Skip to contentSkip to conversation

SCSI core patches

PATCH Superseded Bart Van Assche 7 participants 23 messages May 3, 2023 – May 17, 2023

Subsystems (2)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v2 cover SCSI core patches

Hi Martin,

Please consider these SCSI core patches for the next merge window.

Thanks,

Bart.

Changes compared to v1:
- Improved the SCSI tracing patch as requested by Steven Rostedt and Niklas Cassel.
- Added patch "scsi: core: Delay running the queue if the host is blocked"

Bart Van Assche (5):
scsi: core: Use min() instead of open-coding it
scsi: core: Update a source code comment
scsi: core: Trace SCSI sense data
scsi: core: Only kick the requeue list if necessary
scsi: core: Delay running the queue if the host is blocked

 drivers/scsi/scsi_common.c  |  3 +--
 drivers/scsi/scsi_lib.c     | 20 ++++++++++++++------
 include/scsi/scsi_host.h    |  2 +-
 include/trace/events/scsi.h | 21 +++++++++++++++++++--
 4 files changed, 35 insertions(+), 11 deletions(-)
#1 lore raw

PATCH v2 1/5 scsi: core: Use min() instead of open-coding it

Use min() instead of open-coding it in scsi_normalize_sense().

---
 drivers/scsi/scsi_common.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/scsi/scsi_common.c +1 -2
index 6e50e81a8216..24dec80a6253 100644
@@ -176,8 +176,7 @@ bool scsi_normalize_sense(const u8 *sense_buffer, int sb_len,
 		if (sb_len > 2)
 			sshdr->sense_key = (sense_buffer[2] & 0xf);
 		if (sb_len > 7) {
-			sb_len = (sb_len < (sense_buffer[7] + 8)) ?
-					 sb_len : (sense_buffer[7] + 8);
+			sb_len = min(sb_len, sense_buffer[7] + 8);
 			if (sb_len > 12)
 				sshdr->asc = sense_buffer[12];
 			if (sb_len > 13)
Reviewed-by on v2 1/5
#3 lore raw

Re: [PATCH v2 1/5] scsi: core: Use min() instead of open-coding it

In reply to Bart Van Assche
Quoted 13 lines · Bart Van Assche

Use min() instead of open-coding it in scsi_normalize_sense().

---
 drivers/scsi/scsi_common.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/scsi/scsi_common.c +0 -0
index 6e50e81a8216..24dec80a6253 100644

drivers/scsi/scsi_common.c bool scsi_normalize_sense(const u8 *sense_buffer, int sb_len, · line 182

Quoted 9 lines
drivers/scsi/scsi_common.c +1 -2
@@ -176,8 +176,7 @@ bool scsi_normalize_sense(const u8 *sense_buffer, int sb_len,
 		if (sb_len > 2)
 			sshdr->sense_key = (sense_buffer[2] & 0xf);
 		if (sb_len > 7) {
-			sb_len = (sb_len < (sense_buffer[7] + 8)) ?
-					 sb_len : (sense_buffer[7] + 8);
+			sb_len = min(sb_len, sense_buffer[7] + 8);
 			if (sb_len > 12)
 				sshdr->asc = sense_buffer[12];
 			if (sb_len > 13)

Looks good to me.

Signature · 4 lines
Best Regards, Benjamin Block        /        Linux on IBM Z Kernel Development
IBM Deutschland Research & Development GmbH    /   https://www.ibm.com/privacy
Vors. Aufs.-R.: Gregor Pillen         /         Geschäftsführung: David Faller
Sitz der Ges.: Böblingen     /    Registergericht: AmtsG Stuttgart, HRB 243294
#4 lore raw

PATCH v2 2/5 scsi: core: Update a source code comment

In reply to Bart Van Assche

The proc_name SCSI host template attribute is used for:
- The name of the /proc directory if CONFIG_SCSI_PROC_FS=y.
- The contents of the proc_name SCSI host sysfs attribute.

The current comment in include/scsi/scsi_host.h is not correct if CONFIG_SCSI_PROC_FS=n. Hence, change that comment.

Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 include/scsi/scsi_host.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
include/scsi/scsi_host.h +1 -1
index 0f29799efa02..18632adca17e 100644
@@ -365,7 +365,7 @@ struct scsi_host_template {
 
 
 	/*
-	 * Name of proc directory
+	 * Name reported via the proc_name SCSI host sysfs attribute.
 	 */
 	const char *proc_name;
 
#6 lore raw

Re: [PATCH v2 2/5] scsi: core: Update a source code comment

In reply to Bart Van Assche
Quoted 2 lines · Bart Van Assche

The proc_name SCSI host template attribute is used for:
- The name of the /proc directory if CONFIG_SCSI_PROC_FS=y.

But now you remove that case completely? It seems kinda strange to bother touching the comment, but then only switching from one incomplete form to an other?

Quoted 28 lines

- The contents of the proc_name SCSI host sysfs attribute.

The current comment in include/scsi/scsi_host.h is not correct if CONFIG_SCSI_PROC_FS=n. Hence, change that comment.

Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 include/scsi/scsi_host.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
include/scsi/scsi_host.h +1 -1
index 0f29799efa02..18632adca17e 100644
@@ -365,7 +365,7 @@ struct scsi_host_template {
 
 
 	/*
-	 * Name of proc directory
+	 * Name reported via the proc_name SCSI host sysfs attribute.
 	 */
 	const char *proc_name;
 
Signature · 4 lines
Best Regards, Benjamin Block        /        Linux on IBM Z Kernel Development
IBM Deutschland Research & Development GmbH    /   https://www.ibm.com/privacy
Vors. Aufs.-R.: Gregor Pillen         /         Geschäftsführung: David Faller
Sitz der Ges.: Böblingen     /    Registergericht: AmtsG Stuttgart, HRB 243294
#7 lore raw

Re: [PATCH v2 2/5] scsi: core: Update a source code comment

Quoted 7 lines · Benjamin Block
Quoted 2 lines · Bart Van Assche

The proc_name SCSI host template attribute is used for:
- The name of the /proc directory if CONFIG_SCSI_PROC_FS=y.

But now you remove that case completely? It seems kinda strange to bother touching the comment, but then only switching from one incomplete form to an other?

Using procfs as an interface to other information than process information is deprecated, so I'm not sure it's worth mentioning how this name is used if CONFIG_SCSI_PROC_FS=y.

Bart.

#8 lore raw

PATCH v2 3/5 scsi: core: Trace SCSI sense data

In reply to Bart Van Assche

If a command fails, SCSI sense data is essential to determine why it failed. Hence make the sense key, ASC and ASCQ codes available in the ftrace output.

Cc Niklas Cassel <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 include/trace/events/scsi.h | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)
include/trace/events/scsi.h +19 -2
index a2c7befd451a..6c4be1ebe268 100644
@@ -269,9 +269,14 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		__field( unsigned int,	prot_sglen )
 		__field( unsigned char,	prot_op )
 		__dynamic_array(unsigned char,	cmnd, cmd->cmd_len)
+		__field( u8, sense_key )
+		__field( u8, asc )
+		__field( u8, ascq )
 	),
 
 	TP_fast_assign(
+		struct scsi_sense_hdr sshdr;
+
 		__entry->host_no	= cmd->device->host->host_no;
 		__entry->channel	= cmd->device->channel;
 		__entry->id		= cmd->device->id;
@@ -285,11 +290,22 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		__entry->prot_sglen	= scsi_prot_sg_count(cmd);
 		__entry->prot_op	= scsi_get_prot_op(cmd);
 		memcpy(__get_dynamic_array(cmnd), cmd->cmnd, cmd->cmd_len);
+		if (cmd->sense_buffer && SCSI_SENSE_VALID(cmd) &&
+		    scsi_command_normalize_sense(cmd, &sshdr)) {
+			__entry->sense_key = sshdr.sense_key;
+			__entry->asc = sshdr.asc;
+			__entry->ascq = sshdr.ascq;
+		} else {
+			__entry->sense_key = 0;
+			__entry->asc = 0;
+			__entry->ascq = 0;
+		}
 	),
 
