syscall: add AcpiVerb::SetS3WakingVector and AcpiVerb::EnterS3 (Phase II.X.W)

Phase II.X.W: extend the AcpiVerb enum with two new variants
for the S3 round-trip:

* AcpiVerb::SetS3WakingVector (= 5): acpid writes the
  kernel's S3 resume trampoline address to FACS.
  Mirrors Linux 7.1's acpi_set_firmware_waking_vector.
  The 8-byte write payload is the address in
  little-endian. A zero payload is a sentinel for 'use the
  kernel's default trampoline address' (s3_trampoline
  symbol).
* AcpiVerb::EnterS3 (= 6): acpid requests the kernel to
  enter S3. The kernel's stop::enter_s3() reads the
  SLP_TYP value from S3_SLP_TYP and does the PM1 register
  write. The actual S3 entry happens via acpid writing
  to /scheme/sys/kstop.

Hardware-agnostic: works on any x86_64 system with
standard ACPI S3 support (Dell, HP, Lenovo, LG Gram 14).
On Modern Standby-only systems (LG Gram 16 (2025)), the
kernel never enters S3 so these verbs are no-ops.

.cargo-ok

@@ -1 +0,0 @@
-{"v":1}

.cargo_vcs_info.json

@@ -1,6 +0,0 @@
-{
-  "git": {
-    "sha1": "7040cf71b3a5d15d91802810d0a50aa197970c43"
-  },
-  "path_in_vcs": ""
-}

.gitignore

@@ -1,2 +1,2 @@
-/target
+Cargo.lock
-/Cargo.lock
+target

.gitlab-ci.yml

@@ -0,0 +1,17 @@
+image: "redoxos/redoxer"
+
+stages:
+    - build
+
+workflow:
+  rules:
+    - if: '$CI_COMMIT_BRANCH == "master" && $CI_PROJECT_NAMESPACE == "redox-os"'
+    - if: '$CI_MERGE_REQUEST_TARGET_BRANCH_NAME == "master"'
+
+build:linux:
+    stage: build
+    script: cargo +nightly build
+
+build:redox:
+    stage: build
+    script: redoxer build

Cargo.toml

@@ -1,73 +1,25 @@
-# THIS FILE IS AUTOMATICALLY GENERATED BY CARGO
-#
-# When uploading crates to the registry Cargo will automatically
-# "normalize" Cargo.toml files for maximal compatibility
-# with all versions of Cargo and also rewrite `path` dependencies
-# to registry (e.g., crates.io) dependencies.
-#
-# If you are reading this file be aware that the original Cargo.toml
-# will likely look very different (and much more reasonable).
-# See Cargo.toml.orig for the original contents.
-
 [package]
-edition = "2021"
+name = "redox_syscall"
-name = "libredox"
+version = "0.8.1"
-version = "0.1.18"
+description = "A Rust library to access raw Redox system calls"
-authors = ["4lDO2 <4lDO2@protonmail.com>"]
-build = false
-exclude = ["target"]
-autolib = false
-autobins = false
-autoexamples = false
-autotests = false
-autobenches = false
-description = "Redox stable ABI"
-readme = false
 license = "MIT"
-repository = "https://gitlab.redox-os.org/redox-os/libredox.git"
+authors = ["Jeremy Soller <jackpot51@gmail.com>"]
-
+repository = "https://gitlab.redox-os.org/redox-os/syscall"
-[features]
+documentation = "https://docs.rs/redox_syscall"
-base = ["libc"]
+edition = "2021"
-call = ["base"]
-default = [
-    "base",
-    "call",
-    "std",
-    "protocol",
-]
-mkns = ["ioslice"]
-protocol = [
-    "plain",
-    "bitflags",
-]
-std = ["base"]
 
 [lib]
-name = "libredox"
+name = "syscall"
-path = "src/lib.rs"
 
-[dependencies.bitflags]
+[features]
-version = "2"
+default = ["userspace"]
-optional = true
+rustc-dep-of-std = ["core", "bitflags/rustc-dep-of-std"]
+userspace = []
+std = []
 
-[dependencies.ioslice]
+[dependencies]
-version = "0.6"
+bitflags = "2.4"
-optional = true
+core = { version = "1.0.0", optional = true, package = "rustc-std-workspace-core" }
 
-[dependencies.libc]
+[target.'cfg(loom)'.dev-dependencies]
-version = "0.2"
+loom = "0.7"
-optional = true
-
-[dependencies.plain]
-version = "0.2"
-optional = true
-
-# Red Bear OS Phase J: path override to the local syscall
-# fork. The local fork at ../syscall/ adds the
-# EnterS2Idle/ExitS2Idle AcPiVerb variants. This breaks
-# the libredox::error::Error <-> syscall::Error type-
-# identity barrier that previously caused E0277 errors in
-# scheme-utils and daemon.
-[dependencies.redox_syscall]
-path = "../syscall"
-version = "0.8"

Cargo.toml.orig

@@ -1,44 +0,0 @@
-[package]
-name = "libredox"
-authors = ["4lDO2 <4lDO2@protonmail.com>"]
-# Red Bear OS Phase J: version is 0.1.18 upstream. The
-# redox_syscall dep is now required (not optional) because
-# the local fork's acpi module (added in this commit) re-
-# exports AcPiVerb from redox_syscall, and downstream recipes
-# that don't enable the redox_syscall feature get an
-# "unresolved import" error. Making the dep non-optional
-# also matches the upstream 0.1.18 Cargo.toml pattern where
-# the redox_syscall dep is unconditional.
-version = "0.1.18"
-edition = "2021"
-license = "MIT"
-description = "Redox stable ABI"
-# Red Bear OS fork lives at the canonical outer repo
-# (gitea.redbearos.org/vasilito/RedBear-OS).
-repository = "https://gitea.redbearos.org/vasilito/RedBear-OS"
-exclude = ["target"]
-
-# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
-
-[features]
-default = ["base", "call", "std", "protocol"]
-base = ["libc"]
-call = ["base"]
-std = ["base"]
-protocol = ["plain", "bitflags", "redox_syscall"]
-mkns = ["ioslice"]
-
-[dependencies]
-bitflags = { version = "2", optional = true }
-libc = { version = "0.2", optional = true }
-# Phase J: path override to the local fork (../syscall
-# relative to the libredox fork's local/sources/libredox/
-# path). This gives libredox access to the EnterS2Idle /
-# ExitS2Idle AcpiVerb variants. Cargo's [patch.crates-io]
-# in the workspace's outer Cargo.toml (in base/ and kernel/)
-# is what wires this path through to the actual
-# redox_syscall crate; this path entry is the libredox-
-# side patch override for the same crate.
-redox_syscall = { path = "../syscall", version = "0.8" }
-ioslice = { version = "0.6", optional = true }
-plain = { version = "0.2", optional = true }

LICENSE

@@ -1,21 +1,22 @@
+Copyright (c) 2017 Redox OS Developers
+
 MIT License
 
-Copyright (c) 2023 4lDO2
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
 
-Permission is hereby granted, free of charge, to any person obtaining a copy
+The above copyright notice and this permission notice shall be
-of this software and associated documentation files (the "Software"), to deal
+included in all copies or substantial portions of the Software.
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
 
-The above copyright notice and this permission notice shall be included in all
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-copies or substantial portions of the Software.
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.

README.md

@@ -0,0 +1,7 @@
+# syscall
+
+This crate contains the system call numbers and Rust wrappers for the inline Assembly code of system calls.
+
+[![MIT licensed](https://img.shields.io/badge/license-MIT-blue.svg)](./LICENSE)
+[![crates.io](http://meritbadge.herokuapp.com/redox_syscall)](https://crates.io/crates/redox_syscall)
+[![docs.rs](https://docs.rs/redox_syscall/badge.svg)](https://docs.rs/redox_syscall)

src/arch/aarch64.rs

@@ -0,0 +1,207 @@
+use core::{
+    mem,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+use super::error::{Error, Result};
+
+pub const PAGE_SIZE: usize = 4096;
+/// Size of the metadata region used to transfer information from the kernel to the bootstrapper.
+pub const KERNEL_METADATA_SIZE: usize = 4 * PAGE_SIZE;
+
+#[cfg(feature = "userspace")]
+macro_rules! syscall {
+    ($($name:ident($a:ident, $($b:ident, $($c:ident, $($d:ident, $($e:ident, $($f:ident, $($g:ident, )?)?)?)?)?)?);)+) => {
+        $(
+            pub unsafe fn $name($a: usize, $($b: usize, $($c: usize, $($d: usize, $($e: usize, $($f: usize, $($g: usize)?)?)?)?)?)?) -> Result<usize> {
+                let ret: usize;
+
+                core::arch::asm!(
+                    "svc 0",
+                    in("x8") $a,
+                    $(
+                        in("x0") $b,
+                        $(
+                            in("x1") $c,
+                            $(
+                                in("x2") $d,
+                                $(
+                                    in("x3") $e,
+                                    $(
+                                        in("x4") $f,
+                                        $(
+                                            in("x5") $g,
+                                        )?
+                                    )?
+                                )?
+                            )?
+                        )?
+                    )?
+                    lateout("x0") ret,
+                    options(nostack),
+                );
+
+                Error::demux(ret)
+            }
+        )+
+    };
+}
+
+#[cfg(feature = "userspace")]
+syscall! {
+    syscall0(a,);
+    syscall1(a, b,);
+    syscall2(a, b, c,);
+    syscall3(a, b, c, d,);
+    syscall4(a, b, c, d, e,);
+    syscall5(a, b, c, d, e, f,);
+    syscall6(a, b, c, d, e, f, g,);
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct IntRegisters {
+    pub x30: usize,
+    pub x29: usize,
+    pub x28: usize,
+    pub x27: usize,
+    pub x26: usize,
+    pub x25: usize,
+    pub x24: usize,
+    pub x23: usize,
+    pub x22: usize,
+    pub x21: usize,
+    pub x20: usize,
+    pub x19: usize,
+    pub x18: usize,
+    pub x17: usize,
+    pub x16: usize,
+    pub x15: usize,
+    pub x14: usize,
+    pub x13: usize,
+    pub x12: usize,
+    pub x11: usize,
+    pub x10: usize,
+    pub x9: usize,
+    pub x8: usize,
+    pub x7: usize,
+    pub x6: usize,
+    pub x5: usize,
+    pub x4: usize,
+    pub x3: usize,
+    pub x2: usize,
+    pub x1: usize,
+    pub x0: usize,
+}
+
+impl Deref for IntRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const IntRegisters as *const u8,
+                mem::size_of::<IntRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for IntRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut IntRegisters as *mut u8,
+                mem::size_of::<IntRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct FloatRegisters {
+    pub fp_simd_regs: [u128; 32],
+    pub fpsr: u32,
+    pub fpcr: u32,
+}
+
+impl Deref for FloatRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const FloatRegisters as *const u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for FloatRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut FloatRegisters as *mut u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct EnvRegisters {
+    pub tpidr_el0: usize,
+    pub tpidrro_el0: usize,
+}
+impl Deref for EnvRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const EnvRegisters as *const u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for EnvRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut EnvRegisters as *mut u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct Exception {
+    pub kind: usize,
+    // TODO
+}
+impl Deref for Exception {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const Exception as *const u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for Exception {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut Exception as *mut u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}

src/arch/riscv64.rs

@@ -0,0 +1,206 @@
+use super::error::{Error, Result};
+use core::arch::asm;
+use core::{
+    mem,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+pub const PAGE_SIZE: usize = 4096;
+/// Size of the metadata region used to transfer information from the kernel to the bootstrapper.
+pub const KERNEL_METADATA_SIZE: usize = 4 * PAGE_SIZE;
+
+#[cfg(feature = "userspace")]
+macro_rules! syscall {
+    ($($name:ident($a:ident, $($b:ident, $($c:ident, $($d:ident, $($e:ident, $($f:ident, $($g:ident, )?)?)?)?)?)?);)+) => {
+        $(
+            pub unsafe fn $name($a: usize, $($b: usize, $($c: usize, $($d: usize, $($e: usize, $($f: usize, $($g: usize)?)?)?)?)?)?) -> Result<usize> {
+                let ret: usize;
+
+                asm!(
+                    "ecall",
+                    in("a7") $a,
+                    $(
+                        in("a0") $b,
+                        $(
+                            in("a1") $c,
+                            $(
+                                in("a2") $d,
+                                $(
+                                    in("a3") $e,
+                                    $(
+                                        in("a4") $f,
+                                        $(
+                                            in("a5") $g,
+                                        )?
+                                    )?
+                                )?
+                            )?
+                        )?
+                    )?
+                    lateout("a0") ret,
+                    options(nostack),
+                );
+
+                Error::demux(ret)
+            }
+        )+
+    };
+}
+
+#[cfg(feature = "userspace")]
+syscall! {
+    syscall0(a,);
+    syscall1(a, b,);
+    syscall2(a, b, c,);
+    syscall3(a, b, c, d,);
+    syscall4(a, b, c, d, e,);
+    syscall5(a, b, c, d, e, f,);
+    syscall6(a, b, c, d, e, f, g,);
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct IntRegisters {
+    pub pc: usize,
+    pub x31: usize,
+    pub x30: usize,
+    pub x29: usize,
+    pub x28: usize,
+    pub x27: usize,
+    pub x26: usize,
+    pub x25: usize,
+    pub x24: usize,
+    pub x23: usize,
+    pub x22: usize,
+    pub x21: usize,
+    pub x20: usize,
+    pub x19: usize,
+    pub x18: usize,
+    pub x17: usize,
+    pub x16: usize,
+    pub x15: usize,
+    pub x14: usize,
+    pub x13: usize,
+    pub x12: usize,
+    pub x11: usize,
+    pub x10: usize,
+    pub x9: usize,
+    pub x8: usize,
+    pub x7: usize,
+    pub x6: usize,
+    pub x5: usize,
+    // x4(tp) is in env
+    // x3(gp) is a platform scratch register
+    pub x2: usize,
+    pub x1: usize,
+}
+
+impl Deref for IntRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const IntRegisters as *const u8,
+                mem::size_of::<IntRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for IntRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut IntRegisters as *mut u8,
+                mem::size_of::<IntRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct FloatRegisters {
+    pub fregs: [u64; 32],
+    pub fcsr: u32,
+}
+
+impl Deref for FloatRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const FloatRegisters as *const u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for FloatRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut FloatRegisters as *mut u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(packed)]
+pub struct EnvRegisters {
+    pub tp: usize,
+}
+impl Deref for EnvRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const EnvRegisters as *const u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for EnvRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut EnvRegisters as *mut u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct Exception {
+    pub kind: usize,
+    // TODO
+}
+impl Deref for Exception {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const Exception as *const u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for Exception {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut Exception as *mut u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}

