diff options
63 files changed, 989 insertions, 725 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index bcee98a5f..52a1fd492 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -1,5 +1,6 @@ # CMake 3.2 required for cmake to know the right flags for CXX standard on OSX cmake_minimum_required(VERSION 3.2) +set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${PROJECT_SOURCE_DIR}/CMakeModules) function(download_bundled_external remote_path lib_name prefix_var) set(prefix "${CMAKE_BINARY_DIR}/externals/${lib_name}") diff --git a/CMakeModules/CopyCitraQt5Deps.cmake b/CMakeModules/CopyCitraQt5Deps.cmake new file mode 100644 index 000000000..05f58cf9a --- /dev/null +++ b/CMakeModules/CopyCitraQt5Deps.cmake @@ -0,0 +1,17 @@ +function(copy_citra_Qt5_deps target_dir) + include(WindowsCopyFiles) + set(DLL_DEST "${CMAKE_BINARY_DIR}/bin/$<CONFIG>/") + set(Qt5_DLL_DIR "${Qt5_DIR}/../../../bin") + set(Qt5_PLATFORMS_DIR "${Qt5_DIR}/../../../plugins/platforms/") + set(PLATFORMS ${DLL_DEST}platforms/) + windows_copy_files(${target_dir} ${Qt5_DLL_DIR} ${DLL_DEST} + icudt*.dll + icuin*.dll + icuuc*.dll + Qt5Core$<$<CONFIG:Debug>:d>.* + Qt5Gui$<$<CONFIG:Debug>:d>.* + Qt5OpenGL$<$<CONFIG:Debug>:d>.* + Qt5Widgets$<$<CONFIG:Debug>:d>.* + ) + windows_copy_files(citra-qt ${Qt5_PLATFORMS_DIR} ${PLATFORMS} qwindows$<$<CONFIG:Debug>:d>.*) +endfunction(copy_citra_Qt5_deps) diff --git a/CMakeModules/CopyCitraSDLDeps.cmake b/CMakeModules/CopyCitraSDLDeps.cmake new file mode 100644 index 000000000..4f9e4aeb9 --- /dev/null +++ b/CMakeModules/CopyCitraSDLDeps.cmake @@ -0,0 +1,5 @@ +function(copy_citra_SDL_deps target_dir) + include(WindowsCopyFiles) + set(DLL_DEST "${CMAKE_BINARY_DIR}/bin/$<CONFIG>/") + windows_copy_files(${target_dir} ${SDL2_DLL_DIR} ${DLL_DEST} SDL2.dll) +endfunction(copy_citra_SDL_deps) diff --git a/src/citra/CMakeLists.txt b/src/citra/CMakeLists.txt index f9c488a1a..ecb5d2dfe 100644 --- a/src/citra/CMakeLists.txt +++ b/src/citra/CMakeLists.txt @@ -1,3 +1,5 @@ +set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${PROJECT_SOURCE_DIR}/CMakeModules) + set(SRCS emu_window/emu_window_sdl2.cpp citra.cpp @@ -28,11 +30,6 @@ if(UNIX AND NOT APPLE) endif() if (MSVC) - include(WindowsCopyFiles) - - set(DLL_DEST "${CMAKE_BINARY_DIR}/bin/$<CONFIG>/") - - windows_copy_files(citra ${SDL2_DLL_DIR} ${DLL_DEST} SDL2.dll) - - unset(DLL_DEST) + include(CopyCitraSDLDeps) + copy_citra_SDL_deps(citra) endif() diff --git a/src/citra_qt/CMakeLists.txt b/src/citra_qt/CMakeLists.txt index a9dacd5f1..e1b3566bf 100644 --- a/src/citra_qt/CMakeLists.txt +++ b/src/citra_qt/CMakeLists.txt @@ -1,5 +1,6 @@ set(CMAKE_AUTOMOC ON) set(CMAKE_INCLUDE_CURRENT_DIR ON) +set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${PROJECT_SOURCE_DIR}/CMakeModules) set(SRCS config.cpp @@ -107,27 +108,9 @@ if(UNIX AND NOT APPLE) install(TARGETS citra-qt RUNTIME DESTINATION "${CMAKE_INSTALL_PREFIX}/bin") endif() -if (Qt5_FOUND AND MSVC) - include(WindowsCopyFiles) - - set(Qt5_DLL_DIR "${Qt5_DIR}/../../../bin") - set(Qt5_PLATFORMS_DIR "${Qt5_DIR}/../../../plugins/platforms/") - set(DLL_DEST "${CMAKE_BINARY_DIR}/bin/$<CONFIG>/") - set(PLATFORMS ${DLL_DEST}platforms/) - - windows_copy_files(citra-qt ${Qt5_DLL_DIR} ${DLL_DEST} - icudt*.dll - icuin*.dll - icuuc*.dll - Qt5Core$<$<CONFIG:Debug>:d>.* - Qt5Gui$<$<CONFIG:Debug>:d>.* - Qt5OpenGL$<$<CONFIG:Debug>:d>.* - Qt5Widgets$<$<CONFIG:Debug>:d>.* - ) - windows_copy_files(citra-qt ${Qt5_PLATFORMS_DIR} ${PLATFORMS} qwindows$<$<CONFIG:Debug>:d>.*) - - unset(Qt5_DLL_DIR) - unset(Qt5_PLATFORMS_DIR) - unset(DLL_DEST) - unset(PLATFORMS) +if (MSVC) + include(CopyCitraQt5Deps) + include(CopyCitraSDLDeps) + copy_citra_Qt5_deps(citra-qt) + copy_citra_SDL_deps(citra-qt) endif() diff --git a/src/citra_qt/bootmanager.cpp b/src/citra_qt/bootmanager.cpp index 7699ca8d0..c7eb2aafc 100644 --- a/src/citra_qt/bootmanager.cpp +++ b/src/citra_qt/bootmanager.cpp @@ -19,11 +19,6 @@ #include "video_core/debug_utils/debug_utils.h" #include "video_core/video_core.h" -#define APP_NAME "citra" -#define APP_VERSION "0.1-" VERSION -#define APP_TITLE APP_NAME " " APP_VERSION -#define COPYRIGHT "Copyright (C) 2013-2014 Citra Team" - EmuThread::EmuThread(GRenderWindow* render_window) : exec_step(false), running(false), stop_run(false), render_window(render_window) {} diff --git a/src/citra_qt/debugger/wait_tree.cpp b/src/citra_qt/debugger/wait_tree.cpp index 51e70fae3..5a308bf7f 100644 --- a/src/citra_qt/debugger/wait_tree.cpp +++ b/src/citra_qt/debugger/wait_tree.cpp @@ -229,7 +229,8 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const { list.push_back(std::make_unique<WaitTreeMutexList>(thread.held_mutexes)); } if (thread.status == THREADSTATUS_WAIT_SYNCH) { - list.push_back(std::make_unique<WaitTreeObjectList>(thread.wait_objects, thread.wait_all)); + list.push_back(std::make_unique<WaitTreeObjectList>(thread.wait_objects, + thread.IsSleepingOnWaitAll())); } return list; diff --git a/src/citra_qt/game_list.cpp b/src/citra_qt/game_list.cpp index e536628dd..09469f3c5 100644 --- a/src/citra_qt/game_list.cpp +++ b/src/citra_qt/game_list.cpp @@ -3,6 +3,7 @@ // Refer to the license.txt file included. #include <QHeaderView> +#include <QMenu> #include <QThreadPool> #include <QVBoxLayout> #include "common/common_paths.h" @@ -28,6 +29,7 @@ GameList::GameList(QWidget* parent) : QWidget{parent} { tree_view->setSortingEnabled(true); tree_view->setEditTriggers(QHeaderView::NoEditTriggers); tree_view->setUniformRowHeights(true); + tree_view->setContextMenuPolicy(Qt::CustomContextMenu); item_model->insertColumns(0, COLUMN_COUNT); item_model->setHeaderData(COLUMN_NAME, Qt::Horizontal, "Name"); @@ -35,10 +37,10 @@ GameList::GameList(QWidget* parent) : QWidget{parent} { item_model->setHeaderData(COLUMN_SIZE, Qt::Horizontal, "Size"); connect(tree_view, &QTreeView::activated, this, &GameList::ValidateEntry); + connect(tree_view, &QTreeView::customContextMenuRequested, this, &GameList::PopupContextMenu); // We must register all custom types with the Qt Automoc system so that we are able to use it - // with - // signals/slots. In this case, QList falls under the umbrells of custom types. + // with signals/slots. In this case, QList falls under the umbrells of custom types. qRegisterMetaType<QList<QStandardItem*>>("QList<QStandardItem*>"); layout->addWidget(tree_view); @@ -71,6 +73,23 @@ void GameList::DonePopulating() { tree_view->setEnabled(true); } +void GameList::PopupContextMenu(const QPoint& menu_location) { + QModelIndex item = tree_view->indexAt(menu_location); + if (!item.isValid()) + return; + + int row = item_model->itemFromIndex(item)->row(); + QStandardItem* child_file = item_model->invisibleRootItem()->child(row, COLUMN_NAME); + u64 program_id = child_file->data(GameListItemPath::ProgramIdRole).toULongLong(); + + QMenu context_menu; + QAction* open_save_location = context_menu.addAction(tr("Open Save Data Location")); + open_save_location->setEnabled(program_id != 0); + connect(open_save_location, &QAction::triggered, + [&]() { emit OpenSaveFolderRequested(program_id); }); + context_menu.exec(tree_view->viewport()->mapToGlobal(menu_location)); +} + void GameList::PopulateAsync(const QString& dir_path, bool deep_scan) { if (!FileUtil::Exists(dir_path.toStdString()) || !FileUtil::IsDirectory(dir_path.toStdString())) { @@ -128,8 +147,11 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign std::vector<u8> smdh; loader->ReadIcon(smdh); + u64 program_id = 0; + loader->ReadProgramId(program_id); + emit EntryReady({ - new GameListItemPath(QString::fromStdString(physical_name), smdh), + new GameListItemPath(QString::fromStdString(physical_name), smdh, program_id), new GameListItem( QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))), new GameListItemSize(FileUtil::GetSize(physical_name)), diff --git a/src/citra_qt/game_list.h b/src/citra_qt/game_list.h index 30b2c79a8..1abf10051 100644 --- a/src/citra_qt/game_list.h +++ b/src/citra_qt/game_list.h @@ -36,12 +36,15 @@ public: signals: void GameChosen(QString game_path); void ShouldCancelWorker(); + void OpenSaveFolderRequested(u64 program_id); private: void AddEntry(const QList<QStandardItem*>& entry_items); void ValidateEntry(const QModelIndex& item); void DonePopulating(); + void PopupContextMenu(const QPoint& menu_location); + QTreeView* tree_view = nullptr; QStandardItemModel* item_model = nullptr; GameListWorker* current_worker = nullptr; diff --git a/src/citra_qt/game_list_p.h b/src/citra_qt/game_list_p.h index 5ca3fe991..a15f06c5f 100644 --- a/src/citra_qt/game_list_p.h +++ b/src/citra_qt/game_list_p.h @@ -71,10 +71,13 @@ class GameListItemPath : public GameListItem { public: static const int FullPathRole = Qt::UserRole + 1; static const int TitleRole = Qt::UserRole + 2; + static const int ProgramIdRole = Qt::UserRole + 3; GameListItemPath() : GameListItem() {} - GameListItemPath(const QString& game_path, const std::vector<u8>& smdh_data) : GameListItem() { + GameListItemPath(const QString& game_path, const std::vector<u8>& smdh_data, u64 program_id) + : GameListItem() { setData(game_path, FullPathRole); + setData(qulonglong(program_id), ProgramIdRole); if (!Loader::IsValidSMDH(smdh_data)) { // SMDH is not valid, set a default icon diff --git a/src/citra_qt/main.cpp b/src/citra_qt/main.cpp index a3887f9ab..4486f0870 100644 --- a/src/citra_qt/main.cpp +++ b/src/citra_qt/main.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <cinttypes> #include <clocale> #include <memory> #include <thread> @@ -41,6 +42,7 @@ #include "common/string_util.h" #include "core/arm/disassembler/load_symbol_map.h" #include "core/core.h" +#include "core/file_sys/archive_source_sd_savedata.h" #include "core/gdbstub/gdbstub.h" #include "core/loader/loader.h" #include "core/settings.h" @@ -58,6 +60,36 @@ GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) { ui.setupUi(this); statusBar()->hide(); + InitializeWidgets(); + InitializeDebugMenuActions(); + InitializeRecentFileMenuActions(); + InitializeHotkeys(); + + SetDefaultUIGeometry(); + RestoreUIState(); + + ConnectWidgetEvents(); + + setWindowTitle(QString("Citra | %1-%2").arg(Common::g_scm_branch, Common::g_scm_desc)); + show(); + + game_list->PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan); + + QStringList args = QApplication::arguments(); + if (args.length() >= 2) { + BootGame(args[1].toStdString()); + } +} + +GMainWindow::~GMainWindow() { + // will get automatically deleted otherwise + if (render_window->parent() == nullptr) + delete render_window; + + Pica::g_debug_context.reset(); +} + +void GMainWindow::InitializeWidgets() { render_window = new GRenderWindow(this, emu_thread.get()); render_window->hide(); @@ -93,25 +125,27 @@ GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) { addDockWidget(Qt::RightDockWidgetArea, graphicsCommandsWidget); graphicsCommandsWidget->hide(); - auto graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(Pica::g_debug_context, this); + graphicsBreakpointsWidget = new GraphicsBreakPointsWidget(Pica::g_debug_context, this); addDockWidget(Qt::RightDockWidgetArea, graphicsBreakpointsWidget); graphicsBreakpointsWidget->hide(); - auto graphicsVertexShaderWidget = new GraphicsVertexShaderWidget(Pica::g_debug_context, this); + graphicsVertexShaderWidget = new GraphicsVertexShaderWidget(Pica::g_debug_context, this); addDockWidget(Qt::RightDockWidgetArea, graphicsVertexShaderWidget); graphicsVertexShaderWidget->hide(); - auto graphicsTracingWidget = new GraphicsTracingWidget(Pica::g_debug_context, this); + graphicsTracingWidget = new GraphicsTracingWidget(Pica::g_debug_context, this); addDockWidget(Qt::RightDockWidgetArea, graphicsTracingWidget); graphicsTracingWidget->hide(); - auto graphicsSurfaceViewerAction = new QAction(tr("Create Pica Surface Viewer"), this); - connect(graphicsSurfaceViewerAction, SIGNAL(triggered()), this, - SLOT(OnCreateGraphicsSurfaceViewer())); - waitTreeWidget = new WaitTreeWidget(this); addDockWidget(Qt::LeftDockWidgetArea, waitTreeWidget); waitTreeWidget->hide(); +} + +void GMainWindow::InitializeDebugMenuActions() { + auto graphicsSurfaceViewerAction = new QAction(tr("Create Pica Surface Viewer"), this); + connect(graphicsSurfaceViewerAction, SIGNAL(triggered()), this, + SLOT(OnCreateGraphicsSurfaceViewer())); QMenu* debug_menu = ui.menu_View->addMenu(tr("Debugging")); debug_menu->addAction(graphicsSurfaceViewerAction); @@ -129,19 +163,47 @@ GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) { debug_menu->addAction(graphicsVertexShaderWidget->toggleViewAction()); debug_menu->addAction(graphicsTracingWidget->toggleViewAction()); debug_menu->addAction(waitTreeWidget->toggleViewAction()); +} + +void GMainWindow::InitializeRecentFileMenuActions() { + for (int i = 0; i < max_recent_files_item; ++i) { + actions_recent_files[i] = new QAction(this); + actions_recent_files[i]->setVisible(false); + connect(actions_recent_files[i], SIGNAL(triggered()), this, SLOT(OnMenuRecentFile())); + + ui.menu_recent_files->addAction(actions_recent_files[i]); + } + + UpdateRecentFiles(); +} + +void GMainWindow::InitializeHotkeys() { + RegisterHotkey("Main Window", "Load File", QKeySequence::Open); + RegisterHotkey("Main Window", "Swap Screens", QKeySequence::NextChild); + RegisterHotkey("Main Window", "Start Emulation"); + LoadHotkeys(); + + connect(GetHotkey("Main Window", "Load File", this), SIGNAL(activated()), this, + SLOT(OnMenuLoadFile())); + connect(GetHotkey("Main Window", "Start Emulation", this), SIGNAL(activated()), this, + SLOT(OnStartGame())); + connect(GetHotkey("Main Window", "Swap Screens", render_window), SIGNAL(activated()), this, + SLOT(OnSwapScreens())); +} - // Set default UI state +void GMainWindow::SetDefaultUIGeometry() { // geometry: 55% of the window contents are in the upper screen half, 45% in the lower half - QDesktopWidget* desktop = ((QApplication*)QApplication::instance())->desktop(); - QRect screenRect = desktop->screenGeometry(this); - int x, y, w, h; - w = screenRect.width() * 2 / 3; - h = screenRect.height() / 2; - x = (screenRect.x() + screenRect.width()) / 2 - w / 2; - y = (screenRect.y() + screenRect.height()) / 2 - h * 55 / 100; + const QRect screenRect = QApplication::desktop()->screenGeometry(this); + + const int w = screenRect.width() * 2 / 3; + const int h = screenRect.height() / 2; + const int x = (screenRect.x() + screenRect.width()) / 2 - w / 2; + const int y = (screenRect.y() + screenRect.height()) / 2 - h * 55 / 100; + setGeometry(x, y, w, h); +} - // Restore UI state +void GMainWindow::RestoreUIState() { restoreGeometry(UISettings::values.geometry); restoreState(UISettings::values.state); render_window->restoreGeometry(UISettings::values.renderwindow_geometry); @@ -157,20 +219,13 @@ GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) { ui.actionDisplay_widget_title_bars->setChecked(UISettings::values.display_titlebar); OnDisplayTitleBars(ui.actionDisplay_widget_title_bars->isChecked()); +} - // Prepare actions for recent files - for (int i = 0; i < max_recent_files_item; ++i) { - actions_recent_files[i] = new QAction(this); - actions_recent_files[i]->setVisible(false); - connect(actions_recent_files[i], SIGNAL(triggered()), this, SLOT(OnMenuRecentFile())); - - ui.menu_recent_files->addAction(actions_recent_files[i]); - } - UpdateRecentFiles(); - - // Setup connections +void GMainWindow::ConnectWidgetEvents() { connect(game_list, SIGNAL(GameChosen(QString)), this, SLOT(OnGameListLoadFile(QString)), Qt::DirectConnection); + connect(game_list, SIGNAL(OpenSaveFolderRequested(u64)), this, + SLOT(OnGameListOpenSaveFolder(u64)), Qt::DirectConnection); connect(ui.action_Configure, SIGNAL(triggered()), this, SLOT(OnConfigure())); connect(ui.action_Load_File, SIGNAL(triggered()), this, SLOT(OnMenuLoadFile()), Qt::DirectConnection); @@ -197,40 +252,6 @@ GMainWindow::GMainWindow() : config(new Config()), emu_thread(nullptr) { connect(this, SIGNAL(EmulationStarting(EmuThread*)), waitTreeWidget, SLOT(OnEmulationStarting(EmuThread*))); connect(this, SIGNAL(EmulationStopping()), waitTreeWidget, SLOT(OnEmulationStopping())); - - // Setup hotkeys - RegisterHotkey("Main Window", "Load File", QKeySequence::Open); - RegisterHotkey("Main Window", "Swap Screens", QKeySequence::NextChild); - RegisterHotkey("Main Window", "Start Emulation"); - LoadHotkeys(); - - connect(GetHotkey("Main Window", "Load File", this), SIGNAL(activated()), this, - SLOT(OnMenuLoadFile())); - connect(GetHotkey("Main Window", "Start Emulation", this), SIGNAL(activated()), this, - SLOT(OnStartGame())); - connect(GetHotkey("Main Window", "Swap Screens", this), SIGNAL(activated()), this, - SLOT(OnSwapScreens())); - - std::string window_title = - Common::StringFromFormat("Citra | %s-%s", Common::g_scm_branch, Common::g_scm_desc); - setWindowTitle(window_title.c_str()); - - show(); - - game_list->PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan); - - QStringList args = QApplication::arguments(); - if (args.length() >= 2) { - BootGame(args[1].toStdString()); - } -} - -GMainWindow::~GMainWindow() { - // will get automatically deleted otherwise - if (render_window->parent() == nullptr) - delete render_window; - - Pica::g_debug_context.reset(); } void GMainWindow::OnDisplayTitleBars(bool show) { @@ -386,6 +407,7 @@ void GMainWindow::BootGame(const std::string& filename) { game_list->hide(); } render_window->show(); + render_window->setFocus(); emulation_running = true; OnStartGame(); @@ -460,6 +482,21 @@ void GMainWindow::OnGameListLoadFile(QString game_path) { BootGame(game_path.toStdString()); } +void GMainWindow::OnGameListOpenSaveFolder(u64 program_id) { + std::string sdmc_dir = FileUtil::GetUserPath(D_SDMC_IDX); + std::string path = FileSys::ArchiveSource_SDSaveData::GetSaveDataPathFor(sdmc_dir, program_id); + QString qpath = QString::fromStdString(path); + + QDir dir(qpath); + if (!dir.exists()) { + QMessageBox::critical(this, tr("Error Opening Save Folder"), tr("Folder does not exist!")); + return; + } + + LOG_INFO(Frontend, "Opening save data path for program_id=%" PRIu64, program_id); + QDesktopServices::openUrl(QUrl::fromLocalFile(qpath)); +} + void GMainWindow::OnMenuLoadFile() { QString filename = QFileDialog::getOpenFileName(this, tr("Load File"), UISettings::values.roms_path, diff --git a/src/citra_qt/main.h b/src/citra_qt/main.h index f87178227..a2fd45c47 100644 --- a/src/citra_qt/main.h +++ b/src/citra_qt/main.h @@ -9,18 +9,21 @@ #include <QMainWindow> #include "ui_main.h" +class CallstackWidget; class Config; +class DisassemblerWidget; +class EmuThread; class GameList; class GImageInfo; +class GPUCommandStreamWidget; +class GPUCommandListWidget; +class GraphicsBreakPointsWidget; +class GraphicsTracingWidget; +class GraphicsVertexShaderWidget; class GRenderWindow; -class EmuThread; -class ProfilerWidget; class MicroProfileDialog; -class DisassemblerWidget; +class ProfilerWidget; class RegistersWidget; -class CallstackWidget; -class GPUCommandStreamWidget; -class GPUCommandListWidget; class WaitTreeWidget; class GMainWindow : public QMainWindow { @@ -60,6 +63,16 @@ signals: void EmulationStopping(); private: + void InitializeWidgets(); + void InitializeDebugMenuActions(); + void InitializeRecentFileMenuActions(); + void InitializeHotkeys(); + + void SetDefaultUIGeometry(); + void RestoreUIState(); + + void ConnectWidgetEvents(); + /** * Initializes the emulation system. * @param system_mode The system mode with which to intialize the kernel. @@ -105,6 +118,7 @@ private slots: void OnStopGame(); /// Called whenever a user selects a game in the game list widget. void OnGameListLoadFile(QString game_path); + void OnGameListOpenSaveFolder(u64 program_id); void OnMenuLoadFile(); void OnMenuLoadSymbolMap(); /// Called whenever a user selects the "File->Select Game List Root" menu item @@ -135,6 +149,9 @@ private: CallstackWidget* callstackWidget; GPUCommandStreamWidget* graphicsWidget; GPUCommandListWidget* graphicsCommandsWidget; + GraphicsBreakPointsWidget* graphicsBreakpointsWidget; + GraphicsVertexShaderWidget* graphicsVertexShaderWidget; + GraphicsTracingWidget* graphicsTracingWidget; WaitTreeWidget* waitTreeWidget; QAction* actions_recent_files[max_recent_files_item]; diff --git a/src/common/file_util.cpp b/src/common/file_util.cpp index b6161f2dc..c618495f7 100644 --- a/src/common/file_util.cpp +++ b/src/common/file_util.cpp @@ -40,9 +40,20 @@ #endif #if defined(__APPLE__) +// CFURL contains __attribute__ directives that gcc does not know how to parse, so we need to just +// ignore them if we're not using clang. The macro is only used to prevent linking against +// functions that don't exist on older versions of macOS, and the worst case scenario is a linker +// error, so this is perfectly safe, just inconvenient. +#ifndef __clang__ +#define availability(...) +#endif #include <CoreFoundation/CFBundle.h> #include <CoreFoundation/CFString.h> #include <CoreFoundation/CFURL.h> +#ifdef availability +#undef availability +#endif + #endif #include <algorithm> diff --git a/src/common/scm_rev.cpp.in b/src/common/scm_rev.cpp.in index 09366e801..79b404bb8 100644 --- a/src/common/scm_rev.cpp.in +++ b/src/common/scm_rev.cpp.in @@ -1,5 +1,5 @@ // Copyright 2014 Citra Emulator Project -// Licensed under GPLv2 +// Licensed under GPLv2 or any later version // Refer to the license.txt file included. #include "common/scm_rev.h" diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index 7b3b4389a..af224166a 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -53,8 +53,9 @@ set(SRCS hle/kernel/timer.cpp hle/kernel/vm_manager.cpp hle/service/ac_u.cpp - hle/service/act_a.cpp - hle/service/act_u.cpp + hle/service/act/act.cpp + hle/service/act/act_a.cpp + hle/service/act/act_u.cpp hle/service/am/am.cpp hle/service/am/am_app.cpp hle/service/am/am_net.cpp @@ -74,6 +75,7 @@ set(SRCS hle/service/cam/cam_s.cpp hle/service/cam/cam_u.cpp hle/service/cecd/cecd.cpp + hle/service/cecd/cecd_ndm.cpp hle/service/cecd/cecd_s.cpp hle/service/cecd/cecd_u.cpp hle/service/cfg/cfg.cpp @@ -218,8 +220,9 @@ set(HEADERS hle/kernel/vm_manager.h hle/result.h hle/service/ac_u.h - hle/service/act_a.h - hle/service/act_u.h + hle/service/act/act.h + hle/service/act/act_a.h + hle/service/act/act_u.h hle/service/am/am.h hle/service/am/am_app.h hle/service/am/am_net.h @@ -239,6 +242,7 @@ set(HEADERS hle/service/cam/cam_s.h hle/service/cam/cam_u.h hle/service/cecd/cecd.h + hle/service/cecd/cecd_ndm.h hle/service/cecd/cecd_s.h hle/service/cecd/cecd_u.h hle/service/cfg/cfg.h diff --git a/src/core/arm/dyncom/arm_dyncom_interpreter.cpp b/src/core/arm/dyncom/arm_dyncom_interpreter.cpp index 7b8616702..67c45640a 100644 --- a/src/core/arm/dyncom/arm_dyncom_interpreter.cpp +++ b/src/core/arm/dyncom/arm_dyncom_interpreter.cpp @@ -953,7 +953,7 @@ unsigned InterpreterMainLoop(ARMul_State* cpu) { #define GDB_BP_CHECK \ cpu->Cpsr &= ~(1 << 5); \ cpu->Cpsr |= cpu->TFlag << 5; \ - if (GDBStub::g_server_enabled) { \ + if (GDBStub::IsServerEnabled()) { \ if (GDBStub::IsMemoryBreak() || (breakpoint_data.type != GDBStub::BreakpointType::None && \ PC == breakpoint_data.address)) { \ GDBStub::Break(); \ @@ -1649,7 +1649,7 @@ DISPATCH : { } // Find breakpoint if one exists within the block - if (GDBStub::g_server_enabled && GDBStub::IsConnected()) { + if (GDBStub::IsConnected()) { breakpoint_data = GDBStub::GetNextBreakpointFromAddress(cpu->Reg[15], GDBStub::BreakpointType::Execute); } diff --git a/src/core/arm/dyncom/arm_dyncom_trans.h b/src/core/arm/dyncom/arm_dyncom_trans.h index b1ec90662..632ff2cd6 100644 --- a/src/core/arm/dyncom/arm_dyncom_trans.h +++ b/src/core/arm/dyncom/arm_dyncom_trans.h @@ -1,3 +1,5 @@ +#pragma once + #include <cstddef> #include "common/common_types.h" diff --git a/src/core/arm/skyeye_common/armstate.cpp b/src/core/arm/skyeye_common/armstate.cpp index 1465b074e..c36b0208f 100644 --- a/src/core/arm/skyeye_common/armstate.cpp +++ b/src/core/arm/skyeye_common/armstate.cpp @@ -182,7 +182,7 @@ void ARMul_State::ResetMPCoreCP15Registers() { } static void CheckMemoryBreakpoint(u32 address, GDBStub::BreakpointType type) { - if (GDBStub::g_server_enabled && GDBStub::CheckBreakpoint(address, type)) { + if (GDBStub::IsServerEnabled() && GDBStub::CheckBreakpoint(address, type)) { LOG_DEBUG(Debug, "Found memory breakpoint @ %08x", address); GDBStub::Break(true); } diff --git a/src/core/core.cpp b/src/core/core.cpp index 49ac8be6e..6efa18159 100644 --- a/src/core/core.cpp +++ b/src/core/core.cpp @@ -22,7 +22,7 @@ std::unique_ptr<ARM_Interface> g_sys_core; ///< ARM11 system (OS) core /// Run the core CPU loop void RunLoop(int tight_loop) { - if (GDBStub::g_server_enabled) { + if (GDBStub::IsServerEnabled()) { GDBStub::HandlePacket(); // If the loop is halted and we want to step, use a tiny (1) number of instructions to diff --git a/src/core/file_sys/archive_source_sd_savedata.cpp b/src/core/file_sys/archive_source_sd_savedata.cpp index 2d8a950a3..287322d3e 100644 --- a/src/core/file_sys/archive_source_sd_savedata.cpp +++ b/src/core/file_sys/archive_source_sd_savedata.cpp @@ -90,4 +90,9 @@ ResultVal<ArchiveFormatInfo> ArchiveSource_SDSaveData::GetFormatInfo(u64 program return MakeResult<ArchiveFormatInfo>(info); } +std::string ArchiveSource_SDSaveData::GetSaveDataPathFor(const std::string& mount_point, + u64 program_id) { + return GetSaveDataPath(GetSaveDataContainerPath(mount_point), program_id); +} + } // namespace FileSys diff --git a/src/core/file_sys/archive_source_sd_savedata.h b/src/core/file_sys/archive_source_sd_savedata.h index b33126c31..b5fe43cc1 100644 --- a/src/core/file_sys/archive_source_sd_savedata.h +++ b/src/core/file_sys/archive_source_sd_savedata.h @@ -23,6 +23,8 @@ public: ResultCode Format(u64 program_id, const FileSys::ArchiveFormatInfo& format_info); ResultVal<ArchiveFormatInfo> GetFormatInfo(u64 program_id) const; + static std::string GetSaveDataPathFor(const std::string& mount_point, u64 program_id); + private: std::string mount_point; }; diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp index 21d941363..f96cbde64 100644 --- a/src/core/gdbstub/gdbstub.cpp +++ b/src/core/gdbstub/gdbstub.cpp @@ -5,6 +5,7 @@ // Originally written by Sven Peter <sven@fail0verflow.com> for anergistic. #include <algorithm> +#include <atomic> #include <climits> #include <csignal> #include <cstdarg> @@ -130,7 +131,10 @@ static u16 gdbstub_port = 24689; static bool halt_loop = true; static bool step_loop = false; -std::atomic<bool> g_server_enabled(false); + +// If set to false, the server will never be started and no +// gdbstub-related functions will be executed. +static std::atomic<bool> server_enabled(false); #ifdef _WIN32 WSADATA InitData; @@ -181,11 +185,10 @@ static u8 NibbleToHex(u8 n) { /** * Converts input hex string characters into an array of equivalent of u8 bytes. * -* @param dest Pointer to buffer to store u8 bytes. * @param src Pointer to array of output hex string characters. * @param len Length of src array. */ -static u32 HexToInt(u8* src, u32 len) { +static u32 HexToInt(const u8* src, size_t len) { u32 output = 0; while (len-- > 0) { output = (output << 4) | HexCharToValue(src[0]); @@ -201,7 +204,7 @@ static u32 HexToInt(u8* src, u32 len) { * @param src Pointer to array of u8 bytes. * @param len Length of src array. */ -static void MemToGdbHex(u8* dest, u8* src, u32 len) { +static void MemToGdbHex(u8* dest, const u8* src, size_t len) { while (len-- > 0) { u8 tmp = *src++; *dest++ = NibbleToHex(tmp >> 4); @@ -216,7 +219,7 @@ static void MemToGdbHex(u8* dest, u8* src, u32 len) { * @param src Pointer to array of output hex string characters. * @param len Length of src array. */ -static void GdbHexToMem(u8* dest, u8* src, u32 len) { +static void GdbHexToMem(u8* dest, const u8* src, size_t len) { while (len-- > 0) { *dest++ = (HexCharToValue(src[0]) << 4) | HexCharToValue(src[1]); src += 2; @@ -240,7 +243,7 @@ static void IntToGdbHex(u8* dest, u32 v) { * * @param src Pointer to hex string. */ -static u32 GdbHexToInt(u8* src) { +static u32 GdbHexToInt(const u8* src) { u32 output = 0; for (int i = 0; i < 8; i += 2) { @@ -264,7 +267,7 @@ static u8 ReadByte() { } /// Calculate the checksum of the current command buffer. -static u8 CalculateChecksum(u8* buffer, u32 length) { +static u8 CalculateChecksum(const u8* buffer, size_t length) { return static_cast<u8>(std::accumulate(buffer, buffer + length, 0, std::plus<u8>())); } @@ -582,7 +585,7 @@ static void ReadRegisters() { /// Modify data of register specified by gdb client. static void WriteRegister() { - u8* buffer_ptr = command_buffer + 3; + const u8* buffer_ptr = command_buffer + 3; u32 id = HexCharToValue(command_buffer[1]); if (command_buffer[2] != '=') { @@ -608,7 +611,7 @@ static void WriteRegister() { /// Modify all registers with data received from the client. static void WriteRegisters() { - u8* buffer_ptr = command_buffer + 1; + const u8* buffer_ptr = command_buffer + 1; if (command_buffer[0] != 'G') return SendReply("E01"); @@ -653,7 +656,7 @@ static void ReadMemory() { SendReply("E01"); } - u8* data = Memory::GetPointer(addr); + const u8* data = Memory::GetPointer(addr); if (!data) { return SendReply("E00"); } @@ -902,7 +905,7 @@ void SetServerPort(u16 port) { void ToggleServer(bool status) { if (status) { - g_server_enabled = status; + server_enabled = status; // Start server if (!IsConnected() && Core::g_sys_core != nullptr) { @@ -914,12 +917,12 @@ void ToggleServer(bool status) { Shutdown(); } - g_server_enabled = status; + server_enabled = status; } } static void Init(u16 port) { - if (!g_server_enabled) { + if (!server_enabled) { // Set the halt loop to false in case the user enabled the gdbstub mid-execution. // This way the CPU can still execute normally. halt_loop = false; @@ -998,7 +1001,7 @@ void Init() { } void Shutdown() { - if (!g_server_enabled) { + if (!server_enabled) { return; } @@ -1015,8 +1018,12 @@ void Shutdown() { LOG_INFO(Debug_GDBStub, "GDB stopped."); } +bool IsServerEnabled() { + return server_enabled; +} + bool IsConnected() { - return g_server_enabled && gdbserver_socket != -1; + return IsServerEnabled() && gdbserver_socket != -1; } bool GetCpuHaltFlag() { diff --git a/src/core/gdbstub/gdbstub.h b/src/core/gdbstub/gdbstub.h index a7483bb10..38177e32c 100644 --- a/src/core/gdbstub/gdbstub.h +++ b/src/core/gdbstub/gdbstub.h @@ -5,7 +5,7 @@ // Originally written by Sven Peter <sven@fail0verflow.com> for anergistic. #pragma once -#include <atomic> + #include "common/common_types.h" namespace GDBStub { @@ -24,10 +24,6 @@ struct BreakpointAddress { BreakpointType type; }; -/// If set to false, the server will never be started and no gdbstub-related functions will be -/// executed. -extern std::atomic<bool> g_server_enabled; - /** * Set the port the gdbstub should use to listen for connections. * @@ -36,7 +32,7 @@ extern std::atomic<bool> g_server_enabled; void SetServerPort(u16 port); /** - * Set the g_server_enabled flag and start or stop the server if possible. + * Starts or stops the server if possible. * * @param status Set the server to enabled or disabled. */ @@ -48,6 +44,9 @@ void Init(); /// Stop gdbstub server. void Shutdown(); +/// Checks if the gdbstub server is enabled. +bool IsServerEnabled(); + /// Returns true if there is an active socket connection. bool IsConnected(); diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp index 37eec4c84..b5a0cc3a3 100644 --- a/src/core/hle/kernel/address_arbiter.cpp +++ b/src/core/hle/kernel/address_arbiter.cpp @@ -79,8 +79,6 @@ ResultCode AddressArbiter::ArbitrateAddress(ArbitrationType type, VAddr address, ErrorSummary::WrongArgument, ErrorLevel::Usage); } - HLE::Reschedule(__func__); - // The calls that use a timeout seem to always return a Timeout error even if they did not put // the thread to sleep if (type == ArbitrationType::WaitIfLessThanWithTimeout || diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp index 0c8752670..1db8e102f 100644 --- a/src/core/hle/kernel/kernel.cpp +++ b/src/core/hle/kernel/kernel.cpp @@ -3,6 +3,7 @@ // Refer to the license.txt file included. #include <algorithm> +#include <boost/range/algorithm_ext/erase.hpp> #include "common/assert.h" #include "common/logging/log.h" #include "core/hle/config_mem.h" @@ -31,13 +32,61 @@ void WaitObject::RemoveWaitingThread(Thread* thread) { waiting_threads.erase(itr); } -void WaitObject::WakeupAllWaitingThreads() { - for (auto thread : waiting_threads) - thread->ResumeFromWait(); +SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() { + // Remove the threads that are ready or already running from our waitlist + boost::range::remove_erase_if(waiting_threads, [](const SharedPtr<Thread>& thread) { + return thread->status == THREADSTATUS_RUNNING || thread->status == THREADSTATUS_READY || + thread->status == THREADSTATUS_DEAD; + }); + + // TODO(Subv): This call should be performed inside the loop below to check if an object can be + // acquired by a particular thread. This is useful for things like recursive locking of Mutexes. + if (ShouldWait()) + return nullptr; + + Thread* candidate = nullptr; + s32 candidate_priority = THREADPRIO_LOWEST + 1; + + for (const auto& thread : waiting_threads) { + if (thread->current_priority >= candidate_priority) + continue; - waiting_threads.clear(); + bool ready_to_run = + std::none_of(thread->wait_objects.begin(), thread->wait_objects.end(), + [](const SharedPtr<WaitObject>& object) { return object->ShouldWait(); }); + if (ready_to_run) { + candidate = thread.get(); + candidate_priority = thread->current_priority; + } + } + + return candidate; +} - HLE::Reschedule(__func__); +void WaitObject::WakeupAllWaitingThreads() { + while (auto thread = GetHighestPriorityReadyThread()) { + if (!thread->IsSleepingOnWaitAll()) { + Acquire(); + // Set the output index of the WaitSynchronizationN call to the index of this object. + if (thread->wait_set_output) { + thread->SetWaitSynchronizationOutput(thread->GetWaitObjectIndex(this)); + thread->wait_set_output = false; + } + } else { + for (auto& object : thread->wait_objects) { + object->Acquire(); + object->RemoveWaitingThread(thread.get()); + } + // Note: This case doesn't update the output index of WaitSynchronizationN. + // Clear the thread's waitlist + thread->wait_objects.clear(); + } + + thread->SetWaitSynchronizationResult(RESULT_SUCCESS); + thread->ResumeFromWait(); + // Note: Removing the thread from the object's waitlist will be + // done by GetHighestPriorityReadyThread. + } } const std::vector<SharedPtr<Thread>>& WaitObject::GetWaitingThreads() const { diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h index 0b811c5a7..1adcf6c71 100644 --- a/src/core/hle/kernel/kernel.h +++ b/src/core/hle/kernel/kernel.h @@ -151,9 +151,15 @@ public: */ void RemoveWaitingThread(Thread* thread); - /// Wake up all threads waiting on this object + /** + * Wake up all threads waiting on this object that can be awoken, in priority order, + * and set the synchronization result and output of the thread. + */ void WakeupAllWaitingThreads(); + /// Obtains the highest priority thread that is ready to run from this object's waiting list. + SharedPtr<Thread> GetHighestPriorityReadyThread(); + /// Get a const reference to the waiting threads list for debug use const std::vector<SharedPtr<Thread>>& GetWaitingThreads() const; diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp index 84d6d24c6..18b696f72 100644 --- a/src/core/hle/kernel/thread.cpp +++ b/src/core/hle/kernel/thread.cpp @@ -46,7 +46,7 @@ static std::vector<SharedPtr<Thread>> thread_list; // Lists only ready thread ids. static Common::ThreadQueueList<Thread*, THREADPRIO_LOWEST + 1> ready_queue; -static Thread* current_thread; +static SharedPtr<Thread> current_thread; // The first available thread id at startup static u32 next_thread_id; @@ -63,7 +63,7 @@ Thread::Thread() {} Thread::~Thread() {} Thread* GetCurrentThread() { - return current_thread; + return current_thread.get(); } /** @@ -120,8 +120,6 @@ void Thread::Stop() { u32 tls_slot = ((tls_address - Memory::TLS_AREA_VADDR) % Memory::PAGE_SIZE) / Memory::TLS_ENTRY_SIZE; Kernel::g_current_process->tls_slots[tls_page].reset(tls_slot); - - HLE::Reschedule(__func__); } Thread* ArbitrateHighestPriorityThread(u32 address) { @@ -181,50 +179,6 @@ static void PriorityBoostStarvedThreads() { } /** - * Gets the registers for timeout parameter of the next WaitSynchronization call. - * @param thread a pointer to the thread that is ready to call WaitSynchronization - * @returns a tuple of two register pointers to low and high part of the timeout parameter - */ -static std::tuple<u32*, u32*> GetWaitSynchTimeoutParameterRegister(Thread* thread) { - bool thumb_mode = (thread->context.cpsr & TBIT) != 0; - u16 thumb_inst = Memory::Read16(thread->context.pc & 0xFFFFFFFE); - u32 inst = Memory::Read32(thread->context.pc & 0xFFFFFFFC) & 0x0FFFFFFF; - - if ((thumb_mode && thumb_inst == 0xDF24) || (!thumb_mode && inst == 0x0F000024)) { - // svc #0x24 (WaitSynchronization1) - return std::make_tuple(&thread->context.cpu_registers[2], - &thread->context.cpu_registers[3]); - } else if ((thumb_mode && thumb_inst == 0xDF25) || (!thumb_mode && inst == 0x0F000025)) { - // svc #0x25 (WaitSynchronizationN) - return std::make_tuple(&thread->context.cpu_registers[0], - &thread->context.cpu_registers[4]); - } - - UNREACHABLE(); -} - -/** - * Updates the WaitSynchronization timeout parameter according to the difference - * between ticks of the last WaitSynchronization call and the incoming one. - * @param timeout_low a pointer to the register for the low part of the timeout parameter - * @param timeout_high a pointer to the register for the high part of the timeout parameter - * @param last_tick tick of the last WaitSynchronization call - */ -static void UpdateTimeoutParameter(u32* timeout_low, u32* timeout_high, u64 last_tick) { - s64 timeout = ((s64)*timeout_high << 32) | *timeout_low; - - if (timeout != -1) { - timeout -= cyclesToUs(CoreTiming::GetTicks() - last_tick) * 1000; // in nanoseconds - - if (timeout < 0) - timeout = 0; - - *timeout_low = timeout & 0xFFFFFFFF; - *timeout_high = timeout >> 32; - } -} - -/** * Switches the CPU's active thread context to that of the specified thread * @param new_thread The thread to switch to */ @@ -254,32 +208,6 @@ static void SwitchContext(Thread* new_thread) { current_thread = new_thread; - // If the thread was waited by a svcWaitSynch call, step back PC by one instruction to rerun - // the SVC when the thread wakes up. This is necessary to ensure that the thread can acquire - // the requested wait object(s) before continuing. - if (new_thread->waitsynch_waited) { - // CPSR flag indicates CPU mode - bool thumb_mode = (new_thread->context.cpsr & TBIT) != 0; - - // SVC instruction is 2 bytes for THUMB, 4 bytes for ARM - new_thread->context.pc -= thumb_mode ? 2 : 4; - - // Get the register for timeout parameter - u32 *timeout_low, *timeout_high; - std::tie(timeout_low, timeout_high) = GetWaitSynchTimeoutParameterRegister(new_thread); - - // Update the timeout parameter - UpdateTimeoutParameter(timeout_low, timeout_high, new_thread->last_running_ticks); - } - - // Clean up the thread's wait_objects, they'll be restored if needed during - // the svcWaitSynchronization call - for (size_t i = 0; i < new_thread->wait_objects.size(); ++i) { - SharedPtr<WaitObject> object = new_thread->wait_objects[i]; - object->RemoveWaitingThread(new_thread); - } - new_thread->wait_objects.clear(); - ready_queue.remove(new_thread->current_priority, new_thread); new_thread->status = THREADSTATUS_RUNNING; @@ -319,17 +247,13 @@ static Thread* PopNextReadyThread() { void WaitCurrentThread_Sleep() { Thread* thread = GetCurrentThread(); thread->status = THREADSTATUS_WAIT_SLEEP; - - HLE::Reschedule(__func__); } void WaitCurrentThread_WaitSynchronization(std::vector<SharedPtr<WaitObject>> wait_objects, - bool wait_set_output, bool wait_all) { + bool wait_set_output) { Thread* thread = GetCurrentThread(); thread->wait_set_output = wait_set_output; - thread->wait_all = wait_all; thread->wait_objects = std::move(wait_objects); - thread->waitsynch_waited = true; thread->status = THREADSTATUS_WAIT_SYNCH; } @@ -339,6 +263,13 @@ void WaitCurrentThread_ArbitrateAddress(VAddr wait_address) { thread->status = THREADSTATUS_WAIT_ARB; } +void ExitCurrentThread() { + Thread* thread = GetCurrentThread(); + thread->Stop(); + thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread), + thread_list.end()); +} + /** * Callback that will wake up the thread it was scheduled for * @param thread_handle The handle of the thread that's been awoken @@ -351,15 +282,15 @@ static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) { return; } - thread->waitsynch_waited = false; - if (thread->status == THREADSTATUS_WAIT_SYNCH || thread->status == THREADSTATUS_WAIT_ARB) { + thread->wait_set_output = false; + // Remove the thread from each of its waiting objects' waitlists + for (auto& object : thread->wait_objects) + object->RemoveWaitingThread(thread.get()); + thread->wait_objects.clear(); thread->SetWaitSynchronizationResult(ResultCode(ErrorDescription::Timeout, ErrorModule::OS, ErrorSummary::StatusChanged, ErrorLevel::Info)); - - if (thread->wait_set_output) - thread->SetWaitSynchronizationOutput(-1); } thread->ResumeFromWait(); @@ -399,6 +330,7 @@ void Thread::ResumeFromWait() { ready_queue.push_back(current_priority, this); status = THREADSTATUS_READY; + HLE::Reschedule(__func__); } /** @@ -494,13 +426,11 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point, thread->last_running_ticks = CoreTiming::GetTicks(); thread->processor_id = processor_id; thread->wait_set_output = false; - thread->wait_all = false; thread->wait_objects.clear(); thread->wait_address = 0; thread->name = std::move(name); thread->callback_handle = wakeup_callback_handle_table.Create(thread).MoveFrom(); thread->owner_process = g_current_process; - thread->waitsynch_waited = false; // Find the next available TLS index, and mark it as used auto& tls_slots = Kernel::g_current_process->tls_slots; @@ -555,8 +485,6 @@ ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point, ready_queue.push_back(thread->current_priority, thread.get()); thread->status = THREADSTATUS_READY; - HLE::Reschedule(__func__); - return MakeResult<SharedPtr<Thread>>(std::move(thread)); } @@ -619,14 +547,6 @@ void Reschedule() { HLE::DoneRescheduling(); - // Don't bother switching to the same thread. - // But if the thread was waiting on objects, we still need to switch it - // to perform PC modification, change state to RUNNING, etc. - // This occurs in the case when an object the thread is waiting on immediately wakes up - // the current thread before Reschedule() is called. - if (next == cur && (next == nullptr || next->waitsynch_waited == false)) - return; - if (cur && next) { LOG_TRACE(Kernel, "context switch %u -> %u", cur->GetObjectId(), next->GetObjectId()); } else if (cur) { diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h index e0ffcea8a..d4fefc573 100644 --- a/src/core/hle/kernel/thread.h +++ b/src/core/hle/kernel/thread.h @@ -5,7 +5,9 @@ #pragma once #include <string> +#include <unordered_map> #include <vector> +#include <boost/container/flat_map.hpp> #include <boost/container/flat_set.hpp> #include "common/common_types.h" #include "core/core.h" @@ -125,6 +127,16 @@ public: void SetWaitSynchronizationOutput(s32 output); /** + * Retrieves the index that this particular object occupies in the list of objects + * that the thread passed to WaitSynchronizationN. + * It is used to set the output value of WaitSynchronizationN when the thread is awakened. + * @param object Object to query the index of. + */ + s32 GetWaitObjectIndex(const WaitObject* object) const { + return wait_objects_index.at(object->GetObjectId()); + } + + /** * Stops a thread, invalidating it from further use */ void Stop(); @@ -137,6 +149,15 @@ public: return tls_address; } + /** + * Returns whether this thread is waiting for all the objects in + * its wait list to become ready, as a result of a WaitSynchronizationN call + * with wait_all = true, or a ReplyAndReceive call. + */ + bool IsSleepingOnWaitAll() const { + return !wait_objects.empty(); + } + Core::ThreadContext context; u32 thread_id; @@ -154,16 +175,22 @@ public: VAddr tls_address; ///< Virtual address of the Thread Local Storage of the thread - bool waitsynch_waited; ///< Set to true if the last svcWaitSynch call caused the thread to wait - /// Mutexes currently held by this thread, which will be released when it exits. boost::container::flat_set<SharedPtr<Mutex>> held_mutexes; - SharedPtr<Process> owner_process; ///< Process that owns this thread - std::vector<SharedPtr<WaitObject>> wait_objects; ///< Objects that the thread is waiting on - VAddr wait_address; ///< If waiting on an AddressArbiter, this is the arbitration address - bool wait_all; ///< True if the thread is waiting on all objects before resuming - bool wait_set_output; ///< True if the output parameter should be set on thread wakeup + SharedPtr<Process> owner_process; ///< Process that owns this thread + + /// Objects that the thread is waiting on. + /// This is only populated when the thread should wait for all the objects to become ready. + std::vector<SharedPtr<WaitObject>> wait_objects; + + /// Mapping of Object ids to their position in the last waitlist that this object waited on. + boost::container::flat_map<int, s32> wait_objects_index; + + VAddr wait_address; ///< If waiting on an AddressArbiter, this is the arbitration address + + /// True if the WaitSynchronizationN output parameter should be set on thread wakeup. + bool wait_set_output; std::string name; @@ -215,10 +242,9 @@ void WaitCurrentThread_Sleep(); * @param wait_objects Kernel objects that we are waiting on * @param wait_set_output If true, set the output parameter on thread wakeup (for * WaitSynchronizationN only) - * @param wait_all If true, wait on all objects before resuming (for WaitSynchronizationN only) */ void WaitCurrentThread_WaitSynchronization(std::vector<SharedPtr<WaitObject>> wait_objects, - bool wait_set_output, bool wait_all); + bool wait_set_output); /** * Waits the current thread from an ArbitrateAddress call @@ -227,6 +253,11 @@ void WaitCurrentThread_WaitSynchronization(std::vector<SharedPtr<WaitObject>> wa void WaitCurrentThread_ArbitrateAddress(VAddr wait_address); /** + * Stops the current thread and removes it from the thread_list + */ +void ExitCurrentThread(); + +/** * Initialize threading */ void ThreadingInit(); diff --git a/src/core/hle/kernel/timer.cpp b/src/core/hle/kernel/timer.cpp index eac181f4e..b50cf520d 100644 --- a/src/core/hle/kernel/timer.cpp +++ b/src/core/hle/kernel/timer.cpp @@ -60,14 +60,10 @@ void Timer::Set(s64 initial, s64 interval) { u64 initial_microseconds = initial / 1000; CoreTiming::ScheduleEvent(usToCycles(initial_microseconds), timer_callback_event_type, callback_handle); - - HLE::Reschedule(__func__); } void Timer::Cancel() { CoreTiming::UnscheduleEvent(timer_callback_event_type, callback_handle); - - HLE::Reschedule(__func__); } void Timer::Clear() { diff --git a/src/core/hle/service/act/act.cpp b/src/core/hle/service/act/act.cpp new file mode 100644 index 000000000..9600036c0 --- /dev/null +++ b/src/core/hle/service/act/act.cpp @@ -0,0 +1,18 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/act/act.h" +#include "core/hle/service/act/act_a.h" +#include "core/hle/service/act/act_u.h" + +namespace Service { +namespace ACT { + +void Init() { + AddService(new ACT_A); + AddService(new ACT_U); +} + +} // namespace ACT +} // namespace Service diff --git a/src/core/hle/service/act/act.h b/src/core/hle/service/act/act.h new file mode 100644 index 000000000..1425291aa --- /dev/null +++ b/src/core/hle/service/act/act.h @@ -0,0 +1,14 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +namespace Service { +namespace ACT { + +/// Initializes all ACT services +void Init(); + +} // namespace ACT +} // namespace Service diff --git a/src/core/hle/service/act_a.cpp b/src/core/hle/service/act/act_a.cpp index 9880aafff..5c523368f 100644 --- a/src/core/hle/service/act_a.cpp +++ b/src/core/hle/service/act/act_a.cpp @@ -2,7 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "core/hle/service/act_a.h" +#include "core/hle/service/act/act.h" +#include "core/hle/service/act/act_a.h" namespace Service { namespace ACT { diff --git a/src/core/hle/service/act_a.h b/src/core/hle/service/act/act_a.h index e3adb03e5..e3adb03e5 100644 --- a/src/core/hle/service/act_a.h +++ b/src/core/hle/service/act/act_a.h diff --git a/src/core/hle/service/act_u.cpp b/src/core/hle/service/act/act_u.cpp index b4f69c57d..cf98aa1d6 100644 --- a/src/core/hle/service/act_u.cpp +++ b/src/core/hle/service/act/act_u.cpp @@ -2,7 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "core/hle/service/act_u.h" +#include "core/hle/service/act/act.h" +#include "core/hle/service/act/act_u.h" namespace Service { namespace ACT { diff --git a/src/core/hle/service/act_u.h b/src/core/hle/service/act/act_u.h index 9d8538fbf..9d8538fbf 100644 --- a/src/core/hle/service/act_u.h +++ b/src/core/hle/service/act/act_u.h diff --git a/src/core/hle/service/cecd/cecd.cpp b/src/core/hle/service/cecd/cecd.cpp index 515b344e6..eb04273db 100644 --- a/src/core/hle/service/cecd/cecd.cpp +++ b/src/core/hle/service/cecd/cecd.cpp @@ -5,6 +5,7 @@ #include "common/logging/log.h" #include "core/hle/kernel/event.h" #include "core/hle/service/cecd/cecd.h" +#include "core/hle/service/cecd/cecd_ndm.h" #include "core/hle/service/cecd/cecd_s.h" #include "core/hle/service/cecd/cecd_u.h" #include "core/hle/service/service.h" @@ -43,12 +44,13 @@ void GetChangeStateEventHandle(Service::Interface* self) { } void Init() { - AddService(new CECD_S_Interface); - AddService(new CECD_U_Interface); + AddService(new CECD_NDM); + AddService(new CECD_S); + AddService(new CECD_U); - cecinfo_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "CECD_U::cecinfo_event"); + cecinfo_event = Kernel::Event::Create(Kernel::ResetType::OneShot, "CECD::cecinfo_event"); change_state_event = - Kernel::Event::Create(Kernel::ResetType::OneShot, "CECD_U::change_state_event"); + Kernel::Event::Create(Kernel::ResetType::OneShot, "CECD::change_state_event"); } void Shutdown() { diff --git a/src/core/hle/service/cecd/cecd_ndm.cpp b/src/core/hle/service/cecd/cecd_ndm.cpp new file mode 100644 index 000000000..7baf93750 --- /dev/null +++ b/src/core/hle/service/cecd/cecd_ndm.cpp @@ -0,0 +1,23 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/service/cecd/cecd.h" +#include "core/hle/service/cecd/cecd_ndm.h" + +namespace Service { +namespace CECD { + +static const Interface::FunctionInfo FunctionTable[] = { + {0x00010000, nullptr, "Initialize"}, + {0x00020000, nullptr, "Deinitialize"}, + {0x00030000, nullptr, "ResumeDaemon"}, + {0x00040040, nullptr, "SuspendDaemon"}, +}; + +CECD_NDM::CECD_NDM() { + Register(FunctionTable); +} + +} // namespace CECD +} // namespace Service diff --git a/src/core/hle/service/cecd/cecd_ndm.h b/src/core/hle/service/cecd/cecd_ndm.h new file mode 100644 index 000000000..2e2e50ada --- /dev/null +++ b/src/core/hle/service/cecd/cecd_ndm.h @@ -0,0 +1,22 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "core/hle/service/service.h" + +namespace Service { +namespace CECD { + +class CECD_NDM : public Interface { +public: + CECD_NDM(); + + std::string GetPortName() const override { + return "cecd:ndm"; + } +}; + +} // namespace CECD +} // namespace Service diff --git a/src/core/hle/service/cecd/cecd_s.cpp b/src/core/hle/service/cecd/cecd_s.cpp index 7477b9320..eacda7d41 100644 --- a/src/core/hle/service/cecd/cecd_s.cpp +++ b/src/core/hle/service/cecd/cecd_s.cpp @@ -2,16 +2,34 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include "core/hle/service/cecd/cecd.h" #include "core/hle/service/cecd/cecd_s.h" namespace Service { namespace CECD { -// Empty arrays are illegal -- commented out until an entry is added. -// const Interface::FunctionInfo FunctionTable[] = { }; +static const Interface::FunctionInfo FunctionTable[] = { + // cecd:u shared commands + {0x000100C2, nullptr, "OpenRawFile"}, + {0x00020042, nullptr, "ReadRawFile"}, + {0x00030104, nullptr, "ReadMessage"}, + {0x00040106, nullptr, "ReadMessageWithHMAC"}, + {0x00050042, nullptr, "WriteRawFile"}, + {0x00060104, nullptr, "WriteMessage"}, + {0x00070106, nullptr, "WriteMessageWithHMAC"}, + {0x00080102, nullptr, "Delete"}, + {0x000A00C4, nullptr, "GetSystemInfo"}, + {0x000B0040, nullptr, "RunCommand"}, + {0x000C0040, nullptr, "RunCommandAlt"}, + {0x000E0000, GetCecStateAbbreviated, "GetCecStateAbbreviated"}, + {0x000F0000, GetCecInfoEventHandle, "GetCecInfoEventHandle"}, + {0x00100000, GetChangeStateEventHandle, "GetChangeStateEventHandle"}, + {0x00110104, nullptr, "OpenAndWrite"}, + {0x00120104, nullptr, "OpenAndRead"}, +}; -CECD_S_Interface::CECD_S_Interface() { - // Register(FunctionTable); +CECD_S::CECD_S() { + Register(FunctionTable); } } // namespace CECD diff --git a/src/core/hle/service/cecd/cecd_s.h b/src/core/hle/service/cecd/cecd_s.h index df5c01849..ab6c6789a 100644 --- a/src/core/hle/service/cecd/cecd_s.h +++ b/src/core/hle/service/cecd/cecd_s.h @@ -9,9 +9,9 @@ namespace Service { namespace CECD { -class CECD_S_Interface : public Interface { +class CECD_S : public Interface { public: - CECD_S_Interface(); + CECD_S(); std::string GetPortName() const override { return "cecd:s"; diff --git a/src/core/hle/service/cecd/cecd_u.cpp b/src/core/hle/service/cecd/cecd_u.cpp index 7d98ba6e9..3ed864f0b 100644 --- a/src/core/hle/service/cecd/cecd_u.cpp +++ b/src/core/hle/service/cecd/cecd_u.cpp @@ -9,6 +9,7 @@ namespace Service { namespace CECD { static const Interface::FunctionInfo FunctionTable[] = { + // cecd:u shared commands {0x000100C2, nullptr, "OpenRawFile"}, {0x00020042, nullptr, "ReadRawFile"}, {0x00030104, nullptr, "ReadMessage"}, @@ -27,7 +28,7 @@ static const Interface::FunctionInfo FunctionTable[] = { {0x00120104, nullptr, "OpenAndRead"}, }; -CECD_U_Interface::CECD_U_Interface() { +CECD_U::CECD_U() { Register(FunctionTable); } diff --git a/src/core/hle/service/cecd/cecd_u.h b/src/core/hle/service/cecd/cecd_u.h index 394030ffc..16e874ff5 100644 --- a/src/core/hle/service/cecd/cecd_u.h +++ b/src/core/hle/service/cecd/cecd_u.h @@ -9,9 +9,9 @@ namespace Service { namespace CECD { -class CECD_U_Interface : public Interface { +class CECD_U : public Interface { public: - CECD_U_Interface(); + CECD_U(); std::string GetPortName() const override { return "cecd:u"; diff --git a/src/core/hle/service/dlp/dlp.h b/src/core/hle/service/dlp/dlp.h index ec2fe46e8..3185fe322 100644 --- a/src/core/hle/service/dlp/dlp.h +++ b/src/core/hle/service/dlp/dlp.h @@ -2,6 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#pragma once + namespace Service { namespace DLP { diff --git a/src/core/hle/service/fs/archive.cpp b/src/core/hle/service/fs/archive.cpp index f21934108..09205e4b2 100644 --- a/src/core/hle/service/fs/archive.cpp +++ b/src/core/hle/service/fs/archive.cpp @@ -104,8 +104,8 @@ void File::HandleSyncRequest(Kernel::SharedPtr<Kernel::ServerSession> server_ses u64 offset = cmd_buff[1] | ((u64)cmd_buff[2]) << 32; u32 length = cmd_buff[3]; u32 address = cmd_buff[5]; - LOG_TRACE(Service_FS, "Read %s %s: offset=0x%llx length=%d address=0x%x", - GetTypeName().c_str(), GetName().c_str(), offset, length, address); + LOG_TRACE(Service_FS, "Read %s: offset=0x%llx length=%d address=0x%x", GetName().c_str(), + offset, length, address); if (offset + length > backend->GetSize()) { LOG_ERROR(Service_FS, @@ -130,8 +130,8 @@ void File::HandleSyncRequest(Kernel::SharedPtr<Kernel::ServerSession> server_ses u32 length = cmd_buff[3]; u32 flush = cmd_buff[4]; u32 address = cmd_buff[6]; - LOG_TRACE(Service_FS, "Write %s %s: offset=0x%llx length=%d address=0x%x, flush=0x%x", - GetTypeName().c_str(), GetName().c_str(), offset, length, address, flush); + LOG_TRACE(Service_FS, "Write %s: offset=0x%llx length=%d address=0x%x, flush=0x%x", + GetName().c_str(), offset, length, address, flush); std::vector<u8> data(length); Memory::ReadBlock(address, data.data(), data.size()); @@ -145,7 +145,7 @@ void File::HandleSyncRequest(Kernel::SharedPtr<Kernel::ServerSession> server_ses } case FileCommand::GetSize: { - LOG_TRACE(Service_FS, "GetSize %s %s", GetTypeName().c_str(), GetName().c_str()); + LOG_TRACE(Service_FS, "GetSize %s", GetName().c_str()); u64 size = backend->GetSize(); cmd_buff[2] = (u32)size; cmd_buff[3] = size >> 32; @@ -154,14 +154,13 @@ void File::HandleSyncRequest(Kernel::SharedPtr<Kernel::ServerSession> server_ses case FileCommand::SetSize: { u64 size = cmd_buff[1] | ((u64)cmd_buff[2] << 32); - LOG_TRACE(Service_FS, "SetSize %s %s size=%llu", GetTypeName().c_str(), GetName().c_str(), - size); + LOG_TRACE(Service_FS, "SetSize %s size=%llu", GetName().c_str(), size); backend->SetSize(size); break; } case FileCommand::Close: { - LOG_TRACE(Service_FS, "Close %s %s", GetTypeName().c_str(), GetName().c_str()); + LOG_TRACE(Service_FS, "Close %s", GetName().c_str()); backend->Close(); break; } @@ -219,8 +218,7 @@ void Directory::HandleSyncRequest(Kernel::SharedPtr<Kernel::ServerSession> serve u32 count = cmd_buff[1]; u32 address = cmd_buff[3]; std::vector<FileSys::Entry> entries(count); - LOG_TRACE(Service_FS, "Read %s %s: count=%d", GetTypeName().c_str(), GetName().c_str(), - count); + LOG_TRACE(Service_FS, "Read %s: count=%d", GetName().c_str(), count); // Number of entries actually read u32 read = backend->Read(entries.size(), entries.data()); @@ -230,7 +228,7 @@ void Directory::HandleSyncRequest(Kernel::SharedPtr<Kernel::ServerSession> serve } case DirectoryCommand::Close: { - LOG_TRACE(Service_FS, "Close %s %s", GetTypeName().c_str(), GetName().c_str()); + LOG_TRACE(Service_FS, "Close %s", GetName().c_str()); backend->Close(); break; } diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp index 99baded11..18a1b6a16 100644 --- a/src/core/hle/service/hid/hid.cpp +++ b/src/core/hle/service/hid/hid.cpp @@ -37,7 +37,8 @@ static int enable_gyroscope_count = 0; // positive means enabled static PadState GetCirclePadDirectionState(s16 circle_pad_x, s16 circle_pad_y) { // 30 degree and 60 degree are angular thresholds for directions - constexpr float TAN30 = 0.577350269, TAN60 = 1 / TAN30; + constexpr float TAN30 = 0.577350269f; + constexpr float TAN60 = 1 / TAN30; // a circle pad radius greater than 40 will trigger circle pad direction constexpr int CIRCLE_PAD_THRESHOLD_SQUARE = 40 * 40; PadState state; diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index 2bc3fdc82..25a7aeea8 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -9,8 +9,7 @@ #include "core/hle/kernel/server_port.h" #include "core/hle/service/ac_u.h" -#include "core/hle/service/act_a.h" -#include "core/hle/service/act_u.h" +#include "core/hle/service/act/act.h" #include "core/hle/service/am/am.h" #include "core/hle/service/apt/apt.h" #include "core/hle/service/boss/boss.h" @@ -139,6 +138,7 @@ void Init() { AddNamedPort(new ERR::ERR_F); FS::ArchiveInit(); + ACT::Init(); AM::Init(); APT::Init(); BOSS::Init(); @@ -158,8 +158,6 @@ void Init() { QTM::Init(); AddService(new AC::AC_U); - AddService(new ACT::ACT_A); - AddService(new ACT::ACT_U); AddService(new CSND::CSND_SND); AddService(new DSP_DSP::Interface); AddService(new GSP::GSP_GPU); diff --git a/src/core/hle/service/srv.h b/src/core/hle/service/srv.h index 6041ca42d..d3a9de879 100644 --- a/src/core/hle/service/srv.h +++ b/src/core/hle/service/srv.h @@ -2,6 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#pragma once + #include "core/hle/service/service.h" namespace Service { diff --git a/src/core/hle/svc.cpp b/src/core/hle/svc.cpp index e5ba9a484..5839d7230 100644 --- a/src/core/hle/svc.cpp +++ b/src/core/hle/svc.cpp @@ -43,6 +43,9 @@ const ResultCode ERR_PORT_NAME_TOO_LONG(ErrorDescription(30), ErrorModule::OS, ErrorSummary::InvalidArgument, ErrorLevel::Usage); // 0xE0E0181E +const ResultCode ERR_SYNC_TIMEOUT(ErrorDescription::Timeout, ErrorModule::OS, + ErrorSummary::StatusChanged, ErrorLevel::Info); + const ResultCode ERR_MISALIGNED_ADDRESS{// 0xE0E01BF1 ErrorDescription::MisalignedAddress, ErrorModule::OS, ErrorSummary::InvalidArgument, ErrorLevel::Usage}; @@ -260,27 +263,30 @@ static ResultCode WaitSynchronization1(Handle handle, s64 nano_seconds) { auto object = Kernel::g_handle_table.GetWaitObject(handle); Kernel::Thread* thread = Kernel::GetCurrentThread(); - thread->waitsynch_waited = false; - if (object == nullptr) return ERR_INVALID_HANDLE; LOG_TRACE(Kernel_SVC, "called handle=0x%08X(%s:%s), nanoseconds=%lld", handle, object->GetTypeName().c_str(), object->GetName().c_str(), nano_seconds); - HLE::Reschedule(__func__); - - // Check for next thread to schedule if (object->ShouldWait()) { + if (nano_seconds == 0) + return ERR_SYNC_TIMEOUT; + object->AddWaitingThread(thread); - Kernel::WaitCurrentThread_WaitSynchronization({object}, false, false); + // TODO(Subv): Perform things like update the mutex lock owner's priority to + // prevent priority inversion. Currently this is done in Mutex::ShouldWait, + // but it should be moved to a function that is called from here. + thread->status = THREADSTATUS_WAIT_SYNCH; // Create an event to wake the thread up after the specified nanosecond delay has passed thread->WakeAfterDelay(nano_seconds); - // NOTE: output of this SVC will be set later depending on how the thread resumes - return HLE::RESULT_INVALID; + // Note: The output of this SVC will be set to RESULT_SUCCESS if the thread + // resumes due to a signal in its wait objects. + // Otherwise we retain the default value of timeout. + return ERR_SYNC_TIMEOUT; } object->Acquire(); @@ -291,11 +297,7 @@ static ResultCode WaitSynchronization1(Handle handle, s64 nano_seconds) { /// Wait for the given handles to synchronize, timeout after the specified nanoseconds static ResultCode WaitSynchronizationN(s32* out, Handle* handles, s32 handle_count, bool wait_all, s64 nano_seconds) { - bool wait_thread = !wait_all; - int handle_index = 0; Kernel::Thread* thread = Kernel::GetCurrentThread(); - bool was_waiting = thread->waitsynch_waited; - thread->waitsynch_waited = false; // Check if 'handles' is invalid if (handles == nullptr) @@ -311,90 +313,113 @@ static ResultCode WaitSynchronizationN(s32* out, Handle* handles, s32 handle_cou return ResultCode(ErrorDescription::OutOfRange, ErrorModule::OS, ErrorSummary::InvalidArgument, ErrorLevel::Usage); - // If 'handle_count' is non-zero, iterate through each handle and wait the current thread if - // necessary - if (handle_count != 0) { - bool selected = false; // True once an object has been selected - - Kernel::SharedPtr<Kernel::WaitObject> wait_object; - - for (int i = 0; i < handle_count; ++i) { - auto object = Kernel::g_handle_table.GetWaitObject(handles[i]); - if (object == nullptr) - return ERR_INVALID_HANDLE; - - // Check if the current thread should wait on this object... - if (object->ShouldWait()) { - - // Check we are waiting on all objects... - if (wait_all) - // Wait the thread - wait_thread = true; - } else { - // Do not wait on this object, check if this object should be selected... - if (!wait_all && (!selected || (wait_object == object && was_waiting))) { - // Do not wait the thread - wait_thread = false; - handle_index = i; - wait_object = object; - selected = true; - } - } - } - } else { - // If no handles were passed in, put the thread to sleep only when 'wait_all' is false - // NOTE: This should deadlock the current thread if no timeout was specified - if (!wait_all) { - wait_thread = true; - } + using ObjectPtr = Kernel::SharedPtr<Kernel::WaitObject>; + std::vector<ObjectPtr> objects(handle_count); + + for (int i = 0; i < handle_count; ++i) { + auto object = Kernel::g_handle_table.GetWaitObject(handles[i]); + if (object == nullptr) + return ERR_INVALID_HANDLE; + objects[i] = object; } - SCOPE_EXIT({ - HLE::Reschedule("WaitSynchronizationN"); - }); // Reschedule after putting the threads to sleep. + // Clear the mapping of wait object indices. + // We don't want any lingering state in this map. + // It will be repopulated later in the wait_all = false case. + thread->wait_objects_index.clear(); + + if (wait_all) { + bool all_available = + std::all_of(objects.begin(), objects.end(), + [](const ObjectPtr& object) { return !object->ShouldWait(); }); + if (all_available) { + // We can acquire all objects right now, do so. + for (auto& object : objects) + object->Acquire(); + // Note: In this case, the `out` parameter is not set, + // and retains whatever value it had before. + return RESULT_SUCCESS; + } + + // Not all objects were available right now, prepare to suspend the thread. - // If thread should wait, then set its state to waiting - if (wait_thread) { + // If a timeout value of 0 was provided, just return the Timeout error code instead of + // suspending the thread. + if (nano_seconds == 0) + return ERR_SYNC_TIMEOUT; - // Actually wait the current thread on each object if we decided to wait... - std::vector<SharedPtr<Kernel::WaitObject>> wait_objects; - wait_objects.reserve(handle_count); + // Put the thread to sleep + thread->status = THREADSTATUS_WAIT_SYNCH; - for (int i = 0; i < handle_count; ++i) { - auto object = Kernel::g_handle_table.GetWaitObject(handles[i]); - object->AddWaitingThread(Kernel::GetCurrentThread()); - wait_objects.push_back(object); + // Add the thread to each of the objects' waiting threads. + for (auto& object : objects) { + object->AddWaitingThread(thread); + // TODO(Subv): Perform things like update the mutex lock owner's priority to + // prevent priority inversion. Currently this is done in Mutex::ShouldWait, + // but it should be moved to a function that is called from here. } - Kernel::WaitCurrentThread_WaitSynchronization(std::move(wait_objects), true, wait_all); + // Set the thread's waitlist to the list of objects passed to WaitSynchronizationN + thread->wait_objects = std::move(objects); // Create an event to wake the thread up after the specified nanosecond delay has passed - Kernel::GetCurrentThread()->WakeAfterDelay(nano_seconds); - - // NOTE: output of this SVC will be set later depending on how the thread resumes - return HLE::RESULT_INVALID; - } + thread->WakeAfterDelay(nano_seconds); - // Acquire objects if we did not wait... - for (int i = 0; i < handle_count; ++i) { - auto object = Kernel::g_handle_table.GetWaitObject(handles[i]); + // This value gets set to -1 by default in this case, it is not modified after this. + *out = -1; + // Note: The output of this SVC will be set to RESULT_SUCCESS if the thread resumes due to + // a signal in one of its wait objects. + return ERR_SYNC_TIMEOUT; + } else { + // Find the first object that is acquirable in the provided list of objects + auto itr = std::find_if(objects.begin(), objects.end(), + [](const ObjectPtr& object) { return !object->ShouldWait(); }); - // Acquire the object if it is not waiting... - if (!object->ShouldWait()) { + if (itr != objects.end()) { + // We found a ready object, acquire it and set the result value + Kernel::WaitObject* object = itr->get(); object->Acquire(); + *out = std::distance(objects.begin(), itr); + return RESULT_SUCCESS; + } + + // No objects were ready to be acquired, prepare to suspend the thread. + + // If a timeout value of 0 was provided, just return the Timeout error code instead of + // suspending the thread. + if (nano_seconds == 0) + return ERR_SYNC_TIMEOUT; + + // Put the thread to sleep + thread->status = THREADSTATUS_WAIT_SYNCH; - // If this was the first non-waiting object and 'wait_all' is false, don't acquire - // any other objects - if (!wait_all) - break; + // Clear the thread's waitlist, we won't use it for wait_all = false + thread->wait_objects.clear(); + + // Add the thread to each of the objects' waiting threads. + for (size_t i = 0; i < objects.size(); ++i) { + Kernel::WaitObject* object = objects[i].get(); + // Set the index of this object in the mapping of Objects -> index for this thread. + thread->wait_objects_index[object->GetObjectId()] = static_cast<int>(i); + object->AddWaitingThread(thread); + // TODO(Subv): Perform things like update the mutex lock owner's priority to + // prevent priority inversion. Currently this is done in Mutex::ShouldWait, + // but it should be moved to a function that is called from here. } - } - // TODO(bunnei): If 'wait_all' is true, this is probably wrong. However, real hardware does - // not seem to set it to any meaningful value. - *out = handle_count != 0 ? (wait_all ? -1 : handle_index) : 0; + // Note: If no handles and no timeout were given, then the thread will deadlock, this is + // consistent with hardware behavior. - return RESULT_SUCCESS; + // Create an event to wake the thread up after the specified nanosecond delay has passed + thread->WakeAfterDelay(nano_seconds); + + // Note: The output of this SVC will be set to RESULT_SUCCESS if the thread resumes due to a + // signal in one of its wait objects. + // Otherwise we retain the default value of timeout, and -1 in the out parameter + thread->wait_set_output = true; + *out = -1; + return ERR_SYNC_TIMEOUT; + } } /// Create an address arbiter (to allocate access to shared resources) @@ -559,7 +584,7 @@ static ResultCode CreateThread(Handle* out_handle, s32 priority, u32 entry_point static void ExitThread() { LOG_TRACE(Kernel_SVC, "called, pc=0x%08X", Core::g_app_core->GetPC()); - Kernel::GetCurrentThread()->Stop(); + Kernel::ExitCurrentThread(); } /// Gets the priority for the specified thread @@ -1159,6 +1184,8 @@ void CallSVC(u32 immediate) { if (info) { if (info->func) { info->func(); + // TODO(Subv): Not all service functions should cause a reschedule in all cases. + HLE::Reschedule(__func__); } else { LOG_ERROR(Kernel_SVC, "unimplemented SVC function %s(..)", info->name); } diff --git a/src/core/loader/loader.h b/src/core/loader/loader.h index 5e3d46638..a6c2a745f 100644 --- a/src/core/loader/loader.h +++ b/src/core/loader/loader.h @@ -144,6 +144,15 @@ public: } /** + * Get the program id of the application + * @param out_program_id Reference to store program id into + * @return ResultStatus result of function + */ + virtual ResultStatus ReadProgramId(u64& out_program_id) { + return ResultStatus::ErrorNotImplemented; + } + + /** * Get the RomFS of the application * Since the RomFS can be huge, we return a file reference instead of copying to a buffer * @param romfs_file The file containing the RomFS diff --git a/src/core/loader/ncch.cpp b/src/core/loader/ncch.cpp index d4be61e0e..6f2164428 100644 --- a/src/core/loader/ncch.cpp +++ b/src/core/loader/ncch.cpp @@ -344,6 +344,18 @@ ResultStatus AppLoader_NCCH::ReadLogo(std::vector<u8>& buffer) { return LoadSectionExeFS("logo", buffer); } +ResultStatus AppLoader_NCCH::ReadProgramId(u64& out_program_id) { + if (!file.IsOpen()) + return ResultStatus::Error; + + ResultStatus result = LoadExeFS(); + if (result != ResultStatus::Success) + return result; + + out_program_id = ncch_header.program_id; + return ResultStatus::Success; +} + ResultStatus AppLoader_NCCH::ReadRomFS(std::shared_ptr<FileUtil::IOFile>& romfs_file, u64& offset, u64& size) { if (!file.IsOpen()) diff --git a/src/core/loader/ncch.h b/src/core/loader/ncch.h index bcf3ae6e3..6c93d46d8 100644 --- a/src/core/loader/ncch.h +++ b/src/core/loader/ncch.h @@ -220,6 +220,13 @@ public: ResultStatus ReadLogo(std::vector<u8>& buffer) override; /** + * Get the program id of the application + * @param out_program_id Reference to store program id into + * @return ResultStatus result of function + */ + ResultStatus ReadProgramId(u64& out_program_id) override; + + /** * Get the RomFS of the application * @param romfs_file Reference to buffer to store data * @param offset Offset in the file to the RomFS diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index 9aa446a8f..6ca319b59 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -39,6 +39,7 @@ set(HEADERS rasterizer.h rasterizer_interface.h renderer_base.h + shader/debug_data.h shader/shader.h shader/shader_interpreter.h swrasterizer.h diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp index db4fe659c..ea58e9f54 100644 --- a/src/video_core/command_processor.cpp +++ b/src/video_core/command_processor.cpp @@ -59,7 +59,10 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { regs[id] = (old_value & ~write_mask) | (value & write_mask); - DebugUtils::OnPicaRegWrite({(u16)id, (u16)mask, regs[id]}); + // Double check for is_pica_tracing to avoid call overhead + if (DebugUtils::IsPicaTracing()) { + DebugUtils::OnPicaRegWrite({(u16)id, (u16)mask, regs[id]}); + } if (g_debug_context) g_debug_context->OnEvent(DebugContext::Event::PicaCommandLoaded, @@ -139,7 +142,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { MICROPROFILE_SCOPE(GPU_Drawing); immediate_attribute_id = 0; - Shader::UnitState<false> shader_unit; + Shader::UnitState shader_unit; g_state.vs.Setup(); // Send to vertex shader @@ -240,7 +243,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) { unsigned int vertex_cache_pos = 0; vertex_cache_ids.fill(-1); - Shader::UnitState<false> shader_unit; + Shader::UnitState shader_unit; g_state.vs.Setup(); for (unsigned int index = 0; index < regs.num_vertices; ++index) { diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp index 8806464d9..c44b3d95a 100644 --- a/src/video_core/debug_utils/debug_utils.cpp +++ b/src/video_core/debug_utils/debug_utils.cpp @@ -276,10 +276,10 @@ void DumpShader(const std::string& filename, const Regs::ShaderConfig& config, static std::unique_ptr<PicaTrace> pica_trace; static std::mutex pica_trace_mutex; -static int is_pica_tracing = false; +bool g_is_pica_tracing = false; void StartPicaTracing() { - if (is_pica_tracing) { + if (g_is_pica_tracing) { LOG_WARNING(HW_GPU, "StartPicaTracing called even though tracing already running!"); return; } @@ -287,34 +287,26 @@ void StartPicaTracing() { std::lock_guard<std::mutex> lock(pica_trace_mutex); pica_trace = std::make_unique<PicaTrace>(); - is_pica_tracing = true; -} - -bool IsPicaTracing() { - return is_pica_tracing != 0; + g_is_pica_tracing = true; } void OnPicaRegWrite(PicaTrace::Write write) { - // Double check for is_pica_tracing to avoid pointless locking overhead - if (!is_pica_tracing) - return; - std::lock_guard<std::mutex> lock(pica_trace_mutex); - if (!is_pica_tracing) + if (!g_is_pica_tracing) return; pica_trace->writes.push_back(write); } std::unique_ptr<PicaTrace> FinishPicaTracing() { - if (!is_pica_tracing) { + if (!g_is_pica_tracing) { LOG_WARNING(HW_GPU, "FinishPicaTracing called even though tracing isn't running!"); return {}; } // signalize that no further tracing should be performed - is_pica_tracing = false; + g_is_pica_tracing = false; // Wait until running tracing is finished std::lock_guard<std::mutex> lock(pica_trace_mutex); diff --git a/src/video_core/debug_utils/debug_utils.h b/src/video_core/debug_utils/debug_utils.h index 189c93abb..46ea8d9c7 100644 --- a/src/video_core/debug_utils/debug_utils.h +++ b/src/video_core/debug_utils/debug_utils.h @@ -196,8 +196,12 @@ struct PicaTrace { std::vector<Write> writes; }; +extern bool g_is_pica_tracing; + void StartPicaTracing(); -bool IsPicaTracing(); +inline bool IsPicaTracing() { + return g_is_pica_tracing; +} void OnPicaRegWrite(PicaTrace::Write write); std::unique_ptr<PicaTrace> FinishPicaTracing(); diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 22b8a0db2..5a306a5c8 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -886,10 +886,10 @@ bool RasterizerOpenGL::AccelerateFill(const GPU::Regs::MemoryFillConfig& config) } } - cur_state.color_mask.red_enabled = true; - cur_state.color_mask.green_enabled = true; - cur_state.color_mask.blue_enabled = true; - cur_state.color_mask.alpha_enabled = true; + cur_state.color_mask.red_enabled = GL_TRUE; + cur_state.color_mask.green_enabled = GL_TRUE; + cur_state.color_mask.blue_enabled = GL_TRUE; + cur_state.color_mask.alpha_enabled = GL_TRUE; cur_state.Apply(); glClearBufferfv(GL_COLOR, 0, color_values); } else if (dst_type == SurfaceType::Depth) { diff --git a/src/video_core/shader/debug_data.h b/src/video_core/shader/debug_data.h new file mode 100644 index 000000000..9e82122e1 --- /dev/null +++ b/src/video_core/shader/debug_data.h @@ -0,0 +1,186 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <vector> +#include "common/common_types.h" +#include "common/vector_math.h" +#include "video_core/pica_types.h" + +namespace Pica { +namespace Shader { + +/// Helper structure used to keep track of data useful for inspection of shader emulation +template <bool full_debugging> +struct DebugData; + +template <> +struct DebugData<false> { + // TODO: Hide these behind and interface and move them to DebugData<true> + u32 max_offset = 0; ///< maximum program counter ever reached + u32 max_opdesc_id = 0; ///< maximum swizzle pattern index ever used +}; + +template <> +struct DebugData<true> { + /// Records store the input and output operands of a particular instruction. + struct Record { + enum Type { + // Floating point arithmetic operands + SRC1 = 0x1, + SRC2 = 0x2, + SRC3 = 0x4, + + // Initial and final output operand value + DEST_IN = 0x8, + DEST_OUT = 0x10, + + // Current and next instruction offset (in words) + CUR_INSTR = 0x20, + NEXT_INSTR = 0x40, + + // Output address register value + ADDR_REG_OUT = 0x80, + + // Result of a comparison instruction + CMP_RESULT = 0x100, + + // Input values for conditional flow control instructions + COND_BOOL_IN = 0x200, + COND_CMP_IN = 0x400, + + // Input values for a loop + LOOP_INT_IN = 0x800, + }; + + Math::Vec4<float24> src1; + Math::Vec4<float24> src2; + Math::Vec4<float24> src3; + + Math::Vec4<float24> dest_in; + Math::Vec4<float24> dest_out; + + s32 address_registers[2]; + bool conditional_code[2]; + bool cond_bool; + bool cond_cmp[2]; + Math::Vec4<u8> loop_int; + + u32 instruction_offset; + u32 next_instruction; + + /// set of enabled fields (as a combination of Type flags) + unsigned mask = 0; + }; + + u32 max_offset = 0; ///< maximum program counter ever reached + u32 max_opdesc_id = 0; ///< maximum swizzle pattern index ever used + + /// List of records for each executed shader instruction + std::vector<DebugData<true>::Record> records; +}; + +/// Type alias for better readability +using DebugDataRecord = DebugData<true>::Record; + +/// Helper function to set a DebugData<true>::Record field based on the template enum parameter. +template <DebugDataRecord::Type type, typename ValueType> +inline void SetField(DebugDataRecord& record, ValueType value); + +template <> +inline void SetField<DebugDataRecord::SRC1>(DebugDataRecord& record, float24* value) { + record.src1.x = value[0]; + record.src1.y = value[1]; + record.src1.z = value[2]; + record.src1.w = value[3]; +} + +template <> +inline void SetField<DebugDataRecord::SRC2>(DebugDataRecord& record, float24* value) { + record.src2.x = value[0]; + record.src2.y = value[1]; + record.src2.z = value[2]; + record.src2.w = value[3]; +} + +template <> +inline void SetField<DebugDataRecord::SRC3>(DebugDataRecord& record, float24* value) { + record.src3.x = value[0]; + record.src3.y = value[1]; + record.src3.z = value[2]; + record.src3.w = value[3]; +} + +template <> +inline void SetField<DebugDataRecord::DEST_IN>(DebugDataRecord& record, float24* value) { + record.dest_in.x = value[0]; + record.dest_in.y = value[1]; + record.dest_in.z = value[2]; + record.dest_in.w = value[3]; +} + +template <> +inline void SetField<DebugDataRecord::DEST_OUT>(DebugDataRecord& record, float24* value) { + record.dest_out.x = value[0]; + record.dest_out.y = value[1]; + record.dest_out.z = value[2]; + record.dest_out.w = value[3]; +} + +template <> +inline void SetField<DebugDataRecord::ADDR_REG_OUT>(DebugDataRecord& record, s32* value) { + record.address_registers[0] = value[0]; + record.address_registers[1] = value[1]; +} + +template <> +inline void SetField<DebugDataRecord::CMP_RESULT>(DebugDataRecord& record, bool* value) { + record.conditional_code[0] = value[0]; + record.conditional_code[1] = value[1]; +} + +template <> +inline void SetField<DebugDataRecord::COND_BOOL_IN>(DebugDataRecord& record, bool value) { + record.cond_bool = value; +} + +template <> +inline void SetField<DebugDataRecord::COND_CMP_IN>(DebugDataRecord& record, bool* value) { + record.cond_cmp[0] = value[0]; + record.cond_cmp[1] = value[1]; +} + +template <> +inline void SetField<DebugDataRecord::LOOP_INT_IN>(DebugDataRecord& record, Math::Vec4<u8> value) { + record.loop_int = value; +} + +template <> +inline void SetField<DebugDataRecord::CUR_INSTR>(DebugDataRecord& record, u32 value) { + record.instruction_offset = value; +} + +template <> +inline void SetField<DebugDataRecord::NEXT_INSTR>(DebugDataRecord& record, u32 value) { + record.next_instruction = value; +} + +/// Helper function to set debug information on the current shader iteration. +template <DebugDataRecord::Type type, typename ValueType> +inline void Record(DebugData<false>& debug_data, u32 offset, ValueType value) { + // Debugging disabled => nothing to do +} + +template <DebugDataRecord::Type type, typename ValueType> +inline void Record(DebugData<true>& debug_data, u32 offset, ValueType value) { + if (offset >= debug_data.records.size()) + debug_data.records.resize(offset + 1); + + SetField<type, ValueType>(debug_data.records[offset], value); + debug_data.records[offset].mask |= type; +} + +} // namespace Shader +} // namespace Pica diff --git a/src/video_core/shader/shader.cpp b/src/video_core/shader/shader.cpp index c7f23dab9..a4aa3c9e0 100644 --- a/src/video_core/shader/shader.cpp +++ b/src/video_core/shader/shader.cpp @@ -109,15 +109,12 @@ void ShaderSetup::Setup() { MICROPROFILE_DEFINE(GPU_Shader, "GPU", "Shader", MP_RGB(50, 50, 240)); -void ShaderSetup::Run(UnitState<false>& state, const InputVertex& input, int num_attributes) { +void ShaderSetup::Run(UnitState& state, const InputVertex& input, int num_attributes) { auto& config = g_state.regs.vs; auto& setup = g_state.vs; MICROPROFILE_SCOPE(GPU_Shader); - state.debug.max_offset = 0; - state.debug.max_opdesc_id = 0; - // Setup input register table const auto& attribute_register_map = config.input_register_map; @@ -128,22 +125,23 @@ void ShaderSetup::Run(UnitState<false>& state, const InputVertex& input, int num state.conditional_code[1] = false; #ifdef ARCHITECTURE_x86_64 - if (VideoCore::g_shader_jit_enabled) + if (VideoCore::g_shader_jit_enabled) { jit_shader->Run(setup, state, config.main_offset); - else - RunInterpreter(setup, state, config.main_offset); + } else { + DebugData<false> dummy_debug_data; + RunInterpreter(setup, state, dummy_debug_data, config.main_offset); + } #else - RunInterpreter(setup, state, config.main_offset); + DebugData<false> dummy_debug_data; + RunInterpreter(setup, state, dummy_debug_data, config.main_offset); #endif // ARCHITECTURE_x86_64 } DebugData<true> ShaderSetup::ProduceDebugInfo(const InputVertex& input, int num_attributes, const Regs::ShaderConfig& config, const ShaderSetup& setup) { - UnitState<true> state; - - state.debug.max_offset = 0; - state.debug.max_opdesc_id = 0; + UnitState state; + DebugData<true> debug_data; // Setup input register table boost::fill(state.registers.input, Math::Vec4<float24>::AssignToAll(float24::Zero())); @@ -154,8 +152,8 @@ DebugData<true> ShaderSetup::ProduceDebugInfo(const InputVertex& input, int num_ state.conditional_code[0] = false; state.conditional_code[1] = false; - RunInterpreter(setup, state, config.main_offset); - return state.debug; + RunInterpreter(setup, state, debug_data, config.main_offset); + return debug_data; } } // namespace Shader diff --git a/src/video_core/shader/shader.h b/src/video_core/shader/shader.h index 0111d8c0f..2b07759b9 100644 --- a/src/video_core/shader/shader.h +++ b/src/video_core/shader/shader.h @@ -8,8 +8,6 @@ #include <cstddef> #include <memory> #include <type_traits> -#include <vector> -#include <boost/container/static_vector.hpp> #include <nihstro/shader_bytecode.h> #include "common/assert.h" #include "common/common_funcs.h" @@ -17,6 +15,7 @@ #include "common/vector_math.h" #include "video_core/pica.h" #include "video_core/pica_types.h" +#include "video_core/shader/debug_data.h" using nihstro::RegisterType; using nihstro::SourceRegister; @@ -89,183 +88,12 @@ struct OutputRegisters { }; static_assert(std::is_pod<OutputRegisters>::value, "Structure is not POD"); -// Helper structure used to keep track of data useful for inspection of shader emulation -template <bool full_debugging> -struct DebugData; - -template <> -struct DebugData<false> { - // TODO: Hide these behind and interface and move them to DebugData<true> - u32 max_offset; // maximum program counter ever reached - u32 max_opdesc_id; // maximum swizzle pattern index ever used -}; - -template <> -struct DebugData<true> { - // Records store the input and output operands of a particular instruction. - struct Record { - enum Type { - // Floating point arithmetic operands - SRC1 = 0x1, - SRC2 = 0x2, - SRC3 = 0x4, - - // Initial and final output operand value - DEST_IN = 0x8, - DEST_OUT = 0x10, - - // Current and next instruction offset (in words) - CUR_INSTR = 0x20, - NEXT_INSTR = 0x40, - - // Output address register value - ADDR_REG_OUT = 0x80, - - // Result of a comparison instruction - CMP_RESULT = 0x100, - - // Input values for conditional flow control instructions - COND_BOOL_IN = 0x200, - COND_CMP_IN = 0x400, - - // Input values for a loop - LOOP_INT_IN = 0x800, - }; - - Math::Vec4<float24> src1; - Math::Vec4<float24> src2; - Math::Vec4<float24> src3; - - Math::Vec4<float24> dest_in; - Math::Vec4<float24> dest_out; - - s32 address_registers[2]; - bool conditional_code[2]; - bool cond_bool; - bool cond_cmp[2]; - Math::Vec4<u8> loop_int; - - u32 instruction_offset; - u32 next_instruction; - - // set of enabled fields (as a combination of Type flags) - unsigned mask = 0; - }; - - u32 max_offset; // maximum program counter ever reached - u32 max_opdesc_id; // maximum swizzle pattern index ever used - - // List of records for each executed shader instruction - std::vector<DebugData<true>::Record> records; -}; - -// Type alias for better readability -using DebugDataRecord = DebugData<true>::Record; - -// Helper function to set a DebugData<true>::Record field based on the template enum parameter. -template <DebugDataRecord::Type type, typename ValueType> -inline void SetField(DebugDataRecord& record, ValueType value); - -template <> -inline void SetField<DebugDataRecord::SRC1>(DebugDataRecord& record, float24* value) { - record.src1.x = value[0]; - record.src1.y = value[1]; - record.src1.z = value[2]; - record.src1.w = value[3]; -} - -template <> -inline void SetField<DebugDataRecord::SRC2>(DebugDataRecord& record, float24* value) { - record.src2.x = value[0]; - record.src2.y = value[1]; - record.src2.z = value[2]; - record.src2.w = value[3]; -} - -template <> -inline void SetField<DebugDataRecord::SRC3>(DebugDataRecord& record, float24* value) { - record.src3.x = value[0]; - record.src3.y = value[1]; - record.src3.z = value[2]; - record.src3.w = value[3]; -} - -template <> -inline void SetField<DebugDataRecord::DEST_IN>(DebugDataRecord& record, float24* value) { - record.dest_in.x = value[0]; - record.dest_in.y = value[1]; - record.dest_in.z = value[2]; - record.dest_in.w = value[3]; -} - -template <> -inline void SetField<DebugDataRecord::DEST_OUT>(DebugDataRecord& record, float24* value) { - record.dest_out.x = value[0]; - record.dest_out.y = value[1]; - record.dest_out.z = value[2]; - record.dest_out.w = value[3]; -} - -template <> -inline void SetField<DebugDataRecord::ADDR_REG_OUT>(DebugDataRecord& record, s32* value) { - record.address_registers[0] = value[0]; - record.address_registers[1] = value[1]; -} - -template <> -inline void SetField<DebugDataRecord::CMP_RESULT>(DebugDataRecord& record, bool* value) { - record.conditional_code[0] = value[0]; - record.conditional_code[1] = value[1]; -} - -template <> -inline void SetField<DebugDataRecord::COND_BOOL_IN>(DebugDataRecord& record, bool value) { - record.cond_bool = value; -} - -template <> -inline void SetField<DebugDataRecord::COND_CMP_IN>(DebugDataRecord& record, bool* value) { - record.cond_cmp[0] = value[0]; - record.cond_cmp[1] = value[1]; -} - -template <> -inline void SetField<DebugDataRecord::LOOP_INT_IN>(DebugDataRecord& record, Math::Vec4<u8> value) { - record.loop_int = value; -} - -template <> -inline void SetField<DebugDataRecord::CUR_INSTR>(DebugDataRecord& record, u32 value) { - record.instruction_offset = value; -} - -template <> -inline void SetField<DebugDataRecord::NEXT_INSTR>(DebugDataRecord& record, u32 value) { - record.next_instruction = value; -} - -// Helper function to set debug information on the current shader iteration. -template <DebugDataRecord::Type type, typename ValueType> -inline void Record(DebugData<false>& debug_data, u32 offset, ValueType value) { - // Debugging disabled => nothing to do -} - -template <DebugDataRecord::Type type, typename ValueType> -inline void Record(DebugData<true>& debug_data, u32 offset, ValueType value) { - if (offset >= debug_data.records.size()) - debug_data.records.resize(offset + 1); - - SetField<type, ValueType>(debug_data.records[offset], value); - debug_data.records[offset].mask |= type; -} - /** * This structure contains the state information that needs to be unique for a shader unit. The 3DS * has four shader units that process shaders in parallel. At the present, Citra only implements a * single shader unit that processes all shaders serially. Putting the state information in a struct * here will make it easier for us to parallelize the shader processing later. */ -template <bool Debug> struct UnitState { struct Registers { // The registers are accessed by the shader JIT using SSE instructions, and are therefore @@ -283,8 +111,6 @@ struct UnitState { // TODO: How many bits do these actually have? s32 address_registers[3]; - DebugData<Debug> debug; - static size_t InputOffset(const SourceRegister& reg) { switch (reg.GetRegisterType()) { case RegisterType::Input: @@ -332,21 +158,16 @@ struct ShaderSetup { std::array<Math::Vec4<u8>, 4> i; } uniforms; - static size_t UniformOffset(RegisterType type, unsigned index) { - switch (type) { - case RegisterType::FloatUniform: - return offsetof(ShaderSetup, uniforms.f) + index * sizeof(Math::Vec4<float24>); - - case RegisterType::BoolUniform: - return offsetof(ShaderSetup, uniforms.b) + index * sizeof(bool); + static size_t GetFloatUniformOffset(unsigned index) { + return offsetof(ShaderSetup, uniforms.f) + index * sizeof(Math::Vec4<float24>); + } - case RegisterType::IntUniform: - return offsetof(ShaderSetup, uniforms.i) + index * sizeof(Math::Vec4<u8>); + static size_t GetBoolUniformOffset(unsigned index) { + return offsetof(ShaderSetup, uniforms.b) + index * sizeof(bool); + } - default: - UNREACHABLE(); - return 0; - } + static size_t GetIntUniformOffset(unsigned index) { + return offsetof(ShaderSetup, uniforms.i) + index * sizeof(Math::Vec4<u8>); } std::array<u32, 1024> program_code; @@ -364,7 +185,7 @@ struct ShaderSetup { * @param input Input vertex into the shader * @param num_attributes The number of vertex shader attributes */ - void Run(UnitState<false>& state, const InputVertex& input, int num_attributes); + void Run(UnitState& state, const InputVertex& input, int num_attributes); /** * Produce debug information based on the given shader and input vertex diff --git a/src/video_core/shader/shader_interpreter.cpp b/src/video_core/shader/shader_interpreter.cpp index 6abb6761f..70db4167e 100644 --- a/src/video_core/shader/shader_interpreter.cpp +++ b/src/video_core/shader/shader_interpreter.cpp @@ -6,6 +6,7 @@ #include <array> #include <cmath> #include <numeric> +#include <boost/container/static_vector.hpp> #include <nihstro/shader_bytecode.h> #include "common/assert.h" #include "common/common_types.h" @@ -38,12 +39,42 @@ struct CallStackElement { }; template <bool Debug> -void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned offset) { +void RunInterpreter(const ShaderSetup& setup, UnitState& state, DebugData<Debug>& debug_data, + unsigned offset) { // TODO: Is there a maximal size for this? boost::container::static_vector<CallStackElement, 16> call_stack; - u32 program_counter = offset; + auto call = [&program_counter, &call_stack](u32 offset, u32 num_instructions, u32 return_offset, + u8 repeat_count, u8 loop_increment) { + // -1 to make sure when incrementing the PC we end up at the correct offset + program_counter = offset - 1; + ASSERT(call_stack.size() < call_stack.capacity()); + call_stack.push_back( + {offset + num_instructions, return_offset, repeat_count, loop_increment, offset}); + }; + + auto evaluate_condition = [&state](Instruction::FlowControlType flow_control) { + using Op = Instruction::FlowControlType::Op; + + bool result_x = flow_control.refx.Value() == state.conditional_code[0]; + bool result_y = flow_control.refy.Value() == state.conditional_code[1]; + + switch (flow_control.op) { + case Op::Or: + return result_x || result_y; + case Op::And: + return result_x && result_y; + case Op::JustX: + return result_x; + case Op::JustY: + return result_y; + default: + UNREACHABLE(); + return false; + } + }; + const auto& uniforms = g_state.vs.uniforms; const auto& swizzle_data = g_state.vs.swizzle_data; const auto& program_code = g_state.vs.program_code; @@ -74,20 +105,11 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned const Instruction instr = {program_code[program_counter]}; const SwizzlePattern swizzle = {swizzle_data[instr.common.operand_desc_id]}; - auto call = [&program_counter, &call_stack](UnitState<Debug>& state, u32 offset, - u32 num_instructions, u32 return_offset, - u8 repeat_count, u8 loop_increment) { - // -1 to make sure when incrementing the PC we end up at the correct offset - program_counter = offset - 1; - ASSERT(call_stack.size() < call_stack.capacity()); - call_stack.push_back( - {offset + num_instructions, return_offset, repeat_count, loop_increment, offset}); - }; - Record<DebugDataRecord::CUR_INSTR>(state.debug, iteration, program_counter); + Record<DebugDataRecord::CUR_INSTR>(debug_data, iteration, program_counter); if (iteration > 0) - Record<DebugDataRecord::NEXT_INSTR>(state.debug, iteration - 1, program_counter); + Record<DebugDataRecord::NEXT_INSTR>(debug_data, iteration - 1, program_counter); - state.debug.max_offset = std::max<u32>(state.debug.max_offset, 1 + program_counter); + debug_data.max_offset = std::max<u32>(debug_data.max_offset, 1 + program_counter); auto LookupSourceRegister = [&](const SourceRegister& source_reg) -> const float24* { switch (source_reg.GetRegisterType()) { @@ -155,54 +177,54 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned ? &state.registers.temporary[instr.common.dest.Value().GetIndex()][0] : dummy_vec4_float24; - state.debug.max_opdesc_id = - std::max<u32>(state.debug.max_opdesc_id, 1 + instr.common.operand_desc_id); + debug_data.max_opdesc_id = + std::max<u32>(debug_data.max_opdesc_id, 1 + instr.common.operand_desc_id); switch (instr.opcode.Value().EffectiveOpCode()) { case OpCode::Id::ADD: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; dest[i] = src1[i] + src2[i]; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } case OpCode::Id::MUL: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; dest[i] = src1[i] * src2[i]; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } case OpCode::Id::FLR: - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; dest[i] = float24::FromFloat32(std::floor(src1[i].ToFloat32())); } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; case OpCode::Id::MAX: - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; @@ -212,13 +234,13 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned // max(NaN, 0) -> 0 dest[i] = (src1[i] > src2[i]) ? src1[i] : src2[i]; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; case OpCode::Id::MIN: - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; @@ -228,16 +250,16 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned // min(NaN, 0) -> 0 dest[i] = (src1[i] < src2[i]) ? src1[i] : src2[i]; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; case OpCode::Id::DP3: case OpCode::Id::DP4: case OpCode::Id::DPH: case OpCode::Id::DPHI: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); OpCode::Id opcode = instr.opcode.Value().EffectiveOpCode(); if (opcode == OpCode::Id::DPH || opcode == OpCode::Id::DPHI) @@ -253,14 +275,14 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = dot; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } // Reciprocal case OpCode::Id::RCP: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); float24 rcp_res = float24::FromFloat32(1.0f / src1[0].ToFloat32()); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) @@ -268,14 +290,14 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = rcp_res; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } // Reciprocal Square Root case OpCode::Id::RSQ: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); float24 rsq_res = float24::FromFloat32(1.0f / std::sqrt(src1[0].ToFloat32())); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) @@ -283,12 +305,12 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = rsq_res; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } case OpCode::Id::MOVA: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); for (int i = 0; i < 2; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; @@ -296,29 +318,29 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned // TODO: Figure out how the rounding is done on hardware state.address_registers[i] = static_cast<s32>(src1[i].ToFloat32()); } - Record<DebugDataRecord::ADDR_REG_OUT>(state.debug, iteration, + Record<DebugDataRecord::ADDR_REG_OUT>(debug_data, iteration, state.address_registers); break; } case OpCode::Id::MOV: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; dest[i] = src1[i]; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } case OpCode::Id::SGE: case OpCode::Id::SGEI: - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; @@ -326,14 +348,14 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = (src1[i] >= src2[i]) ? float24::FromFloat32(1.0f) : float24::FromFloat32(0.0f); } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; case OpCode::Id::SLT: case OpCode::Id::SLTI: - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; @@ -341,12 +363,12 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = (src1[i] < src2[i]) ? float24::FromFloat32(1.0f) : float24::FromFloat32(0.0f); } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; case OpCode::Id::CMP: - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); for (int i = 0; i < 2; ++i) { // TODO: Can you restrict to one compare via dest masking? @@ -383,12 +405,12 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned break; } } - Record<DebugDataRecord::CMP_RESULT>(state.debug, iteration, state.conditional_code); + Record<DebugDataRecord::CMP_RESULT>(debug_data, iteration, state.conditional_code); break; case OpCode::Id::EX2: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); // EX2 only takes first component exp2 and writes it to all dest components float24 ex2_res = float24::FromFloat32(std::exp2(src1[0].ToFloat32())); @@ -399,13 +421,13 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = ex2_res; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } case OpCode::Id::LG2: { - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); // LG2 only takes the first component log2 and writes it to all dest components float24 lg2_res = float24::FromFloat32(std::log2(src1[0].ToFloat32())); @@ -416,7 +438,7 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned dest[i] = lg2_res; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); break; } @@ -498,17 +520,17 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned ? &state.registers.temporary[instr.mad.dest.Value().GetIndex()][0] : dummy_vec4_float24; - Record<DebugDataRecord::SRC1>(state.debug, iteration, src1); - Record<DebugDataRecord::SRC2>(state.debug, iteration, src2); - Record<DebugDataRecord::SRC3>(state.debug, iteration, src3); - Record<DebugDataRecord::DEST_IN>(state.debug, iteration, dest); + Record<DebugDataRecord::SRC1>(debug_data, iteration, src1); + Record<DebugDataRecord::SRC2>(debug_data, iteration, src2); + Record<DebugDataRecord::SRC3>(debug_data, iteration, src3); + Record<DebugDataRecord::DEST_IN>(debug_data, iteration, dest); for (int i = 0; i < 4; ++i) { if (!swizzle.DestComponentEnabled(i)) continue; dest[i] = src1[i] * src2[i] + src3[i]; } - Record<DebugDataRecord::DEST_OUT>(state.debug, iteration, dest); + Record<DebugDataRecord::DEST_OUT>(debug_data, iteration, dest); } else { LOG_ERROR(HW_GPU, "Unhandled multiply-add instruction: 0x%02x (%s): 0x%08x", (int)instr.opcode.Value().EffectiveOpCode(), @@ -518,26 +540,6 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned } default: { - static auto evaluate_condition = [](const UnitState<Debug>& state, bool refx, bool refy, - Instruction::FlowControlType flow_control) { - bool results[2] = {refx == state.conditional_code[0], - refy == state.conditional_code[1]}; - - switch (flow_control.op) { - case flow_control.Or: - return results[0] || results[1]; - - case flow_control.And: - return results[0] && results[1]; - - case flow_control.JustX: - return results[0]; - - case flow_control.JustY: - return results[1]; - } - }; - // Handle each instruction on its own switch (instr.opcode.Value()) { case OpCode::Id::END: @@ -545,17 +547,15 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned break; case OpCode::Id::JMPC: - Record<DebugDataRecord::COND_CMP_IN>(state.debug, iteration, - state.conditional_code); - if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, - instr.flow_control)) { + Record<DebugDataRecord::COND_CMP_IN>(debug_data, iteration, state.conditional_code); + if (evaluate_condition(instr.flow_control)) { program_counter = instr.flow_control.dest_offset - 1; } break; case OpCode::Id::JMPU: Record<DebugDataRecord::COND_BOOL_IN>( - state.debug, iteration, uniforms.b[instr.flow_control.bool_uniform_id]); + debug_data, iteration, uniforms.b[instr.flow_control.bool_uniform_id]); if (uniforms.b[instr.flow_control.bool_uniform_id] == !(instr.flow_control.num_instructions & 1)) { @@ -564,25 +564,23 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned break; case OpCode::Id::CALL: - call(state, instr.flow_control.dest_offset, instr.flow_control.num_instructions, + call(instr.flow_control.dest_offset, instr.flow_control.num_instructions, program_counter + 1, 0, 0); break; case OpCode::Id::CALLU: Record<DebugDataRecord::COND_BOOL_IN>( - state.debug, iteration, uniforms.b[instr.flow_control.bool_uniform_id]); + debug_data, iteration, uniforms.b[instr.flow_control.bool_uniform_id]); if (uniforms.b[instr.flow_control.bool_uniform_id]) { - call(state, instr.flow_control.dest_offset, instr.flow_control.num_instructions, + call(instr.flow_control.dest_offset, instr.flow_control.num_instructions, program_counter + 1, 0, 0); } break; case OpCode::Id::CALLC: - Record<DebugDataRecord::COND_CMP_IN>(state.debug, iteration, - state.conditional_code); - if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, - instr.flow_control)) { - call(state, instr.flow_control.dest_offset, instr.flow_control.num_instructions, + Record<DebugDataRecord::COND_CMP_IN>(debug_data, iteration, state.conditional_code); + if (evaluate_condition(instr.flow_control)) { + call(instr.flow_control.dest_offset, instr.flow_control.num_instructions, program_counter + 1, 0, 0); } break; @@ -592,14 +590,13 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned case OpCode::Id::IFU: Record<DebugDataRecord::COND_BOOL_IN>( - state.debug, iteration, uniforms.b[instr.flow_control.bool_uniform_id]); + debug_data, iteration, uniforms.b[instr.flow_control.bool_uniform_id]); if (uniforms.b[instr.flow_control.bool_uniform_id]) { - call(state, program_counter + 1, - instr.flow_control.dest_offset - program_counter - 1, + call(program_counter + 1, instr.flow_control.dest_offset - program_counter - 1, instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0); } else { - call(state, instr.flow_control.dest_offset, instr.flow_control.num_instructions, + call(instr.flow_control.dest_offset, instr.flow_control.num_instructions, instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0); } @@ -609,16 +606,13 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned case OpCode::Id::IFC: { // TODO: Do we need to consider swizzlers here? - Record<DebugDataRecord::COND_CMP_IN>(state.debug, iteration, - state.conditional_code); - if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, - instr.flow_control)) { - call(state, program_counter + 1, - instr.flow_control.dest_offset - program_counter - 1, + Record<DebugDataRecord::COND_CMP_IN>(debug_data, iteration, state.conditional_code); + if (evaluate_condition(instr.flow_control)) { + call(program_counter + 1, instr.flow_control.dest_offset - program_counter - 1, instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0); } else { - call(state, instr.flow_control.dest_offset, instr.flow_control.num_instructions, + call(instr.flow_control.dest_offset, instr.flow_control.num_instructions, instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0); } @@ -633,9 +627,8 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned uniforms.i[instr.flow_control.int_uniform_id].w); state.address_registers[2] = loop_param.y; - Record<DebugDataRecord::LOOP_INT_IN>(state.debug, iteration, loop_param); - call(state, program_counter + 1, - instr.flow_control.dest_offset - program_counter + 1, + Record<DebugDataRecord::LOOP_INT_IN>(debug_data, iteration, loop_param); + call(program_counter + 1, instr.flow_control.dest_offset - program_counter + 1, instr.flow_control.dest_offset + 1, loop_param.x, loop_param.z); break; } @@ -657,8 +650,8 @@ void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned } // Explicit instantiation -template void RunInterpreter(const ShaderSetup& setup, UnitState<false>& state, unsigned offset); -template void RunInterpreter(const ShaderSetup& setup, UnitState<true>& state, unsigned offset); +template void RunInterpreter(const ShaderSetup&, UnitState&, DebugData<false>&, unsigned offset); +template void RunInterpreter(const ShaderSetup&, UnitState&, DebugData<true>&, unsigned offset); } // namespace diff --git a/src/video_core/shader/shader_interpreter.h b/src/video_core/shader/shader_interpreter.h index 48ede0a2e..d31dcd7a6 100644 --- a/src/video_core/shader/shader_interpreter.h +++ b/src/video_core/shader/shader_interpreter.h @@ -8,11 +8,14 @@ namespace Pica { namespace Shader { -template <bool Debug> struct UnitState; template <bool Debug> -void RunInterpreter(const ShaderSetup& setup, UnitState<Debug>& state, unsigned offset); +struct DebugData; + +template <bool Debug> +void RunInterpreter(const ShaderSetup& setup, UnitState& state, DebugData<Debug>& debug_data, + unsigned offset); } // namespace diff --git a/src/video_core/shader/shader_jit_x64.cpp b/src/video_core/shader/shader_jit_x64.cpp index cfdeb8d6a..c588b778b 100644 --- a/src/video_core/shader/shader_jit_x64.cpp +++ b/src/video_core/shader/shader_jit_x64.cpp @@ -185,10 +185,10 @@ void JitShader::Compile_SwizzleSrc(Instruction instr, unsigned src_num, SourceRe if (src_reg.GetRegisterType() == RegisterType::FloatUniform) { src_ptr = SETUP; - src_offset = ShaderSetup::UniformOffset(RegisterType::FloatUniform, src_reg.GetIndex()); + src_offset = ShaderSetup::GetFloatUniformOffset(src_reg.GetIndex()); } else { src_ptr = STATE; - src_offset = UnitState<false>::InputOffset(src_reg); + src_offset = UnitState::InputOffset(src_reg); } int src_offset_disp = (int)src_offset; @@ -222,7 +222,7 @@ void JitShader::Compile_SwizzleSrc(Instruction instr, unsigned src_num, SourceRe movaps(dest, xword[src_ptr + ADDROFFS_REG_1 + src_offset_disp]); break; case 3: // address offset 3 - movaps(dest, xword[src_ptr + LOOPCOUNT_REG + src_offset_disp]); + movaps(dest, xword[src_ptr + LOOPCOUNT_REG.cvt64() + src_offset_disp]); break; default: UNREACHABLE(); @@ -266,9 +266,7 @@ void JitShader::Compile_DestEnable(Instruction instr, Xmm src) { SwizzlePattern swiz = {g_state.vs.swizzle_data[operand_desc_id]}; - int dest_offset_disp = (int)UnitState<false>::OutputOffset(dest); - ASSERT_MSG(dest_offset_disp == UnitState<false>::OutputOffset(dest), - "Destinaton offset too large for int type"); + size_t dest_offset_disp = UnitState::OutputOffset(dest); // If all components are enabled, write the result to the destination register if (swiz.dest_mask == NO_DEST_REG_MASK) { @@ -348,8 +346,7 @@ void JitShader::Compile_EvaluateCondition(Instruction instr) { } void JitShader::Compile_UniformCondition(Instruction instr) { - size_t offset = - ShaderSetup::UniformOffset(RegisterType::BoolUniform, instr.flow_control.bool_uniform_id); + size_t offset = ShaderSetup::GetBoolUniformOffset(instr.flow_control.bool_uniform_id); cmp(byte[SETUP + offset], 0); } @@ -732,8 +729,7 @@ void JitShader::Compile_LOOP(Instruction instr) { // This decodes the fields from the integer uniform at index instr.flow_control.int_uniform_id. // The Y (LOOPCOUNT_REG) and Z (LOOPINC) component are kept multiplied by 16 (Left shifted by // 4 bits) to be used as an offset into the 16-byte vector registers later - size_t offset = - ShaderSetup::UniformOffset(RegisterType::IntUniform, instr.flow_control.int_uniform_id); + size_t offset = ShaderSetup::GetIntUniformOffset(instr.flow_control.int_uniform_id); mov(LOOPCOUNT, dword[SETUP + offset]); mov(LOOPCOUNT_REG, LOOPCOUNT); shr(LOOPCOUNT_REG, 4); diff --git a/src/video_core/shader/shader_jit_x64.h b/src/video_core/shader/shader_jit_x64.h index e0ecde3f2..f37548306 100644 --- a/src/video_core/shader/shader_jit_x64.h +++ b/src/video_core/shader/shader_jit_x64.h @@ -34,7 +34,7 @@ class JitShader : public Xbyak::CodeGenerator { public: JitShader(); - void Run(const ShaderSetup& setup, UnitState<false>& state, unsigned offset) const { + void Run(const ShaderSetup& setup, UnitState& state, unsigned offset) const { program(&setup, &state, instruction_labels[offset].getAddress()); } |