summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/citra_qt/game_list.cpp68
-rw-r--r--src/citra_qt/game_list.h5
-rw-r--r--src/citra_qt/game_list_p.h8
-rw-r--r--src/core/hle/service/hid/hid.cpp19
-rw-r--r--src/core/hle/service/hid/hid.h10
-rw-r--r--src/core/hle/service/ir/ir.cpp5
-rw-r--r--src/core/hle/service/ir/ir.h3
-rw-r--r--src/core/hle/service/ir/ir_rst.cpp186
-rw-r--r--src/core/hle/service/ir/ir_rst.h3
-rw-r--r--src/core/hle/service/ir/ir_user.cpp2
-rw-r--r--src/core/hle/service/ir/ir_user.h2
-rw-r--r--src/core/settings.cpp2
-rw-r--r--src/video_core/command_processor.cpp212
-rw-r--r--src/video_core/shader/shader.h7
-rw-r--r--src/video_core/shader/shader_interpreter.cpp2
-rw-r--r--src/video_core/shader/shader_jit_x64.cpp2
-rw-r--r--src/video_core/shader/shader_jit_x64_compiler.cpp4
-rw-r--r--src/video_core/shader/shader_jit_x64_compiler.h14
18 files changed, 414 insertions, 140 deletions
diff --git a/src/citra_qt/game_list.cpp b/src/citra_qt/game_list.cpp
index 51257520b..a8e3541cd 100644
--- a/src/citra_qt/game_list.cpp
+++ b/src/citra_qt/game_list.cpp
@@ -2,6 +2,7 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <QApplication>
#include <QFileInfo>
#include <QHeaderView>
#include <QKeyEvent>
@@ -194,6 +195,9 @@ void GameList::onFilterCloseClicked() {
}
GameList::GameList(GMainWindow* parent) : QWidget{parent} {
+ watcher = new QFileSystemWatcher(this);
+ connect(watcher, &QFileSystemWatcher::directoryChanged, this, &GameList::RefreshGameDirectory);
+
this->main_window = parent;
layout = new QVBoxLayout;
tree_view = new QTreeView;
@@ -218,7 +222,6 @@ GameList::GameList(GMainWindow* parent) : QWidget{parent} {
connect(tree_view, &QTreeView::activated, this, &GameList::ValidateEntry);
connect(tree_view, &QTreeView::customContextMenuRequested, this, &GameList::PopupContextMenu);
- connect(&watcher, &QFileSystemWatcher::directoryChanged, this, &GameList::RefreshGameDirectory);
// 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.
@@ -269,7 +272,22 @@ void GameList::ValidateEntry(const QModelIndex& item) {
emit GameChosen(file_path);
}
-void GameList::DonePopulating() {
+void GameList::DonePopulating(QStringList watch_list) {
+ // Clear out the old directories to watch for changes and add the new ones
+ auto watch_dirs = watcher->directories();
+ if (!watch_dirs.isEmpty()) {
+ watcher->removePaths(watch_dirs);
+ }
+ // Workaround: Add the watch paths in chunks to allow the gui to refresh
+ // This prevents the UI from stalling when a large number of watch paths are added
+ // Also artificially caps the watcher to a certain number of directories
+ constexpr int LIMIT_WATCH_DIRECTORIES = 5000;
+ constexpr int SLICE_SIZE = 25;
+ int len = std::min(watch_list.length(), LIMIT_WATCH_DIRECTORIES);
+ for (int i = 0; i < len; i += SLICE_SIZE) {
+ watcher->addPaths(watch_list.mid(i, i + SLICE_SIZE));
+ QCoreApplication::processEvents();
+ }
tree_view->setEnabled(true);
int rowCount = tree_view->model()->rowCount();
search_field->setFilterResult(rowCount, rowCount);
@@ -309,11 +327,6 @@ void GameList::PopulateAsync(const QString& dir_path, bool deep_scan) {
emit ShouldCancelWorker();
- auto watch_dirs = watcher.directories();
- if (!watch_dirs.isEmpty()) {
- watcher.removePaths(watch_dirs);
- }
- UpdateWatcherList(dir_path.toStdString(), deep_scan ? 256 : 0);
GameListWorker* worker = new GameListWorker(dir_path, deep_scan);
connect(worker, &GameListWorker::EntryReady, this, &GameList::AddEntry, Qt::QueuedConnection);
@@ -359,38 +372,6 @@ void GameList::RefreshGameDirectory() {
}
}
-/**
- * Adds the game list folder to the QFileSystemWatcher to check for updates.
- *
- * The file watcher will fire off an update to the game list when a change is detected in the game
- * list folder.
- *
- * Notice: This method is run on the UI thread because QFileSystemWatcher is not thread safe and
- * this function is fast enough to not stall the UI thread. If performance is an issue, it should
- * be moved to another thread and properly locked to prevent concurrency issues.
- *
- * @param dir folder to check for changes in
- * @param recursion 0 if recursion is disabled. Any positive number passed to this will add each
- * directory recursively to the watcher and will update the file list if any of the folders
- * change. The number determines how deep the recursion should traverse.
- */
-void GameList::UpdateWatcherList(const std::string& dir, unsigned int recursion) {
- const auto callback = [this, recursion](unsigned* num_entries_out, const std::string& directory,
- const std::string& virtual_name) -> bool {
- std::string physical_name = directory + DIR_SEP + virtual_name;
-
- if (FileUtil::IsDirectory(physical_name)) {
- UpdateWatcherList(physical_name, recursion - 1);
- }
- return true;
- };
-
- watcher.addPath(QString::fromStdString(dir));
- if (recursion > 0) {
- FileUtil::ForeachDirectoryEntry(nullptr, dir, callback);
- }
-}
-
void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion) {
const auto callback = [this, recursion](unsigned* num_entries_out, const std::string& directory,
const std::string& virtual_name) -> bool {
@@ -399,7 +380,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
if (stop_processing)
return false; // Breaks the callback loop.
- if (!FileUtil::IsDirectory(physical_name) && HasSupportedFileExtension(physical_name)) {
+ bool is_dir = FileUtil::IsDirectory(physical_name);
+ if (!is_dir && HasSupportedFileExtension(physical_name)) {
std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(physical_name);
if (!loader)
return true;
@@ -416,7 +398,8 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
new GameListItemSize(FileUtil::GetSize(physical_name)),
});
- } else if (recursion > 0) {
+ } else if (is_dir && recursion > 0) {
+ watch_list.append(QString::fromStdString(physical_name));
AddFstEntriesToGameList(physical_name, recursion - 1);
}
@@ -428,8 +411,9 @@ void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsign
void GameListWorker::run() {
stop_processing = false;
+ watch_list.append(dir_path);
AddFstEntriesToGameList(dir_path.toStdString(), deep_scan ? 256 : 0);
- emit Finished();
+ emit Finished(watch_list);
}
void GameListWorker::Cancel() {
diff --git a/src/citra_qt/game_list.h b/src/citra_qt/game_list.h
index d8f8bc5b6..4823a1296 100644
--- a/src/citra_qt/game_list.h
+++ b/src/citra_qt/game_list.h
@@ -85,10 +85,9 @@ private slots:
private:
void AddEntry(const QList<QStandardItem*>& entry_items);
void ValidateEntry(const QModelIndex& item);
- void DonePopulating();
+ void DonePopulating(QStringList watch_list);
void PopupContextMenu(const QPoint& menu_location);
- void UpdateWatcherList(const std::string& path, unsigned int recursion);
void RefreshGameDirectory();
bool containsAllWords(QString haystack, QString userinput);
@@ -98,5 +97,5 @@ private:
QTreeView* tree_view = nullptr;
QStandardItemModel* item_model = nullptr;
GameListWorker* current_worker = nullptr;
- QFileSystemWatcher watcher;
+ QFileSystemWatcher* watcher = nullptr;
};
diff --git a/src/citra_qt/game_list_p.h b/src/citra_qt/game_list_p.h
index 3c11b6dd1..d1118ff7f 100644
--- a/src/citra_qt/game_list_p.h
+++ b/src/citra_qt/game_list_p.h
@@ -170,9 +170,15 @@ signals:
* @param entry_items a list with `QStandardItem`s that make up the columns of the new entry.
*/
void EntryReady(QList<QStandardItem*> entry_items);
- void Finished();
+
+ /**
+ * After the worker has traversed the game directory looking for entries, this signal is emmited
+ * with a list of folders that should be watched for changes as well.
+ */
+ void Finished(QStringList watch_list);
private:
+ QStringList watch_list;
QString dir_path;
bool deep_scan;
std::atomic_bool stop_processing;
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index b19e831fe..64d01cdd7 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -53,30 +53,29 @@ static std::array<std::unique_ptr<Input::ButtonDevice>, Settings::NativeButton::
buttons;
static std::unique_ptr<Input::AnalogDevice> circle_pad;
-static PadState GetCirclePadDirectionState(s16 circle_pad_x, s16 circle_pad_y) {
+DirectionState GetStickDirectionState(s16 circle_pad_x, s16 circle_pad_y) {
// 30 degree and 60 degree are angular thresholds for directions
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;
- state.hex = 0;
+ DirectionState state{false, false, false, false};
if (circle_pad_x * circle_pad_x + circle_pad_y * circle_pad_y > CIRCLE_PAD_THRESHOLD_SQUARE) {
float t = std::abs(static_cast<float>(circle_pad_y) / circle_pad_x);
if (circle_pad_x != 0 && t < TAN60) {
if (circle_pad_x > 0)
- state.circle_right.Assign(1);
+ state.right = true;
else
- state.circle_left.Assign(1);
+ state.left = true;
}
if (circle_pad_x == 0 || t > TAN30) {
if (circle_pad_y > 0)
- state.circle_up.Assign(1);
+ state.up = true;
else
- state.circle_down.Assign(1);
+ state.down = true;
}
}
@@ -125,7 +124,11 @@ static void UpdatePadCallback(u64 userdata, int cycles_late) {
constexpr int MAX_CIRCLEPAD_POS = 0x9C; // Max value for a circle pad position
s16 circle_pad_x = static_cast<s16>(circle_pad_x_f * MAX_CIRCLEPAD_POS);
s16 circle_pad_y = static_cast<s16>(circle_pad_y_f * MAX_CIRCLEPAD_POS);
- state.hex |= GetCirclePadDirectionState(circle_pad_x, circle_pad_y).hex;
+ const DirectionState direction = GetStickDirectionState(circle_pad_x, circle_pad_y);
+ state.circle_up.Assign(direction.up);
+ state.circle_down.Assign(direction.down);
+ state.circle_left.Assign(direction.left);
+ state.circle_right.Assign(direction.right);
mem->pad.current_state.hex = state.hex;
mem->pad.index = next_pad_index;
diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h
index b505cdcd5..1ef972e70 100644
--- a/src/core/hle/service/hid/hid.h
+++ b/src/core/hle/service/hid/hid.h
@@ -176,6 +176,16 @@ ASSERT_REG_POSITION(touch.index_reset_ticks, 0x2A);
#undef ASSERT_REG_POSITION
#endif // !defined(_MSC_VER)
+struct DirectionState {
+ bool up;
+ bool down;
+ bool left;
+ bool right;
+};
+
+/// Translates analog stick axes to directions. This is exposed for ir_rst module to use.
+DirectionState GetStickDirectionState(s16 circle_pad_x, s16 circle_pad_y);
+
/**
* HID::GetIPCHandles service function
* Inputs:
diff --git a/src/core/hle/service/ir/ir.cpp b/src/core/hle/service/ir/ir.cpp
index 7ac34a990..f06dd552f 100644
--- a/src/core/hle/service/ir/ir.cpp
+++ b/src/core/hle/service/ir/ir.cpp
@@ -25,6 +25,11 @@ void Shutdown() {
ShutdownRST();
}
+void ReloadInputDevices() {
+ ReloadInputDevicesUser();
+ ReloadInputDevicesRST();
+}
+
} // namespace IR
} // namespace Service
diff --git a/src/core/hle/service/ir/ir.h b/src/core/hle/service/ir/ir.h
index c741498e2..6be3e950c 100644
--- a/src/core/hle/service/ir/ir.h
+++ b/src/core/hle/service/ir/ir.h
@@ -16,5 +16,8 @@ void Init();
/// Shutdown IR service
void Shutdown();
+/// Reload input devices. Used when input configuration changed
+void ReloadInputDevices();
+
} // namespace IR
} // namespace Service
diff --git a/src/core/hle/service/ir/ir_rst.cpp b/src/core/hle/service/ir/ir_rst.cpp
index 3f1275c53..53807cd91 100644
--- a/src/core/hle/service/ir/ir_rst.cpp
+++ b/src/core/hle/service/ir/ir_rst.cpp
@@ -2,16 +2,135 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <atomic>
+#include "common/bit_field.h"
+#include "core/core_timing.h"
+#include "core/frontend/input.h"
#include "core/hle/kernel/event.h"
#include "core/hle/kernel/shared_memory.h"
+#include "core/hle/service/hid/hid.h"
#include "core/hle/service/ir/ir.h"
#include "core/hle/service/ir/ir_rst.h"
+#include "core/settings.h"
namespace Service {
namespace IR {
-static Kernel::SharedPtr<Kernel::Event> handle_event;
+union PadState {
+ u32_le hex;
+
+ BitField<14, 1, u32_le> zl;
+ BitField<15, 1, u32_le> zr;
+
+ BitField<24, 1, u32_le> c_stick_right;
+ BitField<25, 1, u32_le> c_stick_left;
+ BitField<26, 1, u32_le> c_stick_up;
+ BitField<27, 1, u32_le> c_stick_down;
+};
+
+struct PadDataEntry {
+ PadState current_state;
+ PadState delta_additions;
+ PadState delta_removals;
+
+ s16_le c_stick_x;
+ s16_le c_stick_y;
+};
+
+struct SharedMem {
+ u64_le index_reset_ticks; ///< CPU tick count for when HID module updated entry index 0
+ u64_le index_reset_ticks_previous; ///< Previous `index_reset_ticks`
+ u32_le index;
+ INSERT_PADDING_WORDS(1);
+ std::array<PadDataEntry, 8> entries; ///< Last 8 pad entries
+};
+
+static_assert(sizeof(SharedMem) == 0x98, "SharedMem has wrong size!");
+
+static Kernel::SharedPtr<Kernel::Event> update_event;
static Kernel::SharedPtr<Kernel::SharedMemory> shared_memory;
+static u32 next_pad_index;
+static int update_callback_id;
+static std::unique_ptr<Input::ButtonDevice> zl_button;
+static std::unique_ptr<Input::ButtonDevice> zr_button;
+static std::unique_ptr<Input::AnalogDevice> c_stick;
+static std::atomic<bool> is_device_reload_pending;
+static bool raw_c_stick;
+static int update_period;
+
+static void LoadInputDevices() {
+ zl_button = Input::CreateDevice<Input::ButtonDevice>(
+ Settings::values.buttons[Settings::NativeButton::ZL]);
+ zr_button = Input::CreateDevice<Input::ButtonDevice>(
+ Settings::values.buttons[Settings::NativeButton::ZR]);
+ c_stick = Input::CreateDevice<Input::AnalogDevice>(
+ Settings::values.analogs[Settings::NativeAnalog::CStick]);
+}
+
+static void UnloadInputDevices() {
+ zl_button = nullptr;
+ zr_button = nullptr;
+ c_stick = nullptr;
+}
+
+static void UpdateCallback(u64 userdata, int cycles_late) {
+ SharedMem* mem = reinterpret_cast<SharedMem*>(shared_memory->GetPointer());
+
+ if (is_device_reload_pending.exchange(false))
+ LoadInputDevices();
+
+ PadState state;
+ state.zl.Assign(zl_button->GetStatus());
+ state.zr.Assign(zr_button->GetStatus());
+
+ // Get current c-stick position and update c-stick direction
+ float c_stick_x_f, c_stick_y_f;
+ std::tie(c_stick_x_f, c_stick_y_f) = c_stick->GetStatus();
+ constexpr int MAX_CSTICK_RADIUS = 0x9C; // Max value for a c-stick radius
+ const s16 c_stick_x = static_cast<s16>(c_stick_x_f * MAX_CSTICK_RADIUS);
+ const s16 c_stick_y = static_cast<s16>(c_stick_y_f * MAX_CSTICK_RADIUS);
+
+ if (!raw_c_stick) {
+ const HID::DirectionState direction = HID::GetStickDirectionState(c_stick_x, c_stick_y);
+ state.c_stick_up.Assign(direction.up);
+ state.c_stick_down.Assign(direction.down);
+ state.c_stick_left.Assign(direction.left);
+ state.c_stick_right.Assign(direction.right);
+ }
+
+ // TODO (wwylele): implement raw C-stick data for raw_c_stick = true
+
+ const u32 last_entry_index = mem->index;
+ mem->index = next_pad_index;
+ next_pad_index = (next_pad_index + 1) % mem->entries.size();
+
+ // Get the previous Pad state
+ PadState old_state{mem->entries[last_entry_index].current_state};
+
+ // Compute bitmask with 1s for bits different from the old state
+ PadState changed = {state.hex ^ old_state.hex};
+
+ // Get the current Pad entry
+ PadDataEntry& pad_entry = mem->entries[mem->index];
+
+ // Update entry properties
+ pad_entry.current_state.hex = state.hex;
+ pad_entry.delta_additions.hex = changed.hex & state.hex;
+ pad_entry.delta_removals.hex = changed.hex & old_state.hex;
+ pad_entry.c_stick_x = c_stick_x;
+ pad_entry.c_stick_y = c_stick_y;
+
+ // If we just updated index 0, provide a new timestamp
+ if (mem->index == 0) {
+ mem->index_reset_ticks_previous = mem->index_reset_ticks;
+ mem->index_reset_ticks = CoreTiming::GetTicks();
+ }
+
+ update_event->Signal();
+
+ // Reschedule recurrent event
+ CoreTiming::ScheduleEvent(msToCycles(update_period) - cycles_late, update_callback_id);
+}
/**
* IR::GetHandles service function
@@ -22,18 +141,52 @@ static Kernel::SharedPtr<Kernel::SharedMemory> shared_memory;
* 4 : Event handle
*/
static void GetHandles(Interface* self) {
- u32* cmd_buff = Kernel::GetCommandBuffer();
+ IPC::RequestParser rp(Kernel::GetCommandBuffer(), 0x01, 0, 0);
+ IPC::RequestBuilder rb = rp.MakeBuilder(1, 3);
+ rb.Push(RESULT_SUCCESS);
+ rb.PushMoveHandles(Kernel::g_handle_table.Create(Service::IR::shared_memory).MoveFrom(),
+ Kernel::g_handle_table.Create(Service::IR::update_event).MoveFrom());
+}
+
+/**
+ * IR::Initialize service function
+ * Inputs:
+ * 1 : pad state update period in ms
+ * 2 : bool output raw c-stick data
+ */
+static void Initialize(Interface* self) {
+ IPC::RequestParser rp(Kernel::GetCommandBuffer(), 0x02, 2, 0);
+ update_period = static_cast<int>(rp.Pop<u32>());
+ raw_c_stick = rp.Pop<bool>();
- cmd_buff[1] = RESULT_SUCCESS.raw;
- cmd_buff[2] = 0x4000000;
- cmd_buff[3] = Kernel::g_handle_table.Create(Service::IR::shared_memory).MoveFrom();
- cmd_buff[4] = Kernel::g_handle_table.Create(Service::IR::handle_event).MoveFrom();
+ if (raw_c_stick)
+ LOG_ERROR(Service_IR, "raw C-stick data is not implemented!");
+
+ next_pad_index = 0;
+ is_device_reload_pending.store(true);
+ CoreTiming::ScheduleEvent(msToCycles(update_period), update_callback_id);
+
+ IPC::RequestBuilder rb = rp.MakeBuilder(1, 0);
+ rb.Push(RESULT_SUCCESS);
+
+ LOG_DEBUG(Service_IR, "called. update_period=%d, raw_c_stick=%d", update_period, raw_c_stick);
+}
+
+static void Shutdown(Interface* self) {
+ IPC::RequestParser rp(Kernel::GetCommandBuffer(), 0x03, 1, 0);
+
+ CoreTiming::UnscheduleEvent(update_callback_id, 0);
+ UnloadInputDevices();
+
+ IPC::RequestBuilder rb = rp.MakeBuilder(1, 0);
+ rb.Push(RESULT_SUCCESS);
+ LOG_DEBUG(Service_IR, "called");
}
const Interface::FunctionInfo FunctionTable[] = {
{0x00010000, GetHandles, "GetHandles"},
- {0x00020080, nullptr, "Initialize"},
- {0x00030000, nullptr, "Shutdown"},
+ {0x00020080, Initialize, "Initialize"},
+ {0x00030000, Shutdown, "Shutdown"},
{0x00090000, nullptr, "WriteToTwoFields"},
};
@@ -43,17 +196,24 @@ IR_RST_Interface::IR_RST_Interface() {
void InitRST() {
using namespace Kernel;
-
+ // Note: these two kernel objects are even available before Initialize service function is
+ // called.
shared_memory =
- SharedMemory::Create(nullptr, 0x1000, MemoryPermission::ReadWrite,
- MemoryPermission::ReadWrite, 0, MemoryRegion::BASE, "IR:SharedMemory");
+ SharedMemory::Create(nullptr, 0x1000, MemoryPermission::ReadWrite, MemoryPermission::Read,
+ 0, MemoryRegion::BASE, "IRRST:SharedMemory");
+ update_event = Event::Create(ResetType::OneShot, "IRRST:UpdateEvent");
- handle_event = Event::Create(ResetType::OneShot, "IR:HandleEvent");
+ update_callback_id = CoreTiming::RegisterEvent("IRRST:UpdateCallBack", UpdateCallback);
}
void ShutdownRST() {
shared_memory = nullptr;
- handle_event = nullptr;
+ update_event = nullptr;
+ UnloadInputDevices();
+}
+
+void ReloadInputDevicesRST() {
+ is_device_reload_pending.store(true);
}
} // namespace IR
diff --git a/src/core/hle/service/ir/ir_rst.h b/src/core/hle/service/ir/ir_rst.h
index 75b732627..d932bb7e5 100644
--- a/src/core/hle/service/ir/ir_rst.h
+++ b/src/core/hle/service/ir/ir_rst.h
@@ -21,5 +21,8 @@ public:
void InitRST();
void ShutdownRST();
+/// Reload input devices. Used when input configuration changed
+void ReloadInputDevicesRST();
+
} // namespace IR
} // namespace Service
diff --git a/src/core/hle/service/ir/ir_user.cpp b/src/core/hle/service/ir/ir_user.cpp
index bccf6bce7..226af0083 100644
--- a/src/core/hle/service/ir/ir_user.cpp
+++ b/src/core/hle/service/ir/ir_user.cpp
@@ -542,7 +542,7 @@ void ShutdownUser() {
receive_event = nullptr;
}
-void ReloadInputDevices() {
+void ReloadInputDevicesUser() {
if (extra_hid)
extra_hid->RequestInputDevicesReload();
}
diff --git a/src/core/hle/service/ir/ir_user.h b/src/core/hle/service/ir/ir_user.h
index 2401346e8..930650406 100644
--- a/src/core/hle/service/ir/ir_user.h
+++ b/src/core/hle/service/ir/ir_user.h
@@ -52,7 +52,7 @@ void InitUser();
void ShutdownUser();
/// Reload input devices. Used when input configuration changed
-void ReloadInputDevices();
+void ReloadInputDevicesUser();
} // namespace IR
} // namespace Service
diff --git a/src/core/settings.cpp b/src/core/settings.cpp
index 3d22c0afa..d2e7c6b97 100644
--- a/src/core/settings.cpp
+++ b/src/core/settings.cpp
@@ -5,7 +5,7 @@
#include "audio_core/audio_core.h"
#include "core/gdbstub/gdbstub.h"
#include "core/hle/service/hid/hid.h"
-#include "core/hle/service/ir/ir_user.h"
+#include "core/hle/service/ir/ir.h"
#include "settings.h"
#include "video_core/video_core.h"
diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp
index 2e32ff905..9a09f81dc 100644
--- a/src/video_core/command_processor.cpp
+++ b/src/video_core/command_processor.cpp
@@ -32,12 +32,13 @@ namespace Pica {
namespace CommandProcessor {
-static int float_regs_counter = 0;
+static int vs_float_regs_counter = 0;
+static u32 vs_uniform_write_buffer[4];
-static u32 uniform_write_buffer[4];
+static int gs_float_regs_counter = 0;
+static u32 gs_uniform_write_buffer[4];
static int default_attr_counter = 0;
-
static u32 default_attr_write_buffer[3];
// Expand a 4-bit mask to 4-byte mask, e.g. 0b0101 -> 0x00FF00FF
@@ -48,6 +49,97 @@ static const u32 expand_bits_to_bytes[] = {
MICROPROFILE_DEFINE(GPU_Drawing, "GPU", "Drawing", MP_RGB(50, 50, 240));
+static const char* GetShaderSetupTypeName(Shader::ShaderSetup& setup) {
+ if (&setup == &g_state.vs) {
+ return "vertex shader";
+ }
+ if (&setup == &g_state.gs) {
+ return "geometry shader";
+ }
+ return "unknown shader";
+}
+
+static void WriteUniformBoolReg(Shader::ShaderSetup& setup, u32 value) {
+ for (unsigned i = 0; i < setup.uniforms.b.size(); ++i)
+ setup.uniforms.b[i] = (value & (1 << i)) != 0;
+}
+
+static void WriteUniformIntReg(Shader::ShaderSetup& setup, unsigned index,
+ const Math::Vec4<u8>& values) {
+ ASSERT(index < setup.uniforms.i.size());
+ setup.uniforms.i[index] = values;
+ LOG_TRACE(HW_GPU, "Set %s integer uniform %d to %02x %02x %02x %02x",
+ GetShaderSetupTypeName(setup), index, values.x, values.y, values.z, values.w);
+}
+
+static void WriteUniformFloatReg(ShaderRegs& config, Shader::ShaderSetup& setup,
+ int& float_regs_counter, u32 uniform_write_buffer[4], u32 value) {
+ auto& uniform_setup = config.uniform_setup;
+
+ // TODO: Does actual hardware indeed keep an intermediate buffer or does
+ // it directly write the values?
+ uniform_write_buffer[float_regs_counter++] = value;
+
+ // Uniforms are written in a packed format such that four float24 values are encoded in
+ // three 32-bit numbers. We write to internal memory once a full such vector is
+ // written.
+ if ((float_regs_counter >= 4 && uniform_setup.IsFloat32()) ||
+ (float_regs_counter >= 3 && !uniform_setup.IsFloat32())) {
+ float_regs_counter = 0;
+
+ auto& uniform = setup.uniforms.f[uniform_setup.index];
+
+ if (uniform_setup.index >= 96) {
+ LOG_ERROR(HW_GPU, "Invalid %s float uniform index %d", GetShaderSetupTypeName(setup),
+ (int)uniform_setup.index);
+ } else {
+
+ // NOTE: The destination component order indeed is "backwards"
+ if (uniform_setup.IsFloat32()) {
+ for (auto i : {0, 1, 2, 3})
+ uniform[3 - i] = float24::FromFloat32(*(float*)(&uniform_write_buffer[i]));
+ } else {
+ // TODO: Untested
+ uniform.w = float24::FromRaw(uniform_write_buffer[0] >> 8);
+ uniform.z = float24::FromRaw(((uniform_write_buffer[0] & 0xFF) << 16) |
+ ((uniform_write_buffer[1] >> 16) & 0xFFFF));
+ uniform.y = float24::FromRaw(((uniform_write_buffer[1] & 0xFFFF) << 8) |
+ ((uniform_write_buffer[2] >> 24) & 0xFF));
+ uniform.x = float24::FromRaw(uniform_write_buffer[2] & 0xFFFFFF);
+ }
+
+ LOG_TRACE(HW_GPU, "Set %s float uniform %x to (%f %f %f %f)",
+ GetShaderSetupTypeName(setup), (int)uniform_setup.index,
+ uniform.x.ToFloat32(), uniform.y.ToFloat32(), uniform.z.ToFloat32(),
+ uniform.w.ToFloat32());
+
+ // TODO: Verify that this actually modifies the register!
+ uniform_setup.index.Assign(uniform_setup.index + 1);
+ }
+ }
+}
+
+static void WriteProgramCode(ShaderRegs& config, Shader::ShaderSetup& setup,
+ unsigned max_program_code_length, u32 value) {
+ if (config.program.offset >= max_program_code_length) {
+ LOG_ERROR(HW_GPU, "Invalid %s program offset %d", GetShaderSetupTypeName(setup),
+ (int)config.program.offset);
+ } else {
+ setup.program_code[config.program.offset] = value;
+ config.program.offset++;
+ }
+}
+
+static void WriteSwizzlePatterns(ShaderRegs& config, Shader::ShaderSetup& setup, u32 value) {
+ if (config.swizzle_patterns.offset >= setup.swizzle_data.size()) {
+ LOG_ERROR(HW_GPU, "Invalid %s swizzle pattern offset %d", GetShaderSetupTypeName(setup),
+ (int)config.swizzle_patterns.offset);
+ } else {
+ setup.swizzle_data[config.swizzle_patterns.offset] = value;
+ config.swizzle_patterns.offset++;
+ }
+}
+
static void WritePicaReg(u32 id, u32 value, u32 mask) {
auto& regs = g_state.regs;
@@ -330,21 +422,70 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
break;
}
- case PICA_REG_INDEX(vs.bool_uniforms):
- for (unsigned i = 0; i < 16; ++i)
- g_state.vs.uniforms.b[i] = (regs.vs.bool_uniforms.Value() & (1 << i)) != 0;
+ case PICA_REG_INDEX(gs.bool_uniforms):
+ WriteUniformBoolReg(g_state.gs, value);
+ break;
+ case PICA_REG_INDEX_WORKAROUND(gs.int_uniforms[0], 0x281):
+ case PICA_REG_INDEX_WORKAROUND(gs.int_uniforms[1], 0x282):
+ case PICA_REG_INDEX_WORKAROUND(gs.int_uniforms[2], 0x283):
+ case PICA_REG_INDEX_WORKAROUND(gs.int_uniforms[3], 0x284): {
+ unsigned index = (id - PICA_REG_INDEX_WORKAROUND(gs.int_uniforms[0], 0x281));
+ auto values = regs.gs.int_uniforms[index];
+ WriteUniformIntReg(g_state.gs, index,
+ Math::Vec4<u8>(values.x, values.y, values.z, values.w));
+ break;
+ }
+
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[0], 0x291):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[1], 0x292):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[2], 0x293):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[3], 0x294):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[4], 0x295):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[5], 0x296):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[6], 0x297):
+ case PICA_REG_INDEX_WORKAROUND(gs.uniform_setup.set_value[7], 0x298): {
+ WriteUniformFloatReg(g_state.regs.gs, g_state.gs, gs_float_regs_counter,
+ gs_uniform_write_buffer, value);
+ break;
+ }
+
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[0], 0x29c):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[1], 0x29d):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[2], 0x29e):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[3], 0x29f):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[4], 0x2a0):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[5], 0x2a1):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[6], 0x2a2):
+ case PICA_REG_INDEX_WORKAROUND(gs.program.set_word[7], 0x2a3): {
+ WriteProgramCode(g_state.regs.gs, g_state.gs, 4096, value);
+ break;
+ }
+
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[0], 0x2a6):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[1], 0x2a7):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[2], 0x2a8):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[3], 0x2a9):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[4], 0x2aa):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[5], 0x2ab):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[6], 0x2ac):
+ case PICA_REG_INDEX_WORKAROUND(gs.swizzle_patterns.set_word[7], 0x2ad): {
+ WriteSwizzlePatterns(g_state.regs.gs, g_state.gs, value);
+ break;
+ }
+
+ case PICA_REG_INDEX(vs.bool_uniforms):
+ WriteUniformBoolReg(g_state.vs, value);
break;
case PICA_REG_INDEX_WORKAROUND(vs.int_uniforms[0], 0x2b1):
case PICA_REG_INDEX_WORKAROUND(vs.int_uniforms[1], 0x2b2):
case PICA_REG_INDEX_WORKAROUND(vs.int_uniforms[2], 0x2b3):
case PICA_REG_INDEX_WORKAROUND(vs.int_uniforms[3], 0x2b4): {
- int index = (id - PICA_REG_INDEX_WORKAROUND(vs.int_uniforms[0], 0x2b1));
+ unsigned index = (id - PICA_REG_INDEX_WORKAROUND(vs.int_uniforms[0], 0x2b1));
auto values = regs.vs.int_uniforms[index];
- g_state.vs.uniforms.i[index] = Math::Vec4<u8>(values.x, values.y, values.z, values.w);
- LOG_TRACE(HW_GPU, "Set integer uniform %d to %02x %02x %02x %02x", index, values.x.Value(),
- values.y.Value(), values.z.Value(), values.w.Value());
+ WriteUniformIntReg(g_state.vs, index,
+ Math::Vec4<u8>(values.x, values.y, values.z, values.w));
break;
}
@@ -356,51 +497,11 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
case PICA_REG_INDEX_WORKAROUND(vs.uniform_setup.set_value[5], 0x2c6):
case PICA_REG_INDEX_WORKAROUND(vs.uniform_setup.set_value[6], 0x2c7):
case PICA_REG_INDEX_WORKAROUND(vs.uniform_setup.set_value[7], 0x2c8): {
- auto& uniform_setup = regs.vs.uniform_setup;
-
- // TODO: Does actual hardware indeed keep an intermediate buffer or does
- // it directly write the values?
- uniform_write_buffer[float_regs_counter++] = value;
-
- // Uniforms are written in a packed format such that four float24 values are encoded in
- // three 32-bit numbers. We write to internal memory once a full such vector is
- // written.
- if ((float_regs_counter >= 4 && uniform_setup.IsFloat32()) ||
- (float_regs_counter >= 3 && !uniform_setup.IsFloat32())) {
- float_regs_counter = 0;
-
- auto& uniform = g_state.vs.uniforms.f[uniform_setup.index];
-
- if (uniform_setup.index > 95) {
- LOG_ERROR(HW_GPU, "Invalid VS uniform index %d", (int)uniform_setup.index);
- break;
- }
-
- // NOTE: The destination component order indeed is "backwards"
- if (uniform_setup.IsFloat32()) {
- for (auto i : {0, 1, 2, 3})
- uniform[3 - i] = float24::FromFloat32(*(float*)(&uniform_write_buffer[i]));
- } else {
- // TODO: Untested
- uniform.w = float24::FromRaw(uniform_write_buffer[0] >> 8);
- uniform.z = float24::FromRaw(((uniform_write_buffer[0] & 0xFF) << 16) |
- ((uniform_write_buffer[1] >> 16) & 0xFFFF));
- uniform.y = float24::FromRaw(((uniform_write_buffer[1] & 0xFFFF) << 8) |
- ((uniform_write_buffer[2] >> 24) & 0xFF));
- uniform.x = float24::FromRaw(uniform_write_buffer[2] & 0xFFFFFF);
- }
-
- LOG_TRACE(HW_GPU, "Set uniform %x to (%f %f %f %f)", (int)uniform_setup.index,
- uniform.x.ToFloat32(), uniform.y.ToFloat32(), uniform.z.ToFloat32(),
- uniform.w.ToFloat32());
-
- // TODO: Verify that this actually modifies the register!
- uniform_setup.index.Assign(uniform_setup.index + 1);
- }
+ WriteUniformFloatReg(g_state.regs.vs, g_state.vs, vs_float_regs_counter,
+ vs_uniform_write_buffer, value);
break;
}
- // Load shader program code
case PICA_REG_INDEX_WORKAROUND(vs.program.set_word[0], 0x2cc):
case PICA_REG_INDEX_WORKAROUND(vs.program.set_word[1], 0x2cd):
case PICA_REG_INDEX_WORKAROUND(vs.program.set_word[2], 0x2ce):
@@ -409,12 +510,10 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
case PICA_REG_INDEX_WORKAROUND(vs.program.set_word[5], 0x2d1):
case PICA_REG_INDEX_WORKAROUND(vs.program.set_word[6], 0x2d2):
case PICA_REG_INDEX_WORKAROUND(vs.program.set_word[7], 0x2d3): {
- g_state.vs.program_code[regs.vs.program.offset] = value;
- regs.vs.program.offset++;
+ WriteProgramCode(g_state.regs.vs, g_state.vs, 512, value);
break;
}
- // Load swizzle pattern data
case PICA_REG_INDEX_WORKAROUND(vs.swizzle_patterns.set_word[0], 0x2d6):
case PICA_REG_INDEX_WORKAROUND(vs.swizzle_patterns.set_word[1], 0x2d7):
case PICA_REG_INDEX_WORKAROUND(vs.swizzle_patterns.set_word[2], 0x2d8):
@@ -423,8 +522,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
case PICA_REG_INDEX_WORKAROUND(vs.swizzle_patterns.set_word[5], 0x2db):
case PICA_REG_INDEX_WORKAROUND(vs.swizzle_patterns.set_word[6], 0x2dc):
case PICA_REG_INDEX_WORKAROUND(vs.swizzle_patterns.set_word[7], 0x2dd): {
- g_state.vs.swizzle_data[regs.vs.swizzle_patterns.offset] = value;
- regs.vs.swizzle_patterns.offset++;
+ WriteSwizzlePatterns(g_state.regs.vs, g_state.vs, value);
break;
}
diff --git a/src/video_core/shader/shader.h b/src/video_core/shader/shader.h
index 38ea717ab..e156f6aef 100644
--- a/src/video_core/shader/shader.h
+++ b/src/video_core/shader/shader.h
@@ -24,6 +24,9 @@ namespace Pica {
namespace Shader {
+constexpr unsigned MAX_PROGRAM_CODE_LENGTH = 4096;
+constexpr unsigned MAX_SWIZZLE_DATA_LENGTH = 4096;
+
struct AttributeBuffer {
alignas(16) Math::Vec4<float24> attr[16];
};
@@ -144,8 +147,8 @@ struct ShaderSetup {
return offsetof(ShaderSetup, uniforms.i) + index * sizeof(Math::Vec4<u8>);
}
- std::array<u32, 1024> program_code;
- std::array<u32, 1024> swizzle_data;
+ std::array<u32, MAX_PROGRAM_CODE_LENGTH> program_code;
+ std::array<u32, MAX_SWIZZLE_DATA_LENGTH> swizzle_data;
/// Data private to ShaderEngines
struct EngineData {
diff --git a/src/video_core/shader/shader_interpreter.cpp b/src/video_core/shader/shader_interpreter.cpp
index f4d1c46c5..aa1cec81f 100644
--- a/src/video_core/shader/shader_interpreter.cpp
+++ b/src/video_core/shader/shader_interpreter.cpp
@@ -653,7 +653,7 @@ static void RunInterpreter(const ShaderSetup& setup, UnitState& state, DebugData
}
void InterpreterEngine::SetupBatch(ShaderSetup& setup, unsigned int entry_point) {
- ASSERT(entry_point < 1024);
+ ASSERT(entry_point < MAX_PROGRAM_CODE_LENGTH);
setup.engine_data.entry_point = entry_point;
}
diff --git a/src/video_core/shader/shader_jit_x64.cpp b/src/video_core/shader/shader_jit_x64.cpp
index 0ee0dd9ef..73c21871c 100644
--- a/src/video_core/shader/shader_jit_x64.cpp
+++ b/src/video_core/shader/shader_jit_x64.cpp
@@ -15,7 +15,7 @@ JitX64Engine::JitX64Engine() = default;
JitX64Engine::~JitX64Engine() = default;
void JitX64Engine::SetupBatch(ShaderSetup& setup, unsigned int entry_point) {
- ASSERT(entry_point < 1024);
+ ASSERT(entry_point < MAX_PROGRAM_CODE_LENGTH);
setup.engine_data.entry_point = entry_point;
u64 code_hash = Common::ComputeHash64(&setup.program_code, sizeof(setup.program_code));
diff --git a/src/video_core/shader/shader_jit_x64_compiler.cpp b/src/video_core/shader/shader_jit_x64_compiler.cpp
index 2dbc8b147..5d9b6448c 100644
--- a/src/video_core/shader/shader_jit_x64_compiler.cpp
+++ b/src/video_core/shader/shader_jit_x64_compiler.cpp
@@ -834,8 +834,8 @@ void JitShader::FindReturnOffsets() {
std::sort(return_offsets.begin(), return_offsets.end());
}
-void JitShader::Compile(const std::array<u32, 1024>* program_code_,
- const std::array<u32, 1024>* swizzle_data_) {
+void JitShader::Compile(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_code_,
+ const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>* swizzle_data_) {
program_code = program_code_;
swizzle_data = swizzle_data_;
diff --git a/src/video_core/shader/shader_jit_x64_compiler.h b/src/video_core/shader/shader_jit_x64_compiler.h
index f27675560..31af0ca48 100644
--- a/src/video_core/shader/shader_jit_x64_compiler.h
+++ b/src/video_core/shader/shader_jit_x64_compiler.h
@@ -22,8 +22,8 @@ namespace Pica {
namespace Shader {
-/// Memory allocated for each compiled shader (64Kb)
-constexpr size_t MAX_SHADER_SIZE = 1024 * 64;
+/// Memory allocated for each compiled shader
+constexpr size_t MAX_SHADER_SIZE = MAX_PROGRAM_CODE_LENGTH * 64;
/**
* This class implements the shader JIT compiler. It recompiles a Pica shader program into x86_64
@@ -37,8 +37,8 @@ public:
program(&setup, &state, instruction_labels[offset].getAddress());
}
- void Compile(const std::array<u32, 1024>* program_code,
- const std::array<u32, 1024>* swizzle_data);
+ void Compile(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_code,
+ const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>* swizzle_data);
void Compile_ADD(Instruction instr);
void Compile_DP3(Instruction instr);
@@ -104,11 +104,11 @@ private:
*/
void FindReturnOffsets();
- const std::array<u32, 1024>* program_code = nullptr;
- const std::array<u32, 1024>* swizzle_data = nullptr;
+ const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_code = nullptr;
+ const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>* swizzle_data = nullptr;
/// Mapping of Pica VS instructions to pointers in the emitted code
- std::array<Xbyak::Label, 1024> instruction_labels;
+ std::array<Xbyak::Label, MAX_PROGRAM_CODE_LENGTH> instruction_labels;
/// Offsets in code where a return needs to be inserted
std::vector<unsigned> return_offsets;
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-22 15:25:03 +0100