godot/drivers/vulkan/vulkan_context.cpp
Rémi Verschelde 98f88a4102
Merge pull request #36585 from disks86/master
Enhanced physical device selection
2020-03-04 13:00:52 +01:00

1540 lines
55 KiB
C++

/*************************************************************************/
/* vulkan_context.cpp */
/*************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* https://godotengine.org */
/*************************************************************************/
/* Copyright (c) 2007-2020 Juan Linietsky, Ariel Manzur. */
/* Copyright (c) 2014-2020 Godot Engine contributors (cf. AUTHORS.md). */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/*************************************************************************/
#include "vulkan_context.h"
#include "core/engine.h"
#include "core/project_settings.h"
#include "core/ustring.h"
#include "core/version.h"
#include "vk_enum_string_helper.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#define ARRAY_SIZE(a) (sizeof(a) / sizeof(a[0]))
#define APP_SHORT_NAME "GodotEngine"
VKAPI_ATTR VkBool32 VKAPI_CALL VulkanContext::_debug_messenger_callback(VkDebugUtilsMessageSeverityFlagBitsEXT messageSeverity,
VkDebugUtilsMessageTypeFlagsEXT messageType,
const VkDebugUtilsMessengerCallbackDataEXT *pCallbackData,
void *pUserData) {
// This error needs to be ignored because the AMD allocator will mix up memory types on IGP processors.
if (strstr(pCallbackData->pMessage, "Mapping an image with layout") != NULL &&
strstr(pCallbackData->pMessage, "can result in undefined behavior if this memory is used by the device") != NULL) {
return VK_FALSE;
}
// This needs to be ignored because Validator is wrong here.
if (strstr(pCallbackData->pMessage, "SPIR-V module not valid: Pointer operand") != NULL &&
strstr(pCallbackData->pMessage, "must be a memory object") != NULL) {
return VK_FALSE;
}
// Workaround for Vulkan-Loader usability bug: https://github.com/KhronosGroup/Vulkan-Loader/issues/262.
if (strstr(pCallbackData->pMessage, "wrong ELF class: ELFCLASS32") != NULL) {
return VK_FALSE;
}
if (pCallbackData->pMessageIdName && strstr(pCallbackData->pMessageIdName, "UNASSIGNED-CoreValidation-DrawState-ClearCmdBeforeDraw") != NULL) {
return VK_FALSE;
}
String severity_string;
switch (messageSeverity) {
case VK_DEBUG_UTILS_MESSAGE_SEVERITY_VERBOSE_BIT_EXT:
severity_string = "VERBOSE : ";
break;
case VK_DEBUG_UTILS_MESSAGE_SEVERITY_INFO_BIT_EXT:
severity_string = "INFO : ";
break;
case VK_DEBUG_UTILS_MESSAGE_SEVERITY_WARNING_BIT_EXT:
severity_string = "WARNING : ";
break;
case VK_DEBUG_UTILS_MESSAGE_SEVERITY_ERROR_BIT_EXT:
severity_string = "ERROR : ";
break;
case VK_DEBUG_UTILS_MESSAGE_SEVERITY_FLAG_BITS_MAX_ENUM_EXT:
break;
}
String type_string;
switch (messageType) {
case (VK_DEBUG_UTILS_MESSAGE_TYPE_GENERAL_BIT_EXT):
type_string = "GENERAL";
break;
case (VK_DEBUG_UTILS_MESSAGE_TYPE_VALIDATION_BIT_EXT):
type_string = "VALIDATION";
break;
case (VK_DEBUG_UTILS_MESSAGE_TYPE_PERFORMANCE_BIT_EXT):
type_string = "PERFORMANCE";
break;
case (VK_DEBUG_UTILS_MESSAGE_TYPE_VALIDATION_BIT_EXT & VK_DEBUG_UTILS_MESSAGE_TYPE_PERFORMANCE_BIT_EXT):
type_string = "VALIDATION|PERFORMANCE";
break;
}
String objects_string;
if (pCallbackData->objectCount > 0) {
objects_string = "\n\tObjects - " + String::num_int64(pCallbackData->objectCount);
for (uint32_t object = 0; object < pCallbackData->objectCount; ++object) {
objects_string +=
"\n\t\tObject[" + String::num_int64(object) + "]" +
" - " + string_VkObjectType(pCallbackData->pObjects[object].objectType) +
", Handle " + String::num_int64(pCallbackData->pObjects[object].objectHandle);
if (NULL != pCallbackData->pObjects[object].pObjectName && strlen(pCallbackData->pObjects[object].pObjectName) > 0) {
objects_string += ", Name \"" + String(pCallbackData->pObjects[object].pObjectName) + "\"";
}
}
}
String labels_string;
if (pCallbackData->cmdBufLabelCount > 0) {
labels_string = "\n\tCommand Buffer Labels - " + String::num_int64(pCallbackData->cmdBufLabelCount);
for (uint32_t cmd_buf_label = 0; cmd_buf_label < pCallbackData->cmdBufLabelCount; ++cmd_buf_label) {
labels_string +=
"\n\t\tLabel[" + String::num_int64(cmd_buf_label) + "]" +
" - " + pCallbackData->pCmdBufLabels[cmd_buf_label].pLabelName +
"{ ";
for (int color_idx = 0; color_idx < 4; ++color_idx) {
labels_string += String::num(pCallbackData->pCmdBufLabels[cmd_buf_label].color[color_idx]);
if (color_idx < 3) {
labels_string += ", ";
}
}
labels_string += " }";
}
}
String error_message(severity_string + type_string +
" - Message Id Number: " + String::num_int64(pCallbackData->messageIdNumber) +
" | Message Id Name: " + pCallbackData->pMessageIdName +
"\n\t" + pCallbackData->pMessage +
objects_string + labels_string);
ERR_PRINT(error_message);
CRASH_COND_MSG(Engine::get_singleton()->is_abort_on_gpu_errors_enabled(),
"Crashing, because abort on GPU errors is enabled.");
return VK_FALSE;
}
VkBool32 VulkanContext::_check_layers(uint32_t check_count, const char **check_names, uint32_t layer_count, VkLayerProperties *layers) {
for (uint32_t i = 0; i < check_count; i++) {
VkBool32 found = 0;
for (uint32_t j = 0; j < layer_count; j++) {
if (!strcmp(check_names[i], layers[j].layerName)) {
found = 1;
break;
}
}
if (!found) {
ERR_PRINT("Cant find layer: " + String(check_names[i]));
return 0;
}
}
return 1;
}
Error VulkanContext::_create_validation_layers() {
VkResult err;
uint32_t instance_layer_count = 0;
uint32_t validation_layer_count = 0;
const char *instance_validation_layers_alt1[] = { "VK_LAYER_LUNARG_standard_validation" };
const char *instance_validation_layers_alt2[] = { "VK_LAYER_GOOGLE_threading", "VK_LAYER_LUNARG_parameter_validation",
"VK_LAYER_LUNARG_object_tracker", "VK_LAYER_LUNARG_core_validation",
"VK_LAYER_GOOGLE_unique_objects" };
VkBool32 validation_found = 0;
err = vkEnumerateInstanceLayerProperties(&instance_layer_count, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
const char **instance_validation_layers = instance_validation_layers_alt1;
if (instance_layer_count > 0) {
VkLayerProperties *instance_layers = (VkLayerProperties *)malloc(sizeof(VkLayerProperties) * instance_layer_count);
err = vkEnumerateInstanceLayerProperties(&instance_layer_count, instance_layers);
if (err) {
free(instance_layers);
ERR_FAIL_V(ERR_CANT_CREATE);
}
