16 files changed, 335 insertions, 52 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 1f0af2d41..a61dee6e0 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -92,10 +92,13 @@ else()
     # /W3 - Level 3 warnings
     # /MP - Multi-threaded compilation
     # /Zi - Output debugging information
-    # /Zo - enahnced debug info for optimized builds
-    set(CMAKE_C_FLAGS   "/W3 /MP /Zi /Zo" CACHE STRING "" FORCE)
+    # /Zo - enhanced debug info for optimized builds
+    # /permissive- - enables stricter C++ standards conformance checks
+    set(CMAKE_C_FLAGS   "/W3 /MP /Zi /Zo /permissive-" CACHE STRING "" FORCE)
     # /EHsc - C++-only exception handling semantics
-    set(CMAKE_CXX_FLAGS "${CMAKE_C_FLAGS} /EHsc" CACHE STRING "" FORCE)
+    # /Zc:throwingNew - let codegen assume `operator new` will never return null
+    # /Zc:inline - let codegen omit inline functions in object files
+    set(CMAKE_CXX_FLAGS "${CMAKE_C_FLAGS} /EHsc /Zc:throwingNew,inline" CACHE STRING "" FORCE)
 
     # /MDd - Multi-threaded Debug Runtime DLL
     set(CMAKE_C_FLAGS_DEBUG   "/Od /MDd" CACHE STRING "" FORCE)
diff --git a/externals/CMakeLists.txt b/externals/CMakeLists.txt
index 1e04931ee..02e02350c 100644
--- a/externals/CMakeLists.txt
+++ b/externals/CMakeLists.txt
@@ -46,7 +46,5 @@ if (ARCHITECTURE_x86_64)
     # Defined before "dynarmic" above
     # add_library(xbyak INTERFACE)
     target_include_directories(xbyak INTERFACE ./xbyak/xbyak)
-    if (NOT MSVC)
-        target_compile_options(xbyak INTERFACE -fno-operator-names)
-    endif()
+    target_compile_definitions(xbyak INTERFACE XBYAK_NO_OP_NAMES)
 endif()
diff --git a/externals/dynarmic b/externals/dynarmic
-Subproject 7707ff13e981b0aecf87f3156ee0b641469f7bb
+Subproject 8f15e3f70cb96e56705e5de6ba97b5d09423a56
diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp
index 6cf1886cf..1cac1d0c9 100644
--- a/src/core/hle/kernel/hle_ipc.cpp
+++ b/src/core/hle/kernel/hle_ipc.cpp
@@ -23,6 +23,11 @@ void SessionRequestHandler::ClientDisconnected(SharedPtr<ServerSession> server_s
     boost::range::remove_erase(connected_sessions, server_session);
 }
 
+HLERequestContext::HLERequestContext(SharedPtr<ServerSession> session)
+    : session(std::move(session)) {
+    cmd_buf[0] = 0;
+}
+
 HLERequestContext::~HLERequestContext() = default;
 
 SharedPtr<Object> HLERequestContext::GetIncomingHandle(u32 id_from_cmdbuf) const {
diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h
index cbb109d8f..35795fc1d 100644
--- a/src/core/hle/kernel/hle_ipc.h
+++ b/src/core/hle/kernel/hle_ipc.h
@@ -84,6 +84,7 @@ protected:
  */
 class HLERequestContext {
 public:
+    HLERequestContext(SharedPtr<ServerSession> session);
     ~HLERequestContext();
 
     /// Returns a pointer to the IPC command buffer for this request.
@@ -118,14 +119,14 @@ public:
      */
     void ClearIncomingObjects();
 
-private:
-    friend class Service::ServiceFrameworkBase;
-
+    /// Populates this context with data from the requesting process/thread.
     ResultCode PopulateFromIncomingCommandBuffer(const u32_le* src_cmdbuf, Process& src_process,
                                                  HandleTable& src_table);
+    /// Writes data from this context back to the requesting process/thread.
     ResultCode WriteToOutgoingCommandBuffer(u32_le* dst_cmdbuf, Process& dst_process,
                                             HandleTable& dst_table) const;
 
