| /* | |
| * Copyright 2022 Google LLC | |
| * | |
| * Use of this source code is governed by a BSD-style license that can be | |
| * found in the LICENSE file. | |
| */ | |
| #include "include/core/SkAlphaType.h" | |
| #include "include/core/SkCanvas.h" | |
| #include "include/core/SkColor.h" | |
| #include "include/core/SkColorType.h" | |
| #include "include/core/SkImageInfo.h" | |
| #include "include/core/SkRefCnt.h" | |
| #include "include/core/SkSurface.h" | |
| #include "include/core/SkTypes.h" | |
| #include "include/gpu/GrDirectContext.h" | |
| #include "include/gpu/vk/GrVkBackendContext.h" | |
| #include "include/gpu/vk/VulkanExtensions.h" | |
| #include "tools/gpu/vk/VkTestUtils.h" | |
| #include <string.h> | |
| #include <vulkan/vulkan_core.h> | |
| #include <functional> | |
| #include <memory> | |
| #define ACQUIRE_INST_VK_PROC(name) \ | |
| do { \ | |
| fVk##name = reinterpret_cast<PFN_vk##name>(getProc("vk" #name, backendContext.fInstance, \ | |
| VK_NULL_HANDLE)); \ | |
| if (fVk##name == nullptr) { \ | |
| SkDebugf("Function ptr for vk%s could not be acquired\n", #name); \ | |
| return 1; \ | |
| } \ | |
| } while(false) | |
| int main(int argc, char** argv) { | |
| GrVkBackendContext backendContext; | |
| VkDebugReportCallbackEXT debugCallback; | |
| std::unique_ptr<skgpu::VulkanExtensions> extensions(new skgpu::VulkanExtensions()); | |
| std::unique_ptr<VkPhysicalDeviceFeatures2> features(new VkPhysicalDeviceFeatures2); | |
| // First we need to create a GrVkBackendContext so that we can make a Vulkan GrDirectContext. | |
| // The vast majority of this chunk of code is setting up the VkInstance and VkDevice objects. | |
| // Normally a client will have their own way of creating these objects. This example uses Skia's | |
| // test helper sk_gpu_test::CreateVkBackendContext to aid in this. Clients can look at this | |
| // function as a guide on things to consider when setting up Vulkan for themselves, but they | |
| // should not depend on that function. We may arbitrarily change it as it is meant only for Skia | |
| // internal testing. Additionally it may do some odd things that a normal Vulkan user wouldn't | |
| // do because it is againt meant for Skia testing. | |
| { | |
| PFN_vkGetInstanceProcAddr instProc; | |
| if (!sk_gpu_test::LoadVkLibraryAndGetProcAddrFuncs(&instProc)) { | |
| return 1; | |
| } | |
| memset(features.get(), 0, sizeof(VkPhysicalDeviceFeatures2)); | |
| features->sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FEATURES_2; | |
| features->pNext = nullptr; | |
| // Fill in features you want to enable here | |
| backendContext.fInstance = VK_NULL_HANDLE; | |
| backendContext.fDevice = VK_NULL_HANDLE; | |
| if (!sk_gpu_test::CreateVkBackendContext(instProc, &backendContext, extensions.get(), | |
| features.get(), &debugCallback)) { | |
| return 1; | |
| } | |
| } | |
| auto getProc = backendContext.fGetProc; | |
| PFN_vkDestroyInstance fVkDestroyInstance; | |
| PFN_vkDestroyDebugReportCallbackEXT fVkDestroyDebugReportCallbackEXT = nullptr; | |
| PFN_vkDestroyDevice fVkDestroyDevice; | |
| ACQUIRE_INST_VK_PROC(DestroyInstance); | |
| if (debugCallback != VK_NULL_HANDLE) { | |
| ACQUIRE_INST_VK_PROC(DestroyDebugReportCallbackEXT); | |
| } | |
| ACQUIRE_INST_VK_PROC(DestroyDevice); | |
| // Create a GrDirectContext with our GrVkBackendContext | |
| sk_sp<GrDirectContext> context = GrDirectContext::MakeVulkan(backendContext); | |
| if (!context) { | |
| fVkDestroyDevice(backendContext.fDevice, nullptr); | |
| if (debugCallback != VK_NULL_HANDLE) { | |
| fVkDestroyDebugReportCallbackEXT(backendContext.fInstance, debugCallback, nullptr); | |
| } | |
| fVkDestroyInstance(backendContext.fInstance, nullptr); | |
| return 1; | |
| } | |
| SkImageInfo imageInfo = SkImageInfo::Make(16, 16, kRGBA_8888_SkColorType, kPremul_SkAlphaType); | |
| // Create an SkSurface backed by a Vulkan VkImage. Often clients will be getting VkImages from | |
| // swapchains. In those cases they should use SkSurface::MakeFromBackendTexture or | |
| // SkSurface::MakeFromBackendRenderTarget to wrap those premade VkImages in Skia. See the | |
| // HelloWorld example app to see how this is done. | |
| sk_sp<SkSurface> surface = | |
| SkSurface::MakeRenderTarget(context.get(), skgpu::Budgeted::kYes, imageInfo); | |
| if (!surface) { | |
| context.reset(); | |
| fVkDestroyDevice(backendContext.fDevice, nullptr); | |
| if (debugCallback != VK_NULL_HANDLE) { | |
| fVkDestroyDebugReportCallbackEXT(backendContext.fInstance, debugCallback, nullptr); | |
| } fVkDestroyInstance(backendContext.fInstance, nullptr); | |
| return 1; | |
| } | |
| surface->getCanvas()->clear(SK_ColorRED); | |
| // After drawing to our surface, we must first flush the recorded work (i.e. convert all our | |
| // recorded SkCanvas calls into a VkCommandBuffer). Then we call submit to submit our | |
| // VkCommandBuffers to the gpu queue. | |
| surface->flush(); | |
| context->submit(); | |
| surface.reset(); | |
| context.reset(); | |
| // Skia doesn't own the VkDevice or VkInstance so the client must manage their lifetime. The | |
| // client must not delete these objects until cleaning up all Skia objects that may have used | |
| // them first. | |
| fVkDestroyDevice(backendContext.fDevice, nullptr); | |
| if (debugCallback != VK_NULL_HANDLE) { | |
| fVkDestroyDebugReportCallbackEXT(backendContext.fInstance, debugCallback, nullptr); | |
| } fVkDestroyInstance(backendContext.fInstance, nullptr); | |
| return 0; | |
| } |