src/arch/x86.rs

@@ -0,0 +1,288 @@
+use core::{
+    arch::asm,
+    mem,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+use super::error::{Error, Result};
+
+pub const PAGE_SIZE: usize = 4096;
+/// Size of the metadata region used to transfer information from the kernel to the bootstrapper.
+pub const KERNEL_METADATA_SIZE: usize = 4 * PAGE_SIZE;
+
+#[cfg(feature = "userspace")]
+macro_rules! syscall {
+    ($($name:ident($a:ident, $($b:ident, $($c:ident, $($d:ident, $($e:ident, $($f:ident, )?)?)?)?)?);)+) => {
+        $(
+            pub unsafe fn $name(mut $a: usize, $($b: usize, $($c: usize, $($d: usize, $($e: usize, $($f: usize)?)?)?)?)?) -> Result<usize> {
+                asm!(
+                    "int 0x80",
+                    inout("eax") $a,
+                    $(
+                        in("ebx") $b,
+                        $(
+                            in("ecx") $c,
+                            $(
+                                in("edx") $d,
+                                $(
+                                    in("esi") $e,
+                                    $(
+                                        in("edi") $f,
+                                    )?
+                                )?
+                            )?
+                        )?
+                    )?
+                    options(nostack),
+                );
+
+                Error::demux($a)
+            }
+        )+
+    };
+}
+
+#[cfg(feature = "userspace")]
+syscall! {
+    syscall0(a,);
+    syscall1(a, b,);
+    syscall2(a, b, c,);
+    syscall3(a, b, c, d,);
+    // Must be done custom because LLVM reserves ESI
+    //syscall4(a, b, c, d, e,);
+    //syscall5(a, b, c, d, e, f,);
+    //syscall6(a, b, c, d, e, f, g,);
+}
+
+#[cfg(feature = "userspace")]
+pub unsafe fn syscall4(mut a: usize, b: usize, c: usize, d: usize, e: usize) -> Result<usize> {
+    asm!(
+        "xchg esi, {e}
+        int 0x80
+        xchg esi, {e}",
+        e = in(reg) e,
+        inout("eax") a,
+        in("ebx") b,
+        in("ecx") c,
+        in("edx") d,
+        options(nostack),
+    );
+
+    Error::demux(a)
+}
+
+#[cfg(feature = "userspace")]
+pub unsafe fn syscall5(
+    mut a: usize,
+    b: usize,
+    c: usize,
+    d: usize,
+    e: usize,
+    f: usize,
+) -> Result<usize> {
+    asm!(
+        "xchg esi, {e}
+        int 0x80
+        xchg esi, {e}",
+        e = in(reg) e,
+        inout("eax") a,
+        in("ebx") b,
+        in("ecx") c,
+        in("edx") d,
+        in("edi") f,
+        options(nostack),
+    );
+
+    Error::demux(a)
+}
+
+#[cfg(feature = "userspace")]
+pub unsafe fn syscall6(
+    mut a: usize,
+    b: usize,
+    c: usize,
+    d: usize,
+    e: usize,
+    f: usize,
+    g: usize,
+) -> Result<usize> {
+    #[repr(C)]
+    struct PackedArgs {
+        arg4: usize,
+        arg6: usize,
+        nr: usize,
+    }
+    let args = PackedArgs {
+        arg4: e,
+        arg6: g,
+        nr: a,
+    };
+    let args_ptr = &args as *const PackedArgs;
+    asm!(
+        "push ebp",
+        "push esi",
+        "mov esi, [eax + 0]", // arg4 -> esi
+        "mov ebp, [eax + 4]", // arg6 -> ebp
+        "mov eax, [eax + 8]", // nr -> eax
+        "int 0x80",
+        "pop esi",
+        "pop ebp",
+        inout("eax") args_ptr => a,
+        in("ebx") b,
+        in("ecx") c,
+        in("edx") d,
+        in("edi") f,
+        options(nostack),
+    );
+
+    Error::demux(a)
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct IntRegisters {
+    // TODO: Some of these don't get set by Redox yet. Should they?
+    pub ebp: usize,
+    pub esi: usize,
+    pub edi: usize,
+    pub ebx: usize,
+    pub eax: usize,
+    pub ecx: usize,
+    pub edx: usize,
+    // pub orig_rax: usize,
+    pub eip: usize,
+    pub cs: usize,
+    pub eflags: usize,
+    pub esp: usize,
+    pub ss: usize,
+    // pub fs_base: usize,
+    // pub gs_base: usize,
+    // pub ds: usize,
+    // pub es: usize,
+    pub fs: usize,
+    // pub gs: usize
+}
+
+impl Deref for IntRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const IntRegisters as *const u8,
+                mem::size_of::<IntRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for IntRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut IntRegisters as *mut u8,
+                mem::size_of::<IntRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct FloatRegisters {
+    pub fcw: u16,
+    pub fsw: u16,
+    pub ftw: u8,
+    pub _reserved: u8,
+    pub fop: u16,
+    pub fip: u64,
+    pub fdp: u64,
+    pub mxcsr: u32,
+    pub mxcsr_mask: u32,
+    pub st_space: [u128; 8],
+    pub xmm_space: [u128; 16],
+    // TODO: YMM/ZMM
+}
+
+impl Deref for FloatRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const FloatRegisters as *const u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for FloatRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut FloatRegisters as *mut u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct EnvRegisters {
+    pub fsbase: u32,
+    pub gsbase: u32,
+}
+
+impl Deref for EnvRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const EnvRegisters as *const u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for EnvRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut EnvRegisters as *mut u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct Exception {
+    pub kind: usize,
+    pub code: usize,
+    pub address: usize,
+}
+impl Deref for Exception {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const Exception as *const u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for Exception {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut Exception as *mut u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}

src/arch/x86_64.rs

@@ -0,0 +1,194 @@
+use core::{
+    mem,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+pub const PAGE_SIZE: usize = 4096;
+/// Size of the metadata region used to transfer information from the kernel to the bootstrapper.
+pub const KERNEL_METADATA_SIZE: usize = 4 * PAGE_SIZE;
+
+#[cfg(feature = "userspace")]
+macro_rules! syscall {
+    ($($name:ident($a:ident, $($b:ident, $($c:ident, $($d:ident, $($e:ident, $($f:ident, $($g:ident, )?)?)?)?)?)?);)+) => {
+        $(
+            pub unsafe fn $name(mut $a: usize, $($b: usize, $($c: usize, $($d: usize, $($e: usize, $($f: usize, $($g: usize)?)?)?)?)?)?) -> crate::error::Result<usize> {
+                core::arch::asm!(
+                    "syscall",
+                    inout("rax") $a,
+                    $(
+                        in("rdi") $b,
+                        $(
+                            in("rsi") $c,
+                            $(
+                                in("rdx") $d,
+                                $(
+                                    in("r10") $e,
+                                    $(
+                                        in("r8") $f,
+                                        $(
+                                            in("r9") $g,
+                                        )?
+                                    )?
+                                )?
+                            )?
+                        )?
+                    )?
+                    out("rcx") _,
+                    out("r11") _,
+                    options(nostack),
+                );
+
+                crate::error::Error::demux($a)
+            }
+        )+
+    };
+}
+
+#[cfg(feature = "userspace")]
+syscall! {
+    syscall0(a,);
+    syscall1(a, b,);
+    syscall2(a, b, c,);
+    syscall3(a, b, c, d,);
+    syscall4(a, b, c, d, e,);
+    syscall5(a, b, c, d, e, f,);
+    syscall6(a, b, c, d, e, f, g,);
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct IntRegisters {
+    pub r15: usize,
+    pub r14: usize,
+    pub r13: usize,
+    pub r12: usize,
+    pub rbp: usize,
+    pub rbx: usize,
+    pub r11: usize,
+    pub r10: usize,
+    pub r9: usize,
+    pub r8: usize,
+    pub rax: usize,
+    pub rcx: usize,
+    pub rdx: usize,
+    pub rsi: usize,
+    pub rdi: usize,
+    pub rip: usize,
+    pub cs: usize,
+    pub rflags: usize,
+    pub rsp: usize,
+    pub ss: usize,
+}
+
+impl Deref for IntRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Self as *const u8, mem::size_of::<Self>()) }
+    }
+}
+
+impl DerefMut for IntRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Self as *mut u8, mem::size_of::<Self>()) }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct FloatRegisters {
+    pub fcw: u16,
+    pub fsw: u16,
+    pub ftw: u8,
+    pub _reserved: u8,
+    pub fop: u16,
+    pub fip: u64,
+    pub fdp: u64,
+    pub mxcsr: u32,
+    pub mxcsr_mask: u32,
+    pub st_space: [u128; 8],
+    pub xmm_space: [u128; 16],
+    // TODO: YMM/ZMM
+}
+
+impl Deref for FloatRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const FloatRegisters as *const u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for FloatRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut FloatRegisters as *mut u8,
+                mem::size_of::<FloatRegisters>(),
+            )
+        }
+    }
+}
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct EnvRegisters {
+    pub fsbase: u64,
+    pub gsbase: u64,
+    // TODO: PKRU?
+}
+impl Deref for EnvRegisters {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const EnvRegisters as *const u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for EnvRegisters {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut EnvRegisters as *mut u8,
+                mem::size_of::<EnvRegisters>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C, packed)]
+pub struct Exception {
+    pub kind: usize,
+    pub code: usize,
+    pub address: usize,
+}
+impl Deref for Exception {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const Exception as *const u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for Exception {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut Exception as *mut u8,
+                mem::size_of::<Exception>(),
+            )
+        }
+    }
+}

