The-Homebrew-Cloud/Cemu
2
0
Fork 0
mirror of https://github.com/cemu-project/Cemu.git synced 2024-11-29 12:34:17 +01:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'main' into metal

Browse Source
This commit is contained in:
SamoZ256 2024-11-01 15:56:10 +01:00 committed by GitHub
commit ed48fbfd55
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
35 changed files with 485 additions and 243 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CMakeLists.txt
View File

@ -238,7 +238,7 @@ if (ENABLE_CUBEB)
option(BUILD_TOOLS "" OFF)
option(BUNDLE_SPEEX "" OFF)
set(USE_WINMM OFF CACHE BOOL "")
add_subdirectory("dependencies/cubeb" EXCLUDE_FROM_ALL)
add_subdirectory("dependencies/cubeb" EXCLUDE_FROM_ALL SYSTEM)
set_property(TARGET cubeb PROPERTY MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>")
add_library(cubeb::cubeb ALIAS cubeb)
endif()

2
src/Cafe/CafeSystem.cpp
View File

@ -403,7 +403,7 @@ void cemu_initForGame()
// replace any known function signatures with our HLE implementations and patch bugs in the games
GamePatch_scan();
}
LatteGPUState.alwaysDisplayDRC = ActiveSettings::DisplayDRCEnabled();
LatteGPUState.isDRCPrimary = ActiveSettings::DisplayDRCEnabled();
InfoLog_PrintActiveSettings();
Latte_Start();
// check for debugger entrypoint bp

2
src/Cafe/HW/Latte/Core/Latte.h
View File

@ -52,7 +52,7 @@ struct LatteGPUState_t
uint32 gx2InitCalled; // incremented every time GX2Init() is called
// OpenGL control
uint32 glVendor; // GLVENDOR_*
bool alwaysDisplayDRC = false;
bool isDRCPrimary = false;
// temporary (replace with proper solution later)
bool tvBufferUsesSRGB;
bool drcBufferUsesSRGB;

47
src/Cafe/HW/Latte/Core/LatteRenderTarget.cpp
View File

@ -989,8 +989,6 @@ void LatteRenderTarget_copyToBackbuffer(LatteTextureView* textureView, bool isPa
g_renderer->ImguiEnd();
}
bool ctrlTabHotkeyPressed = false;
void LatteRenderTarget_itHLECopyColorBufferToScanBuffer(MPTR colorBufferPtr, uint32 colorBufferWidth, uint32 colorBufferHeight, uint32 colorBufferSliceIndex, uint32 colorBufferFormat, uint32 colorBufferPitch, Latte::E_HWTILEMODE colorBufferTilemode, uint32 colorBufferSwizzle, uint32 renderTarget)
{
cemu_assert_debug(colorBufferSliceIndex == 0); // todo - support for non-zero slice
@ -1000,38 +998,31 @@ void LatteRenderTarget_itHLECopyColorBufferToScanBuffer(MPTR colorBufferPtr, uin
return;
}
auto getVPADScreenActive = [](size_t n) -> std::pair<bool, bool> {
auto controller = InputManager::instance().get_vpad_controller(n);
if (!controller)
return {false,false};
auto pressed = controller->is_screen_active();
auto toggle = controller->is_screen_active_toggle();
return {pressed && !toggle, pressed && toggle};
};
const bool tabPressed = gui_isKeyDown(PlatformKeyCodes::TAB);
const bool ctrlPressed = gui_isKeyDown(PlatformKeyCodes::LCONTROL);
const auto [vpad0Active, vpad0Toggle] = getVPADScreenActive(0);
const auto [vpad1Active, vpad1Toggle] = getVPADScreenActive(1);
bool showDRC = swkbd_hasKeyboardInputHook() == false && tabPressed;
bool& alwaysDisplayDRC = LatteGPUState.alwaysDisplayDRC;
const bool altScreenRequested = (!ctrlPressed && tabPressed) || vpad0Active || vpad1Active;
const bool togglePressed = (ctrlPressed && tabPressed) || vpad0Toggle || vpad1Toggle;
static bool togglePressedLast = false;
if (ctrlPressed && tabPressed)
{
if (ctrlTabHotkeyPressed == false)
{
alwaysDisplayDRC = !alwaysDisplayDRC;
ctrlTabHotkeyPressed = true;
}
}
else
ctrlTabHotkeyPressed = false;
bool& isDRCPrimary = LatteGPUState.isDRCPrimary;
if (alwaysDisplayDRC)
showDRC = !tabPressed;
if(togglePressed && !togglePressedLast)
isDRCPrimary = !isDRCPrimary;
togglePressedLast = togglePressed;
if (!showDRC)
{
auto controller = InputManager::instance().get_vpad_controller(0);
if (controller && controller->is_screen_active())
showDRC = true;
if (!showDRC)
{
controller = InputManager::instance().get_vpad_controller(1);
if (controller && controller->is_screen_active())
showDRC = true;
}
}
bool showDRC = swkbd_hasKeyboardInputHook() == false && (isDRCPrimary ^ altScreenRequested);
if ((renderTarget & RENDER_TARGET_DRC) && g_renderer->IsPadWindowActive())
LatteRenderTarget_copyToBackbuffer(texView, true);

34
src/Cafe/HW/Latte/Renderer/Vulkan/SwapchainInfoVk.cpp
View File

@ -146,8 +146,17 @@ void SwapchainInfoVk::Create()
UnrecoverableError("Failed to create semaphore for swapchain acquire");
}
VkFenceCreateInfo fenceInfo = {};
fenceInfo.sType = VK_STRUCTURE_TYPE_FENCE_CREATE_INFO;
fenceInfo.flags = VK_FENCE_CREATE_SIGNALED_BIT;
result = vkCreateFence(m_logicalDevice, &fenceInfo, nullptr, &m_imageAvailableFence);
if (result != VK_SUCCESS)
UnrecoverableError("Failed to create fence for swapchain");
m_acquireIndex = 0;
hasDefinedSwapchainImage = false;
m_queueDepth = 0;
}
void SwapchainInfoVk::Cleanup()
@ -177,6 +186,12 @@ void SwapchainInfoVk::Cleanup()
m_swapchainFramebuffers.clear();
if (m_imageAvailableFence)
{
WaitAvailableFence();
vkDestroyFence(m_logicalDevice, m_imageAvailableFence, nullptr);
m_imageAvailableFence = nullptr;
}
if (m_swapchain)
{
vkDestroySwapchainKHR(m_logicalDevice, m_swapchain, nullptr);
@ -189,6 +204,18 @@ bool SwapchainInfoVk::IsValid() const
return m_swapchain && !m_acquireSemaphores.empty();
}
void SwapchainInfoVk::WaitAvailableFence()
{
if(m_awaitableFence != VK_NULL_HANDLE)
vkWaitForFences(m_logicalDevice, 1, &m_awaitableFence, VK_TRUE, UINT64_MAX);
m_awaitableFence = VK_NULL_HANDLE;
}
void SwapchainInfoVk::ResetAvailableFence() const
{
vkResetFences(m_logicalDevice, 1, &m_imageAvailableFence);
}
VkSemaphore SwapchainInfoVk::ConsumeAcquireSemaphore()
{
VkSemaphore ret = m_currentSemaphore;
@ -198,8 +225,10 @@ VkSemaphore SwapchainInfoVk::ConsumeAcquireSemaphore()
bool SwapchainInfoVk::AcquireImage()
{
ResetAvailableFence();
VkSemaphore acquireSemaphore = m_acquireSemaphores[m_acquireIndex];
VkResult result = vkAcquireNextImageKHR(m_logicalDevice, m_swapchain, 1'000'000'000, acquireSemaphore, nullptr, &swapchainImageIndex);
VkResult result = vkAcquireNextImageKHR(m_logicalDevice, m_swapchain, 1'000'000'000, acquireSemaphore, m_imageAvailableFence, &swapchainImageIndex);
if (result == VK_ERROR_OUT_OF_DATE_KHR || result == VK_SUBOPTIMAL_KHR)
m_shouldRecreate = true;
if (result == VK_TIMEOUT)
@ -216,6 +245,7 @@ bool SwapchainInfoVk::AcquireImage()
return false;
}
m_currentSemaphore = acquireSemaphore;
m_awaitableFence = m_imageAvailableFence;
m_acquireIndex = (m_acquireIndex + 1) % m_swapchainImages.size();
return true;
@ -319,6 +349,7 @@ VkExtent2D SwapchainInfoVk::ChooseSwapExtent(const VkSurfaceCapabilitiesKHR& cap
VkPresentModeKHR SwapchainInfoVk::ChoosePresentMode(const std::vector<VkPresentModeKHR>& modes)
{
m_maxQueued = 0;
const auto vsyncState = (VSync)GetConfig().vsync.GetValue();
if (vsyncState == VSync::MAILBOX)
{
@ -345,6 +376,7 @@ VkPresentModeKHR SwapchainInfoVk::ChoosePresentMode(const std::vector<VkPresentM
return VK_PRESENT_MODE_FIFO_KHR;
}
m_maxQueued = 1;
return VK_PRESENT_MODE_FIFO_KHR;
}