validation_found = _check_layers(ARRAY_SIZE(instance_validation_layers_alt1), instance_validation_layers,
instance_layer_count, instance_layers);
if (validation_found) {
enabled_layer_count = ARRAY_SIZE(instance_validation_layers_alt1);
enabled_layers[0] = "VK_LAYER_LUNARG_standard_validation";
validation_layer_count = 1;
} else {
// use alternative set of validation layers
instance_validation_layers = instance_validation_layers_alt2;
enabled_layer_count = ARRAY_SIZE(instance_validation_layers_alt2);
validation_found = _check_layers(ARRAY_SIZE(instance_validation_layers_alt2), instance_validation_layers,
instance_layer_count, instance_layers);
validation_layer_count = ARRAY_SIZE(instance_validation_layers_alt2);
for (uint32_t i = 0; i < validation_layer_count; i++) {
enabled_layers[i] = instance_validation_layers[i];
}
}
free(instance_layers);
}
if (!validation_found) {
return ERR_CANT_CREATE;
}
return OK;
}
Error VulkanContext::_initialize_extensions() {
VkResult err;
uint32_t instance_extension_count = 0;
enabled_extension_count = 0;
enabled_layer_count = 0;
/* Look for instance extensions */
VkBool32 surfaceExtFound = 0;
VkBool32 platformSurfaceExtFound = 0;
memset(extension_names, 0, sizeof(extension_names));
err = vkEnumerateInstanceExtensionProperties(NULL, &instance_extension_count, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
if (instance_extension_count > 0) {
VkExtensionProperties *instance_extensions = (VkExtensionProperties *)malloc(sizeof(VkExtensionProperties) * instance_extension_count);
err = vkEnumerateInstanceExtensionProperties(NULL, &instance_extension_count, instance_extensions);
if (err) {
free(instance_extensions);
ERR_FAIL_V(ERR_CANT_CREATE);
}
for (uint32_t i = 0; i < instance_extension_count; i++) {
if (!strcmp(VK_KHR_SURFACE_EXTENSION_NAME, instance_extensions[i].extensionName)) {
surfaceExtFound = 1;
extension_names[enabled_extension_count++] = VK_KHR_SURFACE_EXTENSION_NAME;
}
if (!strcmp(_get_platform_surface_extension(), instance_extensions[i].extensionName)) {
platformSurfaceExtFound = 1;
extension_names[enabled_extension_count++] = _get_platform_surface_extension();
}
if (!strcmp(VK_EXT_DEBUG_REPORT_EXTENSION_NAME, instance_extensions[i].extensionName)) {
if (use_validation_layers) {
extension_names[enabled_extension_count++] = VK_EXT_DEBUG_REPORT_EXTENSION_NAME;
}
}
if (!strcmp(VK_EXT_DEBUG_UTILS_EXTENSION_NAME, instance_extensions[i].extensionName)) {
if (use_validation_layers) {
extension_names[enabled_extension_count++] = VK_EXT_DEBUG_UTILS_EXTENSION_NAME;
}
}
if (enabled_extension_count >= MAX_EXTENSIONS) {
free(instance_extensions);
ERR_FAIL_V_MSG(ERR_BUG, "Enabled extension count reaches MAX_EXTENSIONS, BUG");
}
}
free(instance_extensions);
}
ERR_FAIL_COND_V_MSG(!surfaceExtFound, ERR_CANT_CREATE, "No surface extension found, is a driver installed?");
ERR_FAIL_COND_V_MSG(!platformSurfaceExtFound, ERR_CANT_CREATE, "No platform surface extension found, is a driver installed?");
return OK;
}
Error VulkanContext::_create_physical_device() {
/* Look for validation layers */
if (use_validation_layers) {
_create_validation_layers();
}
{
Error err = _initialize_extensions();
if (err != OK) {
return err;
}
}
CharString cs = ProjectSettings::get_singleton()->get("application/config/name").operator String().utf8();
String name = "GodotEngine " + String(VERSION_FULL_NAME);
CharString namecs = name.utf8();
const VkApplicationInfo app = {
/*sType*/ VK_STRUCTURE_TYPE_APPLICATION_INFO,
/*pNext*/ NULL,
/*pApplicationName*/ cs.get_data(),
/*applicationVersion*/ 0,
/*pEngineName*/ namecs.get_data(),
/*engineVersion*/ 0,
/*apiVersion*/ VK_API_VERSION_1_0,
};
VkInstanceCreateInfo inst_info = {
/*sType*/ VK_STRUCTURE_TYPE_INSTANCE_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
/*pApplicationInfo*/ &app,
/*enabledLayerCount*/ enabled_layer_count,
/*ppEnabledLayerNames*/ (const char *const *)instance_validation_layers,
/*enabledExtensionCount*/ enabled_extension_count,
/*ppEnabledExtensionNames*/ (const char *const *)extension_names,
};
/*
* This is info for a temp callback to use during CreateInstance.
* After the instance is created, we use the instance-based
* function to register the final callback.
*/
VkDebugUtilsMessengerCreateInfoEXT dbg_messenger_create_info;
if (use_validation_layers) {
// VK_EXT_debug_utils style
dbg_messenger_create_info.sType = VK_STRUCTURE_TYPE_DEBUG_UTILS_MESSENGER_CREATE_INFO_EXT;
dbg_messenger_create_info.pNext = NULL;
dbg_messenger_create_info.flags = 0;
dbg_messenger_create_info.messageSeverity =
VK_DEBUG_UTILS_MESSAGE_SEVERITY_WARNING_BIT_EXT | VK_DEBUG_UTILS_MESSAGE_SEVERITY_ERROR_BIT_EXT;
dbg_messenger_create_info.messageType = VK_DEBUG_UTILS_MESSAGE_TYPE_GENERAL_BIT_EXT |
VK_DEBUG_UTILS_MESSAGE_TYPE_VALIDATION_BIT_EXT |
VK_DEBUG_UTILS_MESSAGE_TYPE_PERFORMANCE_BIT_EXT;
dbg_messenger_create_info.pfnUserCallback = _debug_messenger_callback;
dbg_messenger_create_info.pUserData = this;
inst_info.pNext = &dbg_messenger_create_info;
}
uint32_t gpu_count;
VkResult err = vkCreateInstance(&inst_info, NULL, &inst);
ERR_FAIL_COND_V_MSG(err == VK_ERROR_INCOMPATIBLE_DRIVER, ERR_CANT_CREATE,
"Cannot find a compatible Vulkan installable client driver (ICD).\n\n"
"vkCreateInstance Failure");
ERR_FAIL_COND_V_MSG(err == VK_ERROR_EXTENSION_NOT_PRESENT, ERR_CANT_CREATE,
"Cannot find a specified extension library.\n"
"Make sure your layers path is set appropriately.\n"
"vkCreateInstance Failure");
ERR_FAIL_COND_V_MSG(err, ERR_CANT_CREATE,
"vkCreateInstance failed.\n\n"
"Do you have a compatible Vulkan installable client driver (ICD) installed?\n"
"Please look at the Getting Started guide for additional information.\n"
"vkCreateInstance Failure");
/* Make initial call to query gpu_count, then second call for gpu info*/
err = vkEnumeratePhysicalDevices(inst, &gpu_count, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
ERR_FAIL_COND_V_MSG(gpu_count == 0, ERR_CANT_CREATE,
"vkEnumeratePhysicalDevices reported zero accessible devices.\n\n"
"Do you have a compatible Vulkan installable client driver (ICD) installed?\n"
"vkEnumeratePhysicalDevices Failure");
VkPhysicalDevice *physical_devices = (VkPhysicalDevice *)malloc(sizeof(VkPhysicalDevice) * gpu_count);
err = vkEnumeratePhysicalDevices(inst, &gpu_count, physical_devices);
if (err) {
free(physical_devices);
ERR_FAIL_V(ERR_CANT_CREATE);