src/call.rs

@@ -0,0 +1,410 @@
+use super::{
+    arch::*,
+    data::{Map, Stat, StatVfs, StdFsCallMeta, TimeSpec},
+    error::Result,
+    flag::*,
+    number::*,
+};
+
+use core::mem;
+
+/// Close a file
+pub fn close(fd: usize) -> Result<usize> {
+    unsafe { syscall1(SYS_CLOSE, fd) }
+}
+
+/// Get the current system time
+pub fn clock_gettime(clock: usize, tp: &mut TimeSpec) -> Result<usize> {
+    unsafe { syscall2(SYS_CLOCK_GETTIME, clock, tp as *mut TimeSpec as usize) }
+}
+
+/// Copy and transform a file descriptor
+pub fn dup(fd: usize, buf: &[u8]) -> Result<usize> {
+    unsafe { syscall3(SYS_DUP, fd, buf.as_ptr() as usize, buf.len()) }
+}
+
+/// Copy and transform a file descriptor
+pub fn dup2(fd: usize, newfd: usize, buf: &[u8]) -> Result<usize> {
+    unsafe { syscall4(SYS_DUP2, fd, newfd, buf.as_ptr() as usize, buf.len()) }
+}
+
+/// Change file permissions
+pub fn fchmod(fd: usize, mode: u16) -> Result<usize> {
+    unsafe { syscall2(SYS_FCHMOD, fd, mode as usize) }
+}
+
+/// Change file ownership
+pub fn fchown(fd: usize, uid: u32, gid: u32) -> Result<usize> {
+    unsafe { syscall3(SYS_FCHOWN, fd, uid as usize, gid as usize) }
+}
+
+/// Change file descriptor flags
+pub fn fcntl(fd: usize, cmd: usize, arg: usize) -> Result<usize> {
+    unsafe { syscall3(SYS_FCNTL, fd, cmd, arg) }
+}
+
+/// Map a file into memory, but with the ability to set the address to map into, either as a hint
+/// or as a requirement of the map.
+///
+/// # Errors
+/// `EACCES` - the file descriptor was not open for reading
+/// `EBADF` - if the file descriptor was invalid
+/// `ENODEV` - mmapping was not supported
+/// `EINVAL` - invalid combination of flags
+/// `EEXIST` - if [`MapFlags::MAP_FIXED`] was set, and the address specified was already in use.
+///
+pub unsafe fn fmap(fd: usize, map: &Map) -> Result<usize> {
+    syscall3(
+        SYS_FMAP,
+        fd,
+        map as *const Map as usize,
+        mem::size_of::<Map>(),
+    )
+}
+
+/// Unmap whole (or partial) continous memory-mapped files
+pub unsafe fn funmap(addr: usize, len: usize) -> Result<usize> {
+    syscall2(SYS_FUNMAP, addr, len)
+}
+
+/// Retrieve the canonical path of a file
+pub fn fpath(fd: usize, buf: &mut [u8]) -> Result<usize> {
+    unsafe { syscall3(SYS_FPATH, fd, buf.as_mut_ptr() as usize, buf.len()) }
+}
+
+/// Create a link to a file
+pub fn flink<T: AsRef<str>>(fd: usize, path: T) -> Result<usize> {
+    let path = path.as_ref();
+    unsafe { syscall3(SYS_FLINK, fd, path.as_ptr() as usize, path.len()) }
+}
+
+/// Rename a file
+pub fn frename<T: AsRef<str>>(fd: usize, path: T) -> Result<usize> {
+    let path = path.as_ref();
+    unsafe { syscall3(SYS_FRENAME, fd, path.as_ptr() as usize, path.len()) }
+}
+
+/// Get metadata about a file
+pub fn fstat(fd: usize, stat: &mut Stat) -> Result<usize> {
+    unsafe {
+        syscall3(
+            SYS_FSTAT,
+            fd,
+            stat as *mut Stat as usize,
+            mem::size_of::<Stat>(),
+        )
+    }
+}
+
+/// Get metadata about a filesystem
+pub fn fstatvfs(fd: usize, stat: &mut StatVfs) -> Result<usize> {
+    unsafe {
+        syscall3(
+            SYS_FSTATVFS,
+            fd,
+            stat as *mut StatVfs as usize,
+            mem::size_of::<StatVfs>(),
+        )
+    }
+}
+
+/// Sync a file descriptor to its underlying medium
+pub fn fsync(fd: usize) -> Result<usize> {
+    unsafe { syscall1(SYS_FSYNC, fd) }
+}
+
+/// Truncate or extend a file to a specified length
+pub fn ftruncate(fd: usize, len: usize) -> Result<usize> {
+    unsafe { syscall2(SYS_FTRUNCATE, fd, len) }
+}
+
+// Change modify and/or access times
+pub fn futimens(fd: usize, times: &[TimeSpec]) -> Result<usize> {
+    unsafe {
+        syscall3(
+            SYS_FUTIMENS,
+            fd,
+            times.as_ptr() as usize,
+            mem::size_of_val(times),
+        )
+    }
+}
+
+/// Fast userspace mutex
+pub unsafe fn futex(
+    addr: *mut i32,
+    op: usize,
+    val: i32,
+    val2: usize,
+    addr2: *mut i32,
+) -> Result<usize> {
+    syscall5(
+        SYS_FUTEX,
+        addr as usize,
+        op,
+        (val as isize) as usize,
+        val2,
+        addr2 as usize,
+    )
+}
+
+/// Seek to `offset` bytes in a file descriptor
+pub fn lseek(fd: usize, offset: isize, whence: usize) -> Result<usize> {
+    unsafe { syscall3(SYS_LSEEK, fd, offset as usize, whence) }
+}
+
+/// Make a new scheme namespace
+pub fn mkns(schemes: &[[usize; 2]]) -> Result<usize> {
+    unsafe { syscall2(SYS_MKNS, schemes.as_ptr() as usize, schemes.len()) }
+}
+
+/// Change mapping flags
+pub unsafe fn mprotect(addr: usize, size: usize, flags: MapFlags) -> Result<usize> {
+    syscall3(SYS_MPROTECT, addr, size, flags.bits())
+}
+
+/// Sleep for the time specified in `req`
+pub fn nanosleep(req: &TimeSpec, rem: &mut TimeSpec) -> Result<usize> {
+    unsafe {
+        syscall2(
+            SYS_NANOSLEEP,
+            req as *const TimeSpec as usize,
+            rem as *mut TimeSpec as usize,
+        )
+    }
+}
+
+/// Open a file at a specific path
+pub fn openat<T: AsRef<str>>(
+    fd: usize,
+    path: T,
+    flags: usize,
+    fcntl_flags: usize,
+) -> Result<usize> {
+    let path = path.as_ref();
+    unsafe {
+        syscall5(
+            SYS_OPENAT,
+            fd,
+            path.as_ptr() as usize,
+            path.len(),
+            flags,
+            fcntl_flags,
+        )
+    }
+}
+/// Open a file at a specific path with filter
+pub fn openat_with_filter<T: AsRef<str>>(
+    fd: usize,
+    path: T,
+    flags: usize,
+    fcntl_flags: usize,
+    euid: u32,
+    egid: u32,
+) -> Result<usize> {
+    let path = path.as_ref();
+    unsafe {
+        syscall6(
+            SYS_OPENAT_WITH_FILTER,
+            fd,
+            path.as_ptr() as usize,
+            path.len(),
+            flags | fcntl_flags,
+            // NOTE: Short-term solution to allow namespace management.
+            // In the long term, we need to figure out how we should best handle
+            // Unix permissions using capabilities.
+            euid as usize,
+            egid as usize,
+        )
+    }
+}
+
+/// Remove a file at at specific path
+pub fn unlinkat<T: AsRef<str>>(fd: usize, path: T, flags: usize) -> Result<usize> {
+    let path = path.as_ref();
+    unsafe { syscall4(SYS_UNLINKAT, fd, path.as_ptr() as usize, path.len(), flags) }
+}
+/// Remove a file at at specific path with filter
+pub fn unlinkat_with_filter<T: AsRef<str>>(
+    fd: usize,
+    path: T,
+    flags: usize,
+    euid: u32,
+    egid: u32,
+) -> Result<usize> {
+    let path = path.as_ref();
+    unsafe {
+        syscall6(
+            SYS_UNLINKAT_WITH_FILTER,
+            fd,
+            path.as_ptr() as usize,
+            path.len(),
+            flags,
+            // NOTE: Short-term solution to allow namespace management.
+            // In the long term, we need to figure out how we should best handle
+            // Unix permissions using capabilities.
+            euid as usize,
+            egid as usize,
+        )
+    }
+}
+
+/// Read from a file descriptor into a buffer
+pub fn read(fd: usize, buf: &mut [u8]) -> Result<usize> {
+    unsafe { syscall3(SYS_READ, fd, buf.as_mut_ptr() as usize, buf.len()) }
+}
+
+/// Write a buffer to a file descriptor
+///
+/// The kernel will attempt to write the bytes in `buf` to the file descriptor `fd`, returning
+/// either an `Err`, explained below, or `Ok(count)` where `count` is the number of bytes which
+/// were written.
+///
+/// # Errors
+///
+/// * `EAGAIN` - the file descriptor was opened with `O_NONBLOCK` and writing would block
+/// * `EBADF` - the file descriptor is not valid or is not open for writing
+/// * `EFAULT` - `buf` does not point to the process's addressible memory
+/// * `EIO` - an I/O error occurred
+/// * `ENOSPC` - the device containing the file descriptor has no room for data
+/// * `EPIPE` - the file descriptor refers to a pipe or socket whose reading end is closed
+pub fn write(fd: usize, buf: &[u8]) -> Result<usize> {
+    unsafe { syscall3(SYS_WRITE, fd, buf.as_ptr() as usize, buf.len()) }
+}
+
+/// Yield the process's time slice to the kernel
+///
+/// This function will return Ok(0) on success
+pub fn sched_yield() -> Result<usize> {
+    unsafe { syscall0(SYS_YIELD) }
+}
+
+/// Send a file descriptor `fd`, handled by the scheme providing `receiver_socket`. `flags` is
+/// currently unused (must be zero), and `arg` is included in the scheme call.
+///
+/// The scheme can return an arbitrary value.
+pub fn sendfd(receiver_socket: usize, fd: usize, flags: usize, arg: u64) -> Result<usize> {
+    #[cfg(target_pointer_width = "32")]
+    unsafe {
+        syscall5(
+            SYS_SENDFD,
+            receiver_socket,
+            fd,
+            flags,
+            arg as u32 as usize,
+            (arg >> 32) as u32 as usize,
+        )
+    }
+
+    #[cfg(target_pointer_width = "64")]
+    unsafe {
+        syscall4(SYS_SENDFD, receiver_socket, fd, flags, arg as usize)
+    }
+}
+
+pub trait Call {
+    unsafe fn raw_call(
+        &self,
+        payload_ptr: *const u8,
+        len: usize,
+        flags: CallFlags,
+        metadata: &[u64],
+    ) -> Result<usize>;
+}
+
+impl Call for usize {
+    unsafe fn raw_call(
+        &self,
+        payload_ptr: *const u8,
+        len: usize,
+        flags: CallFlags,
+        metadata: &[u64],
+    ) -> Result<usize> {
+        unsafe {
+            syscall5(
+                SYS_CALL,
+                *self,
+                payload_ptr as usize,
+                len,
+                metadata.len() | flags.bits(),
+                metadata.as_ptr() as usize,
+            )
+        }
+    }
+}
+
+impl Call for &[usize] {
+    unsafe fn raw_call(
+        &self,
+        payload_ptr: *const u8,
+        len: usize,
+        flags: CallFlags,
+        metadata: &[u64],
+    ) -> Result<usize> {
+        let combined_flags = flags | CallFlags::MULTIPLE_FDS;
+        unsafe {
+            syscall6(
+                SYS_CALL,
+                self.as_ptr() as usize,
+                payload_ptr as usize,
+                len,
+                metadata.len() | combined_flags.bits(),
+                metadata.as_ptr() as usize,
+                self.len() * mem::size_of::<usize>(),
+            )
+        }
+    }
+}
+
+/// SYS_CALL interface, read-only variant
+pub fn call_ro<T: Call>(
+    fd: T,
+    payload: &mut [u8],
+    flags: CallFlags,
+    metadata: &[u64],
+) -> Result<usize> {
+    unsafe {
+        fd.raw_call(
+            payload.as_mut_ptr(),
+            payload.len(),
+            flags | CallFlags::READ,
+            metadata,
+        )
+    }
+}
+/// SYS_CALL interface, write-only variant
+pub fn call_wo<T: Call>(
+    fd: T,
+    payload: &[u8],
+    flags: CallFlags,
+    metadata: &[u64],
+) -> Result<usize> {
+    unsafe {
+        fd.raw_call(
+            payload.as_ptr(),
+            payload.len(),
+            flags | CallFlags::WRITE,
+            metadata,
+        )
+    }
+}
+/// SYS_CALL interface, read-write variant
+pub fn call_rw<T: Call>(
+    fd: T,
+    payload: &mut [u8],
+    flags: CallFlags,
+    metadata: &[u64],
+) -> Result<usize> {
+    unsafe {
+        fd.raw_call(
+            payload.as_mut_ptr(),
+            payload.len(),
+            flags | CallFlags::READ | CallFlags::WRITE,
+            metadata,
+        )
+    }
+}
+
+pub fn std_fs_call<T: Call>(fd: T, payload: &mut [u8], metadata: &StdFsCallMeta) -> Result<usize> {
+    call_rw(fd, payload, CallFlags::STD_FS, metadata)
+}

