[PATCH AUTOSEL 4.9 45/66] scsi: libsas: always unregister the old device if going to discover new

[PATCH AUTOSEL 4.9 45/66] scsi: libsas: always unregister the old device if going to discover new

[Sasha Levin]

From: Jason Yan <yanaijie@huawei.com>

[ Upstream commit 32c850bf587f993b2620b91e5af8a64a7813f504 ]

If we went into sas_rediscover_dev() the attached_sas_addr was already insured
not to be zero. So it's unnecessary to check if the attached_sas_addr is zero.

And although if the sas address is not changed, we always have to unregister
the old device when we are going to register a new one. We cannot just leave
the device there and bring up the new.

Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: chenxiang <chenxiang66@hisilicon.com>
CC: John Garry <john.garry@huawei.com>
CC: Johannes Thumshirn <jthumshirn@suse.de>
CC: Ewan Milne <emilne@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Tomas Henzl <thenzl@redhat.com>
CC: Dan Williams <dan.j.williams@intel.com>
CC: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/libsas/sas_expander.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

diff --git a/drivers/scsi/libsas/sas_expander.c b/drivers/scsi/libsas/sas_expander.c
index 400eee9d77832..d44f18f773c0f 100644
--- a/drivers/scsi/libsas/sas_expander.c
+++ b/drivers/scsi/libsas/sas_expander.c
@@ -2049,14 +2049,11 @@ static int sas_rediscover_dev(struct domain_device *dev, int phy_id, bool last)
 		return res;
 	}
 
-	/* delete the old link */
-	if (SAS_ADDR(phy->attached_sas_addr) &&
-	    SAS_ADDR(sas_addr) != SAS_ADDR(phy->attached_sas_addr)) {
-		SAS_DPRINTK("ex %016llx phy 0x%x replace %016llx\n",
-			    SAS_ADDR(dev->sas_addr), phy_id,
-			    SAS_ADDR(phy->attached_sas_addr));
-		sas_unregister_devs_sas_addr(dev, phy_id, last);
-	}
+	/* we always have to delete the old device when we went here */
+	SAS_DPRINTK("ex %016llx phy 0x%x replace %016llx\n",
+		    SAS_ADDR(dev->sas_addr), phy_id,
+		    SAS_ADDR(phy->attached_sas_addr));
+	sas_unregister_devs_sas_addr(dev, phy_id, last);
 
 	return sas_discover_new(dev, phy_id);
 }
-- 
2.20.1

[PATCH AUTOSEL 4.9 55/66] scsi: NCR5380: Clear all unissued commands on host reset

[Sasha Levin]

From: Hannes Reinecke <hare@suse.com>

[ Upstream commit 1aeeeed7f03c576f096eede7b0384f99a98f588c ]

When doing a host reset we should be clearing all outstanding commands, not
just the command triggering the reset.

[mkp: adjusted Hannes' SoB address]

Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Ondrey Zary <linux@rainbow-software.org>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/NCR5380.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index 3cfab8868c989..ba7b3aef3ef0b 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -2392,7 +2392,7 @@ static int NCR5380_bus_reset(struct scsi_cmnd *cmd)
 	spin_lock_irqsave(&hostdata->lock, flags);
 
 #if (NDEBUG & NDEBUG_ANY)
-	scmd_printk(KERN_INFO, cmd, __func__);
+	shost_printk(KERN_INFO, instance, __func__);
 #endif
 	NCR5380_dprint(NDEBUG_ANY, instance);
 	NCR5380_dprint_phase(NDEBUG_ANY, instance);
@@ -2410,10 +2410,13 @@ static int NCR5380_bus_reset(struct scsi_cmnd *cmd)
 	 * commands!
 	 */
 
-	if (list_del_cmd(&hostdata->unissued, cmd)) {
+	list_for_each_entry(ncmd, &hostdata->unissued, list) {
+		struct scsi_cmnd *cmd = NCR5380_to_scmd(ncmd);
+
 		cmd->result = DID_RESET << 16;
 		cmd->scsi_done(cmd);
 	}
+	INIT_LIST_HEAD(&hostdata->unissued);
 
 	if (hostdata->selecting) {
 		hostdata->selecting->result = DID_RESET << 16;
-- 
2.20.1

[PATCH AUTOSEL 4.9 56/66] scsi: NCR5380: Use DRIVER_SENSE to indicate valid sense data

[Sasha Levin]

From: Finn Thain <fthain@telegraphics.com.au>

[ Upstream commit 070356513963be6196142acff56acc8359069fa1 ]