+private:
     std::array<u32, IPC::COMMAND_BUFFER_LENGTH> cmd_buf;
     SharedPtr<ServerSession> session;
     // TODO(yuriks): Check common usage of this and optimize size accordingly
diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp
index 791a65c19..6754cfeea 100644
--- a/src/core/hle/service/service.cpp
+++ b/src/core/hle/service/service.cpp
@@ -173,8 +173,7 @@ void ServiceFrameworkBase::HandleSyncRequest(SharedPtr<ServerSession> server_ses
 
     // TODO(yuriks): The kernel should be the one handling this as part of translation after
     // everything else is migrated
-    Kernel::HLERequestContext context;
-    context.session = std::move(server_session);
+    Kernel::HLERequestContext context(std::move(server_session));
     context.PopulateFromIncomingCommandBuffer(cmd_buf, *Kernel::g_current_process,
                                               Kernel::g_handle_table);
 
diff --git a/src/tests/CMakeLists.txt b/src/tests/CMakeLists.txt
index 00d7c636a..a14df325a 100644
--- a/src/tests/CMakeLists.txt
+++ b/src/tests/CMakeLists.txt
@@ -1,8 +1,9 @@
 set(SRCS
-            glad.cpp
-            tests.cpp
             common/param_package.cpp
             core/file_sys/path_parser.cpp
+            core/hle/kernel/hle_ipc.cpp
+            glad.cpp
+            tests.cpp
             )
 
