Merge branch 'master' of https://github.com/xenia-project/xenia into canary_new

.gitmodules

@@ -67,3 +67,9 @@
 [submodule "third_party/premake-cmake"]
 	path = third_party/premake-cmake
 	url = https://github.com/Enhex/premake-cmake.git
+[submodule "third_party/premake-androidmk"]
+	path = third_party/premake-androidmk
+	url = https://github.com/Triang3l/premake-androidmk.git
+[submodule "third_party/date"]
+	path = third_party/date
+	url = https://github.com/HowardHinnant/date.git

README.md

@@ -69,8 +69,8 @@ that there are some major work areas still untouched:
 <!--
 * Help work through [missing functionality/bugs in games](https://github.com/xenia-project/xenia/labels/compat)
 * Add input drivers for [DualShock4 (PS4) controllers](https://github.com/xenia-project/xenia/issues/60) (or anything else)
-* Skilled with Linux? A strong contributor is needed to [help with porting](https://github.com/xenia-project/xenia/labels/cross%20platform)
 <!--
+* Skilled with Linux? A strong contributor is needed to [help with porting](https://github.com/xenia-project/xenia/labels/platform-linux)
 See more projects [good for contributors](https://github.com/xenia-project/xenia/labels/good%20first%20issue). It's a good idea to ask on Discord and check the issues page before beginning work on
 something.
 -->

android/android_studio_project/.gitignore

@@ -0,0 +1,15 @@
+*.iml
+.gradle
+/local.properties
+/.idea/caches
+/.idea/libraries
+/.idea/modules.xml
+/.idea/workspace.xml
+/.idea/navEditor.xml
+/.idea/assetWizardSettings.xml
+.DS_Store
+/build
+/captures
+.externalNativeBuild
+.cxx
+local.properties

android/android_studio_project/app/.gitignore

@@ -0,0 +1 @@
+/build

android/android_studio_project/app/build.gradle

@@ -0,0 +1,79 @@
+plugins {
+    id 'com.android.application'
+}
+
+android {
+    compileSdkVersion 30
+    buildToolsVersion "30.0.2"
+    ndkVersion '22.0.6917172 rc1'
+
+    defaultConfig {
+        applicationId "jp.xenia.emulator"
+        // 24 (7.0) - Vulkan.
+        minSdkVersion 24
+        targetSdkVersion 30
+        versionCode 1
+        versionName "Prototype"
+        externalNativeBuild {
+            ndkBuild {
+                arguments "NDK_APPLICATION_MK:=../../../build/xenia_Application.mk"
+            }
+        }
+        ndk {
+            abiFilters 'arm64-v8a'
+        }
+    }
+
+    buildTypes {
+        release {
+            externalNativeBuild {
+                ndkBuild {
+                    arguments "PM5_CONFIG:=release_android"
+                }
+            }
+            minifyEnabled true
+            proguardFiles getDefaultProguardFile('proguard-android-optimize.txt'), 'proguard-rules.pro'
+        }
+        debug {
+            applicationIdSuffix ".debug"
+            debuggable true
+            externalNativeBuild {
+                ndkBuild {
+                    arguments "PM5_CONFIG:=debug_android"
+                }
+            }
+        }
+        checked {
+            applicationIdSuffix ".checked"
+            debuggable true
+            externalNativeBuild {
+                ndkBuild {
+                    arguments "PM5_CONFIG:=checked_android"
+                }
+            }
+        }
+    }
+
+    flavorDimensions "distribution"
+    productFlavors {
+        github {
+            dimension "distribution"
+            applicationIdSuffix ".github"
+        }
+        googlePlay {
+            dimension "distribution"
+            // TODO(Triang3l): Provide a signing config for core contributors only.
+        }
+    }
+
+    compileOptions {
+        sourceCompatibility JavaVersion.VERSION_1_8
+        targetCompatibility JavaVersion.VERSION_1_8
+    }
+
+    externalNativeBuild {
+        ndkBuild {
+            path file('../../../build/xenia_Android.mk')
+        }
+    }
+}

android/android_studio_project/app/proguard-rules.pro

@@ -0,0 +1,21 @@
+# Add project specific ProGuard rules here.
+# You can control the set of applied configuration files using the
+# proguardFiles setting in build.gradle.
+#
+# For more details, see
+#   http://developer.android.com/guide/developing/tools/proguard.html
+
+# If your project uses WebView with JS, uncomment the following
+# and specify the fully qualified class name to the JavaScript interface
+# class:
+#-keepclassmembers class fqcn.of.javascript.interface.for.webview {
+#   public *;
+#}
+
+# Uncomment this to preserve the line number information for
+# debugging stack traces.
+#-keepattributes SourceFile,LineNumberTable
+
+# If you keep the line number information, uncomment this to
+# hide the original source file name.
+#-renamesourcefileattribute SourceFile

android/android_studio_project/app/src/main/AndroidManifest.xml

@@ -0,0 +1,28 @@
+<?xml version="1.0" encoding="utf-8"?>
+<manifest xmlns:android="http://schemas.android.com/apk/res/android"
+    package="jp.xenia.emulator">
+
+    <uses-feature android:name="android.hardware.vulkan.level" android:version="0" android:required="true" />
+    <uses-feature android:name="android.hardware.vulkan.version" android:version="0x400000" android:required="true" />
+    <!-- Granted automatically - guest sockets -->
+    <uses-permission android:name="android.permission.INTERNET" />
+    <!-- Needs to be requested - loading games from outside the app data directory -->
+    <!-- WRITE_EXTERNAL_STORAGE is not required to write to the external app data directory since API 19 -->
+    <uses-permission android:name="android.permission.READ_EXTERNAL_STORAGE" />
+
+    <application
+        android:allowBackup="true"
+        android:icon="@mipmap/ic_launcher"
+        android:label="@string/app_name"
+        android:roundIcon="@mipmap/ic_launcher_round"
+        android:supportsRtl="true"
+        android:theme="@android:style/Theme.Material.Light">
+        <activity android:name="jp.xenia.emulator.DemoActivity">
+            <intent-filter>
+                <action android:name="android.intent.action.MAIN" />
+                <category android:name="android.intent.category.LAUNCHER" />
+            </intent-filter>
+        </activity>
+    </application>
+
+</manifest>

android/android_studio_project/app/src/main/java/jp/xenia/emulator/DemoActivity.java

@@ -0,0 +1,12 @@
+package jp.xenia.emulator;
+
+import android.app.Activity;
+import android.os.Bundle;
+
+public class DemoActivity extends Activity {
+    @Override
+    protected void onCreate(Bundle savedInstanceState) {
+        super.onCreate(savedInstanceState);
+        setContentView(R.layout.activity_demo);
+    }
+}

android/android_studio_project/app/src/main/res/drawable-v24/ic_launcher_foreground.xml

@@ -0,0 +1,30 @@
+<vector xmlns:android="http://schemas.android.com/apk/res/android"
+    xmlns:aapt="http://schemas.android.com/aapt"
+    android:width="108dp"
+    android:height="108dp"
+    android:viewportWidth="108"
+    android:viewportHeight="108">
+    <path android:pathData="M31,63.928c0,0 6.4,-11 12.1,-13.1c7.2,-2.6 26,-1.4 26,-1.4l38.1,38.1L107,108.928l-32,-1L31,63.928z">
+        <aapt:attr name="android:fillColor">
+            <gradient
+                android:endX="85.84757"
+                android:endY="92.4963"
+                android:startX="42.9492"
+                android:startY="49.59793"
+                android:type="linear">
+                <item
+                    android:color="#44000000"
+                    android:offset="0.0" />
+                <item
+                    android:color="#00000000"
+                    android:offset="1.0" />
+            </gradient>
+        </aapt:attr>
+    </path>
+    <path
+        android:fillColor="#FFFFFF"
+        android:fillType="nonZero"
+        android:pathData="M65.3,45.828l3.8,-6.6c0.2,-0.4 0.1,-0.9 -0.3,-1.1c-0.4,-0.2 -0.9,-0.1 -1.1,0.3l-3.9,6.7c-6.3,-2.8 -13.4,-2.8 -19.7,0l-3.9,-6.7c-0.2,-0.4 -0.7,-0.5 -1.1,-0.3C38.8,38.328 38.7,38.828 38.9,39.228l3.8,6.6C36.2,49.428 31.7,56.028 31,63.928h46C76.3,56.028 71.8,49.428 65.3,45.828zM43.4,57.328c-0.8,0 -1.5,-0.5 -1.8,-1.2c-0.3,-0.7 -0.1,-1.5 0.4,-2.1c0.5,-0.5 1.4,-0.7 2.1,-0.4c0.7,0.3 1.2,1 1.2,1.8C45.3,56.528 44.5,57.328 43.4,57.328L43.4,57.328zM64.6,57.328c-0.8,0 -1.5,-0.5 -1.8,-1.2s-0.1,-1.5 0.4,-2.1c0.5,-0.5 1.4,-0.7 2.1,-0.4c0.7,0.3 1.2,1 1.2,1.8C66.5,56.528 65.6,57.328 64.6,57.328L64.6,57.328z"
+        android:strokeWidth="1"
+        android:strokeColor="#00000000" />
+</vector>

android/android_studio_project/app/src/main/res/drawable/ic_launcher_background.xml

@@ -0,0 +1,170 @@
+<?xml version="1.0" encoding="utf-8"?>
+<vector xmlns:android="http://schemas.android.com/apk/res/android"
+    android:width="108dp"
+    android:height="108dp"
+    android:viewportWidth="108"
+    android:viewportHeight="108">
+    <path
+        android:fillColor="#3DDC84"
+        android:pathData="M0,0h108v108h-108z" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M9,0L9,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,0L19,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M29,0L29,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M39,0L39,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M49,0L49,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M59,0L59,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M69,0L69,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M79,0L79,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M89,0L89,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M99,0L99,108"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,9L108,9"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,19L108,19"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,29L108,29"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,39L108,39"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,49L108,49"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,59L108,59"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,69L108,69"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,79L108,79"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,89L108,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M0,99L108,99"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,29L89,29"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,39L89,39"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,49L89,49"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,59L89,59"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,69L89,69"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M19,79L89,79"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M29,19L29,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M39,19L39,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M49,19L49,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M59,19L59,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M69,19L69,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+    <path
+        android:fillColor="#00000000"
+        android:pathData="M79,19L79,89"
+        android:strokeWidth="0.8"
+        android:strokeColor="#33FFFFFF" />
+</vector>

android/android_studio_project/app/src/main/res/layout/activity_demo.xml

@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="utf-8"?>
+<RelativeLayout xmlns:android="http://schemas.android.com/apk/res/android"
+    xmlns:tools="http://schemas.android.com/tools"
+    android:layout_width="match_parent"
+    android:layout_height="match_parent"
+    tools:context="jp.xenia.emulator.DemoActivity">
+
+</RelativeLayout>

android/android_studio_project/app/src/main/res/mipmap-anydpi-v26/ic_launcher.xml

@@ -0,0 +1,5 @@
+<?xml version="1.0" encoding="utf-8"?>
+<adaptive-icon xmlns:android="http://schemas.android.com/apk/res/android">
+    <background android:drawable="@drawable/ic_launcher_background" />
+    <foreground android:drawable="@drawable/ic_launcher_foreground" />
+</adaptive-icon>

android/android_studio_project/app/src/main/res/mipmap-anydpi-v26/ic_launcher_round.xml

@@ -0,0 +1,5 @@
+<?xml version="1.0" encoding="utf-8"?>
+<adaptive-icon xmlns:android="http://schemas.android.com/apk/res/android">
+    <background android:drawable="@drawable/ic_launcher_background" />
+    <foreground android:drawable="@drawable/ic_launcher_foreground" />
+</adaptive-icon>

android/android_studio_project/app/src/main/res/values/strings.xml

@@ -0,0 +1,3 @@
+<resources>
+    <string name="app_name">Xenia</string>
+</resources>

android/android_studio_project/build.gradle

@@ -0,0 +1,24 @@
+// Top-level build file where you can add configuration options common to all sub-projects/modules.
+buildscript {
+    repositories {
+        google()
+        jcenter()
+    }
+    dependencies {
+        classpath "com.android.tools.build:gradle:4.1.1"
+
+        // NOTE: Do not place your application dependencies here; they belong
+        // in the individual module build.gradle files
+    }
+}
+
+allprojects {
+    repositories {
+        google()
+        jcenter()
+    }
+}
+
+task clean(type: Delete) {
+    delete rootProject.buildDir
+}

android/android_studio_project/gradle.properties

@@ -0,0 +1,19 @@
+# Project-wide Gradle settings.
+# IDE (e.g. Android Studio) users:
+# Gradle settings configured through the IDE *will override*
+# any settings specified in this file.
+# For more details on how to configure your build environment visit
+# http://www.gradle.org/docs/current/userguide/build_environment.html
+# Specifies the JVM arguments used for the daemon process.
+# The setting is particularly useful for tweaking memory settings.
+org.gradle.jvmargs=-Xmx2048m -Dfile.encoding=UTF-8
+# When configured, Gradle will run in incubating parallel mode.
+# This option should only be used with decoupled projects. More details, visit
+# http://www.gradle.org/docs/current/userguide/multi_project_builds.html#sec:decoupled_projects
+# org.gradle.parallel=true
+# AndroidX package structure to make it clearer which packages are bundled with the
+# Android operating system, and which are packaged with your app"s APK
+# https://developer.android.com/topic/libraries/support-library/androidx-rn
+android.useAndroidX=true
+# Automatically convert third-party libraries to use AndroidX
+android.enableJetifier=true

android/android_studio_project/gradle/wrapper/gradle-wrapper.properties

@@ -0,0 +1,6 @@
+#Sat Nov 21 20:44:11 MSK 2020
+distributionBase=GRADLE_USER_HOME
+distributionPath=wrapper/dists
+zipStoreBase=GRADLE_USER_HOME
+zipStorePath=wrapper/dists
+distributionUrl=https\://services.gradle.org/distributions/gradle-6.5-bin.zip

android/android_studio_project/gradlew

@@ -0,0 +1,172 @@
+#!/usr/bin/env sh
+
+##############################################################################
+##
+##  Gradle start up script for UN*X
+##
+##############################################################################
+
+# Attempt to set APP_HOME
+# Resolve links: $0 may be a link
+PRG="$0"
+# Need this for relative symlinks.
+while [ -h "$PRG" ] ; do
+    ls=`ls -ld "$PRG"`
+    link=`expr "$ls" : '.*-> \(.*\)$'`
+    if expr "$link" : '/.*' > /dev/null; then
+        PRG="$link"
+    else
+        PRG=`dirname "$PRG"`"/$link"
+    fi
+done
+SAVED="`pwd`"
+cd "`dirname \"$PRG\"`/" >/dev/null
+APP_HOME="`pwd -P`"
+cd "$SAVED" >/dev/null
+
+APP_NAME="Gradle"
+APP_BASE_NAME=`basename "$0"`
+
+# Add default JVM options here. You can also use JAVA_OPTS and GRADLE_OPTS to pass JVM options to this script.
+DEFAULT_JVM_OPTS=""
+
+# Use the maximum available, or set MAX_FD != -1 to use that value.
+MAX_FD="maximum"
+
+warn () {
+    echo "$*"
+}
+
+die () {
+    echo
+    echo "$*"
+    echo
+    exit 1
+}
+
+# OS specific support (must be 'true' or 'false').
+cygwin=false
+msys=false
+darwin=false
+nonstop=false
+case "`uname`" in
+  CYGWIN* )
+    cygwin=true
+    ;;
+  Darwin* )
+    darwin=true
+    ;;
+  MINGW* )
+    msys=true
+    ;;
+  NONSTOP* )
+    nonstop=true
+    ;;
+esac
+
+CLASSPATH=$APP_HOME/gradle/wrapper/gradle-wrapper.jar
+
+# Determine the Java command to use to start the JVM.
+if [ -n "$JAVA_HOME" ] ; then
+    if [ -x "$JAVA_HOME/jre/sh/java" ] ; then
+        # IBM's JDK on AIX uses strange locations for the executables
+        JAVACMD="$JAVA_HOME/jre/sh/java"
+    else
+        JAVACMD="$JAVA_HOME/bin/java"
+    fi
+    if [ ! -x "$JAVACMD" ] ; then
+        die "ERROR: JAVA_HOME is set to an invalid directory: $JAVA_HOME
+
+Please set the JAVA_HOME variable in your environment to match the
+location of your Java installation."
+    fi
+else
+    JAVACMD="java"
+    which java >/dev/null 2>&1 || die "ERROR: JAVA_HOME is not set and no 'java' command could be found in your PATH.
+
+Please set the JAVA_HOME variable in your environment to match the
+location of your Java installation."
+fi
+
+# Increase the maximum file descriptors if we can.
+if [ "$cygwin" = "false" -a "$darwin" = "false" -a "$nonstop" = "false" ] ; then
+    MAX_FD_LIMIT=`ulimit -H -n`
+    if [ $? -eq 0 ] ; then
+        if [ "$MAX_FD" = "maximum" -o "$MAX_FD" = "max" ] ; then
+            MAX_FD="$MAX_FD_LIMIT"
+        fi
+        ulimit -n $MAX_FD
+        if [ $? -ne 0 ] ; then
+            warn "Could not set maximum file descriptor limit: $MAX_FD"
+        fi
+    else
+        warn "Could not query maximum file descriptor limit: $MAX_FD_LIMIT"
+    fi
+fi
+
+# For Darwin, add options to specify how the application appears in the dock
+if $darwin; then
+    GRADLE_OPTS="$GRADLE_OPTS \"-Xdock:name=$APP_NAME\" \"-Xdock:icon=$APP_HOME/media/gradle.icns\""
+fi
+
+# For Cygwin, switch paths to Windows format before running java
+if $cygwin ; then
+    APP_HOME=`cygpath --path --mixed "$APP_HOME"`
+    CLASSPATH=`cygpath --path --mixed "$CLASSPATH"`
+    JAVACMD=`cygpath --unix "$JAVACMD"`
+
+    # We build the pattern for arguments to be converted via cygpath
+    ROOTDIRSRAW=`find -L / -maxdepth 1 -mindepth 1 -type d 2>/dev/null`
+    SEP=""
+    for dir in $ROOTDIRSRAW ; do
+        ROOTDIRS="$ROOTDIRS$SEP$dir"
+        SEP="|"
+    done
+    OURCYGPATTERN="(^($ROOTDIRS))"
+    # Add a user-defined pattern to the cygpath arguments
+    if [ "$GRADLE_CYGPATTERN" != "" ] ; then
+        OURCYGPATTERN="$OURCYGPATTERN|($GRADLE_CYGPATTERN)"
+    fi
+    # Now convert the arguments - kludge to limit ourselves to /bin/sh
+    i=0
+    for arg in "$@" ; do
+        CHECK=`echo "$arg"|egrep -c "$OURCYGPATTERN" -`
+        CHECK2=`echo "$arg"|egrep -c "^-"`                                 ### Determine if an option
+
+        if [ $CHECK -ne 0 ] && [ $CHECK2 -eq 0 ] ; then                    ### Added a condition
+            eval `echo args$i`=`cygpath --path --ignore --mixed "$arg"`
+        else
+            eval `echo args$i`="\"$arg\""
+        fi
+        i=$((i+1))
+    done
+    case $i in
+        (0) set -- ;;
+        (1) set -- "$args0" ;;
+        (2) set -- "$args0" "$args1" ;;
+        (3) set -- "$args0" "$args1" "$args2" ;;
+        (4) set -- "$args0" "$args1" "$args2" "$args3" ;;
+        (5) set -- "$args0" "$args1" "$args2" "$args3" "$args4" ;;
+        (6) set -- "$args0" "$args1" "$args2" "$args3" "$args4" "$args5" ;;
+        (7) set -- "$args0" "$args1" "$args2" "$args3" "$args4" "$args5" "$args6" ;;
+        (8) set -- "$args0" "$args1" "$args2" "$args3" "$args4" "$args5" "$args6" "$args7" ;;
+        (9) set -- "$args0" "$args1" "$args2" "$args3" "$args4" "$args5" "$args6" "$args7" "$args8" ;;
+    esac
+fi
+
+# Escape application args
+save () {
+    for i do printf %s\\n "$i" | sed "s/'/'\\\\''/g;1s/^/'/;\$s/\$/' \\\\/" ; done
+    echo " "
+}
+APP_ARGS=$(save "$@")
+
+# Collect all arguments for the java command, following the shell quoting and substitution rules
+eval set -- $DEFAULT_JVM_OPTS $JAVA_OPTS $GRADLE_OPTS "\"-Dorg.gradle.appname=$APP_BASE_NAME\"" -classpath "\"$CLASSPATH\"" org.gradle.wrapper.GradleWrapperMain "$APP_ARGS"
+
+# by default we should be in the correct project dir, but when run from Finder on Mac, the cwd is wrong
+if [ "$(uname)" = "Darwin" ] && [ "$HOME" = "$PWD" ]; then
+  cd "$(dirname "$0")"
+fi
+
+exec "$JAVACMD" "$@"

android/android_studio_project/gradlew.bat

@@ -0,0 +1,84 @@
+@if "%DEBUG%" == "" @echo off
+@rem ##########################################################################
+@rem
+@rem  Gradle startup script for Windows
+@rem
+@rem ##########################################################################
+
+@rem Set local scope for the variables with windows NT shell
+if "%OS%"=="Windows_NT" setlocal
+
+set DIRNAME=%~dp0
+if "%DIRNAME%" == "" set DIRNAME=.
+set APP_BASE_NAME=%~n0
+set APP_HOME=%DIRNAME%
+
+@rem Add default JVM options here. You can also use JAVA_OPTS and GRADLE_OPTS to pass JVM options to this script.
+set DEFAULT_JVM_OPTS=
+
+@rem Find java.exe
+if defined JAVA_HOME goto findJavaFromJavaHome
+
+set JAVA_EXE=java.exe
+%JAVA_EXE% -version >NUL 2>&1
+if "%ERRORLEVEL%" == "0" goto init
+
+echo.
+echo ERROR: JAVA_HOME is not set and no 'java' command could be found in your PATH.
+echo.
+echo Please set the JAVA_HOME variable in your environment to match the
+echo location of your Java installation.
+
+goto fail
+
+:findJavaFromJavaHome
+set JAVA_HOME=%JAVA_HOME:"=%
+set JAVA_EXE=%JAVA_HOME%/bin/java.exe
+
+if exist "%JAVA_EXE%" goto init
+
+echo.
+echo ERROR: JAVA_HOME is set to an invalid directory: %JAVA_HOME%
+echo.
+echo Please set the JAVA_HOME variable in your environment to match the
+echo location of your Java installation.
+
+goto fail
+
+:init
+@rem Get command-line arguments, handling Windows variants
+
+if not "%OS%" == "Windows_NT" goto win9xME_args
+
+:win9xME_args
+@rem Slurp the command line arguments.
+set CMD_LINE_ARGS=
+set _SKIP=2
+
+:win9xME_args_slurp
+if "x%~1" == "x" goto execute
+
+set CMD_LINE_ARGS=%*
+
+:execute
+@rem Setup the command line
+
+set CLASSPATH=%APP_HOME%\gradle\wrapper\gradle-wrapper.jar
+
+@rem Execute Gradle
+"%JAVA_EXE%" %DEFAULT_JVM_OPTS% %JAVA_OPTS% %GRADLE_OPTS% "-Dorg.gradle.appname=%APP_BASE_NAME%" -classpath "%CLASSPATH%" org.gradle.wrapper.GradleWrapperMain %CMD_LINE_ARGS%
+
+:end
+@rem End local scope for the variables with windows NT shell
+if "%ERRORLEVEL%"=="0" goto mainEnd
+
+:fail
+rem Set variable GRADLE_EXIT_CONSOLE if you need the _script_ return code instead of
+rem the _cmd.exe /c_ return code!
+if  not "" == "%GRADLE_EXIT_CONSOLE%" exit 1
+exit /b 1
+
+:mainEnd
+if "%OS%"=="Windows_NT" endlocal
+
+:omega

android/android_studio_project/settings.gradle

@@ -0,0 +1,2 @@
+include ':app'
+rootProject.name = "Xenia"

docs/gpu.md

@@ -119,6 +119,87 @@ Registers documented at [src/xenia/gpu/register_table.inc](../src/xenia/gpu/regi
 
 PM4 commands documented at [src/xenia/gpu/xenos.h](../src/xenia/gpu/xenos.h#L521).
 
+#### Performance Counters that may be read back by D3D
+
+They are 64-bit values and have a high and low 32-bit register as well as a `SELECT` register each:
+
+- CP_PERFCOUNTER0
+
+- RBBM_PERFCOUNTER0
+- RBBM_PERFCOUNTER1
+
+- SQ_PERFCOUNTER0
+- SQ_PERFCOUNTER1
+- SQ_PERFCOUNTER2
+- SQ_PERFCOUNTER3
+
+- VGT_PERFCOUNTER0
+- VGT_PERFCOUNTER1
+- VGT_PERFCOUNTER2
+- VGT_PERFCOUNTER3
+
+- VC_PERFCOUNTER0
+- VC_PERFCOUNTER1
+- VC_PERFCOUNTER2
+- VC_PERFCOUNTER3
+
+- PA_SU_PERFCOUNTER0
+- PA_SU_PERFCOUNTER1
+- PA_SU_PERFCOUNTER2
+- PA_SU_PERFCOUNTER3
+
+- PA_SC_PERFCOUNTER0
+- PA_SC_PERFCOUNTER1
+- PA_SC_PERFCOUNTER2
+- PA_SC_PERFCOUNTER3
+
+- HZ_PERFCOUNTER0
+- HZ_PERFCOUNTER1
+
+- TCR_PERFCOUNTER0
+- TCR_PERFCOUNTER1
+
+- TCM_PERFCOUNTER0
+- TCM_PERFCOUNTER1
+
+- TCF_PERFCOUNTER0
+- TCF_PERFCOUNTER1
+- TCF_PERFCOUNTER2
+- TCF_PERFCOUNTER3
+- TCF_PERFCOUNTER4
+- TCF_PERFCOUNTER5
+- TCF_PERFCOUNTER6
+- TCF_PERFCOUNTER7
+- TCF_PERFCOUNTER8
+- TCF_PERFCOUNTER9
+- TCF_PERFCOUNTER10
+- TCF_PERFCOUNTER11
+
+- TP0_PERFCOUNTER0
+- TP0_PERFCOUNTER1
+- TP1_PERFCOUNTER0
+- TP1_PERFCOUNTER1
+- TP2_PERFCOUNTER0
+- TP2_PERFCOUNTER1
+- TP3_PERFCOUNTER0
+- TP3_PERFCOUNTER1
+
+- SX_PERFCOUNTER0
+
+- BC_PERFCOUNTER0
+- BC_PERFCOUNTER1
+- BC_PERFCOUNTER2
+- BC_PERFCOUNTER3
+
+- MC0_PERFCOUNTER0
+- MC1_PERFCOUNTER0
+
+- MH_PERFCOUNTER0
+- MH_PERFCOUNTER1
+- MH_PERFCOUNTER2
+
+- BIF_PERFCOUNTER0
+
 ### Shaders
 
 * [LLVM R600 Tables](https://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AMDGPU/R600Instructions.td)

premake5.lua

@@ -1,6 +1,9 @@
 include("tools/build")
 require("third_party/premake-export-compile-commands/export-compile-commands")
 require("third_party/premake-cmake/cmake")
+-- gmake required for androidmk.
+require("gmake")
+require("third_party/premake-androidmk/androidmk")
 
 location(build_root)
 targetdir(build_bin)
@@ -26,6 +29,8 @@ defines({
 })
 
 cppdialect("C++17")
+exceptionhandling("On")
+rtti("On")
 symbols("On")
 
 -- TODO(DrChat): Find a way to disable this on other architectures.
@@ -81,10 +86,15 @@ filter("configurations:Release")
   })
   optimize("Speed")
   inlining("Auto")
-  floatingpoint("Fast")
   flags({
     "LinkTimeOptimization",
   })
+  -- Not using floatingpoint("Fast") - NaN checks are used in some places
+  -- (though rarely), overall preferable to avoid any functional differences
+  -- between debug and release builds, and to have calculations involved in GPU
+  -- (especially anything that may affect vertex position invariance) and CPU
+  -- (such as constant propagation) emulation as predictable as possible,
+  -- including handling of specials since games make assumptions about them.
 filter("platforms:Linux")
   system("linux")
   toolset("clang")
@@ -138,6 +148,13 @@ filter({"platforms:Linux", "language:C++", "toolset:clang", "files:*.cc or *.cpp
     "-stdlib=libstdc++",
   })
 
+filter("platforms:Android")
+  system("android")
+  links({
+    "android",
+    "dl",
+  })
+
 filter("platforms:Windows")
   system("windows")
   toolset("msc")
@@ -189,18 +206,24 @@ end
 solution("xenia")
   uuid("931ef4b0-6170-4f7a-aaf2-0fece7632747")
   startproject("xenia-app")
-  architecture("x86_64")
+  if os.istarget("android") then
-  if os.istarget("linux") then
+    -- Not setting architecture as that's handled by ndk-build itself.
-    platforms({"Linux"})
+    platforms({"Android"})
-  elseif os.istarget("windows") then
+    ndkstl("c++_static")
-    platforms({"Windows"})
+  else
-    -- 10.0.15063.0: ID3D12GraphicsCommandList1::SetSamplePositions.
+    architecture("x86_64")
-    -- 10.0.19041.0: D3D12_HEAP_FLAG_CREATE_NOT_ZEROED.
+    if os.istarget("linux") then
-    filter("action:vs2017")
+      platforms({"Linux"})
-      systemversion("10.0.19041.0")
+    elseif os.istarget("windows") then
-    filter("action:vs2019")
+      platforms({"Windows"})
-      systemversion("10.0")
+      -- 10.0.15063.0: ID3D12GraphicsCommandList1::SetSamplePositions.
-    filter({})
+      -- 10.0.19041.0: D3D12_HEAP_FLAG_CREATE_NOT_ZEROED.
+      filter("action:vs2017")
+        systemversion("10.0.19041.0")
+      filter("action:vs2019")
+        systemversion("10.0")
+      filter({})
+    end
   end
   configurations({"Checked", "Debug", "Release"})
 