 	TP_printk("host_no=%u channel=%u id=%u lun=%u data_sgl=%u prot_sgl=%u " \
 		  "prot_op=%s driver_tag=%d scheduler_tag=%d cmnd=(%s %s raw=%s) " \
-		  "result=(driver=%s host=%s message=%s status=%s)",
+		  "result=(driver=%s host=%s message=%s status=%s "
+		  "sense_key=%u asc=%#x ascq=%#x))",
 		  __entry->host_no, __entry->channel, __entry->id,
 		  __entry->lun, __entry->data_sglen, __entry->prot_sglen,
 		  show_prot_op_name(__entry->prot_op), __entry->driver_tag,
@@ -299,7 +315,8 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		  "DRIVER_OK",
 		  show_hostbyte_name(((__entry->result) >> 16) & 0xff),
 		  "COMMAND_COMPLETE",
-		  show_statusbyte_name(__entry->result & 0xff))
+		  show_statusbyte_name(__entry->result & 0xff),
+		  __entry->sense_key, __entry->asc, __entry->ascq)
 );
 
 DEFINE_EVENT(scsi_cmd_done_timeout_template, scsi_dispatch_cmd_done,
#9 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

Quoted 40 lines · Bart Van Assche

If a command fails, SCSI sense data is essential to determine why it failed. Hence make the sense key, ASC and ASCQ codes available in the ftrace output.

Cc Niklas Cassel <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
  include/trace/events/scsi.h | 21 +++++++++++++++++++--
  1 file changed, 19 insertions(+), 2 deletions(-)
include/trace/events/scsi.h +7 -0
index a2c7befd451a..6c4be1ebe268 100644
@@ -269,9 +269,14 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
  		__field( unsigned int,	prot_sglen )
  		__field( unsigned char,	prot_op )
  		__dynamic_array(unsigned char,	cmnd, cmd->cmd_len)
+		__field( u8, sense_key )
+		__field( u8, asc )
+		__field( u8, ascq )
  	),
 
  	TP_fast_assign(
+		struct scsi_sense_hdr sshdr;
+
  		__entry->host_no	= cmd->device->host->host_no;
  		__entry->channel	= cmd->device->channel;
  		__entry->id		= cmd->device->id;
@@ -285,11 +290,22 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
  		__entry->prot_sglen	= scsi_prot_sg_count(cmd);
  		__entry->prot_op	= scsi_get_prot_op(cmd);
  		memcpy(__get_dynamic_array(cmnd), cmd->cmnd, cmd->cmd_len);
+		if (cmd->sense_buffer && SCSI_SENSE_VALID(cmd) &&
+		    scsi_command_normalize_sense(cmd, &sshdr)) {

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 304

Quoted 10 lines
include/trace/events/scsi.h +8 -0
+			__entry->sense_key = sshdr.sense_key;
+			__entry->asc = sshdr.asc;
+			__entry->ascq = sshdr.ascq;
+		} else {
+			__entry->sense_key = 0;
+			__entry->asc = 0;
+			__entry->ascq = 0;
+		}
  	),
 

Hmm. All this business with scsi_normalize_sense(); there are 15 instances calling scsi_normalize_sense(), and half of it are calling it with scmd->sense-buffer.
Makes one wonder if we shouldn't add fields in the scmd structure, and
do the decoding always...

But for another time, I guess.

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 308

Quoted 5 lines
include/trace/events/scsi.h +2 -1
  	TP_printk("host_no=%u channel=%u id=%u lun=%u data_sgl=%u prot_sgl=%u " \
  		  "prot_op=%s driver_tag=%d scheduler_tag=%d cmnd=(%s %s raw=%s) " \
-		  "result=(driver=%s host=%s message=%s status=%s)",
+		  "result=(driver=%s host=%s message=%s status=%s "
+		  "sense_key=%u asc=%#x ascq=%#x))",

Why two closing braces?

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 322

Quoted 13 lines
include/trace/events/scsi.h +2 -1
  		  __entry->host_no, __entry->channel, __entry->id,
  		  __entry->lun, __entry->data_sglen, __entry->prot_sglen,
  		  show_prot_op_name(__entry->prot_op), __entry->driver_tag,
@@ -299,7 +315,8 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
  		  "DRIVER_OK",
  		  show_hostbyte_name(((__entry->result) >> 16) & 0xff),
  		  "COMMAND_COMPLETE",
-		  show_statusbyte_name(__entry->result & 0xff))
+		  show_statusbyte_name(__entry->result & 0xff),
+		  __entry->sense_key, __entry->asc, __entry->ascq)
  );
 
  DEFINE_EVENT(scsi_cmd_done_timeout_template, scsi_dispatch_cmd_done,

Cheers,

Hannes

#10 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

Quoted 8 lines · Hannes Reinecke
Quoted 5 lines · Bart Van Assche

      TP_printk("host_no=%u channel=%u id=%u lun=%u data_sgl=%u prot_sgl=%u " \
            "prot_op=%s driver_tag=%d scheduler_tag=%d cmnd=(%s %s raw=%s) " \
-          "result=(driver=%s host=%s message=%s status=%s)",
+          "result=(driver=%s host=%s message=%s status=%s "
+          "sense_key=%u asc=%#x ascq=%#x))",

Why two closing braces?