8
src/Cafe/HW/Latte/Renderer/Vulkan/SwapchainInfoVk.h
View File

@ -26,6 +26,9 @@ struct SwapchainInfoVk
bool IsValid() const;
void WaitAvailableFence();
void ResetAvailableFence() const;
bool AcquireImage();
// retrieve semaphore of last acquire for submitting a wait operation
// only one wait operation must be submitted per acquire (which submits a single signal operation)
@ -68,6 +71,9 @@ struct SwapchainInfoVk
VkSwapchainKHR m_swapchain{};
Vector2i m_desiredExtent{};
uint32 swapchainImageIndex = (uint32)-1;
uint64 m_presentId = 1;
uint64 m_queueDepth = 0; // number of frames with pending presentation requests
uint64 m_maxQueued = 0; // the maximum number of frames with presentation requests.
// swapchain image ringbuffer (indexed by swapchainImageIndex)
@ -81,6 +87,8 @@ struct SwapchainInfoVk
private:
uint32 m_acquireIndex = 0;
std::vector<VkSemaphore> m_acquireSemaphores; // indexed by m_acquireIndex
VkFence m_imageAvailableFence{};
VkFence m_awaitableFence = VK_NULL_HANDLE;
VkSemaphore m_currentSemaphore = VK_NULL_HANDLE;
std::array<uint32, 2> m_swapchainQueueFamilyIndices;

3
src/Cafe/HW/Latte/Renderer/Vulkan/VulkanAPI.h
View File

@ -188,6 +188,9 @@ VKFUNC_DEVICE(vkCmdPipelineBarrier2KHR);
VKFUNC_DEVICE(vkCmdBeginRenderingKHR);
VKFUNC_DEVICE(vkCmdEndRenderingKHR);
// khr_present_wait
VKFUNC_DEVICE(vkWaitForPresentKHR);
// transform feedback extension
VKFUNC_DEVICE(vkCmdBindTransformFeedbackBuffersEXT);
VKFUNC_DEVICE(vkCmdBeginTransformFeedbackEXT);

81
src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.cpp
View File

@ -47,7 +47,9 @@ const std::vector<const char*> kOptionalDeviceExtensions =
VK_EXT_FILTER_CUBIC_EXTENSION_NAME, // not supported by any device yet
VK_EXT_EXTERNAL_MEMORY_HOST_EXTENSION_NAME,
VK_KHR_SYNCHRONIZATION_2_EXTENSION_NAME,
VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME
VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME,
VK_KHR_PRESENT_WAIT_EXTENSION_NAME,
VK_KHR_PRESENT_ID_EXTENSION_NAME
};
const std::vector<const char*> kRequiredDeviceExtensions =
@ -252,12 +254,24 @@ void VulkanRenderer::GetDeviceFeatures()
pcc.pNext = prevStruct;
prevStruct = &pcc;
VkPhysicalDevicePresentIdFeaturesKHR pidf{};
pidf.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PRESENT_ID_FEATURES_KHR;
pidf.pNext = prevStruct;
prevStruct = &pidf;
VkPhysicalDevicePresentWaitFeaturesKHR pwf{};
pwf.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PRESENT_WAIT_FEATURES_KHR;
pwf.pNext = prevStruct;
prevStruct = &pwf;
VkPhysicalDeviceFeatures2 physicalDeviceFeatures2{};
physicalDeviceFeatures2.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FEATURES_2;
physicalDeviceFeatures2.pNext = prevStruct;
vkGetPhysicalDeviceFeatures2(m_physicalDevice, &physicalDeviceFeatures2);
cemuLog_log(LogType::Force, "Vulkan: present_wait extension: {}", (pwf.presentWait && pidf.presentId) ? "supported" : "unsupported");
/* Get Vulkan device properties and limits */
VkPhysicalDeviceFloatControlsPropertiesKHR pfcp{};
prevStruct = nullptr;
@ -490,6 +504,24 @@ VulkanRenderer::VulkanRenderer()
customBorderColorFeature.customBorderColors = VK_TRUE;
customBorderColorFeature.customBorderColorWithoutFormat = VK_TRUE;
}
// enable VK_KHR_present_id
VkPhysicalDevicePresentIdFeaturesKHR presentIdFeature{};
if(m_featureControl.deviceExtensions.present_wait)
{
presentIdFeature.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PRESENT_ID_FEATURES_KHR;
presentIdFeature.pNext = deviceExtensionFeatures;
deviceExtensionFeatures = &presentIdFeature;
presentIdFeature.presentId = VK_TRUE;
}
// enable VK_KHR_present_wait
VkPhysicalDevicePresentWaitFeaturesKHR presentWaitFeature{};
if(m_featureControl.deviceExtensions.present_wait)
{
presentWaitFeature.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PRESENT_WAIT_FEATURES_KHR;
presentWaitFeature.pNext = deviceExtensionFeatures;
deviceExtensionFeatures = &presentWaitFeature;
presentWaitFeature.presentWait = VK_TRUE;
}
std::vector<const char*> used_extensions;
VkDeviceCreateInfo createInfo = CreateDeviceCreateInfo(queueCreateInfos, deviceFeatures, deviceExtensionFeatures, used_extensions);
@ -1047,6 +1079,10 @@ VkDeviceCreateInfo VulkanRenderer::CreateDeviceCreateInfo(const std::vector<VkDe
used_extensions.emplace_back(VK_KHR_DYNAMIC_RENDERING_EXTENSION_NAME);
if (m_featureControl.deviceExtensions.shader_float_controls)
used_extensions.emplace_back(VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME);
if (m_featureControl.deviceExtensions.present_wait)
used_extensions.emplace_back(VK_KHR_PRESENT_ID_EXTENSION_NAME);
if (m_featureControl.deviceExtensions.present_wait)
used_extensions.emplace_back(VK_KHR_PRESENT_WAIT_EXTENSION_NAME);
VkDeviceCreateInfo createInfo{};
createInfo.sType = VK_STRUCTURE_TYPE_DEVICE_CREATE_INFO;
@ -1144,6 +1180,7 @@ bool VulkanRenderer::CheckDeviceExtensionSupport(const VkPhysicalDevice device,
info.deviceExtensions.shader_float_controls = isExtensionAvailable(VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME);
info.deviceExtensions.dynamic_rendering = false; // isExtensionAvailable(VK_KHR_DYNAMIC_RENDERING_EXTENSION_NAME);
// dynamic rendering doesn't provide any benefits for us right now. Driver implementations are very unoptimized as of Feb 2022
info.deviceExtensions.present_wait = isExtensionAvailable(VK_KHR_PRESENT_WAIT_EXTENSION_NAME) && isExtensionAvailable(VK_KHR_PRESENT_ID_EXTENSION_NAME);
// check for framedebuggers
info.debugMarkersSupported = false;
@ -1855,6 +1892,7 @@ void VulkanRenderer::ProcessFinishedCommandBuffers()
if (fenceStatus == VK_SUCCESS)
{
ProcessDestructionQueue();
m_uniformVarBufferReadIndex = m_cmdBufferUniformRingbufIndices[m_commandBufferSyncIndex];
m_commandBufferSyncIndex = (m_commandBufferSyncIndex + 1) % m_commandBuffers.size();
memoryManager->cleanupBuffers(m_countCommandBufferFinished);
m_countCommandBufferFinished++;
@ -1948,6 +1986,7 @@ void VulkanRenderer::SubmitCommandBuffer(VkSemaphore signalSemaphore, VkSemaphor
cemuLog_logDebug(LogType::Force, "Vulkan: Waiting for available command buffer...");
WaitForNextFinishedCommandBuffer();
}
m_cmdBufferUniformRingbufIndices[nextCmdBufferIndex] = m_cmdBufferUniformRingbufIndices[m_commandBufferIndex];
m_commandBufferIndex = nextCmdBufferIndex;
@ -2695,11 +2734,21 @@ void VulkanRenderer::SwapBuffer(bool mainWindow)
ClearColorImageRaw(chainInfo.m_swapchainImages[chainInfo.swapchainImageIndex], 0, 0, clearColor, VK_IMAGE_LAYOUT_UNDEFINED, VK_IMAGE_LAYOUT_PRESENT_SRC_KHR);
}
const size_t currentFrameCmdBufferID = GetCurrentCommandBufferId();
VkSemaphore presentSemaphore = chainInfo.m_presentSemaphores[chainInfo.swapchainImageIndex];
SubmitCommandBuffer(presentSemaphore); // submit all command and signal semaphore
cemu_assert_debug(m_numSubmittedCmdBuffers > 0);
// wait for the previous frame to finish rendering
WaitCommandBufferFinished(m_commandBufferIDOfPrevFrame);
m_commandBufferIDOfPrevFrame = currentFrameCmdBufferID;
chainInfo.WaitAvailableFence();
VkPresentIdKHR presentId = {};
VkPresentInfoKHR presentInfo = {};
presentInfo.sType = VK_STRUCTURE_TYPE_PRESENT_INFO_KHR;
presentInfo.swapchainCount = 1;
@ -2709,6 +2758,24 @@ void VulkanRenderer::SwapBuffer(bool mainWindow)
presentInfo.waitSemaphoreCount = 1;
presentInfo.pWaitSemaphores = &presentSemaphore;
// if present_wait is available and enabled, add frame markers to present requests
// and limit the number of queued present operations
if (m_featureControl.deviceExtensions.present_wait && chainInfo.m_maxQueued > 0)
{
presentId.sType = VK_STRUCTURE_TYPE_PRESENT_ID_KHR;
presentId.swapchainCount = 1;
presentId.pPresentIds = &chainInfo.m_presentId;
presentInfo.pNext = &presentId;
if(chainInfo.m_queueDepth >= chainInfo.m_maxQueued)
{
uint64 waitFrameId = chainInfo.m_presentId - chainInfo.m_queueDepth;
vkWaitForPresentKHR(m_logicalDevice, chainInfo.m_swapchain, waitFrameId, 40'000'000);
chainInfo.m_queueDepth--;
}
}
VkResult result = vkQueuePresentKHR(m_presentQueue, &presentInfo);
if (result < 0 && result != VK_ERROR_OUT_OF_DATE_KHR)
{
@ -2717,6 +2784,12 @@ void VulkanRenderer::SwapBuffer(bool mainWindow)
if(result == VK_ERROR_OUT_OF_DATE_KHR || result == VK_SUBOPTIMAL_KHR)
chainInfo.m_shouldRecreate = true;
if(result >= 0)
{
chainInfo.m_queueDepth++;
chainInfo.m_presentId++;
}
chainInfo.hasDefinedSwapchainImage = false;
chainInfo.swapchainImageIndex = -1;
@ -3491,13 +3564,13 @@ void VulkanRenderer::buffer_bindUniformBuffer(LatteConst::ShaderType shaderType,
switch (shaderType)
{
case LatteConst::ShaderType::Vertex:
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_VERTEX].unformBufferOffset[bufferIndex] = offset;
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_VERTEX].uniformBufferOffset[bufferIndex] = offset;
break;
case LatteConst::ShaderType::Geometry:
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_GEOMETRY].unformBufferOffset[bufferIndex] = offset;
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_GEOMETRY].uniformBufferOffset[bufferIndex] = offset;
break;
case LatteConst::ShaderType::Pixel:
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_FRAGMENT].unformBufferOffset[bufferIndex] = offset;
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_FRAGMENT].uniformBufferOffset[bufferIndex] = offset;
break;
default:
cemu_assert_debug(false);

6
src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h
View File

@ -449,6 +449,7 @@ private:
bool synchronization2 = false; // VK_KHR_synchronization2
bool dynamic_rendering = false; // VK_KHR_dynamic_rendering
bool shader_float_controls = false; // VK_KHR_shader_float_controls
bool present_wait = false; // VK_KHR_present_wait
}deviceExtensions;
struct
@ -591,6 +592,7 @@ private:
bool m_uniformVarBufferMemoryIsCoherent{false};
uint8* m_uniformVarBufferPtr = nullptr;
uint32 m_uniformVarBufferWriteIndex = 0;
uint32 m_uniformVarBufferReadIndex = 0;
// transform feedback ringbuffer
VkBuffer m_xfbRingBuffer = VK_NULL_HANDLE;
@ -636,6 +638,8 @@ private:
size_t m_commandBufferIndex = 0; // current buffer being filled
size_t m_commandBufferSyncIndex = 0; // latest buffer that finished execution (updated on submit)
size_t m_commandBufferIDOfPrevFrame = 0;
std::array<size_t, kCommandBufferPoolSize> m_cmdBufferUniformRingbufIndices {}; // index in the uniform ringbuffer
std::array<VkFence, kCommandBufferPoolSize> m_cmd_buffer_fences;
std::array<VkCommandBuffer, kCommandBufferPoolSize> m_commandBuffers;
std::array<VkSemaphore, kCommandBufferPoolSize> m_commandBufferSemaphores;
@ -658,7 +662,7 @@ private:
uint32 uniformVarBufferOffset[VulkanRendererConst::SHADER_STAGE_INDEX_COUNT];
struct
{
uint32 unformBufferOffset[LATTE_NUM_MAX_UNIFORM_BUFFERS];
uint32 uniformBufferOffset[LATTE_NUM_MAX_UNIFORM_BUFFERS];
}shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_COUNT];
}dynamicOffsetInfo{};

99
src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRendererCore.cpp
View File