@@ -215,18 +238,15 @@ solution("xenia")
   include("third_party/imgui.lua")
   include("third_party/libav.lua")
   include("third_party/mspack.lua")
-  include("third_party/SDL2.lua")
   include("third_party/snappy.lua")
   include("third_party/spirv-tools.lua")
   include("third_party/volk.lua")
   include("third_party/xxhash.lua")
 
   include("src/xenia")
-  include("src/xenia/app")
   include("src/xenia/app/discord")
   include("src/xenia/apu")
   include("src/xenia/apu/nop")
-  include("src/xenia/apu/sdl")
   include("src/xenia/base")
   include("src/xenia/cpu")
   include("src/xenia/cpu/backend/x64")
@@ -234,10 +254,8 @@ solution("xenia")
   include("src/xenia/gpu")
   include("src/xenia/gpu/null")
   include("src/xenia/gpu/vulkan")
-  include("src/xenia/helper/sdl")
   include("src/xenia/hid")
   include("src/xenia/hid/nop")
-  include("src/xenia/hid/sdl")
   include("src/xenia/kernel")
   include("src/xenia/patcher")
   include("src/xenia/ui")
@@ -245,6 +263,24 @@ solution("xenia")
   include("src/xenia/ui/vulkan")
   include("src/xenia/vfs")
 
+  if not os.istarget("android") then
+    -- SDL2 requires sdl2-config, and as of November 2020 isn't high-quality on
+    -- Android yet, most importantly in game controllers - the keycode and axis
+    -- enums are being ruined during conversion to SDL2 enums resulting in only
+    -- one controller (Nvidia Shield) being supported, digital triggers are also
+    -- not supported; lifecycle management (especially surface loss) is also
+    -- complicated.
+    include("third_party/SDL2.lua")
+
+    include("src/xenia/apu/sdl")
+    include("src/xenia/helper/sdl")
+    include("src/xenia/hid/sdl")
+
+    -- TODO(Triang3l): src/xenia/app has a dependency on xenia-helper-sdl, bring
+    -- it back later.
+    include("src/xenia/app")
+  end
+
   if os.istarget("windows") then
     include("src/xenia/apu/xaudio2")
     include("src/xenia/gpu/d3d12")

src/xenia/app/discord/discord_presence.cc

@@ -43,9 +43,10 @@ void DiscordPresence::NotPlaying() {
 }
 
 void DiscordPresence::PlayingTitle(const std::string_view game_title) {
+  auto details = std::string(game_title);
   DiscordRichPresence discordPresence = {};
   discordPresence.state = "In Game";
-  discordPresence.details = std::string(game_title).c_str();
+  discordPresence.details = details.c_str();
   // TODO(gibbed): we don't have state icons yet.
   // discordPresence.smallImageKey = "app";
   // discordPresence.largeImageKey = "state_ingame";

src/xenia/app/xenia_main.cc

@@ -216,7 +216,7 @@ int xenia_main(const std::vector<std::string>& args) {
     if (!cvars::portable &&
         !std::filesystem::exists(storage_root / "portable.txt")) {
       storage_root = xe::filesystem::GetUserFolder();
-#if defined(XE_PLATFORM_WIN32) || defined(XE_PLATFORM_LINUX)
+#if defined(XE_PLATFORM_WIN32) || defined(XE_PLATFORM_GNU_LINUX)
       storage_root = storage_root / "Xenia";
 #else
 #warning Unhandled platform for the data root.

src/xenia/base/byte_order.h

@@ -15,6 +15,10 @@
 #include "xenia/base/assert.h"
 #include "xenia/base/platform.h"
 
+#if XE_PLATFORM_LINUX
+#include <byteswap.h>
+#endif
+
 namespace xe {
 
 #if XE_PLATFORM_WIN32
@@ -26,9 +30,9 @@ namespace xe {
 #define XENIA_BASE_BYTE_SWAP_32 OSSwapInt32
 #define XENIA_BASE_BYTE_SWAP_64 OSSwapInt64
 #else
-#define XENIA_BASE_BYTE_SWAP_16 __bswap_16
+#define XENIA_BASE_BYTE_SWAP_16 bswap_16
-#define XENIA_BASE_BYTE_SWAP_32 __bswap_32
+#define XENIA_BASE_BYTE_SWAP_32 bswap_32
-#define XENIA_BASE_BYTE_SWAP_64 __bswap_64
+#define XENIA_BASE_BYTE_SWAP_64 bswap_64
 #endif  // XE_PLATFORM_WIN32
 
 inline int8_t byte_swap(int8_t value) { return value; }

src/xenia/base/logging.cc

@@ -36,10 +36,8 @@
 
 #include "third_party/fmt/include/fmt/format.h"
 
-DEFINE_path(
+DEFINE_path(log_file, "", "Logs are written to the given file", "Logging");
-    log_file, "",
+DEFINE_bool(log_to_stdout, true, "Write log output to stdout", "Logging");
-    "Logs are written to the given file (specify stdout for command line)",
-    "Logging");
 DEFINE_bool(log_to_debugprint, false, "Dump the log to DebugPrint.", "Logging");
 DEFINE_bool(flush_log, true, "Flush log file after each log line batch.",
             "Logging");
@@ -66,31 +64,27 @@ struct LogLine {
 
 thread_local char thread_log_buffer_[64 * 1024];
 
+void FileLogSink::Write(const char* buf, size_t size) {
+  if (file_) {
+    fwrite(buf, 1, size, file_);
+  }
+}
+
+void FileLogSink::Flush() {
+  if (file_) {
+    fflush(file_);
+  }
+}
+
 class Logger {
  public:
   explicit Logger(const std::string_view app_name)
-      : file_(nullptr),
+      : wait_strategy_(),
-        running_(true),
-        wait_strategy_(),
         claim_strategy_(kBlockCount, wait_strategy_),
-        consumed_(wait_strategy_) {
+        consumed_(wait_strategy_),
+        running_(true) {
     claim_strategy_.add_claim_barrier(consumed_);
 
-    if (cvars::log_file.empty()) {
-      // Default to app name.
-      auto file_name = fmt::format("{}.log", app_name);
-      auto file_path = std::filesystem::path(file_name);
-      xe::filesystem::CreateParentFolder(file_path);
-      file_ = xe::filesystem::OpenFile(file_path, "wt");
-    } else {
-      if (cvars::log_file == "stdout") {
-        file_ = stdout;
-      } else {
-        xe::filesystem::CreateParentFolder(cvars::log_file);
-        file_ = xe::filesystem::OpenFile(cvars::log_file, "wt");
-      }
-    }
-
     write_thread_ =
         xe::threading::Thread::Create({}, [this]() { WriteThread(); });
     write_thread_->set_name("Logging Writer");
@@ -99,8 +93,10 @@ class Logger {
   ~Logger() {
     running_ = false;
     xe::threading::Wait(write_thread_.get(), true);
-    fflush(file_);
+  }
-    fclose(file_);
+
+  void AddLogSink(std::unique_ptr<LogSink>&& sink) {
+    sinks_.push_back(std::move(sink));
   }
 
  private:
@@ -126,14 +122,14 @@ class Logger {
   dp::multi_threaded_claim_strategy<dp::spin_wait_strategy> claim_strategy_;
   dp::sequence_barrier<dp::spin_wait_strategy> consumed_;
 
-  FILE* file_;
+  std::vector<std::unique_ptr<LogSink>> sinks_;
 
   std::atomic<bool> running_;
   std::unique_ptr<xe::threading::Thread> write_thread_;
 
   void Write(const char* buf, size_t size) {
-    if (file_) {
+    for (const auto& sink : sinks_) {
-      fwrite(buf, 1, size, file_);
+      sink->Write(buf, size);
     }
     if (cvars::log_to_debugprint) {
       debugging::DebugPrint("{}", std::string_view(buf, size));
@@ -246,7 +242,9 @@ class Logger {
         desired_count = 1;
 
         if (cvars::flush_log) {
-          fflush(file_);
+          for (const auto& sink : sinks_) {
+            sink->Flush();
+          }
         }
 
         idle_loops = 0;
@@ -291,6 +289,27 @@ class Logger {
 void InitializeLogging(const std::string_view app_name) {
   auto mem = memory::AlignedAlloc<Logger>(0x10);
   logger_ = new (mem) Logger(app_name);
+
+  FILE* log_file = nullptr;
+
+  if (cvars::log_file.empty()) {
+    // Default to app name.
+    auto file_name = fmt::format("{}.log", app_name);
+    auto file_path = std::filesystem::path(file_name);
+    xe::filesystem::CreateParentFolder(file_path);
+
+    log_file = xe::filesystem::OpenFile(file_path, "wt");
+  } else {
+    xe::filesystem::CreateParentFolder(cvars::log_file);
+    log_file = xe::filesystem::OpenFile(cvars::log_file, "wt");
+  }
+  auto sink = std::make_unique<FileLogSink>(log_file);
+  logger_->AddLogSink(std::move(sink));
+
+  if (cvars::log_to_stdout) {
+    auto stdout_sink = std::make_unique<FileLogSink>(stdout);
+    logger_->AddLogSink(std::move(stdout_sink));
+  }
 }
 
 void ShutdownLogging() {

src/xenia/base/logging.h

@@ -34,6 +34,31 @@ enum class LogLevel {
   Trace,
 };
 
+class LogSink {
+ public:
+  virtual ~LogSink() = default;
+
+  virtual void Write(const char* buf, size_t size) = 0;
+  virtual void Flush() = 0;
+};
+
+class FileLogSink final : public LogSink {
+ public:
+  explicit FileLogSink(FILE* file) : file_(file) {}
+  virtual ~FileLogSink() {
+    if (file_) {
+      fflush(file_);
+      fclose(file_);
+    }
+  }
+
+  void Write(const char* buf, size_t size) override;
+  void Flush() override;
+
+ private:
+  FILE* file_;
+};
+
 // Initializes the logging system and any outputs requested.
 // Must be called on startup.
 void InitializeLogging(const std::string_view app_name);

src/xenia/base/main_win.cc

@@ -29,6 +29,8 @@
 
 DEFINE_bool(win32_high_freq, true,
             "Requests high performance from the NT kernel", "Kernel");
+DEFINE_bool(enable_console, false, "Open a console window with the main window",
+            "General");
 
 namespace xe {
 
@@ -37,27 +39,23 @@ bool has_console_attached_ = true;
 bool has_console_attached() { return has_console_attached_; }
 
 void AttachConsole() {
-  bool has_console = ::AttachConsole(ATTACH_PARENT_PROCESS) == TRUE;
+  if (!cvars::enable_console) {
-  if (!has_console) {
-    // We weren't launched from a console, so just return.
-    // We could alloc our own console, but meh:
-    // has_console = AllocConsole() == TRUE;
-    has_console_attached_ = false;
     return;
   }
+
+  AllocConsole();
+
   has_console_attached_ = true;
 
   auto std_handle = (intptr_t)GetStdHandle(STD_OUTPUT_HANDLE);
   auto con_handle = _open_osfhandle(std_handle, _O_TEXT);
   auto fp = _fdopen(con_handle, "w");
-  *stdout = *fp;
+  freopen_s(&fp, "CONOUT$", "w", stdout);
-  setvbuf(stdout, nullptr, _IONBF, 0);
 
   std_handle = (intptr_t)GetStdHandle(STD_ERROR_HANDLE);
   con_handle = _open_osfhandle(std_handle, _O_TEXT);
   fp = _fdopen(con_handle, "w");
-  *stderr = *fp;
+  freopen_s(&fp, "CONOUT$", "w", stderr);
-  setvbuf(stderr, nullptr, _IONBF, 0);
 }
 
 static void RequestHighPerformance() {
@@ -125,6 +123,10 @@ int Main() {
     return 1;
   }
 
+  // Attach a console so we can write output to stdout. If the user hasn't
+  // redirected output themselves it'll pop up a window.
+  xe::AttachConsole();
+
   // Setup COM on the main thread.
   // NOTE: this may fail if COM has already been initialized - that's OK.
 #pragma warning(suppress : 6031)
@@ -163,10 +165,6 @@ int main(int argc_ignored, char** argv_ignored) { return xe::Main(); }
 
 // Used in windowed apps; automatically picked based on subsystem.
 int WINAPI wWinMain(HINSTANCE, HINSTANCE, LPWSTR command_line, int) {
-  // Attach a console so we can write output to stdout. If the user hasn't
-  // redirected output themselves it'll pop up a window.
-  xe::AttachConsole();
-
   // Run normal entry point.
   return xe::Main();
 }

src/xenia/base/mapped_memory_win.cc

@@ -22,6 +22,11 @@ namespace xe {
 
 class Win32MappedMemory : public MappedMemory {
  public:
+  // CreateFile returns INVALID_HANDLE_VALUE in case of failure.
+  static constexpr HANDLE kFileHandleInvalid = INVALID_HANDLE_VALUE;
+  // CreateFileMapping returns nullptr in case of failure.
+  static constexpr HANDLE kMappingHandleInvalid = nullptr;
+
   Win32MappedMemory(const std::filesystem::path& path, Mode mode)
       : MappedMemory(path, mode) {}
 
@@ -29,10 +34,10 @@ class Win32MappedMemory : public MappedMemory {
     if (data_) {
       UnmapViewOfFile(data_);
     }
-    if (mapping_handle != INVALID_HANDLE_VALUE) {
+    if (mapping_handle != kMappingHandleInvalid) {
       CloseHandle(mapping_handle);
     }
-    if (file_handle != INVALID_HANDLE_VALUE) {
+    if (file_handle != kFileHandleInvalid) {
       CloseHandle(file_handle);
     }
   }
@@ -42,11 +47,11 @@ class Win32MappedMemory : public MappedMemory {
       UnmapViewOfFile(data_);
       data_ = nullptr;
     }
-    if (mapping_handle != INVALID_HANDLE_VALUE) {
+    if (mapping_handle != kMappingHandleInvalid) {
       CloseHandle(mapping_handle);
-      mapping_handle = INVALID_HANDLE_VALUE;
+      mapping_handle = kMappingHandleInvalid;
     }
-    if (file_handle != INVALID_HANDLE_VALUE) {
+    if (file_handle != kFileHandleInvalid) {
       if (truncate_size) {
         LONG distance_high = truncate_size >> 32;
         SetFilePointer(file_handle, truncate_size & 0xFFFFFFFF, &distance_high,
@@ -55,7 +60,7 @@ class Win32MappedMemory : public MappedMemory {
       }
 
       CloseHandle(file_handle);
-      file_handle = INVALID_HANDLE_VALUE;
+      file_handle = kFileHandleInvalid;
     }
   }
 
@@ -65,8 +70,13 @@ class Win32MappedMemory : public MappedMemory {
     size_t aligned_length = length + (offset - aligned_offset);
 
     UnmapViewOfFile(data_);
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
     data_ = MapViewOfFile(mapping_handle, view_access_, aligned_offset >> 32,
                           aligned_offset & 0xFFFFFFFF, aligned_length);
+#else
+    data_ = MapViewOfFileFromApp(mapping_handle, ULONG(view_access_),
+                                 ULONG64(aligned_offset), aligned_length);
+#endif
     if (!data_) {
       return false;
     }
@@ -83,8 +93,8 @@ class Win32MappedMemory : public MappedMemory {
     return true;
   }
 
-  HANDLE file_handle = INVALID_HANDLE_VALUE;
+  HANDLE file_handle = kFileHandleInvalid;
-  HANDLE mapping_handle = INVALID_HANDLE_VALUE;
+  HANDLE mapping_handle = kMappingHandleInvalid;
   DWORD view_access_ = 0;
 };
 
@@ -125,20 +135,32 @@ std::unique_ptr<MappedMemory> MappedMemory::Open(
 
   mm->file_handle = CreateFile(path.c_str(), file_access, file_share, nullptr,
                                create_mode, FILE_ATTRIBUTE_NORMAL, nullptr);
-  if (mm->file_handle == INVALID_HANDLE_VALUE) {
+  if (mm->file_handle == Win32MappedMemory::kFileHandleInvalid) {
     return nullptr;
   }
 
-  mm->mapping_handle = CreateFileMapping(mm->file_handle, nullptr,
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
-                                         mapping_protect, aligned_length >> 32,
+  mm->mapping_handle = CreateFileMapping(
-                                         aligned_length & 0xFFFFFFFF, nullptr);
+      mm->file_handle, nullptr, mapping_protect, DWORD(aligned_length >> 32),
-  if (mm->mapping_handle == INVALID_HANDLE_VALUE) {
+      DWORD(aligned_length), nullptr);
+#else
+  mm->mapping_handle =
+      CreateFileMappingFromApp(mm->file_handle, nullptr, ULONG(mapping_protect),
+                               ULONG64(aligned_length), nullptr);
+#endif
+  if (mm->mapping_handle == Win32MappedMemory::kMappingHandleInvalid) {
     return nullptr;
   }
 
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
   mm->data_ = reinterpret_cast<uint8_t*>(MapViewOfFile(
-      mm->mapping_handle, view_access, static_cast<DWORD>(aligned_offset >> 32),
+      mm->mapping_handle, view_access, DWORD(aligned_offset >> 32),
-      static_cast<DWORD>(aligned_offset & 0xFFFFFFFF), aligned_length));
+      DWORD(aligned_offset), aligned_length));
+#else
+  mm->data_ = reinterpret_cast<uint8_t*>(
+      MapViewOfFileFromApp(mm->mapping_handle, ULONG(view_access),
+                           ULONG64(aligned_offset), aligned_length));
+#endif
   if (!mm->data_) {
     return nullptr;
   }
@@ -203,8 +225,8 @@ class Win32ChunkedMappedMemoryWriter : public ChunkedMappedMemoryWriter {
   class Chunk {
    public:
     explicit Chunk(size_t capacity)
-        : file_handle_(0),
+        : file_handle_(Win32MappedMemory::kFileHandleInvalid),
-          mapping_handle_(0),
+          mapping_handle_(Win32MappedMemory::kMappingHandleInvalid),
           data_(nullptr),
           offset_(0),
           capacity_(capacity),
@@ -214,10 +236,10 @@ class Win32ChunkedMappedMemoryWriter : public ChunkedMappedMemoryWriter {
       if (data_) {
         UnmapViewOfFile(data_);
       }
-      if (mapping_handle_) {
+      if (mapping_handle_ != Win32MappedMemory::kMappingHandleInvalid) {
         CloseHandle(mapping_handle_);
       }
-      if (file_handle_) {
+      if (file_handle_ != Win32MappedMemory::kFileHandleInvalid) {
         CloseHandle(file_handle_);
       }
     }
@@ -231,14 +253,20 @@ class Win32ChunkedMappedMemoryWriter : public ChunkedMappedMemoryWriter {
 
       file_handle_ = CreateFile(path.c_str(), file_access, file_share, nullptr,
                                 create_mode, FILE_ATTRIBUTE_NORMAL, nullptr);
-      if (!file_handle_) {
+      if (file_handle_ == Win32MappedMemory::kFileHandleInvalid) {
         return false;
       }
 
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
       mapping_handle_ =
-          CreateFileMapping(file_handle_, nullptr, mapping_protect, 0,
+          CreateFileMapping(file_handle_, nullptr, mapping_protect,
-                            static_cast<DWORD>(capacity_), nullptr);
+                            DWORD(capacity_ >> 32), DWORD(capacity_), nullptr);
-      if (!mapping_handle_) {
+#else
+      mapping_handle_ = CreateFileMappingFromApp(file_handle_, nullptr,
+                                                 ULONG(mapping_protect),
+                                                 ULONG64(capacity_), nullptr);
+#endif
+      if (mapping_handle_ == Win32MappedMemory::kMappingHandleInvalid) {
         return false;
       }
 
@@ -247,10 +275,32 @@ class Win32ChunkedMappedMemoryWriter : public ChunkedMappedMemoryWriter {
       if (low_address_space) {
         bool successful = false;
         data_ = reinterpret_cast<uint8_t*>(0x10000000);
+#if !WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
+        HANDLE process = GetCurrentProcess();
+#endif
         for (int i = 0; i < 1000; ++i) {
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
           if (MapViewOfFileEx(mapping_handle_, view_access, 0, 0, capacity_,
                               data_)) {
             successful = true;
+          }
+#else
+          // VirtualAlloc2FromApp and MapViewOfFile3FromApp were added in
+          // 10.0.17134.0.
+          // https://docs.microsoft.com/en-us/uwp/win32-and-com/win32-apis
+          if (VirtualAlloc2FromApp(process, data_, capacity_,
+                                   MEM_RESERVE | MEM_RESERVE_PLACEHOLDER,
+                                   PAGE_NOACCESS, nullptr, 0)) {
+            if (MapViewOfFile3FromApp(mapping_handle_, process, data_, 0,
+                                      capacity_, MEM_REPLACE_PLACEHOLDER,
+                                      ULONG(mapping_protect), nullptr, 0)) {
+              successful = true;
+            } else {
+              VirtualFree(data_, capacity_, MEM_RELEASE);
+            }
+          }
+#endif
+          if (successful) {
             break;
           }
           data_ += capacity_;
@@ -261,8 +311,13 @@ class Win32ChunkedMappedMemoryWriter : public ChunkedMappedMemoryWriter {
           }
         }
       } else {
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
         data_ = reinterpret_cast<uint8_t*>(
             MapViewOfFile(mapping_handle_, view_access, 0, 0, capacity_));
+#else
+        data_ = reinterpret_cast<uint8_t*>(MapViewOfFileFromApp(
+            mapping_handle_, ULONG(view_access), 0, capacity_));
+#endif
       }
       if (!data_) {
         return false;

src/xenia/base/memory.cc

@@ -8,11 +8,26 @@
  */
 
 #include "xenia/base/memory.h"
+#include "xenia/base/cvar.h"
 #include "xenia/base/platform.h"
 
 #include <algorithm>
 
+DEFINE_bool(
+    writable_executable_memory, true,
+    "Allow mapping memory with both write and execute access, for simulating "
+    "behavior on platforms where that's not supported",
+    "Memory");
+
 namespace xe {
+namespace memory {
+
+bool IsWritableExecutableMemoryPreferred() {
+  return IsWritableExecutableMemorySupported() &&
+         cvars::writable_executable_memory;
+}
+
+}  // namespace memory
 
 // TODO(benvanik): fancy AVX versions.
 // https://github.com/gnuradio/volk/blob/master/kernels/volk/volk_16u_byteswap.h

src/xenia/base/memory.h

@@ -18,6 +18,7 @@
 
 #include "xenia/base/assert.h"
 #include "xenia/base/byte_order.h"
+#include "xenia/base/platform.h"
 
 namespace xe {
 namespace memory {
@@ -34,6 +35,7 @@ enum class PageAccess {
   kNoAccess = 0,
   kReadOnly = 1 << 0,
   kReadWrite = kReadOnly | 1 << 1,
+  kExecuteReadOnly = kReadOnly | 1 << 2,
   kExecuteReadWrite = kReadWrite | 1 << 2,
 };
 
@@ -48,6 +50,16 @@ enum class DeallocationType {
   kDecommit = 1 << 1,
 };
 
+// Whether the host allows the pages to be allocated or mapped with
+// PageAccess::kExecuteReadWrite - if not, separate mappings backed by the same
+// memory-mapped file must be used to write to executable pages.
+bool IsWritableExecutableMemorySupported();
+
+// Whether PageAccess::kExecuteReadWrite is a supported and preferred way of
+// writing executable memory, useful for simulating how Xenia would work without
+// writable executable memory on a system with it.
+bool IsWritableExecutableMemoryPreferred();
+
 // Allocates a block of memory at the given page-aligned base address.
 // Fails if the memory is not available.
 // Specify nullptr for base_address to leave it up to the system.
@@ -96,12 +108,21 @@ void AlignedFree(T* ptr) {
 #endif  // XE_COMPILER_MSVC
 }
 
+#if XE_PLATFORM_WIN32
+// HANDLE.
 typedef void* FileMappingHandle;
+constexpr FileMappingHandle kFileMappingHandleInvalid = nullptr;
+#else
+// File descriptor.
+typedef int FileMappingHandle;
+constexpr FileMappingHandle kFileMappingHandleInvalid = -1;
+#endif
 
 FileMappingHandle CreateFileMappingHandle(const std::filesystem::path& path,
                                           size_t length, PageAccess access,
                                           bool commit);
-void CloseFileMappingHandle(FileMappingHandle handle);
+void CloseFileMappingHandle(FileMappingHandle handle,
+                            const std::filesystem::path& path);
 void* MapFileView(FileMappingHandle handle, void* base_address, size_t length,
                   PageAccess access, size_t file_offset);
 bool UnmapFileView(FileMappingHandle handle, void* base_address, size_t length);

src/xenia/base/memory_posix.cc

@@ -8,12 +8,23 @@
  */
 
 #include "xenia/base/memory.h"
-#include "xenia/base/string.h"
 
 #include <fcntl.h>
 #include <sys/mman.h>
 #include <unistd.h>
 
+#include "xenia/base/math.h"
+#include "xenia/base/platform.h"
+#include "xenia/base/string.h"
+
+#if XE_PLATFORM_ANDROID
+#include <linux/ashmem.h>
+#include <string.h>
+#include <sys/ioctl.h>
+
+#include "xenia/base/platform_android.h"
+#endif
+
 namespace xe {
 namespace memory {
 
@@ -28,6 +39,8 @@ uint32_t ToPosixProtectFlags(PageAccess access) {
       return PROT_READ;
     case PageAccess::kReadWrite:
       return PROT_READ | PROT_WRITE;
+    case PageAccess::kExecuteReadOnly:
+      return PROT_READ | PROT_EXEC;
     case PageAccess::kExecuteReadWrite:
       return PROT_READ | PROT_WRITE | PROT_EXEC;
     default:
@@ -36,11 +49,19 @@ uint32_t ToPosixProtectFlags(PageAccess access) {
   }
 }
 
+bool IsWritableExecutableMemorySupported() { return true; }
+
 void* AllocFixed(void* base_address, size_t length,
                  AllocationType allocation_type, PageAccess access) {
   // mmap does not support reserve / commit, so ignore allocation_type.
   uint32_t prot = ToPosixProtectFlags(access);
-  return mmap(base_address, length, prot, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+  void* result = mmap(base_address, length, prot,
+                      MAP_PRIVATE | MAP_FIXED | MAP_ANONYMOUS, -1, 0);
+  if (result == MAP_FAILED) {
+    return nullptr;
+  } else {
+    return result;
+  }
 }
 
 bool DeallocFixed(void* base_address, size_t length,
@@ -64,12 +85,38 @@ bool QueryProtect(void* base_address, size_t& length, PageAccess& access_out) {
 FileMappingHandle CreateFileMappingHandle(const std::filesystem::path& path,
                                           size_t length, PageAccess access,
                                           bool commit) {
+#if XE_PLATFORM_ANDROID
+  if (xe::platform::android::api_level() >= 26) {
+    // TODO(Triang3l): Check if memfd can be used instead on API 30+.
+    int sharedmem_fd =
+        xe::platform::android::api_functions().api_26.ASharedMemory_create(
+            path.c_str(), length);
+    return sharedmem_fd >= 0 ? sharedmem_fd : kFileMappingHandleInvalid;
+  }
+
+  // Use /dev/ashmem on API versions below 26, which added ASharedMemory.
+  // /dev/ashmem was disabled on API 29 for apps targeting it.
+  // https://chromium.googlesource.com/chromium/src/+/master/third_party/ashmem/ashmem-dev.c
+  int ashmem_fd = open("/" ASHMEM_NAME_DEF, O_RDWR);
+  if (ashmem_fd < 0) {
+    return kFileMappingHandleInvalid;
+  }
+  char ashmem_name[ASHMEM_NAME_LEN];
+  strlcpy(ashmem_name, path.c_str(), xe::countof(ashmem_name));
+  if (ioctl(ashmem_fd, ASHMEM_SET_NAME, ashmem_name) < 0 ||
+      ioctl(ashmem_fd, ASHMEM_SET_SIZE, length) < 0) {
+    close(ashmem_fd);
+    return kFileMappingHandleInvalid;
+  }
+  return ashmem_fd;
+#else
   int oflag;
   switch (access) {
     case PageAccess::kNoAccess:
       oflag = 0;
       break;
     case PageAccess::kReadOnly:
+    case PageAccess::kExecuteReadOnly:
       oflag = O_RDONLY;
       break;
     case PageAccess::kReadWrite:
@@ -78,27 +125,33 @@ FileMappingHandle CreateFileMappingHandle(const std::filesystem::path& path,
       break;
     default:
       assert_always();
-      return nullptr;
+      return kFileMappingHandleInvalid;
   }
-
   oflag |= O_CREAT;
-  int ret = shm_open(path.c_str(), oflag, 0777);
+  auto full_path = "/" / path;
-  if (ret > 0) {
+  int ret = shm_open(full_path.c_str(), oflag, 0777);
-    ftruncate64(ret, length);
+  if (ret < 0) {
+    return kFileMappingHandleInvalid;
   }
-
+  ftruncate64(ret, length);
-  return ret <= 0 ? nullptr : reinterpret_cast<FileMappingHandle>(ret);
+  return ret;
+#endif
 }
 
-void CloseFileMappingHandle(FileMappingHandle handle) {
+void CloseFileMappingHandle(FileMappingHandle handle,
-  close((intptr_t)handle);
+                            const std::filesystem::path& path) {
+  close(handle);
+#if !XE_PLATFORM_ANDROID
+  auto full_path = "/" / path;
+  shm_unlink(full_path.c_str());
+#endif
 }
 
 void* MapFileView(FileMappingHandle handle, void* base_address, size_t length,
                   PageAccess access, size_t file_offset) {
   uint32_t prot = ToPosixProtectFlags(access);
-  return mmap64(base_address, length, prot, MAP_PRIVATE | MAP_ANONYMOUS,
+  return mmap64(base_address, length, prot, MAP_PRIVATE | MAP_ANONYMOUS, handle,
-                reinterpret_cast<intptr_t>(handle), file_offset);
+                file_offset);
 }
 
 bool UnmapFileView(FileMappingHandle handle, void* base_address,

src/xenia/base/memory_win.cc

@@ -42,6 +42,8 @@ DWORD ToWin32ProtectFlags(PageAccess access) {
       return PAGE_READONLY;
     case PageAccess::kReadWrite:
       return PAGE_READWRITE;
+    case PageAccess::kExecuteReadOnly:
+      return PAGE_EXECUTE_READ;
     case PageAccess::kExecuteReadWrite:
       return PAGE_EXECUTE_READWRITE;
     default:
@@ -63,6 +65,8 @@ PageAccess ToXeniaProtectFlags(DWORD access) {
       return PageAccess::kReadOnly;
     case PAGE_READWRITE:
       return PageAccess::kReadWrite;
+    case PAGE_EXECUTE_READ:
+      return PageAccess::kExecuteReadOnly;
     case PAGE_EXECUTE_READWRITE:
       return PageAccess::kExecuteReadWrite;
     default:
@@ -70,6 +74,17 @@ PageAccess ToXeniaProtectFlags(DWORD access) {
   }
 }
 
