[PATCH v2 0/4] target: Updates related to UASP

[PATCH v2 0/4] target: Updates related to UASP

[Thinh Nguyen]

Here are some changes splitted from the series below [*]:
"[PATCH 00/36] usb: gadget: f_tcm: Enhance UASP driver"

Cc: linux-usb@vger.kernel.org

Changes in v2:
 - From review, remove the following patches either because they are incorrect
   or not needed:
     target: Don't drain empty list
     target: Does tmr notify on aborted command
     target: Don't remove command too early
     target: Return Function Complete
     target: Don't do tmr_notify on empty aborted list
     target: Refactor core_tmr_abort_task
     target: Don't respond TMR_LUN_DOES_NOT_EXIST for all TMR failure
     target: Introduce target_submit_tmr_fail_response
     target: Include INQUIRY length
 - Minor fixes to the remaining patches


[*] https://lore.kernel.org/linux-usb/cover.1657149962.git.Thinh.Nguyen@synopsys.com/





Thinh Nguyen (4):
  target: Handle MI_REPORT_SUPPORTED_OPERATION_CODES
  target: Implement TMR_ABORT_TASK_SET
  target: Properly set Sense Data Length of CHECK CONDITION
  target: Properly set Sense data length when copy sense

 drivers/target/target_core_alua.c      | 70 ++++++++++++++++++++++++++
 drivers/target/target_core_alua.h      |  2 +
 drivers/target/target_core_spc.c       | 14 +++++-
 drivers/target/target_core_tmr.c       | 16 ++++--
 drivers/target/target_core_transport.c | 21 ++++++--
 5 files changed, 111 insertions(+), 12 deletions(-)


base-commit: 88a15fbb47db483d06b12b1ae69f114b96361a96
-- 
2.28.0

[PATCH v2 2/4] target: Implement TMR_ABORT_TASK_SET

[Thinh Nguyen]

Task ABORT TASK SET function is required by SCSI transport protocol
standards (SAM-4 r14 section 7.3). It is similar to ABORT TASK
function, but it applies to all commands received on a specified I_T
nexus rather than a specific referenced command. Modify
core_tmr_abort_task() to support TMR_ABORT_TASK_SET.

Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
---
 Changes in v2:
 - None

 drivers/target/target_core_tmr.c       | 16 +++++++++++-----
 drivers/target/target_core_transport.c |  2 +-
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/drivers/target/target_core_tmr.c b/drivers/target/target_core_tmr.c
index bac111456fa1..1ea72e15f872 100644
--- a/drivers/target/target_core_tmr.c
+++ b/drivers/target/target_core_tmr.c
@@ -131,11 +131,13 @@ void core_tmr_abort_task(
 				continue;
 
 			ref_tag = se_cmd->tag;
-			if (tmr->ref_task_tag != ref_tag)
-				continue;
+			if (tmr->function == TMR_ABORT_TASK) {
+				if (tmr->ref_task_tag != ref_tag)
+					continue;
 
-			pr_err("ABORT_TASK: Found referenced %s task_tag: %llu\n",
-			       se_cmd->se_tfo->fabric_name, ref_tag);
+				pr_err("ABORT_TASK: Found referenced %s task_tag: %llu\n",
+				       se_cmd->se_tfo->fabric_name, ref_tag);
+			}
 
 			spin_lock(&se_sess->sess_cmd_lock);
 			rc = __target_check_io_state(se_cmd, se_sess, 0);
@@ -158,7 +160,11 @@ void core_tmr_abort_task(
 			       ref_tag);
 			tmr->response = TMR_FUNCTION_COMPLETE;
 			atomic_long_inc(&dev->aborts_complete);
-			return;
+
+			if (tmr->function == TMR_ABORT_TASK)
+				return;
+
+			spin_lock_irqsave(&dev->queues[i].lock, flags);
 		}
 		spin_unlock_irqrestore(&dev->queues[i].lock, flags);
 	}
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 7838dc20f713..8c386142ef90 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -3519,9 +3519,9 @@ static void target_tmr_work(struct work_struct *work)
 
 	switch (tmr->function) {
 	case TMR_ABORT_TASK:
+	case TMR_ABORT_TASK_SET:
 		core_tmr_abort_task(dev, tmr, cmd->se_sess);
 		break;
-	case TMR_ABORT_TASK_SET:
 	case TMR_CLEAR_ACA:
 	case TMR_CLEAR_TASK_SET:
 		tmr->response = TMR_TASK_MGMT_FUNCTION_NOT_SUPPORTED;
-- 
2.28.0

Re: [PATCH v2 2/4] target: Implement TMR_ABORT_TASK_SET

[Mike Christie]

On 7/18/22 9:07 PM, Thinh Nguyen wrote:
> Task ABORT TASK SET function is required by SCSI transport protocol

What OS is using this and how do they use it? For the latter, does the
OS try an abort for each cmd first, then try an abort task set if the
aborts fail (does fail mean get a response that indicates failure and
also does a timeout count)? Or does it start with the abort task set?

