Merge tag 'ata-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/libata

Pull ATA fixes from Damien Le Moal:
 "A larger than usual set of fixes for 6.6-rc4 due to the unexpected
  number of fixes needed to address ATA disks suspend/resume issues.

  In more detail:

   - Add missing additionalProperties on child nodes to the pata-common
     DT bindings (Rob)

   - Fix handling of the REPORT SUPPORTED OPERATION CODES command to
     ignore reserved bits (Niklas)

   - Increase port multiplier soft reset timeout to accomodate slow
     devices and avoid issues on wakeup (Matthias)

   - A couple of minor code fixes to avoid compilation warnings in
     libata-core and libata-eh (me)

   - Many patches from me to address suspend/resume issues, and in
     particular a potential deadlock on resume due to the SCSI disk
     driver resume operation not being synchronized with libata EH port
     resume handling.

     This is addressed by changing the scsi disk driver disk start/stop
     control to allow libata to execute disk suspend (spin down) and
     resume (spin up) on its own during system suspend/resume. Runtime
     suspend/resume control remains with the SCSI disk driver.

     Other fixes include:
      - Fix libata power management request issuing to avoid races
      - Establish a link between ATA ports and SCSI devices to order PM
        operations
      - Fix device removal to avoid issues with driver rmmod removal
      - Fix synchronization of libata device rescan and SCSI disk resume
        operation
      - Remove libsas PM operations as suspend/resume is handled
        directly by the sas controller resume
      - Fix the SCSI disk driver to not issue commands to suspended
        disks, thus avoiding potential system lock-up on resume"

* tag 'ata-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/libata:
  ata: libata-eh: Fix compilation warning in ata_eh_link_report()
  ata: libata-core: Fix compilation warning in ata_dev_config_ncq()
  scsi: sd: Do not issue commands to suspended disks on shutdown
  ata: libata-core: Do not register PM operations for SAS ports
  ata: libata-scsi: Fix delayed scsi_rescan_device() execution
  scsi: Do not attempt to rescan suspended devices
  ata: libata-scsi: Disable scsi device manage_system_start_stop
  scsi: sd: Differentiate system and runtime start/stop management
  ata: libata-scsi: link ata port and scsi device
  ata: libata-core: Fix port and device removal
  ata: libata-core: Fix ata_port_request_pm() locking
  ata: libata-sata: increase PMP SRST timeout to 10s
  ata: libata-scsi: ignore reserved bits for REPORT SUPPORTED OPERATION CODES
  dt-bindings: ata: pata-common: Add missing additionalProperties on child nodes

3 files changed, 100 insertions, 20 deletions

diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
index eaa972bee6c0..902655d75947 100644
--- a/drivers/scsi/scsi_scan.c
+++ b/drivers/scsi/scsi_scan.c
@@ -1619,12 +1619,24 @@ int scsi_add_device(struct Scsi_Host *host, uint channel,
 }
 EXPORT_SYMBOL(scsi_add_device);
 
-void scsi_rescan_device(struct scsi_device *sdev)
+int scsi_rescan_device(struct scsi_device *sdev)
 {
 	struct device *dev = &sdev->sdev_gendev;
+	int ret = 0;
 
 	device_lock(dev);
 
+	/*
+	 * Bail out if the device is not running. Otherwise, the rescan may
+	 * block waiting for commands to be executed, with us holding the
+	 * device lock. This can result in a potential deadlock in the power
+	 * management core code when system resume is on-going.
+	 */
+	if (sdev->sdev_state != SDEV_RUNNING) {
+		ret = -EWOULDBLOCK;
+		goto unlock;
+	}
+
 	scsi_attach_vpd(sdev);
 	scsi_cdl_check(sdev);
 
@@ -1638,7 +1650,11 @@ void scsi_rescan_device(struct scsi_device *sdev)
 			drv->rescan(dev);
 		module_put(dev->driver->owner);
 	}
+
+unlock:
 	device_unlock(dev);
+
+	return ret;
 }
 EXPORT_SYMBOL(scsi_rescan_device);
 