+bool IsWritableExecutableMemorySupported() {
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
+  return true;
+#else
+  // To test FromApp functions on desktop, replace
+  // WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP) with 0 in the #ifs and
+  // link to WindowsApp.lib.
+  return false;
+#endif
+}
+
 void* AllocFixed(void* base_address, size_t length,
                  AllocationType allocation_type, PageAccess access) {
   DWORD alloc_type = 0;
@@ -88,7 +103,12 @@ void* AllocFixed(void* base_address, size_t length,
       break;
   }
   DWORD protect = ToWin32ProtectFlags(access);
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
   return VirtualAlloc(base_address, length, alloc_type, protect);
+#else
+  return VirtualAllocFromApp(base_address, length, ULONG(alloc_type),
+                             ULONG(protect));
+#endif
 }
 
 bool DeallocFixed(void* base_address, size_t length,
@@ -115,13 +135,19 @@ bool Protect(void* base_address, size_t length, PageAccess access,
     *out_old_access = PageAccess::kNoAccess;
   }
   DWORD new_protect = ToWin32ProtectFlags(access);
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
   DWORD old_protect = 0;
   BOOL result = VirtualProtect(base_address, length, new_protect, &old_protect);
+#else
+  ULONG old_protect = 0;
+  BOOL result = VirtualProtectFromApp(base_address, length, ULONG(new_protect),
+                                      &old_protect);
+#endif
   if (!result) {
     return false;
   }
   if (out_old_access) {
-    *out_old_access = ToXeniaProtectFlags(old_protect);
+    *out_old_access = ToXeniaProtectFlags(DWORD(old_protect));
   }
   return true;
 }
@@ -147,15 +173,25 @@ FileMappingHandle CreateFileMappingHandle(const std::filesystem::path& path,
                                           bool commit) {
   DWORD protect =
       ToWin32ProtectFlags(access) | (commit ? SEC_COMMIT : SEC_RESERVE);
-  return CreateFileMappingW(INVALID_HANDLE_VALUE, NULL, protect,
+  auto full_path = "Local" / path;
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
+  return CreateFileMappingW(INVALID_HANDLE_VALUE, nullptr, protect,
                             static_cast<DWORD>(length >> 32),
-                            static_cast<DWORD>(length), path.c_str());
+                            static_cast<DWORD>(length), full_path.c_str());
+#else
+  return CreateFileMappingFromApp(INVALID_HANDLE_VALUE, nullptr, ULONG(protect),
+                                  ULONG64(length), full_path.c_str());
+#endif
 }
 
-void CloseFileMappingHandle(FileMappingHandle handle) { CloseHandle(handle); }
+void CloseFileMappingHandle(FileMappingHandle handle,
+                            const std::filesystem::path& path) {
+  CloseHandle(handle);
+}
 
 void* MapFileView(FileMappingHandle handle, void* base_address, size_t length,
                   PageAccess access, size_t file_offset) {
+#if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
   DWORD target_address_low = static_cast<DWORD>(file_offset);
   DWORD target_address_high = static_cast<DWORD>(file_offset >> 32);
   DWORD file_access = 0;
@@ -166,6 +202,9 @@ void* MapFileView(FileMappingHandle handle, void* base_address, size_t length,
     case PageAccess::kReadWrite:
       file_access = FILE_MAP_ALL_ACCESS;
       break;
+    case PageAccess::kExecuteReadOnly:
+      file_access = FILE_MAP_READ | FILE_MAP_EXECUTE;
+      break;
     case PageAccess::kExecuteReadWrite:
       file_access = FILE_MAP_ALL_ACCESS | FILE_MAP_EXECUTE;
       break;
@@ -176,6 +215,25 @@ void* MapFileView(FileMappingHandle handle, void* base_address, size_t length,
   }
   return MapViewOfFileEx(handle, file_access, target_address_high,
                          target_address_low, length, base_address);
+#else
+  // VirtualAlloc2FromApp and MapViewOfFile3FromApp were added in 10.0.17134.0.
+  // https://docs.microsoft.com/en-us/uwp/win32-and-com/win32-apis
+  HANDLE process = GetCurrentProcess();
+  void* placeholder = VirtualAlloc2FromApp(
+      process, base_address, length, MEM_RESERVE | MEM_RESERVE_PLACEHOLDER,
+      PAGE_NOACCESS, nullptr, 0);
+  if (!placeholder) {
+    return nullptr;
+  }
+  void* mapping = MapViewOfFile3FromApp(
+      handle, process, placeholder, ULONG64(file_offset), length,
+      MEM_REPLACE_PLACEHOLDER, ULONG(ToWin32ProtectFlags(access)), nullptr, 0);
+  if (!mapping) {
+    VirtualFree(placeholder, length, MEM_RELEASE);
+    return nullptr;
+  }
+  return mapping;
+#endif
 }
 
 bool UnmapFileView(FileMappingHandle handle, void* base_address,

src/xenia/base/platform.h

@@ -31,8 +31,14 @@
 #define XE_PLATFORM_MAC 1
 #elif defined(WIN32) || defined(_WIN32)
 #define XE_PLATFORM_WIN32 1
-#else
+#elif defined(__ANDROID__)
+#define XE_PLATFORM_ANDROID 1
 #define XE_PLATFORM_LINUX 1
+#elif defined(__gnu_linux__)
+#define XE_PLATFORM_GNU_LINUX 1
+#define XE_PLATFORM_LINUX 1
+#else
+#error Unsupported target OS.
 #endif
 
 #if defined(__clang__)
@@ -51,8 +57,11 @@
 
 #if defined(_M_AMD64) || defined(__amd64__)
 #define XE_ARCH_AMD64 1
-#elif defined(_M_IX86)
+#elif defined(_M_ARM64) || defined(__aarch64__)
-#error "Xenia is not supported on 32-bit platforms."
+#define XE_ARCH_ARM64 1
+#elif defined(_M_IX86) || defined(__i386__) || defined(_M_ARM) || \
+    defined(__arm__)
+#error Xenia is not supported on 32-bit platforms.
 #elif defined(_M_PPC) || defined(__powerpc__)
 #define XE_ARCH_PPC 1
 #endif

src/xenia/base/platform_android.cc

@@ -0,0 +1,65 @@
+/**
+ ******************************************************************************
+ * Xenia : Xbox 360 Emulator Research Project                                 *
+ ******************************************************************************
+ * Copyright 2020 Ben Vanik. All rights reserved.                             *
+ * Released under the BSD license - see LICENSE in the root for more details. *
+ ******************************************************************************
+ */
+
+#include "xenia/base/platform_android.h"
+
+#include <android/configuration.h>
+#include <dlfcn.h>
+
+#include "xenia/base/assert.h"
+
+namespace xe {
+namespace platform {
+namespace android {
+
+static bool initialized = false;
+
+static int32_t api_level_ = __ANDROID_API__;
+
+static ApiFunctions api_functions_;
+
+void Initialize(const ANativeActivity* activity) {
+  if (initialized) {
+    return;
+  }
+
+  AConfiguration* configuration = AConfiguration_new();
+  AConfiguration_fromAssetManager(configuration, activity->assetManager);
+  api_level_ = AConfiguration_getSdkVersion(configuration);
+  AConfiguration_delete(configuration);
+
+  if (api_level_ >= 26) {
+    // Leaked intentionally as these will be usable anywhere, already loaded
+    // into the address space as the application is linked against them.
+    // https://chromium.googlesource.com/chromium/src/+/master/third_party/ashmem/ashmem-dev.c#201
+    void* libandroid = dlopen("libandroid.so", RTLD_NOW);
+    assert_not_null(libandroid);
+    void* libc = dlopen("libc.so", RTLD_NOW);
+    assert_not_null(libc);
+#define XE_PLATFORM_ANDROID_LOAD_API_FUNCTION(lib, name, api)    \
+  api_functions_.api_##api.name =                                \
+      reinterpret_cast<decltype(api_functions_.api_##api.name)>( \
+          dlsym(lib, #name));                                    \
+  assert_not_null(api_functions_.api_##api.name);
+    XE_PLATFORM_ANDROID_LOAD_API_FUNCTION(libandroid, ASharedMemory_create, 26);
+    // pthreads are a part of Bionic libc on Android.
+    XE_PLATFORM_ANDROID_LOAD_API_FUNCTION(libc, pthread_getname_np, 26);
+#undef XE_PLATFORM_ANDROID_LOAD_API_FUNCTION
+  }
+
+  initialized = true;
+}
+
+int32_t api_level() { return api_level_; }
+
+const ApiFunctions& api_functions() { return api_functions_; }
+
+}  // namespace android
+}  // namespace platform
+}  // namespace xe

src/xenia/base/platform_android.h

@@ -0,0 +1,48 @@
+/**
+ ******************************************************************************
+ * Xenia : Xbox 360 Emulator Research Project                                 *
+ ******************************************************************************
+ * Copyright 2020 Ben Vanik. All rights reserved.                             *
+ * Released under the BSD license - see LICENSE in the root for more details. *
+ ******************************************************************************
+ */
+
+#ifndef XENIA_BASE_PLATFORM_ANDROID_H_
+#define XENIA_BASE_PLATFORM_ANDROID_H_
+
+// NOTE: if you're including this file it means you are explicitly depending
+// on Android-specific headers. This is bad for portability and should be
+// avoided!
+
+#include <android/native_activity.h>
+#include <pthread.h>
+#include <cstddef>
+#include <cstdint>
+
+namespace xe {
+namespace platform {
+namespace android {
+
+// Must be called in onCreate of the first activity.
+void Initialize(const ANativeActivity* activity);
+
+// Returns the device API level - if not initialized, will return the minimum
+// level supported by Xenia.
+int32_t api_level();
+
+// Android API functions added after the minimum supported API version.
+struct ApiFunctions {
+  struct {
+    // libandroid
+    int (*ASharedMemory_create)(const char* name, size_t size);
+    // libc
+    int (*pthread_getname_np)(pthread_t pthread, char* buf, size_t n);
+  } api_26;
+};
+const ApiFunctions& api_functions();
+
+}  // namespace android
+}  // namespace platform
+}  // namespace xe
+
+#endif  // XENIA_BASE_PLATFORM_ANDROID_H_

src/xenia/base/system_win.cc

@@ -15,7 +15,7 @@ namespace xe {
 
 void LaunchWebBrowser(const std::string& url) {
   auto temp = xe::to_utf16(url);
-  ShellExecuteW(nullptr, L"open", reinterpret_cast<LPCWSTR>(url.c_str()),
+  ShellExecuteW(nullptr, L"open", reinterpret_cast<LPCWSTR>(temp.c_str()),
                 nullptr, nullptr, SW_SHOWNORMAL);
 }
 

src/xenia/base/testing/memory_test.cc

@@ -10,6 +10,9 @@
 #include "xenia/base/memory.h"
 
 #include "third_party/catch/include/catch.hpp"
+#include "third_party/fmt/include/fmt/format.h"
+
+#include "xenia/base/clock.h"
 
 namespace xe {
 namespace base {
@@ -414,6 +417,58 @@ TEST_CASE("copy_and_swap_16_in_32_unaligned", "Copy and Swap") {
   REQUIRE(true == true);
 }
 
+TEST_CASE("create_and_close_file_mapping", "Virtual Memory Mapping") {
+  auto path = fmt::format("xenia_test_{}", Clock::QueryHostTickCount());
+  auto memory = xe::memory::CreateFileMappingHandle(
+      path, 0x100, xe::memory::PageAccess::kReadWrite, true);
+  REQUIRE(memory != xe::memory::kFileMappingHandleInvalid);
+  xe::memory::CloseFileMappingHandle(memory, path);
+}
+
+TEST_CASE("map_view", "Virtual Memory Mapping") {
+  auto path = fmt::format("xenia_test_{}", Clock::QueryHostTickCount());
+  const size_t length = 0x100;
+  auto memory = xe::memory::CreateFileMappingHandle(
+      path, length, xe::memory::PageAccess::kReadWrite, true);
+  REQUIRE(memory != xe::memory::kFileMappingHandleInvalid);
+
+  uintptr_t address = 0x100000000;
+  auto view =
+      xe::memory::MapFileView(memory, reinterpret_cast<void*>(address), length,
+                              xe::memory::PageAccess::kReadWrite, 0);
+  REQUIRE(reinterpret_cast<uintptr_t>(view) == address);
+
+  xe::memory::UnmapFileView(memory, reinterpret_cast<void*>(address), length);
+  xe::memory::CloseFileMappingHandle(memory, path);
+}
+
+TEST_CASE("read_write_view", "Virtual Memory Mapping") {
+  const size_t length = 0x100;
+  auto path = fmt::format("xenia_test_{}", Clock::QueryHostTickCount());
+  auto memory = xe::memory::CreateFileMappingHandle(
+      path, length, xe::memory::PageAccess::kReadWrite, true);
+  REQUIRE(memory != xe::memory::kFileMappingHandleInvalid);
+
+  uintptr_t address = 0x100000000;
+  auto view =
+      xe::memory::MapFileView(memory, reinterpret_cast<void*>(address), length,
+                              xe::memory::PageAccess::kReadWrite, 0);
+  REQUIRE(reinterpret_cast<uintptr_t>(view) == address);
+
+  for (uint32_t i = 0; i < length; i += sizeof(uint8_t)) {
+    auto p_value = reinterpret_cast<uint8_t*>(address + i);
+    *p_value = i;
+  }
+  for (uint32_t i = 0; i < length; i += sizeof(uint8_t)) {
+    auto p_value = reinterpret_cast<uint8_t*>(address + i);
+    uint8_t value = *p_value;
+    REQUIRE(value == i);
+  }
+
+  xe::memory::UnmapFileView(memory, reinterpret_cast<void*>(address), length);
+  xe::memory::CloseFileMappingHandle(memory, path);
+}
+
 }  // namespace test
 }  // namespace base
 }  // namespace xe

src/xenia/base/testing/threading_test.cc

@@ -174,8 +174,8 @@ TEST_CASE("HighResolutionTimer") {
   {
     const auto interval1 = 100ms;
     const auto interval2 = 200ms;
-    std::atomic<uint64_t> counter1;
+    std::atomic<uint64_t> counter1(0);
-    std::atomic<uint64_t> counter2;
+    std::atomic<uint64_t> counter2(0);
     auto start = std::chrono::steady_clock::now();
     auto cb1 = [&counter1] { ++counter1; };
     auto cb2 = [&counter2] { ++counter2; };
@@ -813,7 +813,7 @@ TEST_CASE("Create and Run Thread", "Thread") {
   result = Wait(Thread::GetCurrentThread(), false, 50ms);
   REQUIRE(result == WaitResult::kTimeout);
 
-  params.stack_size = 16 * 1024;
+  params.stack_size = 16 * 1024 * 1024;
   thread = Thread::Create(params, [] {
     while (true) {
       Thread::Exit(-1);

src/xenia/base/threading.h

@@ -95,9 +95,6 @@ void set_current_thread_id(uint32_t id);
 
 // Sets the current thread name.
 void set_name(const std::string_view name);
-// Sets the target thread name.
-void set_name(std::thread::native_handle_type handle,
-              const std::string_view name);
 
 // Yields the current thread to the scheduler. Maybe.
 void MaybeYield();

src/xenia/base/threading_linux.cc

@@ -1,17 +0,0 @@
-/**
- ******************************************************************************
- * Xenia : Xbox 360 Emulator Research Project                                 *
- ******************************************************************************
- * Copyright 2015 Ben Vanik. All rights reserved.                             *
- * Released under the BSD license - see LICENSE in the root for more details. *
- ******************************************************************************
- */
-
-#include "xenia/base/threading.h"
-
-#include <pthread.h>
-#include <time.h>
-
-namespace xe {
-namespace threading {}  // namespace threading
-}  // namespace xe

src/xenia/base/threading_mac.cc

@@ -26,10 +26,6 @@ uint32_t current_thread_id() {
 
 void set_name(const std::string& name) { pthread_setname_np(name.c_str()); }
 
-void set_name(std::thread::native_handle_type handle, const std::string& name) {
-  // ?
-}
-
 void MaybeYield() { pthread_yield_np(); }
 
 void Sleep(std::chrono::microseconds duration) {

src/xenia/base/threading_posix.cc

@@ -11,6 +11,7 @@
 
 #include "xenia/base/assert.h"
 #include "xenia/base/logging.h"
+#include "xenia/base/platform.h"
 
 #include <pthread.h>
 #include <signal.h>
@@ -19,9 +20,18 @@
 #include <sys/time.h>
 #include <sys/types.h>
 #include <unistd.h>
+#include <array>
+#include <cstring>
 #include <ctime>
 #include <memory>
 
+#if XE_PLATFORM_ANDROID
+#include <sched.h>
+
+#include "xenia/base/platform_android.h"
+#include "xenia/base/string_util.h"
+#endif
+
 namespace xe {
 namespace threading {
 
@@ -43,6 +53,12 @@ enum class SignalType {
   kTimer,
   kThreadSuspend,
   kThreadUserCallback,
+#if XE_PLATFORM_ANDROID
+  // pthread_cancel is not available on Android, using a signal handler for
+  // simplified PTHREAD_CANCEL_ASYNCHRONOUS-like behavior - not disabling
+  // cancellation currently, so should be enough.
+  kThreadTerminate,
+#endif
   k_Count
 };
 
@@ -80,15 +96,12 @@ uint32_t current_thread_system_id() {
   return static_cast<uint32_t>(syscall(SYS_gettid));
 }
 
-void set_name(std::thread::native_handle_type handle,
-              const std::string_view name) {
-  pthread_setname_np(handle, std::string(name).c_str());
-}
-
-void set_name(const std::string_view name) { set_name(pthread_self(), name); }
-
 void MaybeYield() {
+#if XE_PLATFORM_ANDROID
+  sched_yield();
+#else
   pthread_yield();
+#endif
   __sync_synchronize();
 }
 
@@ -482,7 +495,11 @@ class PosixCondition<Thread> : public PosixConditionBase {
         signaled_(false),
         exit_code_(0),
         state_(State::kUninitialized),
-        suspend_count_(0) {}
+        suspend_count_(0) {
+#if XE_PLATFORM_ANDROID
+    android_pre_api_26_name_[0] = '\0';
+#endif
+  }
   bool Initialize(Thread::CreationParameters params,
                   ThreadStartData* start_data) {
     start_data->create_suspended = params.create_suspended;
@@ -505,6 +522,7 @@ class PosixCondition<Thread> : public PosixConditionBase {
       }
     }
     if (pthread_create(&thread_, &attr, ThreadStartRoutine, start_data) != 0) {
+      pthread_attr_destroy(&attr);
       return false;
     }
     pthread_attr_destroy(&attr);
@@ -517,13 +535,24 @@ class PosixCondition<Thread> : public PosixConditionBase {
       : thread_(thread),
         signaled_(false),
         exit_code_(0),
-        state_(State::kRunning) {}
+        state_(State::kRunning) {
+#if XE_PLATFORM_ANDROID
+    android_pre_api_26_name_[0] = '\0';
+#endif
+  }
 
   virtual ~PosixCondition() {
     if (thread_ && !signaled_) {
+#if XE_PLATFORM_ANDROID
+      if (pthread_kill(thread_,
+                       GetSystemSignal(SignalType::kThreadTerminate)) != 0) {
+        assert_always();
+      }
+#else
       if (pthread_cancel(thread_) != 0) {
         assert_always();
       }
+#endif
       if (pthread_join(thread_, nullptr) != 0) {
         assert_always();
       }
@@ -537,8 +566,24 @@ class PosixCondition<Thread> : public PosixConditionBase {
     auto result = std::array<char, 17>{'\0'};
     std::unique_lock<std::mutex> lock(state_mutex_);
     if (state_ != State::kUninitialized && state_ != State::kFinished) {
-      if (pthread_getname_np(thread_, result.data(), result.size() - 1) != 0)
+#if XE_PLATFORM_ANDROID
+      // pthread_getname_np was added in API 26 - below that, store the name in
+      // this object, which may be only modified through Xenia threading, but
+      // should be enough in most cases.
+      if (xe::platform::android::api_level() >= 26) {
+        if (xe::platform::android::api_functions().api_26.pthread_getname_np(
+                thread_, result.data(), result.size() - 1) != 0) {
+          assert_always();
+        }
+      } else {
+        std::lock_guard<std::mutex> lock(android_pre_api_26_name_mutex_);
+        std::strcpy(result.data(), android_pre_api_26_name_);
+      }
+#else
+      if (pthread_getname_np(thread_, result.data(), result.size() - 1) != 0) {
         assert_always();
+      }
+#endif
     }
     return std::string(result.data());
   }
@@ -547,18 +592,39 @@ class PosixCondition<Thread> : public PosixConditionBase {
     WaitStarted();
     std::unique_lock<std::mutex> lock(state_mutex_);
     if (state_ != State::kUninitialized && state_ != State::kFinished) {
-      threading::set_name(static_cast<std::thread::native_handle_type>(thread_),
+      pthread_setname_np(thread_, std::string(name).c_str());
-                          name);
+#if XE_PLATFORM_ANDROID
+      SetAndroidPreApi26Name(name);
+#endif
     }
   }
 
+#if XE_PLATFORM_ANDROID
+  void SetAndroidPreApi26Name(const std::string_view name) {
+    if (xe::platform::android::api_level() >= 26) {
+      return;
+    }
+    std::lock_guard<std::mutex> lock(android_pre_api_26_name_mutex_);
+    xe::string_util::copy_truncating(android_pre_api_26_name_, name,
+                                     xe::countof(android_pre_api_26_name_));
+  }
+#endif
+
   uint32_t system_id() const { return static_cast<uint32_t>(thread_); }
 
   uint64_t affinity_mask() {
     WaitStarted();
     cpu_set_t cpu_set;
-    if (pthread_getaffinity_np(thread_, sizeof(cpu_set_t), &cpu_set) != 0)
+#if XE_PLATFORM_ANDROID
+    if (sched_getaffinity(pthread_gettid_np(thread_), sizeof(cpu_set_t),
+                          &cpu_set) != 0) {
       assert_always();
+    }
+#else
+    if (pthread_getaffinity_np(thread_, sizeof(cpu_set_t), &cpu_set) != 0) {
+      assert_always();
+    }
+#endif
     uint64_t result = 0;
     auto cpu_count = std::min(CPU_SETSIZE, 64);
     for (auto i = 0u; i < cpu_count; i++) {
@@ -577,9 +643,16 @@ class PosixCondition<Thread> : public PosixConditionBase {
         CPU_SET(i, &cpu_set);
       }
     }
+#if XE_PLATFORM_ANDROID
+    if (sched_setaffinity(pthread_gettid_np(thread_), sizeof(cpu_set_t),
+                          &cpu_set) != 0) {
+      assert_always();
+    }
+#else
     if (pthread_setaffinity_np(thread_, sizeof(cpu_set_t), &cpu_set) != 0) {
       assert_always();
     }
+#endif
   }
 
   int priority() {
@@ -608,8 +681,13 @@ class PosixCondition<Thread> : public PosixConditionBase {
     user_callback_ = std::move(callback);
     sigval value{};
     value.sival_ptr = this;
+#if XE_PLATFORM_ANDROID
+    sigqueue(pthread_gettid_np(thread_),
+             GetSystemSignal(SignalType::kThreadUserCallback), value);
+#else
     pthread_sigqueue(thread_, GetSystemSignal(SignalType::kThreadUserCallback),
                      value);
+#endif
   }
 
   void CallUserCallback() {
@@ -665,7 +743,16 @@ class PosixCondition<Thread> : public PosixConditionBase {
     signaled_ = true;
     cond_.notify_all();
 
-    if (pthread_cancel(thread) != 0) assert_always();
+#ifdef XE_PLATFORM_ANDROID
+    if (pthread_kill(thread, GetSystemSignal(SignalType::kThreadTerminate)) !=
+        0) {
+      assert_always();
+    }
+#else
+    if (pthread_cancel(thread) != 0) {
+      assert_always();
+    }
+#endif
   }
 
   void WaitStarted() const {
@@ -703,6 +790,12 @@ class PosixCondition<Thread> : public PosixConditionBase {
   mutable std::mutex callback_mutex_;
   mutable std::condition_variable state_signal_;
   std::function<void()> user_callback_;
+#if XE_PLATFORM_ANDROID
+  // Name accessible via name() on Android before API 26 which added
+  // pthread_getname_np.
+  mutable std::mutex android_pre_api_26_name_mutex_;
+  char android_pre_api_26_name_[16];
+#endif
 };
 
 class PosixWaitHandle {
@@ -722,7 +815,7 @@ class PosixConditionHandle : public T, public PosixWaitHandle {
   PosixConditionHandle(uint32_t initial_count, uint32_t maximum_count);
   ~PosixConditionHandle() override = default;
 
-  PosixConditionBase& condition() override { return handle_; }
+  PosixCondition<T>& condition() override { return handle_; }
   void* native_handle() const override { return handle_.native_handle(); }
 
  protected:
@@ -941,9 +1034,11 @@ class PosixThread : public PosixConditionHandle<Thread> {
 thread_local PosixThread* current_thread_ = nullptr;
 
 void* PosixCondition<Thread>::ThreadStartRoutine(void* parameter) {
+#if !XE_PLATFORM_ANDROID
   if (pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, nullptr) != 0) {
     assert_always();
   }
+#endif
   threading::set_name("");
 
   auto start_data = static_cast<ThreadStartData*>(parameter);
@@ -990,6 +1085,9 @@ std::unique_ptr<Thread> Thread::Create(CreationParameters params,
                                        std::function<void()> start_routine) {
   install_signal_handler(SignalType::kThreadSuspend);
   install_signal_handler(SignalType::kThreadUserCallback);
+#if XE_PLATFORM_ANDROID
+  install_signal_handler(SignalType::kThreadTerminate);
+#endif
   auto thread = std::make_unique<PosixThread>();
   if (!thread->Initialize(params, std::move(start_routine))) return nullptr;
   assert_not_null(thread);
@@ -1006,7 +1104,10 @@ Thread* Thread::GetCurrentThread() {
   pthread_t handle = pthread_self();
 
   current_thread_ = new PosixThread(handle);
-  atexit([] { delete current_thread_; });
+  // TODO(bwrsandman): Disabling deleting thread_local current thread to prevent
+  //                   assert in destructor. Since this is thread local, the
+  //                   "memory leaking" is controlled.
+  // atexit([] { delete current_thread_; });
 
   return current_thread_;
 }
@@ -1023,6 +1124,15 @@ void Thread::Exit(int exit_code) {
   }
 }
 
+void set_name(const std::string_view name) {
+  pthread_setname_np(pthread_self(), std::string(name).c_str());
+#if XE_PLATFORM_ANDROID
+  if (xe::platform::android::api_level() < 26 && current_thread_) {
+    current_thread_->condition().SetAndroidPreApi26Name(name);
+  }
+#endif
+}
+
 static void signal_handler(int signal, siginfo_t* info, void* /*context*/) {
   switch (GetSystemSignalType(signal)) {
     case SignalType::kHighResolutionTimer: {
@@ -1049,6 +1159,11 @@ static void signal_handler(int signal, siginfo_t* info, void* /*context*/) {
         p_thread->CallUserCallback();
       }
     } break;
+#if XE_PLATFORM_ANDROID
+    case SignalType::kThreadTerminate: {
+      pthread_exit(reinterpret_cast<void*>(-1));
+    } break;
+#endif
     default:
       assert_always();
   }

src/xenia/base/threading_win.cc

@@ -57,7 +57,7 @@ void raise_thread_name_exception(HANDLE thread, const std::string& name) {
   }
 }
 
-void set_name(HANDLE thread, const std::string_view name) {
+static void set_name(HANDLE thread, const std::string_view name) {
   auto kernel = GetModuleHandleW(L"kernel32.dll");
   if (kernel) {
     auto func =

src/xenia/cpu/backend/code_cache.h

@@ -10,6 +10,8 @@
 #ifndef XENIA_CPU_BACKEND_CODE_CACHE_H_
 #define XENIA_CPU_BACKEND_CODE_CACHE_H_
 
+#include <cstddef>
+#include <cstdint>
 #include <string>
 
 #include "xenia/cpu/function.h"
@@ -24,8 +26,8 @@ class CodeCache {
   virtual ~CodeCache() = default;
 
   virtual const std::filesystem::path& file_name() const = 0;
-  virtual uint32_t base_address() const = 0;
+  virtual uintptr_t execute_base_address() const = 0;
-  virtual uint32_t total_size() const = 0;
+  virtual size_t total_size() const = 0;
 
   // Finds a function based on the given host PC (that may be within a
   // function).

src/xenia/cpu/backend/x64/x64_code_cache.cc

@@ -40,11 +40,18 @@ X64CodeCache::~X64CodeCache() {
   }
 
   // Unmap all views and close mapping.
-  if (mapping_) {
+  if (mapping_ != xe::memory::kFileMappingHandleInvalid) {
-    xe::memory::UnmapFileView(mapping_, generated_code_base_,
+    if (generated_code_write_base_ &&
-                              kGeneratedCodeSize);
+        generated_code_write_base_ != generated_code_execute_base_) {
-    xe::memory::CloseFileMappingHandle(mapping_);
+      xe::memory::UnmapFileView(mapping_, generated_code_write_base_,
-    mapping_ = nullptr;
+                                kGeneratedCodeSize);
+    }
+    if (generated_code_execute_base_) {
+      xe::memory::UnmapFileView(mapping_, generated_code_execute_base_,
+                                kGeneratedCodeSize);
+    }
+    xe::memory::CloseFileMappingHandle(mapping_, file_name_);
+    mapping_ = xe::memory::kFileMappingHandleInvalid;
   }
 }
 
@@ -63,28 +70,51 @@ bool X64CodeCache::Initialize() {
   }
 
   // Create mmap file. This allows us to share the code cache with the debugger.
-  file_name_ =
+  file_name_ = fmt::format("xenia_code_cache_{}", Clock::QueryHostTickCount());
-      fmt::format("Local\\xenia_code_cache_{}", Clock::QueryHostTickCount());
   mapping_ = xe::memory::CreateFileMappingHandle(
       file_name_, kGeneratedCodeSize, xe::memory::PageAccess::kExecuteReadWrite,
       false);
-  if (!mapping_) {
+  if (mapping_ == xe::memory::kFileMappingHandleInvalid) {
     XELOGE("Unable to create code cache mmap");
     return false;
   }
 
   // Map generated code region into the file. Pages are committed as required.
-  generated_code_base_ = reinterpret_cast<uint8_t*>(xe::memory::MapFileView(
+  if (xe::memory::IsWritableExecutableMemoryPreferred()) {
-      mapping_, reinterpret_cast<void*>(kGeneratedCodeBase), kGeneratedCodeSize,
+    generated_code_execute_base_ =
-      xe::memory::PageAccess::kExecuteReadWrite, 0));
+        reinterpret_cast<uint8_t*>(xe::memory::MapFileView(
-  if (!generated_code_base_) {
+            mapping_, reinterpret_cast<void*>(kGeneratedCodeExecuteBase),
-    XELOGE("Unable to allocate code cache generated code storage");
+            kGeneratedCodeSize, xe::memory::PageAccess::kExecuteReadWrite, 0));
-    XELOGE(
+    generated_code_write_base_ = generated_code_execute_base_;
-        "This is likely because the {:X}-{:X} range is in use by some other "
+    if (!generated_code_execute_base_ || !generated_code_write_base_) {
-        "system DLL",
+      XELOGE("Unable to allocate code cache generated code storage");
-        static_cast<uint64_t>(kGeneratedCodeBase),
+      XELOGE(
-        kGeneratedCodeBase + kGeneratedCodeSize);
+          "This is likely because the {:X}-{:X} range is in use by some other "
-    return false;
+          "system DLL",
+          uint64_t(kGeneratedCodeExecuteBase),
+          uint64_t(kGeneratedCodeExecuteBase + kGeneratedCodeSize));
+      return false;
+    }
+  } else {
+    generated_code_execute_base_ =
+        reinterpret_cast<uint8_t*>(xe::memory::MapFileView(
+            mapping_, reinterpret_cast<void*>(kGeneratedCodeExecuteBase),
+            kGeneratedCodeSize, xe::memory::PageAccess::kExecuteReadOnly, 0));
+    generated_code_write_base_ =
+        reinterpret_cast<uint8_t*>(xe::memory::MapFileView(
+            mapping_, reinterpret_cast<void*>(kGeneratedCodeWriteBase),
+            kGeneratedCodeSize, xe::memory::PageAccess::kReadWrite, 0));
+    if (!generated_code_execute_base_ || !generated_code_write_base_) {
+      XELOGE("Unable to allocate code cache generated code storage");
+      XELOGE(
+          "This is likely because the {:X}-{:X} and {:X}-{:X} ranges are in "
+          "use by some other system DLL",
+          uint64_t(kGeneratedCodeExecuteBase),
+          uint64_t(kGeneratedCodeExecuteBase + kGeneratedCodeSize),
+          uint64_t(kGeneratedCodeWriteBase),
+          uint64_t(kGeneratedCodeWriteBase + kGeneratedCodeSize));
+      return false;
+    }
   }
 
   // Preallocate the function map to a large, reasonable size.
@@ -118,7 +148,7 @@ void X64CodeCache::CommitExecutableRange(uint32_t guest_low,
   xe::memory::AllocFixed(
       indirection_table_base_ + (guest_low - kIndirectionTableBase),
       guest_high - guest_low, xe::memory::AllocationType::kCommit,
-      xe::memory::PageAccess::kExecuteReadWrite);
+      xe::memory::PageAccess::kReadWrite);
 
   // Fill memory with the default value.
   uint32_t* p = reinterpret_cast<uint32_t*>(indirection_table_base_);
@@ -127,21 +157,26 @@ void X64CodeCache::CommitExecutableRange(uint32_t guest_low,
   }
 }
 
-void* X64CodeCache::PlaceHostCode(uint32_t guest_address, void* machine_code,
+void X64CodeCache::PlaceHostCode(uint32_t guest_address, void* machine_code,
-                                  const EmitFunctionInfo& func_info) {
+                                 const EmitFunctionInfo& func_info,
+                                 void*& code_execute_address_out,
+                                 void*& code_write_address_out) {
   // Same for now. We may use different pools or whatnot later on, like when
   // we only want to place guest code in a serialized cache on disk.
-  return PlaceGuestCode(guest_address, machine_code, func_info, nullptr);
+  PlaceGuestCode(guest_address, machine_code, func_info, nullptr,
+                 code_execute_address_out, code_write_address_out);
 }
 
-void* X64CodeCache::PlaceGuestCode(uint32_t guest_address, void* machine_code,
+void X64CodeCache::PlaceGuestCode(uint32_t guest_address, void* machine_code,
-                                   const EmitFunctionInfo& func_info,
+                                  const EmitFunctionInfo& func_info,
-                                   GuestFunction* function_info) {
+                                  GuestFunction* function_info,
+                                  void*& code_execute_address_out,
+                                  void*& code_write_address_out) {
   // Hold a lock while we bump the pointers up. This is important as the
   // unwind table requires entries AND code to be sorted in order.
   size_t low_mark;
   size_t high_mark;
-  uint8_t* code_address;
+  uint8_t* code_execute_address;
   UnwindReservation unwind_reservation;
   {
     auto global_lock = global_critical_region_.Acquire();
@@ -150,26 +185,33 @@ void* X64CodeCache::PlaceGuestCode(uint32_t guest_address, void* machine_code,
 
     // Reserve code.
     // Always move the code to land on 16b alignment.
-    code_address = generated_code_base_ + generated_code_offset_;
+    code_execute_address =
+        generated_code_execute_base_ + generated_code_offset_;
+    code_execute_address_out = code_execute_address;
+    uint8_t* code_write_address =
+        generated_code_write_base_ + generated_code_offset_;
+    code_write_address_out = code_write_address;
     generated_code_offset_ += xe::round_up(func_info.code_size.total, 16);
 
-    auto tail_address = generated_code_base_ + generated_code_offset_;
+    auto tail_write_address =
+        generated_code_write_base_ + generated_code_offset_;
 
     // Reserve unwind info.
     // We go on the high size of the unwind info as we don't know how big we
     // need it, and a few extra bytes of padding isn't the worst thing.
-    unwind_reservation =
+    unwind_reservation = RequestUnwindReservation(generated_code_write_base_ +
-        RequestUnwindReservation(generated_code_base_ + generated_code_offset_);
+                                                  generated_code_offset_);
     generated_code_offset_ += xe::round_up(unwind_reservation.data_size, 16);
 
-    auto end_address = generated_code_base_ + generated_code_offset_;
+    auto end_write_address =
+        generated_code_write_base_ + generated_code_offset_;
 
     high_mark = generated_code_offset_;
 
     // Store in map. It is maintained in sorted order of host PC dependent on
     // us also being append-only.
     generated_code_map_.emplace_back(
-        (uint64_t(code_address - generated_code_base_) << 32) |
+        (uint64_t(code_execute_address - generated_code_execute_base_) << 32) |
             generated_code_offset_,
         function_info);
 
@@ -186,21 +228,30 @@ void* X64CodeCache::PlaceGuestCode(uint32_t guest_address, void* machine_code,
       if (high_mark <= old_commit_mark) break;
 
       new_commit_mark = old_commit_mark + 16 * 1024 * 1024;
-      xe::memory::AllocFixed(generated_code_base_, new_commit_mark,
+      if (generated_code_execute_base_ == generated_code_write_base_) {
-                             xe::memory::AllocationType::kCommit,
+        xe::memory::AllocFixed(generated_code_execute_base_, new_commit_mark,
-                             xe::memory::PageAccess::kExecuteReadWrite);
+                               xe::memory::AllocationType::kCommit,
+                               xe::memory::PageAccess::kExecuteReadWrite);
+      } else {
+        xe::memory::AllocFixed(generated_code_execute_base_, new_commit_mark,
+                               xe::memory::AllocationType::kCommit,
+                               xe::memory::PageAccess::kExecuteReadOnly);
+        xe::memory::AllocFixed(generated_code_write_base_, new_commit_mark,
+                               xe::memory::AllocationType::kCommit,
+                               xe::memory::PageAccess::kReadWrite);
+      }
     } while (generated_code_commit_mark_.compare_exchange_weak(
         old_commit_mark, new_commit_mark));
 
     // Copy code.
-    std::memcpy(code_address, machine_code, func_info.code_size.total);
+    std::memcpy(code_write_address, machine_code, func_info.code_size.total);
 
     // Fill unused slots with 0xCC
-    std::memset(tail_address, 0xCC,
+    std::memset(tail_write_address, 0xCC,
-                static_cast<size_t>(end_address - tail_address));
+                static_cast<size_t>(end_write_address - tail_write_address));
 
     // Notify subclasses of placed code.
-    PlaceCode(guest_address, machine_code, func_info, code_address,
+    PlaceCode(guest_address, machine_code, func_info, code_execute_address,
               unwind_reservation);
   }
 
@@ -215,7 +266,7 @@ void* X64CodeCache::PlaceGuestCode(uint32_t guest_address, void* machine_code,
 
     iJIT_Method_Load_V2 method = {0};
     method.method_id = iJIT_GetNewMethodID();
-    method.method_load_address = code_address;
+    method.method_load_address = code_execute_address;
     method.method_size = uint32_t(code_size);
     method.method_name = const_cast<char*>(method_name.data());
     method.module_name = function_info
@@ -231,10 +282,9 @@ void* X64CodeCache::PlaceGuestCode(uint32_t guest_address, void* machine_code,
   if (guest_address && indirection_table_base_) {
     uint32_t* indirection_slot = reinterpret_cast<uint32_t*>(
         indirection_table_base_ + (guest_address - kIndirectionTableBase));
-    *indirection_slot = uint32_t(reinterpret_cast<uint64_t>(code_address));
+    *indirection_slot =
+        uint32_t(reinterpret_cast<uint64_t>(code_execute_address));
   }
-
-  return code_address;
 }
 
 uint32_t X64CodeCache::PlaceData(const void* data, size_t length) {
@@ -246,7 +296,7 @@ uint32_t X64CodeCache::PlaceData(const void* data, size_t length) {
 
     // Reserve code.
     // Always move the code to land on 16b alignment.
-    data_address = generated_code_base_ + generated_code_offset_;
+    data_address = generated_code_write_base_ + generated_code_offset_;
     generated_code_offset_ += xe::round_up(length, 16);
 
     high_mark = generated_code_offset_;
@@ -261,9 +311,18 @@ uint32_t X64CodeCache::PlaceData(const void* data, size_t length) {
     if (high_mark <= old_commit_mark) break;
 
     new_commit_mark = old_commit_mark + 16 * 1024 * 1024;
-    xe::memory::AllocFixed(generated_code_base_, new_commit_mark,
+    if (generated_code_execute_base_ == generated_code_write_base_) {
-                           xe::memory::AllocationType::kCommit,
+      xe::memory::AllocFixed(generated_code_execute_base_, new_commit_mark,
-                           xe::memory::PageAccess::kExecuteReadWrite);
+                             xe::memory::AllocationType::kCommit,
+                             xe::memory::PageAccess::kExecuteReadWrite);
+    } else {
+      xe::memory::AllocFixed(generated_code_execute_base_, new_commit_mark,
+                             xe::memory::AllocationType::kCommit,
+                             xe::memory::PageAccess::kExecuteReadOnly);
+      xe::memory::AllocFixed(generated_code_write_base_, new_commit_mark,
+                             xe::memory::AllocationType::kCommit,
+                             xe::memory::PageAccess::kReadWrite);
+    }
   } while (generated_code_commit_mark_.compare_exchange_weak(old_commit_mark,
                                                              new_commit_mark));
 
@@ -274,7 +333,7 @@ uint32_t X64CodeCache::PlaceData(const void* data, size_t length) {
 }
 
 GuestFunction* X64CodeCache::LookupFunction(uint64_t host_pc) {
-  uint32_t key = uint32_t(host_pc - kGeneratedCodeBase);
+  uint32_t key = uint32_t(host_pc - kGeneratedCodeExecuteBase);
   void* fn_entry = std::bsearch(
       &key, generated_code_map_.data(), generated_code_map_.size() + 1,
       sizeof(std::pair<uint32_t, Function*>),

src/xenia/cpu/backend/x64/x64_code_cache.h

@@ -11,6 +11,8 @@
 #define XENIA_CPU_BACKEND_X64_X64_CODE_CACHE_H_
 
 #include <atomic>
+#include <cstddef>
+#include <cstdint>
 #include <memory>
 #include <string>
 #include <utility>
@@ -46,8 +48,10 @@ class X64CodeCache : public CodeCache {
   virtual bool Initialize();
 
   const std::filesystem::path& file_name() const override { return file_name_; }
-  uint32_t base_address() const override { return kGeneratedCodeBase; }
+  uintptr_t execute_base_address() const override {
-  uint32_t total_size() const override { return kGeneratedCodeSize; }
+    return kGeneratedCodeExecuteBase;
+  }
+  size_t total_size() const override { return kGeneratedCodeSize; }
 
   // TODO(benvanik): ELF serialization/etc
   // TODO(benvanik): keep track of code blocks
@@ -59,11 +63,15 @@ class X64CodeCache : public CodeCache {
 
   void CommitExecutableRange(uint32_t guest_low, uint32_t guest_high);
 
-  void* PlaceHostCode(uint32_t guest_address, void* machine_code,
+  void PlaceHostCode(uint32_t guest_address, void* machine_code,
-                      const EmitFunctionInfo& func_info);
+                     const EmitFunctionInfo& func_info,
-  void* PlaceGuestCode(uint32_t guest_address, void* machine_code,
+                     void*& code_execute_address_out,
-                       const EmitFunctionInfo& func_info,
+                     void*& code_write_address_out);
-                       GuestFunction* function_info);
+  void PlaceGuestCode(uint32_t guest_address, void* machine_code,
+                      const EmitFunctionInfo& func_info,
+                      GuestFunction* function_info,
+                      void*& code_execute_address_out,
+                      void*& code_write_address_out);
   uint32_t PlaceData(const void* data, size_t length);
 
   GuestFunction* LookupFunction(uint64_t host_pc) override;
@@ -71,13 +79,16 @@ class X64CodeCache : public CodeCache {
  protected:
   // All executable code falls within 0x80000000 to 0x9FFFFFFF, so we can
   // only map enough for lookups within that range.
-  static const uint64_t kIndirectionTableBase = 0x80000000;
+  static const size_t kIndirectionTableSize = 0x1FFFFFFF;
-  static const uint64_t kIndirectionTableSize = 0x1FFFFFFF;
+  static const uintptr_t kIndirectionTableBase = 0x80000000;
   // The code range is 512MB, but we know the total code games will have is
   // pretty small (dozens of mb at most) and our expansion is reasonablish
   // so 256MB should be more than enough.
-  static const uint64_t kGeneratedCodeBase = 0xA0000000;
+  static const size_t kGeneratedCodeSize = 0x0FFFFFFF;
-  static const uint64_t kGeneratedCodeSize = 0x0FFFFFFF;
+  static const uintptr_t kGeneratedCodeExecuteBase = 0xA0000000;
+  // Used for writing when PageAccess::kExecuteReadWrite is not supported.
+  static const uintptr_t kGeneratedCodeWriteBase =
+      kGeneratedCodeExecuteBase + kGeneratedCodeSize + 1;
 
   // This is picked to be high enough to cover whatever we can reasonably
   // expect. If we hit issues with this it probably means some corner case
@@ -96,11 +107,13 @@ class X64CodeCache : public CodeCache {
     return UnwindReservation();
   }
   virtual void PlaceCode(uint32_t guest_address, void* machine_code,
-                         const EmitFunctionInfo& func_info, void* code_address,
+                         const EmitFunctionInfo& func_info,
+                         void* code_execute_address,
                          UnwindReservation unwind_reservation) {}
 
   std::filesystem::path file_name_;
-  xe::memory::FileMappingHandle mapping_ = nullptr;
+  xe::memory::FileMappingHandle mapping_ =
+      xe::memory::kFileMappingHandleInvalid;
 
   // NOTE: the global critical region must be held when manipulating the offsets
   // or counts of anything, to keep the tables consistent and ordered.
@@ -113,9 +126,13 @@ class X64CodeCache : public CodeCache {
   // the generated code table that correspond to the PPC functions in guest
   // space.
   uint8_t* indirection_table_base_ = nullptr;
-  // Fixed at kGeneratedCodeBase and holding all generated code, growing as
+  // Fixed at kGeneratedCodeExecuteBase and holding all generated code, growing
-  // needed.
+  // as needed.
-  uint8_t* generated_code_base_ = nullptr;
+  uint8_t* generated_code_execute_base_ = nullptr;
+  // View of the memory that backs generated_code_execute_base_ when
+  // PageAccess::kExecuteReadWrite is not supported, for writing the generated
+  // code. Equals to generated_code_execute_base_ when it's supported.
+  uint8_t* generated_code_write_base_ = nullptr;
   // Current offset to empty space in generated code.
   size_t generated_code_offset_ = 0;
   // Current high water mark of COMMITTED code.

src/xenia/cpu/backend/x64/x64_code_cache_posix.cc

@@ -27,7 +27,7 @@ class PosixX64CodeCache : public X64CodeCache {
   /*
   UnwindReservation RequestUnwindReservation(uint8_t* entry_address) override;
   void PlaceCode(uint32_t guest_address, void* machine_code, size_t code_size,
-                 size_t stack_size, void* code_address,
+                 size_t stack_size, void* code_execute_address,
                  UnwindReservation unwind_reservation) override;
 
   void InitializeUnwindEntry(uint8_t* unwind_entry_address,

src/xenia/cpu/backend/x64/x64_code_cache_win.cc

@@ -107,11 +107,12 @@ class Win32X64CodeCache : public X64CodeCache {
  private:
   UnwindReservation RequestUnwindReservation(uint8_t* entry_address) override;
   void PlaceCode(uint32_t guest_address, void* machine_code,
-                 const EmitFunctionInfo& func_info, void* code_address,
+                 const EmitFunctionInfo& func_info, void* code_execute_address,
                  UnwindReservation unwind_reservation) override;
 
   void InitializeUnwindEntry(uint8_t* unwind_entry_address,
-                             size_t unwind_table_slot, void* code_address,
+                             size_t unwind_table_slot,
+                             void* code_execute_address,
                              const EmitFunctionInfo& func_info);
 
   // Growable function table system handle.
@@ -140,9 +141,9 @@ Win32X64CodeCache::~Win32X64CodeCache() {
       delete_growable_table_(unwind_table_handle_);
     }
   } else {
-    if (generated_code_base_) {
+    if (generated_code_execute_base_) {
       RtlDeleteFunctionTable(reinterpret_cast<PRUNTIME_FUNCTION>(
-          reinterpret_cast<DWORD64>(generated_code_base_) | 0x3));
+          reinterpret_cast<DWORD64>(generated_code_execute_base_) | 0x3));
     }
   }
 }
@@ -176,11 +177,12 @@ bool Win32X64CodeCache::Initialize() {
   // Create table and register with the system. It's empty now, but we'll grow
   // it as functions are added.
   if (supports_growable_table_) {
-    if (add_growable_table_(&unwind_table_handle_, unwind_table_.data(),
+    if (add_growable_table_(
-                            unwind_table_count_, DWORD(unwind_table_.size()),
+            &unwind_table_handle_, unwind_table_.data(), unwind_table_count_,
-                            reinterpret_cast<ULONG_PTR>(generated_code_base_),
+            DWORD(unwind_table_.size()),
-                            reinterpret_cast<ULONG_PTR>(generated_code_base_ +
+            reinterpret_cast<ULONG_PTR>(generated_code_execute_base_),
-                                                        kGeneratedCodeSize))) {
+            reinterpret_cast<ULONG_PTR>(generated_code_execute_base_ +
+                                        kGeneratedCodeSize))) {
       XELOGE("Unable to create unwind function table");
       return false;
     }
@@ -188,8 +190,9 @@ bool Win32X64CodeCache::Initialize() {
     // Install a callback that the debugger will use to lookup unwind info on
     // demand.
     if (!RtlInstallFunctionTableCallback(
-            reinterpret_cast<DWORD64>(generated_code_base_) | 0x3,
+            reinterpret_cast<DWORD64>(generated_code_execute_base_) | 0x3,
-            reinterpret_cast<DWORD64>(generated_code_base_), kGeneratedCodeSize,
+            reinterpret_cast<DWORD64>(generated_code_execute_base_),
+            kGeneratedCodeSize,
             [](DWORD64 control_pc, PVOID context) {
               auto code_cache = reinterpret_cast<Win32X64CodeCache*>(context);
               return reinterpret_cast<PRUNTIME_FUNCTION>(
@@ -216,11 +219,12 @@ Win32X64CodeCache::RequestUnwindReservation(uint8_t* entry_address) {
 
 void Win32X64CodeCache::PlaceCode(uint32_t guest_address, void* machine_code,
                                   const EmitFunctionInfo& func_info,
-                                  void* code_address,
+                                  void* code_execute_address,
                                   UnwindReservation unwind_reservation) {
   // Add unwind info.
   InitializeUnwindEntry(unwind_reservation.entry_address,
-                        unwind_reservation.table_slot, code_address, func_info);
+                        unwind_reservation.table_slot, code_execute_address,
+                        func_info);
 
   if (supports_growable_table_) {
     // Notify that the unwind table has grown.
@@ -229,13 +233,15 @@ void Win32X64CodeCache::PlaceCode(uint32_t guest_address, void* machine_code,
   }
 
   // This isn't needed on x64 (probably), but is convention.
-  FlushInstructionCache(GetCurrentProcess(), code_address,
+  // On UWP, FlushInstructionCache available starting from 10.0.16299.0.
+  // https://docs.microsoft.com/en-us/uwp/win32-and-com/win32-apis
+  FlushInstructionCache(GetCurrentProcess(), code_execute_address,
                         func_info.code_size.total);
 }
 
 void Win32X64CodeCache::InitializeUnwindEntry(
-    uint8_t* unwind_entry_address, size_t unwind_table_slot, void* code_address,
+    uint8_t* unwind_entry_address, size_t unwind_table_slot,
-    const EmitFunctionInfo& func_info) {
+    void* code_execute_address, const EmitFunctionInfo& func_info) {
   auto unwind_info = reinterpret_cast<UNWIND_INFO*>(unwind_entry_address);
   UNWIND_CODE* unwind_code = nullptr;
 
@@ -299,10 +305,12 @@ void Win32X64CodeCache::InitializeUnwindEntry(
   // Add entry.
   auto& fn_entry = unwind_table_[unwind_table_slot];
   fn_entry.BeginAddress =
-      (DWORD)(reinterpret_cast<uint8_t*>(code_address) - generated_code_base_);
+      DWORD(reinterpret_cast<uint8_t*>(code_execute_address) -
+            generated_code_execute_base_);
   fn_entry.EndAddress =
-      (DWORD)(fn_entry.BeginAddress + func_info.code_size.total);
+      DWORD(fn_entry.BeginAddress + func_info.code_size.total);
-  fn_entry.UnwindData = (DWORD)(unwind_entry_address - generated_code_base_);
+  fn_entry.UnwindData =
+      DWORD(unwind_entry_address - generated_code_execute_base_);
 }
 
 void* Win32X64CodeCache::LookupUnwindInfo(uint64_t host_pc) {
@@ -310,8 +318,8 @@ void* Win32X64CodeCache::LookupUnwindInfo(uint64_t host_pc) {
       &host_pc, unwind_table_.data(), unwind_table_count_,
       sizeof(RUNTIME_FUNCTION),
       [](const void* key_ptr, const void* element_ptr) {
-        auto key =
+        auto key = *reinterpret_cast<const uintptr_t*>(key_ptr) -
-            *reinterpret_cast<const uintptr_t*>(key_ptr) - kGeneratedCodeBase;
+                   kGeneratedCodeExecuteBase;
         auto element = reinterpret_cast<const RUNTIME_FUNCTION*>(element_ptr);
         if (key < element->BeginAddress) {
           return -1;

src/xenia/cpu/backend/x64/x64_emitter.cc

@@ -125,20 +125,26 @@ void* X64Emitter::Emplace(const EmitFunctionInfo& func_info,
   // top_ points to the Xbyak buffer, and since we are in AutoGrow mode
   // it has pending relocations. We copy the top_ to our buffer, swap the
   // pointer, relocate, then return the original scratch pointer for use.
+  // top_ is used by Xbyak's ready() as both write base pointer and the absolute
+  // address base, which would not work on platforms not supporting writable
+  // executable memory, but Xenia doesn't use absolute label addresses in the
+  // generated code.
   uint8_t* old_address = top_;
-  void* new_address;
+  void* new_execute_address;
+  void* new_write_address;
   assert_true(func_info.code_size.total == size_);
   if (function) {
-    new_address = code_cache_->PlaceGuestCode(function->address(), top_,
+    code_cache_->PlaceGuestCode(function->address(), top_, func_info, function,
-                                              func_info, function);
+                                new_execute_address, new_write_address);
   } else {
-    new_address = code_cache_->PlaceHostCode(0, top_, func_info);
+    code_cache_->PlaceHostCode(0, top_, func_info, new_execute_address,
+                               new_write_address);
   }
-  top_ = reinterpret_cast<uint8_t*>(new_address);
+  top_ = reinterpret_cast<uint8_t*>(new_write_address);
   ready();
   top_ = old_address;
   reset();
-  return new_address;
+  return new_execute_address;
 }
 
 bool X64Emitter::Emit(HIRBuilder* builder, EmitFunctionInfo& func_info) {

src/xenia/cpu/ppc/testing/ppc_testing_native_main.cc

@@ -177,6 +177,9 @@ class TestRunner {
  public:
   TestRunner() {
     memory_size_ = 64 * 1024 * 1024;
+    // FIXME(Triang3l): If this is ever compiled for a platform without
+    // xe::memory::IsWritableExecutableMemorySupported, two memory mappings must
+    // be used.
     memory_ = memory::AllocFixed(nullptr, memory_size_,
                                  memory::AllocationType::kReserveCommit,
                                  memory::PageAccess::kExecuteReadWrite);

src/xenia/cpu/stack_walker_win.cc

@@ -9,6 +9,7 @@
 
 #include "xenia/cpu/stack_walker.h"
 
+#include <cstdint>
 #include <mutex>
 
 #include "xenia/base/logging.h"
@@ -120,8 +121,8 @@ class Win32StackWalker : public StackWalker {
     // They never change, so it's fine even if they are touched from multiple
     // threads.
     code_cache_ = code_cache;
-    code_cache_min_ = code_cache_->base_address();
+    code_cache_min_ = code_cache_->execute_base_address();
-    code_cache_max_ = code_cache_->base_address() + code_cache_->total_size();
+    code_cache_max_ = code_cache_min_ + code_cache_->total_size();
   }
 
   bool Initialize() {
@@ -297,13 +298,13 @@ class Win32StackWalker : public StackWalker {
   std::mutex dbghelp_mutex_;
 
   static xe::cpu::backend::CodeCache* code_cache_;
-  static uint32_t code_cache_min_;
+  static uintptr_t code_cache_min_;
-  static uint32_t code_cache_max_;
+  static uintptr_t code_cache_max_;
 };
 
 xe::cpu::backend::CodeCache* Win32StackWalker::code_cache_ = nullptr;
-uint32_t Win32StackWalker::code_cache_min_ = 0;
+uintptr_t Win32StackWalker::code_cache_min_ = 0;
-uint32_t Win32StackWalker::code_cache_max_ = 0;
+uintptr_t Win32StackWalker::code_cache_max_ = 0;
 
 std::unique_ptr<StackWalker> StackWalker::Create(
     backend::CodeCache* code_cache) {

src/xenia/debug/ui/debug_window.cc

@@ -1454,7 +1454,7 @@ void DebugWindow::UpdateCache() {
   // Fetch module listing.
   // We hold refs so that none are unloaded.
   cache_.modules =
-      object_table->GetObjectsByType<XModule>(XObject::Type::kTypeModule);
+      object_table->GetObjectsByType<XModule>(XObject::Type::Module);
 
   cache_.thread_debug_infos = processor_->QueryThreadDebugInfos();
 

src/xenia/emulator.cc

@@ -394,7 +394,7 @@ void Emulator::Pause() {
   auto lock = global_critical_region::AcquireDirect();
   auto threads =
       kernel_state()->object_table()->GetObjectsByType<kernel::XThread>(
-          kernel::XObject::kTypeThread);
+          kernel::XObject::Type::Thread);
   auto current_thread = kernel::XThread::IsInThread()
                             ? kernel::XThread::GetCurrentThread()
                             : nullptr;
@@ -424,7 +424,7 @@ void Emulator::Resume() {
 
   auto threads =
       kernel_state()->object_table()->GetObjectsByType<kernel::XThread>(
-          kernel::XObject::kTypeThread);
+          kernel::XObject::Type::Thread);
   for (auto thread : threads) {
     if (!thread->can_debugger_suspend()) {
       // Don't pause host threads.
@@ -578,7 +578,7 @@ bool Emulator::ExceptionCallbackThunk(Exception* ex, void* data) {
 bool Emulator::ExceptionCallback(Exception* ex) {
   // Check to see if the exception occurred in guest code.
   auto code_cache = processor()->backend()->code_cache();
-  auto code_base = code_cache->base_address();
+  auto code_base = code_cache->execute_base_address();
   auto code_end = code_base + code_cache->total_size();
 
   if (!processor()->is_debugger_attached() && debugging::IsDebuggerAttached()) {

src/xenia/gpu/command_processor.cc

@@ -1182,9 +1182,11 @@ bool CommandProcessor::ExecutePacketType3_DRAW_INDX(RingBuffer* reader,
   // initiate fetch of index buffer and draw
   // if dword0 != 0, this is a conditional draw based on viz query.
   // This ID matches the one issued in PM4_VIZ_QUERY
-  // ID = dword0 & 0x3F;
-  // use = dword0 & 0x40;
   uint32_t dword0 = reader->ReadAndSwap<uint32_t>();  // viz query info
+  // uint32_t viz_id = dword0 & 0x3F;
+  // when true, render conditionally based on query result
+  // uint32_t viz_use = dword0 & 0x100;
+
   reg::VGT_DRAW_INITIATOR vgt_draw_initiator;
   vgt_draw_initiator.value = reader->ReadAndSwap<uint32_t>();
   WriteRegister(XE_GPU_REG_VGT_DRAW_INITIATOR, vgt_draw_initiator.value);
@@ -1221,6 +1223,14 @@ bool CommandProcessor::ExecutePacketType3_DRAW_INDX(RingBuffer* reader,
     } break;
   }
 