I'm asking because it looks like if it does an abort first, then the
abort task set will always return TMR_TASK_DOES_NOT_EXIST. For the abort
we will remove the cmds from the state_list so if the abort task set runs
after the initiator has tried to abort all the commands it will never
find any.

> standards (SAM-4 r14 section 7.3). It is similar to ABORT TASK
> function, but it applies to all commands received on a specified I_T
> nexus rather than a specific referenced command. Modify
> core_tmr_abort_task() to support TMR_ABORT_TASK_SET.
> 

Re: [PATCH v2 2/4] target: Implement TMR_ABORT_TASK_SET

[Thinh Nguyen]

On 7/19/2022, Mike Christie wrote:
> On 7/18/22 9:07 PM, Thinh Nguyen wrote:
>> Task ABORT TASK SET function is required by SCSI transport protocol
> What OS is using this and how do they use it? For the latter, does the
> OS try an abort for each cmd first, then try an abort task set if the
> aborts fail (does fail mean get a response that indicates failure and
> also does a timeout count)? Or does it start with the abort task set?

It's not from any real driver. It's from the USB Compliant Verification 
(https://www.usb.org/document-library/usb3cv). It uses the command for 
UASP compliant test.

The test only ever aborts a single command at a time, so I can't confirm 
your following questions. The SAM4-r14 wasn't clear on those questions 
either.

> I'm asking because it looks like if it does an abort first, then the
> abort task set will always return TMR_TASK_DOES_NOT_EXIST. For the abort
> we will remove the cmds from the state_list so if the abort task set runs
> after the initiator has tried to abort all the commands it will never
> find any.

I didn't notice since I dropped a patch where I removed the 
TMR_TASK_DOES_NOT_EXIST and UASP converts this to RC_TMF_COMPLETE. UASP 
respond to FUNCTION COMPLETE with RC_TMF_COMPLETE. I'll can make a fix 
to that.

If there's any suggestion to implement this, please advise.

Thanks,
Thinh

>> standards (SAM-4 r14 section 7.3). It is similar to ABORT TASK
>> function, but it applies to all commands received on a specified I_T
>> nexus rather than a specific referenced command. Modify
>> core_tmr_abort_task() to support TMR_ABORT_TASK_SET.
>>

Re: [PATCH v2 2/4] target: Implement TMR_ABORT_TASK_SET

[Konstantin Shelekhin]

On Tue, Jul 19, 2022 at 10:56:07AM -0500, Mike Christie wrote:
> «Внимание! Данное письмо от внешнего адресата!»
> 
> On 7/18/22 9:07 PM, Thinh Nguyen wrote:
> > Task ABORT TASK SET function is required by SCSI transport protocol
> 
> What OS is using this and how do they use it? For the latter, does the
> OS try an abort for each cmd first, then try an abort task set if the
> aborts fail (does fail mean get a response that indicates failure and
> also does a timeout count)? Or does it start with the abort task set?

AIX IIRC. However, this feature also requires valid bits in one of the
VPD pages.

[PATCH v2 3/4] target: Properly set Sense Data Length of CHECK CONDITION

[Thinh Nguyen]

CHECK CONDITION returns sense data, and sense data is minimum 8 bytes
long plus additional sense data length in the data buffer. Don't just
set the allocated buffer length to the cmd->scsi_sense_length and check
the sense data for that.

See SPC4-r37 section 4.5.2.1.

Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
---
 Changes in v2:
 - Make sure it doesn't overflow and properly pass TRANSPORT_SENSE_BUFFER to
   scsi_set_sense_information()

 drivers/target/target_core_transport.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 8c386142ef90..db53b326048d 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -3459,12 +3459,19 @@ static void translate_sense_reason(struct se_cmd *cmd, sense_reason_t reason)
 
 	cmd->se_cmd_flags |= SCF_EMULATED_TASK_SENSE;
 	cmd->scsi_status = SAM_STAT_CHECK_CONDITION;
-	cmd->scsi_sense_length  = TRANSPORT_SENSE_BUFFER;
+
 	scsi_build_sense_buffer(desc_format, buffer, key, asc, ascq);
 	if (sd->add_sense_info)
 		WARN_ON_ONCE(scsi_set_sense_information(buffer,
-							cmd->scsi_sense_length,
+							TRANSPORT_SENSE_BUFFER,
 							cmd->sense_info) < 0);
+	/*
+	 * CHECK CONDITION returns sense data, and sense data is minimum 8
+	 * bytes long plus additional Sense Data Length.
+	 * See SPC4-r37 section 4.5.2.1.
+	 */
+	cmd->scsi_sense_length = min_t(u16, buffer[7] + 8,
+				       TRANSPORT_SENSE_BUFFER);
 }
 
 int
-- 
2.28.0