}
/*Find the first discrete GPU with the most VRAM.*/
{
print_line("Selecting primary GPU.");
VkPhysicalDeviceProperties device_properties;
VkPhysicalDeviceMemoryProperties memory_properties;
gpu = physical_devices[0];
uint32_t largest_vram_size = 0;
VkPhysicalDeviceType gpu_type = VK_PHYSICAL_DEVICE_TYPE_OTHER;
for (uint32_t i = 0; i < gpu_count; i++) {
vkGetPhysicalDeviceProperties(physical_devices[i], &device_properties);
/*Skip virtual and CPU devices for now.*/
if (device_properties.deviceType > VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU) {
continue;
}
vkGetPhysicalDeviceMemoryProperties(physical_devices[i], &memory_properties);
/*Total all heaps in case of 3GB+1GB configurations and similar.*/
uint32_t memory_size = 0;
for (uint32_t j = 0; j < memory_properties.memoryHeapCount; j++) {
memory_size += memory_properties.memoryHeaps[j].size;
}
if ((device_properties.deviceType >= gpu_type) || (device_properties.deviceType == gpu_type && memory_size > largest_vram_size)) {
gpu = physical_devices[i];
gpu_type = device_properties.deviceType;
largest_vram_size = memory_size;
print_line(device_properties.deviceName);
}
}
}
free(physical_devices);
/* Look for device extensions */
uint32_t device_extension_count = 0;
VkBool32 swapchainExtFound = 0;
enabled_extension_count = 0;
memset(extension_names, 0, sizeof(extension_names));
err = vkEnumerateDeviceExtensionProperties(gpu, NULL, &device_extension_count, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
if (device_extension_count > 0) {
VkExtensionProperties *device_extensions = (VkExtensionProperties *)malloc(sizeof(VkExtensionProperties) * device_extension_count);
err = vkEnumerateDeviceExtensionProperties(gpu, NULL, &device_extension_count, device_extensions);
if (err) {
free(device_extensions);
ERR_FAIL_V(ERR_CANT_CREATE);
}
for (uint32_t i = 0; i < device_extension_count; i++) {
if (!strcmp(VK_KHR_SWAPCHAIN_EXTENSION_NAME, device_extensions[i].extensionName)) {
swapchainExtFound = 1;
extension_names[enabled_extension_count++] = VK_KHR_SWAPCHAIN_EXTENSION_NAME;
}
if (enabled_extension_count >= MAX_EXTENSIONS) {
free(device_extensions);
ERR_FAIL_V_MSG(ERR_BUG, "Enabled extension count reaches MAX_EXTENSIONS, BUG");
}
}
if (VK_KHR_incremental_present_enabled) {
// Even though the user "enabled" the extension via the command
// line, we must make sure that it's enumerated for use with the
// device. Therefore, disable it here, and re-enable it again if
// enumerated.
VK_KHR_incremental_present_enabled = false;
for (uint32_t i = 0; i < device_extension_count; i++) {
if (!strcmp(VK_KHR_INCREMENTAL_PRESENT_EXTENSION_NAME, device_extensions[i].extensionName)) {
extension_names[enabled_extension_count++] = VK_KHR_INCREMENTAL_PRESENT_EXTENSION_NAME;
VK_KHR_incremental_present_enabled = true;
}
if (enabled_extension_count >= MAX_EXTENSIONS) {
free(device_extensions);
ERR_FAIL_V_MSG(ERR_BUG, "Enabled extension count reaches MAX_EXTENSIONS, BUG");
}
}
}
if (VK_GOOGLE_display_timing_enabled) {
// Even though the user "enabled" the extension via the command
// line, we must make sure that it's enumerated for use with the
// device. Therefore, disable it here, and re-enable it again if
// enumerated.
VK_GOOGLE_display_timing_enabled = false;
for (uint32_t i = 0; i < device_extension_count; i++) {
if (!strcmp(VK_GOOGLE_DISPLAY_TIMING_EXTENSION_NAME, device_extensions[i].extensionName)) {
extension_names[enabled_extension_count++] = VK_GOOGLE_DISPLAY_TIMING_EXTENSION_NAME;
VK_GOOGLE_display_timing_enabled = true;
}
if (enabled_extension_count >= MAX_EXTENSIONS) {
free(device_extensions);
ERR_FAIL_V_MSG(ERR_BUG, "Enabled extension count reaches MAX_EXTENSIONS, BUG");
}
}
}
free(device_extensions);
}
ERR_FAIL_COND_V_MSG(!swapchainExtFound, ERR_CANT_CREATE,
"vkEnumerateDeviceExtensionProperties failed to find the " VK_KHR_SWAPCHAIN_EXTENSION_NAME
" extension.\n\nDo you have a compatible Vulkan installable client driver (ICD) installed?\n"
"vkCreateInstance Failure");
if (use_validation_layers) {
// Setup VK_EXT_debug_utils function pointers always (we use them for
// debug labels and names).
CreateDebugUtilsMessengerEXT =
(PFN_vkCreateDebugUtilsMessengerEXT)vkGetInstanceProcAddr(inst, "vkCreateDebugUtilsMessengerEXT");
DestroyDebugUtilsMessengerEXT =
(PFN_vkDestroyDebugUtilsMessengerEXT)vkGetInstanceProcAddr(inst, "vkDestroyDebugUtilsMessengerEXT");
SubmitDebugUtilsMessageEXT =
(PFN_vkSubmitDebugUtilsMessageEXT)vkGetInstanceProcAddr(inst, "vkSubmitDebugUtilsMessageEXT");
CmdBeginDebugUtilsLabelEXT =
(PFN_vkCmdBeginDebugUtilsLabelEXT)vkGetInstanceProcAddr(inst, "vkCmdBeginDebugUtilsLabelEXT");
CmdEndDebugUtilsLabelEXT =
(PFN_vkCmdEndDebugUtilsLabelEXT)vkGetInstanceProcAddr(inst, "vkCmdEndDebugUtilsLabelEXT");
CmdInsertDebugUtilsLabelEXT =
(PFN_vkCmdInsertDebugUtilsLabelEXT)vkGetInstanceProcAddr(inst, "vkCmdInsertDebugUtilsLabelEXT");
SetDebugUtilsObjectNameEXT =
(PFN_vkSetDebugUtilsObjectNameEXT)vkGetInstanceProcAddr(inst, "vkSetDebugUtilsObjectNameEXT");
if (NULL == CreateDebugUtilsMessengerEXT || NULL == DestroyDebugUtilsMessengerEXT ||
NULL == SubmitDebugUtilsMessageEXT || NULL == CmdBeginDebugUtilsLabelEXT ||
NULL == CmdEndDebugUtilsLabelEXT || NULL == CmdInsertDebugUtilsLabelEXT ||
NULL == SetDebugUtilsObjectNameEXT) {
ERR_FAIL_V_MSG(ERR_CANT_CREATE,
"GetProcAddr: Failed to init VK_EXT_debug_utils\n"
"GetProcAddr: Failure");
}
err = CreateDebugUtilsMessengerEXT(inst, &dbg_messenger_create_info, NULL, &dbg_messenger);
switch (err) {
case VK_SUCCESS:
break;
case VK_ERROR_OUT_OF_HOST_MEMORY:
ERR_FAIL_V_MSG(ERR_CANT_CREATE,
"CreateDebugUtilsMessengerEXT: out of host memory\n"
"CreateDebugUtilsMessengerEXT Failure");
break;
default:
ERR_FAIL_V_MSG(ERR_CANT_CREATE,
"CreateDebugUtilsMessengerEXT: unknown failure\n"
"CreateDebugUtilsMessengerEXT Failure");
ERR_FAIL_V(ERR_CANT_CREATE);
break;
}
}
vkGetPhysicalDeviceProperties(gpu, &gpu_props);
/* Call with NULL data to get count */
vkGetPhysicalDeviceQueueFamilyProperties(gpu, &queue_family_count, NULL);
ERR_FAIL_COND_V(queue_family_count == 0, ERR_CANT_CREATE);
queue_props = (VkQueueFamilyProperties *)malloc(queue_family_count * sizeof(VkQueueFamilyProperties));
vkGetPhysicalDeviceQueueFamilyProperties(gpu, &queue_family_count, queue_props);
// Query fine-grained feature support for this device.
