Hi Krzysztof,
...
> @@ -88,24 +89,25 @@ int __i915_subtests(const char *caller,
> int (*teardown)(int err, void *data),
> const struct i915_subtest *st,
> unsigned int count,
> - void *data);
> -#define i915_subtests(T, data) \
> + void *data,
> + struct drm_device *device);
> +#define i915_subtests(T, data, device) \
> __i915_subtests(__func__, \
> __i915_nop_setup, __i915_nop_teardown, \
> - T, ARRAY_SIZE(T), data)
> -#define i915_live_subtests(T, data) ({ \
> + T, ARRAY_SIZE(T), data, device)
> +#define i915_live_subtests(T, data, device) ({ \
> typecheck(struct drm_i915_private *, data); \
> (data)->gt[0]->uc.guc.submission_state.sched_disable_delay_ms = 0; \
> __i915_subtests(__func__, \
> __i915_live_setup, __i915_live_teardown, \
> - T, ARRAY_SIZE(T), data); \
> + T, ARRAY_SIZE(T), data, device); \
> })
> -#define intel_gt_live_subtests(T, data) ({ \
> +#define intel_gt_live_subtests(T, data, device) ({ \
As I stated in my previous review, here perhaps it's better to
use gt_*() and we can drop this change. What do you think?
Andi
> typecheck(struct intel_gt *, data); \
> (data)->uc.guc.submission_state.sched_disable_delay_ms = 0; \
> __i915_subtests(__func__, \
> __intel_gt_live_setup, __intel_gt_live_teardown, \
> - T, ARRAY_SIZE(T), data); \
> + T, ARRAY_SIZE(T), data, device); \
> })