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index c92a317ba547..83b6a3f3863b 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -201,18 +201,32 @@ cache_type_store(struct device *dev, struct device_attribute *attr,
 }
 
 static ssize_t
-manage_start_stop_show(struct device *dev, struct device_attribute *attr,
-		       char *buf)
+manage_start_stop_show(struct device *dev,
+		       struct device_attribute *attr, char *buf)
 {
 	struct scsi_disk *sdkp = to_scsi_disk(dev);
 	struct scsi_device *sdp = sdkp->device;
 
-	return sprintf(buf, "%u\n", sdp->manage_start_stop);
+	return sysfs_emit(buf, "%u\n",
+			  sdp->manage_system_start_stop &&
+			  sdp->manage_runtime_start_stop);
 }
+static DEVICE_ATTR_RO(manage_start_stop);
 
 static ssize_t
-manage_start_stop_store(struct device *dev, struct device_attribute *attr,
-			const char *buf, size_t count)
+manage_system_start_stop_show(struct device *dev,
+			      struct device_attribute *attr, char *buf)
+{
+	struct scsi_disk *sdkp = to_scsi_disk(dev);
+	struct scsi_device *sdp = sdkp->device;
+
+	return sysfs_emit(buf, "%u\n", sdp->manage_system_start_stop);
+}
+
+static ssize_t
+manage_system_start_stop_store(struct device *dev,
+			       struct device_attribute *attr,
+			       const char *buf, size_t count)
 {
 	struct scsi_disk *sdkp = to_scsi_disk(dev);
 	struct scsi_device *sdp = sdkp->device;
@@ -224,11 +238,42 @@ manage_start_stop_store(struct device *dev, struct device_attribute *attr,
 	if (kstrtobool(buf, &v))
 		return -EINVAL;
 
-	sdp->manage_start_stop = v;
+	sdp->manage_system_start_stop = v;
 
 	return count;
 }
-static DEVICE_ATTR_RW(manage_start_stop);
+static DEVICE_ATTR_RW(manage_system_start_stop);
+
+static ssize_t
+manage_runtime_start_stop_show(struct device *dev,
+			       struct device_attribute *attr, char *buf)
+{
+	struct scsi_disk *sdkp = to_scsi_disk(dev);
+	struct scsi_device *sdp = sdkp->device;
+
+	return sysfs_emit(buf, "%u\n", sdp->manage_runtime_start_stop);
+}
+
+static ssize_t
+manage_runtime_start_stop_store(struct device *dev,
+				struct device_attribute *attr,
+				const char *buf, size_t count)
+{
+	struct scsi_disk *sdkp = to_scsi_disk(dev);
+	struct scsi_device *sdp = sdkp->device;
+	bool v;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EACCES;
+
+	if (kstrtobool(buf, &v))
+		return -EINVAL;
+
+	sdp->manage_runtime_start_stop = v;
+
+	return count;
+}
+static DEVICE_ATTR_RW(manage_runtime_start_stop);
 
 static ssize_t
 allow_restart_show(struct device *dev, struct device_attribute *attr, char *buf)
@@ -560,6 +605,8 @@ static struct attribute *sd_disk_attrs[] = {
 	&dev_attr_FUA.attr,
 	&dev_attr_allow_restart.attr,
 	&dev_attr_manage_start_stop.attr,
+	&dev_attr_manage_system_start_stop.attr,
+	&dev_attr_manage_runtime_start_stop.attr,
 	&dev_attr_protection_type.attr,
 	&dev_attr_protection_mode.attr,
 	&dev_attr_app_tag_own.attr,
@@ -3694,7 +3741,8 @@ static int sd_remove(struct device *dev)
 
 	device_del(&sdkp->disk_dev);
 	del_gendisk(sdkp->disk);
-	sd_shutdown(dev);
+	if (!sdkp->suspended)
+		sd_shutdown(dev);
 
 	put_disk(sdkp->disk);
 	return 0;
@@ -3771,13 +3819,20 @@ static void sd_shutdown(struct device *dev)
 		sd_sync_cache(sdkp, NULL);
 	}
 