// If app has specific feature requirements it should check supported
// features based on this query
vkGetPhysicalDeviceFeatures(gpu, &physical_device_features);
#define GET_INSTANCE_PROC_ADDR(inst, entrypoint) \
{ \
fp##entrypoint = (PFN_vk##entrypoint)vkGetInstanceProcAddr(inst, "vk" #entrypoint); \
ERR_FAIL_COND_V_MSG(fp##entrypoint == NULL, ERR_CANT_CREATE, \
"vkGetInstanceProcAddr failed to find vk" #entrypoint); \
}
GET_INSTANCE_PROC_ADDR(inst, GetPhysicalDeviceSurfaceSupportKHR);
GET_INSTANCE_PROC_ADDR(inst, GetPhysicalDeviceSurfaceCapabilitiesKHR);
GET_INSTANCE_PROC_ADDR(inst, GetPhysicalDeviceSurfaceFormatsKHR);
GET_INSTANCE_PROC_ADDR(inst, GetPhysicalDeviceSurfacePresentModesKHR);
GET_INSTANCE_PROC_ADDR(inst, GetSwapchainImagesKHR);
return OK;
}
Error VulkanContext::_create_device() {
VkResult err;
float queue_priorities[1] = { 0.0 };
VkDeviceQueueCreateInfo queues[2];
queues[0].sType = VK_STRUCTURE_TYPE_DEVICE_QUEUE_CREATE_INFO;
queues[0].pNext = NULL;
queues[0].queueFamilyIndex = graphics_queue_family_index;
queues[0].queueCount = 1;
queues[0].pQueuePriorities = queue_priorities;
queues[0].flags = 0;
VkDeviceCreateInfo sdevice = {
/*sType*/ VK_STRUCTURE_TYPE_DEVICE_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
/*queueCreateInfoCount*/ 1,
/*pQueueCreateInfos*/ queues,
/*enabledLayerCount*/ 0,
/*ppEnabledLayerNames*/ NULL,
/*enabledExtensionCount*/ enabled_extension_count,
/*ppEnabledExtensionNames*/ (const char *const *)extension_names,
/*pEnabledFeatures*/ &physical_device_features, // If specific features are required, pass them in here
};
if (separate_present_queue) {
queues[1].sType = VK_STRUCTURE_TYPE_DEVICE_QUEUE_CREATE_INFO;
queues[1].pNext = NULL;
queues[1].queueFamilyIndex = present_queue_family_index;
queues[1].queueCount = 1;
queues[1].pQueuePriorities = queue_priorities;
queues[1].flags = 0;
sdevice.queueCreateInfoCount = 2;
}
err = vkCreateDevice(gpu, &sdevice, NULL, &device);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
return OK;
}
Error VulkanContext::_initialize_queues(VkSurfaceKHR surface) {
// Iterate over each queue to learn whether it supports presenting:
VkBool32 *supportsPresent = (VkBool32 *)malloc(queue_family_count * sizeof(VkBool32));
for (uint32_t i = 0; i < queue_family_count; i++) {
fpGetPhysicalDeviceSurfaceSupportKHR(gpu, i, surface, &supportsPresent[i]);
}
// Search for a graphics and a present queue in the array of queue
// families, try to find one that supports both
uint32_t graphicsQueueFamilyIndex = UINT32_MAX;
uint32_t presentQueueFamilyIndex = UINT32_MAX;
for (uint32_t i = 0; i < queue_family_count; i++) {
if ((queue_props[i].queueFlags & VK_QUEUE_GRAPHICS_BIT) != 0) {
if (graphicsQueueFamilyIndex == UINT32_MAX) {
graphicsQueueFamilyIndex = i;
}
if (supportsPresent[i] == VK_TRUE) {
graphicsQueueFamilyIndex = i;
presentQueueFamilyIndex = i;
break;
}
}
}
if (presentQueueFamilyIndex == UINT32_MAX) {
// If didn't find a queue that supports both graphics and present, then
// find a separate present queue.
for (uint32_t i = 0; i < queue_family_count; ++i) {
if (supportsPresent[i] == VK_TRUE) {
presentQueueFamilyIndex = i;
break;
}
}
}
free(supportsPresent);
// Generate error if could not find both a graphics and a present queue
ERR_FAIL_COND_V_MSG(graphicsQueueFamilyIndex == UINT32_MAX || presentQueueFamilyIndex == UINT32_MAX, ERR_CANT_CREATE,
"Could not find both graphics and present queues\n");
graphics_queue_family_index = graphicsQueueFamilyIndex;
present_queue_family_index = presentQueueFamilyIndex;
separate_present_queue = (graphics_queue_family_index != present_queue_family_index);
_create_device();
static PFN_vkGetDeviceProcAddr g_gdpa = NULL;
#define GET_DEVICE_PROC_ADDR(dev, entrypoint) \
{ \
if (!g_gdpa) g_gdpa = (PFN_vkGetDeviceProcAddr)vkGetInstanceProcAddr(inst, "vkGetDeviceProcAddr"); \
fp##entrypoint = (PFN_vk##entrypoint)g_gdpa(dev, "vk" #entrypoint); \
ERR_FAIL_COND_V_MSG(fp##entrypoint == NULL, ERR_CANT_CREATE, \
"vkGetDeviceProcAddr failed to find vk" #entrypoint); \
}
GET_DEVICE_PROC_ADDR(device, CreateSwapchainKHR);
GET_DEVICE_PROC_ADDR(device, DestroySwapchainKHR);
GET_DEVICE_PROC_ADDR(device, GetSwapchainImagesKHR);
GET_DEVICE_PROC_ADDR(device, AcquireNextImageKHR);
GET_DEVICE_PROC_ADDR(device, QueuePresentKHR);
if (VK_GOOGLE_display_timing_enabled) {
GET_DEVICE_PROC_ADDR(device, GetRefreshCycleDurationGOOGLE);
GET_DEVICE_PROC_ADDR(device, GetPastPresentationTimingGOOGLE);
}
vkGetDeviceQueue(device, graphics_queue_family_index, 0, &graphics_queue);
if (!separate_present_queue) {
present_queue = graphics_queue;
} else {
vkGetDeviceQueue(device, present_queue_family_index, 0, &present_queue);
}
// Get the list of VkFormat's that are supported:
uint32_t formatCount;
VkResult err = fpGetPhysicalDeviceSurfaceFormatsKHR(gpu, surface, &formatCount, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
VkSurfaceFormatKHR *surfFormats = (VkSurfaceFormatKHR *)malloc(formatCount * sizeof(VkSurfaceFormatKHR));
err = fpGetPhysicalDeviceSurfaceFormatsKHR(gpu, surface, &formatCount, surfFormats);
if (err) {
free(surfFormats);
ERR_FAIL_V(ERR_CANT_CREATE);
}
// If the format list includes just one entry of VK_FORMAT_UNDEFINED,
// the surface has no preferred format. Otherwise, at least one
// supported format will be returned.
