[PATCH V4 0/2] driver core: Add ability to delete device links of unregistered devices

[PATCH V4 0/2] driver core: Add ability to delete device links of unregistered devices

[Adrian Hunter]

Hi

There is an issue with the SCSI UFS driver when the optional
BOOT well-known LUN fails to probe, which is not a fatal error.
The issue is that the device and its "managed" device link do not
then get deleted.  The device because the device link has a
reference to it.  The device link because it can only be deleted
by device_del(), but device_add() was never called, so device_del()
never will be either.

Since V2, these patches fix the issue by amending device link removal to
accept removal of a link with an unregistered consumer device, as suggested
by Rafael.


Changes in V4:
    driver core: Prevent warning when removing a device link from unregistered consumer
	Add stable tag and Rafael's Reviewed-by

    driver core: Add ability to delete device links of unregistered devices
	Amend comment "discover an error" -> "discovering an error"
	Merge with next patch

    scsi: ufshcd: Fix device links when BOOT WLUN fails to probe
	Merge with previous patch
	Add Rafael's Reviewed-by


Changes in V3:

    driver core: Prevent warning when removing a device link from unregistered consumer
	New patch split from "driver core: Add ability to delete device
	links of unregistered devices" except first chunk from that patch
	dropped as unnecessary

    driver core: Add ability to delete device links of unregistered devices
	Move warning fix to separate patch.


Changes in V2:

    Take approach suggested by Rafael


Adrian Hunter (2):
      driver core: Prevent warning when removing a device link from unregistered consumer
      scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

 drivers/base/core.c       |  8 ++++++--
 drivers/scsi/ufs/ufshcd.c | 23 +++++++++++++++++++++--
 2 files changed, 27 insertions(+), 4 deletions(-)


Regards
Adrian

[PATCH V4 1/2] driver core: Prevent warning when removing a device link from unregistered consumer

[Adrian Hunter]

sysfs_remove_link() causes a warning if the parent directory does not
exist. That can happen if the device link consumer has not been registered.
So do not attempt sysfs_remove_link() in that case.

Fixes: 287905e68dd29 ("driver core: Expose device link details in sysfs")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: stable@vger.kernel.org # 5.9+
Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
---
 drivers/base/core.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/base/core.c b/drivers/base/core.c
index ea5b85354526..2de8f7d8cf54 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -575,8 +575,10 @@ static void devlink_remove_symlinks(struct device *dev,
 		return;
 	}
 
-	snprintf(buf, len, "supplier:%s:%s", dev_bus_name(sup), dev_name(sup));
-	sysfs_remove_link(&con->kobj, buf);
+	if (device_is_registered(con)) {
+		snprintf(buf, len, "supplier:%s:%s", dev_bus_name(sup), dev_name(sup));
+		sysfs_remove_link(&con->kobj, buf);
+	}
 	snprintf(buf, len, "consumer:%s:%s", dev_bus_name(con), dev_name(con));
 	sysfs_remove_link(&sup->kobj, buf);
 	kfree(buf);
-- 
2.17.1

[PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Adrian Hunter]

Managed device links are deleted by device_del(). However it is possible to
add a device link to a consumer before device_add(), and then discovering
an error prevents the device from being used. In that case normally
references to the device would be dropped and the device would be deleted.
However the device link holds a reference to the device, so the device link
and device remain indefinitely (unless the supplier is deleted).

For UFSHCD, if a LUN fails to probe (e.g. absent BOOT WLUN), the device
will not have been registered but can still have a device link holding a
reference to the device. The unwanted device link will prevent runtime
suspend indefinitely.

Amend device link removal to accept removal of a link with an unregistered
consumer device (suggested by Rafael), and fix UFSHCD by explicitly
deleting the device link when SCSI destroys the SCSI device.