-	if (system_state != SYSTEM_RESTART && sdkp->device->manage_start_stop) {
+	if (system_state != SYSTEM_RESTART &&
+	    sdkp->device->manage_system_start_stop) {
 		sd_printk(KERN_NOTICE, sdkp, "Stopping disk\n");
 		sd_start_stop_device(sdkp, 0);
 	}
 }
 
-static int sd_suspend_common(struct device *dev, bool ignore_stop_errors)
+static inline bool sd_do_start_stop(struct scsi_device *sdev, bool runtime)
+{
+	return (sdev->manage_system_start_stop && !runtime) ||
+		(sdev->manage_runtime_start_stop && runtime);
+}
+
+static int sd_suspend_common(struct device *dev, bool runtime)
 {
 	struct scsi_disk *sdkp = dev_get_drvdata(dev);
 	struct scsi_sense_hdr sshdr;
@@ -3809,15 +3864,18 @@ static int sd_suspend_common(struct device *dev, bool ignore_stop_errors)
 		}
 	}
 
-	if (sdkp->device->manage_start_stop) {
+	if (sd_do_start_stop(sdkp->device, runtime)) {
 		if (!sdkp->device->silence_suspend)
 			sd_printk(KERN_NOTICE, sdkp, "Stopping disk\n");
 		/* an error is not worth aborting a system sleep */
 		ret = sd_start_stop_device(sdkp, 0);
-		if (ignore_stop_errors)
+		if (!runtime)
 			ret = 0;
 	}
 
+	if (!ret)
+		sdkp->suspended = true;
+
 	return ret;
 }
 
@@ -3826,15 +3884,15 @@ static int sd_suspend_system(struct device *dev)
 	if (pm_runtime_suspended(dev))
 		return 0;
 
-	return sd_suspend_common(dev, true);
+	return sd_suspend_common(dev, false);
 }
 
 static int sd_suspend_runtime(struct device *dev)
 {
-	return sd_suspend_common(dev, false);
+	return sd_suspend_common(dev, true);
 }
 
-static int sd_resume(struct device *dev)
+static int sd_resume(struct device *dev, bool runtime)
 {
 	struct scsi_disk *sdkp = dev_get_drvdata(dev);
 	int ret = 0;
@@ -3842,16 +3900,21 @@ static int sd_resume(struct device *dev)
 	if (!sdkp)	/* E.g.: runtime resume at the start of sd_probe() */
 		return 0;
 
-	if (!sdkp->device->manage_start_stop)
+	if (!sd_do_start_stop(sdkp->device, runtime)) {
+		sdkp->suspended = false;
 		return 0;
+	}
 
 	if (!sdkp->device->no_start_on_resume) {
 		sd_printk(KERN_NOTICE, sdkp, "Starting disk\n");
 		ret = sd_start_stop_device(sdkp, 1);
 	}
 
-	if (!ret)
+	if (!ret) {
 		opal_unlock_from_suspend(sdkp->opal_dev);
+		sdkp->suspended = false;
+	}
+
 	return ret;
 }
 
@@ -3860,7 +3923,7 @@ static int sd_resume_system(struct device *dev)
 	if (pm_runtime_suspended(dev))
 		return 0;
 
-	return sd_resume(dev);
+	return sd_resume(dev, false);
 }
 
 static int sd_resume_runtime(struct device *dev)
@@ -3887,7 +3950,7 @@ static int sd_resume_runtime(struct device *dev)
 				  "Failed to clear sense data\n");
 	}
 
-	return sd_resume(dev);
+	return sd_resume(dev, true);
 }
 
 static const struct dev_pm_ops sd_pm_ops = {
diff --git a/drivers/scsi/sd.h b/drivers/scsi/sd.h
index 5eea762f84d1..409dda5350d1 100644
--- a/drivers/scsi/sd.h
+++ b/drivers/scsi/sd.h
@@ -131,6 +131,7 @@ struct scsi_disk {
 	u8		provisioning_mode;
 	u8		zeroing_mode;
 	u8		nr_actuators;		/* Number of actuators */
+	bool		suspended;	/* Disk is suspended (stopped) */
 	unsigned	ATO : 1;	/* state of disk ATO bit */
 	unsigned	cache_override : 1; /* temp override of WCE,RCD */
 	unsigned	WCE : 1;	/* state of disk WCE bit */