When sense data is valid, call set_driver_byte(cmd, DRIVER_SENSE).  Otherwise
some callers of scsi_execute() will ignore sense data.  Don't set DID_ERROR or
DID_RESET just because sense data is missing.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/NCR5380.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index ba7b3aef3ef0b..35419ba32c467 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -617,11 +617,12 @@ static void complete_cmd(struct Scsi_Host *instance,
 
 	if (hostdata->sensing == cmd) {
 		/* Autosense processing ends here */
-		if ((cmd->result & 0xff) != SAM_STAT_GOOD) {
+		if (status_byte(cmd->result) != GOOD) {
 			scsi_eh_restore_cmnd(cmd, &hostdata->ses);
-			set_host_byte(cmd, DID_ERROR);
-		} else
+		} else {
 			scsi_eh_restore_cmnd(cmd, &hostdata->ses);
+			set_driver_byte(cmd, DRIVER_SENSE);
+		}
 		hostdata->sensing = NULL;
 	}
 
@@ -2356,7 +2357,6 @@ static int NCR5380_abort(struct scsi_cmnd *cmd)
 	if (list_del_cmd(&hostdata->autosense, cmd)) {
 		dsprintk(NDEBUG_ABORT, instance,
 		         "abort: removed %p from sense queue\n", cmd);
-		set_host_byte(cmd, DID_ERROR);
 		complete_cmd(instance, cmd);
 	}
 
@@ -2435,7 +2435,6 @@ static int NCR5380_bus_reset(struct scsi_cmnd *cmd)
 	list_for_each_entry(ncmd, &hostdata->autosense, list) {
 		struct scsi_cmnd *cmd = NCR5380_to_scmd(ncmd);
 
-		set_host_byte(cmd, DID_RESET);
 		cmd->scsi_done(cmd);
 	}
 	INIT_LIST_HEAD(&hostdata->autosense);
-- 
2.20.1

[PATCH AUTOSEL 4.9 57/66] scsi: NCR5380: Check for invalid reselection target

[Sasha Levin]

From: Finn Thain <fthain@telegraphics.com.au>

[ Upstream commit 7ef55f6744c45e3d7c85a3f74ada39b67ac741dd ]

The X3T9.2 specification (draft) says, under "6.1.4.1 RESELECTION", that "the
initiator shall not respond to a RESELECTION phase if other than two SCSI ID
bits are on the DATA BUS." This issue (too many bits set) has been observed in
the wild, so add a check.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/NCR5380.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index 35419ba32c467..15c0fff7519be 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -2102,6 +2102,11 @@ static void NCR5380_reselect(struct Scsi_Host *instance)
 	NCR5380_write(MODE_REG, MR_BASE);
 
 	target_mask = NCR5380_read(CURRENT_SCSI_DATA_REG) & ~(hostdata->id_mask);
+	if (!target_mask || target_mask & (target_mask - 1)) {
+		shost_printk(KERN_WARNING, instance,
+			     "reselect: bad target_mask 0x%02x\n", target_mask);
+		return;
+	}
 
 	dsprintk(NDEBUG_RESELECTION, instance, "reselect\n");
 
-- 
2.20.1

[PATCH AUTOSEL 4.9 58/66] scsi: NCR5380: Don't clear busy flag when abort fails

[Sasha Levin]

From: Finn Thain <fthain@telegraphics.com.au>

[ Upstream commit 45ddc1b24806cc8f1a09f23dd4e7b6e4a8ae36e1 ]

When NCR5380_abort() returns FAILED, the driver forgets that the target is
still busy. Hence, further commands may be sent to the target, which may fail
during selection and produce the error message, "reselection after won
arbitration?". Prevent this by leaving the busy flag set when NCR5380_abort()
fails.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/NCR5380.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index 15c0fff7519be..a4856eeac26f6 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -626,8 +626,6 @@ static void complete_cmd(struct Scsi_Host *instance,
 		hostdata->sensing = NULL;
 	}
 
-	hostdata->busy[scmd_id(cmd)] &= ~(1 << cmd->device->lun);
-
 	cmd->scsi_done(cmd);
 }
 
@@ -1799,6 +1797,7 @@ static void NCR5380_information_transfer(struct Scsi_Host *instance)
 				cmd->result = DID_ERROR << 16;
 				complete_cmd(instance, cmd);
 				hostdata->connected = NULL;
+				hostdata->busy[scmd_id(cmd)] &= ~(1 << cmd->device->lun);
 				return;
 #endif
 			case PHASE_DATAIN:
@@ -1881,6 +1880,7 @@ static void NCR5380_information_transfer(struct Scsi_Host *instance)
 					         cmd, scmd_id(cmd), cmd->device->lun);
 
 					hostdata->connected = NULL;
+					hostdata->busy[scmd_id(cmd)] &= ~(1 << cmd->device->lun);
 