if (true || (formatCount == 1 && surfFormats[0].format == VK_FORMAT_UNDEFINED)) {
format = VK_FORMAT_B8G8R8A8_UNORM;
} else {
if (formatCount < 1) {
free(surfFormats);
ERR_FAIL_V_MSG(ERR_CANT_CREATE, "formatCount less than 1");
}
format = surfFormats[0].format;
}
color_space = surfFormats[0].colorSpace;
free(surfFormats);
Error serr = _create_semaphores();
if (serr) {
return serr;
}
queues_initialized = true;
return OK;
}
Error VulkanContext::_create_semaphores() {
VkResult err;
// Create semaphores to synchronize acquiring presentable buffers before
// rendering and waiting for drawing to be complete before presenting
VkSemaphoreCreateInfo semaphoreCreateInfo = {
/*sType*/ VK_STRUCTURE_TYPE_SEMAPHORE_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
};
// Create fences that we can use to throttle if we get too far
// ahead of the image presents
VkFenceCreateInfo fence_ci = {
/*sType*/ VK_STRUCTURE_TYPE_FENCE_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ VK_FENCE_CREATE_SIGNALED_BIT
};
for (uint32_t i = 0; i < FRAME_LAG; i++) {
err = vkCreateFence(device, &fence_ci, NULL, &fences[i]);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
err = vkCreateSemaphore(device, &semaphoreCreateInfo, NULL, &image_acquired_semaphores[i]);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
err = vkCreateSemaphore(device, &semaphoreCreateInfo, NULL, &draw_complete_semaphores[i]);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
if (separate_present_queue) {
err = vkCreateSemaphore(device, &semaphoreCreateInfo, NULL, &image_ownership_semaphores[i]);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
}
}
frame_index = 0;
// Get Memory information and properties
vkGetPhysicalDeviceMemoryProperties(gpu, &memory_properties);
return OK;
}
int VulkanContext::_window_create(VkSurfaceKHR p_surface, int p_width, int p_height) {
if (!queues_initialized) {
// We use a single GPU, but we need a surface to initialize the
// queues, so this process must be deferred until a surface
// is created.
_initialize_queues(p_surface);
}
Window window;
window.surface = p_surface;
window.width = p_width;
window.height = p_height;
Error err = _update_swap_chain(&window);
ERR_FAIL_COND_V(err != OK, -1);
int id = last_window_id;
windows[id] = window;
last_window_id++;
return id;
}
void VulkanContext::window_resize(int p_window, int p_width, int p_height) {
ERR_FAIL_COND(!windows.has(p_window));
windows[p_window].width = p_width;
windows[p_window].height = p_height;
_update_swap_chain(&windows[p_window]);
}
int VulkanContext::window_get_width(int p_window) {
ERR_FAIL_COND_V(!windows.has(p_window), -1);
return windows[p_window].width;
}
int VulkanContext::window_get_height(int p_window) {
ERR_FAIL_COND_V(!windows.has(p_window), -1);
return windows[p_window].height;
}
VkRenderPass VulkanContext::window_get_render_pass(int p_window) {
ERR_FAIL_COND_V(!windows.has(p_window), VK_NULL_HANDLE);
Window *w = &windows[p_window];
//vulkan use of currentbuffer
return w->render_pass;
}
VkFramebuffer VulkanContext::window_get_framebuffer(int p_window) {
ERR_FAIL_COND_V(!windows.has(p_window), VK_NULL_HANDLE);
ERR_FAIL_COND_V(!buffers_prepared, VK_NULL_HANDLE);
Window *w = &windows[p_window];
//vulkan use of currentbuffer
return w->swapchain_image_resources[w->current_buffer].framebuffer;
}
void VulkanContext::window_destroy(int p_window_id) {
ERR_FAIL_COND(!windows.has(p_window_id));
_clean_up_swap_chain(&windows[p_window_id]);
vkDestroySurfaceKHR(inst, windows[p_window_id].surface, NULL);
windows.erase(p_window_id);
}
Error VulkanContext::_clean_up_swap_chain(Window *window) {
if (!window->swapchain) {
return OK;
}
vkDeviceWaitIdle(device);
//this destroys images associated it seems
fpDestroySwapchainKHR(device, window->swapchain, NULL);
window->swapchain = VK_NULL_HANDLE;
vkDestroyRenderPass(device, window->render_pass, NULL);
if (window->swapchain_image_resources) {
for (uint32_t i = 0; i < swapchainImageCount; i++) {
vkDestroyImageView(device, window->swapchain_image_resources[i].view, NULL);
vkDestroyFramebuffer(device, window->swapchain_image_resources[i].framebuffer, NULL);
}
free(window->swapchain_image_resources);
window->swapchain_image_resources = NULL;
}
if (separate_present_queue) {
vkDestroyCommandPool(device, window->present_cmd_pool, NULL);
}
return OK;
}
Error VulkanContext::_update_swap_chain(Window *window) {
VkResult err;
if (window->swapchain) {
_clean_up_swap_chain(window);
}
// Check the surface capabilities and formats
VkSurfaceCapabilitiesKHR surfCapabilities;
err = fpGetPhysicalDeviceSurfaceCapabilitiesKHR(gpu, window->surface, &surfCapabilities);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
uint32_t presentModeCount;
err = fpGetPhysicalDeviceSurfacePresentModesKHR(gpu, window->surface, &presentModeCount, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
VkPresentModeKHR *presentModes = (VkPresentModeKHR *)malloc(presentModeCount * sizeof(VkPresentModeKHR));
ERR_FAIL_COND_V(!presentModes, ERR_CANT_CREATE);
err = fpGetPhysicalDeviceSurfacePresentModesKHR(gpu, window->surface, &presentModeCount, presentModes);
if (err) {
free(presentModes);
ERR_FAIL_V(ERR_CANT_CREATE);
}
VkExtent2D swapchainExtent;
// width and height are either both 0xFFFFFFFF, or both not 0xFFFFFFFF.
if (surfCapabilities.currentExtent.width == 0xFFFFFFFF) {
// If the surface size is undefined, the size is set to the size
// of the images requested, which must fit within the minimum and
// maximum values.
swapchainExtent.width = window->width;
swapchainExtent.height = window->height;
if (swapchainExtent.width < surfCapabilities.minImageExtent.width) {
swapchainExtent.width = surfCapabilities.minImageExtent.width;
} else if (swapchainExtent.width > surfCapabilities.maxImageExtent.width) {
swapchainExtent.width = surfCapabilities.maxImageExtent.width;
}
if (swapchainExtent.height < surfCapabilities.minImageExtent.height) {
swapchainExtent.height = surfCapabilities.minImageExtent.height;
} else if (swapchainExtent.height > surfCapabilities.maxImageExtent.height) {
swapchainExtent.height = surfCapabilities.maxImageExtent.height;
}
} else {
// If the surface size is defined, the swap chain size must match
swapchainExtent = surfCapabilities.currentExtent;
window->width = surfCapabilities.currentExtent.width;
window->height = surfCapabilities.currentExtent.height;
}
if (window->width == 0 || window->height == 0) {
free(presentModes);
//likely window minimized, no swapchain created
return OK;
}
// The FIFO present mode is guaranteed by the spec to be supported
// and to have no tearing. It's a great default present mode to use.
VkPresentModeKHR swapchainPresentMode = VK_PRESENT_MODE_FIFO_KHR;
// There are times when you may wish to use another present mode. The
// following code shows how to select them, and the comments provide some
// reasons you may wish to use them.
//
// It should be noted that Vulkan 1.0 doesn't provide a method for
// synchronizing rendering with the presentation engine's display. There
// is a method provided for throttling rendering with the display, but
// there are some presentation engines for which this method will not work.
// If an application doesn't throttle its rendering, and if it renders much
// faster than the refresh rate of the display, this can waste power on
// mobile devices. That is because power is being spent rendering images
// that may never be seen.
// VK_PRESENT_MODE_IMMEDIATE_KHR is for applications that don't care about
// tearing, or have some way of synchronizing their rendering with the
// display.
// VK_PRESENT_MODE_MAILBOX_KHR may be useful for applications that
// generally render a new presentable image every refresh cycle, but are
// occasionally early. In this case, the application wants the new image
// to be displayed instead of the previously-queued-for-presentation image
// that has not yet been displayed.
// VK_PRESENT_MODE_FIFO_RELAXED_KHR is for applications that generally
// render a new presentable image every refresh cycle, but are occasionally
// late. In this case (perhaps because of stuttering/latency concerns),
// the application wants the late image to be immediately displayed, even
// though that may mean some tearing.
if (window->presentMode != swapchainPresentMode) {
for (size_t i = 0; i < presentModeCount; ++i) {
if (presentModes[i] == window->presentMode) {
swapchainPresentMode = window->presentMode;
break;
}
}
}
free(presentModes);
ERR_FAIL_COND_V_MSG(swapchainPresentMode != window->presentMode, ERR_CANT_CREATE, "Present mode specified is not supported\n");
// Determine the number of VkImages to use in the swap chain.
// Application desires to acquire 3 images at a time for triple
// buffering
uint32_t desiredNumOfSwapchainImages = 3;
if (desiredNumOfSwapchainImages < surfCapabilities.minImageCount) {
desiredNumOfSwapchainImages = surfCapabilities.minImageCount;
}
// If maxImageCount is 0, we can ask for as many images as we want;
// otherwise we're limited to maxImageCount
if ((surfCapabilities.maxImageCount > 0) && (desiredNumOfSwapchainImages > surfCapabilities.maxImageCount)) {
// Application must settle for fewer images than desired:
desiredNumOfSwapchainImages = surfCapabilities.maxImageCount;
}
VkSurfaceTransformFlagsKHR preTransform;
if (surfCapabilities.supportedTransforms & VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR) {
preTransform = VK_SURFACE_TRANSFORM_IDENTITY_BIT_KHR;
} else {
preTransform = surfCapabilities.currentTransform;
}
// Find a supported composite alpha mode - one of these is guaranteed to be set
VkCompositeAlphaFlagBitsKHR compositeAlpha = VK_COMPOSITE_ALPHA_OPAQUE_BIT_KHR;
VkCompositeAlphaFlagBitsKHR compositeAlphaFlags[4] = {
VK_COMPOSITE_ALPHA_OPAQUE_BIT_KHR,
VK_COMPOSITE_ALPHA_PRE_MULTIPLIED_BIT_KHR,
VK_COMPOSITE_ALPHA_POST_MULTIPLIED_BIT_KHR,
VK_COMPOSITE_ALPHA_INHERIT_BIT_KHR,
};
for (uint32_t i = 0; i < ARRAY_SIZE(compositeAlphaFlags); i++) {
if (surfCapabilities.supportedCompositeAlpha & compositeAlphaFlags[i]) {
compositeAlpha = compositeAlphaFlags[i];
break;
}
}
VkSwapchainCreateInfoKHR swapchain_ci = {
/*sType*/ VK_STRUCTURE_TYPE_SWAPCHAIN_CREATE_INFO_KHR,
/*pNext*/ NULL,
/*flags*/ 0,
/*surface*/ window->surface,
/*minImageCount*/ desiredNumOfSwapchainImages,
/*imageFormat*/ format,
/*imageColorSpace*/ color_space,
/*imageExtent*/ {
/*width*/ swapchainExtent.width,
/*height*/ swapchainExtent.height,
},
/*imageArrayLayers*/ 1,
/*imageUsage*/ VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT,
/*imageSharingMode*/ VK_SHARING_MODE_EXCLUSIVE,
/*queueFamilyIndexCount*/ 0,
/*pQueueFamilyIndices*/ NULL,
/*preTransform*/ (VkSurfaceTransformFlagBitsKHR)preTransform,
/*compositeAlpha*/ compositeAlpha,
/*presentMode*/ swapchainPresentMode,
/*clipped*/ true,
/*oldSwapchain*/ NULL,
};
err = fpCreateSwapchainKHR(device, &swapchain_ci, NULL, &window->swapchain);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
uint32_t sp_image_count;
err = fpGetSwapchainImagesKHR(device, window->swapchain, &sp_image_count, NULL);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
if (swapchainImageCount == 0) {
//assign here for the first time.
swapchainImageCount = sp_image_count;
} else {
ERR_FAIL_COND_V(swapchainImageCount != sp_image_count, ERR_BUG);
}
VkImage *swapchainImages = (VkImage *)malloc(swapchainImageCount * sizeof(VkImage));
ERR_FAIL_COND_V(!swapchainImages, ERR_CANT_CREATE);
err = fpGetSwapchainImagesKHR(device, window->swapchain, &swapchainImageCount, swapchainImages);
if (err) {
free(swapchainImages);
ERR_FAIL_V(ERR_CANT_CREATE);
}
window->swapchain_image_resources =
(SwapchainImageResources *)malloc(sizeof(SwapchainImageResources) * swapchainImageCount);
if (!window->swapchain_image_resources) {
free(swapchainImages);
ERR_FAIL_V(ERR_CANT_CREATE);
}
for (uint32_t i = 0; i < swapchainImageCount; i++) {
VkImageViewCreateInfo color_image_view = {
/*sType*/ VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
/*image*/ swapchainImages[i],
/*viewType*/ VK_IMAGE_VIEW_TYPE_2D,
/*format*/ format,
/*components*/ {
/*r*/ VK_COMPONENT_SWIZZLE_R,
/*g*/ VK_COMPONENT_SWIZZLE_G,
/*b*/ VK_COMPONENT_SWIZZLE_B,
/*a*/ VK_COMPONENT_SWIZZLE_A,
},
/*subresourceRange*/ { /*aspectMask*/ VK_IMAGE_ASPECT_COLOR_BIT,
/*baseMipLevel*/ 0,
/*levelCount*/ 1,
/*baseArrayLayer*/ 0,
/*layerCount*/ 1 },
};
window->swapchain_image_resources[i].image = swapchainImages[i];
color_image_view.image = window->swapchain_image_resources[i].image;
err = vkCreateImageView(device, &color_image_view, NULL, &window->swapchain_image_resources[i].view);
if (err) {
free(swapchainImages);
ERR_FAIL_V(ERR_CANT_CREATE);
}
}
free(swapchainImages);
/******** FRAMEBUFFER ************/
{
const VkAttachmentDescription attachment = {
/*flags*/ 0,
/*format*/ format,
/*samples*/ VK_SAMPLE_COUNT_1_BIT,
/*loadOp*/ VK_ATTACHMENT_LOAD_OP_CLEAR,
/*storeOp*/ VK_ATTACHMENT_STORE_OP_STORE,
/*stencilLoadOp*/ VK_ATTACHMENT_LOAD_OP_DONT_CARE,
/*stencilStoreOp*/ VK_ATTACHMENT_STORE_OP_DONT_CARE,
/*initialLayout*/ VK_IMAGE_LAYOUT_UNDEFINED,
/*finalLayout*/ VK_IMAGE_LAYOUT_PRESENT_SRC_KHR,
};
const VkAttachmentReference color_reference = {
/*attachment*/ 0,
/*layout*/ VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL,
};
const VkSubpassDescription subpass = {
/*flags*/ 0,
/*pipelineBindPoint*/ VK_PIPELINE_BIND_POINT_GRAPHICS,
/*inputAttachmentCount*/ 0,
/*pInputAttachments*/ NULL,
/*colorAttachmentCount*/ 1,
/*pColorAttachments*/ &color_reference,
/*pResolveAttachments*/ NULL,
/*pDepthStencilAttachment*/ NULL,
/*preserveAttachmentCount*/ 0,
/*pPreserveAttachments*/ NULL,
};
const VkRenderPassCreateInfo rp_info = {
/*sTyp*/ VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
/*attachmentCount*/ 1,
/*pAttachments*/ &attachment,
/*subpassCount*/ 1,
/*pSubpasses*/ &subpass,
/*dependencyCount*/ 0,
/*pDependencies*/ NULL,
};
err = vkCreateRenderPass(device, &rp_info, NULL, &window->render_pass);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
for (uint32_t i = 0; i < swapchainImageCount; i++) {
const VkFramebufferCreateInfo fb_info = {
/*sType*/ VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
/*renderPass*/ window->render_pass,
/*attachmentCount*/ 1,
/*pAttachments*/ &window->swapchain_image_resources[i].view,