+  auto viz_query = register_file_->Get<reg::PA_SC_VIZ_QUERY>();
+  if (viz_query.viz_query_ena && viz_query.kill_pix_post_hi_z) {
+    // TODO(Triang3l): Don't drop the draw call completely if the vertex shader
+    // has memexport.
+    // TODO(Triang3l || JoelLinn): Handle this properly in the render backends.
+    return true;
+  }
+
   bool success =
       IssueDraw(vgt_draw_initiator.prim_type, vgt_draw_initiator.num_indices,
                 is_indexed ? &index_buffer_info : nullptr,
@@ -1254,6 +1264,14 @@ bool CommandProcessor::ExecutePacketType3_DRAW_INDX_2(RingBuffer* reader,
   // TODO(Triang3l): VGT_IMMED_DATA.
   reader->AdvanceRead((count - 1) * sizeof(uint32_t));
 
+  auto viz_query = register_file_->Get<reg::PA_SC_VIZ_QUERY>();
+  if (viz_query.viz_query_ena && viz_query.kill_pix_post_hi_z) {
+    // TODO(Triang3l): Don't drop the draw call completely if the vertex shader
+    // has memexport.
+    // TODO(Triang3l || JoelLinn): Handle this properly in the render backends.
+    return true;
+  }
+
   bool success = IssueDraw(
       vgt_draw_initiator.prim_type, vgt_draw_initiator.num_indices, nullptr,
       xenos::IsMajorModeExplicit(vgt_draw_initiator.major_mode,
@@ -1450,15 +1468,26 @@ bool CommandProcessor::ExecutePacketType3_VIZ_QUERY(RingBuffer* reader,
   uint32_t dword0 = reader->ReadAndSwap<uint32_t>();
 
   uint32_t id = dword0 & 0x3F;
-  uint32_t end = dword0 & 0x80;
+  uint32_t end = dword0 & 0x100;
   if (!end) {
     // begin a new viz query @ id
+    // On hardware this clears the internal state of the scan converter (which
+    // is different to the register)
     WriteRegister(XE_GPU_REG_VGT_EVENT_INITIATOR, VIZQUERY_START);
     XELOGGPU("Begin viz query ID {:02X}", id);
   } else {
     // end the viz query
     WriteRegister(XE_GPU_REG_VGT_EVENT_INITIATOR, VIZQUERY_END);
     XELOGGPU("End viz query ID {:02X}", id);
+    // The scan converter writes the internal result back to the register here.
+    // We just fake it and say it was visible in case it is read back.
+    if (id < 32) {
+      register_file_->values[XE_GPU_REG_PA_SC_VIZ_QUERY_STATUS_0].u32 |=
+          uint32_t(1) << id;
+    } else {
+      register_file_->values[XE_GPU_REG_PA_SC_VIZ_QUERY_STATUS_1].u32 |=
+          uint32_t(1) << (id - 32);
+    }
   }
 
   return true;

src/xenia/gpu/d3d12/d3d12_command_processor.cc

@@ -1876,14 +1876,14 @@ bool D3D12CommandProcessor::IssueDraw(xenos::PrimitiveType primitive_type,
       !pixel_shader->memexport_stream_constants().empty();
   bool memexport_used = memexport_used_vertex || memexport_used_pixel;
 
-  bool primitive_two_faced =
+  bool primitive_polygonal =
-      xenos::IsPrimitiveTwoFaced(tessellated, primitive_type);
+      xenos::IsPrimitivePolygonal(tessellated, primitive_type);
   auto sq_program_cntl = regs.Get<reg::SQ_PROGRAM_CNTL>();
   auto pa_su_sc_mode_cntl = regs.Get<reg::PA_SU_SC_MODE_CNTL>();
   if (!memexport_used_vertex &&
       (sq_program_cntl.vs_export_mode ==
            xenos::VertexShaderExportMode::kMultipass ||
-       (primitive_two_faced && pa_su_sc_mode_cntl.cull_front &&
+       (primitive_polygonal && pa_su_sc_mode_cntl.cull_front &&
         pa_su_sc_mode_cntl.cull_back))) {
     // All faces are culled - can't be expressed in the pipeline.
     return true;
@@ -1996,15 +1996,45 @@ bool D3D12CommandProcessor::IssueDraw(xenos::PrimitiveType primitive_type,
     current_external_pipeline_ = nullptr;
   }
 
+  // Get dynamic rasterizer state.
+  // Supersampling replacing multisampling due to difficulties of emulating
+  // EDRAM with multisampling with RTV/DSV (with ROV, there's MSAA), and also
+  // resolution scale.
+  uint32_t pixel_size_x, pixel_size_y;
+  if (edram_rov_used_) {
+    pixel_size_x = 1;
+    pixel_size_y = 1;
+  } else {
+    xenos::MsaaSamples msaa_samples =
+        regs.Get<reg::RB_SURFACE_INFO>().msaa_samples;
+    pixel_size_x = msaa_samples >= xenos::MsaaSamples::k4X ? 2 : 1;
+    pixel_size_y = msaa_samples >= xenos::MsaaSamples::k2X ? 2 : 1;
+  }
+  if (texture_cache_->IsResolutionScale2X()) {
+    pixel_size_x *= 2;
+    pixel_size_y *= 2;
+  }
+  draw_util::ViewportInfo viewport_info;
+  draw_util::GetHostViewportInfo(regs, float(pixel_size_x), float(pixel_size_y),
+                                 true, float(D3D12_VIEWPORT_BOUNDS_MAX),
+                                 float(D3D12_VIEWPORT_BOUNDS_MAX), false,
+                                 viewport_info);
+  draw_util::Scissor scissor;
+  draw_util::GetScissor(regs, scissor);
+  scissor.left *= pixel_size_x;
+  scissor.top *= pixel_size_y;
+  scissor.width *= pixel_size_x;
+  scissor.height *= pixel_size_y;
+
   // Update viewport, scissor, blend factor and stencil reference.
-  UpdateFixedFunctionState(primitive_two_faced);
+  UpdateFixedFunctionState(viewport_info, scissor, primitive_polygonal);
 
   // Update system constants before uploading them.
   UpdateSystemConstantValues(
-      memexport_used, primitive_two_faced, line_loop_closing_index,
+      memexport_used, primitive_polygonal, line_loop_closing_index,
       indexed ? index_buffer_info->endianness : xenos::Endian::kNone,
-      used_texture_mask, early_z, GetCurrentColorMask(pixel_shader),
+      viewport_info, pixel_size_x, pixel_size_y, used_texture_mask, early_z,
-      pipeline_render_targets);
+      GetCurrentColorMask(pixel_shader), pipeline_render_targets);
 
   // Update constant buffers, descriptors and root parameters.
   if (!UpdateBindings(vertex_shader, pixel_shader, root_signature)) {
@@ -2753,87 +2783,21 @@ void D3D12CommandProcessor::ClearCommandAllocatorCache() {
   command_allocator_writable_last_ = nullptr;
 }
 
-void D3D12CommandProcessor::UpdateFixedFunctionState(bool primitive_two_faced) {
+void D3D12CommandProcessor::UpdateFixedFunctionState(
+    const draw_util::ViewportInfo& viewport_info,
+    const draw_util::Scissor& scissor, bool primitive_polygonal) {
 #if XE_UI_D3D12_FINE_GRAINED_DRAW_SCOPES
   SCOPE_profile_cpu_f("gpu");
 #endif  // XE_UI_D3D12_FINE_GRAINED_DRAW_SCOPES
 
-  const RegisterFile& regs = *register_file_;
-
-  // Window parameters.
-  // http://ftp.tku.edu.tw/NetBSD/NetBSD-current/xsrc/external/mit/xf86-video-ati/dist/src/r600_reg_auto_r6xx.h
-  // See r200UpdateWindow:
-  // https://github.com/freedreno/mesa/blob/master/src/mesa/drivers/dri/r200/r200_state.c
-  auto pa_sc_window_offset = regs.Get<reg::PA_SC_WINDOW_OFFSET>();
-
-  // Supersampling replacing multisampling due to difficulties of emulating
-  // EDRAM with multisampling with RTV/DSV (with ROV, there's MSAA), and also
-  // resolution scale.
-  uint32_t pixel_size_x, pixel_size_y;
-  if (edram_rov_used_) {
-    pixel_size_x = 1;
-    pixel_size_y = 1;
-  } else {
-    xenos::MsaaSamples msaa_samples =
-        regs.Get<reg::RB_SURFACE_INFO>().msaa_samples;
-    pixel_size_x = msaa_samples >= xenos::MsaaSamples::k4X ? 2 : 1;
-    pixel_size_y = msaa_samples >= xenos::MsaaSamples::k2X ? 2 : 1;
-  }
-  if (texture_cache_->IsResolutionScale2X()) {
-    pixel_size_x *= 2;
-    pixel_size_y *= 2;
-  }
-
   // Viewport.
-  // PA_CL_VTE_CNTL contains whether offsets and scales are enabled.
-  // http://www.x.org/docs/AMD/old/evergreen_3D_registers_v2.pdf
-  // In games, either all are enabled (for regular drawing) or none are (for
-  // rectangle lists usually).
-  //
-  // If scale/offset is enabled, the Xenos shader is writing (neglecting W
-  // division) position in the NDC (-1, -1, dx_clip_space_def - 1) -> (1, 1, 1)
-  // box. If it's not, the position is in screen space. Since we can only use
-  // the NDC in PC APIs, we use a viewport of the largest possible size, and
-  // divide the position by it in translated shaders.
-  auto pa_cl_vte_cntl = regs.Get<reg::PA_CL_VTE_CNTL>();
-  float viewport_scale_x =
-      pa_cl_vte_cntl.vport_x_scale_ena
-          ? std::abs(regs[XE_GPU_REG_PA_CL_VPORT_XSCALE].f32)
-          : 4096.0f;
-  float viewport_scale_y =
-      pa_cl_vte_cntl.vport_y_scale_ena
-          ? std::abs(regs[XE_GPU_REG_PA_CL_VPORT_YSCALE].f32)
-          : 4096.0f;
-  float viewport_scale_z = pa_cl_vte_cntl.vport_z_scale_ena
-                               ? regs[XE_GPU_REG_PA_CL_VPORT_ZSCALE].f32
-                               : 1.0f;
-  float viewport_offset_x = pa_cl_vte_cntl.vport_x_offset_ena
-                                ? regs[XE_GPU_REG_PA_CL_VPORT_XOFFSET].f32
-                                : std::abs(viewport_scale_x);
-  float viewport_offset_y = pa_cl_vte_cntl.vport_y_offset_ena
-                                ? regs[XE_GPU_REG_PA_CL_VPORT_YOFFSET].f32
-                                : std::abs(viewport_scale_y);
-  float viewport_offset_z = pa_cl_vte_cntl.vport_z_offset_ena
-                                ? regs[XE_GPU_REG_PA_CL_VPORT_ZOFFSET].f32
-                                : 0.0f;
-  if (regs.Get<reg::PA_SU_SC_MODE_CNTL>().vtx_window_offset_enable) {
-    viewport_offset_x += float(pa_sc_window_offset.window_x_offset);
-    viewport_offset_y += float(pa_sc_window_offset.window_y_offset);
-  }
   D3D12_VIEWPORT viewport;
-  viewport.TopLeftX =
+  viewport.TopLeftX = viewport_info.left;
-      (viewport_offset_x - viewport_scale_x) * float(pixel_size_x);
+  viewport.TopLeftY = viewport_info.top;
-  viewport.TopLeftY =
+  viewport.Width = viewport_info.width;
-      (viewport_offset_y - viewport_scale_y) * float(pixel_size_y);
+  viewport.Height = viewport_info.height;
-  viewport.Width = viewport_scale_x * 2.0f * float(pixel_size_x);
+  viewport.MinDepth = viewport_info.z_min;
-  viewport.Height = viewport_scale_y * 2.0f * float(pixel_size_y);
+  viewport.MaxDepth = viewport_info.z_max;
-  viewport.MinDepth = viewport_offset_z;
-  viewport.MaxDepth = viewport_offset_z + viewport_scale_z;
-  if (viewport_scale_z < 0.0f) {
-    // MinDepth > MaxDepth doesn't work on Nvidia, emulating it in vertex
-    // shaders and when applying polygon offset.
-    std::swap(viewport.MinDepth, viewport.MaxDepth);
-  }
   ff_viewport_update_needed_ |= ff_viewport_.TopLeftX != viewport.TopLeftX;
   ff_viewport_update_needed_ |= ff_viewport_.TopLeftY != viewport.TopLeftY;
   ff_viewport_update_needed_ |= ff_viewport_.Width != viewport.Width;
@@ -2847,13 +2811,11 @@ void D3D12CommandProcessor::UpdateFixedFunctionState(bool primitive_two_faced) {
   }
 
   // Scissor.
-  draw_util::Scissor scissor;
-  draw_util::GetScissor(regs, scissor);
   D3D12_RECT scissor_rect;
-  scissor_rect.left = LONG(scissor.left * pixel_size_x);
+  scissor_rect.left = LONG(scissor.left);
-  scissor_rect.top = LONG(scissor.top * pixel_size_y);
+  scissor_rect.top = LONG(scissor.top);
-  scissor_rect.right = LONG((scissor.left + scissor.width) * pixel_size_x);
+  scissor_rect.right = LONG(scissor.left + scissor.width);
-  scissor_rect.bottom = LONG((scissor.top + scissor.height) * pixel_size_y);
+  scissor_rect.bottom = LONG(scissor.top + scissor.height);
   ff_scissor_update_needed_ |= ff_scissor_.left != scissor_rect.left;
   ff_scissor_update_needed_ |= ff_scissor_.top != scissor_rect.top;
   ff_scissor_update_needed_ |= ff_scissor_.right != scissor_rect.right;
@@ -2865,6 +2827,8 @@ void D3D12CommandProcessor::UpdateFixedFunctionState(bool primitive_two_faced) {
   }
 
   if (!edram_rov_used_) {
+    const RegisterFile& regs = *register_file_;
+
     // Blend factor.
     ff_blend_factor_update_needed_ |=
         ff_blend_factor_[0] != regs[XE_GPU_REG_RB_BLEND_RED].f32;
@@ -2887,7 +2851,7 @@ void D3D12CommandProcessor::UpdateFixedFunctionState(bool primitive_two_faced) {
     // choose the back face one only if drawing only back faces.
     Register stencil_ref_mask_reg;
     auto pa_su_sc_mode_cntl = regs.Get<reg::PA_SU_SC_MODE_CNTL>();
-    if (primitive_two_faced &&
+    if (primitive_polygonal &&
         regs.Get<reg::RB_DEPTHCONTROL>().backface_enable &&
         pa_su_sc_mode_cntl.cull_front && !pa_su_sc_mode_cntl.cull_back) {
       stencil_ref_mask_reg = XE_GPU_REG_RB_STENCILREFMASK_BF;
@@ -2906,9 +2870,11 @@ void D3D12CommandProcessor::UpdateFixedFunctionState(bool primitive_two_faced) {
 }
 
 void D3D12CommandProcessor::UpdateSystemConstantValues(
-    bool shared_memory_is_uav, bool primitive_two_faced,
+    bool shared_memory_is_uav, bool primitive_polygonal,
     uint32_t line_loop_closing_index, xenos::Endian index_endian,
-    uint32_t used_texture_mask, bool early_z, uint32_t color_mask,
+    const draw_util::ViewportInfo& viewport_info, uint32_t pixel_size_x,
+    uint32_t pixel_size_y, uint32_t used_texture_mask, bool early_z,
+    uint32_t color_mask,
     const RenderTargetCache::PipelineRenderTarget render_targets[4]) {
 #if XE_UI_D3D12_FINE_GRAINED_DRAW_SCOPES
   SCOPE_profile_cpu_f("gpu");
@@ -2920,7 +2886,6 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
   auto pa_su_point_minmax = regs.Get<reg::PA_SU_POINT_MINMAX>();
   auto pa_su_point_size = regs.Get<reg::PA_SU_POINT_SIZE>();
   auto pa_su_sc_mode_cntl = regs.Get<reg::PA_SU_SC_MODE_CNTL>();
-  auto pa_su_vtx_cntl = regs.Get<reg::PA_SU_VTX_CNTL>();
   float rb_alpha_ref = regs[XE_GPU_REG_RB_ALPHA_REF].f32;
   auto rb_colorcontrol = regs.Get<reg::RB_COLORCONTROL>();
   auto rb_depth_info = regs.Get<reg::RB_DEPTH_INFO>();
@@ -2986,11 +2951,6 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
     }
   }
 
-  // Get viewport Z scale - needed for flags and ROV output.
-  float viewport_scale_z = pa_cl_vte_cntl.vport_z_scale_ena
-                               ? regs[XE_GPU_REG_PA_CL_VPORT_ZSCALE].f32
-                               : 1.0f;
-
   bool dirty = false;
 
   // Flags.
@@ -3023,13 +2983,9 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
     flags |= (pa_cl_clip_cntl.value & 0b111111)
              << DxbcShaderTranslator::kSysFlag_UserClipPlane0_Shift;
   }
-  // Reversed depth.
+  // Whether the primitive is polygonal and SV_IsFrontFace matters.
-  if (viewport_scale_z < 0.0f) {
+  if (primitive_polygonal) {
-    flags |= DxbcShaderTranslator::kSysFlag_ReverseZ;
+    flags |= DxbcShaderTranslator::kSysFlag_PrimitivePolygonal;
-  }
-  // Whether SV_IsFrontFace matters.
-  if (primitive_two_faced) {
-    flags |= DxbcShaderTranslator::kSysFlag_PrimitiveTwoFaced;
   }
   // Primitive killing condition.
   if (pa_cl_clip_cntl.vtx_kill_or) {
@@ -3122,81 +3078,24 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
   }
 
   // Conversion to Direct3D 12 normalized device coordinates.
-  // See viewport configuration in UpdateFixedFunctionState for explanations.
-  // X and Y scale/offset is to convert unnormalized coordinates generated by
-  // shaders (for rectangle list drawing, for instance) to the viewport of the
-  // largest possible render target size that is used to emulate unnormalized
-  // coordinates. Z scale/offset is to convert from OpenGL NDC to Direct3D NDC
-  // if needed. Also apply half-pixel offset to reproduce Direct3D 9
-  // rasterization rules - must be done before clipping, not through the
-  // viewport, for SSAA and resolution scale to work correctly.
-  float viewport_scale_x = regs[XE_GPU_REG_PA_CL_VPORT_XSCALE].f32;
-  float viewport_scale_y = regs[XE_GPU_REG_PA_CL_VPORT_YSCALE].f32;
   // Kill all primitives if multipass or both faces are culled, but still need
   // to do memexport.
   if (sq_program_cntl.vs_export_mode ==
           xenos::VertexShaderExportMode::kMultipass ||
-      (primitive_two_faced && pa_su_sc_mode_cntl.cull_front &&
+      (primitive_polygonal && pa_su_sc_mode_cntl.cull_front &&
        pa_su_sc_mode_cntl.cull_back)) {
-    dirty |= !std::isnan(system_constants_.ndc_scale[0]);
-    dirty |= !std::isnan(system_constants_.ndc_scale[1]);
-    dirty |= !std::isnan(system_constants_.ndc_scale[2]);
-    dirty |= !std::isnan(system_constants_.ndc_offset[0]);
-    dirty |= !std::isnan(system_constants_.ndc_offset[1]);
-    dirty |= !std::isnan(system_constants_.ndc_offset[2]);
     float nan_value = std::nanf("");
-    system_constants_.ndc_scale[0] = nan_value;
+    for (uint32_t i = 0; i < 3; ++i) {
-    system_constants_.ndc_scale[1] = nan_value;
+      dirty |= !std::isnan(system_constants_.ndc_scale[i]);
-    system_constants_.ndc_scale[2] = nan_value;
+      system_constants_.ndc_scale[i] = nan_value;
-    system_constants_.ndc_offset[0] = nan_value;
+    }
-    system_constants_.ndc_offset[1] = nan_value;
+  } else {
-    system_constants_.ndc_offset[2] = nan_value;
+    for (uint32_t i = 0; i < 3; ++i) {
-  } else {
+      dirty |= system_constants_.ndc_scale[i] != viewport_info.ndc_scale[i];
-    // When VPORT_Z_SCALE_ENA is disabled, Z/W is directly what is expected to
+      dirty |= system_constants_.ndc_offset[i] != viewport_info.ndc_offset[i];
-    // be written to the depth buffer, and for some reason DX_CLIP_SPACE_DEF
+      system_constants_.ndc_scale[i] = viewport_info.ndc_scale[i];
-    // isn't set in this case in draws in games.
+      system_constants_.ndc_offset[i] = viewport_info.ndc_offset[i];
-    bool gl_clip_space_def =
-        !pa_cl_clip_cntl.dx_clip_space_def && pa_cl_vte_cntl.vport_z_scale_ena;
-    float ndc_scale_x = pa_cl_vte_cntl.vport_x_scale_ena
-                            ? (viewport_scale_x >= 0.0f ? 1.0f : -1.0f)
-                            : (1.0f / 4096.0f);
-    float ndc_scale_y = pa_cl_vte_cntl.vport_y_scale_ena
-                            ? (viewport_scale_y >= 0.0f ? -1.0f : 1.0f)
-                            : (-1.0f / 4096.0f);
-    float ndc_scale_z = gl_clip_space_def ? 0.5f : 1.0f;
-    float ndc_offset_x = pa_cl_vte_cntl.vport_x_offset_ena ? 0.0f : -1.0f;
-    float ndc_offset_y = pa_cl_vte_cntl.vport_y_offset_ena ? 0.0f : 1.0f;
-    float ndc_offset_z = gl_clip_space_def ? 0.5f : 0.0f;
-    if (cvars::half_pixel_offset && !pa_su_vtx_cntl.pix_center) {
-      // Signs are hopefully correct here, tested in GTA IV on both clearing
-      // (without a viewport) and drawing things near the edges of the screen.
-      if (pa_cl_vte_cntl.vport_x_scale_ena) {
-        if (viewport_scale_x != 0.0f) {
-          ndc_offset_x += 0.5f / viewport_scale_x;
-        }
-      } else {
-        ndc_offset_x += 1.0f / xenos::kTexture2DCubeMaxWidthHeight;
-      }
-      if (pa_cl_vte_cntl.vport_y_scale_ena) {
-        if (viewport_scale_y != 0.0f) {
-          ndc_offset_y += 0.5f / viewport_scale_y;
-        }
-      } else {
-        ndc_offset_y -= 1.0f / xenos::kTexture2DCubeMaxWidthHeight;
-      }
     }
-    dirty |= system_constants_.ndc_scale[0] != ndc_scale_x;
-    dirty |= system_constants_.ndc_scale[1] != ndc_scale_y;
-    dirty |= system_constants_.ndc_scale[2] != ndc_scale_z;
-    dirty |= system_constants_.ndc_offset[0] != ndc_offset_x;
-    dirty |= system_constants_.ndc_offset[1] != ndc_offset_y;
-    dirty |= system_constants_.ndc_offset[2] != ndc_offset_z;
-    system_constants_.ndc_scale[0] = ndc_scale_x;
-    system_constants_.ndc_scale[1] = ndc_scale_y;
-    system_constants_.ndc_scale[2] = ndc_scale_z;
-    system_constants_.ndc_offset[0] = ndc_offset_x;
-    system_constants_.ndc_offset[1] = ndc_offset_y;
-    system_constants_.ndc_offset[2] = ndc_offset_z;
   }
 
   // Point size.
@@ -3212,19 +3111,10 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
   system_constants_.point_size[1] = point_size_y;
   system_constants_.point_size_min_max[0] = point_size_min;
   system_constants_.point_size_min_max[1] = point_size_max;
-  float point_screen_to_ndc_x, point_screen_to_ndc_y;
+  float point_screen_to_ndc_x =
-  if (pa_cl_vte_cntl.vport_x_scale_ena) {
+      (0.5f * 2.0f * pixel_size_x) / viewport_info.width;
-    point_screen_to_ndc_x =
+  float point_screen_to_ndc_y =
-        (viewport_scale_x != 0.0f) ? (0.5f / viewport_scale_x) : 0.0f;
+      (0.5f * 2.0f * pixel_size_y) / viewport_info.height;
-  } else {
-    point_screen_to_ndc_x = 1.0f / xenos::kTexture2DCubeMaxWidthHeight;
-  }
-  if (pa_cl_vte_cntl.vport_y_scale_ena) {
-    point_screen_to_ndc_y =
-        (viewport_scale_y != 0.0f) ? (-0.5f / viewport_scale_y) : 0.0f;
-  } else {
-    point_screen_to_ndc_y = -1.0f / xenos::kTexture2DCubeMaxWidthHeight;
-  }
   dirty |= system_constants_.point_screen_to_ndc[0] != point_screen_to_ndc_x;
   dirty |= system_constants_.point_screen_to_ndc[1] != point_screen_to_ndc_y;
   system_constants_.point_screen_to_ndc[0] = point_screen_to_ndc_x;
@@ -3374,27 +3264,18 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
     dirty |= system_constants_.edram_depth_base_dwords != depth_base_dwords;
     system_constants_.edram_depth_base_dwords = depth_base_dwords;
 
-    // The Z range is reversed in the vertex shader if it's reverse - use the
+    float depth_range_scale = viewport_info.z_max - viewport_info.z_min;
-    // absolute value of the scale.
-    float depth_range_scale = std::abs(viewport_scale_z);
     dirty |= system_constants_.edram_depth_range_scale != depth_range_scale;
     system_constants_.edram_depth_range_scale = depth_range_scale;
-    float depth_range_offset = pa_cl_vte_cntl.vport_z_offset_ena
+    dirty |= system_constants_.edram_depth_range_offset != viewport_info.z_min;
-                                   ? regs[XE_GPU_REG_PA_CL_VPORT_ZOFFSET].f32
+    system_constants_.edram_depth_range_offset = viewport_info.z_min;
-                                   : 0.0f;
-    if (viewport_scale_z < 0.0f) {
-      // Similar to MinDepth in fixed-function viewport calculation.
-      depth_range_offset += viewport_scale_z;
-    }
-    dirty |= system_constants_.edram_depth_range_offset != depth_range_offset;
-    system_constants_.edram_depth_range_offset = depth_range_offset;
 
     // For non-polygons, front polygon offset is used, and it's enabled if
     // POLY_OFFSET_PARA_ENABLED is set, for polygons, separate front and back
     // are used.
     float poly_offset_front_scale = 0.0f, poly_offset_front_offset = 0.0f;
     float poly_offset_back_scale = 0.0f, poly_offset_back_offset = 0.0f;
-    if (primitive_two_faced) {
+    if (primitive_polygonal) {
       if (pa_su_sc_mode_cntl.poly_offset_front_enable) {
         poly_offset_front_scale =
             regs[XE_GPU_REG_PA_SU_POLY_OFFSET_FRONT_SCALE].f32;
@@ -3457,7 +3338,7 @@ void D3D12CommandProcessor::UpdateSystemConstantValues(
           system_constants_.edram_stencil_front_func_ops != stencil_func_ops;
       system_constants_.edram_stencil_front_func_ops = stencil_func_ops;
 
-      if (primitive_two_faced && rb_depthcontrol.backface_enable) {
+      if (primitive_polygonal && rb_depthcontrol.backface_enable) {
         dirty |= system_constants_.edram_stencil_back_reference !=
                  rb_stencilrefmask_bf.stencilref;
         system_constants_.edram_stencil_back_reference =

src/xenia/gpu/d3d12/d3d12_command_processor.h

@@ -27,6 +27,7 @@
 #include "xenia/gpu/d3d12/primitive_converter.h"
 #include "xenia/gpu/d3d12/render_target_cache.h"
 #include "xenia/gpu/d3d12/texture_cache.h"
+#include "xenia/gpu/draw_util.h"
 #include "xenia/gpu/dxbc_shader_translator.h"
 #include "xenia/gpu/xenos.h"
 #include "xenia/kernel/kernel_state.h"
@@ -349,11 +350,15 @@ class D3D12CommandProcessor : public CommandProcessor {
       D3D12_CPU_DESCRIPTOR_HANDLE& cpu_handle_out,
       D3D12_GPU_DESCRIPTOR_HANDLE& gpu_handle_out);
 
-  void UpdateFixedFunctionState(bool primitive_two_faced);
+  void UpdateFixedFunctionState(const draw_util::ViewportInfo& viewport_info,
+                                const draw_util::Scissor& scissor,
+                                bool primitive_polygonal);
   void UpdateSystemConstantValues(
-      bool shared_memory_is_uav, bool primitive_two_faced,
+      bool shared_memory_is_uav, bool primitive_polygonal,
       uint32_t line_loop_closing_index, xenos::Endian index_endian,
-      uint32_t used_texture_mask, bool early_z, uint32_t color_mask,
+      const draw_util::ViewportInfo& viewport_info, uint32_t pixel_size_x,
+      uint32_t pixel_size_y, uint32_t used_texture_mask, bool early_z,
+      uint32_t color_mask,
       const RenderTargetCache::PipelineRenderTarget render_targets[4]);
   bool UpdateBindings(const D3D12Shader* vertex_shader,
                       const D3D12Shader* pixel_shader,

src/xenia/gpu/d3d12/pipeline_cache.cc

@@ -1285,7 +1285,7 @@ bool PipelineCache::GetCurrentStateDescription(
         uint32_t(regs.Get<reg::VGT_HOS_CNTL>().tess_mode);
   }
 
