Re: [PATCH v10 26/34] drm/xe/vf: Replay GuC submission state on pause / unpause

[Raag Jadav <raag.jadav@intel.com>]

On Wed, Oct 08, 2025 at 02:45:24PM -0700, Matthew Brost wrote:
> Fixup GuC submission pause / unpause functions to properly replay any
> possible state lost during VF post migration recovery.
> 
> v3:
>  - Add helpers for revert / replay (Tomasz)
>  - Add comment around WQ NOPs (Tomasz)
> v7:
>  - Only fixup / replay parallel queues once (Testing)
>  - Skip unpause step on queues created after resfix done (Testing)
> 
> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> Reviewed-by: Tomasz Lis <tomasz.lis@intel.com>

...

>  /**
>   * xe_guc_submit_pause - Stop further runs of submission tasks on given GuC.
>   * @guc: the &xe_guc struct instance whose scheduler is to be disabled
> @@ -2018,8 +2145,17 @@ void xe_guc_submit_pause(struct xe_guc *guc)
>  	struct xe_exec_queue *q;
>  	unsigned long index;
>  
> -	xa_for_each(&guc->submission_state.exec_queue_lookup, index, q)
> -		xe_sched_submission_stop_async(&q->guc->sched);
> +	xe_gt_assert(guc_to_gt(guc), vf_recovery(guc));

I'm trying to reuse this into GuC runtime PM flows[1] but hitting this
assert in runtime suspend path. I'm guessing because runtime PM flows are
not applicable for VFs, but any insight into how can I do this properly
without regressing?

[1] https://patchwork.freedesktop.org/series/154017/

Raag

> +
> +	mutex_lock(&guc->submission_state.lock);
> +	xa_for_each(&guc->submission_state.exec_queue_lookup, index, q) {
> +		/* Prevent redundant attempts to stop parallel queues */
> +		if (q->guc->id != index)
> +			continue;
> +
> +		guc_exec_queue_pause(guc, q);
> +	}
> +	mutex_unlock(&guc->submission_state.lock);
>  }