Skip to contentSkip to conversation

scsi: More detailed I/O errors

PATCH In review Hannes Reinecke 2 participants 6 messages Jul 1, 2013 – Jul 2, 2013

Subsystems (4)
VFS SCSI Block layer Device mapper
lore
Share & tools
lore

Hannes Reinecke

author
#0 lore raw

v3 cover scsi: More detailed I/O errors

As discussed at LSF the SCSI stack already knows about several error conditions, it's just not well documented. Also there are some more conditions upper layers might be interested in.

This patchset improves the documentation of the existing I/O error codes and adds two more error codes, ENOSPC for thin provisioning failure and ENODATA for medium error.

Changes since the initial version:
- Correctly document all return codes in
__scsi_error_from_host_byte()
- Set hostbyte status in scsi_check_sense()

Changes since v2:
- Modify dm-mpath to handle ENOSPC

Hannes Reinecke (4):
scsi: Document enhanced error codes
scsi: Set hostbyte status in scsi_check_sense()
scsi: return ENOSPC on thin provisioning failure
scsi: Return ENODATA on medium error

 block/blk-core.c          |  6 ++++++
 drivers/md/dm-mpath.c     | 16 +++++++++++++++-
 drivers/scsi/scsi_error.c | 28 +++++++++++++---------------
 drivers/scsi/scsi_lib.c   | 22 ++++++++++++++++++++++
 include/scsi/scsi.h       |  3 ++-
 5 files changed, 58 insertions(+), 17 deletions(-)
Signature · 1 line
1.7.12.4
author
#1 lore raw

PATCH v3 1/4 scsi: Document enhanced error codes

Document the various error codes returned on I/O failure.

Signed-off-by Hannes Reinecke <[email protected]>
---
 drivers/scsi/scsi_lib.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)
drivers/scsi/scsi_lib.c +12 -0
index 86d5220..d3fc50f 100644
@@ -700,6 +700,18 @@ void scsi_release_buffers(struct scsi_cmnd *cmd)
 }
 EXPORT_SYMBOL(scsi_release_buffers);
 
+/**
+ * __scsi_error_from_host_byte - translate SCSI error code into errno
+ * @cmd:	SCSI command (unused)
+ * @result:	scsi error code
+ *
+ * Translate SCSI error code into standard UNIX errno.
+ * Return values:
+ * -ENOLINK	temporary transport failure
+ * -EREMOTEIO	permanent target failure, do not retry
+ * -EBADE	permanent nexus failure, retry on other path
+ * -EIO		unspecified I/O error
+ */
 static int __scsi_error_from_host_byte(struct scsi_cmnd *cmd, int result)
 {
 	int error = 0;
Signature · 1 line
1.7.12.4
author
#2 lore raw

PATCH v3 2/4 scsi: Set hostbyte status in scsi_check_sense()

In reply to Hannes Reinecke

We should be modifying the host_byte status in scsi_check_sense() directly; this saves us to introduce a special return code for each and every condition.

Signed-off-by Hannes Reinecke <[email protected]>
---
 drivers/scsi/scsi_error.c | 21 +++++++--------------
 include/scsi/scsi.h       |  1 -
 2 files changed, 7 insertions(+), 15 deletions(-)
drivers/scsi/scsi_error.c +7 -14
index 84369f2..e5cb162 100644
@@ -248,7 +248,7 @@ static inline void scsi_eh_prt_fail_stats(struct Scsi_Host *shost,
  * @scmd:	Cmd to have sense checked.
  *
  * Return value:
- *	SUCCESS or FAILED or NEEDS_RETRY or TARGET_ERROR
+ *	SUCCESS or FAILED or NEEDS_RETRY or ADD_TO_MLQUEUE
  *
  * Notes:
  *	When a deferred error is detected the current command has

@@ -379,13 +379,15 @@ static int scsi_check_sense(struct scsi_cmnd *scmd)
 	case MISCOMPARE:
 	case BLANK_CHECK:
 	case DATA_PROTECT:
-		return TARGET_ERROR;
+		set_host_byte(scmd, DID_TARGET_FAILURE);
+		return SUCCESS;
 
 	case MEDIUM_ERROR:
 		if (sshdr.asc == 0x11 || /* UNRECOVERED READ ERR */
 		    sshdr.asc == 0x13 || /* AMNF DATA FIELD */
 		    sshdr.asc == 0x14) { /* RECORD NOT FOUND */
-			return TARGET_ERROR;
+			set_host_byte(scmd, DID_TARGET_FAILURE);
+			return SUCCESS;
 		}
 		return NEEDS_RETRY;
 

