<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Tue, Feb 20, 2018 at 1:57 PM, Nanley Chery <<a href="mailto:nanleychery@gmail.com" target="_blank">nanleychery@gmail.com</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="HOEnZb"><div class="h5">On Tue, Feb 20, 2018 at 11:31:08AM -0800, Jason Ekstrand wrote: > On Tue, Feb 20, 2018 at 11:26 AM, Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> > wrote: > > > On Tue, Feb 20, 2018 at 11:25 AM, Nanley Chery <<a href="mailto:nanleychery@gmail.com">nanleychery@gmail.com</a>> > > wrote: > > > >> On Fri, Feb 16, 2018 at 09:28:43AM -0800, Jason Ekstrand wrote: > >> > --- > >> > src/intel/vulkan/anv_formats.c | 9 +++++++ > >> > src/intel/vulkan/anv_image.c | 53 ++++++++++++++++++++++++++---- > >> ------------ > >> > 2 files changed, 42 insertions(+), 20 deletions(-) > >> > > >> > diff --git a/src/intel/vulkan/anv_formats.c > >> b/src/intel/vulkan/anv_formats.c > >> > index 9c52ad5..3c17366 100644 > >> > --- a/src/intel/vulkan/anv_formats.c > >> > +++ b/src/intel/vulkan/anv_formats.c > >> > @@ -671,9 +671,18 @@ get_wsi_format_modifier_properties_list(const > >> struct anv_physical_device *physic > >> > DRM_FORMAT_MOD_LINEAR, > >> > I915_FORMAT_MOD_X_TILED, > >> > I915_FORMAT_MOD_Y_TILED, > >> > + I915_FORMAT_MOD_Y_TILED_CCS, > >> > }; > >> > > >> > for (uint32_t i = 0; i < ARRAY_SIZE(modifiers); i++) { > >> > + const struct isl_drm_modifier_info *mod_info = > >> > + isl_drm_modifier_get_info(modifiers[i]); > >> > + > >> > + if (mod_info->aux_usage == ISL_AUX_USAGE_CCS_E && > >> > + !isl_format_supports_ccs_e(&physical_device->info, > >> > + anv_format->planes[0].isl_for > >> mat)) > >> > + continue; > >> > + > >> > vk_outarray_append(&out, mod_props) { > >> > mod_props->modifier = modifiers[i]; > >> > if (isl_drm_modifier_has_aux(modifiers[i])) > >> > diff --git a/src/intel/vulkan/anv_image.c b/src/intel/vulkan/anv_image.c > >> > index a2bae7b..d7c2e55 100644 > >> > --- a/src/intel/vulkan/anv_image.c > >> > +++ b/src/intel/vulkan/anv_image.c > >> > @@ -515,6 +515,7 @@ score_drm_format_mod(uint64_t modifier) > >> > case DRM_FORMAT_MOD_LINEAR: return 1; > >> > case I915_FORMAT_MOD_X_TILED: return 2; > >> > case I915_FORMAT_MOD_Y_TILED: return 3; > >> > + case I915_FORMAT_MOD_Y_TILED_CCS: return 4; > >> > default: unreachable("bad DRM format modifier"); > >> > } > >> > } > >> > @@ -746,8 +747,13 @@ void anv_GetImageSubresourceLayout( > >> > VkSubresourceLayout* layout) > >> > { > >> > ANV_FROM_HANDLE(anv_image, image, _image); > >> > - const struct anv_surface *surface = > >> > - get_surface(image, subresource->aspectMask); > >> > + > >> > + const struct anv_surface *surface; > >> > + if (subresource->aspectMask == VK_IMAGE_ASPECT_PLANE_1_BIT_KHR && > >> > + isl_drm_modifier_has_aux(image->drm_format_mod)) > >> > + surface = &image->planes[0].aux_surface; > >> > + else > >> > + surface = get_surface(image, subresource->aspectMask); > >> > > >> > assert(__builtin_popcount(subresource->aspectMask) == 1); > >> > > >> > @@ -862,25 +868,20 @@ anv_layout_to_aux_usage(const struct > >> gen_device_info * const devinfo, > >> > } > >> > > >> > > >> > - case VK_IMAGE_LAYOUT_PRESENT_SRC_KHR: > >> > + case VK_IMAGE_LAYOUT_PRESENT_SRC_KHR: { > >> > assert(image->aspects == VK_IMAGE_ASPECT_COLOR_BIT); > >> > > >> > - /* On SKL+, the render buffer can be decompressed by the > >> presentation > >> > - * engine. Support for this feature has not yet landed in the > >> wider > >> > - * ecosystem. TODO: Update this code