Fixes: b294ff3e34490 ("scsi: ufs: core: Enable power management for wlun")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
---
 drivers/base/core.c       |  2 ++
 drivers/scsi/ufs/ufshcd.c | 23 +++++++++++++++++++++--
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/drivers/base/core.c b/drivers/base/core.c
index 2de8f7d8cf54..983e895d4ced 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -887,6 +887,8 @@ static void device_link_put_kref(struct device_link *link)
 {
 	if (link->flags & DL_FLAG_STATELESS)
 		kref_put(&link->kref, __device_link_del);
+	else if (!device_is_registered(link->consumer))
+		__device_link_del(&link->kref);
 	else
 		WARN(1, "Unable to drop a managed device link reference\n");
 }
diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index 708b3b62fc4d..9864a8ee0263 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -5020,15 +5020,34 @@ static int ufshcd_slave_configure(struct scsi_device *sdev)
 static void ufshcd_slave_destroy(struct scsi_device *sdev)
 {
 	struct ufs_hba *hba;
+	unsigned long flags;
 
 	hba = shost_priv(sdev->host);
 	/* Drop the reference as it won't be needed anymore */
 	if (ufshcd_scsi_to_upiu_lun(sdev->lun) == UFS_UPIU_UFS_DEVICE_WLUN) {
-		unsigned long flags;
-
 		spin_lock_irqsave(hba->host->host_lock, flags);
 		hba->sdev_ufs_device = NULL;
 		spin_unlock_irqrestore(hba->host->host_lock, flags);
+	} else if (hba->sdev_ufs_device) {
+		struct device *supplier = NULL;
+
+		/* Ensure UFS Device WLUN exists and does not disappear */
+		spin_lock_irqsave(hba->host->host_lock, flags);
+		if (hba->sdev_ufs_device) {
+			supplier = &hba->sdev_ufs_device->sdev_gendev;
+			get_device(supplier);
+		}
+		spin_unlock_irqrestore(hba->host->host_lock, flags);
+
+		if (supplier) {
+			/*
+			 * If a LUN fails to probe (e.g. absent BOOT WLUN), the
+			 * device will not have been registered but can still
+			 * have a device link holding a reference to the device.
+			 */
+			device_link_remove(&sdev->sdev_gendev, supplier);
+			put_device(supplier);
+		}
 	}
 }
 
-- 
2.17.1

RE: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Avri Altman]

> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index 708b3b62fc4d..9864a8ee0263 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -5020,15 +5020,34 @@ static int ufshcd_slave_configure(struct scsi_device
> *sdev)
>  static void ufshcd_slave_destroy(struct scsi_device *sdev)
>  {
>         struct ufs_hba *hba;
> +       unsigned long flags;
> 
>         hba = shost_priv(sdev->host);
>         /* Drop the reference as it won't be needed anymore */
>         if (ufshcd_scsi_to_upiu_lun(sdev->lun) == UFS_UPIU_UFS_DEVICE_WLUN) {
> -               unsigned long flags;
> -
>                 spin_lock_irqsave(hba->host->host_lock, flags);
>                 hba->sdev_ufs_device = NULL;
>                 spin_unlock_irqrestore(hba->host->host_lock, flags);
> +       } else if (hba->sdev_ufs_device) {
> +               struct device *supplier = NULL;
> +
> +               /* Ensure UFS Device WLUN exists and does not disappear */
> +               spin_lock_irqsave(hba->host->host_lock, flags);
> +               if (hba->sdev_ufs_device) {
Was just checked in the outer clause?

Thanks,
Avri

> +                       supplier = &hba->sdev_ufs_device->sdev_gendev;
> +                       get_device(supplier);
> +               }
> +               spin_unlock_irqrestore(hba->host->host_lock, flags);
> +
> +               if (supplier) {
> +                       /*
> +                        * If a LUN fails to probe (e.g. absent BOOT WLUN), the
> +                        * device will not have been registered but can still
> +                        * have a device link holding a reference to the device.
> +                        */
> +                       device_link_remove(&sdev->sdev_gendev, supplier);
> +                       put_device(supplier);
> +               }
>         }
>  }
> 
> --
> 2.17.1

Re: [PATCH V4 1/2] driver core: Prevent warning when removing a device link from unregistered consumer