 set(HEADERS
diff --git a/src/tests/core/hle/kernel/hle_ipc.cpp b/src/tests/core/hle/kernel/hle_ipc.cpp
new file mode 100644
index 000000000..e07a28c5b
--- /dev/null
+++ b/src/tests/core/hle/kernel/hle_ipc.cpp
@@ -0,0 +1,193 @@
+// Copyright 2017 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <catch.hpp>
+#include "core/hle/ipc.h"
+#include "core/hle/kernel/client_port.h"
+#include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/event.h"
+#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/server_session.h"
+
+namespace Kernel {
+
+static SharedPtr<Object> MakeObject() {
+    return Event::Create(ResetType::OneShot);
+}
+
+TEST_CASE("HLERequestContext::PopoulateFromIncomingCommandBuffer", "[core][kernel]") {
+    auto session = std::get<SharedPtr<ServerSession>>(ServerSession::CreateSessionPair());
+    HLERequestContext context(std::move(session));
+
+    auto process = Process::Create(CodeSet::Create("", 0));
+    HandleTable handle_table;
+
+    SECTION("works with empty cmdbuf") {
+        const u32_le input[]{
+            IPC::MakeHeader(0x1234, 0, 0),
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        REQUIRE(context.CommandBuffer()[0] == 0x12340000);
+    }
+
+    SECTION("translates regular params") {
+        const u32_le input[]{
+            IPC::MakeHeader(0, 3, 0), 0x12345678, 0x21122112, 0xAABBCCDD,
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        auto* output = context.CommandBuffer();
+        REQUIRE(output[1] == 0x12345678);
+        REQUIRE(output[2] == 0x21122112);
+        REQUIRE(output[3] == 0xAABBCCDD);
+    }
+
+    SECTION("translates move handles") {
+        auto a = MakeObject();
+        Handle a_handle = handle_table.Create(a).Unwrap();
+        const u32_le input[]{
+            IPC::MakeHeader(0, 0, 2), IPC::MoveHandleDesc(1), a_handle,
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        auto* output = context.CommandBuffer();
+        REQUIRE(context.GetIncomingHandle(output[2]) == a);
+        REQUIRE(handle_table.GetGeneric(a_handle) == nullptr);
+    }
+
+    SECTION("translates copy handles") {
+        auto a = MakeObject();
+        Handle a_handle = handle_table.Create(a).Unwrap();
+        const u32_le input[]{
+            IPC::MakeHeader(0, 0, 2), IPC::CopyHandleDesc(1), a_handle,
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        auto* output = context.CommandBuffer();
+        REQUIRE(context.GetIncomingHandle(output[2]) == a);
+        REQUIRE(handle_table.GetGeneric(a_handle) == a);
+    }
+
+    SECTION("translates multi-handle descriptors") {
+        auto a = MakeObject();
+        auto b = MakeObject();
+        auto c = MakeObject();
+        const u32_le input[]{
+            IPC::MakeHeader(0, 0, 5),        IPC::MoveHandleDesc(2),
+            handle_table.Create(a).Unwrap(), handle_table.Create(b).Unwrap(),
+            IPC::MoveHandleDesc(1),          handle_table.Create(c).Unwrap(),
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        auto* output = context.CommandBuffer();
+        REQUIRE(context.GetIncomingHandle(output[2]) == a);
+        REQUIRE(context.GetIncomingHandle(output[3]) == b);
+        REQUIRE(context.GetIncomingHandle(output[5]) == c);
+    }
+
+    SECTION("translates CallingPid descriptors") {
+        const u32_le input[]{
+            IPC::MakeHeader(0, 0, 2), IPC::CallingPidDesc(), 0x98989898,
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        REQUIRE(context.CommandBuffer()[2] == process->process_id);
+    }
+
+    SECTION("translates mixed params") {
+        auto a = MakeObject();
+        const u32_le input[]{
+            IPC::MakeHeader(0, 2, 4),
+            0x12345678,
+            0xABCDEF00,
+            IPC::MoveHandleDesc(1),
+            handle_table.Create(a).Unwrap(),
+            IPC::CallingPidDesc(),
+            0,
+        };
+
+        context.PopulateFromIncomingCommandBuffer(input, *process, handle_table);
+
+        auto* output = context.CommandBuffer();
+        REQUIRE(output[1] == 0x12345678);
+        REQUIRE(output[2] == 0xABCDEF00);
+        REQUIRE(context.GetIncomingHandle(output[4]) == a);
+        REQUIRE(output[6] == process->process_id);
+    }
+}
+
+TEST_CASE("HLERequestContext::WriteToOutgoingCommandBuffer", "[core][kernel]") {
+    auto session = std::get<SharedPtr<ServerSession>>(ServerSession::CreateSessionPair());
+    HLERequestContext context(std::move(session));
+
+    auto process = Process::Create(CodeSet::Create("", 0));
+    HandleTable handle_table;
+    auto* input = context.CommandBuffer();
+    u32_le output[IPC::COMMAND_BUFFER_LENGTH];
+
+    SECTION("works with empty cmdbuf") {
+        input[0] = IPC::MakeHeader(0x1234, 0, 0);
+
+        context.WriteToOutgoingCommandBuffer(output, *process, handle_table);
+
+        REQUIRE(output[0] == 0x12340000);
+    }
+
+    SECTION("translates regular params") {
+        input[0] = IPC::MakeHeader(0, 3, 0);
+        input[1] = 0x12345678;
+        input[2] = 0x21122112;
+        input[3] = 0xAABBCCDD;
+
+        context.WriteToOutgoingCommandBuffer(output, *process, handle_table);
+
+        REQUIRE(output[1] == 0x12345678);
+        REQUIRE(output[2] == 0x21122112);
+        REQUIRE(output[3] == 0xAABBCCDD);
+    }
+
+    SECTION("translates move/copy handles") {
+        auto a = MakeObject();
+        auto b = MakeObject();
+        input[0] = IPC::MakeHeader(0, 0, 4);
+        input[1] = IPC::MoveHandleDesc(1);
+        input[2] = context.AddOutgoingHandle(a);
+        input[3] = IPC::CopyHandleDesc(1);
+        input[4] = context.AddOutgoingHandle(b);
+
+        context.WriteToOutgoingCommandBuffer(output, *process, handle_table);
+
+        REQUIRE(handle_table.GetGeneric(output[2]) == a);
+        REQUIRE(handle_table.GetGeneric(output[4]) == b);
+    }
+
+    SECTION("translates multi-handle descriptors") {
+        auto a = MakeObject();
+        auto b = MakeObject();
+        auto c = MakeObject();
+        input[0] = IPC::MakeHeader(0, 0, 5);
+        input[1] = IPC::MoveHandleDesc(2);
+        input[2] = context.AddOutgoingHandle(a);
+        input[3] = context.AddOutgoingHandle(b);
+        input[4] = IPC::CopyHandleDesc(1);
+        input[5] = context.AddOutgoingHandle(c);
+
+        context.WriteToOutgoingCommandBuffer(output, *process, handle_table);
+
+        REQUIRE(handle_table.GetGeneric(output[2]) == a);
+        REQUIRE(handle_table.GetGeneric(output[3]) == b);
+        REQUIRE(handle_table.GetGeneric(output[5]) == c);
+    }
+}
+
+} // namespace Kernel
diff --git a/src/video_core/regs_lighting.h b/src/video_core/regs_lighting.h
index fbfebc0a7..7221d1688 100644
--- a/src/video_core/regs_lighting.h
+++ b/src/video_core/regs_lighting.h
@@ -84,7 +84,7 @@ struct LightingRegs {
         NV = 2, // Cosine of the angle between the normal and the view vector
         LN = 3, // Cosine of the angle between the light and the normal vectors
         SP = 4, // Cosine of the angle between the light and the inverse spotlight vectors
-        CP = 5, // TODO: document and implement
+        CP = 5, // Cosine of the angle between the tangent and projection of half-angle vectors
     };
 
