[Mesa-dev] [PATCH 1/3] intel: fix check for 48b ppgtt support

[Chris Wilson]

Quoting Kenneth Graunke (2018-04-18 22:22:54)
> From: Scott D Phillips <scott.d.phillips at intel.com>
> 
> The previous logic of the supports_48b_addresses wasn't actually
> checking if i915.ko was running with full_48bit_ppgtt. The ENOENT
> it was checking for was actually coming from the invalid context
> id provided in the test execbuffer.  There is no path in the
> kernel driver where the presence of
> EXEC_OBJECT_SUPPORTS_48B_ADDRESS leads to an error.
> 
> Instead, check the HAS_ALIASING_PPGTT param for the value 3 which
> corresponds with full_48bit_ppgtt.
> 
> v2 (Ken): Fix in i965 as well.
> ---
>  src/intel/vulkan/anv_device.c          |  4 ++--
>  src/intel/vulkan/anv_gem.c             | 18 ------------------
>  src/intel/vulkan/anv_gem_stubs.c       |  6 ------
>  src/intel/vulkan/anv_private.h         |  1 -
>  src/mesa/drivers/dri/i965/brw_bufmgr.c | 22 ++--------------------
>  5 files changed, 4 insertions(+), 47 deletions(-)
> 
> diff --git a/src/intel/vulkan/anv_device.c b/src/intel/vulkan/anv_device.c
> index 7522b7865c2..116574cb978 100644
> --- a/src/intel/vulkan/anv_device.c
> +++ b/src/intel/vulkan/anv_device.c
> @@ -107,8 +107,8 @@ anv_physical_device_init_heaps(struct anv_physical_device *device, int fd)
>      * EXEC_OBJECT_SUPPORTS_48B_ADDRESS flag and not whether or not the
>      * hardware has actual 48bit address support.
>      */
> -   device->supports_48bit_addresses =
> -      (device->info.gen >= 8) && anv_gem_supports_48b_addresses(fd);
> +   device->supports_48bit_addresses = (device->info.gen >= 8) &&
> +      anv_gem_get_param(fd, I915_PARAM_HAS_ALIASING_PPGTT) == 3;

Just query whether you have move than 4G of GTT. The context query is the
same age as the 48b support and was intended to be used for this purpose.
-Chris