There should only be one closing brace. I will fix this.

Thanks,

Bart.

Reviewed-by on v2 3/5
#11 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

In reply to Bart Van Assche
Quoted 45 lines · Bart Van Assche

If a command fails, SCSI sense data is essential to determine why it failed. Hence make the sense key, ASC and ASCQ codes available in the ftrace output.

Cc Niklas Cassel <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 include/trace/events/scsi.h | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)
include/trace/events/scsi.h +12 -0
index a2c7befd451a..6c4be1ebe268 100644
@@ -269,9 +269,14 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		__field( unsigned int,	prot_sglen )
 		__field( unsigned char,	prot_op )
 		__dynamic_array(unsigned char,	cmnd, cmd->cmd_len)
+		__field( u8, sense_key )
+		__field( u8, asc )
+		__field( u8, ascq )
 	),
 
 	TP_fast_assign(
+		struct scsi_sense_hdr sshdr;
+
 		__entry->host_no	= cmd->device->host->host_no;
 		__entry->channel	= cmd->device->channel;
 		__entry->id		= cmd->device->id;
@@ -285,11 +290,22 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		__entry->prot_sglen	= scsi_prot_sg_count(cmd);
 		__entry->prot_op	= scsi_get_prot_op(cmd);
 		memcpy(__get_dynamic_array(cmnd), cmd->cmnd, cmd->cmd_len);
+		if (cmd->sense_buffer && SCSI_SENSE_VALID(cmd) &&
+		    scsi_command_normalize_sense(cmd, &sshdr)) {
+			__entry->sense_key = sshdr.sense_key;
+			__entry->asc = sshdr.asc;
+			__entry->ascq = sshdr.ascq;
+		} else {
+			__entry->sense_key = 0;

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 308

Quoted 10 lines
include/trace/events/scsi.h +5 -1
+			__entry->asc = 0;
+			__entry->ascq = 0;
+		}
 	),
 
 	TP_printk("host_no=%u channel=%u id=%u lun=%u data_sgl=%u prot_sgl=%u " \
 		  "prot_op=%s driver_tag=%d scheduler_tag=%d cmnd=(%s %s raw=%s) " \
-		  "result=(driver=%s host=%s message=%s status=%s)",
+		  "result=(driver=%s host=%s message=%s status=%s "
+		  "sense_key=%u asc=%#x ascq=%#x))",

With the extra parentheses removed (as mentioned by Hannes):

Reviewed-by on v2 3/5
#12 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

In reply to Bart Van Assche
Quoted 3 lines · Bart Van Assche

If a command fails, SCSI sense data is essential to determine why it failed. Hence make the sense key, ASC and ASCQ codes available in the ftrace output.

Looks good:

Although, I'd also love to see pre-decoded ASC and ASCQ codes in the scsi_cmnd at some point.

#13 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

Christoph,

Quoted 2 lines

Although, I'd also love to see pre-decoded ASC and ASCQ codes in the scsi_cmnd at some point.

Yep! It would be nice to have these readily available.

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#14 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

Quoted 4 lines · Martin K. Petersen
Quoted 2 lines

Although, I'd also love to see pre-decoded ASC and ASCQ codes in the scsi_cmnd at some point.

Yep! It would be nice to have these readily available.

Implementing this seems nontrivial and risky to me. As an example, whether or not scsi_decide_disposition() calls scsi_check_sense() depends on the command result. Guaranteeing that scsi_command_normalize_sense() is called independent of the command result implies moving this call from scsi_check_sense() to somewhere else, e.g. scsi_decide_disposition(). Unfortunately scsi_check_sense() is not only called from the SCSI completion path but also by ATA code that is not related to the SCSI completion path (ata_eh_analyze_tf()). So I consider this change as out-of-scope for this patch series.

Bart.

#15 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

In reply to Bart Van Assche
Quoted 35 lines · Bart Van Assche

If a command fails, SCSI sense data is essential to determine why it failed. Hence make the sense key, ASC and ASCQ codes available in the ftrace output.