-  bool primitive_two_faced = xenos::IsPrimitiveTwoFaced(
+  bool primitive_polygonal = xenos::IsPrimitivePolygonal(
       host_vertex_shader_type != Shader::HostVertexShaderType::kVertex,
       primitive_type);
 
@@ -1305,16 +1305,11 @@ bool PipelineCache::GetCurrentStateDescription(
   // Here we also assume that only one side is culled - if two sides are culled,
   // the D3D12 command processor will drop such draw early.
   bool cull_front, cull_back;
-  if (primitive_two_faced) {
+  float poly_offset = 0.0f, poly_offset_scale = 0.0f;
+  if (primitive_polygonal) {
+    description_out.front_counter_clockwise = pa_su_sc_mode_cntl.face == 0;
     cull_front = pa_su_sc_mode_cntl.cull_front != 0;
     cull_back = pa_su_sc_mode_cntl.cull_back != 0;
-  } else {
-    cull_front = false;
-    cull_back = false;
-  }
-  float poly_offset = 0.0f, poly_offset_scale = 0.0f;
-  if (primitive_two_faced) {
-    description_out.front_counter_clockwise = pa_su_sc_mode_cntl.face == 0;
     if (cull_front) {
       description_out.cull_mode = PipelineCullMode::kFront;
     } else if (cull_back) {
@@ -1354,9 +1349,9 @@ bool PipelineCache::GetCurrentStateDescription(
       description_out.fill_mode_wireframe = 0;
     }
   } else {
-    // Filled front faces only.
+    // Filled front faces only, without culling.
-    // Use front depth bias if POLY_OFFSET_PARA_ENABLED
+    cull_front = false;
-    // (POLY_OFFSET_FRONT_ENABLED is for two-sided primitives).
+    cull_back = false;
     if (!edram_rov_used_ && pa_su_sc_mode_cntl.poly_offset_para_enable) {
       poly_offset = regs[XE_GPU_REG_PA_SU_POLY_OFFSET_FRONT_OFFSET].f32;
       poly_offset_scale = regs[XE_GPU_REG_PA_SU_POLY_OFFSET_FRONT_SCALE].f32;
@@ -1413,7 +1408,7 @@ bool PipelineCache::GetCurrentStateDescription(
       if (rb_depthcontrol.stencil_enable) {
         description_out.stencil_enable = 1;
         bool stencil_backface_enable =
-            primitive_two_faced && rb_depthcontrol.backface_enable;
+            primitive_polygonal && rb_depthcontrol.backface_enable;
         // Per-face masks not supported by Direct3D 12, choose the back face
         // ones only if drawing only back faces.
         Register stencil_ref_mask_reg;

src/xenia/gpu/d3d12/shaders/dxbc/primitive_point_list_gs.h

@@ -1,11 +1,11 @@
 // generated from `xb buildhlsl`
 // source: primitive_point_list.gs.hlsl
 const uint8_t primitive_point_list_gs[] = {
-    0x44, 0x58, 0x42, 0x43, 0x6F, 0x7A, 0xE0, 0xA0, 0x82, 0xF0, 0x8E, 0x77,
+    0x44, 0x58, 0x42, 0x43, 0x16, 0x84, 0x10, 0x1C, 0xE9, 0xAD, 0x76, 0xF9,
-    0x2B, 0x62, 0x44, 0x00, 0xA3, 0x34, 0x47, 0x40, 0x01, 0x00, 0x00, 0x00,
+    0x92, 0xF2, 0xD5, 0x65, 0x7C, 0x8A, 0x5F, 0xC5, 0x01, 0x00, 0x00, 0x00,
-    0x0C, 0x1E, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x34, 0x00, 0x00, 0x00,
+    0x20, 0x1E, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x34, 0x00, 0x00, 0x00,
     0xD0, 0x0A, 0x00, 0x00, 0x28, 0x0D, 0x00, 0x00, 0xAC, 0x0F, 0x00, 0x00,
-    0x70, 0x1D, 0x00, 0x00, 0x52, 0x44, 0x45, 0x46, 0x94, 0x0A, 0x00, 0x00,
+    0x84, 0x1D, 0x00, 0x00, 0x52, 0x44, 0x45, 0x46, 0x94, 0x0A, 0x00, 0x00,
     0x01, 0x00, 0x00, 0x00, 0x78, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
     0x3C, 0x00, 0x00, 0x00, 0x01, 0x05, 0x53, 0x47, 0x00, 0x05, 0x00, 0x00,
     0x6A, 0x0A, 0x00, 0x00, 0x13, 0x13, 0x44, 0x25, 0x3C, 0x00, 0x00, 0x00,
@@ -335,8 +335,8 @@ const uint8_t primitive_point_list_gs[] = {
     0x54, 0x45, 0x58, 0x43, 0x4F, 0x4F, 0x52, 0x44, 0x00, 0x53, 0x56, 0x5F,
     0x50, 0x6F, 0x73, 0x69, 0x74, 0x69, 0x6F, 0x6E, 0x00, 0x53, 0x56, 0x5F,
     0x43, 0x6C, 0x69, 0x70, 0x44, 0x69, 0x73, 0x74, 0x61, 0x6E, 0x63, 0x65,
-    0x00, 0xAB, 0xAB, 0xAB, 0x53, 0x48, 0x45, 0x58, 0xBC, 0x0D, 0x00, 0x00,
+    0x00, 0xAB, 0xAB, 0xAB, 0x53, 0x48, 0x45, 0x58, 0xD0, 0x0D, 0x00, 0x00,
-    0x51, 0x00, 0x02, 0x00, 0x6F, 0x03, 0x00, 0x00, 0x6A, 0x08, 0x00, 0x01,
+    0x51, 0x00, 0x02, 0x00, 0x74, 0x03, 0x00, 0x00, 0x6A, 0x08, 0x00, 0x01,
     0x59, 0x00, 0x00, 0x07, 0x46, 0x8E, 0x30, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x5F, 0x00, 0x00, 0x04, 0xF2, 0x10, 0x20, 0x00,
@@ -369,7 +369,7 @@ const uint8_t primitive_point_list_gs[] = {
     0x13, 0x00, 0x00, 0x00, 0x5F, 0x00, 0x00, 0x04, 0x32, 0x10, 0x20, 0x00,
     0x01, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x5F, 0x00, 0x00, 0x04,
     0x42, 0x10, 0x20, 0x00, 0x01, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00,
-    0x68, 0x00, 0x00, 0x02, 0x02, 0x00, 0x00, 0x00, 0x5D, 0x08, 0x00, 0x01,
+    0x68, 0x00, 0x00, 0x02, 0x03, 0x00, 0x00, 0x00, 0x5D, 0x08, 0x00, 0x01,
     0x8F, 0x00, 0x00, 0x03, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x5C, 0x28, 0x00, 0x01, 0x65, 0x00, 0x00, 0x03, 0xF2, 0x20, 0x10, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x65, 0x00, 0x00, 0x03, 0xF2, 0x20, 0x10, 0x00,
@@ -426,113 +426,13 @@ const uint8_t primitive_point_list_gs[] = {
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
     0x38, 0x00, 0x18, 0x08, 0x32, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x46, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0xF6, 0x1F, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x38, 0x00, 0x78, 0x0A,
+    0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x38, 0x06,
-    0xF2, 0x00, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x46, 0x04, 0x10, 0x00,
+    0x72, 0x00, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x06, 0x01, 0x10, 0x80,
-    0x00, 0x00, 0x00, 0x00, 0x02, 0x40, 0x00, 0x00, 0x00, 0x00, 0x80, 0xBF,
+    0x41, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x40, 0x05,
-    0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x80, 0xBF,
+    0x82, 0x00, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x1A, 0x00, 0x10, 0x00,
-    0x00, 0x00, 0x78, 0x08, 0xF2, 0x00, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x78, 0x08, 0xF2, 0x00, 0x10, 0x00,
-    0x46, 0x0E, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x46, 0x14, 0x20, 0x00,
+    0x02, 0x00, 0x00, 0x00, 0xC6, 0x09, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0x46, 0x14, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0xF2, 0x20, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x02, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x03, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x04, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x05, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x06, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x07, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x08, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x09, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0A, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0A, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0B, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0B, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0C, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0C, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0D, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0D, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0E, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0E, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0F, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0F, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x08,
-    0x32, 0x20, 0x10, 0x00, 0x10, 0x00, 0x00, 0x00, 0x02, 0x40, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0x42, 0x20, 0x10, 0x00,
-    0x10, 0x00, 0x00, 0x00, 0x2A, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x10, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00,
-    0x11, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x11, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x05, 0x32, 0x20, 0x10, 0x00,
-    0x12, 0x00, 0x00, 0x00, 0x46, 0x00, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00,
-    0x36, 0x00, 0x00, 0x06, 0xC2, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0xA6, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x13, 0x00, 0x00, 0x00,
-    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00,
-    0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00, 0x14, 0x00, 0x00, 0x00,
-    0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00,
-    0x75, 0x00, 0x00, 0x03, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x60, 0x08, 0xC2, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x06, 0x04, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x14, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x02, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x03, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x04, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x05, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x06, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x07, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x08, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x09, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0A, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0A, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0B, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0B, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0C, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0C, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0D, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0D, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0E, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0E, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
-    0xF2, 0x20, 0x10, 0x00, 0x0F, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x0F, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x08,
-    0x32, 0x20, 0x10, 0x00, 0x10, 0x00, 0x00, 0x00, 0x02, 0x40, 0x00, 0x00,
-    0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0x42, 0x20, 0x10, 0x00,
-    0x10, 0x00, 0x00, 0x00, 0x2A, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x10, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00,
-    0x11, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x11, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x05, 0x32, 0x20, 0x10, 0x00,
-    0x12, 0x00, 0x00, 0x00, 0xE6, 0x0A, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x36, 0x00, 0x00, 0x06, 0xC2, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0xA6, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x13, 0x00, 0x00, 0x00,
-    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00,
-    0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00, 0x14, 0x00, 0x00, 0x00,
-    0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00,
-    0x75, 0x00, 0x00, 0x03, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x18, 0x09, 0x32, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x46, 0x00, 0x10, 0x80, 0x41, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00,
     0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00,
@@ -573,7 +473,7 @@ const uint8_t primitive_point_list_gs[] = {
     0x32, 0x20, 0x10, 0x00, 0x11, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x05,
     0x32, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00, 0x46, 0x00, 0x10, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xC2, 0x20, 0x10, 0x00,
+    0x02, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xC2, 0x20, 0x10, 0x00,
     0x12, 0x00, 0x00, 0x00, 0xA6, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
     0x13, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
@@ -613,6 +513,56 @@ const uint8_t primitive_point_list_gs[] = {
     0x0E, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
     0x0F, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x0F, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x08, 0x32, 0x20, 0x10, 0x00,
+    0x10, 0x00, 0x00, 0x00, 0x02, 0x40, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0x42, 0x20, 0x10, 0x00, 0x10, 0x00, 0x00, 0x00,
+    0x2A, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00, 0x11, 0x00, 0x00, 0x00,
+    0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x05, 0x32, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00,
+    0xE6, 0x0A, 0x10, 0x00, 0x02, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0xC2, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00, 0xA6, 0x1E, 0x20, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0xF2, 0x20, 0x10, 0x00, 0x13, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0x32, 0x20, 0x10, 0x00, 0x14, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x75, 0x00, 0x00, 0x03,
+    0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x50, 0x08,
+    0xA2, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x04, 0x10, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x06, 0x14, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x01, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x01, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x02, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x02, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x03, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x03, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x04, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x04, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x05, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x05, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x06, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x06, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x07, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x07, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x08, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x08, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x09, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x09, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x0A, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x0A, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x0B, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x0B, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x0C, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x0C, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x0D, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x0D, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x0E, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x0E, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x0F, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x0F, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x08, 0x32, 0x20, 0x10, 0x00,
     0x10, 0x00, 0x00, 0x00, 0x02, 0x40, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3F,
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x36, 0x00, 0x00, 0x06, 0x42, 0x20, 0x10, 0x00, 0x10, 0x00, 0x00, 0x00,
@@ -620,26 +570,78 @@ const uint8_t primitive_point_list_gs[] = {
     0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00, 0x11, 0x00, 0x00, 0x00,
     0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00,
     0x36, 0x00, 0x00, 0x05, 0x32, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0xE6, 0x0A, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0xD6, 0x05, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
     0xC2, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00, 0xA6, 0x1E, 0x20, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
     0xF2, 0x20, 0x10, 0x00, 0x13, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x13, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
     0x32, 0x20, 0x10, 0x00, 0x14, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x75, 0x00, 0x00, 0x03,
-    0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x76, 0x00, 0x00, 0x03,
+    0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x20, 0x05,
-    0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3E, 0x00, 0x00, 0x01,
+    0x42, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2A, 0x00, 0x10, 0x00,
-    0x53, 0x54, 0x41, 0x54, 0x94, 0x00, 0x00, 0x00, 0x74, 0x00, 0x00, 0x00,
+    0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x18, 0x08, 0x32, 0x00, 0x10, 0x00,
-    0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2B, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x86, 0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x0B, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
+    0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00,
-    0x02, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x02, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x03, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x04, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x05, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x06, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x07, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x07, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x08, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x09, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x09, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x0A, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0A, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x0B, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0B, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x0C, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0C, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x0D, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0D, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x0E, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0E, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00, 0x0F, 0x00, 0x00, 0x00,
+    0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0F, 0x00, 0x00, 0x00,
+    0x36, 0x00, 0x00, 0x08, 0x32, 0x20, 0x10, 0x00, 0x10, 0x00, 0x00, 0x00,
+    0x02, 0x40, 0x00, 0x00, 0x00, 0x00, 0x80, 0x3F, 0x00, 0x00, 0x80, 0x3F,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0x42, 0x20, 0x10, 0x00, 0x10, 0x00, 0x00, 0x00, 0x2A, 0x10, 0x20, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06,
+    0x32, 0x20, 0x10, 0x00, 0x11, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x05,
+    0x32, 0x20, 0x10, 0x00, 0x12, 0x00, 0x00, 0x00, 0x46, 0x00, 0x10, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xC2, 0x20, 0x10, 0x00,
+    0x12, 0x00, 0x00, 0x00, 0xA6, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x12, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0xF2, 0x20, 0x10, 0x00,
+    0x13, 0x00, 0x00, 0x00, 0x46, 0x1E, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x13, 0x00, 0x00, 0x00, 0x36, 0x00, 0x00, 0x06, 0x32, 0x20, 0x10, 0x00,
+    0x14, 0x00, 0x00, 0x00, 0x46, 0x10, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x14, 0x00, 0x00, 0x00, 0x75, 0x00, 0x00, 0x03, 0x00, 0x00, 0x11, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x76, 0x00, 0x00, 0x03, 0x00, 0x00, 0x11, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x3E, 0x00, 0x00, 0x01, 0x53, 0x54, 0x41, 0x54,
+    0x94, 0x00, 0x00, 0x00, 0x76, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x2B, 0x00, 0x00, 0x00, 0x0B, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
+    0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00,
-    0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00,
+    0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-    0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x01, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x04, 0x00, 0x00, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
     0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
 };

src/xenia/gpu/d3d12/shaders/dxbc/primitive_point_list_gs.txt

@@ -130,7 +130,7 @@ dcl_input_siv v[1][18].xyzw, position
 dcl_input v[1][19].xyzw
 dcl_input v[1][20].xy
 dcl_input v[1][20].z
-dcl_temps 2
+dcl_temps 3
 dcl_inputprimitive point 
 dcl_stream m0
 dcl_outputtopology trianglestrip 
@@ -170,58 +170,9 @@ max [precise(xy)] r0.xy, r0.xyxx, CB0[0][2].xxxx
 min [precise(xy)] r0.xy, r0.xyxx, CB0[0][2].yyyy
 mul [precise(xy)] r0.xy, r0.xyxx, CB0[0][2].zwzz
 mul [precise(xy)] r0.xy, r0.xyxx, v[0][18].wwww
-mul [precise] r1.xyzw, r0.xyxy, l(-1.000000, 1.000000, 1.000000, -1.000000)
+mov [precise(xyz)] r1.xyz, -r0.xxyx
-add [precise] r1.xyzw, r1.xyzw, v[0][18].xyxy
+mov [precise(w)] r1.w, r0.y
-mov o0.xyzw, v[0][0].xyzw
+add [precise] r2.xyzw, r1.xwyz, v[0][18].xyxy
-mov o1.xyzw, v[0][1].xyzw
-mov o2.xyzw, v[0][2].xyzw
-mov o3.xyzw, v[0][3].xyzw
-mov o4.xyzw, v[0][4].xyzw
-mov o5.xyzw, v[0][5].xyzw
-mov o6.xyzw, v[0][6].xyzw
-mov o7.xyzw, v[0][7].xyzw
-mov o8.xyzw, v[0][8].xyzw
-mov o9.xyzw, v[0][9].xyzw
-mov o10.xyzw, v[0][10].xyzw
-mov o11.xyzw, v[0][11].xyzw
-mov o12.xyzw, v[0][12].xyzw
-mov o13.xyzw, v[0][13].xyzw
-mov o14.xyzw, v[0][14].xyzw
-mov o15.xyzw, v[0][15].xyzw
-mov o16.xy, l(0,1.000000,0,0)
-mov o16.z, v[0][16].z
-mov o17.xy, v[0][17].xyxx
-mov o18.xy, r1.xyxx
-mov o18.zw, v[0][18].zzzw
-mov o19.xyzw, v[0][19].xyzw
-mov o20.xy, v[0][20].xyxx
-emit_stream m0
-add [precise(zw)] r0.zw, r0.xxxy, v[0][18].xxxy
-mov o0.xyzw, v[0][0].xyzw
-mov o1.xyzw, v[0][1].xyzw
-mov o2.xyzw, v[0][2].xyzw
-mov o3.xyzw, v[0][3].xyzw
-mov o4.xyzw, v[0][4].xyzw
-mov o5.xyzw, v[0][5].xyzw
-mov o6.xyzw, v[0][6].xyzw
-mov o7.xyzw, v[0][7].xyzw
-mov o8.xyzw, v[0][8].xyzw
-mov o9.xyzw, v[0][9].xyzw
-mov o10.xyzw, v[0][10].xyzw
-mov o11.xyzw, v[0][11].xyzw
-mov o12.xyzw, v[0][12].xyzw
-mov o13.xyzw, v[0][13].xyzw
-mov o14.xyzw, v[0][14].xyzw
-mov o15.xyzw, v[0][15].xyzw
-mov o16.xy, l(1.000000,1.000000,0,0)
-mov o16.z, v[0][16].z
-mov o17.xy, v[0][17].xyxx
-mov o18.xy, r0.zwzz
-mov o18.zw, v[0][18].zzzw
-mov o19.xyzw, v[0][19].xyzw
-mov o20.xy, v[0][20].xyxx
-emit_stream m0
-add [precise(xy)] r0.xy, -r0.xyxx, v[0][18].xyxx
 mov o0.xyzw, v[0][0].xyzw
 mov o1.xyzw, v[0][1].xyzw
 mov o2.xyzw, v[0][2].xyzw
@@ -241,7 +192,7 @@ mov o15.xyzw, v[0][15].xyzw
 mov o16.xy, l(0,0,0,0)
 mov o16.z, v[0][16].z
 mov o17.xy, v[0][17].xyxx
-mov o18.xy, r0.xyxx
+mov o18.xy, r2.xyxx
 mov o18.zw, v[0][18].zzzw
 mov o19.xyzw, v[0][19].xyzw
 mov o20.xy, v[0][20].xyxx
@@ -262,14 +213,65 @@ mov o12.xyzw, v[0][12].xyzw
 mov o13.xyzw, v[0][13].xyzw
 mov o14.xyzw, v[0][14].xyzw
 mov o15.xyzw, v[0][15].xyzw
+mov o16.xy, l(0,1.000000,0,0)
+mov o16.z, v[0][16].z
+mov o17.xy, v[0][17].xyxx
+mov o18.xy, r2.zwzz
+mov o18.zw, v[0][18].zzzw
+mov o19.xyzw, v[0][19].xyzw
+mov o20.xy, v[0][20].xyxx
+emit_stream m0
+add [precise(yw)] r0.yw, r0.xxxy, v[0][18].xxxy
+mov o0.xyzw, v[0][0].xyzw
+mov o1.xyzw, v[0][1].xyzw
+mov o2.xyzw, v[0][2].xyzw
+mov o3.xyzw, v[0][3].xyzw
+mov o4.xyzw, v[0][4].xyzw
+mov o5.xyzw, v[0][5].xyzw
+mov o6.xyzw, v[0][6].xyzw
+mov o7.xyzw, v[0][7].xyzw
+mov o8.xyzw, v[0][8].xyzw
+mov o9.xyzw, v[0][9].xyzw
+mov o10.xyzw, v[0][10].xyzw
+mov o11.xyzw, v[0][11].xyzw
+mov o12.xyzw, v[0][12].xyzw
+mov o13.xyzw, v[0][13].xyzw
+mov o14.xyzw, v[0][14].xyzw
+mov o15.xyzw, v[0][15].xyzw
 mov o16.xy, l(1.000000,0,0,0)
 mov o16.z, v[0][16].z
 mov o17.xy, v[0][17].xyxx
-mov o18.xy, r1.zwzz
+mov o18.xy, r0.ywyy
+mov o18.zw, v[0][18].zzzw
+mov o19.xyzw, v[0][19].xyzw
+mov o20.xy, v[0][20].xyxx
+emit_stream m0
+mov [precise(z)] r0.z, r1.z
+add [precise(xy)] r0.xy, r0.xzxx, v[0][18].xyxx
+mov o0.xyzw, v[0][0].xyzw
+mov o1.xyzw, v[0][1].xyzw
+mov o2.xyzw, v[0][2].xyzw
+mov o3.xyzw, v[0][3].xyzw
+mov o4.xyzw, v[0][4].xyzw
+mov o5.xyzw, v[0][5].xyzw
+mov o6.xyzw, v[0][6].xyzw
+mov o7.xyzw, v[0][7].xyzw
+mov o8.xyzw, v[0][8].xyzw
+mov o9.xyzw, v[0][9].xyzw
+mov o10.xyzw, v[0][10].xyzw
+mov o11.xyzw, v[0][11].xyzw
+mov o12.xyzw, v[0][12].xyzw
+mov o13.xyzw, v[0][13].xyzw
+mov o14.xyzw, v[0][14].xyzw
+mov o15.xyzw, v[0][15].xyzw
+mov o16.xy, l(1.000000,1.000000,0,0)
+mov o16.z, v[0][16].z
+mov o17.xy, v[0][17].xyxx
+mov o18.xy, r0.xyxx
 mov o18.zw, v[0][18].zzzw
 mov o19.xyzw, v[0][19].xyzw
 mov o20.xy, v[0][20].xyxx
 emit_stream m0
 cut_stream m0
 ret 
-// Approximately 116 instruction slots used
+// Approximately 118 instruction slots used

src/xenia/gpu/d3d12/shaders/primitive_point_list.gs.hlsl

@@ -26,19 +26,22 @@ void main(point XeVertexPreGS xe_in[1],
       clamp(point_size, xe_point_size_min_max.xx, xe_point_size_min_max.yy) *
       xe_point_screen_to_ndc * xe_in[0].post_gs.position.w;
 
-  xe_out.point_params.xy = float2(0.0, 1.0);
-  xe_out.position.xy =
-      xe_in[0].post_gs.position.xy + float2(-1.0, 1.0) * point_size;
-  xe_stream.Append(xe_out);
-  xe_out.point_params.xy = float2(1.0, 1.0);
-  xe_out.position.xy = xe_in[0].post_gs.position.xy + point_size;
-  xe_stream.Append(xe_out);
   xe_out.point_params.xy = float2(0.0, 0.0);
+  // TODO(Triang3l): On Vulkan, sign of Y needs to inverted because of
+  // upper-left origin.
+  // TODO(Triang3l): Investigate the true signs of point sprites.
+  xe_out.position.xy =
+      xe_in[0].post_gs.position.xy + float2(-point_size.x, point_size.y);
+  xe_stream.Append(xe_out);
+  xe_out.point_params.xy = float2(0.0, 1.0);
   xe_out.position.xy = xe_in[0].post_gs.position.xy - point_size;
   xe_stream.Append(xe_out);
   xe_out.point_params.xy = float2(1.0, 0.0);
+  xe_out.position.xy = xe_in[0].post_gs.position.xy + point_size;
+  xe_stream.Append(xe_out);
+  xe_out.point_params.xy = float2(1.0, 1.0);
   xe_out.position.xy =
-      xe_in[0].post_gs.position.xy + float2(1.0, -1.0) * point_size;
+      xe_in[0].post_gs.position.xy + float2(point_size.x, -point_size.y);
   xe_stream.Append(xe_out);
   xe_stream.RestartStrip();
 }

src/xenia/gpu/draw_util.cc

@@ -111,6 +111,179 @@ int32_t FloatToD3D11Fixed16p8(float f32) {
   return result.s;
 }
 
+void GetHostViewportInfo(const RegisterFile& regs, float pixel_size_x,
+                         float pixel_size_y, bool origin_bottom_left,
+                         float x_max, float y_max, bool allow_reverse_z,
+                         ViewportInfo& viewport_info_out) {
+  assert_true(pixel_size_x >= 1.0f);
+  assert_true(pixel_size_y >= 1.0f);
+  assert_true(x_max >= 1.0f);
+  assert_true(y_max >= 1.0f);
+
+  // PA_CL_VTE_CNTL contains whether offsets and scales are enabled.
+  // http://www.x.org/docs/AMD/old/evergreen_3D_registers_v2.pdf
+  // In games, either all are enabled (for regular drawing) or none are (for
+  // rectangle lists usually).
+  //
+  // If scale/offset is enabled, the Xenos shader is writing (neglecting W
+  // division) position in the NDC (-1, -1, dx_clip_space_def - 1) -> (1, 1, 1)
+  // box. If it's not, the position is in screen space. Since we can only use
+  // the NDC in PC APIs, we use a viewport of the largest possible size, and
+  // divide the position by it in translated shaders.
+
+  auto pa_cl_clip_cntl = regs.Get<reg::PA_CL_CLIP_CNTL>();
+  auto pa_cl_vte_cntl = regs.Get<reg::PA_CL_VTE_CNTL>();
+  auto pa_su_sc_mode_cntl = regs.Get<reg::PA_SU_SC_MODE_CNTL>();
+  auto pa_su_vtx_cntl = regs.Get<reg::PA_SU_VTX_CNTL>();
+
+  float viewport_left, viewport_top;
+  float viewport_width, viewport_height;
+  float ndc_scale_x, ndc_scale_y;
+  float ndc_offset_x, ndc_offset_y;
+  // To avoid zero size viewports, which would harm division and aren't allowed
+  // on Vulkan. Nothing will ever be covered by a viewport of this size - this
+  // is 2 orders of magnitude smaller than a .8 subpixel, and thus shouldn't
+  // have any effect on rounding, n and n + 1 / 1024 would be rounded to the
+  // same .8 fixed-point value, thus in fixed-point, the viewport would have
+  // zero size.
+  const float size_min = 1.0f / 1024.0f;
+
+  float viewport_offset_x = pa_cl_vte_cntl.vport_x_offset_ena
+                                ? regs[XE_GPU_REG_PA_CL_VPORT_XOFFSET].f32
+                                : 0.0f;
+  float viewport_offset_y = pa_cl_vte_cntl.vport_y_offset_ena
+                                ? regs[XE_GPU_REG_PA_CL_VPORT_YOFFSET].f32
+                                : 0.0f;
+  if (pa_su_sc_mode_cntl.vtx_window_offset_enable) {
+    auto pa_sc_window_offset = regs.Get<reg::PA_SC_WINDOW_OFFSET>();
+    viewport_offset_x += float(pa_sc_window_offset.window_x_offset);
+    viewport_offset_y += float(pa_sc_window_offset.window_y_offset);
+  }
+
+  if (pa_cl_vte_cntl.vport_x_scale_ena) {
+    float pa_cl_vport_xscale = regs[XE_GPU_REG_PA_CL_VPORT_XSCALE].f32;
+    float viewport_scale_x_abs = std::abs(pa_cl_vport_xscale) * pixel_size_x;
+    viewport_left = viewport_offset_x * pixel_size_x - viewport_scale_x_abs;
+    float viewport_right = viewport_left + viewport_scale_x_abs * 2.0f;
+    // Keep the viewport in the positive quarter-plane for simplicity of
+    // clamping to the maximum supported bounds.
+    float cutoff_left = std::fmax(-viewport_left, 0.0f);
+    float cutoff_right = std::fmax(viewport_right - x_max, 0.0f);
+    viewport_left = std::fmax(viewport_left, 0.0f);
+    viewport_right = std::fmin(viewport_right, x_max);
+    viewport_width = viewport_right - viewport_left;
+    if (viewport_width > size_min) {
+      ndc_scale_x =
+          (viewport_width + cutoff_left + cutoff_right) / viewport_width;
+      if (pa_cl_vport_xscale < 0.0f) {
+        ndc_scale_x = -ndc_scale_x;
+      }
+      ndc_offset_x =
+          ((cutoff_right - cutoff_left) * (0.5f * 2.0f)) / viewport_width;
+    } else {
+      // Empty viewport, but don't pass 0 because that's against the Vulkan
+      // specification.
+      viewport_left = 0.0f;
+      viewport_width = size_min;
+      ndc_scale_x = 0.0f;
+      ndc_offset_x = 0.0f;
+    }
+  } else {
+    // Drawing without a viewport and without clipping to one - use a viewport
+    // covering the entire potential guest render target or the positive part of
+    // the host viewport area, whichever is smaller, and apply the offset, if
+    // enabled, via the shader.
+    viewport_left = 0.0f;
+    viewport_width = std::min(
+        float(xenos::kTexture2DCubeMaxWidthHeight) * pixel_size_x, x_max);
+    ndc_scale_x = (2.0f * pixel_size_x) / viewport_width;
+    ndc_offset_x = viewport_offset_x * ndc_scale_x - 1.0f;
+  }
+
+  if (pa_cl_vte_cntl.vport_y_scale_ena) {
+    float pa_cl_vport_yscale = regs[XE_GPU_REG_PA_CL_VPORT_YSCALE].f32;
+    float viewport_scale_y_abs = std::abs(pa_cl_vport_yscale) * pixel_size_y;
+    viewport_top = viewport_offset_y * pixel_size_y - viewport_scale_y_abs;
+    float viewport_bottom = viewport_top + viewport_scale_y_abs * 2.0f;
+    float cutoff_top = std::fmax(-viewport_top, 0.0f);
+    float cutoff_bottom = std::fmax(viewport_bottom - y_max, 0.0f);
+    viewport_top = std::fmax(viewport_top, 0.0f);
+    viewport_bottom = std::fmin(viewport_bottom, y_max);
+    viewport_height = viewport_bottom - viewport_top;
+    if (viewport_height > size_min) {
+      ndc_scale_y =
+          (viewport_height + cutoff_top + cutoff_bottom) / viewport_height;
+      if (pa_cl_vport_yscale < 0.0f) {
+        ndc_scale_y = -ndc_scale_y;
+      }
+      ndc_offset_y =
+          ((cutoff_bottom - cutoff_top) * (0.5f * 2.0f)) / viewport_height;
+    } else {
+      // Empty viewport, but don't pass 0 because that's against the Vulkan
+      // specification.
+      viewport_top = 0.0f;
+      viewport_height = size_min;
+      ndc_scale_y = 0.0f;
+      ndc_offset_y = 0.0f;
+    }
+  } else {
+    viewport_height = std::min(
+        float(xenos::kTexture2DCubeMaxWidthHeight) * pixel_size_y, y_max);
+    ndc_scale_y = (2.0f * pixel_size_y) / viewport_height;
+    ndc_offset_y = viewport_offset_y * ndc_scale_y - 1.0f;
+  }
+
+  // Apply the vertex half-pixel offset via the shader (it must not affect
+  // clipping, otherwise with SSAA or resolution scale, samples in the left/top
+  // half will never be covered).
+  if (cvars::half_pixel_offset && !pa_su_vtx_cntl.pix_center) {
+    ndc_offset_x += (0.5f * 2.0f * pixel_size_x) / viewport_width;
+    ndc_offset_y += (0.5f * 2.0f * pixel_size_y) / viewport_height;
+  }
+
+  if (origin_bottom_left) {
+    ndc_scale_y = -ndc_scale_y;
+    ndc_offset_y = -ndc_offset_y;
+  }
+
+  float viewport_scale_z = pa_cl_vte_cntl.vport_z_scale_ena
+                               ? regs[XE_GPU_REG_PA_CL_VPORT_ZSCALE].f32
+                               : 1.0f;
+  float viewport_offset_z = pa_cl_vte_cntl.vport_z_offset_ena
+                                ? regs[XE_GPU_REG_PA_CL_VPORT_ZOFFSET].f32
+                                : 0.0f;
+  // Vulkan requires the depth bounds to be in the 0 to 1 range without
+  // VK_EXT_depth_range_unrestricted (which isn't used on the Xbox 360).
+  float viewport_z_min = std::min(std::fmax(viewport_offset_z, 0.0f), 1.0f);
+  float viewport_z_max =
+      std::min(std::fmax(viewport_offset_z + viewport_scale_z, 0.0f), 1.0f);
+  // When VPORT_Z_SCALE_ENA is disabled, Z/W is directly what is expected to be
+  // written to the depth buffer, and for some reason DX_CLIP_SPACE_DEF isn't
+  // set in this case in draws in games.
+  bool gl_clip_space_def =
+      !pa_cl_clip_cntl.dx_clip_space_def && pa_cl_vte_cntl.vport_z_scale_ena;
+  float ndc_scale_z = gl_clip_space_def ? 0.5f : 1.0f;
+  float ndc_offset_z = gl_clip_space_def ? 0.5f : 0.0f;
+  if (viewport_z_min > viewport_z_max && !allow_reverse_z) {
+    std::swap(viewport_z_min, viewport_z_max);
+    ndc_scale_z = -ndc_scale_z;
+    ndc_offset_z = 1.0f - ndc_offset_z;
+  }
+
+  viewport_info_out.left = viewport_left;
+  viewport_info_out.top = viewport_top;
+  viewport_info_out.width = viewport_width;
+  viewport_info_out.height = viewport_height;
+  viewport_info_out.z_min = viewport_z_min;
+  viewport_info_out.z_max = viewport_z_max;
+  viewport_info_out.ndc_scale[0] = ndc_scale_x;
+  viewport_info_out.ndc_scale[1] = ndc_scale_y;
+  viewport_info_out.ndc_scale[2] = ndc_scale_z;
+  viewport_info_out.ndc_offset[0] = ndc_offset_x;
+  viewport_info_out.ndc_offset[1] = ndc_offset_y;
+  viewport_info_out.ndc_offset[2] = ndc_offset_z;
+}
+
 void GetScissor(const RegisterFile& regs, Scissor& scissor_out) {
   // FIXME(Triang3l): Screen scissor isn't applied here, but it seems to be
   // unused on Xbox 360 Direct3D 9.

src/xenia/gpu/draw_util.h

@@ -33,6 +33,28 @@ namespace draw_util {
 // for use with the top-left rasterization rule later.
 int32_t FloatToD3D11Fixed16p8(float f32);
 
+struct ViewportInfo {
+  // The returned viewport will always be in the positive quarter-plane for
+  // simplicity of clamping to the maximum size supported by the host, negative
+  // offset will be applied via ndc_offset.
+  float left;
+  float top;
+  float width;
+  float height;
+  float z_min;
+  float z_max;
+  float ndc_scale[3];
+  float ndc_offset[3];
+};
+// Converts the guest viewport (or fakes one if drawing without a viewport) to
+// a viewport, plus values to multiply-add the returned position by, usable on
+// host graphics APIs such as Direct3D 11+ and Vulkan, also forcing it to the
+// Direct3D clip space with 0...W Z rather than -W...W.
+void GetHostViewportInfo(const RegisterFile& regs, float pixel_size_x,
+                         float pixel_size_y, bool origin_bottom_left,
+                         float x_max, float y_max, bool allow_reverse_z,
+                         ViewportInfo& viewport_info_out);
+
 struct Scissor {
   uint32_t left;
   uint32_t top;

src/xenia/gpu/dxbc_shader_translator.cc

@@ -770,7 +770,7 @@ void DxbcShaderTranslator::StartPixelShader() {
                             uint32_t(CbufferRegister::kSystemConstants),
                             kSysConst_Flags_Vec)
                     .Select(kSysConst_Flags_Comp),
-                DxbcSrc::LU(kSysFlag_PrimitiveTwoFaced));
+                DxbcSrc::LU(kSysFlag_PrimitivePolygonal));
       DxbcOpIf(true, DxbcSrc::R(param_gen_temp, DxbcSrc::kZZZZ));
       {
         // Negate modifier flips the sign bit even for 0 - set it to minus for
@@ -1044,10 +1044,9 @@ void DxbcShaderTranslator::CompleteVertexOrDomainShader() {
     DxbcOpEndIf();
   }
 
-  // Apply scale for drawing without a viewport, and also remap from OpenGL
+  // Apply scale for guest to host viewport and clip space conversion. Also, if
-  // Z clip space to Direct3D if needed. Also, if the vertex shader is
+  // the vertex shader is multipass, the NDC scale constant can be used to set
-  // multipass, the NDC scale constant can be used to set position to NaN to
+  // position to NaN to kill all primitives.
-  // kill all primitives.
   system_constants_used_ |= 1ull << kSysConst_NDCScale_Index;
   DxbcOpMul(DxbcDest::R(system_temp_position_, 0b0111),
             DxbcSrc::R(system_temp_position_),
@@ -1056,16 +1055,7 @@ void DxbcShaderTranslator::CompleteVertexOrDomainShader() {
                         kSysConst_NDCScale_Vec,
                         kSysConst_NDCScale_Comp * 0b010101 + 0b100100));
 
-  // Reverse Z (Z = W - Z) if the viewport depth is inverted.
+  // Apply offset (multiplied by W) used for the same purposes.
-  DxbcOpAnd(temp_x_dest, flags_src, DxbcSrc::LU(kSysFlag_ReverseZ));
-  DxbcOpIf(true, temp_x_src);
-  DxbcOpAdd(DxbcDest::R(system_temp_position_, 0b0100),
-            DxbcSrc::R(system_temp_position_, DxbcSrc::kWWWW),
-            -DxbcSrc::R(system_temp_position_, DxbcSrc::kZZZZ));
-  DxbcOpEndIf();
-
-  // Apply offset (multiplied by W) for drawing without a viewport and for half
-  // pixel offset.
   system_constants_used_ |= 1ull << kSysConst_NDCOffset_Index;
   DxbcOpMAd(DxbcDest::R(system_temp_position_, 0b0111),
             DxbcSrc::CB(cbuffer_index_system_constants_,

src/xenia/gpu/dxbc_shader_translator.h

@@ -123,9 +123,8 @@ class DxbcShaderTranslator : public ShaderTranslator {
     kSysFlag_UserClipPlane3_Shift,
     kSysFlag_UserClipPlane4_Shift,
     kSysFlag_UserClipPlane5_Shift,
-    kSysFlag_ReverseZ_Shift,
     kSysFlag_KillIfAnyVertexKilled_Shift,
-    kSysFlag_PrimitiveTwoFaced_Shift,
+    kSysFlag_PrimitivePolygonal_Shift,
     kSysFlag_AlphaPassIfLess_Shift,
     kSysFlag_AlphaPassIfEqual_Shift,
     kSysFlag_AlphaPassIfGreater_Shift,
@@ -165,9 +164,8 @@ class DxbcShaderTranslator : public ShaderTranslator {
     kSysFlag_UserClipPlane3 = 1u << kSysFlag_UserClipPlane3_Shift,
     kSysFlag_UserClipPlane4 = 1u << kSysFlag_UserClipPlane4_Shift,
     kSysFlag_UserClipPlane5 = 1u << kSysFlag_UserClipPlane5_Shift,
-    kSysFlag_ReverseZ = 1u << kSysFlag_ReverseZ_Shift,
     kSysFlag_KillIfAnyVertexKilled = 1u << kSysFlag_KillIfAnyVertexKilled_Shift,
-    kSysFlag_PrimitiveTwoFaced = 1u << kSysFlag_PrimitiveTwoFaced_Shift,
+    kSysFlag_PrimitivePolygonal = 1u << kSysFlag_PrimitivePolygonal_Shift,
     kSysFlag_AlphaPassIfLess = 1u << kSysFlag_AlphaPassIfLess_Shift,
     kSysFlag_AlphaPassIfEqual = 1u << kSysFlag_AlphaPassIfEqual_Shift,
     kSysFlag_AlphaPassIfGreater = 1u << kSysFlag_AlphaPassIfGreater_Shift,
@@ -220,8 +218,7 @@ class DxbcShaderTranslator : public ShaderTranslator {
     float point_size[2];
 
     float point_size_min_max[2];
-    // Inverse scale of the host viewport (but not supersampled), with signs
+    // Screen point size * 2 (but not supersampled) -> size in NDC.
-    // pre-applied.
     float point_screen_to_ndc[2];
 
     float user_clip_planes[6][4];

src/xenia/gpu/register_table.inc

@@ -15,9 +15,25 @@
 
 //#define XE_GPU_REGISTER(index, type, name)
 
+XE_GPU_REGISTER(0x0048, kDword, BIF_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0049, kDword, BIF_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x004A, kDword, BIF_PERFCOUNTER0_LOW)
+
 XE_GPU_REGISTER(0x01DD, kDword, SCRATCH_ADDR)
 XE_GPU_REGISTER(0x01DC, kDword, SCRATCH_UMSK)
 
+XE_GPU_REGISTER(0x01E6, kDword, CP_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x01E7, kDword, CP_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x01E8, kDword, CP_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x01F5, kDword, CP_PERFMON_CNTL)
+
+XE_GPU_REGISTER(0x0395, kDword, RBBM_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0396, kDword, RBBM_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0397, kDword, RBBM_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0398, kDword, RBBM_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0399, kDword, RBBM_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x039A, kDword, RBBM_PERFCOUNTER1_HI)
+
 XE_GPU_REGISTER(0x045E, kDword, CALLBACK_ACK)
 
 XE_GPU_REGISTER(0x0578, kDword, SCRATCH_REG0)  // interrupt sync
@@ -31,33 +47,226 @@ XE_GPU_REGISTER(0x057F, kDword, SCRATCH_REG7)
 
 XE_GPU_REGISTER(0x05C8, kDword, WAIT_UNTIL)
 
+XE_GPU_REGISTER(0x0815, kDword, MC0_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0816, kDword, MC0_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0817, kDword, MC0_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0855, kDword, MC1_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0856, kDword, MC1_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0857, kDword, MC1_PERFCOUNTER0_LOW)
+
 XE_GPU_REGISTER(0x0A02, kDword, UNKNOWN_0A02)
 XE_GPU_REGISTER(0x0A03, kDword, UNKNOWN_0A03)
 XE_GPU_REGISTER(0x0A04, kDword, UNKNOWN_0A04)
 XE_GPU_REGISTER(0x0A05, kDword, UNKNOWN_0A05)
 
+XE_GPU_REGISTER(0x0A18, kDword, MH_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0A19, kDword, MH_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0A1A, kDword, MH_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0A1B, kDword, MH_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0A1C, kDword, MH_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0A1D, kDword, MH_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0A1E, kDword, MH_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0A1F, kDword, MH_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0A20, kDword, MH_PERFCOUNTER2_LOW)
+
 XE_GPU_REGISTER(0x0A2F, kDword, COHER_SIZE_HOST)
 XE_GPU_REGISTER(0x0A30, kDword, COHER_BASE_HOST)
 XE_GPU_REGISTER(0x0A31, kDword, COHER_STATUS_HOST)
 
+// Status flags of viz queries, doesn't seem to be read back by d3d
+// queries 0x00 to 0x1f (be), bit set when visible
+XE_GPU_REGISTER(0x0C44, kDword, PA_SC_VIZ_QUERY_STATUS_0)
+// queries 0x20 to 0x3f (be)
+XE_GPU_REGISTER(0x0C45, kDword, PA_SC_VIZ_QUERY_STATUS_1)
+
+XE_GPU_REGISTER(0x0C48, kDword, VGT_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0C49, kDword, VGT_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0C4A, kDword, VGT_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0C4B, kDword, VGT_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0C4C, kDword, VGT_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0C4D, kDword, VGT_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0C4E, kDword, VGT_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0C4F, kDword, VGT_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0C50, kDword, VGT_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0C51, kDword, VGT_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0C52, kDword, VGT_PERFCOUNTER3_LOW)
+XE_GPU_REGISTER(0x0C53, kDword, VGT_PERFCOUNTER3_HI)
+
+XE_GPU_REGISTER(0x0C85, kDword, PA_CL_ENHANCE)
+
+XE_GPU_REGISTER(0x0C88, kDword, PA_SU_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0C89, kDword, PA_SU_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0C8A, kDword, PA_SU_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0C8B, kDword, PA_SU_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0C8C, kDword, PA_SU_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0C8D, kDword, PA_SU_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0C8E, kDword, PA_SU_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0C8F, kDword, PA_SU_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0C90, kDword, PA_SU_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0C91, kDword, PA_SU_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0C92, kDword, PA_SU_PERFCOUNTER3_LOW)
+XE_GPU_REGISTER(0x0C93, kDword, PA_SU_PERFCOUNTER3_HI)
+XE_GPU_REGISTER(0x0C98, kDword, PA_SC_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0C99, kDword, PA_SC_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0C9A, kDword, PA_SC_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0C9B, kDword, PA_SC_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0C9C, kDword, PA_SC_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0C9D, kDword, PA_SC_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0C9E, kDword, PA_SC_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0C9F, kDword, PA_SC_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0CA0, kDword, PA_SC_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0CA1, kDword, PA_SC_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0CA2, kDword, PA_SC_PERFCOUNTER3_LOW)
+XE_GPU_REGISTER(0x0CA3, kDword, PA_SC_PERFCOUNTER3_HI)
+
 XE_GPU_REGISTER(0x0D00, kDword, SQ_GPR_MANAGEMENT)
 XE_GPU_REGISTER(0x0D01, kDword, SQ_FLOW_CONTROL)
 XE_GPU_REGISTER(0x0D02, kDword, SQ_INST_STORE_MANAGMENT)
 
 XE_GPU_REGISTER(0x0D04, kDword, SQ_EO_RT)
 
-XE_GPU_REGISTER(0x0C85, kDword, PA_CL_ENHANCE)
+XE_GPU_REGISTER(0x0DC8, kDword, SQ_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0DC9, kDword, SQ_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0DCA, kDword, SQ_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0DCB, kDword, SQ_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0DCC, kDword, SQ_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0DCD, kDword, SQ_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0DCE, kDword, SQ_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0DCF, kDword, SQ_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0DD0, kDword, SQ_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0DD1, kDword, SQ_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0DD2, kDword, SQ_PERFCOUNTER3_LOW)
+XE_GPU_REGISTER(0x0DD3, kDword, SQ_PERFCOUNTER3_HI)
+XE_GPU_REGISTER(0x0DD4, kDword, SX_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0DD8, kDword, SX_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0DD9, kDword, SX_PERFCOUNTER0_HI)
 
 // Set with WAIT_UNTIL = WAIT_3D_IDLECLEAN
 XE_GPU_REGISTER(0x0E00, kDword, UNKNOWN_0E00)
+
+XE_GPU_REGISTER(0x0E05, kDword, TCR_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E06, kDword, TCR_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E07, kDword, TCR_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E08, kDword, TCR_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E09, kDword, TCR_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E0A, kDword, TCR_PERFCOUNTER1_LOW)
+
+XE_GPU_REGISTER(0x0E1F, kDword, TP0_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E20, kDword, TP0_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E21, kDword, TP0_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E22, kDword, TP0_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E23, kDword, TP0_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E24, kDword, TP0_PERFCOUNTER1_LOW)
+
+XE_GPU_REGISTER(0x0E28, kDword, TP1_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E29, kDword, TP1_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E2A, kDword, TP1_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E2B, kDword, TP1_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E2C, kDword, TP1_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E2D, kDword, TP1_PERFCOUNTER1_LOW)
+
+XE_GPU_REGISTER(0x0E31, kDword, TP2_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E32, kDword, TP2_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E33, kDword, TP2_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E34, kDword, TP2_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E35, kDword, TP2_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E36, kDword, TP2_PERFCOUNTER1_LOW)
+
+XE_GPU_REGISTER(0x0E3A, kDword, TP3_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E3B, kDword, TP3_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E3C, kDword, TP3_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E3D, kDword, TP3_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E3E, kDword, TP3_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E3F, kDword, TP3_PERFCOUNTER1_LOW)
+
+// Set with WAIT_UNTIL = WAIT_3D_IDLECLEAN
 XE_GPU_REGISTER(0x0E40, kDword, UNKNOWN_0E40)
 
+// Set during GPU initialization by D3D
 XE_GPU_REGISTER(0x0E42, kDword, UNKNOWN_0E42)
 
+XE_GPU_REGISTER(0x0E48, kDword, VC_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E49, kDword, VC_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E4A, kDword, VC_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E4B, kDword, VC_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E4C, kDword, VC_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E4D, kDword, VC_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0E4E, kDword, VC_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0E4F, kDword, VC_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0E50, kDword, VC_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0E51, kDword, VC_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0E52, kDword, VC_PERFCOUNTER3_HI)
+XE_GPU_REGISTER(0x0E53, kDword, VC_PERFCOUNTER3_LOW)
+
+XE_GPU_REGISTER(0x0E54, kDword, TCM_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E55, kDword, TCM_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E56, kDword, TCM_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E57, kDword, TCM_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E58, kDword, TCM_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E59, kDword, TCM_PERFCOUNTER1_LOW)
+
+XE_GPU_REGISTER(0x0E5A, kDword, TCF_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0E5B, kDword, TCF_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0E5C, kDword, TCF_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0E5D, kDword, TCF_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0E5E, kDword, TCF_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0E5F, kDword, TCF_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0E60, kDword, TCF_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0E61, kDword, TCF_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0E62, kDword, TCF_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0E63, kDword, TCF_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0E64, kDword, TCF_PERFCOUNTER3_HI)
+XE_GPU_REGISTER(0x0E65, kDword, TCF_PERFCOUNTER3_LOW)
+XE_GPU_REGISTER(0x0E66, kDword, TCF_PERFCOUNTER4_SELECT)
+XE_GPU_REGISTER(0x0E67, kDword, TCF_PERFCOUNTER4_HI)
+XE_GPU_REGISTER(0x0E68, kDword, TCF_PERFCOUNTER4_LOW)
+XE_GPU_REGISTER(0x0E69, kDword, TCF_PERFCOUNTER5_SELECT)
+XE_GPU_REGISTER(0x0E6A, kDword, TCF_PERFCOUNTER5_HI)
+XE_GPU_REGISTER(0x0E6B, kDword, TCF_PERFCOUNTER5_LOW)
+XE_GPU_REGISTER(0x0E6C, kDword, TCF_PERFCOUNTER6_SELECT)
+XE_GPU_REGISTER(0x0E6D, kDword, TCF_PERFCOUNTER6_HI)
+XE_GPU_REGISTER(0x0E6E, kDword, TCF_PERFCOUNTER6_LOW)
+XE_GPU_REGISTER(0x0E6F, kDword, TCF_PERFCOUNTER7_SELECT)
+XE_GPU_REGISTER(0x0E70, kDword, TCF_PERFCOUNTER7_HI)
+XE_GPU_REGISTER(0x0E71, kDword, TCF_PERFCOUNTER7_LOW)
+XE_GPU_REGISTER(0x0E72, kDword, TCF_PERFCOUNTER8_SELECT)
+XE_GPU_REGISTER(0x0E73, kDword, TCF_PERFCOUNTER8_HI)
+XE_GPU_REGISTER(0x0E74, kDword, TCF_PERFCOUNTER8_LOW)
+XE_GPU_REGISTER(0x0E75, kDword, TCF_PERFCOUNTER9_SELECT)
+XE_GPU_REGISTER(0x0E76, kDword, TCF_PERFCOUNTER9_HI)
+XE_GPU_REGISTER(0x0E77, kDword, TCF_PERFCOUNTER9_LOW)
+XE_GPU_REGISTER(0x0E78, kDword, TCF_PERFCOUNTER10_SELECT)
+XE_GPU_REGISTER(0x0E79, kDword, TCF_PERFCOUNTER10_HI)
+XE_GPU_REGISTER(0x0E7A, kDword, TCF_PERFCOUNTER10_LOW)
+XE_GPU_REGISTER(0x0E7B, kDword, TCF_PERFCOUNTER11_SELECT)
+XE_GPU_REGISTER(0x0E7C, kDword, TCF_PERFCOUNTER11_HI)
+XE_GPU_REGISTER(0x0E7D, kDword, TCF_PERFCOUNTER11_LOW)
+
 XE_GPU_REGISTER(0x0F01, kDword, RB_BC_CONTROL)
 XE_GPU_REGISTER(0x0F02, kDword, RB_EDRAM_INFO)
 
-// D1*, LUT, and AVIVO registers taken from libxenon and https://www.x.org/docs/AMD/old/RRG-216M56-03oOEM.pdf
+XE_GPU_REGISTER(0x0F04, kDword, BC_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x0F05, kDword, BC_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x0F06, kDword, BC_PERFCOUNTER2_SELECT)
+XE_GPU_REGISTER(0x0F07, kDword, BC_PERFCOUNTER3_SELECT)
+XE_GPU_REGISTER(0x0F08, kDword, BC_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x0F09, kDword, BC_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x0F0A, kDword, BC_PERFCOUNTER1_LOW)
+XE_GPU_REGISTER(0x0F0B, kDword, BC_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x0F0C, kDword, BC_PERFCOUNTER2_LOW)
+XE_GPU_REGISTER(0x0F0D, kDword, BC_PERFCOUNTER2_HI)
+XE_GPU_REGISTER(0x0F0E, kDword, BC_PERFCOUNTER3_LOW)
+XE_GPU_REGISTER(0x0F0F, kDword, BC_PERFCOUNTER3_HI)
+
+XE_GPU_REGISTER(0x1004, kDword, HZ_PERFCOUNTER0_SELECT)
+XE_GPU_REGISTER(0x1005, kDword, HZ_PERFCOUNTER0_HI)
+XE_GPU_REGISTER(0x1006, kDword, HZ_PERFCOUNTER0_LOW)
+XE_GPU_REGISTER(0x1007, kDword, HZ_PERFCOUNTER1_SELECT)
+XE_GPU_REGISTER(0x1008, kDword, HZ_PERFCOUNTER1_HI)
+XE_GPU_REGISTER(0x1009, kDword, HZ_PERFCOUNTER1_LOW)
+
+// D1*, LUT, and AVIVO registers taken from libxenon and
+// https://www.x.org/docs/AMD/old/RRG-216M56-03oOEM.pdf
 XE_GPU_REGISTER(0x1838, kDword, D1MODE_MASTER_UPDATE_LOCK)
 
 XE_GPU_REGISTER(0x1841, kDword, D1GRPH_CONTROL)

src/xenia/gpu/registers.h

@@ -289,12 +289,16 @@ union PA_SC_MPASS_PS_CNTL {
   static constexpr Register register_index = XE_GPU_REG_PA_SC_MPASS_PS_CNTL;
 };
 
-// Scanline converter viz query
+// Scanline converter viz query, used by D3D for gpu side conditional rendering
 union PA_SC_VIZ_QUERY {
   struct {
-    uint32_t viz_query_ena : 1;          // +0
+    // the visibility of draws should be evaluated
-    uint32_t viz_query_id : 6;           // +1
+    uint32_t viz_query_ena : 1;  // +0
-    uint32_t kill_pix_post_early_z : 1;  // +7
+    uint32_t viz_query_id : 6;   // +1
+    // discard geometry after test (but use for testing)
+    uint32_t kill_pix_post_hi_z : 1;  // +7
+    // not used with d3d
+    uint32_t kill_pix_detail_mask : 1;  // +8
   };
   uint32_t value;
   static constexpr Register register_index = XE_GPU_REG_PA_SC_VIZ_QUERY;

src/xenia/gpu/xenos.h

@@ -64,7 +64,15 @@ enum class PrimitiveType : uint32_t {
   kQuadPatch = 0x12,
 };
 
-inline bool IsPrimitiveTwoFaced(bool tessellated, PrimitiveType type) {
+// Polygonal primitive types (not including points and lines) are rasterized as
+// triangles, have front and back faces, and also support face culling and fill
+// modes (polymode_front_ptype, polymode_back_ptype). Other primitive types are
+// always "front" (but don't support front face and back face culling, according
+// to OpenGL and Vulkan specifications - even if glCullFace is
+// GL_FRONT_AND_BACK, points and lines are still drawn), and may in some cases
+// use the "para" registers instead of "front" or "back" (for "parallelogram" -
+// like poly_offset_para_enable).
+constexpr bool IsPrimitivePolygonal(bool tessellated, PrimitiveType type) {
   if (tessellated && (type == PrimitiveType::kTrianglePatch ||
                       type == PrimitiveType::kQuadPatch)) {
     return true;
@@ -81,6 +89,10 @@ inline bool IsPrimitiveTwoFaced(bool tessellated, PrimitiveType type) {
     default:
       break;
   }
+  // TODO(Triang3l): Investigate how kRectangleList should be treated - possibly
+  // actually drawn as two polygons on the console, however, the current
+  // geometry shader doesn't care about the winding order - allowing backface
+  // culling for rectangles currently breaks Gears of War 2.
   return false;
 }
 

src/xenia/kernel/kernel_state.cc

@@ -782,13 +782,13 @@ bool KernelState::Save(ByteStream* stream) {
   for (auto object : objects) {
     auto prev_offset = stream->offset();
 
-    if (object->is_host_object() || object->type() == XObject::kTypeThread) {
+    if (object->is_host_object() || object->type() == XObject::Type::Thread) {
       // Don't save host objects or save XThreads again
       num_objects--;
       continue;
     }
 
-    stream->Write<uint32_t>(object->type());
+    stream->Write<uint32_t>(static_cast<uint32_t>(object->type()));
     if (!object->Save(stream)) {
       XELOGD("Did not save object of type {}", object->type());
       assert_always();
@@ -823,7 +823,7 @@ bool KernelState::Restore(ByteStream* stream) {
   uint32_t num_threads = stream->Read<uint32_t>();
   XELOGD("Loading {} threads...", num_threads);
   for (uint32_t i = 0; i < num_threads; i++) {
-    auto thread = XObject::Restore(this, XObject::kTypeThread, stream);
+    auto thread = XObject::Restore(this, XObject::Type::Thread, stream);
     if (!thread) {
       // Can't continue the restore or we risk misalignment.
       assert_always();

src/xenia/kernel/util/object_table.h

@@ -51,7 +51,7 @@ class ObjectTable {
   object_ref<T> LookupObject(X_HANDLE handle) {
     auto object = LookupObject(handle, false);
     if (object) {
-      assert_true(object->type() == T::kType);
+      assert_true(object->type() == T::kObjectType);
     }
     auto result = object_ref<T>(reinterpret_cast<T*>(object));
     return result;
@@ -72,7 +72,7 @@ class ObjectTable {
   std::vector<object_ref<T>> GetObjectsByType() {
     std::vector<object_ref<T>> results;
     GetObjectsByType(
-        T::kType,
+        T::kObjectType,
         reinterpret_cast<std::vector<object_ref<XObject>>*>(&results));
     return results;
   }

src/xenia/kernel/util/shim_utils.h

@@ -148,7 +148,7 @@ class Param {
 
  protected:
   Param() : ordinal_(-1) {}
-  explicit Param(Init& init) : ordinal_(--init.ordinal) {}
+  explicit Param(Init& init) : ordinal_(init.ordinal++) {}
 
   template <typename V>
   void LoadValue(Init& init, V* out_value) {
@@ -519,10 +519,13 @@ xe::cpu::Export* RegisterExport(R (*fn)(Ps&...), const char* name,
       ++export_entry->function_data.call_count;
       Param::Init init = {
           ppc_context,
-          sizeof...(Ps),
           0,
       };
-      auto params = std::make_tuple<Ps...>(Ps(init)...);
+      // Using braces initializer instead of make_tuple because braces
+      // enforce execution order across compilers.
+      // The make_tuple order is undefined per the C++ standard and
+      // cause inconsitencies between msvc and clang.
+      std::tuple<Ps...> params = {Ps(init)...};
       if (export_entry->tags & xe::cpu::ExportTag::kLog &&
           (!(export_entry->tags & xe::cpu::ExportTag::kHighFrequency) ||
            cvars::log_high_frequency_kernel_calls)) {
@@ -554,9 +557,13 @@ xe::cpu::Export* RegisterExport(void (*fn)(Ps&...), const char* name,
       ++export_entry->function_data.call_count;
       Param::Init init = {
           ppc_context,
-          sizeof...(Ps),
+          0,
       };
-      auto params = std::make_tuple<Ps...>(Ps(init)...);
+      // Using braces initializer instead of make_tuple because braces
+      // enforce execution order across compilers.
+      // The make_tuple order is undefined per the C++ standard and
+      // cause inconsitencies between msvc and clang.
+      std::tuple<Ps...> params = {Ps(init)...};
       if (export_entry->tags & xe::cpu::ExportTag::kLog &&
           (!(export_entry->tags & xe::cpu::ExportTag::kHighFrequency) ||
            cvars::log_high_frequency_kernel_calls)) {

src/xenia/kernel/xam/user_profile.h

@@ -314,6 +314,7 @@ class UserProfile {
   uint64_t xuid() const { return account_.xuid_online; }
   std::string name() const { return to_utf8(account_.GetGamertagString()); }
   uint32_t signin_state() const { return 1; }
+  uint32_t type() const { return 2; /* online profile? */ }
 
   void AddSetting(std::unique_ptr<Setting> setting);
   Setting* GetSetting(uint32_t setting_id);

src/xenia/kernel/xam/xam_content.cc

@@ -39,15 +39,6 @@ struct DeviceInfo {
   char16_t name[28];
 };
 
-typedef struct {
-  xe::be<uint32_t> device_id;
-  xe::be<uint32_t> device_type;
-  xe::be<uint64_t> total_bytes;
-  xe::be<uint64_t> free_bytes;
-  xe::be<uint16_t> name[28];
-} X_CONTENT_DEVICE_DATA;
-//static_assert_size(X_CONTENT_DEVICE_DATA, 0x50);
-
 // TODO(gibbed): real information.
 //
 // Until we expose real information about a HDD device, we
@@ -57,15 +48,12 @@ typedef struct {
 // they incorrectly only look at the lower 32-bits of free_bytes,
 // when it is a 64-bit value. Which means any size above ~4GB
 // will not be recognized properly.
-//
-// Rapala fishing only detected the hdd if the size was
-// raised to 20GB (the smallest xbox hdd)
 #define ONE_GB (1024ull * 1024ull * 1024ull)
 static const DeviceInfo dummy_device_info_ = {
-    0x00000001, // first device seems to be storage
+    0x00000001,      // id
-    1,        // hdd type
+    1,               // 1=HDD
-    20ull * ONE_GB,     // 20GB
+    20ull * ONE_GB,  // 20GB
-    12ull * ONE_GB,     // 12GB, so it looks a little used.
+    3ull * ONE_GB,   // 3GB, so it looks a little used.
     u"Dummy HDD",
 };
 #undef ONE_GB
@@ -132,6 +120,15 @@ dword_result_t XamContentGetDeviceState(dword_t device_id,
 }
 DECLARE_XAM_EXPORT1(XamContentGetDeviceState, kContent, kStub);
 
+typedef struct {
+  xe::be<uint32_t> device_id;
+  xe::be<uint32_t> device_type;
+  xe::be<uint64_t> total_bytes;
+  xe::be<uint64_t> free_bytes;
+  xe::be<uint16_t> name[28];
+} X_CONTENT_DEVICE_DATA;
+static_assert_size(X_CONTENT_DEVICE_DATA, 0x50);
+
 dword_result_t XamContentGetDeviceData(
     dword_t device_id, pointer_t<X_CONTENT_DEVICE_DATA> device_data) {
   if ((device_id & 0x0000000F) != dummy_device_info_.device_id) {
@@ -144,10 +141,10 @@ dword_result_t XamContentGetDeviceData(
 
   device_data.Zero();
   const auto& device_info = dummy_device_info_;
-  xe::store_and_swap(&device_data->device_id, device_info.device_id);
+  device_data->device_id = device_info.device_id;
-  xe::store_and_swap(&device_data->device_type, device_info.device_type);
+  device_data->device_type = device_info.device_type;
-  xe::store(&device_data->total_bytes, device_info.total_bytes);
+  device_data->total_bytes = device_info.total_bytes;
-  xe::store(&device_data->free_bytes, device_info.free_bytes);
+  device_data->free_bytes = device_info.free_bytes;
   xe::store_and_swap<std::u16string>(&device_data->name[0], device_info.name);
   return X_ERROR_SUCCESS;
 }

src/xenia/kernel/xam/xam_user.cc

@@ -102,25 +102,25 @@ DECLARE_XAM_EXPORT1(XamProfileEnumerate, kUserProfiles, kImplemented);
 
 X_HRESULT_result_t XamUserGetXUID(dword_t user_index, dword_t type_mask,
                                   lpqword_t xuid_ptr) {
+  assert_true(type_mask == 1 || type_mask == 2 || type_mask == 3 ||
+              type_mask == 4 || type_mask == 7);
   if (!xuid_ptr) {
     return X_E_INVALIDARG;
   }
-
-  assert_true(type_mask == 1 || type_mask == 2 || type_mask == 3 ||
-              type_mask == 4 || type_mask == 7);
   uint32_t result = X_E_NO_SUCH_USER;
   uint64_t xuid = 0;
   if (user_index < 4) {
     if (user_index == 0) {
       const auto& user_profile = kernel_state()->user_profile();
-      if (type_mask & (2 | 4)) {
+      auto type = user_profile->type() & type_mask;
+      if (type & (2 | 4)) {
+        // maybe online profile?
         xuid = user_profile->xuid();
         result = X_E_SUCCESS;
-      } else if (type_mask & 1) {
+      } else if (type & 1) {
+        // maybe offline profile?
         xuid = user_profile->xuid();
         result = X_E_SUCCESS;
-      } else {
-        result = X_E_INVALIDARG;
       }
     }
   } else {
@@ -178,24 +178,44 @@ DECLARE_XAM_EXPORT1(XamUserGetSigninInfo, kUserProfiles, kImplemented);
 
 dword_result_t XamUserGetName(dword_t user_index, lpstring_t buffer,
                               dword_t buffer_len) {
-  if (user_index) {
+  if (user_index >= 4) {
-    return X_ERROR_NO_SUCH_USER;
+    return X_E_INVALIDARG;
   }
 
-  if (!buffer_len) {
+  if (user_index) {
-    return X_ERROR_SUCCESS;
+    return X_E_NO_SUCH_USER;
   }
 
   const auto& user_profile = kernel_state()->user_profile();
   const auto& user_name = user_profile->name();
-
-  // Real XAM will only copy a maximum of 15 characters out.
   xe::string_util::copy_truncating(buffer, user_name,
-                                   std::min(buffer_len.value(), uint32_t(15)));
+                                   std::min(buffer_len.value(), uint32_t(16)));
-  return X_ERROR_SUCCESS;
+  return X_E_SUCCESS;
 }
 DECLARE_XAM_EXPORT1(XamUserGetName, kUserProfiles, kImplemented);
 
+dword_result_t XamUserGetGamerTag(dword_t user_index, lpu16string_t buffer,
+                                  dword_t buffer_len) {
+  if (user_index >= 4) {
+    return X_E_INVALIDARG;
+  }
+
+  if (user_index) {
+    return X_E_NO_SUCH_USER;
+  }
+
+  if (!buffer || buffer_len < 16) {
+    return X_E_INVALIDARG;
+  }
+
+  const auto& user_profile = kernel_state()->user_profile();
+  auto user_name = xe::to_utf16(user_profile->name());
+  xe::string_util::copy_and_swap_truncating(
+      buffer, user_name, std::min(buffer_len.value(), uint32_t(16)));
+  return X_E_SUCCESS;
+}
+DECLARE_XAM_EXPORT1(XamUserGetGamerTag, kUserProfiles, kImplemented);
+
 typedef struct {
   xe::be<uint32_t> setting_count;
   xe::be<uint32_t> settings_ptr;
@@ -481,8 +501,15 @@ DECLARE_XAM_EXPORT1(XamUserWriteProfileSettings, kUserProfiles, kImplemented);
 
 dword_result_t XamUserCheckPrivilege(dword_t user_index, dword_t mask,
                                      lpdword_t out_value) {
-  if (user_index) {
+  // checking all users?
-    return X_ERROR_NO_SUCH_USER;
+  if (user_index != 0xFF) {
+    if (user_index >= 4) {
+      return X_ERROR_INVALID_PARAMETER;
+    }
+
+    if (user_index) {
+      return X_ERROR_NO_SUCH_USER;
+    }
   }
 
   // If we deny everything, games should hopefully not try to do stuff.
@@ -539,6 +566,17 @@ DECLARE_XAM_EXPORT1(XamUserContentRestrictionCheckAccess, kUserProfiles, kStub);
 dword_result_t XamUserIsOnlineEnabled(dword_t user_index) { return 1; }
 DECLARE_XAM_EXPORT1(XamUserIsOnlineEnabled, kUserProfiles, kStub);
 
+dword_result_t XamUserGetMembershipTier(dword_t user_index) {
+  if (user_index >= 4) {
+    return X_ERROR_INVALID_PARAMETER;
+  }
+  if (user_index) {
+    return X_ERROR_NO_SUCH_USER;
+  }
+  return 6 /* 6 appears to be Gold */;
+}
+DECLARE_XAM_EXPORT1(XamUserGetMembershipTier, kUserProfiles, kStub);
+
 dword_result_t XamUserAreUsersFriends(dword_t user_index, dword_t unk1,
                                       dword_t unk2, lpdword_t out_value,
                                       dword_t overlapped_ptr) {

src/xenia/kernel/xboxkrnl/xboxkrnl_debug.cc

@@ -47,8 +47,15 @@ void HandleSetThreadName(pointer_t<X_EXCEPTION_RECORD> record) {
     return;
   }
 
-  auto name =
+  // Shadowrun (and its demo) has a bug where it ends up passing freed memory
-      kernel_memory()->TranslateVirtual<const char*>(thread_info->name_ptr);
+  // for the name, so at the point of SetThreadName it's filled with junk.
+
+  // TODO(gibbed): cvar for thread name encoding for conversion, some games use
+  // SJIS and there's no way to automatically know this.
+  auto name = std::string(
+      kernel_memory()->TranslateVirtual<const char*>(thread_info->name_ptr));
+  std::replace_if(
+      name.begin(), name.end(), [](auto c) { return c < 32 || c > 127; }, '?');
 
   object_ref<XThread> thread;
   if (thread_info->thread_id == -1) {

src/xenia/kernel/xboxkrnl/xboxkrnl_io.cc

@@ -1,4 +1,4 @@
-/**
+/**
  ******************************************************************************
  * Xenia : Xbox 360 Emulator Research Project                                 *
  ******************************************************************************
@@ -29,10 +29,23 @@ namespace xboxkrnl {
 
 static bool IsValidPath(const std::string_view s, bool is_pattern) {
   // TODO(gibbed): validate path components individually
+  bool got_asterisk = false;
   for (const auto& c : s) {
     if (c <= 31 || c >= 127) {
       return false;
     }
+    if (got_asterisk) {
+      // * must be followed by a . (*.)
+      //
+      // Viva Piñata: Party Animals (4D530819) has a bug in its game code where
+      // it attempts to FindFirstFile() with filters of "Game:\\*_X3.rkv",
+      // "Game:\\m*_X3.rkv", and "Game:\\w*_X3.rkv" and will infinite loop if
+      // the path filter is allowed.
+      if (c != '.') {
+        return false;
+      }
+      got_asterisk = false;
+    }
     switch (c) {
       case '"':
       // case '*':
@@ -47,12 +60,20 @@ static bool IsValidPath(const std::string_view s, bool is_pattern) {
       case '|': {
         return false;
       }
-      case '*':
+      case '*': {
+        // Pattern-specific (for NtQueryDirectoryFile)
+        if (!is_pattern) {
+          return false;
+        }
+        got_asterisk = true;
+        break;
+      }
       case '?': {
         // Pattern-specific (for NtQueryDirectoryFile)
         if (!is_pattern) {
           return false;
         }
+        break;
       }
       default: {
         break;
@@ -98,7 +119,7 @@ dword_result_t NtCreateFile(lpdword_t handle_out, dword_t desired_access,
     auto root_file = kernel_state()->object_table()->LookupObject<XFile>(
         object_attrs->root_directory);
     assert_not_null(root_file);
-    assert_true(root_file->type() == XObject::Type::kTypeFile);
+    assert_true(root_file->type() == XObject::Type::File);
 
     root_entry = root_file->entry();
   }
@@ -368,7 +389,7 @@ dword_result_t NtQueryFullAttributesFile(
     root_file = kernel_state()->object_table()->LookupObject<XFile>(
         obj_attribs->root_directory);
     assert_not_null(root_file);
-    assert_true(root_file->type() == XObject::Type::kTypeFile);
+    assert_true(root_file->type() == XObject::Type::File);
     assert_always();
   }
 
@@ -415,7 +436,7 @@ dword_result_t NtQueryDirectoryFile(
 
   // Enforce that the path is ASCII.
   if (!IsValidPath(name, true)) {
-    return X_STATUS_OBJECT_NAME_INVALID;
+    return X_STATUS_INVALID_PARAMETER;
   }
 
   if (file) {

src/xenia/kernel/xboxkrnl/xboxkrnl_memory.cc

@@ -418,6 +418,11 @@ DECLARE_XBOXKRNL_EXPORT2(MmQueryAddressProtect, kMemory, kImplemented,
 
 void MmSetAddressProtect(lpvoid_t base_address, dword_t region_size,
                          dword_t protect_bits) {
+  if (!protect_bits) {
+    XELOGE("MmSetAddressProtect: Failed due to incorrect protect_bits");
+    return;
+  }
+
   uint32_t protect = FromXdkProtectFlags(protect_bits);
   auto heap = kernel_memory()->LookupHeap(base_address);
   heap->Protect(base_address.guest_address(), region_size, protect);

src/xenia/kernel/xboxkrnl/xboxkrnl_module.cc

@@ -157,12 +157,15 @@ XboxkrnlModule::XboxkrnlModule(Emulator* emulator, KernelState* kernel_state)
   // disassembly.
   // aomega08 says the value is 0x02000817, bit 27: debug mode on.
   // When that is set, though, allocs crash in weird ways.
+  //
+  // From kernel dissasembly, after storage is initialized
+  // XboxHardwareInfo flags is set with flag 5 (0x20).
   uint32_t pXboxHardwareInfo = memory_->SystemHeapAlloc(16);
   auto lpXboxHardwareInfo = memory_->TranslateVirtual(pXboxHardwareInfo);
   export_resolver_->SetVariableMapping(
       "xboxkrnl.exe", ordinals::XboxHardwareInfo, pXboxHardwareInfo);
-  xe::store_and_swap<uint32_t>(lpXboxHardwareInfo + 0, 0x20);    // flags //test
+  xe::store_and_swap<uint32_t>(lpXboxHardwareInfo + 0, 0x20);  // flags
-  xe::store_and_swap<uint8_t>(lpXboxHardwareInfo + 4, 0x06);  // cpu count
+  xe::store_and_swap<uint8_t>(lpXboxHardwareInfo + 4, 0x06);   // cpu count
   // Remaining 11b are zeroes?
 
   // ExConsoleGameRegion, probably same values as keyvault region uses?

src/xenia/kernel/xboxkrnl/xboxkrnl_ob.cc

@@ -78,22 +78,21 @@ DECLARE_XBOXKRNL_EXPORT1(ObLookupThreadByThreadId, kNone, kImplemented);
 dword_result_t ObReferenceObjectByHandle(dword_t handle,
                                          dword_t object_type_ptr,
                                          lpdword_t out_object_ptr) {
-  const static std::unordered_map<XObject::Type, uint32_t> obj_type_match = {
+  // These values come from how Xenia handles uninitialized kernel data exports.
-      {XObject::kTypeEvent, 0xD00EBEEF},
+  // D###BEEF where ### is the ordinal.
-      {XObject::kTypeSemaphore, 0xD017BEEF},
+  const static std::unordered_map<XObject::Type, uint32_t> object_types = {
-      {XObject::kTypeThread, 0xD01BBEEF}};
+      {XObject::Type::Event, 0xD00EBEEF},
-
+      {XObject::Type::Semaphore, 0xD017BEEF},
+      {XObject::Type::Thread, 0xD01BBEEF}};
   auto object = kernel_state()->object_table()->LookupObject<XObject>(handle);
-
   if (!object) {
     return X_STATUS_INVALID_HANDLE;
   }
 
   uint32_t native_ptr = object->guest_object();
-  auto obj_type = obj_type_match.find(object->type());
+  auto object_type = object_types.find(object->type());
-
+  if (object_type != object_types.end()) {
-  if (obj_type != obj_type_match.end()) {
+    if (object_type_ptr && object_type_ptr != object_type->second) {
-    if (object_type_ptr && object_type_ptr != obj_type->second) {
       return X_STATUS_OBJECT_TYPE_MISMATCH;
     }
   } else {

src/xenia/kernel/xboxkrnl/xboxkrnl_rtl.cc

@@ -21,16 +21,10 @@
 #include "xenia/kernel/util/shim_utils.h"
 #include "xenia/kernel/xboxkrnl/xboxkrnl_private.h"
 #include "xenia/kernel/xboxkrnl/xboxkrnl_threading.h"
+#include "xenia/kernel/xclock.h"
 #include "xenia/kernel/xevent.h"
 #include "xenia/kernel/xthread.h"
 
-#if XE_PLATFORM_WIN32
-#include "xenia/base/platform_win.h"
-#define timegm _mkgmtime
-#endif
-
-static constexpr uint64_t MAX_TIME64_T = 0x793406FFF;
-
 namespace xe {
 namespace kernel {
 namespace xboxkrnl {
@@ -509,52 +503,51 @@ struct X_TIME_FIELDS {
   xe::be<uint16_t> milliseconds;
   xe::be<uint16_t> weekday;
 };
-static_assert(sizeof(X_TIME_FIELDS) == 16, "Must be LARGEINTEGER");
+static_assert_size(X_TIME_FIELDS, 16);
 
-// https://support.microsoft.com/en-us/kb/167296
+// https://docs.microsoft.com/en-us/windows-hardware/drivers/ddi/wdm/nf-wdm-rtltimetotimefields
 void RtlTimeToTimeFields(lpqword_t time_ptr,
                          pointer_t<X_TIME_FIELDS> time_fields_ptr) {
-  int64_t time_ms = time_ptr.value() / 10000 - 11644473600000LL;
+  auto tp = XClock::to_sys(XClock::from_file_time(time_ptr.value()));
-  time_t timet = time_ms / 1000;
+  auto dp = date::floor<date::days>(tp);
-
+  auto year_month_day = date::year_month_day{dp};
-  // https://docs.microsoft.com/en-us/cpp/c-runtime-library/reference/localtime-s-localtime32-s-localtime64-s?view=vs-2019
+  auto weekday = date::weekday{dp};
-  // __time64_t structure
+  auto time = date::hh_mm_ss{date::floor<std::chrono::milliseconds>(tp - dp)};
-  // allows dates to be expressed up through 23:59:59, January 18, 3001
+  time_fields_ptr->year = static_cast<int>(year_month_day.year());
-  if (uint64_t(timet) > MAX_TIME64_T) {
+  time_fields_ptr->month = static_cast<unsigned>(year_month_day.month());
-    return;
+  time_fields_ptr->day = static_cast<unsigned>(year_month_day.day());
-  }
+  time_fields_ptr->weekday = weekday.c_encoding();
-
+  time_fields_ptr->hour = time.hours().count();
-  struct tm* tm = gmtime(&timet);
+  time_fields_ptr->minute = time.minutes().count();
-
+  time_fields_ptr->second = static_cast<uint16_t>(time.seconds().count());
-  time_fields_ptr->year = tm->tm_year + 1900;
+  time_fields_ptr->milliseconds =
-  time_fields_ptr->month = tm->tm_mon + 1;
+      static_cast<uint16_t>(time.subseconds().count());
-  time_fields_ptr->day = tm->tm_mday;
-  time_fields_ptr->hour = tm->tm_hour;
-  time_fields_ptr->minute = tm->tm_min;
-  time_fields_ptr->second = tm->tm_sec;
-  time_fields_ptr->milliseconds = time_ms % 1000;
-  time_fields_ptr->weekday = tm->tm_wday;
 }
 DECLARE_XBOXKRNL_EXPORT1(RtlTimeToTimeFields, kNone, kImplemented);
 
+// https://docs.microsoft.com/en-us/windows-hardware/drivers/ddi/wdm/nf-wdm-rtltimefieldstotime
 dword_result_t RtlTimeFieldsToTime(pointer_t<X_TIME_FIELDS> time_fields_ptr,
                                    lpqword_t time_ptr) {
-  struct tm tm;
+  if (time_fields_ptr->year < 1601 || time_fields_ptr->month < 1 ||
-  tm.tm_year = time_fields_ptr->year - 1900;
+      time_fields_ptr->month > 11 || time_fields_ptr->day < 1 ||
-  tm.tm_mon = time_fields_ptr->month - 1;
+      time_fields_ptr->hour > 23 || time_fields_ptr->minute > 59 ||
-  tm.tm_mday = time_fields_ptr->day;
+      time_fields_ptr->second > 59 || time_fields_ptr->milliseconds > 999) {
-  tm.tm_hour = time_fields_ptr->hour;
-  tm.tm_min = time_fields_ptr->minute;
-  tm.tm_sec = time_fields_ptr->second;
-  tm.tm_isdst = 0;
-  time_t timet = timegm(&tm);
-  if (timet == -1) {
-    // set last error = ERROR_INVALID_PARAMETER
     return 0;
   }
-  uint64_t time =
+  auto year = date::year{time_fields_ptr->year};
-      ((timet + 11644473600LL) * 1000 + time_fields_ptr->milliseconds) * 10000;
+  auto month = date::month{time_fields_ptr->month};
-  *time_ptr = time;
+  auto day = date::day{time_fields_ptr->day};
+  auto year_month_day = date::year_month_day{year, month, day};
+  if (!year_month_day.ok()) {
+    return 0;
+  }
+  auto dp = static_cast<date::sys_days>(year_month_day);
+  std::chrono::system_clock::time_point time = dp;
+  time += std::chrono::hours{time_fields_ptr->hour};
+  time += std::chrono::minutes{time_fields_ptr->minute};
+  time += std::chrono::seconds{time_fields_ptr->second};
+  time += std::chrono::milliseconds{time_fields_ptr->milliseconds};
+  *time_ptr = XClock::to_file_time(XClock::from_sys(time));
   return 1;
 }
 DECLARE_XBOXKRNL_EXPORT1(RtlTimeFieldsToTime, kNone, kImplemented);

src/xenia/kernel/xboxkrnl/xboxkrnl_threading.cc

@@ -434,7 +434,7 @@ dword_result_t NtCreateEvent(lpdword_t handle_ptr,
   auto existing_object =
       LookupNamedObject<XEvent>(kernel_state(), obj_attributes_ptr);
   if (existing_object) {
-    if (existing_object->type() == XObject::kTypeEvent) {
+    if (existing_object->type() == XObject::Type::Event) {
       if (handle_ptr) {
         existing_object->RetainHandle();
         *handle_ptr = existing_object->handle();
@@ -561,7 +561,7 @@ dword_result_t NtCreateSemaphore(lpdword_t handle_ptr,
   auto existing_object =
       LookupNamedObject<XSemaphore>(kernel_state(), obj_attributes_ptr);
   if (existing_object) {
-    if (existing_object->type() == XObject::kTypeSemaphore) {
+    if (existing_object->type() == XObject::Type::Semaphore) {
       if (handle_ptr) {
         existing_object->RetainHandle();
         *handle_ptr = existing_object->handle();
@@ -615,7 +615,7 @@ dword_result_t NtCreateMutant(lpdword_t handle_out,
   auto existing_object = LookupNamedObject<XMutant>(
       kernel_state(), obj_attributes.guest_address());
   if (existing_object) {
-    if (existing_object->type() == XObject::kTypeMutant) {
+    if (existing_object->type() == XObject::Type::Mutant) {
       if (handle_out) {
         existing_object->RetainHandle();
         *handle_out = existing_object->handle();
@@ -676,7 +676,7 @@ dword_result_t NtCreateTimer(lpdword_t handle_ptr, lpvoid_t obj_attributes_ptr,
   auto existing_object =
       LookupNamedObject<XTimer>(kernel_state(), obj_attributes_ptr);
   if (existing_object) {
-    if (existing_object->type() == XObject::kTypeTimer) {
+    if (existing_object->type() == XObject::Type::Timer) {
       if (handle_ptr) {
         existing_object->RetainHandle();
         *handle_ptr = existing_object->handle();

src/xenia/kernel/xclock.h

@@ -0,0 +1,78 @@
+/**
+ ******************************************************************************
+ * Xenia : Xbox 360 Emulator Research Project                                 *
+ ******************************************************************************
+ * Copyright 2020 Ben Vanik. All rights reserved.                             *
+ * Released under the BSD license - see LICENSE in the root for more details. *
+ ******************************************************************************
+ */
+
+#ifndef XENIA_KERNEL_XCLOCK_H_
+#define XENIA_KERNEL_XCLOCK_H_
+
+#include <chrono>
+
+#include "xenia/base/clock.h"
+
+#include "third_party/date/include/date/date.h"
+
+namespace xe {
+namespace kernel {
+
+struct XClock {
+  using rep = int64_t;
+  using period = std::ratio_multiply<std::ratio<100>, std::nano>;
+  using duration = std::chrono::duration<rep, period>;
+  using time_point = std::chrono::time_point<XClock>;
+  static constexpr bool is_steady = false;
+
+  static time_point now() noexcept {
+    return from_file_time(Clock::QueryGuestSystemTime());
+  }
+
+  static uint64_t to_file_time(time_point const& tp) noexcept {
+    return static_cast<uint64_t>(tp.time_since_epoch().count());
+  }
+
+  static time_point from_file_time(uint64_t const& tp) noexcept {
+    return time_point{duration{tp}};
+  }
+
+  static std::chrono::system_clock::time_point to_sys(time_point const& tp) {
+    // TODO(gibbed): verify behavior under Linux
+    using sys_duration = std::chrono::system_clock::duration;
+    using sys_time = std::chrono::system_clock::time_point;
+    auto dp = tp;
+    dp += system_clock_delta();
+    auto cdp = std::chrono::time_point_cast<sys_duration>(dp);
+    return sys_time{cdp.time_since_epoch()};
+  }
+
+  static time_point from_sys(std::chrono::system_clock::time_point const& tp) {
+    // TODO(gibbed): verify behavior under Linux
+    auto ctp = std::chrono::time_point_cast<duration>(tp);
+    auto dp = time_point{ctp.time_since_epoch()};
+    dp -= system_clock_delta();
+    return dp;
+  }
+
+ private:
+  // The delta between std::chrono::system_clock (Jan 1 1970) and Xenon file
+  // time (Jan 1 1601), in seconds. In the spec std::chrono::system_clock's
+  // epoch is undefined, but C++20 cements it as Jan 1 1970.
+  static constexpr std::chrono::seconds system_clock_delta() {
+    auto filetime_epoch = date::year{1601} / date::month{1} / date::day{1};
+    auto system_clock_epoch = date::year{1970} / date::month{1} / date::day{1};
+    std::chrono::system_clock::time_point fp{
+        static_cast<date::sys_days>(filetime_epoch)};
+    std::chrono::system_clock::time_point sp{
+        static_cast<date::sys_days>(system_clock_epoch)};
+    return std::chrono::floor<std::chrono::seconds>(fp.time_since_epoch() -
+                                                    sp.time_since_epoch());
+  }
+};
+
+}  // namespace kernel
+}  // namespace xe
+
+#endif  // XENIA_KERNEL_XCLOCK_H_

src/xenia/kernel/xenumerator.cc

@@ -14,7 +14,7 @@ namespace kernel {
 
 XEnumerator::XEnumerator(KernelState* kernel_state, size_t items_per_enumerate,
                          size_t item_size)
-    : XObject(kernel_state, kType),
+    : XObject(kernel_state, kObjectType),
       items_per_enumerate_(items_per_enumerate),
       item_size_(item_size) {}
 

src/xenia/kernel/xenumerator.h

@@ -21,7 +21,7 @@ namespace kernel {
 
 class XEnumerator : public XObject {
  public:
-  static const Type kType = kTypeEnumerator;
+  static const XObject::Type kObjectType = XObject::Type::Enumerator;
 
   XEnumerator(KernelState* kernel_state, size_t items_per_enumerate,
               size_t item_size);

src/xenia/kernel/xevent.cc

@@ -15,7 +15,8 @@
 namespace xe {
 namespace kernel {
 
-XEvent::XEvent(KernelState* kernel_state) : XObject(kernel_state, kType) {}
+XEvent::XEvent(KernelState* kernel_state)
+    : XObject(kernel_state, kObjectType) {}
 
 XEvent::~XEvent() = default;
 

src/xenia/kernel/xevent.h

@@ -25,7 +25,7 @@ static_assert_size(X_KEVENT, 0x10);
 
 class XEvent : public XObject {
  public:
-  static const Type kType = kTypeEvent;
+  static const XObject::Type kObjectType = XObject::Type::Event;
 
   explicit XEvent(KernelState* kernel_state);
   ~XEvent() override;

src/xenia/kernel/xfile.cc

@@ -22,11 +22,13 @@ namespace xe {
 namespace kernel {
 
 XFile::XFile(KernelState* kernel_state, vfs::File* file, bool synchronous)
-    : XObject(kernel_state, kType), file_(file), is_synchronous_(synchronous) {
+    : XObject(kernel_state, kObjectType),
+      file_(file),
+      is_synchronous_(synchronous) {
   async_event_ = threading::Event::CreateAutoResetEvent(false);
 }
 
-XFile::XFile() : XObject(kType) {
+XFile::XFile() : XObject(kObjectType) {
   async_event_ = threading::Event::CreateAutoResetEvent(false);
 }