src/data.rs

@@ -0,0 +1,501 @@
+use core::{
+    mem,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+use crate::flag::{EventFlags, MapFlags, PtraceFlags, StdFsCallKind};
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct Event {
+    pub id: usize,
+    pub flags: EventFlags,
+    pub data: usize,
+}
+
+impl Deref for Event {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Event as *const u8, mem::size_of::<Event>()) }
+    }
+}
+
+impl DerefMut for Event {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Event as *mut u8, mem::size_of::<Event>()) }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct ITimerSpec {
+    pub it_interval: TimeSpec,
+    pub it_value: TimeSpec,
+}
+
+impl Deref for ITimerSpec {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const ITimerSpec as *const u8,
+                mem::size_of::<ITimerSpec>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for ITimerSpec {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut ITimerSpec as *mut u8,
+                mem::size_of::<ITimerSpec>(),
+            )
+        }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct OldMap {
+    pub offset: usize,
+    pub size: usize,
+    pub flags: MapFlags,
+}
+
+impl Deref for OldMap {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(self as *const OldMap as *const u8, mem::size_of::<OldMap>())
+        }
+    }
+}
+
+impl DerefMut for OldMap {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(self as *mut OldMap as *mut u8, mem::size_of::<OldMap>())
+        }
+    }
+}
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct Map {
+    /// The offset inside the file that is being mapped.
+    pub offset: usize,
+
+    /// The size of the memory map.
+    pub size: usize,
+
+    /// Contains both prot and map flags.
+    pub flags: MapFlags,
+
+    /// Functions as a hint to where in the virtual address space of the running process, to place
+    /// the memory map. If [`MapFlags::MAP_FIXED`] is set, then this address must be the address to
+    /// map to.
+    pub address: usize,
+}
+
+impl Deref for Map {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Map as *const u8, mem::size_of::<Map>()) }
+    }
+}
+
+impl DerefMut for Map {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Map as *mut u8, mem::size_of::<Map>()) }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default, PartialEq)]
+#[repr(C)]
+pub struct Stat {
+    pub st_dev: u64,
+    pub st_ino: u64,
+    pub st_mode: u16,
+    pub st_nlink: u32,
+    pub st_uid: u32,
+    pub st_gid: u32,
+    pub st_size: u64,
+    pub st_blksize: u32,
+    pub st_blocks: u64,
+    pub st_mtime: u64,
+    pub st_mtime_nsec: u32,
+    pub st_atime: u64,
+    pub st_atime_nsec: u32,
+    pub st_ctime: u64,
+    pub st_ctime_nsec: u32,
+}
+
+impl Deref for Stat {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Stat as *const u8, mem::size_of::<Stat>()) }
+    }
+}
+
+impl DerefMut for Stat {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Stat as *mut u8, mem::size_of::<Stat>()) }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default, PartialEq)]
+#[repr(C)]
+pub struct StatVfs {
+    pub f_bsize: u32,
+    pub f_blocks: u64,
+    pub f_bfree: u64,
+    pub f_bavail: u64,
+}
+
+impl Deref for StatVfs {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const StatVfs as *const u8,
+                mem::size_of::<StatVfs>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for StatVfs {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(self as *mut StatVfs as *mut u8, mem::size_of::<StatVfs>())
+        }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default, PartialEq)]
+#[repr(C, packed)]
+pub struct StdFsCallMeta {
+    pub kind: u8, // enum StdFsCallKind
+    _rsvd: [u8; 7],
+    pub arg1: u64,
+    pub arg2: u64,
+}
+
+impl StdFsCallMeta {
+    pub fn new(kind: StdFsCallKind, arg1: u64, arg2: u64) -> Self {
+        Self {
+            kind: kind as u8,
+            _rsvd: [0; 7],
+            arg1,
+            arg2,
+        }
+    }
+}
+
+impl Deref for StdFsCallMeta {
+    type Target = [u64];
+    fn deref(&self) -> &[u64] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const StdFsCallMeta as *const u64,
+                mem::size_of::<StdFsCallMeta>() / mem::size_of::<u64>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for StdFsCallMeta {
+    fn deref_mut(&mut self) -> &mut [u64] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut StdFsCallMeta as *mut u64,
+                mem::size_of::<StdFsCallMeta>() / mem::size_of::<u64>(),
+            )
+        }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default, PartialEq)]
+#[repr(C)]
+pub struct TimeSpec {
+    pub tv_sec: i64,
+    pub tv_nsec: i32,
+}
+
+const NANOS_PER_SEC: u128 = 1_000_000_000;
+
+impl TimeSpec {
+    pub fn from_nanos(nanos: u128) -> Self {
+        Self {
+            tv_sec: i64::try_from(nanos / NANOS_PER_SEC).unwrap_or(i64::MAX),
+            tv_nsec: (nanos % NANOS_PER_SEC) as i32, // guaranteed to never overflow
+        }
+    }
+    pub fn to_nanos(&self) -> u128 {
+        self.tv_sec as u128 * NANOS_PER_SEC + self.tv_nsec as u128
+    }
+}
+
+impl Deref for TimeSpec {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const TimeSpec as *const u8,
+                mem::size_of::<TimeSpec>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for TimeSpec {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(self as *mut TimeSpec as *mut u8, mem::size_of::<TimeSpec>())
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C)]
+pub struct PtraceEvent {
+    pub cause: PtraceFlags,
+    pub a: usize,
+    pub b: usize,
+    pub c: usize,
+    pub d: usize,
+    pub e: usize,
+    pub f: usize,
+}
+
+impl Deref for PtraceEvent {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const PtraceEvent as *const u8,
+                mem::size_of::<PtraceEvent>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for PtraceEvent {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut PtraceEvent as *mut u8,
+                mem::size_of::<PtraceEvent>(),
+            )
+        }
+    }
+}
+
+#[macro_export]
+macro_rules! ptrace_event {
+    ($cause:expr $(, $a:expr $(, $b:expr $(, $c:expr)?)?)?) => {
+        $crate::data::PtraceEvent {
+            cause: $cause,
+            $(a: $a,
+              $(b: $b,
+                $(c: $c,)?
+              )?
+            )?
+            ..Default::default()
+        }
+    }
+}
+
+bitflags::bitflags! {
+    #[derive(PartialEq, Eq, PartialOrd, Ord, Hash, Debug, Clone, Copy, Default)]
+    pub struct GrantFlags: usize {
+        const GRANT_READ = 0x0000_0001;
+        const GRANT_WRITE = 0x0000_0002;
+        const GRANT_EXEC = 0x0000_0004;
+
+        const GRANT_SHARED = 0x0000_0008;
+        const GRANT_LAZY = 0x0000_0010;
+        const GRANT_SCHEME = 0x0000_0020;
+        const GRANT_PHYS = 0x0000_0040;
+        const GRANT_PINNED = 0x0000_0080;
+        const GRANT_PHYS_CONTIGUOUS = 0x0000_0100;
+    }
+}
+
+impl GrantFlags {
+    #[deprecated = "use the safe `from_bits_retain` method instead"]
+    pub unsafe fn from_bits_unchecked(bits: usize) -> Self {
+        Self::from_bits_retain(bits)
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C)]
+pub struct GrantDesc {
+    pub base: usize,
+    pub size: usize,
+    pub flags: GrantFlags,
+    pub offset: u64,
+}
+
+impl Deref for GrantDesc {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe {
+            slice::from_raw_parts(
+                self as *const GrantDesc as *const u8,
+                mem::size_of::<GrantDesc>(),
+            )
+        }
+    }
+}
+
+impl DerefMut for GrantDesc {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut GrantDesc as *mut u8,
+                mem::size_of::<GrantDesc>(),
+            )
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(C)]
+pub struct SetSighandlerData {
+    pub user_handler: usize,
+    pub excp_handler: usize,
+    pub thread_control_addr: usize,
+    pub proc_control_addr: usize,
+}
+
+impl Deref for SetSighandlerData {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Self as *const u8, mem::size_of::<Self>()) }
+    }
+}
+
+impl DerefMut for SetSighandlerData {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Self as *mut u8, mem::size_of::<Self>()) }
+    }
+}
+pub use crate::sigabi::*;
+
+/// UNSTABLE
+#[derive(Copy, Clone, Debug, Default, PartialEq)]
+#[repr(C)]
+pub struct ProcSchemeAttrs {
+    pub pid: u32,
+    pub euid: u32,
+    pub egid: u32,
+    pub prio: u32,
+    pub debug_name: [u8; 32],
+}
+impl Deref for ProcSchemeAttrs {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Self as *const u8, mem::size_of::<Self>()) }
+    }
+}
+impl DerefMut for ProcSchemeAttrs {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut ProcSchemeAttrs as *mut u8,
+                mem::size_of::<ProcSchemeAttrs>(),
+            )
+        }
+    }
+}
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct CtxtStsBuf {
+    pub status: usize,
+    pub excp: crate::Exception,
+}
+impl Deref for CtxtStsBuf {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Self as *const u8, mem::size_of::<Self>()) }
+    }
+}
+impl DerefMut for CtxtStsBuf {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe {
+            slice::from_raw_parts_mut(
+                self as *mut CtxtStsBuf as *mut u8,
+                mem::size_of::<CtxtStsBuf>(),
+            )
+        }
+    }
+}
+
+#[derive(Copy, Clone, Debug, Default)]
+#[repr(C)]
+pub struct NewFdParams {
+    pub offset: u64,
+    pub number: usize,
+    pub flags: usize,
+    pub internal_flags: u8,
+}
+
+#[repr(u8)]
+#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
+pub enum GlobalSchemes {
+    Debug = 1,
+    Event = 2,
+    Memory = 3,
+    Pipe = 4,
+    Serio = 5,
+    Irq = 6,
+    Time = 7,
+    Sys = 8,
+    Proc = 9,
+    Acpi = 10,
+    Dtb = 11,
+}
+impl GlobalSchemes {
+    pub fn try_from_raw(raw: u8) -> Option<Self> {
+        match raw {
+            1 => Some(Self::Debug),
+            2 => Some(Self::Event),
+            3 => Some(Self::Memory),
+            4 => Some(Self::Pipe),
+            5 => Some(Self::Serio),
+            6 => Some(Self::Irq),
+            7 => Some(Self::Time),
+            8 => Some(Self::Sys),
+            9 => Some(Self::Proc),
+            10 => Some(Self::Acpi),
+            11 => Some(Self::Dtb),
+            _ => None,
+        }
+    }
+    pub fn as_str(&self) -> &'static str {
+        match self {
+            Self::Debug => "debug",
+            Self::Event => "event",
+            Self::Memory => "memory",
+            Self::Pipe => "pipe",
+            Self::Serio => "serio",
+            Self::Irq => "irq",
+            Self::Time => "time",
+            Self::Sys => "sys",
+            Self::Proc => "kernel.proc",
+            Self::Acpi => "kernel.acpi",
+            Self::Dtb => "kernel.dtb",
+        }
+    }
+}
+
+#[repr(C)]
+#[derive(Debug, Clone, Copy, Default)]
+pub struct KernelSchemeInfo {
+    pub scheme_id: u8,
+    pub fd: usize,
+}

src/dirent.rs

@@ -0,0 +1,231 @@
+use core::{
+    mem::size_of,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+use crate::{
+    error::{Error, Result, EINVAL},
+    ENAMETOOLONG,
+};
+
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(packed)]
+pub struct DirentHeader {
+    pub inode: u64,
+    /// A filesystem-specific opaque value used to uniquely identify directory entries. This value,
+    /// in the last returned entry from a SYS_GETDENTS invocation, shall be passed to the next
+    /// call.
+    pub next_opaque_id: u64,
+    // This struct intentionally does not include a "next" offset field, unlike Linux, to easily
+    // guarantee the iterator will be reasonably deterministic, even if the scheme is adversarial.
+    pub record_len: u16,
+    /// A `DirentKind`.
+    ///
+    /// May not be directly available (Unspecified), and if so needs to be looked using fstat.
+    pub kind: u8,
+}
+
+impl Deref for DirentHeader {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Self as *const u8, size_of::<Self>()) }
+    }
+}
+
+impl DerefMut for DirentHeader {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Self as *mut u8, size_of::<Self>()) }
+    }
+}
+
+// Note: Must match relibc/include/bits/dirent.h
+#[derive(Clone, Copy, Debug, Default)]
+#[repr(u8)]
+pub enum DirentKind {
+    #[default]
+    Unspecified = 0,
+
+    CharDev = 2,
+    Directory = 4,
+    BlockDev = 6,
+    Regular = 8,
+    Symlink = 10,
+    Socket = 12,
+}
+
+impl DirentKind {
+    // TODO: derive(FromPrimitive)
+    pub fn try_from_raw(raw: u8) -> Option<Self> {
+        Some(match raw {
+            0 => Self::Unspecified,
+
+            2 => Self::CharDev,
+            4 => Self::Directory,
+            6 => Self::BlockDev,
+            8 => Self::Regular,
+            10 => Self::Symlink,
+            12 => Self::Socket,
+
+            _ => return None,
+        })
+    }
+}
+
+pub struct DirentIter<'a>(&'a [u8]);
+
+impl<'a> DirentIter<'a> {
+    pub const fn new(buffer: &'a [u8]) -> Self {
+        Self(buffer)
+    }
+}
+#[derive(Debug)]
+pub struct Invalid;
+
+impl<'a> Iterator for DirentIter<'a> {
+    type Item = Result<(&'a DirentHeader, &'a [u8]), Invalid>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        if self.0.len() < size_of::<DirentHeader>() {
+            return None;
+        }
+        let header = unsafe { &*(self.0.as_ptr().cast::<DirentHeader>()) };
+        if self.0.len() < usize::from(header.record_len) {
+            return Some(Err(Invalid));
+        }
+        let (this, remaining) = self.0.split_at(usize::from(header.record_len));
+        self.0 = remaining;
+
+        let name_and_nul = &this[size_of::<DirentHeader>()..];
+        let name = &name_and_nul[..name_and_nul.len() - 1];
+
+        Some(Ok((header, name)))
+    }
+}
+
+#[derive(Debug)]
+pub struct DirentBuf<B> {
+    buffer: B,
+
+    // Exists in order to allow future extensions to the DirentHeader struct.
+
+    // TODO: Might add an upper bound to protect against cache miss DoS. The kernel currently
+    // forbids any other value than size_of::<DirentHeader>().
+    header_size: u16,
+
+    written: usize,
+}
+/// Abstraction between &mut [u8] and the kernel's UserSliceWo.
+pub trait Buffer<'a>: Sized + 'a {
+    fn empty() -> Self;
+    fn length(&self) -> usize;
+
+    /// Split all of `self` into two disjoint contiguous subbuffers of lengths `index` and `length
+    /// - index` respectively.
+    ///
+    /// Returns None if and only if `index > length`.
+    fn split_at(self, index: usize) -> Option<[Self; 2]>;
+
+    /// Copy from `src`, lengths must match exactly.
+    ///
+    /// Allowed to overwrite subsequent buffer space, for performance reasons. Can be changed in
+    /// the future if too restrictive.
+    fn copy_from_slice_exact(self, src: &[u8]) -> Result<()>;
+
+    /// Write zeroes to this part of the buffer.
+    ///
+    /// Allowed to overwrite subsequent buffer space, for performance reasons. Can be changed in
+    /// the future if too restrictive.
+    fn zero_out(self) -> Result<()>;
+}
+impl<'a> Buffer<'a> for &'a mut [u8] {
+    fn empty() -> Self {
+        &mut []
+    }
+    fn length(&self) -> usize {
+        self.len()
+    }
+
+    fn split_at(self, index: usize) -> Option<[Self; 2]> {
+        self.split_at_mut_checked(index).map(|(a, b)| [a, b])
+    }
+    fn copy_from_slice_exact(self, src: &[u8]) -> Result<()> {
+        self.copy_from_slice(src);
+        Ok(())
+    }
+    fn zero_out(self) -> Result<()> {
+        self.fill(0);
+        Ok(())
+    }
+}
+
+pub struct DirEntry<'name> {
+    pub inode: u64,
+    pub next_opaque_id: u64,
+    pub name: &'name str,
+    pub kind: DirentKind,
+}
+
+impl<'a, B: Buffer<'a>> DirentBuf<B> {
+    pub fn new(buffer: B, header_size: u16) -> Option<Self> {
+        if usize::from(header_size) < size_of::<DirentHeader>() {
+            return None;
+        }
+
+        Some(Self {
+            buffer,
+            header_size,
+            written: 0,
+        })
+    }
+    pub fn entry(&mut self, entry: DirEntry<'_>) -> Result<()> {
+        let name16 = u16::try_from(entry.name.len()).map_err(|_| Error::new(EINVAL))?;
+        let record_align = align_of::<*const DirentHeader>();
+        let record_len = self
+            .header_size
+            .checked_add(name16)
+            // XXX: NUL byte. Unfortunately this is probably the only performant way to be
+            // compatible with C.
+            .and_then(|l| l.checked_add(1))
+            // Align length so next header is aligned
+            .and_then(|l| l.checked_next_multiple_of(record_align as u16))
+            .ok_or(Error::new(ENAMETOOLONG))?;
+
+        let [this, remaining] = core::mem::replace(&mut self.buffer, B::empty())
+            .split_at(usize::from(record_len))
+            .ok_or(Error::new(EINVAL))?;
+
+        let [this_header_variable, this_name_and_nul] = this
+            .split_at(usize::from(self.header_size))
+            .expect("already know header_size + ... >= header_size");
+
+        let [this_name, this_name_nul] = this_name_and_nul
+            .split_at(usize::from(name16))
+            .expect("already know name.len() <= name.len() + 1");
+
+        // Every write here is currently sequential, allowing the buffer trait to do optimizations
+        // where subbuffer writes are out-of-bounds (but inside the total buffer).
+
+        let [this_header, this_header_extra] = this_header_variable
+            .split_at(size_of::<DirentHeader>())
+            .expect("already checked header_size <= size_of Header");
+
+        this_header.copy_from_slice_exact(&DirentHeader {
+            record_len,
+            next_opaque_id: entry.next_opaque_id,
+            inode: entry.inode,
+            kind: entry.kind as u8,
+        })?;
+        this_header_extra.zero_out()?;
+        this_name.copy_from_slice_exact(entry.name.as_bytes())?;
+        this_name_nul.zero_out()?;
+
+        self.written += usize::from(record_len);
+        self.buffer = remaining;
+
+        Ok(())
+    }
+    pub fn finalize(self) -> usize {
+        self.written
+    }
+}