@@ -393,14 +395,14 @@ static int scsi_check_sense(struct scsi_cmnd *scmd)
 		if (scmd->device->retry_hwerror)
 			return ADD_TO_MLQUEUE;
 		else
-			return TARGET_ERROR;
+			set_host_byte(scmd, DID_TARGET_FAILURE);
 
 	case ILLEGAL_REQUEST:
 		if (sshdr.asc == 0x20 || /* Invalid command operation code */
 		    sshdr.asc == 0x21 || /* Logical block address out of range */
 		    sshdr.asc == 0x24 || /* Invalid field in cdb */
 		    sshdr.asc == 0x26) { /* Parameter value invalid */
-			return TARGET_ERROR;
+			set_host_byte(scmd, DID_TARGET_FAILURE);
 		}
 		return SUCCESS;
 

@@ -863,7 +865,6 @@ retry:
 		case SUCCESS:
 		case NEEDS_RETRY:
 		case FAILED:
-		case TARGET_ERROR:
 			break;
 		case ADD_TO_MLQUEUE:
 			rtn = NEEDS_RETRY;

@@ -1690,14 +1691,6 @@ int scsi_decide_disposition(struct scsi_cmnd *scmd)
 		rtn = scsi_check_sense(scmd);
 		if (rtn == NEEDS_RETRY)
 			goto maybe_retry;
-		else if (rtn == TARGET_ERROR) {
-			/*
-			 * Need to modify host byte to signal a
-			 * permanent target failure
-			 */
-			set_host_byte(scmd, DID_TARGET_FAILURE);
-			rtn = SUCCESS;
-		}
 		/* if rtn == FAILED, we have no sense information;
 		 * returning FAILED will wake the error handler thread
 		 * to collect the sense and redo the decide
include/scsi/scsi.h +0 -1
index 66216c1..07d4bd916 100644
@@ -481,7 +481,6 @@ static inline int scsi_is_wlun(unsigned int lun)
 #define TIMEOUT_ERROR   0x2007
 #define SCSI_RETURN_NOT_HANDLED   0x2008
 #define FAST_IO_FAIL	0x2009
-#define TARGET_ERROR    0x200A
 
 /*
  * Midlevel queue return values.
Signature · 1 line
1.7.12.4
author
#3 lore raw

PATCH v3 3/4 scsi: return ENOSPC on thin provisioning failure

In reply to Hannes Reinecke

When the thin provisioning hard threshold is reached we should return ENOSPC to inform upper layers about this fact.

Signed-off-by Hannes Reinecke <[email protected]>
---
 block/blk-core.c          | 3 +++
 drivers/scsi/scsi_error.c | 7 ++++++-
 drivers/scsi/scsi_lib.c   | 5 +++++
 include/scsi/scsi.h       | 1 +
 4 files changed, 15 insertions(+), 1 deletion(-)
block/blk-core.c +3 -0
index d5745b5..170607c 100644
@@ -2315,6 +2315,9 @@ bool blk_update_request(struct request *req, int error, unsigned int nr_bytes)
 		case -EBADE:
 			error_type = "critical nexus";
 			break;
+		case -ENOSPC:
+			error_type = "critical space allocation";
+			break;
 		case -EIO:
 		default:
 			error_type = "I/O";
drivers/scsi/scsi_error.c +6 -1
index e5cb162..aeec0e7 100644
@@ -374,11 +374,16 @@ static int scsi_check_sense(struct scsi_cmnd *scmd)
 		return SUCCESS;
 
