[PATCH] accel/ivpu: Trigger device recovery on engine reset/resume failure

[Jacek Lawrynowicz]

Applied to drm-misc-fixes

On 5/28/2025 5:42 PM, Jacek Lawrynowicz wrote:
> From: Karol Wachowski <karol.wachowski at intel.com>
> 
> Trigger full device recovery when the driver fails to restore device state
> via engine reset and resume operations. This is necessary because, even if
> submissions from a faulty context are blocked, the NPU may still process
> previously submitted faulty jobs if the engine reset fails to abort them.
> Such jobs can continue to generate faults and occupy device resources.
> When engine reset is ineffective, the only way to recover is to perform
> a full device recovery.
> 
> Fixes: dad945c27a42 ("accel/ivpu: Add handling of VPU_JSM_STATUS_MVNCI_CONTEXT_VIOLATION_HW")
> Cc: <stable at vger.kernel.org> # v6.15+
> Signed-off-by: Karol Wachowski <karol.wachowski at intel.com>
> Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz at linux.intel.com>
> ---
>  drivers/accel/ivpu/ivpu_job.c     | 6 ++++--
>  drivers/accel/ivpu/ivpu_jsm_msg.c | 9 +++++++--
>  2 files changed, 11 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c
> index 1c8e283ad9854..fae8351aa3309 100644
> --- a/drivers/accel/ivpu/ivpu_job.c
> +++ b/drivers/accel/ivpu/ivpu_job.c
> @@ -986,7 +986,8 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
>  		return;
>  
>  	if (vdev->fw->sched_mode == VPU_SCHEDULING_MODE_HW)
> -		ivpu_jsm_reset_engine(vdev, 0);
> +		if (ivpu_jsm_reset_engine(vdev, 0))
> +			return;
>  
>  	mutex_lock(&vdev->context_list_lock);
>  	xa_for_each(&vdev->context_xa, ctx_id, file_priv) {
> @@ -1009,7 +1010,8 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
>  	if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW)
>  		goto runtime_put;
>  
> -	ivpu_jsm_hws_resume_engine(vdev, 0);
> +	if (ivpu_jsm_hws_resume_engine(vdev, 0))
> +		return;
>  	/*
>  	 * In hardware scheduling mode NPU already has stopped processing jobs
>  	 * and won't send us any further notifications, thus we have to free job related resources
> diff --git a/drivers/accel/ivpu/ivpu_jsm_msg.c b/drivers/accel/ivpu/ivpu_jsm_msg.c
> index 219ab8afefabd..0256b2dfefc10 100644
> --- a/drivers/accel/ivpu/ivpu_jsm_msg.c
> +++ b/drivers/accel/ivpu/ivpu_jsm_msg.c
> @@ -7,6 +7,7 @@
>  #include "ivpu_hw.h"
>  #include "ivpu_ipc.h"
>  #include "ivpu_jsm_msg.h"
> +#include "ivpu_pm.h"
>  #include "vpu_jsm_api.h"
>  
>  const char *ivpu_jsm_msg_type_to_str(enum vpu_ipc_msg_type type)
> @@ -163,8 +164,10 @@ int ivpu_jsm_reset_engine(struct ivpu_device *vdev, u32 engine)
>  
>  	ret = ivpu_ipc_send_receive(vdev, &req, VPU_JSM_MSG_ENGINE_RESET_DONE, &resp,
>  				    VPU_IPC_CHAN_ASYNC_CMD, vdev->timeout.jsm);
> -	if (ret)
> +	if (ret) {
>  		ivpu_err_ratelimited(vdev, "Failed to reset engine %d: %d\n", engine, ret);
> +		ivpu_pm_trigger_recovery(vdev, "Engine reset failed");
> +	}
>  
>  	return ret;
>  }
> @@ -354,8 +357,10 @@ int ivpu_jsm_hws_resume_engine(struct ivpu_device *vdev, u32 engine)
>  
>  	ret = ivpu_ipc_send_receive(vdev, &req, VPU_JSM_MSG_HWS_RESUME_ENGINE_DONE, &resp,
>  				    VPU_IPC_CHAN_ASYNC_CMD, vdev->timeout.jsm);
> -	if (ret)
> +	if (ret) {
>  		ivpu_err_ratelimited(vdev, "Failed to resume engine %d: %d\n", engine, ret);
> +		ivpu_pm_trigger_recovery(vdev, "Engine resume failed");
> +	}
>  
>  	return ret;
>  }