 					cmd->result &= ~0xffff;
 					cmd->result |= cmd->SCp.Status;
@@ -2040,6 +2040,7 @@ static void NCR5380_information_transfer(struct Scsi_Host *instance)
 				NCR5380_transfer_pio(instance, &phase, &len, &data);
 				if (msgout == ABORT) {
 					hostdata->connected = NULL;
+					hostdata->busy[scmd_id(cmd)] &= ~(1 << cmd->device->lun);
 					cmd->result = DID_ERROR << 16;
 					complete_cmd(instance, cmd);
 					maybe_release_dma_irq(instance);
@@ -2194,13 +2195,16 @@ static void NCR5380_reselect(struct Scsi_Host *instance)
 		dsprintk(NDEBUG_RESELECTION | NDEBUG_QUEUES, instance,
 		         "reselect: removed %p from disconnected queue\n", tmp);
 	} else {
+		int target = ffs(target_mask) - 1;
+
 		shost_printk(KERN_ERR, instance, "target bitmask 0x%02x lun %d not in disconnected queue.\n",
 		             target_mask, lun);
 		/*
 		 * Since we have an established nexus that we can't do anything
 		 * with, we must abort it.
 		 */
-		do_abort(instance);
+		if (do_abort(instance) == 0)
+			hostdata->busy[target] &= ~(1 << lun);
 		return;
 	}
 
@@ -2368,8 +2372,10 @@ static int NCR5380_abort(struct scsi_cmnd *cmd)
 out:
 	if (result == FAILED)
 		dsprintk(NDEBUG_ABORT, instance, "abort: failed to abort %p\n", cmd);
-	else
+	else {
+		hostdata->busy[scmd_id(cmd)] &= ~(1 << cmd->device->lun);
 		dsprintk(NDEBUG_ABORT, instance, "abort: successfully aborted %p\n", cmd);
+	}
 
 	queue_work(hostdata->work_q, &hostdata->main_task);
 	maybe_release_dma_irq(instance);
-- 
2.20.1

[PATCH AUTOSEL 4.9 59/66] scsi: NCR5380: Don't call dsprintk() following reselection interrupt

[Sasha Levin]

From: Finn Thain <fthain@telegraphics.com.au>

[ Upstream commit 08267216b3f8aa5adc204bdccf8deb72c1cd7665 ]

The X3T9.2 specification (draft) says, under "6.1.4.1 RESELECTION",

    ... The reselected initiator shall then assert the BSY signal
    within a selection abort time of its most recent detection of being
    reselected; this is required for correct operation of the time-out
    procedure.

The selection abort time is only 200 us which may be insufficient time for a
printk() call. Move the diagnostics to the error paths.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/NCR5380.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index a4856eeac26f6..03f9ddbc37fbb 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -2109,8 +2109,6 @@ static void NCR5380_reselect(struct Scsi_Host *instance)
 		return;
 	}
 
-	dsprintk(NDEBUG_RESELECTION, instance, "reselect\n");
-
 	/*
 	 * At this point, we have detected that our SCSI ID is on the bus,
 	 * SEL is true and BSY was false for at least one bus settle delay
@@ -2123,6 +2121,7 @@ static void NCR5380_reselect(struct Scsi_Host *instance)
 	NCR5380_write(INITIATOR_COMMAND_REG, ICR_BASE | ICR_ASSERT_BSY);
 	if (NCR5380_poll_politely(instance,
 	                          STATUS_REG, SR_SEL, 0, 2 * HZ) < 0) {
+		shost_printk(KERN_ERR, instance, "reselect: !SEL timeout\n");
 		NCR5380_write(INITIATOR_COMMAND_REG, ICR_BASE);
 		return;
 	}
@@ -2134,6 +2133,7 @@ static void NCR5380_reselect(struct Scsi_Host *instance)
 
 	if (NCR5380_poll_politely(instance,
 	                          STATUS_REG, SR_REQ, SR_REQ, 2 * HZ) < 0) {
+		shost_printk(KERN_ERR, instance, "reselect: REQ timeout\n");
 		do_abort(instance);
 		return;
 	}
-- 
2.20.1

[PATCH AUTOSEL 4.9 60/66] scsi: NCR5380: Handle BUS FREE during reselection

[Sasha Levin]

From: Finn Thain <fthain@telegraphics.com.au>

[ Upstream commit ca694afad707cb3ae2fdef3b28454444d9ac726e ]

The X3T9.2 specification (draft) says, under "6.1.4.2 RESELECTION time-out
procedure", that a target may assert RST or go to BUS FREE phase if the
initiator does not respond within 200 us. Something like this has been
observed with AztecMonster II target. When it happens, all we can do is wait
for the target to try again.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/scsi/NCR5380.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index 03f9ddbc37fbb..27270631c70c2 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -2133,6 +2133,9 @@ static void NCR5380_reselect(struct Scsi_Host *instance)
 
 	if (NCR5380_poll_politely(instance,
 	                          STATUS_REG, SR_REQ, SR_REQ, 2 * HZ) < 0) {
+		if ((NCR5380_read(STATUS_REG) & (SR_BSY | SR_SEL)) == 0)
+			/* BUS FREE phase */
+			return;
 		shost_printk(KERN_ERR, instance, "reselect: REQ timeout\n");
 		do_abort(instance);
 		return;
-- 
2.20.1