src/error.rs

@@ -0,0 +1,327 @@
+use core::{fmt, result};
+
+#[derive(Clone, Copy, Eq, PartialEq)]
+pub struct Error {
+    pub errno: i32,
+}
+
+pub type Result<T, E = Error> = result::Result<T, E>;
+
+impl Error {
+    pub fn new(errno: i32) -> Error {
+        Error { errno }
+    }
+
+    pub fn mux(result: Result<usize>) -> usize {
+        match result {
+            Ok(value) => value,
+            Err(error) => -error.errno as usize,
+        }
+    }
+
+    pub fn demux(value: usize) -> Result<usize> {
+        let errno = -(value as i32);
+        if errno >= 1 && errno < STR_ERROR.len() as i32 {
+            Err(Error::new(errno))
+        } else {
+            Ok(value)
+        }
+    }
+
+    pub fn text(&self) -> &'static str {
+        STR_ERROR
+            .get(self.errno as usize)
+            .map(|&x| x)
+            .unwrap_or("Unknown Error")
+    }
+}
+
+impl fmt::Debug for Error {
+    fn fmt(&self, f: &mut fmt::Formatter) -> result::Result<(), fmt::Error> {
+        f.write_str(self.text())
+    }
+}
+
+impl fmt::Display for Error {
+    fn fmt(&self, f: &mut fmt::Formatter) -> result::Result<(), fmt::Error> {
+        f.write_str(self.text())
+    }
+}
+#[cfg(feature = "std")]
+impl std::error::Error for Error {}
+
+#[cfg(feature = "std")]
+impl From<Error> for std::io::Error {
+    fn from(value: Error) -> Self {
+        std::io::Error::from_raw_os_error(value.errno)
+    }
+}
+
+pub const EPERM: i32 = 1; /* Operation not permitted */
+pub const ENOENT: i32 = 2; /* No such file or directory */
+pub const ESRCH: i32 = 3; /* No such process */
+pub const EINTR: i32 = 4; /* Interrupted system call */
+pub const EIO: i32 = 5; /* I/O error */
+pub const ENXIO: i32 = 6; /* No such device or address */
+pub const E2BIG: i32 = 7; /* Argument list too long */
+pub const ENOEXEC: i32 = 8; /* Exec format error */
+pub const EBADF: i32 = 9; /* Bad file number */
+pub const ECHILD: i32 = 10; /* No child processes */
+pub const EAGAIN: i32 = 11; /* Try again */
+pub const ENOMEM: i32 = 12; /* Out of memory */
+pub const EACCES: i32 = 13; /* Permission denied */
+pub const EFAULT: i32 = 14; /* Bad address */
+pub const ENOTBLK: i32 = 15; /* Block device required */
+pub const EBUSY: i32 = 16; /* Device or resource busy */
+pub const EEXIST: i32 = 17; /* File exists */
+pub const EXDEV: i32 = 18; /* Cross-device link */
+pub const ENODEV: i32 = 19; /* No such device */
+pub const ENOTDIR: i32 = 20; /* Not a directory */
+pub const EISDIR: i32 = 21; /* Is a directory */
+pub const EINVAL: i32 = 22; /* Invalid argument */
+pub const ENFILE: i32 = 23; /* File table overflow */
+pub const EMFILE: i32 = 24; /* Too many open files */
+pub const ENOTTY: i32 = 25; /* Not a typewriter */
+pub const ETXTBSY: i32 = 26; /* Text file busy */
+pub const EFBIG: i32 = 27; /* File too large */
+pub const ENOSPC: i32 = 28; /* No space left on device */
+pub const ESPIPE: i32 = 29; /* Illegal seek */
+pub const EROFS: i32 = 30; /* Read-only file system */
+pub const EMLINK: i32 = 31; /* Too many links */
+pub const EPIPE: i32 = 32; /* Broken pipe */
+pub const EDOM: i32 = 33; /* Math argument out of domain of func */
+pub const ERANGE: i32 = 34; /* Math result not representable */
+pub const EDEADLK: i32 = 35; /* Resource deadlock would occur */
+pub const ENAMETOOLONG: i32 = 36; /* File name too long */
+pub const ENOLCK: i32 = 37; /* No record locks available */
+pub const ENOSYS: i32 = 38; /* Function not implemented */
+pub const ENOTEMPTY: i32 = 39; /* Directory not empty */
+pub const ELOOP: i32 = 40; /* Too many symbolic links encountered */
+pub const EWOULDBLOCK: i32 = 41; /* Operation would block */
+pub const ENOMSG: i32 = 42; /* No message of desired type */
+pub const EIDRM: i32 = 43; /* Identifier removed */
+pub const ECHRNG: i32 = 44; /* Channel number out of range */
+pub const EL2NSYNC: i32 = 45; /* Level 2 not synchronized */
+pub const EL3HLT: i32 = 46; /* Level 3 halted */
+pub const EL3RST: i32 = 47; /* Level 3 reset */
+pub const ELNRNG: i32 = 48; /* Link number out of range */
+pub const EUNATCH: i32 = 49; /* Protocol driver not attached */
+pub const ENOCSI: i32 = 50; /* No CSI structure available */
+pub const EL2HLT: i32 = 51; /* Level 2 halted */
+pub const EBADE: i32 = 52; /* Invalid exchange */
+pub const EBADR: i32 = 53; /* Invalid request descriptor */
+pub const EXFULL: i32 = 54; /* Exchange full */
+pub const ENOANO: i32 = 55; /* No anode */
+pub const EBADRQC: i32 = 56; /* Invalid request code */
+pub const EBADSLT: i32 = 57; /* Invalid slot */
+pub const EDEADLOCK: i32 = 58; /* Resource deadlock would occur */
+pub const EBFONT: i32 = 59; /* Bad font file format */
+pub const ENOSTR: i32 = 60; /* Device not a stream */
+pub const ENODATA: i32 = 61; /* No data available */
+pub const ETIME: i32 = 62; /* Timer expired */
+pub const ENOSR: i32 = 63; /* Out of streams resources */
+pub const ENONET: i32 = 64; /* Machine is not on the network */
+pub const ENOPKG: i32 = 65; /* Package not installed */
+pub const EREMOTE: i32 = 66; /* Object is remote */
+pub const ENOLINK: i32 = 67; /* Link has been severed */
+pub const EADV: i32 = 68; /* Advertise error */
+pub const ESRMNT: i32 = 69; /* Srmount error */
+pub const ECOMM: i32 = 70; /* Communication error on send */
+pub const EPROTO: i32 = 71; /* Protocol error */
+pub const EMULTIHOP: i32 = 72; /* Multihop attempted */
+pub const EDOTDOT: i32 = 73; /* RFS specific error */
+pub const EBADMSG: i32 = 74; /* Not a data message */
+pub const EOVERFLOW: i32 = 75; /* Value too large for defined data type */
+pub const ENOTUNIQ: i32 = 76; /* Name not unique on network */
+pub const EBADFD: i32 = 77; /* File descriptor in bad state */
+pub const EREMCHG: i32 = 78; /* Remote address changed */
+pub const ELIBACC: i32 = 79; /* Can not access a needed shared library */
+pub const ELIBBAD: i32 = 80; /* Accessing a corrupted shared library */
+pub const ELIBSCN: i32 = 81; /* .lib section in a.out corrupted */
+pub const ELIBMAX: i32 = 82; /* Attempting to link in too many shared libraries */
+pub const ELIBEXEC: i32 = 83; /* Cannot exec a shared library directly */
+pub const EILSEQ: i32 = 84; /* Illegal byte sequence */
+pub const ERESTART: i32 = 85; /* Interrupted system call should be restarted */
+pub const ESTRPIPE: i32 = 86; /* Streams pipe error */
+pub const EUSERS: i32 = 87; /* Too many users */
+pub const ENOTSOCK: i32 = 88; /* Socket operation on non-socket */
+pub const EDESTADDRREQ: i32 = 89; /* Destination address required */
+pub const EMSGSIZE: i32 = 90; /* Message too long */
+pub const EPROTOTYPE: i32 = 91; /* Protocol wrong type for socket */
+pub const ENOPROTOOPT: i32 = 92; /* Protocol not available */
+pub const EPROTONOSUPPORT: i32 = 93; /* Protocol not supported */
+pub const ESOCKTNOSUPPORT: i32 = 94; /* Socket type not supported */
+pub const EOPNOTSUPP: i32 = 95; /* Operation not supported on transport endpoint */
+pub const EPFNOSUPPORT: i32 = 96; /* Protocol family not supported */
+pub const EAFNOSUPPORT: i32 = 97; /* Address family not supported by protocol */
+pub const EADDRINUSE: i32 = 98; /* Address already in use */
+pub const EADDRNOTAVAIL: i32 = 99; /* Cannot assign requested address */
+pub const ENETDOWN: i32 = 100; /* Network is down */
+pub const ENETUNREACH: i32 = 101; /* Network is unreachable */
+pub const ENETRESET: i32 = 102; /* Network dropped connection because of reset */
+pub const ECONNABORTED: i32 = 103; /* Software caused connection abort */
+pub const ECONNRESET: i32 = 104; /* Connection reset by peer */
+pub const ENOBUFS: i32 = 105; /* No buffer space available */
+pub const EISCONN: i32 = 106; /* Transport endpoint is already connected */
+pub const ENOTCONN: i32 = 107; /* Transport endpoint is not connected */
+pub const ESHUTDOWN: i32 = 108; /* Cannot send after transport endpoint shutdown */
+pub const ETOOMANYREFS: i32 = 109; /* Too many references: cannot splice */
+pub const ETIMEDOUT: i32 = 110; /* Connection timed out */
+pub const ECONNREFUSED: i32 = 111; /* Connection refused */
+pub const EHOSTDOWN: i32 = 112; /* Host is down */
+pub const EHOSTUNREACH: i32 = 113; /* No route to host */
+pub const EALREADY: i32 = 114; /* Operation already in progress */
+pub const EINPROGRESS: i32 = 115; /* Operation now in progress */
+pub const ESTALE: i32 = 116; /* Stale NFS file handle */
+pub const EUCLEAN: i32 = 117; /* Structure needs cleaning */
+pub const ENOTNAM: i32 = 118; /* Not a XENIX named type file */
+pub const ENAVAIL: i32 = 119; /* No XENIX semaphores available */
+pub const EISNAM: i32 = 120; /* Is a named type file */
+pub const EREMOTEIO: i32 = 121; /* Remote I/O error */
+pub const EDQUOT: i32 = 122; /* Quota exceeded */
+pub const ENOMEDIUM: i32 = 123; /* No medium found */
+pub const EMEDIUMTYPE: i32 = 124; /* Wrong medium type */
+pub const ECANCELED: i32 = 125; /* Operation Canceled */
+pub const ENOKEY: i32 = 126; /* Required key not available */
+pub const EKEYEXPIRED: i32 = 127; /* Key has expired */
+pub const EKEYREVOKED: i32 = 128; /* Key has been revoked */
+pub const EKEYREJECTED: i32 = 129; /* Key was rejected by service */
+pub const EOWNERDEAD: i32 = 130; /* Owner died */
+pub const ENOTRECOVERABLE: i32 = 131; /* State not recoverable */
+pub const ERSVD: i32 = 132; /* Reserved (formerly "scheme-kernel message code") */
+
+pub static STR_ERROR: [&'static str; 133] = [
+    "Success",
+    "Operation not permitted",
+    "No such file or directory",
+    "No such process",
+    "Interrupted system call",
+    "I/O error",
+    "No such device or address",
+    "Argument list too long",
+    "Exec format error",
+    "Bad file number",
+    "No child processes",
+    "Try again",
+    "Out of memory",
+    "Permission denied",
+    "Bad address",
+    "Block device required",
+    "Device or resource busy",
+    "File exists",
+    "Cross-device link",
+    "No such device",
+    "Not a directory",
+    "Is a directory",
+    "Invalid argument",
+    "File table overflow",
+    "Too many open files",
+    "Not a typewriter",
+    "Text file busy",
+    "File too large",
+    "No space left on device",
+    "Illegal seek",
+    "Read-only file system",
+    "Too many links",
+    "Broken pipe",
+    "Math argument out of domain of func",
+    "Math result not representable",
+    "Resource deadlock would occur",
+    "File name too long",
+    "No record locks available",
+    "Function not implemented",
+    "Directory not empty",
+    "Too many symbolic links encountered",
+    "Operation would block",
+    "No message of desired type",
+    "Identifier removed",
+    "Channel number out of range",
+    "Level 2 not synchronized",
+    "Level 3 halted",
+    "Level 3 reset",
+    "Link number out of range",
+    "Protocol driver not attached",
+    "No CSI structure available",
+    "Level 2 halted",
+    "Invalid exchange",
+    "Invalid request descriptor",
+    "Exchange full",
+    "No anode",
+    "Invalid request code",
+    "Invalid slot",
+    "Resource deadlock would occur",
+    "Bad font file format",
+    "Device not a stream",
+    "No data available",
+    "Timer expired",
+    "Out of streams resources",
+    "Machine is not on the network",
+    "Package not installed",
+    "Object is remote",
+    "Link has been severed",
+    "Advertise error",
+    "Srmount error",
+    "Communication error on send",
+    "Protocol error",
+    "Multihop attempted",
+    "RFS specific error",
+    "Not a data message",
+    "Value too large for defined data type",
+    "Name not unique on network",
+    "File descriptor in bad state",
+    "Remote address changed",
+    "Can not access a needed shared library",
+    "Accessing a corrupted shared library",
+    ".lib section in a.out corrupted",
+    "Attempting to link in too many shared libraries",
+    "Cannot exec a shared library directly",
+    "Illegal byte sequence",
+    "Interrupted system call should be restarted",
+    "Streams pipe error",
+    "Too many users",
+    "Socket operation on non-socket",
+    "Destination address required",
+    "Message too long",
+    "Protocol wrong type for socket",
+    "Protocol not available",
+    "Protocol not supported",
+    "Socket type not supported",
+    "Operation not supported on transport endpoint",
+    "Protocol family not supported",
+    "Address family not supported by protocol",
+    "Address already in use",
+    "Cannot assign requested address",
+    "Network is down",
+    "Network is unreachable",
+    "Network dropped connection because of reset",
+    "Software caused connection abort",
+    "Connection reset by peer",
+    "No buffer space available",
+    "Transport endpoint is already connected",
+    "Transport endpoint is not connected",
+    "Cannot send after transport endpoint shutdown",
+    "Too many references: cannot splice",
+    "Connection timed out",
+    "Connection refused",
+    "Host is down",
+    "No route to host",
+    "Operation already in progress",
+    "Operation now in progress",
+    "Stale NFS file handle",
+    "Structure needs cleaning",
+    "Not a XENIX named type file",
+    "No XENIX semaphores available",
+    "Is a named type file",
+    "Remote I/O error",
+    "Quota exceeded",
+    "No medium found",
+    "Wrong medium type",
+    "Operation Canceled",
+    "Required key not available",
+    "Key has expired",
+    "Key has been revoked",
+    "Key was rejected by service",
+    "Owner died",
+    "State not recoverable",
+    "Reserved (formerly scheme-kernel message code)",
+];

src/flag.rs

@@ -0,0 +1,567 @@
+use bitflags::bitflags as inner_bitflags;
+use core::{mem, ops::Deref, slice};
+
+macro_rules! bitflags {
+    (
+        $(#[$outer:meta])*
+        pub struct $BitFlags:ident: $T:ty {
+            $(
+                $(#[$inner:ident $($args:tt)*])*
+                const $Flag:ident = $value:expr;
+            )+
+        }
+    ) => {
+        // First, use the inner bitflags
+        inner_bitflags! {
+            #[derive(PartialEq, Eq, PartialOrd, Ord, Hash, Debug, Clone, Copy, Default)]
+            $(#[$outer])*
+            pub struct $BitFlags: $T {
+                $(
+                    $(#[$inner $($args)*])*
+                    const $Flag = $value;
+                )+
+            }
+        }
+
+        impl $BitFlags {
+            #[deprecated = "use the safe `from_bits_retain` method instead"]
+            pub unsafe fn from_bits_unchecked(bits: $T) -> Self {
+                Self::from_bits_retain(bits)
+            }
+        }
+
+        // Secondly, re-export all inner constants
+        // (`pub use self::Struct::*` doesn't work)
+        $(
+            $(#[$inner $($args)*])*
+            pub const $Flag: $BitFlags = $BitFlags::$Flag;
+        )+
+    }
+}
+
+pub const CLOCK_REALTIME: usize = 1;
+pub const CLOCK_MONOTONIC: usize = 4;
+
+bitflags! {
+    pub struct EventFlags: usize {
+        const EVENT_NONE = 0;
+        const EVENT_READ = 1;
+        const EVENT_WRITE = 2;
+    }
+}
+
+pub const F_DUPFD: usize = 0;
+pub const F_GETFD: usize = 1;
+pub const F_SETFD: usize = 2;
+pub const F_GETFL: usize = 3;
+pub const F_SETFL: usize = 4;
+pub const F_DUPFD_CLOEXEC: usize = 1030;
+
+pub const FUTEX_WAIT: usize = 0;
+pub const FUTEX_WAKE: usize = 1;
+pub const FUTEX_REQUEUE: usize = 2;
+pub const FUTEX_WAIT64: usize = 3;
+
+// TODO: Split SendFdFlags into caller flags and flags that the scheme receives?
+bitflags::bitflags! {
+    #[derive(Clone, Copy, Debug)]
+    pub struct SendFdFlags: usize {
+        /// If set, the kernel will enforce that the file descriptors are exclusively owned.
+        ///
+        /// That is, there will no longer exist any other reference to those FDs when removed from
+        /// the file table (sendfd always removes the FDs from the file table, but without this
+        /// flag, it can be retained by SYS_DUPing them first).
+        const EXCLUSIVE = 1;
+
+        /// If set, the file descriptors will be cloned and *not* removed from the sender's file table.
+        /// By default, `SYS_SENDFD` moves the file descriptors, removing them from the sender.
+        const CLONE = 2;
+    }
+}
+bitflags::bitflags! {
+    #[derive(Clone, Copy, Debug)]
+    pub struct FobtainFdFlags: usize {
+        /// If set, the SYS_CALL payload specifies the destination file descriptor slots, otherwise the lowest
+        /// available slots will be selected, and placed in the usize pointed to by SYS_CALL
+        /// payload.
+        const MANUAL_FD = 1;
+
+        /// If set, the file descriptors received are guaranteed to be exclusively owned (by the file
+        /// table the obtainer is running in).
+        const EXCLUSIVE = 2;
+
+        /// If set, the file descriptors received will be placed into the *upper* file table.
+        const UPPER_TBL = 4;
+
+        /// If set, the received file descriptors are marked as close-on-exec.
+        const CLOEXEC = 8;
+
+        // No, cloexec won't be stored in the kernel in the future, when the stable ABI is moved to
+        // relibc, so no flag for that!
+    }
+}
+bitflags::bitflags! {
+    #[derive(Clone, Copy, Debug)]
+    pub struct RecvFdFlags: usize {
+        /// If set, the SYS_CALL payload specifies the destination file descriptor slots, otherwise the lowest
+        /// available slots will be selected, and placed in the usize pointed to by SYS_CALL
+        /// payload.
+        const MANUAL_FD = 1;
+
+        /// If set, the file descriptors received will be placed into the *upper* file table.
+        const UPPER_TBL = 2;
+
+        /// If set, the received file descriptors are marked as close-on-exec.
+        const CLOEXEC = 4;
+    }
+}
+bitflags::bitflags! {
+    #[derive(Clone, Copy, Debug)]
+    pub struct FmoveFdFlags: usize {
+        /// If set, the kernel will enforce that the file descriptors are exclusively owned.
+        ///
+        /// That is, there will no longer exist any other reference to those FDs when removed from
+        /// the file table (SYS_CALL always removes the FDs from the file table, but without this
+        /// flag, it can be retained by SYS_DUPing them first).
+        const EXCLUSIVE = 1;
+
+        /// If set, the file descriptors will be cloned and *not* removed from the sender's file table.
+        /// By default, sendfd moves the file descriptors, removing them from the sender.
+        const CLONE = 2;
+    }
+}
+
+bitflags! {
+    pub struct MapFlags: usize {
+        // TODO: Downgrade PROT_NONE to global constant? (bitflags specifically states zero flags
+        // can cause buggy behavior).
+        const PROT_NONE = 0x0000_0000;
+
+        const PROT_EXEC = 0x0001_0000;
+        const PROT_WRITE = 0x0002_0000;
+        const PROT_READ = 0x0004_0000;
+
+        const MAP_SHARED = 0x0001;
+        const MAP_PRIVATE = 0x0002;
+
+        const MAP_FIXED = 0x0004;
+        const MAP_FIXED_NOREPLACE = 0x000C;
+
+        /// For *userspace-backed mmaps*, return from the mmap call before all pages have been
+        /// provided by the scheme. This requires the scheme to be trusted, as the current context
+        /// can block indefinitely, if the scheme does not respond to the page fault handler's
+        /// request, as it tries to map the page by requesting it from the scheme.
+        ///
+        /// In some cases however, such as the program loader, the data needs to be trusted as much
+        /// with or without MAP_LAZY, and if so, mapping lazily will not cause insecureness by
+        /// itself.
+        ///
+        /// For kernel-backed mmaps, this flag has no effect at all. It is unspecified whether
+        /// kernel mmaps are lazy or not.
+        const MAP_LAZY = 0x0010;
+    }
+}
+bitflags! {
+    pub struct MunmapFlags: usize {
+        /// Indicates whether the funmap call must implicitly do an msync, for the changes to
+        /// become visible later.
+        ///
+        /// This flag will currently be set if and only if MAP_SHARED | PROT_WRITE are set.
+        const NEEDS_SYNC = 1;
+    }
+}
+
+pub const MODE_TYPE: u16 = 0xF000;
+pub const MODE_DIR: u16 = 0x4000;
+pub const MODE_FILE: u16 = 0x8000;
+pub const MODE_SYMLINK: u16 = 0xA000;
+pub const MODE_FIFO: u16 = 0x1000;
+pub const MODE_CHR: u16 = 0x2000;
+pub const MODE_SOCK: u16 = 0xC000;
+
+pub const MODE_PERM: u16 = 0x0FFF;
+pub const MODE_SETUID: u16 = 0o4000;
+pub const MODE_SETGID: u16 = 0o2000;
+
+pub const O_RDONLY: usize = 0x0001_0000;
+pub const O_WRONLY: usize = 0x0002_0000;
+pub const O_RDWR: usize = 0x0003_0000;
+pub const O_NONBLOCK: usize = 0x0004_0000;
+pub const O_APPEND: usize = 0x0008_0000;
+pub const O_SHLOCK: usize = 0x0010_0000;
+pub const O_EXLOCK: usize = 0x0020_0000;
+pub const O_ASYNC: usize = 0x0040_0000;
+pub const O_FSYNC: usize = 0x0080_0000;
+pub const O_CLOEXEC: usize = 0x0100_0000;
+pub const O_CREAT: usize = 0x0200_0000;
+pub const O_TRUNC: usize = 0x0400_0000;
+pub const O_EXCL: usize = 0x0800_0000;
+pub const O_DIRECTORY: usize = 0x1000_0000;
+pub const O_STAT: usize = 0x2000_0000;
+pub const O_SYMLINK: usize = 0x4000_0000;
+pub const O_NOFOLLOW: usize = 0x8000_0000;
+pub const O_ACCMODE: usize = O_RDONLY | O_WRONLY | O_RDWR;
+pub const O_FCNTL_MASK: usize = O_NONBLOCK | O_APPEND | O_ASYNC | O_FSYNC;
+
+/// Remove directory instead of unlinking file.
+pub const AT_REMOVEDIR: usize = 0x200;
+
+// The top 48 bits of PTRACE_* are reserved, for now
+
+// NOT ABI STABLE!
+#[derive(Clone, Copy, Debug, Default, Eq, PartialEq)]
+#[repr(usize)]
+pub enum ContextStatus {
+    Runnable,
+    Blocked,
+    NotYetStarted,
+    Dead,
+    ForceKilled,
+    Stopped,
+    UnhandledExcp,
+    #[default]
+    Other, // reserved
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[repr(usize)]
+pub enum ContextVerb {
+    Stop = 1,
+    Unstop = 2,
+    Interrupt = 3,
+    ForceKill = usize::MAX,
+}
+impl ContextVerb {
+    pub fn try_from_raw(raw: usize) -> Option<Self> {
+        Some(match raw {
+            1 => Self::Stop,
+            2 => Self::Unstop,
+            3 => Self::Interrupt,
+            usize::MAX => Self::ForceKill,
+            _ => return None,
+        })
+    }
+}
+
+// NOT ABI STABLE!
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[repr(u8)]
+pub enum AddrSpaceVerb {
+    MmapMin = 255,
+}
+impl AddrSpaceVerb {
+    pub fn try_from_raw(verb: u8) -> Option<Self> {
+        Some(match verb {
+            255 => Self::MmapMin,
+
+            _ => return None,
+        })
+    }
+}
+
+// NOT ABI STABLE!
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[repr(u8)]
+pub enum ProcSchemeVerb {
+    RegsInt = 250,
+    RegsFloat = 251,
+    RegsEnv = 252,
+    SchedAffinity = 253,
+    Start = 254,
+    Iopl = 255,
+}
+impl ProcSchemeVerb {
+    pub fn try_from_raw(verb: u8) -> Option<Self> {
+        Some(match verb {
+            250 => Self::RegsInt,
+            251 => Self::RegsFloat,
+            252 => Self::RegsEnv,
+            253 => Self::SchedAffinity,
+            254 => Self::Start,
+            255 => Self::Iopl,
+            _ => return None,
+        })
+    }
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub enum FileTableVerb {
+    Close = 1,
+    Dup2 = 2,
+    CloseCloExec = 3,
+}
+impl FileTableVerb {
+    pub fn try_from_raw(value: u8) -> Option<Self> {
+        Some(match value {
+            1 => Self::Close,
+            2 => Self::Dup2,
+            3 => Self::CloseCloExec,
+            _ => return None,
+        })
+    }
+}
+
+// NOT ABI-STABLE!
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[repr(u64)]
+pub enum AcpiVerb {
+    // copies the rsdt/xsdt to the payload buffer (the number of bytes that fit), and returns the
+    // rsdt/xsdt length regardless
+    ReadRxsdt = 1,
+    // no payload, just returns 0 or 1
+    CheckShutdown = 2,
+    /// Red Bear OS extension (Phase I): acpid requests the kernel
+    /// enter s2idle (Modern Standby / S0ix). The kernel sets
+    /// `S2IDLE_REQUESTED`; the idle path calls `mwait_loop()`. Read
+    /// payload (1 byte) returns the *previous* value of the flag.
+    /// Write payload is opaque (ignored by current kernel).
+    /// Mirrors Linux 7.1 `s2idle_enter()` in
+    /// `kernel/power/suspend.c:91`. Hardware-agnostic — works on
+    /// any platform with Modern Standby firmware (Dell, HP, Lenovo,
+    /// LG Gram, etc.), not just LG Gram.
+    EnterS2Idle = 3,
+    /// Red Bear OS extension (Phase I): acpid signals s2idle
+    /// exit. Kernel clears `S2IDLE_REQUESTED`. Read payload (1
+    /// byte) always returns 0. Mirrors Linux 7.1 `s2idle_wake()` in
+    /// `kernel/power/suspend.c:133`. Hardware-agnostic.
+    ExitS2Idle = 4,
+    /// Red Bear OS extension (Phase II.X.W): acpid writes the
+    /// 64-bit kernel S3 resume trampoline address to
+    /// FACS.xfirmware_waking_vector. The kernel's
+    /// `arch/x86_shared/s3_resume.rs` trampoline is at this
+    /// address; the platform firmware jumps to it on S3 wake.
+    /// Write payload: 8-byte little-endian u64 (the trampoline
+    /// address). Read payload: 0 (no error condition; this verb
+    /// is one-way). Mirrors Linux 7.1's
+    /// `acpi_set_firmware_waking_vector` in ACPICA.
+    /// Hardware-agnostic: works on any x86_64 system with
+    /// standard ACPI S3 support (Dell, HP, Lenovo, LG Gram 14).
+    SetS3WakingVector = 5,
+    /// Red Bear OS extension (Phase II.X.W): acpid requests
+    /// the kernel to enter S3. The kernel's kstop handler
+    /// dispatches on the "s3" string arg, dispatches on the
+    /// SLP_TYP byte, and does the PM1 register write. The
+    /// acpid has already done the AML prep (`_TTS(3)`, `_PTS(3)`,
+    /// `_SST(3)`) and written the trampoline address to FACS
+    /// via `SetS3WakingVector`. No payload needed.
+    /// Mirrors Linux 7.1's `enter_sleep_state` /
+    /// `acpi_hw_legacy_sleep` for S3.
+    EnterS3 = 6,
+}
+impl AcpiVerb {
+    pub const fn try_from_raw(value: u64) -> Option<Self> {
+        Some(match value {
+            1 => Self::ReadRxsdt,
+            2 => Self::CheckShutdown,
+            3 => Self::EnterS2Idle,
+            4 => Self::ExitS2Idle,
+            5 => Self::SetS3WakingVector,
+            6 => Self::EnterS3,
+            _ => return None,
+        })
+    }
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[repr(usize)]
+pub enum SchemeSocketCall {
+    ObtainFd = 0,
+    MoveFd = 1,
+}
+impl SchemeSocketCall {
+    pub fn try_from_raw(raw: usize) -> Option<Self> {
+        Some(match raw {
+            0 => Self::ObtainFd,
+            1 => Self::MoveFd,
+            _ => return None,
+        })
+    }
+}
+
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+#[repr(usize)]
+#[non_exhaustive]
+pub enum FsCall {
+    Connect = 0,
+}
+impl FsCall {
+    pub fn try_from_raw(raw: usize) -> Option<Self> {
+        Some(match raw {
+            0 => Self::Connect,
+            _ => return None,
+        })
+    }
+}
+
+bitflags! {
+    pub struct PtraceFlags: u64 {
+        /// Stop before a syscall is handled. Send PTRACE_FLAG_IGNORE to not
+        /// handle the syscall.
+        const PTRACE_STOP_PRE_SYSCALL = 0x0000_0000_0000_0001;
+        /// Stop after a syscall is handled.
+        const PTRACE_STOP_POST_SYSCALL = 0x0000_0000_0000_0002;
+        /// Stop after exactly one instruction. TODO: This may not handle
+        /// fexec/signal boundaries. Should it?
+        const PTRACE_STOP_SINGLESTEP = 0x0000_0000_0000_0004;
+        /// Stop before a signal is handled. Send PTRACE_FLAG_IGNORE to not
+        /// handle signal.
+        const PTRACE_STOP_SIGNAL = 0x0000_0000_0000_0008;
+        /// Stop on a software breakpoint, such as the int3 instruction for
+        /// x86_64.
+        const PTRACE_STOP_BREAKPOINT = 0x0000_0000_0000_0010;
+        /// Stop just before exiting for good.
+        const PTRACE_STOP_EXIT = 0x0000_0000_0000_0020;
+
+        const PTRACE_STOP_MASK = 0x0000_0000_0000_00FF;
+
+
+        /// Sent when a child is cloned, giving you the opportunity to trace it.
+        /// If you don't catch this, the child is started as normal.
+        const PTRACE_EVENT_CLONE = 0x0000_0000_0000_0100;
+
+        /// Sent when current-addrspace is changed, allowing the tracer to reopen the memory file.
+        const PTRACE_EVENT_ADDRSPACE_SWITCH = 0x0000_0000_0000_0200;
+
+        const PTRACE_EVENT_MASK = 0x0000_0000_0000_0F00;
+
+        /// Special meaning, depending on the event. Usually, when fired before
+        /// an action, it will skip performing that action.
+        const PTRACE_FLAG_IGNORE = 0x0000_0000_0000_1000;
+
+        const PTRACE_FLAG_MASK = 0x0000_0000_0000_F000;
+    }
+}
+impl Deref for PtraceFlags {
+    type Target = [u8];
+    fn deref(&self) -> &Self::Target {
+        // Same as to_ne_bytes but in-place
+        unsafe {
+            slice::from_raw_parts(&self.bits() as *const _ as *const u8, mem::size_of::<u64>())
+        }
+    }
+}
+
+pub const SEEK_SET: usize = 0;
+pub const SEEK_CUR: usize = 1;
+pub const SEEK_END: usize = 2;
+
+pub const SIGCHLD: usize = 17;
+pub const SIGTSTP: usize = 20;
+pub const SIGTTIN: usize = 21;
+pub const SIGTTOU: usize = 22;
+
+pub const ADDRSPACE_OP_MMAP: usize = 0;
+pub const ADDRSPACE_OP_MUNMAP: usize = 1;
+pub const ADDRSPACE_OP_MPROTECT: usize = 2;
+pub const ADDRSPACE_OP_TRANSFER: usize = 3;
+
+bitflags! {
+    pub struct MremapFlags: usize {
+        const FIXED = 1;
+        const FIXED_REPLACE = 3;
+        /// Alias's memory region at `old_address` to `new_address` such that both regions share
+        /// the same frames.
+        const KEEP_OLD = 1 << 2;
+        // TODO: MAYMOVE, DONTUNMAP
+    }
+}
+bitflags! {
+    pub struct RwFlags: u32 {
+        const NONBLOCK = 1;
+        const APPEND = 2;
+        // TODO: sync/dsync
+        // TODO: O_DIRECT?
+    }
+}
+bitflags! {
+    pub struct SigcontrolFlags: usize {
+        /// Prevents the kernel from jumping the context to the signal trampoline, but otherwise
+        /// has absolutely no effect on which signals are blocked etc. Meant to be used for
+        /// short-lived critical sections inside libc.
+        const INHIBIT_DELIVERY = 1;
+    }
+}
+bitflags! {
+    pub struct CallFlags: usize {
+        // reserved
+        const RSVD0 = 1 << 0;
+        const RSVD1 = 1 << 1;
+        const RSVD2 = 1 << 2;
+        const RSVD3 = 1 << 3;
+        const RSVD4 = 1 << 4;
+        const RSVD5 = 1 << 5;
+        const RSVD6 = 1 << 6;
+        const RSVD7 = 1 << 7;
+
+        /// Remove the fd from the caller's file table before sending the message.
+        const CONSUME = 1 << 8;
+
+        const WRITE = 1 << 9;
+        const READ = 1 << 10;
+
+        /// Indicates the request is a bulk fd passing request.
+        const FD = 1 << 11;
+        /// Flags for the fd passing request.
+        const FD_EXCLUSIVE = 1 << 12;
+        const FD_CLONE = 1 << 13;
+        const FD_UPPER = 1 << 14;
+        const FD_CLOEXEC = 1 << 15;
+
+        /// Call is a standard fs call, with metadata defined in `StdFsCallMeta`
+        const STD_FS = 1 << 16;
+
+        /// Call is taking multiple fds as an argument
+        const MULTIPLE_FDS = 1 << 17;
+    }
+}
+
+#[repr(u8)]
+#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
+pub enum StdFsCallKind {
+    // TODO: remove old syscalls
+    Fchmod = 1,
+    Fchown = 2,
+    Getdents = 3,
+    Fstat = 4,
+    Fstatvfs = 5,
+    Fsync = 6,
+    Ftruncate = 7,
+    Futimens = 8,
+    // 9 reserved in fscall RFC
+    // Unlinkat = 10,
+    Relpathat = 11,
+    Lock = 12,
+    Unlock = 13,
+    GetLock = 14,
+}
+
+impl StdFsCallKind {
+    pub fn try_from_raw(raw: u8) -> Option<Self> {
+        use StdFsCallKind::*;
+
+        // TODO: Use a library where this match can be automated.
+        Some(match raw {
+            1 => Fchmod,
+            2 => Fchown,
+            3 => Getdents,
+            4 => Fstat,
+            5 => Fstatvfs,
+            6 => Fsync,
+            7 => Ftruncate,
+            8 => Futimens,
+            // 9 reserved in fscall RFC
+            // 10 => Unlinkat,
+            11 => Relpathat,
+            12 => Lock,
+            13 => Unlock,
+            14 => GetLock,
+            _ => return None,
+        })
+    }
+}
+
+/// The tag for the fd number in the upper file descriptor table.
+pub const UPPER_FDTBL_TAG: usize = 1 << (usize::BITS - 2);
+
+/// The identifier for registering event timeout
+pub const EVENT_TIMEOUT_ID: usize = usize::MAX - 2;

src/io/io.rs

@@ -0,0 +1,73 @@
+use core::{
+    cmp::PartialEq,
+    ops::{BitAnd, BitOr, Not},
+};
+
+pub trait Io {
+    type Value: Copy
+        + PartialEq
+        + BitAnd<Output = Self::Value>
+        + BitOr<Output = Self::Value>
+        + Not<Output = Self::Value>;
+
+    fn read(&self) -> Self::Value;
+    fn write(&mut self, value: Self::Value);
+
+    #[inline(always)]
+    fn readf(&self, flags: Self::Value) -> bool {
+        (self.read() & flags) as Self::Value == flags
+    }
+
+    #[inline(always)]
+    fn writef(&mut self, flags: Self::Value, value: bool) {
+        let tmp: Self::Value = match value {
+            true => self.read() | flags,
+            false => self.read() & !flags,
+        };
+        self.write(tmp);
+    }
+}
+
+pub struct ReadOnly<I> {
+    inner: I,
+}
+
+impl<I> ReadOnly<I> {
+    pub const fn new(inner: I) -> ReadOnly<I> {
+        ReadOnly { inner: inner }
+    }
+}
+
+impl<I: Io> ReadOnly<I> {
+    #[inline(always)]
+    pub fn read(&self) -> I::Value {
+        self.inner.read()
+    }
+
+    #[inline(always)]
+    pub fn readf(&self, flags: I::Value) -> bool {
+        self.inner.readf(flags)
+    }
+}
+
+pub struct WriteOnly<I> {
+    inner: I,
+}
+
+impl<I> WriteOnly<I> {
+    pub const fn new(inner: I) -> WriteOnly<I> {
+        WriteOnly { inner: inner }
+    }
+}
+
+impl<I: Io> WriteOnly<I> {
+    #[inline(always)]
+    pub fn write(&mut self, value: I::Value) {
+        self.inner.write(value)
+    }
+
+    #[inline(always)]
+    pub fn writef(&mut self, flags: I::Value, value: bool) {
+        self.inner.writef(flags, value)
+    }
+}

src/io/mmio.rs

@@ -0,0 +1,165 @@
+#[cfg(not(any(target_arch = "x86", target_arch = "x86_64")))]
+use core::ops::{BitAnd, BitOr, Not};
+use core::{mem::MaybeUninit, ptr};
+
+use super::io::Io;
+
+#[repr(transparent)]
+pub struct Mmio<T> {
+    value: MaybeUninit<T>,
+}
+
+impl<T> Mmio<T> {
+    pub unsafe fn zeroed() -> Self {
+        Self {
+            value: MaybeUninit::zeroed(),
+        }
+    }
+    pub unsafe fn uninit() -> Self {
+        Self {
+            value: MaybeUninit::uninit(),
+        }
+    }
+    pub const fn from(value: T) -> Self {
+        Self {
+            value: MaybeUninit::new(value),
+        }
+    }
+}
+
+// Generic implementation (WARNING: requires aligned pointers!)
+#[cfg(not(any(target_arch = "x86", target_arch = "x86_64")))]
+impl<T> Io for Mmio<T>
+where
+    T: Copy + PartialEq + BitAnd<Output = T> + BitOr<Output = T> + Not<Output = T>,
+{
+    type Value = T;
+
+    fn read(&self) -> T {
+        unsafe { ptr::read_volatile(ptr::addr_of!(self.value).cast::<T>()) }
+    }
+
+    fn write(&mut self, value: T) {
+        unsafe { ptr::write_volatile(ptr::addr_of_mut!(self.value).cast::<T>(), value) };
+    }
+}
+
+// x86 u8 implementation
+#[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+impl Io for Mmio<u8> {
+    type Value = u8;
+
+    fn read(&self) -> Self::Value {
+        unsafe {
+            let value: Self::Value;
+            let ptr: *const Self::Value = ptr::addr_of!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov {}, [{}]",
+                out(reg_byte) value,
+                in(reg) ptr
+            );
+            value
+        }
+    }
+
+    fn write(&mut self, value: Self::Value) {
+        unsafe {
+            let ptr: *mut Self::Value = ptr::addr_of_mut!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov [{}], {}",
+                in(reg) ptr,
+                in(reg_byte) value,
+            );
+        }
+    }
+}
+
+// x86 u16 implementation
+#[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+impl Io for Mmio<u16> {
+    type Value = u16;
+
+    fn read(&self) -> Self::Value {
+        unsafe {
+            let value: Self::Value;
+            let ptr: *const Self::Value = ptr::addr_of!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov {:x}, [{}]",
+                out(reg) value,
+                in(reg) ptr
+            );
+            value
+        }
+    }
+
+    fn write(&mut self, value: Self::Value) {
+        unsafe {
+            let ptr: *mut Self::Value = ptr::addr_of_mut!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov [{}], {:x}",
+                in(reg) ptr,
+                in(reg) value,
+            );
+        }
+    }
+}
+
+// x86 u32 implementation
+#[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+impl Io for Mmio<u32> {
+    type Value = u32;
+
+    fn read(&self) -> Self::Value {
+        unsafe {
+            let value: Self::Value;
+            let ptr: *const Self::Value = ptr::addr_of!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov {:e}, [{}]",
+                out(reg) value,
+                in(reg) ptr
+            );
+            value
+        }
+    }
+
+    fn write(&mut self, value: Self::Value) {
+        unsafe {
+            let ptr: *mut Self::Value = ptr::addr_of_mut!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov [{}], {:e}",
+                in(reg) ptr,
+                in(reg) value,
+            );
+        }
+    }
+}
+
+// x86 u64 implementation (x86_64 only)
+#[cfg(target_arch = "x86_64")]
+impl Io for Mmio<u64> {
+    type Value = u64;
+
+    fn read(&self) -> Self::Value {
+        unsafe {
+            let value: Self::Value;
+            let ptr: *const Self::Value = ptr::addr_of!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov {:r}, [{}]",
+                out(reg) value,
+                in(reg) ptr
+            );
+            value
+        }
+    }
+
+    fn write(&mut self, value: Self::Value) {
+        unsafe {
+            let ptr: *mut Self::Value = ptr::addr_of_mut!(self.value).cast::<Self::Value>();
+            core::arch::asm!(
+                "mov [{}], {:r}",
+                in(reg) ptr,
+                in(reg) value,
+            );
+        }
+    }
+}

src/io/mod.rs

@@ -0,0 +1,12 @@
+//! I/O functions
+
+pub use self::{io::*, mmio::*};
+
+#[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+pub use self::pio::*;
+
+mod io;
+mod mmio;
+
+#[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+mod pio;

src/io/pio.rs

@@ -0,0 +1,89 @@
+use core::{arch::asm, marker::PhantomData};
+
+use super::io::Io;
+
+/// Generic PIO
+#[derive(Copy, Clone)]
+pub struct Pio<T> {
+    port: u16,
+    value: PhantomData<T>,
+}
+
+impl<T> Pio<T> {
+    /// Create a PIO from a given port
+    pub const fn new(port: u16) -> Self {
+        Pio::<T> {
+            port,
+            value: PhantomData,
+        }
+    }
+}
+
+/// Read/Write for byte PIO
+impl Io for Pio<u8> {
+    type Value = u8;
+
+    /// Read
+    #[inline(always)]
+    fn read(&self) -> u8 {
+        let value: u8;
+        unsafe {
+            asm!("in al, dx", in("dx") self.port, out("al") value, options(nostack, nomem, preserves_flags));
+        }
+        value
+    }
+
+    /// Write
+    #[inline(always)]
+    fn write(&mut self, value: u8) {
+        unsafe {
+            asm!("out dx, al", in("dx") self.port, in("al") value, options(nostack, nomem, preserves_flags));
+        }
+    }
+}
+
+/// Read/Write for word PIO
+impl Io for Pio<u16> {
+    type Value = u16;
+
+    /// Read
+    #[inline(always)]
+    fn read(&self) -> u16 {
+        let value: u16;
+        unsafe {
+            asm!("in ax, dx", in("dx") self.port, out("ax") value, options(nostack, nomem, preserves_flags));
+        }
+        value
+    }
+
+    /// Write
+    #[inline(always)]
+    fn write(&mut self, value: u16) {
+        unsafe {
+            asm!("out dx, ax", in("dx") self.port, in("ax") value, options(nostack, nomem, preserves_flags));
+        }
+    }
+}
+
+/// Read/Write for doubleword PIO
+impl Io for Pio<u32> {
+    type Value = u32;
+
+    /// Read
+    #[inline(always)]
+    fn read(&self) -> u32 {
+        let value: u32;
+        unsafe {
+            asm!("in eax, dx", in("dx") self.port, out("eax") value, options(nostack, nomem, preserves_flags));
+        }
+        value
+    }
+
+    /// Write
+    #[inline(always)]
+    fn write(&mut self, value: u32) {
+        unsafe {
+            asm!("out dx, eax", in("dx") self.port, in("eax") value, options(nostack, nomem, preserves_flags));
+        }
+    }
+}

src/number.rs

@@ -0,0 +1,58 @@
+pub const SYS_CLASS: usize = 0xF000_0000;
+pub const SYS_CLASS_PATH: usize = 0x1000_0000;
+pub const SYS_CLASS_FILE: usize = 0x2000_0000;
+
+pub const SYS_ARG: usize = 0x0F00_0000;
+pub const SYS_ARG_SLICE: usize = 0x0100_0000;
+pub const SYS_ARG_MSLICE: usize = 0x0200_0000;
+pub const SYS_ARG_PATH: usize = 0x0300_0000;
+
+pub const SYS_RET: usize = 0x00F0_0000;
+pub const SYS_RET_FILE: usize = 0x0010_0000;
+
+pub const SYS_OPENAT: usize = SYS_CLASS_PATH | SYS_RET_FILE | 7;
+pub const SYS_OPENAT_WITH_FILTER: usize = SYS_CLASS_PATH | SYS_RET_FILE | 985;
+pub const SYS_UNLINKAT: usize = SYS_CLASS_PATH | 263;
+pub const SYS_UNLINKAT_WITH_FILTER: usize = SYS_CLASS_PATH | 986;
+
+pub const SYS_CLOSE: usize = SYS_CLASS_FILE | 6;
+pub const SYS_DUP: usize = SYS_CLASS_FILE | SYS_RET_FILE | 41;
+pub const SYS_DUP2: usize = SYS_CLASS_FILE | SYS_RET_FILE | 63;
+pub const SYS_READ: usize = SYS_CLASS_FILE | SYS_ARG_MSLICE | 3;
+pub const SYS_READ2: usize = SYS_CLASS_FILE | SYS_ARG_MSLICE | 35;
+pub const SYS_WRITE: usize = SYS_CLASS_FILE | SYS_ARG_SLICE | 4;
+pub const SYS_WRITE2: usize = SYS_CLASS_FILE | SYS_ARG_SLICE | 45;
+pub const SYS_LSEEK: usize = SYS_CLASS_FILE | 19;
+pub const SYS_FCHMOD: usize = SYS_CLASS_FILE | 94;
+pub const SYS_FCHOWN: usize = SYS_CLASS_FILE | 207;
+pub const SYS_FCNTL: usize = SYS_CLASS_FILE | 55;
+pub const SYS_FEVENT: usize = SYS_CLASS_FILE | 927;
+
+// SYS_CALL, fd, inout buf ptr, inout buf len, flags, metadata buf ptr, metadata buf len
+// TODO: new number for SYS_CALL where flags are sent as 6th argument (using syscall6)
+pub const SYS_CALL: usize = SYS_CLASS_FILE | SYS_ARG_SLICE | SYS_ARG_MSLICE | 0xCA11;
+
+pub const SYS_SENDFD: usize = SYS_CLASS_FILE | 34;
+pub const SYS_GETDENTS: usize = SYS_CLASS_FILE | 43;
+
+// TODO: Rename FMAP/FUNMAP to MMAP/MUNMAP
+pub const SYS_FMAP: usize = SYS_CLASS_FILE | SYS_ARG_SLICE | 900;
+// TODO: SYS_FUNMAP should be SYS_CLASS_FILE
+pub const SYS_FUNMAP: usize = SYS_CLASS_FILE | 92;
+pub const SYS_MREMAP: usize = 155;
+
+pub const SYS_FLINK: usize = SYS_CLASS_FILE | SYS_ARG_PATH | 9;
+pub const SYS_FPATH: usize = SYS_CLASS_FILE | SYS_ARG_MSLICE | 928;
+pub const SYS_FRENAME: usize = SYS_CLASS_FILE | SYS_ARG_PATH | 38;
+pub const SYS_FSTAT: usize = SYS_CLASS_FILE | SYS_ARG_MSLICE | 28;
+pub const SYS_FSTATVFS: usize = SYS_CLASS_FILE | SYS_ARG_MSLICE | 100;
+pub const SYS_FSYNC: usize = SYS_CLASS_FILE | 118;
+pub const SYS_FTRUNCATE: usize = SYS_CLASS_FILE | 93;
+pub const SYS_FUTIMENS: usize = SYS_CLASS_FILE | SYS_ARG_SLICE | 320;
+
+pub const SYS_CLOCK_GETTIME: usize = 265;
+pub const SYS_FUTEX: usize = 240;
+pub const SYS_MPROTECT: usize = 125;
+pub const SYS_MKNS: usize = 984;
+pub const SYS_NANOSLEEP: usize = 162;
+pub const SYS_YIELD: usize = 158;

src/schemev2.rs

@@ -0,0 +1,213 @@
+use core::{
+    mem,
+    ops::{Deref, DerefMut},
+    slice,
+};
+
+use bitflags::bitflags;
+
+pub struct CallerCtx {
+    pub pid: usize,
+    pub uid: u32,
+    pub gid: u32,
+}
+
+pub enum OpenResult {
+    ThisScheme { number: usize },
+    OtherScheme { fd: usize },
+}
+
+#[repr(C)]
+#[derive(Clone, Copy, Debug, Default)]
+pub struct Sqe {
+    pub opcode: u8,
+    pub sqe_flags: SqeFlags,
+    pub _rsvd: u16, // TODO: priority
+    pub tag: u32,
+    pub args: [u64; 6],
+    pub caller: u64,
+}
+impl Deref for Sqe {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Sqe as *const u8, mem::size_of::<Sqe>()) }
+    }
+}
+
+impl DerefMut for Sqe {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Sqe as *mut u8, mem::size_of::<Sqe>()) }
+    }
+}
+
+bitflags! {
+    #[derive(Clone, Copy, Debug, Default)]
+    pub struct SqeFlags: u8 {
+        // If zero, the message is bidirectional, and the scheme is expected to pass the Ksmsg's
+        // tag field to the Skmsg. Some opcodes require this flag to be set.
+        const ONEWAY = 1;
+
+        // If this flag is set, index 0 of Sqe's args stores the IDs buffer address,
+        // and index 1 stores the IDs buffer length.
+        const MULTIPLE_IDS = 1 << 1;
+    }
+}
+
+#[repr(C)]
+#[derive(Clone, Copy, Debug, Default)]
+pub struct Cqe {
+    pub flags: u8, // bits 2:0 are CqeOpcode
+    pub extra_raw: [u8; 3],
+    pub tag: u32,
+    pub result: u64,
+}
+impl Deref for Cqe {
+    type Target = [u8];
+    fn deref(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self as *const Cqe as *const u8, mem::size_of::<Cqe>()) }
+    }
+}
+
+impl DerefMut for Cqe {
+    fn deref_mut(&mut self) -> &mut [u8] {
+        unsafe { slice::from_raw_parts_mut(self as *mut Cqe as *mut u8, mem::size_of::<Cqe>()) }
+    }
+}
+
+bitflags! {
+    #[derive(Clone, