Docs: Update for release FEX-2109

Fix unaligned CASPair on ARMv8.0

.gitmodules

@@ -33,3 +33,12 @@
 [submodule "External/jemalloc"]
 	path = External/jemalloc
 	url = https://github.com/FEX-Emu/jemalloc.git
+[submodule "External/fmt"]
+	path = External/fmt
+	url = https://github.com/fmtlib/fmt.git
+[submodule "External/drm-headers"]
+	path = External/drm-headers
+	url = https://github.com/FEX-Emu/drm-headers.git
+[submodule "External/xxhash"]
+	path = External/xxhash
+	url = https://github.com/FEX-Emu/xxHash.git

CMakeLists.txt

@@ -14,6 +14,9 @@ option(ENABLE_ASSERTIONS "Enables assertions in build" FALSE)
 option(ENABLE_VISUAL_DEBUGGER "Enables the visual debugger for compiling" FALSE)
 option(ENABLE_STRICT_WERROR "Enables stricter -Werror for CI" FALSE)
 option(ENABLE_WERROR "Enables -Werror" FALSE)
+option(ENABLE_STATIC_PIE "Enables static-pie build" FALSE)
+option(ENABLE_JEMALLOC "Enables jemalloc allocator" TRUE)
+option(ENABLE_OFFLINE_TELEMETRY "Enables FEX offline telemetry" TRUE)
 
 set (X86_C_COMPILER "x86_64-linux-gnu-gcc" CACHE STRING "c compiler for compiling x86 guest libs")
 set (X86_CXX_COMPILER "x86_64-linux-gnu-g++" CACHE STRING "c++ compiler for compiling x86 guest libs")
@@ -44,38 +47,6 @@ else()
   set(CMAKE_INTERPROCEDURAL_OPTIMIZATION FALSE)
 endif()
 
-find_program(CCACHE_PROGRAM ccache)
-if(CCACHE_PROGRAM)
-  message(STATUS "CCache enabled")
-  set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE "${CCACHE_PROGRAM}")
-endif()
-
-if (ENABLE_XRAY)
-  add_compile_options(-fxray-instrument)
-  link_libraries(-fxray-instrument)
-endif()
-
-if (ENABLE_LLD)
-  link_libraries(-fuse-ld=lld)
-endif()
-
-if (ENABLE_ASAN)
-  add_definitions(-DENABLE_ASAN=1)
-  add_compile_options(-fno-omit-frame-pointer -fsanitize=address -fsanitize-address-use-after-scope)
-  link_libraries(-fno-omit-frame-pointer -fsanitize=address -fsanitize-address-use-after-scope)
-endif()
-
-if (ENABLE_TSAN)
-  add_compile_options(-fno-omit-frame-pointer -fsanitize=thread)
-  link_libraries(-fno-omit-frame-pointer -fsanitize=thread)
-endif()
-
-set (CMAKE_CXX_FLAGS_RELWITHDEBINFO "${CMAKE_CXX_FLAGS_RELWITHDEBINFO} -fno-omit-frame-pointer")
-set (CMAKE_LINKER_FLAGS_RELWITHDEBINFO "${CMAKE_LINKER_FLAGS_RELWITHDEBINFO} -fno-omit-frame-pointer")
-
-set (CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -fomit-frame-pointer")
-set (CMAKE_LINKER_FLAGS_RELEASE "${CMAKE_LINKER_FLAGS_RELEASE} -fomit-frame-pointer")
-
 if (CMAKE_SYSTEM_PROCESSOR MATCHES "x86_64")
   option(ENABLE_X86_HOST_DEBUG "Enables compiling on x86_64 host" FALSE)
   if (NOT ENABLE_X86_HOST_DEBUG)
@@ -95,6 +66,181 @@ if (CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64")
   add_definitions(-D_M_ARM_64=1)
 endif()
 
+find_program(CCACHE_PROGRAM ccache)
+if(CCACHE_PROGRAM)
+  message(STATUS "CCache enabled")
+  set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE "${CCACHE_PROGRAM}")
+endif()
+
+if (ENABLE_XRAY)
+  add_compile_options(-fxray-instrument)
+  link_libraries(-fxray-instrument)
+endif()
+
+set (PTHREAD_LIB pthread)
+if (ENABLE_LLD)
+  set (LD_OVERRIDE "-fuse-ld=lld")
+  link_libraries(${LD_OVERRIDE})
+endif()
+
+if (NOT ENABLE_OFFLINE_TELEMETRY)
+  # Disable FEX offline telemetry entirely if asked
+  add_definitions(-DFEX_DISABLE_TELEMETRY=1)
+endif()
+
+if (ENABLE_STATIC_PIE)
+  if (_M_ARM_64 AND ENABLE_LLD)
+    message (FATAL_ERROR "Static linking does not currently work with AArch64+LLD. Use GNU ld for now.")
+  endif()
+
+  file(WRITE ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp/Determine_iplt.c
+    "int main(int argc, char* argv[])
+      {
+        return 0;
+      }")
+
+  # Compile the test application with our LD_OVERRIDE and static-pie options
+  try_compile(
+    COMPILE_RESULT
+    ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp
+    ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp/Determine_iplt.c
+    COMPILE_DEFINITIONS "-fPIE ${LD_OVERRIDE}"
+    LINK_LIBRARIES "-static-pie ${LD_OVERRIDE}"
+    COPY_FILE ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp/Determine_iplt
+  )
+
+  if (${COMPILE_RESULT})
+    # Read the symbols from the elf
+    execute_process(COMMAND
+      readelf -s ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp/Determine_iplt
+      OUTPUT_FILE ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp/plt_out.txt
+      OUTPUT_VARIABLE PLT_SYMBOLS)
+
+    # Pull out the __rela_iplt_{start,end} symbols if they exist
+    execute_process(COMMAND
+      "grep" "__rela_iplt" ${PROJECT_BINARY_DIR}/CMakeFiles/CMakeTmp/plt_out.txt
+      OUTPUT_VARIABLE PLT_SYMBOLS)
+
+    set (SYMBOLS_FINE TRUE)
+    set (HAS_IPLT -1)
+    # Check if we have any symbols in our grep output
+    # The symbols must either not exist at all OR the symbols are zero
+    if (PLT_SYMBOLS)
+      string(FIND ${PLT_SYMBOLS} "__rela_iplt_start" HAS_IPLT)
+    endif()
+
+    if (NOT HAS_IPLT EQUAL -1)
+      # We have some symbols from readelf. Let's parse the results to check if they are zero
+      # Format: '35: 0000000000000000     0 NOTYPE  LOCAL  HIDDEN   UND __rela_iplt_start'
+      string(REPLACE "\n" ";" SYMBOL_LIST ${PLT_SYMBOLS})
+      foreach (SYMBOL ${SYMBOL_LIST})
+        # strip any leading and trailing whitespace
+        string (STRIP ${SYMBOL} SYMBOL)
+        # Convert string to a list
+        string(REPLACE " " ";" SYMBOL_VALUES ${SYMBOL}})
+        # Pull out the address argument
+        list(GET SYMBOL_VALUES 1 OFFSET)
+
+        # Check against integer zero
+        if (NOT ${OFFSET} EQUAL 0)
+          # Symbol wasn't zero, this now fails
+          set (SYMBOLS_FINE FALSE)
+        endif()
+      endforeach()
+    endif()
+
+    if (SYMBOLS_FINE)
+      # We can now exnable static-pie
+      set (STATIC_PIE_OPTIONS "-static-pie")
+      # Pthreads has an issue with exposing symbols
+      # We need to make some concessions to the pthread gods
+      if (ENABLE_LLD)
+        set (PTHREAD_LIB
+          -Wl,--undefined-glob=pthread_*
+          -Wl,--undefined=__cxa_finalize
+          -Wl,--undefined=_pthread_cleanup_push_defer
+          -Wl,--undefined=_pthread_cleanup_pop_restore
+          -Wl,--undefined=__pthread_cleanup_upto
+          pthread)
+      else()
+        set (PTHREAD_LIB
+          -Wl,--undefined=pthread_join
+          -Wl,--undefined=pthread_attr_getdetachstate
+          -Wl,--undefined=pthread_sigmask
+          -Wl,--undefined=pthread_mutex_lock
+          -Wl,--undefined=pthread_cond_init
+          -Wl,--undefined=pthread_attr_init
+          -Wl,--undefined=pthread_mutex_unlock
+          -Wl,--undefined=pthread_mutexattr_destroy
+          -Wl,--undefined=pthread_detach
+          -Wl,--undefined=pthread_mutex_init
+          -Wl,--undefined=pthread_getattr_np
+          -Wl,--undefined=pthread_cond_timedwait
+          -Wl,--undefined=pthread_attr_destroy
+          -Wl,--undefined=pthread_mutexattr_settype
+          -Wl,--undefined=pthread_rwlock_unlock
+          -Wl,--undefined=pthread_rwlock_wrlock
+          -Wl,--undefined=pthread_setspecific
+          -Wl,--undefined=pthread_create
+          -Wl,--undefined=pthread_cond_clockwait
+          -Wl,--undefined=pthread_key_create
+          -Wl,--undefined=pthread_rwlock_rdlock
+          -Wl,--undefined=pthread_setname_np
+          -Wl,--undefined=pthread_cond_signal
+          -Wl,--undefined=pthread_mutexattr_init
+          -Wl,--undefined=pthread_attr_setstack
+          -Wl,--undefined=pthread_self
+          -Wl,--undefined=pthread_getaffinity_np
+          -Wl,--undefined=pthread_cond_wait
+          -Wl,--undefined=pthread_mutex_trylock
+          -Wl,--undefined=pthread_cond_broadcast
+          -Wl,--undefined=pthread_cond_destroy
+          -Wl,--undefined=pthread_getspecific
+          -Wl,--undefined=pthread_key_delete
+          -Wl,--undefined=pthread_once
+          -Wl,--undefined=__cxa_finalize
+          -Wl,--undefined=_pthread_cleanup_push_defer
+          -Wl,--undefined=_pthread_cleanup_pop_restore
+          -Wl,--undefined=__pthread_cleanup_upto
+          pthread)
+      endif()
+    else()
+      message (FATAL_ERROR "Application has __rela_iplt_{start,end} symbols. Which means static-pie can't be enabled")
+    endif()
+  else()
+    message (FATAL_ERROR "Couldn't compile static-pie test. Static-pie can't be enabled!")
+  endif()
+endif()
+
+if (ENABLE_ASAN)
+  add_definitions(-DENABLE_ASAN=1)
+  add_compile_options(-fno-omit-frame-pointer -fsanitize=address -fsanitize-address-use-after-scope)
+  link_libraries(-fno-omit-frame-pointer -fsanitize=address -fsanitize-address-use-after-scope)
+endif()
+
+if (ENABLE_TSAN)
+  add_compile_options(-fno-omit-frame-pointer -fsanitize=thread)
+  link_libraries(-fno-omit-frame-pointer -fsanitize=thread)
+endif()
+
+if (ENABLE_JEMALLOC)
+  add_definitions(-DENABLE_JEMALLOC=1)
+  add_subdirectory(External/jemalloc/)
+  include_directories(External/jemalloc/pregen/include/)
+else()
+  message (STATUS
+    " jemalloc disabled!\n"
+    " This is not a recommended configuration!\n"
+    " This will very explicitly break 32-bit application execution!\n"
+    " Use at your own risk!")
+endif()
+
+set (CMAKE_CXX_FLAGS_RELWITHDEBINFO "${CMAKE_CXX_FLAGS_RELWITHDEBINFO} -fno-omit-frame-pointer")
+set (CMAKE_LINKER_FLAGS_RELWITHDEBINFO "${CMAKE_LINKER_FLAGS_RELWITHDEBINFO} -fno-omit-frame-pointer")
+
+set (CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -fomit-frame-pointer")
+set (CMAKE_LINKER_FLAGS_RELEASE "${CMAKE_LINKER_FLAGS_RELEASE} -fomit-frame-pointer")
+
 add_subdirectory(External/vixl/)
 include_directories(External/vixl/src/)
 
@@ -105,17 +251,22 @@ endif()
 
 find_package(PkgConfig REQUIRED)
 find_package(Python 3.0 REQUIRED COMPONENTS Interpreter)
-pkg_check_modules(XXHASH libxxhash REQUIRED)
+pkg_check_modules(XXHASH libxxhash>=0.8.0 QUIET)
+
+if (NOT XXHASH_FOUND)
+  message(STATUS "xxHash not found. Using Externals")
+  add_subdirectory(External/xxhash/)
+  include_directories(External/xxhash/)
+endif()
 
 add_definitions(-Wno-trigraphs)
 add_definitions(-DGLOBAL_DATA_DIRECTORY="${DATA_DIRECTORY}/")
 
-add_subdirectory(External/jemalloc/)
-include_directories(External/jemalloc/pregen/include/)
-
 add_subdirectory(External/cpp-optparse/)
 include_directories(External/cpp-optparse/)
 
+add_subdirectory(External/fmt/)
+
 add_subdirectory(External/imgui/)
 include_directories(External/imgui/)
 
@@ -250,7 +401,7 @@ add_compile_options(-Wall)
 
 configure_file(
   ${CMAKE_CURRENT_SOURCE_DIR}/include/Config.h.in
-  ${CMAKE_BINARY_DIR}/generated/Config.h)
+  ${CMAKE_BINARY_DIR}/generated/ConfigDefines.h)
 
 if (BUILD_TESTS)
   include(CTest)
@@ -259,6 +410,9 @@ if (BUILD_TESTS)
 endif()
 add_subdirectory(External/FEXCore)
 
+# Binfmt_misc files must be installed prior to Source/ installs
+add_subdirectory(Data/binfmts/)
+
 add_subdirectory(Source/)
 add_subdirectory(Data/AppConfig/)
 
@@ -305,3 +459,64 @@ if (BUILD_THUNKS)
     DEPENDS guest-libs
   )
 endif()
+
+set(FEX_VERSION_MAJOR "0")
+set(FEX_VERSION_MINOR "0")
+set(FEX_VERSION_PATCH "0")
+
+find_package(Git)
+if (GIT_FOUND)
+  execute_process(
+    COMMAND ${GIT_EXECUTABLE} describe --abbrev=0
+    WORKING_DIRECTORY "${CMAKE_SOURCE_DIR}"
+    OUTPUT_VARIABLE GIT_DESCRIBE_STRING
+    RESULT_VARIABLE GIT_ERROR
+    ERROR_QUIET
+    OUTPUT_STRIP_TRAILING_WHITESPACE
+  )
+
+  if (NOT ${GIT_ERROR} EQUAL 0)
+    # Likely built in a way that doesn't have tags
+    # Setup a version tag that is unknown
+    set(GIT_DESCRIBE_STRING "FEX-0000")
+  endif()
+
+  # Change something like `FEX-2106.1-76-<hash>` in to a list
+  string(REPLACE "-" ";" DESCRIBE_LIST ${GIT_DESCRIBE_STRING})
+
+  # Extract the `2106.1` element
+  list(GET DESCRIBE_LIST 1 DESCRIBE_LIST)
+
+  # Change `2106.1` in to a list
+  string(REPLACE "." ";" DESCRIBE_LIST ${DESCRIBE_LIST})
+
+  # Calculate list size
+  list(LENGTH DESCRIBE_LIST LIST_SIZE)
+
+  # Pull out the major version
+  list(GET DESCRIBE_LIST 0 FEX_VERSION_MAJOR)
+
+  # Minor version only exists if there is a .1 at the end
+  # eg: 2106 versus 2106.1
+  if (LIST_SIZE GREATER 1)
+    list(GET DESCRIBE_LIST 1 FEX_VERSION_MINOR)
+  endif()
+endif()
+
+# Package creation
+set (CPACK_GENERATOR "DEB")
+set (CPACK_PACKAGE_NAME fex-emu)
+set (CPACK_PACKAGE_CONTACT "team@fex-emu.org")
+set (CPACK_PACKAGE_VERSION_MAJOR "${FEX_VERSION_MAJOR}")
+set (CPACK_PACKAGE_VERSION_MINOR "${FEX_VERSION_MINOR}")
+set (CPACK_PACKAGE_VERSION_PATCH "${FEX_VERSION_PATCH}")
+
+# Debian defines
+set (CPACK_DEBIAN_PACKAGE_DEPENDS "libstdc++6")
+set (CPACK_DEBIAN_PACKAGE_CONTROL_EXTRA "${CMAKE_CURRENT_SOURCE_DIR}/CPack/postinst;${CMAKE_CURRENT_SOURCE_DIR}/CPack/prerm")
+if (CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64")
+  # binfmt_misc conflicts with qemu-user-static
+  # We also only install binfmt_misc on aarch64 hosts
+  set (CPACK_DEBIAN_PACKAGE_CONFLICTS "qemu-user-static")
+endif()
+include (CPack)

CPack/postinst

@@ -0,0 +1,18 @@
+#!/bin/sh
+set -e
+update_binfmt() {
+  # Check for update-binfmts
+  command -v update-binfmts >/dev/null || return 0
+
+  # Setup binfmt_misc
+  update-binfmts --import FEX-x86
+  update-binfmts --import FEX-x86_64
+}
+
+# Install FEXInterpreter hardlink
+# Needs to be done before setting up binfmt_misc
+ln -f /usr/bin/FEXLoader /usr/bin/FEXInterpreter
+
+if [ $(uname -m) = 'aarch64' ]; then
+  update_binfmt
+fi

CPack/prerm

@@ -0,0 +1,17 @@
+#!/bin/sh
+set -e
+update_binfmt() {
+  # Check for update-binfmts
+  command -v update-binfmts >/dev/null || return 0
+
+  # Uninstall
+  update-binfmts --unimport FEX-x86
+  update-binfmts --unimport FEX-x86_64
+}
+
+if [ $(uname -m) = 'aarch64' ]; then
+  update_binfmt
+fi
+
+# Remove FEXInterpreter hardlink
+unlink /usr/bin/FEXInterpreter

Data/AppConfig/steamwebhelper.json

@@ -0,0 +1,5 @@
+{
+  "Config": {
+    "StallProcess": "1"
+  }
+}

Data/binfmts/CMakeLists.txt

@@ -0,0 +1,17 @@
+function(GenBinFmt Name)
+  # Get the filename only component
+  get_filename_component(FMT_NAME ${Name} NAME_WE)
+
+  # Configure it
+  configure_file(
+    ${Name}
+    ${CMAKE_BINARY_DIR}/Data/binfmts/${FMT_NAME})
+
+  # Then install the configured binfmt
+  install(
+    FILES ${CMAKE_BINARY_DIR}/Data/binfmts/${FMT_NAME}
+    DESTINATION ${CMAKE_INSTALL_PREFIX}/share/binfmts/)
+endfunction()
+
+GenBinFmt(FEX-x86.in)
+GenBinFmt(FEX-x86_64.in)

Data/binfmts/FEX-x86.in

@@ -0,0 +1,8 @@
+package fex
+interpreter @CMAKE_INSTALL_PREFIX@/bin/FEXInterpreter
+magic \x7fELF\x01\x01\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00\x02\x00\x03\x00
+offset 0
+mask \xff\xff\xff\xff\xff\xfe\xfe\x00\x00\x00\x00\xff\xff\xff\xff\xff\xfe\xff\xff\xff
+credentials yes
+fix_binary yes
+preserve yes

Data/binfmts/FEX-x86_64.in

@@ -0,0 +1,8 @@
+package fex
+interpreter @CMAKE_INSTALL_PREFIX@/bin/FEXInterpreter
+magic \x7fELF\x02\x01\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00\x02\x00\x3e\x00
+offset 0
+mask \xff\xff\xff\xff\xff\xfe\xfe\x00\x00\x00\x00\xff\xff\xff\xff\xff\xfe\xff\xff\xff
+credentials yes
+fix_binary yes
+preserve yes

Dockerfile

@@ -3,7 +3,7 @@ FROM ubuntu:20.04 as builder
 
 RUN DEBIAN_FRONTEND="noninteractive" apt-get update
 RUN DEBIAN_FRONTEND="noninteractive" apt install -y cmake \
-clang-10 llvm-10 nasm ninja-build libnuma-dev \
+clang-10 llvm-10 nasm ninja-build \
 libcap-dev libglfw3-dev libepoxy-dev python3-dev \
 python3 linux-headers-generic
 
@@ -23,7 +23,7 @@ FROM ubuntu:20.04
 
 RUN DEBIAN_FRONTEND="noninteractive" apt-get update
 RUN DEBIAN_FRONTEND="noninteractive" apt install -y \
-libnuma-dev libcap-dev libglfw3-dev libepoxy-dev
+libcap-dev libglfw3-dev libepoxy-dev
 
 COPY --from=builder /opt/FEX/build/Bin/* /usr/bin/
 

External/FEXCore/Scripts/config_generator.py

@@ -98,14 +98,19 @@ def print_man_option(short, long, desc, default):
     output_man.write("\\fBdefault:\\fR {0}\n".format(default))
     output_man.write(".Pp\n\n")
 
-def print_man_env_option(name, desc, default):
-    output_man.write("\\fBFEX_{0}\\fR\n".format(name))
+def print_man_env_option(name, desc, default, no_json_key):
+    output_man.write("\\fBFEX_{0}\\fR\n".format(name.upper()))
 
     # Print description
     for line in desc:
         output_man.write(".Pp\n")
         output_man.write("{0}\n".format(line))
 
+    if (not no_json_key):
+        output_man.write(".Pp\n")
+        output_man.write("\\fBJSON key:\\fR '{0}'\n".format(name))
+        output_man.write(".Pp\n\n")
+
     output_man.write(".Pp\n")
     output_man.write("\\fBdefault:\\fR {0}\n".format(default))
     output_man.write(".Pp\n\n")
@@ -154,9 +159,10 @@ def print_man_environment(options):
                 # Wrap the string argument in quotes
                 default = "'" + default + "'"
             print_man_env_option(
-                op_key.upper(),
+                op_key,
                 op_vals["Desc"],
-                default
+                default,
+                False
             )
 
     print_man_environment_tail()
@@ -172,7 +178,7 @@ def print_man_environment_tail():
     "By default FEX will look in {$HOME, $XDG_CONFIG_HOME}/.fex-emu/",
     "This will override the full path",
     ],
-    "''")
+    "''", True)
 
     print_man_env_option(
     "FEX_APP_CONFIG",
@@ -183,7 +189,7 @@ def print_man_environment_tail():
     "One must be careful with this option as it will override any applications that load with execve as well"
     "If you need to support applications that execve then use FEX_APP_CONFIG_LOCATION instead"
     ],
-    "''")
+    "''", True)
 
     print_man_env_option(
     "FEX_APP_DATA_LOCATION",
@@ -193,7 +199,7 @@ def print_man_environment_tail():
     "This will override the full path",
     "This is the folder where FEX stores generated files like IR cache"
     ],
-    "''")
+    "''", True)
 
 def print_man_header():
     header ='''.Dd {0}

External/FEXCore/Source/CMakeLists.txt

@@ -81,7 +81,12 @@ set (SRCS
   Interface/Core/Frontend.cpp
   Interface/Core/GdbServer.cpp
   Interface/Core/HostFeatures.cpp
+  Interface/Core/OpcodeDispatcher/Crypto.cpp
+  Interface/Core/OpcodeDispatcher/Flags.cpp
+  Interface/Core/OpcodeDispatcher/Vector.cpp
+  Interface/Core/OpcodeDispatcher/X87.cpp
   Interface/Core/OpcodeDispatcher.cpp
+  Interface/Core/SignalDelegator.cpp
   Interface/Core/X86Tables.cpp
   Interface/Core/X86DebugInfo.cpp
   Interface/Core/X86HelperGen.cpp
@@ -124,9 +129,8 @@ set (SRCS
   Interface/IR/Passes/SyscallOptimization.cpp
   Utils/Allocator.cpp
   Utils/Allocator/64BitAllocator.cpp
-  Utils/ELFContainer.cpp
-  Utils/ELFSymbolDatabase.cpp
   Utils/LogManager.cpp
+  Utils/Telemetry.cpp
   Utils/Threads.cpp
   )
 
@@ -135,7 +139,7 @@ if(_M_ARM_64)
     Interface/Core/ArchHelpers/Arm64.cpp)
 endif()
 
-set(DEFINES )
+set(DEFINES -DTHREAD_LOCAL=_Thread_local)
 
 if (_M_X86_64)
   list(APPEND DEFINES -D_M_X86_64=1)
@@ -181,6 +185,11 @@ if (ENABLE_JITSYMBOLS)
   list(APPEND DEFINES -DENABLE_JITSYMBOLS=1)
 endif()
 
+set (LIBS vixl dl fmt::fmt xxhash tiny-json)
+if (ENABLE_JEMALLOC)
+  list (APPEND LIBS FEX_jemalloc)
+endif()
+
 # Generate IR include file
 set(OUTPUT_IR_FOLDER "${CMAKE_BINARY_DIR}/include/FEXCore/IR")
 set(OUTPUT_NAME "${OUTPUT_IR_FOLDER}/IRDefines.inc")
@@ -267,7 +276,7 @@ function(AddObject Name Type)
   add_dependencies(${Name} IR_INC)
   add_dependencies(${Name} CONFIG_INC)
 
-  target_link_libraries(${Name} pthread vixl dl xxhash FEX_jemalloc)
+  target_link_libraries(${Name} ${LIBS})
   set_target_properties(${Name} PROPERTIES OUTPUT_NAME FEXCore)
   set_target_properties(${Name} PROPERTIES C_VISIBILITY_PRESET hidden)
   set_target_properties(${Name} PROPERTIES CXX_VISIBILITY_PRESET hidden)
@@ -286,6 +295,8 @@ function(AddObject Name Type)
   target_compile_options(${Name}
     PRIVATE
     -Wall
+    -Werror=cast-qual
+    -Werror=ignored-qualifiers
     -Werror=implicit-fallthrough
 
     -Wno-trigraphs
@@ -306,7 +317,7 @@ endfunction()
 
 function(AddLibrary Name Type)
   add_library(${Name} ${Type} $<TARGET_OBJECTS:${PROJECT_NAME}_object>)
-  target_link_libraries(${Name} pthread vixl dl xxhash FEX_jemalloc)
+  target_link_libraries(${Name} ${LIBS})
   set_target_properties(${Name} PROPERTIES OUTPUT_NAME FEXCore)
   set_target_properties(${Name} PROPERTIES C_VISIBILITY_PRESET hidden)
   set_target_properties(${Name} PROPERTIES CXX_VISIBILITY_PRESET hidden)

External/FEXCore/Source/Common/NetStream.cpp

@@ -1,11 +1,7 @@
 #include "NetStream.h"
 
-#include <cstring>
-
-#include <sys/types.h>
 #include <sys/socket.h>
 
-#include <stdio.h>
 #include <unistd.h>
 
 int NetStream::NetBuf::flushBuffer(const char *buffer, size_t size) {
@@ -29,7 +25,7 @@ std::streamsize NetStream::NetBuf::xsputn(const char* buffer, std::streamsize si
 
     // Check if the string fits neatly in our buffer
     if (size <= buf_remaining) {
-        std::memcpy(pptr(), buffer, size);
+        ::memcpy(pptr(), buffer, size);
         pbump(size);
         return size;
     }

External/FEXCore/Source/Common/NetStream.h

@@ -2,6 +2,7 @@
 
 #include <array>
 #include <iostream>
+#include <iterator>
 #include <string.h>
 
 class NetStream : public std::iostream {

External/FEXCore/Source/Common/Paths.cpp

@@ -3,13 +3,48 @@
 
 #include <cstdlib>
 #include <filesystem>
-#include <sys/stat.h>
+#include <memory>
+#include <pwd.h>
+#include <system_error>
+#include <unistd.h>
 
 namespace FEXCore::Paths {
-  std::string CachePath;
-  std::string EntryCache;
+  std::unique_ptr<std::string> CachePath;
+  std::unique_ptr<std::string> EntryCache;
+
+  char const* FindUserHomeThroughUID() {
+    auto passwd = getpwuid(geteuid());
+    if (passwd) {
+      return passwd->pw_dir;
+    }
+    return nullptr;
+  }
+
+  const char *GetHomeDirectory() {
+    char const *HomeDir = getenv("HOME");
+
+    // Try to get home directory from uid
+    if (!HomeDir) {
+      HomeDir = FindUserHomeThroughUID();
+    }
+
+    // try the PWD
+    if (!HomeDir) {
+      HomeDir = getenv("PWD");
+    }
+
+    // Still doesn't exit? You get local
+    if (!HomeDir) {
+      HomeDir = ".";
+    }
+
+    return HomeDir;
+  }
 
   void InitializePaths() {
+    CachePath = std::make_unique<std::string>();
+    EntryCache = std::make_unique<std::string>();
+
     char const *HomeDir = getenv("HOME");
 
     if (!HomeDir) {
@@ -22,29 +57,35 @@ namespace FEXCore::Paths {
 
     char *XDGDataDir = getenv("XDG_DATA_DIR");
     if (XDGDataDir) {
-      CachePath = XDGDataDir;
+      *CachePath = XDGDataDir;
     }
     else {
       if (HomeDir) {
-        CachePath = HomeDir;
+        *CachePath = HomeDir;
       }
     }
 
-    CachePath += "/.fex-emu/";
-    EntryCache = CachePath + "/EntryCache/";
+    *CachePath += "/.fex-emu/";
+    *EntryCache = *CachePath + "/EntryCache/";
 
+    std::error_code ec{};
     // Ensure the folder structure is created for our Data
-    if (!std::filesystem::exists(EntryCache) &&
-        !std::filesystem::create_directories(EntryCache)) {
-      LogMan::Msg::D("Couldn't create EntryCache directory: '%s'", EntryCache.c_str());
+    if (!std::filesystem::exists(*EntryCache, ec) &&
+        !std::filesystem::create_directories(*EntryCache, ec)) {
+      LogMan::Msg::D("Couldn't create EntryCache directory: '%s'", EntryCache->c_str());
     }
   }
 
+  void ShutdownPaths() {
+    CachePath.reset();
+    EntryCache.reset();
+  }
+
   std::string GetCachePath() {
-    return CachePath;
+    return *CachePath;
   }
 
   std::string GetEntryCachePath() {
-    return EntryCache;
+    return *EntryCache;
   }
 }

External/FEXCore/Source/Common/Paths.h

@@ -3,6 +3,10 @@
 
 namespace FEXCore::Paths {
   void InitializePaths();
+  void ShutdownPaths();
+
+  const char *GetHomeDirectory();
+
   std::string GetCachePath();
   std::string GetEntryCachePath();
 }

External/FEXCore/Source/Common/SoftFloat.h

@@ -1,4 +1,6 @@
 #pragma once
+
+#include <FEXCore/Utils/BitUtils.h>
 #include <FEXCore/Utils/LogManager.h>
 
 #include <cmath>
@@ -158,18 +160,18 @@ struct X80SoftFloat {
   }
 
   operator float() const {
-    float32_t Result = extF80_to_f32(*this);
-    return *(float*)&Result;
+    const float32_t Result = extF80_to_f32(*this);
+    return FEXCore::BitCast<float>(Result);
   }
 
   operator double() const {
-    float64_t Result = extF80_to_f64(*this);
-    return *(double*)&Result;
+    const float64_t Result = extF80_to_f64(*this);
+    return FEXCore::BitCast<double>(Result);
   }
 
   operator BIGFLOAT() const {
-    float128_t Result = extF80_to_f128(*this);
-    return *(BIGFLOAT*)&Result;
+    const float128_t Result = extF80_to_f128(*this);
+    return FEXCore::BitCast<BIGFLOAT>(Result);
   }
 
   operator int16_t() const {
@@ -196,11 +198,11 @@ struct X80SoftFloat {
   }
 
   void operator=(const float rhs) {
-    *this = f32_to_extF80(*(float32_t*)&rhs);
+    *this = f32_to_extF80(FEXCore::BitCast<float32_t>(rhs));
   }
 
   void operator=(const double rhs) {
-    *this = f64_to_extF80(*(float64_t*)&rhs);
+    *this = f64_to_extF80(FEXCore::BitCast<float64_t>(rhs));
   }
 
   void operator=(const int16_t rhs) {
@@ -226,15 +228,15 @@ struct X80SoftFloat {
   }
 
   X80SoftFloat(const float rhs) {
-    *this = f32_to_extF80(*(float32_t*)&rhs);
+    *this = f32_to_extF80(FEXCore::BitCast<float32_t>(rhs));
   }
 
   X80SoftFloat(const double rhs) {
-    *this = f64_to_extF80(*(float64_t*)&rhs);
+    *this = f64_to_extF80(FEXCore::BitCast<float64_t>(rhs));
   }
 
   X80SoftFloat(BIGFLOAT rhs) {
-    *this = f128_to_extF80(*(float128_t*)&rhs);
+    *this = f128_to_extF80(FEXCore::BitCast<float128_t>(rhs));
   }
 
   X80SoftFloat(const int16_t rhs) {

External/FEXCore/Source/Interface/Config/Config.cpp

@@ -1,43 +1,164 @@
 #include "Common/StringConv.h"
-
-#include <FEXCore/Utils/LogManager.h>
-#include "Interface/Context/Context.h"
+#include "Common/Paths.h"
 
 #include <FEXCore/Config/Config.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <array>
+#include <assert.h>
+#include <cstdlib>
 #include <filesystem>
-#include <pwd.h>
+#include <fstream>
+#include <functional>
 #include <map>
+#include <memory>
+#include <list>
+#include <optional>
+#include <stddef.h>
+#include <stdint.h>
+#include <string>
+#include <string_view>
 #include <sys/sysinfo.h>
-#include <unistd.h>
+#include <system_error>
+#include <type_traits>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+#include <tiny-json.h>
+
+namespace FEXCore::Context {
+  struct Context;
+}
 
 namespace FEXCore::Config {
-  char const* FindUserHomeThroughUID() {
-    auto passwd = getpwuid(geteuid());
-    if (passwd) {
-      return passwd->pw_dir;
+namespace DefaultValues {
+#define P(x) x
+#define OPT_BASE(type, group, enum, json, default) const P(type) P(enum) = P(default);
+#define OPT_STR(group, enum, json, default) const std::string_view P(enum) = P(default);
+#define OPT_STRARRAY(group, enum, json, default) OPT_STR(group, enum, json, default)
+#include <FEXCore/Config/ConfigValues.inl>
+}
+
+namespace JSON {
+  static bool LoadConfigFile(std::vector<char> &Data, const std::string &Config) {
+    std::fstream ConfigFile;
+    ConfigFile.open(Config, std::ios::in);
+
+    if (!ConfigFile.is_open()) {
+      return false;
     }
-    return nullptr;
+
+    if (!ConfigFile.seekg(0, std::fstream::end)) {
+      LogMan::Msg::D("Couldn't load configuration file: Seek end");
+      return false;
+    }
+
+    auto FileSize = ConfigFile.tellg();
+    if (ConfigFile.fail()) {
+      LogMan::Msg::D("Couldn't load configuration file: tellg");
+      return false;
+    }
+
+    if (!ConfigFile.seekg(0, std::fstream::beg)) {
+      LogMan::Msg::D("Couldn't load configuration file: Seek beginning");
+      return false;
+    }
+
+    if (FileSize > 0) {
+      Data.resize(FileSize);
+      if (!ConfigFile.read(&Data.at(0), FileSize)) {
+        // Probably means permissions aren't set. Just early exit
+        return false;
+      }
+      ConfigFile.close();
+    }
+    else {
+      return false;
+    }
+
+    return true;
   }
 
-  const char *GetHomeDirectory() {
-    char const *HomeDir = getenv("HOME");
+  struct JsonAllocator {
+    jsonPool_t PoolObject;
+    std::unique_ptr<std::list<json_t>> json_objects;
+  };
+  static_assert(offsetof(JsonAllocator, PoolObject) == 0, "This needs to be at offset zero");
 
-    // Try to get home directory from uid
-    if (!HomeDir) {
-      HomeDir = FindUserHomeThroughUID();
+  json_t* PoolInit(jsonPool_t* Pool) {
+    JsonAllocator* alloc = reinterpret_cast<JsonAllocator*>(Pool);
+    alloc->json_objects = std::make_unique<std::list<json_t>>();
+    return &*alloc->json_objects->emplace(alloc->json_objects->end());
+  }
+
+  json_t* PoolAlloc(jsonPool_t* Pool) {
+    JsonAllocator* alloc = reinterpret_cast<JsonAllocator*>(Pool);
+    return &*alloc->json_objects->emplace(alloc->json_objects->end());
+  }
+
+  static void LoadJSonConfig(const std::string &Config, std::function<void(const char *Name, const char *ConfigSring)> Func) {
+    std::vector<char> Data;
+    if (!LoadConfigFile(Data, Config)) {
+      return;
     }
 
-    // try the PWD
-    if (!HomeDir) {
-      HomeDir = getenv("PWD");
+    JsonAllocator Pool {
+      .PoolObject = {
+        .init = PoolInit,
+        .alloc = PoolAlloc,
+      },
+    };
+
+    json_t const *json = json_createWithPool(&Data.at(0), &Pool.PoolObject);
+    if (!json) {
+      LogMan::Msg::E("Couldn't create json");
+      return;
     }
 
-    // Still doesn't exit? You get local
-    if (!HomeDir) {
-      HomeDir = ".";
+    json_t const* ConfigList = json_getProperty(json, "Config");
+
+    if (!ConfigList) {
+      LogMan::Msg::E("Couldn't get config list");
+      return;
     }
 
-    return HomeDir;
+    for (json_t const* ConfigItem = json_getChild(ConfigList);
+      ConfigItem != nullptr;
+      ConfigItem = json_getSibling(ConfigItem)) {
+      const char* ConfigName = json_getName(ConfigItem);
+      const char* ConfigString = json_getValue(ConfigItem);
+
+      if (!ConfigName) {
+        LogMan::Msg::E("Couldn't get config name");
+        return;
+      }
+
+      if (!ConfigString) {
+        LogMan::Msg::E("Couldn't get ConfigString for '%s'", ConfigName);
+        return;
+      }
+
+      Func(ConfigName, ConfigString);
+    }
+  }
+}
+
+  std::string GetDataDirectory() {
+    std::string DataDir{};
+
+    char const *HomeDir = Paths::GetHomeDirectory();
+    char const *DataXDG = getenv("XDG_DATA_HOME");
+    char const *DataOverride = getenv("FEX_APP_DATA_LOCATION");
+    if (DataOverride) {
+      // Data override will override the complete directory
+      DataDir = DataOverride;
+    }
+    else {
+      DataDir = DataXDG ?: HomeDir;
+      DataDir += "/.fex-emu/";
+    }
+    return DataDir;
   }
 
   std::string GetConfigDirectory(bool Global) {
@@ -46,7 +167,7 @@ namespace FEXCore::Config {
       ConfigDir = GLOBAL_DATA_DIRECTORY;
     }
     else {
-      char const *HomeDir = GetHomeDirectory();
+      char const *HomeDir = Paths::GetHomeDirectory();
       char const *ConfigXDG = getenv("XDG_CONFIG_HOME");
       char const *ConfigOverride = getenv("FEX_APP_CONFIG_LOCATION");
       if (ConfigOverride) {
@@ -59,9 +180,9 @@ namespace FEXCore::Config {
       }
 
       // Ensure the folder structure is created for our configuration
-      if (!std::filesystem::exists(ConfigDir) &&
-          !std::filesystem::create_directories(ConfigDir)) {
-        LogMan::Msg::D("Couldn't create config directory: '%s'", ConfigDir.c_str());
+      std::error_code ec{};
+      if (!std::filesystem::exists(ConfigDir, ec) &&
+          !std::filesystem::create_directories(ConfigDir, ec)) {
         // Let's go local in this case
         return "./";
       }
@@ -83,37 +204,32 @@ namespace FEXCore::Config {
     return ConfigFile;
   }
 
-  std::string GetApplicationConfig(std::string &Filename, bool Global) {
+  std::string GetApplicationConfig(const std::string &Filename, bool Global) {
     std::string ConfigFile = GetConfigDirectory(Global);
+
+    std::error_code ec{};
     if (!Global &&
-        !std::filesystem::exists(ConfigFile) &&
-        !std::filesystem::create_directories(ConfigFile)) {
+        !std::filesystem::exists(ConfigFile, ec) &&
+        !std::filesystem::create_directories(ConfigFile, ec)) {
       LogMan::Msg::D("Couldn't create config directory: '%s'", ConfigFile.c_str());
       // Let's go local in this case
-      return "./";
+      return "./" + Filename + ".json";
     }
 
-    ConfigFile += "AppConfig/" + Filename + ".json";
+    ConfigFile += "AppConfig/";
+
+    // Attempt to create the local folder if it doesn't exist
+    if (!Global &&
+        !std::filesystem::exists(ConfigFile, ec) &&
+        !std::filesystem::create_directories(ConfigFile, ec)) {
+      // Let's go local in this case
+      return "./" + Filename + ".json";
+    }
+
+    ConfigFile += Filename + ".json";
     return ConfigFile;
   }
 
-  std::string GetDataDirectory() {
-    std::string DataDir{};
-
-    char const *HomeDir = GetHomeDirectory();
-    char const *DataXDG = getenv("XDG_DATA_HOME");
-    char const *DataOverride = getenv("FEX_APP_DATA_LOCATION");
-    if (DataOverride) {
-      // Data override will override the complete directory
-      DataDir = DataOverride;
-    }
-    else {
-      DataDir = DataXDG ?: HomeDir;
-      DataDir += "/.fex-emu/";
-    }
-    return DataDir;
-  }
-
   void SetConfig(FEXCore::Context::Context *CTX, ConfigOption Option, uint64_t Config) {
   }
 
@@ -262,7 +378,8 @@ namespace FEXCore::Config {
       Path = std::filesystem::absolute(Path);
 
       // Only return if it exists
-      if (std::filesystem::exists(Path)) {
+      std::error_code ec{};
+      if (std::filesystem::exists(Path, ec)) {
         return Path;
       }
     }
@@ -298,7 +415,8 @@ namespace FEXCore::Config {
       else if (!PathName().empty()) {
         // If the filesystem doesn't exist then let's see if it exists in the fex-emu folder
         std::string NamedRootFS = GetDataDirectory() + "RootFS/" + PathName();
-        if (std::filesystem::exists(NamedRootFS)) {
+        std::error_code ec{};
+        if (std::filesystem::exists(NamedRootFS, ec)) {
           FEXCore::Config::EraseSet(FEXCore::Config::CONFIG_ROOTFS, NamedRootFS);
         }
       }
@@ -313,7 +431,19 @@ namespace FEXCore::Config {
     }
     if (FEXCore::Config::Exists(FEXCore::Config::CONFIG_THUNKCONFIG)) {
       FEX_CONFIG_OPT(PathName, THUNKCONFIG);
-      ExpandPathIfExists(FEXCore::Config::CONFIG_THUNKCONFIG, PathName());
+      auto ExpandedString = ExpandPath(PathName());
+      if (!ExpandedString.empty()) {
+        // Adjust the path if it ended up being relative
+        FEXCore::Config::EraseSet(FEXCore::Config::CONFIG_THUNKCONFIG, ExpandedString);
+      }
+      else if (!PathName().empty()) {
+        // If the filesystem doesn't exist then let's see if it exists in the fex-emu folder
+        std::string NamedConfig = GetDataDirectory() + "ThunkConfigs/" + PathName();
+        std::error_code ec{};
+        if (std::filesystem::exists(NamedConfig, ec)) {
+          FEXCore::Config::EraseSet(FEXCore::Config::CONFIG_THUNKCONFIG, NamedConfig);
+        }
+      }
     }
     if (FEXCore::Config::Exists(FEXCore::Config::CONFIG_OUTPUTLOG)) {
       FEX_CONFIG_OPT(PathName, OUTPUTLOG);
@@ -391,6 +521,17 @@ namespace FEXCore::Config {
     }
   }
 
+  template<>
+  std::string Value<std::string>::GetIfExists(FEXCore::Config::ConfigOption Option, std::string_view Default) {
+    auto Value = FEXCore::Config::Get(Option);
+    if (Value) {
+      return **Value;
+    }
+    else {
+      return std::string(Default);
+    }
+  }
+
   template bool     Value<bool>::GetIfExists(FEXCore::Config::ConfigOption Option, bool Default);
   template int8_t   Value<int8_t>::GetIfExists(FEXCore::Config::ConfigOption Option, int8_t Default);
   template uint8_t  Value<uint8_t>::GetIfExists(FEXCore::Config::ConfigOption Option, uint8_t Default);
@@ -416,5 +557,143 @@ namespace FEXCore::Config {
     }
   }
   template void Value<std::string>::GetListIfExists(FEXCore::Config::ConfigOption Option, std::list<std::string> *List);
+
+  // Application loaders
+  class MainLoader final : public FEXCore::Config::OptionMapper {
+  public:
+    explicit MainLoader();
+    explicit MainLoader(std::string ConfigFile);
+    void Load() override;
+
+  private:
+    std::string Config;
+  };
+
+  class AppLoader final : public FEXCore::Config::OptionMapper {
+  public:
+    explicit AppLoader(const std::string& Filename, bool Global);
+    void Load();
+
+  private:
+    std::string Config;
+  };
+
+  class EnvLoader final : public FEXCore::Config::Layer {
+  public:
+    explicit EnvLoader(char *const _envp[]);
+    void Load() override;
+
+  private:
+    char *const *envp;
+  };
+
+  static const std::map<std::string, FEXCore::Config::ConfigOption, std::less<>> ConfigLookup = {{
+#define OPT_BASE(type, group, enum, json, default) {#json, FEXCore::Config::ConfigOption::CONFIG_##enum},
+#include <FEXCore/Config/ConfigValues.inl>
+  }};
+  static const std::vector<std::pair<const char*, FEXCore::Config::ConfigOption>> EnvConfigLookup = {{
+#define OPT_BASE(type, group, enum, json, default) {"FEX_" #enum, FEXCore::Config::ConfigOption::CONFIG_##enum},
+#include <FEXCore/Config/ConfigValues.inl>
+  }};
+
+  OptionMapper::OptionMapper(FEXCore::Config::LayerType Layer)
+    : FEXCore::Config::Layer(Layer) {
+  }
+
+  void OptionMapper::MapNameToOption(const char *ConfigName, const char *ConfigString) {
+    auto it = ConfigLookup.find(ConfigName);
+    if (it != ConfigLookup.end()) {
+      Set(it->second, ConfigString);
+    }
+  }
+
+  MainLoader::MainLoader()
+    : FEXCore::Config::OptionMapper(FEXCore::Config::LayerType::LAYER_MAIN)
+    , Config{FEXCore::Config::GetConfigFileLocation()} {
+  }
+
+  MainLoader::MainLoader(std::string ConfigFile)
+    : FEXCore::Config::OptionMapper(FEXCore::Config::LayerType::LAYER_MAIN)
+    , Config{std::move(ConfigFile)} {
+  }
+
+  void MainLoader::Load() {
+    JSON::LoadJSonConfig(Config, [this](const char *Name, const char *ConfigString) {
+      MapNameToOption(Name, ConfigString);
+    });
+  }
+
+  AppLoader::AppLoader(const std::string& Filename, bool Global)
+    : FEXCore::Config::OptionMapper(Global ? FEXCore::Config::LayerType::LAYER_GLOBAL_APP : FEXCore::Config::LayerType::LAYER_LOCAL_APP) {
+    Config = FEXCore::Config::GetApplicationConfig(Filename, Global);
+
+    // Immediately load so we can reload the meta layer
+    Load();
+  }
+
+  void AppLoader::Load() {
+    JSON::LoadJSonConfig(Config, [this](const char *Name, const char *ConfigString) {
+      MapNameToOption(Name, ConfigString);
+    });
+  }
+
+  EnvLoader::EnvLoader(char *const _envp[])
+    : FEXCore::Config::Layer(FEXCore::Config::LayerType::LAYER_ENVIRONMENT)
+    , envp {_envp} {
+  }
+
+  void EnvLoader::Load() {
+    std::unordered_map<std::string_view, std::string_view> EnvMap;
+
+    for(const char *const *pvar=envp; pvar && *pvar; pvar++) {
+      std::string_view Var(*pvar);
+      size_t pos = Var.rfind('=');
+      if (std::string::npos == pos)
+        continue;
+
+      std::string_view Ident = Var.substr(0,pos);
+      std::string_view Value = Var.substr(pos+1);
+      EnvMap[Ident]=Value;
+    }
+
+    std::function GetVar = [=](const std::string_view id)  -> std::optional<std::string_view> {
+      if (EnvMap.find(id) != EnvMap.end())
+        return EnvMap.at(id);
+
+      // If envp[] was empty, search using std::getenv()
+      const char* vs = std::getenv(id.data());
+      if (vs) {
+        return vs;
+      }
+      else {
+        return std::nullopt;
+      }
+    };
+
+    std::optional<std::string_view> Value;
+
+    for (auto &it : EnvConfigLookup) {
+      if ((Value = GetVar(it.first)).has_value()) {
+        Set(it.second, std::string(*Value));
+      }
+    }
+  }
+
+  std::unique_ptr<FEXCore::Config::Layer> CreateMainLayer(std::string const *File) {
+    if (File) {
+      return std::make_unique<FEXCore::Config::MainLoader>(*File);
+    }
+    else {
+      return std::make_unique<FEXCore::Config::MainLoader>();
+    }
+  }
+
+  std::unique_ptr<FEXCore::Config::Layer> CreateAppLayer(const std::string& Filename, bool Global) {
+    return std::make_unique<FEXCore::Config::AppLoader>(Filename, Global);
+  }
+
+  std::unique_ptr<FEXCore::Config::Layer> CreateEnvironmentLayer(char *const _envp[]) {
+    return std::make_unique<FEXCore::Config::EnvLoader>(_envp);
+  }
 }
 

External/FEXCore/Source/Interface/Config/Config.json

@@ -77,7 +77,14 @@
         "Default": "",
         "ShortArg": "k",
         "Desc": [
-          "A json file specifying where to overlay the thunks."
+          "A json file specifying where to overlay the thunks.",
+          "This can be a filesystem path",
+          "\teg: ~/MyThunkConfig.json",
+          "Or this can be a named of a Thunk config file",
+          "If the named config file exists in the FEX data folder folder the it will use that one",
+          "\teg: $HOME/.fex-emu/ThunkConfigs/<ThunkConfig name>",
+          "Or if you have XDG_DATA_HOME the config will search in that directory",
+          "\teg: $XDG_DATA_HOME/.fex-emu/ThunkConfigs/<ThunkConfig name>"
         ]
       },
       "Env": {
@@ -130,6 +137,13 @@
           "Disables optimizations passes for debugging."
         ]
       },
+      "SRA": {
+        "Type": "bool",
+        "Default": "true",
+        "Desc": [
+          "Set to false to disable Static Register Allocation"
+        ]
+      },
       "Force32BitAllocator": {
         "Type": "bool",
         "Default": "false",
@@ -207,6 +221,14 @@
           "Makes TSO operations even more strict.",
           "Forces vector loadstores to also become atomic."
         ]
+      },
+      "StallProcess": {
+        "Type": "bool",
+        "Default": "false",
+        "Desc": [
+          "Forces a process to stall out on initialization",
+          "Useful for a process that keeps restarting and doesn't work"
+        ]
       }
     },
     "Misc": {

External/FEXCore/Source/Interface/Context/Context.cpp

@@ -2,10 +2,20 @@
 #include "Interface/Context/Context.h"
 #include "Interface/Core/Core.h"
 #include "Interface/Core/OpcodeDispatcher.h"
+#include "Interface/Core/X86Tables/X86Tables.h"
 
-#include <FEXCore/Config/Config.h>
 #include <FEXCore/Core/CoreState.h>
-#include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/Core/Context.h>
+#include <FEXCore/Core/CPUID.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include "FEXCore/Debug/InternalThreadState.h"
+
+#include <string.h>
+#include <utility>
+
+namespace FEXCore::HLE {
+  class SyscallVisitor;
+}
 
 namespace FEXCore::Context {
   void InitializeStaticTables(OperatingMode Mode) {
@@ -14,6 +24,10 @@ namespace FEXCore::Context {
     IR::InstallOpcodeHandlers(Mode);
   }
 
+  void ShutdownStaticTables() {
+    FEXCore::Paths::ShutdownPaths();
+  }
+
   FEXCore::Context::Context *CreateNewContext() {
     return new FEXCore::Context::Context{};
   }
@@ -29,16 +43,15 @@ namespace FEXCore::Context {
     delete CTX;
   }
 
-  bool InitCore(FEXCore::Context::Context *CTX, FEXCore::CodeLoader *Loader) {
+  FEXCore::Core::InternalThreadState* InitCore(FEXCore::Context::Context *CTX, FEXCore::CodeLoader *Loader) {
     return CTX->InitCore(Loader);
   }
 
-  void SetExitHandler(FEXCore::Context::Context *CTX,
-      std::function<void(uint64_t ThreadId, FEXCore::Context::ExitReason)> handler) {
-    CTX->CustomExitHandler = handler;
+  void SetExitHandler(FEXCore::Context::Context *CTX, ExitHandler handler) {
+    CTX->CustomExitHandler = std::move(handler);
   }
 
-  std::function<void(uint64_t ThreadId, FEXCore::Context::ExitReason)> GetExitHandler(FEXCore::Context::Context *CTX) {
+  ExitHandler GetExitHandler(FEXCore::Context::Context *CTX) {
     return CTX->CustomExitHandler;
   }
 
@@ -50,8 +63,8 @@ namespace FEXCore::Context {
     CTX->Step();
   }
 
-  void CompileRIP(FEXCore::Context::Context *CTX, uint64_t GuestRIP) {
-    CTX->CompileBlock(CTX->ParentThread->CurrentFrame, GuestRIP);
+  void CompileRIP(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP) {
+    Thread->CTX->CompileBlock(Thread->CurrentFrame, GuestRIP);
   }
 
   FEXCore::Context::ExitReason RunUntilExit(FEXCore::Context::Context *CTX) {
@@ -97,22 +110,26 @@ namespace FEXCore::Context {
   void RegisterExternalSyscallVisitor(FEXCore::Context::Context *CTX, [[maybe_unused]] uint64_t Syscall, [[maybe_unused]] FEXCore::HLE::SyscallVisitor *Visitor) {
   }
 
-  void HandleCallback(FEXCore::Context::Context *CTX, uint64_t RIP) {
-    CTX->HandleCallback(RIP);
+  void HandleCallback(FEXCore::Context::Context *CTX, FEXCore::Core::InternalThreadState *Thread, uint64_t RIP) {
+    CTX->HandleCallback(Thread, RIP);
   }
 
-  void RegisterHostSignalHandler(FEXCore::Context::Context *CTX, int Signal, HostSignalDelegatorFunction Func) {
-      CTX->RegisterHostSignalHandler(Signal, Func);
+  void RegisterHostSignalHandler(FEXCore::Context::Context *CTX, int Signal, HostSignalDelegatorFunction Func, bool Required) {
+      CTX->RegisterHostSignalHandler(Signal, Func, Required);
   }
 
-  void RegisterFrontendHostSignalHandler(FEXCore::Context::Context *CTX, int Signal, HostSignalDelegatorFunction Func) {
-    CTX->RegisterFrontendHostSignalHandler(Signal, Func);
+  void RegisterFrontendHostSignalHandler(FEXCore::Context::Context *CTX, int Signal, HostSignalDelegatorFunction Func, bool Required) {
+    CTX->RegisterFrontendHostSignalHandler(Signal, Func, Required);
   }
 
   FEXCore::Core::InternalThreadState* CreateThread(FEXCore::Context::Context *CTX, FEXCore::Core::CPUState *NewThreadState, uint64_t ParentTID) {
     return CTX->CreateThread(NewThreadState, ParentTID);
   }
 
+  void ExecutionThread(FEXCore::Context::Context *CTX, FEXCore::Core::InternalThreadState *Thread) {
+    return CTX->ExecutionThread(Thread);
+  }
+
   void InitializeThread(FEXCore::Context::Context *CTX, FEXCore::Core::InternalThreadState *Thread) {
     return CTX->InitializeThread(Thread);
   }
@@ -149,8 +166,12 @@ namespace FEXCore::Context {
     CTX->AOTIRLoader = CacheReader;
   }
 
-  bool WriteAOTIR(FEXCore::Context::Context *CTX, std::function<std::unique_ptr<std::ostream>(const std::string&)> CacheWriter) {
-    return CTX->WriteAOTIRCache(CacheWriter);
+  void SetAOTIRWriter(FEXCore::Context::Context *CTX, std::function<std::unique_ptr<std::ostream>(const std::string&)> CacheWriter) {
+    CTX->AOTIRWriter = CacheWriter;
+  }
+
+  void FinalizeAOTIRCache(FEXCore::Context::Context *CTX) {
+    CTX->FinalizeAOTIRCache();
   }
 
   void WriteFilesWithCode(FEXCore::Context::Context *CTX, std::function<void(const std::string& fileid, const std::string& filename)> Writer) {

External/FEXCore/Source/Interface/Context/Context.h

@@ -1,44 +1,47 @@
 #pragma once
-#include "Common/JitSymbols.h"
+
 #include "Interface/Core/CPUID.h"
-#include "Interface/Core/Frontend.h"
 #include "Interface/Core/HostFeatures.h"
-#include "Interface/Core/InternalThreadState.h"
 #include "Interface/Core/X86HelperGen.h"
-#include "Interface/IR/PassManager.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
 #include <FEXCore/Config/Config.h>
-#include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Core/Context.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <FEXCore/Utils/CompilerDefs.h>
 #include <FEXCore/Utils/Event.h>
 #include <stdint.h>
 
+#include <atomic>
+#include <condition_variable>
 #include <functional>
 #include <istream>
 #include <map>
 #include <memory>
 #include <mutex>
-#include <optional>
-#include <ostream>
-#include <set>
+#include <shared_mutex>
+#include <stddef.h>
+#include <string>
 #include <unordered_map>
+#include <queue>
+#include <vector>
 
 namespace FEXCore {
+class CodeLoader;
 class ThunkHandler;
-class BlockSamplingData;
 class GdbServer;
-class SiganlDelegator;
 
 namespace CPU {
   class Arm64JITCore;
   class X86JITCore;
 }
 namespace HLE {
+struct SyscallArguments;
 class SyscallHandler;
 }
 }
 
 namespace FEXCore::IR {
-  class RegisterAllocationPass;
   class RegisterAllocationData;
   class IRListView;
 namespace Validation {
@@ -52,14 +55,6 @@ namespace FEXCore::Context {
     MODE_SINGLESTEP = 1,
   };
 
-  struct AOTIRCaptureCacheEntry {
-    uint64_t start;
-    uint64_t len;
-    uint64_t crc;
-    IR::IRListView *IR;
-    IR::RegisterAllocationData *RAData;
-  };
-
   struct AOTIRInlineEntry {
     uint64_t GuestHash;
     uint64_t GuestLength;
@@ -67,8 +62,8 @@ namespace FEXCore::Context {
     /* RAData followed by IRData */
     uint8_t InlineData[0];
 
-      IR::RegisterAllocationData *GetRAData();
-      IR::IRListView *GetIRData();
+    IR::RegisterAllocationData *GetRAData();
+    IR::IRListView *GetIRData();
   };
 
   struct AOTIRInlineIndexEntry {
@@ -85,6 +80,13 @@ namespace FEXCore::Context {
     AOTIRInlineEntry *GetInlineEntry(uint64_t DataOffset);
   };
 
+  struct AOTIRCaptureCacheEntry {
+    std::unique_ptr<std::ostream> Stream;
+    std::map<uint64_t, uint64_t> Index;
+
+    void AppendAOTIRCaptureCache(uint64_t GuestRIP, uint64_t Start, uint64_t Length, uint64_t Hash, FEXCore::IR::IRListView *IRList, FEXCore::IR::RegisterAllocationData *RAData);
+  };
+
   struct Context {
     friend class FEXCore::HLE::SyscallHandler;
   #ifdef JIT_ARM64
@@ -119,9 +121,10 @@ namespace FEXCore::Context {
       FEX_CONFIG_OPT(RootFSPath, ROOTFS);
       FEX_CONFIG_OPT(ThunkHostLibsPath, THUNKHOSTLIBS);
       FEX_CONFIG_OPT(DumpIR, DUMPIR);
+      FEX_CONFIG_OPT(StaticRegisterAllocation, SRA);
     } Config;
 
-    using IntCallbackReturn =  __attribute__((naked)) void(*)(FEXCore::Core::InternalThreadState *Thread, volatile void *Host_RSP);
+    using IntCallbackReturn =  FEX_NAKED void(*)(FEXCore::Core::InternalThreadState *Thread, volatile void *Host_RSP);
     IntCallbackReturn InterpreterCallbackReturn;
 
     FEXCore::HostFeatures HostFeatures;
@@ -144,7 +147,7 @@ namespace FEXCore::Context {
     std::unique_ptr<FEXCore::ThunkHandler> ThunkHandler;
 
     CustomCPUFactoryType CustomCPUFactory;
-    std::function<void(uint64_t ThreadId, FEXCore::Context::ExitReason)> CustomExitHandler;
+    FEXCore::Context::ExitHandler CustomExitHandler;
 
     struct AOTIRCacheEntry {
       AOTIRInlineIndex *Array;
@@ -154,7 +157,8 @@ namespace FEXCore::Context {
 
     std::unordered_map<std::string, AOTIRCacheEntry> AOTIRCache;
     std::function<int(const std::string&)> AOTIRLoader;
-    std::unordered_map<std::string, std::map<uint64_t, AOTIRCaptureCacheEntry>> AOTIRCaptureCache;
+    std::function<std::unique_ptr<std::ostream>(const std::string&)> AOTIRWriter;
+    std::unordered_map<std::string, AOTIRCaptureCacheEntry> AOTIRCaptureCache;
 
     struct AddrToFileEntry {
       uint64_t Start;
@@ -179,9 +183,9 @@ namespace FEXCore::Context {
     Context();
     ~Context();
 
-    bool InitCore(FEXCore::CodeLoader *Loader);
+    FEXCore::Core::InternalThreadState* InitCore(FEXCore::CodeLoader *Loader);
     FEXCore::Context::ExitReason RunUntilExit();
-    int GetProgramStatus();
+    int GetProgramStatus() const;
     bool IsPaused() const { return !Running; }
     void Pause();
     void Run();
@@ -192,12 +196,12 @@ namespace FEXCore::Context {
     void StopThread(FEXCore::Core::InternalThreadState *Thread);
     void SignalThread(FEXCore::Core::InternalThreadState *Thread, FEXCore::Core::SignalEvent Event);
 
-    bool GetGdbServerStatus() { return (bool)DebugServer; }
+    bool GetGdbServerStatus() const { return DebugServer != nullptr; }
     void StartGdbServer();
     void StopGdbServer();
-    void HandleCallback(uint64_t RIP);
-    void RegisterHostSignalHandler(int Signal, HostSignalDelegatorFunction Func);
-    void RegisterFrontendHostSignalHandler(int Signal, HostSignalDelegatorFunction Func);
+    void HandleCallback(FEXCore::Core::InternalThreadState *Thread, uint64_t RIP);
+    void RegisterHostSignalHandler(int Signal, HostSignalDelegatorFunction Func, bool Required);
+    void RegisterFrontendHostSignalHandler(int Signal, HostSignalDelegatorFunction Func, bool Required);
 
     static void RemoveCodeEntry(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP);
 
@@ -213,16 +217,35 @@ namespace FEXCore::Context {
     bool GetDebugDataForRIP(uint64_t RIP, FEXCore::Core::DebugData *Data);
     bool FindHostCodeForRIP(uint64_t RIP, uint8_t **Code);
 
-    std::tuple<FEXCore::IR::IRListView *, FEXCore::IR::RegisterAllocationData *, uint64_t, uint64_t, uint64_t, uint64_t> GenerateIR(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP);
+    struct GenerateIRResult {
+      FEXCore::IR::IRListView* IRList;
+      // User's responsibility to deallocate this.
+      FEXCore::IR::RegisterAllocationData* RAData;
+      uint64_t TotalInstructions;
+      uint64_t TotalInstructionsLength;
+      uint64_t StartAddr;
+      uint64_t Length;
+    };
+    [[nodiscard]] GenerateIRResult GenerateIR(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP);
 
-    std::tuple<void *, FEXCore::IR::IRListView *, FEXCore::Core::DebugData *, FEXCore::IR::RegisterAllocationData *, bool, uint64_t, uint64_t> CompileCode(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP);
+    struct CompileCodeResult {
+      void* CompiledCode;
+      FEXCore::IR::IRListView* IRData;
+      FEXCore::Core::DebugData* DebugData;
+      // User's responsibility to deallocate this.
+      FEXCore::IR::RegisterAllocationData* RAData;
+      bool GeneratedIR;
+      uint64_t StartAddr;
+      uint64_t Length;
+    };
+    [[nodiscard]] CompileCodeResult CompileCode(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP);
     uintptr_t CompileBlock(FEXCore::Core::CpuStateFrame *Frame, uint64_t GuestRIP);
     
     // same as CompileBlock, but aborts on failure
     void CompileBlockJit(FEXCore::Core::CpuStateFrame *Frame, uint64_t GuestRIP);
 
     bool LoadAOTIRCache(int streamfd);
-    bool WriteAOTIRCache(std::function<std::unique_ptr<std::ostream>(const std::string&)> CacheWriter);
+    void FinalizeAOTIRCache();
     void WriteFilesWithCode(std::function<void(const std::string& fileid, const std::string& filename)> Writer);
     
     // Used for thread creation from syscalls
@@ -236,7 +259,9 @@ namespace FEXCore::Context {
     void DestroyThread(FEXCore::Core::InternalThreadState *Thread);
     void CleanupAfterFork(FEXCore::Core::InternalThreadState *ExceptForThread);
 
-    std::vector<FEXCore::Core::InternalThreadState*> *const GetThreads() { return &Threads; }
+    std::vector<FEXCore::Core::InternalThreadState*>* GetThreads() { return &Threads; }
+
+    uint8_t GetGPRSize() const { return Config.Is64BitMode ? 8 : 4; }
 
     void AddNamedRegion(uintptr_t Base, uintptr_t Size, uintptr_t Offset, const std::string &filename);
     void RemoveNamedRegion(uintptr_t Base, uintptr_t Size);
@@ -257,7 +282,6 @@ namespace FEXCore::Context {
     void NotifyPause();
 
     void AddBlockMapping(FEXCore::Core::InternalThreadState *Thread, uint64_t Address, void *Ptr, uint64_t Start, uint64_t Length);
-
     FEXCore::CodeLoader *LocalLoader{};
 
     // Entry Cache
@@ -265,6 +289,14 @@ namespace FEXCore::Context {
     std::mutex ExitMutex;
     std::unique_ptr<GdbServer> DebugServer;
 
+    std::shared_mutex AOTIRCacheLock;
+    std::shared_mutex AOTIRCaptureCacheWriteoutLock;
+    std::atomic<bool> AOTIRCaptureCacheWriteoutFlusing;
+
+    std::queue<std::function<void()>> AOTIRCaptureCacheWriteoutQueue;
+    void AOTIRCaptureCacheWriteoutQueue_Flush();
+    void AOTIRCaptureCacheWriteoutQueue_Append(const std::function<void()> &fn);
+
     bool StartPaused = false;
     FEX_CONFIG_OPT(AppFilename, APP_FILENAME);
   };

External/FEXCore/Source/Interface/Core/ArchHelpers/Arm64.h

@@ -18,6 +18,36 @@ namespace FEXCore::ArchHelpers::Arm64 {
   constexpr uint32_t STLXP_MASK = 0xBF'E0'80'00;
   constexpr uint32_t STLXP_INST = 0x88'20'80'00;
 
+  constexpr uint32_t LDAXR_MASK = 0x3F'FF'FC'00;
+  constexpr uint32_t LDAXR_INST = 0x08'5F'FC'00;
+
+  constexpr uint32_t STLXR_MASK = 0x3F'E0'FC'00;
+  constexpr uint32_t STLXR_INST = 0x08'00'FC'00;
+
+  constexpr uint32_t CBNZ_MASK = 0x7F'00'00'00;
+  constexpr uint32_t CBNZ_INST = 0x35'00'00'00;
+
+  constexpr uint32_t ALU_OP_MASK = 0x7F'00'00'00;
+  constexpr uint32_t ADD_INST    = 0x0B'00'00'00;
+  constexpr uint32_t SUB_INST    = 0x4B'00'00'00;
+  constexpr uint32_t CMP_INST    = 0x6B'00'00'00;
+  constexpr uint32_t AND_INST    = 0x0A'00'00'00;
+  constexpr uint32_t OR_INST     = 0x2A'00'00'00;
+  constexpr uint32_t EOR_INST    = 0x4A'00'00'00;
+  
+  constexpr uint32_t CCMP_MASK   = 0x7F'E0'0C'10;
+  constexpr uint32_t CCMP_INST   = 0x7A'40'00'00;
+  
+  enum ExclusiveAtomicPairType {
+    TYPE_SWAP,
+    TYPE_ADD,
+    TYPE_SUB,
+    TYPE_AND,
+    TYPE_OR,
+    TYPE_EOR,
+    TYPE_NEG, // This is just a sub with zero. Need to know the differences
+  };
+
   // Load ops are 4 bits
   // Acquire and release bits are independent on the instruction
   constexpr uint32_t ATOMIC_ADD_OP  = 0b0000;
@@ -30,7 +60,29 @@ namespace FEXCore::ArchHelpers::Arm64 {
   constexpr uint32_t ATOMIC_UMIN_OP = 0b0111;
   constexpr uint32_t ATOMIC_SWAP_OP = 0b1000;
 
+  constexpr uint32_t REGISTER_MASK = 0b11111;
+  constexpr uint32_t RD_OFFSET = 0;
+  constexpr uint32_t RN_OFFSET = 5;
+  constexpr uint32_t RM_OFFSET = 16;
+
+  inline uint32_t GetRdReg(uint32_t Instr) {
+    return (Instr >> RD_OFFSET) & REGISTER_MASK;
+  }
+
+  inline uint32_t GetRnReg(uint32_t Instr) {
+    return (Instr >> RN_OFFSET) & REGISTER_MASK;
+  }
+
+  inline uint32_t GetRmReg(uint32_t Instr) {
+    return (Instr >> RM_OFFSET) & REGISTER_MASK;
+  }
+
+  bool HandleAtomicLoad(void *_ucontext, void *_info, uint32_t Instr);
+  bool HandleAtomicStore(void *_ucontext, void *_info, uint32_t Instr);
+  bool HandleAtomicLoad128(void *_ucontext, void *_info, uint32_t Instr);
+  uint64_t HandleAtomicLoadstoreExclusive(void *_ucontext, void *_info);
   bool HandleCASPAL(void *_ucontext, void *_info, uint32_t Instr);
+  uint64_t HandleCASPAL_ARMv8(void *_ucontext, void *_info, uint32_t Instr);
   bool HandleCASAL(void *_ucontext, void *_info, uint32_t Instr);
   bool HandleAtomicMemOp(void *_ucontext, void *_info, uint32_t Instr);
 }

External/FEXCore/Source/Interface/Core/ArchHelpers/Arm64Emitter.cpp

@@ -4,6 +4,11 @@
 #include <FEXCore/Core/CoreState.h>
 
 #include "aarch64/cpu-aarch64.h"
+#include "cpu-features.h"
+#include "aarch64/instructions-aarch64.h"
+#include "utils-vixl.h"
+
+#include <tuple>
 
 namespace FEXCore::CPU {
 #define STATE x28
@@ -143,22 +148,26 @@ void Arm64Emitter::PopCalleeSavedRegisters() {
 
 
 void Arm64Emitter::SpillStaticRegs() {
-  for (size_t i = 0; i < SRA64.size(); i+=2) {
-      stp(SRA64[i], SRA64[i+1], MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.gregs[i])));
-  }
+  if (StaticRegisterAllocation()) {
+    for (size_t i = 0; i < SRA64.size(); i+=2) {
+        stp(SRA64[i], SRA64[i+1], MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.gregs[i])));
+    }
 
-  for (size_t i = 0; i < SRAFPR.size(); i+=2) {
-    stp(SRAFPR[i].Q(), SRAFPR[i+1].Q(), MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.xmm[i][0])));
+    for (size_t i = 0; i < SRAFPR.size(); i+=2) {
+      stp(SRAFPR[i].Q(), SRAFPR[i+1].Q(), MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.xmm[i][0])));
+    }
   }
 }
 
 void Arm64Emitter::FillStaticRegs() {
-  for (size_t i = 0; i < SRA64.size(); i+=2) {
-    ldp(SRA64[i], SRA64[i+1], MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.gregs[i])));
-  }
+  if (StaticRegisterAllocation()) {
+    for (size_t i = 0; i < SRA64.size(); i+=2) {
+      ldp(SRA64[i], SRA64[i+1], MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.gregs[i])));
+    }
 
-  for (size_t i = 0; i < SRAFPR.size(); i+=2) {
-    ldp(SRAFPR[i].Q(), SRAFPR[i+1].Q(), MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.xmm[i][0])));
+    for (size_t i = 0; i < SRAFPR.size(); i+=2) {
+      ldp(SRAFPR[i].Q(), SRAFPR[i+1].Q(), MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, State.xmm[i][0])));
+    }
   }
 }
 
@@ -222,7 +231,7 @@ void Arm64Emitter::ResetStack() {
 }
 
 void Arm64Emitter::Align16B() {
-  uint64_t CurrentOffset = GetBuffer()->GetOffsetAddress<uint64_t>(GetCursorOffset());
+  uint64_t CurrentOffset = GetCursorAddress<uint64_t>();
   for (uint64_t i = (16 - (CurrentOffset & 0xF)); i != 0; i -= 4) {
       nop();
   }

External/FEXCore/Source/Interface/Core/ArchHelpers/Arm64Emitter.h

@@ -1,7 +1,16 @@
 #pragma once
 
 #include "aarch64/assembler-aarch64.h"
+#include "aarch64/constants-aarch64.h"
 #include "aarch64/cpu-aarch64.h"
+#include "aarch64/operands-aarch64.h"
+#include "platform-vixl.h"
+#include "FEXCore/Config/Config.h"
+
+#include <array>
+#include <stddef.h>
+#include <stdint.h>
+#include <utility>
 
 namespace FEXCore::CPU {
 using namespace vixl;
@@ -72,6 +81,8 @@ protected:
 
   uint32_t DCacheLineSize{};
   uint32_t ICacheLineSize{};
+
+  FEX_CONFIG_OPT(StaticRegisterAllocation, SRA);
 };
 
 }

External/FEXCore/Source/Interface/Core/ArchHelpers/Arm64_stubs.cpp

@@ -1,6 +1,7 @@
 #include "Interface/Core/ArchHelpers/Arm64.h"
 
 #include <FEXCore/Utils/LogManager.h>
+#include <stdint.h>
 
 namespace FEXCore::ArchHelpers::Arm64 {
 
@@ -23,4 +24,4 @@ bool HandleAtomicMemOp(void *_ucontext, void *_info, uint32_t Instr) {
 }
 #endif
 
-}
+}

External/FEXCore/Source/Interface/Core/ArchHelpers/MContext.h

@@ -18,6 +18,7 @@ struct X86ContextBackup {
   // RIP and RSP is stored in GPRs here
   uint64_t GPRs[23];
   FEXCore::x86_64::_libc_fpstate FPRState;
+  uint64_t sa_mask;
 
   // Guest state
   int Signal;
@@ -35,6 +36,7 @@ struct ArmContextBackup {
   uint32_t FPSR;
   uint32_t FPCR;
   __uint128_t FPRs[32];
+  uint64_t sa_mask;
 
   // Guest state
   int Signal;
@@ -44,6 +46,11 @@ struct ArmContextBackup {
   static constexpr int RedZoneSize = 0;
 };
 
+static inline ucontext_t* GetUContext(void* ucontext) {
+  ucontext_t* _context = (ucontext_t*)ucontext;
+  return _context;
+}
+
 static inline mcontext_t* GetMContext(void* ucontext) {
   ucontext_t* _context = (ucontext_t*)ucontext;
   return &_context->uc_mcontext;
@@ -102,6 +109,7 @@ using ContextBackup = ArmContextBackup;
 template <typename T>
 static inline void BackupContext(void* ucontext, T *Backup) {
   if constexpr (std::is_same<T, ArmContextBackup>::value) {
+    auto _ucontext = GetUContext(ucontext);
     auto _mcontext = GetMContext(ucontext);
 
     memcpy(&Backup->GPRs[0], &_mcontext->regs[0], 31 * sizeof(uint64_t));
@@ -115,6 +123,9 @@ static inline void BackupContext(void* ucontext, T *Backup) {
     Backup->FPSR = HostState->FPSR;
     Backup->FPCR = HostState->FPCR;
     memcpy(&Backup->FPRs[0], &HostState->FPRs[0], 32 * sizeof(__uint128_t));
+
+    // Save the signal mask so we can restore it
+    memcpy(&Backup->sa_mask, &_ucontext->uc_sigmask, sizeof(uint64_t));
   } else {
       ERROR_AND_DIE("Wrong context type"); // This must be a runtime error
   }
@@ -123,6 +134,7 @@ static inline void BackupContext(void* ucontext, T *Backup) {
 template <typename T>
 static inline void RestoreContext(void* ucontext, T *Backup) {
   if constexpr (std::is_same<T, ArmContextBackup>::value) {
+    auto _ucontext = GetUContext(ucontext);
    auto _mcontext = GetMContext(ucontext);
 
     HostFPRState *HostState = reinterpret_cast<HostFPRState*>(&_mcontext->__reserved[0]);
@@ -136,6 +148,9 @@ static inline void RestoreContext(void* ucontext, T *Backup) {
     ArchHelpers::Context::SetPc(ucontext, Backup->PrevPC);
     ArchHelpers::Context::SetSp(ucontext, Backup->PrevSP);
     memcpy(&_mcontext->regs[0], &Backup->GPRs[0], 31 * sizeof(uint64_t));
+
+    // Restore the signal mask now
+    memcpy(&_ucontext->uc_sigmask, &Backup->sa_mask, sizeof(uint64_t));
   } else {
       ERROR_AND_DIE("Wrong context type"); // This must be a runtime error
   }
@@ -181,6 +196,7 @@ using ContextBackup = X86ContextBackup;
 template <typename T>
 static inline void BackupContext(void* ucontext, T *Backup) {
   if constexpr (std::is_same<T, X86ContextBackup>::value) {
+    auto _ucontext = GetUContext(ucontext);
     auto _mcontext = GetMContext(ucontext);
 
     // Copy the GPRs
@@ -188,6 +204,9 @@ static inline void BackupContext(void* ucontext, T *Backup) {
     // Copy the FPRState
     memcpy(&Backup->FPRState, _mcontext->fpregs, sizeof(X86ContextBackup::FPRState));
     // XXX: Save 256bit and 512bit AVX register state
+
+    // Save the signal mask so we can restore it
+    memcpy(&Backup->sa_mask, &_ucontext->uc_sigmask, sizeof(uint64_t));
   } else {
     ERROR_AND_DIE("Wrong context type"); // This must be a runtime error
   }
@@ -196,12 +215,16 @@ static inline void BackupContext(void* ucontext, T *Backup) {
 template <typename T>
 static inline void RestoreContext(void* ucontext, T *Backup) {
   if constexpr (std::is_same<T, X86ContextBackup>::value) {
+    auto _ucontext = GetUContext(ucontext);
     auto _mcontext = GetMContext(ucontext);
 
     // Copy the GPRs
     memcpy(&_mcontext->gregs[0], &Backup->GPRs[0], sizeof(X86ContextBackup::GPRs));
     // Copy the FPRState
     memcpy(_mcontext->fpregs, &Backup->FPRState, sizeof(X86ContextBackup::FPRState));
+
+    // Restore the signal mask now
+    memcpy(&_ucontext->uc_sigmask, &Backup->sa_mask, sizeof(uint64_t));
   } else {
     ERROR_AND_DIE("Wrong context type"); // This must be a runtime error
   }
@@ -209,4 +232,4 @@ static inline void RestoreContext(void* ucontext, T *Backup) {
 
 #endif
 
-} // namespace FEXCore::ArchHelpers::Context
+} // namespace FEXCore::ArchHelpers::Context

External/FEXCore/Source/Interface/Core/BlockSamplingData.cpp

@@ -2,6 +2,7 @@
 #include <FEXCore/Utils/LogManager.h>
 #include <cstring>
 #include <fstream>
+#include <utility>
 
 namespace FEXCore {
   void BlockSamplingData::DumpBlockData() {

External/FEXCore/Source/Interface/Core/BlockSamplingData.h

@@ -1,6 +1,7 @@
 #pragma once
+
+#include <cstdint>
 #include <unordered_map>
-#include <stdint.h>
 
 namespace FEXCore {
 class BlockSamplingData {

External/FEXCore/Source/Interface/Core/CPUID.cpp

@@ -5,8 +5,11 @@ desc: Handles presented capability bits for guest cpu
 $end_info$
 */
 
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Core/CPUID.h>
 #include "Interface/Context/Context.h"
 #include "Interface/Core/CPUID.h"
+#include "Interface/Core/HostFeatures.h"
 #include "git_version.h"
 
 #include <cstring>
@@ -15,7 +18,26 @@ $end_info$
 #endif
 
 namespace FEXCore {
-//#define CPUID_AMD
+constexpr uint32_t SUPPORTS_AVX = 0;
+// #define CPUID_AMD
+#ifdef CPUID_AMD
+constexpr uint32_t FAMILY_IDENTIFIER =
+  0 |          // Stepping
+  (0xA << 4) | // Model
+  (0xF << 8) | // Family ID
+  (0 << 12) |  // Processor type
+  (0 << 16) |  // Extended model ID
+  (1 << 20);   // Extended family ID
+#else
+constexpr uint32_t FAMILY_IDENTIFIER =
+  0 |          // Stepping
+  (0x7 << 4) |   // Model
+  (0x6 << 8) | // Family ID
+  (0 << 12) |  // Processor type
+  (1 << 16) |  // Extended model ID
+  (0x0 << 20);   // Extended family ID
+#endif
+
 #ifdef _M_ARM_64
 static uint32_t GetCycleCounterFrequency() {
   uint64_t Result{};
@@ -38,7 +60,7 @@ static uint32_t GetCycleCounterFrequency() {
 }
 #endif
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_0h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_0h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
 
   // EBX, EDX, ECX become the manufacturer id string
@@ -57,29 +79,27 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_0h() {
 }
 
 // Processor Info and Features bits
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_01h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_01h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
+  uint32_t CoreCount = Cores();
+
+  Res.eax = FAMILY_IDENTIFIER;
 
-  Res.eax = 0 | // Stepping
-    (0 << 4) | // Model
-    (0xF << 8) | // Family ID
-    (0 << 12) | // Processor type
-    (0 << 16) | // Extended model ID
-    (0 << 20); // Extended family ID
   Res.ebx = 0 | // Brand index
     (8 << 8) | // Cache line size in bytes
-    (8 << 16) | // Number of addressable IDs for the logical cores in the physical CPU
+    (CoreCount << 16) | // Number of addressable IDs for the logical cores in the physical CPU
     (0 << 24); // Local APIC ID
+
   Res.ecx =
     (1 <<  0) | // SSE3
     (0 <<  1) | // PCLMULQDQ
     (1 <<  2) | // DS area supports 64bit layout
     (1 <<  3) | // MWait
-    (1 <<  4) | // DS-CPL
+    (0 <<  4) | // DS-CPL
     (0 <<  5) | // VMX
     (0 <<  6) | // SMX
     (0 <<  7) | // Intel SpeedStep
-    (0 <<  8) | // Thermal Monitor 2
+    (1 <<  8) | // Thermal Monitor 2
     (1 <<  9) | // SSSE3
     (0 << 10) | // L1 context ID
     (0 << 11) | // Silicon debug
@@ -89,8 +109,8 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_01h() {
     (0 << 15) | // Perfmon and debug capability
     (0 << 16) | // Reserved
     (0 << 17) | // Process-context identifiers
-    (1 << 18) | // Prefetching from memory mapped device
-    (0 << 19) | // SSE4.1
+    (0 << 18) | // Prefetching from memory mapped device
+    (1 << 19) | // SSE4.1
     (0 << 20) | // SSE4.2
     (0 << 21) | // X2APIC
     (1 << 22) | // MOVBE
@@ -99,7 +119,7 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_01h() {
     (CTX->HostFeatures.SupportsAES << 25) | // AES
     (0 << 26) | // XSAVE
     (0 << 27) | // OSXSAVE
-    (0 << 28) | // AVX
+    (SUPPORTS_AVX << 28) | // AVX
     (0 << 29) | // F16C
     (0 << 30) | // RDRAND
     (0 << 31);  // Hypervisor always returns zero
@@ -132,16 +152,16 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_01h() {
     (1 << 24) | // FXSAVE/FXRSTOR
     (1 << 25) | // SSE
     (1 << 26) | // SSE2
-    (1 << 27) | // Self Snoop
+    (0 << 27) | // Self Snoop
     (1 << 28) | // Max APIC IDs reserved field is valid
     (1 << 29) | // Thermal monitor
     (0 << 30) | // Reserved
-    (1 << 31);  // Pending break enable
+    (0 << 31);  // Pending break enable
   return Res;
 }
 
 // 2: Cache and TLB information
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_02h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_02h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
 
   // returns default values from i7 model 1Ah
@@ -165,124 +185,286 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_02h() {
   return Res;
 }
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_06h() {
+// 4: Deterministic cache parameters for each level
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_04h(uint32_t Leaf) {
+  FEXCore::CPUID::FunctionResults Res{};
+  constexpr uint32_t CacheType_Data = 1;
+  constexpr uint32_t CacheType_Instruction = 2;
+  constexpr uint32_t CacheType_Unified = 3;
+
+  if (Leaf == 0) {
+    // Report L1D
+    uint32_t CoreCount = Cores() - 1;
+
+    Res.eax = CacheType_Data | // Cache type
+      (0b001 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (0 << 14) |         // Maximum number of addressable IDs for logical processors sharing this cache (With SMT this would be 1)
+      (CoreCount << 26);  // Maximum number of addressable IDs for processor cores in the physical package
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 32KB
+    Res.ecx = 63; // Number of sets - 1 : Claiming 64 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1) | // Cache inclusiveness - Includes lower caches
+      (0 << 2);  // Complex cache indexing - 0: Direct, 1: Complex
+  }
+  else if (Leaf == 1) {
+    // Report L1I
+    uint32_t CoreCount = Cores() - 1;
+
+    Res.eax = CacheType_Instruction | // Cache type
+      (0b001 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (0 << 14) |         // Maximum number of addressable IDs for logical processors sharing this cache (With SMT this would be 1)
+      (CoreCount << 26); // Maximum number of addressable IDs for processor cores in the physical package
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 32KB
+    Res.ecx = 63; // Number of sets - 1 : Claiming 64 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1) | // Cache inclusiveness - Includes lower caches
+      (0 << 2);  // Complex cache indexing - 0: Direct, 1: Complex
+  }
+  else if (Leaf == 2) {
+    // Report L2
+    uint32_t CoreCount = Cores() - 1;
+
+    Res.eax = CacheType_Unified | // Cache type
+      (0b010 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (0 << 14) |         // Maximum number of addressable IDs for logical processors sharing this cache
+      (CoreCount << 26);  // Maximum number of addressable IDs for processor cores in the physical package
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 512KB
+    Res.ecx = 0x3FF; // Number of sets - 1 : Claiming 1024 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1) | // Cache inclusiveness - Includes lower caches
+      (0 << 2);  // Complex cache indexing - 0: Direct, 1: Complex
+  }
+  else if (Leaf == 3) {
+    // Report L3
+    uint32_t CoreCount = Cores() - 1;
+
+    Res.eax = CacheType_Unified | // Cache type
+      (0b011 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (CoreCount << 14) | // Maximum number of addressable IDs for logical processors sharing this cache
+      (CoreCount << 26);  // Maximum number of addressable IDs for processor cores in the physical package
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 8MB
+    Res.ecx = 0x4000; // Number of sets - 1 : Claiming 16384 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1) | // Cache inclusiveness - Includes lower caches
+      (1 << 2);  // Complex cache indexing - 0: Direct, 1: Complex
+  }
+
+  return Res;
+}
+
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_06h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   Res.eax = (1 << 2); // Always running APIC
   Res.ecx = (0 << 3); // Intel performance energy bias preference (EPB)
   return Res;
 }
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_07h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_07h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
+  if (Leaf == 0) {
+    // Number of subfunctions
+    Res.eax = 0x0;
+    Res.ebx =
+      (1 <<  0) | // FS/GS support
+      (0 <<  1) | // TSC adjust MSR
+      (0 <<  2) | // SGX
+      (0 <<  3) | // BMI1
+      (0 <<  4) | // Intel Hardware Lock Elison
+      (0 <<  5) | // AVX2 support
+      (1 <<  6) | // FPU data pointer updated only on exception
+      (1 <<  7) | // SMEP support
+      (0 <<  8) | // BMI2
+      (0 <<  9) | // Enhanced REP MOVSB/STOSB
+      (1 << 10) | // INVPCID for system software control of process-context
+      (0 << 11) | // Restricted transactional memory
+      (0 << 12) | // Intel resource directory technology Monitoring
+      (1 << 13) | // Deprecates FPU CS and DS
+      (0 << 14) | // Intel MPX
+      (0 << 15) | // Intel Resource Directory Technology Allocation
+      (0 << 16) | // Reserved
+      (0 << 17) | // Reserved
+      (0 << 18) | // RDSEED
+      (0 << 19) | // ADCX and ADOX instructions
+      (0 << 20) | // SMAP Supervisor mode access prevention and CLAC/STAC instructions
+      (0 << 21) | // Reserved
+      (0 << 22) | // Reserved
+      (0 << 23) | // CLFLUSHOPT instruction
+      (0 << 24) | // CLWB instruction
+      (0 << 25) | // Intel processor trace
+      (0 << 26) | // Reserved
+      (0 << 27) | // Reserved
+      (0 << 28) | // Reserved
+      (0 << 29) | // SHA instructions
+      (0 << 30) | // Reserved
+      (0 << 31);  // Reserved
 
-  // Number of subfunctions
-  Res.eax = 0x0;
-  Res.ebx =
-    (1 <<  0) | // FS/GS support
-    (0 <<  1) | // TSC adjust MSR
-    (0 <<  2) | // SGX
-    (0 <<  3) | // BMI1
-    (0 <<  4) | // Intel Hardware Lock Elison
-    (0 <<  5) | // AVX2 support
-    (1 <<  6) | // FPU data pointer updated only on exception
-    (1 <<  7) | // SMEP support
-    (0 <<  8) | // BMI2
-    (0 <<  9) | // Enhanced REP MOVSB/STOSB
-    (1 << 10) | // INVPCID for system software control of process-context
-    (0 << 11) | // Restricted transactional memory
-    (0 << 12) | // Intel resource directory technology Monitoring
-    (1 << 13) | // Deprecates FPU CS and DS
-    (0 << 14) | // Intel MPX
-    (0 << 15) | // Intel Resource Directory Technology Allocation
-    (0 << 16) | // Reserved
-    (0 << 17) | // Reserved
-    (0 << 18) | // RDSEED
-    (0 << 19) | // ADCX and ADOX instructions
-    (0 << 20) | // SMAP Supervisor mode access prevention and CLAC/STAC instructions
-    (0 << 21) | // Reserved
-    (0 << 22) | // Reserved
-    (0 << 23) | // CLFLUSHOPT instruction
-    (0 << 24) | // CLWB instruction
-    (0 << 25) | // Intel processor trace
-    (0 << 26) | // Reserved
-    (0 << 27) | // Reserved
-    (0 << 28) | // Reserved
-    (0 << 29) | // SHA instructions
-    (0 << 30) | // Reserved
-    (0 << 31);  // Reserved
+    Res.ecx =
+      (1 <<  0) | // PREFETCHWT1
+      (0 <<  1) | // AVX512VBMI
+      (0 <<  2) | // Usermode instruction prevention
+      (0 <<  3) | // Protection keys for user mode pages
+      (0 <<  4) | // OS protection keys
+      (0 <<  5) | // waitpkg
+      (0 <<  6) | // AVX512_VBMI2
+      (0 <<  7) | // CET shadow stack
+      (0 <<  8) | // GFNI
+      (0 <<  9) | // VAES
+      (0 << 10) | // VPCLMULQDQ
+      (0 << 11) | // AVX512_VNNI
+      (0 << 12) | // AVX512_BITALG
+      (0 << 13) | // Intel Total Memory Encryption
+      (0 << 14) | // AVX512_VPOPCNTDQ
+      (0 << 15) | // Reserved
+      (0 << 16) | // 5 Level page tables
+      (0 << 17) | // MPX MAWAU
+      (0 << 18) | // MPX MAWAU
+      (0 << 19) | // MPX MAWAU
+      (0 << 20) | // MPX MAWAU
+      (0 << 21) | // MPX MAWAU
+      (0 << 22) | // RDPID Read Processor ID
+      (0 << 23) | // Reserved
+      (0 << 24) | // Reserved
+      (0 << 25) | // CLDEMOTE
+      (0 << 26) | // Reserved
+      (0 << 27) | // MOVDIRI
+      (0 << 28) | // MOVDIR64B
+      (0 << 29) | // Reserved
+      (0 << 30) | // SGX Launch configuration
+      (0 << 31);  // Reserved
 
-  Res.ecx =
-    (1 <<  0) | // PREFETCHWT1
-    (0 <<  1) | // AVX512VBMI
-    (0 <<  2) | // Usermode instruction prevention
-    (0 <<  3) | // Protection keys for user mode pages
-    (1 <<  4) | // OS protection keys
-    (0 <<  5) | // waitpkg
-    (0 <<  6) | // AVX512_VBMI2
-    (0 <<  7) | // CET shadow stack
-    (0 <<  8) | // GFNI
-    (0 <<  9) | // VAES
-    (0 << 10) | // VPCLMULQDQ
-    (0 << 11) | // AVX512_VNNI
-    (0 << 12) | // AVX512_BITALG
-    (0 << 13) | // Intel Total Memory Encryption
-    (0 << 14) | // AVX512_VPOPCNTDQ
-    (0 << 15) | // Reserved
-    (0 << 16) | // 5 Level page tables
-    (0 << 17) | // MPX MAWAU
-    (0 << 18) | // MPX MAWAU
-    (0 << 19) | // MPX MAWAU
-    (0 << 20) | // MPX MAWAU
-    (0 << 21) | // MPX MAWAU
-    (0 << 22) | // RDPID Read Processor ID
-    (0 << 23) | // Reserved
-    (0 << 24) | // Reserved
-    (0 << 25) | // CLDEMOTE
-    (0 << 26) | // Reserved
-    (0 << 27) | // MOVDIRI
-    (0 << 28) | // MOVDIR64B
-    (0 << 29) | // Reserved
-    (0 << 30) | // SGX Launch configuration
-    (0 << 31);  // Reserved
-
-  Res.edx =
-    (0 <<  0) | // Reserved
-    (0 <<  1) | // Reserved
-    (0 <<  2) | // AVX512_4VNNIW
-    (0 <<  3) | // AVX512_4FMAPS
-    (0 <<  4) | // Fast Short Rep Mov
-    (0 <<  5) | // Reserved
-    (0 <<  6) | // Reserved
-    (0 <<  7) | // Reserved
-    (0 <<  8) | // AVX512_VP2INTERSECT
-    (0 <<  9) | // Reserved
-    (0 << 10) | // VERW clears CPU buffers
-    (0 << 11) | // Reserved
-    (0 << 12) | // Reserved
-    (0 << 13) | // Reserved
-    (0 << 14) | // SERIALIZE instruction
-    (0 << 15) | // Reserved
-    (0 << 16) | // Reserved
-    (0 << 17) | // Reserved
-    (0 << 18) | // Intel PCONFIG
-    (0 << 19) | // Intel Architectural LBR
-    (0 << 20) | // Intel CET
-    (0 << 21) | // Reserved
-    (0 << 22) | // Reserved
-    (0 << 23) | // Reserved
-    (0 << 24) | // Reserved
-    (0 << 25) | // Reserved
-    (0 << 26) | // Reserved
-    (0 << 27) | // Reserved
-    (0 << 28) | // L1D Flush
-    (0 << 29) | // Arch capabilities
-    (0 << 30) | // Reserved
-    (0 << 31);  // Reserved
+    Res.edx =
+      (0 <<  0) | // Reserved
+      (0 <<  1) | // Reserved
+      (0 <<  2) | // AVX512_4VNNIW
+      (0 <<  3) | // AVX512_4FMAPS
+      (0 <<  4) | // Fast Short Rep Mov
+      (0 <<  5) | // Reserved
+      (0 <<  6) | // Reserved
+      (0 <<  7) | // Reserved
+      (0 <<  8) | // AVX512_VP2INTERSECT
+      (0 <<  9) | // Reserved
+      (0 << 10) | // VERW clears CPU buffers
+      (0 << 11) | // Reserved
+      (0 << 12) | // Reserved
+      (0 << 13) | // Reserved
+      (0 << 14) | // SERIALIZE instruction
+      (0 << 15) | // Reserved
+      (0 << 16) | // Reserved
+      (0 << 17) | // Reserved
+      (0 << 18) | // Intel PCONFIG
+      (0 << 19) | // Intel Architectural LBR
+      (0 << 20) | // Intel CET
+      (0 << 21) | // Reserved
+      (0 << 22) | // Reserved
+      (0 << 23) | // Reserved
+      (0 << 24) | // Reserved
+      (0 << 25) | // Reserved
+      (0 << 26) | // Reserved
+      (0 << 27) | // Reserved
+      (0 << 28) | // L1D Flush
+      (0 << 29) | // Arch capabilities
+      (0 << 30) | // Reserved
+      (0 << 31);  // Reserved
+  }
 
   return Res;
 }
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_15h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_0Dh(uint32_t Leaf) {
+  // Leaf 0
+  FEXCore::CPUID::FunctionResults Res{};
+
+  uint32_t XFeatureSupportedSizeMax = SUPPORTS_AVX ? 0x0000'0340 : 0x0000'0240; // XFeatureEnabledSizeMax: Legacy Header + FPU/SSE + AVX
+  if (Leaf == 0) {
+    // XFeatureSupportedMask[31:0]
+    Res.eax =
+      (1 << 0) |            // X87 support
+      (1 << 1) |            // 128-bit SSE support
+      (SUPPORTS_AVX << 2) | // 256-bit AVX support
+      (0b00 << 3) |         // MPX State
+      (0b000 << 5) |        // AVX-512 state
+      (0 << 8) |            // "Used for IA32_XSS" ... Used for what?
+      (0 << 9);             // PKRU state
+
+    // EBX and ECX doesn't need to match if a feature is supported but not enabled
+    Res.ebx = XFeatureSupportedSizeMax;
+    Res.ecx = XFeatureSupportedSizeMax; // XFeatureSupportedSizeMax: Size in bytes of XSAVE/XRSTOR area
+
+    // XFeatureSupportedMask[63:32]
+    Res.edx = 0; // Upper 32-bits of XFeatureSupportedMask
+  }
+  else if (Leaf == 1) {
+    Res.eax =
+      (0 << 0) | // XSAVEOPT
+      (0 << 1) | // XSAVEC (and XRSTOR)
+      (0 << 2) | // XGETBV - XGETBV with ECX=1 supported
+      (0 << 3);  // XSAVES - XSAVES, XRSTORS, and IA32_XSS supported
+
+    // Same information as Leaf 0 for ebx
+    Res.ebx = XFeatureSupportedSizeMax;
+
+    // Lower supported 32bits of IA32_XSS MSR. IA32_XSS[n] can only be set to 1 if ECX[n] is 1
+    Res.ecx =
+      (0b0000'0000 << 0) | // Used for XCR0
+      (0 << 8) |           // PT state
+      (0 << 9);            // Used for XCR0
+
+    // Upper supported 32bits of IA32_XSS MSR. IA32_XSS[n+32] can only be set to 1 if EDX[n] is 1
+    // Entirely reserved atm
+    Res.edx = 0;
+  }
+  else if (Leaf == 2) {
+    Res.eax = SUPPORTS_AVX ? 0x0000'0100 : 0; // YmmSaveStateSize
+    Res.ebx = SUPPORTS_AVX ? 0x0000'0240 : 0; // YmmSaveStateOffset
+
+    // Reserved
+    Res.ecx = 0;
+    Res.edx = 0;
+  }
+  return Res;
+}
+
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_15h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   // TSC frequency = ECX * EBX / EAX
   uint32_t FrequencyHz = GetCycleCounterFrequency();
@@ -295,7 +477,7 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_15h() {
 }
 
 // Highest extended function implemented
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0000h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0000h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   Res.eax = 0x8000001F;
 
@@ -314,15 +496,10 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0000h() {
 }
 
 // Extended processor and feature bits
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0001h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0001h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
 
-  Res.eax = 0 | // Stepping
-    (0 << 4) | // Model
-    (0 << 8) | // Family ID
-    (0 << 12) | // Processor type
-    (0 << 16) | // Extended model ID
-    (0 << 20); // Extended family ID
+  Res.eax = FAMILY_IDENTIFIER;
 
   Res.ecx =
     (1 <<  0) | // LAHF/SAHF
@@ -347,13 +524,13 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0001h() {
     (0 << 19) | // Reserved
     (0 << 20) | // Reserved
     (0 << 21) | // Reserved
-    (1 << 22) | // Topology extensions support
-    (1 << 23) | // Core performance counter extensions
-    (1 << 24) | // NB performance counter extensions
+    (0 << 22) | // Topology extensions support
+    (0 << 23) | // Core performance counter extensions
+    (0 << 24) | // NB performance counter extensions
     (0 << 25) | // Reserved
     (0 << 26) | // Data breakpoints extensions
-    (1 << 27) | // Performance TSC
-    (1 << 28) | // L2 perf counter extensions
+    (0 << 27) | // Performance TSC
+    (0 << 28) | // L2 perf counter extensions
     (0 << 29) | // Reserved
     (0 << 30) | // Reserved
     (0 << 31);  // Reserved
@@ -385,7 +562,7 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0001h() {
     (1 << 23) | // MMX
     (1 << 24) | // FXSAVE/FXRSTOR
     (1 << 25) | // FXSAVE/FXRSTOR Optimizations
-    (1 << 26) | // 1 gigabit pages
+    (0 << 26) | // 1 gigabit pages
     (0 << 27) | // RDTSCP
     (0 << 28) | // Reserved
     (1 << 29) | // Long Mode
@@ -400,26 +577,26 @@ constexpr char ProcessorBrand[48] = {
 };
 
 //Processor brand string
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0002h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0002h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   memcpy(&Res, &ProcessorBrand[0], sizeof(FEXCore::CPUID::FunctionResults));
   return Res;
 }
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0003h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0003h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   memcpy(&Res, &ProcessorBrand[16], sizeof(FEXCore::CPUID::FunctionResults));
   return Res;
 }
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0004h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0004h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   memcpy(&Res, &ProcessorBrand[32], sizeof(FEXCore::CPUID::FunctionResults));
   return Res;
 }
 
 // L1 Cache and TLB identifiers
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0005h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0005h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
 
   // L1 TLB Information for 2MB and 4MB pages
@@ -454,7 +631,7 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0005h() {
 }
 
 // L2 Cache identifiers
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0006h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0006h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
 
   // L2 TLB Information for 2MB and 4MB pages
@@ -488,7 +665,7 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0006h() {
 }
 
 // Advanced power management
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0007h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0007h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   Res.eax = (1 << 2); // APIC timer not affected by p-state
   Res.edx =
@@ -497,7 +674,7 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0007h() {
 }
 
 // Virtual and physical address sizes
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0008h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0008h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   Res.eax =
     (48 << 0) | // PhysAddrSize = 48-bit
@@ -512,14 +689,14 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0008h() {
   uint32_t CoreCount = Cores() - 1;
   Res.ecx =
     (0 << 16) |       // PerfTscSize: Performance timestamp count size
-    (0 << 12) |       // ApicIdSize: Number of bits in ApicID
+    ((uint32_t)std::log2(CoreCount + 1) << 12) |       // ApicIdSize: Number of bits in ApicID
     (CoreCount << 0); // Count count subtract one
 
   return Res;
 }
 
 // TLB 1GB page identifiers
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0019h() {
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0019h(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   Res.eax =
     (0xF << 28) | // L1 DTLB associativity for 1GB pages
@@ -535,27 +712,128 @@ FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_0019h() {
   return Res;
 }
 
-FEXCore::CPUID::FunctionResults CPUIDEmu::Function_Reserved() {
+// Deterministic cache parameters for each level
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_8000_001Dh(uint32_t Leaf) {
+  // This is nearly a copy of CPUID function 4h
+  // There are some minor changes though
+
+  FEXCore::CPUID::FunctionResults Res{};
+  constexpr uint32_t CacheType_Data = 1;
+  constexpr uint32_t CacheType_Instruction = 2;
+  constexpr uint32_t CacheType_Unified = 3;
+
+  if (Leaf == 0) {
+    // Report L1D
+    Res.eax = CacheType_Data | // Cache type
+      (0b001 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (0 << 14);          // Maximum number of addressable IDs for logical processors sharing this cache (With SMT this would be 1)
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 32KB
+    Res.ecx = 63; // Number of sets - 1 : Claiming 64 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1);  // Cache inclusiveness - Includes lower caches
+  }
+  else if (Leaf == 1) {
+    // Report L1I
+    Res.eax = CacheType_Instruction | // Cache type
+      (0b001 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (0 << 14);          // Maximum number of addressable IDs for logical processors sharing this cache (With SMT this would be 1)
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 32KB
+    Res.ecx = 63; // Number of sets - 1 : Claiming 64 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1);  // Cache inclusiveness - Includes lower caches
+  }
+  else if (Leaf == 2) {
+    // Report L2
+    Res.eax = CacheType_Unified | // Cache type
+      (0b010 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (0 << 14);          // Maximum number of addressable IDs for logical processors sharing this cache
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 512KB
+    Res.ecx = 0x3FF; // Number of sets - 1 : Claiming 1024 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1);  // Cache inclusiveness - Includes lower caches
+  }
+  else if (Leaf == 3) {
+    // Report L3
+    uint32_t CoreCount = Cores() - 1;
+
+    Res.eax = CacheType_Unified | // Cache type
+      (0b011 << 5) |      // Cache level
+      (1 << 8)  |         // Self initializing cache level
+      (0 << 9)  |         // Fully associative
+      (CoreCount << 14);  // Maximum number of addressable IDs for logical processors sharing this cache
+
+    Res.ebx =
+      (63 << 0) | // Line Size - 1 : Claiming 64 byte
+      (0 << 12) | // Physical Line partitions
+      (7 << 22);  // Associativity - 1 : Claiming 8 way
+
+    // 8MB
+    Res.ecx = 0x4000; // Number of sets - 1 : Claiming 16384 sets
+
+    Res.edx =
+      (0 << 0) | // Write-back invalidate
+      (0 << 1);  // Cache inclusiveness - Includes lower caches
+  }
+
+  return Res;
+}
+
+FEXCore::CPUID::FunctionResults CPUIDEmu::Function_Reserved(uint32_t Leaf) {
   FEXCore::CPUID::FunctionResults Res{};
   return Res;
 }
 
 void CPUIDEmu::Init(FEXCore::Context::Context *ctx) {
   CTX = ctx;
-  RegisterFunction(0, std::bind(&CPUIDEmu::Function_0h, this));
-  RegisterFunction(1, std::bind(&CPUIDEmu::Function_01h, this));
-  RegisterFunction(2, std::bind(&CPUIDEmu::Function_02h, this));
+  using namespace std::placeholders;
+  RegisterFunction(0, std::bind(&CPUIDEmu::Function_0h, this, _1));
+  RegisterFunction(1, std::bind(&CPUIDEmu::Function_01h, this, _1));
+  RegisterFunction(2, std::bind(&CPUIDEmu::Function_02h, this, _1));
   // 3: Serial Number(previously), now reserved
-  // 4: Deterministic cache parameters for each level
+#ifndef CPUID_AMD
+  // Deterministic cache parameters for each level
+  RegisterFunction(0x4, std::bind(&CPUIDEmu::Function_04h, this, _1));
+#endif
   // 5: Monitor/mwait
   // Thermal and power management
-  RegisterFunction(6, std::bind(&CPUIDEmu::Function_06h, this));
+  RegisterFunction(6, std::bind(&CPUIDEmu::Function_06h, this, _1));
   // Extended feature flags
-  RegisterFunction(7, std::bind(&CPUIDEmu::Function_07h, this));
+  RegisterFunction(7, std::bind(&CPUIDEmu::Function_07h, this, _1));
   // 9: Direct Cache Access information
   // 0x0A: Architectural performance monitoring
   // 0x0B: Extended topology enumeration
   // 0x0D: Processor extended state enumeration
+  RegisterFunction(0x0D, std::bind(&CPUIDEmu::Function_0Dh, this, _1));
   // 0x0F: Intel RDT monitoring
   // 0x10: Intel RDT allocation enumeration
   // 0x12: Intel SGX capability enumeration
@@ -564,43 +842,47 @@ void CPUIDEmu::Init(FEXCore::Context::Context *ctx) {
 #ifndef CPUID_AMD
   // Timestamp counter information
   // Doesn't exist on AMD hardware
-  RegisterFunction(0x15, std::bind(&CPUIDEmu::Function_15h, this));
+  RegisterFunction(0x15, std::bind(&CPUIDEmu::Function_15h, this, _1));
 #endif
   // 0x16: Processor frequency information
   // 0x17: SoC vendor attribute enumeration
 
   // Largest extended function number
-  RegisterFunction(0x8000'0000, std::bind(&CPUIDEmu::Function_8000_0000h, this));
+  RegisterFunction(0x8000'0000, std::bind(&CPUIDEmu::Function_8000_0000h, this, _1));
   // Processor vendor
-  RegisterFunction(0x8000'0001, std::bind(&CPUIDEmu::Function_8000_0001h, this));
+  RegisterFunction(0x8000'0001, std::bind(&CPUIDEmu::Function_8000_0001h, this, _1));
   // Processor brand string
-  RegisterFunction(0x8000'0002, std::bind(&CPUIDEmu::Function_8000_0002h, this));
+  RegisterFunction(0x8000'0002, std::bind(&CPUIDEmu::Function_8000_0002h, this, _1));
   // Processor brand string continued
-  RegisterFunction(0x8000'0003, std::bind(&CPUIDEmu::Function_8000_0003h, this));
+  RegisterFunction(0x8000'0003, std::bind(&CPUIDEmu::Function_8000_0003h, this, _1));
   // Processor brand string continued
-  RegisterFunction(0x8000'0004, std::bind(&CPUIDEmu::Function_8000_0004h, this));
+  RegisterFunction(0x8000'0004, std::bind(&CPUIDEmu::Function_8000_0004h, this, _1));
   // 0x8000'0005: L1 Cache and TLB identifiers
 #ifdef CPUID_AMD
-  RegisterFunction(0x8000'0005, std::bind(&CPUIDEmu::Function_8000_0005h, this));
+  RegisterFunction(0x8000'0005, std::bind(&CPUIDEmu::Function_8000_0005h, this, _1));
 #else
   // This is full reserved on Intel platforms
-  RegisterFunction(0x8000'0005, std::bind(&CPUIDEmu::Function_Reserved, this));
+  RegisterFunction(0x8000'0005, std::bind(&CPUIDEmu::Function_Reserved, this, _1));
 #endif
   // 0x8000'0006: L2 Cache identifiers
-  RegisterFunction(0x8000'0006, std::bind(&CPUIDEmu::Function_8000_0006h, this));
+  RegisterFunction(0x8000'0006, std::bind(&CPUIDEmu::Function_8000_0006h, this, _1));
   // Advanced power management information
-  RegisterFunction(0x8000'0007, std::bind(&CPUIDEmu::Function_8000_0007h, this));
+  RegisterFunction(0x8000'0007, std::bind(&CPUIDEmu::Function_8000_0007h, this, _1));
   // Virtual and physical address sizes
-  RegisterFunction(0x8000'0008, std::bind(&CPUIDEmu::Function_8000_0008h, this));
+  RegisterFunction(0x8000'0008, std::bind(&CPUIDEmu::Function_8000_0008h, this, _1));
 
   // 0x8000'000A: SVM Revision
   // TLB 1GB page identifiers
-  RegisterFunction(0x8000'0019, std::bind(&CPUIDEmu::Function_8000_0019h, this));
+  RegisterFunction(0x8000'0019, std::bind(&CPUIDEmu::Function_8000_0019h, this, _1));
 
   // 0x8000'001A: Performance optimization identifiers
   // 0x8000'001B: Instruction based sampling identifiers
   // 0x8000'001C: Lightweight profiling capabilities
   // 0x8000'001D: Cache properties
+#ifdef CPUID_AMD
+  // Deterministic cache parameters for each level
+  RegisterFunction(0x8000'001D, std::bind(&CPUIDEmu::Function_8000_001Dh, this, _1));
+#endif
   // 0x8000'001E: Extended APIC ID
   // 0x8000'001F: AMD Secure Encryption
 }

External/FEXCore/Source/Interface/Core/CPUID.h

@@ -4,7 +4,9 @@
 
 #include <FEXCore/Core/CPUID.h>
 #include <FEXCore/Config/Config.h>
-#include <FEXCore/Utils/LogManager.h>
+
+#include <cstdint>
+#include <utility>
 
 namespace FEXCore {
 namespace Context {
@@ -24,23 +26,20 @@ private:
 public:
   void Init(FEXCore::Context::Context *ctx);
 
-  FEXCore::CPUID::FunctionResults RunFunction(uint32_t Function, [[maybe_unused]] uint32_t Leaf) {
+  FEXCore::CPUID::FunctionResults RunFunction(uint32_t Function, uint32_t Leaf) {
     auto Handler = FunctionHandlers.find(Function);
 
     if (Handler == FunctionHandlers.end()) {
-      #ifndef NDEBUG
-        LogMan::Msg::E("Unhandled CPU ID function, 0x%x", Function);
-      #endif
-      return Function_Reserved();
+      return Function_Reserved(Leaf);
     }
 
-    return Handler->second();
+    return Handler->second(Leaf);
   }
 private:
   FEXCore::Context::Context *CTX;
   FEX_CONFIG_OPT(Cores, THREADS);
 
-  using FunctionHandler = std::function<FEXCore::CPUID::FunctionResults()>;
+  using FunctionHandler = std::function<FEXCore::CPUID::FunctionResults(uint32_t Leaf)>;
   void RegisterFunction(uint32_t Function, FunctionHandler Handler) {
     FunctionHandlers[Function] = Handler;
   }
@@ -48,23 +47,26 @@ private:
   std::unordered_map<uint32_t, FunctionHandler> FunctionHandlers;
 
   // Functions
-  FEXCore::CPUID::FunctionResults Function_0h();
-  FEXCore::CPUID::FunctionResults Function_01h();
-  FEXCore::CPUID::FunctionResults Function_02h();
-  FEXCore::CPUID::FunctionResults Function_06h();
-  FEXCore::CPUID::FunctionResults Function_07h();
-  FEXCore::CPUID::FunctionResults Function_15h();
-  FEXCore::CPUID::FunctionResults Function_8000_0000h();
-  FEXCore::CPUID::FunctionResults Function_8000_0001h();
-  FEXCore::CPUID::FunctionResults Function_8000_0002h();
-  FEXCore::CPUID::FunctionResults Function_8000_0003h();
-  FEXCore::CPUID::FunctionResults Function_8000_0004h();
-  FEXCore::CPUID::FunctionResults Function_8000_0005h();
-  FEXCore::CPUID::FunctionResults Function_8000_0006h();
-  FEXCore::CPUID::FunctionResults Function_8000_0007h();
-  FEXCore::CPUID::FunctionResults Function_8000_0008h();
-  FEXCore::CPUID::FunctionResults Function_8000_0009h();
-  FEXCore::CPUID::FunctionResults Function_8000_0019h();
-  FEXCore::CPUID::FunctionResults Function_Reserved();
+  FEXCore::CPUID::FunctionResults Function_0h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_01h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_02h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_04h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_06h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_07h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_0Dh(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_15h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0000h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0001h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0002h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0003h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0004h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0005h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0006h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0007h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0008h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0009h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_0019h(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_8000_001Dh(uint32_t Leaf);
+  FEXCore::CPUID::FunctionResults Function_Reserved(uint32_t Leaf);
 };
 }

External/FEXCore/Source/Interface/Core/CompileService.cpp

@@ -1,8 +1,21 @@
 #include "Interface/Context/Context.h"
 #include "Interface/Core/LookupCache.h"
 #include "Interface/Core/CompileService.h"
-#include "Interface/Core/InternalThreadState.h"
 #include "Interface/Core/OpcodeDispatcher.h"
+#include "FEXCore/Debug/InternalThreadState.h"
+#include "FEXCore/HLE/Linux/ThreadManagement.h"
+#include "Interface/IR/PassManager.h"
+
+#include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Utils/Event.h>
+#include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/Utils/Threads.h>
+
+#include <memory>
+#include <pthread.h>
+#include <stdio.h>
 
 namespace FEXCore {
   static void* ThreadHandler(void *Arg) {

External/FEXCore/Source/Interface/Core/CompileService.h

@@ -1,23 +1,22 @@
 #pragma once
 
-#include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Debug/InternalThreadState.h>
 #include <FEXCore/Utils/Event.h>
 #include <FEXCore/Utils/Threads.h>
 
+#include <atomic>
 #include <memory>
-#include <thread>
-#include <unordered_map>
+#include <mutex>
 #include <queue>
+#include <stdint.h>
+#include <vector>
 
 namespace FEXCore {
 namespace Context {
   struct Context;
 }
-namespace Core {
-  struct InternalThreadState;
-}
-
 namespace IR {
+  class IRListView;
   class RegisterAllocationData;
 };
 class CompileService final {

External/FEXCore/Source/Interface/Core/Core.cpp

@@ -7,42 +7,70 @@ desc: Glues Frontend, OpDispatcher and IR Opts & Compilation, LookupCache, Dispa
 $end_info$
 */
 
-#include "Common/MathUtils.h"
-#include "Common/Paths.h"
-
 #include "Interface/Context/Context.h"
 #include "Interface/Core/LookupCache.h"
-#include "Interface/Core/BlockSamplingData.h"
 #include "Interface/Core/CompileService.h"
 #include "Interface/Core/Core.h"
-#include "Interface/Core/DebugData.h"
+#include "Interface/Core/CPUID.h"
+#include "Interface/Core/Frontend.h"
+#include "Interface/Core/GdbServer.h"
 #include "Interface/Core/OpcodeDispatcher.h"
 #include "Interface/Core/Interpreter/InterpreterCore.h"
 #include "Interface/Core/JIT/JITCore.h"
+#include "Interface/HLE/Thunks/Thunks.h"
 #include "Interface/IR/Passes/RegisterAllocationPass.h"
 #include "Interface/IR/Passes.h"
+#include "Interface/IR/PassManager.h"
 
 #include <FEXCore/Config/Config.h>
 #include <FEXCore/Core/CodeLoader.h>
+#include <FEXCore/Core/Context.h>
 #include <FEXCore/Core/CoreState.h>
 #include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Core/SignalDelegator.h>
 #include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <FEXCore/Debug/X86Tables.h>
 #include <FEXCore/HLE/SyscallHandler.h>
+#include <FEXCore/HLE/Linux/ThreadManagement.h>
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/IR/IREmitter.h>
+#include <FEXCore/IR/IntrusiveIRList.h>
+#include <FEXCore/IR/RegisterAllocationData.h>
 #include <FEXCore/Utils/Allocator.h>
+#include <FEXCore/Utils/Event.h>
+#include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/Utils/Threads.h>
 
-#include "Interface/HLE/Thunks/Thunks.h"
-#include "FEXCore/Utils/Allocator.h"
-
-#include <xxh3.h>
-#include <fstream>
-#include <unistd.h>
-#include <filesystem>
 #include <algorithm>
+#include <array>
+#include <atomic>
+#include <chrono>
+#include <condition_variable>
+#include <cstdint>
+#include <filesystem>
+#include <functional>
+#include <map>
+#include <memory>
+#include <mutex>
+#include <queue>
+#include <set>
+#include <shared_mutex>
+#include <signal.h>
+#include <stdio.h>
+#include <string.h>
+#include <string>
+#include <string_view>
+#include <sstream>
 #include <sys/mman.h>
-#include <unistd.h>
 #include <sys/stat.h>
-
-#include "Interface/Core/GdbServer.h"
+#include <sys/syscall.h>
+#include <type_traits>
+#include <unistd.h>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+#include <xxhash.h>
 
 namespace FEXCore::CPU {
   bool CreateCPUCore(FEXCore::Context::Context *CTX) {
@@ -52,8 +80,6 @@ namespace FEXCore::CPU {
   }
 }
 
-static std::mutex AOTIRCacheLock;
-
 namespace FEXCore::Core {
 struct ThreadLocalData {
   FEXCore::Core::InternalThreadState* Thread;
@@ -112,41 +138,56 @@ constexpr std::array<std::string_view const, 16> RegNames = {
 std::string_view const& GetGRegName(unsigned Reg) {
   return RegNames[Reg];
 }
-
-namespace DefaultFallbackCore {
-  class DefaultFallbackCore final : public FEXCore::CPU::CPUBackend {
-  public:
-    explicit DefaultFallbackCore(FEXCore::Core::ThreadState *Thread)
-      : ThreadState {reinterpret_cast<FEXCore::Core::InternalThreadState*>(Thread)} {
-    }
-    ~DefaultFallbackCore() override = default;
-
-    std::string GetName() override { return "Default Fallback"; }
-
-    void *MapRegion(void *HostPtr, uint64_t VirtualGuestPtr, uint64_t Size) override {
-      return HostPtr;
-    }
-
-    void Initialize() override {}
-    bool NeedsOpDispatch() override { return false; }
-
-    void *CompileCode(uint64_t Entry, FEXCore::IR::IRListView const *IR, FEXCore::Core::DebugData *DebugData, FEXCore::IR::RegisterAllocationData *RAData) override {
-      LogMan::Msg::E("Fell back to default code handler at RIP: 0x%lx", ThreadState->CurrentFrame->State.rip);
-      return nullptr;
-    }
-
-  private:
-    FEXCore::Core::InternalThreadState *ThreadState;
-  };
-
-  FEXCore::CPU::CPUBackend *CPUCreationFactory(FEXCore::Context::Context* CTX, FEXCore::Core::ThreadState *Thread) {
-    return new DefaultFallbackCore(Thread);
-  }
-}
-
-}
+} // namespace FEXCore::Core
 
 namespace FEXCore::Context {
+  void Context::AOTIRCaptureCacheWriteoutQueue_Flush() {
+    {
+      std::shared_lock lk{AOTIRCaptureCacheWriteoutLock};
+      if (AOTIRCaptureCacheWriteoutQueue.size() == 0) {
+        AOTIRCaptureCacheWriteoutFlusing.store(false);
+        return;
+      }
+    }
+
+    for (;;) {
+      AOTIRCaptureCacheWriteoutLock.lock();
+      std::function<void()> fn = std::move(AOTIRCaptureCacheWriteoutQueue.front());
+      bool MaybeEmpty = false;
+      AOTIRCaptureCacheWriteoutQueue.pop();
+      MaybeEmpty = AOTIRCaptureCacheWriteoutQueue.size() == 0;
+      AOTIRCaptureCacheWriteoutLock.unlock();
+
+      fn();
+      if (MaybeEmpty) {
+        std::shared_lock lk{AOTIRCaptureCacheWriteoutLock};
+        if (AOTIRCaptureCacheWriteoutQueue.size() == 0) {
+          AOTIRCaptureCacheWriteoutFlusing.store(false);
+          return;
+        }
+      }
+    }
+
+    LOGMAN_MSG_A("Must never get here");
+  }
+
+  void Context::AOTIRCaptureCacheWriteoutQueue_Append(const std::function<void()> &fn) {
+    bool Flush = false;
+
+    {
+      std::unique_lock lk{AOTIRCaptureCacheWriteoutLock};
+      AOTIRCaptureCacheWriteoutQueue.push(fn);
+      if (AOTIRCaptureCacheWriteoutQueue.size() > 10000) {
+        Flush = true;
+      }
+    }
+
+    bool test_val = false;
+    if (Flush && AOTIRCaptureCacheWriteoutFlusing.compare_exchange_strong(test_val, true)) {
+      AOTIRCaptureCacheWriteoutQueue_Flush();
+    }
+  }
+
   Context::Context() {
 #ifdef BLOCKSTATS
     BlockData = std::make_unique<FEXCore::BlockSamplingData>();
@@ -177,20 +218,12 @@ namespace FEXCore::Context {
       Threads.clear();
     }
 
-    // AOTIRCaptureCache needs manual clear
-    for (auto &Mod: AOTIRCaptureCache) {
-      for (auto &Entry: Mod.second) {
-        delete Entry.second.IR;
-        FEXCore::Allocator::free(Entry.second.RAData);
-      }
-    }
-
     for (auto &Mod: AOTIRCache) {
       FEXCore::Allocator::munmap(Mod.second.mapping, Mod.second.size);
     }
   }
 
-  bool Context::InitCore(FEXCore::CodeLoader *Loader) {
+  FEXCore::Core::InternalThreadState* Context::InitCore(FEXCore::CodeLoader *Loader) {
     ThunkHandler.reset(FEXCore::ThunkHandler::Create());
 
     LocalLoader = Loader;
@@ -211,6 +244,7 @@ namespace FEXCore::Context {
     NewThreadState.flags[1] = 1;
     NewThreadState.flags[9] = 1;
     NewThreadState.FCW = 0x37F;
+    NewThreadState.FTW = 0xFFFF;
 
     FEXCore::Core::InternalThreadState *Thread = CreateThread(&NewThreadState, 0);
 
@@ -222,8 +256,7 @@ namespace FEXCore::Context {
     Thread->CurrentFrame->State.rip = StartingRIP = Loader->DefaultRIP();
 
     InitializeThreadData(Thread);
-
-    return true;
+    return Thread;
   }
 
   void Context::StartGdbServer() {
@@ -237,17 +270,16 @@ namespace FEXCore::Context {
     DebugServer.reset();
   }
 
-  void Context::HandleCallback(uint64_t RIP) {
-    auto Thread = Core::ThreadData.Thread;
+  void Context::HandleCallback(FEXCore::Core::InternalThreadState *Thread, uint64_t RIP) {
     Thread->CPUBackend->CallbackPtr(Thread->CurrentFrame, RIP);
   }
 
-  void Context::RegisterHostSignalHandler(int Signal, HostSignalDelegatorFunction Func) {
-      SignalDelegation->RegisterHostSignalHandler(Signal, Func);
+  void Context::RegisterHostSignalHandler(int Signal, HostSignalDelegatorFunction Func, bool Required) {
+      SignalDelegation->RegisterHostSignalHandler(Signal, Func, Required);
   }
 
-  void Context::RegisterFrontendHostSignalHandler(int Signal, HostSignalDelegatorFunction Func) {
-    SignalDelegation->RegisterFrontendHostSignalHandler(Signal, Func);
+  void Context::RegisterFrontendHostSignalHandler(int Signal, HostSignalDelegatorFunction Func, bool Required) {
+    SignalDelegation->RegisterFrontendHostSignalHandler(Signal, Func, Required);
   }
 
   void Context::WaitForIdle() {
@@ -282,7 +314,7 @@ namespace FEXCore::Context {
     // Tell all the threads that they should pause
     std::lock_guard<std::mutex> lk(ThreadCreationMutex);
     for (auto &Thread : Threads) {
-      Thread->SignalReason.store(FEXCore::Core::SignalEvent::SIGNALEVENT_PAUSE);
+      Thread->SignalReason.store(FEXCore::Core::SignalEvent::Pause);
       if (Thread->RunningEvents.Running.load()) {
         // Only attempt to stop this thread if it is running
         tgkill(Thread->ThreadManager.PID, Thread->ThreadManager.TID, SignalDelegator::SIGNAL_FOR_PAUSE);
@@ -303,7 +335,7 @@ namespace FEXCore::Context {
     // Spin up all the threads
     std::lock_guard<std::mutex> lk(ThreadCreationMutex);
     for (auto &Thread : Threads) {
-      Thread->SignalReason.store(FEXCore::Core::SignalEvent::SIGNALEVENT_RETURN);
+      Thread->SignalReason.store(FEXCore::Core::SignalEvent::Return);
       Thread->RunningEvents.WaitingToStart.store(true);
     }
 
@@ -369,8 +401,6 @@ namespace FEXCore::Context {
         }
         if (Thread->RunningEvents.Running.load()) {
           StopThread(Thread);
-        } else {
-          LogMan::Msg::D("Skipping thread %p: Already stopped", Thread);
         }
       }
     }
@@ -383,7 +413,7 @@ namespace FEXCore::Context {
 
   void Context::StopThread(FEXCore::Core::InternalThreadState *Thread) {
     if (Thread->RunningEvents.Running.exchange(false)) {
-      Thread->SignalReason.store(FEXCore::Core::SignalEvent::SIGNALEVENT_STOP);
+      Thread->SignalReason.store(FEXCore::Core::SignalEvent::Stop);
       tgkill(Thread->ThreadManager.PID, Thread->ThreadManager.TID, SignalDelegator::SIGNAL_FOR_PAUSE);
     }
   }
@@ -411,7 +441,7 @@ namespace FEXCore::Context {
     }
   }
 
-  int Context::GetProgramStatus() {
+  int Context::GetProgramStatus() const {
     return ParentThread->StatusCode;
   }
 
@@ -442,8 +472,6 @@ namespace FEXCore::Context {
   }
 
   void Context::InitializeThread(FEXCore::Core::InternalThreadState *Thread) {
-    InitializeThreadData(Thread);
-
     // This will create the execution thread but it won't actually start executing
     ExecutionThreadHandler *Arg = reinterpret_cast<ExecutionThreadHandler*>(FEXCore::Allocator::malloc(sizeof(ExecutionThreadHandler)));
     Arg->This = this;
@@ -469,8 +497,10 @@ namespace FEXCore::Context {
         Stop(false /* Ignore current thread */);
     });
 
+    State->CTX = this;
+
     #if _M_ARM_64
-    bool DoSRA = true;
+    bool DoSRA = State->CTX->Config.StaticRegisterAllocation;
     #else
     bool DoSRA = false;
     #endif
@@ -480,26 +510,26 @@ namespace FEXCore::Context {
 
     State->PassManager->RegisterSyscallHandler(SyscallHandler);
 
-    State->CTX = this;
-
     // Create CPU backend
     switch (Config.Core) {
     case FEXCore::Config::CONFIG_INTERPRETER:
-      State->CPUBackend.reset(FEXCore::CPU::CreateInterpreterCore(this, State, CompileThread));
+      State->CPUBackend = FEXCore::CPU::CreateInterpreterCore(this, State, CompileThread);
       break;
     case FEXCore::Config::CONFIG_IRJIT:
       State->PassManager->InsertRegisterAllocationPass(DoSRA);
 
 #if (_M_X86_64 && JIT_X86_64)
-      State->CPUBackend.reset(FEXCore::CPU::CreateX86JITCore(this, State, CompileThread));
+      State->CPUBackend = FEXCore::CPU::CreateX86JITCore(this, State, CompileThread);
 #elif (_M_ARM_64 && JIT_ARM64)
-      State->CPUBackend.reset(FEXCore::CPU::CreateArm64JITCore(this, State, CompileThread));
+      State->CPUBackend = FEXCore::CPU::CreateArm64JITCore(this, State, CompileThread);
 #else
       ERROR_AND_DIE("FEXCore has been compiled without a viable JIT core");
 #endif
 
       break;
-    case FEXCore::Config::CONFIG_CUSTOM:      State->CPUBackend.reset(CustomCPUFactory(this, State)); break;
+    case FEXCore::Config::CONFIG_CUSTOM:
+      State->CPUBackend = CustomCPUFactory(this, State);
+      break;
     default: ERROR_AND_DIE("Unknown core configuration");
     }
   }
@@ -522,6 +552,7 @@ namespace FEXCore::Context {
     Thread->ThreadManager.parent_tid = ParentTID;
 
     InitializeCompiler(Thread, false);
+    InitializeThreadData(Thread);
 
     return Thread;
   }
@@ -579,6 +610,15 @@ namespace FEXCore::Context {
 
     // We now only have one thread
     IdleWaitRefCount = 1;
+
+    // Clean up dead stacks
+    FEXCore::Threads::Thread::CleanupAfterFork();
+
+    if (LiveThread->CompileService) {
+      // If this live thread had a compile service then it no longer exists
+      // Erase the shared_ptr
+      LiveThread->CompileService.reset();
+    }
   }
 
   void Context::AddBlockMapping(FEXCore::Core::InternalThreadState *Thread, uint64_t Address, void *Ptr, uint64_t Start, uint64_t Length) {
@@ -597,7 +637,7 @@ namespace FEXCore::Context {
     }
   }
 
-  std::tuple<FEXCore::IR::IRListView *, FEXCore::IR::RegisterAllocationData *, uint64_t, uint64_t, uint64_t, uint64_t> Context::GenerateIR(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP) {
+  Context::GenerateIRResult Context::GenerateIR(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP) {
     uint8_t const *GuestCode{};
     GuestCode = reinterpret_cast<uint8_t const*>(GuestRIP);
 
@@ -607,14 +647,14 @@ namespace FEXCore::Context {
     uint64_t TotalInstructionsLength {0};
 
     if (!Thread->FrontendDecoder->DecodeInstructionsAtEntry(GuestCode, GuestRIP)) {
-      return { nullptr, nullptr, 0, 0, 0, 0 };
+      return {};
     }
 
     auto CodeBlocks = Thread->FrontendDecoder->GetDecodedBlocks();
 
     Thread->OpDispatcher->BeginFunction(GuestRIP, CodeBlocks);
 
-    uint8_t GPRSize = Config.Is64BitMode ? 8 : 4;
+    const uint8_t GPRSize = GetGPRSize();
 
     for (size_t j = 0; j < CodeBlocks->size(); ++j) {
       FEXCore::Frontend::Decoder::DecodedBlocks const &Block = CodeBlocks->at(j);
@@ -694,7 +734,7 @@ namespace FEXCore::Context {
             return { nullptr, nullptr, 0, 0, 0, 0 };
           }
           else {
-            uint8_t GPRSize = Config.Is64BitMode ? 8 : 4;
+            const uint8_t GPRSize = GetGPRSize();
 
             // We had some instructions. Early exit
             Thread->OpDispatcher->_ExitFunction(Thread->OpDispatcher->_EntrypointOffset(Block.Entry + BlockInstructionsLength - GuestRIP, GPRSize));
@@ -764,7 +804,6 @@ namespace FEXCore::Context {
           LogMan::Msg::I("two:\n %s", out2.str().c_str());
           LOGMAN_MSG_A("Parsed ir doesn't match\n");
         }
-        delete reparsed;
       }
     }
     // Run the passmanager over the IR from the dispatcher
@@ -786,7 +825,14 @@ namespace FEXCore::Context {
 
     Thread->OpDispatcher->ResetWorkingList();
 
-    return {IRList, RAData.release(), TotalInstructions, TotalInstructionsLength, Thread->FrontendDecoder->DecodedMinAddress, Thread->FrontendDecoder->DecodedMaxAddress - Thread->FrontendDecoder->DecodedMinAddress };
+    return {
+      .IRList = IRList,
+      .RAData = RAData.release(),
+      .TotalInstructions = TotalInstructions,
+      .TotalInstructionsLength = TotalInstructionsLength,
+      .StartAddr = Thread->FrontendDecoder->DecodedMinAddress,
+      .Length = Thread->FrontendDecoder->DecodedMaxAddress - Thread->FrontendDecoder->DecodedMinAddress,
+    };
   }
 
   AOTIRInlineEntry *AOTIRInlineIndex::GetInlineEntry(uint64_t DataOffset) {
@@ -824,7 +870,29 @@ namespace FEXCore::Context {
     return (IR::IRListView *)&InlineData[Offset];
   }
 
-  std::tuple<void *, FEXCore::IR::IRListView *, FEXCore::Core::DebugData *, FEXCore::IR::RegisterAllocationData *, bool, uint64_t, uint64_t> Context::CompileCode(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP) {
+  void AOTIRCaptureCacheEntry::AppendAOTIRCaptureCache(uint64_t GuestRIP, uint64_t Start, uint64_t Length, uint64_t Hash, FEXCore::IR::IRListView *IRList, FEXCore::IR::RegisterAllocationData *RAData) {
+    auto Inserted = Index.emplace(GuestRIP, Stream->tellp());
+
+    if (Inserted.second) {
+      //GuestHash
+      Stream->write((const char*)&Hash, sizeof(Hash));
+
+      //GuestLength
+      Stream->write((const char*)&Length, sizeof(Length));
+
+      // RAData (inline)
+      // In file, IsShared is always set
+      auto Shared = RAData->IsShared;
+      RAData->IsShared = true;
+      Stream->write((const char*)RAData, RAData->Size(RAData->MapCount));
+      RAData->IsShared = Shared;
+
+      // IRData (inline)
+      IRList->Serialize(*Stream);
+    }
+  }
+
+  Context::CompileCodeResult Context::CompileCode(FEXCore::Core::InternalThreadState *Thread, uint64_t GuestRIP) {
     FEXCore::IR::IRListView *IRList {};
     FEXCore::Core::DebugData *DebugData {};
     FEXCore::IR::RegisterAllocationData *RAData {};
@@ -848,7 +916,7 @@ namespace FEXCore::Context {
     }
 
     {
-      std::lock_guard<std::mutex> lk(AOTIRCacheLock);
+      std::shared_lock lk(AOTIRCacheLock);
       auto file = AddrToFile.lower_bound(GuestRIP);
       if (file != AddrToFile.begin()) {
         --file;
@@ -860,7 +928,7 @@ namespace FEXCore::Context {
     }
 
     if (IRList == nullptr && Config.AOTIRLoad) {
-      std::lock_guard<std::mutex> lk(AOTIRCacheLock);
+      std::shared_lock lk(AOTIRCacheLock);
       auto file = AddrToFile.lower_bound(GuestRIP);
       if (file != AddrToFile.begin()) {
         --file;
@@ -922,10 +990,18 @@ namespace FEXCore::Context {
     }
 
     if (IRList == nullptr) {
-      return { nullptr, nullptr, nullptr, nullptr, false, 0, 0 };
+      return {};
     }
     // Attempt to get the CPU backend to compile this code
-    return { Thread->CPUBackend->CompileCode(GuestRIP, IRList, DebugData, RAData), IRList, DebugData, RAData, GeneratedIR, StartAddr, Length};
+    return {
+      .CompiledCode = Thread->CPUBackend->CompileCode(GuestRIP, IRList, DebugData, RAData),
+      .IRData = IRList,
+      .DebugData = DebugData,
+      .RAData = RAData,
+      .GeneratedIR = GeneratedIR,
+      .StartAddr = StartAddr,
+      .Length = Length,
+    };
   }
 
   static bool readAll(int fd, void *data, size_t size) {
@@ -939,31 +1015,44 @@ namespace FEXCore::Context {
 
   bool Context::LoadAOTIRCache(int streamfd) {
     uint64_t tag;
-    
-    if (!readAll(streamfd, (char*)&tag, sizeof(tag)) || tag != 0xDEADBEEFC0D30003)
+
+    if (!readAll(streamfd, (char*)&tag, sizeof(tag)) || tag != 0xDEADBEEFC0D30004)
       return false;
-    
+
     std::string Module;
     uint64_t ModSize;
-    
+    uint64_t IndexSize;
+
+    lseek(streamfd, -sizeof(ModSize), SEEK_END);
+
     if (!readAll(streamfd,  (char*)&ModSize, sizeof(ModSize)))
       return false;
 
     Module.resize(ModSize);
+
+    lseek(streamfd, -sizeof(ModSize) - ModSize, SEEK_END);
+
     if (!readAll(streamfd,  (char*)&Module[0], Module.size()))
       return false;
 
+    lseek(streamfd, -sizeof(ModSize) - ModSize - sizeof(IndexSize), SEEK_END);
+
+    if (!readAll(streamfd,  (char*)&IndexSize, sizeof(IndexSize)))
+      return false;
+
     struct stat fileinfo;
     if (fstat(streamfd, &fileinfo) < 0)
       return false;
     size_t Size = (fileinfo.st_size + 4095) & ~4095;
 
+    size_t IndexOffset = fileinfo.st_size - IndexSize -sizeof(ModSize) - ModSize - sizeof(IndexSize);
+
     void *FilePtr = FEXCore::Allocator::mmap(nullptr, Size, PROT_READ, MAP_SHARED, streamfd, 0);
 
     if (FilePtr == MAP_FAILED)
       return false;
 
-    auto Array = (AOTIRInlineIndex *)((char*)FilePtr + sizeof(tag) + sizeof(ModSize) + ((ModSize+31) & ~31));
+    auto Array = (AOTIRInlineIndex *)((char*)FilePtr + IndexOffset);
 
     AOTIRCache.insert({Module, {Array, FilePtr, Size}});
 
@@ -974,83 +1063,53 @@ namespace FEXCore::Context {
   }
 
   void Context::WriteFilesWithCode(std::function<void(const std::string& fileid, const std::string& filename)> Writer) {
-    std::lock_guard<std::mutex> lk(AOTIRCacheLock);
+    std::shared_lock lk(AOTIRCacheLock);
     for( const auto &File: FilesWithCode) {
       Writer(File.first, File.second);
     }
   }
 
-  bool Context::WriteAOTIRCache(std::function<std::unique_ptr<std::ostream>(const std::string&)> CacheWriter) {
-    std::lock_guard<std::mutex> lk(AOTIRCacheLock);
+  void Context::FinalizeAOTIRCache() {
+    AOTIRCaptureCacheWriteoutQueue_Flush();
 
-    bool rv = true;
+    std::unique_lock lk(AOTIRCacheLock);
 
-    for (auto AOTModule: AOTIRCaptureCache) {
-      if (AOTModule.second.size() == 0) {
+    for (auto& [String, Entry] : AOTIRCaptureCache) {
+      if (!Entry.Stream) {
         continue;
       }
 
-      auto stream = CacheWriter(AOTModule.first);
-      if (!*stream) {
-        rv = false;
-      }
-      uint64_t tag = 0xDEADBEEFC0D30003;
-      stream->write((char*)&tag, sizeof(tag));
+      const auto ModSize = String.size();
+      auto &stream = Entry.Stream;
 
-      auto ModSize = AOTModule.first.size();
-      stream->write((char*)&ModSize, sizeof(ModSize));
-      stream->write((char*)&AOTModule.first[0], ModSize);
-
-      auto Skip = ((ModSize + 31) & ~31) - ModSize;
-      char Zero = 0;
-      for (int i = 0; i < Skip; i++)
+      // pad to 32 bytes
+      constexpr char Zero = 0;
+      while(stream->tellp() & 31)
         stream->write(&Zero, 1);
 
       // AOTIRInlineIndex
-      
+      const auto FnCount = Entry.Index.size();
+      const size_t DataBase = -stream->tellp();
 
-      auto FnCount = AOTModule.second.size();
-      stream->write((char*)&FnCount, sizeof(FnCount));
+      stream->write((const char*)&FnCount, sizeof(FnCount));
+      stream->write((const char*)&DataBase, sizeof(DataBase));
 
-      size_t DataBase = sizeof(FnCount) + sizeof(DataBase) + FnCount * sizeof(AOTIRInlineIndexEntry);
-      stream->write((char*)&DataBase, sizeof(DataBase));
-
-      size_t DataOffset = 0;
-      for (auto entry: AOTModule.second) {
+      for (const auto& [GuestStart, DataOffset] : Entry.Index) {
         //AOTIRInlineIndexEntry
 
         // GuestStart
-        stream->write((char*)&entry.first, sizeof(entry.first));
+        stream->write((const char*)&GuestStart, sizeof(GuestStart));
 
         // DataOffset
-        stream->write((char*)&DataOffset, sizeof(DataOffset));
-
-
-        DataOffset += sizeof(entry.second.crc);
-        DataOffset += sizeof(entry.second.len);
-
-        DataOffset += entry.second.RAData->Size(entry.second.RAData->MapCount);
-
-        DataOffset += entry.second.IR->GetInlineSize();
+        stream->write((const char*)&DataOffset, sizeof(DataOffset));
       }
 
-      // AOTIRInlineEntry
-      for (auto entry: AOTModule.second) {
-        //GuestHash
-        stream->write((char*)&entry.second.crc, sizeof(entry.second.crc));
-
-        //GuestLength
-        stream->write((char*)&entry.second.len, sizeof(entry.second.len));
-        
-        // RAData (inline)
-        stream->write((char*)entry.second.RAData, entry.second.RAData->Size(entry.second.RAData->MapCount));
-        
-        // IRData (inline)
-        entry.second.IR->Serialize(*stream);
-      }
+      // End of file header
+      const auto IndexSize = FnCount * sizeof(AOTIRInlineIndexEntry) + sizeof(DataBase) + sizeof(FnCount);
+      stream->write((const char*)&IndexSize, sizeof(IndexSize));
+      stream->write(String.c_str(), ModSize);
+      stream->write((const char*)&ModSize, sizeof(ModSize));
     }
-
-    return rv;
   }
 
   void Context::CompileBlockJit(FEXCore::Core::CpuStateFrame *Frame, uint64_t GuestRIP) {
@@ -1058,10 +1117,12 @@ namespace FEXCore::Context {
 
     if (NewBlock == 0) {
       LogMan::Msg::E("CompileBlockJit: Failed to compile code %lX - aborting process", GuestRIP);
-      abort();
+      // Return similar behaviour of SIGILL abort
+      Frame->Thread->StatusCode = 128 + SIGILL;
+      Stop(false /* Ignore current thread */);
     }
   }
-  
+
   uintptr_t Context::CompileBlock(FEXCore::Core::CpuStateFrame *Frame, uint64_t GuestRIP) {
     auto Thread = Frame->Thread;
 
@@ -1135,26 +1196,49 @@ namespace FEXCore::Context {
 
     // Insert to caches if we generated IR
     if (GeneratedIR) {
-      Core::LocalIREntry Entry = {StartAddr, Length, decltype(Entry.IR)(IRList), decltype(Entry.RAData)(RAData), decltype(Entry.DebugData)(DebugData)};
-      Thread->LocalIRCache.insert({GuestRIP, std::move(Entry)});
-
       // Add to AOT cache if aot generation is enabled
       if ((Config.AOTIRCapture() || Config.AOTIRGenerate()) && RAData) {
-        std::lock_guard<std::mutex> lk(AOTIRCacheLock);
-
-        RAData->IsShared = true;
-        IRList->SetShared(true);
-
         auto hash = XXH3_64bits((void*)StartAddr, Length);
 
+        std::shared_lock lk(AOTIRCacheLock);
+
         auto file = AddrToFile.lower_bound(StartAddr);
         if (file != AddrToFile.begin()) {
           --file;
           if (file->second.Start <= StartAddr && (file->second.Start + file->second.Len) >= (StartAddr + Length)) {
-            AOTIRCaptureCache[file->second.fileid].insert({GuestRIP - file->second.Start + file->second.Offset, {StartAddr - file->second.Start + file->second.Offset, Length, hash, IRList, RAData}});
+            auto LocalRIP = GuestRIP - file->second.Start + file->second.Offset;
+            auto LocalStartAddr = StartAddr - file->second.Start + file->second.Offset;
+            auto fileid = file->second.fileid;
+            AOTIRCaptureCacheWriteoutQueue_Append([this, LocalRIP, LocalStartAddr, Length, hash, IRList, RAData, fileid]() {
+              auto *AotFile = &AOTIRCaptureCache[fileid];
+
+              if (!AotFile->Stream) {
+                AotFile->Stream = AOTIRWriter(fileid);
+                uint64_t tag = 0xDEADBEEFC0D30004;
+                AotFile->Stream->write((char*)&tag, sizeof(tag));
+              }
+              AotFile->AppendAOTIRCaptureCache(LocalRIP, LocalStartAddr, Length, hash, IRList, RAData);
+              delete IRList;
+              FEXCore::Allocator::free(RAData);
+            });
           }
         }
+
+        if (Config.AOTIRGenerate()) {
+          // cleanup memory and early exit here -- we're not running the application
+
+          if (DecrementRefCount)
+            --Thread->CompileBlockReentrantRefCount;
+
+          Thread->CPUBackend->ClearCache();
+
+          return (uintptr_t)CodePtr;
+        }
       }
+
+      // Add to thread local ir cache
+      Core::LocalIREntry Entry = {StartAddr, Length, decltype(Entry.IR)(IRList), decltype(Entry.RAData)(RAData), decltype(Entry.DebugData)(DebugData)};
+      Thread->LocalIRCache.insert({GuestRIP, std::move(Entry)});
     }
 
     if (DecrementRefCount)
@@ -1178,8 +1262,6 @@ namespace FEXCore::Context {
 
     ++IdleWaitRefCount;
 
-    LogMan::Msg::D("[%d] Waiting to run", Thread->ThreadManager.TID.load());
-
     // Now notify the thread that we are initialized
     Thread->ThreadWaiting.NotifyAll();
 
@@ -1188,8 +1270,6 @@ namespace FEXCore::Context {
       Thread->StartRunning.Wait();
     }
 
-    LogMan::Msg::D("[%d] Running", Thread->ThreadManager.TID.load());
-
     Thread->ExitReason = FEXCore::Context::ExitReason::EXIT_NONE;
 
     Thread->RunningEvents.Running = true;
@@ -1303,7 +1383,7 @@ namespace FEXCore::Context {
       fileid += Config.ABILocalFlags ? "L" : "l";
       fileid += Config.ABINoPF ? "p" : "P";
 
-      std::lock_guard<std::mutex> lk(AOTIRCacheLock);
+      std::unique_lock lk(AOTIRCacheLock);
 
       AddrToFile.insert({ Base, { Base, Size, Offset, fileid, filename, nullptr, false} });
 
@@ -1318,13 +1398,13 @@ namespace FEXCore::Context {
   }
 
   void Context::RemoveNamedRegion(uintptr_t Base, uintptr_t Size) {
-    std::lock_guard<std::mutex> lk(AOTIRCacheLock);
+    std::unique_lock lk(AOTIRCacheLock);
     // TODO: Support partial removing
     AddrToFile.erase(Base);
   }
 
-  void ConfigureAOTGen(FEXCore::Context::Context *CTX, std::set<uint64_t> *ExternalBranches, uint64_t SectionMaxAddress) {
-    CTX->ParentThread->FrontendDecoder->SetExternalBranches(ExternalBranches);
-    CTX->ParentThread->FrontendDecoder->SetSectionMaxAddress(SectionMaxAddress);
+  void ConfigureAOTGen(FEXCore::Core::InternalThreadState *Thread, std::set<uint64_t> *ExternalBranches, uint64_t SectionMaxAddress) {
+    Thread->FrontendDecoder->SetExternalBranches(ExternalBranches);
+    Thread->FrontendDecoder->SetSectionMaxAddress(SectionMaxAddress);
   }
 }

External/FEXCore/Source/Interface/Core/Dispatcher/Arm64Dispatcher.cpp

@@ -1,15 +1,29 @@
+#include "Interface/Core/LookupCache.h"
+
 #include "Interface/Core/ArchHelpers/MContext.h"
 #include "Interface/Core/Dispatcher/Arm64Dispatcher.h"
-
 #include "Interface/Core/Interpreter/InterpreterClass.h"
-#include "Interface/Context/Context.h"
 
+#include "Interface/Context/Context.h"
+#include "Interface/Core/X86HelperGen.h"
+
+#include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Core/CoreState.h>
 #include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <array>
+#include <bit>
 #include <cmath>
+#include <cstdint>
+#include <memory>
+#include <stddef.h>
 
 #include "aarch64/assembler-aarch64.h"
+#include "aarch64/constants-aarch64.h"
+#include "aarch64/operands-aarch64.h"
 #include "aarch64/cpu-aarch64.h"
-#include "aarch64/disasm-aarch64.h"
+#include "code-buffer-vixl.h"
+#include "platform-vixl.h"
 
 namespace FEXCore::CPU {
 
@@ -24,8 +38,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
   SRAEnabled = config.StaticRegisterAssignment;
   SetAllowAssembler(true);
 
-  auto Buffer = GetBuffer();
-  DispatchPtr = Buffer->GetOffsetAddress<CPUBackend::AsmDispatch>(GetCursorOffset());
+  DispatchPtr = GetCursorAddress<CPUBackend::AsmDispatch>();
 
   // while (true) {
   //    Ptr = FindBlock(RIP)
@@ -58,7 +71,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
   add(x0, sp, 0);
   str(x0, MemOperand(STATE, offsetof(FEXCore::Core::CpuStateFrame, ReturningStackLocation)));
 
-  AbsoluteLoopTopAddressFillSRA = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+  AbsoluteLoopTopAddressFillSRA = GetCursorAddress<uint64_t>();
 
   if (SRAEnabled) {
     FillStaticRegs();
@@ -103,7 +116,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
   ldr(x0, &l_PagePtr);
 
   // Mask the address by the virtual address size so we can check for aliases
-  if (__builtin_popcountl(VirtualMemorySize) == 1) {
+  if (std::popcount(VirtualMemorySize) == 1) {
     and_(x3, RipReg, Thread->LookupCache->GetVirtualMemorySize() - 1);
   }
   else {
@@ -177,11 +190,11 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
 
   {
     bind(&ExitSpillSRA);
-    ThreadStopHandlerAddressSpillSRA = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    ThreadStopHandlerAddressSpillSRA = GetCursorAddress<uint64_t>();
     if (SRAEnabled)
       SpillStaticRegs();
 
-    ThreadStopHandlerAddress = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    ThreadStopHandlerAddress = GetCursorAddress<uint64_t>();
 
     PopCalleeSavedRegisters();
 
@@ -191,7 +204,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
   }
 
   {
-    ExitFunctionLinkerAddress = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    ExitFunctionLinkerAddress = GetCursorAddress<uint64_t>();
     if (SRAEnabled)
       SpillStaticRegs();
 
@@ -228,7 +241,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
   }
 
   {
-    SignalHandlerReturnAddress = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    SignalHandlerReturnAddress = GetCursorAddress<uint64_t>();
 
     // Now to get back to our old location we need to do a fault dance
     // We can't use SIGTRAP here since gdb catches it and never gives it to the application!
@@ -236,12 +249,12 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
   }
 
   {
-    ThreadPauseHandlerAddressSpillSRA = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    ThreadPauseHandlerAddressSpillSRA = GetCursorAddress<uint64_t>();
     if (SRAEnabled)
       SpillStaticRegs();
 
     bind(&ThreadPauseHandler);
-    ThreadPauseHandlerAddress = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    ThreadPauseHandlerAddress = GetCursorAddress<uint64_t>();
     // We are pausing, this means the frontend should be waiting for this thread to idle
     // We will have faulted and jumped to this location at this point
 
@@ -251,7 +264,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
     ldr(x2, &l_Sleep);
     blr(x2);
 
-    PauseReturnInstruction = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+    PauseReturnInstruction = GetCursorAddress<uint64_t>();
     // Fault to start running again
     hlt(0);
   }
@@ -272,7 +285,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
     // On return to the thunk, the thunk can get whatever its return value is from the thread context depending on ABI handling on its end
     // When the thunk itself returns, it'll do its regular return logic there
     // void ReentrantCallback(FEXCore::Core::InternalThreadState *Thread, uint64_t RIP);
-    CallbackPtr = Buffer->GetOffsetAddress<CPUBackend::JITCallback>(GetCursorOffset());
+    CallbackPtr = GetCursorAddress<CPUBackend::JITCallback>();
 
     // We expect the thunk to have previously pushed the registers it was using
     PushCalleeSavedRegisters();
@@ -321,7 +334,7 @@ Arm64Dispatcher::Arm64Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::
 
   FinalizeCode();
   Start = reinterpret_cast<uint64_t>(DispatchPtr);
-  End = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+  End = GetCursorAddress<uint64_t>();
   vixl::aarch64::CPU::EnsureIAndDCacheCoherency(reinterpret_cast<void*>(DispatchPtr), End - reinterpret_cast<uint64_t>(DispatchPtr));
   GetBuffer()->SetExecutable();
 

External/FEXCore/Source/Interface/Core/Dispatcher/Arm64Dispatcher.h

@@ -3,7 +3,13 @@
 #include "Interface/Core/ArchHelpers/Arm64Emitter.h"
 #include "Interface/Core/Dispatcher/Dispatcher.h"
 
-#include "aarch64/assembler-aarch64.h"
+namespace FEXCore::Context {
+struct Context;
+}
+
+namespace FEXCore::Core {
+struct InternalThreadState;
+}
 
 namespace FEXCore::CPU {
 
@@ -15,4 +21,4 @@ class Arm64Dispatcher final : public Dispatcher, public Arm64Emitter {
     void SpillSRA(void *ucontext) override;
 };
 
-}
+}

External/FEXCore/Source/Interface/Core/Dispatcher/Dispatcher.cpp

@@ -1,8 +1,23 @@
-#include "Interface/Core/Dispatcher/Dispatcher.h"
 
 #include "Common/MathUtils.h"
 #include "Interface/Core/ArchHelpers/MContext.h"
+#include "Interface/Core/Dispatcher/Dispatcher.h"
+#include "Interface/Core/X86HelperGen.h"
+
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Core/UContext.h>
 #include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <FEXCore/Utils/Event.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <atomic>
+#include <condition_variable>
+#include <bits/types/siginfo_t.h>
+#include <signal.h>
+#include <string.h>
 
 namespace FEXCore::CPU {
 
@@ -54,6 +69,7 @@ void Dispatcher::StoreThreadState(int Signal, void *ucontext) {
 }
 
 void Dispatcher::RestoreThreadState(void *ucontext) {
+  LOGMAN_THROW_A(!SignalFrames.empty(), "Trying to restore a signal frame when we don't have any");
   uint64_t OldSP = SignalFrames.top();
   SignalFrames.pop();
   uintptr_t NewSP = OldSP;
@@ -64,10 +80,38 @@ void Dispatcher::RestoreThreadState(void *ucontext) {
 
   // Now restore host state
   ArchHelpers::Context::RestoreContext(ucontext, Context);
+}
 
-  // Restore the previous signal state
-  // This allows recursive signals to properly handle signal masking as we are walking back up the list of signals
-  CTX->SignalDelegation->SetCurrentSignal(Context->Signal);
+static uint32_t ConvertSignalToTrapNo(int Signal, siginfo_t *HostSigInfo) {
+  switch (Signal) {
+    case SIGSEGV:
+      if (HostSigInfo->si_code == SEGV_MAPERR ||
+          HostSigInfo->si_code == SEGV_ACCERR) {
+        // Protection fault
+        return X86State::X86_TRAPNO_PF;
+      }
+      break;
+  }
+
+  // Unknown mapping, fall back to old behaviour and just pass signal
+  return Signal;
+}
+
+static uint32_t ConvertSignalToError(int Signal, siginfo_t *HostSigInfo) {
+  switch (Signal) {
+    case SIGSEGV:
+      if (HostSigInfo->si_code == SEGV_MAPERR ||
+          HostSigInfo->si_code == SEGV_ACCERR) {
+        // Protection fault
+        // Always a user fault for us
+        // XXX: PF_PROT and PF_WRITE
+        return X86State::X86_PF_USER;
+      }
+      break;
+  }
+
+  // Not a page fault issue
+  return 0;
 }
 
 bool Dispatcher::HandleGuestSignal(int Signal, void *info, void *ucontext, GuestSigAction *GuestAction, stack_t *GuestStack) {
@@ -78,60 +122,97 @@ bool Dispatcher::HandleGuestSignal(int Signal, void *info, void *ucontext, Guest
   // We use this to track if it is safe to clear cache
   ++SignalHandlerRefCounter;
 
+  uint64_t OldPC = ArchHelpers::Context::GetPc(ucontext);
   // Set the new PC
   ArchHelpers::Context::SetPc(ucontext, AbsoluteLoopTopAddressFillSRA);
   // Set our state register to point to our guest thread data
   ArchHelpers::Context::SetState(ucontext, reinterpret_cast<uint64_t>(Frame));
 
-
   uint64_t OldGuestSP = Frame->State.gregs[X86State::REG_RSP];
   uint64_t NewGuestSP = OldGuestSP;
 
-  if (!(GuestStack->ss_flags & SS_DISABLE)) {
-    // If our guest is already inside of the alternative stack
-    // Then that means we are hitting recursive signals and we need to walk back the stack correctly
-    uint64_t AltStackBase = reinterpret_cast<uint64_t>(GuestStack->ss_sp);
-    uint64_t AltStackEnd = AltStackBase + GuestStack->ss_size;
-    if (OldGuestSP >= AltStackBase &&
-        OldGuestSP <= AltStackEnd) {
-      // We are already in the alt stack, the rest of the code will handle adjusting this
-    }
-    else {
-      NewGuestSP = AltStackEnd;
+  // Pulling from context here
+  bool Is64BitMode = CTX->Config.Is64BitMode;
+  uint64_t SignalReturn = CTX->X86CodeGen.SignalReturn;
+
+  // Spill the SRA regardless of signal handler type
+  // We are going to be returning to the top of the dispatcher which will fill again
+  // Otherwise we might load garbage
+  if (SRAEnabled) {
+    if (IsAddressInJITCode(OldPC, false)) {
+      // We are in jit, SRA must be spilled
+      SpillSRA(ucontext);
+    } else {
+      if (!IsAddressInJITCode(OldPC, true)) {
+        // This is likely to cause issues but in some cases it isn't fatal
+        // This can also happen if we have put a signal on hold, then we just reenabled the signal
+        // So we are in the syscall handler
+        // Only throw a log message in this case
+        LogMan::Msg::E("Signals in dispatcher have unsynchronized context");
+      }
     }
   }
 
-  // Back up past the redzone, which is 128bytes
-  // Don't need this offset if we aren't going to be putting siginfo in to it
-  NewGuestSP -= 128;
-
-  if (GuestAction->sa_flags & SA_SIGINFO) {
-    if (SRAEnabled) {
-      if (!IsAddressInJITCode(ArchHelpers::Context::GetPc(ucontext), false)) {
-        LOGMAN_THROW_A(!IsAddressInJITCode(ArchHelpers::Context::GetPc(ucontext), true), "Signals in dispatcher have unsynchronized context");
-      } else {
-        // We are in jit, SRA must be spilled
-        SpillSRA(ucontext);
+  // altstack is only used if the signal handler was setup with SA_ONSTACK
+  if (GuestAction->sa_flags & SA_ONSTACK) {
+    // Additionally the altstack is only used if the enabled (SS_DISABLE flag is not set)
+    if (!(GuestStack->ss_flags & SS_DISABLE)) {
+      // If our guest is already inside of the alternative stack
+      // Then that means we are hitting recursive signals and we need to walk back the stack correctly
+      uint64_t AltStackBase = reinterpret_cast<uint64_t>(GuestStack->ss_sp);
+      uint64_t AltStackEnd = AltStackBase + GuestStack->ss_size;
+      if (OldGuestSP >= AltStackBase &&
+          OldGuestSP <= AltStackEnd) {
+        // We are already in the alt stack, the rest of the code will handle adjusting this
+      }
+      else {
+        NewGuestSP = AltStackEnd;
       }
     }
+  }
+
+  if (Is64BitMode) {
+    // Back up past the redzone, which is 128bytes
+    // 32-bit doesn't have a redzone
+    NewGuestSP -= 128;
+  }
+
+  // siginfo_t
+  siginfo_t *HostSigInfo = reinterpret_cast<siginfo_t*>(info);
+
+  if (GuestAction->sa_flags & SA_SIGINFO) {
 
     // Setup ucontext a bit
-    if (CTX->Config.Is64BitMode) {
+    if (Is64BitMode) {
+      NewGuestSP -= sizeof(FEXCore::x86_64::_libc_fpstate);
+      NewGuestSP = AlignDown(NewGuestSP, alignof(FEXCore::x86_64::_libc_fpstate));
+      uint64_t FPStateLocation = NewGuestSP;
 
       NewGuestSP -= sizeof(FEXCore::x86_64::ucontext_t);
+      NewGuestSP = AlignDown(NewGuestSP, alignof(FEXCore::x86_64::ucontext_t));
       uint64_t UContextLocation = NewGuestSP;
 
       NewGuestSP -= sizeof(siginfo_t);
+      NewGuestSP = AlignDown(NewGuestSP, alignof(siginfo_t));
       uint64_t SigInfoLocation = NewGuestSP;
 
       FEXCore::x86_64::ucontext_t *guest_uctx = reinterpret_cast<FEXCore::x86_64::ucontext_t*>(UContextLocation);
       siginfo_t *guest_siginfo = reinterpret_cast<siginfo_t*>(SigInfoLocation);
 
       // We have extended float information
-      guest_uctx->uc_flags |= FEXCore::x86_64::UC_FP_XSTATE;
+      guest_uctx->uc_flags = FEXCore::x86_64::UC_FP_XSTATE;
 
       // Pointer to where the fpreg memory is
-      guest_uctx->uc_mcontext.fpregs = &guest_uctx->__fpregs_mem;
+      guest_uctx->uc_mcontext.fpregs = reinterpret_cast<FEXCore::x86_64::_libc_fpstate*>(FPStateLocation);
+      FEXCore::x86_64::_libc_fpstate *fpstate = reinterpret_cast<FEXCore::x86_64::_libc_fpstate*>(FPStateLocation);
+
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_RIP] = Frame->State.rip;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_EFL] = 0;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_CSGSFS] = 0;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_ERR] = ConvertSignalToError(Signal, HostSigInfo);
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_TRAPNO] = ConvertSignalToTrapNo(Signal, HostSigInfo);
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_OLDMASK] = 0;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_CR2] = 0;
 
 #define COPY_REG(x) \
       guest_uctx->uc_mcontext.gregs[FEXCore::x86_64::FEX_REG_##x] = Frame->State.gregs[X86State::REG_##x];
@@ -154,14 +235,15 @@ bool Dispatcher::HandleGuestSignal(int Signal, void *info, void *ucontext, Guest
 #undef COPY_REG
 
       // Copy float registers
-      memcpy(guest_uctx->__fpregs_mem._st, Frame->State.mm, sizeof(Frame->State.mm));
-      memcpy(guest_uctx->__fpregs_mem._xmm, Frame->State.xmm, sizeof(Frame->State.xmm));
+      memcpy(fpstate->_st, Frame->State.mm, sizeof(Frame->State.mm));
+      memcpy(fpstate->_xmm, Frame->State.xmm, sizeof(Frame->State.xmm));
 
       // FCW store default
-      guest_uctx->__fpregs_mem.fcw = Frame->State.FCW;
+      fpstate->fcw = Frame->State.FCW;
+      fpstate->ftw = Frame->State.FTW;
 
       // Reconstruct FSW
-      guest_uctx->__fpregs_mem.fsw =
+      fpstate->fsw =
         (Frame->State.flags[FEXCore::X86State::X87FLAG_TOP_LOC] << 11) |
         (Frame->State.flags[FEXCore::X86State::X87FLAG_C0_LOC] << 8) |
         (Frame->State.flags[FEXCore::X86State::X87FLAG_C1_LOC] << 9) |
@@ -173,57 +255,165 @@ bool Dispatcher::HandleGuestSignal(int Signal, void *info, void *ucontext, Guest
       guest_uctx->uc_stack.ss_sp = GuestStack->ss_sp;
       guest_uctx->uc_stack.ss_size = GuestStack->ss_size;
 
-      // siginfo_t
-      siginfo_t *HostSigInfo = reinterpret_cast<siginfo_t*>(info);
-      guest_siginfo->si_signo = Signal;
-      switch (Signal) {
-      case SIGSEGV:
-      case SIGBUS:
-        guest_siginfo->si_code = HostSigInfo->si_code;
-        guest_siginfo->si_errno = HostSigInfo->si_errno;
-        // Macro expansion to get the si_addr
-        guest_siginfo->si_addr = HostSigInfo->si_addr;
-        break;
-      default: LogMan::Msg::D("Unhandled siginfo_t signal: %d", Signal); break;
-      }
+      // aarch64 and x86_64 siginfo_t matches. We can just copy this over
+      // SI_USER could also potentially have random data in it, needs to be bit perfect
+      // For guest faults we don't have a real way to reconstruct state to a real guest RIP
+      *guest_siginfo = *HostSigInfo;
 
       Frame->State.gregs[X86State::REG_RSI] = SigInfoLocation;
       Frame->State.gregs[X86State::REG_RDX] = UContextLocation;
     }
     else {
-      // XXX: 32bit Support
+      NewGuestSP -= sizeof(FEXCore::x86::_libc_fpstate);
+      NewGuestSP = AlignDown(NewGuestSP, alignof(FEXCore::x86::_libc_fpstate));
+      uint64_t FPStateLocation = NewGuestSP;
+
       NewGuestSP -= sizeof(FEXCore::x86::ucontext_t);
-      uint64_t UContextLocation = 0; // NewGuestSP;
+      NewGuestSP = AlignDown(NewGuestSP, alignof(FEXCore::x86::ucontext_t));
+      uint64_t UContextLocation = NewGuestSP;
+
       NewGuestSP -= sizeof(FEXCore::x86::siginfo_t);
-      uint64_t SigInfoLocation = 0; // NewGuestSP;
+      NewGuestSP = AlignDown(NewGuestSP, alignof(FEXCore::x86::siginfo_t));
+      uint64_t SigInfoLocation = NewGuestSP;
+
+      FEXCore::x86::ucontext_t *guest_uctx = reinterpret_cast<FEXCore::x86::ucontext_t*>(UContextLocation);
+      FEXCore::x86::siginfo_t *guest_siginfo = reinterpret_cast<FEXCore::x86::siginfo_t*>(SigInfoLocation);
+
+      // We have extended float information
+      guest_uctx->uc_flags = FEXCore::x86::UC_FP_XSTATE;
+
+      // Pointer to where the fpreg memory is
+      guest_uctx->uc_mcontext.fpregs = static_cast<uint32_t>(FPStateLocation);
+      FEXCore::x86::_libc_fpstate *fpstate = reinterpret_cast<FEXCore::x86::_libc_fpstate*>(FPStateLocation);
+
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_GS] = Frame->State.gs;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_FS] = Frame->State.fs;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_ES] = Frame->State.es;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_DS] = Frame->State.ds;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_TRAPNO] = ConvertSignalToTrapNo(Signal, HostSigInfo);
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_ERR] = ConvertSignalToError(Signal, HostSigInfo);
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_EIP] = Frame->State.rip;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_CS] = Frame->State.cs;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_EFL] = 0;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_UESP] = 0;
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_SS] = Frame->State.ss;
+
+#define COPY_REG(x) \
+      guest_uctx->uc_mcontext.gregs[FEXCore::x86::FEX_REG_##x] = Frame->State.gregs[X86State::REG_##x];
+      COPY_REG(RDI);
+      COPY_REG(RSI);
+      COPY_REG(RBP);
+      COPY_REG(RBX);
+      COPY_REG(RDX);
+      COPY_REG(RAX);
+      COPY_REG(RCX);
+      COPY_REG(RSP);
+#undef COPY_REG
+
+      // Copy float registers
+      for (size_t i = 0; i < 8; ++i) {
+        // 32-bit st register size is only 10 bytes. Not padded to 16byte like x86-64
+        memcpy(&fpstate->_st[i], &Frame->State.mm[i], 10);
+      }
+
+      // Extended XMM state
+      fpstate->status = FEXCore::x86::fpstate_magic::MAGIC_XFPSTATE;
+      memcpy(fpstate->_xmm, Frame->State.xmm, sizeof(Frame->State.xmm));
+
+      // FCW store default
+      fpstate->fcw = Frame->State.FCW;
+      fpstate->ftw = Frame->State.FTW;
+      // Reconstruct FSW
+      fpstate->fsw =
+        (Frame->State.flags[FEXCore::X86State::X87FLAG_TOP_LOC] << 11) |
+        (Frame->State.flags[FEXCore::X86State::X87FLAG_C0_LOC] << 8) |
+        (Frame->State.flags[FEXCore::X86State::X87FLAG_C1_LOC] << 9) |
+        (Frame->State.flags[FEXCore::X86State::X87FLAG_C2_LOC] << 10) |
+        (Frame->State.flags[FEXCore::X86State::X87FLAG_C3_LOC] << 14);
+
+      // Copy over signal stack information
+      guest_uctx->uc_stack.ss_flags = GuestStack->ss_flags;
+      guest_uctx->uc_stack.ss_sp = static_cast<uint32_t>(reinterpret_cast<uint64_t>(GuestStack->ss_sp));
+      guest_uctx->uc_stack.ss_size = GuestStack->ss_size;
+
+      // These three elements are in every siginfo
+      guest_siginfo->si_signo = HostSigInfo->si_signo;
+      guest_siginfo->si_errno = HostSigInfo->si_errno;
+      guest_siginfo->si_code = HostSigInfo->si_code;
+
+      switch (Signal) {
+        case SIGSEGV:
+        case SIGBUS:
+          // Macro expansion to get the si_addr
+          // This is the address trying to be accessed, not the RIP
+          guest_siginfo->_sifields._sigfault.addr = static_cast<uint32_t>(reinterpret_cast<uintptr_t>(HostSigInfo->si_addr));
+          break;
+        case SIGFPE:
+        case SIGILL:
+          // Macro expansion to get the si_addr
+          // Can't really give a real result here. Pull from the context for now
+          guest_siginfo->_sifields._sigfault.addr = Frame->State.rip;
+          break;
+        case SIGCHLD:
+          guest_siginfo->_sifields._sigchld.pid = HostSigInfo->si_pid;
+          guest_siginfo->_sifields._sigchld.uid = HostSigInfo->si_uid;
+          guest_siginfo->_sifields._sigchld.status = HostSigInfo->si_status;
+          guest_siginfo->_sifields._sigchld.utime = HostSigInfo->si_utime;
+          guest_siginfo->_sifields._sigchld.stime = HostSigInfo->si_stime;
+          break;
+        case SIGALRM:
+        case SIGVTALRM:
+          guest_siginfo->_sifields._timer.tid = HostSigInfo->si_timerid;
+          guest_siginfo->_sifields._timer.overrun = HostSigInfo->si_overrun;
+          guest_siginfo->_sifields._timer.sigval.sival_int = HostSigInfo->si_int;
+          break;
+        default:
+          LogMan::Msg::E("Unhandled siginfo_t for signal: %d\n", Signal);
+          break;
+      }
 
       NewGuestSP -= 4;
       *(uint32_t*)NewGuestSP = UContextLocation;
       NewGuestSP -= 4;
       *(uint32_t*)NewGuestSP = SigInfoLocation;
+      NewGuestSP -= 4;
+      *(uint32_t*)NewGuestSP = Signal;
     }
 
     Frame->State.rip = reinterpret_cast<uint64_t>(GuestAction->sigaction_handler.sigaction);
   }
   else {
+    if (!Is64BitMode) {
+      NewGuestSP -= 4;
+      *(uint32_t*)NewGuestSP = Signal;
+    }
+
     Frame->State.rip = reinterpret_cast<uint64_t>(GuestAction->sigaction_handler.handler);
   }
 
-  if (CTX->Config.Is64BitMode) {
-    Frame->State.gregs[X86State::REG_RDI] = Signal;
+  if (Is64BitMode) {
+    Frame->State.gregs[FEXCore::X86State::REG_RDI] = Signal;
 
     // Set up the new SP for stack handling
     NewGuestSP -= 8;
-    *(uint64_t*)NewGuestSP = CTX->X86CodeGen.SignalReturn;
-    Frame->State.gregs[X86State::REG_RSP] = NewGuestSP;
+    *(uint64_t*)NewGuestSP = SignalReturn;
+    Frame->State.gregs[FEXCore::X86State::REG_RSP] = NewGuestSP;
   }
   else {
     NewGuestSP -= 4;
-    *(uint32_t*)NewGuestSP = CTX->X86CodeGen.SignalReturn;
-    LOGMAN_THROW_A(CTX->X86CodeGen.SignalReturn < 0x1'0000'0000ULL, "This needs to be below 4GB");
-    Frame->State.gregs[X86State::REG_RSP] = NewGuestSP;
+    *(uint32_t*)NewGuestSP = SignalReturn;
+    LOGMAN_THROW_A(SignalReturn < 0x1'0000'0000ULL, "This needs to be below 4GB");
+    Frame->State.gregs[FEXCore::X86State::REG_RSP] = NewGuestSP;
   }
 
+  // The guest starts its signal frame with a zero initialized FPU
+  // Set that up now. Little bit costly but it's a requirement
+  // This state will be restored on rt_sigreturn
+  memset(Frame->State.xmm, 0, sizeof(Frame->State.xmm));
+  memset(Frame->State.mm, 0, sizeof(Frame->State.mm));
+  Frame->State.FCW = 0x37F;
+  Frame->State.FTW = 0xFFFF;
+
   return true;
 }
 
@@ -254,7 +444,7 @@ bool Dispatcher::HandleSignalPause(int Signal, void *info, void *ucontext) {
   FEXCore::Core::SignalEvent SignalReason = ThreadState->SignalReason.load();
   auto Frame = ThreadState->CurrentFrame;
 
-  if (SignalReason == FEXCore::Core::SignalEvent::SIGNALEVENT_PAUSE) {
+  if (SignalReason == FEXCore::Core::SignalEvent::Pause) {
     // Store our thread state so we can come back to this
     StoreThreadState(Signal, ucontext);
 
@@ -269,9 +459,6 @@ bool Dispatcher::HandleSignalPause(int Signal, void *info, void *ucontext) {
       ArchHelpers::Context::SetPc(ucontext, ThreadPauseHandlerAddress);
     }
 
-    // Set the new PC
-    ArchHelpers::Context::SetPc(ucontext, ThreadPauseHandlerAddress);
-
     // Set our state register to point to our guest thread data
     ArchHelpers::Context::SetState(ucontext, reinterpret_cast<uint64_t>(Frame));
 
@@ -279,11 +466,11 @@ bool Dispatcher::HandleSignalPause(int Signal, void *info, void *ucontext) {
     // We use this to track if it is safe to clear cache
     ++SignalHandlerRefCounter;
 
-    ThreadState->SignalReason.store(FEXCore::Core::SIGNALEVENT_NONE);
+    ThreadState->SignalReason.store(FEXCore::Core::SignalEvent::Nothing);
     return true;
   }
 
-  if (SignalReason == FEXCore::Core::SignalEvent::SIGNALEVENT_STOP) {
+  if (SignalReason == FEXCore::Core::SignalEvent::Stop) {
     // Our thread is stopping
     // We don't care about anything at this point
     // Set the stack to our starting location when we entered the core and get out safely
@@ -304,18 +491,18 @@ bool Dispatcher::HandleSignalPause(int Signal, void *info, void *ucontext) {
       ArchHelpers::Context::SetPc(ucontext, ThreadStopHandlerAddress);
     }
 
-    ThreadState->SignalReason.store(FEXCore::Core::SIGNALEVENT_NONE);
+    ThreadState->SignalReason.store(FEXCore::Core::SignalEvent::Nothing);
     return true;
   }
 
-  if (SignalReason == FEXCore::Core::SignalEvent::SIGNALEVENT_RETURN) {
+  if (SignalReason == FEXCore::Core::SignalEvent::Return) {
     RestoreThreadState(ucontext);
 
     // Ref count our faults
     // We use this to track if it is safe to clear cache
     --SignalHandlerRefCounter;
 
-    ThreadState->SignalReason.store(FEXCore::Core::SIGNALEVENT_NONE);
+    ThreadState->SignalReason.store(FEXCore::Core::SignalEvent::Nothing);
     return true;
   }
 
@@ -344,7 +531,7 @@ void Dispatcher::RemoveCodeBuffer(uint8_t* start_to_remove) {
   }
 }
 
-bool Dispatcher::IsAddressInJITCode(uint64_t Address, bool IncludeDispatcher) {
+bool Dispatcher::IsAddressInJITCode(uint64_t Address, bool IncludeDispatcher) const {
   for (auto [start, end] : CodeBuffers) {
     if (Address >= start && Address < end) {
       return true;

External/FEXCore/Source/Interface/Core/Dispatcher/Dispatcher.h

@@ -1,11 +1,24 @@
 #pragma once
 
 #include <FEXCore/Core/CPUBackend.h>
-#include <FEXCore/Core/SignalDelegator.h>
 
 #include "Interface/Context/Context.h"
 
+#include <bits/types/stack_t.h>
+#include <cstdint>
+#include <stddef.h>
 #include <stack>
+#include <tuple>
+#include <vector>
+
+namespace FEXCore {
+struct GuestSigAction;
+}
+
+namespace FEXCore::Core {
+struct CpuStateFrame;
+struct InternalThreadState;
+}
 
 namespace FEXCore::CPU {
 
@@ -54,8 +67,8 @@ public:
 
   void RemoveCodeBuffer(uint8_t* start);
 
-  bool IsAddressInJITCode(uint64_t Address, bool IncludeDispatcher = true);
-  bool IsAddressInDispatcher(uint64_t Address) {
+  bool IsAddressInJITCode(uint64_t Address, bool IncludeDispatcher = true) const;
+  bool IsAddressInDispatcher(uint64_t Address) const {
     return Address >= Start && Address < End;
   }
 

External/FEXCore/Source/Interface/Core/Dispatcher/X86Dispatcher.cpp

@@ -1,10 +1,23 @@
+#include "Interface/Core/LookupCache.h"
+
 #include "Interface/Core/Dispatcher/X86Dispatcher.h"
 
 #include "Interface/Core/Interpreter/InterpreterClass.h"
+#include "Interface/Core/X86HelperGen.h"
 #include "Interface/Context/Context.h"
 
 #include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <FEXCore/Utils/Allocator.h>
+
 #include <cmath>
+#include <memory>
+#include <stddef.h>
+#include <stdint.h>
+#include <sys/mman.h>
+#include "xbyak/xbyak.h"
 
 namespace FEXCore::CPU {
 static constexpr size_t MAX_DISPATCHER_CODE_SIZE = 4096;
@@ -12,7 +25,9 @@ static constexpr size_t MAX_DISPATCHER_CODE_SIZE = 4096;
 
 X86Dispatcher::X86Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, DispatcherConfig &config)
   : Dispatcher(ctx, Thread)
-  , Xbyak::CodeGenerator(MAX_DISPATCHER_CODE_SIZE, nullptr, this) {
+  , Xbyak::CodeGenerator(MAX_DISPATCHER_CODE_SIZE,
+      FEXCore::Allocator::mmap(nullptr, MAX_DISPATCHER_CODE_SIZE, PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0),
+      nullptr) {
 
   using namespace Xbyak;
   using namespace Xbyak::util;
@@ -266,7 +281,7 @@ X86Dispatcher::X86Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::Inte
 
   {
     ReturnPtr = getCurr<FEXCore::Context::Context::IntCallbackReturn>();
-//  using CallbackReturn =  __attribute__((naked)) void(*)(FEXCore::Core::InternalThreadState *Thread, volatile void *Host_RSP);
+//  using CallbackReturn =  FEX_NAKED void(*)(FEXCore::Core::InternalThreadState *Thread, volatile void *Host_RSP);
 
     // rdi = thread
     // rsi = rsp
@@ -293,12 +308,12 @@ X86Dispatcher::X86Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::Inte
 
   #if ENABLE_JITSYMBOLS
     std::string Name = "Dispatch_" + std::to_string(::gettid());
-    CTX->Symbols.Register(Start, End-Start, Name);
+    CTX->Symbols.Register(reinterpret_cast<void*>(Start), End-Start, Name);
   #endif
 }
 
 X86Dispatcher::~X86Dispatcher() {
-
+  FEXCore::Allocator::munmap(top_, MAX_DISPATCHER_CODE_SIZE);
 }
 
 #ifdef _M_X86_64

External/FEXCore/Source/Interface/Core/Dispatcher/X86Dispatcher.h

@@ -2,26 +2,24 @@
 
 #include "Interface/Core/Dispatcher/Dispatcher.h"
 
-#include <FEXCore/Utils/Allocator.h>
-
 #define XBYAK64
 #include <xbyak/xbyak.h>
 
+namespace FEXCore::Context {
+  struct Context;
+}
+
+namespace FEXCore::Core {
+  struct InternalThreadState;
+}
+
 namespace FEXCore::CPU {
 
-class X86Dispatcher final : public Dispatcher, public Xbyak::CodeGenerator, public Xbyak::Allocator {
+class X86Dispatcher final : public Dispatcher, public Xbyak::CodeGenerator {
   public:
     X86Dispatcher(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, DispatcherConfig &config);
 
     virtual ~X86Dispatcher() override;
-
-    // Xbyak::Allocator
-    Xbyak::uint8 *alloc(size_t size) override { Size = size; return reinterpret_cast<uint8_t*>(FEXCore::Allocator::mmap(nullptr, size, PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0)); }
-    void free(Xbyak::uint8 *p) override { FEXCore::Allocator::munmap(p, Size); }
-    bool useProtect() const override { return false; }
-
-  private:
-    size_t Size{};
 };
 
 }

External/FEXCore/Source/Interface/Core/Frontend.cpp

@@ -7,21 +7,30 @@ $end_info$
 
 #include "Interface/Context/Context.h"
 #include "Interface/Core/Frontend.h"
-#include "Interface/Core/InternalThreadState.h"
 
 #include <array>
+#include <assert.h>
 #include <algorithm>
 #include <cstring>
+#include <FEXCore/Config/Config.h>
 #include <FEXCore/Core/X86Enums.h>
 #include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/HLE/SyscallHandler.h>
+#include <FEXCore/Utils/Allocator.h>
 #include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/Utils/Telemetry.h>
 #include <set>
+#include <sys/mman.h>
 
 namespace FEXCore::Frontend {
+#include "Interface/Core/VSyscall/VSyscall.inc"
+
 using namespace FEXCore::X86Tables;
 
 static uint32_t MapModRMToReg(uint8_t REX, uint8_t bits, bool HighBits, bool HasREX, bool HasXMM, bool HasMM, uint8_t InvalidOffset = 16) {
-  constexpr std::array<uint64_t, 16> GPRIndexes = {
+  using GPRArray = std::array<uint32_t, 16>;
+
+  static constexpr GPRArray GPRIndexes = {
     // Classical ordering?
     FEXCore::X86State::REG_RAX,
     FEXCore::X86State::REG_RCX,
@@ -41,7 +50,7 @@ static uint32_t MapModRMToReg(uint8_t REX, uint8_t bits, bool HighBits, bool Has
     FEXCore::X86State::REG_R15,
   };
 
-  constexpr std::array<uint64_t, 16> GPR8BitHighIndexes = {
+  static constexpr GPRArray GPR8BitHighIndexes = {
     // Classical ordering?
     FEXCore::X86State::REG_RAX,
     FEXCore::X86State::REG_RCX,
@@ -61,7 +70,7 @@ static uint32_t MapModRMToReg(uint8_t REX, uint8_t bits, bool HighBits, bool Has
     FEXCore::X86State::REG_R15,
   };
 
-  constexpr std::array<uint64_t, 16> XMMIndexes = {
+  static constexpr GPRArray XMMIndexes = {
     FEXCore::X86State::REG_XMM_0,
     FEXCore::X86State::REG_XMM_1,
     FEXCore::X86State::REG_XMM_2,
@@ -80,7 +89,7 @@ static uint32_t MapModRMToReg(uint8_t REX, uint8_t bits, bool HighBits, bool Has
     FEXCore::X86State::REG_XMM_15,
   };
 
-  constexpr std::array<uint64_t, 16> MMIndexes = {
+  static constexpr GPRArray MMIndexes = {
     FEXCore::X86State::REG_MM_0,
     FEXCore::X86State::REG_MM_1,
     FEXCore::X86State::REG_MM_2,
@@ -99,7 +108,7 @@ static uint32_t MapModRMToReg(uint8_t REX, uint8_t bits, bool HighBits, bool Has
     FEXCore::X86State::REG_INVALID
   };
 
-  const std::array<uint64_t, 16> *GPRs = &GPRIndexes;
+  const GPRArray *GPRs = &GPRIndexes;
   if (HasXMM) {
     GPRs = &XMMIndexes;
   }
@@ -119,8 +128,18 @@ static uint32_t MapModRMToReg(uint8_t REX, uint8_t bits, bool HighBits, bool Has
 }
 
 Decoder::Decoder(FEXCore::Context::Context *ctx)
-  : CTX {ctx} {
-  DecodedBuffer.resize(DefaultDecodedBufferSize);
+  : CTX {ctx}
+  , OSABI { ctx->SyscallHandler ? ctx->SyscallHandler->GetOSABI() : FEXCore::HLE::SyscallOSABI::OS_UNKNOWN } {
+  // Using mmap is a start-up time optimization
+  // Take advantage of page faulting to reduce startup time for minimal runtime cost
+  DecodedBuffer =
+    reinterpret_cast<FEXCore::X86Tables::DecodedInst *>(
+      FEXCore::Allocator::mmap(0, sizeof(FEXCore::X86Tables::DecodedInst) * DefaultDecodedBufferSize,
+      PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0));
+}
+
+Decoder::~Decoder() {
+  FEXCore::Allocator::munmap(DecodedBuffer, sizeof(FEXCore::X86Tables::DecodedInst) * DefaultDecodedBufferSize);
 }
 
 uint8_t Decoder::ReadByte() {
@@ -131,7 +150,7 @@ uint8_t Decoder::ReadByte() {
   return Byte;
 }
 
-uint8_t Decoder::PeekByte(uint8_t Offset) {
+uint8_t Decoder::PeekByte(uint8_t Offset) const {
   uint8_t Byte = InstStream[InstructionSize + Offset];
   return Byte;
 }
@@ -197,9 +216,9 @@ void Decoder::DecodeModRM_16(X86Tables::DecodedOperand *Operand, X86Tables::ModR
     }
   }
 
-  Operand->TypeSIB.Type = DecodedOperand::TYPE_SIB;
-  Operand->TypeSIB.Scale = 1;
-  Operand->TypeSIB.Offset = Literal;
+  Operand->Type = DecodedOperand::OpType::SIB;
+  Operand->Data.SIB.Scale = 1;
+  Operand->Data.SIB.Offset = Literal;
 
   // Only called when ModRM.mod != 0b11
   struct Encodings {
@@ -212,34 +231,34 @@ void Decoder::DecodeModRM_16(X86Tables::DecodedOperand *Operand, X86Tables::ModR
     {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_RDI},
     {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_RSI},
     {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_RDI},
-    {FEXCore::X86State::REG_RSI, 255},
-    {FEXCore::X86State::REG_RDI, 255},
-    {255, 255},
-    {FEXCore::X86State::REG_RBX, 255},
+    {FEXCore::X86State::REG_RSI, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RDI, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_INVALID, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_INVALID},
     // Mod = 0b01
     {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_RSI},
     {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_RDI},
     {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_RSI},
     {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_RDI},
-    {FEXCore::X86State::REG_RSI, 255},
-    {FEXCore::X86State::REG_RDI, 255},
-    {FEXCore::X86State::REG_RBP, 255},
-    {FEXCore::X86State::REG_RBX, 255},
+    {FEXCore::X86State::REG_RSI, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RDI, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_INVALID},
     // Mod = 0b10
     {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_RSI},
     {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_RDI},
     {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_RSI},
     {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_RDI},
-    {FEXCore::X86State::REG_RSI, 255},
-    {FEXCore::X86State::REG_RDI, 255},
-    {FEXCore::X86State::REG_RBP, 255},
-    {FEXCore::X86State::REG_RBX, 255},
+    {FEXCore::X86State::REG_RSI, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RDI, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RBP, FEXCore::X86State::REG_INVALID},
+    {FEXCore::X86State::REG_RBX, FEXCore::X86State::REG_INVALID},
   }};
 
   uint8_t LookupIndex = ModRM.mod << 3 | ModRM.rm;
   auto it = Lookup[LookupIndex];
-  Operand->TypeSIB.Base = it.Base;
-  Operand->TypeSIB.Index = it.Index;
+  Operand->Data.SIB.Base = it.Base;
+  Operand->Data.SIB.Index = it.Index;
 }
 
 void Decoder::DecodeModRM_64(X86Tables::DecodedOperand *Operand, X86Tables::ModRMDecoded ModRM) {
@@ -277,12 +296,12 @@ void Decoder::DecodeModRM_64(X86Tables::DecodedOperand *Operand, X86Tables::ModR
     }
 
     // SIB
-    Operand->TypeSIB.Type = DecodedOperand::TYPE_SIB;
-    Operand->TypeSIB.Scale = 1 << SIB.scale;
+    Operand->Type = DecodedOperand::OpType::SIB;
+    Operand->Data.SIB.Scale = 1 << SIB.scale;
 
     // The invalid encoding types are described at Table 1-12. "promoted nsigned is always non-zero"
-    Operand->TypeSIB.Index = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_X ? 1 : 0, SIB.index, false, false, false, false, 0b100);
-    Operand->TypeSIB.Base  = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, SIB.base, false, false, false, false, ModRM.mod == 0 ? 0b101 : 16);
+    Operand->Data.SIB.Index = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_X ? 1 : 0, SIB.index, false, false, false, false, 0b100);
+    Operand->Data.SIB.Base  = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, SIB.base, false, false, false, false, ModRM.mod == 0 ? 0b101 : 16);
 
     uint64_t Literal {0};
     LOGMAN_THROW_A(Displacement <= 4, "Number of bytes should be <= 4 for literal src");
@@ -291,7 +310,7 @@ void Decoder::DecodeModRM_64(X86Tables::DecodedOperand *Operand, X86Tables::ModR
     if (Displacement == 1) {
       Literal = static_cast<int8_t>(Literal);
     }
-    Operand->TypeSIB.Offset = Literal;
+    Operand->Data.SIB.Offset = Literal;
   }
   else if (ModRM.mod == 0) {
     // Explained in Table 1-14. "Operand Addressing Using ModRM and SIB Bytes"
@@ -300,13 +319,13 @@ void Decoder::DecodeModRM_64(X86Tables::DecodedOperand *Operand, X86Tables::ModR
       uint32_t Literal;
       Literal = ReadData(4);
 
-      Operand->TypeRIPLiteral.Type = DecodedOperand::TYPE_RIP_RELATIVE;
-      Operand->TypeRIPLiteral.Literal.u = Literal;
+      Operand->Type = DecodedOperand::OpType::RIPRelative;
+      Operand->Data.RIPLiteral.Value.u = Literal;
     }
     else {
       // Register-direct addressing
-      Operand->TypeGPR.Type = DecodedOperand::TYPE_GPR_DIRECT;
-      Operand->TypeGPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, ModRM.rm, false, false, false, false);
+      Operand->Type = DecodedOperand::OpType::GPRDirect;
+      Operand->Data.GPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, ModRM.rm, false, false, false, false);
     }
   }
   else {
@@ -318,9 +337,9 @@ void Decoder::DecodeModRM_64(X86Tables::DecodedOperand *Operand, X86Tables::ModR
     }
     Displacement = DisplacementSize;
 
-    Operand->TypeGPRIndirect.Type = DecodedOperand::TYPE_GPR_INDIRECT;
-    Operand->TypeGPRIndirect.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, ModRM.rm, false, false, false, false);
-    Operand->TypeGPRIndirect.Displacement = Literal;
+    Operand->Type = DecodedOperand::OpType::GPRIndirect;
+    Operand->Data.GPRIndirect.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, ModRM.rm, false, false, false, false);
+    Operand->Data.GPRIndirect.Displacement = Literal;
   }
 }
 
@@ -460,9 +479,9 @@ bool Decoder::NormalOp(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op)
   if (HAS_NON_XMM_SUBFLAG(Info->Flags, FEXCore::X86Tables::InstFlags::FLAGS_SF_DST_RAX) ||
       HAS_NON_XMM_SUBFLAG(Info->Flags, FEXCore::X86Tables::InstFlags::FLAGS_SF_DST_RDX)) {
     // Some instructions hardcode their destination as RAX
-    CurrentDest->TypeGPR.Type = DecodedOperand::TYPE_GPR;
-    CurrentDest->TypeGPR.HighBits = false;
-    CurrentDest->TypeGPR.GPR = HAS_NON_XMM_SUBFLAG(Info->Flags, FEXCore::X86Tables::InstFlags::FLAGS_SF_DST_RAX) ? FEXCore::X86State::REG_RAX : FEXCore::X86State::REG_RDX;
+    CurrentDest->Type = DecodedOperand::OpType::GPR;
+    CurrentDest->Data.GPR.HighBits = false;
+    CurrentDest->Data.GPR.GPR = HAS_NON_XMM_SUBFLAG(Info->Flags, FEXCore::X86Tables::InstFlags::FLAGS_SF_DST_RAX) ? FEXCore::X86State::REG_RAX : FEXCore::X86State::REG_RDX;
     CurrentDest = &DecodeInst->Src[0];
   }
 
@@ -473,11 +492,11 @@ bool Decoder::NormalOp(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op)
     // This also means that the destination is always a GPR on these ones
     // ADDITIONALLY:
     // If there is a REX prefix then that allows extended GPR usage
-    CurrentDest->TypeGPR.Type = DecodedOperand::TYPE_GPR;
-    DecodeInst->Dest.TypeGPR.HighBits = (Is8BitDest && !HasREX && (Op & 0b111) >= 0b100) || HasHighXMM;
-    CurrentDest->TypeGPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, Op & 0b111, Is8BitDest, HasREX, false, false);
+    CurrentDest->Type = DecodedOperand::OpType::GPR;
+    DecodeInst->Dest.Data.GPR.HighBits = (Is8BitDest && !HasREX && (Op & 0b111) >= 0b100) || HasHighXMM;
+    CurrentDest->Data.GPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, Op & 0b111, Is8BitDest, HasREX, false, false);
 
-    if (CurrentDest->TypeGPR.GPR  == FEXCore::X86State::REG_INVALID)
+    if (CurrentDest->Data.GPR.GPR == FEXCore::X86State::REG_INVALID)
       return false;
   }
 
@@ -501,20 +520,20 @@ bool Decoder::NormalOp(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op)
     ModRM.Hex = DecodeInst->ModRM;
 
     // Decode the GPR source first
-    GPR.TypeGPR.Type = DecodedOperand::TYPE_GPR;
-    GPR.TypeGPR.HighBits = (GPR8Bit && ModRM.reg >= 0b100 && !HasREX) || HasHighXMM;
-    GPR.TypeGPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_R ? 1 : 0, ModRM.reg, GPR8Bit, HasREX, HasXMMGPR, HasMMGPR);
+    GPR.Type = DecodedOperand::OpType::GPR;
+    GPR.Data.GPR.HighBits = (GPR8Bit && ModRM.reg >= 0b100 && !HasREX) || HasHighXMM;
+    GPR.Data.GPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_R ? 1 : 0, ModRM.reg, GPR8Bit, HasREX, HasXMMGPR, HasMMGPR);
 
-    if (GPR.TypeGPR.GPR == FEXCore::X86State::REG_INVALID)
+    if (GPR.Data.GPR.GPR == FEXCore::X86State::REG_INVALID)
       return false;
 
     // ModRM.mod == 0b11 == Register
     // ModRM.Mod != 0b11 == Register-direct addressing
     if (ModRM.mod == 0b11) {
-      NonGPR.TypeGPR.Type = DecodedOperand::TYPE_GPR;
-      NonGPR.TypeGPR.HighBits = (NonGPR8Bit && ModRM.rm >= 0b100 && !HasREX) || HasHighXMM;
-      NonGPR.TypeGPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, ModRM.rm, NonGPR8Bit, HasREX, HasXMMNonGPR, HasMMNonGPR);
-      if (NonGPR.TypeGPR.GPR == FEXCore::X86State::REG_INVALID)
+      NonGPR.Type = DecodedOperand::OpType::GPR;
+      NonGPR.Data.GPR.HighBits = (NonGPR8Bit && ModRM.rm >= 0b100 && !HasREX) || HasHighXMM;
+      NonGPR.Data.GPR.GPR = MapModRMToReg(DecodeInst->Flags & DecodeFlags::FLAG_REX_XGPR_B ? 1 : 0, ModRM.rm, NonGPR8Bit, HasREX, HasXMMNonGPR, HasMMNonGPR);
+      if (NonGPR.Data.GPR.GPR == FEXCore::X86State::REG_INVALID)
         return false;
     }
     else {
@@ -540,25 +559,24 @@ bool Decoder::NormalOp(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op)
   }
 
   if (HAS_NON_XMM_SUBFLAG(Info->Flags, FEXCore::X86Tables::InstFlags::FLAGS_SF_SRC_RAX)) {
-    DecodeInst->Src[CurrentSrc].TypeGPR.Type = DecodedOperand::TYPE_GPR;
-    DecodeInst->Src[CurrentSrc].TypeGPR.HighBits = false;
-    DecodeInst->Src[CurrentSrc].TypeGPR.GPR = FEXCore::X86State::REG_RAX;
+    DecodeInst->Src[CurrentSrc].Type = DecodedOperand::OpType::GPR;
+    DecodeInst->Src[CurrentSrc].Data.GPR.HighBits = false;
+    DecodeInst->Src[CurrentSrc].Data.GPR.GPR = FEXCore::X86State::REG_RAX;
     ++CurrentSrc;
   }
   else if (HAS_NON_XMM_SUBFLAG(Info->Flags, FEXCore::X86Tables::InstFlags::FLAGS_SF_SRC_RCX)) {
-    DecodeInst->Src[CurrentSrc].TypeGPR.Type = DecodedOperand::TYPE_GPR;
-    DecodeInst->Src[CurrentSrc].TypeGPR.HighBits = false;
-    DecodeInst->Src[CurrentSrc].TypeGPR.GPR = FEXCore::X86State::REG_RCX;
+    DecodeInst->Src[CurrentSrc].Type = DecodedOperand::OpType::GPR;
+    DecodeInst->Src[CurrentSrc].Data.GPR.HighBits = false;
+    DecodeInst->Src[CurrentSrc].Data.GPR.GPR = FEXCore::X86State::REG_RCX;
     ++CurrentSrc;
   }
 
   if (Bytes != 0) {
     LOGMAN_THROW_A(Bytes <= 8, "Number of bytes should be <= 8 for literal src");
 
-    DecodeInst->Src[CurrentSrc].TypeLiteral.Size = Bytes;
+    DecodeInst->Src[CurrentSrc].Data.Literal.Size = Bytes;
 
-    uint64_t Literal {0};
-    Literal = ReadData(Bytes);
+    uint64_t Literal = ReadData(Bytes);
 
     if ((Info->Flags & FEXCore::X86Tables::InstFlags::FLAGS_SRC_SEXT) ||
         (DecodeFlags::GetSizeDstFlags(DecodeInst->Flags) == DecodeFlags::SIZE_64BIT && Info->Flags & FEXCore::X86Tables::InstFlags::FLAGS_SRC_SEXT64BIT)) {
@@ -571,12 +589,12 @@ bool Decoder::NormalOp(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op)
       else {
         Literal = static_cast<int32_t>(Literal);
       }
-      DecodeInst->Src[CurrentSrc].TypeLiteral.Size = DestSize;
+      DecodeInst->Src[CurrentSrc].Data.Literal.Size = DestSize;
     }
 
     Bytes = 0;
-    DecodeInst->Src[CurrentSrc].TypeLiteral.Type = DecodedOperand::TYPE_LITERAL;
-    DecodeInst->Src[CurrentSrc].TypeLiteral.Literal = Literal;
+    DecodeInst->Src[CurrentSrc].Type = DecodedOperand::OpType::Literal;
+    DecodeInst->Src[CurrentSrc].Data.Literal.Value = Literal;
   }
 
   LOGMAN_THROW_A(Bytes == 0, "Inst at 0x%lx: 0x%04x '%s' Had an instruction of size %d with %d remaining", DecodeInst->PC, DecodeInst->OP, DecodeInst->TableInfo->Name, InstructionSize, Bytes);
@@ -682,6 +700,7 @@ bool Decoder::NormalOpHeader(FEXCore::X86Tables::X86InstInfo const *Info, uint16
     return NormalOp(&X87Ops[X87Op], X87Op);
   }
   else if (Info->Type == FEXCore::X86Tables::TYPE_VEX_TABLE_PREFIX) {
+    FEXCORE_TELEMETRY_SET(VEXOpTelem, 1);
     uint16_t map_select = 1;
     uint16_t pp = 0;
 
@@ -709,6 +728,7 @@ bool Decoder::NormalOpHeader(FEXCore::X86Tables::X86InstInfo const *Info, uint16
 
     if (LocalInfo->Type >= FEXCore::X86Tables::TYPE_VEX_GROUP_12 &&
         LocalInfo->Type <= FEXCore::X86Tables::TYPE_VEX_GROUP_17) {
+    FEXCORE_TELEMETRY_SET(VEXOpTelem, 1);
     // We have ModRM
     uint8_t ModRMByte = ReadByte();
     DecodeInst->ModRM = ModRMByte;
@@ -726,6 +746,8 @@ bool Decoder::NormalOpHeader(FEXCore::X86Tables::X86InstInfo const *Info, uint16
       return NormalOp(LocalInfo, Op);
   }
   else if (Info->Type == FEXCore::X86Tables::TYPE_GROUP_EVEX) {
+    FEXCORE_TELEMETRY_SET(EVEXOpTelem, 1);
+
     /* uint8_t P1 = */ ReadByte();
     /* uint8_t P2 = */ ReadByte();
     /* uint8_t P3 = */ ReadByte();
@@ -927,8 +949,8 @@ bool Decoder::DecodeInstruction(uint64_t PC) {
 
   }
 
-  if (DecodeInst->Dest.TypeNone.Type == FEXCore::X86Tables::DecodedOperand::TYPE_GPR) {
-    assert(DecodeInst->Dest.TypeGPR.GPR != 255);
+  if (DecodeInst->Dest.IsGPR()) {
+    assert(DecodeInst->Dest.Data.GPR.GPR != FEXCore::X86State::REG_INVALID);
   }
 
   return true;
@@ -940,7 +962,7 @@ void Decoder::BranchTargetInMultiblockRange() {
 
   // If the RIP setting is conditional AND within our symbol range then it can be considered for multiblock
   uint64_t TargetRIP = 0;
-  uint8_t GPRSize = CTX->Config.Is64BitMode ? 8 : 4;
+  const uint8_t GPRSize = CTX->GetGPRSize();
   bool Conditional = true;
 
   switch (DecodeInst->OP) {
@@ -950,14 +972,14 @@ void Decoder::BranchTargetInMultiblockRange() {
       // auto RIPOffset = LoadSource(Op, Op->Src[0], Op->Flags);
       // auto RIPTargetConst = _Constant(Op->PC + Op->InstSize);
       // Target offset is PC + InstSize + Literal
-      LOGMAN_THROW_A(DecodeInst->Src[0].TypeNone.Type == DecodedOperand::TYPE_LITERAL, "Had wrong operand type");
-      TargetRIP = DecodeInst->PC + DecodeInst->InstSize + DecodeInst->Src[0].TypeLiteral.Literal;
+      LOGMAN_THROW_A(DecodeInst->Src[0].IsLiteral(), "Had wrong operand type");
+      TargetRIP = DecodeInst->PC + DecodeInst->InstSize + DecodeInst->Src[0].Data.Literal.Value;
     break;
     }
     case 0xE9:
     case 0xEB: // Both are unconditional JMP instructions
-      LOGMAN_THROW_A(DecodeInst->Src[0].TypeNone.Type == DecodedOperand::TYPE_LITERAL, "Had wrong operand type");
-      TargetRIP = DecodeInst->PC + DecodeInst->InstSize + DecodeInst->Src[0].TypeLiteral.Literal;
+      LOGMAN_THROW_A(DecodeInst->Src[0].IsLiteral(), "Had wrong operand type");
+      TargetRIP = DecodeInst->PC + DecodeInst->InstSize + DecodeInst->Src[0].Data.Literal.Value;
       Conditional = false;
     break;
     case 0xE8: // Call - Immediate target, We don't want to inline calls
@@ -1003,6 +1025,25 @@ void Decoder::BranchTargetInMultiblockRange() {
   }
 }
 
+const uint8_t *Decoder::AdjustAddrForSpecialRegion(uint8_t const* _InstStream, uint64_t EntryPoint, uint64_t RIP) {
+  constexpr uint64_t VSyscall_Base = 0xFFFF'FFFF'FF60'0000ULL;
+  constexpr uint64_t VSyscall_End = VSyscall_Base + 0x1000;
+
+  if (OSABI == FEXCore::HLE::SyscallOSABI::OS_LINUX64 &&
+      RIP >= VSyscall_Base &&
+      RIP < VSyscall_End) {
+    // VSyscall
+    // This doesn't exist on AArch64 and on x86_64 hosts this is emulated with faults to a region mapped with --xp permissions
+    // Offset     0: vgettimeofday
+    // Offset 0x400: vtime
+    // Offset 0x800: vgetcpu
+    uint64_t Offset = RIP - VSyscall_Base;
+    return VSyscallData + Offset;
+  }
+
+  return _InstStream - EntryPoint + RIP;
+}
+
 bool Decoder::DecodeInstructionsAtEntry(uint8_t const* _InstStream, uint64_t PC) {
   Blocks.clear();
   BlocksToDecode.clear();
@@ -1046,7 +1087,7 @@ bool Decoder::DecodeInstructionsAtEntry(uint8_t const* _InstStream, uint64_t PC)
     uint64_t BlockStartOffset = DecodedSize;
 
     // Do a bit of pointer math to figure out where we are in code
-    InstStream = _InstStream - EntryPoint + RIPToDecode;
+    InstStream = AdjustAddrForSpecialRegion(_InstStream, EntryPoint, RIPToDecode);
 
     while (1) {
       ErrorDuringDecoding = !DecodeInstruction(RIPToDecode + PCOffset);
@@ -1090,7 +1131,7 @@ bool Decoder::DecodeInstructionsAtEntry(uint8_t const* _InstStream, uint64_t PC)
       }
 
       if (DecodedSize >= CTX->Config.MaxInstPerBlock ||
-          DecodedSize >= DecodedBuffer.size()) {
+          DecodedSize >= DefaultDecodedBufferSize) {
         break;
       }
 
@@ -1107,7 +1148,7 @@ bool Decoder::DecodeInstructionsAtEntry(uint8_t const* _InstStream, uint64_t PC)
 
     // Copy over only the number of instructions we decoded
     CurrentBlockDecoding.NumInstructions = BlockNumberOfInstructions;
-    CurrentBlockDecoding.DecodedInstructions = &DecodedBuffer.at(BlockStartOffset);
+    CurrentBlockDecoding.DecodedInstructions = &DecodedBuffer[BlockStartOffset];
   }
 
 

External/FEXCore/Source/Interface/Core/Frontend.h

@@ -1,11 +1,13 @@
 #pragma once
 
 #include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/HLE/SyscallHandler.h>
+#include <FEXCore/Utils/Telemetry.h>
+
 #include <array>
 #include <cstdint>
-#include <utility>
 #include <set>
-#include <stack>
+#include <stddef.h>
 #include <vector>
 
 namespace FEXCore::Context {
@@ -24,9 +26,10 @@ public:
   };
 
   Decoder(FEXCore::Context::Context *ctx);
+  ~Decoder();
   bool DecodeInstructionsAtEntry(uint8_t const* InstStream, uint64_t PC);
 
-  std::vector<DecodedBlocks> const *GetDecodedBlocks() {
+  std::vector<DecodedBlocks> const *GetDecodedBlocks() const {
     return &Blocks;
   }
 
@@ -37,20 +40,21 @@ public:
   void SetExternalBranches(std::set<uint64_t> *v) { ExternalBranches = v; }
 private:
   FEXCore::Context::Context *CTX;
+  const FEXCore::HLE::SyscallOSABI OSABI{};
 
   bool DecodeInstruction(uint64_t PC);
 
   void BranchTargetInMultiblockRange();
 
   uint8_t ReadByte();
-  uint8_t PeekByte(uint8_t Offset);
+  uint8_t PeekByte(uint8_t Offset) const;
   uint64_t ReadData(uint8_t Size);
   void SkipBytes(uint8_t Size) { InstructionSize += Size; }
   bool NormalOp(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op);
   bool NormalOpHeader(FEXCore::X86Tables::X86InstInfo const *Info, uint16_t Op);
 
   static constexpr size_t DefaultDecodedBufferSize = 0x10000;
-  std::vector<FEXCore::X86Tables::DecodedInst> DecodedBuffer;
+  FEXCore::X86Tables::DecodedInst *DecodedBuffer{};
   size_t DecodedSize {};
 
   uint8_t const *InstStream;
@@ -83,5 +87,10 @@ private:
     &FEXCore::Frontend::Decoder::DecodeModRM_64,
     &FEXCore::Frontend::Decoder::DecodeModRM_16,
   };
+
+  const uint8_t *AdjustAddrForSpecialRegion(uint8_t const* _InstStream, uint64_t EntryPoint, uint64_t RIP);
+
+  FEXCORE_TELEMETRY_INIT(VEXOpTelem, TYPE_USES_VEX_OPS);
+  FEXCORE_TELEMETRY_INIT(EVEXOpTelem, TYPE_USES_EVEX_OPS);
 };
 }

External/FEXCore/Source/Interface/Core/GdbServer.cpp

@@ -8,46 +8,61 @@ $end_info$
 #include <cstdlib>
 #include <cstdio>
 #include <iomanip>
-#include <iostream>
 #include <sstream>
 #include <string>
 #include <memory>
 #include <optional>
 #include "Common/NetStream.h"
 #include "Common/SoftFloat.h"
-#include <FEXCore/Utils/LogManager.h>
+#include "Interface/Context/Context.h"
 
-#include <sys/types.h>
-#include <sys/socket.h>
-#include <netdb.h>
-#include <string.h>
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Core/Context.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <FEXCore/HLE/Linux/ThreadManagement.h>
+#include <FEXCore/Utils/CompilerDefs.h>
+#include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/Utils/Threads.h>
+
+#include <atomic>
+#include <cstring>
+#include <errno.h>
 #include <fcntl.h>
-#include <unistd.h>
 #include <fstream>
+#include <fmt/format.h>
+#include <netdb.h>
+#include <signal.h>
+#include <stddef.h>
+#include <string_view>
+#include <sys/socket.h>
+#include <unistd.h>
+#include <utility>
+#include <vector>
 
 #include "GdbServer.h"
-#include <FEXCore/Core/CodeLoader.h>
-#include <FEXCore/Core/X86Enums.h>
 
 namespace FEXCore
 {
 
 void GdbServer::Break(int signal) {
     std::lock_guard lk(sendMutex);
+    if (!CommsStream) {
+      return;
+    }
 
-    std::ostringstream ss;
-    ss << "S" << std::setfill('0') << std::setw(2) << std::hex << signal;
-
-    if (CommsStream)
-        SendPacket(*CommsStream, ss.str());
+    const auto str = fmt::format("S{:02x}", signal);
+    SendPacket(*CommsStream, str);
 }
 
 GdbServer::GdbServer(FEXCore::Context::Context *ctx) : CTX(ctx) {
-    ctx->CustomExitHandler = [this](uint64_t ThreadId, FEXCore::Context::ExitReason ExitReason) {
+    Context::SetExitHandler(ctx, [this](uint64_t ThreadId, FEXCore::Context::ExitReason ExitReason) {
         if (ExitReason == FEXCore::Context::ExitReason::EXIT_DEBUG) {
             this->Break(SIGTRAP);
         }
-    };
+    });
 
     // This is a total hack as there is currently no way to resume once hitting a segfault
     // But it's semi-useful for debugging.
@@ -60,12 +75,12 @@ GdbServer::GdbServer(FEXCore::Context::Context *ctx) : CTX(ctx) {
           usleep(100000);
 
         return true;
-    });
+    }, true);
 
     StartThread();
 }
 
-static int calculateChecksum(std::string &packet) {
+static int calculateChecksum(const std::string &packet) {
     unsigned char checksum = 0;
     for (const char &c : packet) {
         checksum += c;
@@ -99,11 +114,9 @@ static std::string encodeHex(unsigned char *data, size_t length) {
 }
 
 static std::string getThreadName(uint32_t ThreadID) {
-  std::fstream fs;
-  std::ostringstream ThreadFile;
-  ThreadFile << "/proc/" << getpid() << "/task/" << ThreadID << "/comm";
+  const auto ThreadFile = fmt::format("/proc/{}/task/{}/comm", getpid(), ThreadID);
+  std::fstream fs(ThreadFile, std::fstream::in | std::fstream::binary);
 
-  fs.open(ThreadFile.str(), std::fstream::in | std::fstream::binary);
   if (fs.is_open()) {
     std::string ThreadName;
     fs >> ThreadName;
@@ -135,7 +148,7 @@ std::string GdbServer::ReadPacket(std::iostream &stream) {
         switch(c) {
         case '$': // start of packet
             if (packet.size() != 0)
-                LogMan::Msg::E("Dropping unexpected data: \"%s\"", packet.c_str());
+                LogMan::Msg::EFmt("Dropping unexpected data: \"{}\"", packet);
 
             // clear any existing data, must have been a mistake.
             packet = std::string();
@@ -156,7 +169,7 @@ std::string GdbServer::ReadPacket(std::iostream &stream) {
             if (calculateChecksum(packet) == expected_checksum) {
                 return packet;
             } else {
-                LogMan::Msg::E("Received Invalid Packet: $%s#%02x %c%c", packet.c_str(), expected_checksum);
+                LogMan::Msg::EFmt("Received Invalid Packet: ${}#{:02x}", packet, expected_checksum);
             }
             break;
         }
@@ -169,10 +182,10 @@ std::string GdbServer::ReadPacket(std::iostream &stream) {
     return "";
 }
 
-static std::string escapePacket(std::string packet) {
+static std::string escapePacket(const std::string& packet) {
     std::ostringstream ss;
 
-    for(auto &c : packet) {
+    for(const auto &c : packet) {
         switch (c) {
         case '$':
         case '#':
@@ -191,13 +204,11 @@ static std::string escapePacket(std::string packet) {
     return ss.str();
 }
 
-void GdbServer::SendPacket(std::ostream &stream, std::string packet) {
-  auto escaped = escapePacket(packet);
-  std::ostringstream ss;
+void GdbServer::SendPacket(std::ostream &stream, const std::string& packet) {
+  const auto escaped = escapePacket(packet);
+  const auto str = fmt::format("${}#{:02x}", escaped, calculateChecksum(escaped));
 
-  ss << '$' << escaped << '#';
-  ss << std::setfill('0') << std::setw(2) << std::hex << (int)calculateChecksum(escaped);
-  stream << ss.str() << std::flush;
+  stream << str << std::flush;
 }
 
 void GdbServer::SendACK(std::ostream &stream, bool NACK) {
@@ -218,7 +229,7 @@ void GdbServer::SendACK(std::ostream &stream, bool NACK) {
   }
 }
 
-struct __attribute__((packed)) GDBContextDefinition {
+struct FEX_PACKED GDBContextDefinition {
   uint64_t gregs[16];
   uint64_t rip;
   uint32_t eflags;
@@ -279,7 +290,7 @@ std::string GdbServer::readRegs() {
   return encodeHex((unsigned char *)&GDB, sizeof(GDBContextDefinition));
 }
 
-GdbServer::HandledPacketType GdbServer::readReg(std::string& packet) {
+GdbServer::HandledPacketType GdbServer::readReg(const std::string& packet) {
 	size_t addr;
 	auto ss = std::istringstream(packet);
 	ss.get(); // Drop first letter
@@ -357,7 +368,7 @@ GdbServer::HandledPacketType GdbServer::readReg(std::string& packet) {
     return {encodeHex((unsigned char *)(&Empty), sizeof(uint32_t)), HandledPacketType::TYPE_ACK};
   }
 
-  LogMan::Msg::E("Unknown GDB register 0x%lx", addr);
+  LogMan::Msg::EFmt("Unknown GDB register 0x{:x}", addr);
   return {"E00", HandledPacketType::TYPE_ACK};
 }
 
@@ -462,7 +473,7 @@ std::string buildTargetXML() {
     return xml.str();
 }
 
-GdbServer::HandledPacketType GdbServer::handleXfer(std::string &packet) {
+GdbServer::HandledPacketType GdbServer::handleXfer(const std::string &packet) {
     std::string object;
     std::string rw;
     std::string annex;
@@ -548,10 +559,9 @@ GdbServer::HandledPacketType GdbServer::handleXfer(std::string &packet) {
 
 static size_t CheckMemMapping(uint64_t Address, size_t Size) {
   uint64_t AddressEnd = Address + Size;
-
-  std::fstream fs;
-  fs.open("/proc/self/maps", std::fstream::in | std::fstream::binary);
+  std::fstream fs("/proc/self/maps", std::fstream::in | std::fstream::binary);
   std::string Line;
+
   while (std::getline(fs, Line)) {
     if (fs.eof()) break;
     uint64_t Begin, End;
@@ -568,32 +578,29 @@ static size_t CheckMemMapping(uint64_t Address, size_t Size) {
     }
   }
 
-  fs.close();
   return 0;
 }
 
 GdbServer::HandledPacketType GdbServer::handleProgramOffsets() {
-  std::fstream fs;
-  fs.open("/proc/self/maps", std::fstream::in | std::fstream::binary);
+  std::fstream fs("/proc/self/maps", std::fstream::in | std::fstream::binary);
   std::string Line;
   std::string const &RuntimeExecutable = Filename();
+
   while (std::getline(fs, Line)) {
     uint64_t Begin, End;
     char Filename[255];
     if (sscanf(Line.c_str(), "%lx-%lx %*c%*c%*c%*c %*x %*x:%*x %*d%s", &Begin, &End, Filename) == 3) {
       if (RuntimeExecutable == Filename) {
-        std::ostringstream ss;
-        ss << "Text=" << std::hex << Begin << ";Data=" << std::hex << Begin << ";Bss=" << std::hex << Begin;
-        ss << std::flush;
-        return {ss.str(), HandledPacketType::TYPE_ACK};
+        auto str = fmt::format("Text={:x};Data={:x};Bss={:x}", Begin, Begin, Begin);
+        return {std::move(str), HandledPacketType::TYPE_ACK};
       }
     }
   }
-  fs.close();
+
   return {"Text=0;Data=0;Bss=0", HandledPacketType::TYPE_ACK};
 }
 
-GdbServer::HandledPacketType GdbServer::handleMemory(std::string &packet) {
+GdbServer::HandledPacketType GdbServer::handleMemory(const std::string &packet) {
     bool write;
     size_t addr;
     size_t length;
@@ -634,8 +641,8 @@ GdbServer::HandledPacketType GdbServer::handleMemory(std::string &packet) {
 }
 
 
-GdbServer::HandledPacketType GdbServer::handleQuery(std::string &packet) {
-  auto match = [&](const char *str) -> bool { return packet.rfind(str, 0) == 0; };
+GdbServer::HandledPacketType GdbServer::handleQuery(const std::string &packet) {
+  const auto match = [&](const char *str) -> bool { return packet.rfind(str, 0) == 0; };
 
   if (match("qSupported")) {
     return {"PacketSize=5000;xmlRegisters=i386;qXfer:exec-file:read+;qXfer:features:read+;", HandledPacketType::TYPE_ACK};
@@ -693,8 +700,8 @@ GdbServer::HandledPacketType GdbServer::handleQuery(std::string &packet) {
   return {"", HandledPacketType::TYPE_UNKNOWN};
 }
 
-GdbServer::HandledPacketType GdbServer::handleV(std::string& packet) {
-    auto match = [&](std::string str) -> std::optional<std::istringstream> {
+GdbServer::HandledPacketType GdbServer::handleV(const std::string& packet) {
+    const auto match = [&](const std::string& str) -> std::optional<std::istringstream> {
         if (packet.rfind(str, 0) == 0) {
             auto ss = std::istringstream(packet);
             ss.seekg(str.size());
@@ -703,18 +710,11 @@ GdbServer::HandledPacketType GdbServer::handleV(std::string& packet) {
         return std::nullopt;
     };
 
-    auto F = [](int result) {
-        std::ostringstream ss;
-        ss << "F" << std::hex << result;
-        return ss.str(); };
-    auto F_error = [&]() {
-        std::ostringstream ss;
-        ss << "F-1," << std::hex << errno;
-        return ss.str(); };
-    auto F_data = [&](int result, std::string data) {
-        std::ostringstream ss;
-        ss << "F" << std::hex << result << ";" << data;
-        return ss.str(); };
+    const auto F       = [](int result) { return fmt::format("F{:x}", result); };
+    const auto F_error = [] { return fmt::format("F-1,{:x}", errno); };
+    const auto F_data  = [](int result, const std::string& data) {
+        return fmt::format("F{:x};{}", result, data);
+    };
 
     std::optional<std::istringstream> ss;
     if((ss = match("vFile:open:"))) {
@@ -736,11 +736,11 @@ GdbServer::HandledPacketType GdbServer::handleV(std::string& packet) {
         return {F(pid == 0 ? 0 : -1), HandledPacketType::TYPE_ACK}; // Only support the common filesystem
     }
     if((ss = match("vFile:close:"))) {
-			int fd;
-			*ss >> std::hex >> fd;
-			close(fd);
-			return {F(0), HandledPacketType::TYPE_ACK};
-		}
+      int fd;
+      *ss >> std::hex >> fd;
+      close(fd);
+      return {F(0), HandledPacketType::TYPE_ACK};
+    }
     if((ss = match("vFile:pread:"))) {
         int fd, count, offset;
 
@@ -777,7 +777,7 @@ GdbServer::HandledPacketType GdbServer::handleV(std::string& packet) {
         }
 
         if (ss->fail()) {
- 						return {"E00", HandledPacketType::TYPE_ACK};
+            return {"E00", HandledPacketType::TYPE_ACK};
         }
 
         switch (action) {
@@ -787,27 +787,25 @@ GdbServer::HandledPacketType GdbServer::handleV(std::string& packet) {
           }
         case 's': {
             CTX->Step();
-						SendPacketPair({"OK", HandledPacketType::TYPE_ACK});
-            std::ostringstream ss;
-            ss << "T05thread:" << std::setfill('0') << std::setw(2) << std::hex << getpid() << ";core:2c;";
-
-            SendPacketPair({ss.str(), HandledPacketType::TYPE_ACK});
+            SendPacketPair({"OK", HandledPacketType::TYPE_ACK});
+            auto str = fmt::format("T05thread:{:02x};core:2c;", getpid());
+            SendPacketPair({std::move(str), HandledPacketType::TYPE_ACK});
             return {"OK", HandledPacketType::TYPE_ACK};
           }
         case 't':
             // This thread isn't part of the thread pool
             CTX->Stop(false /* Ignore current thread */);
- 						return {"OK", HandledPacketType::TYPE_ACK};
+            return {"OK", HandledPacketType::TYPE_ACK};
         default:
- 						return {"E00", HandledPacketType::TYPE_ACK};
+            return {"E00", HandledPacketType::TYPE_ACK};
         }
 
     }
-		return {"", HandledPacketType::TYPE_ACK};
+    return {"", HandledPacketType::TYPE_ACK};
 }
 
