renderer_vulkan: Make unconditional use of VK_KHR_timeline_semaphore

This reworks how host<->device synchronization works on the Vulkan
backend. Instead of "protecting" resources with a fence and signalling
these as free when the fence is known to be signalled by the host GPU,
use timeline semaphores.

Vulkan timeline semaphores allow use to work on a subset of D3D12
fences. As far as we are concerned, timeline semaphores are a value set
by the host or the device that can be waited by either of them.

Taking advantange of this, we can have a monolithically increasing
atomic value for each submission to the graphics queue. Instead of
protecting resources with a fence, we simply store the current logical
tick (the atomic value stored in CPU memory). When we want to know if a
resource is free, it can be compared to the current GPU tick.

This greatly simplifies resource management code and the free status of
resources should have less false negatives.

To workaround bugs in validation layers, when these are attached there's
a thread waiting for timeline semaphores.

1 files changed, 8 insertions, 12 deletions

diff --git a/src/video_core/renderer_vulkan/vk_swapchain.cpp b/src/video_core/renderer_vulkan/vk_swapchain.cpp
index 6bfd2abae..9636a7c65 100644
--- a/src/video_core/renderer_vulkan/vk_swapchain.cpp
+++ b/src/video_core/renderer_vulkan/vk_swapchain.cpp
@@ -12,7 +12,7 @@
 #include "core/core.h"
 #include "core/frontend/framebuffer_layout.h"
 #include "video_core/renderer_vulkan/vk_device.h"
-#include "video_core/renderer_vulkan/vk_resource_manager.h"
+#include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_swapchain.h"
 #include "video_core/renderer_vulkan/wrapper.h"
 
@@ -56,8 +56,8 @@ VkExtent2D ChooseSwapExtent(const VkSurfaceCapabilitiesKHR& capabilities, u32 wi
 
 } // Anonymous namespace
 
-VKSwapchain::VKSwapchain(VkSurfaceKHR surface, const VKDevice& device)
-    : surface{surface}, device{device} {}
+VKSwapchain::VKSwapchain(VkSurfaceKHR surface_, const VKDevice& device_, VKScheduler& scheduler_)
+    : surface{surface_}, device{device_}, scheduler{scheduler_} {}
 
 VKSwapchain::~VKSwapchain() = default;
 
@@ -75,21 +75,18 @@ void VKSwapchain::Create(u32 width, u32 height, bool srgb) {
     CreateSemaphores();
     CreateImageViews();
 
-    fences.resize(image_count, nullptr);
+    resource_ticks.clear();
+    resource_ticks.resize(image_count);
 }
 
 void VKSwapchain::AcquireNextImage() {
     device.GetLogical().AcquireNextImageKHR(*swapchain, std::numeric_limits<u64>::max(),
                                             *present_semaphores[frame_index], {}, &image_index);
 
-    if (auto& fence = fences[image_index]; fence) {
-        fence->Wait();
-        fence->Release();
-        fence = nullptr;
-    }
+    scheduler.Wait(resource_ticks[image_index]);
 }
 
-bool VKSwapchain::Present(VkSemaphore render_semaphore, VKFence& fence) {
+bool VKSwapchain::Present(VkSemaphore render_semaphore) {
     const VkSemaphore present_semaphore{*present_semaphores[frame_index]};
     const std::array<VkSemaphore, 2> semaphores{present_semaphore, render_semaphore};
     const auto present_queue{device.GetPresentQueue()};
@@ -123,8 +120,7 @@ bool VKSwapchain::Present(VkSemaphore render_semaphore, VKFence& fence) {
         break;
     }
 
-    ASSERT(fences[image_index] == nullptr);
-    fences[image_index] = &fence;
+    resource_ticks[image_index] = scheduler.CurrentTick();
     frame_index = (frame_index + 1) % static_cast<u32>(image_count);
     return recreated;
 }