Quoting Lionel Landwerlin (2017-11-02 16:29:48)
> +/* Query various aspects of the topology of the GPU. Below is a list of the
> + * currently supported queries. The motivation of this more detailed query
> + * mechanism is to expose asynmetric properties of the GPU. Starting with CNL,
> + * slices might have different sizes (for example 3 subslices in slice0 and 2
> + * subslices in slice1+). This means we cannot rely on PARAM_SUBSLICE_MASK
> + * anymore.
> + *
> + * When using this parameter, getparam value should point to a structure of
> + * type drm_i915_topology_t. Call this once with query set to the relevant
> + * information to be queried and data_size set to 0. The kernel will then set
> + * params and data_size to the expected length of data[]. The should make sure
> + * memory is allocated to the right length before making a second getparam
> + * with data_size already set. The kernel will then populate data[]. The
> + * meaning of params[] elements is described for each query below.
> + */
> +#define I915_PARAM_TOPOLOGY 51
> +typedef struct drm_i915_topology {
Oh crumbs. Please join the intel_engine_info_ioctl discussion.
As it stands, lets not introduce multiplexing into an already multiplexed
getparam ioctl.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