-GdbServer::HandledPacketType GdbServer::handleThreadOp(std::string &packet) {
-  auto match = [&](const char *str) -> bool { return packet.rfind(str, 0) == 0; };
+GdbServer::HandledPacketType GdbServer::handleThreadOp(const std::string &packet) {
+  const auto match = [&](const char *str) -> bool { return packet.rfind(str, 0) == 0; };
 
   if (match("Hc")) {
     // Sets thread to this ID for stepping
@@ -823,7 +821,7 @@ GdbServer::HandledPacketType GdbServer::handleThreadOp(std::string &packet) {
   if (match("Hg")) {
     // Sets thread for "other" operations
     auto ss = std::istringstream(packet);
-    ss.seekg(std::string("Hg").size());
+    ss.seekg(std::string_view("Hg").size());
     ss >> std::hex >> CurrentDebuggingThread;
 
     // This must return quick otherwise IDA complains
@@ -834,7 +832,7 @@ GdbServer::HandledPacketType GdbServer::handleThreadOp(std::string &packet) {
   return {"", HandledPacketType::TYPE_UNKNOWN};
 }
 
-GdbServer::HandledPacketType GdbServer::handleBreakpoint(std::string &packet) {
+GdbServer::HandledPacketType GdbServer::handleBreakpoint(const std::string &packet) {
   auto ss = std::istringstream(packet);
 
   bool Set{};
@@ -850,17 +848,15 @@ GdbServer::HandledPacketType GdbServer::handleBreakpoint(std::string &packet) {
   return {"OK", HandledPacketType::TYPE_ACK};
 }
 
-GdbServer::HandledPacketType GdbServer::ProcessPacket(std::string &packet) {
+GdbServer::HandledPacketType GdbServer::ProcessPacket(const std::string &packet) {
   switch (packet[0]) {
     case '?': {
       // Indicates the reason that the thread has stopped
       // Behaviour changes if the target is in non-stop mode
       // Binja doesn't support S response here
       //return {"S00", HandledPacketType::TYPE_ACK};
-      std::ostringstream ss;
-      ss << "T00thread:" << std::setfill('0') << std::setw(2) << std::hex << getpid() << ";core:2c;";
-
-      return {ss.str(), HandledPacketType::TYPE_ACK};
+      auto str = fmt::format("T00thread:{:02x};core:2c;", getpid());
+      return {std::move(str), HandledPacketType::TYPE_ACK};
     }
     case 'g':
       return {readRegs(), HandledPacketType::TYPE_ACK};
@@ -890,14 +886,14 @@ GdbServer::HandledPacketType GdbServer::ProcessPacket(std::string &packet) {
   }
 }
 
-void GdbServer::SendPacketPair(HandledPacketType response) {
+void GdbServer::SendPacketPair(const HandledPacketType& response) {
   std::lock_guard lk(sendMutex);
   if (response.TypeResponse == HandledPacketType::TYPE_ACK ||
       response.TypeResponse == HandledPacketType::TYPE_ONLYACK) {
     SendACK(*CommsStream, false);
   }
   else if (response.TypeResponse == HandledPacketType::TYPE_NACK ||
-      response.TypeResponse == HandledPacketType::TYPE_ONLYNACK) {
+           response.TypeResponse == HandledPacketType::TYPE_ONLYNACK) {
     SendACK(*CommsStream, true);
   }
 
@@ -905,8 +901,8 @@ void GdbServer::SendPacketPair(HandledPacketType response) {
     SendPacket(*CommsStream, "");
   }
   else if (response.TypeResponse != HandledPacketType::TYPE_ONLYNACK &&
-      response.TypeResponse != HandledPacketType::TYPE_ONLYACK &&
-      response.TypeResponse != HandledPacketType::TYPE_NONE) {
+           response.TypeResponse != HandledPacketType::TYPE_ONLYACK &&
+           response.TypeResponse != HandledPacketType::TYPE_NONE) {
     SendPacket(*CommsStream, response.Response);
   }
 }
@@ -927,7 +923,7 @@ void GdbServer::GdbServerLoop() {
             response = ProcessPacket(packet);
             SendPacketPair(response);
             if (response.TypeResponse == HandledPacketType::TYPE_UNKNOWN) {
-              LogMan::Msg::D("Unknown packet %s", packet.c_str());
+              LogMan::Msg::DFmt("Unknown packet {}", packet);
             }
             break;
         }
@@ -943,19 +939,18 @@ void GdbServer::GdbServerLoop() {
             break;
         case '\x03': { // ASCII EOT
             CTX->Pause();
-            std::ostringstream ss;
-            ss << "T02thread:" << std::setfill('0') << std::setw(2) << std::hex << getpid() << ";core:2c;";
-            SendPacketPair({ss.str(), HandledPacketType::TYPE_ACK});
+            auto str = fmt::format("T02thread:{:02x};core:2c;", getpid());
+            SendPacketPair({std::move(str), HandledPacketType::TYPE_ACK});
             break;
           }
         default:
-            LogMan::Msg::D("GdbServer: Unexpected byte %c (%02x)", c, c);
+            LogMan::Msg::DFmt("GdbServer: Unexpected byte {} ({:02x})", static_cast<char>(c), c);
         }
     }
 
     {
         std::lock_guard lk(sendMutex);
-        CommsStream.release();
+        CommsStream.reset();
     }
   }
 }
@@ -1003,7 +998,7 @@ std::unique_ptr<std::iostream> GdbServer::OpenSocket() {
 
     // Block until a connection arrives
 
-    LogMan::Msg::I("GdbServer, waiting for connection on localhost:8086");
+    LogMan::Msg::IFmt("GdbServer, waiting for connection on localhost:8086");
     listen(sockfd, 1);
 
     new_fd = accept(sockfd, (struct sockaddr *)&their_addr, &addr_size);

External/FEXCore/Source/Interface/Core/GdbServer.h

@@ -5,18 +5,21 @@ $end_info$
 */
 #pragma once
 
-#include <mutex>
-#include <thread>
-
-#include "Interface/Context/Context.h"
-#include "Common/NetStream.h"
-
+#include <FEXCore/Config/Config.h>
 #include <FEXCore/Utils/Threads.h>
 
+#include <istream>
+#include <memory>
 #include <mutex>
+#include <stdint.h>
+#include <string>
 
 namespace FEXCore {
 
+namespace Context {
+  struct Context;
+}
+
 class GdbServer {
 public:
     GdbServer(FEXCore::Context::Context *ctx);
@@ -30,7 +33,7 @@ private:
     std::unique_ptr<std::iostream> OpenSocket();
     void StartThread();
     std::string ReadPacket(std::iostream &stream);
-    void SendPacket(std::ostream &stream, std::string packet);
+    void SendPacket(std::ostream &stream, const std::string& packet);
 
     void SendACK(std::ostream &stream, bool NACK);
 
@@ -47,18 +50,18 @@ private:
       ResponseType TypeResponse{};
     };
 
-    void SendPacketPair(HandledPacketType packetPair);
-    HandledPacketType ProcessPacket(std::string &packet);
-    HandledPacketType handleQuery(std::string &packet);
-    HandledPacketType handleXfer(std::string &packet);
-    HandledPacketType handleMemory(std::string &packet);
-    HandledPacketType handleV(std::string& packet);
-    HandledPacketType handleThreadOp(std::string &packet);
-    HandledPacketType handleBreakpoint(std::string &packet);
+    void SendPacketPair(const HandledPacketType& packetPair);
+    HandledPacketType ProcessPacket(const std::string &packet);
+    HandledPacketType handleQuery(const std::string &packet);
+    HandledPacketType handleXfer(const std::string &packet);
+    HandledPacketType handleMemory(const std::string &packet);
+    HandledPacketType handleV(const std::string& packet);
+    HandledPacketType handleThreadOp(const std::string &packet);
+    HandledPacketType handleBreakpoint(const std::string &packet);
     HandledPacketType handleProgramOffsets();
 
     std::string readRegs();
-    HandledPacketType readReg(std::string& packet);
+    HandledPacketType readReg(const std::string& packet);
 
     FEXCore::Context::Context *CTX;
     std::unique_ptr<FEXCore::Threads::Thread> gdbServerThread;

External/FEXCore/Source/Interface/Core/Interpreter/InterpreterClass.h

@@ -1,6 +1,5 @@
 #pragma once
 
-#include "Interface/Core/LookupCache.h"
 #include "Interface/Core/InternalThreadState.h"
 #include "Interface/Core/Dispatcher/Dispatcher.h"
 

External/FEXCore/Source/Interface/Core/Interpreter/InterpreterCore.cpp

@@ -1,30 +1,31 @@
-#include "Common/MathUtils.h"
-#include "Common/SoftFloat.h"
 #include "Interface/Context/Context.h"
 
 #include "Interface/Core/ArchHelpers/Arm64.h"
 #include "Interface/Core/ArchHelpers/MContext.h"
-#include "Interface/Core/LookupCache.h"
-#include "Interface/Core/DebugData.h"
-#include "Interface/Core/InternalThreadState.h"
+#include "Interface/Core/Dispatcher/Dispatcher.h"
 #include "Interface/Core/Interpreter/InterpreterClass.h"
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Debug/InternalThreadState.h>
 #include <FEXCore/Utils/LogManager.h>
 
-#include <FEXCore/Core/CPUBackend.h>
-#include <FEXCore/HLE/SyscallHandler.h>
-#include <FEXCore/IR/IR.h>
-#include <FEXCore/IR/IntrusiveIRList.h>
-
-#include "Interface/HLE/Thunks/Thunks.h"
-
-#include <atomic>
-#include <cmath>
-#include <limits>
-#include <vector>
+#include <memory>
+#include <bits/types/stack_t.h>
+#include <signal.h>
+#include <stdint.h>
+#include <unordered_map>
+#include <utility>
 
 #include "InterpreterOps.h"
 
+namespace FEXCore::IR {
+  class IRListView;
+  class RegisterAllocationData;
+}
+
 namespace FEXCore::CPU {
+class CPUBackend;
 
 static void InterpreterExecution(FEXCore::Core::CpuStateFrame *Frame) {
   auto Thread = Frame->Thread;
@@ -78,6 +79,18 @@ bool InterpreterCore::HandleSIGBUS(int Signal, void *info, void *ucontext) {
         return false;
       }
     }
+    else if ((Instr & FEXCore::ArchHelpers::Arm64::LDAXR_MASK) == FEXCore::ArchHelpers::Arm64::LDAXR_INST) { // LDAXR*
+      uint64_t BytesToSkip = FEXCore::ArchHelpers::Arm64::HandleAtomicLoadstoreExclusive(ucontext, info);
+      if (BytesToSkip) {
+        // Skip this instruction now
+        ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) + BytesToSkip);
+        return true;
+      }
+      else {
+        LogMan::Msg::E("Unhandled JIT SIGBUS LDAXR: PC: %p Instruction: 0x%08x\n", PC, PC[0]);
+        return false;
+      }
+    }
   }
   return false;
 }
@@ -93,19 +106,19 @@ InterpreterCore::InterpreterCore(FEXCore::Context::Context *ctx, FEXCore::Core::
     CTX->SignalDelegation->RegisterHostSignalHandler(SignalDelegator::SIGNAL_FOR_PAUSE, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       InterpreterCore *Core = reinterpret_cast<InterpreterCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleSignalPause(Signal, info, ucontext);
-    });
+    }, true);
 
     CTX->SignalDelegation->RegisterHostSignalHandler(SIGBUS, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       InterpreterCore *Core = reinterpret_cast<InterpreterCore*>(Thread->CPUBackend.get());
       return Core->HandleSIGBUS(Signal, info, ucontext);
-    });
+    }, true);
 
     auto GuestSignalHandler = [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext, GuestSigAction *GuestAction, stack_t *GuestStack) -> bool {
       InterpreterCore *Core = reinterpret_cast<InterpreterCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleGuestSignal(Signal, info, ucontext, GuestAction, GuestStack);
     };
 
-    for (uint32_t Signal = 0; Signal < SignalDelegator::MAX_SIGNALS; ++Signal) {
+    for (uint32_t Signal = 0; Signal <= SignalDelegator::MAX_SIGNALS; ++Signal) {
       CTX->SignalDelegation->RegisterHostSignalHandlerForGuest(Signal, GuestSignalHandler);
     }
   }
@@ -115,8 +128,8 @@ void *InterpreterCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR:
   return reinterpret_cast<void*>(InterpreterExecution);
 }
 
-FEXCore::CPU::CPUBackend *CreateInterpreterCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread) {
-  return new InterpreterCore(ctx, Thread, CompileThread);
+std::unique_ptr<CPUBackend> CreateInterpreterCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread) {
+  return std::make_unique<InterpreterCore>(ctx, Thread, CompileThread);
 }
 
 }

External/FEXCore/Source/Interface/Core/Interpreter/InterpreterCore.h

@@ -1,5 +1,7 @@
 #pragma once
 
+#include <memory>
+
 namespace FEXCore::Context {
 struct Context;
 }
@@ -11,6 +13,6 @@ namespace FEXCore::Core {
 namespace FEXCore::CPU {
 class CPUBackend;
 
-FEXCore::CPU::CPUBackend *CreateInterpreterCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread);
+std::unique_ptr<CPUBackend> CreateInterpreterCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread);
 
 }

External/FEXCore/Source/Interface/Core/Interpreter/InterpreterOps.h

@@ -1,9 +1,13 @@
+#pragma once
+#include <stdint.h>
+
 namespace FEXCore::Core {
   struct InternalThreadState;
 }
 
 namespace FEXCore::IR {
     class IRListView;
+    struct IROp_Header;
 }
 
 namespace FEXCore::Core{
@@ -32,11 +36,11 @@ namespace FEXCore::CPU {
     FallbackABI ABI;
     void *fn;
   };
-  
+
   class InterpreterOps {
 
     public:
       static void InterpretIR(FEXCore::Core::InternalThreadState *Thread, uint64_t Entry, FEXCore::IR::IRListView *CurrentIR, FEXCore::Core::DebugData *DebugData);
       static bool GetFallbackHandler(IR::IROp_Header *IROp, FallbackInfo *Info);
   };
-};
+};

External/FEXCore/Source/Interface/Core/JIT/Arm64/ALUOps.cpp

@@ -46,7 +46,7 @@ DEF_OP(TruncElementPair) {
       mov(Dst.second, Src.second);
       break;
     }
-    default: LOGMAN_MSG_A("Unhandled Truncation size: %d", Op->Size); break;
+    default: LOGMAN_MSG_A_FMT("Unhandled Truncation size: {}", Op->Size); break;
   }
 }
 
@@ -95,7 +95,7 @@ DEF_OP(Add) {
       case 8:
         add(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), Const);
         break;
-      default: LOGMAN_MSG_A("Unsupported Add size: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unsupported Add size: {}", OpSize);
     }
   } else {
     switch (OpSize) {
@@ -105,7 +105,7 @@ DEF_OP(Add) {
       case 8:
         add(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
         break;
-      default: LOGMAN_MSG_A("Unsupported Add size: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unsupported Add size: {}", OpSize);
     }
   }
 }
@@ -121,7 +121,7 @@ DEF_OP(Sub) {
       case 8:
         sub(GRS(Node), GRS(Op->Header.Args[0].ID()), Const);
         break;
-      default: LOGMAN_MSG_A("Unsupported Sub size: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unsupported Sub size: {}", OpSize);
     }
   } else {
     switch (OpSize) {
@@ -131,7 +131,7 @@ DEF_OP(Sub) {
       case 8:
         sub(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
         break;
-      default: LOGMAN_MSG_A("Unsupported Sub size: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unsupported Sub size: {}", OpSize);
     }
   }
 
@@ -147,7 +147,7 @@ DEF_OP(Neg) {
     case 8:
       neg(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()));
       break;
-    default: LOGMAN_MSG_A("Unsupported Not size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unsupported Neg size: {}", OpSize);
   }
 }
 
@@ -163,7 +163,7 @@ DEF_OP(Mul) {
     case 8:
       mul(Dst, GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
     break;
-    default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Mul size: %d", OpSize);
   }
 }
 
@@ -179,7 +179,7 @@ DEF_OP(UMul) {
     case 8:
       mul(Dst, GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
     break;
-    default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unknown UMul size: {}", OpSize);
   }
 }
 
@@ -216,7 +216,7 @@ DEF_OP(Div) {
       sdiv(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
     break;
     }
-    default: LOGMAN_MSG_A("Unknown DIV Size: %d", Size); break;
+    default: LOGMAN_MSG_A_FMT("Unknown DIV Size: {}", Size); break;
   }
 }
 
@@ -243,7 +243,7 @@ DEF_OP(UDiv) {
       udiv(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
       break;
     }
-    default: LOGMAN_MSG_A("Unknown UDIV Size: %d", Size); break;
+    default: LOGMAN_MSG_A_FMT("Unknown UDIV Size: {}", Size); break;
   }
 }
 
@@ -290,7 +290,7 @@ DEF_OP(Rem) {
       msub(GetReg<RA_64>(Node), TMP1, Divisor, Dividend);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown REM Size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown REM Size: {}", OpSize); break;
   }
 }
 
@@ -332,7 +332,7 @@ DEF_OP(URem) {
       msub(GetReg<RA_64>(Node), TMP1, Divisor, Dividend);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown UREM Size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown UREM Size: {}", OpSize); break;
   }
 }
 
@@ -349,7 +349,7 @@ DEF_OP(MulH) {
     case 8:
       smulh(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
     break;
-    default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Sext size: {}", OpSize);
   }
 }
 
@@ -366,7 +366,7 @@ DEF_OP(UMulH) {
     case 8:
       umulh(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()), GetReg<RA_64>(Op->Header.Args[1].ID()));
     break;
-    default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Sext size: {}", OpSize);
   }
 }
 
@@ -462,7 +462,7 @@ DEF_OP(Ror) {
       break;
       }
 
-      default: LOGMAN_MSG_A("Unhandled ROR size: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unhandled ROR size: {}", OpSize);
     }
   } else {
     switch (OpSize) {
@@ -475,7 +475,7 @@ DEF_OP(Ror) {
       break;
       }
 
-      default: LOGMAN_MSG_A("Unhandled ROR size: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unhandled ROR size: {}", OpSize);
     }
   }
 }
@@ -494,7 +494,7 @@ DEF_OP(Extr) {
     break;
     }
 
-    default: LOGMAN_MSG_A("Unhandled EXTR size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unhandled EXTR size: {}", OpSize);
   }
 }
 
@@ -539,7 +539,7 @@ DEF_OP(LDiv) {
       mov(GetReg<RA_64>(Node), x0);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown LDIV Size: %d", Size); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LDIV Size: {}", Size); break;
   }
 }
 
@@ -582,7 +582,7 @@ DEF_OP(LUDiv) {
       mov(GetReg<RA_64>(Node), x0);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown LUDIV Size: %d", Size); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LUDIV Size: {}", Size); break;
   }
 }
 
@@ -635,7 +635,7 @@ DEF_OP(LRem) {
       mov(GetReg<RA_64>(Node), x0);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown LREM Size: %d", Size); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LREM Size: {}", Size); break;
   }
 }
 
@@ -685,7 +685,7 @@ DEF_OP(LURem) {
       mov(GetReg<RA_64>(Node), x0);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown LUREM Size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LUREM Size: {}", OpSize); break;
   }
 }
 
@@ -699,7 +699,7 @@ DEF_OP(Not) {
     case 8:
       mvn(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()));
       break;
-    default: LOGMAN_MSG_A("Unsupported Not size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unsupported Not size: {}", OpSize);
   }
 }
 
@@ -730,7 +730,7 @@ DEF_OP(Popcount) {
       // fmov has zero extended, unused bytes are zero
       addv(VTMP1.B(), VTMP1.V8B());
       break;
-    default: LOGMAN_MSG_A("Unsupported Popcount size: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unsupported Popcount size: {}", OpSize);
   }
 
   auto Dst = GetReg<RA_32>(Node);
@@ -779,7 +779,7 @@ DEF_OP(FindMSB) {
       clz(Dst, GetReg<RA_64>(Op->Header.Args[0].ID()));
       sub(Dst, TMP1, Dst);
       break;
-    default: LOGMAN_MSG_A("Unknown REV size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown FindMSB size: {}", OpSize); break;
   }
 }
 
@@ -800,7 +800,7 @@ DEF_OP(FindTrailingZeros) {
       rbit(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()));
       clz(GetReg<RA_64>(Node), GetReg<RA_64>(Node));
       break;
-    default: LOGMAN_MSG_A("Unknown size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown FindTrailingZeros size: {}", OpSize); break;
   }
 }
 
@@ -819,7 +819,7 @@ DEF_OP(CountLeadingZeroes) {
     case 8:
       clz(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()));
       break;
-    default: LOGMAN_MSG_A("Unknown size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown CountLeadingZeroes size: {}", OpSize); break;
   }
 }
 
@@ -837,7 +837,7 @@ DEF_OP(Rev) {
     case 8:
       rev(GetReg<RA_64>(Node), GetReg<RA_64>(Op->Header.Args[0].ID()));
       break;
-    default: LOGMAN_MSG_A("Unknown REV size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown REV size: {}", OpSize); break;
   }
 }
 
@@ -859,15 +859,14 @@ DEF_OP(Bfi) {
       bfi(TMP1, GetReg<RA_64>(Op->Header.Args[1].ID()), Op->lsb, Op->Width);
       mov(GetReg<RA_64>(Node), TMP1);
       break;
-    default: LOGMAN_MSG_A("Unknown BFI size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown BFI size: {}", OpSize); break;
   }
 }
 
 DEF_OP(Bfe) {
   auto Op = IROp->C<IR::IROp_Bfe>();
-  uint8_t OpSize = IROp->Size;
-  LOGMAN_THROW_A(OpSize <= 8, "OpSize is too large for BFE: %d", OpSize);
-  LOGMAN_THROW_A(Op->Width != 0, "Invalid BFE width of 0");
+  LOGMAN_THROW_A_FMT(IROp->Size <= 8, "OpSize is too large for BFE: {}", IROp->Size);
+  LOGMAN_THROW_A_FMT(Op->Width != 0, "Invalid BFE width of 0");
 
   auto Dst = GetReg<RA_64>(Node);
   ubfx(Dst, GetReg<RA_64>(Op->Header.Args[0].ID()), Op->lsb, Op->Width);
@@ -881,7 +880,7 @@ DEF_OP(Sbfe) {
   if (OpSize == 8) {
     sbfx(Dst, GetReg<RA_64>(Op->Header.Args[0].ID()), Op->lsb, Op->Width);
   } else {
-    LogMan::Msg::D("Unimplemented Sbfe size");
+    LogMan::Msg::DFmt("Unimplemented Sbfe size");
   }
 }
 
@@ -912,7 +911,7 @@ Condition MapSelectCC(IR::CondClassType Cond) {
   case FEXCore::IR::COND_MI:
   case FEXCore::IR::COND_PL:
   default:
-  LOGMAN_MSG_A("Unsupported compare type");
+  LOGMAN_MSG_A_FMT("Unsupported compare type");
   return Condition::nv;
   }
 }
@@ -930,7 +929,7 @@ DEF_OP(Select) {
   } else if (IsFPR(Op->Cmp1.ID())) {
     fcmp(GRFCMP(Op->Cmp1.ID()), GRFCMP(Op->Cmp2.ID()));
   } else {
-    LOGMAN_MSG_A("Select: Expected GPR or FPR");
+    LOGMAN_MSG_A_FMT("Select: Expected GPR or FPR");
   }
 
   auto cc = MapSelectCC(Op->Cond);
@@ -941,7 +940,7 @@ DEF_OP(Select) {
 
   if (is_const_true || is_const_false) {
     if (is_const_false != true || is_const_true != true || const_true != 1 || const_false != 0) {
-      LOGMAN_MSG_A("Select: Unsupported compare inline parameters");
+      LOGMAN_MSG_A_FMT("Select: Unsupported compare inline parameters");
     }
     cset(GRS(Node), cc);
   } else {
@@ -965,38 +964,53 @@ DEF_OP(VExtractToGPR) {
     case 8:
       umov(GetReg<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()).V2D(), Op->Idx);
     break;
-    default:  LOGMAN_MSG_A("Unhandled ExtractElementSize: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled ExtractElementSize: {}", OpSize);
   }
 }
 
-DEF_OP(Float_ToGPR_ZU) {
-  LogMan::Msg::D("Unimplemented");
-}
-
 DEF_OP(Float_ToGPR_ZS) {
   auto Op = IROp->C<IR::IROp_Float_ToGPR_ZS>();
-  if (Op->Header.ElementSize == 8) {
-    fcvtzs(GetReg<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()).D());
+  aarch64::Register Dst{};
+  aarch64::VRegister Src{};
+  if (Op->SrcElementSize == 8) {
+    Src = GetSrc(Op->Header.Args[0].ID()).D();
   }
   else {
-    fcvtzs(GetReg<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()).S());
+    Src = GetSrc(Op->Header.Args[0].ID()).S();
   }
-}
 
-DEF_OP(Float_ToGPR_U) {
-  LogMan::Msg::D("Unimplemented");
+  if (IROp->Size == 8) {
+    Dst = GetReg<RA_64>(Node);
+  }
+  else {
+    Dst = GetReg<RA_32>(Node);
+  }
+
+  fcvtzs(Dst, Src);
 }
 
 DEF_OP(Float_ToGPR_S) {
   auto Op = IROp->C<IR::IROp_Float_ToGPR_S>();
-  if (Op->Header.ElementSize == 8) {
+
+  aarch64::Register Dst{};
+  aarch64::VRegister Src{};
+  if (Op->SrcElementSize == 8) {
     frinti(VTMP1.D(), GetSrc(Op->Header.Args[0].ID()).D());
-    fcvtzs(GetReg<RA_64>(Node), VTMP1.D());
+    Src = VTMP1.D();
   }
   else {
     frinti(VTMP1.S(), GetSrc(Op->Header.Args[0].ID()).S());
-    fcvtzs(GetReg<RA_32>(Node), VTMP1.S());
+    Src = VTMP1.S();
   }
+
+  if (IROp->Size == 8) {
+    Dst = GetReg<RA_64>(Node);
+  }
+  else {
+    Dst = GetReg<RA_32>(Node);
+  }
+
+  fcvtzs(Dst, Src);
 }
 
 DEF_OP(FCmp) {
@@ -1013,7 +1027,7 @@ DEF_OP(FCmp) {
   bool set = false;
 
   if (Op->Flags & (1 << IR::FCMP_FLAG_EQ)) {
-    LOGMAN_THROW_A(IR::FCMP_FLAG_EQ == 0, "IR::FCMP_FLAG_EQ must equal 0");
+    LOGMAN_THROW_A_FMT(IR::FCMP_FLAG_EQ == 0, "IR::FCMP_FLAG_EQ must equal 0");
     // EQ or unordered
     cset(Dst, Condition::eq); // Z = 1
     csinc(Dst, Dst, xzr, Condition::vc); // IF !V ? Z : 1
@@ -1087,9 +1101,7 @@ void Arm64JITCore::RegisterALUHandlers() {
   REGISTER_OP(SBFE,              Sbfe);
   REGISTER_OP(SELECT,            Select);
   REGISTER_OP(VEXTRACTTOGPR,     VExtractToGPR);
-  REGISTER_OP(FLOAT_TOGPR_ZU,    Float_ToGPR_ZU);
   REGISTER_OP(FLOAT_TOGPR_ZS,    Float_ToGPR_ZS);
-  REGISTER_OP(FLOAT_TOGPR_U,     Float_ToGPR_U);
   REGISTER_OP(FLOAT_TOGPR_S,     Float_ToGPR_S);
   REGISTER_OP(FCMP,              FCmp);
 

External/FEXCore/Source/Interface/Core/JIT/Arm64/AtomicOps.cpp

@@ -34,7 +34,7 @@ DEF_OP(CASPair) {
       mov(Dst.first, TMP3);
       mov(Dst.second, TMP4);
       break;
-    default: LOGMAN_MSG_A("Unsupported: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unsupported: {}", OpSize);
     }
   }
   else {
@@ -44,6 +44,7 @@ DEF_OP(CASPair) {
         aarch64::Label LoopNotExpected;
         aarch64::Label LoopExpected;
         bind(&LoopTop);
+
         ldaxp(TMP2.W(), TMP3.W(), MemOperand(MemSrc));
         cmp(TMP2.W(), Expected.first.W());
         ccmp(TMP3.W(), Expected.second.W(), NoFlag, Condition::eq);
@@ -69,6 +70,7 @@ DEF_OP(CASPair) {
         aarch64::Label LoopNotExpected;
         aarch64::Label LoopExpected;
         bind(&LoopTop);
+
         ldaxp(TMP2.X(), TMP3.X(), MemOperand(MemSrc));
         cmp(TMP2.X(), Expected.first.X());
         ccmp(TMP3.X(), Expected.second.X(), NoFlag, Condition::eq);
@@ -89,7 +91,7 @@ DEF_OP(CASPair) {
         bind(&LoopExpected);
         break;
       }
-      default: LOGMAN_MSG_A("Unsupported: %d", OpSize);
+      default: LOGMAN_MSG_A_FMT("Unsupported: {}", OpSize);
     }
   }
 }
@@ -115,7 +117,7 @@ DEF_OP(CAS) {
     case 2: casalh(TMP2.W(), Desired.W(), MemOperand(MemSrc)); break;
     case 4: casal(TMP2.W(), Desired.W(), MemOperand(MemSrc)); break;
     case 8: casal(TMP2.X(), Desired.X(), MemOperand(MemSrc)); break;
-    default: LOGMAN_MSG_A("Unsupported: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unsupported: {}", OpSize);
     }
     mov(GetReg<RA_64>(Node), TMP2);
   }
@@ -206,7 +208,7 @@ DEF_OP(CAS) {
 
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", OpSize);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", OpSize);
     }
   }
 }
@@ -222,7 +224,7 @@ DEF_OP(AtomicAdd) {
     case 2: staddlh(GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
     case 4: staddl(GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
     case 8: staddl(GetReg<RA_64>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -264,7 +266,7 @@ DEF_OP(AtomicAdd) {
         cbnz(TMP2, &LoopTop);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -281,7 +283,7 @@ DEF_OP(AtomicSub) {
     case 2: staddlh(TMP2.W(), MemOperand(MemSrc)); break;
     case 4: staddl(TMP2.W(), MemOperand(MemSrc)); break;
     case 8: staddl(TMP2.X(), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -323,7 +325,7 @@ DEF_OP(AtomicSub) {
         cbnz(TMP2, &LoopTop);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -340,7 +342,7 @@ DEF_OP(AtomicAnd) {
     case 2: stclrlh(TMP2.W(), MemOperand(MemSrc)); break;
     case 4: stclrl(TMP2.W(), MemOperand(MemSrc)); break;
     case 8: stclrl(TMP2.X(), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -382,7 +384,7 @@ DEF_OP(AtomicAnd) {
         cbnz(TMP2, &LoopTop);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -398,7 +400,7 @@ DEF_OP(AtomicOr) {
     case 2: stsetlh(GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
     case 4: stsetl(GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
     case 8: stsetl(GetReg<RA_64>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -440,7 +442,7 @@ DEF_OP(AtomicOr) {
         cbnz(TMP2, &LoopTop);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -456,7 +458,7 @@ DEF_OP(AtomicXor) {
     case 2: steorlh(GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
     case 4: steorl(GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
     case 8: steorl(GetReg<RA_64>(Op->Header.Args[1].ID()), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -498,7 +500,7 @@ DEF_OP(AtomicXor) {
         cbnz(TMP2, &LoopTop);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -515,36 +517,35 @@ DEF_OP(AtomicSwap) {
     case 2: swplh(TMP2.W(), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 4: swpl(TMP2.W(), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 8: swpl(TMP2.X(), GetReg<RA_64>(Node), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
     // TMP2-TMP3
-    mov(TMP3, GetReg<RA_64>(Op->Header.Args[1].ID()));
     switch (Op->Size) {
       case 1: {
         aarch64::Label LoopTop;
         bind(&LoopTop);
         ldaxrb(TMP2.W(), MemOperand(MemSrc));
-        stlxrb(TMP4.W(), TMP3.W(), MemOperand(MemSrc));
+        stlxrb(TMP4.W(), GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc));
         cbnz(TMP4.W(), &LoopTop);
-        uxtb(GetReg<RA_64>(Node), TMP2.W());
+        uxtb(GetReg<RA_32>(Node), TMP2.W());
         break;
       }
       case 2: {
         aarch64::Label LoopTop;
         bind(&LoopTop);
         ldaxrh(TMP2.W(), MemOperand(MemSrc));
-        stlxrh(TMP4.W(), TMP3.W(), MemOperand(MemSrc));
+        stlxrh(TMP4.W(), GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc));
         cbnz(TMP4.W(), &LoopTop);
-        uxtw(GetReg<RA_64>(Node), TMP2.W());
+        uxtw(GetReg<RA_32>(Node), TMP2.W());
         break;
       }
       case 4: {
         aarch64::Label LoopTop;
         bind(&LoopTop);
         ldaxr(TMP2.W(), MemOperand(MemSrc));
-        stlxr(TMP4.W(), TMP3.W(), MemOperand(MemSrc));
+        stlxr(TMP4.W(), GetReg<RA_32>(Op->Header.Args[1].ID()), MemOperand(MemSrc));
         cbnz(TMP4.W(), &LoopTop);
         mov(GetReg<RA_32>(Node), TMP2.W());
         break;
@@ -553,12 +554,12 @@ DEF_OP(AtomicSwap) {
         aarch64::Label LoopTop;
         bind(&LoopTop);
         ldaxr(TMP2, MemOperand(MemSrc));
-        stlxr(TMP4, TMP3.X(), MemOperand(MemSrc));
+        stlxr(TMP4, GetReg<RA_64>(Op->Header.Args[1].ID()), MemOperand(MemSrc));
         cbnz(TMP4, &LoopTop);
         mov(GetReg<RA_64>(Node), TMP2.X());
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -573,7 +574,7 @@ DEF_OP(AtomicFetchAdd) {
     case 2: ldaddalh(GetReg<RA_32>(Op->Header.Args[1].ID()), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 4: ldaddal(GetReg<RA_32>(Op->Header.Args[1].ID()), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 8: ldaddal(GetReg<RA_64>(Op->Header.Args[1].ID()), GetReg<RA_64>(Node), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -619,7 +620,7 @@ DEF_OP(AtomicFetchAdd) {
         mov(GetReg<RA_64>(Node), TMP2);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -635,7 +636,7 @@ DEF_OP(AtomicFetchSub) {
     case 2: ldaddalh(TMP2.W(), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 4: ldaddal(TMP2.W(), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 8: ldaddal(TMP2.X(), GetReg<RA_64>(Node), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -681,7 +682,7 @@ DEF_OP(AtomicFetchSub) {
         mov(GetReg<RA_64>(Node), TMP2);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -697,7 +698,7 @@ DEF_OP(AtomicFetchAnd) {
     case 2: ldclralh(TMP2.W(), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 4: ldclral(TMP2.W(), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 8: ldclral(TMP2.X(), GetReg<RA_64>(Node), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -743,7 +744,7 @@ DEF_OP(AtomicFetchAnd) {
         mov(GetReg<RA_64>(Node), TMP2);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -758,7 +759,7 @@ DEF_OP(AtomicFetchOr) {
     case 2: ldsetalh(GetReg<RA_32>(Op->Header.Args[1].ID()), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 4: ldsetal(GetReg<RA_32>(Op->Header.Args[1].ID()), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 8: ldsetal(GetReg<RA_64>(Op->Header.Args[1].ID()), GetReg<RA_64>(Node), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -804,7 +805,7 @@ DEF_OP(AtomicFetchOr) {
         mov(GetReg<RA_64>(Node), TMP2);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
@@ -819,7 +820,7 @@ DEF_OP(AtomicFetchXor) {
     case 2: ldeoralh(GetReg<RA_32>(Op->Header.Args[1].ID()), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 4: ldeoral(GetReg<RA_32>(Op->Header.Args[1].ID()), GetReg<RA_32>(Node), MemOperand(MemSrc)); break;
     case 8: ldeoral(GetReg<RA_64>(Op->Header.Args[1].ID()), GetReg<RA_64>(Node), MemOperand(MemSrc)); break;
-    default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
   else {
@@ -865,11 +866,61 @@ DEF_OP(AtomicFetchXor) {
         mov(GetReg<RA_64>(Node), TMP2);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled Atomic size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
     }
   }
 }
 
+DEF_OP(AtomicFetchNeg) {
+  auto Op = IROp->C<IR::IROp_AtomicFetchNeg>();
+  auto MemSrc = GetReg<RA_64>(Op->Header.Args[0].ID());
+
+  // TMP2-TMP3
+  switch (Op->Size) {
+    case 1: {
+      aarch64::Label LoopTop;
+      bind(&LoopTop);
+      ldaxrb(TMP2.W(), MemOperand(MemSrc));
+      neg(TMP3.W(), TMP2.W());
+      stlxrb(TMP4.W(), TMP3.W(), MemOperand(MemSrc));
+      cbnz(TMP4.W(), &LoopTop);
+      mov(GetReg<RA_32>(Node), TMP2.W());
+      break;
+    }
+    case 2: {
+      aarch64::Label LoopTop;
+      bind(&LoopTop);
+      ldaxrh(TMP2.W(), MemOperand(MemSrc));
+      neg(TMP3.W(), TMP2.W());
+      stlxrh(TMP4.W(), TMP3.W(), MemOperand(MemSrc));
+      cbnz(TMP4.W(), &LoopTop);
+      mov(GetReg<RA_32>(Node), TMP2.W());
+      break;
+    }
+    case 4: {
+      aarch64::Label LoopTop;
+      bind(&LoopTop);
+      ldaxr(TMP2.W(), MemOperand(MemSrc));
+      neg(TMP3.W(), TMP2.W());
+      stlxr(TMP4.W(), TMP3.W(), MemOperand(MemSrc));
+      cbnz(TMP4.W(), &LoopTop);
+      mov(GetReg<RA_32>(Node), TMP2.W());
+      break;
+    }
+    case 8: {
+      aarch64::Label LoopTop;
+      bind(&LoopTop);
+      ldaxr(TMP2, MemOperand(MemSrc));
+      neg(TMP3, TMP2);
+      stlxr(TMP4, TMP3, MemOperand(MemSrc));
+      cbnz(TMP4, &LoopTop);
+      mov(GetReg<RA_64>(Node), TMP2);
+      break;
+    }
+    default:  LOGMAN_MSG_A_FMT("Unhandled Atomic size: {}", Op->Size);
+  }
+}
+
 #undef DEF_OP
 void Arm64JITCore::RegisterAtomicHandlers() {
 #define REGISTER_OP(op, x) OpHandlers[FEXCore::IR::IROps::OP_##op] = &Arm64JITCore::Op_##x
@@ -886,6 +937,7 @@ void Arm64JITCore::RegisterAtomicHandlers() {
   REGISTER_OP(ATOMICFETCHAND, AtomicFetchAnd);
   REGISTER_OP(ATOMICFETCHOR,  AtomicFetchOr);
   REGISTER_OP(ATOMICFETCHXOR, AtomicFetchXor);
+  REGISTER_OP(ATOMICFETCHNEG, AtomicFetchNeg);
 #undef REGISTER_OP
 }
 }

External/FEXCore/Source/Interface/Core/JIT/Arm64/BranchOps.cpp

@@ -4,6 +4,8 @@ tags: backend|arm64
 $end_info$
 */
 
+#include "Interface/Core/LookupCache.h"
+
 #include "Interface/Core/JIT/Arm64/JITClass.h"
 #include "Interface/Core/InternalThreadState.h"
 
@@ -16,15 +18,15 @@ using namespace vixl;
 using namespace vixl::aarch64;
 #define DEF_OP(x) void Arm64JITCore::Op_##x(FEXCore::IR::IROp_Header *IROp, uint32_t Node)
 DEF_OP(GuestCallDirect) {
-  LogMan::Msg::D("Unimplemented");
+  LogMan::Msg::DFmt("Unimplemented");
 }
 
 DEF_OP(GuestCallIndirect) {
-  LogMan::Msg::D("Unimplemented");
+  LogMan::Msg::DFmt("Unimplemented");
 }
 
 DEF_OP(GuestReturn) {
-  LogMan::Msg::D("Unimplemented");
+  LogMan::Msg::DFmt("Unimplemented");
 }
 
 DEF_OP(SignalReturn) {
@@ -142,7 +144,7 @@ Condition MapBranchCC(IR::CondClassType Cond) {
   case FEXCore::IR::COND_MI:
   case FEXCore::IR::COND_PL:
   default:
-  LOGMAN_MSG_A("Unsupported compare type");
+  LOGMAN_MSG_A_FMT("Unsupported compare type");
   return Condition::nv;
   }
 }
@@ -169,10 +171,10 @@ DEF_OP(CondJump) {
   bool isConst = IsInlineConstant(Op->Cmp2, &Const);
 
   if (isConst && Const == 0 && Op->Cond.Val == FEXCore::IR::COND_EQ) {
-    LOGMAN_THROW_A(IsGPR(Op->Cmp1.ID()), "CondJump: Expected GPR");
+    LOGMAN_THROW_A_FMT(IsGPR(Op->Cmp1.ID()), "CondJump: Expected GPR");
     cbz(GRCMP(Op->Cmp1.ID()), TrueTargetLabel);
   } else if (isConst && Const == 0 && Op->Cond.Val == FEXCore::IR::COND_NEQ) {
-    LOGMAN_THROW_A(IsGPR(Op->Cmp1.ID()), "CondJump: Expected GPR");
+    LOGMAN_THROW_A_FMT(IsGPR(Op->Cmp1.ID()), "CondJump: Expected GPR");
     cbnz(GRCMP(Op->Cmp1.ID()), TrueTargetLabel);
   } else {
     if (IsGPR(Op->Cmp1.ID())) {
@@ -183,7 +185,7 @@ DEF_OP(CondJump) {
     } else if (IsFPR(Op->Cmp1.ID())) {
       fcmp(GRFCMP(Op->Cmp1.ID()), GRFCMP(Op->Cmp2.ID()));
     } else {
-      LOGMAN_MSG_A("CondJump: Expected GPR or FPR");
+      LOGMAN_MSG_A_FMT("CondJump: Expected GPR or FPR");
     }
 
     b(TrueTargetLabel, MapBranchCC(Op->Cond));
@@ -257,7 +259,7 @@ DEF_OP(Thunk) {
 
 DEF_OP(ValidateCode) {
   auto Op = IROp->C<IR::IROp_ValidateCode>();
-  uint8_t *OldCode = (uint8_t *)&Op->CodeOriginalLow;
+  const auto *OldCode = (const uint8_t *)&Op->CodeOriginalLow;
   int len = Op->CodeLength;
   int idx = 0;
 
@@ -268,7 +270,7 @@ DEF_OP(ValidateCode) {
   while (len >= 8)
   {
     ldr(x2, MemOperand(x0, idx));
-    LoadConstant(x3, *(uint32_t *)(OldCode + idx));
+    LoadConstant(x3, *(const uint32_t *)(OldCode + idx));
     cmp(x2, x3);
     csel(GetReg<RA_64>(Node), GetReg<RA_64>(Node), x1, Condition::eq);
     len -= 8;
@@ -277,7 +279,7 @@ DEF_OP(ValidateCode) {
   while (len >= 4)
   {
     ldr(w2, MemOperand(x0, idx));
-    LoadConstant(w3, *(uint32_t *)(OldCode + idx));
+    LoadConstant(w3, *(const uint32_t *)(OldCode + idx));
     cmp(w2, w3);
     csel(GetReg<RA_64>(Node), GetReg<RA_64>(Node), x1, Condition::eq);
     len -= 4;
@@ -286,7 +288,7 @@ DEF_OP(ValidateCode) {
   while (len >= 2)
   {
     ldrh(w2, MemOperand(x0, idx));
-    LoadConstant(w3, *(uint16_t *)(OldCode + idx));
+    LoadConstant(w3, *(const uint16_t *)(OldCode + idx));
     cmp(w2, w3);
     csel(GetReg<RA_64>(Node), GetReg<RA_64>(Node), x1, Condition::eq);
     len -= 2;
@@ -295,7 +297,7 @@ DEF_OP(ValidateCode) {
   while (len >= 1)
   {
     ldrb(w2, MemOperand(x0, idx));
-    LoadConstant(w3, *(uint8_t *)(OldCode + idx));
+    LoadConstant(w3, *(const uint8_t *)(OldCode + idx));
     cmp(w2, w3);
     csel(GetReg<RA_64>(Node), GetReg<RA_64>(Node), x1, Condition::eq);
     len -= 1;

External/FEXCore/Source/Interface/Core/JIT/Arm64/ConversionOps.cpp

@@ -31,7 +31,7 @@ DEF_OP(VInsGPR) {
       ins(GetDst(Node).V2D(), Op->Index, GetReg<RA_64>(Op->Header.Args[1].ID()));
     break;
     }
-    default: LOGMAN_MSG_A("Unknown Element Size: %d", Op->Header.ElementSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Element Size: {}", Op->Header.ElementSize); break;
   }
 }
 
@@ -52,14 +52,10 @@ DEF_OP(VCastFromGPR) {
     case 8:
       fmov(GetDst(Node).D(), GetReg<RA_64>(Op->Header.Args[0].ID()).X());
       break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown castGPR element size: {}", Op->Header.ElementSize);
   }
 }
 
-DEF_OP(Float_FromGPR_U) {
-  LOGMAN_MSG_A("Unimplemented");
-}
-
 DEF_OP(Float_FromGPR_S) {
   auto Op = IROp->C<IR::IROp_Float_FromGPR_S>();
   uint16_t Conv = (Op->Header.ElementSize << 8) | Op->SrcElementSize;
@@ -95,20 +91,7 @@ DEF_OP(Float_FToF) {
       fcvt(GetDst(Node).S(), GetSrc(Op->Header.Args[0].ID()).D());
       break;
     }
-    default: LOGMAN_MSG_A("Unknown FCVT sizes: 0x%x", Conv);
-  }
-}
-
-DEF_OP(Vector_UToF) {
-  auto Op = IROp->C<IR::IROp_Vector_UToF>();
-  switch (Op->Header.ElementSize) {
-    case 4:
-      ucvtf(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
-    break;
-    case 8:
-      ucvtf(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
-    break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown FCVT sizes: 0x{:x}", Conv);
   }
 }
 
@@ -121,20 +104,7 @@ DEF_OP(Vector_SToF) {
     case 8:
       scvtf(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
     break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
-  }
-}
-
-DEF_OP(Vector_FToZU) {
-  auto Op = IROp->C<IR::IROp_Vector_FToZU>();
-  switch (Op->Header.ElementSize) {
-    case 4:
-      fcvtzu(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
-    break;
-    case 8:
-      fcvtzu(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
-    break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_SToF element size: {}", Op->Header.ElementSize);
   }
 }
 
@@ -147,22 +117,7 @@ DEF_OP(Vector_FToZS) {
     case 8:
       fcvtzs(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
     break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
-  }
-}
-
-DEF_OP(Vector_FToU) {
-  auto Op = IROp->C<IR::IROp_Vector_FToU>();
-  switch (Op->Header.ElementSize) {
-    case 4:
-      frinti(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
-      fcvtzu(GetDst(Node).V4S(), GetDst(Node).V4S());
-    break;
-    case 8:
-      frinti(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
-      fcvtzu(GetDst(Node).V2D(), GetDst(Node).V2D());
-    break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_FToZS element size: {}", Op->Header.ElementSize);
   }
 }
 
@@ -177,7 +132,7 @@ DEF_OP(Vector_FToS) {
       frinti(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
       fcvtzs(GetDst(Node).V2D(), GetDst(Node).V2D());
     break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_FToS element size: {}", Op->Header.ElementSize);
   }
 }
 
@@ -194,7 +149,63 @@ DEF_OP(Vector_FToF) {
       fcvtn(GetDst(Node).V2S(), GetSrc(Op->Header.Args[0].ID()).V2D());
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Conversion Type : 0%04x", Conv); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_FToF Type : 0x{:04x}", Conv); break;
+  }
+}
+
+DEF_OP(Vector_FToI) {
+  auto Op = IROp->C<IR::IROp_Vector_FToI>();
+  switch (Op->Round) {
+    case FEXCore::IR::Round_Nearest.Val:
+      switch (Op->Header.ElementSize) {
+        case 4:
+          frintn(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
+        break;
+        case 8:
+          frintn(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
+        break;
+      }
+    break;
+    case FEXCore::IR::Round_Negative_Infinity.Val:
+      switch (Op->Header.ElementSize) {
+        case 4:
+          frintm(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
+        break;
+        case 8:
+          frintm(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
+        break;
+      }
+    break;
+    case FEXCore::IR::Round_Positive_Infinity.Val:
+      switch (Op->Header.ElementSize) {
+        case 4:
+          frintp(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
+        break;
+        case 8:
+          frintp(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
+        break;
+      }
+    break;
+    case FEXCore::IR::Round_Towards_Zero.Val:
+      switch (Op->Header.ElementSize) {
+        case 4:
+          frintz(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
+        break;
+        case 8:
+          frintz(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
+        break;
+      }
+    break;
+    case FEXCore::IR::Round_Host.Val:
+      switch (Op->Header.ElementSize) {
+        case 4:
+          frinti(GetDst(Node).V4S(), GetSrc(Op->Header.Args[0].ID()).V4S());
+        break;
+        case 8:
+          frinti(GetDst(Node).V2D(), GetSrc(Op->Header.Args[0].ID()).V2D());
+        break;
+      }
+    break;
   }
 }
 
@@ -203,16 +214,13 @@ void Arm64JITCore::RegisterConversionHandlers() {
 #define REGISTER_OP(op, x) OpHandlers[FEXCore::IR::IROps::OP_##op] = &Arm64JITCore::Op_##x
   REGISTER_OP(VINSGPR,         VInsGPR);
   REGISTER_OP(VCASTFROMGPR,    VCastFromGPR);
-  REGISTER_OP(FLOAT_FROMGPR_U, Float_FromGPR_U);
   REGISTER_OP(FLOAT_FROMGPR_S, Float_FromGPR_S);
   REGISTER_OP(FLOAT_FTOF,      Float_FToF);
-  REGISTER_OP(VECTOR_UTOF,     Vector_UToF);
   REGISTER_OP(VECTOR_STOF,     Vector_SToF);
-  REGISTER_OP(VECTOR_FTOZU,    Vector_FToZU);
   REGISTER_OP(VECTOR_FTOZS,    Vector_FToZS);
-  REGISTER_OP(VECTOR_FTOU,     Vector_FToU);
   REGISTER_OP(VECTOR_FTOS,     Vector_FToS);
   REGISTER_OP(VECTOR_FTOF,     Vector_FToF);
+  REGISTER_OP(VECTOR_FTOI,     Vector_FToI);
 #undef REGISTER_OP
 }
 }

External/FEXCore/Source/Interface/Core/JIT/Arm64/JIT.cpp

@@ -11,6 +11,7 @@ $end_info$
 */
 
 #include "Interface/Context/Context.h"
+#include "Interface/Core/LookupCache.h"
 
 #include "Interface/Core/ArchHelpers/Arm64.h"
 #include "Interface/Core/ArchHelpers/MContext.h"
@@ -23,6 +24,7 @@ $end_info$
 #include <FEXCore/Core/X86Enums.h>
 #include <FEXCore/Core/UContext.h>
 #include <FEXCore/Utils/Allocator.h>
+#include <FEXCore/Utils/CompilerDefs.h>
 #include "Interface/Core/Interpreter/InterpreterOps.h"
 
 #include <sys/mman.h>
@@ -43,8 +45,9 @@ using namespace vixl::aarch64;
 void Arm64JITCore::Op_Unhandled(FEXCore::IR::IROp_Header *IROp, uint32_t Node) {
   FallbackInfo Info;
   if (!InterpreterOps::GetFallbackHandler(IROp, &Info)) {
-    auto Name = FEXCore::IR::GetName(IROp->Op);
-    LOGMAN_MSG_A("Unhandled IR Op: %s", std::string(Name).c_str());
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
+    LOGMAN_MSG_A_FMT("Unhandled IR Op: {}", FEXCore::IR::GetName(IROp->Op));
+#endif
   } else {
     switch(Info.ABI) {
       case FABI_VOID_U16:{
@@ -291,8 +294,10 @@ void Arm64JITCore::Op_Unhandled(FEXCore::IR::IROp_Header *IROp, uint32_t Node) {
 
       case FABI_UNKNOWN:
       default:
-      auto Name = FEXCore::IR::GetName(IROp->Op);
-        LOGMAN_MSG_A("Unhandled IR Fallback abi: %s %d", std::string(Name).c_str(), Info.ABI);
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
+        LOGMAN_MSG_A_FMT("Unhandled IR Fallback ABI: {} {}", FEXCore::IR::GetName(IROp->Op), Info.ABI);
+#endif
+      break;
     }
   }
 }
@@ -309,7 +314,7 @@ Arm64JITCore::CodeBuffer Arm64JITCore::AllocateNewCodeBuffer(size_t Size) {
                     PROT_READ | PROT_WRITE | PROT_EXEC,
                     MAP_PRIVATE | MAP_ANONYMOUS,
                     -1, 0));
-  LOGMAN_THROW_A(!!Buffer.Ptr, "Couldn't allocate code buffer");
+  LOGMAN_THROW_A_FMT(!!Buffer.Ptr, "Couldn't allocate code buffer");
   Dispatcher->RegisterCodeBuffer(Buffer.Ptr, Buffer.Size);
   return Buffer;
 }
@@ -339,54 +344,70 @@ bool Arm64JITCore::HandleSIGBUS(int Signal, void *info, void *ucontext) {
     0b1011'0000'0000; // Inner shareable all
   if ((Instr & 0x3F'FF'FC'00) == 0x08'DF'FC'00 || // LDAR*
       (Instr & 0x3F'FF'FC'00) == 0x38'BF'C0'00) { // LDAPR*
-    uint32_t LDR = 0b0011'1000'0111'1111'0110'1000'0000'0000;
-    LDR |= Size << 30;
-    LDR |= AddrReg << 5;
-    LDR |= DataReg;
-    PC[-1] = DMB;
-    PC[0] = LDR;
-    PC[1] = DMB;
-    // Back up one instruction and have another go
-    ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) - 4);
+    if (ParanoidTSO()) {
+      if (FEXCore::ArchHelpers::Arm64::HandleAtomicLoad(ucontext, info, Instr)) {
+        // Skip this instruction now
+        ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) + 4);
+        return true;
+      }
+      else {
+        LogMan::Msg::EFmt("Unhandled JIT SIGBUS LDAR*: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
+        return false;
+      }
+    }
+    else {
+      uint32_t LDR = 0b0011'1000'0111'1111'0110'1000'0000'0000;
+      LDR |= Size << 30;
+      LDR |= AddrReg << 5;
+      LDR |= DataReg;
+      PC[-1] = DMB;
+      PC[0] = LDR;
+      PC[1] = DMB;
+      // Back up one instruction and have another go
+      ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) - 4);
+    }
   }
   else if ( (Instr & 0x3F'FF'FC'00) == 0x08'9F'FC'00) { // STLR*
-    uint32_t STR = 0b0011'1000'0011'1111'0110'1000'0000'0000;
-    STR |= Size << 30;
-    STR |= AddrReg << 5;
-    STR |= DataReg;
-    PC[-1] = DMB;
-    PC[0] = STR;
-    PC[1] = DMB;
-    // Back up one instruction and have another go
-    ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) - 4);
+    if (ParanoidTSO()) {
+      if (FEXCore::ArchHelpers::Arm64::HandleAtomicStore(ucontext, info, Instr)) {
+        // Skip this instruction now
+        ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) + 4);
+        return true;
+      }
+      else {
+        LogMan::Msg::EFmt("Unhandled JIT SIGBUS STLR*: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
+        return false;
+      }
+    }
+    else {
+      uint32_t STR = 0b0011'1000'0011'1111'0110'1000'0000'0000;
+      STR |= Size << 30;
+      STR |= AddrReg << 5;
+      STR |= DataReg;
+      PC[-1] = DMB;
+      PC[0] = STR;
+      PC[1] = DMB;
+      // Back up one instruction and have another go
+      ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) - 4);
+    }
   }
   else if ((Instr & FEXCore::ArchHelpers::Arm64::LDAXP_MASK) == FEXCore::ArchHelpers::Arm64::LDAXP_INST) { // LDAXP
-    uint32_t DataReg2 = (Instr >> 10) & 0x1F;
-    // Convert to LDP
-    uint32_t LDP = 0b0010'1001'0100'0000'0000'0000'0000'0000;
-    LDP |= Size << 31;
-    LDP |= DataReg2 << 10;
-    LDP |= AddrReg << 5;
-    LDP |= DataReg;
-    PC[-1] = DMB;
-    PC[0] = LDP;
-    PC[1] = DMB;
-    // Back up one instruction and have another go
-    ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) - 4);
+    //Should be compare and swap pair only. LDAXP not used elsewhere
+    uint64_t BytesToSkip = FEXCore::ArchHelpers::Arm64::HandleCASPAL_ARMv8(ucontext, info, Instr);
+    if (BytesToSkip) {
+      // Skip this instruction now
+      ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) + BytesToSkip);
+      return true;
+    }
+    else {
+      LogMan::Msg::EFmt("Unhandled JIT SIGBUS LDAXP: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
+      return false;
+    }
   }
   else if ((Instr & FEXCore::ArchHelpers::Arm64::STLXP_MASK) == FEXCore::ArchHelpers::Arm64::STLXP_INST) { // STLXP
-    uint32_t DataReg2 = (Instr >> 10) & 0x1F;
-    // Convert to STP
-    uint32_t STP = 0b0010'1001'0000'0000'0000'0000'0000'0000;
-    STP |= Size << 31;
-    STP |= DataReg2 << 10;
-    STP |= AddrReg << 5;
-    STP |= DataReg;
-    PC[-1] = DMB;
-    PC[0] = STP;
-    PC[1] = DMB;
-    // Back up one instruction and have another go
-    ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) - 4);
+    //Should not trigger - middle of an LDAXP/STAXP pair.
+    LogMan::Msg::EFmt("Unhandled JIT SIGBUS STLXP: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
+    return false;
   }
   else if ((Instr & FEXCore::ArchHelpers::Arm64::CASPAL_MASK) == FEXCore::ArchHelpers::Arm64::CASPAL_INST) { // CASPAL
     if (FEXCore::ArchHelpers::Arm64::HandleCASPAL(ucontext, info, Instr)) {
@@ -395,7 +416,7 @@ bool Arm64JITCore::HandleSIGBUS(int Signal, void *info, void *ucontext) {
       return true;
     }
     else {
-      LogMan::Msg::E("Unhandled JIT SIGBUS CASPAL: PC: %p Instruction: 0x%08x\n", PC, PC[0]);
+      LogMan::Msg::EFmt("Unhandled JIT SIGBUS CASPAL: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
       return false;
     }
   }
@@ -406,7 +427,7 @@ bool Arm64JITCore::HandleSIGBUS(int Signal, void *info, void *ucontext) {
       return true;
     }
     else {
-      LogMan::Msg::E("Unhandled JIT SIGBUS CASAL: PC: %p Instruction: 0x%08x\n", PC, PC[0]);
+      LogMan::Msg::EFmt("Unhandled JIT SIGBUS CASAL: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
       return false;
     }
   }
@@ -418,12 +439,24 @@ bool Arm64JITCore::HandleSIGBUS(int Signal, void *info, void *ucontext) {
     }
     else {
       uint8_t Op = (PC[0] >> 12) & 0xF;
-      LogMan::Msg::E("Unhandled JIT SIGBUS Atomic mem op 0x%02x: PC: %p Instruction: 0x%08x\n", Op, PC, PC[0]);
+      LogMan::Msg::EFmt("Unhandled JIT SIGBUS Atomic mem op 0x{:02x}: PC: {} Instruction: 0x{:08x}\n", Op, fmt::ptr(PC), PC[0]);
+      return false;
+    }
+  }
+  else if ((Instr & FEXCore::ArchHelpers::Arm64::LDAXR_MASK) == FEXCore::ArchHelpers::Arm64::LDAXR_INST) { // LDAXR*
+    uint64_t BytesToSkip = FEXCore::ArchHelpers::Arm64::HandleAtomicLoadstoreExclusive(ucontext, info);
+    if (BytesToSkip) {
+      // Skip this instruction now
+      ArchHelpers::Context::SetPc(ucontext, ArchHelpers::Context::GetPc(ucontext) + BytesToSkip);
+      return true;
+    }
+    else {
+      LogMan::Msg::EFmt("Unhandled JIT SIGBUS LDAXR: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
       return false;
     }
   }
   else {
-    LogMan::Msg::E("Unhandled JIT SIGBUS: PC: %p Instruction: 0x%08x\n", PC, PC[0]);
+    LogMan::Msg::EFmt("Unhandled JIT SIGBUS: PC: {} Instruction: 0x{:08x}\n", fmt::ptr(PC), PC[0]);
     return false;
   }
 
@@ -439,7 +472,7 @@ Arm64JITCore::Arm64JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::Intern
     DispatcherConfig config;
     config.ExitFunctionLink = reinterpret_cast<uintptr_t>(&ExitFunctionLink);
     config.ExitFunctionLinkThis = reinterpret_cast<uintptr_t>(this);
-    config.StaticRegisterAssignment = true;
+    config.StaticRegisterAssignment = ctx->Config.StaticRegisterAllocation;
 
     Dispatcher = std::make_unique<Arm64Dispatcher>(CTX, ThreadState, config);
     DispatchPtr = Dispatcher->DispatchPtr;
@@ -501,24 +534,24 @@ Arm64JITCore::Arm64JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::Intern
     CTX->SignalDelegation->RegisterHostSignalHandler(SIGILL, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       Arm64JITCore *Core = reinterpret_cast<Arm64JITCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleSIGILL(Signal, info, ucontext);
-    });
+    }, true);
 
     CTX->SignalDelegation->RegisterHostSignalHandler(SIGBUS, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       Arm64JITCore *Core = reinterpret_cast<Arm64JITCore*>(Thread->CPUBackend.get());
       return Core->HandleSIGBUS(Signal, info, ucontext);
-    });
+    }, true);
 
     CTX->SignalDelegation->RegisterHostSignalHandler(SignalDelegator::SIGNAL_FOR_PAUSE, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       Arm64JITCore *Core = reinterpret_cast<Arm64JITCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleSignalPause(Signal, info, ucontext);
-    });
+    }, true);
 
     auto GuestSignalHandler = [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext, GuestSigAction *GuestAction, stack_t *GuestStack) -> bool {
       Arm64JITCore *Core = reinterpret_cast<Arm64JITCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleGuestSignal(Signal, info, ucontext, GuestAction, GuestStack);
     };
 
-    for (uint32_t Signal = 0; Signal < SignalDelegator::MAX_SIGNALS; ++Signal) {
+    for (uint32_t Signal = 0; Signal <= SignalDelegator::MAX_SIGNALS; ++Signal) {
       CTX->SignalDelegation->RegisterHostSignalHandlerForGuest(Signal, GuestSignalHandler);
     }
   }
@@ -575,16 +608,16 @@ Arm64JITCore::~Arm64JITCore() {
   FreeCodeBuffer(InitialCodeBuffer);
 }
 
-IR::PhysicalRegister Arm64JITCore::GetPhys(uint32_t Node) {
+IR::PhysicalRegister Arm64JITCore::GetPhys(uint32_t Node) const {
   auto PhyReg = RAData->GetNodeRegister(Node);
 
-  LOGMAN_THROW_A(!PhyReg.IsInvalid(), "Couldn't Allocate register for node: ssa%d. Class: %d", Node, PhyReg.Class);
+  LOGMAN_THROW_A_FMT(!PhyReg.IsInvalid(), "Couldn't Allocate register for node: ssa{}. Class: {}", Node, PhyReg.Class);
 
   return PhyReg;
 }
 
 template<>
-aarch64::Register Arm64JITCore::GetReg<Arm64JITCore::RA_32>(uint32_t Node) {
+aarch64::Register Arm64JITCore::GetReg<Arm64JITCore::RA_32>(uint32_t Node) const {
   auto Reg = GetPhys(Node);
 
   if (Reg.Class == IR::GPRFixedClass.Val) {
@@ -592,13 +625,14 @@ aarch64::Register Arm64JITCore::GetReg<Arm64JITCore::RA_32>(uint32_t Node) {
   } else if (Reg.Class == IR::GPRClass.Val) {
     return RA64[Reg.Reg].W();
   } else {
-    LOGMAN_THROW_A(false, "Unexpected Class: %d", Reg.Class);
+    LOGMAN_THROW_A_FMT(false, "Unexpected Class: {}", Reg.Class);
   }
-  __builtin_unreachable();
+
+  FEX_UNREACHABLE;
 }
 
 template<>
-aarch64::Register Arm64JITCore::GetReg<Arm64JITCore::RA_64>(uint32_t Node) {
+aarch64::Register Arm64JITCore::GetReg<Arm64JITCore::RA_64>(uint32_t Node) const {
   auto Reg = GetPhys(Node);
 
   if (Reg.Class == IR::GPRFixedClass.Val) {
@@ -606,24 +640,25 @@ aarch64::Register Arm64JITCore::GetReg<Arm64JITCore::RA_64>(uint32_t Node) {
   } else if (Reg.Class == IR::GPRClass.Val) {
     return RA64[Reg.Reg];
   } else {
-    LOGMAN_THROW_A(false, "Unexpected Class: %d", Reg.Class);
+    LOGMAN_THROW_A_FMT(false, "Unexpected Class: {}", Reg.Class);
   }
-  __builtin_unreachable();
+
+  FEX_UNREACHABLE;
 }
 
 template<>
-std::pair<aarch64::Register, aarch64::Register> Arm64JITCore::GetSrcPair<Arm64JITCore::RA_32>(uint32_t Node) {
+std::pair<aarch64::Register, aarch64::Register> Arm64JITCore::GetSrcPair<Arm64JITCore::RA_32>(uint32_t Node) const {
   uint32_t Reg = GetPhys(Node).Reg;
   return RA32Pair[Reg];
 }
 
 template<>
-std::pair<aarch64::Register, aarch64::Register> Arm64JITCore::GetSrcPair<Arm64JITCore::RA_64>(uint32_t Node) {
+std::pair<aarch64::Register, aarch64::Register> Arm64JITCore::GetSrcPair<Arm64JITCore::RA_64>(uint32_t Node) const {
   uint32_t Reg = GetPhys(Node).Reg;
   return RA64Pair[Reg];
 }
 
-aarch64::VRegister Arm64JITCore::GetSrc(uint32_t Node) {
+aarch64::VRegister Arm64JITCore::GetSrc(uint32_t Node) const {
   auto Reg = GetPhys(Node);
 
   if (Reg.Class == IR::FPRFixedClass.Val) {
@@ -631,12 +666,13 @@ aarch64::VRegister Arm64JITCore::GetSrc(uint32_t Node) {
   } else if (Reg.Class == IR::FPRClass.Val) {
     return RAFPR[Reg.Reg];
   } else {
-    LOGMAN_THROW_A(false, "Unexpected Class: %d", Reg.Class);
+    LOGMAN_THROW_A_FMT(false, "Unexpected Class: {}", Reg.Class);
   }
-  __builtin_unreachable();
+
+  FEX_UNREACHABLE;
 }
 
-aarch64::VRegister Arm64JITCore::GetDst(uint32_t Node) {
+aarch64::VRegister Arm64JITCore::GetDst(uint32_t Node) const {
   auto Reg = GetPhys(Node);
 
   if (Reg.Class == IR::FPRFixedClass.Val) {
@@ -644,12 +680,13 @@ aarch64::VRegister Arm64JITCore::GetDst(uint32_t Node) {
   } else if (Reg.Class == IR::FPRClass.Val) {
     return RAFPR[Reg.Reg];
   } else {
-    LOGMAN_THROW_A(false, "Unexpected Class: %d", Reg.Class);
+    LOGMAN_THROW_A_FMT(false, "Unexpected Class: {}", Reg.Class);
   }
-  __builtin_unreachable();
+
+  FEX_UNREACHABLE;
 }
 
-bool Arm64JITCore::IsInlineConstant(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) {
+bool Arm64JITCore::IsInlineConstant(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) const {
   auto OpHeader = IR->GetOp<IR::IROp_Header>(WNode);
 
   if (OpHeader->Op == IR::IROps::OP_INLINECONSTANT) {
@@ -663,7 +700,7 @@ bool Arm64JITCore::IsInlineConstant(const IR::OrderedNodeWrapper& WNode, uint64_
   }
 }
 
-bool Arm64JITCore::IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) {
+bool Arm64JITCore::IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) const {
   auto OpHeader = IR->GetOp<IR::IROp_Header>(WNode);
 
   if (OpHeader->Op == IR::IROps::OP_INLINEENTRYPOINTOFFSET) {
@@ -677,18 +714,18 @@ bool Arm64JITCore::IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode,
   }
 }
 
-FEXCore::IR::RegisterClassType Arm64JITCore::GetRegClass(uint32_t Node) {
+FEXCore::IR::RegisterClassType Arm64JITCore::GetRegClass(uint32_t Node) const {
   return FEXCore::IR::RegisterClassType {GetPhys(Node).Class};
 }
 
 
-bool Arm64JITCore::IsFPR(uint32_t Node) {
+bool Arm64JITCore::IsFPR(uint32_t Node) const {
   auto Class = GetRegClass(Node);
 
   return Class == IR::FPRClass || Class == IR::FPRFixedClass;
 }
 
-bool Arm64JITCore::IsGPR(uint32_t Node) {
+bool Arm64JITCore::IsGPR(uint32_t Node) const {
   auto Class = GetRegClass(Node);
 
   return Class == IR::GPRClass || Class == IR::GPRFixedClass;
@@ -702,8 +739,6 @@ void *Arm64JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IR
   this->Entry = Entry;
   this->RAData = RAData;
 
-  auto HeaderOp = IR->GetHeader();
-
   #ifndef NDEBUG
   LoadConstant(x0, Entry);
   #endif
@@ -736,8 +771,7 @@ void *Arm64JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IR
   // X1-X3 = Temp
   // X4-r18 = RA
 
-  auto Buffer = GetBuffer();
-  auto GuestEntry = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+  auto GuestEntry = GetCursorAddress<uint64_t>();
 
  if (CTX->GetGdbServerStatus()) {
     aarch64::Label RunBlock;
@@ -781,8 +815,10 @@ void *Arm64JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IR
 
   for (auto [BlockNode, BlockHeader] : IR->GetBlocks()) {
     using namespace FEXCore::IR;
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
     auto BlockIROp = BlockHeader->CW<FEXCore::IR::IROp_CodeBlock>();
-    LOGMAN_THROW_A(BlockIROp->Header.Op == IR::OP_CODEBLOCK, "IR type failed to be a code block");
+    LOGMAN_THROW_A_FMT(BlockIROp->Header.Op == IR::OP_CODEBLOCK, "IR type failed to be a code block");
+#endif
 
     {
       uint32_t Node = IR->GetID(BlockNode);
@@ -802,7 +838,7 @@ void *Arm64JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IR
     }
 
     if (DebugData) {
-      DebugData->Subblocks.push_back({Buffer->GetOffsetAddress<uintptr_t>(GetCursorOffset()), 0, IR->GetID(BlockNode)});
+      DebugData->Subblocks.push_back({GetCursorAddress<uintptr_t>(), 0, IR->GetID(BlockNode)});
     }
 
     for (auto [CodeNode, IROp] : IR->GetCode(BlockNode)) {
@@ -814,7 +850,7 @@ void *Arm64JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IR
     }
 
     if (DebugData) {
-      DebugData->Subblocks.back().HostCodeSize = Buffer->GetOffsetAddress<uintptr_t>(GetCursorOffset()) - DebugData->Subblocks.back().HostCodeStart;
+      DebugData->Subblocks.back().HostCodeSize = GetCursorAddress<uintptr_t>() - DebugData->Subblocks.back().HostCodeStart;
     }
   }
 
@@ -827,7 +863,7 @@ void *Arm64JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IR
 
   FinalizeCode();
 
-  auto CodeEnd = Buffer->GetOffsetAddress<uint64_t>(GetCursorOffset());
+  auto CodeEnd = GetCursorAddress<uint64_t>();
   CPU.EnsureIAndDCacheCoherency(reinterpret_cast<void*>(GuestEntry), CodeEnd - reinterpret_cast<uint64_t>(GuestEntry));
 
   if (DebugData) {
@@ -846,7 +882,7 @@ uint64_t Arm64JITCore::ExitFunctionLink(Arm64JITCore *core, FEXCore::Core::CpuSt
   auto HostCode = Thread->LookupCache->FindBlock(GuestRip);
 
   if (!HostCode) {
-    //printf("ExitFunctionLink: Aborting, %lX not in cache\n", GuestRip);
+    //fmt::print("ExitFunctionLink: Aborting, {:X} not in cache\n", GuestRip);
     Frame->State.rip = GuestRip;
     return core->ThreadSharedData.Dispatcher->AbsoluteLoopTopAddress;
   }
@@ -888,7 +924,7 @@ uint64_t Arm64JITCore::ExitFunctionLink(Arm64JITCore *core, FEXCore::Core::CpuSt
   return HostCode;
 }
 
-FEXCore::CPU::CPUBackend *CreateArm64JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread) {
-  return new Arm64JITCore(ctx, Thread, CompileThread);
+std::unique_ptr<CPUBackend> CreateArm64JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread) {
+  return std::make_unique<Arm64JITCore>(ctx, Thread, CompileThread);
 }
 }

External/FEXCore/Source/Interface/Core/JIT/Arm64/JITClass.h

@@ -6,7 +6,6 @@ $end_info$
 
 #pragma once
 
-#include "Interface/Core/LookupCache.h"
 #include "Interface/Core/ArchHelpers/Arm64Emitter.h"
 #include "Interface/Core/Dispatcher/Dispatcher.h"
 
@@ -96,35 +95,35 @@ private:
   constexpr static uint8_t RA_FPR = 2;
 
   template<uint8_t RAType>
-  aarch64::Register GetReg(uint32_t Node);
+  aarch64::Register GetReg(uint32_t Node) const;
 
   template<>
-  aarch64::Register GetReg<RA_32>(uint32_t Node);
+  aarch64::Register GetReg<RA_32>(uint32_t Node) const;
   template<>
-  aarch64::Register GetReg<RA_64>(uint32_t Node);
+  aarch64::Register GetReg<RA_64>(uint32_t Node) const;
 
   template<uint8_t RAType>
-  std::pair<aarch64::Register, aarch64::Register> GetSrcPair(uint32_t Node);
+  std::pair<aarch64::Register, aarch64::Register> GetSrcPair(uint32_t Node) const;
 
   template<>
-  std::pair<aarch64::Register, aarch64::Register> GetSrcPair<RA_32>(uint32_t Node);
+  std::pair<aarch64::Register, aarch64::Register> GetSrcPair<RA_32>(uint32_t Node) const;
   template<>
-  std::pair<aarch64::Register, aarch64::Register> GetSrcPair<RA_64>(uint32_t Node);
+  std::pair<aarch64::Register, aarch64::Register> GetSrcPair<RA_64>(uint32_t Node) const;
 
-  aarch64::VRegister GetSrc(uint32_t Node);
-  aarch64::VRegister GetDst(uint32_t Node);
+  aarch64::VRegister GetSrc(uint32_t Node) const;
+  aarch64::VRegister GetDst(uint32_t Node) const;
 
-  FEXCore::IR::RegisterClassType GetRegClass(uint32_t Node);
+  FEXCore::IR::RegisterClassType GetRegClass(uint32_t Node) const;
 
-  IR::PhysicalRegister GetPhys(uint32_t Node);
+  IR::PhysicalRegister GetPhys(uint32_t Node) const;
 
-  bool IsFPR(uint32_t Node);
-  bool IsGPR(uint32_t Node);
+  bool IsFPR(uint32_t Node) const;
+  bool IsGPR(uint32_t Node) const;
 
   MemOperand GenerateMemOperand(uint8_t AccessSize, aarch64::Register Base, IR::OrderedNodeWrapper Offset, IR::MemOffsetType OffsetType, uint8_t OffsetScale);
 
-  bool IsInlineConstant(const IR::OrderedNodeWrapper& Node, uint64_t* Value = nullptr);
-  bool IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value);
+  bool IsInlineConstant(const IR::OrderedNodeWrapper& Node, uint64_t* Value = nullptr) const;
+  bool IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) const;
 
   struct LiveRange {
     uint32_t Begin;
@@ -240,7 +239,6 @@ private:
   DEF_OP(VExtractToGPR);
   DEF_OP(Float_ToGPR_ZU);
   DEF_OP(Float_ToGPR_ZS);
-  DEF_OP(Float_ToGPR_U);
   DEF_OP(Float_ToGPR_S);
   DEF_OP(FCmp);
 
@@ -258,6 +256,7 @@ private:
   DEF_OP(AtomicFetchAnd);
   DEF_OP(AtomicFetchOr);
   DEF_OP(AtomicFetchXor);
+  DEF_OP(AtomicFetchNeg);
 
   ///< Branch ops
   DEF_OP(GuestCallDirect);
@@ -277,16 +276,13 @@ private:
   ///< Conversion ops
   DEF_OP(VInsGPR);
   DEF_OP(VCastFromGPR);
-  DEF_OP(Float_FromGPR_U);
   DEF_OP(Float_FromGPR_S);
   DEF_OP(Float_FToF);
-  DEF_OP(Vector_UToF);
   DEF_OP(Vector_SToF);
-  DEF_OP(Vector_FToZU);
   DEF_OP(Vector_FToZS);
-  DEF_OP(Vector_FToU);
   DEF_OP(Vector_FToS);
   DEF_OP(Vector_FToF);
+  DEF_OP(Vector_FToI);
 
   ///< Flag ops
   DEF_OP(GetHostFlag);
@@ -336,6 +332,7 @@ private:
   DEF_OP(SplatVector4);
   DEF_OP(VMov);
   DEF_OP(VAnd);
+  DEF_OP(VBic);
   DEF_OP(VOr);
   DEF_OP(VXor);
   DEF_OP(VAdd);
@@ -346,8 +343,10 @@ private:
   DEF_OP(VSQSub);
   DEF_OP(VAddP);
   DEF_OP(VAddV);
+  DEF_OP(VUMinV);
   DEF_OP(VURAvg);
   DEF_OP(VAbs);
+  DEF_OP(VPopcount);
   DEF_OP(VFAdd);
   DEF_OP(VFAddP);
   DEF_OP(VFSub);
@@ -367,6 +366,8 @@ private:
   DEF_OP(VSMax);
   DEF_OP(VZip);
   DEF_OP(VZip2);
+  DEF_OP(VUnZip);
+  DEF_OP(VUnZip2);
   DEF_OP(VBSL);
   DEF_OP(VCMPEQ);
   DEF_OP(VCMPEQZ);
@@ -389,6 +390,7 @@ private:
   DEF_OP(VInsElement);
   DEF_OP(VInsScalarElement);
   DEF_OP(VExtractElement);
+  DEF_OP(VDupElement);
   DEF_OP(VExtr);
   DEF_OP(VSLI);
   DEF_OP(VSRI);
@@ -411,6 +413,7 @@ private:
   DEF_OP(VSMull);
   DEF_OP(VUMull2);
   DEF_OP(VSMull2);
+  DEF_OP(VUABDL);
   DEF_OP(VTBL1);
 
   ///< Encryption ops

External/FEXCore/Source/Interface/Core/JIT/Arm64/MemoryOps.cpp

@@ -5,6 +5,7 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/Arm64/JITClass.h"
+#include <FEXCore/Utils/CompilerDefs.h>
 
 namespace FEXCore::CPU {
 
@@ -29,7 +30,7 @@ DEF_OP(LoadContext) {
     case 8:
       ldr(GetReg<RA_64>(Node), MemOperand(STATE, Op->Offset));
     break;
-    default:  LOGMAN_MSG_A("Unhandled LoadContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled LoadContext size: {}", OpSize);
     }
   }
   else {
@@ -50,7 +51,7 @@ DEF_OP(LoadContext) {
     case 16:
       ldr(Dst, MemOperand(STATE, Op->Offset));
     break;
-    default:  LOGMAN_MSG_A("Unhandled LoadContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled LoadContext size: {}", OpSize);
     }
   }
 }
@@ -72,7 +73,7 @@ DEF_OP(StoreContext) {
     case 8:
       str(GetReg<RA_64>(Op->Header.Args[0].ID()), MemOperand(STATE, Op->Offset));
     break;
-    default:  LOGMAN_MSG_A("Unhandled StoreContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled StoreContext size: {}", OpSize);
     }
   }
   else {
@@ -93,7 +94,7 @@ DEF_OP(StoreContext) {
     case 16:
       str(Src, MemOperand(STATE, Op->Offset));
     break;
-    default:  LOGMAN_MSG_A("Unhandled LoadContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled StoreContext size: {}", OpSize);
     }
   }
 }
@@ -106,29 +107,29 @@ DEF_OP(LoadRegister) {
     auto regId = (Op->Offset - offsetof(FEXCore::Core::CpuStateFrame, State.gregs[0])) / 8;
     auto regOffs = Op->Offset & 7;
 
-    LOGMAN_THROW_A(regId < SRA64.size(), "out of range regId");
+    LOGMAN_THROW_A_FMT(regId < SRA64.size(), "out of range regId");
 
     auto reg = SRA64[regId];
 
     switch(Op->Header.Size) {
       case 1:
-        LOGMAN_THROW_A(regOffs == 0 || regOffs == 1, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0 || regOffs == 1, "unexpected regOffs");
         ubfx(GetReg<RA_64>(Node), reg, regOffs * 8, 8);
         break;
 
       case 2:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         ubfx(GetReg<RA_64>(Node), reg, 0, 16);
         break;
 
       case 4:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         if (GetReg<RA_64>(Node).GetCode() != reg.GetCode())
           mov(GetReg<RA_32>(Node), reg.W());
         break;
 
       case 8:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         if (GetReg<RA_64>(Node).GetCode() != reg.GetCode())
           mov(GetReg<RA_64>(Node), reg);
         break;
@@ -137,24 +138,24 @@ DEF_OP(LoadRegister) {
     auto regId = (Op->Offset - offsetof(FEXCore::Core::CpuStateFrame, State.xmm[0][0])) / 16;
     auto regOffs = Op->Offset & 15;
 
-    LOGMAN_THROW_A(regId < SRAFPR.size(), "out of range regId");
+    LOGMAN_THROW_A_FMT(regId < SRAFPR.size(), "out of range regId");
 
     auto guest = SRAFPR[regId];
     auto host = GetSrc(Node);
 
     switch(Op->Header.Size) {
       case 1:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         mov(host.B(), guest.B());
         break;
 
       case 2:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         fmov(host.H(), guest.H());
         break;
 
       case 4:
-        LOGMAN_THROW_A((regOffs & 3) == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT((regOffs & 3) == 0, "unexpected regOffs");
         if (regOffs == 0) {
           if (host.GetCode() != guest.GetCode())
             fmov(host.S(), guest.S());
@@ -164,7 +165,7 @@ DEF_OP(LoadRegister) {
         break;
 
       case 8:
-        LOGMAN_THROW_A((regOffs & 7) == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT((regOffs & 7) == 0, "unexpected regOffs");
         if (regOffs == 0) {
           if (host.GetCode() != guest.GetCode())
             mov(host.D(), guest.D());
@@ -174,13 +175,13 @@ DEF_OP(LoadRegister) {
         break;
 
       case 16:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         if (host.GetCode() != guest.GetCode())
           mov(host.Q(), guest.Q());
         break;
     }
   } else {
-    LOGMAN_THROW_A(false, "Unhandled Op->Class %d", Op->Class);
+    LOGMAN_THROW_A_FMT(false, "Unhandled Op->Class {}", Op->Class);
   }
 }
 
@@ -191,28 +192,28 @@ DEF_OP(StoreRegister) {
     auto regId = Op->Offset / 8 - 1;
     auto regOffs = Op->Offset & 7;
 
-    LOGMAN_THROW_A(regId < SRA64.size(), "out of range regId");
+    LOGMAN_THROW_A_FMT(regId < SRA64.size(), "out of range regId");
 
     auto reg = SRA64[regId];
 
     switch(Op->Header.Size) {
       case 1:
-        LOGMAN_THROW_A(regOffs == 0 || regOffs == 1, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0 || regOffs == 1, "unexpected regOffs");
         bfi(reg, GetReg<RA_64>(Op->Value.ID()), regOffs * 8, 8);
         break;
 
       case 2:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         bfi(reg, GetReg<RA_64>(Op->Value.ID()), 0, 16);
         break;
 
       case 4:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         bfi(reg, GetReg<RA_64>(Op->Value.ID()), 0, 32);
         break;
 
       case 8:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         if (GetReg<RA_64>(Op->Value.ID()).GetCode() != reg.GetCode())
           mov(reg, GetReg<RA_64>(Op->Value.ID()));
         break;
@@ -221,7 +222,7 @@ DEF_OP(StoreRegister) {
     auto regId = (Op->Offset - offsetof(FEXCore::Core::CpuStateFrame, State.xmm[0][0])) / 16;
     auto regOffs = Op->Offset & 15;
 
-    LOGMAN_THROW_A(regId < SRAFPR.size(), "regId out of range");
+    LOGMAN_THROW_A_FMT(regId < SRAFPR.size(), "regId out of range");
 
     auto guest = SRAFPR[regId];
     auto host = GetSrc(Op->Value.ID());
@@ -232,28 +233,28 @@ DEF_OP(StoreRegister) {
         break;
 
       case 2:
-        LOGMAN_THROW_A((regOffs & 1) == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT((regOffs & 1) == 0, "unexpected regOffs");
         ins(guest.V8H(), regOffs/2, host.V8H(), 0);
         break;
 
       case 4:
-        LOGMAN_THROW_A((regOffs & 3) == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT((regOffs & 3) == 0, "unexpected regOffs");
         ins(guest.V4S(), regOffs/4, host.V4S(), 0);
         break;
 
       case 8:
-        LOGMAN_THROW_A((regOffs & 7) == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT((regOffs & 7) == 0, "unexpected regOffs");
         ins(guest.V2D(), regOffs / 8, host.V2D(), 0);
         break;
 
       case 16:
-        LOGMAN_THROW_A(regOffs == 0, "unexpected regOffs");
+        LOGMAN_THROW_A_FMT(regOffs == 0, "unexpected regOffs");
         if (guest.GetCode() != host.GetCode())
           mov(guest.Q(), host.Q());
         break;
     }
   } else {
-    LOGMAN_THROW_A(false, "Unhandled Op->Class %d", Op->Class);
+    LOGMAN_THROW_A_FMT(false, "Unhandled Op->Class {}", Op->Class);
   }
 }
 
@@ -287,15 +288,17 @@ DEF_OP(LoadContextIndexed) {
         ldr(GetReg<RA_64>(Node), MemOperand(TMP1, Op->BaseOffset));
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
     case 16:
-      LOGMAN_MSG_A("Invalid Class load of size 16");
+      LOGMAN_MSG_A_FMT("Invalid Class load of size 16");
       break;
     default:
-      LOGMAN_MSG_A("Unhandled LoadContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
   else {
@@ -332,12 +335,14 @@ DEF_OP(LoadContextIndexed) {
         }
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
     default:
-      LOGMAN_MSG_A("Unhandled LoadContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
 }
@@ -373,15 +378,17 @@ DEF_OP(StoreContextIndexed) {
         str(value, MemOperand(TMP1, Op->BaseOffset));
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
     case 16:
-      LOGMAN_MSG_A("Invalid Class load of size 16");
+      LOGMAN_MSG_A_FMT("Invalid Class store of size 16");
       break;
     default:
-      LOGMAN_MSG_A("Unhandled LoadContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
   else {
@@ -420,12 +427,14 @@ DEF_OP(StoreContextIndexed) {
         }
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
     default:
-      LOGMAN_MSG_A("Unhandled LoadContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
 }
@@ -433,7 +442,7 @@ DEF_OP(StoreContextIndexed) {
 DEF_OP(SpillRegister) {
   auto Op = IROp->C<IR::IROp_SpillRegister>();
   uint8_t OpSize = IROp->Size;
-  uint32_t SlotOffset = Op->Slot * 16 + 16;
+  uint32_t SlotOffset = Op->Slot * 16;
 
   if (Op->Class == FEXCore::IR::GPRClass) {
     switch (OpSize) {
@@ -453,7 +462,7 @@ DEF_OP(SpillRegister) {
       str(GetReg<RA_64>(Op->Header.Args[0].ID()), MemOperand(sp, SlotOffset));
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled SpillRegister size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled SpillRegister size: {}", OpSize);
     }
   } else if (Op->Class == FEXCore::IR::FPRClass) {
     switch (OpSize) {
@@ -469,17 +478,17 @@ DEF_OP(SpillRegister) {
       str(GetSrc(Op->Header.Args[0].ID()), MemOperand(sp, SlotOffset));
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled SpillRegister size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled SpillRegister size: {}", OpSize);
     }
   } else {
-    LOGMAN_MSG_A("Unhandled SpillRegister class: %d", Op->Class.Val);
+    LOGMAN_MSG_A_FMT("Unhandled SpillRegister class: {}", Op->Class.Val);
   }
 }
 
 DEF_OP(FillRegister) {
   auto Op = IROp->C<IR::IROp_FillRegister>();
   uint8_t OpSize = IROp->Size;
-  uint32_t SlotOffset = Op->Slot * 16 + 16;
+  uint32_t SlotOffset = Op->Slot * 16;
 
   if (Op->Class == FEXCore::IR::GPRClass) {
     switch (OpSize) {
@@ -499,7 +508,7 @@ DEF_OP(FillRegister) {
       ldr(GetReg<RA_64>(Node), MemOperand(sp, SlotOffset));
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled SpillRegister size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled FillRegister size: {}", OpSize);
     }
   } else if (Op->Class == FEXCore::IR::FPRClass) {
     switch (OpSize) {
@@ -515,10 +524,10 @@ DEF_OP(FillRegister) {
       ldr(GetDst(Node), MemOperand(sp, SlotOffset));
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled SpillRegister size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled FillRegister size: {}", OpSize);
     }
   } else {
-    LOGMAN_MSG_A("Unhandled FillRegister class: %d", Op->Class.Val);
+    LOGMAN_MSG_A_FMT("Unhandled FillRegister class: {}", Op->Class.Val);
   }
 }
 
@@ -538,7 +547,7 @@ MemOperand Arm64JITCore::GenerateMemOperand(uint8_t AccessSize, aarch64::Registe
     return MemOperand(Base);
   } else {
     if (OffsetScale != 1 && OffsetScale != AccessSize) {
-        LOGMAN_MSG_A("Unhandled GenerateMemOperand OffsetScale: %d", OffsetScale);
+        LOGMAN_MSG_A_FMT("Unhandled GenerateMemOperand OffsetScale: {}", OffsetScale);
     }
     uint64_t Const;
     if (IsInlineConstant(Offset, &Const)) {
@@ -550,11 +559,12 @@ MemOperand Arm64JITCore::GenerateMemOperand(uint8_t AccessSize, aarch64::Registe
         case IR::MEM_OFFSET_UXTW.Val: return MemOperand(Base, RegOffset.W(), Extend::UXTW, (int)std::log2(OffsetScale) );
         case IR::MEM_OFFSET_SXTW.Val: return MemOperand(Base, RegOffset.W(), Extend::SXTW, (int)std::log2(OffsetScale) );
 
-        default: LOGMAN_MSG_A("Unhandled GenerateMemOperand OffsetType: %d", OffsetType.Val); break;
+        default: LOGMAN_MSG_A_FMT("Unhandled GenerateMemOperand OffsetType: {}", OffsetType.Val); break;
       }
     }
   }
-  __builtin_unreachable();
+
+  FEX_UNREACHABLE;
 }
 
 DEF_OP(LoadMem) {
@@ -578,7 +588,7 @@ DEF_OP(LoadMem) {
       case 8:
         ldr(Dst, MemSrc);
         break;
-      default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled LoadMem size: {}", Op->Size);
     }
   }
   else {
@@ -599,7 +609,7 @@ DEF_OP(LoadMem) {
       case 16:
         ldr(Dst, MemSrc);
         break;
-      default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled LoadMem size: {}", Op->Size);
     }
   }
 }
@@ -610,7 +620,7 @@ DEF_OP(LoadMemTSO) {
   auto MemSrc = MemOperand(GetReg<RA_64>(Op->Header.Args[0].ID()));
 
   if (!Op->Offset.IsInvalid()) {
-    LOGMAN_MSG_A("LoadMemTSO: No offset allowed");
+    LOGMAN_MSG_A_FMT("LoadMemTSO: No offset allowed");
   }
 
   if (SupportsRCPC && Op->Class == FEXCore::IR::GPRClass) {
@@ -633,7 +643,7 @@ DEF_OP(LoadMemTSO) {
         case 8:
           ldapr(Dst, MemSrc);
           break;
-        default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+        default:  LOGMAN_MSG_A_FMT("Unhandled LoadMemTSO size: {}", Op->Size);
       }
       nop();
     }
@@ -658,7 +668,7 @@ DEF_OP(LoadMemTSO) {
         case 8:
           ldar(Dst, MemSrc);
           break;
-        default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+        default:  LOGMAN_MSG_A_FMT("Unhandled LoadMemTSO size: {}", Op->Size);
       }
       nop();
     }
@@ -679,7 +689,7 @@ DEF_OP(LoadMemTSO) {
       case 16:
         ldr(Dst, MemSrc);
         break;
-      default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled LoadMemTSO size: {}", Op->Size);
     }
     dmb(InnerShareable, BarrierAll);
   }
@@ -706,7 +716,7 @@ DEF_OP(StoreMem) {
       case 8:
         str(GetReg<RA_64>(Op->Header.Args[1].ID()), MemSrc);
         break;
-      default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled StoreMem size: {}", Op->Size);
     }
   }
   else {
@@ -727,7 +737,7 @@ DEF_OP(StoreMem) {
       case 16:
         str(Src, MemSrc);
         break;
-      default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled StoreMem size: {}", Op->Size);
     }
   }
 }
@@ -737,7 +747,7 @@ DEF_OP(StoreMemTSO) {
   auto MemSrc = MemOperand(GetReg<RA_64>(Op->Header.Args[0].ID()));
 
   if (!Op->Offset.IsInvalid()) {
-    LOGMAN_MSG_A("StoreMemTSO: No offset allowed");
+    LOGMAN_MSG_A_FMT("StoreMemTSO: No offset allowed");
   }
 
   if (Op->Class == FEXCore::IR::GPRClass) {
@@ -757,7 +767,7 @@ DEF_OP(StoreMemTSO) {
         case 8:
           stlr(GetReg<RA_64>(Op->Header.Args[1].ID()), MemSrc);
           break;
-        default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+        default:  LOGMAN_MSG_A_FMT("Unhandled StoreMemTSO size: {}", Op->Size);
       }
       nop();
     }
@@ -781,7 +791,7 @@ DEF_OP(StoreMemTSO) {
       case 16:
         str(Src, MemSrc);
         break;
-      default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled StoreMemTSO size: {}", Op->Size);
     }
     dmb(InnerShareable, BarrierAll);
   }
@@ -793,7 +803,7 @@ DEF_OP(ParanoidLoadMemTSO) {
   auto MemSrc = MemOperand(GetReg<RA_64>(Op->Header.Args[0].ID()));
 
   if (!Op->Offset.IsInvalid()) {
-    LOGMAN_MSG_A("LoadMemTSO: No offset allowed");
+    LOGMAN_MSG_A_FMT("ParanoidLoadMemTSO: No offset allowed");
   }
 
   if (Op->Class == FEXCore::IR::GPRClass) {
@@ -804,7 +814,6 @@ DEF_OP(ParanoidLoadMemTSO) {
     }
     else {
       auto Dst = GetReg<RA_64>(Node);
-      nop();
       switch (Op->Size) {
         case 2:
           ldarh(Dst, MemSrc);
@@ -815,31 +824,24 @@ DEF_OP(ParanoidLoadMemTSO) {
         case 8:
           ldar(Dst, MemSrc);
           break;
-        default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+        default:  LOGMAN_MSG_A_FMT("Unhandled ParanoidLoadMemTSO size: {}", Op->Size);
       }
-      nop();
     }
   }
   else {
     auto Dst = GetDst(Node);
     switch (Op->Size) {
       case 2:
-        nop();
-        ldarh(TMP1, MemSrc);
-        nop();
-        fmov(Dst, TMP1);
+        ldarh(TMP1.W(), MemSrc);
+        fmov(Dst.H(), TMP1.W());
         break;
       case 4:
-        nop();
         ldar(TMP1.W(), MemSrc);
-        nop();
-        fmov(Dst, TMP1);
+        fmov(Dst.S(), TMP1.W());
         break;
       case 8:
-        nop();
         ldar(TMP1, MemSrc);
-        nop();
-        fmov(Dst, TMP1);
+        fmov(Dst.D(), TMP1);
         break;
       case 16:
         nop();
@@ -848,7 +850,7 @@ DEF_OP(ParanoidLoadMemTSO) {
         mov(Dst.V2D(), 0, TMP1);
         mov(Dst.V2D(), 1, TMP2);
         break;
-      default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled ParanoidLoadMemTSO size: {}", Op->Size);
     }
   }
 }
@@ -858,7 +860,7 @@ DEF_OP(ParanoidStoreMemTSO) {
   auto MemSrc = MemOperand(GetReg<RA_64>(Op->Header.Args[0].ID()));
 
   if (!Op->Offset.IsInvalid()) {
-    LOGMAN_MSG_A("StoreMemTSO: No offset allowed");
+    LOGMAN_MSG_A_FMT("ParanoidStoreMemTSO: No offset allowed");
   }
 
   if (Op->Class == FEXCore::IR::GPRClass) {
@@ -867,7 +869,6 @@ DEF_OP(ParanoidStoreMemTSO) {
       stlrb(GetReg<RA_64>(Op->Header.Args[1].ID()), MemSrc);
     }
     else {
-      nop();
       switch (Op->Size) {
         case 2:
           stlrh(GetReg<RA_64>(Op->Header.Args[1].ID()), MemSrc);
@@ -878,37 +879,30 @@ DEF_OP(ParanoidStoreMemTSO) {
         case 8:
           stlr(GetReg<RA_64>(Op->Header.Args[1].ID()), MemSrc);
           break;
-        default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+        default:  LOGMAN_MSG_A_FMT("Unhandled ParanoidStoreMemTSO size: {}", Op->Size);
       }
-      nop();
     }
   }
   else {
     auto Src = GetSrc(Op->Header.Args[1].ID());
     if (Op->Size == 1) {
       // 8bit load is always aligned to natural alignment
-      mov(TMP1, Src.V4S(), 0);
+      mov(TMP1.W(), Src.V16B(), 0);
       stlrb(TMP1, MemSrc);
     }
     else {
       switch (Op->Size) {
         case 2:
-          mov(TMP1, Src.V4S(), 0);
-          nop();
+          mov(TMP1.W(), Src.V8H(), 0);
           stlrh(TMP1, MemSrc);
-          nop();
           break;
         case 4:
-          mov(TMP1, Src.V4S(), 0);
-          nop();
+          mov(TMP1.W(), Src.V4S(), 0);
           stlr(TMP1.W(), MemSrc);
-          nop();
           break;
         case 8:
           mov(TMP1, Src.V2D(), 0);
-          nop();
           stlr(TMP1, MemSrc);
-          nop();
           break;
         case 16: {
           // Move vector to GPRs
@@ -925,18 +919,18 @@ DEF_OP(ParanoidStoreMemTSO) {
           cbnz(TMP3, &B); // < Overwritten with DMB
           break;
         }
-        default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+        default:  LOGMAN_MSG_A_FMT("Unhandled ParanoidStoreMemTSO size: {}", Op->Size);
       }
     }
   }
 }
 
 DEF_OP(VLoadMemElement) {
-  LOGMAN_MSG_A("Unimplemented");
+  LOGMAN_MSG_A_FMT("Unimplemented");
 }
 
 DEF_OP(VStoreMemElement) {
-  LOGMAN_MSG_A("Unimplemented");
+  LOGMAN_MSG_A_FMT("Unimplemented");
 }
 
 DEF_OP(CacheLineClear) {

External/FEXCore/Source/Interface/Core/JIT/Arm64/MiscOps.cpp

@@ -8,11 +8,11 @@ $end_info$
 
 namespace FEXCore::CPU {
 static void PrintValue(uint64_t Value) {
-  LogMan::Msg::D("Value: 0x%lx", Value);
+  LogMan::Msg::DFmt("Value: 0x{:x}", Value);
 }
 
 static void PrintVectorValue(uint64_t Value, uint64_t ValueUpper) {
-  LogMan::Msg::D("Value: 0x%016lx'%016lx", ValueUpper, Value);
+  LogMan::Msg::DFmt("Value: 0x{:016x}'{:016x}", ValueUpper, Value);
 }
 
 using namespace vixl;
@@ -31,7 +31,7 @@ DEF_OP(Fence) {
     case IR::Fence_Store.Val:
       dmb(FullSystem, BarrierWrites);
       break;
-    default: LOGMAN_MSG_A("Unknown Fence: %d", Op->Fence); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Fence: {}", Op->Fence); break;
   }
 }
 
@@ -69,7 +69,7 @@ DEF_OP(Break) {
       br(TMP1);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Break reason: %d", Op->Reason);
+    default: LOGMAN_MSG_A_FMT("Unknown Break reason: {}", Op->Reason);
   }
 }
 

External/FEXCore/Source/Interface/Core/JIT/Arm64/MoveOps.cpp

@@ -26,7 +26,7 @@ DEF_OP(ExtractElementPair) {
       mov (GetReg<RA_64>(Node), Regs[Op->Element]);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Size"); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Size"); break;
   }
 }
 
@@ -52,7 +52,7 @@ DEF_OP(CreateElementPair) {
       RegTmp = x0;
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Size"); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Size"); break;
   }
 
   if (Dst.first.GetCode() != RegSecond.GetCode()) {

External/FEXCore/Source/Interface/Core/JIT/JITCore.h

@@ -1,5 +1,7 @@
 #pragma once
 
+#include <memory>
+
 namespace FEXCore::Context {
 struct Context;
 }
@@ -11,6 +13,6 @@ struct InternalThreadState;
 namespace FEXCore::CPU {
 class CPUBackend;
 
-FEXCore::CPU::CPUBackend *CreateX86JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread);
-FEXCore::CPU::CPUBackend *CreateArm64JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread);
+std::unique_ptr<CPUBackend> CreateX86JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread);
+std::unique_ptr<CPUBackend> CreateArm64JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread);
 }

External/FEXCore/Source/Interface/Core/JIT/x86_64/ALUOps.cpp

@@ -5,7 +5,14 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <array>
+#include <stdint.h>
+#include <utility>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 #define DEF_OP(x) void X86JITCore::Op_##x(FEXCore::IR::IROp_Header *IROp, uint32_t Node)
@@ -20,7 +27,7 @@ DEF_OP(TruncElementPair) {
       mov(Dst.second, Src.second);
       break;
     }
-    default: LOGMAN_MSG_A("Unhandled Truncation size: %d", Op->Size); break;
+    default: LOGMAN_MSG_A_FMT("Unhandled Truncation size: {}", Op->Size); break;
   }
 }
 
@@ -70,7 +77,7 @@ DEF_OP(Add) {
     case 8:
       add(rax, Const);
       break;
-    default:  LOGMAN_MSG_A("Unhandled Add size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Add size: {}", OpSize);
       break;
     }
   } else {
@@ -81,7 +88,7 @@ DEF_OP(Add) {
     case 8:
       add(rax, GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled Add size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Add size: {}", OpSize);
       break;
     }
   }
@@ -103,7 +110,7 @@ DEF_OP(Sub) {
     case 8:
       sub(rax, Const);
       break;
-    default:  LOGMAN_MSG_A("Unhandled Sub size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Sub size: {}", OpSize);
       break;
     }
   } else {
@@ -114,7 +121,7 @@ DEF_OP(Sub) {
     case 8:
       sub(rax, GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled Sub size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled Sub size: {}", OpSize);
       break;
     }
   }
@@ -136,7 +143,7 @@ DEF_OP(Neg) {
     Src = GetSrc<RA_64>(Op->Header.Args[0].ID());
     Dst = GetDst<RA_64>(Node);
     break;
-  default:  LOGMAN_MSG_A("Unhandled Neg size: %d", OpSize);
+  default:  LOGMAN_MSG_A_FMT("Unhandled Neg size: {}", OpSize);
     break;
   }
   mov(Dst, Src);
@@ -160,7 +167,7 @@ DEF_OP(Mul) {
     imul(rax, GetSrc<RA_64>(Op->Header.Args[1].ID()));
     mov(Dst, rax);
   break;
-  default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+  default: LOGMAN_MSG_A_FMT("Unknown Mul size: {}", OpSize);
   }
 }
 
@@ -179,7 +186,7 @@ DEF_OP(UMul) {
     mul(GetSrc<RA_64>(Op->Header.Args[1].ID()));
     mov(GetDst<RA_64>(Node), rax);
   break;
-  default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+  default: LOGMAN_MSG_A_FMT("Unknown UMul size: {}", OpSize);
   }
 }
 
@@ -218,7 +225,7 @@ DEF_OP(Div) {
     mov(GetDst<RA_64>(Node), rax);
   break;
   }
-  default: LOGMAN_MSG_A("Unknown UDIV Size: %d", Size); break;
+  default: LOGMAN_MSG_A_FMT("Unknown DIV Size: {}", Size); break;
   }
 }
 
@@ -261,7 +268,7 @@ DEF_OP(UDiv) {
     mov(GetDst<RA_64>(Node), rax);
   break;
   }
-  default: LOGMAN_MSG_A("Unknown UDIV OpSize: %d", OpSize); break;
+  default: LOGMAN_MSG_A_FMT("Unknown UDIV OpSize: {}", OpSize); break;
   }
 }
 
@@ -298,7 +305,7 @@ DEF_OP(Rem) {
     mov(GetDst<RA_64>(Node), rdx);
   break;
   }
-  default: LOGMAN_MSG_A("Unknown UDIV Size: %d", OpSize); break;
+  default: LOGMAN_MSG_A_FMT("Unknown Rem Size: {}", OpSize); break;
   }
 }
 
@@ -341,7 +348,7 @@ DEF_OP(URem) {
     mov(GetDst<RA_64>(Node), rdx);
   break;
   }
-  default: LOGMAN_MSG_A("Unknown UDIV OpSize: %d", OpSize); break;
+  default: LOGMAN_MSG_A_FMT("Unknown URem OpSize: {}", OpSize); break;
   }
 }
 
@@ -360,7 +367,7 @@ DEF_OP(MulH) {
     imul(GetSrc<RA_64>(Op->Header.Args[1].ID()));
     mov(GetDst<RA_64>(Node), rdx);
   break;
-  default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+  default: LOGMAN_MSG_A_FMT("Unknown MulH size: {}", OpSize);
   }
 }
 
@@ -379,7 +386,7 @@ DEF_OP(UMulH) {
     mul(GetSrc<RA_64>(Op->Header.Args[1].ID()));
     mov(GetDst<RA_64>(Node), rdx);
   break;
-  default: LOGMAN_MSG_A("Unknown Sext size: %d", OpSize);
+  default: LOGMAN_MSG_A_FMT("Unknown UMulH size: {}", OpSize);
   }
 }
 
@@ -441,7 +448,7 @@ DEF_OP(Lshl) {
         mov(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
         shl(GetDst<RA_64>(Node), Const);
         break;
-      default: LOGMAN_MSG_A("Unknown LSHL Size: %d\n", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown LSHL Size: {}\n", OpSize); break;
     };
   } else {
     mov(rcx, GetSrc<RA_64>(Op->Header.Args[1].ID()));
@@ -456,7 +463,7 @@ DEF_OP(Lshl) {
         mov(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
         shl(GetDst<RA_64>(Node), cl);
         break;
-      default: LOGMAN_MSG_A("Unknown LSHL Size: %d\n", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown LSHL Size: {}\n", OpSize); break;
     };
   }
 }
@@ -488,7 +495,7 @@ DEF_OP(Lshr) {
         mov(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
         shr(GetDst<RA_64>(Node), Const);
         break;
-      default: LOGMAN_MSG_A("Unknown Size: %d\n", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown Size: {}\n", OpSize); break;
     };
 
   } else {
@@ -512,7 +519,7 @@ DEF_OP(Lshr) {
         mov(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
         shr(GetDst<RA_64>(Node), cl);
         break;
-      default: LOGMAN_MSG_A("Unknown Size: %d\n", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown Size: {}\n", OpSize); break;
     };
   }
 }
@@ -546,7 +553,7 @@ DEF_OP(Ashr) {
       mov(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
       sar(GetDst<RA_64>(Node), Const);
     break;
-    default: LOGMAN_MSG_A("Unknown ASHR Size: %d\n", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown ASHR Size: {}\n", OpSize); break;
     };
 
   } else {
@@ -571,7 +578,7 @@ DEF_OP(Ashr) {
       mov(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
       sar(GetDst<RA_64>(Node), cl);
     break;
-    default: LOGMAN_MSG_A("Unknown ASHR Size: %d\n", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown ASHR Size: {}\n", OpSize); break;
     };
   }
 }
@@ -596,7 +603,7 @@ DEF_OP(Ror) {
         ror(rax, Const);
       break;
       }
-      default: LOGMAN_MSG_A("Unknown ROR Size: %d\n", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown ROR Size: {}\n", OpSize); break;
     }
   } else {
     mov (rcx, GetSrc<RA_64>(Op->Header.Args[1].ID()));
@@ -612,7 +619,7 @@ DEF_OP(Ror) {
         ror(rax, cl);
       break;
       }
-      default: LOGMAN_MSG_A("Unknown ROR Size: %d\n", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown ROR Size: {}\n", OpSize); break;
     }
   }
   mov(GetDst<RA_64>(Node), rax);
@@ -668,7 +675,7 @@ DEF_OP(LDiv) {
       mov(GetDst<RA_64>(Node), rax);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown LDIV OpSize: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LDIV OpSize: {}", OpSize); break;
   }
 }
 
@@ -700,7 +707,7 @@ DEF_OP(LUDiv) {
       mov(GetDst<RA_64>(Node), rax);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown LUDIV OpSize: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LUDIV OpSize: {}", OpSize); break;
   }
 }
 
@@ -732,7 +739,7 @@ DEF_OP(LRem) {
       mov(GetDst<RA_64>(Node), rdx);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown LREM OpSize: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LREM OpSize: {}", OpSize); break;
   }
 }
 
@@ -764,7 +771,7 @@ DEF_OP(LURem) {
       mov(GetDst<RA_64>(Node), rdx);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown LUDIV OpSize: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown LUREM OpSize: {}", OpSize); break;
   }
 }
 
@@ -829,7 +836,7 @@ DEF_OP(FindMSB) {
     case 8:
       bsr(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
       break;
-    default: LOGMAN_MSG_A("Unknown OpSize: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unknown FindMSB OpSize: {}", OpSize);
   }
 }
 
@@ -853,7 +860,7 @@ DEF_OP(FindTrailingZeros) {
       mov(rax, 0x40);
       cmovz(GetDst<RA_64>(Node), rax);
       break;
-    default: LOGMAN_MSG_A("Unknown size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown FindTrailingZeros size: {}", OpSize); break;
   }
 }
 
@@ -876,7 +883,7 @@ DEF_OP(CountLeadingZeroes) {
         lzcnt(GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
         break;
       }
-      default: LOGMAN_MSG_A("Unknown size: %d", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown CountLeadingZeros size: {}", OpSize); break;
     }
   }
   else {
@@ -915,7 +922,7 @@ DEF_OP(CountLeadingZeroes) {
         mov(GetDst<RA_64>(Node), rax);
         break;
       }
-      default: LOGMAN_MSG_A("Unknown size: %d", OpSize); break;
+      default: LOGMAN_MSG_A_FMT("Unknown CountLeadingZeros size: {}", OpSize); break;
     }
   }
 }
@@ -937,7 +944,7 @@ DEF_OP(Rev) {
       mov (GetDst<RA_64>(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()));
       bswap(GetDst<RA_64>(Node).cvt64());
       break;
-    default: LOGMAN_MSG_A("Unknown REV size: %d", OpSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown REV size: {}", OpSize); break;
   }
 }
 
@@ -970,9 +977,7 @@ DEF_OP(Bfi) {
 
 DEF_OP(Bfe) {
   auto Op = IROp->C<IR::IROp_Bfe>();
-  uint8_t OpSize = IROp->Size;
-
-  LOGMAN_THROW_A(OpSize <= 8, "OpSize is too large for BFE: %d", OpSize);
+  LOGMAN_THROW_A_FMT(IROp->Size <= 8, "OpSize is too large for BFE: {}", IROp->Size);
 
   auto Dst = GetDst<RA_64>(Node);
 
@@ -1073,7 +1078,7 @@ DEF_OP(Select) {
 
   if (is_const_true || is_const_false) {
     if (is_const_false != true || is_const_true != true || const_true != 1 || const_false != 0) {
-      LOGMAN_MSG_A("Select: Unsupported compare inline parameters");
+      LOGMAN_MSG_A_FMT("Select: Unsupported compare inline parameters");
     }
     (this->*SetCC)(al);
     movzx(Dst, al);
@@ -1104,46 +1109,67 @@ DEF_OP(VExtractToGPR) {
       pextrq(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()), Op->Idx);
     break;
     }
-    default: LOGMAN_MSG_A("Unknown Element Size: %d", Op->Header.ElementSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Element Size: {}", Op->Header.ElementSize); break;
   }
 }
 
-DEF_OP(Float_ToGPR_ZU) {
-  LogMan::Msg::D("Unimplemented");
-}
-
 DEF_OP(Float_ToGPR_ZS) {
   auto Op = IROp->C<IR::IROp_Float_ToGPR_ZS>();
-  if (Op->Header.ElementSize == 8) {
-    cvttsd2si(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()));
-  }
-  else {
-    cvttss2si(GetDst<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()));
-  }
-}
 
-DEF_OP(Float_ToGPR_U) {
-  LogMan::Msg::D("Unimplemented");
+  uint16_t Conv = (IROp->Size << 8) | Op->SrcElementSize;
+  switch (Conv) {
+    case 0x0804: // int64_t <- float
+      cvttss2si(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+    case 0x0808: // int64_t <- double
+      cvttsd2si(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+    case 0x0404: // int32_t <- float
+      cvttss2si(GetDst<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+    case 0x0408: // int32_t <- double
+      cvttsd2si(GetDst<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+  }
 }
 
 DEF_OP(Float_ToGPR_S) {
   auto Op = IROp->C<IR::IROp_Float_ToGPR_S>();
-  if (Op->Header.ElementSize == 8) {
-    cvtsd2si(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()));
-  }
-  else {
-    cvtss2si(GetDst<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()));
+  uint16_t Conv = (IROp->Size << 8) | Op->SrcElementSize;
+  switch (Conv) {
+    case 0x0804: // int64_t <- float
+      cvtss2si(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+    case 0x0808: // int64_t <- double
+      cvtsd2si(GetDst<RA_64>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+    case 0x0404: // int32_t <- float
+      cvtss2si(GetDst<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
+    case 0x0408: // int32_t <- double
+      cvtsd2si(GetDst<RA_32>(Node), GetSrc(Op->Header.Args[0].ID()));
+    break;
   }
 }
 
 DEF_OP(FCmp) {
   auto Op = IROp->C<IR::IROp_FCmp>();
 
-  if (Op->ElementSize == 4) {
-    ucomiss(GetSrc(Op->Header.Args[0].ID()), GetSrc(Op->Header.Args[1].ID()));
+  if (Op->Flags & (1 << IR::FCMP_FLAG_UNORDERED)) {
+    if (Op->ElementSize == 4) {
+      ucomiss(GetSrc(Op->Header.Args[0].ID()), GetSrc(Op->Header.Args[1].ID()));
+    }
+    else {
+      ucomisd(GetSrc(Op->Header.Args[0].ID()), GetSrc(Op->Header.Args[1].ID()));
+    }
   }
   else {
-    ucomisd(GetSrc(Op->Header.Args[0].ID()), GetSrc(Op->Header.Args[1].ID()));
+    if (Op->ElementSize == 4) {
+      comiss(GetSrc(Op->Header.Args[0].ID()), GetSrc(Op->Header.Args[1].ID()));
+    }
+    else {
+      comisd(GetSrc(Op->Header.Args[0].ID()), GetSrc(Op->Header.Args[1].ID()));
+    }
   }
   mov (rdx, 0);
 
@@ -1217,9 +1243,7 @@ void X86JITCore::RegisterALUHandlers() {
   REGISTER_OP(SBFE,              Sbfe);
   REGISTER_OP(SELECT,            Select);
   REGISTER_OP(VEXTRACTTOGPR,     VExtractToGPR);
-  REGISTER_OP(FLOAT_TOGPR_ZU,    Float_ToGPR_ZU);
   REGISTER_OP(FLOAT_TOGPR_ZS,    Float_ToGPR_ZS);
-  REGISTER_OP(FLOAT_TOGPR_U,     Float_ToGPR_U);
   REGISTER_OP(FLOAT_TOGPR_S,     Float_ToGPR_S);
   REGISTER_OP(FCMP,              FCmp);
 #undef REGISTER_OP

External/FEXCore/Source/Interface/Core/JIT/x86_64/AtomicOps.cpp

@@ -5,7 +5,14 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <array>
+#include <stdint.h>
+#include <utility>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 #define DEF_OP(x) void X86JITCore::Op_##x(FEXCore::IR::IROp_Header *IROp, uint32_t Node)
@@ -55,7 +62,7 @@ DEF_OP(CASPair) {
       mov(Dst.second, rdx);
     break;
     }
-    default: LOGMAN_MSG_A("Unsupported: %d", OpSize);
+    default: LOGMAN_MSG_A_FMT("Unsupported: {}", OpSize);
   }
 }
 
@@ -104,7 +111,7 @@ DEF_OP(CAS) {
     mov (GetDst<RA_64>(Node), rax);
   break;
   }
-  default: LOGMAN_MSG_A("Unsupported: %d", OpSize);
+  default: LOGMAN_MSG_A_FMT("Unsupported: {}", OpSize);
   }
 }
 
@@ -127,7 +134,7 @@ DEF_OP(AtomicAdd) {
     case 8:
       add(qword [MemReg], GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicAdd size: {}", Op->Size);
   }
 }
 
@@ -149,7 +156,7 @@ DEF_OP(AtomicSub) {
     case 8:
       sub(qword [MemReg], GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicAdd size: {}", Op->Size);
   }
 }
 
@@ -171,7 +178,7 @@ DEF_OP(AtomicAnd) {
     case 8:
       and_(qword [MemReg], GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicAdd size: {}", Op->Size);
   }
 }
 
@@ -193,7 +200,7 @@ DEF_OP(AtomicOr) {
     case 8:
       or_(qword [MemReg], GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicAdd size: {}", Op->Size);
   }
 }
 
@@ -215,7 +222,7 @@ DEF_OP(AtomicXor) {
     case 8:
       xor_(qword [MemReg], GetSrc<RA_64>(Op->Header.Args[1].ID()));
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicAdd size: {}", Op->Size);
   }
 }
 
@@ -246,7 +253,7 @@ DEF_OP(AtomicSwap) {
       lock();
       xchg(qword [MemReg], GetDst<RA_64>(Node));
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicSwap size: {}", Op->Size);
   }
 }
 
@@ -279,7 +286,7 @@ DEF_OP(AtomicFetchAdd) {
       xadd(qword [MemReg], rcx);
       mov(GetDst<RA_64>(Node), rcx);
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicFetchAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicFetchAdd size: {}", Op->Size);
   }
 }
 
@@ -316,7 +323,7 @@ DEF_OP(AtomicFetchSub) {
       xadd(qword [MemReg], rcx);
       mov(GetDst<RA_64>(Node), rcx);
       break;
-    default:  LOGMAN_MSG_A("Unhandled AtomicFetchAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicFetchSub size: {}", Op->Size);
   }
 }
 
@@ -394,7 +401,7 @@ DEF_OP(AtomicFetchAnd) {
       mov(GetDst<RA_64>(Node), TMP3.cvt64());
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled AtomicFetchAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicFetchAnd size: {}", Op->Size);
   }
 }
 
@@ -471,7 +478,7 @@ DEF_OP(AtomicFetchOr) {
       mov(GetDst<RA_64>(Node), TMP3.cvt64());
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled AtomicFetchAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicFetchOr size: {}", Op->Size);
   }
 }
 
@@ -548,7 +555,83 @@ DEF_OP(AtomicFetchXor) {
       mov(GetDst<RA_64>(Node), TMP3.cvt64());
       break;
     }
-    default:  LOGMAN_MSG_A("Unhandled AtomicFetchAdd size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicFetchXor size: {}", Op->Size);
+  }
+}
+
+DEF_OP(AtomicFetchNeg) {
+  auto Op = IROp->C<IR::IROp_AtomicFetchNeg>();
+
+  Xbyak::Reg MemReg = GetSrc<RA_64>(Op->Header.Args[0].ID());
+  switch (Op->Size) {
+    case 1: {
+      mov(TMP1.cvt8(), byte [MemReg]);
+
+      Label Loop;
+      L(Loop);
+      mov(TMP2.cvt8(), TMP1.cvt8());
+      mov(TMP3.cvt8(), TMP1.cvt8());
+      neg(TMP2.cvt8());
+
+      // Updates RAX with the value from memory
+      lock(); cmpxchg(byte [MemReg], TMP2.cvt8());
+      jne(Loop);
+      // Result is the previous value from memory, which is currently in TMP3
+      movzx(GetDst<RA_64>(Node), TMP3.cvt8());
+      break;
+    }
+    case 2: {
+      mov(TMP1.cvt16(), word [MemReg]);
+
+      Label Loop;
+      L(Loop);
+      mov(TMP2.cvt16(), TMP1.cvt16());
+      mov(TMP3.cvt16(), TMP1.cvt16());
+      neg(TMP2.cvt16());
+
+      // Updates RAX with the value from memory
+      lock(); cmpxchg(word [MemReg], TMP2.cvt16());
+      jne(Loop);
+
+      // Result is the previous value from memory, which is currently in TMP3
+      movzx(GetDst<RA_64>(Node), TMP3.cvt16());
+      break;
+    }
+    case 4: {
+      mov(TMP1.cvt32(), dword [MemReg]);
+
+      Label Loop;
+      L(Loop);
+      mov(TMP2.cvt32(), TMP1.cvt32());
+      mov(TMP3.cvt32(), TMP1.cvt32());
+      neg(TMP2.cvt32());
+
+      // Updates RAX with the value from memory
+      lock(); cmpxchg(dword [MemReg], TMP2.cvt32());
+      jne(Loop);
+
+      // Result is the previous value from memory, which is currently in TMP3
+      mov(GetDst<RA_32>(Node), TMP3.cvt32());
+      break;
+    }
+    case 8: {
+      mov(TMP1.cvt64(), qword [MemReg]);
+
+      Label Loop;
+      L(Loop);
+      mov(TMP2.cvt64(), TMP1.cvt64());
+      mov(TMP3.cvt64(), TMP1.cvt64());
+      neg(TMP2.cvt64());
+
+      // Updates RAX with the value from memory
+      lock(); cmpxchg(qword [MemReg], TMP2.cvt64());
+      jne(Loop);
+
+      // Result is the previous value from memory, which is currently in TMP3
+      mov(GetDst<RA_64>(Node), TMP3.cvt64());
+      break;
+    }
+    default:  LOGMAN_MSG_A_FMT("Unhandled AtomicFetchNeg size: {}", Op->Size);
   }
 }
 
@@ -568,6 +651,7 @@ void X86JITCore::RegisterAtomicHandlers() {
   REGISTER_OP(ATOMICFETCHAND, AtomicFetchAnd);
   REGISTER_OP(ATOMICFETCHOR,  AtomicFetchOr);
   REGISTER_OP(ATOMICFETCHXOR, AtomicFetchXor);
+  REGISTER_OP(ATOMICFETCHNEG, AtomicFetchNeg);
 #undef REGISTER_OP
 }
 }

External/FEXCore/Source/Interface/Core/JIT/x86_64/BranchOps.cpp

@@ -4,25 +4,41 @@ tags: backend|x86-64
 $end_info$
 */
 
+#include "Interface/Context/Context.h"
+#include "Interface/Core/CPUID.h"
+#include "Interface/Core/Dispatcher/Dispatcher.h"
+#include "Interface/Core/LookupCache.h"
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+#include "Interface/HLE/Thunks/Thunks.h"
 
+#include <FEXCore/Core/CPUID.h>
+#include <FEXCore/Core/CoreState.h>
 #include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Debug/InternalThreadState.h>
 #include <FEXCore/HLE/SyscallHandler.h>
-#include <Interface/HLE/Thunks/Thunks.h>
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <array>
+#include <memory>
+#include <stddef.h>
+#include <stdint.h>
+#include <unordered_map>
+#include <utility>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 #define DEF_OP(x) void X86JITCore::Op_##x(FEXCore::IR::IROp_Header *IROp, uint32_t Node)
 DEF_OP(GuestCallDirect) {
-  LogMan::Msg::D("Unimplemented");
+  LogMan::Msg::DFmt("Unimplemented");
 }
 
 DEF_OP(GuestCallIndirect) {
-  LogMan::Msg::D("Unimplemented");
+  LogMan::Msg::DFmt("Unimplemented");
 }
 
 DEF_OP(GuestReturn) {
-  LogMan::Msg::D("Unimplemented");
+  LogMan::Msg::DFmt("Unimplemented");
 }
 
 DEF_OP(SignalReturn) {
@@ -248,7 +264,7 @@ DEF_OP(Thunk) {
 
 DEF_OP(ValidateCode) {
   auto Op = IROp->C<IR::IROp_ValidateCode>();
-  uint8_t* OldCode = (uint8_t*)&Op->CodeOriginalLow;
+  const auto* OldCode = (const uint8_t*)&Op->CodeOriginalLow;
   int len = Op->CodeLength;
   int idx = 0;
 
@@ -256,20 +272,20 @@ DEF_OP(ValidateCode) {
   mov(rax, Entry + Op->Offset);
   mov(rbx, 1);
   while (len >= 4) {
-    cmp(dword[rax + idx], *(uint32_t*)(OldCode + idx));
+    cmp(dword[rax + idx], *(const uint32_t*)(OldCode + idx));
     cmovne(GetDst<RA_64>(Node), rbx);
     len-=4;
     idx+=4;
   }
   while (len >= 2) {
-    mov(rcx, *(uint16_t*)(OldCode + idx));
+    mov(rcx, *(const uint16_t*)(OldCode + idx));
     cmp(word[rax + idx], cx);
     cmovne(GetDst<RA_64>(Node), rbx);
     len-=2;
     idx+=2;
   }
   while (len >= 1) {
-    cmp(byte[rax + idx], *(uint8_t*)(OldCode + idx));
+    cmp(byte[rax + idx], *(const uint8_t*)(OldCode + idx));
     cmovne(GetDst<RA_64>(Node), rbx);
     len-=1;
     idx+=1;
@@ -319,8 +335,9 @@ DEF_OP(CPUID) {
   //
   // Result: RAX, RDX. 4xi32
 
-  mov (rsi, GetSrc<RA_64>(Op->Header.Args[0].ID()));
-  mov (rdx, GetSrc<RA_64>(Op->Header.Args[1].ID()));
+  // rsi can be in the source registers, so copy argument to edx first
+  mov (edx, GetSrc<RA_32>(Op->Header.Args[1].ID()));
+  mov (esi, GetSrc<RA_32>(Op->Header.Args[0].ID()));
   mov (rdi, reinterpret_cast<uint64_t>(&CTX->CPUID));
 
   auto NumPush = RA64.size();

External/FEXCore/Source/Interface/Core/JIT/x86_64/ConversionOps.cpp

@@ -5,7 +5,13 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <array>
+#include <stdint.h>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 
@@ -31,7 +37,7 @@ DEF_OP(VInsGPR) {
       pinsrq(GetDst(Node), GetSrc<RA_64>(Op->Header.Args[1].ID()), Op->Index);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Element Size: %d", Op->Header.ElementSize); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Element Size: {}", Op->Header.ElementSize); break;
   }
 }
 
@@ -52,14 +58,10 @@ DEF_OP(VCastFromGPR) {
     case 8:
       vmovq(GetDst(Node), GetSrc<RA_64>(Op->Header.Args[0].ID()).cvt64());
       break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown VCastFromGPR element size: {}", Op->Header.ElementSize);
   }
 }
 
-DEF_OP(Float_FromGPR_U) {
-  LOGMAN_MSG_A("Unimplemented");
-}
-
 DEF_OP(Float_FromGPR_S) {
   auto Op = IROp->C<IR::IROp_Float_FromGPR_S>();
   uint16_t Conv = (Op->Header.ElementSize << 8) | Op->SrcElementSize;
@@ -95,14 +97,10 @@ DEF_OP(Float_FToF) {
       cvtsd2ss(GetDst(Node), GetSrc(Op->Header.Args[0].ID()));
       break;
     }
-    default: LOGMAN_MSG_A("Unknown FCVT sizes: 0x%x", Conv);
+    default: LOGMAN_MSG_A_FMT("Unknown Float_FToF sizes: 0x{:x}", Conv);
   }
 }
 
-DEF_OP(Vector_UToF) {
-  LOGMAN_MSG_A("Unimplemented");
-}
-
 DEF_OP(Vector_SToF) {
   auto Op = IROp->C<IR::IROp_Vector_SToF>();
   switch (Op->Header.ElementSize) {
@@ -121,14 +119,10 @@ DEF_OP(Vector_SToF) {
       cvtsi2sd(xmm15, rax);
       movlhps(GetDst(Node), xmm15);
     break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_SToF element size: {}", Op->Header.ElementSize);
   }
 }
 
-DEF_OP(Vector_FToZU) {
-  LOGMAN_MSG_A("Unimplemented");
-}
-
 DEF_OP(Vector_FToZS) {
   auto Op = IROp->C<IR::IROp_Vector_FToZS>();
   switch (Op->Header.ElementSize) {
@@ -138,14 +132,10 @@ DEF_OP(Vector_FToZS) {
     case 8:
       cvttpd2dq(GetDst(Node), GetSrc(Op->Header.Args[0].ID()));
     break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_FToZS element size: {}", Op->Header.ElementSize);
   }
 }
 
-DEF_OP(Vector_FToU) {
-  LOGMAN_MSG_A("Unimplemented");
-}
-
 DEF_OP(Vector_FToS) {
   auto Op = IROp->C<IR::IROp_Vector_FToS>();
   switch (Op->Header.ElementSize) {
@@ -155,7 +145,7 @@ DEF_OP(Vector_FToS) {
     case 8:
       cvtpd2dq(GetDst(Node), GetSrc(Op->Header.Args[0].ID()));
     break;
-    default: LOGMAN_MSG_A("Unknown castGPR element size: %d", Op->Header.ElementSize);
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_FToS element size: {}", Op->Header.ElementSize);
   }
 }
 
@@ -172,7 +162,39 @@ DEF_OP(Vector_FToF) {
       cvtpd2ps(GetDst(Node), GetSrc(Op->Header.Args[0].ID()));
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Conversion Type : 0%04x", Conv); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Vector_FToF conversion type : 0x{:04x}", Conv); break;
+  }
+}
+
+DEF_OP(Vector_FToI) {
+  auto Op = IROp->C<IR::IROp_Vector_FToI>();
+  uint8_t RoundMode{};
+
+  switch (Op->Round) {
+    case FEXCore::IR::Round_Nearest.Val:
+      RoundMode = 0b0000'0'0'00;
+    break;
+    case FEXCore::IR::Round_Negative_Infinity.Val:
+      RoundMode = 0b0000'0'0'01;
+    break;
+    case FEXCore::IR::Round_Positive_Infinity.Val:
+      RoundMode = 0b0000'0'0'10;
+    break;
+    case FEXCore::IR::Round_Towards_Zero.Val:
+      RoundMode = 0b0000'0'0'11;
+    break;
+    case FEXCore::IR::Round_Host.Val:
+      RoundMode = 0b0000'0'1'00;
+    break;
+  }
+
+  switch (Op->Header.ElementSize) {
+    case 4:
+      roundps(GetDst(Node), GetSrc(Op->Header.Args[0].ID()), RoundMode);
+    break;
+    case 8:
+      roundpd(GetDst(Node), GetSrc(Op->Header.Args[0].ID()), RoundMode);
+    break;
   }
 }
 
@@ -181,16 +203,13 @@ void X86JITCore::RegisterConversionHandlers() {
 #define REGISTER_OP(op, x) OpHandlers[FEXCore::IR::IROps::OP_##op] = &X86JITCore::Op_##x
   REGISTER_OP(VINSGPR,         VInsGPR);
   REGISTER_OP(VCASTFROMGPR,    VCastFromGPR);
-  REGISTER_OP(FLOAT_FROMGPR_U, Float_FromGPR_U);
   REGISTER_OP(FLOAT_FROMGPR_S, Float_FromGPR_S);
   REGISTER_OP(FLOAT_FTOF,      Float_FToF);
-  REGISTER_OP(VECTOR_UTOF,     Vector_UToF);
   REGISTER_OP(VECTOR_STOF,     Vector_SToF);
-  REGISTER_OP(VECTOR_FTOZU,    Vector_FToZU);
   REGISTER_OP(VECTOR_FTOZS,    Vector_FToZS);
-  REGISTER_OP(VECTOR_FTOU,     Vector_FToU);
   REGISTER_OP(VECTOR_FTOS,     Vector_FToS);
   REGISTER_OP(VECTOR_FTOF,     Vector_FToF);
+  REGISTER_OP(VECTOR_FTOI,     Vector_FToI);
 #undef REGISTER_OP
 }
 }

External/FEXCore/Source/Interface/Core/JIT/x86_64/EncryptionOps.cpp

@@ -5,7 +5,12 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/IR/IR.h>
+
+#include <array>
+#include <stdint.h>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 #define DEF_OP(x) void X86JITCore::Op_##x(FEXCore::IR::IROp_Header *IROp, uint32_t Node)

External/FEXCore/Source/Interface/Core/JIT/x86_64/FlagOps.cpp

@@ -5,7 +5,12 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/IR/IR.h>
+
+#include <array>
+#include <stdint.h>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 

External/FEXCore/Source/Interface/Core/JIT/x86_64/JIT.cpp

@@ -6,21 +6,38 @@ $end_info$
 */
 
 #include "Interface/Context/Context.h"
+#include "Interface/Core/LookupCache.h"
 
+#include "Interface/Core/Dispatcher/Dispatcher.h"
 #include "Interface/Core/Dispatcher/X86Dispatcher.h"
+#include "Interface/Core/Interpreter/InterpreterOps.h"
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/Core/InternalThreadState.h"
-
+#include "Interface/IR/PassManager.h"
 #include "Interface/IR/Passes/RegisterAllocationPass.h"
 
-#include <FEXCore/Core/X86Enums.h>
-#include <FEXCore/Core/UContext.h>
+#include <FEXCore/Core/CPUBackend.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Debug/InternalThreadState.h>
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/IR/IntrusiveIRList.h>
+#include <FEXCore/IR/RegisterAllocationData.h>
 #include <FEXCore/Utils/Allocator.h>
+#include <FEXCore/Utils/LogManager.h>
 
-#include <cmath>
+#include <algorithm>
+#include <array>
+#include <bits/types/stack_t.h>
+#include <memory>
+#include <stddef.h>
+#include <stdint.h>
 #include <signal.h>
-
-#include "Interface/Core/Interpreter/InterpreterOps.h"
+#include <sys/mman.h>
+#include <tuple>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+#include <xbyak/xbyak.h>
 
 // #define DEBUG_RA 1
 // #define DEBUG_CYCLES
@@ -36,7 +53,7 @@ CodeBuffer AllocateNewCodeBuffer(size_t Size) {
                     PROT_READ | PROT_WRITE | PROT_EXEC,
                     MAP_PRIVATE | MAP_ANONYMOUS,
                     -1, 0));
-  LOGMAN_THROW_A(Buffer.Ptr != reinterpret_cast<uint8_t*>(~0ULL), "Couldn't allocate code buffer");
+  LOGMAN_THROW_A_FMT(Buffer.Ptr != reinterpret_cast<uint8_t*>(~0ULL), "Couldn't allocate code buffer");
   return Buffer;
 }
 
@@ -54,9 +71,9 @@ void X86JITCore::CopyNecessaryDataForCompileThread(CPUBackend *Original) {
 }
 
 void X86JITCore::PushRegs() {
-  for (auto &Xmm : RAXMM_x) {
-    sub(rsp, 16);
-    movaps(ptr[rsp], Xmm);
+  sub(rsp, 16 * RAXMM_x.size());
+  for (size_t i = 0; i < RAXMM_x.size(); ++i) {
+    movaps(ptr[rsp + i * 16], RAXMM_x[i]);
   }
 
   for (auto &Reg : RA64)
@@ -75,17 +92,19 @@ void X86JITCore::PopRegs() {
   for (uint32_t i = RA64.size(); i > 0; --i)
     pop(RA64[i - 1]);
 
-  for (uint32_t i = RAXMM_x.size(); i > 0; --i) {
-    movaps(RAXMM_x[i - 1], ptr[rsp]);
-    add(rsp, 16);
+  for (size_t i = 0; i < RAXMM_x.size(); ++i) {
+    movaps(RAXMM_x[i], ptr[rsp + i * 16]);
   }
+
+  add(rsp, 16 * RAXMM_x.size());
 }
 
 void X86JITCore::Op_Unhandled(FEXCore::IR::IROp_Header *IROp, uint32_t Node) {
   FallbackInfo Info;
   if (!InterpreterOps::GetFallbackHandler(IROp, &Info)) {
-    auto Name = FEXCore::IR::GetName(IROp->Op);
-    LOGMAN_MSG_A("Unhandled IR Op: %s", std::string(Name).c_str());
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
+    LOGMAN_MSG_A_FMT("Unhandled IR Op: {}", FEXCore::IR::GetName(IROp->Op));
+#endif
   } else {
     switch(Info.ABI) {
       case FABI_VOID_U16: {
@@ -282,8 +301,10 @@ void X86JITCore::Op_Unhandled(FEXCore::IR::IROp_Header *IROp, uint32_t Node) {
 
       case FABI_UNKNOWN:
       default:
-      auto Name = FEXCore::IR::GetName(IROp->Op);
-        LOGMAN_MSG_A("Unhandled IR Fallback abi: %s %d", std::string(Name).c_str(), Info.ABI);
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
+        LOGMAN_MSG_A_FMT("Unhandled IR Fallback ABI: {} {}", FEXCore::IR::GetName(IROp->Op), Info.ABI);
+#endif
+        break;
     }
   }
 }
@@ -343,19 +364,19 @@ X86JITCore::X86JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalTh
     CTX->SignalDelegation->RegisterHostSignalHandler(SIGILL, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       X86JITCore *Core = reinterpret_cast<X86JITCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleSIGILL(Signal, info, ucontext);
-    });
+    }, true);
 
     CTX->SignalDelegation->RegisterHostSignalHandler(SignalDelegator::SIGNAL_FOR_PAUSE, [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext) -> bool {
       X86JITCore *Core = reinterpret_cast<X86JITCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleSignalPause(Signal, info, ucontext);
-    });
+    }, true);
 
     auto GuestSignalHandler = [](FEXCore::Core::InternalThreadState *Thread, int Signal, void *info, void *ucontext, GuestSigAction *GuestAction, stack_t *GuestStack) -> bool {
       X86JITCore *Core = reinterpret_cast<X86JITCore*>(Thread->CPUBackend.get());
       return Core->Dispatcher->HandleGuestSignal(Signal, info, ucontext, GuestAction, GuestStack);
     };
 
-    for (uint32_t Signal = 0; Signal < SignalDelegator::MAX_SIGNALS; ++Signal) {
+    for (uint32_t Signal = 0; Signal <= SignalDelegator::MAX_SIGNALS; ++Signal) {
       CTX->SignalDelegation->RegisterHostSignalHandlerForGuest(Signal, GuestSignalHandler);
     }
   }
@@ -411,106 +432,106 @@ void X86JITCore::ClearCache() {
   }
 }
 
-IR::PhysicalRegister X86JITCore::GetPhys(uint32_t Node) {
+IR::PhysicalRegister X86JITCore::GetPhys(uint32_t Node) const {
   auto PhyReg = RAData->GetNodeRegister(Node);
 
-  LOGMAN_THROW_A(PhyReg.Raw != 255, "Couldn't Allocate register for node: ssa%d. Class: %d", Node, PhyReg.Class);
+  LOGMAN_THROW_A_FMT(PhyReg.Raw != 255, "Couldn't Allocate register for node: ssa{}. Class: {}", Node, PhyReg.Class);
 
   return PhyReg;
 }
 
-bool X86JITCore::IsFPR(uint32_t Node) {
+bool X86JITCore::IsFPR(uint32_t Node) const {
   return RAData->GetNodeRegister(Node).Class == IR::FPRClass.Val;
 }
 
-bool X86JITCore::IsGPR(uint32_t Node) {
+bool X86JITCore::IsGPR(uint32_t Node) const {
   return RAData->GetNodeRegister(Node).Class == IR::GPRClass.Val;
 }
 
 template<uint8_t RAType>
-Xbyak::Reg X86JITCore::GetSrc(uint32_t Node) {
+Xbyak::Reg X86JITCore::GetSrc(uint32_t Node) const {
   // rax, rcx, rdx, rsi, r8, r9,
   // r10
   // Callee Saved
   // rbx, rbp, r12, r13, r14, r15
   auto PhyReg = GetPhys(Node);
-  if (RAType == RA_64)
+  if constexpr (RAType == RA_64)
     return RA64[PhyReg.Reg].cvt64();
-  else if (RAType == RA_XMM)
+  else if constexpr (RAType == RA_XMM)
     return RAXMM[PhyReg.Reg];
-  else if (RAType == RA_32)
+  else if constexpr (RAType == RA_32)
     return RA64[PhyReg.Reg].cvt32();
-  else if (RAType == RA_16)
+  else if constexpr (RAType == RA_16)
     return RA64[PhyReg.Reg].cvt16();
-  else if (RAType == RA_8)
+  else if constexpr (RAType == RA_8)
     return RA64[PhyReg.Reg].cvt8();
 }
 
 template
-Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_64>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_64>(uint32_t Node) const;
 
 template
-Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_32>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_32>(uint32_t Node) const;
 
 template
-Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_16>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_16>(uint32_t Node) const;
 
 template
-Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_8>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetSrc<X86JITCore::RA_8>(uint32_t Node) const;
 
-Xbyak::Xmm X86JITCore::GetSrc(uint32_t Node) {
+Xbyak::Xmm X86JITCore::GetSrc(uint32_t Node) const {
   auto PhyReg = GetPhys(Node);
   return RAXMM_x[PhyReg.Reg];
 }
 
 template<uint8_t RAType>
-Xbyak::Reg X86JITCore::GetDst(uint32_t Node) {
+Xbyak::Reg X86JITCore::GetDst(uint32_t Node) const {
   auto PhyReg = GetPhys(Node);
-  if (RAType == RA_64)
+  if constexpr (RAType == RA_64)
     return RA64[PhyReg.Reg].cvt64();
-  else if (RAType == RA_XMM)
+  else if constexpr (RAType == RA_XMM)
     return RAXMM[PhyReg.Reg];
-  else if (RAType == RA_32)
+  else if constexpr (RAType == RA_32)
     return RA64[PhyReg.Reg].cvt32();
-  else if (RAType == RA_16)
+  else if constexpr (RAType == RA_16)
     return RA64[PhyReg.Reg].cvt16();
-  else if (RAType == RA_8)
+  else if constexpr (RAType == RA_8)
     return RA64[PhyReg.Reg].cvt8();
 }
 
 template
-Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_64>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_64>(uint32_t Node) const;
 
 template
-Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_32>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_32>(uint32_t Node) const;
 
 template
-Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_16>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_16>(uint32_t Node) const;
 
 template
-Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_8>(uint32_t Node);
+Xbyak::Reg X86JITCore::GetDst<X86JITCore::RA_8>(uint32_t Node) const;
 
 template<uint8_t RAType>
-std::pair<Xbyak::Reg, Xbyak::Reg> X86JITCore::GetSrcPair(uint32_t Node) {
+std::pair<Xbyak::Reg, Xbyak::Reg> X86JITCore::GetSrcPair(uint32_t Node) const {
   auto PhyReg = GetPhys(Node);
-  if (RAType == RA_64)
+  if constexpr (RAType == RA_64)
     return RA64Pair[PhyReg.Reg];
-  else if (RAType == RA_32)
+  else if constexpr (RAType == RA_32)
     return {RA64Pair[PhyReg.Reg].first.cvt32(), RA64Pair[PhyReg.Reg].second.cvt32()};
 }
 
 template
-std::pair<Xbyak::Reg, Xbyak::Reg> X86JITCore::GetSrcPair<X86JITCore::RA_64>(uint32_t Node);
+std::pair<Xbyak::Reg, Xbyak::Reg> X86JITCore::GetSrcPair<X86JITCore::RA_64>(uint32_t Node) const;
 
 template
-std::pair<Xbyak::Reg, Xbyak::Reg> X86JITCore::GetSrcPair<X86JITCore::RA_32>(uint32_t Node);
+std::pair<Xbyak::Reg, Xbyak::Reg> X86JITCore::GetSrcPair<X86JITCore::RA_32>(uint32_t Node) const;
 
-Xbyak::Xmm X86JITCore::GetDst(uint32_t Node) {
+Xbyak::Xmm X86JITCore::GetDst(uint32_t Node) const {
   auto PhyReg = GetPhys(Node);
   return RAXMM_x[PhyReg.Reg];
 }
 
-bool X86JITCore::IsInlineConstant(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) {
+bool X86JITCore::IsInlineConstant(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) const {
   auto OpHeader = IR->GetOp<IR::IROp_Header>(WNode);
 
   if (OpHeader->Op == IR::IROps::OP_INLINECONSTANT) {
@@ -524,7 +545,7 @@ bool X86JITCore::IsInlineConstant(const IR::OrderedNodeWrapper& WNode, uint64_t*
   }
 }
 
-bool X86JITCore::IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) {
+bool X86JITCore::IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) const {
   auto OpHeader = IR->GetOp<IR::IROp_Header>(WNode);
 
   if (OpHeader->Op == IR::IROps::OP_INLINEENTRYPOINTOFFSET) {
@@ -563,7 +584,7 @@ std::tuple<X86JITCore::SetCC, X86JITCore::CMovCC, X86JITCore::JCC> X86JITCore::G
     case FEXCore::IR::COND_VS:
     case FEXCore::IR::COND_VC:
     default:
-      LOGMAN_MSG_A("Unsupported compare type");
+      LOGMAN_MSG_A_FMT("Unsupported compare type");
       break;
   }
 
@@ -606,7 +627,7 @@ void *X86JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IRLi
     L(RunBlock);
   }
 
-  LOGMAN_THROW_A(RAData != nullptr, "Needs RA");
+  LOGMAN_THROW_A_FMT(RAData != nullptr, "Needs RA");
 
   SpillSlots = RAData->SpillSlots();
 
@@ -662,8 +683,10 @@ void *X86JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IRLi
   for (auto [BlockNode, BlockHeader] : IR->GetBlocks()) {
     using namespace FEXCore::IR;
     {
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
       auto BlockIROp = BlockHeader->CW<IROp_CodeBlock>();
-      LOGMAN_THROW_A(BlockIROp->Header.Op == IR::OP_CODEBLOCK, "IR type failed to be a code block");
+      LOGMAN_THROW_A_FMT(BlockIROp->Header.Op == IR::OP_CODEBLOCK, "IR type failed to be a code block");
+#endif
 
       uint32_t Node = IR->GetID(BlockNode);
       auto IsTarget = JumpTargets.find(Node);
@@ -714,7 +737,7 @@ void *X86JITCore::CompileCode(uint64_t Entry, [[maybe_unused]] FEXCore::IR::IRLi
             Inst << "Reg" << GetPhys(ArgNode) << (i + 1 == NumArgs ? "" : ", ");
         }
 
-        LogMan::Msg::D("%s", Inst.str().c_str());
+        LogMan::Msg::DFmt("{}", Inst.str());
       }
       #endif
       uint32_t ID = IR->GetID(CodeNode);
@@ -764,7 +787,7 @@ uint64_t X86JITCore::ExitFunctionLink(X86JITCore *core, FEXCore::Core::CpuStateF
   return HostCode;
 }
 
-FEXCore::CPU::CPUBackend *CreateX86JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread) {
-  return new X86JITCore(ctx, Thread, AllocateNewCodeBuffer(CompileThread ? X86JITCore::MAX_CODE_SIZE : X86JITCore::INITIAL_CODE_SIZE), CompileThread);
+std::unique_ptr<CPUBackend> CreateX86JITCore(FEXCore::Context::Context *ctx, FEXCore::Core::InternalThreadState *Thread, bool CompileThread) {
+  return std::make_unique<X86JITCore>(ctx, Thread, AllocateNewCodeBuffer(CompileThread ? X86JITCore::MAX_CODE_SIZE : X86JITCore::INITIAL_CODE_SIZE), CompileThread);
 }
 }

External/FEXCore/Source/Interface/Core/JIT/x86_64/JITClass.h

@@ -6,7 +6,6 @@ $end_info$
 
 #pragma once
 
-#include "Interface/Core/LookupCache.h"
 #include "Interface/Core/BlockSamplingData.h"
 #include "Interface/Core/Dispatcher/Dispatcher.h"
 
@@ -112,26 +111,26 @@ private:
   constexpr static uint8_t RA_64 = 3;
   constexpr static uint8_t RA_XMM = 4;
 
-  IR::PhysicalRegister GetPhys(uint32_t Node);
+  IR::PhysicalRegister GetPhys(uint32_t Node) const;
 
-  bool IsFPR(uint32_t Node);
-  bool IsGPR(uint32_t Node);
+  bool IsFPR(uint32_t Node) const;
+  bool IsGPR(uint32_t Node) const;
 
   template<uint8_t RAType>
-  Xbyak::Reg GetSrc(uint32_t Node);
+  Xbyak::Reg GetSrc(uint32_t Node) const;
   template<uint8_t RAType>
-  std::pair<Xbyak::Reg, Xbyak::Reg> GetSrcPair(uint32_t Node);
+  std::pair<Xbyak::Reg, Xbyak::Reg> GetSrcPair(uint32_t Node) const;
 
   template<uint8_t RAType>
-  Xbyak::Reg GetDst(uint32_t Node);
+  Xbyak::Reg GetDst(uint32_t Node) const;
 
-  Xbyak::Xmm GetSrc(uint32_t Node);
-  Xbyak::Xmm GetDst(uint32_t Node);
+  Xbyak::Xmm GetSrc(uint32_t Node) const;
+  Xbyak::Xmm GetDst(uint32_t Node) const;
 
-  Xbyak::RegExp GenerateModRM(Xbyak::Reg Base, IR::OrderedNodeWrapper Offset, IR::MemOffsetType OffsetType, uint8_t OffsetScale);
+  Xbyak::RegExp GenerateModRM(Xbyak::Reg Base, IR::OrderedNodeWrapper Offset, IR::MemOffsetType OffsetType, uint8_t OffsetScale) const;
 
-  bool IsInlineConstant(const IR::OrderedNodeWrapper& Node, uint64_t* Value = nullptr);
-  bool IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value);
+  bool IsInlineConstant(const IR::OrderedNodeWrapper& Node, uint64_t* Value = nullptr) const;
+  bool IsInlineEntrypointOffset(const IR::OrderedNodeWrapper& WNode, uint64_t* Value) const;
 
   IR::RegisterAllocationPass *RAPass;
   FEXCore::IR::RegisterAllocationData *RAData;
@@ -241,9 +240,7 @@ private:
   DEF_OP(Sbfe);
   DEF_OP(Select);
   DEF_OP(VExtractToGPR);
-  DEF_OP(Float_ToGPR_ZU);
   DEF_OP(Float_ToGPR_ZS);
-  DEF_OP(Float_ToGPR_U);
   DEF_OP(Float_ToGPR_S);
   DEF_OP(FCmp);
   DEF_OP(F80Cmp);
@@ -262,6 +259,7 @@ private:
   DEF_OP(AtomicFetchAnd);
   DEF_OP(AtomicFetchOr);
   DEF_OP(AtomicFetchXor);
+  DEF_OP(AtomicFetchNeg);
 
   ///< Branch ops
   DEF_OP(GuestCallDirect);
@@ -281,16 +279,14 @@ private:
   ///< Conversion ops
   DEF_OP(VInsGPR);
   DEF_OP(VCastFromGPR);
-  DEF_OP(Float_FromGPR_U);
   DEF_OP(Float_FromGPR_S);
   DEF_OP(Float_FToF);
   DEF_OP(Vector_UToF);
   DEF_OP(Vector_SToF);
-  DEF_OP(Vector_FToZU);
   DEF_OP(Vector_FToZS);
-  DEF_OP(Vector_FToU);
   DEF_OP(Vector_FToS);
   DEF_OP(Vector_FToF);
+  DEF_OP(Vector_FToI);
 
   ///< Flag ops
   DEF_OP(GetHostFlag);
@@ -333,6 +329,7 @@ private:
   DEF_OP(SplatVector);
   DEF_OP(VMov);
   DEF_OP(VAnd);
+  DEF_OP(VBic);
   DEF_OP(VOr);
   DEF_OP(VXor);
   DEF_OP(VAdd);
@@ -343,8 +340,10 @@ private:
   DEF_OP(VSQSub);
   DEF_OP(VAddP);
   DEF_OP(VAddV);
+  DEF_OP(VUMinV);
   DEF_OP(VURAvg);
   DEF_OP(VAbs);
+  DEF_OP(VPopcount);
   DEF_OP(VFAdd);
   DEF_OP(VFAddP);
   DEF_OP(VFSub);
@@ -364,6 +363,8 @@ private:
   DEF_OP(VSMax);
   DEF_OP(VZip);
   DEF_OP(VZip2);
+  DEF_OP(VUnZip);
+  DEF_OP(VUnZip2);
   DEF_OP(VBSL);
   DEF_OP(VCMPEQ);
   DEF_OP(VCMPEQZ);
@@ -386,6 +387,7 @@ private:
   DEF_OP(VInsElement);
   DEF_OP(VInsScalarElement);
   DEF_OP(VExtractElement);
+  DEF_OP(VDupElement);
   DEF_OP(VExtr);
   DEF_OP(VSLI);
   DEF_OP(VSRI);
@@ -408,6 +410,7 @@ private:
   DEF_OP(VSMull);
   DEF_OP(VUMull2);
   DEF_OP(VSMull2);
+  DEF_OP(VUABDL);
   DEF_OP(VTBL1);
 
   ///< Encryption ops

External/FEXCore/Source/Interface/Core/JIT/x86_64/MemoryOps.cpp

@@ -5,9 +5,15 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
 
-#include <cmath>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/IR/IR.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <array>
+#include <stddef.h>
+#include <stdint.h>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 
@@ -36,10 +42,10 @@ DEF_OP(LoadContext) {
     }
     break;
     case 16: {
-      LOGMAN_MSG_A("Invalid GPR load of size 16");
+      LOGMAN_MSG_A_FMT("Invalid GPR load of size 16");
     }
     break;
-    default:  LOGMAN_MSG_A("Unhandled LoadContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled LoadContext size: {}", OpSize);
     }
   }
   else {
@@ -69,7 +75,7 @@ DEF_OP(LoadContext) {
         movups(GetDst(Node), xword [STATE + Op->Offset]);
     }
     break;
-    default:  LOGMAN_MSG_A("Unhandled LoadContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled LoadContext size: {}", OpSize);
     }
   }
 }
@@ -98,9 +104,9 @@ DEF_OP(StoreContext) {
     }
     break;
     case 16:
-      LogMan::Msg::D("Invalid store size of 16");
+      LogMan::Msg::DFmt("Invalid store size of 16");
     break;
-    default:  LOGMAN_MSG_A("Unhandled StoreContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled StoreContext size: {}", OpSize);
     }
   }
   else {
@@ -129,7 +135,7 @@ DEF_OP(StoreContext) {
         movups(xword [STATE + Op->Offset], GetSrc(Op->Header.Args[0].ID()));
     }
     break;
-    default:  LOGMAN_MSG_A("Unhandled StoreContext size: %d", OpSize);
+    default:  LOGMAN_MSG_A_FMT("Unhandled StoreContext size: {}", OpSize);
     }
   }
 }
@@ -160,17 +166,18 @@ DEF_OP(LoadContextIndexed) {
         mov(GetDst<RA_64>(Node),  qword [rax + index * Op->Stride]);
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
     case 16:
-      LOGMAN_MSG_A("Invalid Class load of size 16");
+      LOGMAN_MSG_A_FMT("Invalid Class load of size 16");
       break;
     default:
-      LOGMAN_MSG_A("Unhandled LoadContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed stride: {}", Op->Stride);
+      break;
     }
-
   }
   else {
     switch (Op->Stride) {
@@ -195,7 +202,8 @@ DEF_OP(LoadContextIndexed) {
         vmovq(GetDst(Node),  qword [rax + index * Op->Stride]);
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
@@ -223,12 +231,14 @@ DEF_OP(LoadContextIndexed) {
           movups(GetDst(Node), xword [STATE + rax]);
         break;
       default:
-        LOGMAN_MSG_A("Unhandled LoadContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed size: {}", Op->Size);
+        break;
       }
       break;
     }
     default:
-      LOGMAN_MSG_A("Unhandled LoadContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled LoadContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
 }
@@ -248,13 +258,14 @@ DEF_OP(StoreContextIndexed) {
     case 4:
     case 8: {
       if (!(size == 1 || size == 2 || size == 4 || size == 8)) {
-        LOGMAN_MSG_A("Unhandled StoreContextIndexed size: %d", Op->Size);
+        LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed size: {}", Op->Size);
       }
       mov(AddressFrame(Op->Size * 8) [rax + index * Op->Stride], value);
       break;
     }
     default:
-      LOGMAN_MSG_A("Unhandled StoreContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
   else {
@@ -279,7 +290,8 @@ DEF_OP(StoreContextIndexed) {
         vmovq(AddressFrame(Op->Size * 8) [rax + index * Op->Stride], value);
         break;
       default:
-        LOGMAN_MSG_A("Unhandled StoreContextIndexed size: %d", size);
+        LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed size: {}", size);
+        break;
       }
       break;
     }
@@ -307,12 +319,14 @@ DEF_OP(StoreContextIndexed) {
           movups(xword [STATE + rax], value);
         break;
       default:
-        LOGMAN_MSG_A("Unhandled StoreContextIndexed size: %d", size);
+        LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed size: {}", size);
+        break;
       }
       break;
     }
     default:
-      LOGMAN_MSG_A("Unhandled StoreContextIndexed stride: %d", Op->Stride);
+      LOGMAN_MSG_A_FMT("Unhandled StoreContextIndexed stride: {}", Op->Stride);
+      break;
     }
   }
 }
@@ -340,7 +354,7 @@ DEF_OP(SpillRegister) {
         mov(qword [rsp + SlotOffset], GetSrc<RA_64>(Op->Header.Args[0].ID()));
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled SpillRegister size: %d", OpSize);
+      default:  LOGMAN_MSG_A_FMT("Unhandled SpillRegister size: {}", OpSize);
     }
   } else if (Op->Class == FEXCore::IR::FPRClass) {
     switch (OpSize) {
@@ -356,10 +370,10 @@ DEF_OP(SpillRegister) {
         movaps(xword [rsp + SlotOffset], GetSrc(Op->Header.Args[0].ID()));
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled SpillRegister size: %d", OpSize);
+      default:  LOGMAN_MSG_A_FMT("Unhandled SpillRegister size: {}", OpSize);
     }
   } else {
-    LOGMAN_MSG_A("Unhandled SpillRegister class: %d", Op->Class.Val);
+    LOGMAN_MSG_A_FMT("Unhandled SpillRegister class: {}", Op->Class.Val);
   }
 
 
@@ -388,7 +402,7 @@ DEF_OP(FillRegister) {
         mov(GetDst<RA_64>(Node), qword [rsp + SlotOffset]);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled FillRegister size: %d", OpSize);
+      default:  LOGMAN_MSG_A_FMT("Unhandled FillRegister size: {}", OpSize);
     }
   } else if (Op->Class == FEXCore::IR::FPRClass) {
     switch (OpSize) {
@@ -404,10 +418,10 @@ DEF_OP(FillRegister) {
         movaps(GetDst(Node), xword [rsp + SlotOffset]);
         break;
       }
-      default:  LOGMAN_MSG_A("Unhandled FillRegister size: %d", OpSize);
+      default:  LOGMAN_MSG_A_FMT("Unhandled FillRegister size: {}", OpSize);
     }
   } else {
-    LOGMAN_MSG_A("Unhandled FillRegister class: %d", Op->Class.Val);
+    LOGMAN_MSG_A_FMT("Unhandled FillRegister class: {}", Op->Class.Val);
   }
 }
 
@@ -425,16 +439,16 @@ DEF_OP(StoreFlag) {
   mov(byte [STATE + (offsetof(FEXCore::Core::CPUState, flags[0]) + Op->Flag)], al);
 }
 
-Xbyak::RegExp X86JITCore::GenerateModRM(Xbyak::Reg Base, IR::OrderedNodeWrapper Offset, IR::MemOffsetType OffsetType, uint8_t OffsetScale) {
+Xbyak::RegExp X86JITCore::GenerateModRM(Xbyak::Reg Base, IR::OrderedNodeWrapper Offset, IR::MemOffsetType OffsetType, uint8_t OffsetScale) const {
   if (Offset.IsInvalid()) {
     return Base;
   } else {
     if (OffsetScale != 1 && OffsetScale != 2 && OffsetScale != 4 && OffsetScale != 8) {
-      LOGMAN_MSG_A("Unhandled GenerateModRM OffsetScale: %d", OffsetScale);
+      LOGMAN_MSG_A_FMT("Unhandled GenerateModRM OffsetScale: {}", OffsetScale);
     }
 
     if (OffsetType != IR::MEM_OFFSET_SXTX) {
-      LOGMAN_MSG_A("Unhandled GenerateModRM OffsetType: %d", OffsetType.Val);
+      LOGMAN_MSG_A_FMT("Unhandled GenerateModRM OffsetType: {}", OffsetType.Val);
     }
 
     uint64_t Const;
@@ -475,7 +489,7 @@ DEF_OP(LoadMem) {
         mov(Dst, qword [MemPtr]);
       }
       break;
-      default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled LoadMem size: {}", Op->Size);
     }
   }
   else
@@ -511,7 +525,7 @@ DEF_OP(LoadMem) {
          }
        }
        break;
-      default:  LOGMAN_MSG_A("Unhandled LoadMem size: %d", Op->Size);
+      default:  LOGMAN_MSG_A_FMT("Unhandled LoadMem size: {}", Op->Size);
     }
   }
 }
@@ -537,7 +551,7 @@ DEF_OP(StoreMem) {
     case 8:
       mov(qword [MemPtr], GetSrc<RA_64>(Op->Header.Args[1].ID()));
     break;
-    default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled StoreMem size: {}", Op->Size);
     }
   }
   else {
@@ -560,17 +574,17 @@ DEF_OP(StoreMem) {
       else
         movups(xword [MemPtr], GetSrc(Op->Header.Args[1].ID()));
     break;
-    default:  LOGMAN_MSG_A("Unhandled StoreMem size: %d", Op->Size);
+    default:  LOGMAN_MSG_A_FMT("Unhandled StoreMem size: {}", Op->Size);
     }
   }
 }
 
 DEF_OP(VLoadMemElement) {
-  LOGMAN_MSG_A("Unimplemented");
+  LOGMAN_MSG_A_FMT("Unimplemented");
 }
 
 DEF_OP(VStoreMemElement) {
-  LOGMAN_MSG_A("Unimplemented");
+  LOGMAN_MSG_A_FMT("Unimplemented");
 }
 
 DEF_OP(CacheLineClear) {

External/FEXCore/Source/Interface/Core/JIT/x86_64/MiscOps.cpp

@@ -4,12 +4,26 @@ tags: backend|x86-64
 $end_info$
 */
 
+#include "Interface/Context/Context.h"
+#include "Interface/Core/Dispatcher/Dispatcher.h"
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/IR/IR.h>
+
+#include <array>
+#include <stddef.h>
+#include <stdint.h>
+#include <xbyak/xbyak.h>
 
 namespace FEXCore::CPU {
 static void PrintValue(uint64_t Value) {
-  LogMan::Msg::D("Value: 0x%lx", Value);
+  LogMan::Msg::DFmt("Value: 0x{:x}", Value);
+}
+
+static void PrintVectorValue(uint64_t Value, uint64_t ValueUpper) {
+  LogMan::Msg::DFmt("Value: 0x{:016x}'{:016x}", ValueUpper, Value);
 }
 
 #define DEF_OP(x) void X86JITCore::Op_##x(FEXCore::IR::IROp_Header *IROp, uint32_t Node)
@@ -26,7 +40,7 @@ DEF_OP(Fence) {
     case IR::Fence_Store.Val:
       sfence();
       break;
-    default: LOGMAN_MSG_A("Unknown Fence: %d", Op->Fence); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Fence: {}", Op->Fence); break;
   }
 }
 
@@ -79,7 +93,7 @@ DEF_OP(Break) {
       }
     break;
     }
-    default: LOGMAN_MSG_A("Unknown Break reason: %d", Op->Reason);
+    default: LOGMAN_MSG_A_FMT("Unknown Break reason: {}", Op->Reason);
   }
 }
 
@@ -119,24 +133,22 @@ DEF_OP(SetRoundingMode) {
 DEF_OP(Print) {
   auto Op = IROp->C<IR::IROp_Print>();
 
-  for (auto &Reg : RA64)
-    push(Reg);
+  PushRegs();
+  if (IsGPR(Op->Header.Args[0].ID())) {
+    mov (rdi, GetSrc<RA_64>(Op->Header.Args[0].ID()));
 
-  auto NumPush = RA64.size();
-  if (NumPush & 1)
-    sub(rsp, 8); // Align
+    mov(rax, reinterpret_cast<uintptr_t>(PrintValue));
+  }
+  else {
+    pextrq(rdi, GetSrc(Op->Header.Args[0].ID()), 0);
+    pextrq(rsi, GetSrc(Op->Header.Args[0].ID()), 1);
 
-  mov (rdi, GetSrc<RA_64>(Op->Header.Args[0].ID()));
-
-  mov(rax, reinterpret_cast<uintptr_t>(PrintValue));
+    mov(rax, reinterpret_cast<uintptr_t>(PrintVectorValue));
+  }
 
   call(rax);
 
-  if (NumPush & 1)
-    add(rsp, 8); // Align
-
-  for (uint32_t i = RA64.size(); i > 0; --i)
-    pop(RA64[i - 1]);
+  PopRegs();
 }
 
 #undef DEF_OP

External/FEXCore/Source/Interface/Core/JIT/x86_64/MoveOps.cpp

@@ -5,7 +5,13 @@ $end_info$
 */
 
 #include "Interface/Core/JIT/x86_64/JITClass.h"
-#include "Interface/IR/Passes/RegisterAllocationPass.h"
+
+#include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/IR/IR.h>
+
+#include <array>
+#include <stdint.h>
+#include <utility>
 
 namespace FEXCore::CPU {
 
@@ -25,7 +31,7 @@ DEF_OP(ExtractElementPair) {
       mov (GetDst<RA_64>(Node), Regs[Op->Element]);
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Size"); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Size"); break;
   }
 }
 
@@ -51,7 +57,7 @@ DEF_OP(CreateElementPair) {
       RegTmp = rax;
       break;
     }
-    default: LOGMAN_MSG_A("Unknown Size"); break;
+    default: LOGMAN_MSG_A_FMT("Unknown Size"); break;
   }
 
   if (Dst.first != RegSecond) {

External/FEXCore/Source/Interface/Core/LookupCache.cpp

@@ -5,10 +5,11 @@ desc: Stores information about blocks, and provides C++ implementations to looku
 $end_info$
 */
 
-#include "Interface/Context/Context.h"
-#include "Interface/Core/Core.h"
-#include "Interface/Core/LookupCache.h"
 #include <FEXCore/Utils/Allocator.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include "Interface/Context/Context.h"
+#include "Interface/Core/LookupCache.h"
 
 #include <sys/mman.h>
 

External/FEXCore/Source/Interface/Core/LookupCache.h

@@ -1,10 +1,18 @@
 #pragma once
-#include "Interface/Context/Context.h"
 #include <FEXCore/Utils/LogManager.h>
 
+#include <cstdint>
+#include <functional>
 #include <map>
+#include <stddef.h>
+#include <utility>
+#include <vector>
 
 namespace FEXCore {
+namespace Context {
+  struct Context;
+}
+
 class LookupCache {
 public:
 
@@ -38,7 +46,10 @@ public:
   std::map<uint64_t, std::vector<uint64_t>> CodePages;
 
   void AddBlockMapping(uint64_t Address, void *HostCode, uint64_t Start, uint64_t Length) { 
-    auto InsertPoint = BlockList.emplace(Address, (uintptr_t)HostCode);
+#if defined(ASSERTIONS_ENABLED) && ASSERTIONS_ENABLED
+    auto InsertPoint =
+#endif
+    BlockList.emplace(Address, (uintptr_t)HostCode);
     LOGMAN_THROW_A(InsertPoint.second == true, "Dupplicate block mapping added");
 
     for (auto CurrentPage = Start >> 12, EndPage = (Start + Length) >> 12; CurrentPage <= EndPage; CurrentPage++) {
@@ -98,8 +109,8 @@ public:
 
   void HintUsedRange(uint64_t Address, uint64_t Size);
 
-  uintptr_t GetL1Pointer() { return L1Pointer; }
-  uintptr_t GetPagePointer() { return PagePointer; }
+  uintptr_t GetL1Pointer() const { return L1Pointer; }
+  uintptr_t GetPagePointer() const { return PagePointer; }
   uintptr_t GetVirtualMemorySize() const { return VirtualMemSize; }
 
   constexpr static size_t L1_ENTRIES = 1 * 1024 * 1024; // Must be a power of 2

External/FEXCore/Source/Interface/Core/OpcodeDispatcher.h

@@ -3,7 +3,8 @@
 #include "Interface/Core/Frontend.h"
 #include "Interface/Context/Context.h"
 
-#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Core/Context.h>
 #include <FEXCore/Debug/X86Tables.h>
 #include <FEXCore/IR/IntrusiveIRList.h>
 #include <FEXCore/IR/IR.h>
@@ -12,9 +13,10 @@
 #include <FEXCore/Utils/LogManager.h>
 
 #include <cstdint>
-#include <functional>
 #include <map>
-#include <set>
+#include <stddef.h>
+#include <utility>
+#include <vector>
 
 namespace FEXCore::IR {
 class Pass;
@@ -85,7 +87,7 @@ public:
       auto it = JumpTargets.find(NextRIP);
       if (it == JumpTargets.end()) {
 
-        uint8_t GPRSize = CTX->Config.Is64BitMode ? 8 : 4;
+        const uint8_t GPRSize = CTX->GetGPRSize();
         // If we don't have a jump target to a new block then we have to leave
         // Set the RIP to the next instruction and leave
         auto RelocatedNextRIP = _EntrypointOffset(NextRIP - Entry, GPRSize);
@@ -105,7 +107,7 @@ public:
 
   void ResetWorkingList();
   void ResetDecodeFailure() { DecodeFailure = false; }
-  bool HadDecodeFailure() { return DecodeFailure; }
+  bool HadDecodeFailure() const { return DecodeFailure; }
 
   void BeginFunction(uint64_t RIP, std::vector<FEXCore::Frontend::Decoder::DecodedBlocks> const *Blocks);
   void Finalize();
@@ -260,12 +262,6 @@ public:
   template<size_t ElementSize>
   void PSUBQOp(OpcodeArgs);
   template<size_t ElementSize>
-  void PMINUOp(OpcodeArgs);
-  template<size_t ElementSize>
-  void PMAXUOp(OpcodeArgs);
-  void PMINSWOp(OpcodeArgs);
-  void PMAXSWOp(OpcodeArgs);
-  template<size_t ElementSize>
   void MOVMSKOp(OpcodeArgs);
   void MOVMSKOpOne(OpcodeArgs);
   template<size_t ElementSize>
@@ -275,20 +271,16 @@ public:
   void PSHUFBOp(OpcodeArgs);
   template<size_t ElementSize, bool HalfSize, bool Low>
   void PSHUFDOp(OpcodeArgs);
-  template<size_t ElementSize>
-  void PCMPEQOp(OpcodeArgs);
-  template<size_t ElementSize>
-  void PCMPGTOp(OpcodeArgs);
   void MOVDOp(OpcodeArgs);
-  template<size_t ElementSize, bool Scalar, uint32_t SrcIndex>
+  template<size_t ElementSize>
   void PSRLDOp(OpcodeArgs);
   template<size_t ElementSize>
   void PSRLI(OpcodeArgs);
   template<size_t ElementSize>
   void PSLLI(OpcodeArgs);
-  template<size_t ElementSize, bool Scalar, uint32_t SrcIndex>
+  template<size_t ElementSize>
   void PSLL(OpcodeArgs);
-  template<size_t ElementSize, bool Scalar, uint32_t SrcIndex>
+  template<size_t ElementSize>
   void PSRAOp(OpcodeArgs);
   void PSRLDQ(OpcodeArgs);
   void PSLLDQ(OpcodeArgs);
@@ -297,21 +289,21 @@ public:
   template<size_t ElementSize>
   void PAVGOp(OpcodeArgs);
   void MOVDDUPOp(OpcodeArgs);
-  template<size_t DstElementSize, bool Signed>
+  template<size_t DstElementSize>
   void CVTGPR_To_FPR(OpcodeArgs);
-  template<size_t SrcElementSize, bool Signed, bool HostRoundingMode>
+  template<size_t SrcElementSize, bool HostRoundingMode>
   void CVTFPR_To_GPR(OpcodeArgs);
-  template<size_t SrcElementSize, bool Signed, bool Widen>
+  template<size_t SrcElementSize, bool Widen>
   void Vector_CVT_Int_To_Float(OpcodeArgs);
   template<size_t DstElementSize, size_t SrcElementSize>
   void Scalar_CVT_Float_To_Float(OpcodeArgs);
   template<size_t DstElementSize, size_t SrcElementSize>
   void Vector_CVT_Float_To_Float(OpcodeArgs);
-  template<size_t SrcElementSize, bool Signed, bool Narrow, bool HostRoundingMode>
+  template<size_t SrcElementSize, bool Narrow, bool HostRoundingMode>
   void Vector_CVT_Float_To_Int(OpcodeArgs);
-  template<size_t SrcElementSize, bool Signed, bool Widen>
+  template<size_t SrcElementSize, bool Widen>
   void MMX_To_XMM_Vector_CVT_Int_To_Float(OpcodeArgs);
-  template<size_t SrcElementSize, bool Signed, bool Narrow, bool HostRoundingMode>
+  template<size_t SrcElementSize, bool HostRoundingMode>
   void XMM_To_MMX_Vector_CVT_Float_To_Int(OpcodeArgs);
   void MASKMOVOp(OpcodeArgs);
   void MOVBetweenGPR_FPR(OpcodeArgs);
@@ -325,17 +317,13 @@ public:
   void ANDNOp(OpcodeArgs);
   template<size_t ElementSize>
   void PINSROp(OpcodeArgs);
+  void InsertPSOp(OpcodeArgs);
   template<size_t ElementSize>
   void PExtrOp(OpcodeArgs);
-  template<size_t ElementSize, bool Signed>
-  void PMULOp(OpcodeArgs);
 
   template<size_t ElementSize>
   void PSIGN(OpcodeArgs);
 
-  template<size_t ElementSize>
-  void PABS(OpcodeArgs);
-
   // X87 Ops
   template<size_t width>
   void FLD(OpcodeArgs);
@@ -394,6 +382,8 @@ public:
   void X87FRSTOR(OpcodeArgs);
   void X87FXAM(OpcodeArgs);
   void X87FCMOV(OpcodeArgs);
+  void X87EMMS(OpcodeArgs);
+  void X87FFREE(OpcodeArgs);
 
   void FXCH(OpcodeArgs);
 
@@ -470,6 +460,23 @@ public:
   void AESDecLastOp(OpcodeArgs);
   void AESKeyGenAssist(OpcodeArgs);
 
+  template<size_t ElementSize, size_t DstElementSize, bool Signed>
+  void ExtendVectorElements(OpcodeArgs);
+  template<size_t ElementSize, bool Scalar>
+  void VectorRound(OpcodeArgs);
+
+  template<size_t ElementSize>
+  void VectorBlend(OpcodeArgs);
+
+  template<size_t ElementSize>
+  void VectorVariableBlend(OpcodeArgs);
+  void PTestOp(OpcodeArgs);
+  void PHMINPOSUWOp(OpcodeArgs);
+  template<size_t ElementSize>
+  void DPPOp(OpcodeArgs);
+
+  void MPSADBWOp(OpcodeArgs);
+
   void UnimplementedOp(OpcodeArgs);
 
 #undef OpcodeArgs
@@ -494,13 +501,23 @@ private:
   void StoreResult(FEXCore::IR::RegisterClassType Class, FEXCore::X86Tables::DecodedOp Op, FEXCore::X86Tables::DecodedOperand const& Operand, OrderedNode *const Src, int8_t Align);
   void StoreResult(FEXCore::IR::RegisterClassType Class, FEXCore::X86Tables::DecodedOp Op, OrderedNode *const Src, int8_t Align);
 
-  uint8_t GetDstSize(FEXCore::X86Tables::DecodedOp Op);
-  uint8_t GetSrcSize(FEXCore::X86Tables::DecodedOp Op);
+  uint8_t GetDstSize(FEXCore::X86Tables::DecodedOp Op) const;
+  uint8_t GetSrcSize(FEXCore::X86Tables::DecodedOp Op) const;
 
   template<unsigned BitOffset>
-  void SetRFLAG(OrderedNode *Value);
-  void SetRFLAG(OrderedNode *Value, unsigned BitOffset);
-  OrderedNode *GetRFLAG(unsigned BitOffset);
+  void SetRFLAG(OrderedNode *Value) {
+    flagsOp = FLAGS_OP_NONE;
+    _StoreFlag(_Bfe(1, 0, Value), BitOffset);
+  }
+
+  void SetRFLAG(OrderedNode *Value, unsigned BitOffset) {
+    flagsOp = FLAGS_OP_NONE;
+    _StoreFlag(_Bfe(1, 0, Value), BitOffset);
+  }
+
+  OrderedNode *GetRFLAG(unsigned BitOffset) {
+    return _LoadFlag(BitOffset);
+  }
 
   OrderedNode *SelectCC(uint8_t OP, OrderedNode *TrueValue, OrderedNode *FalseValue);
 
@@ -523,14 +540,22 @@ private:
   void GenerateFlags_RotateLeftImmediate(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, uint64_t Shift);
 
   OrderedNode * GetX87Top();
+  enum X87Tag {
+    TAG_VALID   = 0b00,
+    TAG_ZERO    = 0b01,
+    TAG_SPECIAL = 0b10,
+    TAG_EMPTY   = 0b11
+  };
+  void SetX87TopTag(OrderedNode *Value, uint32_t Tag);
+  OrderedNode *GetX87FTW(OrderedNode *Value);
   void SetX87Top(OrderedNode *Value);
 
-  bool DestIsLockedMem(FEXCore::X86Tables::DecodedOp Op) {
-    return Op->Dest.TypeNone.Type !=FEXCore::X86Tables::DecodedOperand::TYPE_GPR && (Op->Flags & FEXCore::X86Tables::DecodeFlags::FLAG_LOCK);
+  bool DestIsLockedMem(FEXCore::X86Tables::DecodedOp Op) const {
+    return DestIsMem(Op) && (Op->Flags & FEXCore::X86Tables::DecodeFlags::FLAG_LOCK) != 0;
   }
 
-  bool DestIsMem(FEXCore::X86Tables::DecodedOp Op) {
-    return Op->Dest.TypeNone.Type !=FEXCore::X86Tables::DecodedOperand::TYPE_GPR;
+  bool DestIsMem(FEXCore::X86Tables::DecodedOp Op) const {
+    return !Op->Dest.IsGPR();
   }
 
   void CreateJumpBlocks(std::vector<FEXCore::Frontend::Decoder::DecodedBlocks> const *Blocks);

External/FEXCore/Source/Interface/Core/OpcodeDispatcher/Crypto.cpp

@@ -0,0 +1,63 @@
+/*
+$info$
+tags: frontend|x86-to-ir, opcodes|dispatcher-implementations
+desc: Handles x86/64 Crypto instructions to IR
+$end_info$
+*/
+
+#include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/IR/IREmitter.h>
+#include <FEXCore/Utils/LogManager.h>
+#include "Interface/Core/OpcodeDispatcher.h"
+
+#include <stdint.h>
+
+namespace FEXCore::IR {
+class OrderedNode;
+
+#define OpcodeArgs [[maybe_unused]] FEXCore::X86Tables::DecodedOp Op
+
+void OpDispatchBuilder::AESImcOp(OpcodeArgs) {
+  OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, -1);
+  auto Res = _VAESImc(Src);
+  StoreResult(FPRClass, Op, Res, -1);
+}
+
+void OpDispatchBuilder::AESEncOp(OpcodeArgs) {
+  OrderedNode *Dest = LoadSource(FPRClass, Op, Op->Dest, Op->Flags, -1);
+  OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, -1);
+  auto Res = _VAESEnc(Dest, Src);
+  StoreResult(FPRClass, Op, Res, -1);
+}
+
+void OpDispatchBuilder::AESEncLastOp(OpcodeArgs) {
+  OrderedNode *Dest = LoadSource(FPRClass, Op, Op->Dest, Op->Flags, -1);
+  OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, -1);
+  auto Res = _VAESEncLast(Dest, Src);
+  StoreResult(FPRClass, Op, Res, -1);
+}
+
+void OpDispatchBuilder::AESDecOp(OpcodeArgs) {
+  OrderedNode *Dest = LoadSource(FPRClass, Op, Op->Dest, Op->Flags, -1);
+  OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, -1);
+  auto Res = _VAESDec(Dest, Src);
+  StoreResult(FPRClass, Op, Res, -1);
+}
+
+void OpDispatchBuilder::AESDecLastOp(OpcodeArgs) {
+  OrderedNode *Dest = LoadSource(FPRClass, Op, Op->Dest, Op->Flags, -1);
+  OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, -1);
+  auto Res = _VAESDecLast(Dest, Src);
+  StoreResult(FPRClass, Op, Res, -1);
+}
+
+void OpDispatchBuilder::AESKeyGenAssist(OpcodeArgs) {
+  OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, -1);
+  LOGMAN_THROW_A(Op->Src[1].IsLiteral(), "Src1 needs to be literal here");
+  uint64_t RCON = Op->Src[1].Data.Literal.Value;
+
+  auto Res = _VAESKeyGenAssist(Src, RCON);
+  StoreResult(FPRClass, Op, Res, -1);
+}
+
+}

External/FEXCore/Source/Interface/Core/OpcodeDispatcher/Flags.cpp

@@ -0,0 +1,818 @@
+/*
+$info$
+tags: frontend|x86-to-ir, opcodes|dispatcher-implementations
+desc: Handles x86/64 flag generation
+$end_info$
+*/
+
+#include "Interface/Context/Context.h"
+#include "Interface/Core/OpcodeDispatcher.h"
+
+#include <FEXCore/Core/X86Enums.h>
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/IR/IR.h>
+
+#include <array>
+#include <cstdint>
+
+namespace FEXCore::IR {
+constexpr std::array<uint32_t, 17> FlagOffsets = {
+  FEXCore::X86State::RFLAG_CF_LOC,
+  FEXCore::X86State::RFLAG_PF_LOC,
+  FEXCore::X86State::RFLAG_AF_LOC,
+  FEXCore::X86State::RFLAG_ZF_LOC,
+  FEXCore::X86State::RFLAG_SF_LOC,
+  FEXCore::X86State::RFLAG_TF_LOC,
+  FEXCore::X86State::RFLAG_IF_LOC,
+  FEXCore::X86State::RFLAG_DF_LOC,
+  FEXCore::X86State::RFLAG_OF_LOC,
+  FEXCore::X86State::RFLAG_IOPL_LOC,
+  FEXCore::X86State::RFLAG_NT_LOC,
+  FEXCore::X86State::RFLAG_RF_LOC,
+  FEXCore::X86State::RFLAG_VM_LOC,
+  FEXCore::X86State::RFLAG_AC_LOC,
+  FEXCore::X86State::RFLAG_VIF_LOC,
+  FEXCore::X86State::RFLAG_VIP_LOC,
+  FEXCore::X86State::RFLAG_ID_LOC,
+};
+
+void OpDispatchBuilder::SetPackedRFLAG(bool Lower8, OrderedNode *Src) {
+  uint8_t NumFlags = FlagOffsets.size();
+  if (Lower8) {
+    NumFlags = 5;
+  }
+  auto OneConst = _Constant(1);
+  for (int i = 0; i < NumFlags; ++i) {
+    auto Tmp = _And(_Lshr(Src, _Constant(FlagOffsets[i])), OneConst);
+    SetRFLAG(Tmp, FlagOffsets[i]);
+  }
+}
+
+OrderedNode *OpDispatchBuilder::GetPackedRFLAG(bool Lower8) {
+  OrderedNode *Original = _Constant(2);
+  uint8_t NumFlags = FlagOffsets.size();
+  if (Lower8) {
+    NumFlags = 5;
+  }
+
+  for (int i = 0; i < NumFlags; ++i) {
+    OrderedNode *Flag = _LoadFlag(FlagOffsets[i]);
+    Flag = _Bfe(4, 32, 0, Flag);
+    Flag = _Lshl(Flag, _Constant(FlagOffsets[i]));
+    Original = _Or(Original, Flag);
+  }
+  return Original;
+}
+
+void OpDispatchBuilder::GenerateFlags_ADC(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2, OrderedNode *CF) {
+  auto Size = GetSrcSize(Op) * 8;
+  // AF
+  {
+    OrderedNode *AFRes = _Xor(_Xor(Src1, Src2), Res);
+    AFRes = _Bfe(1, 4, AFRes);
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(AFRes);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto PopCountOp = _Popcount(_And(Res, _Constant(0xFF)));
+
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // CF
+  // Unsigned
+  {
+    auto SelectOpLT = _Select(FEXCore::IR::COND_ULT, Res, Src2, _Constant(1), _Constant(0));
+    auto SelectOpLE = _Select(FEXCore::IR::COND_ULE, Res, Src2, _Constant(1), _Constant(0));
+    auto SelectCF   = _Select(FEXCore::IR::COND_EQ, CF, _Constant(1), SelectOpLE, SelectOpLT);
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(SelectCF);
+  }
+
+  // OF
+  // Signed
+  {
+    auto NegOne = _Constant(~0ULL);
+    auto XorOp1 = _Xor(_Xor(Src1, Src2), NegOne);
+    auto XorOp2 = _Xor(Res, Src1);
+    OrderedNode *AndOp1 = _And(XorOp1, XorOp2);
+
+    switch (Size) {
+    case 8:
+      AndOp1 = _Bfe(1, 7, AndOp1);
+    break;
+    case 16:
+      AndOp1 = _Bfe(1, 15, AndOp1);
+    break;
+    case 32:
+      AndOp1 = _Bfe(1, 31, AndOp1);
+    break;
+    case 64:
+      AndOp1 = _Bfe(1, 63, AndOp1);
+    break;
+    default: LOGMAN_MSG_A("Unknown BFESize: %d", Size); break;
+    }
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(AndOp1);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_SBB(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2, OrderedNode *CF) {
+  // AF
+  {
+    OrderedNode *AFRes = _Xor(_Xor(Src1, Src2), Res);
+    AFRes = _Bfe(1, 4, AFRes);
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(AFRes);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto PopCountOp = _Popcount(_And(Res, _Constant(0xFF)));
+
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // CF
+  // Unsigned
+  {
+    auto SelectOpLT = _Select(FEXCore::IR::COND_UGT, Res, Src1, _Constant(1), _Constant(0));
+    auto SelectOpLE = _Select(FEXCore::IR::COND_UGE, Res, Src1, _Constant(1), _Constant(0));
+    auto SelectCF   = _Select(FEXCore::IR::COND_EQ, CF, _Constant(1), SelectOpLE, SelectOpLT);
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(SelectCF);
+  }
+
+  // OF
+  // Signed
+  {
+    auto XorOp1 = _Xor(Src1, Src2);
+    auto XorOp2 = _Xor(Res, Src1);
+    OrderedNode *AndOp1 = _And(XorOp1, XorOp2);
+
+    switch (GetSrcSize(Op)) {
+    case 1:
+      AndOp1 = _Bfe(1, 7, AndOp1);
+    break;
+    case 2:
+      AndOp1 = _Bfe(1, 15, AndOp1);
+    break;
+    case 4:
+      AndOp1 = _Bfe(1, 31, AndOp1);
+    break;
+    case 8:
+      AndOp1 = _Bfe(1, 63, AndOp1);
+    break;
+    default: LOGMAN_MSG_A("Unknown BFESize: %d", GetSrcSize(Op)); break;
+    }
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(AndOp1);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_SUB(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2, bool UpdateCF) {
+  // AF
+  {
+    OrderedNode *AFRes = _Xor(_Xor(Src1, Src2), Res);
+    AFRes = _Bfe(1, 4, AFRes);
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(AFRes);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // ZF
+  {
+    auto ZeroConst = _Constant(0);
+    auto OneConst = _Constant(1);
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, ZeroConst, OneConst, ZeroConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // CF
+  if (UpdateCF) {
+    auto ZeroConst = _Constant(0);
+    auto OneConst = _Constant(1);
+
+    auto SelectOp = _Select(FEXCore::IR::COND_ULT,
+        Src1, Src2, OneConst, ZeroConst);
+
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(SelectOp);
+  }
+  // OF
+  {
+    auto XorOp1 = _Xor(Src1, Src2);
+    auto XorOp2 = _Xor(Res, Src1);
+    OrderedNode *FinalAnd = _And(XorOp1, XorOp2);
+
+    FinalAnd = _Bfe(1, GetSrcSize(Op) * 8 - 1, FinalAnd);
+
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(FinalAnd);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_ADD(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2, bool UpdateCF) {
+  // AF
+  {
+    OrderedNode *AFRes = _Xor(_Xor(Src1, Src2), Res);
+    AFRes = _Bfe(1, 4, AFRes);
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(AFRes);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+  // CF
+  if (UpdateCF) {
+    auto SelectOp = _Select(FEXCore::IR::COND_ULT, Res, Src2, _Constant(1), _Constant(0));
+
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(SelectOp);
+  }
+
+  // OF
+  {
+    auto NegOne = _Constant(~0ULL);
+    auto XorOp1 = _Xor(_Xor(Src1, Src2), NegOne);
+    auto XorOp2 = _Xor(Res, Src1);
+
+    OrderedNode *AndOp1 = _And(XorOp1, XorOp2);
+
+    switch (GetSrcSize(Op)) {
+    case 1:
+      AndOp1 = _Bfe(1, 7, AndOp1);
+    break;
+    case 2:
+      AndOp1 = _Bfe(1, 15, AndOp1);
+    break;
+    case 4:
+      AndOp1 = _Bfe(1, 31, AndOp1);
+    break;
+    case 8:
+      AndOp1 = _Bfe(1, 63, AndOp1);
+    break;
+    default: LOGMAN_MSG_A("Unknown BFESize: %d", GetSrcSize(Op)); break;
+    }
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(AndOp1);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_MUL(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *High) {
+  // PF/AF/ZF/SF
+  // Undefined
+  {
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(_Constant(0));
+  }
+
+  // CF/OF
+  {
+    // CF and OF are set if the result of the operation can't be fit in to the destination register
+    // If the value can fit then the top bits will be zero
+
+    auto SignBit = _Sbfe(1, GetSrcSize(Op) * 8 - 1, Res);
+
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ, High, SignBit, _Constant(0), _Constant(1));
+
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(SelectOp);
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(SelectOp);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_UMUL(FEXCore::X86Tables::DecodedOp Op, OrderedNode *High) {
+  // AF/SF/PF/ZF
+  // Undefined
+  {
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(_Constant(0));
+  }
+
+  // CF/OF
+  {
+    // CF and OF are set if the result of the operation can't be fit in to the destination register
+    // The result register will be all zero if it can't fit due to how multiplication behaves
+
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ, High, _Constant(0), _Constant(0), _Constant(1));
+
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(SelectOp);
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(SelectOp);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_Logical(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2) {
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(_Constant(0));
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // CF/OF
+  {
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(_Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(_Constant(0));
+  }
+}
+
+#define COND_FLAG_SET(cond, flag, newflag) \
+auto oldflag = GetRFLAG(FEXCore::X86State::flag);\
+auto newval = _Select(FEXCore::IR::COND_EQ, cond, _Constant(0), oldflag, newflag);\
+SetRFLAG<FEXCore::X86State::flag>(newval);
+
+void OpDispatchBuilder::GenerateFlags_ShiftLeft(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2) {
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    auto Size = _Constant(GetSrcSize(Op) * 8);
+    auto ShiftAmt = _Sub(Size, Src2);
+    auto LastBit = _And(_Lshr(Src1, ShiftAmt), _Constant(1));
+    COND_FLAG_SET(Src2, RFLAG_CF_LOC, LastBit);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    COND_FLAG_SET(Src2, RFLAG_PF_LOC, XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    COND_FLAG_SET(Src2, RFLAG_AF_LOC, _Constant(0));
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    COND_FLAG_SET(Src2, RFLAG_ZF_LOC, SelectOp);
+  }
+
+  // SF
+  {
+    auto val = _Bfe(1, GetSrcSize(Op) * 8 - 1, Res);
+    COND_FLAG_SET(Src2, RFLAG_SF_LOC, val);
+  }
+
+  // OF
+  {
+    // In the case of left shift. OF is only set from the result of <Top Source Bit> XOR <Top Result Bit>
+    // When Shift > 1 then OF is undefined
+    auto val = _Bfe(1, GetSrcSize(Op) * 8 - 1, _Xor(Src1, Res));
+    COND_FLAG_SET(Src2, RFLAG_OF_LOC, val);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_ShiftRight(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2) {
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    auto ShiftAmt = _Sub(Src2, _Constant(1));
+    auto LastBit = _And(_Lshr(Src1, ShiftAmt), _Constant(1));
+    COND_FLAG_SET(Src2, RFLAG_CF_LOC, LastBit);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    COND_FLAG_SET(Src2, RFLAG_PF_LOC, XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    COND_FLAG_SET(Src2, RFLAG_AF_LOC, _Constant(0));
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    COND_FLAG_SET(Src2, RFLAG_ZF_LOC, SelectOp);
+  }
+
+  // SF
+  {
+    auto val =_Bfe(1, GetSrcSize(Op) * 8 - 1, Res);
+    COND_FLAG_SET(Src2, RFLAG_SF_LOC, val);
+  }
+
+  // OF
+  {
+    // Only defined when Shift is 1 else undefined
+    // OF flag is set if a sign change occurred
+    auto val = _Bfe(1, GetSrcSize(Op) * 8 - 1, _Xor(Src1, Res));
+    COND_FLAG_SET(Src2, RFLAG_OF_LOC, val);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_SignShiftRight(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2) {
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    auto ShiftAmt = _Sub(Src2, _Constant(1));
+    auto LastBit = _And(_Lshr(Src1, ShiftAmt), _Constant(1));
+    COND_FLAG_SET(Src2, RFLAG_CF_LOC, LastBit);
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    COND_FLAG_SET(Src2, RFLAG_PF_LOC, XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    COND_FLAG_SET(Src2, RFLAG_AF_LOC, _Constant(0));
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    COND_FLAG_SET(Src2, RFLAG_ZF_LOC, SelectOp);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    COND_FLAG_SET(Src2, RFLAG_SF_LOC, LshrOp);
+  }
+
+  // OF
+  {
+    COND_FLAG_SET(Src2, RFLAG_OF_LOC, _Constant(0));
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_ShiftLeftImmediate(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, uint64_t Shift) {
+  // No flags changed if shift is zero
+  if (Shift == 0) return;
+
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(_Bfe(1, GetSrcSize(Op) * 8 - Shift, Src1));
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(_Constant(0));
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // SF
+  {
+    auto LshrOp = _Bfe(1, GetSrcSize(Op) * 8 - 1, Res);
+
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+
+    // OF
+    // In the case of left shift. OF is only set from the result of <Top Source Bit> XOR <Top Result Bit>
+    if (Shift == 1) {
+      auto SourceBit = _Bfe(1, GetSrcSize(Op) * 8 - 1, Src1);
+      SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(_Xor(SourceBit, LshrOp));
+    }
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_SignShiftRightImmediate(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, uint64_t Shift) {
+  // No flags changed if shift is zero
+  if (Shift == 0) return;
+
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(_Bfe(1, Shift-1, Src1));
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(_Constant(0));
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+
+    // OF
+    // Only defined when Shift is 1 else undefined
+    // Only is set if the top bit was set to 1 when shifted
+    // So it is set to same value as SF
+    if (Shift == 1) {
+      SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(_Constant(0));
+    }
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_ShiftRightImmediate(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, uint64_t Shift) {
+  // No flags changed if shift is zero
+  if (Shift == 0) return;
+
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(_Bfe(1, Shift-1, Src1));
+  }
+
+  // PF
+  if (!CTX->Config.ABINoPF) {
+    auto EightBitMask = _Constant(0xFF);
+    auto PopCountOp = _Popcount(_And(Res, EightBitMask));
+    auto XorOp = _Xor(PopCountOp, _Constant(1));
+    SetRFLAG<FEXCore::X86State::RFLAG_PF_LOC>(XorOp);
+  } else {
+    _InvalidateFlags(1UL << FEXCore::X86State::RFLAG_PF_LOC);
+  }
+
+  // AF
+  {
+    // Undefined
+    // Set to zero anyway
+    SetRFLAG<FEXCore::X86State::RFLAG_AF_LOC>(_Constant(0));
+  }
+
+  // ZF
+  {
+    auto SelectOp = _Select(FEXCore::IR::COND_EQ,
+        Res, _Constant(0), _Constant(1), _Constant(0));
+    SetRFLAG<FEXCore::X86State::RFLAG_ZF_LOC>(SelectOp);
+  }
+
+  // SF
+  {
+    auto SignBitConst = _Constant(GetSrcSize(Op) * 8 - 1);
+
+    auto LshrOp = _Lshr(Res, SignBitConst);
+    SetRFLAG<FEXCore::X86State::RFLAG_SF_LOC>(LshrOp);
+  }
+
+  // OF
+  {
+    // Only defined when Shift is 1 else undefined
+    // Is set to the MSB of the original value
+    if (Shift == 1) {
+      SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(_Bfe(1, GetSrcSize(Op) * 8 - 1, Src1));
+    }
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_RotateRight(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2) {
+  auto OpSize = GetSrcSize(Op) * 8;
+
+  // Extract the last bit shifted in to CF
+  auto NewCF = _Bfe(1, OpSize - 1, Res);
+
+  // CF
+  {
+    auto OldCF = GetRFLAG(FEXCore::X86State::RFLAG_CF_LOC);
+    auto CF = _Select(FEXCore::IR::COND_EQ, Src2, _Constant(0), OldCF, NewCF);
+
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(CF);
+  }
+
+  // OF
+  {
+    auto OldOF = GetRFLAG(FEXCore::X86State::RFLAG_OF_LOC);
+
+    // OF is set to the XOR of the new CF bit and the most significant bit of the result
+    auto NewOF = _Xor(_Bfe(1, OpSize - 2, Res), NewCF);
+
+    // If shift == 0, don't update flags
+    auto OF = _Select(FEXCore::IR::COND_EQ, Src2, _Constant(0), OldOF, NewOF);
+
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(OF);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_RotateLeft(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, OrderedNode *Src2) {
+  auto OpSize = GetSrcSize(Op) * 8;
+
+  // Extract the last bit shifted in to CF
+  //auto Size = _Constant(GetSrcSize(Res) * 8);
+  //auto ShiftAmt = _Sub(Size, Src2);
+  auto NewCF = _Bfe(1, 0, Res);
+
+  // CF
+  {
+    auto OldCF = GetRFLAG(FEXCore::X86State::RFLAG_CF_LOC);
+    auto CF = _Select(FEXCore::IR::COND_EQ, Src2, _Constant(0), OldCF, NewCF);
+
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(CF);
+  }
+
+  // OF
+  {
+    auto OldOF = GetRFLAG(FEXCore::X86State::RFLAG_OF_LOC);
+    // OF is set to the XOR of the new CF bit and the most significant bit of the result
+    auto NewOF = _Xor(_Bfe(1, OpSize - 1, Res), NewCF);
+
+    auto OF = _Select(FEXCore::IR::COND_EQ, Src2, _Constant(0), OldOF, NewOF);
+
+    // If shift == 0, don't update flags
+    SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(OF);
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_RotateRightImmediate(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, uint64_t Shift) {
+  if (Shift == 0) return;
+
+  auto OpSize = GetSrcSize(Op) * 8;
+
+  auto NewCF = _Bfe(1, OpSize - Shift, Src1);
+
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(NewCF);
+  }
+
+  // OF
+  {
+    if (Shift == 1) {
+      // OF is set to the XOR of the new CF bit and the most significant bit of the result
+      SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(_Xor(_Bfe(1, OpSize - 1, Res), NewCF));
+    }
+  }
+}
+
+void OpDispatchBuilder::GenerateFlags_RotateLeftImmediate(FEXCore::X86Tables::DecodedOp Op, OrderedNode *Res, OrderedNode *Src1, uint64_t Shift) {
+  if (Shift == 0) return;
+
+  auto OpSize = GetSrcSize(Op) * 8;
+
+  // CF
+  {
+    // Extract the last bit shifted in to CF
+    SetRFLAG<FEXCore::X86State::RFLAG_CF_LOC>(_Bfe(1, Shift, Src1));
+  }
+
+  // OF
+  {
+    if (Shift == 1) {
+      // OF is the top two MSBs XOR'd together
+      SetRFLAG<FEXCore::X86State::RFLAG_OF_LOC>(_Xor(_Bfe(1, OpSize - 1, Src1), _Bfe(1, OpSize - 2, Src1)));
+    }
+  }
+}
+
+}

External/FEXCore/Source/Interface/Core/SignalDelegator.cpp

@@ -0,0 +1,113 @@
+#include <FEXCore/Core/SignalDelegator.h>
+#include <FEXCore/Utils/LogManager.h>
+
+#include <unistd.h>
+#include <signal.h>
+
+namespace FEXCore {
+  struct ThreadState {
+    FEXCore::Core::InternalThreadState *Thread{};
+  };
+
+  thread_local ThreadState ThreadData{};
+
+  static bool IsSynchronous(int Signal) {
+    switch (Signal) {
+    case SIGBUS:
+    case SIGFPE:
+    case SIGILL:
+    case SIGSEGV:
+    case SIGTRAP:
+      return true;
+    default: break;
+    };
+    return false;
+  }
+
+  /**
+   * @brief Masks signals from the signal mask
+   *
+   * @param how Argument to sigmask. SIG_{BLOCK, SETMASK, UNBLOCK}
+   * @param Signal Which signal to set or -1 to sweep through them all
+   */
+  static void MaskSignals(int how, int Signal = -1) {
+    // If we have a helper thread, we need to mask a significant amount of signals so the an errant thread doesn't receive a signal that it shouldn't
+    sigset_t SignalSet{};
+    sigemptyset(&SignalSet);
+
+    if (Signal == -1) {
+      for (int i = 0; i <= SignalDelegator::MAX_SIGNALS; ++i) {
+        // If it is a synchronous signal then don't ignore it
+        if (IsSynchronous(i)) {
+          continue;
+        }
+
+        // Add this signal to the ignore list
+        sigaddset(&SignalSet, i);
+      }
+    }
+    else {
+      sigaddset(&SignalSet, Signal);
+    }
+
+    // Be warned, a thread will inherit the signal mask if created from this thread
+    int Result = pthread_sigmask(how, &SignalSet, nullptr);
+    if (Result != 0) {
+      LogMan::Msg::E("Couldn't register thread to mask signals");
+    }
+  }
+
+  void SignalDelegator::MaskThreadSignals() {
+    MaskSignals(SIG_BLOCK);
+  }
+
+  FEXCore::Core::InternalThreadState *SignalDelegator::GetTLSThread() {
+    return ThreadData.Thread;
+  }
+
+  void SignalDelegator::RegisterTLSState(FEXCore::Core::InternalThreadState *Thread) {
+    ThreadData.Thread = Thread;
+    RegisterFrontendTLSState(Thread);
+  }
+
+  void SignalDelegator::UninstallTLSState(FEXCore::Core::InternalThreadState *Thread) {
+    UninstallFrontendTLSState(Thread);
+    ThreadData.Thread = nullptr;
+  }
+
+  void SignalDelegator::RegisterHostSignalHandler(int Signal, HostSignalDelegatorFunction Func, bool Required) {
+    SetHostSignalHandler(Signal, Func, Required);
+    FrontendRegisterHostSignalHandler(Signal, Func, Required);
+  }
+
+  void SignalDelegator::RegisterFrontendHostSignalHandler(int Signal, HostSignalDelegatorFunction Func, bool Required) {
+    SetFrontendHostSignalHandler(Signal, Func, Required);
+    FrontendRegisterFrontendHostSignalHandler(Signal, Func, Required);
+  }
+
+  void SignalDelegator::HandleSignal(int Signal, void *Info, void *UContext) {
+    // Let the host take first stab at handling the signal
+    auto Thread = GetTLSThread();
+    HostSignalHandler &Handler = HostHandlers[Signal];
+
+    if (!Thread) {
+      LogMan::Msg::E("[%d] Thread has received a signal and hasn't registered itself with the delegate! Programming error!", ::gettid());
+    }
+    else {
+      if (Handler.Handler &&
+          Handler.Handler(Thread, Signal, Info, UContext)) {
+        // If the host handler handled the fault then we can continue now
+        return;
+      }
+
+      if (Handler.FrontendHandler &&
+          Handler.FrontendHandler(Thread, Signal, Info, UContext)) {
+        return;
+      }
+
+      // Now let the frontend handle the signal
+      // It's clearly a guest signal and this ends up being an OS specific issue
+      HandleGuestSignal(Thread, Signal, Info, UContext);
+    }
+  }
+}

External/FEXCore/Source/Interface/Core/VSyscall/VSyscall.inc

@@ -0,0 +1,285 @@
+// This is the vsyscall page for x86_64 guest code
+// This was compiled with nasm with the following source then exported to binary
+
+//BITS 64;
+//
+//align 4096, db 0xcc
+//  ; __NR_gettimeofday
+//  mov rax, 96
+//  syscall
+//  ret
+//
+//align 1024, db 0xcc
+//  ; __NR_time
+//  mov rax, 201
+//  syscall
+//  ret
+//
+//align 1024, db 0xcc
+//  ; __NR_getcpu
+//  mov rax, 309
+//  syscall
+//  ret
+//
+//align 4096, db 0xcc
+
+// We only want one of these pages per FEX process
+// One page
+const static uint8_t VSyscallData[0x1000] = {
+	0xB8, 0x60, 0x00, 0x00, 0x00, 0x0F, 0x05, 0xC3, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xB8, 0xC9, 0x00, 0x00, 0x00, 0x0F, 0x05, 0xC3, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xB8, 0x35, 0x01, 0x00, 0x00, 0x0F, 0x05, 0xC3, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC,
+	0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC, 0xCC
+};

External/FEXCore/Source/Interface/Core/X86DebugInfo.cpp

@@ -114,8 +114,6 @@ void InstallDebugInfo() {
   GenerateDebugTable(PrimaryInstGroupOps, PrimaryGroupOpTable);
 
   GenerateDebugTable(SecondInstGroupOps, SecondaryExtensionOpTable);
-
-  LogMan::Msg::D("Installing debug info");
 }
 }
 #endif

External/FEXCore/Source/Interface/Core/X86HelperGen.cpp

@@ -6,12 +6,15 @@ $end_info$
 */
 
 #include "Interface/Core/X86HelperGen.h"
+
+#include <FEXCore/Config/Config.h>
 #include <FEXCore/Utils/Allocator.h>
 
+#include <cstdint>
 #include <cstring>
-#include <stdlib.h>
 #include <vector>
 #include <sys/mman.h>
+#include <bits/mman-map-flags-generic.h>
 
 namespace FEXCore {
 constexpr size_t CODE_SIZE = 0x1000;
@@ -50,14 +53,14 @@ void* X86GeneratedCode::AllocateGuestCodeSpace(size_t Size) {
   // We need to have the sigret handler in the lower 32bits of memory space
   // Scan top down and try to allocate a location
   for (size_t Location = 0xFFFF'E000; Location != 0x0; Location -= 0x1000) {
-    void *Ptr = FEXCore::Allocator::mmap(reinterpret_cast<void*>(Location), Size, PROT_READ | PROT_WRITE, MAP_FIXED_NOREPLACE | MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+    void *Ptr = ::mmap(reinterpret_cast<void*>(Location), Size, PROT_READ | PROT_WRITE, MAP_FIXED_NOREPLACE | MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
 
     if (Ptr != MAP_FAILED &&
         reinterpret_cast<uintptr_t>(Ptr) >= LOCATION_MAX) {
       // Failed to map in the lower 32bits
       // Try again
       // Can happen in the case that host kernel ignores MAP_FIXED_NOREPLACE
-      FEXCore::Allocator::munmap(Ptr, Size);
+      ::munmap(Ptr, Size);
       continue;
     }
 

External/FEXCore/Source/Interface/Core/X86HelperGen.h

@@ -5,8 +5,8 @@ $end_info$
 */
 
 #pragma once
-#include <FEXCore/Config/Config.h>
 
+#include <stddef.h>
 #include <stdint.h>
 
 namespace FEXCore {

External/FEXCore/Source/Interface/Core/X86Tables.cpp

@@ -5,14 +5,8 @@ tags: frontend|x86-tables
 $end_info$
 */
 
-#include <FEXCore/Utils/LogManager.h>
-
 #include <FEXCore/Core/Context.h>
 #include <FEXCore/Debug/X86Tables.h>
-#include <array>
-#include <cstdint>
-#include <tuple>
-#include <vector>
 
 namespace FEXCore::X86Tables {
 
@@ -108,7 +102,6 @@ void InitializeInfoTables(Context::OperatingMode Mode) {
 
 #ifndef NDEBUG
   X86InstDebugInfo::InstallDebugInfo();
-  LogMan::Msg::D("X86Tables had %ld total insts, and %ld labeled as understood", Total, NumInsts);
 #endif
 }
 

External/FEXCore/Source/Interface/Core/X86Tables/BaseTables.cpp

@@ -7,6 +7,9 @@ $end_info$
 #include "Interface/Core/X86Tables/X86Tables.h"
 
 #include <FEXCore/Core/Context.h>
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
 
 namespace FEXCore::X86Tables {
 using namespace InstFlags;

External/FEXCore/Source/Interface/Core/X86Tables/DDDTables.cpp

@@ -6,6 +6,10 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/EVEXTables.cpp

@@ -6,6 +6,10 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/H0F38Tables.cpp

@@ -5,6 +5,10 @@ $end_info$
 */
 
 #include "Interface/Core/X86Tables/X86Tables.h"
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+#include <stdint.h>
 
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
@@ -41,10 +45,10 @@ void InitializeH0F38Tables() {
     {OPD(PF_38_NONE, 0x0B), 1, X86InstInfo{"PMULHRSW",   TYPE_INST, GenFlagsSameSize(SIZE_64BIT)  | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_MMX, 0, nullptr}},
     {OPD(PF_38_66,   0x0B), 1, X86InstInfo{"PMULHRSW",   TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
 
-    {OPD(PF_38_66,   0x10), 1, X86InstInfo{"PBLENDVB",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x14), 1, X86InstInfo{"BLENDVPS",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x15), 1, X86InstInfo{"BLENDVPD",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x17), 1, X86InstInfo{"PTEST",      TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(PF_38_66,   0x10), 1, X86InstInfo{"PBLENDVB",   TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x14), 1, X86InstInfo{"BLENDVPS",   TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x15), 1, X86InstInfo{"BLENDVPD",   TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x17), 1, X86InstInfo{"PTEST",      TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
     {OPD(PF_38_NONE, 0x1C), 1, X86InstInfo{"PABSB",      TYPE_INST, GenFlagsSameSize(SIZE_64BIT)  | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_MMX, 0, nullptr}},
     {OPD(PF_38_66,   0x1C), 1, X86InstInfo{"PABSB",      TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
     {OPD(PF_38_NONE, 0x1D), 1, X86InstInfo{"PABSW",      TYPE_INST, GenFlagsSameSize(SIZE_64BIT)  | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_MMX, 0, nullptr}},
@@ -52,34 +56,34 @@ void InitializeH0F38Tables() {
     {OPD(PF_38_NONE, 0x1E), 1, X86InstInfo{"PABSD",      TYPE_INST, GenFlagsSameSize(SIZE_64BIT)  | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_MMX, 0, nullptr}},
     {OPD(PF_38_66,   0x1E), 1, X86InstInfo{"PABSD",      TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
 
-    {OPD(PF_38_66,   0x20), 1, X86InstInfo{"PMOVSXBW",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x21), 1, X86InstInfo{"PMOVSXBD",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x22), 1, X86InstInfo{"PMOVSXBQ",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x23), 1, X86InstInfo{"PMOVSXWD",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x24), 1, X86InstInfo{"PMOVSXWQ",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x25), 1, X86InstInfo{"PMOVSXDQ",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x28), 1, X86InstInfo{"PMULDQ",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x29), 1, X86InstInfo{"PCMPEQQ",    TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(PF_38_66,   0x20), 1, X86InstInfo{"PMOVSXBW",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_64BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x21), 1, X86InstInfo{"PMOVSXBD",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_32BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x22), 1, X86InstInfo{"PMOVSXBQ",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_16BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x23), 1, X86InstInfo{"PMOVSXWD",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_64BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x24), 1, X86InstInfo{"PMOVSXWQ",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_32BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x25), 1, X86InstInfo{"PMOVSXDQ",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_64BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x28), 1, X86InstInfo{"PMULDQ",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x29), 1, X86InstInfo{"PCMPEQQ",    TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
     {OPD(PF_38_66,   0x2A), 1, X86InstInfo{"MOVNTDQA",   TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_SF_MOD_MEM_ONLY | FLAGS_XMM_FLAGS, 0, nullptr}},
-    {OPD(PF_38_66,   0x2B), 1, X86InstInfo{"PACKUSDW",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(PF_38_66,   0x2B), 1, X86InstInfo{"PACKUSDW",   TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
 
-    {OPD(PF_38_66,   0x30), 1, X86InstInfo{"PMOVZXBW",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x31), 1, X86InstInfo{"PMOVZXBD",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x32), 1, X86InstInfo{"PMOVZXBQ",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x33), 1, X86InstInfo{"PMOVZXWD",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x34), 1, X86InstInfo{"PMOVZXWQ",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x35), 1, X86InstInfo{"PMOVZXDQ",   TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x38), 1, X86InstInfo{"PMINSB",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x39), 1, X86InstInfo{"PMINSD",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x3A), 1, X86InstInfo{"PMINUW",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x3B), 1, X86InstInfo{"PMINUD",     TYPE_INST, FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
-    {OPD(PF_38_66,   0x3C), 1, X86InstInfo{"PMAXSB",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x3D), 1, X86InstInfo{"PMAXSD",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x3E), 1, X86InstInfo{"PMAXUW",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x3F), 1, X86InstInfo{"PMAXUD",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(PF_38_66,   0x30), 1, X86InstInfo{"PMOVZXBW",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_64BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x31), 1, X86InstInfo{"PMOVZXBD",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_32BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x32), 1, X86InstInfo{"PMOVZXBQ",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_16BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x33), 1, X86InstInfo{"PMOVZXWD",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_64BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x34), 1, X86InstInfo{"PMOVZXWQ",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_32BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x35), 1, X86InstInfo{"PMOVZXDQ",   TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_64BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x38), 1, X86InstInfo{"PMINSB",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x39), 1, X86InstInfo{"PMINSD",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x3A), 1, X86InstInfo{"PMINUW",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x3B), 1, X86InstInfo{"PMINUD",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x3C), 1, X86InstInfo{"PMAXSB",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x3D), 1, X86InstInfo{"PMAXSD",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x3E), 1, X86InstInfo{"PMAXUW",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x3F), 1, X86InstInfo{"PMAXUD",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
 
-    {OPD(PF_38_66,   0x40), 1, X86InstInfo{"PMULLD",     TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(PF_38_66,   0x41), 1, X86InstInfo{"PHMINPOSUW", TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(PF_38_66,   0x40), 1, X86InstInfo{"PMULLD",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
+    {OPD(PF_38_66,   0x41), 1, X86InstInfo{"PHMINPOSUW", TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
 
     {OPD(PF_38_66,   0xDB), 1, X86InstInfo{"AESIMC",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},
     {OPD(PF_38_66,   0xDC), 1, X86InstInfo{"AESENC",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 0, nullptr}},

External/FEXCore/Source/Interface/Core/X86Tables/H0F3ATables.cpp

@@ -6,6 +6,12 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/Core/Context.h>
+
+#include <iterator>
+#include <stdint.h>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 
@@ -16,26 +22,26 @@ void InitializeH0F3ATables(Context::OperatingMode Mode) {
 
   const U16U8InfoStruct H0F3ATable[] = {
     {OPD(0, PF_3A_NONE, 0x0F), 1, X86InstInfo{"PALIGNR",         TYPE_INST, GenFlagsSameSize(SIZE_64BIT)  | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_MMX, 1, nullptr}},
-    {OPD(0, PF_3A_66,   0x08), 1, X86InstInfo{"ROUNDPS",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x09), 1, X86InstInfo{"ROUNDPD",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x0A), 1, X86InstInfo{"ROUNDSS",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x0B), 1, X86InstInfo{"ROUNDSD",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x0C), 1, X86InstInfo{"BLENDPS",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x0D), 1, X86InstInfo{"BLENDPD",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x0E), 1, X86InstInfo{"PBLENDW",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(0, PF_3A_66,   0x08), 1, X86InstInfo{"ROUNDPS",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
+    {OPD(0, PF_3A_66,   0x09), 1, X86InstInfo{"ROUNDPD",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
+    {OPD(0, PF_3A_66,   0x0A), 1, X86InstInfo{"ROUNDSS",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
+    {OPD(0, PF_3A_66,   0x0B), 1, X86InstInfo{"ROUNDSD",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
+    {OPD(0, PF_3A_66,   0x0C), 1, X86InstInfo{"BLENDPS",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS,           1, nullptr}},
+    {OPD(0, PF_3A_66,   0x0D), 1, X86InstInfo{"BLENDPD",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS,           1, nullptr}},
+    {OPD(0, PF_3A_66,   0x0E), 1, X86InstInfo{"PBLENDW",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS,           1, nullptr}},
     {OPD(0, PF_3A_66,   0x0F), 1, X86InstInfo{"PALIGNR",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
 
     {OPD(0, PF_3A_66,   0x14), 1, X86InstInfo{"PEXTRB",          TYPE_INST, GenFlagsSizes(SIZE_8BIT, SIZE_128BIT) | FLAGS_MODRM | FLAGS_SF_MOD_DST | FLAGS_SF_DST_GPR | FLAGS_XMM_FLAGS, 1, nullptr}},
     {OPD(0, PF_3A_66,   0x15), 1, X86InstInfo{"PEXTRW",          TYPE_INST, GenFlagsSizes(SIZE_16BIT, SIZE_128BIT) | FLAGS_MODRM | FLAGS_SF_MOD_DST | FLAGS_SF_DST_GPR | FLAGS_XMM_FLAGS, 1, nullptr}},
     {OPD(0, PF_3A_66,   0x16), 1, X86InstInfo{"PEXTRD",          TYPE_INST, GenFlagsSizes(SIZE_32BIT, SIZE_128BIT) | FLAGS_MODRM | FLAGS_SF_MOD_DST | FLAGS_SF_DST_GPR | FLAGS_XMM_FLAGS, 1, nullptr}},
-    {OPD(0, PF_3A_66,   0x17), 1, X86InstInfo{"EXTRACTPS",       TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(0, PF_3A_66,   0x17), 1, X86InstInfo{"EXTRACTPS",       TYPE_INST, GenFlagsSizes(SIZE_32BIT, SIZE_128BIT) | FLAGS_MODRM | FLAGS_SF_MOD_DST | FLAGS_SF_DST_GPR | FLAGS_XMM_FLAGS, 1, nullptr}},
 
-    {OPD(0, PF_3A_66,   0x20), 1, X86InstInfo{"PINSRB",          TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_8BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_SRC_GPR,           1, nullptr}},
-    {OPD(0, PF_3A_66,   0x21), 1, X86InstInfo{"INSERTPS",        TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(0, PF_3A_66,   0x20), 1, X86InstInfo{"PINSRB",          TYPE_INST, GenFlagsDstSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_SRC_GPR,           1, nullptr}},
+    {OPD(0, PF_3A_66,   0x21), 1, X86InstInfo{"INSERTPS",        TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS,           1, nullptr}},
     {OPD(0, PF_3A_66,   0x22), 1, X86InstInfo{"PINSRD",          TYPE_INST, GenFlagsSizes(SIZE_128BIT, SIZE_32BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS | FLAGS_SF_SRC_GPR,           1, nullptr}},
-    {OPD(0, PF_3A_66,   0x40), 1, X86InstInfo{"DPPS",            TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x41), 1, X86InstInfo{"DPPD",            TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
-    {OPD(0, PF_3A_66,   0x42), 1, X86InstInfo{"MPSADBW",         TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
+    {OPD(0, PF_3A_66,   0x40), 1, X86InstInfo{"DPPS",            TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
+    {OPD(0, PF_3A_66,   0x41), 1, X86InstInfo{"DPPD",            TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
+    {OPD(0, PF_3A_66,   0x42), 1, X86InstInfo{"MPSADBW",         TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS, 1, nullptr}},
     {OPD(0, PF_3A_66,   0x44), 1, X86InstInfo{"PCLMULQDQ",       TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},
 
     {OPD(0, PF_3A_66,   0x60), 1, X86InstInfo{"PCMPESTRM",       TYPE_UNDEC, FLAGS_NONE, 0, nullptr}},

External/FEXCore/Source/Interface/Core/X86Tables/PrimaryGroupTables.cpp

@@ -6,6 +6,11 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Core/Context.h>
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/SecondaryGroupTables.cpp

@@ -6,6 +6,11 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+#include <stdint.h>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/SecondaryModRMTables.cpp

@@ -6,6 +6,10 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/SecondaryTables.cpp

@@ -4,7 +4,12 @@ tags: frontend|x86-tables
 $end_info$
 */
 
- #include "Interface/Core/X86Tables/X86Tables.h"
+#include "Interface/Core/X86Tables/X86Tables.h"
+
+#include <FEXCore/Core/Context.h>
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
 
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
@@ -27,7 +32,7 @@ void InitializeSecondaryTables(Context::OperatingMode Mode) {
     {0x0B, 1, X86InstInfo{"UD2",        TYPE_INST, FLAGS_DEBUG | FLAGS_BLOCK_END | FLAGS_NO_OVERLAY,                                                    0, nullptr}},
     {0x0C, 1, X86InstInfo{"",           TYPE_INVALID, FLAGS_NO_OVERLAY,                                                                                 0, nullptr}},
     {0x0D, 1, X86InstInfo{"",           TYPE_GROUP_P, FLAGS_MODRM | FLAGS_NO_OVERLAY,                                                                                 0, nullptr}},
-    {0x0E, 1, X86InstInfo{"FEMMS",      TYPE_INST, FLAGS_BLOCK_END | FLAGS_NO_OVERLAY,                                                            0, nullptr}},
+    {0x0E, 1, X86InstInfo{"FEMMS",      TYPE_INST, FLAGS_NO_OVERLAY,                                                            0, nullptr}},
     {0x0F, 1, X86InstInfo{"",           TYPE_3DNOW_TABLE, FLAGS_NO_OVERLAY,                                                                             0, nullptr}},
 
     {0x10, 1, X86InstInfo{"MOVUPS",     TYPE_INST, GenFlagsSameSize(SIZE_128BIT) | FLAGS_MODRM | FLAGS_XMM_FLAGS,                                                 0, nullptr}},

External/FEXCore/Source/Interface/Core/X86Tables/VEXTables.cpp

@@ -6,6 +6,9 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+#include <iterator>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/X86Tables.h

@@ -6,6 +6,7 @@ $end_info$
 
 #pragma once
 #include <FEXCore/Debug/X86Tables.h>
+#include <FEXCore/Core/Context.h>
 
 #include <FEXCore/Utils/LogManager.h>
 
@@ -113,5 +114,7 @@ static inline void GenerateX87Table(X86InstInfo *FinalTable, U16U8InfoStruct con
   }
 };
 
+void InitializeInfoTables(Context::OperatingMode Mode);
+
 }
 

External/FEXCore/Source/Interface/Core/X86Tables/X87Tables.cpp

@@ -6,6 +6,9 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+#include <iterator>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/Core/X86Tables/XOPTables.cpp

@@ -6,6 +6,11 @@ $end_info$
 
 #include "Interface/Core/X86Tables/X86Tables.h"
 
+#include <FEXCore/Debug/X86Tables.h>
+
+#include <iterator>
+#include <stdint.h>
+
 namespace FEXCore::X86Tables {
 using namespace InstFlags;
 

External/FEXCore/Source/Interface/HLE/Thunks/Thunks.cpp

@@ -5,21 +5,23 @@ tags: glue|thunks
 $end_info$
 */
 
+#include <FEXCore/Config/Config.h>
+#include <FEXCore/Core/CoreState.h>
+#include <FEXCore/Debug/InternalThreadState.h>
 #include <FEXCore/Utils/LogManager.h>
+#include <FEXCore/IR/IR.h>
 #include "Thunks.h"
 
-#include "stdio.h"
 #include <dlfcn.h>
 
-
-#include <string>
-#include <map>
-#include <array>
 #include <Interface/Context/Context.h>
-#include "Interface/Core/InternalThreadState.h"
 #include "FEXCore/Core/X86Enums.h"
-#include <mutex>
+#include <map>
+#include <memory>
 #include <shared_mutex>
+#include <stdint.h>
+#include <string>
+#include <utility>
 
 struct LoadlibArgs {
     const char *Name;
@@ -51,7 +53,7 @@ namespace FEXCore {
             Thread->CurrentFrame->State.gregs[FEXCore::X86State::REG_RDI] = (uintptr_t)arg0;
             Thread->CurrentFrame->State.gregs[FEXCore::X86State::REG_RSI] = (uintptr_t)arg1;
 
-            Thread->CTX->HandleCallback((uintptr_t)callback);
+            Thread->CTX->HandleCallback(Thread, (uintptr_t)callback);
         }
 
         static void LoadLib(void *ArgsV) {

External/FEXCore/Source/Interface/HLE/Thunks/Thunks.h

@@ -5,12 +5,15 @@ $end_info$
 */
 
 #pragma once
-#include <FEXCore/IR/IR.h>
 
 namespace FEXCore::Core {
   struct InternalThreadState;
 }
 
+namespace FEXCore::IR {
+  struct SHA256Sum;
+}
+
 namespace FEXCore {
     typedef void ThunkedFunction(void* ArgsRv);
 
@@ -22,4 +25,4 @@ namespace FEXCore {
 
         static ThunkHandler* Create();
     };
-};
+};