     enum class LightingBumpMode : u32 {
@@ -168,6 +168,8 @@ struct LightingRegs {
         union {
             BitField<0, 1, u32> directional;
             BitField<1, 1, u32> two_sided_diffuse; // When disabled, clamp dot-product to 0
+            BitField<2, 1, u32> geometric_factor_0;
+            BitField<3, 1, u32> geometric_factor_1;
         } config;
 
         BitField<0, 20, u32> dist_atten_bias;
diff --git a/src/video_core/regs_texturing.h b/src/video_core/regs_texturing.h
index 3f5355fa9..0b09f2299 100644
--- a/src/video_core/regs_texturing.h
+++ b/src/video_core/regs_texturing.h
@@ -30,10 +30,10 @@ struct TexturingRegs {
             Repeat = 2,
             MirroredRepeat = 3,
             // Mode 4-7 produces some weird result and may be just invalid:
-            // 4: Positive coord: clamp to edge; negative coord: repeat
-            // 5: Positive coord: clamp to border; negative coord: repeat
-            // 6: Repeat
-            // 7: Repeat
+            ClampToEdge2 = 4,   // Positive coord: clamp to edge; negative coord: repeat
+            ClampToBorder2 = 5, // Positive coord: clamp to border; negative coord: repeat
+            Repeat2 = 6,        // Same as Repeat
+            Repeat3 = 7,        // Same as Repeat
         };
 
         enum TextureFilter : u32 {
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index db53710aa..540cbb9d0 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -73,6 +73,8 @@ PicaShaderConfig PicaShaderConfig::BuildFromRegs(const Pica::Regs& regs) {
         state.lighting.light[light_index].num = num;
         state.lighting.light[light_index].directional = light.config.directional != 0;
         state.lighting.light[light_index].two_sided_diffuse = light.config.two_sided_diffuse != 0;
+        state.lighting.light[light_index].geometric_factor_0 = light.config.geometric_factor_0 != 0;
+        state.lighting.light[light_index].geometric_factor_1 = light.config.geometric_factor_1 != 0;
         state.lighting.light[light_index].dist_atten_enable =
             !regs.lighting.IsDistAttenDisabled(num);
         state.lighting.light[light_index].spot_atten_enable =
@@ -518,14 +520,16 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
            "vec4 specular_sum = vec4(0.0, 0.0, 0.0, 1.0);\n"
            "vec3 light_vector = vec3(0.0);\n"
            "vec3 refl_value = vec3(0.0);\n"
-           "vec3 spot_dir = vec3(0.0);\n;";
+           "vec3 spot_dir = vec3(0.0);\n"
+           "vec3 half_vector = vec3(0.0);\n"
+           "float geo_factor = 1.0;\n";
 