/*width*/ (uint32_t)window->width,
/*height*/ (uint32_t)window->height,
/*layers*/ 1,
};
err = vkCreateFramebuffer(device, &fb_info, NULL, &window->swapchain_image_resources[i].framebuffer);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
}
}
/******** SEPARATE PRESENT QUEUE ************/
if (separate_present_queue) {
const VkCommandPoolCreateInfo present_cmd_pool_info = {
/*sType*/ VK_STRUCTURE_TYPE_COMMAND_POOL_CREATE_INFO,
/*pNext*/ NULL,
/*flags*/ 0,
/*queueFamilyIndex*/ present_queue_family_index,
};
err = vkCreateCommandPool(device, &present_cmd_pool_info, NULL, &window->present_cmd_pool);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
const VkCommandBufferAllocateInfo present_cmd_info = {
/*sType*/ VK_STRUCTURE_TYPE_COMMAND_BUFFER_ALLOCATE_INFO,
/*pNext*/ NULL,
/*commandPool*/ window->present_cmd_pool,
/*level*/ VK_COMMAND_BUFFER_LEVEL_PRIMARY,
/*commandBufferCount*/ 1,
};
for (uint32_t i = 0; i < swapchainImageCount; i++) {
err = vkAllocateCommandBuffers(device, &present_cmd_info,
&window->swapchain_image_resources[i].graphics_to_present_cmd);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
const VkCommandBufferBeginInfo cmd_buf_info = {
/*sType*/ VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO,
/*pNext*/ NULL,
/*flags*/ VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT,
/*pInheritanceInfo*/ NULL,
};
err = vkBeginCommandBuffer(window->swapchain_image_resources[i].graphics_to_present_cmd, &cmd_buf_info);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
VkImageMemoryBarrier image_ownership_barrier = {
/*sType*/ VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
/*pNext*/ NULL,
/*srcAccessMask*/ 0,
/*dstAccessMask*/ VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT,
/*oldLayout*/ VK_IMAGE_LAYOUT_PRESENT_SRC_KHR,
/*newLayout*/ VK_IMAGE_LAYOUT_PRESENT_SRC_KHR,
/*srcQueueFamilyIndex*/ graphics_queue_family_index,
/*dstQueueFamilyIndex*/ present_queue_family_index,
/*image*/ window->swapchain_image_resources[i].image,
/*subresourceRange*/ { VK_IMAGE_ASPECT_COLOR_BIT, 0, 1, 0, 1 }
};
vkCmdPipelineBarrier(window->swapchain_image_resources[i].graphics_to_present_cmd, VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT,
VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT, 0, 0, NULL, 0, NULL, 1, &image_ownership_barrier);
err = vkEndCommandBuffer(window->swapchain_image_resources[i].graphics_to_present_cmd);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
}
}
//reset current buffer
window->current_buffer = 0;
return OK;
}
Error VulkanContext::initialize() {
Error err = _create_physical_device();
if (err) {
return err;
}
return OK;
}
void VulkanContext::set_setup_buffer(const VkCommandBuffer &pCommandBuffer) {
command_buffer_queue.write[0] = pCommandBuffer;
}
void VulkanContext::append_command_buffer(const VkCommandBuffer &pCommandBuffer) {
if (command_buffer_queue.size() <= command_buffer_count) {
command_buffer_queue.resize(command_buffer_count + 1);
}
command_buffer_queue.write[command_buffer_count] = pCommandBuffer;
command_buffer_count++;
}
void VulkanContext::flush(bool p_flush_setup, bool p_flush_pending) {
// ensure everything else pending is executed
vkDeviceWaitIdle(device);
//flush the pending setup buffer
if (p_flush_setup && command_buffer_queue[0]) {
//use a fence to wait for everything done
VkSubmitInfo submit_info;
submit_info.sType = VK_STRUCTURE_TYPE_SUBMIT_INFO;
submit_info.pNext = NULL;
submit_info.pWaitDstStageMask = NULL;
submit_info.waitSemaphoreCount = 0;
submit_info.pWaitSemaphores = NULL;
submit_info.commandBufferCount = 1;
submit_info.pCommandBuffers = command_buffer_queue.ptr();
submit_info.signalSemaphoreCount = 0;
submit_info.pSignalSemaphores = NULL;
VkResult err = vkQueueSubmit(graphics_queue, 1, &submit_info, VK_NULL_HANDLE);
command_buffer_queue.write[0] = NULL;
ERR_FAIL_COND(err);
vkDeviceWaitIdle(device);
}
if (p_flush_pending && command_buffer_count > 1) {
//use a fence to wait for everything done
VkSubmitInfo submit_info;
submit_info.sType = VK_STRUCTURE_TYPE_SUBMIT_INFO;
submit_info.pNext = NULL;
submit_info.pWaitDstStageMask = NULL;
submit_info.waitSemaphoreCount = 0;
submit_info.pWaitSemaphores = NULL;
submit_info.commandBufferCount = command_buffer_count - 1;
submit_info.pCommandBuffers = command_buffer_queue.ptr() + 1;
submit_info.signalSemaphoreCount = 0;
submit_info.pSignalSemaphores = NULL;
VkResult err = vkQueueSubmit(graphics_queue, 1, &submit_info, VK_NULL_HANDLE);
ERR_FAIL_COND(err);
vkDeviceWaitIdle(device);
command_buffer_count = 1;
}
}
Error VulkanContext::prepare_buffers() {
if (!queues_initialized) {
return OK;
}
VkResult err;
// Ensure no more than FRAME_LAG renderings are outstanding
vkWaitForFences(device, 1, &fences[frame_index], VK_TRUE, UINT64_MAX);
vkResetFences(device, 1, &fences[frame_index]);
for (Map<int, Window>::Element *E = windows.front(); E; E = E->next()) {
Window *w = &E->get();
if (w->swapchain == VK_NULL_HANDLE) {
continue;
}
do {
// Get the index of the next available swapchain image:
err =
fpAcquireNextImageKHR(device, w->swapchain, UINT64_MAX,
image_acquired_semaphores[frame_index], VK_NULL_HANDLE, &w->current_buffer);
if (err == VK_ERROR_OUT_OF_DATE_KHR) {
// swapchain is out of date (e.g. the window was resized) and
// must be recreated:
print_line("early out of data");
//resize_notify();
_update_swap_chain(w);
} else if (err == VK_SUBOPTIMAL_KHR) {
print_line("early suboptimal");
// swapchain is not as optimal as it could be, but the platform's
// presentation engine will still present the image correctly.
break;
} else {
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
}
} while (err != VK_SUCCESS);
}
buffers_prepared = true;
return OK;
}
Error VulkanContext::swap_buffers() {
if (!queues_initialized) {
return OK;
}
// print_line("swapbuffers?");
VkResult err;
#if 0
if (VK_GOOGLE_display_timing_enabled) {
// Look at what happened to previous presents, and make appropriate
// adjustments in timing:
DemoUpdateTargetIPD(demo);
// Note: a real application would position its geometry to that it's in
// the correct locatoin for when the next image is presented. It might
// also wait, so that there's less latency between any input and when
// the next image is rendered/presented. This demo program is so
// simple that it doesn't do either of those.