Cc Niklas Cassel <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 include/trace/events/scsi.h | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)
include/trace/events/scsi.h +5 -0
index a2c7befd451a..6c4be1ebe268 100644
@@ -269,9 +269,14 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		__field( unsigned int,	prot_sglen )
 		__field( unsigned char,	prot_op )
 		__dynamic_array(unsigned char,	cmnd, cmd->cmd_len)
+		__field( u8, sense_key )
+		__field( u8, asc )
+		__field( u8, ascq )
 	),
 
 	TP_fast_assign(
+		struct scsi_sense_hdr sshdr;
+
 		__entry->host_no	= cmd->device->host->host_no;
 		__entry->channel	= cmd->device->channel;
 		__entry->id		= cmd->device->id;

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 293

Quoted 4 lines
include/trace/events/scsi.h +1 -0
@@ -285,11 +290,22 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		__entry->prot_sglen	= scsi_prot_sg_count(cmd);
 		__entry->prot_op	= scsi_get_prot_op(cmd);
 		memcpy(__get_dynamic_array(cmnd), cmd->cmnd, cmd->cmd_len);
+		if (cmd->sense_buffer && SCSI_SENSE_VALID(cmd) &&

Can't hurt to have these explicitly here, but these checks are also done by scsi_command_normalize_sense() AFAICT.

Quoted 6 lines
include/trace/events/scsi.h +6 -0
+		    scsi_command_normalize_sense(cmd, &sshdr)) {
+			__entry->sense_key = sshdr.sense_key;
+			__entry->asc = sshdr.asc;
+			__entry->ascq = sshdr.ascq;
+		} else {
+			__entry->sense_key = 0;

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 308

Quoted 10 lines
include/trace/events/scsi.h +5 -1
+			__entry->asc = 0;
+			__entry->ascq = 0;
+		}
 	),
 
 	TP_printk("host_no=%u channel=%u id=%u lun=%u data_sgl=%u prot_sgl=%u " \
 		  "prot_op=%s driver_tag=%d scheduler_tag=%d cmnd=(%s %s raw=%s) " \
-		  "result=(driver=%s host=%s message=%s status=%s)",
+		  "result=(driver=%s host=%s message=%s status=%s "
+		  "sense_key=%u asc=%#x ascq=%#x))",

In SPC, these are all in base 16, and some existing functions in scsi_logging.c format Sense Key as base 16. We probably should keep this consistent and also format Sense Key with %#x.

Quoted 13 lines
include/trace/events/scsi.h +2 -1
 		  __entry->host_no, __entry->channel, __entry->id,
 		  __entry->lun, __entry->data_sglen, __entry->prot_sglen,
 		  show_prot_op_name(__entry->prot_op), __entry->driver_tag,
@@ -299,7 +315,8 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
 		  "DRIVER_OK",
 		  show_hostbyte_name(((__entry->result) >> 16) & 0xff),
 		  "COMMAND_COMPLETE",
-		  show_statusbyte_name(__entry->result & 0xff))
+		  show_statusbyte_name(__entry->result & 0xff),
+		  __entry->sense_key, __entry->asc, __entry->ascq)
 );
 
 DEFINE_EVENT(scsi_cmd_done_timeout_template, scsi_dispatch_cmd_done,
Signature · 4 lines
Best Regards, Benjamin Block        /        Linux on IBM Z Kernel Development
IBM Deutschland Research & Development GmbH    /   https://www.ibm.com/privacy
Vors. Aufs.-R.: Gregor Pillen         /         Geschäftsführung: David Faller
Sitz der Ges.: Böblingen     /    Registergericht: AmtsG Stuttgart, HRB 243294
#16 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

Quoted 9 lines · Benjamin Block

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 293

Quoted 5 lines · Bart Van Assche
include/trace/events/scsi.h +1 -0
@@ -285,11 +290,22 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
  		__entry->prot_sglen	= scsi_prot_sg_count(cmd);
  		__entry->prot_op	= scsi_get_prot_op(cmd);
  		memcpy(__get_dynamic_array(cmnd), cmd->cmnd, cmd->cmd_len);
+		if (cmd->sense_buffer && SCSI_SENSE_VALID(cmd) &&

Can't hurt to have these explicitly here, but these checks are also done by scsi_command_normalize_sense() AFAICT.