-    // Compute fragment normals
+    // Compute fragment normals and tangents
+    const std::string pertubation =
+        "2.0 * (" + SampleTexture(config, lighting.bump_selector) + ").rgb - 1.0";
     if (lighting.bump_mode == LightingRegs::LightingBumpMode::NormalMap) {
-        // Bump mapping is enabled using a normal map, read perturbation vector from the selected
-        // texture
-        out += "vec3 surface_normal = 2.0 * (" + SampleTexture(config, lighting.bump_selector) +
-               ").rgb - 1.0;\n";
+        // Bump mapping is enabled using a normal map
+        out += "vec3 surface_normal = " + pertubation + ";\n";
 
         // Recompute Z-component of perturbation if 'renorm' is enabled, this provides a higher
         // precision result
@@ -534,31 +538,41 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
                 "(1.0 - (surface_normal.x*surface_normal.x + surface_normal.y*surface_normal.y))";
             out += "surface_normal.z = sqrt(max(" + val + ", 0.0));\n";
         }
+
+        // The tangent vector is not perturbed by the normal map and is just a unit vector.
+        out += "vec3 surface_tangent = vec3(1.0, 0.0, 0.0);\n";
     } else if (lighting.bump_mode == LightingRegs::LightingBumpMode::TangentMap) {
         // Bump mapping is enabled using a tangent map
-        LOG_CRITICAL(HW_GPU, "unimplemented bump mapping mode (tangent mapping)");
-        UNIMPLEMENTED();
+        out += "vec3 surface_tangent = " + pertubation + ";\n";
+        // Mathematically, recomputing Z-component of the tangent vector won't affect the relevant
+        // computation below, which is also confirmed on 3DS. So we don't bother recomputing here
+        // even if 'renorm' is enabled.
+
+        // The normal vector is not perturbed by the tangent map and is just a unit vector.
+        out += "vec3 surface_normal = vec3(0.0, 0.0, 1.0);\n";
     } else {
-        // No bump mapping - surface local normal is just a unit normal
+        // No bump mapping - surface local normal and tangent are just unit vectors
         out += "vec3 surface_normal = vec3(0.0, 0.0, 1.0);\n";
+        out += "vec3 surface_tangent = vec3(1.0, 0.0, 0.0);\n";
     }
 
     // Rotate the surface-local normal by the interpolated normal quaternion to convert it to
     // eyespace.
-    out += "vec3 normal = quaternion_rotate(normalize(normquat), surface_normal);\n";
+    out += "vec4 normalized_normquat = normalize(normquat);\n";
+    out += "vec3 normal = quaternion_rotate(normalized_normquat, surface_normal);\n";
+    out += "vec3 tangent = quaternion_rotate(normalized_normquat, surface_tangent);\n";
 
     // Gets the index into the specified lookup table for specular lighting
     auto GetLutIndex = [&lighting](unsigned light_num, LightingRegs::LightingLutInput input,
                                    bool abs) {
-        const std::string half_angle = "normalize(normalize(view) + light_vector)";
         std::string index;
         switch (input) {
         case LightingRegs::LightingLutInput::NH:
-            index = "dot(normal, " + half_angle + ")";
+            index = "dot(normal, normalize(half_vector))";
             break;
 
         case LightingRegs::LightingLutInput::VH:
-            index = std::string("dot(normalize(view), " + half_angle + ")");
+            index = std::string("dot(normalize(view), normalize(half_vector))");
             break;
 
         case LightingRegs::LightingLutInput::NV:
@@ -573,6 +587,22 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
             index = std::string("dot(light_vector, spot_dir)");
             break;
 
+        case LightingRegs::LightingLutInput::CP:
+            // CP input is only available with configuration 7
+            if (lighting.config == LightingRegs::LightingConfig::Config7) {
+                // Note: even if the normal vector is modified by normal map, which is not the
+                // normal of the tangent plane anymore, the half angle vector is still projected
+                // using the modified normal vector.
+                std::string half_angle_proj = "normalize(half_vector) - normal / dot(normal, "
+                                              "normal) * dot(normal, normalize(half_vector))";
+                // Note: the half angle vector projection is confirmed not normalized before the dot
+                // product. The result is in fact not cos(phi) as the name suggested.
+                index = "dot(" + half_angle_proj + ", tangent)";
+            } else {
+                index = "0.0";
+            }
+            break;
+
         default:
             LOG_CRITICAL(HW_GPU, "Unknown lighting LUT input %d\n", (int)input);
             UNIMPLEMENTED();
@@ -610,6 +640,7 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
             out += "light_vector = normalize(" + light_src + ".position + view);\n";
 
         out += "spot_dir = " + light_src + ".spot_direction;\n";
+        out += "half_vector = normalize(view) + light_vector;\n";
 