[Greg Kroah-Hartman]

On Fri, Jul 16, 2021 at 02:44:07PM +0300, Adrian Hunter wrote:
> sysfs_remove_link() causes a warning if the parent directory does not
> exist. That can happen if the device link consumer has not been registered.
> So do not attempt sysfs_remove_link() in that case.
> 
> Fixes: 287905e68dd29 ("driver core: Expose device link details in sysfs")
> Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
> Cc: stable@vger.kernel.org # 5.9+
> Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
> ---
>  drivers/base/core.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/base/core.c b/drivers/base/core.c
> index ea5b85354526..2de8f7d8cf54 100644
> --- a/drivers/base/core.c
> +++ b/drivers/base/core.c
> @@ -575,8 +575,10 @@ static void devlink_remove_symlinks(struct device *dev,
>  		return;
>  	}
>  
> -	snprintf(buf, len, "supplier:%s:%s", dev_bus_name(sup), dev_name(sup));
> -	sysfs_remove_link(&con->kobj, buf);
> +	if (device_is_registered(con)) {
> +		snprintf(buf, len, "supplier:%s:%s", dev_bus_name(sup), dev_name(sup));
> +		sysfs_remove_link(&con->kobj, buf);
> +	}
>  	snprintf(buf, len, "consumer:%s:%s", dev_bus_name(con), dev_name(con));
>  	sysfs_remove_link(&sup->kobj, buf);
>  	kfree(buf);
> -- 
> 2.17.1
> 

I've applied this patch to my tree now.

thanks,

greg k-h

Re: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Adrian Hunter]

On 17/07/21 9:02 pm, Avri Altman wrote:
>> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
>> index 708b3b62fc4d..9864a8ee0263 100644
>> --- a/drivers/scsi/ufs/ufshcd.c
>> +++ b/drivers/scsi/ufs/ufshcd.c
>> @@ -5020,15 +5020,34 @@ static int ufshcd_slave_configure(struct scsi_device
>> *sdev)
>>  static void ufshcd_slave_destroy(struct scsi_device *sdev)
>>  {
>>         struct ufs_hba *hba;
>> +       unsigned long flags;
>>
>>         hba = shost_priv(sdev->host);
>>         /* Drop the reference as it won't be needed anymore */
>>         if (ufshcd_scsi_to_upiu_lun(sdev->lun) == UFS_UPIU_UFS_DEVICE_WLUN) {
>> -               unsigned long flags;
>> -
>>                 spin_lock_irqsave(hba->host->host_lock, flags);
>>                 hba->sdev_ufs_device = NULL;
>>                 spin_unlock_irqrestore(hba->host->host_lock, flags);
>> +       } else if (hba->sdev_ufs_device) {
>> +               struct device *supplier = NULL;
>> +
>> +               /* Ensure UFS Device WLUN exists and does not disappear */
>> +               spin_lock_irqsave(hba->host->host_lock, flags);
>> +               if (hba->sdev_ufs_device) {
> Was just checked in the outer clause?

Yes, but need to re-check with the spinlock locked.

> 
> Thanks,
> Avri
> 
>> +                       supplier = &hba->sdev_ufs_device->sdev_gendev;
>> +                       get_device(supplier);
>> +               }
>> +               spin_unlock_irqrestore(hba->host->host_lock, flags);
>> +
>> +               if (supplier) {
>> +                       /*
>> +                        * If a LUN fails to probe (e.g. absent BOOT WLUN), the
>> +                        * device will not have been registered but can still
>> +                        * have a device link holding a reference to the device.
>> +                        */
>> +                       device_link_remove(&sdev->sdev_gendev, supplier);
>> +                       put_device(supplier);
>> +               }
>>         }
>>  }
>>
>> --
>> 2.17.1
> 

RE: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Avri Altman]

