Bumped version to v0.2.4

Added some opcodes

.github/workflows/manual.yml

@@ -19,6 +19,7 @@ on:
         - RPI4ARM64
         - SD845
         - TEGRAX1
+        - ARM64
       build_type:
         description: 'Build type'
         required: true

.github/workflows/release.yml

@@ -27,7 +27,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        platform: [X64, OTHER_ARM, RISCV, RPI4ARM64, RK3326, RK3399, RK3588, PHYTIUM, SD845]
+        platform: [X64, OTHER_ARM, RISCV, RPI4ARM64, RK3326, RK3399, RK3588, PHYTIUM, SD845, ARM64]
         type: [Release, Trace]
         os: [ubuntu-latest]
         include:

CMakeLists.txt

@@ -14,11 +14,13 @@ option(ODROIDN2 "Set to ON if targeting an Odroid-N2 device" ${ODROIDN2})
 option(TEGRAX1 "Set to ON if targeting an Tegra X1 based device" ${TEGRAX1})
 option(PHYTIUM "Set to ON if targeting an Phytium (D2000 or FT2000/4) based device" ${PHYTIUM})
 option(SD845 "Set to ON if targeting a Snapragon 845 based device" ${SD845})
+option(SD888 "Set to ON if targeting a Snapragon 888 based device" ${SD888})
 option(M1 "Set to ON if targeting a AppleM1 running on Asahi computer" ${M1})
 option(LARCH64 "Set to ON if targeting an Loongarch64 based device" ${LARCH64})
 option(RV64 "Set to ON if targeting an RISC-V RV64GC based device" ${RV64})
 option(PPC64LE "Set to ON if targeting an PowerPC 64 LE based device" ${PPC64LE})
 option(LX2160A "Set to ON if targeting an LX2160A based device" ${LX2160A})
+option(ARM64 "Set to ON if targeting a generic ARM64 based device" ${ARM64})
 option(USE_CCACHE "Set to ON to use ccache if present in the system" ${USE_CCACHE})
 option(HAVE_TRACE "Set to ON to have Trace ability (needs ZydisInfo library)" ${HAVE_TRACE})
 option(NOLOADADDR "Set to ON to avoid fixing the load address of Box64" OFF)
@@ -46,13 +48,13 @@ if(PPC64LE)
     set(RV64_DYNAREC OFF CACHE BOOL "")
     set(PAGE16K OFF CACHE BOOL "")
 endif()
-if(RK3399 OR RK3588 OR ODROIDN2 OR RPI3ARM64 OR RPI4ARM64 OR RK3326 OR TEGRAX1 OR PHYTIUM OR SD845 OR LX2160A OR M1)
+if(RK3399 OR RK3588 OR ODROIDN2 OR RPI3ARM64 OR RPI4ARM64 OR RK3326 OR TEGRAX1 OR PHYTIUM OR SD845 OR SD888 OR LX2160A OR M1 OR ARM64)
     set(LD80BITS OFF CACHE BOOL "")
     set(NOALIGN OFF CACHE BOOL "")
     set(ARM_DYNAREC ON CACHE BOOL "")
     set(RV64_DYNAREC OFF CACHE BOOL "")
 endif()
-if(RK3399 OR RK3588 OR ODROIDN2 OR RPI3ARM64 OR RPI4ARM64 OR RK3326 OR TEGRAX1 OR PHYTIUM OR SD845 OR LX2160A)
+if(RK3399 OR RK3588 OR ODROIDN2 OR RPI3ARM64 OR RPI4ARM64 OR RK3326 OR TEGRAX1 OR PHYTIUM OR SD845 OR SD888 OR LX2160A)
     set(PAGE16K OFF CACHE BOOL "")
 endif()
 if(RK3588)
@@ -137,13 +139,17 @@ elseif(TEGRAX1)
     add_definitions(-pipe -march=armv8-a+crc+simd+crypto -mcpu=cortex-a57+crypto)
     set(CMAKE_ASM_FLAGS  "-pipe -march=armv8-a+crc+simd+crypto -mcpu=cortex-a57+crypto")
 elseif(PHYTIUM)
-    add_definitions(-DPHYSIUM)
+    add_definitions(-DPHYTIUM)
     add_definitions(-pipe -march=armv8.1-a+crc+simd+crypto)
     set(CMAKE_ASM_FLAGS  "-pipe -march=armv8.1-a+crc+simd+crypto+aes")
 elseif(SD845)
     add_definitions(-DSD845)
     add_definitions(-pipe -march=armv8.2-a+simd+crypto -mtune=cortex-a75.cortex-a55)
     set(CMAKE_ASM_FLAGS  "-pipe -march=armv8.2-a+simd+crypto -mtune=cortex-a75.cortex-a55")
+elseif(SD888)
+    add_definitions(-DSD888)
+    add_definitions(-pipe -march=armv8.4-a+crypto)
+    set(CMAKE_ASM_FLAGS  "-pipe -march=armv8.4-a+crypto")
 elseif(M1)
     add_definitions(-DM1)
     add_definitions(-pipe -march=armv8.5-a+simd+crypto)
@@ -164,6 +170,9 @@ elseif(LX2160A)
     set(CMAKE_ASM_FLAGS  "-pipe -march=armv8-a+crypto+crc -mcpu=cortex-a72+crypto")
 elseif(SW64)
     add_definitions(-DSW64)
+elseif(ARM64)
+    #add_definitions(-pipe -march=native)
+    set(CMAKE_ASM_FLAGS  "-pipe -march=armv8-a+simd")
 elseif(ARM_DYNAREC)
     set(CMAKE_ASM_FLAGS  "-pipe -march=armv8-a+crc+simd+crypto")
 endif()
@@ -309,6 +318,7 @@ set(INTERPRETER
     "${BOX64_ROOT}/src/emu/x64run66f0.c"
     "${BOX64_ROOT}/src/emu/x64run67.c"
     "${BOX64_ROOT}/src/emu/x64run67_32.c"
+    "${BOX64_ROOT}/src/emu/x64run6764_32.c"
     "${BOX64_ROOT}/src/emu/x64run670f.c"
     "${BOX64_ROOT}/src/emu/x64run6766.c"
     "${BOX64_ROOT}/src/emu/x64run67660f.c"
@@ -333,6 +343,7 @@ set(WRAPPEDS
     "${BOX64_ROOT}/src/wrapped/wrappedatomic.c"
     "${BOX64_ROOT}/src/wrapped/wrappedatspi.c"
     "${BOX64_ROOT}/src/wrapped/wrappedbz2.c"
+    "${BOX64_ROOT}/src/wrapped/wrappedcap.c"
     "${BOX64_ROOT}/src/wrapped/wrappedcairo.c"
     "${BOX64_ROOT}/src/wrapped/wrappedcairogobject.c"
     "${BOX64_ROOT}/src/wrapped/wrappedcrashhandler.c"
@@ -359,6 +370,7 @@ set(WRAPPEDS
     "${BOX64_ROOT}/src/wrapped/wrappedgmp.c"
     "${BOX64_ROOT}/src/wrapped/wrappedgnutls.c"
     "${BOX64_ROOT}/src/wrapped/wrappedgobject2.c"
+    "${BOX64_ROOT}/src/wrapped/wrappedgomp.c"
     "${BOX64_ROOT}/src/wrapped/wrappedgssapi.c"
     "${BOX64_ROOT}/src/wrapped/wrappedgssapikrb5.c"
     "${BOX64_ROOT}/src/wrapped/wrappedgstapp.c"
@@ -739,10 +751,10 @@ else()
     endif()
 endif()
 
-string(COMPARE EQUAL "${CMAKE_HOST_SYSTEM_PROCESSOR}" "i686"  _x86)
-string(COMPARE EQUAL "${CMAKE_HOST_SYSTEM_PROCESSOR}" "x86_64"  _x86_64)
-string(COMPARE EQUAL "${CMAKE_HOST_SYSTEM_PROCESSOR}" "aarch64"  _aarch64)
-string(COMPARE EQUAL "${CMAKE_HOST_SYSTEM_PROCESSOR}" "riscv64"  _riscv64)
+string(COMPARE EQUAL "${CMAKE_SYSTEM_PROCESSOR}" "i686"  _x86)
+string(COMPARE EQUAL "${CMAKE_SYSTEM_PROCESSOR}" "x86_64"  _x86_64)
+string(COMPARE EQUAL "${CMAKE_SYSTEM_PROCESSOR}" "aarch64"  _aarch64)
+string(COMPARE EQUAL "${CMAKE_SYSTEM_PROCESSOR}" "riscv64"  _riscv64)
 
 if(_x86_64 OR _aarch64)
   add_definitions(-DCONFIG_64BIT)

README.md

@@ -2,7 +2,7 @@
 
 Linux Userspace x86_64 Emulator with a twist
 
-[View changelog](https://github.com/ptitSeb/box64/blob/main/docs/CHANGELOG.md) | [中文](https://github.com/ptitSeb/box64/blob/main/README_CN.md) | [Report an error](https://github.com/ptitSeb/box64/issues/new)
+[View changelog](https://github.com/ptitSeb/box64/blob/main/docs/CHANGELOG.md) | [中文](https://github.com/ptitSeb/box64/blob/main/README_CN.md) | [Українська](https://github.com/ptitSeb/box64/blob/main/README_UK.md) | [Report an error](https://github.com/ptitSeb/box64/issues/new)
 
 ![build](https://app.travis-ci.com/ptitSeb/box64.svg?branch=main) ![stars](https://img.shields.io/github/stars/ptitSeb/box64) ![forks](https://img.shields.io/github/forks/ptitSeb/box64) ![contributors](https://img.shields.io/github/contributors/ptitSeb/box64) ![prs](https://img.shields.io/github/issues-pr/ptitSeb/box64) ![issues](https://img.shields.io/github/issues/ptitSeb/box64)
 
@@ -86,23 +86,22 @@ GTK libraries are now wrapped on box64, both gtk2 and gtk3.
 Notes about Steam
 ----
 
-Note that Steam is a hybrid 32-bit / 64-bit. You NEED box86 to run Steam, as the client app is a 32-bit binary. It also uses a 64-bit local server binaries, but like most stuff using libcef/chromium, it's not working correctly on box64 for now.
-So, no Steam for now on box64.
+Note that Steam is a hybrid 32-bit / 64-bit. You NEED box86 to run Steam, as the client app is a 32-bit binary. It also uses a 64-bit local server binaries, and that steamwebhelper process is now mendatory, even on the "small mode". And that process will eat lots of memory. So machine with less the 6Gb of RAM will need a swapfile tp use Steam.
 
 ----
 
 Notes about Wine
 ----
 
-Wine64 is supported on box64. Proton should also work. Be aware that 64-bit Wine also includes 32-bit components, to be able to run 32-bit Windows programs. The 32-bit apps will need box86 and will not run without it. On a system where both box64 and box86 are present and working, a wine 64-bit setup can run both 32-bit and 64-bit Windows programs (just use `wine` and `wine64` respectively).
-Note that the new 32bits PE in 64bits process that the Wine time in currently implementing in Wine 7.+ is not yet supported. I tested Wine 7.5 64bits and it worked, but more recent version will probably just not work yet.
+Wine64 is supported on box64. Proton too. Be aware that 64-bit Wine also includes 32-bit components, to be able to run 32-bit Windows programs. The 32-bit apps will need box86 and will not run without it. On a system where both box64 and box86 are present and working, a wine 64-bit setup can run both 32-bit and 64-bit Windows programs (just use `wine` and `wine64` respectively).
+Note that the new 32bits PE in 64bits process that the Wine time in currently implementing in Wine 7.+ is now supported, but the support is quite young so there might be some residual issues.
 
 ----
 
 Notes about Vulkan
 ----
 
-Box64 wraps Vulkan libraries, but note that it as only been tested with a RX550 card, so some extensions may be missing depending on your graphics card.
+Box64 wraps Vulkan libraries, but note that it as mostly been tested with a AMD RX550 card and on the Freedreno driver, so some extensions may be missing depending on your graphics card.
 
 ----
 

README_CN.md

@@ -6,7 +6,7 @@ Linux 上用户空间的 x86_64 应用模拟器（支持运行原生库）
 ----
 ![box64 build status](https://app.travis-ci.com/ptitSeb/box64.svg?branch=main)
 
-[English](docs/README.md)
+[English](docs/README.md) | [Українська](https://github.com/ptitSeb/box64/blob/main/README_UK.md) 
 
 Box64 可以在非 x86_64 Linux 系统（比如 ARM64）上运行 x86_64 Linux 程序（比如游戏），注意主机系统需要是 64 位小端。
 

README_UK.md

@@ -0,0 +1,119 @@
+![Official logo](docs/img/Box64Logo.png "Official Logo")
+
+Емулятор простіра користувача x86_64 із родзинкою
+
+[Дивитись журнал змін](https://github.com/ptitSeb/box64/blob/main/docs/CHANGELOG.md) | [中文](https://github.com/ptitSeb/box64/blob/main/README_CN.md) | [Українська](https://github.com/ptitSeb/box64/blob/main/README_UK.md) | [Повідомити про помилку](https://github.com/ptitSeb/box64/issues/new)
+
+![build](https://app.travis-ci.com/ptitSeb/box64.svg?branch=main) ![stars](https://img.shields.io/github/stars/ptitSeb/box64) ![forks](https://img.shields.io/github/forks/ptitSeb/box64) ![contributors](https://img.shields.io/github/contributors/ptitSeb/box64) ![prs](https://img.shields.io/github/issues-pr/ptitSeb/box64) ![issues](https://img.shields.io/github/issues/ptitSeb/box64)
+
+----
+
+Box64 дозволяє запускати x86_64 Linux програми (наприклад, ігри) на системах Linux відмінних від x86_64, наприклад ARM (хост-система має мати 64-розрядний порядок редагування).
+
+Ви можете знайти багато відео про Box64 в YouTube каналах [MicroLinux](https://www.youtube.com/channel/UCwFQAEj1lp3out4n7BeBatQ), [Pi Labs](https://www.youtube.com/channel/UCgfQjdc5RceRlTGfuthBs7g) або [The Byteman](https://www.youtube.com/channel/UCEr8lpIJ3B5Ctc5BvcOHSnA)
+
+Оскільки Box64 використовує рідні версії деяких «системних» бібліотек, таких як libc, libm, SDL і OpenGL, його легко інтегрувати та використовувати з більшістю програм, а продуктивність у багатьох випадках може бути напрочуд високою. Подивіться на цей стендовий аналіз для прикладу [тут](https://box86.org/index.php/2021/06/game-performances/).
+
+Box64 інтегрується з DynaRec (динамічний рекомпілятор) для платформ ARM64 і RV64 забезпечуючи підвищення швидкості в 5-10 разів, ніж використання лише інтерпретатора. Деякі високорівневі відомості про те, як працює DynaRec, можна знайти [тут](https://box86.org/2021/07/inner-workings-a-high%e2%80%91level-view-of-box86-and-a-low%e2%80%91level-view-of-the-dynarec/).
+
+Деякі внутрішні коди операцій x64 використовують частини «Бібліотеки емулятора Realmode X86», дивіться [x64primop.c](src/emu/x64primop.c) для детальної інформації про авторські права.
+
+<img src="docs/img/Box64Icon.png" width="96" height="96">
+
+Логотип і піктограма зроблені @grayduck, дякую!
+
+----
+
+Використання
+----
+
+Є кілька змінних середовища для керування поведінкою Box64.
+
+Дивіться [тут](docs/USAGE.md) для перегляду всіх змінних середовища та того, що вони роблять.
+
+Примітка: Dynarec від Box64 використовує механізм із захистом пам’яті та обробником сигналу помилки сегментів для обробки JIT коду. Простіше кажучи, якщо ви хочете використовувати GDB для налагодження запущеної програми, яка використовує JIT-код (наприклад, mono/Unity3D), ви все одно матимете багато «звичайних» помилок сегментів. Пропонується використовувати щось на зразок `handle SIGSEGV nostop` в GDB, щоб не зупинятися на кожній помилці сегмента, і, можливо, поставити точку зупину всередині `my_memprotectionhandler` в `signals.c`, якщо ви хочете перехопити помилки сегментів.
+
+----
+Компіляція/Встановлення
+----
+> Інструкцію зі складання Box64 можна знайти [тут](docs/COMPILE.md).\
+> Інструкцію з встановлення Wine для Box64 можна знайти [тут](docs/X64WINE.md).
+
+----
+
+Історія версій/Журнал змін
+----
+
+Історія версій доступна [тут](docs/CHANGELOG.md).
+
+----
+
+Примітки щодо 32 бітних платформ
+----
+Оскільки Box64 працює шляхом прямого перекладу викликів функцій з x86_64 на хост-систему, хост-система (та, на якій працює Box64) повинна мати 64-розрядні бібліотеки. Box64 не містить 64-біт <-> 32-біт перекладу.
+
+Також розумійте що Box64 запускатиме лише 64-розрядні двійкові Linux файли. Для 32-розрядних двійкових файлів вам потрібен Box86 (з усіма хитрощами з мультиархітектурною системою або корінними правами). Зауважте, що багато інсталяторів (на основі mojo) повертаються до "x86" при виявленні ОС ARM64, і тому намагатимуться використовувати Box86 для налаштування, навіть якщо існує версія x86_64. Ви можете зламати свій шлях за допомогою фальшивого "uname", який повертає "x86_64", коли аргумент "-m"
+
+----
+
+Примітки щодо конфігурації Box64
+----
+
+Обидва файли мають однаковий синтаксис і в основному є файлами ini. Розділ у квадратних дужках визначає назву процесу, а решта — встановлють змінну середовища. Подивіться [використання](USAGE.md), щоб дізнатися, які параметри можна завдати. Box64 поставляється з файлом за замовчуванням, який слід встановити для кращої стабільності. Файл знаходиться в `system/box64.box64rc` і має бути встановлен в `/etc/box64.box64rc`. Якщо з певних причин ви не хочете встановлювати цей файл у `/etc`, принаймні скопіюйте його в `~/.box64rc` або гра може не працювати належним чином. Зауважте, що пріоритетом є: `~/.bashrc` > `/etc/box64.box64rc` > командний рядок
+Таким чином, ваші налаштування в `~/.bashrc` можуть замінити налаштування з вашого командного рядка...
+
+----
+
+Примітки щодо емуляції ігор Unity
+----
+
+Запуск ігор Unity має працювати, але ви також повинні зауважити, що для багатьох ігор Unity3D потрібен OpenGL 3+, який може бути складно забезпечити на ARM SBC (одноплатних комп’ютерах). Крім того, багато нових ігор Unity3D (наприклад, KSP) використовують стиснуті текстури BC7, які не підтримуються багатьма інтегрованими GPU ARM.
+Підказка: в Pi4 використовуйте `MESA_GL_VERSION_OVERRIDE=3.2`, а в Panfrost — `PAN_MESA_DEBUG=gl3`, щоб використовувати вищий профіль, якщо гра запускається, а потім завершується, перш ніж щось показує.
+
+----
+
+Примітки щодо програм GTK
+----
+
+Бібліотеки GTK2 і GTK3 тепер використовуються в Box64.
+
+----
+
+Примітки щодо Steam
+----
+
+Зауважте, що Steam є гібридом 32-біт / 64-біт. Вам ПОТРІБЕН box86, щоб запустити Steam, оскільки клієнта є 32-розрядною двійковою програмою. Він також використовує 64-розрядні двійкові файли локального сервера, і цей процес steamwebhelper тепер є обов’язковим навіть у «малому режимі». І цей процес з’їсть багато оперативної пам’яті, таким чином машині з 6 Гб оперативної пам’яті знадобиться файл підкачки для використання Steam.
+
+----
+
+Примітки щодо Wine
+----
+
+Wine64 підтримується Box64, Proton теж. Майте на увазі, що 64-розрядний Wine також містить 32-розрядні компоненти щоб мати можливість запускати 32-розрядні програми Windows. Для 32-розрядних програм потрібен Box86, і вони не працюватимуть без нього. У системі, де присутні і працюють Box64, і Box86, 64-розрядне встановлення Wine може запускати як 32-розрядні, так і 64-розрядні програми Windows (просто використовуйте `wine` і `wine64` відповідно).
+Зауважте, що новий 32-бітний PE в 64-бітному процесі, який Wine зараз реалізує у Wine 7+ тепер підтримується, але підтримка є досить молодою тому можуть виникнути деякі проблеми.
+
+----
+
+Примітки щодо Vulkan
+----
+
+Box64 використовує бібліотеки Vulkan, але зауважте що він тестувався з картою AMD Radeon RX550 і драйвером Freedreno, тому деякі розширення можуть бути відсутні залежно від вашої відеокарти.
+
+----
+
+Останнє слово
+----
+
+Я хочу подякувати всім, хто зробив внесок у розвиток box64.
+Є багато способів зробити внесок: внесок коду, фінансовий внесок, апаратне забезпечення та реклама!
+Отже, без особливого порядку, я хочу подякувати:
+* За основний внесок у код: rajdakin, mogery
+* За великий фінансовий внесок: FlyingFathead, stormchaser3000, dennis1248, sll00, [libre-computer-project](https://libre.computer/)
+* За апаратне забезпечення та міграцію LoongArch: [xiaoji](https://www.linuxgame.cn/), команда розробників Deepin Beijing
+* За основний внесок у код RV64 Dynarec: ksco, xctan
+* За внесок у апаратне забезпечення: [Radxa](https://rockpi.org/), [Pine64](https://www.pine64.org/), [StarFive](https://rvspace.org/)
+* За безперервну рекламу Box64: salva ([microLinux](https://www.youtube.com/channel/UCwFQAEj1lp3out4n7BeBatQ)), [PILab](https://www.youtube.com/channel/UCgfQjdc5RceRlTGfuthBs7g)/Команда [TwisterOS](https://twisteros.com/), [The Byteman](https://www.youtube.com/channel/UCEr8lpIJ3B5Ctc5BvcOHSnA), [NicoD](https://www.youtube.com/channel/UCpv7NFr0-9AB5xoklh3Snhg), ekianjo ([Boilingsteam](https://boilingsteam.com/))
+
+І я також дякую багатьом іншим людям, які хоч раз брали участь у цьому проекті.
+
+(Якщо ви використовуєте Box64 у своєму проекті, будь ласка, не забудьте згадати про це!)

docs/CHANGELOG.md

@@ -1,3 +1,28 @@
+v0.2.4
+======
+* Added Dynarec for RISC-V
+* Added partial 32 bits support (WoW64 works)
+* Changed cpuid to display box64
+* Fixed steam, again
+* Improved hacked version for overridden malloc/free functions
+* More wrapped libs
+* More wrapped functions and syscalls
+* Reduced memory footprint
+* Added support for the BIND_NOW flags in ELF files
+* Added RTLD_NOLOAD support
+* Fixed some typos and general cleanup, again
+* Added a cosimulation mode (to compare the behaviour of the dynarec with the interpreter)
+* Added FASTNAN/FASTROUND handling for more opcodes
+* Better float/double/long double handling
+* Better dlopen/dlclose handling
+* More games in the rc file
+* Added BOX64_DYNAREC_MISSING to show only missing opcode, independantly of the log level
+* Added BOX64_RESERVE_HIGH to reserve addresses above 47bits
+* Improved multitreading support
+* Ukrainian translation of the README
+=> Dynarec now support RISC-V, large speedup on RISC-V 64, were games are now playable (like Stardew Valley on a Vision Five 2)
+=> Improved compatibility with more fixes in elf handling, more wrapped libs and functiond, more fixed opcodes and refined profile
+
 v0.2.2
 ======
 * Added a hacked version for overridden malloc/free functions, enabling libcef (and derivative) to work.

docs/COMPILE.md

@@ -7,7 +7,7 @@ You can use [@ryanfortner](https://github.com/ryanfortner)'s apt repository to i
 ```
 sudo wget https://ryanfortner.github.io/box64-debs/box64.list -O /etc/apt/sources.list.d/box64.list
 wget -qO- https://ryanfortner.github.io/box64-debs/KEY.gpg | sudo gpg --dearmor -o /etc/apt/trusted.gpg.d/box64-debs-archive-keyring.gpg
-sudo apt update && sudo apt install box64-generic-arm -y
+sudo apt update && sudo apt install box64-arm64 -y
 ```
 
 Alternatively, you can generate your own package using the [instructions below](https://github.com/ptitSeb/box64/blob/main/docs/COMPILE.md#debian-packaging). 
@@ -111,6 +111,13 @@ Using a 64bit OS:
 -D PHYTIUM=1 -D CMAKE_BUILD_TYPE=RelWithDebInfo
 ```
 
+#### for a generic ARM64 machine
+
+Using a 64bit OS:
+```
+-D ARM64=1 -D CMAKE_BUILD_TYPE=RelWithDebInfo
+```
+
 #### for M1
 
 Only test on Asahi for now, using the default "16K page" kernel

rebuild_wrappers.py

@@ -824,6 +824,10 @@ def main(root: str, files: Iterable[Filename], ver: str):
 	allowed_regs  : str = "cCwWiuIUlLp"
 	allowed_fpr   : str = "fd"
 	
+	# Detect functions which return in an x87 register
+	retx87_wraps: Dict[ClausesStr, List[FunctionType]] = {}
+	return_x87: str = "DK"
+	
 	# Sanity checks
 	forbidden_simple: str = "EDKVOSNMHPAxX"
 	assert(len(allowed_simply) + len(allowed_regs) + len(allowed_fpr) + len(forbidden_simple) == len(allowed_conv.values))
@@ -831,6 +835,7 @@ def main(root: str, files: Iterable[Filename], ver: str):
 	assert(all(c not in allowed_simply + allowed_regs for c in allowed_fpr))
 	assert(all(c not in allowed_simply + allowed_regs + allowed_fpr for c in forbidden_simple))
 	assert(all(c in allowed_simply + allowed_regs + allowed_fpr + forbidden_simple for c in allowed_conv.values))
+	assert(all(c in allowed_conv.values for c in return_x87))
 	
 	def check_simple(v: FunctionType) -> Optional[int]:
 		regs_count: int = 0
@@ -867,6 +872,15 @@ def main(root: str, files: Iterable[Filename], ver: str):
 			simple_wraps[k] = tmp
 	simple_idxs = sorted(simple_wraps.keys(), key=lambda x: Clauses(x).splitdef())
 	
+	def check_return_x87(v: FunctionType) -> bool:
+		return v[0] in return_x87
+    
+	for k in gbls:
+		tmp = [v for v in gbls[k] if check_return_x87(v)]
+		if tmp:
+			retx87_wraps[k] = tmp
+	retx87_idxs = sorted(retx87_wraps.keys(), key=lambda x: Clauses(x).splitdef())
+	
 	# Now the files rebuilding part
 	# File headers and guards
 	files_header = {
@@ -1381,6 +1395,17 @@ def main(root: str, files: Iterable[Filename], ver: str):
 				file.write("#endif\n")
 		file.write("\treturn 0;\n}\n")
 		
+		# Write the isRetX87Wrapper function
+		file.write("\nint isRetX87Wrapper(wrapper_t fun) {\n")
+		for k in retx87_idxs:
+			if k != str(Clauses()):
+				file.write("#if " + k + "\n")
+			for vf in retx87_wraps[k]:
+				file.write("\tif (fun == &" + vf + ") return 1;\n")
+			if k != str(Clauses()):
+				file.write("#endif\n")
+		file.write("\treturn 0;\n}\n")
+		
 		file.write(files_guard["wrapper.c"].format(lbr="{", rbr="}", version=ver))
 	
 	# Rewrite the wrapper.h file:
@@ -1454,6 +1479,6 @@ if __name__ == '__main__':
 		if v == "--":
 			limit.append(i)
 	Define.defines = list(map(DefineType, sys.argv[2:limit[0]]))
-	if main(sys.argv[1], sys.argv[limit[0]+1:], "2.2.0.17") != 0:
+	if main(sys.argv[1], sys.argv[limit[0]+1:], "2.2.0.18") != 0:
 		exit(2)
 	exit(0)

src/box64context.c

@@ -259,6 +259,8 @@ box64context_t *NewBox64Context(int argc)
     context->segtls[4].present = 1;
     context->segtls[4].is32bits = 1;
 
+    context->globdata = NewMapSymbols();
+
     initAllHelpers(context);
 
     return context;
@@ -353,6 +355,8 @@ void FreeBox64Context(box64context_t** context)
     if(ctx->emu_sig)
         FreeX64Emu(&ctx->emu_sig);
 
+    FreeMapSymbols(&ctx->globdata);
+
     finiAllHelpers(ctx);
 
 #ifdef DYNAREC

src/box64version.h

@@ -3,6 +3,6 @@
 
 #define BOX64_MAJOR 0
 #define BOX64_MINOR 2
-#define BOX64_REVISION 3
+#define BOX64_REVISION 4
 
 #endif //__BOX64_VERSION_H_

src/custommem.c

@@ -1463,7 +1463,7 @@ void init_custommem_helper(box64context_t* ctx)
         memprot[i].prot = memprot_default;
     init_mutexes();
 #ifdef DYNAREC
-    if(box64_dynarec)
+    if(box64_dynarec) {
         for(int i=0; i<(1<<JMPTABL_SHIFT3); ++i)
             box64_jmptbl3[i] = box64_jmptbldefault2;
         for(int i=0; i<(1<<JMPTABL_SHIFT2); ++i)
@@ -1472,6 +1472,7 @@ void init_custommem_helper(box64context_t* ctx)
             box64_jmptbldefault1[i] = box64_jmptbldefault0;
         for(int i=0; i<(1<<JMPTABL_SHIFT0); ++i)
             box64_jmptbldefault0[i] = (uintptr_t)native_next;
+    }
     lockaddress = kh_init(lockaddress);
 #endif
     pthread_atfork(NULL, NULL, atfork_child_custommem);

src/dynarec/arm64/arm64_emitter.h

@@ -363,6 +363,9 @@
 #define POP2_32(reg1, reg2)     LDPw_S7_postindex(reg1, reg2, xRSP, 8)
 #define PUSH2_32(reg1, reg2)    STPw_S7_preindex(reg2, reg1, xRSP, -8)
 
+#define POP1_16(reg)            LDRH_S9_postindex(reg, xRSP, 2)
+#define PUSH1_16(reg)           STRH_S9_preindex(reg, xRSP, -2)
+
 #define POP1z(reg)              if(rex.is32bits) {POP1_32(reg);} else {POP1(reg);}
 #define PUSH1z(reg)             if(rex.is32bits) {PUSH1_32(reg);} else {PUSH1(reg);}
 #define POP2z(reg1, reg2)       if(rex.is32bits) {POP2_32(reg1, reg2);} else {POP2(reg1, reg2);}
@@ -487,9 +490,9 @@
 #define CSNEGx(Rd, Rn, Rm, cond)            EMIT(CSNEG_gen(1, Rm, cond, Rn, Rd))
 #define CSNEGw(Rd, Rn, Rm, cond)            EMIT(CSNEG_gen(0, Rm, cond, Rn, Rd))
 #define CSNEGxw(Rd, Rn, Rm, cond)           EMIT(CSNEG_gen(rex.w, Rm, cond, Rn, Rd))
-#define CNEGx(Rd, Rn, cond)                 CSNEGx(Rn, Rn, Rn, invCond(cond))
-#define CNEGw(Rd, Rn, cond)                 CSNEGw(Rn, Rn, Rn, invCond(cond))
-#define CNEGxw(Rd, Rn, cond)                CSNEGxw(Rn, Rn, Rn, invCond(cond))
+#define CNEGx(Rd, Rn, cond)                 CSNEGx(Rn, Rn, Rd, invCond(cond))
+#define CNEGw(Rd, Rn, cond)                 CSNEGw(Rn, Rn, Rd, invCond(cond))
+#define CNEGxw(Rd, Rn, cond)                CSNEGxw(Rn, Rn, Rd, invCond(cond))
 
 // AND / ORR
 #define LOGIC_gen(sf, opc, N, immr, imms, Rn, Rd)  ((sf)<<31 | (opc)<<29 | 0b100100<<23 | (N)<<22 | (immr)<<16 | (imms)<<10 | (Rn)<<5 | Rd)
@@ -534,6 +537,15 @@
 #define EORx_REG_LSR(Rd, Rn, Rm, lsr)   EMIT(LOGIC_REG_gen(1, 0b10, 0b01, 0, Rm, lsr, Rn, Rd))
 #define EORw_REG_LSR(Rd, Rn, Rm, lsr)   EMIT(LOGIC_REG_gen(0, 0b10, 0b01, 0, Rm, lsr, Rn, Rd))
 #define EORxw_REG_LSR(Rd, Rn, Rm, lsr)  EMIT(LOGIC_REG_gen(rex.w, 0b10, 0b01, 0, Rm, lsr, Rn, Rd))
+#define EONx_REG(Rd, Rn, Rm)            EMIT(LOGIC_REG_gen(1, 0b10, 0b00, 1, Rm, 0, Rn, Rd))
+#define EONw_REG(Rd, Rn, Rm)            EMIT(LOGIC_REG_gen(0, 0b10, 0b00, 1, Rm, 0, Rn, Rd))
+#define EONxw_REG(Rd, Rn, Rm)           EMIT(LOGIC_REG_gen(rex.w, 0b10, 0b00, 1, Rm, 0, Rn, Rd))
+#define EONx_REG_LSL(Rd, Rn, Rm, lsl)   EMIT(LOGIC_REG_gen(1, 0b10, 0b00, 1, Rm, lsl, Rn, Rd))
+#define EONw_REG_LSL(Rd, Rn, Rm, lsl)   EMIT(LOGIC_REG_gen(0, 0b10, 0b00, 1, Rm, lsl, Rn, Rd))
+#define EONxw_REG_LSL(Rd, Rn, Rm, lsl)  EMIT(LOGIC_REG_gen(rex.w, 0b10, 0b00, 1, Rm, lsl, Rn, Rd))
+#define EONx_REG_LSR(Rd, Rn, Rm, lsr)   EMIT(LOGIC_REG_gen(1, 0b10, 0b01, 1, Rm, lsr, Rn, Rd))
+#define EONw_REG_LSR(Rd, Rn, Rm, lsr)   EMIT(LOGIC_REG_gen(0, 0b10, 0b01, 1, Rm, lsr, Rn, Rd))
+#define EONxw_REG_LSR(Rd, Rn, Rm, lsr)  EMIT(LOGIC_REG_gen(rex.w, 0b10, 0b01, 1, Rm, lsr, Rn, Rd))
 #define MOVx_REG(Rd, Rm)                ORRx_REG(Rd, xZR, Rm)
 #define MOVw_REG(Rd, Rm)                ORRw_REG(Rd, xZR, Rm)
 #define MOVxw_REG(Rd, Rm)               ORRxw_REG(Rd, xZR, Rm)
@@ -560,6 +572,12 @@
 #define BICx_REG    BICx
 #define BICw_REG    BICw
 #define BICxw_REG   BICxw
+#define BICx_REG_LSL(Rd, Rn, Rm, lsl)   EMIT(LOGIC_REG_gen(1, 0b00, 0b00, 1, Rm, lsl, Rn, Rd))
+#define BICw_REG_LSL(Rd, Rn, Rm, lsl)   EMIT(LOGIC_REG_gen(0, 0b00, 0b00, 1, Rm, lsl, Rn, Rd))
+#define BICxw_REG_LSL(Rd, Rn, Rm, lsl)  EMIT(LOGIC_REG_gen(rex.w, 0b00, 0b00, 1, Rm, lsl, Rn, Rd))
+#define BICx_REG_LSR(Rd, Rn, Rm, lsl)   EMIT(LOGIC_REG_gen(1, 0b00, 0b01, 1, Rm, lsl, Rn, Rd))
+#define BICw_REG_LSR(Rd, Rn, Rm, lsl)   EMIT(LOGIC_REG_gen(0, 0b00, 0b01, 1, Rm, lsl, Rn, Rd))
+#define BICxw_REG_LSR(Rd, Rn, Rm, lsl)  EMIT(LOGIC_REG_gen(rex.w, 0b00, 0b01, 1, Rm, lsl, Rn, Rd))
 #define TSTx_REG(Rn, Rm)                ANDSx_REG(xZR, Rn, Rm)
 #define TSTw_REG(Rn, Rm)                ANDSw_REG(wZR, Rn, Rm)
 #define TSTxw_REG(Rn, Rm)               ANDSxw_REG(xZR, Rn, Rm)
@@ -950,6 +968,10 @@
 #define VSUB_16(Vd, Vn, Vm)                 EMIT(ADDSUB_vector(0, 1, 0b01, Vm, Vn, Vd))
 #define VSUB_32(Vd, Vn, Vm)                 EMIT(ADDSUB_vector(0, 1, 0b10, Vm, Vn, Vd))
 
+#define ADDSUB_scalar(U, size, Rm, Rn, Rd)  (01<<30 | (U)<<29 | 0b11110<<24 | (size)<<22 | 1<<21 | (Rm)<<16 | 0b10000<<11 | 1<<10 | (Rn)<<5 | (Rd))
+#define ADD_64(Vd, Vn, Vm)                  EMIT(ADDSUB_scalar(0, 0b11, Vm, Vn, Vd))
+#define SUB_64(Vd, Vn, Vm)                  EMIT(ADDSUB_scalar(1, 0b11, Vm, Vn, Vd))
+
 #define NEGABS_vector(Q, U, size, Rn, Rd)   ((Q)<<30 | (U)<<29 | 0b01110<<24 | (size)<<22 | 0b10000<<17 | 0b01011<<12 | 0b10<<10 | (Rn)<<5 | (Rd))
 #define NEG_8(Vd, Vn)                       EMIT(NEGABS_vector(0, 1, 0b00, Vn, Vd))
 #define NEG_16(Vd, Vn)                      EMIT(NEGABS_vector(0, 1, 0b01, Vn, Vd))
@@ -1424,14 +1446,23 @@
 #define VUZP1Q_64(Rt, Rn, Rm)       EMIT(UZP_gen(1, 0b11, Rm, 0, Rn, Rt))
 #define VUZP2Q_64(Rt, Rn, Rm)       EMIT(UZP_gen(1, 0b11, Rm, 1, Rn, Rt))
 
-#define DUP_gen(Q, imm5, Rn, Rd)    ((Q)<<30 | 0b01110000<<21 | (imm5)<<16 | 1<<10 | (Rn)<<5 | (Rd))
-#define VDUP_8(Vd, Vn, idx)         EMIT(DUP_gen(0, ((idx)<<1|1), Vn, Vd))
-#define VDUPQ_8(Vd, Vn, idx)        EMIT(DUP_gen(1, ((idx)<<1|1), Vn, Vd))
-#define VDUP_16(Vd, Vn, idx)        EMIT(DUP_gen(0, ((idx)<<2|0b10), Vn, Vd))
-#define VDUPQ_16(Vd, Vn, idx)       EMIT(DUP_gen(1, ((idx)<<2|0b10), Vn, Vd))
-#define VDUP_32(Vd, Vn, idx)        EMIT(DUP_gen(0, ((idx)<<3|0b100), Vn, Vd))
-#define VDUPQ_32(Vd, Vn, idx)       EMIT(DUP_gen(1, ((idx)<<3|0b100), Vn, Vd))
-#define VDUPQ_64(Vd, Vn, idx)       EMIT(DUP_gen(1, ((idx)<<4|0b1000), Vn, Vd))
+#define DUP_element(Q, imm5, Rn, Rd)    ((Q)<<30 | 0b01110000<<21 | (imm5)<<16 | 1<<10 | (Rn)<<5 | (Rd))
+#define VDUP_8(Vd, Vn, idx)         EMIT(DUP_element(0, ((idx)<<1|1), Vn, Vd))
+#define VDUPQ_8(Vd, Vn, idx)        EMIT(DUP_element(1, ((idx)<<1|1), Vn, Vd))
+#define VDUP_16(Vd, Vn, idx)        EMIT(DUP_element(0, ((idx)<<2|0b10), Vn, Vd))
+#define VDUPQ_16(Vd, Vn, idx)       EMIT(DUP_element(1, ((idx)<<2|0b10), Vn, Vd))
+#define VDUP_32(Vd, Vn, idx)        EMIT(DUP_element(0, ((idx)<<3|0b100), Vn, Vd))
+#define VDUPQ_32(Vd, Vn, idx)       EMIT(DUP_element(1, ((idx)<<3|0b100), Vn, Vd))
+#define VDUPQ_64(Vd, Vn, idx)       EMIT(DUP_element(1, ((idx)<<4|0b1000), Vn, Vd))
+
+#define DUP_general(Q, imm5, Rn, Rd)    ((Q)<<30 | 0b01110000<<21 | (imm5)<<16 | 0b11<<10 | (Rn)<<5 | (Rd))
+#define VDUPB(Vd, Wn)         EMIT(DUP_general(0, 0b1, Wn, Vd))
+#define VDUPQB(Vd, Wn)        EMIT(DUP_general(1, 0b1, Wn, Vd))
+#define VDUPH(Vd, Wn)        EMIT(DUP_general(0, 0b10, Wn, Vd))
+#define VDUPQH(Vd, Wn)       EMIT(DUP_general(1, 0b10, Wn, Vd))
+#define VDUPS(Vd, Wn)        EMIT(DUP_general(0, 0b100, Wn, Vd))
+#define VDUPQS(Vd, Wn)       EMIT(DUP_general(1, 0b100, Wn, Vd))
+#define VDUPQD(Vd, Xn)       EMIT(DUP_general(1, 0b1000, Xn, Vd))
 
 // TBL
 #define TBL_gen(Q, Rm, len, op, Rn, Rd) ((Q)<<30 | 0b001110<<24 | (Rm)<<16 | (len)<<13 | (op)<<12 | (Rn)<<5 | (Rd))

src/dynarec/arm64/arm64_lock.h

@@ -27,46 +27,46 @@ extern void arm64_lock_read_dq(uint64_t * a, uint64_t* b, void* addr);
 // STLXRD of ADDR, return 0 if ok, 1 if not
 extern int arm64_lock_write_dq(uint64_t a, uint64_t b, void* addr);
 
-// Atomicaly exchange value at [p] with val, return old p
+// Atomically exchange value at [p] with val, return old p
 extern uintptr_t arm64_lock_xchg_dd(void* p, uintptr_t val);
 
-// Atomicaly exchange value at [p] with val, return old p
+// Atomically exchange value at [p] with val, return old p
 extern uint32_t arm64_lock_xchg_d(void* p, uint32_t val);
 
-// Atomicaly exchange value at [p] with val, return old p
+// Atomically exchange value at [p] with val, return old p
 extern uint32_t arm64_lock_xchg_h(void* p, uint32_t val);
 
-// Atomicaly exchange value at [p] with val, return old p
+// Atomically exchange value at [p] with val, return old p
 extern uint32_t arm64_lock_xchg_b(void* p, uint32_t val);
 
-// Atomicaly store value to [p] only if [p] is NULL. Return old [p] value
+// Atomically store value to [p] only if [p] is NULL. Return old [p] value
 extern uint32_t arm64_lock_storeifnull_d(void*p, uint32_t val);
 
-// Atomicaly store value to [p] only if [p] is NULL. Return old [p] value
+// Atomically store value to [p] only if [p] is NULL. Return old [p] value
 extern void* arm64_lock_storeifnull(void*p, void* val);
 
-// Atomicaly store value to [p] only if [p] is ref. Return new [p] value (so val or old)
+// Atomically store value to [p] only if [p] is ref. Return new [p] value (so val or old)
 extern void* arm64_lock_storeifref(void*p, void* val, void* ref);
 
-// Atomicaly store value to [p] only if [p] is ref. Return new [p] value (so val or old)
+// Atomically store value to [p] only if [p] is ref. Return new [p] value (so val or old)
 extern uint32_t arm64_lock_storeifref_d(void*p, uint32_t val, uint32_t ref);
 
-// Atomicaly store value to [p] only if [p] is ref. Return new [p] value (so val or old)
+// Atomically store value to [p] only if [p] is ref. Return new [p] value (so val or old)
 extern uint32_t arm64_lock_storeifref2_d(void*p, uint32_t val, uint32_t ref);
 
-// decrement atomicaly the byte at [p] (but only if p not 0)
+// decrement atomically the byte at [p] (but only if p not 0)
 extern void arm64_lock_decifnot0b(void*p);
 
 // atomic store (with memory barrier)
 extern void arm64_lock_storeb(void*p, uint8_t b);
 
-// increment atomicaly the int at [p] only if it was 0. Return the old value of [p]
+// increment atomically the int at [p] only if it was 0. Return the old value of [p]
 extern int arm64_lock_incif0(void*p);
 
-// decrement atomicaly the int at [p] (but only if p not 0)
+// decrement atomically the int at [p] (but only if p not 0)
 extern int arm64_lock_decifnot0(void*p);
 
 // atomic store (with memory barrier)
 extern void arm64_lock_store(void*p, uint32_t v);
 
-#endif  //__ARM64_LOCK__H__
+#endif  //__ARM64_LOCK__H__

src/dynarec/arm64/arm64_printer.c

@@ -1107,7 +1107,19 @@ const char* arm64_print(uint32_t opcode, uintptr_t addr)
         snprintf(buff, sizeof(buff), "FSQRT %c%d, %c%d", s, Rd, s, Rn);
         return buff;
     }
-
+    // FPRINTX
+    if(isMask(opcode, "0Q1011100f100001100110nnnnnddddd", &a)) {
+        char s = a.Q?'V':'D';
+        char d = sf?'D':'S';
+        int n = (a.Q && !sf)?4:2;
+        snprintf(buff, sizeof(buff), "VFRINTX %c%d.%d%c, %c%d.%d%c", s, Rd, n, d, s, Rn, n, d);
+        return buff;
+    }
+    if (isMask(opcode, "00011110ff100111010000nnnnnddddd", &a)) {
+        char s = (sf==0)?'S':((sf==1)?'D':'?');
+        snprintf(buff, sizeof(buff), "FRINTX %c%d, %c%d", s, Rd, s, Rn);
+        return buff;
+    }
     // FRINTI
     if(isMask(opcode, "0Q1011101f100001100110nnnnnddddd", &a)) {
         char s = a.Q?'V':'D';
@@ -1487,7 +1499,7 @@ const char* arm64_print(uint32_t opcode, uintptr_t addr)
 
     // DUP
     if(isMask(opcode, "0Q001110000iiiii000001nnnnnddddd", &a)) {
-        const char* Y[] = {"8B", "16B", "4H", "8H", "2S", "4S", "2D", "??"};
+        const char* Y[] = {"8B", "16B", "4H", "8H", "2S", "4S", "??", "2D"};
         const char* Z[] = {"B", "H", "S", "D"};
         int sz = 3;
         if((imm&0b0001)==0b0001) sz=0;
@@ -1499,7 +1511,18 @@ const char* arm64_print(uint32_t opcode, uintptr_t addr)
         snprintf(buff, sizeof(buff), "DUP V%d.%s, V%d.%s[%d]", Rd, Vd, Rn, Vn, sh);
         return buff;
     }
-    
+
+    if(isMask(opcode, "0Q001110000iiiii000011nnnnnddddd", &a)) {
+        const char* Y[] = {"8B", "16B", "4H", "8H", "2S", "4S", "??", "2D"};
+        int sz = 3;
+        if((imm&0b0001)==0b0001) sz=0;
+        else if((imm&0b0011)==0b0010) sz=1;
+        else if((imm&0b0111)==0b0100) sz=2;
+        const char* Vd = Y[(sz<<1)|a.Q];
+        snprintf(buff, sizeof(buff), "DUP V%d.%s, X%d", Rd, Vd, Rn);
+        return buff;
+    }
+
     // AES
     if(isMask(opcode, "0100111000101000010f10nnnnnddddd", &a)) {
         snprintf(buff, sizeof(buff), "AES%c V%d.16B, V%d.16B", sf?'D':'E', Rd, Rn);

src/dynarec/arm64/dynarec_arm64_00.c

@@ -25,6 +25,7 @@
 #include "dynarec_arm64_helper.h"
 
 int isSimpleWrapper(wrapper_t fun);
+int isRetX87Wrapper(wrapper_t fun);
 
 uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int ninst, rex_t rex, int rep, int* ok, int* need_epilog)
 {
@@ -525,6 +526,19 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                 emit_cmp32_0(dyn, ninst, rex, xRAX, x3, x4);
             break;
 
+        case 0x3F:
+            if(rex.is32bits) {
+                INST_NAME("AAS");
+                MESSAGE(LOG_DUMP, "Need Optimization AAS\n");
+                READFLAGS(X_AF);
+                SETFLAGS(X_AF|X_CF|X_PF|X_SF|X_ZF, SF_SET);
+                UXTHw(x1, xRAX);
+                CALL_(aas16, x1, 0);
+                BFIx(xRAX, x1, 0, 16);
+            } else {
+                DEFAULT;
+            }
+            break;
         case 0x40:
         case 0x41:
         case 0x42:
@@ -665,7 +679,16 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                 DEFAULT;
             }
             break;
-
+        case 0x62:
+            if(rex.is32bits) {
+                // BOUND here
+                DEFAULT;                
+            } else {
+                INST_NAME("BOUND Gd, Ed");
+                nextop = F8;
+                FAKEED(0);
+            }
+            break;
         case 0x63:
             if(rex.is32bits) {
                 // ARPL here
@@ -1007,7 +1030,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             nextop=F8;
             GETGD;
             GETED(0);
-            emit_test32(dyn, ninst, rex, ed, gd, x3, x5);
+            emit_test32(dyn, ninst, rex, ed, gd, x3, x5, x6);
             break;
         case 0x86:
             INST_NAME("(LOCK)XCHG Eb, Gb");
@@ -1301,6 +1324,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             else
                 u64 = F64;
             MOV64z(x1, u64);
+            SMREAD();
             LDRB_U12(x2, x1, 0);
             BFIx(xRAX, x2, 0, 8);
             break;
@@ -1311,6 +1335,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             else
                 u64 = F64;
             MOV64z(x1, u64);
+            SMREAD();
             LDRxw_U12(xRAX, x1, 0);
             break;
         case 0xA2:
@@ -1476,7 +1501,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             SETFLAGS(X_ALL, SF_SET_PENDING);
             i64 = F32S;
             MOV64xw(x2, i64);
-            emit_test32(dyn, ninst, rex, xRAX, x2, x3, x4);
+            emit_test32(dyn, ninst, rex, xRAX, x2, x3, x4, x5);
             break;
         case 0xAA:
             if(rep) {
@@ -1522,26 +1547,42 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                 ADDx_REG(xRDI, xRDI, x3);
             }
             break;
-
+        case 0xAC:
+            if(rep) {
+                INST_NAME("REP LODSB");
+            } else {
+                INST_NAME("LODSB");
+            }
+            GETDIR(x1, 1);
+            if(rep) {
+                CBZx_NEXT(xRCX);
+                MARK;
+            }
+            LDRB_U12(x2, xRSI, 0);
+            ADDx_REG(xRSI, xRSI, x1);
+            if(rep) {
+                SUBx_U12(xRCX, xRCX, 1);
+                CBNZx_MARK(xRCX);
+            }
+            BFIx(xRAX, x2, 0, 8);
+            break;
         case 0xAD:
             if(rep) {
                 INST_NAME("REP LODSD");
             } else {
                 INST_NAME("LODSD");
             }
-            LDRxw_U12(xRAX, xRSI, 0);
-            BFIw(x2, xFlags, F_DF, 1);
-            if(rex.w) {
-                MOV32w(x1, 8);
-            } else {
-                MOV32w(x1, 4);
-            }
-            EORxw_REG_LSL(x1, x1, x2, rex.w?63:31);
+            GETDIR(x1, rex.w?8:4);
             if(rep) {
-                MULxw(x1, x1, xRCX);
-                EORw_REG(xRCX, xRCX, xRCX);
+                CBZx_NEXT(xRCX);
+                MARK;
             }
+            LDRxw_U12(xRAX, xRSI, 0);
             ADDx_REG(xRSI, xRSI, x1);
+            if(rep) {
+                SUBx_U12(xRCX, xRCX, 1);
+                CBNZx_MARK(xRCX);
+            }
             break;
         case 0xAE:
             switch(rep) {
@@ -1901,6 +1942,9 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     x87_forget(dyn, ninst, x3, x4, 0);
                     sse_purge07cache(dyn, ninst, x3);
                     tmp = isSimpleWrapper(*(wrapper_t*)(addr));
+                    if(isRetX87Wrapper(*(wrapper_t*)(addr)))
+                        // return value will be on the stack, so the stack depth needs to be updated
+                        x87_purgecache(dyn, ninst, 0, x3, x1, x4);
                     if((box64_log<2 && !cycle_log) && tmp) {
                         //GETIP(ip+3+8+8); // read the 0xCC
                         call_n(dyn, ninst, *(void**)(addr+8), tmp);
@@ -1941,10 +1985,15 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             break;
         case 0xCD:
             u8 = F8;
-            if(box64_wine && u8==0x2D) {
-                INST_NAME("INT 2D");
+            if(box64_wine && (u8==0x2D || u8==0x2C || u8==0x29)) {
+                INST_NAME("INT 29/2c/2d");
                 // lets do nothing
-                MESSAGE(LOG_INFO, "INT 2D Windows anti-debug hack\n");
+                MESSAGE(LOG_INFO, "INT 29/2c/2d Windows interruption\n");
+                GETIP(ip);
+                STORE_XEMU_CALL(xRIP);
+                MOV32w(x1, u8);
+                CALL(native_int, -1);
+                LOAD_XEMU_CALL(xRIP);
             } else if (u8==0x80) {
                 INST_NAME("32bits SYSCALL");
                 NOTEST(x1);
@@ -1961,14 +2010,6 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                 MARK;
                 LOAD_XEMU_REM();
                 jump_to_epilog(dyn, 0, xRIP, ninst);
-            } else if(box64_wine && u8==0x29) {
-                INST_NAME("INT 0x29");
-                // __fastfail ignored!
-                MOV32w(x1, 1);
-                STRw_U12(x1, xEmu, offsetof(x64emu_t, quit));
-                jump_to_epilog(dyn, 0, xRIP, ninst);
-                *need_epilog = 0;
-                *ok = 0;
             } else {
                 INST_NAME("INT n");
                 SETFLAGS(X_ALL, SF_SET);    // Hack to set flags in "don't care" state
@@ -2338,7 +2379,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             READFLAGS(X_ZF);
             i8 = F8S;
             SUBz_U12(xRCX, xRCX, 1);
-            TBNZ_NEXT(xFlags, 1<<F_ZF);
+            TBNZ_NEXT(xFlags, F_ZF);
             GO(0);
             break;
         case 0xE1:
@@ -2346,7 +2387,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             READFLAGS(X_ZF);
             i8 = F8S;
             SUBz_U12(xRCX, xRCX, 1);
-            TBZ_NEXT(xFlags, 1<<F_ZF);
+            TBZ_NEXT(xFlags, F_ZF);
             GO(0);
             break;
         case 0xE2:
@@ -2392,10 +2433,13 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     SKIPTEST(x1);    // disable test as this hack dos 2 instructions for 1
                     // calling a native function
                     sse_purge07cache(dyn, ninst, x3);
-                    if((box64_log<2 && !cycle_log) && dyn->insts[ninst].natcall)
+                    if((box64_log<2 && !cycle_log) && dyn->insts[ninst].natcall) {
                         tmp=isSimpleWrapper(*(wrapper_t*)(dyn->insts[ninst].natcall+2));
-                    else
+                    } else
                         tmp=0;
+                    if(dyn->insts[ninst].natcall && isRetX87Wrapper(*(wrapper_t*)(dyn->insts[ninst].natcall+2)))
+                    // return value will be on the stack, so the stack depth needs to be updated
+                        x87_purgecache(dyn, ninst, 0, x3, x1, x4);
                     if((box64_log<2 && !cycle_log) && dyn->insts[ninst].natcall && tmp) {
                         //GETIP(ip+3+8+8); // read the 0xCC
                         call_n(dyn, ninst, *(void**)(dyn->insts[ninst].natcall+2+8), tmp);
@@ -2604,7 +2648,7 @@ uintptr_t dynarec64_00(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     GETEDH(x1, 4);
                     i64 = F32S;
                     MOV64xw(x2, i64);
-                    emit_test32(dyn, ninst, rex, ed, x2, x3, x4);
+                    emit_test32(dyn, ninst, rex, ed, x2, x3, x4, x5);
                     break;
                 case 2:
                     INST_NAME("NOT Ed");

src/dynarec/arm64/dynarec_arm64_0f.c

@@ -30,8 +30,8 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
     uint8_t opcode = F8;
     uint8_t nextop, u8;
     uint8_t gd, ed;
-    uint8_t wback, wb2;
-    uint8_t eb1, eb2;
+    uint8_t wback, wb1, wb2;
+    uint8_t eb1, eb2, gb1, gb2;
     int32_t i32, i32_;
     int cacheupd = 0;
     int v0, v1;
@@ -42,9 +42,12 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
     int64_t j64;
     int64_t fixedaddress;
     int unscaled;
+    MAYUSE(wb1);
     MAYUSE(wb2);
     MAYUSE(eb1);
     MAYUSE(eb2);
+    MAYUSE(gb1);
+    MAYUSE(gb2);
     MAYUSE(q0);
     MAYUSE(q1);
     MAYUSE(d0);
@@ -1104,7 +1107,7 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<(2+rex.w), (1<<(2+rex.w))-1, rex, NULL, 0, 0);
-                ASRxw(x1, gd, 5+rex.w); // r1 = (gd>>5)
+                ASRx(x1, gd, 5+rex.w); // r1 = (gd>>5)
                 ADDx_REG_LSL(x3, wback, x1, 2+rex.w); //(&ed)+=r1*4;
                 LDxw(x1, x3, fixedaddress);
                 ed = x1;
@@ -1152,7 +1155,7 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<(2+rex.w), (1<<(2+rex.w))-1, rex, NULL, 0, 0);
-                ASRxw(x1, gd, 5+rex.w); // r1 = (gd>>5)
+                ASRx(x1, gd, 5+rex.w); // r1 = (gd>>5)
                 ADDx_REG_LSL(x3, wback, x1, 2+rex.w); //(&ed)+=r1*4;
                 LDxw(x1, x3, fixedaddress);
                 ed = x1;
@@ -1318,7 +1321,7 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<(2+rex.w), (1<<(2+rex.w))-1, rex, NULL, 0, 0);
-                ASRxw(x1, gd, 5+rex.w); // r1 = (gd>>5)
+                ASRx(x1, gd, 5+rex.w); // r1 = (gd>>5)
                 ADDx_REG_LSL(x3, wback, x1, 2+rex.w); //(&ed)+=r1*4;
                 LDxw(x1, x3, fixedaddress);
                 ed = x1;
@@ -1331,15 +1334,14 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             }
             LSRxw_REG(x4, ed, x2);
             if(rex.w) {
-                ANDSx_mask(x4, x4, 1, 0, 0);  //mask=1
+                ANDx_mask(x4, x4, 1, 0, 0);  //mask=1
             } else {
-                ANDSw_mask(x4, x4, 0, 0);  //mask=1
+                ANDw_mask(x4, x4, 0, 0);  //mask=1
             }
             BFIw(xFlags, x4, F_CF, 1);
             MOV32w(x4, 1);
             LSLxw_REG(x4, x4, x2);
-            EORxw_REG(x4, ed, x4);
-            CSELxw(ed, ed, x4, cEQ);
+            BICxw_REG(ed, ed, x4);
             if(wback) {
                 STRxw_U12(ed, wback, fixedaddress);
                 SMWRITE();
@@ -1418,7 +1420,7 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     BFXILxw(xFlags, ed, u8, 1);  // inject 1 bit from u8 to F_CF (i.e. pos 0)
                     TBNZ_MARK3(xFlags, 0); // bit already set, jump to next instruction
                     MOV32w(x4, 1);
-                    EORxw_REG_LSL(ed, ed, x4, u8);
+                    ORRxw_REG_LSL(ed, ed, x4, u8);
                     if(wback) {
                         STxw(ed, wback, fixedaddress);
                         SMWRITE();
@@ -1443,7 +1445,7 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     BFXILxw(xFlags, ed, u8, 1);  // inject 1 bit from u8 to F_CF (i.e. pos 0)
                     TBZ_MARK3(xFlags, 0); // bit already clear, jump to next instruction
                     MOV32w(x4, 1);
-                    EORxw_REG_LSL(ed, ed, x4, u8);
+                    BICxw_REG_LSL(ed, ed, x4, u8);
                     if(wback) {
                         STxw(ed, wback, fixedaddress);
                         SMWRITE();
@@ -1490,7 +1492,7 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<(2+rex.w), (1<<(2+rex.w))-1, rex, NULL, 0, 0);
-                ASRxw(x1, gd, 5+rex.w); // r1 = (gd>>5)
+                ASRx(x1, gd, 5+rex.w); // r1 = (gd>>5)
                 ADDx_REG_LSL(x3, wback, x1, 2+rex.w); //(&ed)+=r1*4;
                 LDxw(x1, x3, fixedaddress);
                 ed = x1;
@@ -1580,7 +1582,29 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                 LDSHxw(gd, ed, fixedaddress);
             }
             break;
-
+        case 0xC0:
+            INST_NAME("XADD Gb, Eb");
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGB(x1);
+            GETEB(x2, 0);
+            gd = x2; ed = x1;    // swap gd/ed
+            emit_add8(dyn, ninst, x1, x2, x4, x5);
+            GBBACK; // gb gets x2 (old ed)
+            EBBACK; // eb gets x1 (sum)
+            break;
+        case 0xC1:
+            INST_NAME("XADD Gd, Ed");
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGD;
+            GETED(0);
+            MOVxw_REG(x3, ed);
+            MOVxw_REG(ed, gd);
+            MOVxw_REG(gd, x3);
+            emit_add32(dyn, ninst, rex, ed, gd, x4, x5);
+            WBACK;
+            break;
         case 0xC2:
             INST_NAME("CMPPS Gx, Ex, Ib");
             nextop = F8;
@@ -1701,7 +1725,56 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             gd = xRAX+(opcode&7)+(rex.b<<3);
             REVxw(gd, gd);
             break;
-
+        case 0xD1:
+            INST_NAME("PSRLW Gm, Em");
+            nextop = F8;
+            GETGM(d0);
+            GETEM(d1, 0);
+            if(MODREG) {
+                q0 = fpu_get_scratch(dyn);
+            }
+            else {
+                q0 = d1;
+            }
+            q1 = fpu_get_scratch(dyn);
+            VMOVBto(x1, d1, 0);
+            MOVZw(x2, 16);
+            SUBSw_REG(x2, x2, x1);
+            B_MARK(cGT);
+            VMOVQDfrom(d0, 0, xZR);
+            B_NEXT_nocond;
+            MARK;
+            VDUPQS(q1, x2);
+            UXTL_16(q0, d0);
+            USHLQ_32(q0, q0, q1);
+            VUZP2Q_16(q0, q0, q0);
+            VMOVeD(d0, 0, q0, 0);
+            break;
+        case 0xD2:
+            INST_NAME("PSRLD Gm, Em");
+            nextop = F8;
+            GETGM(d0);
+            GETEM(d1, 0);
+            if(MODREG) {
+                q0 = fpu_get_scratch(dyn);
+            }
+            else {
+                q0 = d1;
+            }
+            q1 = fpu_get_scratch(dyn);
+            VMOVBto(x1, d1, 0);
+            MOVZw(x2, 32);
+            SUBSw_REG(x2, x2, x1);
+            B_MARK(cGT);
+            VMOVQDfrom(d0, 0, xZR);
+            B_NEXT_nocond;
+            MARK;
+            VDUPQD(q1, x2);
+            UXTL_32(q0, d0);
+            USHLQ_64(q0, q0, q1);
+            VUZP2Q_32(q0, q0, q0);
+            VMOVeD(d0, 0, q0, 0);
+            break;
         case 0xD3:
             INST_NAME("PSRLQ Gm,Em");
             nextop = F8;
@@ -1714,7 +1787,13 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             NEG_64(q0, d1);
             USHL_R_64(d0, d0, q0);
             break;
-
+        case 0xD4:
+            INST_NAME("PADDQ Gm,Em");
+            nextop = F8;
+            GETGM(v0);
+            GETEM(q0, 0);
+            ADD_64(v0, v0, q0);
+            break;
         case 0xD5:
             INST_NAME("PMULLW Gm, Em");
             nextop = F8;
@@ -1908,7 +1987,13 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             SQXTN_32(v0, v0); // 2*q1 in 32bits now
             SSHL_32(d0, d0, v0);
             break;
-
+        case 0xF4:
+            INST_NAME("PMULUDQ Gx,Ex");
+            nextop = F8;
+            GETGX(v0, 1);
+            GETEX(v1, 0, 0);
+            VUMULL_32(v0, v0, v1);
+            break;
         case 0xF5:
             INST_NAME("PMADDWD Gm, Em");
             nextop = F8;
@@ -1961,6 +2046,13 @@ uintptr_t dynarec64_0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             GETEM(v1, 0);
             VADD_8(v0, v0, v1);
             break;
+        case 0xFB:
+            INST_NAME("PSUBQ Gm, Em");
+            nextop = F8;
+            GETGM(v0);
+            GETEM(v1, 0);
+            SUB_64(v0, v0, v1);
+            break;
         case 0xFD:
             INST_NAME("PADDW Gm, Em");
             nextop = F8;

src/dynarec/arm64/dynarec_arm64_64.c

@@ -884,7 +884,7 @@ uintptr_t dynarec64_64(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     GETEDO(x6, 4);
                     i64 = F32S;
                     MOV64xw(x2, i64);
-                    emit_test32(dyn, ninst, rex, ed, x2, x3, x4);
+                    emit_test32(dyn, ninst, rex, ed, x2, x3, x4, x5);
                     break;
                 case 2:
                     INST_NAME("NOT Ed");

src/dynarec/arm64/dynarec_arm64_66.c

@@ -319,35 +319,78 @@ uintptr_t dynarec64_66(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             emit_dec16(dyn, ninst, x1, x2, x3);
             BFIw(gd, x1, 0, 16);
             break;
-
+        case 0x50:
+        case 0x51:
+        case 0x52:
+        case 0x53:
+        case 0x54:
+        case 0x55:
+        case 0x56:
+        case 0x57:
+            if(rex.is32bits) {
+                INST_NAME("PUSH reg");
+                gd = xRAX+(opcode&0x07);
+                if (gd == xRSP) {
+                    MOVw_REG(x1, xRSP);
+                    PUSH1_16(x1);
+                } else {
+                    PUSH1_16(gd);
+                }
+            } else {
+                DEFAULT;
+            }
+            break;
+        case 0x58:
+        case 0x59:
+        case 0x5A:
+        case 0x5B:
+        case 0x5C:
+        case 0x5D:
+        case 0x5E:
+        case 0x5F:
+            if(rex.is32bits) {
+                INST_NAME("POP reg");
+                gd = xRAX+(opcode&0x07);
+                POP1_16(x1);
+                BFIw(gd, x1, 0, 16);
+            } else {
+                DEFAULT;
+            }
+            break;
         case 0x60:
             if(rex.is32bits) {
-                INST_NAME("PUSHA 16 (32bits)");
+                INST_NAME("PUSHA 16bits (32bits)");
                 MOVw_REG(x1, xRSP);
-                STRH_S9_preindex(xRAX, xRSP, -2);
-                STRH_S9_preindex(xRCX, xRSP, -2);
-                STRH_S9_preindex(xRDX, xRSP, -2);
-                STRH_S9_preindex(xRBX, xRSP, -2);
-                STRH_S9_preindex(x1, xRSP, -2);
-                STRH_S9_preindex(xRBP, xRSP, -2);
-                STRH_S9_preindex(xRSI, xRSP, -2);
-                STRH_S9_preindex(xRDI, xRSP, -2);
+                PUSH1_16(xRAX);
+                PUSH1_16(xRCX);
+                PUSH1_16(xRDX);
+                PUSH1_16(xRBX);
+                PUSH1_16(x1);
+                PUSH1_16(xRBP);
+                PUSH1_16(xRSI);
+                PUSH1_16(xRDI);
             } else {
                 DEFAULT;
             }
             break;
         case 0x61:
             if(rex.is32bits) {
-                INST_NAME("POPA 16 (32bits)");
-                MOVw_REG(x1, xRSP);
-                LDRH_S9_postindex(xRDI, xRSP, 2);
-                LDRH_S9_postindex(xRSI, xRSP, 2);
-                LDRH_S9_postindex(xRBP, xRSP, 2);
-                LDRH_S9_postindex(x1, xRSP, 2);
-                LDRH_S9_postindex(xRBX, xRSP, 2);
-                LDRH_S9_postindex(xRDX, xRSP, 2);
-                LDRH_S9_postindex(xRCX, xRSP, 2);
-                LDRH_S9_postindex(xRAX, xRSP, 2);
+                INST_NAME("POPA 16bits (32bits)");
+                POP1_16(x1);
+                BFIw(xRDI, x1, 0, 16);
+                POP1_16(x1);
+                BFIw(xRSI, x1, 0, 16);
+                POP1_16(x1);
+                BFIw(xRBP, x1, 0, 16);
+                POP1_16(x1); // RSP ignored
+                POP1_16(x1);
+                BFIw(xRBX, x1, 0, 16);
+                POP1_16(x1);
+                BFIw(xRDX, x1, 0, 16);
+                POP1_16(x1);
+                BFIw(xRCX, x1, 0, 16);
+                POP1_16(x1);
+                BFIw(xRAX, x1, 0, 16);
             } else {
                 DEFAULT;
             }
@@ -589,7 +632,30 @@ uintptr_t dynarec64_66(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
         case 0x98:
             INST_NAME("CBW");
             SXTBw(x1, xRAX);
-            BFIw(xRAX, x1, 0, 16);
+            BFIx(xRAX, x1, 0, 16);
+            break;
+
+        case 0x9C:
+            INST_NAME("PUSHF");
+            READFLAGS(X_ALL);
+            PUSH1_16(xFlags);
+            break;
+        case 0x9D:
+            INST_NAME("POPF");
+            SETFLAGS(X_ALL, SF_SET);
+            POP1_16(x1);    // probably not usefull...
+            BFIw(xFlags, x1, 0, 16);
+            MOV32w(x1, 0x3F7FD7);
+            ANDw_REG(xFlags, xFlags, x1);
+            ORRw_mask(xFlags, xFlags, 0b011111, 0);   //mask=0x00000002
+            SET_DFNONE(x1);
+            if(box64_wine) {    // should this be done all the time?
+                TBZ_NEXT(xFlags, F_TF);
+                MOV64x(x1, addr);
+                STORE_XEMU_CALL(x1);
+                CALL(native_singlestep, -1);
+                BFCw(xFlags, F_TF, 1);
+            }
             break;
 
         case 0xA1:
@@ -599,6 +665,7 @@ uintptr_t dynarec64_66(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             else
                 u64 = F64;
             MOV64z(x1, u64);
+            SMREAD();
             LDRH_U12(x2, x1, 0);
             BFIx(xRAX, x2, 0, 16);
             break;
@@ -719,20 +786,23 @@ uintptr_t dynarec64_66(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                 CBZx_NEXT(xRCX);
                 TBNZ_MARK2(xFlags, F_DF);
                 MARK;   // Part with DF==0
-                LDRH_S9_postindex(xRAX, xRSI, 2);
+                LDRH_S9_postindex(x2, xRSI, 2);
                 SUBx_U12(xRCX, xRCX, 1);
                 CBNZx_MARK(xRCX);
-                B_NEXT_nocond;
+                B_MARK3_nocond;
                 MARK2;  // Part with DF==1
-                LDRH_S9_postindex(xRAX, xRSI, -2);
+                LDRH_S9_postindex(x2, xRSI, -2);
                 SUBx_U12(xRCX, xRCX, 1);
                 CBNZx_MARK2(xRCX);
+                MARK3;
+                BFIx(xRAX, x2, 0, 16);
                 // done
             } else {
-                INST_NAME("STOSW");
+                INST_NAME("LODSW");
                 GETDIR(x3, 2);
-                LDRH_U12(xRAX, xRSI, 0);
+                LDRH_U12(x2, xRSI, 0);
                 ADDx_REG(xRSI, xRSI, x3);
+                BFIx(xRAX, x2, 0, 16);
             }
             break;
 
@@ -1094,6 +1164,16 @@ uintptr_t dynarec64_66(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     emit_dec16(dyn, ninst, x1, x2, x4);
                     EWBACK;
                     break;
+                case 6: // Push Ew
+                    if(rex.is32bits) {
+                        INST_NAME("PUSH Ew");
+                        GETEW(x1, 0);
+                        PUSH1_16(ed);
+                    } else {
+                        DEFAULT;
+                    }
+                    break;
+
                 default:
                     DEFAULT;
             }

src/dynarec/arm64/dynarec_arm64_660f.c

@@ -747,7 +747,37 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
                         VEORQ(q0, q0, q1);
                     }
                     break;
-
+                case 0xF0:
+                    INST_NAME("MOVBE Gw, Ew");
+                    nextop=F8;
+                    gd = xRAX+((nextop&0x38)>>3)+(rex.r<<3);
+                    if(MODREG) {
+                        ed = xRAX+(nextop&7)+(rex.b<<3);
+                        REV16x(x1, ed);
+                        BFIx(gd, x1, 0, 16);
+                    } else {
+                        SMREAD();
+                        addr = geted(dyn, addr, ninst, nextop, &ed, x2, &fixedaddress, &unscaled, 0xfff<<2, (1<<2)-1, rex, NULL, 0, 0);
+                        LDH(x1, ed, fixedaddress);
+                        REV16x(x1, x1);
+                        BFIx(gd, x1, 0, 16);
+                    }
+                    break;
+                case 0xF1:
+                    INST_NAME("MOVBE Ew, Gw");
+                    nextop=F8;
+                    gd = xRAX+((nextop&0x38)>>3)+(rex.r<<3);
+                    if(MODREG) {
+                        ed = xRAX+(nextop&7)+(rex.b<<3);
+                        REV16x(x1, gd);
+                        BFIx(ed, x1, 0, 16);
+                    } else {
+                        SMREAD();
+                        addr = geted(dyn, addr, ninst, nextop, &ed, x2, &fixedaddress, &unscaled, 0xfff<<2, (1<<2)-1, rex, NULL, 0, 0);
+                        REV16x(x1, gd);
+                        STH(x1, ed, fixedaddress);
+                    }
+                    break;
                 default:
                     DEFAULT;
             }
@@ -1756,7 +1786,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<2, (1<<2)-1, rex, NULL, 0, 0);
-                SBFXw(x1, gd, 4, 12);   // r1 = (gw>>4)
+                SBFXx(x1, gd, 4, 12);   // r1 = (gw>>4)
                 ADDx_REG_LSL(x3, wback, x1, 1); //(&ed)+=r1*2;
                 LDH(x1, x3, fixedaddress);
                 ed = x1;
@@ -1797,9 +1827,10 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
                 wback = 0;
             } else {
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<2, (1<<2)-1, rex, NULL, 0, 0);
-                SBFXw(x4, gd, 4, 12);   // r1 = (gw>>4)
+                SBFXx(x4, gd, 4, 12);   // r1 = (gw>>4)
                 ADDx_REG_LSL(x3, wback, x4, 1); //(&ed)+=r1*2;
                 LDH(x4, x3, fixedaddress);
+                wback = x3;
                 ed = x4;
             }
             ANDw_mask(x2, gd, 0, 0b000011);  // mask=0x0f
@@ -1809,7 +1840,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
             B_NEXT(cNE);
             MOV32w(x1, 1);
             LSLxw_REG(x1, x1, x2);
-            EORx_REG(ed, ed, x1);
+            ORRx_REG(ed, ed, x1);
             if(wback) {
                 STRH_U12(ed, wback, fixedaddress);
             }
@@ -1859,7 +1890,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<2, (1<<2)-1, rex, NULL, 0, 0);
-                SBFXw(x4, gd, 4, 12);   // r1 = (gw>>4)
+                SBFXx(x4, gd, 4, 12);   // r1 = (gw>>4)
                 ADDx_REG_LSL(x3, wback, x4, 1); //(&ed)+=r1*2;
                 LDH(x4, x3, fixedaddress);
                 wback = x3;
@@ -1872,7 +1903,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
             B_NEXT(cEQ);
             MOV32w(x1, 1);
             LSLxw_REG(x1, x1, x2);
-            EORx_REG(ed, ed, x1);
+            BICx_REG(ed, ed, x1);
             if(wback) {
                 STH(ed, wback, fixedaddress);
                 SMWRITE();
@@ -1939,7 +1970,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
                     BFXILxw(xFlags, ed, u8, 1);  // inject 1 bit from u8 to F_CF (i.e. pos 0)
                     TBNZ_MARK3(xFlags, 0); // bit already set, jump to next instruction
                     MOV32w(x4, 1);
-                    EORxw_REG_LSL(ed, ed, x4, u8);
+                    ORRxw_REG_LSL(ed, ed, x4, u8);
                     EWBACK(x1);
                     MARK3;
                     break;
@@ -1953,7 +1984,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
                     BFXILxw(xFlags, ed, u8, 1);  // inject 1 bit from u8 to F_CF (i.e. pos 0)
                     TBZ_MARK3(xFlags, 0); // bit already clear, jump to next instruction
                     MOV32w(x4, 1);
-                    EORxw_REG_LSL(ed, ed, x4, u8);
+                    BICxw_REG_LSL(ed, ed, x4, u8);
                     EWBACK(x1);
                     MARK3;
                     break;
@@ -1985,7 +2016,7 @@ uintptr_t dynarec64_660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
             } else {
                 SMREAD();
                 addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, &unscaled, 0xfff<<2, (1<<2)-1, rex, NULL, 0, 0);
-                SBFXw(x4, gd, 4, 12);   // r1 = (gw>>4)
+                SBFXx(x4, gd, 4, 12);   // r1 = (gw>>4)
                 ADDx_REG_LSL(x3, wback, x4, 1); //(&ed)+=r1*2;
                 LDH(x4, x3, fixedaddress);
                 wback = x3;

src/dynarec/arm64/dynarec_arm64_66f0.c

@@ -152,7 +152,28 @@ uintptr_t dynarec64_66F0(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int n
                     DEFAULT;
             }
             break;
-
+        case 0x11:
+            INST_NAME("LOCK ADC Ew, Gw");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGW(x5);
+            SMDMB();
+            if(MODREG) {
+                ed = xRAX+(nextop&7)+(rex.b<<3);
+                UXTHw(x6, ed);
+                emit_adc16(dyn, ninst, x6, x5, x3, x4);
+                BFIx(ed, x6, 0, 16);
+            } else {
+                addr = geted(dyn, addr, ninst, nextop, &wback, x2, &fixedaddress, NULL, 0, 0, rex, LOCK_LOCK, 0, 0);
+                MARKLOCK;
+                LDAXRH(x1, wback);
+                emit_adc16(dyn, ninst, x1, x5, x3, x4);
+                STLXRH(x3, x1, wback);
+                CBNZx_MARKLOCK(x3);
+            }
+            SMDMB();
+            break;
         case 0x81:
         case 0x83:
             nextop = F8;

src/dynarec/arm64/dynarec_arm64_67.c

@@ -166,6 +166,28 @@ uintptr_t dynarec64_67(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
             opcode=F8;
             switch(opcode) {
 
+                case 0x11:
+                    switch(rep) {
+                        case 0:
+                            INST_NAME("MOVUPS Ex,Gx");
+                            nextop = F8;
+                            GETG;
+                            v0 = sse_get_reg(dyn, ninst, x1, gd, 0);
+                            if(MODREG) {
+                                ed = (nextop&7)+(rex.b<<3);
+                                v1 = sse_get_reg_empty(dyn, ninst, x1, ed);
+                                VMOVQ(v1, v0);
+                            } else {
+                                addr = geted32(dyn, addr, ninst, nextop, &ed, x1, &fixedaddress, &unscaled, 0xfff<<4, 15, rex, NULL, 0, 0);
+                                VST128(v0, ed, fixedaddress);
+                                SMWRITE2();
+                            }
+                            break;
+                        default:
+                            DEFAULT;
+                    }
+                    break;
+
                 case 0x2E:
                     // no special check...
                 case 0x2F:
@@ -208,6 +230,20 @@ uintptr_t dynarec64_67(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                                 VLD64(v0, ed, fixedaddress);
                             }
                             break;
+                        case 2:
+                            INST_NAME("MOVDQU Gx,Ex");
+                            nextop = F8;
+                            if(MODREG) {
+                                v1 = sse_get_reg(dyn, ninst, x1, (nextop&7)+(rex.b<<3), 0);
+                                GETGX_empty(v0);
+                                VMOVQ(v0, v1);
+                            } else {
+                                GETGX_empty(v0);
+                                SMREAD();
+                                addr = geted32(dyn, addr, ninst, nextop, &ed, x1, &fixedaddress, &unscaled, 0xfff<<4, 15, rex, NULL, 0, 0);
+                                VLD128(v0, ed, fixedaddress);
+                            }
+                            break;
                         default:
                             DEFAULT;
                     }
@@ -954,7 +990,7 @@ uintptr_t dynarec64_67(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     GETED32H(x1, 4);
                     i64 = F32S;
                     MOV64xw(x2, i64);
-                    emit_test32(dyn, ninst, rex, ed, x2, x3, x4);
+                    emit_test32(dyn, ninst, rex, ed, x2, x3, x4, x5);
                     break;
                 case 2:
                     INST_NAME("NOT Ed");

src/dynarec/arm64/dynarec_arm64_de.c

@@ -184,7 +184,8 @@ uintptr_t dynarec64_DE(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
         case 0xDD:
         case 0xDE:
         case 0xDF:
-            return 0;
+            DEFAULT;
+            break;
         default:
             switch((nextop>>3)&7) {
                 case 0:

src/dynarec/arm64/dynarec_arm64_emit_tests.c

@@ -265,8 +265,8 @@ void emit_cmp8_0(dynarec_arm_t* dyn, int ninst, int s1, int s3, int s4)
     }
 }
 
-// emit TEST32 instruction, from test s1, s2, using s3 and s4 as scratch
-void emit_test32(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, int s3, int s4)
+// emit TEST32 instruction, from test s1, s2, using s3, s4 and s5 as scratch
+void emit_test32(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, int s3, int s4, int s5)
 {
     MAYUSE(s1); MAYUSE(s2); MAYUSE(s3); MAYUSE(s4);
     IFX_PENDOR0 {
@@ -292,18 +292,11 @@ void emit_test32(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, int s
     }
     // PF: (((emu->x64emu_parity_tab[(res) / 32] >> ((res) % 32)) & 1) == 0)
     IFX(X_PF) {
-        ANDw_mask(s3, s3, 0b011011, 0b000010); // 0xE0
-        LSRw(s3, s3, 5);
-        MOV64x(s4, (uintptr_t)GetParityTab());
-        LDRw_REG_LSL2(s4, s4, s3);
-        ANDw_mask(s3, s1, 0, 0b000100);   // 0x1f
-        LSRw_REG(s4, s4, s3);
-        MVNx_REG(s4, s4);
-        BFIw(xFlags, s4, F_PF, 1);
+        emit_pf(dyn, ninst, s3, s4, s5);
     }
 }
 
-// emit TEST16 instruction, from test s1, s2, using s3 and s4 as scratch
+// emit TEST16 instruction, from test s1, s2, using s3, s4 and s5 as scratch
 void emit_test16(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3, int s4, int s5)
 {
     MAYUSE(s1); MAYUSE(s2);
@@ -334,7 +327,7 @@ void emit_test16(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3, int s4,
     }
 }
 
-// emit TEST8 instruction, from test s1, s2, using s3 and s4 as scratch
+// emit TEST8 instruction, from test s1, s2, using s3, s4 and s5 as scratch
 void emit_test8(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3, int s4, int s5)
 {
     MAYUSE(s1); MAYUSE(s2);

src/dynarec/arm64/dynarec_arm64_f0.c

@@ -173,8 +173,7 @@ uintptr_t dynarec64_F0(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                         BFIw(xFlags, x4, F_CF, 1);
                         MOV32w(x4, 1);
                         LSLxw_REG(x4, x4, x2);
-                        EORxw_REG(x4, ed, x4);
-                        CSELxw(ed, ed, x4, cNE);
+                        ORRxw_REG(ed, ed, x4);
                     } else {
                         // Will fetch only 1 byte, to avoid alignment issue
                         if(rex.w) {
@@ -191,12 +190,11 @@ uintptr_t dynarec64_F0(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                         ed = x1;
                         wback = x3;
                         LSRw_REG(x4, ed, x2);
-                        ANDSw_mask(x4, x4, 0, 0);  //mask=1
+                        ANDw_mask(x4, x4, 0, 0);  //mask=1
                         BFIw(xFlags, x4, F_CF, 1);
                         MOV32w(x4, 1);
                         LSLw_REG(x4, x4, x2);
-                        EORw_REG(x4, ed, x4);
-                        CSELw(ed, ed, x4, cNE);
+                        ORRw_REG(ed, ed, x4);
                         STLXRB(x4, ed, wback);
                         CBNZw_MARKLOCK(x4);
                         SMDMB();
@@ -322,15 +320,14 @@ uintptr_t dynarec64_F0(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                         }
                         LSRxw_REG(x4, ed, x2);
                         if(rex.w) {
-                            ANDSx_mask(x4, x4, 1, 0, 0);  //mask=1
+                            ANDx_mask(x4, x4, 1, 0, 0);  //mask=1
                         } else {
-                            ANDSw_mask(x4, x4, 0, 0);  //mask=1
+                            ANDw_mask(x4, x4, 0, 0);  //mask=1
                         }
                         BFIw(xFlags, x4, F_CF, 1);
                         MOV32w(x4, 1);
                         LSLxw_REG(x4, x4, x2);
-                        EORxw_REG(x4, ed, x4);
-                        CSELxw(ed, ed, x4, cEQ);
+                        BICxw_REG(ed, ed, x4);
                     } else {
                         // Will fetch only 1 byte, to avoid alignment issue
                         if(rex.w) {
@@ -340,19 +337,18 @@ uintptr_t dynarec64_F0(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                         }
                         SMDMB();
                         addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, NULL, 0, 0, rex, LOCK_LOCK, 0, 0);
-                        ASRxw(x1, gd, 3); // r1 = (gd>>3)
+                        ASRx(x1, gd, 3); // r1 = (gd>>3)
                         ADDx_REG_LSL(x3, wback, x1, 0); //(&ed)+=r1;
                         MARKLOCK;
                         LDAXRB(x1, wback);
                         ed = x1;
                         wback = x3;
                         LSRw_REG(x4, ed, x2);
-                        ANDSw_mask(x4, x4, 0, 0);  //mask=1
+                        ANDw_mask(x4, x4, 0, 0);  //mask=1
                         BFIw(xFlags, x4, F_CF, 1);
                         MOV32w(x4, 1);
                         LSLw_REG(x4, x4, x2);
-                        EORw_REG(x4, ed, x4);
-                        CSELw(ed, ed, x4, cEQ);
+                        ORRw_REG(ed, ed, x4);
                         STLXRB(x4, ed, wback);
                         CBNZw_MARKLOCK(x4);
                         SMDMB();
@@ -440,7 +436,55 @@ uintptr_t dynarec64_F0(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nin
                     DEFAULT;
             }
             break;
-
+        case 0x10:
+            INST_NAME("LOCK ADC Eb, Gb");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            SMDMB();
+            GETGB(x2);
+            if((nextop&0xC0)==0xC0) {
+                if(rex.rex) {
+                    wback = xRAX + (nextop&7) + (rex.b<<3);
+                    wb2 = 0;
+                } else {
+                    wback = (nextop&7);
+                    wb2 = (wback>>2);
+                    wback = xRAX+(wback&3);
+                }
+                UBFXw(x1, wback, wb2*8, 8);
+                emit_adc8(dyn, ninst, x1, x2, x4, x5);
+                BFIx(wback, x1, wb2*8, 8);
+            } else {
+                addr = geted(dyn, addr, ninst, nextop, &wback, x3, &fixedaddress, NULL, 0, 0, rex, LOCK_LOCK, 0, 0);
+                MARKLOCK;
+                LDAXRB(x1, wback);
+                emit_adc8(dyn, ninst, x1, x2, x4, x5);
+                STLXRB(x4, x1, wback);
+                CBNZx_MARKLOCK(x4);
+            }
+            SMDMB();
+            break;
+        case 0x11:
+            INST_NAME("LOCK ADC Ed, Gd");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGD;
+            SMDMB();
+                if(MODREG) {
+                ed = xRAX+(nextop&7)+(rex.b<<3);
+                emit_adc32(dyn, ninst, rex, ed, gd, x3, x4);
+            } else {
+                addr = geted(dyn, addr, ninst, nextop, &wback, x2, &fixedaddress, NULL, 0, 0, rex, LOCK_LOCK, 0, 0);
+                MARKLOCK;
+                LDAXRxw(x1, wback);
+                emit_adc32(dyn, ninst, rex, x1, gd, x4, x5);
+                STLXRxw(x4, x1, wback);
+                CBNZx_MARKLOCK(x4);
+            }
+            SMDMB();
+            break;
         case 0x21:
             INST_NAME("LOCK AND Ed, Gd");
             SETFLAGS(X_ALL, SF_SET_PENDING);

src/dynarec/arm64/dynarec_arm64_functions.h

@@ -36,7 +36,7 @@ void neoncache_promote_double(dynarec_arm_t* dyn, int ninst, int a);
 // Combine and propagate if needed (pass 1 only)
 int neoncache_combine_st(dynarec_arm_t* dyn, int ninst, int a, int b);  // with stack current dyn->n_stack*
 
-// FPU Cache transformation (for loops) // Specific, need to be written par backend
+// FPU Cache transformation (for loops) // Specific, need to be written by backend
 int fpuCacheNeedsTransform(dynarec_arm_t* dyn, int ninst);
 
 // Undo the changes of a neoncache to get the status before the instruction

src/dynarec/arm64/dynarec_arm64_helper.c

@@ -477,8 +477,8 @@ uintptr_t geted16(dynarec_arm_t* dyn, uintptr_t addr, int ninst, uint8_t nextop,
     int64_t offset = 0;
     int absmin = 0;
     if(s) absmin = -absmax;
-    if(!n && m==6) {
-        offset = F16;
+    if(!n && (m&7)==6) {
+        offset = F16S;
         MOVZw(ret, offset);
     } else {
         switch(n) {

src/dynarec/arm64/dynarec_arm64_helper.h

@@ -34,12 +34,12 @@
 
 // Strong mem emulation helpers
 // Sequence of Read will trigger a DMB on "first" read if strongmem is 2
-// Squence of Write will trigger a DMB on "last" write if strongmem is 1
+// Sequence of Write will trigger a DMB on "last" write if strongmem is 1
 // Opcode will read
 #define SMREAD()    if(!dyn->smread && box64_dynarec_strongmem>1) {SMDMB();}
 // Opcode will read with option forced lock
 #define SMREADLOCK(lock)    if(lock || (!dyn->smread && box64_dynarec_strongmem>1)) {SMDMB();}
-// Opcode migh read (depend on nextop)
+// Opcode might read (depend on nextop)
 #define SMMIGHTREAD()   if(!MODREG) {SMREAD();}
 // Opcode has wrote
 #define SMWRITE()   dyn->smwrite=1
@@ -47,7 +47,7 @@
 #define SMWRITE2()   if(box64_dynarec_strongmem>1) dyn->smwrite=1
 // Opcode has wrote with option forced lock
 #define SMWRITELOCK(lock)   if(lock) {SMDMB();} else dyn->smwrite=1
-// Opcode migh have wrote (depend on nextop)
+// Opcode might have wrote (depend on nextop)
 #define SMMIGHTWRITE()   if(!MODREG) {SMWRITE();}
 // Start of sequence
 #define SMSTART()   SMEND()
@@ -360,18 +360,18 @@
                         gb2 = 0;                \
                     } else {                    \
                         gd = (nextop&0x38)>>3;  \
-                        gb2 = ((gd&4)>>2);      \
+                        gb2 = ((gd&4)<<1);      \
                         gb1 = xRAX+(gd&3);      \
                     }                           \
                     gd = i;                     \
-                    UBFXx(gd, gb1, gb2*8, 8);
-//GETSGB signe extend GB, will use i for gd
+                    UBFXx(gd, gb1, gb2, 8);
+//GETSGB sign extend GB, will use i for gd
 #define GETSGB(i)   if(rex.rex) {               \
                         gb1 = xRAX+((nextop&0x38)>>3)+(rex.r<<3);     \
                         gb2 = 0;                \
                     } else {                    \
                         gd = (nextop&0x38)>>3;  \
-                        gb2 = ((gd&4)>>2);      \
+                        gb2 = ((gd&4)<<1);      \
                         gb1 = xRAX+(gd&3);      \
                     }                           \
                     gd = i;                     \
@@ -1035,7 +1035,7 @@ void emit_cmp16_0(dynarec_arm_t* dyn, int ninst, int s1, int s3, int s4);
 void emit_cmp32_0(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s3, int s4);
 void emit_test8(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3, int s4, int s5);
 void emit_test16(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3, int s4, int s5);
-void emit_test32(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, int s3, int s4);
+void emit_test32(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, int s3, int s4, int s5);
 void emit_add32(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, int s3, int s4);
 void emit_add32c(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int64_t c, int s3, int s4, int s5);
 void emit_add8(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3, int s4);
@@ -1100,7 +1100,7 @@ void emit_shld32c(dynarec_arm_t* dyn, int ninst, rex_t rex, int s1, int s2, uint
 void emit_pf(dynarec_arm_t* dyn, int ninst, int s1, int s3, int s4);
 
 // x87 helper
-// cache of the local stack counter, to avoid upadte at every call
+// cache of the local stack counter, to avoid update at every call
 void x87_stackcount(dynarec_arm_t* dyn, int ninst, int scratch);
 // fpu push. Return the Dd value to be used
 int x87_do_push(dynarec_arm_t* dyn, int ninst, int s1, int t);

src/dynarec/arm64/dynarec_arm64_private.h

@@ -77,8 +77,8 @@ typedef struct instruction_arm64_s {
     uintptr_t           natcall;
     int                 retn;
     int                 barrier_maybe;
-    flagcache_t         f_exit;     // flags status at end of intruction
-    neoncache_t         n;          // neoncache at end of intruction (but before poping)
+    flagcache_t         f_exit;     // flags status at end of instruction
+    neoncache_t         n;          // neoncache at end of instruction (but before poping)
     flagcache_t         f_entry;    // flags status before the instruction begin
 } instruction_arm64_t;
 
@@ -87,12 +87,12 @@ typedef struct dynarec_arm_s {
     int32_t             size;
     int32_t             cap;
     uintptr_t           start;      // start of the block
-    uint32_t            isize;      // size in byte of x64 instructions included
+    uint32_t            isize;      // size in bytes of x64 instructions included
     void*               block;      // memory pointer where next instruction is emitted
     uintptr_t           native_start;  // start of the arm code
     size_t              native_size;   // size of emitted arm code
     uintptr_t           last_ip;    // last set IP in RIP (or NULL if unclean state) TODO: move to a cache something
-    uint64_t*           table64;   // table of 64bits value
+    uint64_t*           table64;    // table of 64bits values
     int                 table64size;// size of table (will be appended at end of executable code)
     int                 table64cap;
     uintptr_t           tablestart;
@@ -106,7 +106,7 @@ typedef struct dynarec_arm_s {
     dynablock_t*        dynablock;
     instsize_t*         instsize;
     size_t              insts_size; // size of the instruction size array (calculated)
-    uint8_t             smread;    // for strongmem model emulation
+    uint8_t             smread;     // for strongmem model emulation
     uint8_t             smwrite;    // for strongmem model emulation
     uintptr_t           forward;    // address of the last end of code while testing forward
     uintptr_t           forward_to; // address of the next jump to (to check if everything is ok)
@@ -122,7 +122,7 @@ uintptr_t get_closest_next(dynarec_arm_t *dyn, uintptr_t addr);
 int is_nops(dynarec_arm_t *dyn, uintptr_t addr, int n);
 int is_instructions(dynarec_arm_t *dyn, uintptr_t addr, int n);
 
-int Table64(dynarec_arm_t *dyn, uint64_t val, int pass);  // add a value to etable64 (if needed) and gives back the imm19 to use in LDR_literal
+int Table64(dynarec_arm_t *dyn, uint64_t val, int pass);  // add a value to table64 (if needed) and gives back the imm19 to use in LDR_literal
 
 void CreateJmpNext(void* addr, void* next);
 

src/dynarec/dynablock.c

@@ -209,7 +209,7 @@ static dynablock_t* internalDBGetBlock(x64emu_t* emu, uintptr_t addr, uintptr_t
     // fill the block
     block->x64_addr = (void*)addr;
     if(sigsetjmp(&dynarec_jmpbuf, 1)) {
-        printf_log(LOG_INFO, "FillBlock at %p triggered a segfault, cancelling\n", (void*)addr);
+        printf_log(LOG_INFO, "FillBlock at %p triggered a segfault, canceling\n", (void*)addr);
         FreeDynablock(block, 0);
         if(need_lock)
             mutex_unlock(&my_context->mutex_dyndump);

src/dynarec/dynarec.c

@@ -73,114 +73,65 @@ void* LinkNext(x64emu_t* emu, uintptr_t addr, void* x2, uintptr_t* x3)
 }
 #endif
 
-#ifdef __GNUC__
-// Disable "clobbered" warnings
-#pragma GCC diagnostic push
-#pragma GCC diagnostic ignored "-Wclobbered"
-#endif
 void DynaCall(x64emu_t* emu, uintptr_t addr)
 {
-    // prepare setjump for signal handling
-    struct __jmp_buf_tag jmpbuf[1] = {0};
-    int skip = 0;
-    struct __jmp_buf_tag *old_jmpbuf = emu->jmpbuf;
-    emu->jmpbuf = jmpbuf;
-
-    if((skip = sigsetjmp(emu->jmpbuf, 1))) {
-        printf_log(LOG_DEBUG, "Setjmp DynaCall, fs=0x%x\n", emu->segs[_FS]);
-        addr = R_RIP;   // not sure if it should still be inside DynaCall!
-        #ifdef DYNAREC
-        if(box64_dynarec_test) {
-            if(emu->test.clean)
-                x64test_check(emu, R_RIP);
-            emu->test.clean = 0;
-        }
-        #endif
+    uint64_t old_rsp = R_RSP;
+    uint64_t old_rbx = R_RBX;
+    uint64_t old_rdi = R_RDI;
+    uint64_t old_rsi = R_RSI;
+    uint64_t old_rbp = R_RBP;
+    uint64_t old_rip = R_RIP;
+    PushExit(emu);
+    R_RIP = addr;
+    emu->df = d_none;
+    DynaRun(emu);
+    emu->quit = 0;  // reset Quit flags...
+    emu->df = d_none;
+    if(emu->flags.quitonlongjmp && emu->flags.longjmp) {
+        if(emu->flags.quitonlongjmp==1)
+            emu->flags.longjmp = 0;   // don't change anything because of the longjmp
+    } else {
+        R_RBX = old_rbx;
+        R_RDI = old_rdi;
+        R_RSI = old_rsi;
+        R_RBP = old_rbp;
+        R_RSP = old_rsp;
+        R_RIP = old_rip;  // and set back instruction pointer
     }
-#ifdef DYNAREC
-    if(!box64_dynarec)
-#endif
-        EmuCall(emu, addr);
-#ifdef DYNAREC
-    else {
-        uint64_t old_rsp = R_RSP;
-        uint64_t old_rbx = R_RBX;
-        uint64_t old_rdi = R_RDI;
-        uint64_t old_rsi = R_RSI;
-        uint64_t old_rbp = R_RBP;
-        uint64_t old_rip = R_RIP;
-        PushExit(emu);
-        R_RIP = addr;
-        emu->df = d_none;
-        while(!emu->quit) {
-            int is32bits = (emu->segs[_CS]==0x23);
-            dynablock_t* block = (skip)?NULL:DBGetBlock(emu, R_RIP, 1, is32bits);
-            if(!block || !block->block || !block->done) {
-                skip = 0;
-                // no block, of block doesn't have DynaRec content (yet, temp is not null)
-                // Use interpreter (should use single instruction step...)
-                dynarec_log(LOG_DEBUG, "%04d|Calling Interpreter @%p, emu=%p\n", GetTID(), (void*)R_RIP, emu);
-                if(box64_dynarec_test)
-                    emu->test.clean = 0;
-                Run(emu, 1);
-            } else {
-                dynarec_log(LOG_DEBUG, "%04d|Calling DynaRec Block @%p (%p) of %d x64 instructions (hash=0x%x) emu=%p\n", GetTID(), (void*)R_RIP, block->block, block->isize ,block->hash, emu);
-                CHECK_FLAGS(emu);
-                // block is here, let's run it!
-                native_prolog(emu, block->block);
-            }
-            if(emu->fork) {
-                int forktype = emu->fork;
-                emu->quit = 0;
-                emu->fork = 0;
-                emu = x64emu_fork(emu, forktype);
-            }
-        }
-        emu->quit = 0;  // reset Quit flags...
-        emu->df = d_none;
-        if(emu->quitonlongjmp && emu->longjmp) {
-            if(emu->quitonlongjmp==1)
-                emu->longjmp = 0;   // don't change anything because of the longjmp
-        } else {
-            R_RBX = old_rbx;
-            R_RDI = old_rdi;
-            R_RSI = old_rsi;
-            R_RBP = old_rbp;
-            R_RSP = old_rsp;
-            R_RIP = old_rip;  // and set back instruction pointer
-        }
-    }
-#endif
-    // clear the setjmp
-    emu->jmpbuf = old_jmpbuf;
 }
 
-int DynaRun(x64emu_t* emu)
+void DynaRun(x64emu_t* emu)
 {
     // prepare setjump for signal handling
     struct __jmp_buf_tag jmpbuf[1] = {0};
     int skip = 0;
     struct __jmp_buf_tag *old_jmpbuf = emu->jmpbuf;
-    emu->jmpbuf = jmpbuf;
+    emu->flags.jmpbuf_ready = 0;
 
-    if((skip=sigsetjmp(emu->jmpbuf, 1))) {
-        printf_log(LOG_DEBUG, "Setjmp DynaRun, fs=0x%x\n", emu->segs[_FS]);
-        #ifdef DYNAREC
-        if(box64_dynarec_test) {
-            if(emu->test.clean)
-                x64test_check(emu, R_RIP);
-            emu->test.clean = 0;
+    while(!(emu->quit)) {
+        if(!emu->jmpbuf || (emu->flags.need_jmpbuf && emu->jmpbuf!=jmpbuf)) {
+            emu->jmpbuf = jmpbuf;
+            emu->flags.jmpbuf_ready = 1;
+            if((skip=sigsetjmp(emu->jmpbuf, 1))) {
+                printf_log(LOG_DEBUG, "Setjmp DynaRun, fs=0x%x\n", emu->segs[_FS]);
+                #ifdef DYNAREC
+                if(box64_dynarec_test) {
+                    if(emu->test.clean)
+                        x64test_check(emu, R_RIP);
+                    emu->test.clean = 0;
+                }
+                #endif
+            }
         }
-        #endif
-    }
+        if(emu->flags.need_jmpbuf)
+            emu->flags.need_jmpbuf = 0;
 
 #ifdef DYNAREC
-    if(!box64_dynarec)
+        if(!box64_dynarec)
 #endif
-        return Run(emu, 0);
+            Run(emu, 0);
 #ifdef DYNAREC
-    else {
-        while(!emu->quit) {
+        else {
             int is32bits = (emu->segs[_CS]==0x23);
             dynablock_t* block = (skip)?NULL:DBGetBlock(emu, R_RIP, 1, is32bits);
             if(!block || !block->block || !block->done) {
@@ -203,12 +154,10 @@ int DynaRun(x64emu_t* emu)
                 emu = x64emu_fork(emu, forktype);
             }
         }
+#endif
+        if(emu->flags.need_jmpbuf)
+            emu->quit = 0;
     }
     // clear the setjmp
     emu->jmpbuf = old_jmpbuf;
-    return 0;
-#endif
 }
-#ifdef __GNUC__
-#pragma GCC diagnostic pop
-#endif

src/dynarec/dynarec_native.c

@@ -78,7 +78,7 @@ void add_next(dynarec_native_t *dyn, uintptr_t addr) {
     dyn->next[dyn->next_sz++] = addr;
 }
 uintptr_t get_closest_next(dynarec_native_t *dyn, uintptr_t addr) {
-    // get closest, but no addresses befores
+    // get closest, but no addresses before
     uintptr_t best = 0;
     int i = 0;
     while((i<dyn->next_sz) && (best!=addr)) {
@@ -124,8 +124,8 @@ int is_nops(dynarec_native_t *dyn, uintptr_t addr, int n)
     return 0;
 }
 
-// return size of next instuciton, -1 is unknown
-// not all instrction are setup
+// return size of next instruction, -1 is unknown
+// not all instructions are setup
 int next_instruction(dynarec_native_t *dyn, uintptr_t addr)
 {
     uint8_t opcode = PK(0);
@@ -293,7 +293,7 @@ int Table64(dynarec_native_t *dyn, uint64_t val, int pass)
 static void fillPredecessors(dynarec_native_t* dyn)
 {
     int pred_sz = 1;    // to be safe
-    // compute total size of predecessor to alocate the array
+    // compute total size of predecessor to allocate the array
     // first compute the jumps
     for(int i=0; i<dyn->size; ++i) {
         if(dyn->insts[i].x64.jmp && (dyn->insts[i].x64.jmp_insts!=-1)) {
@@ -301,6 +301,13 @@ static void fillPredecessors(dynarec_native_t* dyn)
             dyn->insts[dyn->insts[i].x64.jmp_insts].pred_sz++;
         }
     }
+    // remove "has_next" from orphan branch
+    for(int i=0; i<dyn->size-1; ++i) {
+        if(!dyn->insts[i].x64.has_next) {
+            if(dyn->insts[i+1].x64.has_next && !dyn->insts[i+1].pred_sz)
+                dyn->insts[i+1].x64.has_next = 0;
+        }
+    }
     // second the "has_next"
     for(int i=0; i<dyn->size-1; ++i) {
         if(dyn->insts[i].x64.has_next) {
@@ -328,7 +335,7 @@ static void fillPredecessors(dynarec_native_t* dyn)
     }
 }
 
-// updateNeed goes backward, from last intruction to top
+// updateNeed goes backward, from last instruction to top
 static int updateNeed(dynarec_native_t* dyn, int ninst, uint8_t need) {
     while (ninst>=0) {
         // need pending but instruction is only a subset: remove pend and use an X_ALL instead
@@ -368,9 +375,9 @@ static int updateNeed(dynarec_native_t* dyn, int ninst, uint8_t need) {
             else
                 updateNeed(dyn, dyn->insts[ninst].pred[i], need);
         }
-        if(!ok)
-            return ninst - 1;
         --ninst;
+        if(!ok)
+            return ninst;
     }
     return ninst;
 }
@@ -410,7 +417,7 @@ void* CreateEmptyBlock(dynablock_t* block, uintptr_t addr) {
     void* actual_p = (void*)AllocDynarecMap(sz);
     void* p = actual_p + sizeof(void*);
     if(actual_p==NULL) {
-        dynarec_log(LOG_INFO, "AllocDynarecMap(%p, %zu) failed, cancelling block\n", block, sz);
+        dynarec_log(LOG_INFO, "AllocDynarecMap(%p, %zu) failed, canceling block\n", block, sz);
         CancelBlock64(0);
         return NULL;
     }
@@ -437,11 +444,11 @@ void* FillBlock64(dynablock_t* block, uintptr_t addr, int alternate, int is32bit
         B+8 ..  B+15    : 2 Native code for jmpnext (or jmp epilog in case of empty block)
         B+16 .. B+23    : jmpnext (or jmp_epilog) address
         B+24 .. B+31    : empty (in case an architecture needs more than 2 opcodes)
-        B+32 .. B+32+sz : instsize (compressed array with each instruction lenght on x64 and native side)
+        B+32 .. B+32+sz : instsize (compressed array with each instruction length on x64 and native side)
 
     */
     if(IsInHotPage(addr)) {
-        dynarec_log(LOG_DEBUG, "Cancelling dynarec FillBlock on hotpage for %p\n", (void*)addr);
+        dynarec_log(LOG_DEBUG, "Canceling dynarec FillBlock on hotpage for %p\n", (void*)addr);
         return NULL;
     }
     if(addr>=box64_nodynarec_start && addr<box64_nodynarec_end) {
@@ -449,7 +456,7 @@ void* FillBlock64(dynablock_t* block, uintptr_t addr, int alternate, int is32bit
         return CreateEmptyBlock(block, addr);
     }
     if(current_helper) {
-        dynarec_log(LOG_DEBUG, "Cancelling dynarec FillBlock at %p as anothor one is going on\n", (void*)addr);
+        dynarec_log(LOG_DEBUG, "Canceling dynarec FillBlock at %p as another one is going on\n", (void*)addr);
         return NULL;
     }
     // protect the 1st page
@@ -513,7 +520,7 @@ void* FillBlock64(dynablock_t* block, uintptr_t addr, int alternate, int is32bit
     while (pos>=0)
         pos = updateNeed(&helper, pos, 0);
 
-    // pass 1, float optimisations, first pass for flags
+    // pass 1, float optimizations, first pass for flags
     native_pass1(&helper, addr, alternate, is32bits);
     
     // pass 2, instruction size
@@ -531,7 +538,7 @@ void* FillBlock64(dynablock_t* block, uintptr_t addr, int alternate, int is32bit
     void* next = tablestart + helper.table64size*sizeof(uint64_t);
     void* instsize = next + 4*sizeof(void*);
     if(actual_p==NULL) {
-        dynarec_log(LOG_INFO, "AllocDynarecMap(%p, %zu) failed, cancelling block\n", block, sz);
+        dynarec_log(LOG_INFO, "AllocDynarecMap(%p, %zu) failed, canceling block\n", block, sz);
         CancelBlock64(0);
         return NULL;
     }
@@ -581,7 +588,7 @@ void* FillBlock64(dynablock_t* block, uintptr_t addr, int alternate, int is32bit
     // all done...
     __clear_cache(actual_p, actual_p+sz);   // need to clear the cache before execution...
     block->hash = X31_hash_code(block->x64_addr, block->x64_size);
-    // Check if something changed, to abbort if it as
+    // Check if something changed, to abort if it is
     if((block->hash != hash)) {
         dynarec_log(LOG_DEBUG, "Warning, a block changed while being processed hash(%p:%ld)=%x/%x\n", block->x64_addr, block->x64_size, block->hash, hash);
         AddHotPage(addr);
@@ -604,7 +611,7 @@ void* FillBlock64(dynablock_t* block, uintptr_t addr, int alternate, int is32bit
         return NULL;
     }
     if(insts_rsize/sizeof(instsize_t)<helper.insts_size) {
-        printf_log(LOG_NONE, "BOX64: Warning, ists_size difference in block between pass2 (%zu) and pass3 (%zu), allocated: %zu\n", oldinstsize, helper.insts_size, insts_rsize/sizeof(instsize_t));
+        printf_log(LOG_NONE, "BOX64: Warning, insts_size difference in block between pass2 (%zu) and pass3 (%zu), allocated: %zu\n", oldinstsize, helper.insts_size, insts_rsize/sizeof(instsize_t));
     }
     if(!isprotectedDB(addr, end-addr)) {
         dynarec_log(LOG_DEBUG, "Warning, block unprotected while being processed %p:%ld, marking as need_test\n", block->x64_addr, block->x64_size);

src/dynarec/dynarec_native_functions.c

@@ -138,7 +138,7 @@ void native_fistp64(x64emu_t* emu, int64_t* ed)
 
 void native_fistt64(x64emu_t* emu, int64_t* ed)
 {
-    // used of memcpy to avoid aligments issues
+    // used of memcpy to avoid alignments issues
     int64_t tmp = ST0.d;
     memcpy(ed, &tmp, sizeof(tmp));
 }
@@ -163,6 +163,12 @@ void native_priv(x64emu_t* emu)
     emit_signal(emu, SIGSEGV, (void*)R_RIP, 0);
 }
 
+void native_int(x64emu_t* emu, int num)
+{
+    emu->test.test = 0;
+    emit_interruption(emu, num, (void*)R_RIP);
+}
+
 void native_singlestep(x64emu_t* emu)
 {
     emit_signal(emu, SIGTRAP, (void*)R_RIP, 1);

src/dynarec/dynarec_native_functions.h

@@ -46,6 +46,7 @@ void native_clflush(x64emu_t* emu, void* p);
 void native_ud(x64emu_t* emu);
 void native_priv(x64emu_t* emu);
 void native_singlestep(x64emu_t* emu);
+void native_int(x64emu_t* emu, int num);
 
 // Caches transformation (for loops) // Specific, need to be written par backend
 int CacheNeedsTransform(dynarec_native_t* dyn, int i1);

src/dynarec/dynarec_native_pass.c

@@ -47,7 +47,7 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
     ARCH_INIT();
     int reset_n = -1;
     dyn->last_ip = (alternate || (dyn->insts && dyn->insts[0].pred_sz))?0:ip;  // RIP is always set at start of block unless there is a predecessor!
-    int stopblock = 2+(FindElfAddress(my_context, addr)?0:1); // if block is in elf_memory, it can be extended with bligblocks==2, else it needs 3
+    int stopblock = 2+(FindElfAddress(my_context, addr)?0:1); // if block is in elf_memory, it can be extended with box64_dynarec_bigblock==2, else it needs 3
     // ok, go now
     INIT;
     while(ok) {
@@ -74,7 +74,7 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
             }
             reset_n = -1;
         } else if(ninst && (dyn->insts[ninst].pred_sz>1 || (dyn->insts[ninst].pred_sz==1 && dyn->insts[ninst].pred[0]!=ninst-1)))
-            dyn->last_ip = 0;   // reset IP if some jump are comming here
+            dyn->last_ip = 0;   // reset IP if some jump are coming here
         fpu_propagate_stack(dyn, ninst);
         NEW_INST;
         if(!ninst) {
@@ -166,7 +166,7 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
             if((dyn->insts[ii].x64.barrier&BARRIER_FULL)==BARRIER_FULL)
                 reset_n = -2;    // hack to say Barrier!
             else {
-                reset_n = getNominalPred(dyn, ii);  // may get -1 if no predecessor are availble
+                reset_n = getNominalPred(dyn, ii);  // may get -1 if no predecessor are available
                 if(reset_n==-1) {
                     reset_n = -2;
                     MESSAGE(LOG_DEBUG, "Warning, Reset Caches mark not found\n");
@@ -183,7 +183,7 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
                 dyn->forward_size = 0;
                 dyn->forward_ninst = 0;
                 ok = 1; // in case it was 0
-            } else if ((dyn->forward_to < addr) || !ok) {
+            } else if ((dyn->forward_to < addr) || ok<=0) {
                 // something when wrong! rollback
                 if(box64_dynarec_dump) dynarec_log(LOG_NONE, "Could not forward extend block for %d bytes %p -> %p\n", dyn->forward_to-dyn->forward, (void*)dyn->forward, (void*)dyn->forward_to);
                 ok = 0;
@@ -214,12 +214,19 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
                     if(box64_dynarec_dump) dynarec_log(LOG_NONE, "Extend block %p, %p -> %p (ninst=%d, jump from %d)\n", dyn, (void*)addr, (void*)next, ninst, reset_n);
                 } else if(next && (next-addr)<box64_dynarec_forward && (getProtection(next)&PROT_READ)/*box64_dynarec_bigblock>=stopblock*/) {
                     if(!((box64_dynarec_bigblock<stopblock) && !isJumpTableDefault64((void*)next))) {
-                        dyn->forward = addr;
-                        dyn->forward_to = next;
-                        dyn->forward_size = dyn->size;
-                        dyn->forward_ninst = ninst;
-                        reset_n = -2;
-                        ok = 1;
+                        if(dyn->forward) {
+                            if(next<dyn->forward_to)
+                                dyn->forward_to = next;
+                            reset_n = -2;
+                            ok = 1;
+                        } else {
+                            dyn->forward = addr;
+                            dyn->forward_to = next;
+                            dyn->forward_size = dyn->size;
+                            dyn->forward_ninst = ninst;
+                            reset_n = -2;
+                            ok = 1;
+                        }
                     }
                 }
             }
@@ -235,6 +242,16 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
                 dyn->insts[ninst].x64.need_after |= X_PEND;
                 ++ninst;
             }
+            if(dyn->forward) {
+                // stopping too soon
+                dyn->size = dyn->forward_size;
+                ninst = dyn->forward_ninst+1;
+                addr = dyn->forward;
+                dyn->forward = 0;
+                dyn->forward_to = 0;
+                dyn->forward_size = 0;
+                dyn->forward_ninst = 0;
+            }
             #endif
         }
         ++ninst;
@@ -279,7 +296,7 @@ uintptr_t native_pass(dynarec_native_t* dyn, uintptr_t addr, int alternate, int
     if(need_epilog) {
         NOTEST(x3);
         fpu_purgecache(dyn, ninst, 0, x1, x2, x3);
-        jump_to_epilog(dyn, ip, 0, ninst);  // no linker here, it's an unknow instruction
+        jump_to_epilog(dyn, ip, 0, ninst);  // no linker here, it's an unknown instruction
     }
     FINI;
     MESSAGE(LOG_DUMP, "---- END OF BLOCK ---- (%d)\n", dyn->size);

src/dynarec/rv64/dynarec_rv64_00.c

@@ -25,8 +25,6 @@
 #include "dynarec_rv64_functions.h"
 #include "dynarec_rv64_helper.h"
 
-int isSimpleWrapper(wrapper_t fun);
-
 uintptr_t dynarec64_00(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ninst, rex_t rex, int rep, int* ok, int* need_epilog)
 {
     uint8_t opcode;

src/dynarec/rv64/dynarec_rv64_00_0.c

@@ -25,8 +25,6 @@
 #include "dynarec_rv64_functions.h"
 #include "dynarec_rv64_helper.h"
 
-int isSimpleWrapper(wrapper_t fun);
-
 uintptr_t dynarec64_00_0(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ninst, rex_t rex, int rep, int* ok, int* need_epilog)
 {
     uint8_t nextop, opcode;
@@ -169,7 +167,16 @@ uintptr_t dynarec64_00_0(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
                 DEFAULT;
             }
             break;
-
+        case 0x10:
+            INST_NAME("ADC Eb, Gb");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETEB(x1, 0);
+            GETGB(x2);
+            emit_adc8(dyn, ninst, x1, x2, x4, x5, x3);
+            EBBACK(x5, 0);
+            break;
         case 0x11:
             INST_NAME("ADC Ed, Gd");
             READFLAGS(X_CF);
@@ -180,7 +187,43 @@ uintptr_t dynarec64_00_0(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
             emit_adc32(dyn, ninst, rex, ed, gd, x3, x4, x5, x6);
             WBACK;
             break;
-
+        case 0x12:
+            INST_NAME("ADC Gb, Eb");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETEB(x2, 0);
+            GETGB(x1);
+            emit_adc8(dyn, ninst, x1, x2, x4, x3, x5);
+            GBBACK(x5);
+            break;
+        case 0x13:
+            INST_NAME("ADC Gd, Ed");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGD;
+            GETED(0);
+            emit_adc32(dyn, ninst, rex, gd, ed, x3, x4, x5, x6);
+            break;
+        case 0x14:
+            INST_NAME("ADC AL, Ib");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            u8 = F8;
+            ANDI(x1, xRAX, 0xff);
+            emit_adc8c(dyn, ninst, x1, u8, x3, x4, x5, x6);
+            ANDI(xRAX, xRAX, -256);
+            OR(xRAX, xRAX, x1);
+            break;
+        case 0x15:
+            INST_NAME("ADC EAX, Id");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            i64 = F32S;
+            MOV64xw(x1, i64);
+            emit_adc32(dyn, ninst, rex, xRAX, x1, x3, x4, x5, x6);
+            break;
         case 0x18:
             INST_NAME("SBB Eb, Gb");
             READFLAGS(X_CF);

src/dynarec/rv64/dynarec_rv64_00_2.c

@@ -25,8 +25,6 @@
 #include "dynarec_rv64_functions.h"
 #include "dynarec_rv64_helper.h"
 
-int isSimpleWrapper(wrapper_t fun);
-
 uintptr_t dynarec64_00_2(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ninst, rex_t rex, int rep, int* ok, int* need_epilog)
 {
     uint8_t nextop, opcode;

src/dynarec/rv64/dynarec_rv64_00_3.c

@@ -26,6 +26,7 @@
 #include "dynarec_rv64_helper.h"
 
 int isSimpleWrapper(wrapper_t fun);
+int isRetX87Wrapper(wrapper_t fun);
 
 uintptr_t dynarec64_00_3(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ninst, rex_t rex, int rep, int* ok, int* need_epilog)
 {
@@ -307,6 +308,9 @@ uintptr_t dynarec64_00_3(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
                     // disabling isSimpleWrapper because all signed value less than 64bits needs to be sign extended
                     // and return value needs to be cleanned up
                     tmp = 0;//isSimpleWrapper(*(wrapper_t*)(addr));
+                    if(isRetX87Wrapper(*(wrapper_t*)(addr)))
+                        // return value will be on the stack, so the stack depth needs to be updated
+                        x87_purgecache(dyn, ninst, 0, x3, x1, x4);
                     if(tmp<0 || tmp>1)
                         tmp=0;  //TODO: removed when FP is in place
                     if((box64_log<2 && !cycle_log) && tmp) {
@@ -617,6 +621,58 @@ uintptr_t dynarec64_00_3(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
         case 0xDF:
             addr = dynarec64_DF(dyn, addr, ip, ninst, rex, rep, ok, need_epilog);
             break;
+        #define GO(Z)                                                   \
+            BARRIER(BARRIER_MAYBE);                                     \
+            JUMP(addr+i8, 1);                                           \
+            if(dyn->insts[ninst].x64.jmp_insts==-1 ||                   \
+                CHECK_CACHE()) {                                        \
+                /* out of the block */                                  \
+                i32 = dyn->insts[ninst].epilog-(dyn->native_size);      \
+                if(Z) {BNE(xRCX, xZR, i32);} else {BEQ(xRCX, xZR, i32);};      \
+                if(dyn->insts[ninst].x64.jmp_insts==-1) {               \
+                    if(!(dyn->insts[ninst].x64.barrier&BARRIER_FLOAT))  \
+                        fpu_purgecache(dyn, ninst, 1, x1, x2, x3);      \
+                    jump_to_next(dyn, addr+i8, 0, ninst);               \
+                } else {                                                \
+                    CacheTransform(dyn, ninst, cacheupd, x1, x2, x3);   \
+                    i32 = dyn->insts[dyn->insts[ninst].x64.jmp_insts].address-(dyn->native_size);    \
+                    B(i32);                                             \
+                }                                                       \
+            } else {                                                    \
+                /* inside the block */                                  \
+                i32 = dyn->insts[dyn->insts[ninst].x64.jmp_insts].address-(dyn->native_size);    \
+                if(Z) {BEQ(xRCX, xZR, i32);} else {BNE(xRCX, xZR, i32);};      \
+            }
+        case 0xE0:
+            INST_NAME("LOOPNZ");
+            READFLAGS(X_ZF);
+            i8 = F8S;
+            SUBI(xRCX, xRCX, 1);
+            ANDI(x1, xFlags, 1 << F_ZF);
+            CBNZ_NEXT(x1);
+            GO(0);
+            break;
+        case 0xE1:
+            INST_NAME("LOOPZ");
+            READFLAGS(X_ZF);
+            i8 = F8S;
+            SUBI(xRCX, xRCX, 1);
+            ANDI(x1, xFlags, 1 << F_ZF);
+            CBZ_NEXT(x1);
+            GO(0);
+            break;
+        case 0xE2:
+            INST_NAME("LOOP");
+            i8 = F8S;
+            SUBI(xRCX, xRCX, 1);
+            GO(0);
+            break;
+        case 0xE3:
+            INST_NAME("JECXZ");
+            i8 = F8S;
+            GO(1);
+            break;
+        #undef GO
 
         case 0xE8:
             INST_NAME("CALL Id");
@@ -657,6 +713,9 @@ uintptr_t dynarec64_00_3(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
                             tmp=0;  // float paramters not ready!
                     } else
                         tmp=0;
+                    if(dyn->insts[ninst].natcall && isRetX87Wrapper(*(wrapper_t*)(dyn->insts[ninst].natcall+2)))
+                        // return value will be on the stack, so the stack depth needs to be updated
+                        x87_purgecache(dyn, ninst, 0, x3, x1, x4);
                     if((box64_log<2 && !cycle_log) && dyn->insts[ninst].natcall && tmp) {
                         //GETIP(ip+3+8+8); // read the 0xCC
                         call_n(dyn, ninst, *(void**)(dyn->insts[ninst].natcall+2+8), tmp);

src/dynarec/rv64/dynarec_rv64_0f.c

@@ -316,7 +316,134 @@ uintptr_t dynarec64_0F(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ni
             SRLI(xRDX, x3, 32);
             AND(xRAX, x3, 32);   // wipe upper part
             break;
+        case 0x38:
+            //SSE3
+            nextop=F8;
+            switch(nextop) {
+                case 0xF0:
+                    INST_NAME("MOVBE Gd, Ed");
+                    nextop=F8;
+                    GETGD;
+                    SMREAD();
+                    addr = geted(dyn, addr, ninst, nextop, &ed, x2, x1, &fixedaddress, rex, NULL, 1, 0);
+                    LDxw(gd, ed, fixedaddress);
+                    if (rv64_zbb) {
+                        REV8(gd, gd);
+                        if (!rex.w) {
+                            SRLI(gd, gd, 32);
+                        }
+                    } else {
+                        if (rex.w) {
+                            MOV_U12(x2, 0xff);
+                            SLLI(x1, gd, 56);
+                            SRLI(x3, gd, 56);
+                            SRLI(x4, gd, 40);
+                            SLLI(x2, x2, 8);
+                            AND(x4, x4, x2);
+                            OR(x1, x1, x3);
+                            OR(x1, x1, x4);
+                            SLLI(x3, gd, 40);
+                            SLLI(x4, x2, 40);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
 
+                            SRLI(x3, gd, 24);
+                            SLLI(x4, x2, 8);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
+                            SLLI(x3, gd, 24);
+                            SLLI(x4, x2, 32);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
+
+                            SRLI(x3, gd, 8);
+                            SLLI(x4, x2, 16);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);                     
+                            SLLI(x3, gd, 8);
+                            SLLI(x4, x2, 24);
+                            AND(x3, x3, x4);
+                            OR(gd, x1, x3);
+                        } else {
+                            MOV_U12(x2, 0xff);
+                            SLLIW(x2, x2, 8);
+                            SLLIW(x1, gd, 24);
+                            SRLIW(x3, gd, 24);
+                            SRLIW(x4, gd, 8);
+                            AND(x4, x4, x2);
+                            OR(x1, x1, x3);
+                            OR(x1, x1, x4);
+                            SLLIW(gd, gd, 8);
+                            LUI(x2, 0xff0);
+                            AND(gd, gd, x2);
+                            OR(gd, gd, x1);
+                        }                        
+                    }
+                    break;
+                case 0xF1:
+                    INST_NAME("MOVBE Ed, Gd");
+                    nextop=F8;
+                    GETGD;
+                    SMREAD();
+                    addr = geted(dyn, addr, ninst, nextop, &wback, x2, x1, &fixedaddress, rex, NULL, 1, 0);
+                    if (rv64_zbb) {
+                        REV8(x1, gd);
+                        if (!rex.w) {
+                            SRLI(x1, x1, 32);
+                        }
+                    } else {
+                        if (rex.w) {
+                            MOV_U12(x2, 0xff);
+                            SLLI(x1, gd, 56);
+                            SRLI(x3, gd, 56);
+                            SRLI(x4, gd, 40);
+                            SLLI(x2, x2, 8);
+                            AND(x4, x4, x2);
+                            OR(x1, x1, x3);
+                            OR(x1, x1, x4);
+                            SLLI(x3, gd, 40);
+                            SLLI(x4, x2, 40);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
+
+                            SRLI(x3, gd, 24);
+                            SLLI(x4, x2, 8);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
+                            SLLI(x3, gd, 24);
+                            SLLI(x4, x2, 32);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
+
+                            SRLI(x3, gd, 8);
+                            SLLI(x4, x2, 16);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);                     
+                            SLLI(x3, gd, 8);
+                            SLLI(x4, x2, 24);
+                            AND(x3, x3, x4);
+                            OR(x1, x1, x3);
+                        } else {
+                            MOV_U12(x2, 0xff);
+                            SLLIW(x2, x2, 8);
+                            SLLIW(x1, gd, 24);
+                            SRLIW(x3, gd, 24);
+                            SRLIW(x4, gd, 8);
+                            AND(x4, x4, x2);
+                            OR(x1, x1, x3);
+                            OR(x1, x1, x4);
+                            SLLIW(x3, gd, 8);
+                            LUI(x2, 0xff0);
+                            AND(x3, x3, x2);
+                            OR(x1, x1, x3);
+                        }
+                    }
+                    SDxw(x1, wback, fixedaddress);
+                    break;
+                default:
+                    DEFAULT;
+            }
+            break;
 
         #define GO(GETFLAGS, NO, YES, F)            \
             READFLAGS(F);                           \

src/dynarec/rv64/dynarec_rv64_66.c

@@ -117,6 +117,43 @@ uintptr_t dynarec64_66(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ni
         case 0x0F:
             addr = dynarec64_660F(dyn, addr, ip, ninst, rex, ok, need_epilog);
             break;
+        case 0x11:
+            INST_NAME("ADC Ew, Gw");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGW(x2);
+            GETEW(x1, 0);
+            emit_adc16(dyn, ninst, x1, x2, x4, x3, x5);
+            EWBACK;
+            break;
+        case 0x13:
+            INST_NAME("ADC Gw, Ew");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGW(x1);
+            GETEW(x2, 0);
+            emit_adc16(dyn, ninst, x1, x2, x4, x3, x5);
+            GWBACK;
+            break;
+        case 0x15:
+            INST_NAME("ADC AX, Iw");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            i16 = F16;
+            SRLI(x6, xMASK, 16);
+            AND(x1, xRAX, x6);
+            MOV32w(x2, i16);
+            emit_adc16(dyn, ninst, x1, x2, x3, x4, x5);
+            if (rv64_zbb) {
+                ANDN(xRAX, xRAX, x6);
+            } else {
+                NOT(x6, x6);
+                AND(xRAX, xRAX, x6);
+            }
+            OR(xRAX, xRAX, x1);
+            break;
         case 0x19:
             INST_NAME("SBB Ew, Gw");
             READFLAGS(X_CF);
@@ -427,6 +464,67 @@ uintptr_t dynarec64_66(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ni
             GETGW(x2);
             emit_test16(dyn, ninst, x1, x2, x3, x4, x5);
             break;
+        case 0x87:
+            INST_NAME("(LOCK)XCHG Ew, Gw");
+            nextop = F8;
+            if(MODREG) {
+                GETGD;
+                GETED(0);
+                ADDI(x1, gd, 0);
+                LUI(x3, 0xffff0);
+                AND(gd, gd, x3);
+                ZEXTH(x4, ed);
+                OR(gd, gd, x4);
+                AND(ed, ed, x3);
+                ZEXTH(x4, x1);
+                OR(ed, ed, x4);
+            } else {
+                GETGD;
+                addr = geted(dyn, addr, ninst, nextop, &ed, x2, x1, &fixedaddress, rex, LOCK_LOCK, 0, 0);
+                SMDMB();
+
+                ANDI(x3, ed, 1);
+                BNEZ_MARK(x3);
+
+                ANDI(x3, ed, 0b10);
+                LUI(x5, 0xffff0);
+                ZEXTH(x6, gd);
+                BNEZ_MARK3(x3);
+
+                MARKLOCK;
+                LR_W(x1, ed, 1, 0);
+                AND(x3, x1, x5);
+                OR(x3, x3, x6);
+                SC_W(x3, x3, ed, 0, 1);
+                BNEZ_MARKLOCK(x3);
+                ZEXTH(x1, x1);
+                B_MARK2_nocond;
+
+                MARK3;
+                NOT(x5, x5);
+                SLLI(x5, x5, 16);
+                SLLI(x6, x6, 16);
+                NOT(x5, x5);
+                LR_W(x1, ed, 1, 0);
+                AND(x3, x1, x5);
+                OR(x3, x3, x6);
+                SC_W(x3, x3, ed, 0, 1);
+                BNEZ(x3, -4 * 4);
+                SRLI(x1, x1, 16);
+                ZEXTH(x1, x1);
+                B_MARK2_nocond;
+
+                MARK;
+                LHU(x1, ed, 0);
+                SH(gd, ed, 0);
+
+                MARK2;
+                SMDMB();
+                LUI(x5, 0xffff0);
+                AND(gd, gd, x5);
+                OR(gd, gd, x1);
+            }
+            break;
         case 0x89:
             INST_NAME("MOV Ew, Gw");
             nextop = F8;

src/dynarec/rv64/dynarec_rv64_660f.c

@@ -411,7 +411,7 @@ uintptr_t dynarec64_660F(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
                             AND(x4, x4, x6);
                             OR(x3, x3, x4);
                             BNEZ(x3, 8);
-                            ORI(xFlags, xFlags, 1<<F_ZF);
+                            ORI(xFlags, xFlags, 1<<F_CF);
                         }
                     }
                     break;
@@ -715,6 +715,44 @@ uintptr_t dynarec64_660F(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
                     GETEX(x2, 0);
                     SSE_LOOP_Q(x3, x4, XOR(x3, x3, x4));
                     break;
+                case 0xF0:
+                    INST_NAME("MOVBE Gw, Ew");
+                    nextop=F8;
+                    GETGD;
+                    SMREAD();
+                    addr = geted(dyn, addr, ninst, nextop, &ed, x3, x2, &fixedaddress, rex, NULL, 0, 1);
+                    LHU(x1, ed, fixedaddress);
+                    if (rv64_zbb) {
+                        REV8(x1, x1);
+                        SRLI(x1, x1, 48);
+                    } else {
+                        ANDI(x2, x1, 0xff);
+                        SLLI(x2, x2, 8);
+                        SRLI(x1, x1, 8);
+                        OR(x1, x1, x2);
+                    }
+                    LUI(x2, 0xffff0);
+                    AND(gd, gd, x2);
+                    OR(gd, gd, x1);                    
+                    break;
+                case 0xF1:
+                    INST_NAME("MOVBE Ew, Gw");
+                    nextop=F8;
+                    GETGD;
+                    SMREAD();
+                    addr = geted(dyn, addr, ninst, nextop, &wback, x3, x2, &fixedaddress, rex, NULL, 0, 1);  
+                    if (rv64_zbb) {
+                        REV8(x1, gd);
+                        SRLI(x1, x1, 48);
+                    } else {
+                        ANDI(x1, gd, 0xff);
+                        SLLI(x1, x1, 8);
+                        SRLI(x2, gd, 8);
+                        ANDI(x2, x2, 0xff);
+                        OR(x1, x1, x2);
+                    }
+                    SH(x1, wback, fixedaddress);
+                    break;
                 default:
                     DEFAULT;
             }

src/dynarec/rv64/dynarec_rv64_6664.c

@@ -37,6 +37,35 @@ uintptr_t dynarec64_6664(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
     GETREX();
 
     switch(opcode) {
+        case 0x89:
+            INST_NAME("MOV FS:Ew, Gw");
+            nextop = F8;
+            GETGD;  // don't need GETGW here
+            if(MODREG) {
+                ed = xRAX+(nextop&7)+(rex.b<<3);
+                if(rex.w) {
+                    ADDI(ed, gd, 0);
+                } else {
+                    if(ed!=gd) {
+                        LUI(x1, 0xffff0);
+                        AND(gd, gd, x1);
+                        ZEXTH(x1, ed);
+                        OR(gd, gd, x1);
+                    }
+                }
+            } else {
+                grab_segdata(dyn, addr, ninst, x4, seg);
+                addr = geted(dyn, addr, ninst, nextop, &ed, x2, x1, &fixedaddress, rex, NULL, 1, 0);
+                ADD(x4, ed, x4);
+                if(rex.w) {
+                    SD(gd, x4, fixedaddress);
+                } else {
+                    SH(gd, x4, fixedaddress);
+                }
+                SMWRITE();
+            }
+            break;
+
         case 0x8B:
             INST_NAME("MOV Gd, FS:Ed");
             nextop=F8;
@@ -69,7 +98,6 @@ uintptr_t dynarec64_6664(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
             }
             break;
 
-
         default:
             DEFAULT;
     }

src/dynarec/rv64/dynarec_rv64_66f0.c

@@ -8,6 +8,7 @@
 #include "dynarec.h"
 #include "emu/x64emu_private.h"
 #include "emu/x64run_private.h"
+#include "rv64_emitter.h"
 #include "x64run.h"
 #include "x64emu.h"
 #include "box64stack.h"
@@ -48,6 +49,71 @@ uintptr_t dynarec64_66F0(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int
     GETREX();
 
     switch(opcode) {
+        case 0x11:
+            INST_NAME("LOCK ADC Ew, Gw");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGW(x1);
+            SMDMB();
+            if(MODREG) {
+                ed = xRAX+(nextop&7)+(rex.b<<3);
+                MOV32w(x6, 0xFFFF);
+                AND(x2, ed, x6);
+                emit_adc16(dyn, ninst, x1, x2, x3, x4, x5);
+                if (rv64_zbb) {
+                    ANDN(ed, ed, x6);
+                } else {
+                    NOT(x6, x6);
+                    AND(ed, ed, x6);
+                }
+                OR(ed, ed, x2);
+            } else {
+                addr = geted(dyn, addr, ninst, nextop, &wback, x2, x3, &fixedaddress, rex, LOCK_LOCK, 0, 0);
+                ANDI(x3, wback, 0b10);
+                MOV32w(x4, 0xffff); // x4 = mask
+                BNEZ_MARK(x3);
+                // lower 16bits
+                MARKLOCK;
+                LR_W(x5, wback, 1, 1);
+                AND(x6, x5, x4);        // x6 = Ed.h[0]
+                SRLIW(x5, x5, 16);
+                SLLIW(x5, x5, 16);      // x5 = clear Ed.h[0]
+                ADDW(x2, x1, x6);       // x2 = Gw + Ew
+                ANDI(x9, xFlags, 1 << F_CF);
+                ADDW(x2, x2, x9);       // x2 = adc
+                AND(x2, x2, x4);
+                OR(x5, x5, x2);
+                SC_W(x2, x5, wback, 1, 1);
+                BNEZ_MARKLOCK(x2);
+                B_MARK3_nocond;
+                MARK;
+                // upper 16bits
+                SLLIW(x3, x4, 16);     // x3 = mask
+                SUBI(wback, wback, 2); // aligning address
+                SLLI(x1, x1, 16);      // x1 = extended Gw
+                MARK2;
+                LR_W(x6, wback, 1, 1); // x6 = Ed
+                AND(x5, x6, x3);       // x5 = Ed.h[1] << 16
+                ADDW(x5,x5, x1);
+                ANDI(x9, xFlags, 1 << F_CF);
+                SLLIW(x9, x9, 16);
+                ADDW(x5, x5, x9);      // x5 = adc << 16
+                AND(x9, x6, x4);       // x9 = Ed.h[0]
+                OR(x5, x5, x9);
+                SC_W(x9, x5, wback, 1, 1);
+                BNEZ_MARK2(x9);
+                IFX(X_ALL|X_PEND) {
+                    SRLIW(x6, x6, 16);
+                    SRLIW(x1, x1, 16);
+                }
+                MARK3;
+                IFX(X_ALL|X_PEND) {
+                    emit_adc16(dyn, ninst, x1, x6, x3, x4, x5);
+                }
+            }
+            SMDMB();
+            break;
         case 0x81:
         case 0x83:
             nextop = F8;

src/dynarec/rv64/dynarec_rv64_67.c

@@ -567,6 +567,63 @@ uintptr_t dynarec64_67(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ni
                 }
             }
             break;
+        #define GO(Z)                                                   \
+            BARRIER(BARRIER_MAYBE);                                     \
+            JUMP(addr+i8, 1);                                           \
+            if(dyn->insts[ninst].x64.jmp_insts==-1 ||                   \
+                CHECK_CACHE()) {                                        \
+                /* out of the block */                                  \
+                i32 = dyn->insts[ninst].epilog-(dyn->native_size);      \
+                if(Z) {BNE(x1, xZR, i32);} else {BEQ(x1, xZR, i32);};   \
+                if(dyn->insts[ninst].x64.jmp_insts==-1) {               \
+                    if(!(dyn->insts[ninst].x64.barrier&BARRIER_FLOAT))  \
+                        fpu_purgecache(dyn, ninst, 1, x1, x2, x3);      \
+                    jump_to_next(dyn, addr+i8, 0, ninst);               \
+                } else {                                                \
+                    CacheTransform(dyn, ninst, cacheupd, x1, x2, x3);   \
+                    i32 = dyn->insts[dyn->insts[ninst].x64.jmp_insts].address-(dyn->native_size);    \
+                    B(i32);                                             \
+                }                                                       \
+            } else {                                                    \
+                /* inside the block */                                  \
+                i32 = dyn->insts[dyn->insts[ninst].x64.jmp_insts].address-(dyn->native_size);    \
+                if(Z) {BEQ(x1, xZR, i32);} else {BNE(x1, xZR, i32);};   \
+            }
+        case 0xE0:
+            INST_NAME("LOOPNZ (32bits)");
+            READFLAGS(X_ZF);
+            i8 = F8S;
+            SUBI(xRCX, xRCX, 1);
+            ANDI(x1, xFlags, 1 << F_ZF);
+            CBNZ_NEXT(x1);
+            AND(x1, xRCX, xMASK);
+            GO(0);
+            break;
+        case 0xE1:
+            INST_NAME("LOOPZ (32bits)");
+            READFLAGS(X_ZF);
+            i8 = F8S;
+            SUBI(xRCX, xRCX, 1);
+            ANDI(x1, xFlags, 1 << F_ZF);
+            CBZ_NEXT(x1);
+            AND(x1, xRCX, xMASK);
+            GO(0);
+            break;
+        case 0xE2:
+            INST_NAME("LOOP (32bits)");
+            i8 = F8S;
+            SUBI(xRCX, xRCX, 1);
+            AND(x1, xRCX, xMASK);
+            GO(0);
+            break;
+        case 0xE3:
+            INST_NAME("JECXZ (32bits)");
+            i8 = F8S;
+            AND(x1, xRCX, xMASK);
+            GO(1);
+            break;
+        #undef GO
+
         default:
             DEFAULT;
     }

src/dynarec/rv64/dynarec_rv64_f0.c

@@ -339,6 +339,101 @@ uintptr_t dynarec64_F0(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ni
                     DEFAULT;
             }
             break;
+        case 0x10:
+            INST_NAME("LOCK ADC Eb, Gb");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGB(x2);
+            SMDMB();
+            if((nextop&0xC0)==0xC0) {
+                if(rex.rex) {
+                    wback = xRAX + (nextop&7) + (rex.b<<3);
+                    wb2 = 0;
+                } else {
+                    wback = (nextop&7);
+                    wb2 = (wback>>2);
+                    wback = xRAX+(wback&3);
+                }
+                SRLIW(x1, wback, wb2*8);
+                ANDI(x1, x1, 0xFF);
+                emit_adc8(dyn, ninst, x1, x2, x3, x4, x5);
+                SLLI(x1, x1, wb2*8);
+                MOV_U12(x3, 0xFF);
+                SLLI(x3, x3, wb2*8);
+                NOT(x3, x3);
+                AND(wback, wback, x3);
+                OR(wback, wback, x1);
+            } else {
+                addr = geted(dyn, addr, ninst, nextop, &wback, x1, x3, &fixedaddress, rex, LOCK_LOCK, 0, 0);
+                ANDI(x3, wback, 0b11);
+                BNEZ_MARK(x3);
+                MARKLOCK;
+                LR_W(x5, wback, 1, 1);
+                ANDI(x4, x5, 0xff); // x4 = Ed.b[0]
+                ANDI(x5, x5, -256); // x5 = clear Ed.b[0]
+                ADDW(x6, x4, x2);
+                ANDI(x9, xFlags, 1 << F_CF);
+                ADDW(x6, x6, x9);   // x6 = adc
+                ANDI(x6, x6, 0xff);
+                OR(x5, x5, x6);
+                SC_W(x9, x5, wback, 1, 1);
+                BNEZ_MARKLOCK(x9);
+                B_MARK3_nocond;
+                MARK;
+                SLLI(x3, x3, 3);
+                MOV_U12(x4, 0xff);
+                ANDI(wback, wback, ~3); // aligning address
+                SLLI(x4, x4, x3);       // x4 = byte mask
+                NOT(x5, x4);            // x5 = ~mask
+                SLL(x2, x2, x3);        // x2 = extented Gb
+                MARK2;
+                LR_W(x6, wback, 1, 1);  // x6 = Ed
+                AND(x9, x6, x4);        // x9 = extended Ed.b[dest]
+                AND(x6, x6, x5);        // x6 = clear Ed.b[dest]
+                ADDW(x5, x9, x2);
+                ANDI(x4, xFlags, 1 << F_CF);
+                SLL(x4, x4, x3);        // extented
+                ADDW(x5, x5, x4);       // x5 = adc
+                OR(x5, x5, x6);
+                SC_W(x4, x5, wback, 1, 1);
+                BNEZ_MARK2(x4);
+                IFX(X_ALL|X_PEND) {
+                    SRLI(x2, x2, x3);  // Gb
+                    SRLI(x4, x9, x3);  // Eb
+                }
+                MARK3;
+                IFX(X_ALL|X_PEND) {
+                    emit_adc8(dyn, ninst, x4, x2, x3, x5, x6);
+                }
+            }
+            SMDMB();
+            break;
+        case 0x11:
+            INST_NAME("LOCK ADC Ed, Gd");
+            READFLAGS(X_CF);
+            SETFLAGS(X_ALL, SF_SET_PENDING);
+            nextop = F8;
+            GETGD;
+            SMDMB();
+            if(MODREG) {
+                ed = xRAX+(nextop&7)+(rex.b<<3);
+                emit_adc32(dyn, ninst, rex, ed, gd, x3, x4, x5, x6);
+            } else {
+                addr = geted(dyn, addr, ninst, nextop, &wback, x2, x1, &fixedaddress, rex, LOCK_LOCK, 0, 0);
+                MARKLOCK;
+                LRxw(x1, wback, 1, 1);
+                ADDxw(x3, x1, gd);
+                ANDI(x4, xFlags, 1 << F_CF);
+                ADDxw(x3, x3, x4);
+                SCxw(x4, x3, wback, 1, 1);
+                BNEZ_MARKLOCK(x4);
+                IFX(X_ALL|X_PEND) {
+                    emit_adc32(dyn, ninst, rex, x1, gd, x3, x4, x5, x6);
+                }
+            }
+            SMDMB();
+            break;
         case 0x21:
             INST_NAME("LOCK AND Ed, Gd");
             SETFLAGS(X_ALL, SF_SET_PENDING);
@@ -381,6 +476,9 @@ uintptr_t dynarec64_F0(dynarec_rv64_t* dyn, uintptr_t addr, uintptr_t ip, int ni
             }
             SMDMB();
             break;
+        case 0x66:
+            return dynarec64_66F0(dyn, addr, ip, ninst, rex, rep, ok, need_epilog);
+
         case 0x80:
             nextop = F8;
             SMDMB();

src/dynarec/rv64/dynarec_rv64_private.h

@@ -135,7 +135,7 @@ uintptr_t get_closest_next(dynarec_rv64_t *dyn, uintptr_t addr);
 int is_nops(dynarec_rv64_t *dyn, uintptr_t addr, int n);
 int is_instructions(dynarec_rv64_t *dyn, uintptr_t addr, int n);
 
-int Table64(dynarec_rv64_t *dyn, uint64_t val, int pass);  // add a value to etable64 (if needed) and gives back the imm19 to use in LDR_literal
+int Table64(dynarec_rv64_t *dyn, uint64_t val, int pass);  // add a value to table64 (if needed) and gives back the imm19 to use in LDR_literal
 
 void CreateJmpNext(void* addr, void* next);
 

src/dynarec/rv64/rv64_emitter.h

@@ -128,6 +128,7 @@ f28–31  ft8–11  FP temporaries                  Caller
 // RV32I
 // put imm20 in the [31:12] bits of rd, zero [11:0] and sign extend bits31
 #define LUI(rd, imm20)                 EMIT(U_type((imm20)<<12, rd, 0b0110111))
+
 // put PC+imm20 in rd
 #define AUIPC(rd, imm20)               EMIT(U_type((imm20)<<12, rd, 0b0010111))
 

src/dynarec/rv64/rv64_lock.h

@@ -2,49 +2,49 @@
 #define __RV64_LOCK__H__
 #include <stdint.h>
 
-// Atomicaly store val at [p] if old [p] is ref. Return 0 if OK, 1 is not. p needs to be aligned
+// Atomically store val at [p] if old [p] is ref. Return 0 if OK, 1 is not. p needs to be aligned
 extern int rv64_lock_cas_d(void* p, int32_t ref, int32_t val);
 
-// Atomicaly store val at [p] if old [p] is ref. Return 0 if OK, 1 is not. p needs to be aligned
+// Atomically store val at [p] if old [p] is ref. Return 0 if OK, 1 is not. p needs to be aligned
 extern int rv64_lock_cas_dd(void* p, int64_t ref, int64_t val);
 
-// Atomicaly exchange value at [p] with val, return old p
+// Atomically exchange value at [p] with val, return old p
 extern uintptr_t rv64_lock_xchg_dd(void* p, uintptr_t val);
 
-// Atomicaly exchange value at [p] with val, return old p
+// Atomically exchange value at [p] with val, return old p
 extern uint32_t rv64_lock_xchg_d(void* p, uint32_t val);
 
-// Atomicaly store value to [p] only if [p] is NULL. Return old [p] value
+// Atomically store value to [p] only if [p] is NULL. Return old [p] value
 extern uint32_t rv64_lock_storeifnull_d(void*p, uint32_t val);
 
-// Atomicaly store value to [p] only if [p] is NULL. Return old [p] value
+// Atomically store value to [p] only if [p] is NULL. Return old [p] value
 extern void* rv64_lock_storeifnull(void*p, void* val);
 
-// Atomicaly store value to [p] only if [p] is ref. Return new [p] value (so val or old)
+// Atomically store value to [p] only if [p] is ref. Return new [p] value (so val or old)
 extern void* rv64_lock_storeifref(void*p, void* val, void* ref);
 
-// Atomicaly store value to [p] only if [p] is ref. Return new [p] value (so val or old)
+// Atomically store value to [p] only if [p] is ref. Return new [p] value (so val or old)
 extern uint32_t rv64_lock_storeifref_d(void*p, uint32_t val, uint32_t ref);
 
-// Atomicaly store value to [p] only if [p] is ref. Return new [p] value (so val or old)
+// Atomically store value to [p] only if [p] is ref. Return new [p] value (so val or old)
 extern uint32_t rv64_lock_storeifref2_d(void*p, uint32_t val, uint32_t ref);
 
-// decrement atomicaly the byte at [p] (but only if p not 0)
+// decrement atomically the byte at [p] (but only if p not 0)
 extern void rv64_lock_decifnot0b(void*p);
 
 // atomic store (with memory barrier)
 extern void rv64_lock_storeb(void*p, uint8_t b);
 
-// increment atomicaly the int at [p] only if it was 0. Return the old value of [p]
+// increment atomically the int at [p] only if it was 0. Return the old value of [p]
 extern int rv64_lock_incif0(void*p);
 
-// decrement atomicaly the int at [p] (but only if p not 0)
+// decrement atomically the int at [p] (but only if p not 0)
 extern int rv64_lock_decifnot0(void*p);
 
 // atomic store (with memory barrier)
 extern void rv64_lock_store(void*p, uint32_t v);
 
-// (mostly) Atomicaly store val1 and val2 at [p] if old [p] is ref. Return 0 if OK, 1 is not. p needs to be aligned
+// (mostly) Atomically store val1 and val2 at [p] if old [p] is ref. Return 0 if OK, 1 is not. p needs to be aligned
 extern int rv64_lock_cas_dq(void* p, uint64_t ref, uint64_t val1, uint64_t val2);
 
 // Not defined in assembler but in dynarec_rv64_functions

src/elfs/elfloader.c

@@ -311,8 +311,10 @@ void FreeElfMemory(elfheader_t* head)
         for(int i=0; i<head->multiblock_n; ++i) {
 #ifdef DYNAREC
             dynarec_log(LOG_INFO, "Free DynaBlocks for %s\n", head->path);
-            cleanDBFromAddressRange((uintptr_t)head->multiblock[i], head->multiblock_size[i], 1);
-            freeProtection((uintptr_t)head->multiblock[i], head->multiblock_size[i]);
+            if(box64_dynarec) {
+                cleanDBFromAddressRange((uintptr_t)head->multiblock[i], head->multiblock_size[i], 1);
+                freeProtection((uintptr_t)head->multiblock[i], head->multiblock_size[i]);
+            }
 #endif
             munmap(head->multiblock[i], head->multiblock_size[i]);
         }
@@ -374,6 +376,23 @@ int LoadElfMemory(FILE* f, box64context_t* context, elfheader_t* head)
     return 0;
 }
 
+int isElfHasNeededVer(elfheader_t* head, const char* libname, elfheader_t* verneeded)
+{
+    if(!verneeded || !head)
+        return 1;
+    if(!head->VerDef || !verneeded->VerNeed)
+        return 1;
+    int cnt = GetNeededVersionCnt(verneeded, libname);
+    for (int i=0; i<cnt; ++i) {
+        const char* vername = GetNeededVersionString(verneeded, libname, i);
+        if(vername && !GetVersionIndice(head, vername)) {
+            printf_log(/*LOG_DEBUG*/LOG_INFO, "Discarding %s for missing version %s\n", head->path, vername);
+            return 0;   // missing version
+        }
+    }
+    return 1;
+}
+
 int ReloadElfMemory(FILE* f, box64context_t* context, elfheader_t* head)
 {
     (void)context;
@@ -471,11 +490,20 @@ int RelocateElfREL(lib_t *maplib, lib_t *local_maplib, int bindnow, elfheader_t*
                 end = offs + sym->st_size;
             }*/
             // so weak symbol are the one left
-            if(!offs && !end) {
-                if(!offs && !end && local_maplib)
-                    GetGlobalSymbolStartEnd(local_maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
-                if(!offs && !end && local_maplib)
-                    GetGlobalSymbolStartEnd(maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+            if(bind==STB_WEAK) {
+                if(!offs && !end) {
+                    if(!offs && !end && local_maplib)
+                        GetGlobalWeakSymbolStartEnd(local_maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                    if(!offs && !end && local_maplib)
+                        GetGlobalWeakSymbolStartEnd(maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                }
+            } else {
+                if(!offs && !end) {
+                    if(!offs && !end && local_maplib)
+                        GetGlobalSymbolStartEnd(local_maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                    if(!offs && !end && local_maplib)
+                        GetGlobalSymbolStartEnd(maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                }
             }
         }
         uintptr_t globoffs, globend;
@@ -503,7 +531,8 @@ int RelocateElfREL(lib_t *maplib, lib_t *local_maplib, int bindnow, elfheader_t*
                     if(sym->st_size && offs) {
                         printf_dump(LOG_NEVER, "Apply %s R_X86_64_GLOB_DAT with R_X86_64_COPY @%p/%p (%p/%p -> %p/%p) size=%ld on sym=%s \n", (bind==STB_LOCAL)?"Local":"Global", p, globp, (void*)(p?(*p):0), (void*)(globp?(*globp):0), (void*)(offs + head->delta), (void*)globoffs, sym->st_size, symname);
                         memmove((void*)globoffs, (void*)offs, sym->st_size);   // preapply to copy part from lib to main elf
-                        AddUniqueSymbol(GetGlobalData(maplib), symname, offs + head->delta, sym->st_size, version, vername);
+                        AddUniqueSymbol(GetGlobalData(maplib), symname, globoffs, sym->st_size, version, vername);
+                        AddUniqueSymbol(my_context->globdata, symname, offs + head->delta, sym->st_size, version, vername);
                     } else {
                         printf_dump(LOG_NEVER, "Apply %s R_X86_64_GLOB_DAT with R_X86_64_COPY @%p/%p (%p/%p -> %p/%p) null sized on sym=%s \n", (bind==STB_LOCAL)?"Local":"Global", p, globp, (void*)(p?(*p):0), (void*)(globp?(*globp):0), (void*)offs, (void*)globoffs, symname);
                     }
@@ -528,7 +557,7 @@ int RelocateElfREL(lib_t *maplib, lib_t *local_maplib, int bindnow, elfheader_t*
                     uintptr_t old_offs = offs;
                     uintptr_t old_end = end;
                     offs = 0;
-                    GetSizedSymbolStartEnd(GetGlobalData(maplib), symname, &offs, &end, size, version, vername, 1, globdefver); // try globaldata symbols first
+                    GetSizedSymbolStartEnd(my_context->globdata, symname, &offs, &end, size, version, vername, 1, globdefver); // try globaldata symbols first
                     if(offs==0) {
                         GetNoSelfSymbolStartEnd(maplib, symname, &offs, &end, head, size, version, vername, globdefver, weakdefver);   // get original copy if any
                         if(!offs && local_maplib)
@@ -672,13 +701,24 @@ int RelocateElfRELA(lib_t *maplib, lib_t *local_maplib, int bindnow, elfheader_t
                 end = offs + sym->st_size;
             }*/
             // so weak symbol are the one left
-            if(old_version==version && old_bind==bind && old_symname==symname) {
-                offs = old_offs;
-                end = old_end;
+            if(bind==STB_WEAK) {
+                if(old_version==version && old_bind==bind && old_symname==symname) {
+                    offs = old_offs;
+                    end = old_end;
+                } else {
+                    GetGlobalWeakSymbolStartEnd(maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                    if(!offs && !end && local_maplib)
+                        GetGlobalWeakSymbolStartEnd(local_maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                }
             } else {
-                GetGlobalSymbolStartEnd(maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
-                if(!offs && !end && local_maplib)
-                    GetGlobalSymbolStartEnd(local_maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                if(old_version==version && old_bind==bind && old_symname==symname) {
+                    offs = old_offs;
+                    end = old_end;
+                } else {
+                    GetGlobalSymbolStartEnd(maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                    if(!offs && !end && local_maplib)
+                        GetGlobalSymbolStartEnd(local_maplib, symname, &offs, &end, head, version, vername, globdefver, weakdefver);
+                }
             }
         }
         old_bind = bind;
@@ -712,7 +752,7 @@ int RelocateElfRELA(lib_t *maplib, lib_t *local_maplib, int bindnow, elfheader_t
                 globoffs = offs;
                 globend = end;
                 offs = end = 0;
-                GetSizedSymbolStartEnd(GetGlobalData(maplib), symname, &offs, &end, size, version, vername, 1, globdefver); // try globaldata symbols first
+                GetSizedSymbolStartEnd(my_context->globdata, symname, &offs, &end, size, version, vername, 1, globdefver); // try globaldata symbols first
                 if(!offs && local_maplib)
                     GetNoSelfSymbolStartEnd(local_maplib, symname, &offs, &end, head, size, version, vername, globdefver, weakdefver);
                 if(!offs)
@@ -737,7 +777,8 @@ int RelocateElfRELA(lib_t *maplib, lib_t *local_maplib, int bindnow, elfheader_t
                             (bind==STB_LOCAL)?"Local":"Global", p, globp, (void*)(p?(*p):0), 
                             (void*)(globp?(*globp):0), (void*)offs, (void*)globoffs, sym->st_size, symname, version, vername?vername:"(none)");
                         //memmove((void*)globoffs, (void*)offs, sym->st_size);   // preapply to copy part from lib to main elf
-                        AddUniqueSymbol(GetGlobalData(maplib), symname, offs, sym->st_size, version, vername);
+                        AddUniqueSymbol(GetGlobalData(maplib), symname, globoffs, sym->st_size, version, vername);
+                        AddUniqueSymbol(my_context->globdata, symname, offs, sym->st_size, version, vername);
                     } else {
                         printf_dump(LOG_NEVER, "Apply %s R_X86_64_GLOB_DAT with R_X86_64_COPY @%p/%p (%p/%p -> %p/%p) null sized on sym=%s (ver=%d/%s)\n", 
                             (bind==STB_LOCAL)?"Local":"Global", p, globp, (void*)(p?(*p):0), 
@@ -1201,7 +1242,7 @@ int LoadNeededLibs(elfheader_t* h, lib_t *maplib, int local, int bindnow, box64c
             h->needed->names[j++] = h->DynStrTab+h->delta+h->Dynamic[i].d_un.d_val;
 
     // TODO: Add LD_LIBRARY_PATH and RPATH handling
-    if(AddNeededLib(maplib, local, bindnow, h->needed, box64, emu)) {
+    if(AddNeededLib(maplib, local, bindnow, h->needed, h, box64, emu)) {
         printf_log(LOG_INFO, "Error loading one of needed lib\n");
         if(!allow_missing_libs)
             return 1;   //error...
@@ -1246,6 +1287,7 @@ void MarkElfInitDone(elfheader_t* h)
     if(h)
         h->init_done = 1;
 }
+void startMallocHook();
 void RunElfInitPltResolver(elfheader_t* h, x64emu_t *emu)
 {
     if(!h || h->init_done)
@@ -1260,17 +1302,20 @@ void RunElfInitPltResolver(elfheader_t* h, x64emu_t *emu)
     }
     printf_dump(LOG_DEBUG, "Calling Init for %s @%p\n", ElfName(h), (void*)p);
     if(h->initentry)
-        RunSafeFunction(my_context, p, 3, my_context->argc, my_context->argv, my_context->envv);
+        RunSafeFunction(p, 3, my_context->argc, my_context->argv, my_context->envv);
     printf_dump(LOG_DEBUG, "Done Init for %s\n", ElfName(h));
     // and check init array now
     Elf64_Addr *addr = (Elf64_Addr*)(h->initarray + h->delta);
     for (size_t i=0; i<h->initarray_sz; ++i) {
         if(addr[i]) {
             printf_dump(LOG_DEBUG, "Calling Init[%zu] for %s @%p\n", i, ElfName(h), (void*)addr[i]);
-            RunSafeFunction(my_context, (uintptr_t)addr[i], 3, my_context->argc, my_context->argv, my_context->envv);
+            RunSafeFunction((uintptr_t)addr[i], 3, my_context->argc, my_context->argv, my_context->envv);
         }
     }
 
+    if(h->malloc_hook_2)
+        startMallocHook();
+
     h->fini_done = 0;   // can be fini'd now (in case it was re-inited)
     printf_dump(LOG_DEBUG, "All Init Done for %s\n", ElfName(h));
     return;
@@ -1315,6 +1360,9 @@ void RunElfInit(elfheader_t* h, x64emu_t *emu)
         }
     }
 
+    if(h->malloc_hook_2)
+        startMallocHook();
+
     h->fini_done = 0;   // can be fini'd now (in case it was re-inited)
     printf_dump(LOG_DEBUG, "All Init Done for %s\n", ElfName(h));
     return;
@@ -1587,7 +1635,7 @@ static int my_dl_iterate_phdr_##A(struct dl_phdr_info* a, size_t b, void* c)
         return 0;                                                                   \
     if(!a->dlpi_name[0]) /*don't send informations about box64 itself*/             \
         return 0;                                                                   \
-    return (int)RunFunction(my_context, my_dl_iterate_phdr_fct_##A, 3, a, b, c);    \
+    return (int)RunFunction(my_dl_iterate_phdr_fct_##A, 3, a, b, c);                \
 }
 SUPER()
 #undef GO

src/elfs/elfloader_private.h

@@ -87,6 +87,7 @@ struct elfheader_s {
     int         init_done;
     int         fini_done;
     int         refcnt;     // ref count for the elf
+    int         malloc_hook_2;  // this elf hook malloc, hacking it
 
     char*       memory;     // char* and not void* to allow math on memory pointer
     void**      multiblock;

src/elfs/elfparser.c

@@ -416,18 +416,6 @@ int GetVersionIndice(elfheader_t* h, const char* vername)
 {
     if(!vername)
         return 0;
-    if(h->VerNeed) {
-        Elf64_Verneed *ver = (Elf64_Verneed*)((uintptr_t)h->VerNeed + h->delta);
-        while(ver) {
-            Elf64_Vernaux *aux = (Elf64_Vernaux*)((uintptr_t)ver + ver->vn_aux);
-            for(int j=0; j<ver->vn_cnt; ++j) {
-                if(!strcmp(h->DynStr+aux->vna_name, vername)) 
-                    return aux->vna_other;
-                aux = (Elf64_Vernaux*)((uintptr_t)aux + aux->vna_next);
-            }
-            ver = ver->vn_next?((Elf64_Verneed*)((uintptr_t)ver + ver->vn_next)):NULL;
-        }
-    }
     if(h->VerDef) {
         Elf64_Verdef *def = (Elf64_Verdef*)((uintptr_t)h->VerDef + h->delta);
         while(def) {
@@ -438,4 +426,44 @@ int GetVersionIndice(elfheader_t* h, const char* vername)
         }
     }
     return 0;
+}
+
+int GetNeededVersionCnt(elfheader_t* h, const char* libname)
+{
+    if(!libname)
+        return 0;
+    if(h->VerNeed) {
+        Elf64_Verneed *ver = (Elf64_Verneed*)((uintptr_t)h->VerNeed + h->delta);
+        while(ver) {
+            char *filename = h->DynStr + ver->vn_file;
+            Elf64_Vernaux *aux = (Elf64_Vernaux*)((uintptr_t)ver + ver->vn_aux);
+            if(!strcmp(filename, libname))
+                return ver->vn_cnt;
+            ver = ver->vn_next?((Elf64_Verneed*)((uintptr_t)ver + ver->vn_next)):NULL;
+        }
+    }
+    return 0;
+}
+
+const char* GetNeededVersionString(elfheader_t* h, const char* libname, int idx)
+{
+    if(!libname)
+        return 0;
+    if(h->VerNeed) {
+        Elf64_Verneed *ver = (Elf64_Verneed*)((uintptr_t)h->VerNeed + h->delta);
+        while(ver) {
+            char *filename = h->DynStr + ver->vn_file;
+            Elf64_Vernaux *aux = (Elf64_Vernaux*)((uintptr_t)ver + ver->vn_aux);
+            if(!strcmp(filename, libname)) {
+                for(int j=0; j<ver->vn_cnt; ++j) {
+                    if(j==idx) 
+                        return h->DynStr+aux->vna_name;
+                    aux = (Elf64_Vernaux*)((uintptr_t)aux + aux->vna_next);
+                }
+                return NULL;    // idx out of bound, return NULL...
+           }
+            ver = ver->vn_next?((Elf64_Verneed*)((uintptr_t)ver + ver->vn_next)):NULL;
+        }
+    }
+    return NULL;
 }

src/emu/modrm.h

@@ -34,6 +34,7 @@
 #define GETEW(D)            oped=TestEw(test, &addr, rex, nextop, D)
 #define GETEW32(D)          oped=TestEw32O(test, &addr, rex, nextop, D, 0)
 #define GETEW_OFFS(D, O)    oped=TestEdO(test, &addr, rex, nextop, D, O)
+#define GETEW_OFFS_16(O)    oped=TestEw16off(test, &addr, rex, nextop, O)
 #define GETGW               opgd=GetGw(test->emu, &addr, rex, nextop)
 #define GETEX(D)            opex=TestEx(test, &addr, rex, nextop, D)
 #define GETEX32(D)          opex=TestEx32O(test, &addr, rex, nextop, D, 0)
@@ -58,6 +59,7 @@
 #define GETEW(D)            oped=GetEw(emu, &addr, rex, nextop, D)
 #define GETEW32(D)          oped=GetEw32O(emu, &addr, rex, nextop, D, 0)
 #define GETEW_OFFS(D, O)    oped=GetEdO(emu, &addr, rex, nextop, D, O)
+#define GETEW_OFFS_16(O)    oped=GetEw16off(emu, &addr, rex, nextop, O)
 #define GETGW               opgd=GetGw(emu, &addr, rex, nextop)
 #define GETEX(D)            opex=GetEx(emu, &addr, rex, nextop, D)
 #define GETEX32(D)          opex=GetEx32O(emu, &addr, rex, nextop, D, 0)

src/emu/x64emu.c

@@ -567,9 +567,9 @@ void EmuCall(x64emu_t* emu, uintptr_t addr)
     Run(emu, 0);
     emu->quit = 0;  // reset Quit flags...
     emu->df = d_none;
-    if(emu->quitonlongjmp && emu->longjmp) {
-        if(emu->quitonlongjmp==1)
-            emu->longjmp = 0;   // don't change anything because of the longjmp
+    if(emu->flags.quitonlongjmp && emu->flags.longjmp) {
+        if(emu->flags.quitonlongjmp==1)
+            emu->flags.longjmp = 0;   // don't change anything because of the longjmp
     } else {
         R_RBX = old_rbx;
         R_RDI = old_rdi;

src/emu/x64emu_private.h

@@ -36,6 +36,14 @@ typedef struct x64test_s {
     uint8_t     mem[16];
 } x64test_t;
 
+typedef struct emu_flags_s {
+    uint32_t    need_jmpbuf:1;    // need a new jmpbuff for signal handling
+    uint32_t    quitonlongjmp:2;  // quit if longjmp is called
+    uint32_t    quitonexit:2;     // quit if exit/_exit is called
+    uint32_t    longjmp:1;        // if quit because of longjmp
+    uint32_t    jmpbuf_ready:1;   // the jmpbuf in the emu is ok and don't need refresh
+} emu_flags_t;
+
 typedef struct x64emu_s {
     // cpu
 	reg64_t     regs[16];
@@ -48,7 +56,7 @@ typedef struct x64emu_s {
 	mmx87_regs_t mmx[8];
 	x87control_t cw;
 	x87flags_t  sw;
-	uint32_t    top;        // top is part of sw, but it's faster to have it separatly
+	uint32_t    top;        // top is part of sw, but it's faster to have it separately
     int         fpu_stack;
     mmxcontrol_t mxcsr;
     fpu_ld_t    fpu_ld[8]; // for long double emulation / 80bits fld fst
@@ -83,16 +91,14 @@ typedef struct x64emu_s {
     int         quit;
     int         error;
     int         fork;   // quit because need to fork
-    forkpty_t*  forkpty_info;
     int         exit;
-    int         quitonlongjmp;  // quit if longjmp is called
-    int         quitonexit;     // quit if exit/_exit is called
-    int         longjmp;        // if quit because of longjmp
+    forkpty_t*  forkpty_info;
+    emu_flags_t flags;
     x64test_t   test;       // used for dynarec testing
     #ifdef HAVE_TRACE
     sse_regs_t  old_xmm[16];
     #endif
-    // scratch stack, used for alignement of double and 64bits ints on arm. 200 elements should be enough
+    // scratch stack, used for alignment of double and 64bits ints on arm. 200 elements should be enough
     uint64_t    scratch[200];
     // local stack, do be deleted when emu is freed
     void*       stack2free; // this is the stack to free (can be NULL)

src/emu/x64run.c

@@ -227,11 +227,34 @@ x64emurun:
             emu->segs_serial[_DS] = 0;
             break;
 
+        case 0x27:                  /* DAA */
+            if(rex.is32bits) {
+                R_AL = daa8(emu, R_AL);
+            } else {
+                unimp = 1;
+                goto fini;
+            }
+            break;
+        case 0x2F:                  /* DAS */
+            if(rex.is32bits) {
+                R_AL = das8(emu, R_AL);
+            } else {
+                unimp = 1;
+                goto fini;
+            }
+            break;
 	    case 0x2E:	    /* segments are ignored */
         case 0x26:
         case 0x36:          /* SS: (ignored) */
             break;
-
+        case 0x37:                  /* AAA */
+            if(rex.is32bits) {
+                R_AX = aaa16(emu, R_AX);
+            } else {
+                unimp = 1;
+                goto fini;
+            }
+            break;
         case 0x38:
             nextop = F8;
             GETEB(0);
@@ -271,6 +294,15 @@ x64emurun:
             else
                 cmp32(emu, R_EAX, F32);
             break;
+
+        case 0x3F:                  /* AAS */
+            if(rex.is32bits) {
+                R_AX = aas16(emu, R_AX);
+            } else {
+                unimp = 1;
+                goto fini;
+            }
+            break;
         case 0x40:
         case 0x41:
         case 0x42:
@@ -361,7 +393,14 @@ x64emurun:
                 goto fini;
             }
             break;
-
+        case 0x62:                  /* BOUND Gd, Ed */
+            if(rex.is32bits) {
+                FAKEED(0);
+            } else {
+                unimp = 1;
+                goto fini;
+            }
+            break;
         case 0x63:                      /* MOVSXD Gd,Ed */
             nextop = F8;
             GETED(0);
@@ -659,12 +698,12 @@ x64emurun:
             GETEB(0);
             GETGB;
             if(!MODREG)
-                pthread_mutex_lock(&emu->context->mutex_lock); // XCHG always LOCK (but when accessing memory only)
+                pthread_mutex_lock(&my_context->mutex_lock); // XCHG always LOCK (but when accessing memory only)
             tmp8u = GB;
             GB = EB->byte[0];
             EB->byte[0] = tmp8u;
             if(!MODREG)
-                pthread_mutex_unlock(&emu->context->mutex_lock);
+                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif                
             break;
         case 0x87:                      /* XCHG Ed,Gd */
@@ -693,7 +732,7 @@ x64emurun:
             GETED(0);
             GETGD;
             if(!MODREG)
-                pthread_mutex_lock(&emu->context->mutex_lock); // XCHG always LOCK (but when accessing memory only)
+                pthread_mutex_lock(&my_context->mutex_lock); // XCHG always LOCK (but when accessing memory only)
             if(rex.w) {
                 tmp64u = GD->q[0];
                 GD->q[0] = ED->q[0];
@@ -707,7 +746,7 @@ x64emurun:
                     ED->dword[0] = tmp32u;
             }
             if(!MODREG)
-                pthread_mutex_unlock(&emu->context->mutex_lock);
+                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
             break;
         case 0x88:                      /* MOV Eb,Gb */
@@ -1315,7 +1354,32 @@ x64emurun:
             addr = rex.is32bits?Pop32(emu):Pop64(emu);
             STEP2
             break;
-
+        case 0xC4:                      /* LES Gd,Ed */
+            if(rex.is32bits) {
+                nextop = F8;
+                GETED(0);
+                GETGD;
+                emu->segs[_ES] = *(__uint16_t*)(((char*)ED)+4);
+                emu->segs_serial[_ES] = 0;
+                GD->dword[0] = *(uint32_t*)ED;
+            } else {
+                // AVX not supported yet
+                emit_signal(emu, SIGILL, (void*)R_RIP, 0);
+            }
+            break;
+        case 0xC5:                      /* LDS Gd,Ed */
+            if(rex.is32bits) {
+                nextop = F8;
+                GETED(0);
+                GETGD;
+                emu->segs[_DS] = *(__uint16_t*)(((char*)ED)+4);
+                emu->segs_serial[_DS] = 0;
+                GD->dword[0] = *(uint32_t*)ED;
+            } else {
+                // AVX not supported yet
+                emit_signal(emu, SIGILL, (void*)R_RIP, 0);
+            }
+            break;
         case 0xC6:                      /* MOV Eb,Ib */
             nextop = F8;
             GETEB(1);
@@ -1378,12 +1442,14 @@ x64emurun:
             if(box64_wine && tmp8u==0x2D) {
                 // lets ignore the INT 2D
                 printf_log(LOG_DEBUG, "INT 2D called\n");
+                emit_interruption(emu, 0x2d, (void*)R_RIP);
+            } else if(box64_wine && tmp8u==0x2c) {
+                printf_log(LOG_DEBUG, "INT 2c called\n");
+                emit_interruption(emu, 0x2c, (void*)R_RIP);
             } else if(box64_wine && tmp8u==0x29) {
                 // INT 29 is __fastfail
                 printf_log(LOG_DEBUG, "INT 29 called => __fastfail(0x%x)\n", R_ECX);
-                emu->quit = 1;
-                R_RIP = addr;
-                goto fini;
+                emit_interruption(emu, 0x29, (void*)R_RIP);
             } else if (tmp8u==0x80) {
                 // 32bits syscall
                 #ifndef TEST_INTERPRETER
@@ -1470,6 +1536,15 @@ x64emurun:
             }
             break;
 
+        case 0xD6:                      /* SALC */
+            if(rex.is32bits) {
+                CHECK_FLAGS(emu);
+                R_AL = ACCESS_FLAG(F_CF)?0xff:0x00;
+            } else {
+                unimp = 1;
+                goto fini;
+            };
+            break;
         case 0xD7:                      /* XLAT */
             R_AL = *(uint8_t*)(R_RBX + R_AL);
             break;
@@ -1871,9 +1946,8 @@ x64emurun:
                 case 3:                 /* CALL FAR Ed */
                     GETET(0);
                     if(MODREG) {
-                        printf_log(LOG_NONE, "Illegal Opcode %p: %02X %02X %02X %02X\n", (void*)R_RIP, opcode, nextop, PK(2), PK(3));
-                        emu->quit=1;
-                        emu->error |= ERR_ILLEGAL;
+                        printf_log(LOG_NONE, "Illegal Opcode %p: (%02X %02X %02X %02X) %02X %02X %02X %02X\n", (void*)R_RIP, PK(-6), PK(-5), PK(-4), PK(-3), opcode, nextop, PK(0), PK(1));
+                        emit_signal(emu, SIGILL, (void*)R_RIP, 0);
                         goto fini;
                     } else {
                         if(rex.is32bits || !rex.w) {
@@ -1902,9 +1976,8 @@ x64emurun:
                 case 5:                 /* JMP FAR Ed */
                     GETET(0);
                     if(MODREG) {
-                        printf_log(LOG_NONE, "Illegal Opcode %p: 0x%02X 0x%02X %02X %02X\n", (void*)R_RIP, opcode, nextop, PK(2), PK(3));
-                        emu->quit=1;
-                        emu->error |= ERR_ILLEGAL;
+                        printf_log(LOG_NONE, "Illegal Opcode %p: (%02X %02X %02X %02X) %02X %02X %02X %02X\n", (void*)R_RIP, PK(-6), PK(-5), PK(-4), PK(-3), opcode, nextop, PK(0), PK(1));
+                        emit_signal(emu, SIGILL, (void*)R_RIP, 0);
                         goto fini;
                     } else {
                         if(rex.is32bits || !rex.w) {
@@ -1929,9 +2002,8 @@ x64emurun:
                     }
                     break;
                 default:
-                    printf_log(LOG_NONE, "Illegal Opcode %p: %02X %02X %02X %02X %02X %02X\n",(void*)R_RIP, opcode, nextop, PK(2), PK(3), PK(4), PK(5));
-                    emu->quit=1;
-                    emu->error |= ERR_ILLEGAL;
+                    printf_log(LOG_NONE, "Illegal Opcode %p: (%02X %02X %02X %02X) %02X %02X %02X %02X %02X %02X\n", (void*)R_RIP, PK(-6), PK(-5), PK(-4), PK(-3), opcode, nextop, PK(0), PK(1), PK(2), PK(3));
+                    emit_signal(emu, SIGILL, (void*)R_RIP, 0);
                     goto fini;
             }
             break;

src/emu/x64run0f.c

@@ -813,7 +813,19 @@ uintptr_t Run0F(x64emu_t *emu, rex_t rex, uintptr_t addr, int *step)
             , EB->byte[0]=0;
             ,
         )                               /* 0x90 -> 0x9F SETxx Eb */
-
+        case 0xA0:                      /* PUSH FS */
+            if(rex.is32bits)
+                Push32(emu, emu->segs[_FS]);
+            else
+                Push64(emu, emu->segs[_FS]);
+            break;
+        case 0xA1:                      /* POP FS */
+            if(rex.is32bits)
+                emu->segs[_FS] = Pop32(emu);
+            else
+                emu->segs[_FS] = Pop64(emu);
+            emu->segs_serial[_FS] = 0;
+            break;
         case 0xA2:                      /* CPUID */
             tmp32u = R_EAX;
             my_cpuid(emu, tmp32u);
@@ -869,6 +881,20 @@ uintptr_t Run0F(x64emu_t *emu, rex_t rex, uintptr_t addr, int *step)
             }
             break;
 
+        case 0xA8:                      /* PUSH GS */
+            if(rex.is32bits)
+                Push32(emu, emu->segs[_GS]);
+            else
+                Push64(emu, emu->segs[_GS]);
+            break;
+        case 0xA9:                      /* POP GS */
+            if(rex.is32bits)
+                emu->segs[_GS] = Pop32(emu);
+            else
+                emu->segs[_GS] = Pop64(emu);
+            emu->segs_serial[_FS] = 0;
+            break;
+
         case 0xAB:                      /* BTS Ed,Gd */
             CHECK_FLAGS(emu);
             nextop = F8;
@@ -1438,7 +1464,12 @@ uintptr_t Run0F(x64emu_t *emu, rex_t rex, uintptr_t addr, int *step)
             GETGM;
             GM->q = (EM->q > 63) ? 0L : (GM->q >> EM->q);
             break;
-
+        case 0xD4:                   /* PADDQ Gm,Em */
+            nextop = F8;
+            GETEM(0);
+            GETGM;
+            GM->sq += EM->sq;
+            break;
         case 0xD5:                   /* PMULLW Gm,Em */
             nextop = F8;
             GETEM(0);
@@ -1677,7 +1708,12 @@ uintptr_t Run0F(x64emu_t *emu, rex_t rex, uintptr_t addr, int *step)
             GETGM;
             GM->q = (EM->q > 63) ? 0L : (GM->q << EM->ub[0]);
             break;
-
+        case 0xF4:                   /* PMULUDQ Gm,Em */
+            nextop = F8;
+            GETEM(0);
+            GETGM;
+            GM->q = (uint64_t)GM->ud[0] * (uint64_t)EM->ud[0];
+            break;
         case 0xF5:                   /* PMADDWD Gm, Em */
             nextop = F8;
             GETEM(0);
@@ -1720,7 +1756,12 @@ uintptr_t Run0F(x64emu_t *emu, rex_t rex, uintptr_t addr, int *step)
             for(int i=0; i<2; ++i)
                 GM->ud[i] -= EM->ud[i];
             break;
-
+        case 0xFB:                   /* PSUBQ Gm, Em */
+            nextop = F8;
+            GETEM(0);
+            GETGM;
+            GM->sq -= EM->sq;
+            break;
         case 0xFC:                   /* PADDB Gm, Em */
             nextop = F8;
             GETEM(0);

src/emu/x64run64.c

@@ -326,6 +326,11 @@ uintptr_t Run64(x64emu_t *emu, rex_t rex, int seg, uintptr_t addr)
 
         case 0x66:
             return Run6664(emu, rex, seg, addr);
+        case 0x67:
+            if(rex.is32bits)
+                return Run6764_32(emu, rex, seg, seg, addr);
+            else
+                return 0;
 
         case 0x80:                      /* GRP Eb,Ib */
             nextop = F8;

src/emu/x64run66.c

@@ -166,7 +166,33 @@ uintptr_t Run66(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
         tmp8u = opcode&7;
         emu->regs[tmp8u].word[0] = dec16(emu, emu->regs[tmp8u].word[0]);
         break;
-
+    case 0x50:
+    case 0x51:
+    case 0x52:
+    case 0x53:
+    case 0x55:
+    case 0x56:
+    case 0x57:                      /* PUSH Reg */
+        if(rex.is32bits) {
+            tmp8u = opcode&7;
+            Push16(emu, emu->regs[tmp8u].word[0]);
+        } else
+            return 0;
+        break;
+    case 0x58:
+    case 0x59:
+    case 0x5A:
+    case 0x5B:
+    case 0x5C:                      /* POP ESP */
+    case 0x5D:
+    case 0x5E:
+    case 0x5F:                      /* POP Reg */
+        if(rex.is32bits) {
+            tmp8u = opcode&7;
+            emu->regs[tmp8u].word[0] = Pop16(emu);
+        } else
+            return 0;
+        break;
     case 0x60:                              /* PUSHA */
         if(rex.is32bits) {
             tmp16u = R_SP;
@@ -338,27 +364,26 @@ uintptr_t Run66(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
             GD->word[0] = (uint16_t)tmp64u;
         break;
 
-        case 0x90:                      /* NOP or XCHG R8d, AX*/
-        case 0x91:
-        case 0x92:
-        case 0x93:
-        case 0x94:
-        case 0x95:
-        case 0x96:
-        case 0x97:                      /* XCHG reg,AX */
-            tmp8u = _AX+(opcode&7)+(rex.b<<3);
-            if(tmp8u!=_AX) {
-                if(rex.w) {
-                    tmp64u = R_RAX;
-                    R_RAX = emu->regs[tmp8u].q[0];
-                    emu->regs[tmp8u].q[0] = tmp64u;
-                } else {
-                    tmp16u = R_AX;
-                    R_AX = emu->regs[tmp8u].word[0];
-                    emu->regs[tmp8u].word[0] = tmp16u;
-                }
+    case 0x90:                      /* NOP or XCHG R8d, AX*/
+    case 0x91:
+    case 0x92:
+    case 0x93:
+    case 0x94:
+    case 0x95:
+    case 0x96:
+    case 0x97:                      /* XCHG reg,AX */
+        tmp8u = _AX+(opcode&7)+(rex.b<<3);
+        if(tmp8u!=_AX) {
+            if(rex.w) {
+                tmp64u = R_RAX;
+                R_RAX = emu->regs[tmp8u].q[0];
+                emu->regs[tmp8u].q[0] = tmp64u;
+            } else {
+                tmp16u = R_AX;
+                R_AX = emu->regs[tmp8u].word[0];
+                emu->regs[tmp8u].word[0] = tmp16u;
             }
-            break;
+        }
         break;
 
     case 0x98:                               /* CBW */
@@ -368,6 +393,16 @@ uintptr_t Run66(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
         R_DX=((R_AX & 0x8000)?0xFFFF:0x0000);
         break;
 
+    case 0x9C:                              /* PUSHFW */
+        CHECK_FLAGS(emu);
+        Push16(emu, (uint16_t)emu->eflags.x64);
+        break;
+    case 0x9D:                              /* POPFW */
+        CHECK_FLAGS(emu);
+        emu->eflags.x64 &=0xffff0000;
+        emu->eflags.x64 |= (Pop16(emu) & 0x3F7FD7) | 0x2;
+        break;
+
     case 0xA1:                      /* MOV EAX,Od */
         if(rex.is32bits) {
             R_AX = *(uint16_t*)(uintptr_t)F32;
@@ -789,7 +824,6 @@ uintptr_t Run66(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
     case 0xFF:                      /* GRP 5 Ew */
         nextop = F8;
         GETEW(0);
-        GETGW;
         switch((nextop>>3)&7) {
             case 0:                 /* INC Ed */
                 EW->word[0] = inc16(emu, EW->word[0]);
@@ -807,9 +841,12 @@ uintptr_t Run66(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
                 }
                 addr = tmp64u;
                 break;
-           /*case 6:
-                Push16(emu, EW->word[0]);
-                break;*/
+           case 6:                  /* Push Ew */
+                if(rex.is32bits) {
+                    Push16(emu, EW->word[0]);
+                } else 
+                    return 0;
+                break;
             default:
                     printf_log(LOG_NONE, "Illegal Opcode %p: 66 %02X %02X %02X %02X %02X %02X\n",(void*)R_RIP, opcode, nextop, PK(2), PK(3), PK(4), PK(5));
                     emu->quit=1;

src/emu/x64run660f.c

@@ -788,7 +788,18 @@ uintptr_t Run660F(x64emu_t *emu, rex_t rex, uintptr_t addr)
                 GX->q[0] ^= EX->q[0];
                 GX->q[1] ^= EX->q[1];
                 break;
-
+            case 0xF0: /* MOVBE Gw, Ew */
+                nextop = F8;
+                GETEX(0);
+                GETGX;
+                GX->uw[0] = __builtin_bswap16(EX->uw[0]);
+                break;
+            case 0xF1: /* MOVBE Ew, Gw */
+                nextop = F8;
+                GETEX(0);
+                GETGX;
+                EX->uw[0] = __builtin_bswap16(GX->uw[0]);
+                break;
             default:
                 return 0;
         }

src/emu/x64run66f0.c

@@ -79,14 +79,14 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         }
                     } while(tmp32s);
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     cmp16(emu, R_AX, EW->word[0]);
                     if(ACCESS_FLAG(F_ZF)) {
                         EW->word[0] = GW->word[0];
                     } else {
                         R_AX = EW->word[0];
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
 
@@ -110,11 +110,11 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                     }
                     GD->word[0] = tmp16u;
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     tmp16u = add16(emu, ED->word[0], GD->word[0]);
                     GD->word[0] = ED->word[0];
                     ED->word[0] = tmp16u;
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
 
@@ -148,22 +148,22 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
             nextop = F8;                                            \
             GETEW(0);                                               \
             GETGW;                                                  \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             EW->word[0] = OP##16(emu, EW->word[0], GW->word[0]);    \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+3:                                                   \
             nextop = F8;                                            \
             GETEW(0);                                               \
             GETGW;                                                  \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             GW->word[0] = OP##16(emu, GW->word[0], EW->word[0]);    \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+5:                                                   \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             R_AX = OP##16(emu, R_AX, F16);                          \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;
 #endif
         GO(0x00, add)                   /* ADD 0x00 -> 0x05 */
@@ -205,7 +205,7 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                     case 7:                                                  cmp16(emu, ED->word[0], tmp16u); break;
                 }
 #else
-            pthread_mutex_lock(&emu->context->mutex_lock);
+            pthread_mutex_lock(&my_context->mutex_lock);
             switch((nextop>>3)&7) {
                 case 0: ED->word[0] = add16(emu, ED->word[0], tmp16u); break;
                 case 1: ED->word[0] =  or16(emu, ED->word[0], tmp16u); break;
@@ -216,7 +216,7 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                 case 6: ED->word[0] = xor16(emu, ED->word[0], tmp16u); break;
                 case 7:               cmp16(emu, ED->word[0], tmp16u); break;
             }
-            pthread_mutex_unlock(&emu->context->mutex_lock);
+            pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
             break;
 
@@ -241,9 +241,9 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         } while(native_lock_write_h(ED, inc16(emu, tmp16u)));
                     }
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     ED->word[0] = inc16(emu, ED->word[0]);
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 case 1:                 /* DEC Ed */
@@ -252,9 +252,9 @@ uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         tmp16u = native_lock_read_h(ED);
                     } while(native_lock_write_h(ED, dec16(emu, tmp16u)));
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     ED->word[0] = dec16(emu, ED->word[0]);
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 default:

src/emu/x64run67.c

@@ -39,10 +39,14 @@ uintptr_t Run67(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
     #ifdef TEST_INTERPRETER
     x64emu_t* emu = test->emu;
     #endif
-    opcode = F8;
-
     if(rex.is32bits)
+    #ifdef TEST_INTERPRETER
+        return Test67_32(test, rex, rep, addr);
+    #else
         return Run67_32(emu, rex, rep, addr);
+    #endif
+
+    opcode = F8;
 
     while(opcode==0x67)
         opcode = F8;

src/emu/x64run670f.c

@@ -51,6 +51,19 @@ uintptr_t Run670F(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
 
     switch(opcode) {
 
+        case 0x11:
+            switch(rep) {
+                case 0:                      /* MOVUPS Ex,Gx */
+                    nextop = F8;
+                    GETEX32(0);
+                    GETGX;
+                    EX->u128 = GX->u128;
+                    break;
+                default:
+                    return 0;
+            }
+            break;
+
         case 0x2E:
             // same for now
         case 0x2F:                      
@@ -84,6 +97,12 @@ uintptr_t Run670F(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
                     GETGM;
                     GM->q = EM->q;
                     break;
+                case 2:  /* MOVDQU Gx, Ex */
+                    nextop = F8;
+                    GETEX32(0);
+                    GETGX;
+                    memcpy(GX, EX, 16);    // unaligned...
+                    break;
                 default:
                     return 0;
             }

src/emu/x64run6764_32.c

@@ -0,0 +1,67 @@
+#define _GNU_SOURCE
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <math.h>
+#include <string.h>
+#include <signal.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+#include "debug.h"
+#include "box64stack.h"
+#include "x64emu.h"
+#include "x64run.h"
+#include "x64emu_private.h"
+#include "x64run_private.h"
+#include "x64primop.h"
+#include "x64trace.h"
+#include "x87emu_private.h"
+#include "box64context.h"
+#include "bridge.h"
+
+#include "modrm.h"
+
+#ifdef TEST_INTERPRETER
+uintptr_t Test6764_32(x64test_t *test, rex_t rex, int rep, int seg, uintptr_t addr)
+#else
+uintptr_t Run6764_32(x64emu_t *emu, rex_t rex, int rep, int seg, uintptr_t addr)
+#endif
+{
+    uint8_t opcode;
+    uint8_t nextop;
+    int8_t tmp8s;
+    uint8_t tmp8u;
+    uint32_t tmp32u;
+    int32_t tmp32s;
+    uint64_t tmp64u;
+    reg64_t *oped, *opgd;
+    #ifdef TEST_INTERPRETER
+    x64emu_t* emu = test->emu;
+    #endif
+    uintptr_t tlsdata = GetSegmentBaseEmu(emu, seg);
+    opcode = F8;
+
+    while(opcode==0x67)
+        opcode = F8;
+
+    rex.rex = 0;
+    while((opcode==0xF2) || (opcode==0xF3)) {
+        rep = opcode-0xF1;
+        opcode = F8;
+    }
+
+    switch(opcode) {
+
+        case 0x8B:                      /* MOV Gw, FS:Ew */
+            nextop = F8;
+            GETEW_OFFS_16(tlsdata);
+            GETGW;
+            GW->word[0] = EW->word[0];
+            break;
+
+    default:
+        return 0;
+    }
+    return addr;
+}

src/emu/x64run67_32.c

@@ -52,6 +52,21 @@ uintptr_t Run67_32(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr)
 
     switch(opcode) {
 
+    case 0x64:                      /* FS: prefix */
+        #ifdef TEST_INTERPRETER
+        return Test6764_32(test, rex, rep, _FS, addr);
+        #else
+        return Run6764_32(emu, rex, rep, _FS, addr);
+        #endif
+        break;
+    case 0x65:                      /* GS: prefix */
+        #ifdef TEST_INTERPRETER
+        return Test6764_32(test, rex, rep, _GS, addr);
+        #else
+        return Run6764_32(emu, rex, rep, _GS, addr);
+        #endif
+        break;
+
     case 0xE0:                      /* LOOPNZ */
         CHECK_FLAGS(emu);
         tmp8s = F8S;

src/emu/x64run_private.c

@@ -53,9 +53,13 @@ int32_t EXPORT my___libc_start_main(x64emu_t* emu, int *(main) (int, char * *, c
         SetRBP(emu, old_rbp);
         emu->quit = 0;
     } else {
-        RunElfInit(my_context->elfs[0], emu);
+        if(my_context->elfs[0]) {
+            RunElfInit(my_context->elfs[0], emu);
+        }
+    }
+    if(my_context->elfs[0]) {
+        MarkElfInitDone(my_context->elfs[0]);
     }
-    MarkElfInitDone(my_context->elfs[0]);
     printf_log(LOG_DEBUG, "Transfert to main(%d, %p, %p)=>%p from __libc_start_main\n", my_context->argc, my_context->argv, my_context->envv, main);
     // call main and finish
     Push64(emu, GetRBP(emu));   // set frame pointer
@@ -1094,14 +1098,25 @@ void PrintTrace(x64emu_t* emu, uintptr_t ip, int dynarec)
                     printf_log(LOG_NONE, " STACK_TOP: %p ", (void*)(uintptr_t)*(uint32_t*)(R_RSP));
             } else if(peek==0xF3 && PK(1)==0x0F && PK(2)==0x1E && PK(3)==0xFA && !is32bits) {
                 printFunctionAddr(*(uintptr_t*)(R_RSP), " STACK_TOP: ");
-            } else if(peek==0xE8) { // Call
+            } else if(peek==0xE8 || peek==0xE9) { // Call & Jmp
                 uintptr_t nextaddr = ip + 5 + PK32(1);
                 printFunctionAddr(nextaddr, "=> ");
             } else if(peek==0xFF) {
                 if(PK(1)==0x25) {
-                    uintptr_t nextaddr = ip + 6 + PK32(2);
+                    uintptr_t nextaddr = *(uintptr_t*)(ip + 6 + PK32(2));
+                    if(!printFunctionAddr(nextaddr, "=> "))
+                        printf_log(LOG_NONE, " => %p", (void*)nextaddr);
+                } else if((PK(1)==0x14) && (PK(2)==0x25)) {
+                    uintptr_t nextaddr = *(uintptr_t*)(uintptr_t)PK32(3);
+                    printf_log(LOG_NONE, " => %p", (void*)nextaddr);
                     printFunctionAddr(nextaddr, "=> ");
+                } else if((PK(1)==0x14) && (PK(2)==0xC2) && rex.rex==0x41) {
+                    uintptr_t nextaddr = *(uintptr_t*)(R_R10 + R_RAX*8);
+                    printf_log(LOG_NONE, " => %p", (void*)nextaddr);
+                    printFunctionAddr(nextaddr, "=> ");
+
                 }
+
             }
             printf_log(LOG_NONE, "\n");
         }
@@ -1185,7 +1200,7 @@ reg64_t* GetEw16_32(x64emu_t *emu, uintptr_t* addr, uint8_t m, uint32_t base)
         case 7: base+=      R_BX; break;
     }
     switch((m>>6)&3) {
-        case 0: if(m==6) base+= F16(addr); break;
+        case 0: if((m&7)==6) base= F16S(addr); break;
         case 1: base += F8S(addr); break;
         case 2: base += F16S(addr); break;
         // case 3 is C0..C7, already dealt with
@@ -1538,7 +1553,7 @@ reg64_t* GetEw16(x64emu_t *emu, uintptr_t* addr, rex_t rex, uint8_t v)
             case 7: base =      R_BX; break;
         }
         switch((m>>6)&3) {
-            case 0: if(m==6) base = F16(addr); break;
+            case 0: if((m&7)==6) base = F16S(addr); break;
             case 1: base += F8S(addr); break;
             case 2: base += F16S(addr); break;
             // case 3 is C0..C7, already dealt with
@@ -1568,7 +1583,7 @@ reg64_t* TestEw16(x64test_t *test, uintptr_t* addr, rex_t rex, uint8_t v)
             case 7: base =      R_BX; break;
         }
         switch((m>>6)&3) {
-            case 0: if(m==6) base = F16(addr); break;
+            case 0: if((m&7)==6) base = F16S(addr); break;
             case 1: base += F8S(addr); break;
             case 2: base += F16S(addr); break;
             // case 3 is C0..C7, already dealt with
@@ -1600,7 +1615,7 @@ reg64_t* GetEw16off(x64emu_t *emu, uintptr_t* addr, rex_t rex, uint8_t v, uintpt
             case 7: base =      R_BX; break;
         }
         switch((m>>6)&3) {
-            case 0: if(m==6) base = F16(addr); break;
+            case 0: if((m&7)==6) base = F16S(addr); break;
             case 1: base += F8S(addr); break;
             case 2: base += F16S(addr); break;
             // case 3 is C0..C7, already dealt with
@@ -1630,7 +1645,7 @@ reg64_t* TestEw16off(x64test_t *test, uintptr_t* addr, rex_t rex, uint8_t v, uin
             case 7: base =      R_BX; break;
         }
         switch((m>>6)&3) {
-            case 0: if(m==6) base = F16(addr); break;
+            case 0: if((m&7)==6) base = F16S(addr); break;
             case 1: base += F8S(addr); break;
             case 2: base += F16S(addr); break;
             // case 3 is C0..C7, already dealt with

src/emu/x64run_private.h

@@ -135,6 +135,7 @@ uintptr_t Run66DD(x64emu_t *emu, rex_t rex, uintptr_t addr);
 uintptr_t Run66F0(x64emu_t *emu, rex_t rex, uintptr_t addr);
 uintptr_t Run67(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr);
 uintptr_t Run67_32(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr);
+uintptr_t Run6764_32(x64emu_t *emu, rex_t rex, int rep, int seg, uintptr_t addr);
 uintptr_t Run670F(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr);
 uintptr_t Run6766(x64emu_t *emu, rex_t rex, int rep, uintptr_t addr);
 uintptr_t Run67660F(x64emu_t *emu, rex_t rex, uintptr_t addr);
@@ -160,6 +161,7 @@ uintptr_t Test66DD(x64test_t *test, rex_t rex, uintptr_t addr);
 uintptr_t Test66F0(x64test_t *test, rex_t rex, uintptr_t addr);
 uintptr_t Test67(x64test_t *test, rex_t rex, int rep, uintptr_t addr);
 uintptr_t Test67_32(x64test_t *test, rex_t rex, int rep, uintptr_t addr);
+uintptr_t Test6764_32(x64test_t *test, rex_t rex, int rep, int seg, uintptr_t addr);
 uintptr_t Test670F(x64test_t *test, rex_t rex, int rep, uintptr_t addr);
 uintptr_t Test6766(x64test_t *test, rex_t rex, int rep, uintptr_t addr);
 uintptr_t Test67660F(x64test_t *test, rex_t rex, uintptr_t addr);

src/emu/x64runf0.c

@@ -116,15 +116,15 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
             nextop = F8;                                            \
             GETEB(0);                                               \
             GETGB;                                                  \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             EB->byte[0] = OP##8(emu, EB->byte[0], GB);              \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+1:                                                   \
             nextop = F8;                                            \
             GETED(0);                                               \
             GETGD;                                                  \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             if(rex.w)                                               \
                 ED->q[0] = OP##64(emu, ED->q[0], GD->q[0]);         \
             else                                                    \
@@ -132,39 +132,39 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                     ED->q[0] = OP##32(emu, ED->dword[0], GD->dword[0]);     \
                 else                                                        \
                     ED->dword[0] = OP##32(emu, ED->dword[0], GD->dword[0]); \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+2:                                                   \
             nextop = F8;                                            \
             GETEB(0);                                               \
             GETGB;                                                  \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             GB = OP##8(emu, GB, EB->byte[0]);                       \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+3:                                                   \
             nextop = F8;                                            \
             GETED(0);                                               \
             GETGD;                                                  \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             if(rex.w)                                               \
                 GD->q[0] = OP##64(emu, GD->q[0], ED->q[0]);         \
             else                                                    \
                 GD->q[0] = OP##32(emu, GD->dword[0], ED->dword[0]); \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+4:                                                   \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             R_AL = OP##8(emu, R_AL, F8);                            \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;                                                  \
         case B+5:                                                   \
-            pthread_mutex_lock(&emu->context->mutex_lock);          \
+            pthread_mutex_lock(&my_context->mutex_lock);          \
             if(rex.w)                                               \
                 R_RAX = OP##64(emu, R_RAX, F32S64);                 \
             else                                                    \
                 R_RAX = OP##32(emu, R_EAX, F32);                    \
-            pthread_mutex_unlock(&emu->context->mutex_lock);        \
+            pthread_mutex_unlock(&my_context->mutex_lock);        \
             break;
 #endif
         GO(0x00, add)                   /* ADD 0x00 -> 0x05 */
@@ -245,7 +245,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         } while(tmp32s);
                 }
 #else
-                pthread_mutex_lock(&emu->context->mutex_lock);
+                pthread_mutex_lock(&my_context->mutex_lock);
                 if(rex.w) {
                     tmp8u&=63;
                     if(ED->q[0] & (1LL<<tmp8u))
@@ -265,7 +265,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                     if(MODREG)
                         ED->dword[1] = 0;
                 }
-                pthread_mutex_unlock(&emu->context->mutex_lock);
+                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                 break;
 
@@ -286,14 +286,14 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         }
                     } while(tmp32s);
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     cmp8(emu, R_AL, EB->byte[0]);
                     if(ACCESS_FLAG(F_ZF)) {
                         EB->byte[0] = GB;
                     } else {
                         R_AL = EB->byte[0];
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 case 0xB1:                      /* CMPXCHG Ed,Gd */
@@ -343,7 +343,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                             ED->dword[1] = 0;
                     }
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     if(rex.w) {
                         cmp64(emu, R_RAX, ED->q[0]);
                         if(ACCESS_FLAG(F_ZF)) {
@@ -362,7 +362,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         if(MODREG)
                             ED->dword[1] = 0;
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
 
@@ -416,7 +416,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                             ED->dword[1] = 0;
                     }
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     if(rex.w) {
                         if(ED->q[0] & (1<<tmp8u)) {
                             SET_FLAG(F_CF);
@@ -432,7 +432,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         if(MODREG)
                             ED->dword[1] = 0;
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
 
@@ -490,7 +490,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                     } while(tmp32s);
                                 }
 #else
-                                pthread_mutex_lock(&emu->context->mutex_lock);
+                                pthread_mutex_lock(&my_context->mutex_lock);
                                 if(rex.w) {
                                     tmp8u&=63;
                                     if(ED->q[0] & (1LL<<tmp8u)) {
@@ -508,7 +508,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                         CLEAR_FLAG(F_CF);
                                     }
                                 }
-                                pthread_mutex_unlock(&emu->context->mutex_lock);
+                                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                                 break;
                             case 6:             /* BTR Ed, Ib */
@@ -544,7 +544,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                     } while(tmp32s);
                                 }
 #else
-                                pthread_mutex_lock(&emu->context->mutex_lock);
+                                pthread_mutex_lock(&my_context->mutex_lock);
                                 if(rex.w) {
                                     tmp8u&=63;
                                     if(ED->q[0] & (1LL<<tmp8u)) {
@@ -560,7 +560,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                     } else
                                         CLEAR_FLAG(F_CF);
                                 }
-                                pthread_mutex_unlock(&emu->context->mutex_lock);
+                                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                                 break;
                             case 7:             /* BTC Ed, Ib */
@@ -592,7 +592,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                     } while(tmp32s);
                                 }
 #else
-                                pthread_mutex_lock(&emu->context->mutex_lock);
+                                pthread_mutex_lock(&my_context->mutex_lock);
                                 if(rex.w) {
                                     tmp8u&=63;
                                     if(ED->q[0] & (1LL<<tmp8u))
@@ -608,7 +608,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                         CLEAR_FLAG(F_CF);
                                     ED->dword[0] ^= (1<<tmp8u);
                                 }
-                                pthread_mutex_unlock(&emu->context->mutex_lock);
+                                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                                 break;
 
@@ -628,11 +628,11 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                     } while(native_lock_write_b(EB, tmp8u2));
                     GB = tmp8u;
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     tmp8u = add8(emu, EB->byte[0], GB);
                     GB = EB->byte[0];
                     EB->byte[0] = tmp8u;
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 case 0xC1:                      /* XADD Gd,Ed */
@@ -665,7 +665,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                             ED->dword[1] = 0;
                     }
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     if(rex.w) {
                         tmp64u = add64(emu, ED->q[0], GD->q[0]);
                         GD->q[0] = ED->q[0];
@@ -678,7 +678,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         else
                             ED->dword[0] = tmp32u;
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
 
@@ -717,7 +717,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                     }
                                 } while(tmp32s);
 #else
-                            pthread_mutex_lock(&emu->context->mutex_lock);
+                            pthread_mutex_lock(&my_context->mutex_lock);
                             if(rex.w) {
                                 #ifdef TEST_INTERPRETER
                                 test->memsize = 16;
@@ -749,7 +749,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                                     R_RDX = tmp32u2;
                                 }
                             }
-                            pthread_mutex_unlock(&emu->context->mutex_lock);
+                            pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                             break;
                         default:
@@ -785,7 +785,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                 case 7:               cmp8(emu, EB->byte[0], tmp8u); break;
             }
 #else
-            pthread_mutex_lock(&emu->context->mutex_lock);
+            pthread_mutex_lock(&my_context->mutex_lock);
             switch((nextop>>3)&7) {
                 case 0: EB->byte[0] = add8(emu, EB->byte[0], tmp8u); break;
                 case 1: EB->byte[0] =  or8(emu, EB->byte[0], tmp8u); break;
@@ -796,7 +796,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                 case 6: EB->byte[0] = xor8(emu, EB->byte[0], tmp8u); break;
                 case 7:               cmp8(emu, EB->byte[0], tmp8u); break;
             }
-            pthread_mutex_unlock(&emu->context->mutex_lock);
+            pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
             break;
         case 0x81:              /* GRP Ed,Id */
@@ -845,7 +845,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                     }
             }
 #else
-            pthread_mutex_lock(&emu->context->mutex_lock);
+            pthread_mutex_lock(&my_context->mutex_lock);
             if(rex.w) {
                 switch((nextop>>3)&7) {
                     case 0: ED->q[0] = add64(emu, ED->q[0], tmp64u); break;
@@ -881,7 +881,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         case 7:                cmp32(emu, ED->dword[0], tmp64u); break;
                     }
             }
-            pthread_mutex_unlock(&emu->context->mutex_lock);
+            pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
             break;
 
@@ -904,12 +904,12 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
             GETEB(0);
             GETGB;
             if(!MODREG)
-                pthread_mutex_lock(&emu->context->mutex_lock); // XCHG always LOCK (but when accessing memory only)
+                pthread_mutex_lock(&my_context->mutex_lock); // XCHG always LOCK (but when accessing memory only)
             tmp8u = GB;
             GB = EB->byte[0];
             EB->byte[0] = tmp8u;
             if(!MODREG)
-                pthread_mutex_unlock(&emu->context->mutex_lock);
+                pthread_mutex_unlock(&my_context->mutex_lock);
 #endif                
             break;
         case 0x87:                      /* XCHG Ed,Gd */
@@ -937,7 +937,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
 #else
             GETED(0);
             GETGD;
-            pthread_mutex_lock(&emu->context->mutex_lock);
+            pthread_mutex_lock(&my_context->mutex_lock);
             if(rex.w) {
                 tmp64u = GD->q[0];
                 GD->q[0] = ED->q[0];
@@ -950,7 +950,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                 else
                     ED->dword[0] = tmp32u;
             }
-            pthread_mutex_unlock(&emu->context->mutex_lock);
+            pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
             break;            
 
@@ -966,9 +966,9 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         tmp8u2 = not8(emu, tmp8u2);
                     } while(native_lock_write_b(EB, tmp8u2));
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     EB->byte[0] = not8(emu, EB->byte[0]);
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 default:
@@ -986,9 +986,9 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         tmp8u = native_lock_read_b(ED);
                     } while(native_lock_write_b(ED, inc8(emu, tmp8u)));
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     ED->byte[0] = inc8(emu, ED->byte[0]);
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 case 1:                 /* DEC Ed */
@@ -997,9 +997,9 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         tmp8u = native_lock_read_b(ED);
                     } while(native_lock_write_b(ED, dec8(emu, tmp8u)));
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     ED->byte[0] = dec8(emu, ED->byte[0]);
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 default:
@@ -1047,7 +1047,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         if(MODREG) ED->dword[1] = 0;
                     }
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     if(rex.w) {
                         ED->q[0] = inc64(emu, ED->q[0]);
                     } else {
@@ -1056,7 +1056,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         else
                             ED->dword[0] = inc32(emu, ED->dword[0]);
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 case 1:                 /* DEC Ed */
@@ -1082,7 +1082,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         if(MODREG) ED->dword[1] = 0;
                     }
 #else
-                    pthread_mutex_lock(&emu->context->mutex_lock);
+                    pthread_mutex_lock(&my_context->mutex_lock);
                     if(rex.w) {
                         ED->q[0] = dec64(emu, ED->q[0]);
                     } else {
@@ -1091,7 +1091,7 @@ uintptr_t RunF0(x64emu_t *emu, rex_t rex, uintptr_t addr)
                         else
                             ED->dword[0] = dec32(emu, ED->dword[0]);
                     }
-                    pthread_mutex_unlock(&emu->context->mutex_lock);
+                    pthread_mutex_unlock(&my_context->mutex_lock);
 #endif
                     break;
                 default:

src/emu/x64syscall.c

@@ -62,6 +62,7 @@ int32_t my_epoll_pwait(x64emu_t* emu, int32_t epfd, void* events, int32_t maxeve
 pid_t my_vfork(x64emu_t* emu);
 #endif
 int32_t my_fcntl(x64emu_t* emu, int32_t a, int32_t b, void* c);
+int32_t my_execve(x64emu_t* emu, const char* path, char* const argv[], char* const envp[]);
 
 // cannot include <fcntl.h>, it conflict with some asm includes...
 #ifndef O_NONBLOCK
@@ -136,6 +137,7 @@ static scwrap_t syscallwrap[] = {
     { 57, __NR_fork, 0 },    // should wrap this one, because of the struct pt_regs (the only arg)?
     #endif
     //{58, __NR_vfork, 0},
+    //{59, __NR_execve, 3},
     { 60, __NR_exit, 1},    // Nees wrapping?
     { 61, __NR_wait4, 4},
     { 62, __NR_kill, 2 },
@@ -181,9 +183,11 @@ static scwrap_t syscallwrap[] = {
     { 127, __NR_rt_sigpending, 2},
     { 128, __NR_rt_sigtimedwait, 4},
     //{ 131, __NR_sigaltstack, 2},  // wrapped to use my_sigaltstack*
+    { 140, __NR_getpriority, 2},
     { 155, __NR_pivot_root, 2},
     { 157, __NR_prctl, 5 },     // needs wrapping?
     //{ 158, __NR_arch_prctl, 2},   //need wrapping
+    { 160, __NR_setrlimit, 2},
     { 161, __NR_chroot, 1},
     { 186, __NR_gettid, 0 },    //0xBA
     { 200, __NR_tkill, 2 },
@@ -234,6 +238,7 @@ static scwrap_t syscallwrap[] = {
     { 264, __NR_renameat, 4},
     #endif
     { 267, __NR_readlinkat, 4},
+    { 268, __NR_fchmodat, 3},
     { 270, __NR_pselect6, 6},
     { 272, __NR_unshare, 1},
     { 273, __NR_set_robust_list, 2},
@@ -599,6 +604,14 @@ void EXPORT x64Syscall(x64emu_t *emu)
                     R_RAX = (uint64_t)-errno;
             }
             break;
+        case 59:   // execve
+            {
+                int64_t r = my_execve(emu, (const char*)R_RDI, (char* const*)R_RSI, (char* const*)R_RDX);
+                R_RAX = r;
+                if(R_EAX==0xffffffff)
+                    R_RAX = (uint64_t)-errno;
+            }
+            break;
         case 63:    //uname
             {
                 old_utsname_t *old = (old_utsname_t*)R_RDI;
@@ -895,6 +908,8 @@ uintptr_t EXPORT my_syscall(x64emu_t *emu)
         #endif
         case 58:   // vfork
             return my_vfork(emu);
+        case 59:   // execve
+            return my_execve(emu, (const char*)R_RSI, (char* const*)R_RDX, (char* const*)R_RCX);
         case 63:    //uname
             {
                 old_utsname_t *old = (old_utsname_t*)R_RSI;

src/emu/x64tls.c

@@ -130,9 +130,21 @@ uint32_t my_modify_ldt(x64emu_t* emu, int op, thread_area_t* td, int size)
 }
 
 static void* GetSeg43Base();
+static const char* arch_prctl_param(int code)
+{
+    static char ret[10] = {0};
+    switch (code) {
+        case 0x1001: return "ARCH_SET_GS";
+        case 0x1002: return "ARCH_SET_FS";
+        case 0x1003: return "ARCH_GET_FS";
+        case 0x1004: return "ARCH_GET_GS";
+    }
+    sprintf(ret, "0x%x", code);
+    return ret;
+}
 int my_arch_prctl(x64emu_t *emu, int code, void* addr)
 {
-    printf_log(LOG_DEBUG, "%04d| arch_prctl(0x%x, %p) (RSP=%p, FS=0x%x, GS=0x%x)\n", GetTID(), code, addr,(void*)R_RSP, emu->segs[_FS], emu->segs[_GS]);
+    printf_log(LOG_DEBUG, "%04d| arch_prctl(%s, %p) (RSP=%p, FS=0x%x, GS=0x%x)\n", GetTID(), arch_prctl_param(code), addr,(void*)R_RSP, emu->segs[_FS], emu->segs[_GS]);
 
     #define ARCH_SET_GS          0x1001
     #define ARCH_SET_FS          0x1002
@@ -152,12 +164,12 @@ int my_arch_prctl(x64emu_t *emu, int code, void* addr)
             seg=(code==ARCH_SET_FS)?_FS:_GS;
             int idx = -1;
             // search if it's a TLS base
+            if(GetSeg43Base()==addr)
+                idx = 0x43>>3;
             // Is this search only occurs when seg==0?
             for (int i=9; i<15 && idx==-1; ++i)
                 if(my_context->segtls[i].present && my_context->segtls[i].base==(uintptr_t)addr)
                     idx=i;
-            if(idx==-1 && GetSeg43Base()==addr)
-                idx = 0x43>>3;
             // found...
             if(idx!=-1) {
                printf_log(LOG_DEBUG, "Changing segment selector from 0x%x to 0x%x\n", emu->segs[seg], (idx<<3) +3);