Niklas requested these checks as a performance optimization.

Quoted 12 lines
Quoted 12 lines
include/trace/events/scsi.h +9 -0
+		    scsi_command_normalize_sense(cmd, &sshdr)) {
+			__entry->sense_key = sshdr.sense_key;
+			__entry->asc = sshdr.asc;
+			__entry->ascq = sshdr.ascq;
+		} else {
+			__entry->sense_key = 0;
+			__entry->asc = 0;
+			__entry->ascq = 0;
+		}
  	),
 
Quoted 9 lines

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 308

Quoted 5 lines
include/trace/events/scsi.h +2 -1
  	TP_printk("host_no=%u channel=%u id=%u lun=%u data_sgl=%u prot_sgl=%u " \
  		  "prot_op=%s driver_tag=%d scheduler_tag=%d cmnd=(%s %s raw=%s) " \
-		  "result=(driver=%s host=%s message=%s status=%s)",
+		  "result=(driver=%s host=%s message=%s status=%s "
+		  "sense_key=%u asc=%#x ascq=%#x))",

In SPC, these are all in base 16, and some existing functions in scsi_logging.c format Sense Key as base 16. We probably should keep this consistent and also format Sense Key with %#x.

I can make this change.

Thanks,

Bart.

#17 lore raw

Re: [PATCH v2 3/5] scsi: core: Trace SCSI sense data

Quoted 11 lines · Bart Van Assche
Quoted 9 lines · Benjamin Block

include/trace/events/scsi.h DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template, · line 293

Quoted 5 lines · Bart Van Assche
include/trace/events/scsi.h +1 -0
@@ -285,11 +290,22 @@ DECLARE_EVENT_CLASS(scsi_cmd_done_timeout_template,
  		__entry->prot_sglen	= scsi_prot_sg_count(cmd);
  		__entry->prot_op	= scsi_get_prot_op(cmd);
  		memcpy(__get_dynamic_array(cmnd), cmd->cmnd, cmd->cmd_len);
+		if (cmd->sense_buffer && SCSI_SENSE_VALID(cmd) &&

Can't hurt to have these explicitly here, but these checks are also done by scsi_command_normalize_sense() AFAICT.

Quoted 1 line

Niklas requested these checks as a performance optimization.

Oh okay, didn't remember that. Fair enough.

Signature · 4 lines
Best Regards, Benjamin Block        /        Linux on IBM Z Kernel Development
IBM Deutschland Research & Development GmbH    /   https://www.ibm.com/privacy
Vors. Aufs.-R.: Gregor Pillen         /         Geschäftsführung: David Faller
Sitz der Ges.: Böblingen     /    Registergericht: AmtsG Stuttgart, HRB 243294
#18 lore raw

PATCH v2 4/5 scsi: core: Only kick the requeue list if necessary

In reply to Bart Van Assche

Instead of running the request queue of each device associated with a host every 3 ms (BLK_MQ_RESOURCE_DELAY) while host error handling is in progress, run the request queue after error handling has finished.

Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 drivers/scsi/scsi_lib.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)
drivers/scsi/scsi_lib.c +13 -5
index e59eb0cbfc83..a34390d35f1d 100644
@@ -122,11 +122,9 @@ static void scsi_mq_requeue_cmd(struct scsi_cmnd *cmd, unsigned long msecs)
 		WARN_ON_ONCE(true);
 	}
 
-	if (msecs) {
-		blk_mq_requeue_request(rq, false);
+	blk_mq_requeue_request(rq, false);
+	if (!scsi_host_in_recovery(cmd->device->host))
 		blk_mq_delay_kick_requeue_list(rq->q, msecs);
-	} else
-		blk_mq_requeue_request(rq, true);
 }
 
 /**
@@ -165,7 +163,8 @@ static void __scsi_queue_insert(struct scsi_cmnd *cmd, int reason, bool unbusy)
 	 */
 	cmd->result = 0;
 
-	blk_mq_requeue_request(scsi_cmd_to_rq(cmd), true);
+	blk_mq_requeue_request(scsi_cmd_to_rq(cmd),
+			       !scsi_host_in_recovery(cmd->device->host));
 }
 
 /**
@@ -462,10 +461,16 @@ void scsi_requeue_run_queue(struct work_struct *work)
 	scsi_run_queue(q);
 }
 