 		/* these are not supported */
+	case DATA_PROTECT:
+		if (sshdr.asc == 0x27 && sshdr.ascq == 0x07) {
+			/* Thin provisioning hard threshold reached */
+			set_host_byte(scmd, DID_ALLOC_FAILURE);
+			return SUCCESS;
+		}
 	case COPY_ABORTED:
 	case VOLUME_OVERFLOW:
 	case MISCOMPARE:
 	case BLANK_CHECK:
-	case DATA_PROTECT:
 		set_host_byte(scmd, DID_TARGET_FAILURE);
 		return SUCCESS;
drivers/scsi/scsi_lib.c +5 -0
index d3fc50f..8a2f92d 100644
@@ -710,6 +710,7 @@ EXPORT_SYMBOL(scsi_release_buffers);
  * -ENOLINK	temporary transport failure
  * -EREMOTEIO	permanent target failure, do not retry
  * -EBADE	permanent nexus failure, retry on other path
+ * -ENOSPC	No write space available
  * -EIO		unspecified I/O error
  */
 static int __scsi_error_from_host_byte(struct scsi_cmnd *cmd, int result)

@@ -728,6 +729,10 @@ static int __scsi_error_from_host_byte(struct scsi_cmnd *cmd, int result)
 		set_host_byte(cmd, DID_OK);
 		error = -EBADE;
 		break;
+	case DID_ALLOC_FAILURE:
+		set_host_byte(cmd, DID_OK);
+		error = -ENOSPC;
+		break;
 	default:
 		error = -EIO;
 		break;
include/scsi/scsi.h +1 -0
index 07d4bd916..7d5b8e2 100644
@@ -452,6 +452,7 @@ static inline int scsi_is_wlun(unsigned int lun)
 				 * other paths */
 #define DID_NEXUS_FAILURE 0x11  /* Permanent nexus failure, retry on other
 				 * paths might yield different results */
+#define DID_ALLOC_FAILURE 0x12  /* Space allocation on the device failed */
 #define DRIVER_OK       0x00	/* Driver status                           */
 
 /*
Signature · 1 line
1.7.12.4
author
#4 lore raw

PATCH v3 4/4 scsi: Return ENODATA on medium error

In reply to Hannes Reinecke

When a medium error is detected the SCSI stack should return ENODATA to the upper layers.

Cc Jun'ichi Nomura <[email protected]>
Signed-off-by Hannes Reinecke <[email protected]>
---
 block/blk-core.c          |  3 +++
 drivers/md/dm-mpath.c     | 16 +++++++++++++++-
 drivers/scsi/scsi_error.c |  2 +-
 drivers/scsi/scsi_lib.c   |  5 +++++
 include/scsi/scsi.h       |  1 +
 5 files changed, 25 insertions(+), 2 deletions(-)
block/blk-core.c +3 -0
index 170607c..1093ca1 100644
@@ -2318,6 +2318,9 @@ bool blk_update_request(struct request *req, int error, unsigned int nr_bytes)
 		case -ENOSPC:
 			error_type = "critical space allocation";
 			break;
+		case -ENODATA:
+			error_type = "critical medium";
+			break;
 		case -EIO:
 		default:
 			error_type = "I/O";
drivers/md/dm-mpath.c +15 -1
index bdf26f5..57896cea 100644
@@ -1261,6 +1261,20 @@ static void activate_path(struct work_struct *work)
 				pg_init_done, pgpath);
 }
 
+static int noretry_error(int error)
+{
+	switch(error) {
+	case -EOPNOTSUPP:
+	case -EREMOTEIO:
+	case -EILSEQ:
+	case -ENODATA:
+		return 1;
+	}
+
+	/* Anything else could be a path failure, so should be retried */
+	return 0;
+}
+
 /*
  * end_io handling
  */

@@ -1284,7 +1298,7 @@ static int do_end_io(struct multipath *m, struct request *clone,
 	if (!error && !clone->errors)
 		return 0;	/* I/O complete */
 
-	if (error == -EOPNOTSUPP || error == -EREMOTEIO || error == -EILSEQ)
+	if (noretry_error(error))
 		return error;
 
 	if (mpio->pgpath)
drivers/scsi/scsi_error.c +1 -1
index aeec0e7..7f794d4 100644
@@ -391,7 +391,7 @@ static int scsi_check_sense(struct scsi_cmnd *scmd)
 		if (sshdr.asc == 0x11 || /* UNRECOVERED READ ERR */
 		    sshdr.asc == 0x13 || /* AMNF DATA FIELD */
 		    sshdr.asc == 0x14) { /* RECORD NOT FOUND */
-			set_host_byte(scmd, DID_TARGET_FAILURE);
+			set_host_byte(scmd, DID_MEDIUM_ERROR);
 			return SUCCESS;
 		}
 		return NEEDS_RETRY;
drivers/scsi/scsi_lib.c +5 -0
index 8a2f92d..70ca1f6 100644
@@ -711,6 +711,7 @@ EXPORT_SYMBOL(scsi_release_buffers);
  * -EREMOTEIO	permanent target failure, do not retry
  * -EBADE	permanent nexus failure, retry on other path
  * -ENOSPC	No write space available
+ * -ENODATA	Medium error
  * -EIO		unspecified I/O error
  */
 static int __scsi_error_from_host_byte(struct scsi_cmnd *cmd, int result)

@@ -733,6 +734,10 @@ static int __scsi_error_from_host_byte(struct scsi_cmnd *cmd, int result)
 		set_host_byte(cmd, DID_OK);
 		error = -ENOSPC;
 		break;
+	case DID_MEDIUM_ERROR:
+		set_host_byte(cmd, DID_OK);
+		error = -ENODATA;
+		break;
 	default:
 		error = -EIO;
 		break;
include/scsi/scsi.h +1 -0
index 7d5b8e2..95ce204 100644
@@ -453,6 +453,7 @@ static inline int scsi_is_wlun(unsigned int lun)
 #define DID_NEXUS_FAILURE 0x11  /* Permanent nexus failure, retry on other
 				 * paths might yield different results */
 #define DID_ALLOC_FAILURE 0x12  /* Space allocation on the device failed */
+#define DID_MEDIUM_ERROR  0x13  /* Medium error */
 #define DRIVER_OK       0x00	/* Driver status                           */
 
 /*
Signature · 1 line
1.7.12.4
#5 lore raw

Re: [PATCH 4/4] scsi: Return ENODATA on medium error

Hi Hannes,

Quoted 12 lines · Hannes Reinecke

When a medium error is detected the SCSI stack should return ENODATA to the upper layers.

Cc Jun'ichi Nomura <[email protected]>
Signed-off-by Hannes Reinecke <[email protected]>
---
 block/blk-core.c          |  3 +++
 drivers/md/dm-mpath.c     | 16 +++++++++++++++-
 drivers/scsi/scsi_error.c |  2 +-
 drivers/scsi/scsi_lib.c   |  5 +++++
 include/scsi/scsi.h       |  1 +
 5 files changed, 25 insertions(+), 2 deletions(-)

..

drivers/md/dm-mpath.c

Quoted 4 lines
drivers/md/dm-mpath.c +0 -0
index bdf26f5..57896cea 100644

Thank you for including the change.
But I think this change to dm should be a separate patch in this series (prerequisite to the SCSI error code change) and Cc to dm-devel.

drivers/md/dm-mpath.c static void activate_path(struct work_struct *work) · line 1270

Quoted 11 lines
drivers/md/dm-mpath.c +7 -0
@@ -1261,6 +1261,20 @@ static void activate_path(struct work_struct *work)
 				pg_init_done, pgpath);
 }
 
+static int noretry_error(int error)
+{
+	switch(error) {
+	case -EOPNOTSUPP:
+	case -EREMOTEIO:
+	case -EILSEQ:
+	case -ENODATA:

case -ENOSPC:

We don't want to fail the path for -ENOSPC, do we?

Quoted 19 lines
drivers/md/dm-mpath.c +8 -1
+		return 1;
+	}
+
+	/* Anything else could be a path failure, so should be retried */
+	return 0;
+}
+
 /*
  * end_io handling
  */

@@ -1284,7 +1298,7 @@ static int do_end_io(struct multipath *m, struct request *clone,
 	if (!error && !clone->errors)
 		return 0;	/* I/O complete */
 
-	if (error == -EOPNOTSUPP || error == -EREMOTEIO || error == -EILSEQ)
+	if (noretry_error(error))
 		return error;
 
 	if (mpio->pgpath)
Signature · 1 line
Jun'ichi Nomura, NEC Corporation