feat: windows on their own clock, deferred deletion for every handle owner, staged re-uploads, and RTPass push data

Window
- updateInterval runs Update on the window's own clock while the
  compositor sends no frame callbacks (hidden, minimized): the Wayland loop
  is a poll() loop, and the Win32 loop sleeps until the next update. Both
  read the interval each pass.
- SetPresentMode (FIFO, Mailbox, Immediate). A mode the surface lacks, or
  whose swapchain has other than numFrames images (Mesa's Mailbox has four),
  falls back to FIFO until the mode is set again.
- AddPollFd / RemovePollFd wake the Wayland loop on the app's descriptors.
- The swapchain has TRANSFER_DST where the surface allows it.
- CaptureNextFrame replaces SaveFrame: it records the copy in the frame,
  before the present, where SaveFrame read an image already presented. A
  frame unwound past before the write still frees the readback buffer.
- Inter-pass barriers on the swapchain image carry the access masks of the
  stages the passes use.

Deferred deletion (issue #101)
- VulkanBuffer, Mesh, VulkanShader, ComputeShader and PipelineRTVulkan hand
  their handles to the deletion queue from their destructors and moves, as
  a frame in flight may still use them; a static owner goes after the
  driver has torn down. VulkanShader and PipelineRTVulkan move and can't be
  copied; re-Load and re-Init queue the old pipeline.
- The queue and its mutex are never destroyed, and the queue takes handles
  from any thread; Device::frameCounter is atomic. Both the queue and the
  staging rings count frames as one window renders them (documented).
- DrainDeletions also lets every buffer's staging ring start over: without
  a Window nothing else reclaims either, so an app without one calls it
  after waiting for its work.

Meshes and buffers
- A Mesh rebuild queues the old acceleration structure and builds into the
  same storage after a barrier, so the BLAS address stays put when it fits.
  The refit and rebuild barriers cover compute ray queries too.
- RTBuildOptions::noDuplicateAnyHit. Mesh::geometryFlags replaces opaque,
  and the refit and rebuild fallbacks keep the flags.
- A re-upload into a buffer a frame in flight may read (UploadDeviceLocal
  of the same size, UploadDeviceLocalRange) copies on the GPU, after those
  reads, through a per-frame staging ring; on ReBAR it used to write the
  buffer in place. Empty uploads record nothing.
- A compressed Mesh::Build that keeps its buffers decompresses into them
  after the reads of frames in flight, and the hit shaders and ray queries
  wait for the decompress as well as the BLAS build. Its vertex buffer
  takes a ranged refit's copies.
- Mesh move-assigns (queueing the BLAS it replaces), so it can be replaced
  in place or erased from a vector.
- VulkanBuffer::Create asks for a device address only for buffers with
  SHADER_DEVICE_ADDRESS usage, as vkGetBufferDeviceAddress requires; the
  others have address 0. UIRenderer::RegisterBuffer refuses those.
- Build inputs are synchronized as shader reads.

Other
- RTPass pushes the app's own push data and traces at an explicit size.
- DescriptorHeapVulkan hands freed slot ranges out again (they leaked), and
  keeps its free lists sorted.
- ImageVulkan's mip chain clamps each level to 1 texel: a non-square
  image's shorter side went to 0.
- VulkanShader throws on a missing file instead of failing in bad_alloc.
- CRAFTER_GRAPHICS_DEVICE picks the GPU by its number among the distinct
  GPUs Vulkan lists; anything but such a number is an error.
- The debug messenger prints info messages only with
  CRAFTER_GRAPHICS_VALIDATION set.
- Windows: the WinRT gamepad code moved to a C file (mingw-w64's WinRT
  headers don't compile as C++), -lcombase dropped, a reference leak fixed.
- project.cpp honours --local for Crafter.Event, Math and Asset, as Asset
  does for Math, and builds every test through one helper, projectArgs
  included.
- Tests: DebugMessengerSeverity, DeletionQueueThreads, DeviceSelection,
  HeapSlotReuse, MeshLifetime, MipChainNonSquare, PipelineLifetime,
  RTPassPushData, SwapchainTransferDst, UploadInFlight, VulkanBufferAddress,
  VulkanShaderLifetime, WindowUpdateClock; others extended.

Breaking: SaveFrame is CaptureNextFrame; Mesh::opaque is geometryFlags; a
buffer without SHADER_DEVICE_ADDRESS usage has no address; frameCounter is
std::atomic.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Jorijn van der Graaf 2026-10-20 19:07:54 +02:00
commit 079b6b0e1f
Signed by: jorijnvdgraaf
GPG key ID: 2937E59CDCC1BCFB
47 changed files with 3572 additions and 1430 deletions

View file

@ -0,0 +1,9 @@
//SPDX-License-Identifier: LGPL-3.0-only
//SPDX-FileCopyrightText: Copyright (C) 2026 Catcrafts®
#version 460
#extension GL_EXT_ray_tracing : require
// A ray-tracing pipeline to create and release; it never runs.
void main() {
}

