diff mbox series

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

Message ID 20210716114408.17320-3-adrian.hunter@intel.com (mailing list archive)
State Not Applicable, archived
Headers show
Series driver core: Add ability to delete device links of unregistered devices | expand

Commit Message

Adrian Hunter July 16, 2021, 11:44 a.m. UTC
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(-)

Comments

Avri Altman July 17, 2021, 6:02 p.m. UTC | #1
> 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
Adrian Hunter Aug. 2, 2021, 5:45 a.m. UTC | #2
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
>
Avri Altman Aug. 2, 2021, 7:03 a.m. UTC | #3
> 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
> >
Adrian Hunter Aug. 4, 2021, 3:33 p.m. UTC | #4
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);
> +		}
>  	}
>  }
>  
>
Bart Van Assche Aug. 5, 2021, 9:49 p.m. UTC | #5
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.
Martin K. Petersen Aug. 6, 2021, 2:50 a.m. UTC | #6
> 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.
Adrian Hunter Aug. 6, 2021, 12:59 p.m. UTC | #7
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
diff mbox series

Patch

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);
+		}
 	}
 }