 
> On 17/07/21 9:02 pm, Avri Altman wrote:
> >> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> >> index 708b3b62fc4d..9864a8ee0263 100644
> >> --- a/drivers/scsi/ufs/ufshcd.c
> >> +++ b/drivers/scsi/ufs/ufshcd.c
> >> @@ -5020,15 +5020,34 @@ static int ufshcd_slave_configure(struct
> >> scsi_device
> >> *sdev)
> >>  static void ufshcd_slave_destroy(struct scsi_device *sdev)  {
> >>         struct ufs_hba *hba;
> >> +       unsigned long flags;
> >>
> >>         hba = shost_priv(sdev->host);
> >>         /* Drop the reference as it won't be needed anymore */
> >>         if (ufshcd_scsi_to_upiu_lun(sdev->lun) ==
> UFS_UPIU_UFS_DEVICE_WLUN) {
> >> -               unsigned long flags;
> >> -
> >>                 spin_lock_irqsave(hba->host->host_lock, flags);
> >>                 hba->sdev_ufs_device = NULL;
> >>                 spin_unlock_irqrestore(hba->host->host_lock, flags);
> >> +       } else if (hba->sdev_ufs_device) {
> >> +               struct device *supplier = NULL;
> >> +
> >> +               /* Ensure UFS Device WLUN exists and does not disappear */
> >> +               spin_lock_irqsave(hba->host->host_lock, flags);
> >> +               if (hba->sdev_ufs_device) {
> > Was just checked in the outer clause?
> 
> Yes, but need to re-check with the spinlock locked.
OK.
Looks good to me.
Thanks,
Avri

> 
> >
> > Thanks,
> > Avri
> >
> >> +                       supplier = &hba->sdev_ufs_device->sdev_gendev;
> >> +                       get_device(supplier);
> >> +               }
> >> +               spin_unlock_irqrestore(hba->host->host_lock, flags);
> >> +
> >> +               if (supplier) {
> >> +                       /*
> >> +                        * If a LUN fails to probe (e.g. absent BOOT WLUN), the
> >> +                        * device will not have been registered but can still
> >> +                        * have a device link holding a reference to the device.
> >> +                        */
> >> +                       device_link_remove(&sdev->sdev_gendev, supplier);
> >> +                       put_device(supplier);
> >> +               }
> >>         }
> >>  }
> >>
> >> --
> >> 2.17.1
> >

Re: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Adrian Hunter]

Martin, perhaps you could consider picking up this patch if no one objects?