View file

@ -0,0 +1,206 @@
//SPDX-License-Identifier: LGPL-3.0-only
//SPDX-FileCopyrightText: Copyright (C) 2026 Catcrafts®
// ComputeShader and PipelineRTVulkan release their pipelines through the
// deferred-deletion queue (issue #101), like the other owners of Vulkan
// handles. Both used to destroy the pipeline on the spot, while a frame in
// flight could still be running it, and a static one made a Vulkan call at
// exit. Neither can be copied, as two copies would both release the pipeline;
// moves hand it over, and a move assignment queues what it replaces, a ray-
// tracing pipeline's binding table too. A PipelineRTVulkan that was never
// initialised releases nothing: its handle used to start out uninitialised and
// be destroyed regardless. Loading or initialising one again queues the
// pipeline (and binding table) it had, which used to leak.
//
// Needs a Vulkan device and no window: the test steps Device::frameCounter
// itself. The ray-tracing half skips without ray tracing. The validation check
// only means something with CRAFTER_GRAPHICS_VALIDATION set.
#include <cstdlib>
#include "vulkan/vulkan.h"
import Crafter.Graphics;
import std;
using namespace Crafter;
namespace fs = std::filesystem;
namespace {
int failures = 0;
void Check(bool ok, std::string_view what) {
std::println("{} {}", ok ? "PASS" : "FAIL", what);
if (!ok) ++failures;
}
fs::path ExeDir() {
return fs::read_symlink("/proc/self/exe").parent_path();
}
std::size_t QueuedCount(VkPipeline pipeline) {
return static_cast<std::size_t>(std::ranges::count_if(Device::deletionQueue,
[&](const Device::PendingDeletion& e) { return e.pipeline == pipeline; }));
}
std::size_t QueuedCount(VkBuffer buffer) {
return static_cast<std::size_t>(std::ranges::count_if(Device::deletionQueue,
[&](const Device::PendingDeletion& e) { return e.buffer == buffer; }));
}
std::size_t QueuedPipelines() {
return static_cast<std::size_t>(std::ranges::count_if(Device::deletionQueue,
[](const Device::PendingDeletion& e) { return e.pipeline != VK_NULL_HANDLE; }));
}
// Records into a one-off command buffer and runs it to completion.
void Run(const std::function<void(VkCommandBuffer)>& record) {
const VkCommandBufferAllocateInfo allocation {
.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_ALLOCATE_INFO,
.commandPool = Device::commandPool,
.level = VK_COMMAND_BUFFER_LEVEL_PRIMARY,
.commandBufferCount = 1,
};
VkCommandBuffer cmd = VK_NULL_HANDLE;
Device::CheckVkResult(vkAllocateCommandBuffers(Device::device, &allocation, &cmd));
const VkCommandBufferBeginInfo begin {
.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO,
.flags = VK_COMMAND_BUFFER_USAGE_ONE_TIME_SUBMIT_BIT,
};
Device::CheckVkResult(vkBeginCommandBuffer(cmd, &begin));
record(cmd);
Device::CheckVkResult(vkEndCommandBuffer(cmd));
const VkSubmitInfo submit { .sType = VK_STRUCTURE_TYPE_SUBMIT_INFO, .commandBufferCount = 1, .pCommandBuffers = &cmd };
Device::CheckVkResult(vkQueueSubmit(Device::queue, 1, &submit, VK_NULL_HANDLE));
Device::CheckVkResult(vkQueueWaitIdle(Device::queue));
vkFreeCommandBuffers(Device::device, Device::commandPool, 1, &cmd);
}
} // namespace
static_assert(!std::is_copy_constructible_v<ComputeShader> && !std::is_copy_assignable_v<ComputeShader>,
"two copies of a ComputeShader would both release its pipeline");
static_assert(!std::is_copy_constructible_v<PipelineRTVulkan> && !std::is_copy_assignable_v<PipelineRTVulkan>,
"two copies of a PipelineRTVulkan would both release its pipeline");
static_assert(std::is_nothrow_move_constructible_v<PipelineRTVulkan> && std::is_move_assignable_v<PipelineRTVulkan>,
"a PipelineRTVulkan moves, so it can live in a container");
int main() {
Device::Initialize();
Device::validationErrorCount = 0;
Device::framesInFlight = Window::numFrames;
Device::frameCounter = 0;
Device::DrainDeletions();
// What Load needs in descriptor-set mode; the heap build reads nothing from it.
DescriptorHeapVulkan heap;
heap.Initialize(1, 1, 0);
const fs::path compute = ExeDir() / "noop.comp.spv";
// ── ComputeShader ────────────────────────────────────────────────────────
VkPipeline pipeline = VK_NULL_HANDLE;
{
ComputeShader never;
}
Check(Device::deletionQueue.empty(), "a ComputeShader never loaded releases nothing");
{
ComputeShader shader;
shader.Load(compute, heap);
pipeline = shader.pipeline;
Check(pipeline != VK_NULL_HANDLE, "the compute pipeline was created");