@ -375,24 +375,20 @@ float s_vkUniformData[512 * 4];
void VulkanRenderer::uniformData_updateUniformVars(uint32 shaderStageIndex, LatteDecompilerShader* shader)
{
auto GET_UNIFORM_DATA_PTR = [&](size_t index) { return s_vkUniformData + (index / 4); };
auto GET_UNIFORM_DATA_PTR = [](size_t index) { return s_vkUniformData + (index / 4); };
sint32 shaderAluConst;
sint32 shaderUniformRegisterOffset;
switch (shader->shaderType)
{
case LatteConst::ShaderType::Vertex:
shaderAluConst = 0x400;
shaderUniformRegisterOffset = mmSQ_VTX_UNIFORM_BLOCK_START;
break;
case LatteConst::ShaderType::Pixel:
shaderAluConst = 0;
shaderUniformRegisterOffset = mmSQ_PS_UNIFORM_BLOCK_START;
break;
case LatteConst::ShaderType::Geometry:
shaderAluConst = 0; // geometry shader has no ALU const
shaderUniformRegisterOffset = mmSQ_GS_UNIFORM_BLOCK_START;
break;
default:
UNREACHABLE;
@ -445,7 +441,7 @@ void VulkanRenderer::uniformData_updateUniformVars(uint32 shaderStageIndex, Latt
}
if (shader->uniform.loc_verticesPerInstance >= 0)
{
*(int*)(s_vkUniformData + ((size_t)shader->uniform.loc_verticesPerInstance / 4)) = m_streamoutState.verticesPerInstance;
*(int*)GET_UNIFORM_DATA_PTR(shader->uniform.loc_verticesPerInstance) = m_streamoutState.verticesPerInstance;
for (sint32 b = 0; b < LATTE_NUM_STREAMOUT_BUFFER; b++)
{
if (shader->uniform.loc_streamoutBufferBase[b] >= 0)
@ -455,26 +451,63 @@ void VulkanRenderer::uniformData_updateUniformVars(uint32 shaderStageIndex, Latt
}
}
// upload
if ((m_uniformVarBufferWriteIndex + shader->uniform.uniformRangeSize + 1024) > UNIFORMVAR_RINGBUFFER_SIZE)
const uint32 bufferAlignmentM1 = std::max(m_featureControl.limits.minUniformBufferOffsetAlignment, m_featureControl.limits.nonCoherentAtomSize) - 1;
const uint32 uniformSize = (shader->uniform.uniformRangeSize + bufferAlignmentM1) & ~bufferAlignmentM1;
auto waitWhileCondition = [&](std::function<bool()> condition) {
while (condition())
{
if (m_commandBufferSyncIndex == m_commandBufferIndex)
{
if (m_cmdBufferUniformRingbufIndices[m_commandBufferIndex] != m_uniformVarBufferReadIndex)
{
draw_endRenderPass();
SubmitCommandBuffer();
}
else
{
// submitting work would not change readIndex, so there's no way for conditions based on it to change
cemuLog_log(LogType::Force, "draw call overflowed and corrupted uniform ringbuffer. expect visual corruption");
cemu_assert_suspicious();
break;
}
}
WaitForNextFinishedCommandBuffer();
}
};
// wrap around if it doesnt fit consecutively
if (m_uniformVarBufferWriteIndex + uniformSize > UNIFORMVAR_RINGBUFFER_SIZE)
{
waitWhileCondition([&]() {
return m_uniformVarBufferReadIndex > m_uniformVarBufferWriteIndex || m_uniformVarBufferReadIndex == 0;
});
m_uniformVarBufferWriteIndex = 0;
}
uint32 bufferAlignmentM1 = std::max(m_featureControl.limits.minUniformBufferOffsetAlignment, m_featureControl.limits.nonCoherentAtomSize) - 1;
auto ringBufRemaining = [&]() {
ssize_t ringBufferUsedBytes = (ssize_t)m_uniformVarBufferWriteIndex - m_uniformVarBufferReadIndex;
if (ringBufferUsedBytes < 0)
ringBufferUsedBytes += UNIFORMVAR_RINGBUFFER_SIZE;
return UNIFORMVAR_RINGBUFFER_SIZE - 1 - ringBufferUsedBytes;
};
waitWhileCondition([&]() {
return ringBufRemaining() < uniformSize;
});
const uint32 uniformOffset = m_uniformVarBufferWriteIndex;
memcpy(m_uniformVarBufferPtr + uniformOffset, s_vkUniformData, shader->uniform.uniformRangeSize);
m_uniformVarBufferWriteIndex += shader->uniform.uniformRangeSize;
m_uniformVarBufferWriteIndex = (m_uniformVarBufferWriteIndex + bufferAlignmentM1) & ~bufferAlignmentM1;
m_uniformVarBufferWriteIndex += uniformSize;
// update dynamic offset
dynamicOffsetInfo.uniformVarBufferOffset[shaderStageIndex] = uniformOffset;
// flush if not coherent
if (!m_uniformVarBufferMemoryIsCoherent)
{
uint32 nonCoherentAtomSizeM1 = m_featureControl.limits.nonCoherentAtomSize - 1;
VkMappedMemoryRange flushedRange{};
flushedRange.sType = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE;
flushedRange.memory = m_uniformVarBufferMemory;
flushedRange.offset = uniformOffset;
flushedRange.size = (shader->uniform.uniformRangeSize + nonCoherentAtomSizeM1) & ~nonCoherentAtomSizeM1;
flushedRange.size = uniformSize;
vkFlushMappedMemoryRanges(m_logicalDevice, 1, &flushedRange);
}
}
@ -494,7 +527,7 @@ void VulkanRenderer::draw_prepareDynamicOffsetsForDescriptorSet(uint32 shaderSta
{
for (auto& itr : pipeline_info->dynamicOffsetInfo.list_uniformBuffers[shaderStageIndex])
{
dynamicOffsets[numDynOffsets] = dynamicOffsetInfo.shaderUB[shaderStageIndex].unformBufferOffset[itr];
dynamicOffsets[numDynOffsets] = dynamicOffsetInfo.shaderUB[shaderStageIndex].uniformBufferOffset[itr];
numDynOffsets++;
}
}
@ -1357,6 +1390,24 @@ void VulkanRenderer::draw_execute(uint32 baseVertex, uint32 baseInstance, uint32
return;
}
// prepare streamout
m_streamoutState.verticesPerInstance = count;
LatteStreamout_PrepareDrawcall(count, instanceCount);
// update uniform vars
LatteDecompilerShader* vertexShader = LatteSHRC_GetActiveVertexShader();
LatteDecompilerShader* pixelShader = LatteSHRC_GetActivePixelShader();
LatteDecompilerShader* geometryShader = LatteSHRC_GetActiveGeometryShader();
if (vertexShader)
uniformData_updateUniformVars(VulkanRendererConst::SHADER_STAGE_INDEX_VERTEX, vertexShader);
if (pixelShader)
uniformData_updateUniformVars(VulkanRendererConst::SHADER_STAGE_INDEX_FRAGMENT, pixelShader);
if (geometryShader)
uniformData_updateUniformVars(VulkanRendererConst::SHADER_STAGE_INDEX_GEOMETRY, geometryShader);
// store where the read pointer should go after command buffer execution
m_cmdBufferUniformRingbufIndices[m_commandBufferIndex] = m_uniformVarBufferWriteIndex;
// process index data
const LattePrimitiveMode primitiveMode = static_cast<LattePrimitiveMode>(LatteGPUState.contextRegister[mmVGT_PRIMITIVE_TYPE]);
@ -1410,22 +1461,6 @@ void VulkanRenderer::draw_execute(uint32 baseVertex, uint32 baseInstance, uint32
LatteBufferCache_Sync(indexMin + baseVertex, indexMax + baseVertex, baseInstance, instanceCount);
}
// prepare streamout
m_streamoutState.verticesPerInstance = count;
LatteStreamout_PrepareDrawcall(count, instanceCount);
// update uniform vars
LatteDecompilerShader* vertexShader = LatteSHRC_GetActiveVertexShader();
LatteDecompilerShader* pixelShader = LatteSHRC_GetActivePixelShader();
LatteDecompilerShader* geometryShader = LatteSHRC_GetActiveGeometryShader();
if (vertexShader)
uniformData_updateUniformVars(VulkanRendererConst::SHADER_STAGE_INDEX_VERTEX, vertexShader);
if (pixelShader)
uniformData_updateUniformVars(VulkanRendererConst::SHADER_STAGE_INDEX_FRAGMENT, pixelShader);
if (geometryShader)
uniformData_updateUniformVars(VulkanRendererConst::SHADER_STAGE_INDEX_GEOMETRY, geometryShader);
PipelineInfo* pipeline_info;
if (!isFirst)
@ -1613,13 +1648,13 @@ void VulkanRenderer::draw_updateUniformBuffersDirectAccess(LatteDecompilerShader
switch (shaderType)
{
case LatteConst::ShaderType::Vertex:
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_VERTEX].unformBufferOffset[bufferIndex] = physicalAddr - m_importedMemBaseAddress;
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_VERTEX].uniformBufferOffset[bufferIndex] = physicalAddr - m_importedMemBaseAddress;
break;
case LatteConst::ShaderType::Geometry:
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_GEOMETRY].unformBufferOffset[bufferIndex] = physicalAddr - m_importedMemBaseAddress;
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_GEOMETRY].uniformBufferOffset[bufferIndex] = physicalAddr - m_importedMemBaseAddress;
break;
case LatteConst::ShaderType::Pixel:
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_FRAGMENT].unformBufferOffset[bufferIndex] = physicalAddr - m_importedMemBaseAddress;
dynamicOffsetInfo.shaderUB[VulkanRendererConst::SHADER_STAGE_INDEX_FRAGMENT].uniformBufferOffset[bufferIndex] = physicalAddr - m_importedMemBaseAddress;
break;
default:
UNREACHABLE;

4
src/Cafe/IOSU/legacy/iosu_boss.cpp
View File

@ -137,6 +137,10 @@ namespace iosu
this->task_settings.taskType = settings->taskType;
curl = std::shared_ptr<CURL>(curl_easy_init(), curl_easy_cleanup);
if(GetConfig().proxy_server.GetValue() != "")
{
curl_easy_setopt(curl.get(), CURLOPT_PROXY, GetConfig().proxy_server.GetValue().c_str());
}
}
};

4
src/Cafe/OS/libs/nn_olv/nn_olv_OfflineDB.cpp
View File

@ -112,7 +112,7 @@ namespace nn
nnResult _Async_OfflineDB_DownloadPostDataListParam_DownloadPostDataList(coreinit::OSEvent* event, DownloadedTopicData* downloadedTopicData, DownloadedPostData* downloadedPostData, uint32be* postCountOut, uint32 maxCount, DownloadPostDataListParam* param)
{
scope_exit _se([&](){coreinit::OSSignalEvent(event);});
stdx::scope_exit _se([&](){coreinit::OSSignalEvent(event);});
uint64 titleId = CafeSystem::GetForegroundTitleId();
@ -184,7 +184,7 @@ namespace nn
nnResult _Async_OfflineDB_DownloadPostDataListParam_DownloadExternalImageData(coreinit::OSEvent* event, DownloadedDataBase* _this, void* imageDataOut, uint32be* imageSizeOut, uint32 maxSize)
{
scope_exit _se([&](){coreinit::OSSignalEvent(event);});
stdx::scope_exit _se([&](){coreinit::OSSignalEvent(event);});
if (!_this->TestFlags(_this, DownloadedDataBase::FLAGS::HAS_EXTERNAL_IMAGE))
return OLV_RESULT_MISSING_DATA;

6
src/Cafe/OS/libs/nsyshid/Infinity.cpp
View File

@ -1017,11 +1017,7 @@ namespace nsyshid
std::array<uint8, 16> InfinityUSB::GenerateInfinityFigureKey(const std::vector<uint8>& sha1Data)
{
std::array<uint8, 20> digest = {};
SHA_CTX ctx;
SHA1_Init(&ctx);
SHA1_Update(&ctx, sha1Data.data(), sha1Data.size());
SHA1_Final(digest.data(), &ctx);
OPENSSL_cleanse(&ctx, sizeof(ctx));
SHA1(sha1Data.data(), sha1Data.size(), digest.data());
// Infinity AES keys are the first 16 bytes of the SHA1 Digest, every set of 4 bytes need to be
// reversed due to endianness
std::array<uint8, 16> key = {};

2
src/Cafe/OS/libs/ntag/ntag.cpp
View File

@ -509,7 +509,7 @@ namespace ntag
noftHeader->writeCount = _swapEndianU16(_swapEndianU16(noftHeader->writeCount) + 1);
}
memcpy(decryptedBuffer + 0x20, noftHeader, sizeof(noftHeader));
memcpy(decryptedBuffer + 0x20, noftHeader, sizeof(NTAGNoftHeader));
memcpy(decryptedBuffer + _swapEndianU16(rwHeader->offset), data, dataSize);
// Encrypt

8
src/Cafe/OS/libs/snd_core/ax_out.cpp
View File

@ -522,10 +522,10 @@ namespace snd_core
// called periodically to check for AX updates
void AXOut_update()
{
constexpr auto kTimeout = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::milliseconds(((IAudioAPI::kBlockCount * 3) / 4) * (AX_FRAMES_PER_GROUP * 3)));
constexpr auto kWaitDuration = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::milliseconds(3));
constexpr auto kWaitDurationFast = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::microseconds(2900));
constexpr auto kWaitDurationMinimum = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::microseconds(1700));
constexpr static auto kTimeout = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::milliseconds(((IAudioAPI::kBlockCount * 3) / 4) * (AX_FRAMES_PER_GROUP * 3)));
constexpr static auto kWaitDuration = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::milliseconds(3));
constexpr static auto kWaitDurationFast = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::microseconds(2900));
constexpr static auto kWaitDurationMinimum = std::chrono::duration_cast<std::chrono::nanoseconds>(std::chrono::microseconds(1700));
// if we haven't buffered any blocks, we will wait less time than usual
bool additional_blocks_required = false;

4
src/Cemu/Logging/CemuLogging.h
View File

@ -92,7 +92,11 @@ bool cemuLog_log(LogType type, std::basic_string<T> formatStr, TArgs&&... args)
else
{
const auto format_view = fmt::basic_string_view<T>(formatStr);
#if FMT_VERSION >= 110000
const auto text = fmt::vformat(format_view, fmt::make_format_args<fmt::buffered_context<T>>(args...));
#else
const auto text = fmt::vformat(format_view, fmt::make_format_args<fmt::buffer_context<T>>(args...));
#endif
cemuLog_log(type, std::basic_string_view(text.data(), text.size()));
}
return true;

132
src/Common/MemPtr.h
View File

@ -14,25 +14,27 @@ extern uint8* PPCInterpreterGetStackPointer();
extern uint8* PPCInterpreter_PushAndReturnStackPointer(sint32 offset);
extern void PPCInterpreterModifyStackPointer(sint32 offset);
class MEMPTRBase {};
class MEMPTRBase
{
};
template<typename T>
class MEMPTR : MEMPTRBase
{
public:
constexpr MEMPTR()
constexpr MEMPTR() noexcept
: m_value(0) {}
explicit constexpr MEMPTR(uint32 offset)
explicit constexpr MEMPTR(uint32 offset) noexcept
: m_value(offset) {}
explicit constexpr MEMPTR(const uint32be& offset)
explicit constexpr MEMPTR(const uint32be& offset) noexcept
: m_value(offset) {}
constexpr MEMPTR(std::nullptr_t)
constexpr MEMPTR(std::nullptr_t) noexcept
: m_value(0) {}
MEMPTR(T* ptr)
MEMPTR(T* ptr) noexcept
{
if (ptr == nullptr)
m_value = 0;
@ -43,28 +45,23 @@ public:
}
}
constexpr MEMPTR(const MEMPTR& memptr)
: m_value(memptr.m_value) { }
constexpr MEMPTR(const MEMPTR&) noexcept = default;
constexpr MEMPTR& operator=(const MEMPTR& memptr)
{
m_value = memptr.m_value;
return *this;
}
constexpr MEMPTR& operator=(const MEMPTR&) noexcept = default;
constexpr MEMPTR& operator=(const uint32& offset)
constexpr MEMPTR& operator=(const uint32& offset) noexcept
{
m_value = offset;
return *this;
}
constexpr MEMPTR& operator=(const std::nullptr_t rhs)
constexpr MEMPTR& operator=(std::nullptr_t) noexcept
{
m_value = 0;
return *this;
}
MEMPTR& operator=(T* ptr)
MEMPTR& operator=(T* ptr) noexcept
{
if (ptr == nullptr)
m_value = 0;
@ -76,68 +73,108 @@ public:
return *this;
}
bool atomic_compare_exchange(T* comparePtr, T* newPtr)
bool atomic_compare_exchange(T* comparePtr, T* newPtr) noexcept
{
MEMPTR<T> mp_compare = comparePtr;
MEMPTR<T> mp_new = newPtr;
std::atomic<uint32be>* thisValueAtomic = (std::atomic<uint32be>*)&m_value;
auto* thisValueAtomic = reinterpret_cast<std::atomic<uint32be>*>(&m_value);
return thisValueAtomic->compare_exchange_strong(mp_compare.m_value, mp_new.m_value);
}
explicit constexpr operator bool() const noexcept { return m_value != 0; }
constexpr operator T*() const noexcept { return GetPtr(); } // allow implicit cast to wrapped pointer type
explicit constexpr operator bool() const noexcept
{
return m_value != 0;
}
// allow implicit cast to wrapped pointer type
constexpr operator T*() const noexcept
{
return GetPtr();
}
template<typename X>
explicit operator MEMPTR<X>() const { return MEMPTR<X>(this->m_value); }
MEMPTR operator+(const MEMPTR& ptr) { return MEMPTR(this->GetMPTR() + ptr.GetMPTR()); }
MEMPTR operator-(const MEMPTR& ptr) { return MEMPTR(this->GetMPTR() - ptr.GetMPTR()); }
MEMPTR operator+(sint32 v)
explicit operator MEMPTR<X>() const noexcept
{
// pointer arithmetic
return MEMPTR(this->GetMPTR() + v * 4);
return MEMPTR<X>(this->m_value);
}
MEMPTR operator-(sint32 v)
sint32 operator-(const MEMPTR& ptr) noexcept
requires(!std::is_void_v<T>)
{
// pointer arithmetic
return MEMPTR(this->GetMPTR() - v * 4);
return static_cast<sint32>(this->GetMPTR() - ptr.GetMPTR());
}
MEMPTR& operator+=(sint32 v)
MEMPTR operator+(sint32 v) noexcept
requires(!std::is_void_v<T>)
{
// pointer arithmetic
return MEMPTR(this->GetMPTR() + v * sizeof(T));
}
MEMPTR operator-(sint32 v) noexcept
requires(!std::is_void_v<T>)
{
// pointer arithmetic
return MEMPTR(this->GetMPTR() - v * sizeof(T));
}
MEMPTR& operator+=(sint32 v) noexcept
requires(!std::is_void_v<T>)
{
m_value += v * sizeof(T);
return *this;
}
template <class Q = T>
typename std::enable_if<!std::is_same<Q, void>::value, Q>::type&
operator*() const { return *GetPtr(); }
template<typename Q = T>
requires(!std::is_void_v<Q>)
Q& operator*() const noexcept
{
return *GetPtr();
}
T* operator->() const { return GetPtr(); }
constexpr T* operator->() const noexcept
{
return GetPtr();
}
template <class Q = T>
typename std::enable_if<!std::is_same<Q, void>::value, Q>::type&
operator[](int index) { return GetPtr()[index]; }
template<typename Q = T>
requires(!std::is_void_v<Q>)
Q& operator[](int index) noexcept
{
return GetPtr()[index];
}
T* GetPtr() const { return (T*)(m_value == 0 ? nullptr : memory_base + (uint32)m_value); }
T* GetPtr() const noexcept
{
return (T*)(m_value == 0 ? nullptr : memory_base + (uint32)m_value);
}
template<typename C>
C* GetPtr() const { return (C*)(GetPtr()); }
C* GetPtr() const noexcept
{
return static_cast<C*>(GetPtr());
}
constexpr uint32 GetMPTR() const { return m_value.value(); }
constexpr const uint32be& GetBEValue() const { return m_value; }
[[nodiscard]] constexpr uint32 GetMPTR() const noexcept
{
return m_value.value();
}
[[nodiscard]] constexpr const uint32be& GetBEValue() const noexcept
{
return m_value;
}
constexpr bool IsNull() const { return m_value == 0; }
[[nodiscard]] constexpr bool IsNull() const noexcept
{
return m_value == 0;
}
private:
uint32be m_value;
};
static_assert(sizeof(MEMPTR<void*>) == sizeof(uint32be));
static_assert(std::is_trivially_copyable_v<MEMPTR<void*>>);
#include "StackAllocator.h"
#include "SysAllocator.h"
@ -146,5 +183,8 @@ template <typename T>
struct fmt::formatter<MEMPTR<T>> : formatter<string_view>
{
template<typename FormatContext>
auto format(const MEMPTR<T>& v, FormatContext& ctx) const -> format_context::iterator { return fmt::format_to(ctx.out(), "{:#x}", v.GetMPTR()); }
auto format(const MEMPTR<T>& v, FormatContext& ctx) const -> format_context::iterator
{
return fmt::format_to(ctx.out(), "{:#x}", v.GetMPTR());
}
};

48
src/Common/precompiled.h
View File

@ -394,16 +394,10 @@ void vectorRemoveByIndex(std::vector<T>& vec, const size_t index)
vec.erase(vec.begin() + index);
}
template<typename T1, typename T2>
int match_any_of(T1 value, T2 compareTo)
template<typename T1, typename... Types>
bool match_any_of(T1&& value, Types&&... others)
{
return value == compareTo;
}
template<typename T1, typename T2, typename... Types>
bool match_any_of(T1 value, T2 compareTo, Types&&... others)
{
return value == compareTo || match_any_of(value, others...);
return ((value == others) || ...);
}
// we cache the frequency in a static variable
@ -501,13 +495,6 @@ bool future_is_ready(std::future<T>& f)
#endif
}
// replace with std::scope_exit once available
struct scope_exit
{
std::function<void()> f_;
explicit scope_exit(std::function<void()> f) noexcept : f_(std::move(f)) {}
~scope_exit() { if (f_) f_(); }
};
// helper function to cast raw pointers to std::atomic
// this is technically not legal but works on most platforms as long as alignment restrictions are met and the implementation of atomic doesnt come with additional members
@ -515,6 +502,8 @@ struct scope_exit
template<typename T>
std::atomic<T>* _rawPtrToAtomic(T* ptr)
{
static_assert(sizeof(T) == sizeof(std::atomic<T>));
cemu_assert_debug((reinterpret_cast<std::uintptr_t>(ptr) % alignof(std::atomic<T>)) == 0);
return reinterpret_cast<std::atomic<T>*>(ptr);
}
@ -578,13 +567,34 @@ struct fmt::formatter<betype<T>> : fmt::formatter<T>
}
};
// useful C++23 stuff that isn't yet widely supported
// std::to_underlying
// useful future C++ stuff
namespace stdx
{
// std::to_underlying
template <typename EnumT, typename = std::enable_if_t < std::is_enum<EnumT>{} >>
constexpr std::underlying_type_t<EnumT> to_underlying(EnumT e) noexcept {
return static_cast<std::underlying_type_t<EnumT>>(e);
};
// std::scope_exit
template <typename Fn>
class scope_exit
{
Fn m_func;
bool m_released = false;
public:
explicit scope_exit(Fn&& f) noexcept
: m_func(std::forward<Fn>(f))
{}
~scope_exit()
{
if (!m_released) m_func();
}
scope_exit(scope_exit&& other) noexcept
: m_func(std::move(other.m_func)), m_released(std::exchange(other.m_released, true))
{}
scope_exit(const scope_exit&) = delete;
scope_exit& operator=(scope_exit) = delete;
void release() { m_released = true;}
};
}

10
src/config/CemuConfig.h
View File

@ -195,7 +195,7 @@ ENABLE_ENUM_ITERATORS(CrashDump, CrashDump::Disabled, CrashDump::Enabled);
template <>
struct fmt::formatter<PrecompiledShaderOption> : formatter<string_view> {
template <typename FormatContext>
auto format(const PrecompiledShaderOption c, FormatContext &ctx) {
auto format(const PrecompiledShaderOption c, FormatContext &ctx) const {
string_view name;
switch (c)
{
@ -210,7 +210,7 @@ struct fmt::formatter<PrecompiledShaderOption> : formatter<string_view> {
template <>
struct fmt::formatter<AccurateShaderMulOption> : formatter<string_view> {
template <typename FormatContext>
auto format(const AccurateShaderMulOption c, FormatContext &ctx) {
auto format(const AccurateShaderMulOption c, FormatContext &ctx) const {
string_view name;
switch (c)
{
@ -224,7 +224,7 @@ struct fmt::formatter<AccurateShaderMulOption> : formatter<string_view> {
template <>
struct fmt::formatter<CPUMode> : formatter<string_view> {
template <typename FormatContext>
auto format(const CPUMode c, FormatContext &ctx) {
auto format(const CPUMode c, FormatContext &ctx) const {
string_view name;
switch (c)
{
@ -241,7 +241,7 @@ struct fmt::formatter<CPUMode> : formatter<string_view> {
template <>
struct fmt::formatter<CPUModeLegacy> : formatter<string_view> {
template <typename FormatContext>
auto format(const CPUModeLegacy c, FormatContext &ctx) {
auto format(const CPUModeLegacy c, FormatContext &ctx) const {
string_view name;
switch (c)
{
@ -258,7 +258,7 @@ struct fmt::formatter<CPUModeLegacy> : formatter<string_view> {
template <>
struct fmt::formatter<CafeConsoleRegion> : formatter<string_view> {
template <typename FormatContext>
auto format(const CafeConsoleRegion v, FormatContext &ctx) {
auto format(const CafeConsoleRegion v, FormatContext &ctx) const {
string_view name;
switch (v)
{

4
src/config/LaunchSettings.cpp
View File

@ -199,7 +199,11 @@ bool LaunchSettings::HandleCommandline(const std::vector<std::wstring>& args)
std::string errorMsg;
errorMsg.append("Error while trying to parse command line parameter:\n");
errorMsg.append(ex.what());
#if BOOST_OS_WINDOWS
wxMessageBox(errorMsg, "Parameter error", wxICON_ERROR);
#else
std::cout << errorMsg << std::endl;
#endif
return false;
}

3
src/gui/CemuApp.cpp
View File

@ -15,6 +15,9 @@
#if BOOST_OS_LINUX && HAS_WAYLAND
#include "gui/helpers/wxWayland.h"
#endif
#if __WXGTK__
#include <glib.h>
#endif
#include <wx/image.h>
#include <wx/filename.h>

11
src/gui/components/wxGameList.cpp
View File

@ -1392,7 +1392,6 @@ void wxGameList::CreateShortcut(GameInfo2& gameInfo)
const auto outputPath = shortcutDialog.GetPath();
std::optional<fs::path> icon_path = std::nullopt;
[&]()
{
int iconIdx;
int smallIconIdx;
@ -1402,15 +1401,13 @@ void wxGameList::CreateShortcut(GameInfo2& gameInfo)
return;
}
const auto icon = m_image_list->GetIcon(iconIdx);
PWSTR localAppData;
const auto hres = SHGetKnownFolderPath(FOLDERID_LocalAppData, 0, NULL, &localAppData);
wxBitmap bitmap{};
auto folder = fs::path(localAppData) / "Cemu" / "icons";
if (!SUCCEEDED(hres) || (!fs::exists(folder) && !fs::create_directories(folder)))
const auto folder = ActiveSettings::GetUserDataPath("icons");
if (!fs::exists(folder) && !fs::create_directories(folder))
{
cemuLog_log(LogType::Force, "Failed to create icon directory");
return;
}
wxBitmap bitmap{};
if (!bitmap.CopyFromIcon(icon))
{
cemuLog_log(LogType::Force, "Failed to copy icon");
@ -1426,7 +1423,7 @@ void wxGameList::CreateShortcut(GameInfo2& gameInfo)
icon_path = std::nullopt;
cemuLog_log(LogType::Force, "Icon failed to save");
}
}();
}
IShellLinkW* shellLink;
HRESULT hres = CoCreateInstance(CLSID_ShellLink, nullptr, CLSCTX_INPROC_SERVER, IID_IShellLink, reinterpret_cast<LPVOID*>(&shellLink));

2
src/gui/helpers/wxHelpers.h
View File

@ -8,7 +8,7 @@ template <>
struct fmt::formatter<wxString> : formatter<string_view>
{
template <typename FormatContext>
auto format(const wxString& str, FormatContext& ctx)
auto format(const wxString& str, FormatContext& ctx) const
{
return formatter<string_view>::format(str.c_str().AsChar(), ctx);
}

29
src/gui/input/InputAPIAddWindow.cpp
View File

@ -114,6 +114,11 @@ InputAPIAddWindow::InputAPIAddWindow(wxWindow* parent, const wxPoint& position,
this->Bind(wxControllersRefreshed, &InputAPIAddWindow::on_controllers_refreshed, this);
}
InputAPIAddWindow::~InputAPIAddWindow()
{
discard_thread_result();
}
void InputAPIAddWindow::on_add_button(wxCommandEvent& event)
{
const auto selection = m_input_api->GetSelection();
@ -159,6 +164,8 @@ std::unique_ptr<ControllerProviderSettings> InputAPIAddWindow::get_settings() co
void InputAPIAddWindow::on_api_selected(wxCommandEvent& event)
{
discard_thread_result();
if (m_input_api->GetSelection() == wxNOT_FOUND)
return;
@ -239,19 +246,25 @@ void InputAPIAddWindow::on_controller_dropdown(wxCommandEvent& event)
m_controller_list->Append(_("Searching for controllers..."), (wxClientData*)nullptr);
m_controller_list->SetSelection(wxNOT_FOUND);
std::thread([this, provider, selected_uuid]()
m_search_thread_data = std::make_unique<AsyncThreadData>();
std::thread([this, provider, selected_uuid](std::shared_ptr<AsyncThreadData> data)
{
auto available_controllers = provider->get_controllers();
{
std::lock_guard lock{data->mutex};
if(!data->discardResult)
{
wxCommandEvent event(wxControllersRefreshed);
event.SetEventObject(m_controller_list);
event.SetClientObject(new wxCustomData(std::move(available_controllers)));
event.SetInt(provider->api());
event.SetString(selected_uuid);
wxPostEvent(this, event);
m_search_running = false;
}).detach();
}
}
}, m_search_thread_data).detach();
}
void InputAPIAddWindow::on_controller_selected(wxCommandEvent& event)
@ -301,3 +314,13 @@ void InputAPIAddWindow::on_controllers_refreshed(wxCommandEvent& event)
}
}
}
void InputAPIAddWindow::discard_thread_result()
{
m_search_running = false;
if(m_search_thread_data)
{
std::lock_guard lock{m_search_thread_data->mutex};
m_search_thread_data->discardResult = true;
}
}

9
src/gui/input/InputAPIAddWindow.h
View File

@ -19,6 +19,7 @@ class InputAPIAddWindow : public wxDialog
{
public:
InputAPIAddWindow(wxWindow* parent, const wxPoint& position, const std::vector<ControllerPtr>& controllers);
~InputAPIAddWindow();
bool is_valid() const { return m_type.has_value() && m_controller != nullptr; }
InputAPI::Type get_type() const { return m_type.value(); }
@ -38,6 +39,8 @@ private:
void on_controller_selected(wxCommandEvent& event);
void on_controllers_refreshed(wxCommandEvent& event);
void discard_thread_result();
wxChoice* m_input_api;
wxComboBox* m_controller_list;
wxButton* m_ok_button;
@ -50,4 +53,10 @@ private:
std::vector<ControllerPtr> m_controllers;
std::atomic_bool m_search_running = false;
struct AsyncThreadData
{
std::atomic_bool discardResult = false;
std::mutex mutex;
};
std::shared_ptr<AsyncThreadData> m_search_thread_data;
};

24
src/gui/input/panels/VPADInputPanel.cpp
View File

@ -5,6 +5,7 @@
#include <wx/statline.h>
#include <wx/textctrl.h>
#include <wx/slider.h>
#include <wx/checkbox.h>
#include "gui/helpers/wxControlObject.h"
@ -131,11 +132,23 @@ VPADInputPanel::VPADInputPanel(wxWindow* parent)
}
// Blow Mic
row = 9;
row = 8;
add_button_row(main_sizer, row, column, VPADController::kButtonId_Mic, _("blow mic"));
row++;
add_button_row(main_sizer, row, column, VPADController::kButtonId_Screen, _("show screen"));
row++;
auto toggleScreenText = new wxStaticText(this, wxID_ANY, _("toggle screen"));
main_sizer->Add(toggleScreenText,
wxGBPosition(row, column),
wxDefaultSpan,
wxALL | wxALIGN_CENTER_VERTICAL, 5);
m_togglePadViewCheckBox = new wxCheckBox(this, wxID_ANY, {}, wxDefaultPosition, wxDefaultSize, wxTAB_TRAVERSAL);
wxString toggleScreenTT = _("Makes the \"show screen\" button toggle between the TV and gamepad screens");
m_togglePadViewCheckBox->SetToolTip(toggleScreenTT);
toggleScreenText->SetToolTip(toggleScreenTT);
main_sizer->Add(m_togglePadViewCheckBox, wxGBPosition(row,column+1), wxDefaultSpan, wxALL | wxEXPAND, 5);
//////////////////////////////////////////////////////////////////
@ -168,6 +181,8 @@ void VPADInputPanel::on_timer(const EmulatedControllerPtr& emulated_controller,
{
InputPanel::on_timer(emulated_controller, controller_base);
static_cast<VPADController*>(emulated_controller.get())->set_screen_toggle(m_togglePadViewCheckBox->GetValue());
if(emulated_controller)
{
const auto axis = emulated_controller->get_axis();
@ -182,3 +197,10 @@ void VPADInputPanel::OnVolumeChange(wxCommandEvent& event)
{
}
void VPADInputPanel::load_controller(const EmulatedControllerPtr& controller)
{
InputPanel::load_controller(controller);
const bool isToggle = static_cast<VPADController*>(controller.get())->is_screen_active_toggle();
m_togglePadViewCheckBox->SetValue(isToggle);
}

3
src/gui/input/panels/VPADInputPanel.h
View File

@ -4,6 +4,7 @@
#include "gui/input/panels/InputPanel.h"
class wxInputDraw;
class wxCheckBox;
class VPADInputPanel : public InputPanel
{
@ -11,11 +12,13 @@ public:
VPADInputPanel(wxWindow* parent);
void on_timer(const EmulatedControllerPtr& emulated_controller, const ControllerPtr& controller) override;
virtual void load_controller(const EmulatedControllerPtr& controller) override;
private:
void OnVolumeChange(wxCommandEvent& event);
wxInputDraw* m_left_draw, * m_right_draw;
wxCheckBox* m_togglePadViewCheckBox;
void add_button_row(wxGridBagSizer *sizer, sint32 row, sint32 column, const VPADController::ButtonId &button_id);
void add_button_row(wxGridBagSizer *sizer, sint32 row, sint32 column, const VPADController::ButtonId &button_id, const wxString &label);

2
src/input/emulated/EmulatedController.h
View File

@ -127,7 +127,7 @@ using EmulatedControllerPtr = std::shared_ptr<EmulatedController>;
template <>
struct fmt::formatter<EmulatedController::Type> : formatter<string_view> {
template <typename FormatContext>
auto format(EmulatedController::Type v, FormatContext& ctx) {
auto format(EmulatedController::Type v, FormatContext& ctx) const {
switch (v)
{
case EmulatedController::Type::VPAD: return formatter<string_view>::format("Wii U Gamepad", ctx);

11
src/input/emulated/VPADController.cpp
View File

@ -686,3 +686,14 @@ bool VPADController::set_default_mapping(const std::shared_ptr<ControllerBase>&
return mapping_updated;
}
void VPADController::load(const pugi::xml_node& node)
{
if (const auto value = node.child("toggle_display"))
m_screen_active_toggle = ConvertString<bool>(value.child_value());
}
void VPADController::save(pugi::xml_node& node)
{
node.append_child("toggle_display").append_child(pugi::node_pcdata).set_value(fmt::format("{}", (int)m_screen_active_toggle).c_str());
}

6
src/input/emulated/VPADController.h
View File

@ -66,6 +66,8 @@ public:
bool is_mic_active() { return m_mic_active; }
bool is_screen_active() { return m_screen_active; }
bool is_screen_active_toggle() { return m_screen_active_toggle; }
void set_screen_toggle(bool toggle) {m_screen_active_toggle = toggle;}
static std::string_view get_button_name(ButtonId id);
@ -86,9 +88,13 @@ public:
bool set_default_mapping(const std::shared_ptr<ControllerBase>& controller) override;
void load(const pugi::xml_node& node) override;
void save(pugi::xml_node& node) override;
private:
bool m_mic_active = false;
bool m_screen_active = false;
bool m_screen_active_toggle = false;
uint32be m_last_holdvalue = 0;
std::chrono::high_resolution_clock::time_point m_last_hold_change{}, m_last_pulse{};

1
src/util/CMakeLists.txt
View File

@ -50,7 +50,6 @@ add_library(CemuUtil
MemMapper/MemMapper.h
SystemInfo/SystemInfo.cpp
SystemInfo/SystemInfo.h
ThreadPool/ThreadPool.cpp
ThreadPool/ThreadPool.h
tinyxml2/tinyxml2.cpp
tinyxml2/tinyxml2.h

0
src/util/ThreadPool/ThreadPool.cpp
View File

2
src/util/containers/robin_hood.h
View File

@ -194,7 +194,7 @@ namespace robin_hood {
// workaround missing "is_trivially_copyable" in g++ < 5.0
// See https://stackoverflow.com/a/31798726/48181
#if defined(__GNUC__) && __GNUC__ < 5
#if defined(__GNUC__) && __GNUC__ < 5 && !defined(__clang__)
# define ROBIN_HOOD_IS_TRIVIALLY_COPYABLE(...) __has_trivial_copy(__VA_ARGS__)
#else
# define ROBIN_HOOD_IS_TRIVIALLY_COPYABLE(...) std::is_trivially_copyable<__VA_ARGS__>::value

55
src/util/crypto/crc32.cpp
View File

@ -1,29 +1,6 @@
#include "crc32.h"
#if defined(_MSC_VER) || defined(__MINGW32__)
#define __LITTLE_ENDIAN 1234
#define __BIG_ENDIAN 4321
#define __BYTE_ORDER __LITTLE_ENDIAN
#include <xmmintrin.h>
#ifdef __MINGW32__
#define PREFETCH(location) __builtin_prefetch(location)
#else
#define PREFETCH(location) _mm_prefetch(location, _MM_HINT_T0)
#endif
#else
// defines __BYTE_ORDER as __LITTLE_ENDIAN or __BIG_ENDIAN
#include <sys/param.h>
#ifdef __GNUC__
#define PREFETCH(location) __builtin_prefetch(location)
#else
// no prefetching
#define PREFETCH(location) ;
#endif
#endif
unsigned int Crc32Lookup[8][256] =
constexpr uint32 Crc32Lookup[8][256] =
{
{
0x00000000,0x77073096,0xEE0E612C,0x990951BA,0x076DC419,0x706AF48F,0xE963A535,0x9E6495A3,
@ -301,20 +278,7 @@ unsigned int Crc32Lookup[8][256] =
}
};
/// swap endianess
static inline uint32_t swap(uint32_t x)
{
#if defined(__GNUC__) || defined(__clang__)
return __builtin_bswap32(x);
#else
return (x >> 24) |
((x >> 8) & 0x0000FF00) |
((x << 8) & 0x00FF0000) |
(x << 24);
#endif
}
unsigned int crc32_calc_slice_by_8(unsigned int previousCrc32, const void* data, int length)
uint32 crc32_calc_slice_by_8(uint32 previousCrc32, const void* data, size_t length)
{
uint32_t crc = ~previousCrc32; // same as previousCrc32 ^ 0xFFFFFFFF
const uint32_t* current = (const uint32_t*)data;
@ -323,7 +287,7 @@ unsigned int crc32_calc_slice_by_8(unsigned int previousCrc32, const void* data,
while (length >= 8)
{
if constexpr (std::endian::native == std::endian::big){
uint32_t one = *current++ ^ swap(crc);
uint32_t one = *current++ ^ _swapEndianU32(crc);
uint32_t two = *current++;
crc = Crc32Lookup[0][two & 0xFF] ^
Crc32Lookup[1][(two >> 8) & 0xFF] ^
@ -348,13 +312,14 @@ unsigned int crc32_calc_slice_by_8(unsigned int previousCrc32, const void* data,
Crc32Lookup[7][one & 0xFF];
}
else {
cemu_assert(false);
static_assert(std::endian::native == std::endian::big || std::endian::native == std::endian::little,
"Platform byte-order is unsupported");
}
length -= 8;
}
const uint8_t* currentChar = (const uint8_t*)current;
const uint8* currentChar = (const uint8*)current;
// remaining 1 to 7 bytes (standard algorithm)
while (length-- != 0)
crc = (crc >> 8) ^ Crc32Lookup[0][(crc & 0xFF) ^ *currentChar++];
@ -362,20 +327,20 @@ unsigned int crc32_calc_slice_by_8(unsigned int previousCrc32, const void* data,
return ~crc; // same as crc ^ 0xFFFFFFFF
}
unsigned int crc32_calc(unsigned int c, const void* data, int length)
uint32 crc32_calc(uint32 c, const void* data, size_t length)
{
if (length >= 16)
{
return crc32_calc_slice_by_8(c, data, length);
}
unsigned char* p = (unsigned char*)data;
const uint8* p = (const uint8*)data;
if (length == 0)
return c;
c ^= 0xFFFFFFFF;
while (length)
{
unsigned char temp = *p;
temp ^= (unsigned char)c;
uint8 temp = *p;
temp ^= (uint8)c;
c = (c >> 8) ^ Crc32Lookup[0][temp];
// next
length--;

4
src/util/crypto/crc32.h
View File

@ -1,8 +1,8 @@
#pragma once
unsigned int crc32_calc(unsigned int c, const void* data, int length);
uint32 crc32_calc(uint32 c, const void* data, size_t length);
inline unsigned int crc32_calc(const void* data, int length)
inline uint32 crc32_calc(const void* data, size_t length)
{
return crc32_calc(0, data, length);
}