On 16/07/21 2:44 pm, Adrian Hunter wrote:
> Managed device links are deleted by device_del(). However it is possible to
> add a device link to a consumer before device_add(), and then discovering
> an error prevents the device from being used. In that case normally
> references to the device would be dropped and the device would be deleted.
> However the device link holds a reference to the device, so the device link
> and device remain indefinitely (unless the supplier is deleted).
> 
> For UFSHCD, if a LUN fails to probe (e.g. absent BOOT WLUN), the device
> will not have been registered but can still have a device link holding a
> reference to the device. The unwanted device link will prevent runtime
> suspend indefinitely.
> 
> Amend device link removal to accept removal of a link with an unregistered
> consumer device (suggested by Rafael), and fix UFSHCD by explicitly
> deleting the device link when SCSI destroys the SCSI device.
> 
> Fixes: b294ff3e34490 ("scsi: ufs: core: Enable power management for wlun")
> Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
> Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
> ---
>  drivers/base/core.c       |  2 ++
>  drivers/scsi/ufs/ufshcd.c | 23 +++++++++++++++++++++--
>  2 files changed, 23 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/base/core.c b/drivers/base/core.c
> index 2de8f7d8cf54..983e895d4ced 100644
> --- a/drivers/base/core.c
> +++ b/drivers/base/core.c
> @@ -887,6 +887,8 @@ static void device_link_put_kref(struct device_link *link)
>  {
>  	if (link->flags & DL_FLAG_STATELESS)
>  		kref_put(&link->kref, __device_link_del);
> +	else if (!device_is_registered(link->consumer))
> +		__device_link_del(&link->kref);
>  	else
>  		WARN(1, "Unable to drop a managed device link reference\n");
>  }
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index 708b3b62fc4d..9864a8ee0263 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -5020,15 +5020,34 @@ static int ufshcd_slave_configure(struct scsi_device *sdev)
>  static void ufshcd_slave_destroy(struct scsi_device *sdev)
>  {
>  	struct ufs_hba *hba;
> +	unsigned long flags;
>  
>  	hba = shost_priv(sdev->host);
>  	/* Drop the reference as it won't be needed anymore */
>  	if (ufshcd_scsi_to_upiu_lun(sdev->lun) == UFS_UPIU_UFS_DEVICE_WLUN) {
> -		unsigned long flags;
> -
>  		spin_lock_irqsave(hba->host->host_lock, flags);
>  		hba->sdev_ufs_device = NULL;
>  		spin_unlock_irqrestore(hba->host->host_lock, flags);
> +	} else if (hba->sdev_ufs_device) {
> +		struct device *supplier = NULL;
> +
> +		/* Ensure UFS Device WLUN exists and does not disappear */
> +		spin_lock_irqsave(hba->host->host_lock, flags);
> +		if (hba->sdev_ufs_device) {
> +			supplier = &hba->sdev_ufs_device->sdev_gendev;
> +			get_device(supplier);
> +		}
> +		spin_unlock_irqrestore(hba->host->host_lock, flags);
> +
> +		if (supplier) {
> +			/*
> +			 * If a LUN fails to probe (e.g. absent BOOT WLUN), the
> +			 * device will not have been registered but can still
> +			 * have a device link holding a reference to the device.
> +			 */
> +			device_link_remove(&sdev->sdev_gendev, supplier);
> +			put_device(supplier);
> +		}
>  	}
>  }
>  
> 

Re: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Bart Van Assche]

On 8/4/21 8:33 AM, Adrian Hunter wrote:
> Martin, perhaps you could consider picking up this patch if no one objects?

Since patch 1/2 went in through Greg's tree, if patch 2/2 goes upstream 
via Martin's tree, will the resulting kernel be bisectable if Linus 
pulls Martin's tree before he pulls Greg's tree?

Thanks,

Bart.

Re: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Martin K. Petersen]

> Since patch 1/2 went in through Greg's tree, if patch 2/2 goes
> upstream via Martin's tree, will the resulting kernel be bisectable if
> Linus pulls Martin's tree before he pulls Greg's tree?

Also, this patch will need to be rebased on top of 5.15/scsi-staging due
to all the UFS churn in this release.

-- 
Martin K. Petersen	Oracle Linux Engineering

Re: [PATCH V4 2/2] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Adrian Hunter]

On 6/08/21 5:50 am, Martin K. Petersen wrote:
> 
>> Since patch 1/2 went in through Greg's tree, if patch 2/2 goes
>> upstream via Martin's tree, will the resulting kernel be bisectable if
>> Linus pulls Martin's tree before he pulls Greg's tree?
> 
> Also, this patch will need to be rebased on top of 5.15/scsi-staging due
> to all the UFS churn in this release.
> 

Patch "driver core: Prevent warning when removing a device link from
unregistered consumer" is already in Linus' tree:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e64daad660a0c9ace3acdc57099fffe5ed83f977


I will rebase on 5.15/scsi-staging

[PATCH V5] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Adrian Hunter]

Managed device links are deleted by device_del(). However it is possible to
add a device link to a consumer before device_add(), and then discovering
an error prevents the device from being used. In that case normally
references to the device would be dropped and the device would be deleted.
However the device link holds a reference to the device, so the device link
and device remain indefinitely (unless the supplier is deleted).

For UFSHCD, if a LUN fails to probe (e.g. absent BOOT WLUN), the device
will not have been registered but can still have a device link holding a
reference to the device. The unwanted device link will prevent runtime
suspend indefinitely.

Amend device link removal to accept removal of a link with an unregistered
consumer device (suggested by Rafael), and fix UFSHCD by explicitly
deleting the device link when SCSI destroys the SCSI device.