when support lands. > >> > - * > >> > - * From the BDW PRM, Vol 7, Render Target Resolve: > >> > - * > >> > - * If the MCS is enabled on a non-multisampled render target, > >> the > >> > - * render target must be resolved before being used for other > >> > - * purposes (display, texture, CPU lock) The clear value from > >> > - * SURFACE_STATE is written into pixels in the render target > >> > - * indicated as clear in the MCS. > >> > - * > >> > - * Pre-SKL, the render buffer must be resolved before being used > >> for > >> > - * presentation. We can infer that the auxiliary buffer is not > >> used. > >> > + /* When handing the image off to the presentation engine, we > >> need to > >> > + * ensure that things are properly resolved. For images with no > >> > + * modifier, we assume that they follow the old rules and always > >> need > >> > + * a full resolve because the PE doesn't understand any form of > >> > + * compression. For images with modifiers, we use the aux usage > >> from > >> > + * the modifier. > >> > */ > >> > - return ISL_AUX_USAGE_NONE; > >> > + const struct isl_drm_modifier_info *mod_info = > >> > + isl_drm_modifier_get_info(image->drm_format_mod); > >> > + return mod_info ? mod_info->aux_usage : ISL_AUX_USAGE_NONE; > >> > + } > >> > > >> > > >> > /* Rendering Layouts */ > >> > @@ -960,8 +961,20 @@ anv_layout_to_fast_clear_type(const struct > >> gen_device_info * const devinfo, > >> > case VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL: > >> > return ANV_FAST_CLEAR_ANY; > >> > > >> > - case VK_IMAGE_LAYOUT_PRESENT_SRC_KHR: > >> > - return ANV_FAST_CLEAR_NONE; > >> > + case VK_IMAGE_LAYOUT_PRESENT_SRC_KHR: { > >> > + assert(image->aspects == VK_IMAGE_ASPECT_COLOR_BIT); > >> > + > >> > + /* When handing the image off to the presentation engine, we > >> need to > >> > + * ensure that things are properly resolved. For images with no > >> > + * modifier, we assume that they follow the old rules and always > >> need > >> > + * a full resolve because the PE doesn't understand any form of > >> > + * compression. For images with modifiers, we use the value > >> from the > >> > + * modifier. > >> > + */ > >> > + const struct isl_drm_modifier_info *mod_info = > >> > + isl_drm_modifier_get_info(image->drm_format_mod); > >> > + return mod_info && mod_info->supports_clear_color; > >> > >> This should return an enum value, not a boolean, right?= > >> > > > > Yup. Rebase fail. I'll rework and re-send. > > > > It's worth noting that it doesn't actually matter today because we don't > have any modifiers that support fast-clear. Maybe we should just leave it > returning NONE? > </div></div>It is also outside the scope of this patch, so I guess so. </blockquote><div> </div><div>Ok, I'll do that and add an assert so that we catch it when we add the Y_TILED_CCS_CLEAR modifier. </div><div>--Jason </div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> -Nanley <div class="HOEnZb"><div class="h5"> > > > -Nanley > >> > >> > + } > >> > > >> > default: > >> > /* If the image has CCS_E enabled all the time then we can use > >> > -- > >> > 2.5.0.400.gff86faf > >> > > >> > _______________________________________________ > >> > mesa-dev mailing list > >> > <a href="mailto:mesa-dev@lists.freedesktop.org">mesa-dev@lists.freedesktop.org</a> > >> > <a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev" rel="noreferrer" target="_blank">https://lists.freedesktop.org/mailman/listinfo/mesa-dev</a> > >> > > > > </div></div></blockquote></div> </div></div>