         // Compute dot product of light_vector and normal, adjust if lighting is one-sided or
         // two-sided
@@ -643,6 +674,12 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
         std::string clamp_highlights =
             lighting.clamp_highlights ? "(dot(light_vector, normal) <= 0.0 ? 0.0 : 1.0)" : "1.0";
 
+        if (light_config.geometric_factor_0 || light_config.geometric_factor_1) {
+            out += "geo_factor = dot(half_vector, half_vector);\n"
+                   "geo_factor = geo_factor == 0.0 ? 0.0 : min(" +
+                   dot_product + " / geo_factor, 1.0);\n";
+        }
+
         // Specular 0 component
         std::string d0_lut_value = "1.0";
         if (lighting.lut_d0.enable &&
@@ -655,6 +692,9 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
                            GetLutValue(LightingRegs::LightingSampler::Distribution0, index) + ")";
         }
         std::string specular_0 = "(" + d0_lut_value + " * " + light_src + ".specular_0)";
+        if (light_config.geometric_factor_0) {
+            specular_0 = "(" + specular_0 + " * geo_factor)";
+        }
 
         // If enabled, lookup ReflectRed value, otherwise, 1.0 is used
         if (lighting.lut_rr.enable &&
@@ -710,6 +750,9 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) {
         }
         std::string specular_1 =
             "(" + d1_lut_value + " * refl_value * " + light_src + ".specular_1)";
+        if (light_config.geometric_factor_1) {
+            specular_1 = "(" + specular_1 + " * geo_factor)";
+        }
 
         // Fresnel
         if (lighting.lut_fr.enable &&
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.h b/src/video_core/renderer_opengl/gl_shader_gen.h
index 9c90eadf9..2302ae453 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.h
+++ b/src/video_core/renderer_opengl/gl_shader_gen.h
@@ -94,6 +94,8 @@ union PicaShaderConfig {
                 bool two_sided_diffuse;
                 bool dist_atten_enable;
                 bool spot_atten_enable;
+                bool geometric_factor_0;
+                bool geometric_factor_1;
             } light[8];
 
             bool enable;
diff --git a/src/video_core/renderer_opengl/pica_to_gl.h b/src/video_core/renderer_opengl/pica_to_gl.h
index 93d7b0b71..70298e211 100644
--- a/src/video_core/renderer_opengl/pica_to_gl.h
+++ b/src/video_core/renderer_opengl/pica_to_gl.h
@@ -55,6 +55,12 @@ inline GLenum WrapMode(Pica::TexturingRegs::TextureConfig::WrapMode mode) {
         GL_CLAMP_TO_BORDER, // WrapMode::ClampToBorder
         GL_REPEAT,          // WrapMode::Repeat
         GL_MIRRORED_REPEAT, // WrapMode::MirroredRepeat
+        // TODO(wwylele): ClampToEdge2 and ClampToBorder2 are not properly implemented here. See the
+        // comments in enum WrapMode.
+        GL_CLAMP_TO_EDGE,   // WrapMode::ClampToEdge2
+        GL_CLAMP_TO_BORDER, // WrapMode::ClampToBorder2
+        GL_REPEAT,          // WrapMode::Repeat2
+        GL_REPEAT,          // WrapMode::Repeat3
     };
 
     // Range check table for input
@@ -65,6 +71,13 @@ inline GLenum WrapMode(Pica::TexturingRegs::TextureConfig::WrapMode mode) {
         return GL_CLAMP_TO_EDGE;
     }
 
