[PATCH 4.9 094/144] libata: Fix command retry decision

[Greg Kroah-Hartman]

4.9-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Damien Le Moal <damien.lemoal@xxxxxxx>

[ Upstream commit 804689ad2d9b66d0d3920b48cf05881049d44589 ]

For failed commands with valid sense data (e.g. NCQ commands),
scsi_check_sense() is used in ata_analyze_tf() to determine if the
command can be retried. In such case, rely on this decision and ignore
the command error mask based decision done in ata_worth_retry().

This fixes useless retries of commands such as unaligned writes on zoned
disks (TYPE_ZAC).

Signed-off-by: Damien Le Moal <damien.lemoal@xxxxxxx>
Reviewed-by: Hannes Reinecke <hare@xxxxxxxx>
Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
Signed-off-by: Sasha Levin <alexander.levin@xxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
 drivers/ata/libata-eh.c |   12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

--- a/drivers/ata/libata-eh.c
+++ b/drivers/ata/libata-eh.c
@@ -2282,12 +2282,16 @@ static void ata_eh_link_autopsy(struct a
 		if (qc->err_mask & ~AC_ERR_OTHER)
 			qc->err_mask &= ~AC_ERR_OTHER;
 
-		/* SENSE_VALID trumps dev/unknown error and revalidation */
+		/*
+		 * SENSE_VALID trumps dev/unknown error and revalidation. Upper
+		 * layers will determine whether the command is worth retrying
+		 * based on the sense data and device class/type. Otherwise,
+		 * determine directly if the command is worth retrying using its
+		 * error mask and flags.
+		 */
 		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
 			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);
-
-		/* determine whether the command is worth retrying */
-		if (ata_eh_worth_retry(qc))
+		else if (ata_eh_worth_retry(qc))
 			qc->flags |= ATA_QCFLAG_RETRY;
 
 		/* accumulate error info */