libos+LICENSE: Update copyright year

Happy new year!

.gitea/workflows/build.yaml

@@ -11,11 +11,13 @@ jobs:
       - name: Download dependencies
         run: |
           apt update
-          apt install -y cmake ninja-build nasm genext2fs qemu-system build-essential wget git
+          apt install -y cmake ninja-build nasm genext2fs qemu-system build-essential wget git clang-format
       - name: Set up the toolchain
         run: |
           wget https://pub.cloudapio.eu/luna/toolchains/ci-toolchain-arm64.tar.gz --quiet
           tar xf ci-toolchain-arm64.tar.gz
           rm ci-toolchain-arm64.tar.gz
+      - name: Check formatting
+        run: tools/check-formatting.sh
       - name: Build and run tests
         run: tools/run-tests.sh

LICENSE

@@ -1,6 +1,6 @@
 BSD 2-Clause License
 
-Copyright (c) 2022-2024, apio.
+Copyright (c) 2022-2025, apio.
 All rights reserved.
 
 Redistribution and use in source and binary forms, with or without

README.md

@@ -20,22 +20,20 @@ A simple POSIX-based operating system for 64-bit computers, written in C++.
 ## Screenshot
 ![Screenshot as of 0.6.0](docs/screenshots/screenshot-0.6.0.png)
 
+## System requirements and dependencies
+
+Read [docs/dependencies.md](docs/dependencies.md) for the full information. In short, all modern Unixes should work, provided the dependencies are available.
+
 ## Setup
 
 To build and run Luna, you will need to build a [cross-compiler](https://wiki.osdev.org/Why_do_I_need_a_Cross_Compiler) and cross-binutils for `x86_64-luna`.
 
-For this, you should start by installing the [required dependencies](https://wiki.osdev.org/GCC_Cross_Compiler#Installing_Dependencies).
-
-Then, run `tools/setup.sh` to build the toolchain.
+There is a script provided for this. Run `tools/setup.sh` to build the toolchain.
 
 Please beware that building GCC and Binutils can take some time, depending on your machine.
 
 ## Running
 
-To run Luna in a virtual machine, you should have [QEMU](https://www.qemu.org/) installed.
-
-Additionally, the build process needs some extra dependencies to run: `cmake`, `ninja`, `nasm`, `fakeroot` and `genext2fs`.
-
 `tools/run.sh` is the script you should use in most cases. It will build changed files, install, make an ISO image, and run Luna in QEMU.
 
 If you have no toolchain set up, `run.sh` will build it automatically, which means that you don't necessarily have to run `setup.sh` manually since `run.sh` does it for you.
@@ -44,7 +42,7 @@ If you have no toolchain set up, `run.sh` will build it automatically, which mea
 
 For development convenience, the system automatically starts a GUI session as the default user, without prompting for a password.
 
-Despite this, Luna does have a login window built-in. If you'd like to try this feature out or start a GUI session as a different user, you'll need to edit [base/etc/init/99-login](base/etc/init/99-login) and change the line that says `Command=/usr/bin/loginui --autologin=selene` to `Command=/usr/bin/loginui`.
+Despite this, Luna does have a login window built-in. If you'd like to try this feature out or start a GUI session as a different user, you'll need to edit [base/etc/loginui.conf](base/etc/loginui.conf) and change the line that says `Autologin=true` to `Autologin=false`.
 
 ## Prebuilt images
 

base/etc/init/99-login

@@ -1,6 +1,6 @@
 Name=login
 Description=Start a graphical user session.
-Command=/usr/bin/loginui --autologin=selene
+Command=/usr/bin/loginui
 StandardOutput=/dev/uart0
 StandardError=/dev/uart0
 Restart=true

base/etc/loginui.conf

@@ -0,0 +1,5 @@
+# Configuration file for loginui.
+# If this parameter is set to "true", loginui automatically spawns a UI session as the below user instead of prompting for a username and password.
+Autologin=true
+# The user to create a session for if "Autologin" is set to true (see above). If the username is invalid, loginui will behave as if "Autologin" was set to false.
+AutologinUser=selene

base/etc/shadow

@@ -1,3 +1,3 @@
-root:toor:0:0:99999:7:::
+root:ce5ca673d13b36118d54a7cf13aeb0ca012383bf771e713421b4d1fd841f539a:0:0:99999:7:::
 wind:!:0:0:99999:7:::
-selene:moon:0:0:99999:7:::
+selene:9e78b43ea00edcac8299e0cc8df7f6f913078171335f733a21d5d911b6999132:0:0:99999:7:::

base/etc/skel/welcome

@@ -1,14 +1,26 @@
 Welcome to the Luna operating system!
 You are running on the default user account, selene.
 
-If you are familiar with Unix-style operating systems (like Linux or *BSD), you should be able to use the Luna terminal without much problems.
+If you are familiar with Unix-style operating systems (like Linux or *BSD),
+you should be able to use the Luna terminal without much problems.
 
-Following the traditional Unix filesystem structure, programs are installed in /usr/bin (/bin is a symlink to /usr/bin). The command `ls /bin` will show all commands available on your current Luna installation.
+Following the traditional Unix filesystem structure, 
+programs are installed in /usr/bin (/bin is a symlink to /usr/bin). 
+The command `ls /bin` will show all commands available on 
+your current Luna installation.
 
-Currently, because of driver limitations, the root file system is mounted read-only. Your home folder is writable, but volatile; it is created and populated on boot, and its contents will vanish after a reboot.
+Currently, because of driver limitations, 
+the root file system is mounted read-only.
+Your home folder is writable, but volatile; it is 
+created and populated on boot,
+and its contents will vanish after a reboot.
 
-The system is booted using the 'init' program. You can read its configuration files in the /etc/init directory to learn more about the boot process.
+The system is booted using the 'init' program. 
+You can read its configuration files in the /etc/init directory to 
+learn more about the boot process.
 
-Luna is free software, released under the BSD-2-Clause license. The license is included in the LICENSE file in your home directory.
+Luna is free software, released under the BSD-2-Clause license. 
+The license is included in the LICENSE file in your home directory.
 
-View the source code and read more about Luna at https://git.cloudapio.eu/apio/Luna.
+View the source code and read more about Luna at
+https://git.cloudapio.eu/apio/Luna.

base/etc/user/00-terminal

@@ -1,3 +0,0 @@
-Name=terminal
-Description=Start the terminal.
-Command=/usr/bin/terminal

base/etc/user/00-welcome

@@ -0,0 +1,3 @@
+Name=welcome
+Description=Show a welcome message for the user.
+Command=/usr/bin/editor welcome

base/usr/share/applications/05-editor

@@ -0,0 +1,3 @@
+Name=editor
+Icon=/usr/share/icons/32x32/app-editor.tga
+Command=/usr/bin/editor

docs/boot_process.md

@@ -54,15 +54,15 @@ As soon as the scheduler switches to the `[kinit]` thread, it will never return
 
 `[kinit]` spawns two more kernel threads, `[reap]` and `[oom]`. While `[kinit]` exits before PID 1 is started, `[reap]` and `[oom]` are present throughout the lifetime of a Luna system, and can be seen in the output of `ps`. Let's take a look at what they do.
 
-- `[reap]`: To understand what this thread does, we must take a look at what happens when threads exit on Luna.
+- `[reap]`: To understand what this thread does, we must take a look at what happens when processes exit on Luna.
 
-    _(Relevant files: [kernel/src/main.cpp](../kernel/src/main.cpp#L23), [kernel/src/thread/Scheduler.cpp](../kernel/src/thread/Scheduler.cpp#L205), [kernel/src/thread/Thread.cpp](../kernel/src/thread/Thread.cpp#L105), [kernel/src/sys/waitpid.cpp](../kernel/src/sys/waitpid.cpp#L84))_
+    _(Relevant files: [kernel/src/main.cpp](../kernel/src/main.cpp#L23), [kernel/src/thread/Scheduler.cpp](../kernel/src/thread/Scheduler.cpp#L231), [kernel/src/thread/Thread.cpp](../kernel/src/thread/Thread.cpp#L126), [kernel/src/sys/waitpid.cpp](../kernel/src/sys/waitpid.cpp#L84))_
 
-    When a thread calls the `_exit()` syscall, its state is set to "Exited". This tells the scheduler to avoid switching to it, and the thread's parent is notified, either by sending SIGCHLD or unblocking a blocked `waitpid()` call. The thread remains visible to the rest of the system, and if its parent does not wait for it, it will stay there as a "zombie thread".
+    When a process calls the `_exit()` syscall, all its threads' states are set to "Dying". This tells the scheduler to avoid switching to them, and the process's parent is notified, by sending SIGCHLD and (optionally) unblocking a blocked `waitpid()` call. The process remains visible to the rest of the system, and if its parent does not wait for it, it will stay there as a "zombie process". Meanwhile, the `[reap]` thread runs and collects all the resources from each thread. The process object is still alive (in a "zombie" state), but its threads have been cleaned up.
 
-    When the thread's parent waits for it, its state is instead set to "Dying", and the `[reap]` thread runs. (Kernel threads skip all the "parent notifying" shenanigans and go straight to the "Dying" state when calling `kernel_exit()`).
+    When the process's parent waits for it, it is marked for reaping (by setting its thread count to -1 (PROCESS_SHOULD_REAP)), and the `[reap]` thread runs.
 
-    The `[reap]` thread then "reaps" all the "Dying" threads' resources. It frees up the threads' memory, file descriptors, and unmaps the memory used for the kernel stack corresponding to that thread. After reaping, the thread is deleted, and no trace of it is left.
+    The `[reap]` thread then "reaps" all the dead processes' resources. It frees up their memory, file descriptors, and other resources. After reaping, the process is deleted, and no trace of it is left.
 
 - `[oom]`: This thread handles Out-Of-Memory (OOM) situations. Whenever the kernel has 1/4 or 1/8 of the available physical memory left (thresholds may be tweaked in the future), or it has run out, it runs this thread.
 
@@ -149,7 +149,7 @@ Currently, there are two service files defined by default in `/etc/init`:
 
 `00-home`: This service sets up a `tmpfs` on `/home/selene`, so that the home directory is writable.
 
-`99-login`: This service starts a graphical session, by calling `/usr/bin/startui`. This service will be restarted if necessary.
+`99-login`: This service starts a graphical session, by calling `/usr/bin/loginui`. This service will be restarted if necessary.
 
 ### File system and process layout
 
@@ -172,24 +172,60 @@ After the init stage of the boot process, the system looks like this:
 [x86_64-io]         - PID 3
 [reap]              - PID 4
 [oom]               - PID 5
-/usr/bin/startui    - PID 13
+/usr/bin/loginui    - PID 13
 ```
 
-_Note: startui is PID 13 because the `00-home` service is a shell script, which starts a few subprocesses. Since Luna does not allow for PID reuse right now, startui ends up with PID 13._
+_Note: loginui is PID 13 because the `00-home` service is a shell script, which starts a few subprocesses. Since Luna does not allow for PID reuse right now, loginui ends up with PID 13._
 
-## Stage 4: startui
+## Stage 4: loginui
+_Relevant files: [gui/loginui.cpp](../gui/loginui.cpp), [gui/wind/main.cpp](../gui/wind/main.cpp)_
+
+`/usr/bin/loginui`'s job is quite simple: it prompts the user to log in with their password, after which a graphical session is started.
+
+_Note: On development builds, Autologin=true is added to /etc/loginui.conf which disables password prompting and executes startui directly._
+
+First, loginui starts the display server, `/usr/bin/wind`, so that it can use its capabilities to show a graphical login prompt. It is started with permissions `root:root`, and later drops privileges to `wind:wind`.
+
+After that, loginui prompts for a username and password, checks it against the hashed password stored in `/etc/shadow`, and finally executes `/usr/bin/startui` which does the actual heavy work of starting all the services needed for a UI session.
+
+### File system and process layout
+
+After the loginui stage of the boot process, the system looks like this:
+
+#### File system
+```
+/ - ext2 root partition
+    /dev - device file system
+        /dev/shm - POSIX shared memory file system
+        /dev/pts - POSIX pseudoterminal file system
+    /tmp - system temporary file directory
+    /usr, /etc, /home... - other directories contained in the root partition
+        /home/selene - temporary home directory
+```
+
+#### Processes
+```
+/usr/bin/init           - PID 1
+[x86_64-io]             - PID 3
+[reap]                  - PID 4
+[oom]                   - PID 5
+/usr/bin/startui        - PID 13
+/usr/bin/wind           - PID 14
+```
+
+## Stage 5: startui
 _Relevant files: [system/startui.cpp](../system/startui.cpp), [gui/wind/main.cpp](../gui/wind/main.cpp)_
 
 `/usr/bin/startui` starts a graphical user session.
 
 A Luna graphical user session includes the following components:
 
-- The display server itself, `/usr/bin/wind`. It is started with permissions `root:root`, and later drops privileges to `wind:wind`.
-- The launch server (`/usr/bih/launch`), which starts processes and keeps them alive on behalf of other processes. It is started with the standard permissions `selene:selene`.
-- The taskbar, `/usr/bin/taskbar`. It is started with the standard permissions `selene:selene`, plus an extra group `wsys` to be able to connect to a special display server socket.
+- The display server itself, `/usr/bin/wind`. If not already started by loginui, `startui` makes sure it's running.
+- The execution server (`/usr/bin/execd`), which starts processes and keeps them alive on behalf of other processes. It is started with the standard permissions `selene:selene`.
+- The taskbar, `/usr/bin/taskbar`. It is started with the standard permissions `selene:selene`, plus an extra group `wsys` to be able to connect to a special display server socket (`/tmp/wsys.sock`, as opposed to the standard `/tmp/wind.sock`). This grants it the ability to use advanced wind features, such as placing the taskbar window behind all other windows.
 - The init process corresponding to that session (`/usr/bin/init --user`). This process does the same thing as `init` above (manages services), but runs with user privileges and reads configuration files from `/etc/user` instead (in the future this will be changed to a user-specific directory).
 
-Currently, `init --user` only does one thing: it opens up a terminal window on startup. It can be configured to do whatever the user desires to do on startup, by placing the appropriate configuration files in `/etc/user`.
+Currently, `init --user` only does one thing: it opens up a text editor with a welcome message on startup. It can be configured to do whatever the user desires to do on startup, by placing the appropriate configuration files in `/etc/user`.
 
 ### File system and process layout
 
@@ -214,8 +250,7 @@ After the startui stage of the boot process, the system is fully started up and
 [oom]                   - PID 5
 /usr/bin/startui        - PID 13
 /usr/bin/wind           - PID 14
-/usr/bin/launch         - PID 15
+/usr/bin/execd          - PID 15
 /usr/bin/taskbar        - PID 16
 /usr/bin/init --user    - PID 17
-/usr/bin/terminal       - PID 18
-/bin/sh                 - PID 19
+/usr/bin/editor welcome - PID 18

docs/dependencies.md

@@ -0,0 +1,35 @@
+# Dependencies required to build and run Luna
+
+## System requirements
+
+Any modern UNIX-like system that supports all the tools listed below should work (Hopefully, that will include Luna itself in the future!). 
+
+I personally build and run Luna on an amd64 Fedora Linux 40 machine. CI runs on arm64 Ubuntu 22.04. Any other configurations are untested. Windows is not supported, although you can try using WSL if you really want to.
+
+## Building a cross-compiler toolchain
+For this, you should start by installing the [required dependencies](https://wiki.osdev.org/GCC_Cross_Compiler#Installing_Dependencies) for any OSdev cross-compiler build.
+
+Also make sure you have the perl module `File::Compare` installed, it is required to build autoconf. On Fedora you can install it using the package manager by running `# dnf install perl-File-Compare`. If your distro doesn't have it, you might have to install it via `cpan`.
+
+## Building the actual system
+The build process needs some extra dependencies to run: `cmake`, `ninja`, `nasm`, `fakeroot` and `genext2fs`. On some distributions the `ninja` package is called `ninja-build` instead.
+
+If you want to use `make` instead of `ninja`, create a file called `env-local.sh` in the project root and add the line `USE_MAKE=1`. In this case, ninja does not need to be installed.
+
+## Running the built image in a virtual machine
+The script provided by the project to run the system, `tools/run.sh`, assumes that QEMU is installed and uses that to run the image. Therefore, make sure your system has `qemu-system-x86_64` in the PATH. If it doesn't, install it using the method appropriate for your system, usually installing `qemu` or `qemu-system` from the package manager.
+
+That being said, there's no requirement to use QEMU. If you want to use a different virtualization program, such as Oracle VirtualBox or VMWare, just use `tools/build-iso.sh` instead of `run.sh` and use the built `Luna.iso` in those programs.
+
+## Formatting/linting
+
+Please make sure you have `clang-format` installed. Additionally, if your editor does not support format-on-save or you do not have it configured, please run `tools/run-clang-format.sh` before committing, to make sure all code follows the same style conventions.
+
+## Source dependencies
+TLDR: Luna does not depend on any third-party library.
+
+Every part of Luna is written from scratch and depends only on its own libraries and programs, with two small exceptions (included here for crediting and licensing purposes, but there is no need to download and build them separately):
+
+The bootloader, BOOTBOOT. It is available at [gitlab.com/bztsrc/bootboot](https://gitlab.com/bztsrc/bootboot), under the MIT license. It is automatically pulled and built from source by `tools/setup.sh`.
+
+[libc/src/strtod.cpp](../libc/src/strtod.cpp). Written by Yasuhiro Matsumoto, adapted from https://gist.github.com/mattn/1890186 and available under a public domain license.

gui/CMakeLists.txt

@@ -11,7 +11,7 @@ add_subdirectory(libui)
 add_subdirectory(wind)
 add_subdirectory(apps)
 
-luna_service(launch.cpp launch)
+luna_service(execd.cpp execd)
 luna_service(run.cpp run)
 luna_service(loginui.cpp loginui)
 target_link_libraries(loginui PRIVATE ui)

gui/apps/2048.cpp

@@ -1,4 +1,4 @@
-#include <luna/String.h>
+#include <luna/RefString.h>
 #include <luna/Utf8.h>
 #include <stdlib.h>
 #include <time.h>
@@ -325,7 +325,7 @@ class GameWidget final : public ui::Widget
 
         canvas.fill(colors[tile.color]);
 
-        auto fmt = TRY(String::format("%d"_sv, tile.number));
+        auto fmt = TRY(RefString::format("%d"_sv, tile.number));
 
         auto font = ui::Font::default_bold_font();
         auto rect = ui::align({ 0, 0, canvas.width, canvas.height },

gui/apps/editor/EditorWidget.cpp

@@ -10,10 +10,12 @@
 #include "EditorWidget.h"
 #include <ctype.h>
 #include <luna/PathParser.h>
+#include <luna/RefString.h>
 #include <luna/Utf8.h>
 #include <os/File.h>
 #include <os/FileSystem.h>
 #include <ui/App.h>
+#include <ui/Dialog.h>
 
 EditorWidget::EditorWidget(SharedPtr<ui::Font> font) : ui::TextInput(), m_font(font)
 {
@@ -27,7 +29,8 @@ Result<void> EditorWidget::load_file(const os::Path& path)
 
     if (!rc.has_error() && !S_ISREG(st.st_mode))
     {
-        os::eprintln("editor: not loading %s as it is not a regular file", path.name().chars());
+        auto message = TRY(RefString::format("%s is not a regular file", path.name().chars()));
+        ui::Dialog::show_message("Error", message.view());
         return {};
     }
 
@@ -41,9 +44,9 @@ Result<void> EditorWidget::load_file(const os::Path& path)
 
     m_cursor = m_data.size();
 
-    m_path = path;
+    m_path = TRY(String::from_string_view(path.name()));
 
-    auto basename = TRY(PathParser::basename(m_path.name()));
+    auto basename = TRY(PathParser::basename(m_path.view()));
 
     String title = TRY(String::format("Text Editor - %s"_sv, basename.chars()));
     window()->set_title(title.view());
@@ -126,15 +129,38 @@ Result<ui::EventResult> EditorWidget::handle_key_event(const ui::KeyEventRequest
     return ui::EventResult::DidHandle;
 }
 
+Result<void> EditorWidget::save_file_as()
+{
+    ui::Dialog::show_input_dialog(
+        "Save file as...", "Please enter the path to save this file to:", [this](StringView path) {
+            m_path = String::from_string_view(path).release_value();
+            auto rc = save_file();
+            if (rc.has_error())
+            {
+                os::eprintln("Failed to save file %s: %s", m_path.chars(), rc.error_string());
+                ui::Dialog::show_message("Error", "Failed to save file");
+            }
+            else
+            {
+                auto basename = PathParser::basename(m_path.view()).release_value();
+
+                String title = String::format("Text Editor - %s"_sv, basename.chars()).release_value();
+                window()->set_title(title.view());
+            }
+        });
+
+    return {};
+}
+
 Result<void> EditorWidget::save_file()
 {
-    if (m_path.is_empty_path())
+    if (m_path.is_empty())
     {
-        os::eprintln("editor: no file to save buffer to!");
-        return err(ENOENT);
+        TRY(save_file_as());
+        return {};
     }
 
-    auto file = TRY(os::File::open(m_path, os::File::WriteOnly));
+    auto file = TRY(os::File::open_or_create(m_path.view(), os::File::WriteOnly));
     return file->write(m_data);
 }
 

gui/apps/editor/EditorWidget.h

@@ -21,14 +21,15 @@ class EditorWidget : public ui::TextInput
     Result<void> load_file(const os::Path& path);
 
     Result<void> save_file();
+    Result<void> save_file_as();
 
     Result<ui::EventResult> handle_key_event(const ui::KeyEventRequest& request) override;
 
     Result<void> draw(ui::Canvas& canvas) override;
 
-    os::Path& path()
+    os::Path path()
     {
-        return m_path;
+        return m_path.view();
     }
 
   private:
@@ -41,7 +42,7 @@ class EditorWidget : public ui::TextInput
     };
     Vector<Line> m_lines;
 
-    os::Path m_path { AT_FDCWD };
+    String m_path;
 
     Result<void> recalculate_lines();
     void recalculate_cursor_position();

gui/apps/editor/main.cpp

@@ -11,6 +11,7 @@
 #include <os/ArgumentParser.h>
 #include <os/File.h>
 #include <ui/App.h>
+#include <ui/Dialog.h>
 
 Result<int> luna_main(int argc, char** argv)
 {
@@ -32,15 +33,20 @@ Result<int> luna_main(int argc, char** argv)
 
     auto* editor = TRY(make<EditorWidget>(ui::Font::default_font()));
     window->set_main_widget(*editor);
-    if (!path.is_empty()) TRY(editor->load_file(path));
+    if (!path.is_empty()) editor->load_file(path);
 
     TRY(window->add_keyboard_shortcut({ moon::K_CH26, ui::Mod_Ctrl }, true, [&](ui::Shortcut) {
         auto result = editor->save_file();
-        if (result.has_error()) os::eprintln("editor: failed to save file: %s", result.error_string());
-        else
-            os::println("editor: buffer saved to %s successfully", editor->path().name().chars());
+        if (result.has_error())
+        {
+            os::eprintln("Failed to save file %s: %s", editor->path().name().chars(), result.error_string());
+            ui::Dialog::show_message("Error", "Failed to save file");
+        }
     }));
 
+    TRY(window->add_keyboard_shortcut({ moon::K_CH26, ui::Mod_Ctrl | ui::Mod_Shift }, true,
+                                      [&](ui::Shortcut) { editor->save_file_as(); }));
+
     window->draw();
 
     return app.run();

gui/apps/taskbar.cpp

@@ -19,11 +19,6 @@ static constexpr ui::Color TASKBAR_COLOR = ui::Color::from_rgb(83, 83, 83);
 
 static OwnedPtr<os::IPC::Client> launcher_client;
 
-void sigchld_handler(int)
-{
-    wait(nullptr);
-}
-
 void sigquit_handler(int)
 {
     // Reload the taskbar by exec-ing the executable, resetting everything.
@@ -117,11 +112,11 @@ Result<int> luna_main(int, char**)
 {
     ui::App app;
     TRY(app.init("/tmp/wsys.sock"));
+    app.pledge(ui::Pledge::ExtendedLayers);
 
-    TRY(os::EventLoop::the().register_signal_handler(SIGCHLD, sigchld_handler));
     TRY(os::EventLoop::the().register_signal_handler(SIGQUIT, sigquit_handler));
 
-    launcher_client = TRY(os::IPC::Client::connect("/tmp/launch.sock", false));
+    launcher_client = TRY(os::IPC::Client::connect("/tmp/execd.sock", false));
 
     ui::Rect screen = app.screen_rect();
 
@@ -131,7 +126,8 @@ Result<int> luna_main(int, char**)
     app.set_main_window(window);
 
     window->set_background(TASKBAR_COLOR);
-    window->set_special_attributes(ui::UNFOCUSEABLE);
+    window->set_layer(ui::Layer::Background);
+    app.pledge(0);
 
     ui::HorizontalLayout layout(ui::Margins { 0, 0, 0, 0 }, ui::AdjustHeight::Yes, ui::AdjustWidth::No);
     window->set_main_widget(layout);

gui/execd.cpp

@@ -1,5 +1,5 @@
 /**
- * @file launch.cpp
+ * @file execd.cpp
  * @author apio (cloudapio.eu)
  * @brief Background process that handles detached launching of apps.
  *
@@ -42,24 +42,24 @@ void handle_ipc_message(os::IPC::ClientConnection& client, u8 id, void*)
     switch (id)
     {
     case os::Launcher::LAUNCH_DETACHED_ID: handle_launch_detached_message(client); break;
-    default: os::eprintln("launch: Invalid IPC message from client!"); return;
+    default: os::eprintln("execd: Invalid IPC message from client!"); return;
     }
 }
 
 void sigchld_handler(int)
 {
-    os::Process::wait(os::Process::ANY_CHILD, nullptr);
+    os::Process::wait(os::Process::ANY_CHILD, nullptr, WNOHANG);
 }
 
 Result<int> luna_main(int argc, char** argv)
 {
     TRY(os::Security::pledge("stdio wpath cpath unix proc exec", NULL));
 
-    StringView socket_path = "/tmp/launch.sock";
+    StringView socket_path = "/tmp/execd.sock";
 
     os::ArgumentParser parser;
     parser.add_description("Background process that handles detached launching of apps."_sv);
-    parser.add_system_program_info("launch"_sv);
+    parser.add_system_program_info("execd"_sv);
     parser.parse(argc, argv);
 
     signal(SIGCHLD, sigchld_handler);
@@ -87,7 +87,7 @@ Result<int> luna_main(int argc, char** argv)
         if (fds[0].revents & POLLIN)
         {
             auto client = TRY(server->accept());
-            os::println("launch: New client connected!");
+            os::println("execd: New client connected!");
             TRY(fds.try_append({ .fd = client.fd(), .events = POLLIN, .revents = 0 }));
 
             auto connection = TRY(os::IPC::ClientConnection::adopt_connection(move(client)));
@@ -99,7 +99,7 @@ Result<int> luna_main(int argc, char** argv)
             if (fds[i + 1].revents & POLLIN) clients[i]->check_for_messages();
             if (fds[i + 1].revents & POLLHUP)
             {
-                os::println("launch: Client %zu disconnected", i);
+                os::println("execd: Client %zu disconnected", i);
                 fds.remove_at(i + 1);
                 auto client = clients.remove_at(i);
                 client->disconnect();

gui/libui/CMakeLists.txt

@@ -19,6 +19,7 @@ set(SOURCES
     src/Label.cpp
     src/InputField.cpp
     src/TextInput.cpp
+    src/Dialog.cpp
 )
 
 add_library(ui ${SOURCES})

gui/libui/include/ui/App.h

@@ -49,6 +49,8 @@ namespace ui
             return m_main_window;
         }
 
+        void pledge(i16 pledges);
+
         Result<void> register_window(OwnedPtr<Window>&& window, Badge<Window>);
         void unregister_window(Window* window, Badge<Window>);
 
@@ -61,6 +63,7 @@ namespace ui
         HashMap<int, OwnedPtr<Window>> m_windows;
         bool m_should_close { false };
         os::EventLoop m_loop;
+        Vector<int> m_window_clear_queue;
 
         bool process_events();
 

gui/libui/include/ui/Button.h

@@ -8,7 +8,7 @@
  */
 
 #pragma once
-#include <os/Action.h>
+#include <luna/Action.h>
 #include <ui/Widget.h>
 
 namespace ui
@@ -19,7 +19,7 @@ namespace ui
         Button(Rect rect);
 
         void set_widget(Widget& widget);
-        void set_action(os::Action&& action);
+        void set_action(Action&& action);
 
         Result<EventResult> handle_mouse_move(Point position) override;
         Result<EventResult> handle_mouse_leave() override;
@@ -32,6 +32,6 @@ namespace ui
         bool m_hovered { false };
         bool m_clicked { false };
         Widget* m_child;
-        os::Action m_action;
+        Action m_action;
     };
 }

gui/libui/include/ui/Dialog.h

@@ -0,0 +1,22 @@
+/**
+ * @file Window.h
+ * @author apio (cloudapio.eu)
+ * @brief UI window dialogs.
+ *
+ * @copyright Copyright (c) 2024, the Luna authors.
+ *
+ */
+
+#pragma once
+#include <luna/Action.h>
+#include <ui/Window.h>
+
+namespace ui
+{
+    namespace Dialog
+    {
+        Result<void> show_message(StringView title, StringView message);
+
+        Result<void> show_input_dialog(StringView title, StringView message, Function<StringView> callback);
+    }
+}

gui/libui/include/ui/InputField.h

@@ -8,13 +8,13 @@
  */
 
 #pragma once
-#include <os/Action.h>
+#include <luna/Action.h>
 #include <ui/Font.h>
 #include <ui/TextInput.h>
 
 namespace ui
 {
-    class InputField : public ui::TextInput
+    class InputField final : public ui::TextInput
     {
       public:
         InputField(SharedPtr<ui::Font> font);
@@ -27,7 +27,7 @@ namespace ui
 
         StringView data();
 
-        void on_submit(os::Function<StringView>&& action)
+        void on_submit(Function<StringView>&& action)
         {
             m_on_submit_action = move(action);
             m_has_on_submit_action = true;
@@ -36,7 +36,7 @@ namespace ui
       private:
         SharedPtr<ui::Font> m_font;
 
-        os::Function<StringView> m_on_submit_action;
+        Function<StringView> m_on_submit_action;
         bool m_has_on_submit_action { false };
     };
 }

gui/libui/include/ui/Window.h

@@ -65,7 +65,7 @@ namespace ui
 
         void close();
 
-        void set_special_attributes(WindowAttributes attributes);
+        void set_layer(Layer layer);
 
         Result<void> draw();
         Result<ui::EventResult> handle_mouse_leave();
@@ -73,13 +73,19 @@ namespace ui
         Result<ui::EventResult> handle_mouse_buttons(ui::Point position, int buttons);
         Result<ui::EventResult> handle_key_event(const ui::KeyEventRequest& request);
 
-        Result<void> add_keyboard_shortcut(ui::Shortcut shortcut, bool intercept, os::Function<ui::Shortcut>&& action);
+        Result<void> add_keyboard_shortcut(ui::Shortcut shortcut, bool intercept, Function<ui::Shortcut>&& action);
 
         int id() const
         {
             return m_id;
         }
 
+        void on_close(Action&& action)
+        {
+            m_on_close_action = move(action);
+            m_has_on_close_action = true;
+        }
+
         ~Window();
 
       private:
@@ -93,10 +99,13 @@ namespace ui
         Option<int> m_old_mouse_buttons;
         bool m_decorated { false };
 
+        Action m_on_close_action;
+        bool m_has_on_close_action { false };
+
         struct ShortcutAction
         {
             bool intercept;
-            os::Function<Shortcut> action;
+            Function<Shortcut> action;
         };
 
         HashMap<Shortcut, ShortcutAction> m_shortcuts;

gui/libui/include/ui/ipc/Server.h

@@ -25,7 +25,8 @@ namespace ui
         CLOSE_WINDOW_ID,
         GET_SCREEN_RECT_ID,
         SET_TITLEBAR_HEIGHT_ID,
-        SET_SPECIAL_WINDOW_ATTRIBUTES_ID,
+        SET_WINDOW_LAYER_ID,
+        UPDATE_PLEDGE_REQUEST_ID,
     };
 
     struct CreateWindowRequest
@@ -81,16 +82,32 @@ namespace ui
         int height;
     };
 
-    enum WindowAttributes : u8
+    enum Layer : u8
     {
-        UNFOCUSEABLE = 1,
+        Background,
+        Global,
+        GlobalTop,
+        System,
+        Lock
     };
 
-    struct SetSpecialWindowAttributesRequest
+    struct SetWindowLayer
     {
-        static constexpr u8 ID = SET_SPECIAL_WINDOW_ATTRIBUTES_ID;
+        static constexpr u8 ID = SET_WINDOW_LAYER_ID;
 
         int window;
-        WindowAttributes attributes;
+        Layer layer;
+    };
+
+    enum Pledge : i16
+    {
+        ExtendedLayers = 1,
+    };
+
+    struct UpdatePledgeRequest
+    {
+        static constexpr u8 ID = UPDATE_PLEDGE_REQUEST_ID;
+
+        i16 pledges;
     };
 }

gui/libui/src/App.cpp

@@ -75,7 +75,7 @@ namespace ui
     void App::unregister_window(Window* window, Badge<Window>)
     {
         int id = window->id();
-        check(m_windows.try_remove(id));
+        m_window_clear_queue.try_append(id);
     }
 
     Window* App::find_window(int id)
@@ -124,6 +124,22 @@ namespace ui
     {
         check(m_main_window);
         m_client->check_for_messages().release_value();
+        for (int id : m_window_clear_queue)
+        {
+            check(m_windows.try_remove(id));
+
+            ui::CloseWindowRequest request;
+            request.window = id;
+            client().send_async(request);
+        }
+        m_window_clear_queue.clear_data();
         return !m_should_close;
     }
+
+    void App::pledge(i16 pledges)
+    {
+        ui::UpdatePledgeRequest request;
+        request.pledges = pledges;
+        client().send_async(request);
+    }
 }

gui/libui/src/Button.cpp

@@ -24,7 +24,7 @@ namespace ui
         widget.set_parent(this);
     }
 
-    void Button::set_action(os::Action&& action)
+    void Button::set_action(Action&& action)
     {
         m_action = move(action);
     }

gui/libui/src/Dialog.cpp

@@ -0,0 +1,82 @@
+/**
+ * @file Dialog.cpp
+ * @author apio (cloudapio.eu)
+ * @brief UI window dialogs.
+ *
+ * @copyright Copyright (c) 2024, the Luna authors.
+ *
+ */
+
+#include <luna/Alloc.h>
+#include <ui/App.h>
+#include <ui/Dialog.h>
+#include <ui/InputField.h>
+#include <ui/Label.h>
+#include <ui/Layout.h>
+
+namespace ui::Dialog
+{
+    Result<void> show_message(StringView title, StringView message)
+    {
+        auto rect = ui::App::the().main_window()->canvas().rect();
+        int text_length = (int)message.length() * ui::Font::default_font()->width();
+        int text_height = ui::Font::default_font()->height();
+
+        ui::Rect dialog_rect = { 0, 0, text_length + 20, text_height + 20 };
+
+        auto* dialog = TRY(ui::Window::create(
+            ui::align(rect, dialog_rect, ui::VerticalAlignment::Center, ui::HorizontalAlignment::Center)));
+
+        dialog->set_background(ui::GRAY);
+        dialog->set_title(title);
+
+        ui::Label* text = TRY(make<ui::Label>(message));
+        text->set_color(ui::BLACK);
+        dialog->set_main_widget(*text);
+
+        dialog->on_close([text] { delete text; });
+
+        dialog->draw();
+
+        return {};
+    }
+
+    Result<void> show_input_dialog(StringView title, StringView message, Function<StringView> callback)
+    {
+        auto rect = ui::App::the().main_window()->canvas().rect();
+        int text_length = (int)message.length() * ui::Font::default_font()->width();
+        int text_height = ui::Font::default_font()->height();
+
+        ui::Rect dialog_rect = { 0, 0, max(text_length + 20, 300), text_height * 2 + 30 };
+
+        auto* dialog = TRY(ui::Window::create(
+            ui::align(rect, dialog_rect, ui::VerticalAlignment::Center, ui::HorizontalAlignment::Center)));
+
+        dialog->set_background(ui::GRAY);
+        dialog->set_title(title);
+
+        ui::VerticalLayout* layout = TRY(make<ui::VerticalLayout>());
+        dialog->set_main_widget(*layout);
+
+        ui::Label* text = TRY(make<ui::Label>((message)));
+        text->set_color(ui::BLACK);
+        layout->add_widget(*text);
+
+        ui::InputField* input = TRY(make<ui::InputField>(ui::Font::default_font()));
+        input->on_submit([dialog, callback](StringView s) {
+            callback(s);
+            dialog->close();
+        });
+        layout->add_widget(*input);
+
+        dialog->on_close([layout, text, input] {
+            delete text;
+            delete input;
+            delete layout;
+        });
+
+        dialog->draw();
+
+        return {};
+    }
+}

gui/libui/src/InputField.cpp

@@ -115,7 +115,7 @@ namespace ui
 
     StringView InputField::data()
     {
-        if (!m_data.size()) return StringView {};
-        return StringView { (const char*)m_data.data(), m_data.size() };
+        if (m_data.size() < 2) return StringView {};
+        return StringView { (const char*)m_data.data(), m_data.size() - 1 };
     }
 }

gui/libui/src/Window.cpp

@@ -84,6 +84,8 @@ namespace ui
     Window::~Window()
     {
         if (m_canvas.ptr) munmap(m_canvas.ptr, ((usize)m_canvas.width) * ((usize)m_canvas.height) * 4);
+
+        if (m_has_on_close_action) m_on_close_action();
     }
 
     void Window::set_title(StringView title)
@@ -108,20 +110,16 @@ namespace ui
     {
         App& app = App::the();
 
-        ui::CloseWindowRequest request;
-        request.window = m_id;
-        app.client().send_async(request);
-
         if (this == app.main_window()) app.set_should_close(true);
 
         app.unregister_window(this, {});
     }
 
-    void Window::set_special_attributes(WindowAttributes attributes)
+    void Window::set_layer(Layer layer)
     {
-        ui::SetSpecialWindowAttributesRequest request;
+        ui::SetWindowLayer request;
         request.window = m_id;
-        request.attributes = attributes;
+        request.layer = layer;
         App::the().client().send_async(request);
     }
 
@@ -226,8 +224,7 @@ namespace ui
         return m_main_widget->handle_key_event(request);
     }
 
-    Result<void> Window::add_keyboard_shortcut(ui::Shortcut shortcut, bool intercept,
-                                               os::Function<ui::Shortcut>&& action)
+    Result<void> Window::add_keyboard_shortcut(ui::Shortcut shortcut, bool intercept, Function<ui::Shortcut>&& action)
     {
         TRY(m_shortcuts.try_set(shortcut, { intercept, move(action) }));
 

gui/loginui.cpp

@@ -7,12 +7,15 @@
  *
  */
 
-#include <luna/String.h>
+#include <luna/RefString.h>
+#include <luna/SHA.h>
 #include <os/ArgumentParser.h>
+#include <os/Config.h>
 #include <os/File.h>
 #include <os/FileSystem.h>
 #include <os/IPC.h>
 #include <os/Process.h>
+#include <os/Security.h>
 #include <pwd.h>
 #include <shadow.h>
 #include <sys/stat.h>
@@ -31,18 +34,19 @@ enum Stage
 
 static constexpr ui::Color BACKGROUND_COLOR = ui::Color::from_rgb(89, 89, 89);
 
+Result<String> hash_password(StringView& view)
+{
+    SHA256 sha;
+    sha.append((const u8*)view.chars(), view.length());
+    auto digest = TRY(sha.digest());
+    return digest.to_string();
+}
+
 Result<int> luna_main(int argc, char** argv)
 {
-    StringView username;
-
     os::ArgumentParser parser;
     parser.add_description("Login prompt for a graphical UI session.");
     parser.add_system_program_info("loginui"_sv);
-    // FIXME: Make this a config option instead of a switch.
-    // Also, calling "loginui --autologin=user" is functionally identical to calling "startui --user=user", the only
-    // difference is that it makes the init config easier to change (only adding or removing the autologin flag, instead
-    // of changing the program to use)
-    parser.add_value_argument(username, ' ', "autologin", "login as a specific user without prompting");
     parser.parse(argc, argv);
 
     if (geteuid() != 0)
@@ -51,6 +55,8 @@ Result<int> luna_main(int argc, char** argv)
         return 1;
     }
 
+    TRY(os::Security::pledge("stdio rpath wpath unix proc exec id", nullptr));
+
     setsid();
 
     bool success = os::IPC::Notifier::run_and_wait(
@@ -66,13 +72,20 @@ Result<int> luna_main(int argc, char** argv)
         return 1;
     }
 
-    if (!username.is_empty())
-    {
-        auto flag = String::format("--user=%s"_sv, username.chars()).release_value();
+    auto config = TRY(os::ConfigFile::open("/etc/loginui.conf"));
 
-        StringView startui_command[] = { "/usr/bin/startui", flag.view() };
-        os::Process::exec(startui_command[0], Slice<StringView>(startui_command, 2));
-        unreachable();
+    if (config->read_boolean_or("Autologin", false))
+    {
+        StringView username = config->read_string_or("AutologinUser", "");
+
+        if (!username.is_empty())
+        {
+            auto flag = RefString::format("--user=%s"_sv, username.chars()).release_value();
+
+            StringView startui_command[] = { "/usr/bin/startui", flag.view() };
+            os::Process::exec(startui_command[0], Slice<StringView>(startui_command, 2));
+            unreachable();
+        }
     }
 
     ui::App app;
@@ -117,7 +130,7 @@ Result<int> luna_main(int argc, char** argv)
             stage = Stage::PasswordInput;
             label.set_text("Password:");
 
-            String title = String::format("Log in: %s"_sv, data.chars()).release_value();
+            RefString title = RefString::format("Log in: %s"_sv, data.chars()).release_value();
             window->set_title(title.view());
 
             input.clear();
@@ -152,14 +165,16 @@ Result<int> luna_main(int argc, char** argv)
                 return;
             }
 
-            if (strcmp(data.chars(), passwd))
+            auto result = hash_password(data).release_value();
+
+            if (strcmp(result.chars(), passwd))
             {
                 error.set_text("Incorrect password.");
                 input.clear();
                 return;
             }
 
-            auto flag = String::format("--user=%s"_sv, pw->pw_name).release_value();
+            auto flag = RefString::format("--user=%s"_sv, pw->pw_name).release_value();
 
             StringView startui_command[] = { "/usr/bin/startui", flag.view() };
             os::Process::exec(startui_command[0], Slice<StringView>(startui_command, 2));

gui/run.cpp

@@ -22,7 +22,7 @@ Result<int> luna_main(int argc, char** argv)
     parser.add_positional_argument(program, "program", true);
     parser.parse(argc, argv);
 
-    OwnedPtr<os::IPC::Client> launcher_client = TRY(os::IPC::Client::connect("/tmp/launch.sock", false));
+    OwnedPtr<os::IPC::Client> launcher_client = TRY(os::IPC::Client::connect("/tmp/execd.sock", false));
 
     os::println("Requesting to start program '%s'...", program.chars());
 

gui/wind/CMakeLists.txt

@@ -11,6 +11,9 @@ set(SOURCES
     Keyboard.cpp
     Keyboard.h
     Client.h
+    Client.cpp
+    Layer.cpp
+    Layer.h
 )
 
 add_executable(wind ${SOURCES})

gui/wind/Client.cpp

@@ -0,0 +1,52 @@
+#include "Client.h"
+#include <os/File.h>
+
+Client::Client(OwnedPtr<os::IPC::ClientConnection>&& client, i16 _pledges)
+    : conn(move(client)), windows(), pledges(_pledges)
+{
+    conn->set_message_handler(wind::handle_ipc_message, this);
+}
+
+bool Client::update_pledges(i16 _pledges)
+{
+    if (_pledges < 0)
+    {
+        os::eprintln("wind: Client trying to set an invalid pledge, disconnecting!");
+        should_be_disconnected = true;
+        return false;
+    }
+
+    if (pledges < 0)
+    {
+        pledges = _pledges;
+        return true;
+    }
+
+    if (_pledges & ~pledges)
+    {
+        os::eprintln("wind: Client trying to add pledges, disconnecting!");
+        should_be_disconnected = true;
+        return false;
+    }
+
+    pledges = _pledges;
+    return true;
+}
+
+bool Client::check_pledge(i16 pledge)
+{
+    check(pledge > 0);
+
+    if (pledges < 0)
+    {
+        os::eprintln("wind: Client trying to use pledge-protected functions before pledging anything, disconnecting!");
+        should_be_disconnected = true;
+        return false;
+    }
+
+    if ((pledges & pledge) == pledge) return true;
+
+    os::eprintln("wind: Client trying to use a function they haven't pledged, disconnecting!");
+    should_be_disconnected = true;
+    return false;
+}

gui/wind/Client.h

@@ -3,20 +3,19 @@
 #include "Window.h"
 #include <os/IPC.h>
 
+constexpr i16 HAS_NOT_YET_PLEDGED = -1;
+constexpr i16 EMPTY_PLEDGE = 0;
+
 struct Client
 {
     OwnedPtr<os::IPC::ClientConnection> conn;
     Vector<Window*> windows;
     const bool privileged { false };
     bool should_be_disconnected { false };
+    i16 pledges = 0;
 
-    Client(OwnedPtr<os::IPC::ClientConnection>&& client, bool priv)
-#ifdef CLIENT_IMPLEMENTATION
-        : conn(move(client)), windows(), privileged(priv)
-    {
-        conn->set_message_handler(wind::handle_ipc_message, this);
-    }
-#else
-        ;
-#endif
+    bool update_pledges(i16 pledges);
+    bool check_pledge(i16 pledge);
+
+    Client(OwnedPtr<os::IPC::ClientConnection>&& client, i16 pledges);
 };

gui/wind/IPC.cpp

@@ -1,8 +1,9 @@
 #include "IPC.h"
+#include "Layer.h"
 #include "Mouse.h"
 #include "Screen.h"
 #include <luna/Alignment.h>
-#include <luna/String.h>
+#include <luna/RefString.h>
 #include <os/File.h>
 #include <os/SharedMemory.h>
 #include <sys/mman.h>
@@ -35,9 +36,9 @@ static Result<void> handle_create_window_message(Client& client)
 
     request.rect = request.rect.normalized();
 
-    auto name = TRY_OR_IPC_ERROR(String::from_cstring("Window"));
+    auto name = TRY_OR_IPC_ERROR(RefString::from_cstring("Window"));
 
-    auto shm_path = TRY_OR_IPC_ERROR(String::format("/wind-shm-%d-%lu"_sv, client.conn->fd(), time(NULL)));
+    auto shm_path = TRY_OR_IPC_ERROR(RefString::format("/wind-shm-%d-%lu"_sv, client.conn->fd(), time(NULL)));
 
     auto* window = new (std::nothrow) Window(request.rect, move(name));
     if (!window)
@@ -47,7 +48,7 @@ static Result<void> handle_create_window_message(Client& client)
     }
 
     auto guard = make_scope_guard([window] {
-        g_windows.remove(window);
+        window->layer->windows.remove(window);
         delete window;
     });
 
@@ -88,7 +89,7 @@ static Result<void> handle_set_window_title_message(Client& client)
     ui::SetWindowTitleRequest request;
     if (!TRY(client.conn->read_message(request))) return {};
 
-    auto name = COPY_IPC_STRING(request.title);
+    auto name = TRY(RefString::from_string(COPY_IPC_STRING(request.title)));
 
     os::println("wind: SetWindowTitle(\"%s\") for window %d", name.chars(), request.window);
 
@@ -120,7 +121,7 @@ static Result<void> handle_close_window_message(Client& client)
 
     auto* window = client.windows[request.window];
     client.windows[request.window] = nullptr;
-    g_windows.remove(window);
+    window->layer->windows.remove(window);
     Mouse::the().window_did_close(window);
     delete window;
 
@@ -160,22 +161,48 @@ static Result<void> handle_set_titlebar_height_message(Client& client)
     return {};
 }
 
-static Result<void> handle_set_special_window_attributes_message(Client& client)
+static Result<void> handle_set_window_layer_message(Client& client)
 {
-    ui::SetSpecialWindowAttributesRequest request;
+    ui::SetWindowLayer request;
     if (!TRY(client.conn->read_message(request))) return {};
 
-    if (!client.privileged)
+    if (request.layer != ui::Layer::Global && request.layer != ui::Layer::GlobalTop)
     {
-        os::eprintln(
-            "wind: Unprivileged client trying to call privileged request (SetSpecialWindowAttributes), disconnecting!");
+        if (!client.check_pledge(ui::Pledge::ExtendedLayers)) return {};
+    }
+
+    CHECK_WINDOW_ID(request, "SetWindowLayer");
+
+    auto* window = client.windows[request.window];
+
+    window->layer->windows.remove(window);
+
+    switch (request.layer)
+    {
+    case ui::Layer::Background: window->layer = &l_background; break;
+    case ui::Layer::Global: window->layer = &l_global; break;
+    case ui::Layer::GlobalTop: window->layer = &l_global_top; break;
+    case ui::Layer::System: window->layer = &l_system; break;
+    case ui::Layer::Lock: window->layer = &l_lock; break;
+    default: {
+        window->layer->windows.append(window);
+        os::eprintln("wind: Client trying to set window layer to an invalid layer, disconnecting!");
         client.should_be_disconnected = true;
         return {};
     }
+    }
 
-    CHECK_WINDOW_ID(request, "SetSpecialWindowAttributes");
+    window->layer->windows.append(window);
 
-    client.windows[request.window]->attributes = request.attributes;
+    return {};
+}
+
+static Result<void> handle_update_pledge_request_message(Client& client)
+{
+    ui::UpdatePledgeRequest request;
+    if (!TRY(client.conn->read_message(request))) return {};
+
+    client.update_pledges(request.pledges); // update_pledges does all the checking.
 
     return {};
 }
@@ -194,7 +221,8 @@ namespace wind
         case ui::CLOSE_WINDOW_ID: handle_close_window_message(client); break;
         case ui::GET_SCREEN_RECT_ID: handle_get_screen_rect_message(client); break;
         case ui::SET_TITLEBAR_HEIGHT_ID: handle_set_titlebar_height_message(client); break;
-        case ui::SET_SPECIAL_WINDOW_ATTRIBUTES_ID: handle_set_special_window_attributes_message(client); break;
+        case ui::SET_WINDOW_LAYER_ID: handle_set_window_layer_message(client); break;
+        case ui::UPDATE_PLEDGE_REQUEST_ID: handle_update_pledge_request_message(client); break;
         default: os::eprintln("wind: Invalid IPC message from client!"); return;
         }
     }

gui/wind/Layer.cpp

@@ -0,0 +1,79 @@
+#include "Layer.h"
+#include "Client.h"
+#include "Window.h"
+
+Layer l_background;
+Layer l_global;
+Layer l_global_top;
+Layer l_system;
+Layer l_lock;
+
+constexpr int NUM_LAYERS = 5;
+
+static Layer* const layers_front_to_back[NUM_LAYERS] = { &l_lock, &l_system, &l_global_top, &l_global, &l_background };
+static Layer* const layers_back_to_front[NUM_LAYERS] = { &l_background, &l_global, &l_global_top, &l_system, &l_lock };
+
+Window* Layer::focused_window()
+{
+    for (int i = 0; i < NUM_LAYERS; i++)
+    {
+        Layer* l = layers_front_to_back[i];
+        if (l->windows.last().has_value()) return l->windows.last().value();
+    }
+
+    return nullptr;
+}
+
+void Layer::draw_all_windows(ui::Canvas& canvas)
+{
+    for (int i = 0; i < NUM_LAYERS; i++)
+    {
+        Layer* l = layers_back_to_front[i];
+        for (Window* w : l->windows) { w->draw(canvas); }
+    }
+}
+
+Window* Layer::propagate_mouse_event(ui::Point position, u8 buttons)
+{
+    for (int i = 0; i < NUM_LAYERS; i++)
+    {
+        Layer* l = layers_front_to_back[i];
+        for (Window* window = l->windows.last().value_or(nullptr); window;
+             window = l->windows.previous(window).value_or(nullptr))
+        {
+            if (window->surface.contains(position))
+            {
+                ui::MouseEventRequest request;
+                request.window = window->id;
+                request.position = window->surface.relative(position);
+                request.buttons = buttons;
+                window->client->conn->send_async(request);
+                return window;
+            }
+        }
+    }
+
+    return nullptr;
+}
+
+Window* Layer::propagate_drag_event(ui::Point position)
+{
+    for (int i = 0; i < NUM_LAYERS; i++)
+    {
+        Layer* l = layers_front_to_back[i];
+        for (Window* window = l->windows.last().value_or(nullptr); window;
+             window = l->windows.previous(window).value_or(nullptr))
+        {
+            if (window->surface.contains(position))
+            {
+                window->focus();
+
+                if (window->surface.absolute(window->titlebar).contains(position)) return window;
+
+                return nullptr;
+            }
+        }
+    }
+
+    return nullptr;
+}

gui/wind/Layer.h

@@ -0,0 +1,20 @@
+#pragma once
+#include "Window.h"
+#include <luna/LinkedList.h>
+#include <ui/Canvas.h>
+
+struct Layer
+{
+    LinkedList<Window> windows;
+
+    static Window* focused_window();
+    static void draw_all_windows(ui::Canvas& canvas);
+    static Window* propagate_mouse_event(ui::Point position, u8 buttons);
+    static Window* propagate_drag_event(ui::Point position);
+};
+
+extern Layer l_background;
+extern Layer l_global;
+extern Layer l_global_top;
+extern Layer l_system;
+extern Layer l_lock;

gui/wind/Mouse.cpp

@@ -1,5 +1,6 @@
 #include "Mouse.h"
 #include "Client.h"
+#include "Layer.h"
 #include <os/File.h>
 #include <os/IPC.h>
 #include <ui/Image.h>
@@ -57,45 +58,17 @@ void Mouse::update(const moon::MousePacket& packet)
 
     else if ((packet.buttons & moon::MouseButton::Left) && !m_dragging_window)
     {
-        // Iterate from the end of the list, since windows at the beginning are stacked at the bottom and windows at the
-        // top are at the end.
-        for (Window* window = g_windows.last().value_or(nullptr); window;
-             window = g_windows.previous(window).value_or(nullptr))
+        if (auto* window = Layer::propagate_drag_event(m_position))
         {
-            if (window->surface.contains(m_position))
-            {
-                if (!(window->attributes & ui::UNFOCUSEABLE)) window->focus();
-
-                if (window->surface.absolute(window->titlebar).contains(m_position))
-                {
-                    m_dragging_window = window;
-                    m_initial_drag_position = window->surface.relative(m_position);
-                    os::println("Started drag: window at (%d,%d,%d,%d) with offset (%d,%d)", window->surface.pos.x,
-                                window->surface.pos.y, window->surface.width, window->surface.height,
-                                m_initial_drag_position.x, m_initial_drag_position.y);
-                }
-
-                break;
-            }
+            m_dragging_window = window;
+            m_initial_drag_position = window->surface.relative(m_position);
+            os::println("Started drag: window at (%d,%d,%d,%d) with offset (%d,%d)", window->surface.pos.x,
+                        window->surface.pos.y, window->surface.width, window->surface.height, m_initial_drag_position.x,
+                        m_initial_drag_position.y);
         }
     }
 
-    Window* new_active_window = nullptr;
-
-    for (Window* window = g_windows.last().value_or(nullptr); window;
-         window = g_windows.previous(window).value_or(nullptr))
-    {
-        if (window->surface.contains(m_position))
-        {
-            ui::MouseEventRequest request;
-            request.window = window->id;
-            request.position = window->surface.relative(m_position);
-            request.buttons = packet.buttons;
-            window->client->conn->send_async(request);
-            new_active_window = window;
-            break;
-        }
-    }
+    Window* new_active_window = Layer::propagate_mouse_event(m_position, packet.buttons);
 
     if (m_active_window != new_active_window)
     {

gui/wind/Window.cpp

@@ -1,12 +1,11 @@
 #include "Window.h"
+#include "Layer.h"
 #include <luna/Utf8.h>
 #include <os/File.h>
 #include <sys/mman.h>
 #include <ui/Font.h>
 #include <ui/Image.h>
 
-LinkedList<Window> g_windows;
-
 void Window::draw(ui::Canvas& screen)
 {
     dirty = false;
@@ -18,15 +17,16 @@ void Window::draw(ui::Canvas& screen)
 void Window::focus()
 {
     // Bring the window to the front of the list.
-    g_windows.remove(this);
-    g_windows.append(this);
+    layer->windows.remove(this);
+    layer->windows.append(this);
 }
 
-Window::Window(ui::Rect r, String&& n) : surface(r), name(move(n))
+Window::Window(ui::Rect r, RefString&& n) : surface(r), name(move(n))
 {
     auto font = ui::Font::default_font();
     titlebar = ui::Rect { 0, 0, 0, 0 };
-    g_windows.append(this);
+    l_global.windows.append(this);
+    layer = &l_global;
 }
 
 Window::~Window()

gui/wind/Window.h

@@ -1,31 +1,30 @@
 #pragma once
 #include <luna/LinkedList.h>
-#include <luna/String.h>
+#include <luna/RefString.h>
 #include <ui/Canvas.h>
 #include <ui/Color.h>
 #include <ui/Rect.h>
 #include <ui/ipc/Server.h>
 
 struct Client;
+struct Layer;
 
 struct Window : public LinkedListNode<Window>
 {
     ui::Rect surface;
     ui::Rect titlebar;
     u32* pixels;
-    String name;
-    String shm_path;
+    RefString name;
+    RefString shm_path;
     bool dirty { false };
     Client* client;
+    Layer* layer;
     int id;
-    ui::WindowAttributes attributes { 0 };
 
-    Window(ui::Rect, String&&);
+    Window(ui::Rect, RefString&&);
     ~Window();
 
     void focus();
 
     void draw(ui::Canvas& screen);
 };
-
-extern LinkedList<Window> g_windows;

gui/wind/main.cpp

@@ -1,6 +1,6 @@
-#define CLIENT_IMPLEMENTATION
 #include "Client.h"
 #include "Keyboard.h"
+#include "Layer.h"
 #include "Mouse.h"
 #include "Screen.h"
 #include "Window.h"
@@ -23,42 +23,11 @@ static constexpr uid_t WIND_USER_ID = 2;
 static constexpr gid_t WIND_GROUP_ID = 2;
 static constexpr gid_t WSYS_GROUP_ID = 3;
 
-static void debug(const Vector<OwnedPtr<Client>>& clients)
-{
-    os::println("--- wind: DEBUG OUTPUT ---");
-
-    os::println("-- wind: Listing clients --");
-
-    for (const auto& client : clients)
-    {
-        os::println("Client with fd %d, owns %zu windows", client->conn->fd(), client->windows.size());
-    }
-
-    os::println("-- wind: Listing windows --");
-
-    for (const auto& window : g_windows)
-    {
-        os::println("Window of client (fd %d), id %d, %sdirty (\"%s\") (%d,%d,%d,%d)", window->client->conn->fd(),
-                    window->id, window->dirty ? "" : "not ", window->name.chars(), window->surface.pos.x,
-                    window->surface.pos.y, window->surface.width, window->surface.height);
-    }
-
-    os::println("-- wind: Listing processes --");
-
-    system("ps");
-
-    os::println("-- wind: Listing memory usage --");
-
-    system("free -h");
-
-    os::println("--- wind: END DEBUG OUTPUT ---");
-}
-
 Result<int> luna_main(int argc, char** argv)
 {
     srand((unsigned)time(NULL));
 
-    TRY(os::Security::pledge("stdio rpath wpath cpath unix proc exec tty id", NULL));
+    TRY(os::Security::pledge("stdio rpath wpath cpath unix tty id", NULL));
 
     StringView socket_path = "/tmp/wind.sock";
     StringView system_socket_path = "/tmp/wsys.sock";
@@ -73,8 +42,8 @@ Result<int> luna_main(int argc, char** argv)
 
     if (geteuid() != 0)
     {
-        os::eprintln("error: wind must be run as root to initialize resources, run with --user=<USERNAME> to drop "
-                     "privileges afterwards");
+        os::eprintln("error: wind must be run as root to initialize resources, the server will drop "
+                     "privileges automatically afterwards");
         return 1;
     }
 
@@ -128,12 +97,12 @@ Result<int> luna_main(int argc, char** argv)
     TRY(fds.try_append({ .fd = server->fd(), .events = POLLIN, .revents = 0 }));
     TRY(fds.try_append({ .fd = system_server->fd(), .events = POLLIN, .revents = 0 }));
 
-    TRY(os::Security::pledge("stdio rpath wpath cpath unix proc exec", NULL));
+    TRY(os::Security::pledge("stdio rpath wpath cpath unix", NULL));
 
     while (1)
     {
         screen.canvas().fill(background);
-        for (auto* window : g_windows) window->draw(screen.canvas());
+        Layer::draw_all_windows(screen.canvas());
         mouse_pointer.draw(screen.canvas());
         screen.sync();
 
@@ -153,11 +122,9 @@ Result<int> luna_main(int argc, char** argv)
         {
             moon::KeyboardPacket packet;
             TRY(keyboard->read_typed(packet));
-            if (!packet.released && packet.key == moon::K_Tab) debug(clients);
             auto request = wind::Keyboard::decode_keyboard_event((moon::KeyCode)packet.key, packet.released);
-            if (g_windows.last().has_value())
+            if (auto* window = Layer::focused_window())
             {
-                auto* window = g_windows.last().value();
                 request.window = window->id;
                 window->client->conn->send_async(request);
             }
@@ -170,7 +137,7 @@ Result<int> luna_main(int argc, char** argv)
 
             auto connection = TRY(os::IPC::ClientConnection::adopt_connection(move(client)));
 
-            OwnedPtr<Client> c = TRY(adopt_owned_if_nonnull(new Client(move(connection), false)));
+            OwnedPtr<Client> c = TRY(adopt_owned_if_nonnull(new Client(move(connection), EMPTY_PLEDGE)));
             TRY(clients.try_append(move(c)));
         }
         if (fds[3].revents & POLLIN)
@@ -181,7 +148,7 @@ Result<int> luna_main(int argc, char** argv)
 
             auto connection = TRY(os::IPC::ClientConnection::adopt_connection(move(client)));
 
-            OwnedPtr<Client> c = TRY(adopt_owned_if_nonnull(new Client(move(connection), true)));
+            OwnedPtr<Client> c = TRY(adopt_owned_if_nonnull(new Client(move(connection), HAS_NOT_YET_PLEDGED)));
             TRY(clients.try_append(move(c)));
         }
         for (usize i = 0; i < clients.size(); i++)
@@ -198,7 +165,7 @@ Result<int> luna_main(int argc, char** argv)
                 {
                     if (window)
                     {
-                        g_windows.remove(window);
+                        window->layer->windows.remove(window);
                         mouse_pointer.window_did_close(window);
                         delete window;
                     }

kernel/config.cmake.template

@@ -19,3 +19,6 @@
 # Uncomment the line below to make the kernel also calculate stack traces for userspace addresses on program crashes. 
 # This can aid in debugging, but makes the kernel more unstable as stack tracing will access arbitrary userspace memory.
 # target_compile_definitions(moon PRIVATE MOON_ENABLE_USERSPACE_STACK_TRACES)
+
+# Uncomment the line below to enable all kernel debug messages, and console logging.
+# include(debug.cmake)

kernel/debug.cmake

@@ -12,4 +12,4 @@ target_compile_definitions(moon PRIVATE DEVICE_REGISTRY_DEBUG)
 target_compile_definitions(moon PRIVATE FORK_DEBUG)
 target_compile_definitions(moon PRIVATE MOUNT_DEBUG)
 target_compile_definitions(moon PRIVATE CACHE_DEBUG)
-target_compile_options(moon PRIVATE -fsanitize=undefined)
+#target_compile_options(moon PRIVATE -fsanitize=undefined)

kernel/src/Log.cpp

@@ -77,8 +77,7 @@ static void log_text_console(LogLevel level, const char* format, va_list origin)
         TextConsole::set_foreground(WHITE);
 
     // NOTE: Same as above.
-    auto rc = cstyle_format(
-        format, [](char c, void*) -> Result<void> { return TextConsole::putchar(c); }, nullptr, ap);
+    auto rc = cstyle_format(format, [](char c, void*) -> Result<void> { return TextConsole::putchar(c); }, nullptr, ap);
 
     if (rc.has_error()) { TextConsole::wprint(L"Invalid UTF-8 in log message"); }
 

kernel/src/Pledge.cpp

@@ -1,6 +1,7 @@
 #include "Pledge.h"
 #include "Log.h"
 #include "memory/MemoryManager.h"
+#include "thread/Scheduler.h"
 
 static const char* promise_names[] = {
 #define __enumerate(promise) #promise,
@@ -8,30 +9,34 @@ static const char* promise_names[] = {
 #undef __enumerate
 };
 
-Result<void> check_pledge(Thread* thread, Promise promise)
+Result<void> check_pledge(Process* process, Promise promise)
 {
     // Thread has not called pledge().
-    if (thread->promises < 0) return {};
+    if (process->promises < 0) return {};
     int mask = (1 << (int)promise);
-    if ((thread->promises & mask) != mask)
+    if ((process->promises & mask) != mask)
     {
-        kerrorln("Pledge violation in thread %d! Has not pledged %s", thread->id, promise_names[(int)promise]);
-        if (thread->promises & (1 << (int)Promise::p_error)) return err(ENOSYS);
+        kerrorln("Pledge violation in process %d! Has not pledged %s", process->id, promise_names[(int)promise]);
+        if (process->promises & (1 << (int)Promise::p_error)) return err(ENOSYS);
 
-        // Kill this thread with an uncatchable SIGABRT. For this, we reset the disposition of SIGABRT to the default
-        // (dump core). We could just kill the thread here and be done, but that discards anything on the current stack,
-        // which means that some destructors might not be called. Instead, leave the job to the next call of
-        // Thread::process_pending_signals().
-        thread->signal_handlers[SIGABRT - 1].sa_handler = SIG_DFL;
+        Scheduler::for_each_thread(process, [](Thread* thread) {
+            // Kill this thread with an uncatchable SIGABRT. For this, we reset the disposition of SIGABRT to the
+            // default (dump core). We could just kill the thread here and be done, but that discards anything on the
+            // current stack, which means that some destructors might not be called. Instead, leave the job to the next
+            // call of Thread::process_pending_signals().
+            thread->signal_handlers[SIGABRT - 1].sa_handler = SIG_DFL;
 
-        // Unblock SIGABRT.
-        thread->signal_mask.set(SIGABRT - 1, false);
+            // Unblock SIGABRT.
+            thread->signal_mask.set(SIGABRT - 1, false);
 
-        // If there are any other pending signals, they might be processed before SIGABRT. Avoid that by resetting the
-        // thread's pending signals.
-        thread->pending_signals.clear();
+            // If there are any other pending signals, they might be processed before SIGABRT. Avoid that by resetting
+            // the thread's pending signals.
+            thread->pending_signals.clear();
 
-        thread->send_signal(SIGABRT);
+            thread->send_signal(SIGABRT);
+
+            return true;
+        });
 
         // This should never arrive to userspace, unless we're init and have ignored SIGABRT.
         return err(ENOSYS);

kernel/src/Pledge.h

@@ -14,6 +14,6 @@ enum class Promise
         num_promises,
 };
 
-Result<void> check_pledge(Thread* thread, Promise promise);
+Result<void> check_pledge(Process* process, Promise promise);
 
 Result<int> parse_promises(u64 pledge);

kernel/src/arch/x86_64/CPU.cpp

@@ -91,6 +91,9 @@ void handle_cpu_exception(int signo, const char* err, Registers* regs)
         auto* current = Scheduler::current();
         if (current->check_stack_on_exception(regs->rsp)) return;
 
+        auto space = current->process->address_space.lock();
+        (*space)->debug_log();
+
         current->send_signal(signo);
         current->process_pending_signals(regs);
         return;

kernel/src/arch/x86_64/PCI.cpp

@@ -8,7 +8,7 @@
 
 static inline constexpr u32 make_pci_address(const PCI::Device::Address& address, u32 field)
 {
-    return 0x80000000 | (address.bus << 16) | (address.slot << 11) | (address.function << 8) | ((field)&0xFC);
+    return 0x80000000 | (address.bus << 16) | (address.slot << 11) | (address.function << 8) | ((field) & 0xFC);
 }
 
 namespace PCI

kernel/src/arch/x86_64/Thread.cpp

@@ -15,7 +15,7 @@ void Thread::set_ip(u64 ip)
     regs.rip = ip;
 }
 
-u64 Thread::ip()
+u64 Thread::ip() const
 {
     return regs.rip;
 }
@@ -25,7 +25,7 @@ void Thread::set_sp(u64 sp)
     regs.rsp = sp;
 }
 
-u64 Thread::sp()
+u64 Thread::sp() const
 {
     return regs.rsp;
 }
@@ -71,31 +71,14 @@ void switch_context(Thread* old_thread, Thread* new_thread, Registers* regs)
     memcpy(regs, &new_thread->regs, sizeof(Registers));
 }
 
-// FIXME: Move this function to a common location (also used in ThreadImage)
 Result<u64> Thread::push_mem_on_stack(const u8* mem, usize size)
 {
-    if ((regs.rsp - size) < stack.bottom()) return err(E2BIG);
-
-    if (!MemoryManager::validate_user_write((void*)(regs.rsp - size), size)) return err(EFAULT);
-
-    regs.rsp -= size;
-
-    memcpy((void*)regs.rsp, mem, size);
-
-    return regs.rsp;
+    return MemoryManager::push_mem_on_stack(mem, size, stack, regs.rsp);
 }
 
 Result<u64> Thread::pop_mem_from_stack(u8* mem, usize size)
 {
-    if ((regs.rsp + size) > stack.top()) return err(E2BIG);
-
-    if (!MemoryManager::validate_user_read((void*)regs.rsp, size)) return err(EFAULT);
-
-    memcpy(mem, (void*)regs.rsp, size);
-
-    regs.rsp += size;
-
-    return regs.rsp;
+    return MemoryManager::pop_mem_from_stack(mem, size, stack, regs.rsp);
 }
 
 bool Thread::deliver_signal(int signo, Registers* current_regs)

kernel/src/arch/x86_64/disk/ATA.h

@@ -3,6 +3,7 @@
 #include "fs/devices/BlockDevice.h"
 #include "fs/devices/DeviceRegistry.h"
 #include "lib/Mutex.h"
+#include "thread/Thread.h"
 #include <luna/Atomic.h>
 #include <luna/SharedPtr.h>
 #include <luna/StaticString.h>

kernel/src/binfmt/Script.cpp

@@ -35,11 +35,10 @@ Result<u64> ScriptLoader::load(AddressSpace* space)
     if (!m_interpreter_cmdline.size()) return err(ENOEXEC);
 
     auto& interpreter_path = m_interpreter_cmdline[0];
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
-    auto interpreter = TRY(VFS::resolve_path(interpreter_path.chars(), current->auth, &current->extra_groups,
-                                             current->current_directory, true));
-    if (!VFS::can_execute(interpreter, current->auth, &current->extra_groups)) return err(EACCES);
+    auto interpreter = TRY(VFS::resolve_path(interpreter_path.chars(), current, current->current_directory, true));
+    if (!VFS::can_execute(interpreter, current)) return err(EACCES);
 
     auto loader = TRY(BinaryFormat::create_loader(interpreter, m_recursion_level + 1));
     u64 entry = TRY(loader->load(space));

kernel/src/fs/InitRD.cpp

@@ -20,7 +20,7 @@ void InitRD::initialize()
 
 static Result<void> vfs_create_dir_if_not_exists(const char* path, mode_t mode)
 {
-    auto rc = VFS::create_directory(path, mode & (mode_t)~S_IFMT, Credentials {}, nullptr);
+    auto rc = VFS::create_directory(path, mode & (mode_t)~S_IFMT, nullptr);
     if (rc.has_error())
     {
         if (rc.error() == EEXIST) return {};
@@ -37,8 +37,7 @@ Result<void> InitRD::populate_vfs()
     {
         if (entry.type == TarStream::EntryType::RegularFile)
         {
-            auto file =
-                TRY(VFS::create_file(entry.name.chars(), entry.mode & (mode_t)~S_IFMT, Credentials {}, nullptr));
+            auto file = TRY(VFS::create_file(entry.name.chars(), entry.mode & (mode_t)~S_IFMT, nullptr));
             file->write(entry.data(), 0, entry.size);
         }
         else if (entry.type == TarStream::EntryType::Directory)

kernel/src/fs/Pipe.cpp

@@ -8,7 +8,7 @@ Result<void> Pipe::create(SharedPtr<VFS::Inode>& rpipe, SharedPtr<VFS::Inode>& w
     auto writer = TRY(make_shared<PipeWriter>());
     auto reader = TRY(make_shared<PipeReader>());
 
-    auto auth = Scheduler::current()->auth;
+    auto auth = Process::current()->credentials();
 
     pipe->m_writer = writer.ptr();
     pipe->m_reader = reader.ptr();

kernel/src/fs/VFS.cpp

@@ -17,9 +17,8 @@ namespace VFS
 
     static constexpr int MAX_SYMLINKS = 8;
 
-    Result<SharedPtr<Inode>> resolve_path_impl(const char* path, Credentials auth, const Vector<gid_t>* extra_groups,
-                                               SharedPtr<Inode> current_inode, bool follow_last_symlink,
-                                               int& symlinks_followed)
+    Result<SharedPtr<Inode>> resolve_path_impl(const char* path, Process* process, SharedPtr<Inode> current_inode,
+                                               bool follow_last_symlink, int& symlinks_followed)
     {
         if (symlinks_followed >= MAX_SYMLINKS) return err(ELOOP);
 
@@ -32,7 +31,7 @@ namespace VFS
         const char* section;
         while (parser.next().try_set_value(section))
         {
-            if (!can_execute(current_inode, auth, extra_groups)) return err(EACCES);
+            if (!can_execute(current_inode, process)) return err(EACCES);
             current_inode = TRY(current_inode->find(section));
 
             if (current_inode->type() == VFS::InodeType::Symlink && (follow_last_symlink || parser.has_next()))
@@ -46,8 +45,7 @@ namespace VFS
                     symlink_root = parent_inode;
 
                 symlinks_followed++;
-                current_inode =
-                    TRY(resolve_path_impl(link.chars(), auth, extra_groups, symlink_root, true, symlinks_followed));
+                current_inode = TRY(resolve_path_impl(link.chars(), process, symlink_root, true, symlinks_followed));
                 symlinks_followed--;
             }
 
@@ -57,8 +55,8 @@ namespace VFS
         return current_inode;
     }
 
-    Result<SharedPtr<Inode>> resolve_path(const char* path, Credentials auth, const Vector<gid_t>* extra_groups,
-                                          SharedPtr<VFS::Inode> working_directory, bool follow_last_symlink)
+    Result<SharedPtr<Inode>> resolve_path(const char* path, Process* process, SharedPtr<VFS::Inode> working_directory,
+                                          bool follow_last_symlink)
     {
         SharedPtr<Inode> current_inode;
 
@@ -68,17 +66,17 @@ namespace VFS
 
         int symlinks_followed = 0;
 
-        return resolve_path_impl(path, auth, extra_groups, current_inode, follow_last_symlink, symlinks_followed);
+        return resolve_path_impl(path, process, current_inode, follow_last_symlink, symlinks_followed);
     }
 
-    Result<SharedPtr<Inode>> create_directory(const char* path, mode_t mode, Credentials auth,
-                                              const Vector<gid_t>* extra_groups, SharedPtr<Inode> working_directory)
+    Result<SharedPtr<Inode>> create_directory(const char* path, mode_t mode, Process* process,
+                                              SharedPtr<Inode> working_directory)
     {
         auto parent_path = TRY(PathParser::dirname(path));
 
-        auto parent_inode = TRY(resolve_path(parent_path.chars(), auth, extra_groups, working_directory));
+        auto parent_inode = TRY(resolve_path(parent_path.chars(), process, working_directory));
 
-        if (!can_write(parent_inode, auth, extra_groups)) return err(EACCES);
+        if (!can_write(parent_inode, process)) return err(EACCES);
 
         auto child_name = TRY(PathParser::basename(path));
 
@@ -87,14 +85,14 @@ namespace VFS
         return parent_inode->create_subdirectory(child_name.chars(), mode);
     }
 
-    Result<SharedPtr<Inode>> create_file(const char* path, mode_t mode, Credentials auth,
-                                         const Vector<gid_t>* extra_groups, SharedPtr<Inode> working_directory)
+    Result<SharedPtr<Inode>> create_file(const char* path, mode_t mode, Process* process,
+                                         SharedPtr<Inode> working_directory)
     {
         auto parent_path = TRY(PathParser::dirname(path));
 
-        auto parent_inode = TRY(resolve_path(parent_path.chars(), auth, extra_groups, working_directory));
+        auto parent_inode = TRY(resolve_path(parent_path.chars(), process, working_directory));
 
-        if (!can_write(parent_inode, auth, extra_groups)) return err(EACCES);
+        if (!can_write(parent_inode, process)) return err(EACCES);
 
         auto child_name = TRY(PathParser::basename(path));
 
@@ -135,6 +133,81 @@ namespace VFS
         return {};
     }
 
+    // FIXME: Check all three permissions even if the UID or GID match.
+    bool can_execute(SharedPtr<Inode> inode, Process* process)
+    {
+        const auto& metadata = inode->metadata();
+
+        Credentials auth { 0 };
+        if (process) auth = process->credentials();
+
+        if (auth.euid == 0) return true;
+
+        if (metadata.uid == auth.euid) { return metadata.mode & S_IXUSR; }
+        if (metadata.gid == auth.egid) { return metadata.mode & S_IXGRP; }
+
+        if (process)
+        {
+            auto groups = process->extra_groups.lock();
+            for (gid_t group : *groups)
+            {
+                if (metadata.gid == group) return metadata.mode & S_IXGRP;
+            }
+        }
+
+        return metadata.mode & S_IXOTH;
+    }
+
+    // FIXME: Check all three permissions even if the UID or GID match.
+    bool can_write(SharedPtr<Inode> inode, Process* process)
+    {
+        const auto& metadata = inode->metadata();
+
+        Credentials auth { 0 };
+        if (process) auth = process->credentials();
+
+        if (auth.euid == 0) return true;
+
+        if (metadata.uid == auth.euid) { return metadata.mode & S_IWUSR; }
+        if (metadata.gid == auth.egid) { return metadata.mode & S_IWGRP; }
+
+        if (process)
+        {
+            auto groups = process->extra_groups.lock();
+            for (gid_t group : *groups)
+            {
+                if (metadata.gid == group) return metadata.mode & S_IWGRP;
+            }
+        }
+
+        return metadata.mode & S_IWOTH;
+    }
+
+    // FIXME: Check all three permissions even if the UID or GID match.
+    bool can_read(SharedPtr<Inode> inode, Process* process)
+    {
+        const auto& metadata = inode->metadata();
+
+        Credentials auth { 0 };
+        if (process) auth = process->credentials();
+
+        if (auth.euid == 0) return true;
+
+        if (metadata.uid == auth.euid) { return metadata.mode & S_IRUSR; }
+        if (metadata.gid == auth.egid) { return metadata.mode & S_IRGRP; }
+
+        if (process)
+        {
+            auto groups = process->extra_groups.lock();
+            for (gid_t group : *groups)
+            {
+                if (metadata.gid == group) return metadata.mode & S_IRGRP;
+            }
+        }
+
+        return metadata.mode & S_IROTH;
+    }
+
     // FIXME: Check all three permissions even if the UID or GID match.
     bool can_execute(SharedPtr<Inode> inode, Credentials auth, const Vector<gid_t>* extra_groups)
     {
@@ -232,8 +305,7 @@ namespace VFS
         auto new_root_parent = TRY(PathParser::dirname(new_root));
         auto new_root_path = TRY(PathParser::basename(new_root));
 
-        auto new_root_parent_inode =
-            TRY(VFS::resolve_path(new_root_parent.chars(), Credentials {}, nullptr, working_directory));
+        auto new_root_parent_inode = TRY(VFS::resolve_path(new_root_parent.chars(), nullptr, working_directory));
         auto new_root_inode = TRY(new_root_parent_inode->find(new_root_path.chars()));
 
         if (new_root_inode->type() != VFS::InodeType::Directory) return err(ENOTDIR);
@@ -245,7 +317,7 @@ namespace VFS
 
         kdbgln("vfs: Pivoting root from / to %s, using %s as new root", put_old, new_root);
 
-        auto parent_inode = TRY(resolve_path(parent_path.chars(), Credentials {}, nullptr, working_directory));
+        auto parent_inode = TRY(resolve_path(parent_path.chars(), nullptr, working_directory));
 
         auto inode = TRY(parent_inode->find(child.chars()));
         if (inode->type() != VFS::InodeType::Directory) return err(ENOTDIR);
@@ -265,8 +337,8 @@ namespace VFS
         return {};
     }
 
-    Result<void> mount(const char* path, SharedPtr<VFS::FileSystem> fs, Credentials auth,
-                       const Vector<gid_t>* extra_groups, SharedPtr<VFS::Inode> working_directory)
+    Result<void> mount(const char* path, SharedPtr<VFS::FileSystem> fs, Process* process,
+                       SharedPtr<VFS::Inode> working_directory)
     {
         auto parent_path = TRY(PathParser::dirname(path));
         auto child = TRY(PathParser::basename(path));
@@ -275,7 +347,7 @@ namespace VFS
         kdbgln("vfs: Mounting filesystem on target %s", path);
 #endif
 
-        auto parent_inode = TRY(resolve_path(parent_path.chars(), auth, extra_groups, working_directory));
+        auto parent_inode = TRY(resolve_path(parent_path.chars(), process, working_directory));
 
         auto inode = TRY(parent_inode->find(child.chars()));
         if (inode->type() != VFS::InodeType::Directory) return err(ENOTDIR);
@@ -290,8 +362,7 @@ namespace VFS
         return {};
     }
 
-    Result<void> umount(const char* path, Credentials auth, const Vector<gid_t>* extra_groups,
-                        SharedPtr<VFS::Inode> working_directory)
+    Result<void> umount(const char* path, Process* process, SharedPtr<VFS::Inode> working_directory)
     {
         auto parent_path = TRY(PathParser::dirname(path));
         auto child = TRY(PathParser::basename(path));
@@ -300,7 +371,7 @@ namespace VFS
 
         kinfoln("vfs: Unmounting filesystem on target %s", path);
 
-        auto parent_inode = TRY(resolve_path(parent_path.chars(), auth, extra_groups, working_directory));
+        auto parent_inode = TRY(resolve_path(parent_path.chars(), process, working_directory));
 
         auto inode = TRY(parent_inode->find(child.chars()));
         if (!inode->is_mountpoint()) return err(EINVAL);

kernel/src/fs/VFS.h

@@ -7,6 +7,7 @@
 #include <luna/StringView.h>
 #include <sys/types.h>
 
+struct Process;
 struct Credentials;
 
 namespace VFS
@@ -319,20 +320,21 @@ namespace VFS
         virtual ~DeviceInode() = default;
     };
 
-    Result<SharedPtr<Inode>> resolve_path(const char* path, Credentials auth, const Vector<gid_t>* extra_groups,
+    Result<SharedPtr<Inode>> resolve_path(const char* path, Process* process,
                                           SharedPtr<VFS::Inode> working_directory = {},
                                           bool follow_last_symlink = true);
 
-    Result<SharedPtr<Inode>> create_directory(const char* path, mode_t mode, Credentials auth,
-                                              const Vector<gid_t>* extra_groups,
+    Result<SharedPtr<Inode>> create_directory(const char* path, mode_t mode, Process* process,
                                               SharedPtr<VFS::Inode> working_directory = {});
 
-    Result<SharedPtr<Inode>> create_file(const char* path, mode_t mode, Credentials auth,
-                                         const Vector<gid_t>* extra_groups,
+    Result<SharedPtr<Inode>> create_file(const char* path, mode_t mode, Process* process,
                                          SharedPtr<VFS::Inode> working_directory = {});
 
     Result<void> validate_filename(StringView name);
 
+    bool can_execute(SharedPtr<Inode> inode, Process* process);
+    bool can_read(SharedPtr<Inode> inode, Process* process);
+    bool can_write(SharedPtr<Inode> inode, Process* process);
     bool can_execute(SharedPtr<Inode> inode, Credentials auth, const Vector<gid_t>* extra_groups);
     bool can_read(SharedPtr<Inode> inode, Credentials auth, const Vector<gid_t>* extra_groups);
     bool can_write(SharedPtr<Inode> inode, Credentials auth, const Vector<gid_t>* extra_groups);
@@ -346,9 +348,8 @@ namespace VFS
 
     Result<void> mount_root(SharedPtr<VFS::FileSystem> fs);
     Result<void> pivot_root(const char* new_root, const char* put_old, SharedPtr<VFS::Inode> working_directory);
-    Result<void> mount(const char* path, SharedPtr<VFS::FileSystem> fs, Credentials auth,
-                       const Vector<gid_t>* extra_groups, SharedPtr<Inode> working_directory = {});
+    Result<void> mount(const char* path, SharedPtr<VFS::FileSystem> fs, Process* process,
+                       SharedPtr<Inode> working_directory = {});
 
-    Result<void> umount(const char* path, Credentials auth, const Vector<gid_t>* extra_groups,
-                        SharedPtr<Inode> working_directory = {});
+    Result<void> umount(const char* path, Process* process, SharedPtr<Inode> working_directory = {});
 }

kernel/src/fs/devices/FramebufferDevice.cpp

@@ -3,8 +3,8 @@
 #include "memory/SharedMemory.h"
 #include "video/Framebuffer.h"
 #include <bits/ioctl-defs.h>
-#include <luna/CString.h>
 #include <luna/Alignment.h>
+#include <luna/CString.h>
 
 Result<void> FramebufferDevice::create()
 {

kernel/src/fs/devices/MasterPTY.cpp

@@ -31,8 +31,9 @@ Result<SharedPtr<VFS::Inode>> MasterPTY::create_pair(int index)
 
     slave->m_master = master.ptr();
     slave->m_metadata.devid = luna_dev_makedev(DeviceRegistry::Terminal, index + 2);
-    slave->m_metadata.uid = Scheduler::current()->auth.euid;
-    slave->m_metadata.gid = Scheduler::current()->auth.egid;
+    auto credentials = Process::current()->credentials();
+    slave->m_metadata.uid = credentials.euid;
+    slave->m_metadata.gid = credentials.egid;
     slave->m_metadata.mode = 0620;
     slave->m_metadata.initialize_times();
 
@@ -46,7 +47,7 @@ Result<void> MasterPTY::handle_background_process_group(bool can_succeed, int si
     auto foreground_pgrp = m_foreground_process_group.value();
 
     auto* current = Scheduler::current();
-    if (current->pgid == foreground_pgrp) return {};
+    if (current->process->pgid == foreground_pgrp) return {};
 
     if ((current->signal_mask.get(signo - 1)) || (current->signal_handlers[signo - 1].sa_handler == SIG_IGN))
     {
@@ -112,8 +113,8 @@ Result<void> MasterPTY::handle_input(u8 key)
                 if (!(m_settings.c_lflag & NOFLSH)) m_current_line_buffer.clear();
 
                 if (m_foreground_process_group.has_value())
-                    Scheduler::for_each_in_process_group(*m_foreground_process_group, [](Thread* thread) {
-                        thread->send_signal(SIGINT);
+                    Scheduler::for_each_in_process_group(*m_foreground_process_group, [](Process* p) {
+                        p->send_signal(SIGINT);
                         return true;
                     });
 
@@ -125,8 +126,8 @@ Result<void> MasterPTY::handle_input(u8 key)
                 if (!(m_settings.c_lflag & NOFLSH)) m_current_line_buffer.clear();
 
                 if (m_foreground_process_group.has_value())
-                    Scheduler::for_each_in_process_group(*m_foreground_process_group, [](Thread* thread) {
-                        thread->send_signal(SIGQUIT);
+                    Scheduler::for_each_in_process_group(*m_foreground_process_group, [](Process* p) {
+                        p->send_signal(SIGQUIT);
                         return true;
                     });
 
@@ -202,7 +203,7 @@ Result<usize> MasterPTY::write(const u8* buf, usize, usize length)
 
 Result<u64> MasterPTY::ioctl(int request, void* arg)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_tty));
 
     switch (request)

kernel/src/fs/devices/SlavePTY.cpp

@@ -47,7 +47,7 @@ bool SlavePTY::will_block_if_read() const
 
 Result<u64> SlavePTY::ioctl(int request, void* arg)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_tty));
 
     if (!m_master) return err(EIO);
@@ -69,9 +69,9 @@ Result<u64> SlavePTY::ioctl(int request, void* arg)
 
         bool pgid_exists = false;
         pid_t sid;
-        Scheduler::for_each_in_process_group(pgid, [&pgid_exists, &sid](Thread* thread) {
+        Scheduler::for_each_in_process_group(pgid, [&pgid_exists, &sid](Process* p) {
             pgid_exists = true;
-            sid = thread->sid; // should be the same for all threads in the process group
+            sid = p->sid.load(); // should be the same for all threads in the process group
             return false;
         });
         if (!pgid_exists) return err(EPERM);
@@ -95,13 +95,13 @@ Result<u64> SlavePTY::ioctl(int request, void* arg)
         if (this->m_master->m_session.has_value()) return err(EPERM);
         if (!current->is_session_leader()) return err(EPERM);
 
-        Scheduler::for_each_in_session(current->sid, [this](Thread* thread) {
-            thread->controlling_terminal = this;
+        Scheduler::for_each_in_session(current->sid, [this](Process* p) {
+            p->controlling_terminal = this;
             return true;
         });
 
-        m_master->m_session = current->sid;
-        m_master->m_foreground_process_group = current->pgid;
+        m_master->m_session = current->sid.load();
+        m_master->m_foreground_process_group = current->pgid.load();
 
         return 0;
     }

kernel/src/fs/devices/TTYLink.cpp

@@ -10,7 +10,7 @@ TTYLink::TTYLink()
 
 Result<SharedPtr<VFS::Inode>> TTYLink::open()
 {
-    if (!Scheduler::current()->controlling_terminal) return err(ENXIO);
+    if (!Process::current()->controlling_terminal) return err(ENXIO);
 
-    return Scheduler::current()->controlling_terminal;
+    return Process::current()->controlling_terminal;
 }

kernel/src/lib/Mutex.cpp

@@ -6,7 +6,7 @@
 void Mutex::lock()
 {
     auto* current = Scheduler::current();
-    const pid_t desired = current->id;
+    const pid_t desired = current->tid;
     check(desired > 0); // Why the hell would the idle thread be touching a mutex?
 
     while (true)
@@ -19,7 +19,7 @@ void Mutex::lock()
         {
             if (expected == desired)
             {
-                kerrorln("DEADLOCK! KMutex::lock() recursively called by the same thread (%d)", current->id);
+                kerrorln("DEADLOCK! KMutex::lock() recursively called by the same thread (%d)", current->tid);
                 fail("Mutex deadlock detected");
             }
 
@@ -40,7 +40,7 @@ void Mutex::lock()
 void Mutex::unlock()
 {
     auto* current = Scheduler::current();
-    pid_t expected = current->id;
+    pid_t expected = current->tid;
     check(expected > 0); // Why the hell would the idle thread be touching a mutex?
 
     m_spinlock.lock();
@@ -48,7 +48,7 @@ void Mutex::unlock()
     if (!m_thread.compare_exchange_strong(expected, 0))
     {
         kerrorln("KMutex::unlock() called on a lock already locked by another thread (%d, current is %d)", expected,
-                 current->id);
+                 current->tid);
         fail("Mutex unlock by different thread");
     }
 
@@ -70,7 +70,7 @@ void Mutex::unlock()
 bool Mutex::try_lock()
 {
     auto* current = Scheduler::current();
-    const pid_t desired = current->id;
+    const pid_t desired = current->tid;
     check(desired > 0); // Why the hell would the idle thread be touching a mutex?
 
     // Make sure only one thread is touching the mutex at the same time.
@@ -83,7 +83,7 @@ bool Mutex::try_lock()
     {
         kwarnln("Deadlock avoided! KMutex::try_lock() failed because it was already locked by the same thread "
                 "(%d), this is not supposed to happen",
-                current->id);
+                current->tid);
         CPU::print_stack_trace();
     }
 

kernel/src/lib/Mutex.h

@@ -1,7 +1,10 @@
 #pragma once
-#include "thread/Thread.h"
+#include <luna/Action.h>
 #include <luna/CircularQueue.h>
 #include <luna/Spinlock.h>
+#include <sys/types.h>
+
+class Thread;
 
 class Mutex
 {
@@ -16,6 +19,84 @@ class Mutex
     Atomic<pid_t> m_thread;
 };
 
+template <typename T> class MutexLocked
+{
+    struct MutexLockedGuard
+    {
+        MutexLockedGuard(MutexLocked& value_ref) : m_value_ref(&value_ref)
+        {
+        }
+
+        MutexLockedGuard(const MutexLockedGuard& other) = delete;
+        MutexLockedGuard(MutexLockedGuard&& other)
+        {
+            m_value_ref = other.m_value_ref;
+            other.m_value_ref = nullptr;
+        }
+
+        ~MutexLockedGuard()
+        {
+            if (m_value_ref) m_value_ref->m_lock.unlock();
+        }
+
+        T& ref()
+        {
+            expect(m_value_ref, "MutexLockedGuard::ref() called on a moved MutexLockedGuard");
+            return m_value_ref->m_value;
+        }
+
+        void set(const T& other)
+        {
+            ref() = other;
+        }
+
+        T* operator->()
+        {
+            return &ref();
+        }
+
+        T& operator*()
+        {
+            return ref();
+        }
+
+      private:
+        MutexLocked* m_value_ref;
+    };
+
+  public:
+    MutexLocked() : m_value()
+    {
+    }
+
+    MutexLocked(T value) : m_value(move(value))
+    {
+    }
+
+    MutexLockedGuard lock()
+    {
+        m_lock.lock();
+        return { *this };
+    }
+
+    Option<MutexLockedGuard> try_lock()
+    {
+        if (m_lock.try_lock()) { return { *this }; }
+        return {};
+    }
+
+    void with_lock(Function<T&> callback)
+    {
+        m_lock.lock();
+        callback(m_value);
+        m_lock.unlock();
+    }
+
+  private:
+    T m_value;
+    Mutex m_lock;
+};
+
 class ScopedMutexLock
 {
   public:

kernel/src/main.cpp

@@ -26,9 +26,11 @@ void reap_thread()
     {
         CPU::disable_interrupts();
         auto dying_threads = Scheduler::check_for_dying_threads();
+        auto dead_processes = Scheduler::check_for_dead_processes();
         CPU::enable_interrupts();
 
         dying_threads.consume([](Thread* thread) { Scheduler::reap_thread(thread); });
+        dead_processes.consume([](Process* p) { Scheduler::reap_process(p); });
 
         kernel_wait_for_event();
     }
@@ -65,8 +67,8 @@ void oom_thread()
         mark_critical(BinaryFormat::init(), "Failed to register initial binary formats");
         mark_critical(FSRegistry::init(), "Failed to register initial file systems");
 
-        auto init = mark_critical(VFS::resolve_path("/bin/preinit", Credentials {}, nullptr),
-                                  "Can't find init in the initial ramfs!");
+        auto init =
+            mark_critical(VFS::resolve_path("/bin/preinit", nullptr, nullptr), "Can't find init in the initial ramfs!");
         auto init_thread = mark_critical(Scheduler::create_init_process(init, "/bin/preinit"),
                                          "Failed to create PID 1 process for init");
 

kernel/src/memory/AddressSpace.cpp

@@ -1,4 +1,5 @@
 #include "memory/AddressSpace.h"
+#include "Log.h"
 #include "arch/MMU.h"
 #include "memory/Heap.h"
 #include "memory/MemoryManager.h"
@@ -385,3 +386,13 @@ void VMRegion::sync_shared()
         }
     }
 }
+
+void AddressSpace::debug_log()
+{
+    m_regions.for_each([this](VMRegion* region) {
+        kdbgln("VMRegion start: %p, end: %p, count: %zu, used: %s, persistent: %s, flags: %d, prot: %d, shmid: %lu, "
+               "offset: %ld",
+               (void*)region->start, (void*)region->end, region->count, region->used ? "true" : "false",
+               region->persistent ? "true" : "false", region->flags, region->prot, region->shmid, region->offset);
+    });
+}

kernel/src/memory/AddressSpace.h

@@ -15,7 +15,7 @@ class VMRegion : LinkedListNode<VMRegion>
     bool persistent { false };
     int flags { 0 };
     int prot { 0 };
-    u64 shmid;
+    u64 shmid { 0 };
     off_t offset { 0 };
 
     void cleanup_shared();
@@ -52,6 +52,8 @@ class AddressSpace
 
     static Result<OwnedPtr<AddressSpace>> try_create();
 
+    void debug_log();
+
     Result<OwnedPtr<AddressSpace>> clone();
 
     PageDirectory* page_directory() const

kernel/src/memory/MemoryManager.cpp

@@ -641,4 +641,30 @@ namespace MemoryManager
     {
         return free_mem + used_mem + reserved_mem;
     }
+
+    Result<u64> push_mem_on_stack(const u8* mem, usize size, const Stack& stack, u64& sp)
+    {
+        if ((sp - size) < stack.bottom()) return err(E2BIG);
+
+        if (!MemoryManager::validate_user_write((void*)(sp - size), size)) return err(EFAULT);
+
+        sp -= size;
+
+        memcpy((void*)sp, mem, size);
+
+        return sp;
+    }
+
+    Result<u64> pop_mem_from_stack(u8* mem, usize size, const Stack& stack, u64& sp)
+    {
+        if ((sp + size) > stack.top()) return err(E2BIG);
+
+        if (!MemoryManager::validate_user_read((void*)sp, size)) return err(EFAULT);
+
+        memcpy(mem, (void*)sp, size);
+
+        sp += size;
+
+        return sp;
+    }
 }

kernel/src/memory/MemoryManager.h

@@ -1,6 +1,7 @@
 #pragma once
 #include "arch/MMU.h"
 #include <luna/Result.h>
+#include <luna/Stack.h>
 #include <luna/String.h>
 #include <luna/Types.h>
 
@@ -88,4 +89,7 @@ namespace MemoryManager
     usize used();
     usize reserved();
     usize total();
+
+    Result<u64> push_mem_on_stack(const u8* mem, usize size, const Stack& stack, u64& sp);
+    Result<u64> pop_mem_from_stack(u8* mem, usize size, const Stack& stack, u64& sp);
 }

kernel/src/net/UnixSocket.cpp

@@ -56,14 +56,14 @@ Result<usize> UnixSocket::recv(u8* buf, usize length, int) const
     return m_data.dequeue_data(buf, length);
 }
 
-static Result<void> bind_socket_to_fs(const char* path, Credentials auth, const Vector<gid_t>* extra_groups,
-                                      SharedPtr<VFS::Inode> working_directory, SharedPtr<UnixSocket> socket)
+static Result<void> bind_socket_to_fs(const char* path, Process* process, SharedPtr<VFS::Inode> working_directory,
+                                      SharedPtr<UnixSocket> socket)
 {
     auto parent_path = TRY(PathParser::dirname(path));
 
-    auto parent_inode = TRY(VFS::resolve_path(parent_path.chars(), auth, extra_groups, working_directory));
+    auto parent_inode = TRY(VFS::resolve_path(parent_path.chars(), process, working_directory));
 
-    if (!VFS::can_write(parent_inode, auth, extra_groups)) return err(EACCES);
+    if (!VFS::can_write(parent_inode, process)) return err(EACCES);
 
     auto child_name = TRY(PathParser::basename(path));
 
@@ -89,14 +89,14 @@ Result<void> UnixSocket::bind(struct sockaddr* addr, socklen_t addrlen)
     String path = TRY(String::from_string_view(
         StringView::from_fixed_size_cstring(un_address->sun_path, addrlen - sizeof(sa_family_t))));
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     m_metadata.mode = 0777 & ~current->umask;
-    m_metadata.uid = current->auth.euid;
-    m_metadata.gid = current->auth.egid;
+    auto auth = current->credentials();
+    m_metadata.uid = auth.euid;
+    m_metadata.gid = auth.egid;
 
-    auto rc = bind_socket_to_fs(path.chars(), current->auth, &current->extra_groups, current->current_directory,
-                                SharedPtr<Socket> { this });
+    auto rc = bind_socket_to_fs(path.chars(), current, current->current_directory, SharedPtr<Socket> { this });
     if (rc.has_error())
     {
         if (rc.error() == EEXIST) return err(EADDRINUSE);
@@ -126,13 +126,13 @@ Result<void> UnixSocket::connect(Registers* regs, int flags, struct sockaddr* ad
     String path = TRY(String::from_string_view(
         StringView::from_fixed_size_cstring(un_address->sun_path, addrlen - sizeof(sa_family_t))));
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
+    auto* thread = Scheduler::current();
 
-    auto inode =
-        TRY(VFS::resolve_path(path.chars(), current->auth, &current->extra_groups, current->current_directory));
+    auto inode = TRY(VFS::resolve_path(path.chars(), current, current->current_directory));
     if (inode->type() != VFS::InodeType::Socket)
         return err(ENOTSOCK); // FIXME: POSIX doesn't say what error to return here?
-    if (!VFS::can_write(inode, current->auth, &current->extra_groups)) return err(EACCES);
+    if (!VFS::can_write(inode, current)) return err(EACCES);
 
     auto socket = (SharedPtr<UnixSocket>)inode;
     if (socket->m_state != State::Listening) return err(ECONNREFUSED);
@@ -144,14 +144,14 @@ Result<void> UnixSocket::connect(Registers* regs, int flags, struct sockaddr* ad
 
     while (1)
     {
-        m_blocked_thread = current;
+        m_blocked_thread = thread;
         kernel_wait_for_event();
         m_blocked_thread = nullptr;
-        if (current->interrupted)
+        if (thread->interrupted)
         {
-            if (current->will_ignore_pending_signal())
+            if (thread->will_ignore_pending_signal())
             {
-                current->process_pending_signals(regs);
+                thread->process_pending_signals(regs);
                 continue;
             }
             return err(EINTR);

kernel/src/sys/chdir.cpp

@@ -8,16 +8,16 @@ Result<u64> sys_chdir(Registers*, SyscallArgs args)
 {
     auto path = TRY(MemoryManager::strdup_from_user(args[0]));
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_rpath));
 
     if (PathParser::is_absolute(path.view()))
     {
-        SharedPtr<VFS::Inode> inode = TRY(VFS::resolve_path(path.chars(), current->auth, &current->extra_groups));
+        SharedPtr<VFS::Inode> inode = TRY(VFS::resolve_path(path.chars(), current));
 
         if (inode->type() != VFS::InodeType::Directory) return err(ENOTDIR);
-        if (!VFS::can_execute(inode, current->auth, &current->extra_groups)) return err(EACCES);
+        if (!VFS::can_execute(inode, current)) return err(EACCES);
 
         inode->add_handle();
         if (current->current_directory) current->current_directory->remove_handle();
@@ -29,11 +29,10 @@ Result<u64> sys_chdir(Registers*, SyscallArgs args)
     }
     else
     {
-        SharedPtr<VFS::Inode> inode =
-            TRY(VFS::resolve_path(path.chars(), current->auth, &current->extra_groups, current->current_directory));
+        SharedPtr<VFS::Inode> inode = TRY(VFS::resolve_path(path.chars(), current, current->current_directory));
 
         if (inode->type() != VFS::InodeType::Directory) return err(ENOTDIR);
-        if (!VFS::can_execute(inode, current->auth, &current->extra_groups)) return err(EACCES);
+        if (!VFS::can_execute(inode, current)) return err(EACCES);
 
         auto old_wdir = current->current_directory_path.view();
 
@@ -54,7 +53,7 @@ Result<u64> sys_getcwd(Registers*, SyscallArgs args)
     u8* buf = (u8*)args[0];
     usize size = (usize)args[1];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     StringView cwd = current->current_directory_path.view();
     if (cwd.is_empty()) cwd = "/"_sv;

kernel/src/sys/clock_gettime.cpp

@@ -11,7 +11,7 @@ Result<u64> sys_clock_gettime(Registers*, SyscallArgs args)
     clockid_t id = (clockid_t)args[0];
     struct timespec* ts = (struct timespec*)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 

kernel/src/sys/exec.cpp

@@ -64,14 +64,14 @@ Result<u64> sys_execve(Registers* regs, SyscallArgs args)
     if ((calculate_userspace_stack_size(argv) + calculate_userspace_stack_size(envp)) > MAX_ARGV_STACK_SIZE)
         return err(E2BIG);
 
-    auto current = Scheduler::current();
+    auto current = Process::current();
+    auto thread = Scheduler::current();
 
     TRY(check_pledge(current, Promise::p_exec));
 
-    auto inode =
-        TRY(VFS::resolve_path(path.chars(), current->auth, &current->extra_groups, current->current_directory));
+    auto inode = TRY(VFS::resolve_path(path.chars(), current, current->current_directory));
 
-    if (!VFS::can_execute(inode, current->auth, &current->extra_groups)) return err(EACCES);
+    if (!VFS::can_execute(inode, current)) return err(EACCES);
 
 #ifdef EXEC_DEBUG
     kdbgln("exec: attempting to replace current image with %s", path.chars());
@@ -88,7 +88,7 @@ Result<u64> sys_execve(Registers* regs, SyscallArgs args)
     kdbgln("exec: created loader for binary format %s", loader->format().chars());
 #endif
 
-    auto guard = make_scope_guard([current] { MMU::switch_page_directory(current->self_directory()); });
+    auto guard = make_scope_guard([thread] { MMU::switch_page_directory(thread->self_directory()); });
 
     auto image = TRY(ThreadImage::try_load_from_binary(loader));
 
@@ -108,6 +108,15 @@ Result<u64> sys_execve(Registers* regs, SyscallArgs args)
 
     guard.deactivate();
 
+    // Terminate all other threads.
+    Scheduler::for_each_thread(current, [thread](Thread* t) {
+        if (t != thread) t->quit();
+        return true;
+    });
+    Scheduler::signal_reap_thread();
+
+    current->thread_count = 1;
+
     current->real_timer.disarm();
     current->virtual_timer.disarm();
     current->profiling_timer.disarm();
@@ -120,32 +129,39 @@ Result<u64> sys_execve(Registers* regs, SyscallArgs args)
         }
     }
 
-    for (int i = 0; i < FD_MAX; i++)
     {
-        auto& descriptor = current->fd_table[i];
-        if (!descriptor.has_value()) continue;
-        if (descriptor->flags & O_CLOEXEC) { descriptor = {}; }
+        auto table = current->fd_table.lock();
+        for (int i = 0; i < FD_MAX; i++)
+        {
+            auto& descriptor = (*table)[i];
+            if (!descriptor.has_value()) continue;
+            if (descriptor->flags & O_CLOEXEC) { descriptor = {}; }
+        }
     }
 
-    if (is_setuid) current->auth.euid = current->auth.suid = inode->metadata().uid;
-    if (is_setgid) current->auth.egid = current->auth.sgid = inode->metadata().gid;
+    {
+        auto auth = current->auth.lock();
+        if (is_setuid) (*auth).euid = (*auth).suid = inode->metadata().uid;
+        if (is_setgid) (*auth).egid = (*auth).sgid = inode->metadata().gid;
+    }
 
     current->cmdline = cmdline.chars();
+    thread->cmdline = cmdline.chars();
 
-    image->apply(current);
+    image->apply(thread);
 
-    MMU::switch_page_directory(current->self_directory());
+    MMU::switch_page_directory(thread->self_directory());
 
-    current->set_arguments(user_argc, user_argv, user_envc, user_envp);
+    thread->set_arguments(user_argc, user_argv, user_envc, user_envp);
 
     current->promises = current->execpromises;
     current->execpromises = -1;
 
-    memcpy(regs, &current->regs, sizeof(*regs));
+    memcpy(regs, &thread->regs, sizeof(*regs));
 
     for (int i = 0; i < NSIG; i++)
     {
-        current->signal_handlers[i] = { .sa_handler = SIG_DFL, .sa_mask = 0, .sa_flags = 0 };
+        thread->signal_handlers[i] = { .sa_handler = SIG_DFL, .sa_mask = 0, .sa_flags = 0 };
     }
 
     current->has_called_exec = true;
@@ -157,57 +173,78 @@ Result<u64> sys_execve(Registers* regs, SyscallArgs args)
 
 Result<u64> sys_fork(Registers* regs, SyscallArgs)
 {
-    auto current = Scheduler::current();
+    auto current = Process::current();
+    auto current_thread = Scheduler::current();
 
     TRY(check_pledge(current, Promise::p_proc));
 
-    auto extra_groups = TRY(current->extra_groups.shallow_copy());
+    Vector<gid_t> extra_groups = TRY(current->copy_groups());
+    Credentials auth = current->credentials();
 
-    auto guard = make_scope_guard([current] { MMU::switch_page_directory(current->self_directory()); });
+    auto guard = make_scope_guard([current_thread] { MMU::switch_page_directory(current_thread->self_directory()); });
 
-    memcpy(&current->regs, regs, sizeof(*regs));
+    memcpy(&current_thread->regs, regs, sizeof(*regs));
 
     auto current_directory_path = TRY(current->current_directory_path.clone());
 
-    auto image = TRY(ThreadImage::clone_from_thread(current));
+    auto image = TRY(ThreadImage::clone_from_thread(current_thread));
 
     auto thread = TRY(new_thread());
+    auto process = TRY(make<Process>());
+
+    Option<FileDescriptor> fds[FD_MAX];
+    {
+        auto table = current->fd_table.lock();
+        for (int i = 0; i < FD_MAX; i++) { fds[i] = (*table)[i]; }
+    }
 
     thread->state = ThreadState::Runnable;
-    thread->is_kernel = false;
     thread->fp_data.save();
-    thread->cmdline = current->cmdline;
-    thread->auth = current->auth;
-    thread->current_directory = current->current_directory;
-    thread->current_directory_path = move(current_directory_path);
-    thread->umask = current->umask;
-    thread->parent = current;
-    thread->promises = current->promises;
-    thread->execpromises = current->execpromises;
-    thread->controlling_terminal = current->controlling_terminal;
-    thread->pgid = current->pgid;
-    thread->sid = current->sid;
-    thread->extra_groups = move(extra_groups);
+    thread->cmdline = current_thread->cmdline;
+    thread->process = process;
 
-    thread->virtual_clock.set_resolution(1'000'000);
-    thread->profiling_clock.set_resolution(1'000'000);
+    process->thread_count = 1;
+    process->id = thread->tid;
+    process->current_directory = current->current_directory;
+    process->current_directory_path = move(current_directory_path);
+    process->umask = current->umask;
+    process->parent = current;
+    process->promises = current->promises;
+    process->execpromises = current->execpromises;
+    process->controlling_terminal = current->controlling_terminal;
+    process->pgid = current->pgid;
+    process->sid = current->sid;
+    process->extra_groups = move(extra_groups);
+    process->cmdline = current->cmdline;
 
-    for (int i = 0; i < FD_MAX; i++) { thread->fd_table[i] = current->fd_table[i]; }
+    process->virtual_clock.set_resolution(1'000'000);
+    process->profiling_clock.set_resolution(1'000'000);
+
+    {
+        auto credentials = process->auth.lock();
+        *credentials = auth;
+    }
+
+    {
+        auto table = process->fd_table.lock();
+        for (int i = 0; i < FD_MAX; i++) { (*table)[i] = fds[i]; }
+    }
 
     image->apply(thread);
 
     memcpy(&thread->regs, regs, sizeof(*regs));
 
-    for (int i = 0; i < NSIG; i++) thread->signal_handlers[i] = current->signal_handlers[i];
-    thread->signal_mask = current->signal_mask;
+    for (int i = 0; i < NSIG; i++) thread->signal_handlers[i] = current_thread->signal_handlers[i];
+    thread->signal_mask = current_thread->signal_mask;
 
     thread->set_return(0);
 
     Scheduler::add_thread(thread);
+    Scheduler::add_process(process);
 
 #ifdef FORK_DEBUG
-    kdbgln("fork: thread %d forked into child %d", current->id, thread->id);
+    kdbgln("fork: thread %d forked into child %d", current->id, process->id);
 #endif
 
-    return thread->id;
+    return process->id;
 }

kernel/src/sys/exit.cpp

@@ -5,7 +5,7 @@ Result<u64> sys_exit(Registers*, SyscallArgs args)
 {
     u8 status = (u8)args[0];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
-    current->exit_and_signal_parent(status);
+    current->exit(status);
 }

kernel/src/sys/file.cpp

@@ -25,9 +25,9 @@ Result<u64> sys_read(Registers* regs, SyscallArgs args)
 
     Thread* current = Scheduler::current();
 
-    TRY(check_pledge(current, Promise::p_stdio));
+    TRY(check_pledge(current->process, Promise::p_stdio));
 
-    auto& descriptor = *TRY(current->resolve_fd(fd));
+    auto& descriptor = *TRY(current->process->resolve_fd(fd));
 
     if (!descriptor.is_readable()) return err(EBADF);
 
@@ -66,7 +66,7 @@ Result<u64> sys_write(Registers*, SyscallArgs args)
 
     if (!MemoryManager::validate_user_read(buf, size)) return err(EFAULT);
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -90,7 +90,7 @@ Result<u64> sys_lseek(Registers*, SyscallArgs args)
     off_t offset = (long)args[1];
     int whence = (int)args[2];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -122,7 +122,7 @@ Result<u64> sys_fcntl(Registers*, SyscallArgs args)
     int fd = (int)args[0];
     int cmd = (int)args[1];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -135,13 +135,13 @@ Result<u64> sys_fcntl(Registers*, SyscallArgs args)
     case F_DUPFD: is_cloexec = false; [[fallthrough]];
     case F_DUPFD_CLOEXEC: {
         int arg = (int)args[2];
-        int new_fd = TRY(current->allocate_fd(arg));
+        int new_fd = TRY(current->allocate_fd(arg, descriptor));
 
-        current->fd_table[new_fd] = descriptor;
+        auto table = current->fd_table.lock();
 
-        if (is_cloexec) current->fd_table[new_fd]->flags |= O_CLOEXEC;
+        if (is_cloexec) (*table)[new_fd]->flags |= O_CLOEXEC;
         else
-            current->fd_table[new_fd]->flags &= ~O_CLOEXEC;
+            (*table)[new_fd]->flags &= ~O_CLOEXEC;
 
         return (u64)new_fd;
     }
@@ -174,7 +174,7 @@ Result<u64> sys_ioctl(Registers*, SyscallArgs args)
     int request = (int)args[1];
     void* arg = (void*)args[2];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     auto& descriptor = *TRY(current->resolve_fd(fd));
 
     return descriptor.inode()->ioctl(request, arg);
@@ -184,7 +184,7 @@ Result<u64> sys_isatty(Registers*, SyscallArgs args)
 {
     int fd = (int)args[0];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
     auto& descriptor = *TRY(current->resolve_fd(fd));
 
@@ -196,7 +196,7 @@ Result<u64> sys_dup2(Registers*, SyscallArgs args)
     int oldfd = (int)args[0];
     int newfd = (int)args[1];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -206,8 +206,10 @@ Result<u64> sys_dup2(Registers*, SyscallArgs args)
 
     if (newfd == oldfd) return (u64)newfd;
 
-    current->fd_table[newfd] = descriptor;
-    current->fd_table[newfd]->flags &= ~O_CLOEXEC;
+    auto table = current->fd_table.lock();
+
+    (*table)[newfd] = descriptor;
+    (*table)[newfd]->flags &= ~O_CLOEXEC;
 
     return (u64)newfd;
 }
@@ -216,23 +218,23 @@ Result<u64> sys_pipe(Registers*, SyscallArgs args)
 {
     int* pfds = (int*)args[0];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
-    int rfd = TRY(current->allocate_fd(0));
-    int wfd = TRY(current->allocate_fd(rfd + 1));
-
-    if (!MemoryManager::copy_to_user_typed(pfds, &rfd)) return err(EFAULT);
-    if (!MemoryManager::copy_to_user_typed(pfds + 1, &wfd)) return err(EFAULT);
-
     SharedPtr<VFS::Inode> rpipe;
     SharedPtr<VFS::Inode> wpipe;
 
     TRY(Pipe::create(rpipe, wpipe));
 
-    current->fd_table[rfd] = FileDescriptor { TRY(make_shared<OpenFileDescription>(rpipe, O_RDONLY)), 0 };
-    current->fd_table[wfd] = FileDescriptor { TRY(make_shared<OpenFileDescription>(wpipe, O_WRONLY)), 0 };
+    auto rd = FileDescriptor { TRY(make_shared<OpenFileDescription>(rpipe, O_RDONLY)), 0 };
+    auto wd = FileDescriptor { TRY(make_shared<OpenFileDescription>(wpipe, O_WRONLY)), 0 };
+
+    int rfd = TRY(current->allocate_fd(0, rd));
+    int wfd = TRY(current->allocate_fd(rfd + 1, wd));
+
+    if (!MemoryManager::copy_to_user_typed(pfds, &rfd)) return err(EFAULT);
+    if (!MemoryManager::copy_to_user_typed(pfds + 1, &wfd)) return err(EFAULT);
 
     return 0;
 }
@@ -241,7 +243,7 @@ Result<u64> sys_umask(Registers*, SyscallArgs args)
 {
     mode_t new_umask = (mode_t)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -257,12 +259,12 @@ Result<u64> sys_truncate(Registers*, SyscallArgs args)
     auto path = TRY(MemoryManager::strdup_from_user(args[0]));
     size_t length = (size_t)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_wpath));
-    auto inode =
-        TRY(VFS::resolve_path(path.chars(), current->auth, &current->extra_groups, current->current_directory));
 
-    if (!VFS::can_write(inode, current->auth, &current->extra_groups)) return err(EACCES);
+    auto inode = TRY(VFS::resolve_path(path.chars(), current, current->current_directory));
+
+    if (!VFS::can_write(inode, current)) return err(EACCES);
 
     TRY(inode->truncate(length));
 
@@ -274,7 +276,7 @@ Result<u64> sys_ftruncate(Registers*, SyscallArgs args)
     int fd = (int)args[0];
     size_t length = (size_t)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
     auto description = TRY(current->resolve_fd(fd))->description;
     if (!(description->flags & O_WRONLY)) return err(EBADF);
@@ -291,9 +293,12 @@ Result<u64> sys_utimensat(Registers*, SyscallArgs args)
     const auto* times = (const struct timespec*)args[2];
     int flags = (int)args[3];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_fattr));
-    auto inode = TRY(current->resolve_atfile(dirfd, path, flags & AT_EMPTY_PATH, !(flags & AT_SYMLINK_NOFOLLOW)));
+
+    auto* process = Process::current();
+    auto credentials = process->credentials();
+    auto inode = TRY(process->resolve_atfile(dirfd, path, flags & AT_EMPTY_PATH, !(flags & AT_SYMLINK_NOFOLLOW)));
 
     struct timespec ktimes[2];
     ktimes[0].tv_sec = ktimes[1].tv_sec = 0;
@@ -309,11 +314,10 @@ Result<u64> sys_utimensat(Registers*, SyscallArgs args)
 
     if (allow_write_access)
     {
-        if (!VFS::can_write(inode, current->auth, &current->extra_groups) &&
-            current->auth.euid != inode->metadata().uid && current->auth.euid != 0)
+        if (!VFS::can_write(inode, current) && credentials.euid != inode->metadata().uid && credentials.euid != 0)
             return err(EACCES);
     }
-    else if (current->auth.euid != inode->metadata().uid && current->auth.euid != 0)
+    else if (credentials.euid != inode->metadata().uid && credentials.euid != 0)
         return err(EPERM);
 
     auto metadata = inode->metadata();

kernel/src/sys/getdents.cpp

@@ -11,7 +11,7 @@ Result<u64> sys_getdents(Registers*, SyscallArgs args)
     luna_dirent* ent = (luna_dirent*)args[1];
     usize count = (usize)args[2];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
     auto& descriptor = *TRY(current->resolve_fd(fd));

kernel/src/sys/id.cpp

@@ -6,14 +6,14 @@
 
 Result<u64> sys_getpid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
     return current->id;
 }
 
 Result<u64> sys_getppid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
     auto* parent = current->parent;
     return parent ? parent->id : 0;
@@ -21,48 +21,48 @@ Result<u64> sys_getppid(Registers*, SyscallArgs)
 
 Result<u64> sys_getuid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
-    return current->auth.uid;
+    return current->credentials().uid;
 }
 
 Result<u64> sys_geteuid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
-    return current->auth.euid;
+    return current->credentials().euid;
 }
 
 Result<u64> sys_getgid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
-    return current->auth.gid;
+    return current->credentials().gid;
 }
 
 Result<u64> sys_getegid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
-    return current->auth.egid;
+    return current->credentials().egid;
 }
 
 Result<u64> sys_setuid(Registers*, SyscallArgs args)
 {
     u32 uid = (u32)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_id));
-    Credentials& auth = current->auth;
+    auto auth = current->auth.lock();
 
-    if (auth.euid == 0)
+    if (auth->euid == 0)
     {
-        auth.uid = auth.euid = auth.suid = uid;
+        auth->uid = auth->euid = auth->suid = uid;
         return 0;
     }
 
-    if (uid != auth.uid && uid != auth.suid) return err(EPERM);
-    auth.euid = uid;
+    if (uid != auth->uid && uid != auth->suid) return err(EPERM);
+    auth->euid = uid;
 
     return 0;
 }
@@ -71,12 +71,12 @@ Result<u64> sys_seteuid(Registers*, SyscallArgs args)
 {
     u32 uid = (u32)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_id));
-    Credentials& auth = current->auth;
+    auto auth = current->auth.lock();
 
-    if (auth.euid != 0 && uid != auth.uid && uid != auth.suid) return err(EPERM);
-    auth.euid = uid;
+    if (auth->euid != 0 && uid != auth->uid && uid != auth->suid) return err(EPERM);
+    auth->euid = uid;
 
     return 0;
 }
@@ -85,18 +85,18 @@ Result<u64> sys_setgid(Registers*, SyscallArgs args)
 {
     u32 gid = (u32)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_id));
-    Credentials& auth = current->auth;
+    auto auth = current->auth.lock();
 
-    if (auth.euid == 0)
+    if (auth->euid == 0)
     {
-        auth.gid = auth.egid = auth.sgid = gid;
+        auth->gid = auth->egid = auth->sgid = gid;
         return 0;
     }
 
-    if (gid != auth.gid && gid != auth.sgid) return err(EPERM);
-    auth.egid = gid;
+    if (gid != auth->gid && gid != auth->sgid) return err(EPERM);
+    auth->egid = gid;
 
     return 0;
 }
@@ -105,12 +105,12 @@ Result<u64> sys_setegid(Registers*, SyscallArgs args)
 {
     u32 gid = (u32)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_id));
-    Credentials& auth = current->auth;
+    auto auth = current->auth.lock();
 
-    if (auth.euid != 0 && gid != auth.gid && gid != auth.sgid) return err(EPERM);
-    auth.egid = gid;
+    if (auth->euid != 0 && gid != auth->gid && gid != auth->sgid) return err(EPERM);
+    auth->egid = gid;
 
     return 0;
 }
@@ -120,34 +120,34 @@ Result<u64> sys_setpgid(Registers*, SyscallArgs args)
     pid_t pid = (pid_t)args[0];
     pid_t pgid = (pid_t)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_proc));
     if (pid == 0) pid = current->id;
     if (pgid == 0) pgid = current->id;
 
     if (pgid < 0) return err(EINVAL);
 
-    auto* thread = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
-    if (thread != current && thread->parent != current) return err(ESRCH);
+    auto* target = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
+    if (target != current && target->parent != current) return err(ESRCH);
 
-    if (thread->is_session_leader() || thread->sid != current->sid) return err(EPERM);
+    if (target->is_session_leader() || target->sid != current->sid) return err(EPERM);
 
-    if (thread->has_called_exec) return err(EPERM);
+    if (target->has_called_exec) return err(EACCES);
 
     if (pgid != current->id)
     {
         bool pgid_exists = false;
         pid_t sid;
-        Scheduler::for_each_in_process_group(pgid, [&pgid_exists, &sid](Thread* t) {
+        Scheduler::for_each_in_process_group(pgid, [&pgid_exists, &sid](Process* p) {
             pgid_exists = true;
-            sid = t->sid; // this should be the same for all threads in the process group
+            sid = p->sid; // this should be the same for all threads in the process group
             return false;
         });
         if (!pgid_exists) return err(EPERM);
-        if (sid != thread->sid) return err(EPERM);
+        if (sid != target->sid) return err(EPERM);
     }
 
-    thread->pgid = (u64)pgid;
+    target->pgid = (u64)pgid;
 
     return 0;
 }
@@ -156,20 +156,20 @@ Result<u64> sys_getpgid(Registers*, SyscallArgs args)
 {
     pid_t pid = (pid_t)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
     if (pid == 0) pid = current->id;
 
     if (pid < 0) return err(EINVAL);
 
-    auto* thread = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
+    auto* process = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
 
-    return (u64)thread->pgid;
+    return (u64)process->pgid.load();
 }
 
 Result<u64> sys_setsid(Registers*, SyscallArgs)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_proc));
 
     if (current->pgid == current->id) return err(EPERM);
@@ -177,21 +177,21 @@ Result<u64> sys_setsid(Registers*, SyscallArgs)
     current->sid = current->pgid = current->id;
     current->controlling_terminal = {};
 
-    return current->sid;
+    return current->sid.load();
 }
 
 Result<u64> sys_getsid(Registers*, SyscallArgs args)
 {
     pid_t pid = (pid_t)args[0];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
     if (pid == 0) pid = current->id;
 
-    auto* thread = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
+    auto* p = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
 
-    return thread->sid;
+    return p->sid.load();
 }
 
 Result<u64> sys_fchmodat(Registers*, SyscallArgs args)
@@ -201,12 +201,13 @@ Result<u64> sys_fchmodat(Registers*, SyscallArgs args)
     mode_t mode = (mode_t)args[2];
     int flags = (int)args[3];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_wpath));
+    auto credentials = current->credentials();
 
     auto inode = TRY(current->resolve_atfile(dirfd, path, flags & AT_EMPTY_PATH, !(flags & AT_SYMLINK_NOFOLLOW)));
 
-    if (current->auth.euid != 0 && current->auth.euid != inode->metadata().uid) return err(EPERM);
+    if (credentials.euid != 0 && credentials.euid != inode->metadata().uid) return err(EPERM);
 
     auto metadata = inode->metadata();
     metadata.mode = mode;
@@ -223,12 +224,13 @@ Result<u64> sys_fchownat(Registers*, SyscallArgs args)
     gid_t gid = (u32)args[3];
     int flags = (int)args[4];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_chown));
+    auto credentials = current->credentials();
 
     auto inode = TRY(current->resolve_atfile(dirfd, path, flags & AT_EMPTY_PATH, !(flags & AT_SYMLINK_NOFOLLOW)));
 
-    if (current->auth.euid != 0) return err(EPERM);
+    if (credentials.euid != 0) return err(EPERM);
 
     auto metadata = inode->metadata();
     if (uid != (uid_t)-1) metadata.uid = uid;
@@ -243,20 +245,20 @@ Result<u64> sys_getgroups(Registers*, SyscallArgs args)
     int ngroups = (int)args[0];
     gid_t* grouplist = (gid_t*)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
-    if (!ngroups) return current->extra_groups.size();
+    auto groups = current->extra_groups.lock();
+
+    if (!ngroups) return groups->size();
     if (ngroups < 0) return err(EINVAL);
 
-    if (static_cast<usize>(ngroups) < current->extra_groups.size()) return err(EINVAL);
+    if (static_cast<usize>(ngroups) < groups->size()) return err(EINVAL);
 
-    if (!MemoryManager::copy_to_user(grouplist, current->extra_groups.data(),
-                                     current->extra_groups.size() * sizeof(gid_t)))
-        return err(EFAULT);
+    if (!MemoryManager::copy_to_user(grouplist, groups->data(), groups->size() * sizeof(gid_t))) return err(EFAULT);
 
-    return current->extra_groups.size();
+    return groups->size();
 }
 
 Result<u64> sys_setgroups(Registers*, SyscallArgs args)
@@ -264,26 +266,27 @@ Result<u64> sys_setgroups(Registers*, SyscallArgs args)
     int ngroups = (int)args[0];
     const gid_t* grouplist = (const gid_t*)args[1];
 
-    auto* current = Scheduler::current();
-
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_id));
 
-    Credentials& auth = current->auth;
-    if (auth.euid != 0) return err(EPERM);
+    auto credentials = current->credentials();
+    if (credentials.euid != 0) return err(EPERM);
+
+    auto groups = current->extra_groups.lock();
 
     if (!ngroups)
     {
-        current->extra_groups.clear();
+        groups->clear();
         return 0;
     }
 
     if (ngroups < 0 || ngroups > 32) return err(EINVAL);
 
-    TRY(current->extra_groups.try_reserve(ngroups));
+    TRY(groups->try_reserve(ngroups));
 
-    current->extra_groups.mutate([&](gid_t* list, usize) -> usize {
+    groups->mutate([&](gid_t* list, usize) -> usize {
         if (MemoryManager::copy_from_user(grouplist, list, ngroups * sizeof(gid_t))) return ngroups;
-        return current->extra_groups.size();
+        return groups->size();
     });
 
     return 0;

kernel/src/sys/link.cpp

@@ -12,7 +12,7 @@ Result<u64> sys_unlinkat(Registers*, SyscallArgs args)
     auto path = TRY(MemoryManager::strdup_from_user(args[1]));
     int flags = (int)args[2];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_cpath));
 
     auto dirname = TRY(PathParser::dirname(path.view()));
@@ -23,13 +23,14 @@ Result<u64> sys_unlinkat(Registers*, SyscallArgs args)
     kinfoln("unlinkat: remove %s from directory %s, dirfd is %d", basename.chars(), dirname.chars(), dirfd);
 
     auto inode = TRY(current->resolve_atfile(dirfd, dirname, false, false));
-    if (!VFS::can_write(inode, current->auth, &current->extra_groups)) return err(EACCES);
+    auto auth = current->credentials();
+    if (!VFS::can_write(inode, current)) return err(EACCES);
 
     auto child = TRY(inode->find(basename.chars()));
     if (flags == AT_REMOVEDIR && child->type() != VFS::InodeType::Directory) return err(ENOTDIR);
 
-    if (current->auth.euid != 0 && VFS::is_sticky(inode) && current->auth.euid != inode->metadata().uid &&
-        current->auth.euid != child->metadata().uid)
+    if (auth.euid != 0 && VFS::is_sticky(inode) && auth.euid != inode->metadata().uid &&
+        auth.euid != child->metadata().uid)
         return err(EACCES);
 
     TRY(inode->remove_entry(basename.chars()));
@@ -45,14 +46,14 @@ Result<u64> sys_symlinkat(Registers*, SyscallArgs args)
 
     if (target.is_empty()) return err(ENOENT);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_cpath));
 
     auto parent = TRY(PathParser::dirname(linkpath.view()));
 
     auto parent_inode = TRY(current->resolve_atfile(dirfd, parent, false, true));
 
-    if (!VFS::can_write(parent_inode, current->auth, &current->extra_groups)) return err(EACCES);
+    if (!VFS::can_write(parent_inode, current)) return err(EACCES);
 
     auto child_name = TRY(PathParser::basename(linkpath.view()));
 
@@ -60,8 +61,9 @@ Result<u64> sys_symlinkat(Registers*, SyscallArgs args)
 
     auto inode = TRY(parent_inode->fs()->create_symlink_inode(target.view()));
     auto metadata = inode->metadata();
-    metadata.uid = current->auth.euid;
-    metadata.gid = current->auth.egid;
+    auto auth = current->credentials();
+    metadata.uid = auth.euid;
+    metadata.gid = auth.egid;
     TRY(inode->set_metadata(metadata));
     TRY(parent_inode->add_entry(inode, child_name.chars()));
 
@@ -75,7 +77,7 @@ Result<u64> sys_readlinkat(Registers*, SyscallArgs args)
     char* buf = (char*)args[2];
     usize bufsiz = (usize)args[3];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_rpath));
 
     auto symlink = TRY(current->resolve_atfile(dirfd, path, true, false));
@@ -101,7 +103,7 @@ Result<u64> sys_linkat(Registers*, SyscallArgs args)
     auto newpath = TRY(MemoryManager::strdup_from_user(args[3]));
     int flags = (int)args[4];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_cpath));
 
     auto parent = TRY(PathParser::dirname(newpath.view()));
@@ -115,7 +117,7 @@ Result<u64> sys_linkat(Registers*, SyscallArgs args)
 
     if (target->fs() != parent_inode->fs()) return err(EXDEV);
 
-    if (!VFS::can_write(parent_inode, current->auth, &current->extra_groups)) return err(EACCES);
+    if (!VFS::can_write(parent_inode, current)) return err(EACCES);
 
     auto child_name = TRY(PathParser::basename(newpath.view()));
 

kernel/src/sys/memstat.cpp

@@ -6,7 +6,7 @@
 
 Result<u64> sys_memstat(Registers*, SyscallArgs args)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
     struct membuf buf;

kernel/src/sys/mkdir.cpp

@@ -10,14 +10,14 @@ Result<u64> sys_mkdir(Registers*, SyscallArgs args)
     auto path = TRY(MemoryManager::strdup_from_user(args[0]));
     mode_t mode = (mode_t)args[1];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
+    auto credentials = current->credentials();
     TRY(check_pledge(current, Promise::p_cpath));
 
-    auto inode = TRY(VFS::create_directory(path.chars(), mode & ~current->umask, current->auth, &current->extra_groups,
-                                           current->current_directory));
+    auto inode = TRY(VFS::create_directory(path.chars(), mode & ~current->umask, current, current->current_directory));
     auto metadata = inode->metadata();
-    metadata.uid = current->auth.euid;
-    metadata.gid = current->auth.egid;
+    metadata.uid = credentials.euid;
+    metadata.gid = credentials.egid;
     TRY(inode->set_metadata(metadata));
 
     return 0;

kernel/src/sys/mmap.cpp

@@ -20,7 +20,7 @@ Result<u64> sys_mmap(Registers*, SyscallArgs args)
 
     if (params.flags < 0) return err(EINVAL);
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     if (params.prot & PROT_EXEC) TRY(check_pledge(current, Promise::p_prot_exec));
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -55,15 +55,15 @@ Result<u64> sys_mmap(Registers*, SyscallArgs args)
         shmem->prot |= params.prot;
     }
 
+    auto space = current->address_space.lock();
+
     u64 address;
-    if (!params.addr)
-        address = TRY(current->address_space->alloc_region(pages, params.prot, params.flags, params.offset, shmid));
+    if (!params.addr) address = TRY((*space)->alloc_region(pages, params.prot, params.flags, params.offset, shmid));
     else
     {
         // FIXME: We should be more flexible if MAP_FIXED was not specified.
         address = align_down<ARCH_PAGE_SIZE>((u64)params.addr);
-        if (!TRY(current->address_space->test_and_alloc_region(address, pages, params.prot, params.flags, params.offset,
-                                                               shmid)))
+        if (!TRY((*space)->test_and_alloc_region(address, pages, params.prot, params.flags, params.offset, shmid)))
             return err(ENOMEM);
     }
 
@@ -94,10 +94,12 @@ Result<u64> sys_munmap(Registers*, SyscallArgs args)
     if (size == 0) return err(EINVAL);
     if (!is_aligned<ARCH_PAGE_SIZE>(address)) return err(EINVAL);
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
-    bool ok = TRY(current->address_space->free_region(address, ceil_div(size, ARCH_PAGE_SIZE)));
+    auto space = current->address_space.lock();
+
+    bool ok = TRY((*space)->free_region(address, ceil_div(size, ARCH_PAGE_SIZE)));
 
     // POSIX says munmap should silently do nothing if the memory was not already mapped.
     if (!ok) return 0;
@@ -119,10 +121,12 @@ Result<u64> sys_msync(Registers*, SyscallArgs args)
     if (!size) return 0;
     if (!is_aligned<ARCH_PAGE_SIZE>(address)) return err(EINVAL);
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
-    TRY(current->address_space->sync_regions(address, ceil_div(size, ARCH_PAGE_SIZE)));
+    auto space = current->address_space.lock();
+
+    TRY((*space)->sync_regions(address, ceil_div(size, ARCH_PAGE_SIZE)));
 
     return { 0 };
 }

kernel/src/sys/mount.cpp

@@ -14,13 +14,12 @@ Result<u64> sys_mount(Registers*, SyscallArgs args)
     auto fstype = TRY(MemoryManager::strdup_from_user(args[1]));
     auto source = TRY(MemoryManager::strdup_from_user(args[2]));
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_mount));
-    if (current->auth.euid != 0) return err(EPERM);
+    if (current->credentials().euid != 0) return err(EPERM);
 
     auto get_source = [current, &source]() -> Result<SharedPtr<Device>> {
-        auto inode =
-            TRY(VFS::resolve_path(source.chars(), current->auth, &current->extra_groups, current->current_directory));
+        auto inode = TRY(VFS::resolve_path(source.chars(), current, current->current_directory));
         if (inode->type() != VFS::InodeType::BlockDevice) return err(ENOTBLK);
         dev_t device_id = inode->metadata().devid;
         return TRY(DeviceRegistry::fetch_special_device(luna_dev_major(device_id), luna_dev_minor(device_id)));
@@ -41,7 +40,7 @@ Result<u64> sys_mount(Registers*, SyscallArgs args)
         fs = TRY(factory(device));
     }
 
-    TRY(VFS::mount(target.chars(), fs, current->auth, &current->extra_groups, current->current_directory));
+    TRY(VFS::mount(target.chars(), fs, current, current->current_directory));
 
     return 0;
 }
@@ -50,11 +49,11 @@ Result<u64> sys_umount(Registers*, SyscallArgs args)
 {
     auto target = TRY(MemoryManager::strdup_from_user(args[0]));
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_mount));
-    if (current->auth.euid != 0) return err(EPERM);
+    if (current->credentials().euid != 0) return err(EPERM);
 
-    TRY(VFS::umount(target.chars(), current->auth, &current->extra_groups, current->current_directory));
+    TRY(VFS::umount(target.chars(), current, current->current_directory));
 
     return 0;
 }
@@ -64,9 +63,9 @@ Result<u64> sys_pivot_root(Registers*, SyscallArgs args)
     auto new_root = TRY(MemoryManager::strdup_from_user(args[0]));
     auto put_old = TRY(MemoryManager::strdup_from_user(args[1]));
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_mount));
-    if (current->auth.euid != 0) return err(EPERM);
+    if (current->credentials().euid != 0) return err(EPERM);
 
     TRY(VFS::pivot_root(new_root.chars(), put_old.chars(), current->current_directory));
 

kernel/src/sys/open.cpp

@@ -17,7 +17,7 @@ Result<u64> sys_openat(Registers*, SyscallArgs args)
     int flags = (int)args[2];
     mode_t mode = (mode_t)args[3];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
 
     SharedPtr<VFS::Inode> inode;
 
@@ -44,12 +44,12 @@ Result<u64> sys_openat(Registers*, SyscallArgs args)
     {
         if (error == ENOENT && (flags & O_CREAT) && !path.is_empty())
         {
-            inode = TRY(VFS::create_file(path.chars(), mode & ~current->umask, current->auth, &current->extra_groups,
-                                         parent_inode));
+            auto auth = current->credentials();
+            inode = TRY(VFS::create_file(path.chars(), mode & ~current->umask, current, parent_inode));
             // FIXME: Pass these in create_file().
             auto metadata = inode->metadata();
-            metadata.uid = current->auth.euid;
-            metadata.gid = current->auth.egid;
+            metadata.uid = auth.euid;
+            metadata.gid = auth.egid;
             TRY(inode->set_metadata(metadata));
         }
         else
@@ -59,8 +59,8 @@ Result<u64> sys_openat(Registers*, SyscallArgs args)
         return err(EEXIST);
     else
     {
-        if ((flags & O_RDONLY) && !VFS::can_read(inode, current->auth, &current->extra_groups)) return err(EACCES);
-        if ((flags & O_WRONLY) && !VFS::can_write(inode, current->auth, &current->extra_groups)) return err(EACCES);
+        if ((flags & O_RDONLY) && !VFS::can_read(inode, current)) return err(EACCES);
+        if ((flags & O_WRONLY) && !VFS::can_write(inode, current)) return err(EACCES);
     }
 
     inode = TRY(inode->open());
@@ -72,11 +72,12 @@ Result<u64> sys_openat(Registers*, SyscallArgs args)
 
     if (flags & O_TMPFILE)
     {
+        auto auth = current->credentials();
         if (inode->type() != VFS::InodeType::Directory) return err(EINVAL);
         inode = TRY(inode->fs()->create_file_inode(mode & current->umask));
         auto metadata = inode->metadata();
-        metadata.uid = current->auth.euid;
-        metadata.gid = current->auth.egid;
+        metadata.uid = auth.euid;
+        metadata.gid = auth.egid;
         TRY(inode->set_metadata(metadata));
     }
 
@@ -91,15 +92,15 @@ Result<u64> sys_openat(Registers*, SyscallArgs args)
 
     if ((flags & O_WRONLY) && (flags & O_TRUNC)) inode->truncate(0);
 
-    int fd = TRY(current->allocate_fd(0));
+    auto descriptor =
+        FileDescriptor { TRY(make_shared<OpenFileDescription>(inode, flags & FLAGS_TO_KEEP)), 0, flags & O_CLOEXEC };
+
+    int fd = TRY(current->allocate_fd(0, descriptor));
 
 #ifdef OPEN_DEBUG
     kdbgln("openat: opening file %s from dirfd %d, flags %d, mode %#o = fd %d", path.chars(), dirfd, flags, mode, fd);
 #endif
 
-    current->fd_table[fd] =
-        FileDescriptor { TRY(make_shared<OpenFileDescription>(inode, flags & FLAGS_TO_KEEP)), 0, flags & O_CLOEXEC };
-
     return (u64)fd;
 }
 
@@ -108,10 +109,11 @@ Result<u64> sys_close(Registers*, SyscallArgs args)
     int fd = (int)args[0];
     if (fd < 0 || fd >= FD_MAX) return err(EBADF);
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
-    Option<FileDescriptor>& descriptor = current->fd_table[fd];
+    auto table = current->fd_table.lock();
+    Option<FileDescriptor>& descriptor = (*table)[fd];
 
     if (!descriptor.has_value()) return err(EBADF);
 

kernel/src/sys/pledge.cpp

@@ -7,7 +7,7 @@ Result<u64> sys_pledge(Registers*, SyscallArgs args)
     int promises = TRY(parse_promises(args[0]));
     int execpromises = TRY(parse_promises(args[1]));
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     if (promises >= 0)
     {

kernel/src/sys/poll.cpp

@@ -18,7 +18,8 @@ Result<u64> sys_poll(Registers*, SyscallArgs args)
 
     if (!MemoryManager::copy_from_user(fds, kfds, nfds * sizeof(pollfd))) return err(EFAULT);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
+    auto* thread = Scheduler::current();
     TRY(check_pledge(current, Promise::p_stdio));
 
     Vector<SharedPtr<VFS::Inode>> inodes;
@@ -78,8 +79,8 @@ Result<u64> sys_poll(Registers*, SyscallArgs args)
         if (!fds_with_events && (timeout > 0 || infinite))
         {
             kernel_sleep(10);
-            timeout -= (10 - (int)current->sleep_ticks_left);
-            if (current->interrupted)
+            timeout -= (10 - (int)thread->sleep_ticks_left);
+            if (thread->interrupted)
             {
                 guard.deactivate();
                 free_impl(kfds);

kernel/src/sys/pstat.cpp

@@ -15,28 +15,29 @@ Result<u64> sys_pstat(Registers*, SyscallArgs args)
     pid_t pid = (pid_t)args[0];
     struct process* ps = (struct process*)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_proc));
 
     // If pid == -1, return the PID of the last spawned thread.
-    if (pid == -1) return g_threads.expect_last()->id;
+    if (pid == -1) return g_processes.expect_last()->id;
 
-    auto* thread = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
+    auto* target = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
 
     struct process proc;
-    proc.ps_pid = thread->id;
-    proc.ps_ppid = thread->parent ? thread->parent->id : 0;
-    proc.ps_uid = thread->auth.uid;
-    proc.ps_gid = thread->auth.gid;
-    proc.ps_euid = thread->auth.euid;
-    proc.ps_egid = thread->auth.egid;
-    proc.ps_state = (int)thread->state;
-    proc.ps_flags = thread->is_kernel ? PS_FLAG_KRNL : 0;
-    set_timespec(proc.ps_time, thread->user_ticks_self + thread->kernel_ticks_self);
-    set_timespec(proc.ps_ktime, thread->kernel_ticks_self);
-    set_timespec(proc.ps_utime, thread->kernel_ticks_children);
-    strlcpy(proc.ps_name, thread->cmdline.chars(), sizeof(proc.ps_name));
-    strlcpy(proc.ps_cwd, thread->current_directory_path.is_empty() ? "/" : thread->current_directory_path.chars(),
+    proc.ps_pid = target->id;
+    proc.ps_ppid = target->parent ? target->parent->id : 0;
+    auto auth = target->credentials();
+    proc.ps_uid = auth.uid;
+    proc.ps_gid = auth.gid;
+    proc.ps_euid = auth.euid;
+    proc.ps_egid = auth.egid;
+    proc.ps_state = 0; // FIXME: this is thread-specific now
+    proc.ps_flags = 0; // FIXME: add flags
+    set_timespec(proc.ps_time, target->user_ticks_self + target->kernel_ticks_self);
+    set_timespec(proc.ps_ktime, target->kernel_ticks_self);
+    set_timespec(proc.ps_utime, target->kernel_ticks_children);
+    strlcpy(proc.ps_name, target->cmdline.chars(), sizeof(proc.ps_name));
+    strlcpy(proc.ps_cwd, target->current_directory_path.is_empty() ? "/" : target->current_directory_path.chars(),
             sizeof(proc.ps_cwd));
 
     if (!MemoryManager::copy_to_user_typed(ps, &proc)) return err(EFAULT);

kernel/src/sys/resource.cpp

@@ -16,7 +16,7 @@ Result<u64> sys_getrusage(Registers*, SyscallArgs args)
     int who = (int)args[0];
     struct rusage* ru = (struct rusage*)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     struct rusage kru;
     switch (who)

kernel/src/sys/setitimer.cpp

@@ -16,7 +16,7 @@ Result<u64> sys_setitimer(Registers*, SyscallArgs args)
     const struct itimerval* new_timer = (const struct itimerval*)args[1];
     struct itimerval* old_timer = (struct itimerval*)args[2];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -66,7 +66,7 @@ Result<u64> sys_setitimer(Registers*, SyscallArgs args)
         if (!MemoryManager::copy_from_user_typed(new_timer, &itimer)) return err(EFAULT);
 
         timer->signo = SIGALRM; // FIXME: Also use SIGVTALRM or SIGPROF for other timer types.
-        timer->thread = current;
+        timer->process = current;
 
         if (itimer.it_interval.tv_sec != 0 || itimer.it_interval.tv_usec != 0)
         {
@@ -93,7 +93,7 @@ Result<u64> sys_timer_create(Registers*, SyscallArgs args)
     struct sigevent* sevp = (struct sigevent*)args[1];
     timer_t* timerid = (timer_t*)args[2];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -118,7 +118,6 @@ Result<u64> sys_timer_create(Registers*, SyscallArgs args)
     if (ksevp.sigev_signo <= 0 || ksevp.sigev_signo > NSIG) return err(EINVAL);
 
     int id = TRY(current->allocate_timerid());
-    current->posix_timers[id] = Timer {};
 
     Timer* timer = current->posix_timers[id].value_ptr();
     timer->signo = ksevp.sigev_signo;
@@ -138,7 +137,7 @@ Result<u64> sys_timer_settime(Registers*, SyscallArgs args)
     if (timerid < 0 || timerid >= MAX_POSIX_TIMERS) return err(EINVAL);
     if (flags > 0) return err(ENOTSUP);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -170,7 +169,7 @@ Result<u64> sys_timer_settime(Registers*, SyscallArgs args)
     Clock* clock = timer->designated_clock;
     check(clock);
 
-    timer->thread = current;
+    timer->process = current;
 
     if (itimer.it_interval.tv_sec != 0 || itimer.it_interval.tv_nsec != 0)
     {
@@ -196,7 +195,7 @@ Result<u64> sys_timer_gettime(Registers*, SyscallArgs args)
     struct itimerspec* value = (struct itimerspec*)args[1];
     if (timerid < 0 || timerid >= MAX_POSIX_TIMERS) return err(EINVAL);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 
@@ -225,7 +224,7 @@ Result<u64> sys_timer_delete(Registers*, SyscallArgs args)
     timer_t timerid = (timer_t)args[0];
     if (timerid < 0 || timerid >= MAX_POSIX_TIMERS) return err(EINVAL);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
 
     TRY(check_pledge(current, Promise::p_stdio));
 

kernel/src/sys/signal.cpp

@@ -19,7 +19,7 @@ Result<u64> sys_sigreturn(Registers* regs, SyscallArgs)
 Result<u64> sys_sigaction(Registers*, SyscallArgs args)
 {
     auto* current = Scheduler::current();
-    TRY(check_pledge(current, Promise::p_stdio));
+    TRY(check_pledge(current->process, Promise::p_stdio));
 
     int signo = (int)args[0];
     const struct sigaction* act = (const struct sigaction*)args[1];
@@ -48,15 +48,16 @@ Result<u64> sys_sigaction(Registers*, SyscallArgs args)
 
 Result<u64> sys_kill(Registers*, SyscallArgs args)
 {
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_proc));
 
     pid_t pid = (pid_t)args[0];
     int signo = (int)args[1];
 
-    auto send_signal = [&](Thread* target) -> Result<void> {
-        if (current->auth.euid != 0 && current->auth.euid != target->auth.euid &&
-            current->auth.egid != target->auth.egid)
+    auto send_signal = [&](Process* target) -> Result<void> {
+        auto this_auth = current->credentials();
+        auto other_auth = target->credentials();
+        if (this_auth.euid != 0 && this_auth.euid != other_auth.euid && this_auth.egid != other_auth.egid)
             return err(EPERM);
         if (target->is_kernel) return {};
         if (signo == 0) return {};
@@ -68,14 +69,14 @@ Result<u64> sys_kill(Registers*, SyscallArgs args)
 
     if (pid > 0)
     {
-        auto* target = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
+        auto* target = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
         TRY(send_signal(target));
     }
     else if (pid == 0)
     {
         int errno = -1;
         bool pgid_exists = false;
-        Scheduler::for_each_in_process_group(current->pgid, [&](Thread* target) {
+        Scheduler::for_each_in_process_group(current->pgid, [&](Process* target) {
             pgid_exists = true;
             auto rc = send_signal(target);
             if (rc.has_error())
@@ -90,17 +91,17 @@ Result<u64> sys_kill(Registers*, SyscallArgs args)
     }
     else if (pid == -1)
     {
-        for (auto* thread : g_threads)
+        for (auto* process : g_processes)
         {
             // We ignore permission errors here.
-            if (thread != current && thread->id != 1) send_signal(thread);
+            if (process != current && process->id != 1) send_signal(process);
         }
     }
     else if (pid < -1)
     {
         int errno = -1;
         bool pgid_exists = false;
-        Scheduler::for_each_in_process_group(-pid, [&](Thread* target) {
+        Scheduler::for_each_in_process_group(-pid, [&](Process* target) {
             pgid_exists = true;
             auto rc = send_signal(target);
             if (rc.has_error())
@@ -120,7 +121,7 @@ Result<u64> sys_kill(Registers*, SyscallArgs args)
 Result<u64> sys_sigprocmask(Registers*, SyscallArgs args)
 {
     auto* current = Scheduler::current();
-    TRY(check_pledge(current, Promise::p_stdio));
+    TRY(check_pledge(current->process, Promise::p_stdio));
 
     int how = (int)args[0];
     const sigset_t* set = (const sigset_t*)args[1];

kernel/src/sys/socket.cpp

@@ -15,14 +15,13 @@ Result<u64> sys_socket(Registers*, SyscallArgs args)
     if (type != SOCK_STREAM) return err(EPROTOTYPE);
     if (domain != AF_UNIX) return err(EAFNOSUPPORT);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_unix));
 
     auto socket = TRY(make_shared<UnixSocket>());
 
-    int fd = TRY(current->allocate_fd(0));
-
-    current->fd_table[fd] = FileDescriptor { TRY(make_shared<OpenFileDescription>(socket, O_RDWR)), 0 };
+    auto descriptor = FileDescriptor { TRY(make_shared<OpenFileDescription>(socket, O_RDWR)), 0 };
+    int fd = TRY(current->allocate_fd(0, descriptor));
 
     return fd;
 }
@@ -37,7 +36,7 @@ Result<u64> sys_bind(Registers*, SyscallArgs args)
     if ((usize)addrlen > sizeof(storage)) return err(EINVAL);
     if (!MemoryManager::copy_from_user(addr, &storage, addrlen)) return err(EFAULT);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_unix));
 
     auto inode = TRY(current->resolve_fd(sockfd))->inode();
@@ -61,7 +60,7 @@ Result<u64> sys_connect(Registers* regs, SyscallArgs args)
     if ((usize)addrlen > sizeof(storage)) return err(EINVAL);
     if (!MemoryManager::copy_from_user(addr, &storage, addrlen)) return err(EFAULT);
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_unix));
 
     auto description = TRY(current->resolve_fd(sockfd))->description;
@@ -80,7 +79,7 @@ Result<u64> sys_listen(Registers*, SyscallArgs args)
     int sockfd = (int)args[0];
     int backlog = (int)args[1];
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_unix));
 
     auto inode = TRY(current->resolve_fd(sockfd))->inode();
@@ -108,7 +107,7 @@ Result<u64> sys_accept(Registers* regs, SyscallArgs args)
         if (!MemoryManager::copy_from_user_typed(addrlen, &len)) return err(EFAULT);
     }
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_unix));
 
     auto description = TRY(current->resolve_fd(sockfd))->description;
@@ -121,8 +120,8 @@ Result<u64> sys_accept(Registers* regs, SyscallArgs args)
     socklen_t client_len;
     auto new_description = TRY(socket->accept(regs, description->flags, &client, &client_len));
 
-    int fd = TRY(current->allocate_fd(0));
-    current->fd_table[fd] = FileDescriptor { new_description, 0 };
+    auto descriptor = FileDescriptor { new_description, 0 };
+    int fd = TRY(current->allocate_fd(0, descriptor));
 
     if (client_len < len) len = client_len;
     if (addr)

kernel/src/sys/stat.cpp

@@ -33,7 +33,7 @@ Result<u64> sys_fstatat(Registers*, SyscallArgs args)
     stat* st = (stat*)args[2];
     int flags = (int)args[3];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_rpath));
 
     auto inode = TRY(current->resolve_atfile(dirfd, path, flags & AT_EMPTY_PATH, !(flags & AT_SYMLINK_NOFOLLOW)));
@@ -68,22 +68,24 @@ Result<u64> sys_faccessat(Registers*, SyscallArgs args)
 
     Credentials creds;
 
-    auto* current = Scheduler::current();
+    auto* current = Process::current();
     TRY(check_pledge(current, Promise::p_rpath));
+    auto auth = current->credentials();
 
-    if (flags & AT_EACCESS) creds = current->auth;
+    if (flags & AT_EACCESS) creds = auth;
     else
     {
-        auto auth = current->auth;
         creds.euid = auth.uid;
         creds.egid = auth.gid;
     }
 
     auto inode = TRY(current->resolve_atfile(dirfd, path, false, true));
 
-    if ((amode & R_OK) && !VFS::can_read(inode, creds, &current->extra_groups)) return err(EACCES);
-    if ((amode & W_OK) && !VFS::can_write(inode, creds, &current->extra_groups)) return err(EACCES);
-    if ((amode & X_OK) && !VFS::can_execute(inode, creds, &current->extra_groups)) return err(EACCES);
+    auto groups = current->extra_groups.lock();
+
+    if ((amode & R_OK) && !VFS::can_read(inode, creds, &groups.ref())) return err(EACCES);
+    if ((amode & W_OK) && !VFS::can_write(inode, creds, &groups.ref())) return err(EACCES);
+    if ((amode & X_OK) && !VFS::can_execute(inode, creds, &groups.ref())) return err(EACCES);
 
     // Either all checks succeeded, or amode == F_OK and the file exists, since resolve_atfile() would have failed
     // otherwise.

kernel/src/sys/uname.cpp

@@ -40,9 +40,9 @@ Result<u64> sys_sethostname(Registers*, SyscallArgs args)
     const char* buf = (const char*)args[0];
     usize length = (usize)args[1];
 
-    Thread* current = Scheduler::current();
+    Process* current = Process::current();
     TRY(check_pledge(current, Promise::p_host));
-    if (current->auth.euid != 0) return err(EPERM);
+    if (current->credentials().euid != 0) return err(EPERM);
 
     if (length >= _UTSNAME_LENGTH) return err(EINVAL);
 

kernel/src/sys/usleep.cpp

@@ -8,12 +8,12 @@ Result<u64> sys_usleep(Registers*, SyscallArgs args)
     useconds_t us = (useconds_t)args[0];
 
     auto* current = Scheduler::current();
-    TRY(check_pledge(current, Promise::p_stdio));
+    TRY(check_pledge(current->process, Promise::p_stdio));
 
     // FIXME: Allow usleep() to use a more precise resolution.
     if (us < 1000) return 0;
 
     kernel_sleep(us / 1000);
 
-    return current->sleep_ticks_left;
+    return current->sleep_ticks_left.load();
 }

kernel/src/sys/waitpid.cpp

@@ -12,21 +12,21 @@ Result<u64> sys_waitpid(Registers* regs, SyscallArgs args)
     int options = (int)args[2];
 
     Thread* current = Scheduler::current();
-    TRY(check_pledge(current, Promise::p_stdio));
+    TRY(check_pledge(current->process, Promise::p_stdio));
 
-    Thread* thread;
+    Process* target;
 
     if (pid > 0)
     {
-        thread = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
+        target = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(pid), ESRCH));
 
-        if (thread->parent && thread->parent != current) return err(ECHILD);
+        if (target->parent && target->parent != current->process) return err(ECHILD);
 
         if (options & WNOHANG) return err(EAGAIN);
 
     wait_for_child:
-        if (thread->state != ThreadState::Exited) kernel_wait(pid);
-        if (current->interrupted)
+        if (!target->dead()) kernel_wait(pid);
+        if (current->interrupted && (current->pending_signal_count() > 1 || current->pending_signal() != SIGCHLD))
         {
             kdbgln("signal: waitpid interrupted by signal");
             if (current->will_ignore_pending_signal())
@@ -37,20 +37,20 @@ Result<u64> sys_waitpid(Registers* regs, SyscallArgs args)
             return err(EINTR);
         }
 
-        check(thread->state == ThreadState::Exited);
+        check(target->dead());
     }
     else if (pid == -1)
     {
-        if (!Scheduler::has_children(current)) return err(ECHILD);
+        if (!Scheduler::has_children(current->process)) return err(ECHILD);
 
-        auto child = Scheduler::find_exited_child(current);
+        auto child = Scheduler::find_exited_child(current->process);
         if (!child.has_value())
         {
             if (options & WNOHANG) return err(EAGAIN);
 
         wait_for_any_child:
             kernel_wait(pid);
-            if (current->interrupted)
+            if (current->interrupted && (current->pending_signal_count() > 1 || current->pending_signal() != SIGCHLD))
             {
                 kdbgln("signal: waitpid interrupted by signal");
                 if (current->will_ignore_pending_signal())
@@ -61,26 +61,26 @@ Result<u64> sys_waitpid(Registers* regs, SyscallArgs args)
                 return err(EINTR);
             }
 
-            check(current->child_being_waited_for.value_or(-1) != -1);
+            check(current->child_being_waited_for != -1);
 
-            thread = TRY(Result<Thread*>::from_option(Scheduler::find_by_pid(*current->child_being_waited_for), ESRCH));
-            check(thread->state == ThreadState::Exited);
+            target = TRY(Result<Process*>::from_option(Scheduler::find_by_pid(current->child_being_waited_for), ESRCH));
+            check(target->dead());
         }
         else
-            thread = child.value();
+            target = child.value();
     }
     else // FIXME: Now that we have process groups, implement the cases where pid = 0 and pid < -1.
         return err(ENOTSUP);
 
-    current->child_being_waited_for = {};
+    current->child_being_waited_for = -2;
 
-    int status = (int)thread->status;
-    u64 id = thread->id;
+    int status = (int)target->status;
+    u64 id = target->id;
 
-    current->user_ticks_children += thread->user_ticks_self + thread->user_ticks_children;
-    current->kernel_ticks_children += thread->kernel_ticks_self + thread->kernel_ticks_children;
+    current->process->user_ticks_children += target->user_ticks_self + target->user_ticks_children;
+    current->process->kernel_ticks_children += target->kernel_ticks_self + target->kernel_ticks_children;
 
-    thread->state = ThreadState::Dying;
+    target->thread_count = PROCESS_SHOULD_REAP;
     Scheduler::signal_reap_thread();
 
     if (status_ptr)

kernel/src/thread/Clock.cpp

@@ -124,7 +124,7 @@ void Clock::tick()
         {
             this->m_timer_queue.remove(t);
             t->active_clock = nullptr;
-            t->thread->send_signal(t->signo);
+            t->process->send_signal(t->signo);
             if (t->restart) timers_to_be_restarted.append(t);
             return true;
         }

kernel/src/thread/Scheduler.cpp

@@ -11,8 +11,9 @@
 #include <luna/Stack.h>
 
 static Thread g_idle;
+static Process g_idle_process;
 static Thread* g_current = nullptr;
-static Thread* g_init = nullptr;
+static Process* g_init = nullptr;
 static Thread* g_reap = nullptr;
 static Thread* g_oom = nullptr;
 
@@ -22,15 +23,20 @@ namespace Scheduler
 {
     void init()
     {
-        g_idle.id = 0;
+        g_idle.tid = 0;
         g_idle.init_regs_kernel();
         g_idle.set_ip((u64)CPU::idle_loop);
         g_idle.state = ThreadState::Idle;
         g_idle.is_kernel = true;
-        g_idle.parent = nullptr;
+        g_idle.process = &g_idle_process;
         g_idle.cmdline = "[idle]";
         g_idle.active_directory = nullptr;
 
+        g_idle_process.id = 0;
+        g_idle_process.parent = nullptr;
+        g_idle_process.thread_count = 1;
+        g_idle_process.is_kernel = true;
+
         g_idle.ticks_left = 1;
 
         // Map some stack for the idle task
@@ -42,7 +48,7 @@ namespace Scheduler
 
         g_idle.stack = idle_stack;
 
-        kinfoln("Created idle thread: id %d with ip %#lx and sp %#lx", g_idle.id, g_idle.ip(), g_idle.sp());
+        kinfoln("Created idle thread: id %d with ip %#lx and sp %#lx", g_idle_process.id, g_idle.ip(), g_idle.sp());
 
         g_current = &g_idle;
     }
@@ -57,7 +63,7 @@ namespace Scheduler
         return &g_idle;
     }
 
-    Thread* init_thread()
+    Process* init_process()
     {
         return g_init;
     }
@@ -88,30 +94,38 @@ namespace Scheduler
         // If anything fails, make sure to clean up.
         auto guard = make_scope_guard([&] { delete thread; });
 
+        Process* process = TRY(make<Process>());
+
+        auto guard2 = make_scope_guard([&] { delete process; });
+
         const u64 thread_stack_vm = TRY(MemoryManager::alloc_for_kernel(4, MMU::NoExecute | MMU::ReadWrite));
 
         guard.deactivate();
+        guard2.deactivate();
 
         const Stack thread_stack { thread_stack_vm, ARCH_PAGE_SIZE * 4 };
         thread->set_sp(thread_stack.top());
-
         thread->stack = thread_stack;
-
         thread->cmdline = name;
-
         thread->is_kernel = true;
         thread->active_directory = MMU::kernel_page_directory();
 
-        thread->virtual_clock.set_resolution(1'000'000);
-        thread->profiling_clock.set_resolution(1'000'000);
+        thread->process = process;
 
-        thread->auth = Credentials { .uid = 0, .euid = 0, .suid = 0, .gid = 0, .egid = 0, .sgid = 0 };
+        process->id = thread->tid;
+        process->parent = nullptr;
+        process->thread_count = 1;
+        process->virtual_clock.set_resolution(1'000'000);
+        process->profiling_clock.set_resolution(1'000'000);
+        process->cmdline = name;
+        process->is_kernel = true;
 
         g_threads.append(thread);
+        g_processes.append(process);
 
         thread->state = ThreadState::Runnable;
 
-        kinfoln("Created kernel thread: id %d with ip %#lx and sp %#lx", thread->id, thread->ip(), thread->sp());
+        kinfoln("Created kernel thread: id %d with ip %#lx and sp %#lx", process->id, thread->ip(), thread->sp());
 
         return thread;
     }
@@ -149,14 +163,17 @@ namespace Scheduler
         check(!g_init);
 
         Thread* const thread = TRY(make<Thread>());
+        Process* const process = TRY(make<Process>());
 
         thread->state = ThreadState::None;
-        thread->is_kernel = false;
-        thread->id = 1;
-        thread->pgid = 1;
+        thread->tid = 1;
         thread->cmdline = name;
-        thread->auth = Credentials { .uid = 0, .euid = 0, .suid = 0, .gid = 0, .egid = 0, .sgid = 0 };
-        thread->extra_groups = {};
+        thread->process = process;
+
+        process->id = 1;
+        process->pgid = 1;
+        process->thread_count = 1;
+        process->cmdline = name;
 
         Vector<String> args;
         auto name_string = TRY(String::from_cstring(name));
@@ -164,7 +181,10 @@ namespace Scheduler
 
         Vector<String> env;
 
-        auto guard = make_scope_guard([&] { delete thread; });
+        auto guard = make_scope_guard([&] {
+            delete thread;
+            delete process;
+        });
 
         // Contrary to other programs, which use BinaryFormat::create_loader(), init must be a native executable.
         auto loader = TRY(ELFLoader::create(inode, nullptr, 0));
@@ -188,11 +208,12 @@ namespace Scheduler
             thread->signal_handlers[i] = { .sa_handler = SIG_DFL, .sa_mask = 0, .sa_flags = 0 };
         }
 
-        kinfoln("Created userspace thread: id %d with ip %#.16lx and sp %#.16lx (ksp %#lx)", thread->id, thread->ip(),
+        kinfoln("Created userspace thread: id %d with ip %#.16lx and sp %#.16lx (ksp %#lx)", process->id, thread->ip(),
                 thread->sp(), thread->kernel_stack.top());
 
         g_threads.append(thread);
-        g_init = thread;
+        g_processes.append(process);
+        g_init = process;
 
         return thread;
     }
@@ -202,12 +223,35 @@ namespace Scheduler
         g_threads.append(thread);
     }
 
+    void add_process(Process* process)
+    {
+        g_processes.append(process);
+    }
+
+    void reap_process(Process* process)
+    {
+        CPU::disable_interrupts();
+
+        // FIXME: Shouldn't all this be done when the timers' destructors are called?
+        process->real_timer.disarm();
+        process->virtual_timer.disarm();
+        process->profiling_timer.disarm();
+        for (int i = 0; i < MAX_POSIX_TIMERS; i++)
+        {
+            if (process->posix_timers[i].has_value()) process->posix_timers[i]->disarm();
+        }
+
+        delete process;
+
+        CPU::enable_interrupts();
+    }
+
     void reap_thread(Thread* thread)
     {
         CPU::disable_interrupts();
 
 #ifdef REAP_DEBUG
-        kdbgln("reap: reaping thread with id %d", thread->id);
+        kdbgln("reap: reaping thread with id %d", thread->tid);
 #endif
 
         if (thread->is_kernel)
@@ -221,15 +265,6 @@ namespace Scheduler
             MemoryManager::unmap_owned_and_free_vm(stack.bottom(), stack.bytes() / ARCH_PAGE_SIZE).release_value();
         }
 
-        // FIXME: Shouldn't all this be done when the timers' destructors are called?
-        thread->real_timer.disarm();
-        thread->virtual_timer.disarm();
-        thread->profiling_timer.disarm();
-        for (int i = 0; i < MAX_POSIX_TIMERS; i++)
-        {
-            if (thread->posix_timers[i].has_value()) thread->posix_timers[i]->disarm();
-        }
-
         delete thread;
 
         CPU::enable_interrupts();
@@ -307,14 +342,19 @@ namespace Scheduler
     {
         CPU::disable_interrupts();
 
-        if (is_in_kernel(regs)) g_current->kernel_ticks_self++;
+        if (is_in_kernel(regs))
+        {
+            g_current->process->kernel_ticks_self++;
+            g_current->kernel_ticks_self++;
+        }
         else
         {
-            g_current->virtual_clock.tick();
+            g_current->process->virtual_clock.tick();
+            g_current->process->user_ticks_self++;
             g_current->user_ticks_self++;
         }
 
-        g_current->profiling_clock.tick();
+        g_current->process->profiling_clock.tick();
 
         g_current->ticks_left--;
 
@@ -344,21 +384,46 @@ namespace Scheduler
         return result;
     }
 
-    Option<Thread*> find_by_pid(pid_t pid)
+    LinkedList<Process> check_for_dead_processes()
     {
-        for (auto* const thread : g_threads)
+        LinkedList<Process> result;
+
+        g_processes.delayed_for_each([&](Process* p) {
+            if (p->thread_count == PROCESS_SHOULD_REAP)
+            {
+                g_processes.remove(p);
+                result.append(p);
+            }
+        });
+
+        return result;
+    }
+
+    Option<Process*> find_by_pid(pid_t pid)
+    {
+        for (auto* const process : g_processes)
         {
-            if (thread->id == pid && thread->state != ThreadState::Dying) return thread;
+            if (process->id == pid) return process;
         }
 
         return {};
     }
 
-    bool has_children(Thread* thread)
+    Option<Thread*> find_by_tid(pid_t tid)
+    {
+        for (auto* const thread : g_threads)
+        {
+            if (thread->tid == tid) return thread;
+        }
+
+        return {};
+    }
+
+    bool has_children(Process* process)
     {
         bool result { false };
 
-        for_each_child(thread, [&](Thread*) {
+        for_each_child(process, [&](Process*) {
             result = true;
             return false;
         });
@@ -366,12 +431,12 @@ namespace Scheduler
         return result;
     }
 
-    Option<Thread*> find_exited_child(Thread* thread)
+    Option<Process*> find_exited_child(Process* process)
     {
-        Option<Thread*> result;
+        Option<Process*> result;
 
-        for_each_child(thread, [&](Thread* child) {
-            if (!result.has_value() && child->state == ThreadState::Exited)
+        for_each_child(process, [&](Process* child) {
+            if (!result.has_value() && child->dead())
             {
                 result = child;
                 return false;
@@ -387,16 +452,24 @@ namespace Scheduler
         CPU::disable_interrupts();
 
         kdbgln("--- BEGIN SCHEDULER DUMP ---");
-        kdbgln("current at %p, id = %d", g_current, g_current->id);
+        kdbgln("Current thread at %p, tid = %d", g_current, g_current->tid);
+        kdbgln("Current process at %p, pid = %d", g_current->process, g_current->process->id);
 
         for (const auto* thread : g_threads)
         {
-            kdbgln("%p %c [%-20s] %4d, parent = (%-18p,%d), state = %d, ticks: (k:%04zu,u:%04zu), status = "
-                   "%d, cwd = %s",
-                   thread, thread->is_kernel ? 'k' : 'u', thread->cmdline.chars(), thread->id, thread->parent,
-                   thread->parent ? thread->parent->id : 0, (int)thread->state, thread->kernel_ticks_self,
-                   thread->user_ticks_self, thread->status,
-                   thread->current_directory_path.is_empty() ? "/" : thread->current_directory_path.chars());
+            kdbgln("Thread %p (belongs to pid %4d) %c [%-20s] %4d, state = %d, ip = %p", thread, thread->process->id,
+                   thread->is_kernel ? 'k' : 'u', thread->cmdline.chars(), thread->tid, (int)thread->state,
+                   (void*)thread->ip());
+        }
+
+        for (const auto* process : g_processes)
+        {
+            kdbgln("Process %p (%zu threads) %4d, parent = (%-18p,%d), cwd = %s, ticks: (k:%04zu,u:%04zu), "
+                   "status = %d",
+                   process, process->thread_count.load(), process->id, process->parent,
+                   process->parent ? process->parent->id : 0,
+                   process->current_directory_path.is_empty() ? "/" : process->current_directory_path.chars(),
+                   process->kernel_ticks_self.load(), process->user_ticks_self.load(), process->status);
         }
 
         kdbgln("--- END SCHEDULER DUMP ---");
@@ -428,6 +501,7 @@ void kernel_wait_for_event()
 [[noreturn]] void kernel_exit()
 {
     g_current->state = ThreadState::Dying;
+    g_current->process->thread_count = PROCESS_SHOULD_REAP;
     Scheduler::signal_reap_thread();
     kernel_yield();
     unreachable();

kernel/src/thread/Scheduler.h

@@ -8,7 +8,7 @@ namespace Scheduler
 
     Thread* current();
     Thread* idle();
-    Thread* init_thread();
+    Process* init_process();
 
     void set_reap_thread(Thread*);
     void signal_reap_thread();
@@ -23,24 +23,41 @@ namespace Scheduler
     Result<Thread*> create_init_process(SharedPtr<VFS::Inode> inode, const char* name);
 
     void add_thread(Thread* thread);
+    void add_process(Process* process);
 
     Thread* pick_task();
 
     void reap_thread(Thread* thread);
+    void reap_process(Process* thread);
 
     void switch_task(Registers* regs);
 
     void invoke(Registers* regs);
 
     LinkedList<Thread> check_for_dying_threads();
+    LinkedList<Process> check_for_dead_processes();
 
-    Option<Thread*> find_by_pid(pid_t pid);
+    Option<Process*> find_by_pid(pid_t pid);
+    Option<Thread*> find_by_tid(pid_t tid);
 
-    template <typename Callback> void for_each_child(Thread* thread, Callback callback)
+    template <typename Callback> void for_each_child(Process* process, Callback callback)
     {
-        for (Thread* current = thread; current; current = g_threads.next(current).value_or(nullptr))
+        for (Process* current = process; current; current = g_processes.next(current).value_or(nullptr))
         {
-            if (current->parent == thread)
+            if (current->parent == process)
+            {
+                bool should_continue = callback(current);
+                if (!should_continue) return;
+            }
+        }
+    }
+
+    template <typename Callback> void for_each_thread(Process* process, Callback callback)
+    {
+        for (Thread* current = g_threads.first().value_or(nullptr); current;
+             current = g_threads.next(current).value_or(nullptr))
+        {
+            if (current->process == process)
             {
                 bool should_continue = callback(current);
                 if (!should_continue) return;
@@ -50,8 +67,8 @@ namespace Scheduler
 
     template <typename Callback> void for_each_in_process_group(pid_t group, Callback callback)
     {
-        for (Thread* current = g_threads.first().value_or(nullptr); current;
-             current = g_threads.next(current).value_or(nullptr))
+        for (Process* current = g_processes.first().value_or(nullptr); current;
+             current = g_processes.next(current).value_or(nullptr))
         {
             if (current->pgid == group)
             {
@@ -63,8 +80,8 @@ namespace Scheduler
 
     template <typename Callback> void for_each_in_session(pid_t sid, Callback callback)
     {
-        for (Thread* current = g_threads.first().value_or(nullptr); current;
-             current = g_threads.next(current).value_or(nullptr))
+        for (Process* current = g_processes.first().value_or(nullptr); current;
+             current = g_processes.next(current).value_or(nullptr))
         {
             if (current->sid == sid)
             {
@@ -76,9 +93,9 @@ namespace Scheduler
 
     void dump_state();
 
-    bool has_children(Thread* thread);
+    bool has_children(Process* thread);
 
-    Option<Thread*> find_exited_child(Thread* thread);
+    Option<Process*> find_exited_child(Process* thread);
 }
 
 extern "C" void kernel_yield();

kernel/src/thread/Thread.cpp

@@ -15,6 +15,7 @@
 static Atomic<pid_t> g_next_id;
 
 LinkedList<Thread> g_threads;
+LinkedList<Process> g_processes;
 
 void Thread::init()
 {
@@ -25,7 +26,7 @@ Result<Thread*> new_thread()
 {
     Thread* const thread = TRY(make<Thread>());
 
-    thread->id = g_next_id++;
+    thread->tid = g_next_id++;
 
     return thread;
 }
@@ -35,63 +36,83 @@ pid_t next_thread_id()
     return g_next_id.load();
 }
 
-Result<int> Thread::allocate_fd(int min)
+Result<int> Process::allocate_fd(int min, FileDescriptor& descriptor)
 {
     if (min < 0 || min >= FD_MAX) return err(EINVAL);
+
+    auto table = fd_table.lock();
     for (int i = min; i < FD_MAX; i++)
     {
-        // FIXME: Possible race condition if multiple threads share a FileDescriptorTable? Let's not worry about it for
-        // now, we're still a long way away from reaching that point.
-        if (!fd_table[i].has_value()) { return i; }
+        if (!(*table)[i].has_value())
+        {
+            (*table)[i] = descriptor;
+            return i;
+        }
     }
 
     return err(EMFILE);
 }
 
-Result<FileDescriptor*> Thread::resolve_fd(int fd)
+Result<FileDescriptor*> Process::resolve_fd(int fd)
 {
     if (fd < 0 || fd >= FD_MAX) return err(EBADF);
 
-    Option<FileDescriptor>& maybe_descriptor = fd_table[fd];
+    auto table = fd_table.lock();
+
+    Option<FileDescriptor>& maybe_descriptor = (*table)[fd];
 
     if (!maybe_descriptor.has_value()) return err(EBADF);
 
     return maybe_descriptor.value_ptr();
 }
 
-Result<int> Thread::allocate_timerid()
+Credentials Process::credentials()
 {
+    auto credentials = auth.lock();
+    return *credentials;
+}
+
+Result<Vector<gid_t>> Process::copy_groups()
+{
+    auto groups = extra_groups.lock();
+    return groups->shallow_copy();
+}
+
+Result<int> Process::allocate_timerid()
+{
+    ScopedMutexLock lock(posix_timer_mutex);
+
     for (int i = 0; i < MAX_POSIX_TIMERS; i++)
     {
-        // FIXME: Possible race condition, this should be used alongside a mutex.
-        if (!posix_timers[i].has_value()) { return i; }
+        if (!posix_timers[i].has_value())
+        {
+            posix_timers[i] = Timer {};
+            return i;
+        }
     }
 
     return err(EMFILE);
 }
 
-Result<Timer*> Thread::resolve_timerid(int tid)
+Result<Timer*> Process::resolve_timerid(int _tid)
 {
-    if (tid < 0 || tid >= MAX_POSIX_TIMERS) return err(EBADF);
+    if (_tid < 0 || _tid >= MAX_POSIX_TIMERS) return err(EBADF);
 
-    Option<Timer>& maybe_timer = posix_timers[tid];
+    Option<Timer>& maybe_timer = posix_timers[_tid];
 
     if (!maybe_timer.has_value()) return err(EINVAL);
 
     return maybe_timer.value_ptr();
 }
 
-Result<SharedPtr<VFS::Inode>> Thread::resolve_atfile(int dirfd, const String& path, bool allow_empty_path,
-                                                     bool follow_last_symlink, SharedPtr<VFS::Inode>* parent_inode)
+Result<SharedPtr<VFS::Inode>> Process::resolve_atfile(int dirfd, const String& path, bool allow_empty_path,
+                                                      bool follow_last_symlink, SharedPtr<VFS::Inode>* parent_inode)
 {
     if (parent_inode) *parent_inode = this->current_directory;
 
-    if (PathParser::is_absolute(path.view()))
-        return VFS::resolve_path(path.chars(), this->auth, &this->extra_groups, {}, follow_last_symlink);
+    if (PathParser::is_absolute(path.view())) return VFS::resolve_path(path.chars(), this, {}, follow_last_symlink);
 
-    if (dirfd == AT_FDCWD)
-        return VFS::resolve_path(path.chars(), this->auth, &this->extra_groups, this->current_directory,
-                                 follow_last_symlink);
+    if (dirfd == AT_FDCWD) return VFS::resolve_path(path.chars(), this, this->current_directory, follow_last_symlink);
 
     auto descriptor = TRY(resolve_fd(dirfd));
 
@@ -99,63 +120,92 @@ Result<SharedPtr<VFS::Inode>> Thread::resolve_atfile(int dirfd, const String& pa
 
     if (path.is_empty() && allow_empty_path) return descriptor->inode();
 
-    return VFS::resolve_path(path.chars(), this->auth, &this->extra_groups, descriptor->inode(), follow_last_symlink);
+    return VFS::resolve_path(path.chars(), this, descriptor->inode(), follow_last_symlink);
 }
 
-[[noreturn]] void Thread::exit_and_signal_parent(int _status)
+[[noreturn]] void Process::exit(int _status)
 {
-    check(!is_kernel);
+    check(this == Process::current()); // Process::exit() should only be called by the process itself.
 
 #ifndef MOON_ENABLE_TESTING_FEATURES
-    if (this->id == 1) fail("the init process exited");
+    if (id == 1) fail("the init process exited");
 #else
-    if (this->id == 1) CPU::magic_exit(_status);
+    if (id == 1) CPU::magic_exit(_status);
 #endif
 
-    Scheduler::for_each_child(this, [](Thread* child) {
-        child->parent = Scheduler::init_thread();
+    Scheduler::for_each_thread(this, [](Thread* thread) {
+        thread->quit();
+        return true;
+    });
+    Scheduler::signal_reap_thread();
+
+    thread_count = 0;
+
+    status = _status;
+
+    Scheduler::for_each_child(this, [](Process* child) {
+        child->parent = Scheduler::init_process();
         return true;
     });
 
     if (is_session_leader())
     {
-        kinfoln("thread %d is exiting as a session leader, sending signals to session", id);
+        kinfoln("process %d is exiting as a session leader, sending signals to session", id);
         // FIXME: Send SIGHUP only to the foreground process group if the session has a controlling terminal.
-        Scheduler::for_each_in_session(sid, [this](Thread* thread) {
-            if (thread == this) return true;
-            thread->sid = 0;
-            thread->controlling_terminal = {};
-            thread->send_signal(SIGHUP);
-            kinfoln("reparenting and sending SIGHUP to %d", thread->id);
+        Scheduler::for_each_in_session(sid, [this](Process* p) {
+            if (p == this) return true;
+            p->sid = 0;
+            p->controlling_terminal = {};
+            p->send_signal(SIGHUP);
+            kinfoln("reparenting and sending SIGHUP to %d", p->id);
             return true;
         });
     }
 
     if (parent)
     {
-        if (parent->state == ThreadState::Waiting)
-        {
-            auto child = *parent->child_being_waited_for;
-            if (child == -1 || child == id)
+        Scheduler::for_each_thread(parent, [&](Thread* t) {
+            if (t->state == ThreadState::Waiting)
             {
-                parent->child_being_waited_for = id;
-                parent->wake_up();
+                pid_t expected = -1;
+                if (t->child_being_waited_for.compare_exchange_strong(expected, id))
+                {
+                    t->wake_up();
+                    return false;
+                }
+
+                expected = id;
+                if (t->child_being_waited_for.compare_exchange_strong(expected, id))
+                {
+                    t->wake_up();
+                    return false;
+                }
             }
-        }
-        else
-        {
-            while (parent->pending_signals.get(SIGCHLD - 1)) kernel_yield();
-            parent->send_signal(SIGCHLD);
-        }
+            return true;
+        });
+
+        parent->send_signal(SIGCHLD);
     }
 
-    state = ThreadState::Exited;
-
-    status = _status;
     kernel_yield();
     unreachable();
 }
 
+void Thread::quit()
+{
+    state = ThreadState::Dying;
+}
+
+void Thread::exit(bool yield)
+{
+    quit();
+
+    process->thread_count--;
+    if (process->thread_count == 0) { process->exit(0); }
+
+    if (yield) kernel_yield();
+}
+
 enum class DefaultSignalAction
 {
     Ignore,
@@ -200,7 +250,7 @@ void Thread::process_pending_signals(Registers* current_regs)
             if (handler.sa_handler == SIG_DFL || signo == SIGKILL || signo == SIGSTOP)
             {
             default_signal:
-                if (id == 1)
+                if (process->id == 1)
                 {
                     kwarnln("signal: init got a signal it has no handler for, ignoring");
                     return;
@@ -211,9 +261,10 @@ void Thread::process_pending_signals(Registers* current_regs)
                 {
                 case DefaultSignalAction::Ignore: return;
                 case DefaultSignalAction::Terminate:
-                    kwarnln("Terminating thread %d with signal %d", id, signo);
+                    kwarnln("Terminating thread %d with signal %d", tid, signo);
                     CPU::print_stack_trace_at(current_regs);
-                    exit_and_signal_parent(signo | _SIGBIT);
+                    process->exit(signo | _SIGBIT);
+                    unreachable();
                 case DefaultSignalAction::Stop: stop();
                 default: return;
                 }
@@ -226,6 +277,25 @@ void Thread::process_pending_signals(Registers* current_regs)
     }
 }
 
+int Thread::pending_signal_count()
+{
+    int result = 0;
+    for (int i = 0; i < NSIG; i++)
+    {
+        if (pending_signals.get(i)) { result++; }
+    }
+    return result;
+}
+
+int Thread::pending_signal()
+{
+    for (int i = 0; i < NSIG; i++)
+    {
+        if (pending_signals.get(i)) { return i + 1; }
+    }
+    return 0;
+}
+
 bool Thread::will_ignore_pending_signal()
 {
     for (int i = 0; i < NSIG; i++)
@@ -244,6 +314,14 @@ bool Thread::will_ignore_pending_signal()
     return false;
 }
 
+void Process::send_signal(int signo)
+{
+    Scheduler::for_each_thread(this, [signo](Thread* t) {
+        t->send_signal(signo);
+        return false;
+    });
+}
+
 void Thread::send_signal(int signo)
 {
     if (is_kernel) return;
@@ -297,9 +375,11 @@ bool Thread::check_stack_on_exception(u64 stack_pointer)
             return false;
         }
 
+        auto address_space = process->address_space.lock();
+
         // If we can, we'll add 2 more pages of buffer space, otherwise we use whatever we can.
         usize bytes_to_grow = min(stack_space_remaining, exceeded_bytes + 2 * ARCH_PAGE_SIZE);
-        auto maybe_base = address_space->grow_region(stack.bottom(), bytes_to_grow / ARCH_PAGE_SIZE, true);
+        auto maybe_base = (*address_space)->grow_region(stack.bottom(), bytes_to_grow / ARCH_PAGE_SIZE, true);
         if (maybe_base.has_error())
         {
             kwarnln("Failed to grow stack: could not allocate virtual memory space (%s)", maybe_base.error_string());
@@ -311,7 +391,7 @@ bool Thread::check_stack_on_exception(u64 stack_pointer)
                                                      MMU::ReadWrite | MMU::NoExecute | MMU::User);
         if (result.has_error())
         {
-            address_space->free_region(base, bytes_to_grow / ARCH_PAGE_SIZE);
+            (*address_space)->free_region(base, bytes_to_grow / ARCH_PAGE_SIZE);
             kwarnln("Failed to grow stack: could not allocate physical pages (%s)", result.error_string());
             return false;
         }
@@ -331,3 +411,8 @@ void Thread::stop()
     state = ThreadState::Stopped;
     kernel_yield();
 }
+
+Process* Process::current()
+{
+    return Scheduler::current()->process;
+}

kernel/src/thread/Thread.h

@@ -1,8 +1,8 @@
 #pragma once
-
 #include "arch/MMU.h"
 #include "fs/OpenFileDescription.h"
 #include "fs/VFS.h"
+#include "lib/Mutex.h"
 #include "memory/AddressSpace.h"
 #include <bits/signal.h>
 #include <luna/Bitset.h>
@@ -20,6 +20,7 @@
 #endif
 
 constexpr int MAX_POSIX_TIMERS = 64;
+constexpr i64 PROCESS_SHOULD_REAP = -1;
 
 class Timer;
 
@@ -47,59 +48,33 @@ struct Credentials
     u32 sgid { 0 };
 };
 
-struct Thread : public LinkedListNode<Thread>
+struct Process : public LinkedListNode<Process>
 {
-    Registers regs;
+    Atomic<i64> thread_count;
 
     pid_t id;
-    pid_t pgid { 0 };
-    pid_t sid { 0 };
+    Atomic<pid_t> pgid { 0 };
+    Atomic<pid_t> sid { 0 };
 
-    Credentials auth;
-    Vector<gid_t> extra_groups;
+    bool has_called_exec { false };
 
-    u64 user_ticks_self = 0;
-    u64 kernel_ticks_self = 0;
-    u64 user_ticks_children = 0;
-    u64 kernel_ticks_children = 0;
-
-    u64 ticks_left;
-    u64 sleep_ticks_left;
+    mode_t umask { 0 };
 
     int promises { -1 };
     int execpromises { -1 };
 
-    Stack stack;
-    Stack kernel_stack;
+    Process* parent { nullptr };
 
-    OwnedPtr<AddressSpace> address_space;
-    Option<FileDescriptor> fd_table[FD_MAX] = {};
+    MutexLocked<Credentials> auth { Credentials { 0, 0, 0, 0, 0, 0 } };
 
-    Result<int> allocate_fd(int min);
-    Result<FileDescriptor*> resolve_fd(int fd);
-    Result<SharedPtr<VFS::Inode>> resolve_atfile(int dirfd, const String& path, bool allow_empty_path,
-                                                 bool follow_last_symlink,
-                                                 SharedPtr<VFS::Inode>* parent_inode = nullptr);
+    MutexLocked<Vector<gid_t>> extra_groups { {} };
 
-    struct sigaction signal_handlers[NSIG];
-    Bitset<sigset_t> signal_mask { 0 };
-    Bitset<sigset_t> pending_signals { 0 };
-    bool interrupted { false };
+    Credentials credentials();
+    Result<Vector<gid_t>> copy_groups();
 
-    SharedPtr<VFS::Inode> controlling_terminal;
+    MutexLocked<OwnedPtr<AddressSpace>> address_space;
 
-    bool unrestricted_task { false };
-
-    FPData fp_data;
-
-    ThreadState state = ThreadState::Runnable;
-
-    bool is_kernel { true };
-    bool has_called_exec { false };
-
-    int status { 0 };
-
-    mode_t umask { 0 };
+    MutexLocked<Option<FileDescriptor>[FD_MAX]> fd_table = {};
 
     Timer real_timer;
     Timer virtual_timer;
@@ -108,27 +83,104 @@ struct Thread : public LinkedListNode<Thread>
     Clock virtual_clock;
     Clock profiling_clock;
 
+    bool is_kernel { false };
+
     Option<Timer> posix_timers[MAX_POSIX_TIMERS];
+    Mutex posix_timer_mutex;
+
+    StaticString<128> cmdline;
+
+    Atomic<u64> user_ticks_self = 0;
+    Atomic<u64> kernel_ticks_self = 0;
+    Atomic<u64> user_ticks_children = 0;
+    Atomic<u64> kernel_ticks_children = 0;
 
     Result<int> allocate_timerid();
     Result<Timer*> resolve_timerid(int id);
 
-    StaticString<128> cmdline;
+    Result<int> allocate_fd(int min, FileDescriptor& descriptor);
+    Result<FileDescriptor*> resolve_fd(int fd);
+    Result<SharedPtr<VFS::Inode>> resolve_atfile(int dirfd, const String& path, bool allow_empty_path,
+                                                 bool follow_last_symlink,
+                                                 SharedPtr<VFS::Inode>* parent_inode = nullptr);
 
     String current_directory_path = {};
     SharedPtr<VFS::Inode> current_directory = {};
 
-    Thread* parent { nullptr };
-    Option<pid_t> child_being_waited_for = {};
+    SharedPtr<VFS::Inode> controlling_terminal;
+
+    int status { 0 };
+
+    void send_signal(int signo);
+
+    bool is_session_leader()
+    {
+        return id == sid;
+    }
+
+    bool alive()
+    {
+        return thread_count > 0;
+    }
+
+    bool dead()
+    {
+        return thread_count == 0;
+    }
+
+    static Process* current();
+
+    [[noreturn]] void exit(int status);
+};
+
+struct Thread : public LinkedListNode<Thread>
+{
+    Process* process;
+
+    pid_t tid;
+
+    Registers regs;
+
+    Atomic<u64> ticks_left;
+    Atomic<u64> sleep_ticks_left;
+
+    Atomic<u64> user_ticks_self = 0;
+    Atomic<u64> kernel_ticks_self = 0;
+
+    Stack stack;
+    Stack kernel_stack;
+
+    struct sigaction signal_handlers[NSIG];
+    Bitset<sigset_t> signal_mask { 0 };
+    Bitset<sigset_t> pending_signals { 0 };
+    bool interrupted { false };
+
+    Atomic<pid_t> child_being_waited_for = -2;
+
+    bool unrestricted_task { false };
+
+    FPData fp_data;
+
+    ThreadState state = ThreadState::Runnable;
+
+    bool is_kernel { false };
+
+    StaticString<128> cmdline;
 
     PageDirectory* self_directory() const
     {
-        return address_space->page_directory();
+        PageDirectory* result;
+        auto lambda = Function<OwnedPtr<AddressSpace>&>::wrap([&](OwnedPtr<AddressSpace>& space) {
+                          result = space->page_directory();
+                      }).release_value();
+        process->address_space.with_lock(move(lambda));
+        return result;
     }
 
     PageDirectory* active_directory { nullptr };
 
-    [[noreturn]] void exit_and_signal_parent(int status);
+    void quit();
+    void exit(bool yield = true);
 
     bool is_idle()
     {
@@ -140,27 +192,24 @@ struct Thread : public LinkedListNode<Thread>
         state = ThreadState::Runnable;
     }
 
-    bool is_session_leader()
-    {
-        return id == sid;
-    }
-
     void init_regs_kernel();
     void init_regs_user();
 
     void set_arguments(u64 arg1, u64 arg2, u64 arg3, u64 arg4);
 
     void set_ip(u64 ip);
-    u64 ip();
+    u64 ip() const;
 
     void set_sp(u64 sp);
-    u64 sp();
+    u64 sp() const;
 
     void set_return(u64 ret);
     u64 return_register();
 
     void process_pending_signals(Registers* current_regs);
 
+    int pending_signal_count();
+    int pending_signal();
     bool will_ignore_pending_signal();
 
     bool deliver_signal(int signo, Registers* current_regs);
@@ -188,3 +237,4 @@ Result<Thread*> new_thread();
 pid_t next_thread_id();
 
 extern LinkedList<Thread> g_threads;
+extern LinkedList<Process> g_processes;

kernel/src/thread/ThreadImage.cpp

@@ -51,7 +51,9 @@ Result<OwnedPtr<ThreadImage>> ThreadImage::clone_from_thread(Thread* parent)
 {
     auto image = TRY(make_owned<ThreadImage>());
 
-    auto address_space = TRY(parent->address_space->clone());
+    auto space = parent->process->address_space.lock();
+
+    auto address_space = TRY((*space)->clone());
 
     const u64 kernel_stack_base = TRY(MemoryManager::alloc_for_kernel(4, MMU::ReadWrite | MMU::NoExecute));
     Stack kernel_stack { kernel_stack_base, 4 * ARCH_PAGE_SIZE };
@@ -67,15 +69,7 @@ Result<OwnedPtr<ThreadImage>> ThreadImage::clone_from_thread(Thread* parent)
 
 Result<u64> ThreadImage::push_mem_on_stack(const u8* mem, usize size)
 {
-    if ((m_sp - size) < m_user_stack.bottom()) return err(E2BIG);
-
-    if (!MemoryManager::validate_user_write((void*)(m_sp - size), size)) return err(EFAULT);
-
-    m_sp -= size;
-
-    memcpy((void*)m_sp, mem, size);
-
-    return m_sp;
+    return MemoryManager::push_mem_on_stack(mem, size, m_user_stack, m_sp);
 }
 
 Result<u64> ThreadImage::push_string_vector_on_stack(const Vector<String>& vec)
@@ -106,5 +100,6 @@ void ThreadImage::apply(Thread* thread)
 
     thread->active_directory = m_address_space->page_directory();
 
-    thread->address_space = move(m_address_space);
+    auto space = thread->process->address_space.lock();
+    *space = move(m_address_space);
 }