+    if (static_cast<u32>(mode) > 3) {
+        // It is still unclear whether mode 4-7 are valid, so log it if a game uses them.
+        // TODO(wwylele): telemetry should be added here so we can collect more info about which
+        // game uses this.
+        LOG_WARNING(Render_OpenGL, "Using texture wrap mode %u", static_cast<u32>(mode));
+    }
+
     GLenum gl_mode = wrap_mode_table[mode];
 
     // Check for dummy values indicating an unknown mode
diff --git a/src/video_core/shader/shader_jit_x64_compiler.cpp b/src/video_core/shader/shader_jit_x64_compiler.cpp
index 5d9b6448c..42a57aab1 100644
--- a/src/video_core/shader/shader_jit_x64_compiler.cpp
+++ b/src/video_core/shader/shader_jit_x64_compiler.cpp
@@ -321,27 +321,27 @@ void JitShader::Compile_EvaluateCondition(Instruction instr) {
     case Instruction::FlowControlType::Or:
         mov(eax, COND0);
         mov(ebx, COND1);
-        xor(eax, (instr.flow_control.refx.Value() ^ 1));
-        xor(ebx, (instr.flow_control.refy.Value() ^ 1));
-        or (eax, ebx);
+        xor_(eax, (instr.flow_control.refx.Value() ^ 1));
+        xor_(ebx, (instr.flow_control.refy.Value() ^ 1));
+        or_(eax, ebx);
         break;
 
     case Instruction::FlowControlType::And:
         mov(eax, COND0);
         mov(ebx, COND1);
-        xor(eax, (instr.flow_control.refx.Value() ^ 1));
-        xor(ebx, (instr.flow_control.refy.Value() ^ 1));
-        and(eax, ebx);
+        xor_(eax, (instr.flow_control.refx.Value() ^ 1));
+        xor_(ebx, (instr.flow_control.refy.Value() ^ 1));
+        and_(eax, ebx);
         break;
 
     case Instruction::FlowControlType::JustX:
         mov(eax, COND0);
-        xor(eax, (instr.flow_control.refx.Value() ^ 1));
+        xor_(eax, (instr.flow_control.refx.Value() ^ 1));
         break;
 
     case Instruction::FlowControlType::JustY:
         mov(eax, COND1);
-        xor(eax, (instr.flow_control.refy.Value() ^ 1));
+        xor_(eax, (instr.flow_control.refy.Value() ^ 1));
         break;
     }
 }
@@ -734,10 +734,10 @@ void JitShader::Compile_LOOP(Instruction instr) {
     mov(LOOPCOUNT, dword[SETUP + offset]);
     mov(LOOPCOUNT_REG, LOOPCOUNT);
     shr(LOOPCOUNT_REG, 4);
-    and(LOOPCOUNT_REG, 0xFF0); // Y-component is the start
+    and_(LOOPCOUNT_REG, 0xFF0); // Y-component is the start
     mov(LOOPINC, LOOPCOUNT);
     shr(LOOPINC, 12);
-    and(LOOPINC, 0xFF0);                // Z-component is the incrementer
+    and_(LOOPINC, 0xFF0);               // Z-component is the incrementer
     movzx(LOOPCOUNT, LOOPCOUNT.cvt8()); // X-component is iteration count
     add(LOOPCOUNT, 1);                  // Iteration count is X-component + 1
 
@@ -858,9 +858,9 @@ void JitShader::Compile(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_
     mov(STATE, ABI_PARAM2);
 
     // Zero address/loop  registers
-    xor(ADDROFFS_REG_0.cvt32(), ADDROFFS_REG_0.cvt32());
-    xor(ADDROFFS_REG_1.cvt32(), ADDROFFS_REG_1.cvt32());
-    xor(LOOPCOUNT_REG, LOOPCOUNT_REG);
+    xor_(ADDROFFS_REG_0.cvt32(), ADDROFFS_REG_0.cvt32());
+    xor_(ADDROFFS_REG_1.cvt32(), ADDROFFS_REG_1.cvt32());
+    xor_(LOOPCOUNT_REG, LOOPCOUNT_REG);
 
