On Thu, Aug 29, 2019 at 01:17:22PM +0100, Chris Wilson wrote:
> We would like some freedom to break the user API/ABI for future HW but
> yet still expose the driver for upstream development on that HW.
> Currently, we have the i915.force_probe module parameter to avoid binding
> to HW while the driver is under development, but that is still a little
> too soft with respect to the stringent no-regression rules if we also
> plan to be redesigning the uAPI to go along with the new HW.
>
> To allow the uAPI to be changed during development, only expose that API
> and in development HW under STAGING (and BROKEN). Hopefully, making it
> explicit that such interfaces to that HW are under development and not
> to be blindly enabled by distributions.
>
> Signed-off-by: Chris Wilson <ch...@chris-wilson.co.uk>
> Cc: Daniel Vetter <daniel.vet...@ffwll.ch>
> Cc: Joonas Lahtinen <joonas.lahti...@linux.intel.com>
> Cc: Jani Nikula <jani.nik...@intel.com>
> Cc: Rodrigo Vivi <rodrigo.v...@intel.com>
> Cc: Dave Airlie <airl...@redhat.com>
> ---
> drivers/gpu/drm/i915/Kconfig.debug | 15 +++++++++++++++
> drivers/gpu/drm/i915/i915_gem.c | 10 ++++++++++
> drivers/gpu/drm/i915/intel_device_info.h | 1 +
> 3 files changed, 26 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/Kconfig.debug
> b/drivers/gpu/drm/i915/Kconfig.debug
> index 00786a142ff0..7707d1230b85 100644
> --- a/drivers/gpu/drm/i915/Kconfig.debug
> +++ b/drivers/gpu/drm/i915/Kconfig.debug
> @@ -36,6 +36,9 @@ config DRM_I915_DEBUG
> select DRM_I915_SELFTEST
> select DRM_I915_DEBUG_RUNTIME_PM
> select DRM_I915_DEBUG_MMIO
> + select STAGING
> + select BROKEN
> + select DRM_I915_DEBUG_EARLY_API
I'm not sure about forcing this all all debug.
> default n
> help
> Choose this option to turn on extra driver debugging that may
> affect
> @@ -212,3 +215,15 @@ config DRM_I915_DEBUG_RUNTIME_PM
> driver loading, suspend and resume operations.
>
> If in doubt, say "N"
> +
> +config DRM_I915_DEBUG_EARLY_API
> + bool "Expose new HW uAPIs for early testing"
> + depends on DRM_I915
> + depends on STAGING
> + depends on BROKEN
instead why not
select STAGING
select BROKEN
depends on DRM_I915_DEBUG
?
> + default n
> + help
> + Expose in-development uAPI for early testing on HW. Such uAPI
> + will likely subtly change before release and must not be relied upon.
> +
> + If in doubt, say "N"
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index bb39c64a0a17..eab2219d7a50 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -1344,6 +1344,15 @@ int i915_gem_init(struct drm_i915_private *dev_priv)
> {
> int ret;
>
> + if (INTEL_INFO(dev_priv)->early_render_uapi) {
> + if (!IS_ENABLED(CONFIG_DRM_I915_DEBUG_EARLY_API)) {
> + ret = -EIO;
> + goto err_disable;
> + }
> +
> + add_taint(TAINT_USER, LOCKDEP_STILL_OK);
> + }
> +
> /* We need to fallback to 4K pages if host doesn't support huge gtt. */
> if (intel_vgpu_active(dev_priv) && !intel_vgpu_has_huge_gtt(dev_priv))
> mkwrite_device_info(dev_priv)->page_sizes =
> @@ -1476,6 +1485,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)
> intel_timelines_fini(dev_priv);
> }
>
> +err_disable:
> if (ret == -EIO) {
> /*
> * Allow engines or uC initialisation to fail by marking the GPU
> diff --git a/drivers/gpu/drm/i915/intel_device_info.h
> b/drivers/gpu/drm/i915/intel_device_info.h
> index 92e0c2e0954c..5931f619e66b 100644
> --- a/drivers/gpu/drm/i915/intel_device_info.h
> +++ b/drivers/gpu/drm/i915/intel_device_info.h
> @@ -107,6 +107,7 @@ enum intel_ppgtt_type {
> func(is_mobile); \
> func(is_lp); \
> func(require_force_probe); \
> + func(early_render_uapi); \
> /* Keep has_* in alphabetical order */ \
> func(has_64bit_reloc); \
> func(gpu_reset_clobbers_display); \
> --
> 2.23.0
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