// Load's VulkanShader has queued the shader module, nothing more.
Check(QueuedPipelines() == 0, "a live ComputeShader has queued no pipeline");
}
Check(QueuedCount(pipeline) == 1, "destroying a ComputeShader queues its pipeline");
Device::frameCounter = Window::numFrames;
Device::ReclaimDeletions();
Check(Device::deletionQueue.empty(), "the pipeline is freed on its retire frame");
{
ComputeShader source;
source.Load(compute, heap);
pipeline = source.pipeline;
ComputeShader moved(std::move(source));
Check(source.pipeline == VK_NULL_HANDLE && moved.pipeline == pipeline, "a move constructor hands the compute pipeline over");
ComputeShader target;
target.Load(compute, heap);
const VkPipeline replaced = target.pipeline;
target = std::move(moved);
Check(moved.pipeline == VK_NULL_HANDLE && target.pipeline == pipeline, "a move assignment hands it over");
Check(QueuedCount(replaced) == 1, "and queues the compute pipeline it replaced");
Check(QueuedCount(pipeline) == 0, "the pipeline moved around is not queued while owned");
}
Check(QueuedCount(pipeline) == 1 && QueuedPipelines() == 2, "the compute pipeline is queued once, by its last owner");
Device::DrainDeletions();
{
ComputeShader shader;
shader.Load(compute, heap);
pipeline = shader.pipeline;
shader.Load(compute, heap);
Check(shader.pipeline != pipeline && QueuedCount(pipeline) == 1 && QueuedPipelines() == 1, "loading one again queues the pipeline it had");
}
Device::DrainDeletions();
// ── PipelineRTVulkan ─────────────────────────────────────────────────────
if (!Device::rayTracingSupported) {
std::println("SKIP no Vulkan ray-tracing device: PipelineRTVulkan not checked");
} else {
{
PipelineRTVulkan never;
}
Check(Device::deletionQueue.empty(), "a PipelineRTVulkan never initialised releases nothing");
std::array<VulkanShader, 1> shaders {{ { ExeDir() / "lifetime-raygen.spv", "main", VK_SHADER_STAGE_RAYGEN_BIT_KHR, nullptr } }};
ShaderBindingTableVulkan table;
table.Init(shaders);
std::array<VkRayTracingShaderGroupCreateInfoKHR, 1> raygen {{ {
.sType = VK_STRUCTURE_TYPE_RAY_TRACING_SHADER_GROUP_CREATE_INFO_KHR, .type = VK_RAY_TRACING_SHADER_GROUP_TYPE_GENERAL_KHR,
.generalShader = 0, .closestHitShader = VK_SHADER_UNUSED_KHR, .anyHitShader = VK_SHADER_UNUSED_KHR, .intersectionShader = VK_SHADER_UNUSED_KHR,
} }};
auto build = [&](PipelineRTVulkan& rt) {
Run([&](VkCommandBuffer cmd) { rt.Init(cmd, raygen, {}, {}, table); });
};
{
PipelineRTVulkan rt;
build(rt);
pipeline = rt.pipeline;
Check(pipeline != VK_NULL_HANDLE && QueuedPipelines() == 0, "the ray-tracing pipeline was created, and nothing queued while it lives");
}
Check(QueuedCount(pipeline) == 1, "destroying a PipelineRTVulkan queues its pipeline");
Device::DrainDeletions();
{
PipelineRTVulkan source;
build(source);
pipeline = source.pipeline;
const VkBuffer sourceTable = source.sbtBuffer.buffer;
const VkDeviceAddress raygenAddress = source.raygenRegion.deviceAddress;
PipelineRTVulkan moved(std::move(source));
Check(source.pipeline == VK_NULL_HANDLE && moved.pipeline == pipeline && moved.sbtBuffer.buffer == sourceTable
&& moved.raygenRegion.deviceAddress == raygenAddress,
"a move constructor hands the ray-tracing pipeline and its binding table over");
PipelineRTVulkan target;
build(target);
const VkPipeline replaced = target.pipeline;
const VkBuffer replacedTable = target.sbtBuffer.buffer;
target = std::move(moved);
Check(moved.pipeline == VK_NULL_HANDLE && target.pipeline == pipeline && target.sbtBuffer.buffer == sourceTable,
"a move assignment hands them over");
Check(QueuedCount(replaced) == 1 && QueuedCount(replacedTable) == 1, "and queues the pipeline and binding table it replaced");
Check(QueuedCount(pipeline) == 0 && QueuedCount(sourceTable) == 0, "what was moved around is not queued while owned");
build(target);
Check(target.pipeline != pipeline && QueuedCount(pipeline) == 1 && QueuedCount(sourceTable) == 1,
"initialising one again queues the pipeline and binding table it had");
}
Device::DrainDeletions();
}
if (std::getenv("CRAFTER_GRAPHICS_VALIDATION") != nullptr) {
Check(Device::validationErrorCount == 0,
std::format("no Vulkan validation errors across the run ({} seen)", Device::validationErrorCount));
}
std::println("{}", failures == 0 ? "all checks passed" : "check(s) failed");
return failures == 0 ? EXIT_SUCCESS : EXIT_FAILURE;
}