Fixes: b294ff3e34490 ("scsi: ufs: core: Enable power management for wlun")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
---


Patch "driver core: Prevent warning when removing a device link from
unregistered consumer" is already in Linus' tree:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e64daad660a0c9ace3acdc57099fffe5ed83f977


Changes in V5:

  Rebase on 5.15/scsi-staging



 drivers/base/core.c       |  2 ++
 drivers/scsi/ufs/ufshcd.c | 23 +++++++++++++++++++++--
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/drivers/base/core.c b/drivers/base/core.c
index cadcade65825..9badd7f7fe62 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -884,6 +884,8 @@ static void device_link_put_kref(struct device_link *link)
 {
 	if (link->flags & DL_FLAG_STATELESS)
 		kref_put(&link->kref, __device_link_del);
+	else if (!device_is_registered(link->consumer))
+		__device_link_del(&link->kref);
 	else
 		WARN(1, "Unable to drop a managed device link reference\n");
 }
diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index 6c263e94144b..9f72698ff597 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -5028,6 +5028,7 @@ static int ufshcd_slave_configure(struct scsi_device *sdev)
 static void ufshcd_slave_destroy(struct scsi_device *sdev)
 {
 	struct ufs_hba *hba;
+	unsigned long flags;
 
 	hba = shost_priv(sdev->host);
 
@@ -5035,11 +5036,29 @@ static void ufshcd_slave_destroy(struct scsi_device *sdev)
 
 	/* Drop the reference as it won't be needed anymore */
 	if (ufshcd_scsi_to_upiu_lun(sdev->lun) == UFS_UPIU_UFS_DEVICE_WLUN) {
-		unsigned long flags;
-
 		spin_lock_irqsave(hba->host->host_lock, flags);
 		hba->sdev_ufs_device = NULL;
 		spin_unlock_irqrestore(hba->host->host_lock, flags);
+	} else if (hba->sdev_ufs_device) {
+		struct device *supplier = NULL;
+
+		/* Ensure UFS Device WLUN exists and does not disappear */
+		spin_lock_irqsave(hba->host->host_lock, flags);
+		if (hba->sdev_ufs_device) {
+			supplier = &hba->sdev_ufs_device->sdev_gendev;
+			get_device(supplier);
+		}
+		spin_unlock_irqrestore(hba->host->host_lock, flags);
+
+		if (supplier) {
+			/*
+			 * If a LUN fails to probe (e.g. absent BOOT WLUN), the
+			 * device will not have been registered but can still
+			 * have a device link holding a reference to the device.
+			 */
+			device_link_remove(&sdev->sdev_gendev, supplier);
+			put_device(supplier);
+		}
 	}
 }
 
-- 
2.17.1

Re: [PATCH V5] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Martin K. Petersen]

Adrian,

> Managed device links are deleted by device_del(). However it is
> possible to add a device link to a consumer before device_add(), and
> then discovering an error prevents the device from being used. In that
> case normally references to the device would be dropped and the device
> would be deleted.  However the device link holds a reference to the
> device, so the device link and device remain indefinitely (unless the
> supplier is deleted).

Applied to 5.15/scsi-staging, thanks!

-- 
Martin K. Petersen	Oracle Linux Engineering

Re: [PATCH V5] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe

[Martin K. Petersen]

On Fri, 6 Aug 2021 16:04:41 +0300, Adrian Hunter wrote:

> Managed device links are deleted by device_del(). However it is possible to
> add a device link to a consumer before device_add(), and then discovering
> an error prevents the device from being used. In that case normally
> references to the device would be dropped and the device would be deleted.
> However the device link holds a reference to the device, so the device link
> and device remain indefinitely (unless the supplier is deleted).
> 
> [...]

Applied to 5.15/scsi-queue, thanks!

[1/1] scsi: ufshcd: Fix device links when BOOT WLUN fails to probe
      https://git.kernel.org/mkp/scsi/c/bf25967ac541

-- 
Martin K. Petersen	Oracle Linux Engineering