}
#endif
// Wait for the image acquired semaphore to be signaled to ensure
// that the image won't be rendered to until the presentation
// engine has fully released ownership to the application, and it is
// okay to render to the image.
const VkCommandBuffer *commands_ptr = NULL;
uint32_t commands_to_submit = 0;
if (command_buffer_queue[0] == NULL) {
//no setup command, but commands to submit, submit from the first and skip command
if (command_buffer_count > 1) {
commands_ptr = command_buffer_queue.ptr() + 1;
commands_to_submit = command_buffer_count - 1;
}
} else {
commands_ptr = command_buffer_queue.ptr();
commands_to_submit = command_buffer_count;
}
VkPipelineStageFlags pipe_stage_flags;
VkSubmitInfo submit_info;
submit_info.sType = VK_STRUCTURE_TYPE_SUBMIT_INFO;
submit_info.pNext = NULL;
submit_info.pWaitDstStageMask = &pipe_stage_flags;
pipe_stage_flags = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT;
submit_info.waitSemaphoreCount = 1;
submit_info.pWaitSemaphores = &image_acquired_semaphores[frame_index];
submit_info.commandBufferCount = commands_to_submit;
submit_info.pCommandBuffers = commands_ptr;
submit_info.signalSemaphoreCount = 1;
submit_info.pSignalSemaphores = &draw_complete_semaphores[frame_index];
err = vkQueueSubmit(graphics_queue, 1, &submit_info, fences[frame_index]);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
command_buffer_queue.write[0] = NULL;
command_buffer_count = 1;
if (separate_present_queue) {
// If we are using separate queues, change image ownership to the
// present queue before presenting, waiting for the draw complete
// semaphore and signalling the ownership released semaphore when finished
VkFence nullFence = VK_NULL_HANDLE;
pipe_stage_flags = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT;
submit_info.waitSemaphoreCount = 1;
submit_info.pWaitSemaphores = &draw_complete_semaphores[frame_index];
submit_info.commandBufferCount = 0;
VkCommandBuffer *cmdbufptr = (VkCommandBuffer *)alloca(sizeof(VkCommandBuffer *) * windows.size());
submit_info.pCommandBuffers = cmdbufptr;
for (Map<int, Window>::Element *E = windows.front(); E; E = E->next()) {
Window *w = &E->get();
if (w->swapchain == VK_NULL_HANDLE) {
continue;
}
cmdbufptr[submit_info.commandBufferCount] = w->swapchain_image_resources[w->current_buffer].graphics_to_present_cmd;
submit_info.commandBufferCount++;
}
submit_info.signalSemaphoreCount = 1;
submit_info.pSignalSemaphores = &image_ownership_semaphores[frame_index];
err = vkQueueSubmit(present_queue, 1, &submit_info, nullFence);
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
}
// If we are using separate queues we have to wait for image ownership,
// otherwise wait for draw complete
VkPresentInfoKHR present = {
/*sType*/ VK_STRUCTURE_TYPE_PRESENT_INFO_KHR,
/*pNext*/ NULL,
/*waitSemaphoreCount*/ 1,
/*pWaitSemaphores*/ (separate_present_queue) ? &image_ownership_semaphores[frame_index] : &draw_complete_semaphores[frame_index],
/*swapchainCount*/ 0,
/*pSwapchain*/ NULL,
/*pImageIndices*/ NULL,
/*pResults*/ NULL,
};
VkSwapchainKHR *pSwapchains = (VkSwapchainKHR *)alloca(sizeof(VkSwapchainKHR *) * windows.size());
uint32_t *pImageIndices = (uint32_t *)alloca(sizeof(uint32_t *) * windows.size());
present.pSwapchains = pSwapchains;
present.pImageIndices = pImageIndices;
for (Map<int, Window>::Element *E = windows.front(); E; E = E->next()) {
Window *w = &E->get();
if (w->swapchain == VK_NULL_HANDLE) {
continue;
}
pSwapchains[present.swapchainCount] = w->swapchain;
pImageIndices[present.swapchainCount] = w->current_buffer;
present.swapchainCount++;
}
#if 0
if (VK_KHR_incremental_present_enabled) {
// If using VK_KHR_incremental_present, we provide a hint of the region
// that contains changed content relative to the previously-presented
// image. The implementation can use this hint in order to save
// work/power (by only copying the region in the hint). The
// implementation is free to ignore the hint though, and so we must
// ensure that the entire image has the correctly-drawn content.
uint32_t eighthOfWidth = width / 8;
uint32_t eighthOfHeight = height / 8;
VkRectLayerKHR rect = {
/*offset.x*/ eighthOfWidth,
/*offset.y*/ eighthOfHeight,
/*extent.width*/ eighthOfWidth * 6,
/*extent.height*/ eighthOfHeight * 6,
/*layer*/ 0,
};
VkPresentRegionKHR region = {
/*rectangleCount*/ 1,
/*pRectangles*/ &rect,
};
VkPresentRegionsKHR regions = {
/*sType*/ VK_STRUCTURE_TYPE_PRESENT_REGIONS_KHR,
/*pNext*/ present.pNext,
/*swapchainCount*/ present.swapchainCount,
/*pRegions*/ &region,
};
present.pNext = &regions;
}
#endif
#if 0
if (VK_GOOGLE_display_timing_enabled) {
VkPresentTimeGOOGLE ptime;
if (prev_desired_present_time == 0) {
// This must be the first present for this swapchain.
//
// We don't know where we are relative to the presentation engine's
// display's refresh cycle. We also don't know how long rendering
// takes. Let's make a grossly-simplified assumption that the
// desiredPresentTime should be half way between now and
// now+target_IPD. We will adjust over time.
uint64_t curtime = getTimeInNanoseconds();
if (curtime == 0) {
// Since we didn't find out the current time, don't give a
// desiredPresentTime:
ptime.desiredPresentTime = 0;
} else {
ptime.desiredPresentTime = curtime + (target_IPD >> 1);
}
} else {
ptime.desiredPresentTime = (prev_desired_present_time + target_IPD);
}
ptime.presentID = next_present_id++;
prev_desired_present_time = ptime.desiredPresentTime;
VkPresentTimesInfoGOOGLE present_time = {
/*sType*/ VK_STRUCTURE_TYPE_PRESENT_TIMES_INFO_GOOGLE,
/*pNext*/ present.pNext,
/*swapchainCount*/ present.swapchainCount,
/*pTimes*/ &ptime,
};
if (VK_GOOGLE_display_timing_enabled) {
present.pNext = &present_time;
}
}
#endif
static int total_frames = 0;
total_frames++;
// print_line("current buffer: " + itos(current_buffer));
err = fpQueuePresentKHR(present_queue, &present);
frame_index += 1;
frame_index %= FRAME_LAG;
if (err == VK_ERROR_OUT_OF_DATE_KHR) {
// swapchain is out of date (e.g. the window was resized) and
// must be recreated:
print_line("out of date");
resize_notify();
} else if (err == VK_SUBOPTIMAL_KHR) {
// swapchain is not as optimal as it could be, but the platform's
// presentation engine will still present the image correctly.
print_line("suboptimal");
} else {
ERR_FAIL_COND_V(err, ERR_CANT_CREATE);
}
buffers_prepared = false;
return OK;
}
void VulkanContext::resize_notify() {
}
VkDevice VulkanContext::get_device() {
return device;
}
VkPhysicalDevice VulkanContext::get_physical_device() {
return gpu;
}
int VulkanContext::get_swapchain_image_count() const {
return swapchainImageCount;
}
uint32_t VulkanContext::get_graphics_queue() const {
return graphics_queue_family_index;
}
VkFormat VulkanContext::get_screen_format() const {
return format;
}
VkPhysicalDeviceLimits VulkanContext::get_device_limits() const {
return gpu_props.limits;
}
VulkanContext::VulkanContext() {
command_buffer_count = 0;
instance_validation_layers = NULL;
use_validation_layers = true;
VK_KHR_incremental_present_enabled = true;
VK_GOOGLE_display_timing_enabled = true;
command_buffer_queue.resize(1); //first one is the setup command always
command_buffer_queue.write[0] = NULL;
command_buffer_count = 1;
queues_initialized = false;
buffers_prepared = false;
swapchainImageCount = 0;
last_window_id = 0;
}
VulkanContext::~VulkanContext() {
if (queue_props) {
free(queue_props);
}
}