     // Used to set a register to one
     static const __m128 one = {1.f, 1.f, 1.f, 1.f};
diff --git a/src/video_core/swrasterizer/rasterizer.cpp b/src/video_core/swrasterizer/rasterizer.cpp
index 8b7b1defb..cd7b6c39d 100644
--- a/src/video_core/swrasterizer/rasterizer.cpp
+++ b/src/video_core/swrasterizer/rasterizer.cpp
@@ -357,10 +357,22 @@ static void ProcessTriangleInternal(const Vertex& v0, const Vertex& v1, const Ve
                 int t = (int)(v * float24::FromFloat32(static_cast<float>(texture.config.height)))
                             .ToFloat32();
 
-                if ((texture.config.wrap_s == TexturingRegs::TextureConfig::ClampToBorder &&
-                     (s < 0 || static_cast<u32>(s) >= texture.config.width)) ||
-                    (texture.config.wrap_t == TexturingRegs::TextureConfig::ClampToBorder &&
-                     (t < 0 || static_cast<u32>(t) >= texture.config.height))) {
+                bool use_border_s = false;
+                bool use_border_t = false;
+
+                if (texture.config.wrap_s == TexturingRegs::TextureConfig::ClampToBorder) {
+                    use_border_s = s < 0 || s >= static_cast<int>(texture.config.width);
+                } else if (texture.config.wrap_s == TexturingRegs::TextureConfig::ClampToBorder2) {
+                    use_border_s = s >= static_cast<int>(texture.config.width);
+                }
+
+                if (texture.config.wrap_t == TexturingRegs::TextureConfig::ClampToBorder) {
+                    use_border_t = t < 0 || t >= static_cast<int>(texture.config.height);
+                } else if (texture.config.wrap_t == TexturingRegs::TextureConfig::ClampToBorder2) {
+                    use_border_t = t >= static_cast<int>(texture.config.height);
+                }
+
+                if (use_border_s || use_border_t) {
                     auto border_color = texture.config.border_color;
                     texture_color[i] = {border_color.r, border_color.g, border_color.b,
                                         border_color.a};
diff --git a/src/video_core/swrasterizer/texturing.cpp b/src/video_core/swrasterizer/texturing.cpp
index aeb6aeb8c..4f02b93f2 100644
--- a/src/video_core/swrasterizer/texturing.cpp
+++ b/src/video_core/swrasterizer/texturing.cpp
@@ -18,22 +18,33 @@ using TevStageConfig = TexturingRegs::TevStageConfig;
 
 int GetWrappedTexCoord(TexturingRegs::TextureConfig::WrapMode mode, int val, unsigned size) {
     switch (mode) {
+    case TexturingRegs::TextureConfig::ClampToEdge2:
+        // For negative coordinate, ClampToEdge2 behaves the same as Repeat
+        if (val < 0) {
+            return static_cast<int>(static_cast<unsigned>(val) % size);
+        }
+    // [[fallthrough]]
     case TexturingRegs::TextureConfig::ClampToEdge:
         val = std::max(val, 0);
-        val = std::min(val, (int)size - 1);
+        val = std::min(val, static_cast<int>(size) - 1);
         return val;
 
     case TexturingRegs::TextureConfig::ClampToBorder:
         return val;
 
+    case TexturingRegs::TextureConfig::ClampToBorder2:
+    // For ClampToBorder2, the case of positive coordinate beyond the texture size is already
+    // handled outside. Here we only handle the negative coordinate in the same way as Repeat.
+    case TexturingRegs::TextureConfig::Repeat2:
+    case TexturingRegs::TextureConfig::Repeat3:
     case TexturingRegs::TextureConfig::Repeat:
-        return (int)((unsigned)val % size);
+        return static_cast<int>(static_cast<unsigned>(val) % size);
 
     case TexturingRegs::TextureConfig::MirroredRepeat: {
-        unsigned int coord = ((unsigned)val % (2 * size));
+        unsigned int coord = (static_cast<unsigned>(val) % (2 * size));
         if (coord >= size)
             coord = 2 * size - 1 - coord;
-        return (int)coord;
+        return static_cast<int>(coord);
     }
 
     default: