[PATCH i-g-t] tests/intel/xe_exec_fault_mode: Don't return early

Andrzej Hajda andrzej.hajda at intel.com
Tue Aug 27 14:07:47 UTC 2024



On 27.08.2024 12:52, Nirmoy Das wrote:
> Tests that are causing pagefaults should wait for exec queue to be ban
> otherwise pending engine resets because of on-going pagefaults would
> cause failure in subsequent tests to fail.
>
> Cc: Andrzej Hajda <andrzej.hajda at intel.com>
> Cc: Matthew Brost <matthew.brost at intel.com>
> Cc: Tejas Upadhyay <tejas.upadhyay at intel.com>
> Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1630
> Signed-off-by: Nirmoy Das <nirmoy.das at intel.com>
> ---
>   tests/intel/xe_exec_fault_mode.c | 27 +++++++++++++++++++++++++++
>   1 file changed, 27 insertions(+)
>
> diff --git a/tests/intel/xe_exec_fault_mode.c b/tests/intel/xe_exec_fault_mode.c
> index 1f1f1e50b..339c79353 100644
> --- a/tests/intel/xe_exec_fault_mode.c
> +++ b/tests/intel/xe_exec_fault_mode.c
> @@ -36,6 +36,22 @@
>   #define INVALID_VA	(0x1 << 8)
>   #define ENABLE_SCRATCH  (0x1 << 9)
>   
> +static int get_ban_property(int xe, struct drm_xe_engine_class_instance *eci,
> +			    uint32_t vm, uint32_t exec_queue)
> +{
> +	struct drm_xe_exec_queue_get_property args = {
> +		.value = -1,
> +		.reserved[0] = 0,
> +		.property = DRM_XE_EXEC_QUEUE_GET_PROPERTY_BAN,
> +	};
> +
> +	args.exec_queue_id = exec_queue;
> +
> +	do_ioctl(xe, DRM_IOCTL_XE_EXEC_QUEUE_GET_PROPERTY, &args);
> +
> +	return args.value;
> +}
> +
>   /**
>    * SUBTEST: invalid-va
>    * Description: Access invalid va and check for EIO through user fence.
> @@ -324,6 +340,17 @@ test_exec(int fd, struct drm_xe_engine_class_instance *eci,
>   	xe_wait_ufence(fd, &data[0].vm_sync, USER_FENCE_VALUE,
>   		       bind_exec_queues[0], NSEC_PER_SEC);
>   
> +	if ((flags & INVALID_FAULT)) {
> +		igt_set_timeout(1, "");
You can add "waiting for ban" or sth similar here for free.
> +		for (i = 0; i < n_execs; i++) {
> +			int e = i % n_exec_queues;
> +				while (!get_ban_property(fd, eci, vm,
> +							 exec_queues[e]))
> +					sched_yield();
> +		}
> +		igt_reset_timeout();
> +	}
> +

Why not iterate over exec queues instead of execs?

Regards
Andrzej

>   	if (!(flags & INVALID_FAULT) && !(flags & INVALID_VA)) {
>   		for (i = j; i < n_execs; i++)
>   				igt_assert_eq(data[i].data, 0xc0ffee);



More information about the igt-dev mailing list