[PATCH 1/3] drm/xe/pf: Disable PF restart worker on device removal

[Piotr Piórkowski]

Michal Wajdeczko <michal.wajdeczko at intel.com> wrote on śro [2025-lip-30 19:49:11 +0200]:
> We can't let restart worker run once device is removed, since other
> data that it might want to access could be already released.
> Explicitly disable worker as part of device cleanup action.
> 
> Fixes: a4d1c5d0b99b ("drm/xe/pf: Move VFs reprovisioning to worker")
> Signed-off-by: Michal Wajdeczko <michal.wajdeczko at intel.com>
> ---
>  drivers/gpu/drm/xe/xe_gt_sriov_pf.c | 30 ++++++++++++++++++++++++++++-
>  1 file changed, 29 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
> index 35489fa81825..2761319fdc26 100644
> --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
> +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
> @@ -50,6 +50,11 @@ static void pf_init_workers(struct xe_gt *gt)
>  	INIT_WORK(&gt->sriov.pf.workers.restart, pf_worker_restart_func);
>  }
>  
> +static void pf_fini_workers(struct xe_gt *gt)
> +{

NIT: I wonder if xe_gt_assert should also be here.
It's good that we're keeping an eye on this in pf_init_late, during register fini action,
but theoretically someone could call it directly at some point.

> +	disable_work_sync(&gt->sriov.pf.workers.restart);
> +}
> +
>  /**
>   * xe_gt_sriov_pf_init_early - Prepare SR-IOV PF data structures on PF.
>   * @gt: the &xe_gt to initialize
> @@ -79,6 +84,21 @@ int xe_gt_sriov_pf_init_early(struct xe_gt *gt)
>  	return 0;
>  }
>  
> +static void pf_fini_action(void *arg)
> +{
> +	struct xe_gt *gt = arg;
> +
> +	pf_fini_workers(gt);
> +}
> +
> +static int pf_init_late(struct xe_gt *gt)
> +{
> +	struct xe_device *xe = gt_to_xe(gt);
> +
> +	xe_gt_assert(gt, IS_SRIOV_PF(xe));
> +	return devm_add_action_or_reset(xe->drm.dev, pf_fini_action, gt);
> +}
> +
>  /**
>   * xe_gt_sriov_pf_init - Prepare SR-IOV PF data structures on PF.
>   * @gt: the &xe_gt to initialize
> @@ -95,7 +115,15 @@ int xe_gt_sriov_pf_init(struct xe_gt *gt)
>  	if (err)
>  		return err;
>  
> -	return xe_gt_sriov_pf_migration_init(gt);
> +	err = xe_gt_sriov_pf_migration_init(gt);
> +	if (err)
> +		return err;
> +
> +	err = pf_init_late(gt);
> +	if (err)
> +		return err;
> +
> +	return 0;
>  }
>  

LGTM:
Reviewed-by: Piotr Piórkowski <piotr.piorkowski at intel.com>

>  static bool pf_needs_enable_ggtt_guest_update(struct xe_device *xe)
> -- 
> 2.47.1
> 

--