+/*
+ * Transfer requests from the requeue_list to from where these can be dispatched
+ * and run the request queues.
+ */
 void scsi_run_host_queues(struct Scsi_Host *shost)
 {
 	struct scsi_device *sdev;
 
+	shost_for_each_device(sdev, shost)
+		blk_mq_kick_requeue_list(sdev->request_queue);
 	shost_for_each_device(sdev, shost)
 		scsi_run_queue(sdev->request_queue);
 }
@@ -499,6 +504,9 @@ static void scsi_mq_uninit_cmd(struct scsi_cmnd *cmd)
 
 static void scsi_run_queue_async(struct scsi_device *sdev)
 {
+	if (scsi_host_in_recovery(sdev->host))
+		return;
+
 	if (scsi_target(sdev)->single_lun ||
 	    !list_empty(&sdev->host->starved_list)) {
 		kblockd_schedule_work(&sdev->requeue_work);
#19 lore raw

Re: [PATCH v2 4/5] scsi: core: Only kick the requeue list if necessary

Quoted 48 lines · Bart Van Assche

Instead of running the request queue of each device associated with a host every 3 ms (BLK_MQ_RESOURCE_DELAY) while host error handling is in progress, run the request queue after error handling has finished.

Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 drivers/scsi/scsi_lib.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)
drivers/scsi/scsi_lib.c +6 -5
index e59eb0cbfc83..a34390d35f1d 100644
@@ -122,11 +122,9 @@ static void scsi_mq_requeue_cmd(struct scsi_cmnd *cmd, unsigned long msecs)
 		WARN_ON_ONCE(true);
 	}
 
-	if (msecs) {
-		blk_mq_requeue_request(rq, false);
+	blk_mq_requeue_request(rq, false);
+	if (!scsi_host_in_recovery(cmd->device->host))
 		blk_mq_delay_kick_requeue_list(rq->q, msecs);
-	} else
-		blk_mq_requeue_request(rq, true);
 }
 
 /**
@@ -165,7 +163,8 @@ static void __scsi_queue_insert(struct scsi_cmnd *cmd, int reason, bool unbusy)
 	 */
 	cmd->result = 0;
 
-	blk_mq_requeue_request(scsi_cmd_to_rq(cmd), true);
+	blk_mq_requeue_request(scsi_cmd_to_rq(cmd),
+			       !scsi_host_in_recovery(cmd->device->host));
 }
 
 /**
@@ -462,10 +461,16 @@ void scsi_requeue_run_queue(struct work_struct *work)
 	scsi_run_queue(q);
 }
 
+/*
+ * Transfer requests from the requeue_list to from where these can be dispatched

drivers/scsi/scsi_lib.c void scsi_requeue_run_queue(struct work_struct *work) · line 475

Quoted 10 lines
drivers/scsi/scsi_lib.c +4 -0
+ * and run the request queues.
+ */
 void scsi_run_host_queues(struct Scsi_Host *shost)
 {
 	struct scsi_device *sdev;
 
+	shost_for_each_device(sdev, shost)
+		blk_mq_kick_requeue_list(sdev->request_queue);
 	shost_for_each_device(sdev, shost)
 		scsi_run_queue(sdev->request_queue);

You may move blk_mq_kick_requeue_list() to the following loop, or even inside scsi_run_queue(), which isn't called in fast patch.

Thanks,
Ming

#21 lore raw

PATCH v2 5/5 scsi: core: Delay running the queue if the host is blocked

In reply to Bart Van Assche

Tell the block layer to rerun the queue after a delay instead of immediately if the SCSI LLD decided to block the host.

Cc Christoph Hellwig <[email protected]>
Cc Ming Lei <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc John Garry <[email protected]>
Cc Mike Christie <[email protected]>
---
 drivers/scsi/scsi_lib.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/scsi/scsi_lib.c +1 -1
index a34390d35f1d..370be5994a45 100644
@@ -1772,7 +1772,7 @@ static blk_status_t scsi_queue_rq(struct blk_mq_hw_ctx *hctx,
 		break;
 	case BLK_STS_RESOURCE:
 	case BLK_STS_ZONE_RESOURCE:
-		if (scsi_device_blocked(sdev))
+		if (scsi_device_blocked(sdev) || shost->host_self_blocked)
 			ret = BLK_STS_DEV_RESOURCE;
 		break;
 	case BLK_STS_AGAIN: