
 	/*
 	 * For ATA devices, CDL needs to be enabled with a SET FEATURES command.
 	 */
-	if (is_ata) {
+	if (sdev->is_ata) {
 		struct scsi_mode_data data;
 		struct scsi_sense_hdr sshdr;
 		char *buf_data;
drivers/scsi/scsi_scan.c +2 -1
index 4833b8fe251b..160c2f74c7e7 100644
@@ -909,7 +909,8 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result,
 	sdev->model = (char *) (sdev->inquiry + 16);
 	sdev->rev = (char *) (sdev->inquiry + 32);
 
-	if (strncmp(sdev->vendor, "ATA     ", 8) == 0) {
+	sdev->is_ata = strncmp(sdev->vendor, "ATA     ", 8) == 0;
+	if (sdev->is_ata) {
 		/*
 		 * sata emulation layer device.  This is a hack to work around
 		 * the SATL power management specifications which state that
drivers/scsi/sd.c +4 -9
index 3f6e87705b62..daddef2e9e87 100644
@@ -3459,19 +3459,14 @@ static void sd_read_write_same(struct scsi_disk *sdkp, unsigned char *buffer)
 	}
 
 	if (scsi_report_opcode(sdev, buffer, SD_BUF_SIZE, INQUIRY, 0) < 0) {
-		struct scsi_vpd *vpd;
-
 		sdev->no_report_opcodes = 1;
 
-		/* Disable WRITE SAME if REPORT SUPPORTED OPERATION
-		 * CODES is unsupported and the device has an ATA
-		 * Information VPD page (SAT).
+		/*
+		 * Disable WRITE SAME if REPORT SUPPORTED OPERATION CODES is
+		 * unsupported and this is an ATA device.
 		 */
-		rcu_read_lock();
-		vpd = rcu_dereference(sdev->vpd_pg89);
-		if (vpd)
+		if (sdev->is_ata)
 			sdev->no_write_same = 1;
-		rcu_read_unlock();
 	}
 
 	if (scsi_report_opcode(sdev, buffer, SD_BUF_SIZE, WRITE_SAME_16, 0) == 1)
include/scsi/scsi_device.h +5 -0
index 68dd49947d04..6d6500148c4b 100644
@@ -184,6 +184,11 @@ struct scsi_device {
 	 */
 	unsigned force_runtime_start_on_system_start:1;
 
+	/*
+	 * Set if the device is an ATA device.
+	 */
+	unsigned is_ata:1;
+
 	unsigned removable:1;
 	unsigned changed:1;	/* Data invalid due to media change */
 	unsigned busy:1;	/* Used to prevent races */
Signature · 1 line
2.49.0
#1 lore raw
Quoted 8 lines · Damien Le Moal

scsi_add_lun() tests the device vendor string of SCSI devices to detect if a SCSI device is in fact an ATA device, in order to correctly handle SATL power management. The function scsi_cdl_enable() also requires knowing if a SCSI device is an ATA device to control the state of the device CDL feature but this function does that by testing for the presence of the VPD page 89h (ATA INFORMATION page). sd_read_write_same() also has a similar test.

Quoted 7 lines

Simplify these different methods by adding the is_ata field to struct scsi_device to remember that a SCSI device is in fact an ATA one based on the device vendor name test. This filed can also allow low level SCSI host adapter drivers to take special actions for ATA devices (e.g. to better handle ATA NCQ errors).

With this, simplify scsi_cdl_enable() and sd_read_write_same().

Hi Damien,

There is only one "if (is_ata)" check in the SCSI core as far as I can see. Can it be avoided that ATA code leaks into the SCSI core by introducing a new function pointer, e.g. in struct Scsi_Host, and by calling that new function pointer if it has been set from scsi_cdl_enable()?

Thanks,

Bart.

#2 lore raw
Quoted 15 lines · Bart Van Assche
Quoted 14 lines · Damien Le Moal

scsi_add_lun() tests the device vendor string of SCSI devices to detect if a SCSI device is in fact an ATA device, in order to correctly handle SATL power management. The function scsi_cdl_enable() also requires knowing if a SCSI device is an ATA device to control the state of the device CDL feature but this function does that by testing for the presence of the VPD page 89h (ATA INFORMATION page). sd_read_write_same() also has a similar test.

Simplify these different methods by adding the is_ata field to struct scsi_device to remember that a SCSI device is in fact an ATA one based on the device vendor name test. This filed can also allow low level SCSI host adapter drivers to take special actions for ATA devices (e.g. to better handle ATA NCQ errors).

Quoted 8 lines
Quoted 1 line

With this, simplify scsi_cdl_enable() and sd_read_write_same().

Hi Damien,

There is only one "if (is_ata)" check in the SCSI core as far as I can see. Can it be avoided that ATA code leaks into the SCSI core by introducing a new function pointer, e.g. in struct Scsi_Host, and by calling that new function pointer if it has been set from scsi_cdl_enable()?

You are off by 2 on the count:

git grep "\->is_ata" drivers/scsi/*

drivers/scsi/scsi.c:    if (sdev->is_ata) {
drivers/scsi/scsi_scan.c:       if (sdev->is_ata) {
drivers/scsi/sd.c:              if (sdev->is_ata)

And no, we cannot avoid trying to detect if we are dealing with an ATA/SAT device or a real SCSI device in all 3 places where this "if" is done. 2 of these used VPD page dereference under rcu lock before, which is rather heavy handed. The point of this patch is to simplify that.

Quoted 4 lines

Thanks,

Bart.

Signature · 2 lines
Damien Le Moal
Western Digital Research
#3 lore raw
Quoted 4 lines · Damien Le Moal

And no, we cannot avoid trying to detect if we are dealing with an ATA/SAT device or a real SCSI device in all 3 places where this "if" is done. 2 of these used VPD page dereference under rcu lock before, which is rather heavy handed. The point of this patch is to simplify that.

Hi Damien,

The code under "if (is_ata)" in scsi_cdl_enable() seems very ATA-specific to me. Shouldn't that code be moved under drivers/ata/?

Regarding the following code in scsi_add_lun():

if (strncmp(sdev->vendor, "ATA ", 8) == 0)

		sdev->allow_restart = 1;

Other SCSI LLDs set the 'allow_restart' flag in their sdev_configure callback. Can this be done for ATA disks too?

Regarding the code that sets the no_write_same flag:

	if (sdev->is_ata)
		sdev->no_write_same = 1;

Why is the RCU reader lock held around that code, a lock that protects reading from the VPD pages while the above code does not access the contents of any VPD page?

Other SCSI LLDs set the 'no_write_same' flag in their sdev_configure callback. Can this be done for ATA disks too?

Thanks,

Bart.

#4 lore raw
Quoted 10 lines · Bart Van Assche
Quoted 4 lines · Damien Le Moal

And no, we cannot avoid trying to detect if we are dealing with an ATA/SAT device or a real SCSI device in all 3 places where this "if" is done. 2 of these used VPD page dereference under rcu lock before, which is rather heavy handed. The point of this patch is to simplify that.

Hi Damien,

The code under "if (is_ata)" in scsi_cdl_enable() seems very ATA-specific to me. Shouldn't that code be moved under drivers/ata/?

You seem to be forgetting that when ATA devices are attached to a SAS HBA (one that does not use libsas), then the code in drivers/ata is irrelevant as it is not used at all.

All these "if (is_ata)" cases are addressing shortcomings of the SAT specifications regarding incompatibilities between SPC/SBC and ACS specifications. E.g., for CDL, the CDL feature has an enable/disable action defined in ACS but not in SPC (CDL is always enabled if it is supported).

Quoted 14 lines

Regarding the following code in scsi_add_lun():

if (strncmp(sdev->vendor, "ATA ", 8) == 0)

		sdev->allow_restart = 1;

Other SCSI LLDs set the 'allow_restart' flag in their sdev_configure callback. Can this be done for ATA disks too?

Regarding the code that sets the no_write_same flag:

	if (sdev->is_ata)
		sdev->no_write_same = 1;
Quoted 6 lines

Why is the RCU reader lock held around that code, a lock that protects reading from the VPD pages while the above code does not access the contents of any VPD page?

Other SCSI LLDs set the 'no_write_same' flag in their sdev_configure callback. Can this be done for ATA disks too?

See above. If this is moved to ATA code, then it will never be done for ATA drives connected to SAS HBAs.

Signature · 2 lines
Damien Le Moal
Western Digital Research
Reviewed-by on v2 1/1
#5 lore raw
In reply to Damien Le Moal
Quoted 14 lines · Damien Le Moal

scsi_add_lun() tests the device vendor string of SCSI devices to detect if a SCSI device is in fact an ATA device, in order to correctly handle SATL power management. The function scsi_cdl_enable() also requires knowing if a SCSI device is an ATA device to control the state of the device CDL feature but this function does that by testing for the presence of the VPD page 89h (ATA INFORMATION page). sd_read_write_same() also has a similar test.

Simplify these different methods by adding the is_ata field to struct scsi_device to remember that a SCSI device is in fact an ATA one based on the device vendor name test. This filed can also allow low level SCSI host adapter drivers to take special actions for ATA devices (e.g. to better handle ATA NCQ errors).

Quoted 3 lines

With this, simplify scsi_cdl_enable() and sd_read_write_same().

Thanks,
Igor

#6 lore raw
In reply to Damien Le Moal

Damien,

Quoted 5 lines

Simplify these different methods by adding the is_ata field to struct scsi_device to remember that a SCSI device is in fact an ATA one based on the device vendor name test. This filed can also allow low level SCSI host adapter drivers to take special actions for ATA devices (e.g. to better handle ATA NCQ errors).

Applied to 6.17/scsi-staging, thanks!

Signature · 1 line
Martin K. Petersen
#7 lore raw
In reply to Damien Le Moal
Quoted 9 lines · Damien Le Moal

scsi_add_lun() tests the device vendor string of SCSI devices to detect if a SCSI device is in fact an ATA device, in order to correctly handle SATL power management. The function scsi_cdl_enable() also requires knowing if a SCSI device is an ATA device to control the state of the device CDL feature but this function does that by testing for the presence of the VPD page 89h (ATA INFORMATION page). sd_read_write_same() also has a similar test.

[...]

Applied to 6.17/scsi-queue, thanks!

[1/1] scsi: Remember if a device is an ATA device

      https://git.kernel.org/mkp/scsi/c/b1ba03c49a71
Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering