22 files changed, 451 insertions, 243 deletions

diff --git a/src/citra/emu_window/emu_window_glfw.cpp b/src/citra/emu_window/emu_window_glfw.cpp
index 81231e1e5..997e3bc7d 100644
--- a/src/citra/emu_window/emu_window_glfw.cpp
+++ b/src/citra/emu_window/emu_window_glfw.cpp
@@ -16,18 +16,34 @@ EmuWindow_GLFW* EmuWindow_GLFW::GetEmuWindow(GLFWwindow* win) {
     return static_cast<EmuWindow_GLFW*>(glfwGetWindowUserPointer(win));
 }
 
+void EmuWindow_GLFW::OnMouseButtonEvent(GLFWwindow* win, int button, int action, int mods) {
+    if (button == GLFW_MOUSE_BUTTON_LEFT) {
+        auto emu_window = GetEmuWindow(win);
+        auto layout = emu_window->GetFramebufferLayout();
+        double x, y;
+        glfwGetCursorPos(win, &x, &y);
+
+        if (action == GLFW_PRESS)
+            emu_window->TouchPressed(static_cast<unsigned>(x), static_cast<unsigned>(y));
+        else if (action == GLFW_RELEASE)
+            emu_window->TouchReleased();
+    }
+}
+
+void EmuWindow_GLFW::OnCursorPosEvent(GLFWwindow* win, double x, double y) {
+    GetEmuWindow(win)->TouchMoved(static_cast<unsigned>(x), static_cast<unsigned>(y));
+}
+
 /// Called by GLFW when a key event occurs
 void EmuWindow_GLFW::OnKeyEvent(GLFWwindow* win, int key, int scancode, int action, int mods) {
-
-    int keyboard_id = GetEmuWindow(win)->keyboard_id;
+    auto emu_window = GetEmuWindow(win);
+    int keyboard_id = emu_window->keyboard_id;
 
     if (action == GLFW_PRESS) {
-        EmuWindow::KeyPressed({key, keyboard_id});
+        emu_window->KeyPressed({key, keyboard_id});
     } else if (action == GLFW_RELEASE) {
-        EmuWindow::KeyReleased({key, keyboard_id});
+        emu_window->KeyReleased({key, keyboard_id});
     }
-
-    Service::HID::PadUpdateComplete();
 }
 
 /// Whether the window is still open, and a close request hasn't yet been sent
@@ -88,6 +104,8 @@ EmuWindow_GLFW::EmuWindow_GLFW() {
 
     // Setup callbacks
     glfwSetKeyCallback(m_render_window, OnKeyEvent);
+    glfwSetMouseButtonCallback(m_render_window, OnMouseButtonEvent);
+    glfwSetCursorPosCallback(m_render_window, OnCursorPosEvent);
     glfwSetFramebufferSizeCallback(m_render_window, OnFramebufferResizeEvent);
     glfwSetWindowSizeCallback(m_render_window, OnClientAreaResizeEvent);
 
diff --git a/src/citra/emu_window/emu_window_glfw.h b/src/citra/emu_window/emu_window_glfw.h
index 5252fccc8..16c109b79 100644
--- a/src/citra/emu_window/emu_window_glfw.h
+++ b/src/citra/emu_window/emu_window_glfw.h
@@ -27,6 +27,10 @@ public:
 
     static void OnKeyEvent(GLFWwindow* win, int key, int scancode, int action, int mods);
 
+    static void OnMouseButtonEvent(GLFWwindow* window, int button, int action, int mods);
+
+    static void OnCursorPosEvent(GLFWwindow* window, double x, double y);
+
     /// Whether the window is still open, and a close request hasn't yet been sent
     const bool IsOpen();
 
diff --git a/src/citra_qt/bootmanager.cpp b/src/citra_qt/bootmanager.cpp
index a040e75c1..b81bd6167 100644
--- a/src/citra_qt/bootmanager.cpp
+++ b/src/citra_qt/bootmanager.cpp
@@ -268,14 +268,33 @@ QByteArray GRenderWindow::saveGeometry()
 
 void GRenderWindow::keyPressEvent(QKeyEvent* event)
 {
-    EmuWindow::KeyPressed({event->key(), keyboard_id});
-    Service::HID::PadUpdateComplete();
+    this->KeyPressed({event->key(), keyboard_id});
 }
 
 void GRenderWindow::keyReleaseEvent(QKeyEvent* event)
 {
-    EmuWindow::KeyReleased({event->key(), keyboard_id});
-    Service::HID::PadUpdateComplete();
+    this->KeyReleased({event->key(), keyboard_id});
+}
+
+void GRenderWindow::mousePressEvent(QMouseEvent *event)
+{
+    if (event->button() == Qt::LeftButton)
+    {
+        auto pos = event->pos();
+        this->TouchPressed(static_cast<unsigned>(pos.x()), static_cast<unsigned>(pos.y()));
+    }
+}
+
+void GRenderWindow::mouseMoveEvent(QMouseEvent *event)
+{
+    auto pos = event->pos();
+    this->TouchMoved(static_cast<unsigned>(pos.x()), static_cast<unsigned>(pos.y()));
+}
+
+void GRenderWindow::mouseReleaseEvent(QMouseEvent *event)
+{
+    if (event->button() == Qt::LeftButton)
+        this->TouchReleased();
 }
 
 void GRenderWindow::ReloadSetKeymaps()
diff --git a/src/citra_qt/bootmanager.h b/src/citra_qt/bootmanager.h
index a55db682a..288da45a1 100644
--- a/src/citra_qt/bootmanager.h
+++ b/src/citra_qt/bootmanager.h
@@ -121,6 +121,10 @@ public:
     void keyPressEvent(QKeyEvent* event) override;
     void keyReleaseEvent(QKeyEvent* event) override;
 
+    void mousePressEvent(QMouseEvent *event) override;
+    void mouseMoveEvent(QMouseEvent *event) override;
+    void mouseReleaseEvent(QMouseEvent *event) override;
+
     void ReloadSetKeymaps() override;
 
     void OnClientAreaResized(unsigned width, unsigned height);
diff --git a/src/common/common_paths.h b/src/common/common_paths.h
index eb43d589f..440b06060 100644
--- a/src/common/common_paths.h
+++ b/src/common/common_paths.h
@@ -17,13 +17,12 @@
 
 // The user data dir
 #define ROOT_DIR "."
-#ifdef _WIN32
-    #define USERDATA_DIR "user"
-    #define EMU_DATA_DIR "Citra Emulator"
+#define USERDATA_DIR "user"
+#ifdef USER_DIR
+    #define EMU_DATA_DIR USER_DIR
 #else
-    #define USERDATA_DIR "user"
-    #ifdef USER_DIR
-        #define EMU_DATA_DIR USER_DIR
+    #ifdef _WIN32
+        #define EMU_DATA_DIR "Citra Emulator"
     #else
         #define EMU_DATA_DIR "citra-emu"
     #endif
diff --git a/src/common/emu_window.cpp b/src/common/emu_window.cpp
index 6459d2f32..6516fc633 100644
--- a/src/common/emu_window.cpp
+++ b/src/common/emu_window.cpp
@@ -6,18 +6,61 @@
 #include "video_core/video_core.h"
 
 void EmuWindow::KeyPressed(KeyMap::HostDeviceKey key) {
-    Service::HID::PadState mapped_key = KeyMap::GetPadKey(key);
-
-    Service::HID::PadButtonPress(mapped_key);
+    pad_state.hex |= KeyMap::GetPadKey(key).hex;
 }
 
 void EmuWindow::KeyReleased(KeyMap::HostDeviceKey key) {
-    Service::HID::PadState mapped_key = KeyMap::GetPadKey(key);
+    pad_state.hex &= ~KeyMap::GetPadKey(key).hex;
+}
+
+/**
+ * Check if the given x/y coordinates are within the touchpad specified by the framebuffer layout
+ * @param layout FramebufferLayout object describing the framebuffer size and screen positions
+ * @param framebuffer_x Framebuffer x-coordinate to check
+ * @param framebuffer_y Framebuffer y-coordinate to check
+ * @return True if the coordinates are within the touchpad, otherwise false
+ */
+static bool IsWithinTouchscreen(const EmuWindow::FramebufferLayout& layout, unsigned framebuffer_x,
+                                unsigned framebuffer_y) {
+    return (framebuffer_y >= layout.bottom_screen.top    &&
+            framebuffer_y <  layout.bottom_screen.bottom &&
+            framebuffer_x >= layout.bottom_screen.left   &&
+            framebuffer_x <  layout.bottom_screen.right);
+}
+
+void EmuWindow::TouchPressed(unsigned framebuffer_x, unsigned framebuffer_y) {
+    if (!IsWithinTouchscreen(framebuffer_layout, framebuffer_x, framebuffer_y))
+        return;
+
+    touch_x = VideoCore::kScreenBottomWidth * (framebuffer_x - framebuffer_layout.bottom_screen.left) /
+        (framebuffer_layout.bottom_screen.right - framebuffer_layout.bottom_screen.left);
+    touch_y = VideoCore::kScreenBottomHeight * (framebuffer_y - framebuffer_layout.bottom_screen.top) /
+        (framebuffer_layout.bottom_screen.bottom - framebuffer_layout.bottom_screen.top);
 
-    Service::HID::PadButtonRelease(mapped_key);
+    touch_pressed = true;
+    pad_state.touch = 1;
 }
 
-EmuWindow::FramebufferLayout EmuWindow::FramebufferLayout::DefaultScreenLayout(unsigned width, unsigned height) {
+void EmuWindow::TouchReleased() {
+    touch_pressed = false;
+    touch_x = 0;
+    touch_y = 0;
+    pad_state.touch = 0;
+}
+
+void EmuWindow::TouchMoved(unsigned framebuffer_x, unsigned framebuffer_y) {
+    if (!touch_pressed)
+        return;
+
+    if (IsWithinTouchscreen(framebuffer_layout, framebuffer_x, framebuffer_y))
+        TouchPressed(framebuffer_x, framebuffer_y);
+    else
+        TouchReleased();
+}
+
+EmuWindow::FramebufferLayout EmuWindow::FramebufferLayout::DefaultScreenLayout(unsigned width,
+    unsigned height) {
+
     ASSERT(width > 0);
     ASSERT(height > 0);
 
diff --git a/src/common/emu_window.h b/src/common/emu_window.h
index f6099fdb6..c8e2de04a 100644
--- a/src/common/emu_window.h
+++ b/src/common/emu_window.h
@@ -71,10 +71,48 @@ public:
     virtual void ReloadSetKeymaps() = 0;
 
     /// Signals a key press action to the HID module
-    static void KeyPressed(KeyMap::HostDeviceKey key);
+    void KeyPressed(KeyMap::HostDeviceKey key);
 
     /// Signals a key release action to the HID module
-    static void KeyReleased(KeyMap::HostDeviceKey key);
+    void KeyReleased(KeyMap::HostDeviceKey key);
+
+    /**
+     * Signal that a touch pressed event has occurred (e.g. mouse click pressed)
+     * @param framebuffer_x Framebuffer x-coordinate that was pressed
+     * @param framebuffer_y Framebuffer y-coordinate that was pressed
+     */
+    void TouchPressed(unsigned framebuffer_x, unsigned framebuffer_y);
+
+    /// Signal that a touch released event has occurred (e.g. mouse click released)
+    void TouchReleased();
+
+    /**
+     * Signal that a touch movement event has occurred (e.g. mouse was moved over the emu window)
+     * @param framebuffer_x Framebuffer x-coordinate
+     * @param framebuffer_y Framebuffer y-coordinate
+     */
+    void TouchMoved(unsigned framebuffer_x, unsigned framebuffer_y);
+
+    /**
+     * Gets the current pad state (which buttons are pressed and the circle pad direction).
+     * @note This should be called by the core emu thread to get a state set by the window thread.
+     * @todo Fix this function to be thread-safe.
+     * @return PadState object indicating the current pad state
+     */
+    const Service::HID::PadState GetPadState() const {
+        return pad_state;
+    }
+
+    /**
+     * Gets the current touch screen state (touch X/Y coordinates and whether or not it is pressed).
+     * @note This should be called by the core emu thread to get a state set by the window thread.
+     * @todo Fix this function to be thread-safe.
+     * @return std::tuple of (x, y, pressed) where `x` and `y` are the touch coordinates and
+     *         `pressed` is true if the touch screen is currently being pressed
+     */
+    const std::tuple<u16, u16, bool> GetTouchState() const {
+        return std::make_tuple(touch_x, touch_y, touch_pressed);
+    }
 
     /**
      * Returns currently active configuration.
@@ -100,21 +138,15 @@ public:
         return framebuffer_layout;
     }
 
-    /**
-     * Gets window client area width in logical coordinates.
-     * @note For high-DPI systems, this is smaller than the framebuffer size.
-     * @note This method is thread-safe
-     */
-    std::pair<unsigned,unsigned> GetClientAreaSize() const {
-        return std::make_pair(client_area_width, client_area_height);
-    }
-
 protected:
-    EmuWindow()
-    {
+    EmuWindow() {
         // TODO: Find a better place to set this.
         config.min_client_area_size = std::make_pair(400u, 480u);
         active_config = config;
+        pad_state.hex = 0;
+        touch_x = 0;
+        touch_y = 0;
+        touch_pressed = false;
     }
     virtual ~EmuWindow() {}
 
@@ -168,4 +200,11 @@ private:
 
     WindowConfig config;         ///< Internal configuration (changes pending for being applied in ProcessConfigurationChanges)
     WindowConfig active_config;  ///< Internal active configuration
+
+    bool touch_pressed;          ///< True if touchpad area is currently pressed, otherwise false
+
+    u16 touch_x;    ///< Touchpad X-position in native 3DS pixel coordinates (0-320)
+    u16 touch_y;    ///< Touchpad Y-position in native 3DS pixel coordinates (0-240)
+
+    Service::HID::PadState pad_state;
 };
diff --git a/src/common/platform.h b/src/common/platform.h
index ba1109c9f..e27d6e31f 100644
--- a/src/common/platform.h
+++ b/src/common/platform.h
@@ -83,7 +83,7 @@ inline struct tm* localtime_r(const time_t *clock, struct tm *result) {
 }
 #endif
 
-#else
+#else // EMU_PLATFORM != PLATFORM_WINDOWS
 
 #define EMU_FASTCALL __attribute__((fastcall))
 #define __stdcall
@@ -92,10 +92,6 @@ inline struct tm* localtime_r(const time_t *clock, struct tm *result) {
 #define BOOL bool
 #define DWORD u32
 
-#endif
-
-#if EMU_PLATFORM != PLATFORM_WINDOWS
-
 // TODO: Hacks..
 #include <limits.h>
 
diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h
index ef37ee055..fe1e584ad 100644
--- a/src/core/arm/arm_interface.h
+++ b/src/core/arm/arm_interface.h
@@ -74,12 +74,6 @@ public:
     virtual void SetCPSR(u32 cpsr) = 0;
 
     /**
-     * Returns the number of clock ticks since the last rese
-     * @return Returns number of clock ticks
-     */
-    virtual u64 GetTicks() const = 0;
-
-    /**
      * Advance the CPU core by the specified number of ticks (e.g. to simulate CPU execution time)
      * @param ticks Number of ticks to advance the CPU core
      */
diff --git a/src/core/arm/dyncom/arm_dyncom.cpp b/src/core/arm/dyncom/arm_dyncom.cpp
index bbcbbdd2b..cb1a410a0 100644
--- a/src/core/arm/dyncom/arm_dyncom.cpp
+++ b/src/core/arm/dyncom/arm_dyncom.cpp
@@ -68,11 +68,6 @@ void ARM_DynCom::SetCPSR(u32 cpsr) {
     state->Cpsr = cpsr;
 }
 
-u64 ARM_DynCom::GetTicks() const {
-    // TODO(Subv): Remove ARM_DynCom::GetTicks() and use CoreTiming::GetTicks() directly once ARMemu is gone
-    return CoreTiming::GetTicks();
-}
-
 void ARM_DynCom::AddTicks(u64 ticks) {
     down_count -= ticks;
     if (down_count < 0)
diff --git a/src/core/arm/dyncom/arm_dyncom.h b/src/core/arm/dyncom/arm_dyncom.h
index 213cac1ad..a7f95d307 100644
--- a/src/core/arm/dyncom/arm_dyncom.h
+++ b/src/core/arm/dyncom/arm_dyncom.h
@@ -23,7 +23,6 @@ public:
     u32 GetCPSR() const override;
     void SetCPSR(u32 cpsr) override;
 
-    u64 GetTicks() const override;
     void AddTicks(u64 ticks) override;
 
     void ResetContext(Core::ThreadContext& context, u32 stack_top, u32 entry_point, u32 arg);
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index e0689be2e..138603d9b 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -7,36 +7,30 @@
 #include "core/hle/service/hid/hid_spvr.h"
 #include "core/hle/service/hid/hid_user.h"
 
-#include "core/arm/arm_interface.h"
+#include "core/core_timing.h"
 #include "core/hle/kernel/event.h"
 #include "core/hle/kernel/shared_memory.h"
 #include "core/hle/hle.h"
 
+#include "video_core/video_core.h"
+
 namespace Service {
 namespace HID {
 
-Kernel::SharedPtr<Kernel::SharedMemory> g_shared_mem = nullptr;
-
-Kernel::SharedPtr<Kernel::Event> g_event_pad_or_touch_1;
-Kernel::SharedPtr<Kernel::Event> g_event_pad_or_touch_2;
-Kernel::SharedPtr<Kernel::Event> g_event_accelerometer;
-Kernel::SharedPtr<Kernel::Event> g_event_gyroscope;
-Kernel::SharedPtr<Kernel::Event> g_event_debug_pad;
-
-// Next Pad state update information
-static PadState next_state = {{0}};
-static u32 next_index = 0;
-static s16 next_circle_x = 0;
-static s16 next_circle_y = 0;
-
-/**
- * Gets a pointer to the PadData structure inside HID shared memory
- */
-static inline PadData* GetPadData() {
-    if (g_shared_mem == nullptr)
-        return nullptr;
-    return reinterpret_cast<PadData*>(g_shared_mem->GetPointer().ValueOr(nullptr));
-}
+static const int MAX_CIRCLEPAD_POS = 0x9C; ///< Max value for a circle pad position
+
+// Handle to shared memory region designated to HID_User service
+static Kernel::SharedPtr<Kernel::SharedMemory> shared_mem = nullptr;
+
+// Event handles
+static Kernel::SharedPtr<Kernel::Event> event_pad_or_touch_1 = nullptr;
+static Kernel::SharedPtr<Kernel::Event> event_pad_or_touch_2 = nullptr;
+static Kernel::SharedPtr<Kernel::Event> event_accelerometer = nullptr;
+static Kernel::SharedPtr<Kernel::Event> event_gyroscope = nullptr;
+static Kernel::SharedPtr<Kernel::Event> event_debug_pad = nullptr;
+
+static u32 next_pad_index = 0;
+static u32 next_touch_index = 0;
 
 // TODO(peachum):
 // Add a method for setting analog input from joystick device for the circle Pad.
@@ -51,103 +45,114 @@ static inline PadData* GetPadData() {
 //     * Set PadData.current_state.circle_left = 1 if current PadEntry.circle_pad_x <= -41
 //     * Set PadData.current_state.circle_right = 1 if current PadEntry.circle_pad_y <= -41
 
-/**
- * Circle Pad from keys.
- *
- * This is implemented as "pushed all the way to an edge (max) or centered (0)".
- *
- * Indicate the circle pad is pushed completely to the edge in 1 of 8 directions.
- */
-static void UpdateNextCirclePadState() {
-    static const s16 max_value = 0x9C;
-    next_circle_x = next_state.circle_left ? -max_value : 0x0;
-    next_circle_x += next_state.circle_right ? max_value : 0x0;
-    next_circle_y = next_state.circle_down ? -max_value : 0x0;
-    next_circle_y += next_state.circle_up ? max_value : 0x0;
-}
-
-/**
- * Sets a Pad state (button or button combo) as pressed
- */
-void PadButtonPress(const PadState& pad_state) {
-    next_state.hex |= pad_state.hex;
-    UpdateNextCirclePadState();
-}
-
-/**
- * Sets a Pad state (button or button combo) as released
- */
-void PadButtonRelease(const PadState& pad_state) {
-    next_state.hex &= ~pad_state.hex;
-    UpdateNextCirclePadState();
-}
-
-/**
- * Called after all Pad changes to be included in this update have been made,
- * including both Pad key changes and analog circle Pad changes.
- */
-void PadUpdateComplete() {
-    PadData* pad_data = GetPadData();
+void HIDUpdate() {
+    SharedMem* mem = reinterpret_cast<SharedMem*>(shared_mem->GetPointer().ValueOr(nullptr));
+    const PadState state = VideoCore::g_emu_window->GetPadState();
 
-    if (pad_data == nullptr) {
+    if (mem == nullptr) {
+        LOG_DEBUG(Service_HID, "Cannot update HID prior to mapping shared memory!");
         return;
     }
 
-    // Update PadData struct
-    pad_data->current_state.hex = next_state.hex;
-    pad_data->index = next_index;
-    next_index = (next_index + 1) % pad_data->entries.size();
+    mem->pad.current_state.hex = state.hex;
+    mem->pad.index = next_pad_index;
+    ++next_touch_index %= mem->pad.entries.size();
 
     // Get the previous Pad state
-    u32 last_entry_index = (pad_data->index - 1) % pad_data->entries.size();
-    PadState old_state = pad_data->entries[last_entry_index].current_state;
+    u32 last_entry_index = (mem->pad.index - 1) % mem->pad.entries.size();
+    PadState old_state = mem->pad.entries[last_entry_index].current_state;
 
     // Compute bitmask with 1s for bits different from the old state
-    PadState changed;
-    changed.hex = (next_state.hex ^ old_state.hex);
-
-    // Compute what was added
-    PadState additions;
-    additions.hex = changed.hex & next_state.hex;
-
-    // Compute what was removed
-    PadState removals;
-    removals.hex = changed.hex & old_state.hex;
+    PadState changed = { { (state.hex ^ old_state.hex) } };
 
     // Get the current Pad entry
-    PadDataEntry* current_pad_entry = &pad_data->entries[pad_data->index];
+    PadDataEntry* pad_entry = &mem->pad.entries[mem->pad.index];
 
     // Update entry properties
-    current_pad_entry->current_state.hex = next_state.hex;
-    current_pad_entry->delta_additions.hex = additions.hex;
-    current_pad_entry->delta_removals.hex = removals.hex;
+    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;;
 
     // Set circle Pad
-    current_pad_entry->circle_pad_x = next_circle_x;
-    current_pad_entry->circle_pad_y = next_circle_y;
+    pad_entry->circle_pad_x = state.circle_left  ? -MAX_CIRCLEPAD_POS :
+                              state.circle_right ?  MAX_CIRCLEPAD_POS : 0x0;
+    pad_entry->circle_pad_y = state.circle_down  ? -MAX_CIRCLEPAD_POS :
+                              state.circle_up    ?  MAX_CIRCLEPAD_POS : 0x0;
 
     // If we just updated index 0, provide a new timestamp
-    if (pad_data->index == 0) {
-        pad_data->index_reset_ticks_previous = pad_data->index_reset_ticks;
-        pad_data->index_reset_ticks = (s64)Core::g_app_core->GetTicks();
+    if (mem->pad.index == 0) {
+        mem->pad.index_reset_ticks_previous = mem->pad.index_reset_ticks;
+        mem->pad.index_reset_ticks = (s64)CoreTiming::GetTicks();
+    }
+
+    mem->touch.index = next_touch_index;
+    ++next_touch_index %= mem->touch.entries.size();
+
+    // Get the current touch entry
+    TouchDataEntry* touch_entry = &mem->touch.entries[mem->touch.index];
+    bool pressed = false;
+
+    std::tie(touch_entry->x, touch_entry->y, pressed) = VideoCore::g_emu_window->GetTouchState();
+    touch_entry->valid = pressed ? 1 : 0;
+
+    // TODO(bunnei): We're not doing anything with offset 0xA8 + 0x18 of HID SharedMemory, which
+    // supposedly is "Touch-screen entry, which contains the raw coordinate data prior to being
+    // converted to pixel coordinates." (http://3dbrew.org/wiki/HID_Shared_Memory#Offset_0xA8).
+
+    // If we just updated index 0, provide a new timestamp
+    if (mem->touch.index == 0) {
+        mem->touch.index_reset_ticks_previous = mem->touch.index_reset_ticks;
+        mem->touch.index_reset_ticks = (s64)CoreTiming::GetTicks();
     }
     
     // Signal both handles when there's an update to Pad or touch
-    g_event_pad_or_touch_1->Signal();
-    g_event_pad_or_touch_2->Signal();
+    event_pad_or_touch_1->Signal();
+    event_pad_or_touch_2->Signal();
 }
 
 void GetIPCHandles(Service::Interface* self) {
     u32* cmd_buff = Kernel::GetCommandBuffer();
 
     cmd_buff[1] = 0; // No error
+    cmd_buff[2] = 0x14000000; // IPC Command Structure translate-header
     // TODO(yuriks): Return error from SendSyncRequest is this fails (part of IPC marshalling)
-    cmd_buff[3] = Kernel::g_handle_table.Create(Service::HID::g_shared_mem).MoveFrom();
-    cmd_buff[4] = Kernel::g_handle_table.Create(Service::HID::g_event_pad_or_touch_1).MoveFrom();
-    cmd_buff[5] = Kernel::g_handle_table.Create(Service::HID::g_event_pad_or_touch_2).MoveFrom();
-    cmd_buff[6] = Kernel::g_handle_table.Create(Service::HID::g_event_accelerometer).MoveFrom();
-    cmd_buff[7] = Kernel::g_handle_table.Create(Service::HID::g_event_gyroscope).MoveFrom();
-    cmd_buff[8] = Kernel::g_handle_table.Create(Service::HID::g_event_debug_pad).MoveFrom();
+    cmd_buff[3] = Kernel::g_handle_table.Create(Service::HID::shared_mem).MoveFrom();
+    cmd_buff[4] = Kernel::g_handle_table.Create(Service::HID::event_pad_or_touch_1).MoveFrom();
+    cmd_buff[5] = Kernel::g_handle_table.Create(Service::HID::event_pad_or_touch_2).MoveFrom();
+    cmd_buff[6] = Kernel::g_handle_table.Create(Service::HID::event_accelerometer).MoveFrom();
+    cmd_buff[7] = Kernel::g_handle_table.Create(Service::HID::event_gyroscope).MoveFrom();
+    cmd_buff[8] = Kernel::g_handle_table.Create(Service::HID::event_debug_pad).MoveFrom();
+}
+
+void EnableAccelerometer(Service::Interface* self) {
+    u32* cmd_buff = Kernel::GetCommandBuffer();
+
+    event_accelerometer->Signal();
+
+    cmd_buff[1] = RESULT_SUCCESS.raw;
+
+    LOG_WARNING(Service_HID, "(STUBBED) called");
+}
+
+void EnableGyroscopeLow(Service::Interface* self) {
+    u32* cmd_buff = Kernel::GetCommandBuffer();
+
+    event_gyroscope->Signal();
+
+    cmd_buff[1] = RESULT_SUCCESS.raw;
+
+    LOG_WARNING(Service_HID, "(STUBBED) called");
+}
+
+void GetSoundVolume(Service::Interface* self) {
+    u32* cmd_buff = Kernel::GetCommandBuffer();
+
+    const u8 volume = 0x3F; // TODO(purpasmart): Find out if this is the max value for the volume
+
+    cmd_buff[1] = RESULT_SUCCESS.raw;
+    cmd_buff[2] = volume;
+
+    LOG_WARNING(Service_HID, "(STUBBED) called");
 }
 
 void HIDInit() {
@@ -156,19 +161,22 @@ void HIDInit() {
     AddService(new HID_U_Interface);
     AddService(new HID_SPVR_Interface);
 
-    g_shared_mem = SharedMemory::Create("HID:SharedMem");
+    shared_mem = SharedMemory::Create("HID:SharedMem");
+
+    next_pad_index = 0;
+    next_touch_index = 0;
 
     // Create event handles
-    g_event_pad_or_touch_1 = Event::Create(RESETTYPE_ONESHOT, "HID:EventPadOrTouch1");
-    g_event_pad_or_touch_2 = Event::Create(RESETTYPE_ONESHOT, "HID:EventPadOrTouch2");
-    g_event_accelerometer  = Event::Create(RESETTYPE_ONESHOT, "HID:EventAccelerometer");
-    g_event_gyroscope      = Event::Create(RESETTYPE_ONESHOT, "HID:EventGyroscope");
-    g_event_debug_pad      = Event::Create(RESETTYPE_ONESHOT, "HID:EventDebugPad");
+    event_pad_or_touch_1 = Event::Create(RESETTYPE_ONESHOT, "HID:EventPadOrTouch1");
+    event_pad_or_touch_2 = Event::Create(RESETTYPE_ONESHOT, "HID:EventPadOrTouch2");
+    event_accelerometer  = Event::Create(RESETTYPE_ONESHOT, "HID:EventAccelerometer");
+    event_gyroscope      = Event::Create(RESETTYPE_ONESHOT, "HID:EventGyroscope");
+    event_debug_pad      = Event::Create(RESETTYPE_ONESHOT, "HID:EventDebugPad");
 }
 
 void HIDShutdown() {
-
 }
 
-}
-}
+} // namespace HID
+
+} // namespace Service
diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h
index 9c6e86f77..97462c7f8 100644
--- a/src/core/hle/service/hid/hid.h
+++ b/src/core/hle/service/hid/hid.h
@@ -18,16 +18,6 @@ namespace Kernel {
 namespace Service {
 namespace HID {
 
-// Handle to shared memory region designated to HID_User service
-extern Kernel::SharedPtr<Kernel::SharedMemory> g_shared_mem;
-
-// Event handles
-extern Kernel::SharedPtr<Kernel::Event> g_event_pad_or_touch_1;
-extern Kernel::SharedPtr<Kernel::Event> g_event_pad_or_touch_2;
-extern Kernel::SharedPtr<Kernel::Event> g_event_accelerometer;
-extern Kernel::SharedPtr<Kernel::Event> g_event_gyroscope;
-extern Kernel::SharedPtr<Kernel::Event> g_event_debug_pad;
-
 /**
  * Structure of a Pad controller state.
  */
@@ -65,7 +55,7 @@ struct PadState {
 };
 
 /**
- * Structure of a single entry in the PadData's Pad state history array.
+ * Structure of a single entry of Pad state history within HID shared memory
  */
 struct PadDataEntry {
     PadState current_state;
@@ -77,24 +67,65 @@ struct PadDataEntry {
 };
 
 /**
- * Structure of all data related to the 3DS Pad.
+ * Structure of a single entry of touch state history within HID shared memory
  */
-struct PadData {
-    s64 index_reset_ticks;
-    s64 index_reset_ticks_previous;
-    u32 index; // the index of the last updated Pad state history element
+struct TouchDataEntry {
+    u16 x;                     ///< Y-coordinate of a touchpad press on the lower screen
+    u16 y;                     ///< X-coordinate of a touchpad press on the lower screen
+    BitField<0, 7, u32> valid; ///< Set to 1 when this entry contains actual X/Y data, otherwise 0
+};
+
+/**
+ * Structure of data stored in HID shared memory
+ */
+struct SharedMem {
+    /// Pad data, this is used for buttons and the circle pad
+    struct {
+        s64 index_reset_ticks; ///< CPU tick count for when HID module updated entry index 0
+        s64 index_reset_ticks_previous; ///< Previous `index_reset_ticks`
+        u32 index; ///< Index of the last updated pad state entry
+
+        INSERT_PADDING_WORDS(0x2);
+
+        PadState current_state; ///< Current state of the pad buttons
+
+        // TODO(bunnei): Implement `raw_circle_pad_data` field
+        u32 raw_circle_pad_data; ///< Raw (analog) circle pad data, before being converted
 
-    u32 pad1;
-    u32 pad2;
+        INSERT_PADDING_WORDS(0x1);
 
-    PadState current_state; // same as entries[index].current_state
-    u32 raw_circle_pad_data;
+        std::array<PadDataEntry, 8> entries; ///< Last 8 pad entries
+    } pad;
 
-    u32 pad3;
+    /// Touchpad data, this is used for touchpad input
+    struct {
+        s64 index_reset_ticks; ///< CPU tick count for when HID module updated entry index 0
+        s64 index_reset_ticks_previous; ///< Previous `index_reset_ticks`
+        u32 index; ///< Index of the last updated touch entry
 
-    std::array<PadDataEntry, 8> entries; // Pad state history
+        INSERT_PADDING_WORDS(0x1);
+
+        // TODO(bunnei): Implement `raw_entry` field
+        TouchDataEntry raw_entry; ///< Raw (analog) touch data, before being converted
+
+        std::array<TouchDataEntry, 8> entries; ///< Last 8 touch entries, in pixel coordinates
+    } touch;
 };
 
+// TODO: MSVC does not support using offsetof() on non-static data members even though this
+//       is technically allowed since C++11. This macro should be enabled once MSVC adds
+//       support for that.
+#ifndef _MSC_VER
+#define ASSERT_REG_POSITION(field_name, position)                  \
+    static_assert(offsetof(SharedMem, field_name) == position * 4, \
+                  "Field "#field_name" has invalid position")
+
+ASSERT_REG_POSITION(pad.index_reset_ticks, 0x0);
+ASSERT_REG_POSITION(touch.index_reset_ticks, 0x2A);
+
+#undef ASSERT_REG_POSITION
+#endif // !defined(_MSC_VER)
+
 // Pre-defined PadStates for single button presses
 const PadState PAD_NONE         = {{0}};
 const PadState PAD_A            = {{1u << 0}};
@@ -130,7 +161,7 @@ const PadState PAD_CIRCLE_DOWN  = {{1u << 31}};
  *      None
  *  Outputs:
  *      1 : Result of function, 0 on success, otherwise error code
- *      2 : Unused
+ *      2 : IPC Command Structure translate-header
  *      3 : Handle to HID_User shared memory
  *      4 : Event signaled by HID_User
  *      5 : Event signaled by HID_User
@@ -140,12 +171,41 @@ const PadState PAD_CIRCLE_DOWN  = {{1u << 31}};
  */
 void GetIPCHandles(Interface* self);
 
-// Methods for updating the HID module's state
-void PadButtonPress(const PadState& pad_state);
-void PadButtonRelease(const PadState& pad_state);
-void PadUpdateComplete();
+/**
+ * HID::EnableAccelerometer service function
+ *  Inputs:
+ *      None
+ *  Outputs:
+ *      1 : Result of function, 0 on success, otherwise error code
+ */
+void EnableAccelerometer(Interface* self);
+
+/**
+ * HID::EnableGyroscopeLow service function
+ *  Inputs:
+ *      None
+ *  Outputs:
+ *      1 : Result of function, 0 on success, otherwise error code
+ */
+void EnableGyroscopeLow(Interface* self);
+
+/**
+ * HID::GetSoundVolume service function
+ *  Inputs:
+ *      None
+ *  Outputs:
+ *      1 : Result of function, 0 on success, otherwise error code
+ *      2 : u8 output value
+ */
+void GetSoundVolume(Interface* self);
 
+/// Checks for user input updates
+void HIDUpdate();
+
+/// Initialize HID service
 void HIDInit();
+
+/// Shutdown HID service
 void HIDShutdown();
 
 }
diff --git a/src/core/hle/service/hid/hid_spvr.cpp b/src/core/hle/service/hid/hid_spvr.cpp
index 790dcabbf..f296b076f 100644
--- a/src/core/hle/service/hid/hid_spvr.cpp
+++ b/src/core/hle/service/hid/hid_spvr.cpp
@@ -13,13 +13,13 @@ const Interface::FunctionInfo FunctionTable[] = {
     {0x000A0000, GetIPCHandles,              "GetIPCHandles"},
     {0x000B0000, nullptr,                    "StartAnalogStickCalibration"},
     {0x000E0000, nullptr,                    "GetAnalogStickCalibrateParam"},
-    {0x00110000, nullptr,                    "EnableAccelerometer"},
+    {0x00110000, EnableAccelerometer,        "EnableAccelerometer"},
     {0x00120000, nullptr,                    "DisableAccelerometer"},
-    {0x00130000, nullptr,                    "EnableGyroscopeLow"},
+    {0x00130000, EnableGyroscopeLow,         "EnableGyroscopeLow"},
     {0x00140000, nullptr,                    "DisableGyroscopeLow"},
     {0x00150000, nullptr,                    "GetGyroscopeLowRawToDpsCoefficient"},
     {0x00160000, nullptr,                    "GetGyroscopeLowCalibrateParam"},
-    {0x00170000, nullptr,                    "GetSoundVolume"},
+    {0x00170000, GetSoundVolume,             "GetSoundVolume"},
 };
 
 HID_SPVR_Interface::HID_SPVR_Interface() {
diff --git a/src/core/hle/service/hid/hid_user.cpp b/src/core/hle/service/hid/hid_user.cpp
index 1d0accefe..3682c9416 100644
--- a/src/core/hle/service/hid/hid_user.cpp
+++ b/src/core/hle/service/hid/hid_user.cpp
@@ -3,8 +3,6 @@
 // Refer to the license.txt file included.
 
 #include "core/hle/hle.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/hid/hid_user.h"
 
@@ -12,14 +10,14 @@ namespace Service {
 namespace HID {
 
 const Interface::FunctionInfo FunctionTable[] = {
-    {0x000A0000, GetIPCHandles, "GetIPCHandles"},
-    {0x00110000, nullptr,       "EnableAccelerometer"},
-    {0x00120000, nullptr,       "DisableAccelerometer"},
-    {0x00130000, nullptr,       "EnableGyroscopeLow"},
-    {0x00140000, nullptr,       "DisableGyroscopeLow"},
-    {0x00150000, nullptr,       "GetGyroscopeLowRawToDpsCoefficient"},
-    {0x00160000, nullptr,       "GetGyroscopeLowCalibrateParam"},
-    {0x00170000, nullptr,       "GetSoundVolume"},
+    {0x000A0000, GetIPCHandles,             "GetIPCHandles"},
+    {0x00110000, EnableAccelerometer,       "EnableAccelerometer"},
+    {0x00120000, nullptr,                   "DisableAccelerometer"},
+    {0x00130000, EnableGyroscopeLow,        "EnableGyroscopeLow"},
+    {0x00140000, nullptr,                   "DisableGyroscopeLow"},
+    {0x00150000, nullptr,                   "GetGyroscopeLowRawToDpsCoefficient"},
+    {0x00160000, nullptr,                   "GetGyroscopeLowCalibrateParam"},
+    {0x00170000, GetSoundVolume,            "GetSoundVolume"},
 };
 
 HID_U_Interface::HID_U_Interface() {
diff --git a/src/core/hle/svc.cpp b/src/core/hle/svc.cpp
index 17385f9b2..bbb4eb9cd 100644
--- a/src/core/hle/svc.cpp
+++ b/src/core/hle/svc.cpp
@@ -7,8 +7,9 @@
 #include "common/string_util.h"
 #include "common/symbols.h"
 
-#include "core/arm/arm_interface.h"
+#include "core/core_timing.h"
 #include "core/mem_map.h"
+#include "core/arm/arm_interface.h"
 
 #include "core/hle/kernel/address_arbiter.h"
 #include "core/hle/kernel/event.h"
@@ -551,7 +552,7 @@ static void SleepThread(s64 nanoseconds) {
 
 /// This returns the total CPU ticks elapsed since the CPU was powered-on
 static s64 GetSystemTick() {
-    return (s64)Core::g_app_core->GetTicks();
+    return (s64)CoreTiming::GetTicks();
 }
 
 /// Creates a memory block at the specified address with the specified permissions and size
diff --git a/src/core/hw/gpu.cpp b/src/core/hw/gpu.cpp
index 30318fc06..e6983a225 100644
--- a/src/core/hw/gpu.cpp
+++ b/src/core/hw/gpu.cpp
@@ -14,6 +14,7 @@
 #include "core/hle/hle.h"
 #include "core/hle/service/gsp_gpu.h"
 #include "core/hle/service/dsp_dsp.h"
+#include "core/hle/service/hid/hid.h"
 
 #include "core/hw/hw.h"
 #include "core/hw/gpu.h"
@@ -117,8 +118,14 @@ inline void Write(u32 addr, const T data) {
             u8* src_pointer = Memory::GetPointer(Memory::PhysicalToVirtualAddress(config.GetPhysicalInputAddress()));
             u8* dst_pointer = Memory::GetPointer(Memory::PhysicalToVirtualAddress(config.GetPhysicalOutputAddress()));
 
-            unsigned horizontal_scale = (config.scale_horizontally != 0) ? 2 : 1;
-            unsigned vertical_scale = (config.scale_vertically != 0) ? 2 : 1;
+            if (config.scaling > config.ScaleXY) {
+                LOG_CRITICAL(HW_GPU, "Unimplemented display transfer scaling mode %u", config.scaling.Value());
+                UNIMPLEMENTED();
+                break;
+            }
+
+            unsigned horizontal_scale = (config.scaling != config.NoScale) ? 2 : 1;
+            unsigned vertical_scale = (config.scaling == config.ScaleXY) ? 2 : 1;
 
             u32 output_width = config.output_width / horizontal_scale;
             u32 output_height = config.output_height / vertical_scale;
@@ -139,14 +146,23 @@ inline void Write(u32 addr, const T data) {
                 break;
             }
 
-            // TODO(Subv): Blend the pixels when horizontal / vertical scaling is enabled, 
+            // TODO(Subv): Implement the box filter when scaling is enabled
             // right now we're just skipping the extra pixels.
             for (u32 y = 0; y < output_height; ++y) {
                 for (u32 x = 0; x < output_width; ++x) {
                     Math::Vec4<u8> src_color = { 0, 0, 0, 0 };
 
-                    u32 scaled_x = x * horizontal_scale;
-                    u32 scaled_y = y * vertical_scale;
+                    // Calculate the [x,y] position of the input image 
+                    // based on the current output position and the scale
+                    u32 input_x = x * horizontal_scale;
+                    u32 input_y = y * vertical_scale;
+
+                    if (config.flip_vertically) {
+                        // Flip the y value of the output data, 
+                        // we do this after calculating the [x,y] position of the input image 
+                        // to account for the scaling options.
+                        y = output_height - y - 1;
+                    }
 
                     u32 dst_bytes_per_pixel = GPU::Regs::BytesPerPixel(config.output_format);
                     u32 src_bytes_per_pixel = GPU::Regs::BytesPerPixel(config.input_format);
@@ -158,14 +174,14 @@ inline void Write(u32 addr, const T data) {
                         u32 coarse_y = y & ~7;
                         u32 stride = output_width * dst_bytes_per_pixel;
 
-                        src_offset = (scaled_x + scaled_y * config.input_width) * src_bytes_per_pixel;
+                        src_offset = (input_x + input_y * config.input_width) * src_bytes_per_pixel;
                         dst_offset = VideoCore::GetMortonOffset(x, y, dst_bytes_per_pixel) + coarse_y * stride;
                     } else {
                         // Interpret the input as tiled and the output as linear
-                        u32 coarse_y = scaled_y & ~7;
+                        u32 coarse_y = input_y & ~7;
                         u32 stride = config.input_width * src_bytes_per_pixel;
 
-                        src_offset = VideoCore::GetMortonOffset(scaled_x, scaled_y, src_bytes_per_pixel) + coarse_y * stride;
+                        src_offset = VideoCore::GetMortonOffset(input_x, input_y, src_bytes_per_pixel) + coarse_y * stride;
                         dst_offset = (x + y * output_width) * dst_bytes_per_pixel;
                     }
 
@@ -295,6 +311,9 @@ static void VBlankCallback(u64 userdata, int cycles_late) {
     // this. Certain games expect this to be periodically signaled.
     DSP_DSP::SignalInterrupt();
 
+    // Check for user input updates
+    Service::HID::HIDUpdate();
+
     // Reschedule recurrent event
     CoreTiming::ScheduleEvent(frame_ticks - cycles_late, vblank_event);
 }
diff --git a/src/core/hw/gpu.h b/src/core/hw/gpu.h
index 5b7f0a4e9..c8f884494 100644
--- a/src/core/hw/gpu.h
+++ b/src/core/hw/gpu.h
@@ -188,17 +188,22 @@ struct Regs {
             BitField<16, 16, u32> input_height;
         };
 
+        enum ScalingMode : u32 {
+            NoScale  = 0,  // Doesn't scale the image
+            ScaleX   = 1,  // Downscales the image in half in the X axis and applies a box filter
+            ScaleXY  = 2,  // Downscales the image in half in both the X and Y axes and applies a box filter
+        };
+
         union {
             u32 flags;
 
-            BitField< 0, 1, u32> flip_data;        // flips input data horizontally (TODO) if true
+            BitField< 0, 1, u32> flip_vertically;  // flips input data vertically
             BitField< 1, 1, u32> output_tiled;     // Converts from linear to tiled format
             BitField< 3, 1, u32> raw_copy;         // Copies the data without performing any processing
             BitField< 8, 3, PixelFormat> input_format;
             BitField<12, 3, PixelFormat> output_format;
 
-            BitField<24, 1, u32> scale_horizontally;
-            BitField<25, 1, u32> scale_vertically;
+            BitField<24, 2, ScalingMode> scaling; // Determines the scaling mode of the transfer
         };
 
         INSERT_PADDING_WORDS(0x1);
diff --git a/src/core/loader/ncch.h b/src/core/loader/ncch.h
index 9ae2de99f..f6f670060 100644
--- a/src/core/loader/ncch.h
+++ b/src/core/loader/ncch.h
@@ -20,8 +20,8 @@ struct NCCH_Header {
     u16 version;
     u8 reserved_0[4];
     u8 program_id[8];
-    u8 temp_flag;
-    u8 reserved_1[0x2f];
+    u8 reserved_1[0x10];
+    u8 logo_region_hash[0x20];
     u8 product_code[0x10];
     u8 extended_header_hash[0x20];
     u32 extended_header_size;
@@ -29,15 +29,16 @@ struct NCCH_Header {
     u8 flags[8];
     u32 plain_region_offset;
     u32 plain_region_size;
-    u8 reserved_3[8];
+    u32 logo_region_offset;
+    u32 logo_region_size;
     u32 exefs_offset;
     u32 exefs_size;
     u32 exefs_hash_region_size;
-    u8 reserved_4[4];
+    u8 reserved_3[4];
     u32 romfs_offset;
     u32 romfs_size;
     u32 romfs_hash_region_size;
-    u8 reserved_5[4];
+    u8 reserved_4[4];
     u8 exefs_super_block_hash[0x20];
     u8 romfs_super_block_hash[0x20];
 };
@@ -88,8 +89,7 @@ struct ExHeader_DependencyList{
 };
 
 struct ExHeader_SystemInfo{
-    u32 save_data_size;
-    u8 reserved[4];
+    u64 save_data_size;
     u8 jump_id[8];
     u8 reserved_2[0x30];
 };
@@ -104,11 +104,14 @@ struct ExHeader_StorageInfo{
 
 struct ExHeader_ARM11_SystemLocalCaps{
     u8 program_id[8];
-    u8 flags[8];
-    u8 resource_limit_descriptor[0x10][2];
+    u32 core_version;
+    u8 flags[3];
+    u8 priority;
+    u8 resource_limit_descriptor[0x16][2];
     ExHeader_StorageInfo storage_info;
-    u8 service_access_control[0x20][8];
-    u8 reserved[0x1f];
+    u8 service_access_control[0x32][8];
+    u8 ex_service_access_control[0x2][8];
+    u8 reserved[0xf];
     u8 resource_limit_category;
 };
 
diff --git a/src/video_core/color.h b/src/video_core/color.h
index 14ade74f2..43d635e2c 100644
--- a/src/video_core/color.h
+++ b/src/video_core/color.h
@@ -124,7 +124,7 @@ inline u32 DecodeD24(const u8* bytes) {
  * @return Resulting values stored as a Math::Vec2
  */
 inline const Math::Vec2<u32> DecodeD24S8(const u8* bytes) {
-    return { (bytes[2] << 16) | (bytes[1] << 8) | bytes[0], bytes[3] };
+    return { static_cast<u32>((bytes[2] << 16) | (bytes[1] << 8) | bytes[0]), bytes[3] };
 }
 
 /**
diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp
index 745c4f4ed..83982b4f2 100644
--- a/src/video_core/debug_utils/debug_utils.cpp
+++ b/src/video_core/debug_utils/debug_utils.cpp
@@ -322,7 +322,7 @@ const Math::Vec4<u8> LookupTexture(const u8* source, int x, int y, const Texture
     case Regs::TextureFormat::RGBA8:
     {
         auto res = Color::DecodeRGBA8(source + VideoCore::GetMortonOffset(x, y, 4));
-        return { res.r(), res.g(), res.b(), disable_alpha ? 255 : res.a() };
+        return { res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a()) };
     }
 
     case Regs::TextureFormat::RGB8:
@@ -334,7 +334,7 @@ const Math::Vec4<u8> LookupTexture(const u8* source, int x, int y, const Texture
     case Regs::TextureFormat::RGB5A1:
     {
         auto res = Color::DecodeRGB5A1(source + VideoCore::GetMortonOffset(x, y, 2));
-        return { res.r(), res.g(), res.b(), disable_alpha ? 255 : res.a() };
+        return { res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a()) };
     }
 
     case Regs::TextureFormat::RGB565:
@@ -346,7 +346,7 @@ const Math::Vec4<u8> LookupTexture(const u8* source, int x, int y, const Texture
     case Regs::TextureFormat::RGBA4:
     {
         auto res = Color::DecodeRGBA4(source + VideoCore::GetMortonOffset(x, y, 2));
-        return { res.r(), res.g(), res.b(), disable_alpha ? 255 : res.a() };
+        return { res.r(), res.g(), res.b(), static_cast<u8>(disable_alpha ? 255 : res.a()) };
     }
 
     case Regs::TextureFormat::IA8:
diff --git a/src/video_core/vertex_shader.cpp b/src/video_core/vertex_shader.cpp
index 4eb3e743e..e8d865172 100644
--- a/src/video_core/vertex_shader.cpp
+++ b/src/video_core/vertex_shader.cpp
@@ -72,7 +72,7 @@ struct VertexShaderState {
     u32* program_counter;
 
     const float24* input_register_table[16];
-    float24* output_register_table[7*4];
+    Math::Vec4<float24> output_registers[16];
 
     Math::Vec4<float24> temporary_registers[16];
     bool conditional_code[2];
@@ -198,8 +198,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
                 src2[3] = src2[3] * float24::FromFloat32(-1);
             }
 
-            float24* dest = (instr.common.dest.Value() < 0x08) ? state.output_register_table[4*instr.common.dest.Value().GetIndex()]
-                        : (instr.common.dest.Value() < 0x10) ? dummy_vec4_float24
+            float24* dest = (instr.common.dest.Value() < 0x10) ? &state.output_registers[instr.common.dest.Value().GetIndex()][0]
                         : (instr.common.dest.Value() < 0x20) ? &state.temporary_registers[instr.common.dest.Value().GetIndex()][0]
                         : dummy_vec4_float24;
 
@@ -409,8 +408,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
                     src3[3] = src3[3] * float24::FromFloat32(-1);
                 }
 
-                float24* dest = (instr.mad.dest.Value() < 0x08) ? state.output_register_table[4*instr.mad.dest.Value().GetIndex()]
-                            : (instr.mad.dest.Value() < 0x10) ? dummy_vec4_float24
+                float24* dest = (instr.mad.dest.Value() < 0x10) ? &state.output_registers[instr.mad.dest.Value().GetIndex()][0]
                             : (instr.mad.dest.Value() < 0x20) ? &state.temporary_registers[instr.mad.dest.Value().GetIndex()][0]
                             : dummy_vec4_float24;
 
@@ -587,12 +585,18 @@ OutputVertex RunShader(const InputVertex& input, int num_attributes) {
     if(num_attributes > 14) state.input_register_table[attribute_register_map.attribute14_register] = &input.attr[14].x;
     if(num_attributes > 15) state.input_register_table[attribute_register_map.attribute15_register] = &input.attr[15].x;
 
-    // Setup output register table
-    OutputVertex ret;
-    // Zero output so that attributes which aren't output won't have denormals in them, which will
-    // slow us down later.
-    memset(&ret, 0, sizeof(ret));
+    state.conditional_code[0] = false;
+    state.conditional_code[1] = false;
+
+    ProcessShaderCode(state);
+    DebugUtils::DumpShader(shader_memory.data(), state.debug.max_offset, swizzle_data.data(),
+                           state.debug.max_opdesc_id, registers.vs_main_offset,
+                           registers.vs_output_attributes);
 
+    // Setup output data
+    OutputVertex ret;
+    // TODO(neobrain): Under some circumstances, up to 16 attributes may be output. We need to
+    // figure out what those circumstances are and enable the remaining outputs then.
     for (int i = 0; i < 7; ++i) {
         const auto& output_register_map = registers.vs_output_attributes[i];
 
@@ -601,18 +605,18 @@ OutputVertex RunShader(const InputVertex& input, int num_attributes) {
             output_register_map.map_z, output_register_map.map_w
         };
 
-        for (int comp = 0; comp < 4; ++comp)
-            state.output_register_table[4*i+comp] = ((float24*)&ret) + semantics[comp];
+        for (int comp = 0; comp < 4; ++comp) {
+            float24* out = ((float24*)&ret) + semantics[comp];
+            if (semantics[comp] != Regs::VSOutputAttributes::INVALID) {
+                *out = state.output_registers[i][comp];
+            } else {
+                // Zero output so that attributes which aren't output won't have denormals in them,
+                // which would slow us down later.
+                memset(out, 0, sizeof(*out));
+            }
+        }
     }
 
-    state.conditional_code[0] = false;
-    state.conditional_code[1] = false;
-
-    ProcessShaderCode(state);
-    DebugUtils::DumpShader(shader_memory.data(), state.debug.max_offset, swizzle_data.data(),
-                           state.debug.max_opdesc_id, registers.vs_main_offset,
-                           registers.vs_output_attributes);
-
     LOG_TRACE(Render_Software, "Output vertex: pos (%.2f, %.2f, %.2f, %.2f), col(%.2f, %.2f, %.2f, %.2f), tc0(%.2f, %.2f)",
         ret.pos.x.ToFloat32(), ret.pos.y.ToFloat32(), ret.pos.z.ToFloat32(), ret.pos.w.ToFloat32(),
         ret.color.x.ToFloat32(), ret.color.y.ToFloat32(), ret.color.z.ToFloat32(), ret.color.w.ToFloat32(),