[PATCH v8 8/9] vfio: Add a reset_done callback for vfio-pci driver

Farhan Ali posted 9 patches 2 weeks, 2 days ago
[PATCH v8 8/9] vfio: Add a reset_done callback for vfio-pci driver
Posted by Farhan Ali 2 weeks, 2 days ago
On error recovery for a PCI device bound to vfio-pci driver, we want to
recover the state of the device to its last known saved state. The callback
restores the state of the device to its initial saved state.

Reviewed-by: Julian Ruess <julianr@linux.ibm.com>
Signed-off-by: Farhan Ali <alifm@linux.ibm.com>
---
 drivers/vfio/pci/vfio_pci_core.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
index f677705921e6..c92c6c512b24 100644
--- a/drivers/vfio/pci/vfio_pci_core.c
+++ b/drivers/vfio/pci/vfio_pci_core.c
@@ -2249,6 +2249,17 @@ pci_ers_result_t vfio_pci_core_aer_err_detected(struct pci_dev *pdev,
 }
 EXPORT_SYMBOL_GPL(vfio_pci_core_aer_err_detected);
 
+static void vfio_pci_core_aer_reset_done(struct pci_dev *pdev)
+{
+	struct vfio_pci_core_device *vdev = dev_get_drvdata(&pdev->dev);
+
+	if (!vdev->pci_saved_state)
+		return;
+
+	pci_load_saved_state(pdev, vdev->pci_saved_state);
+	pci_restore_state(pdev);
+}
+
 int vfio_pci_core_sriov_configure(struct vfio_pci_core_device *vdev,
 				  int nr_virtfn)
 {
@@ -2313,6 +2324,7 @@ EXPORT_SYMBOL_GPL(vfio_pci_core_sriov_configure);
 
 const struct pci_error_handlers vfio_pci_core_err_handlers = {
 	.error_detected = vfio_pci_core_aer_err_detected,
+	.reset_done = vfio_pci_core_aer_reset_done,
 };
 EXPORT_SYMBOL_GPL(vfio_pci_core_err_handlers);
 
-- 
2.43.0
Re: [PATCH v8 8/9] vfio: Add a reset_done callback for vfio-pci driver
Posted by Niklas Schnelle 1 week, 4 days ago
On Thu, 2026-01-22 at 11:44 -0800, Farhan Ali wrote:
> On error recovery for a PCI device bound to vfio-pci driver, we want to
> recover the state of the device to its last known saved state. The callback
> restores the state of the device to its initial saved state.

I feel like "its last known saved state" and "its initial saved state"
might not be the same thing. The way vdev->pci_saved_state is used at
the moment it really is the initial saved state. And I think that makes
sense for a simple approach to recovery where user-space drivers would
basically just do their initial setup / probing again.

> 
> Reviewed-by: Julian Ruess <julianr@linux.ibm.com>
> Signed-off-by: Farhan Ali <alifm@linux.ibm.com>
> ---
>  drivers/vfio/pci/vfio_pci_core.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
> index f677705921e6..c92c6c512b24 100644
> --- a/drivers/vfio/pci/vfio_pci_core.c
> +++ b/drivers/vfio/pci/vfio_pci_core.c
> @@ -2249,6 +2249,17 @@ pci_ers_result_t vfio_pci_core_aer_err_detected(struct pci_dev *pdev,
>  }
>  EXPORT_SYMBOL_GPL(vfio_pci_core_aer_err_detected);
>  
> +static void vfio_pci_core_aer_reset_done(struct pci_dev *pdev)
> +{
> +	struct vfio_pci_core_device *vdev = dev_get_drvdata(&pdev->dev);
> +
> +	if (!vdev->pci_saved_state)
> +		return;
> +
> +	pci_load_saved_state(pdev, vdev->pci_saved_state);
> +	pci_restore_state(pdev);
> +}
> +
>  int vfio_pci_core_sriov_configure(struct vfio_pci_core_device *vdev,
>  				  int nr_virtfn)
>  {
> @@ -2313,6 +2324,7 @@ EXPORT_SYMBOL_GPL(vfio_pci_core_sriov_configure);
>  
>  const struct pci_error_handlers vfio_pci_core_err_handlers = {
>  	.error_detected = vfio_pci_core_aer_err_detected,
> +	.reset_done = vfio_pci_core_aer_reset_done,
>  };
>  EXPORT_SYMBOL_GPL(vfio_pci_core_err_handlers);
>  

Code looks good, thanks!

Reviewed-by: Niklas Schnelle <schnelle@linux.ibm.com>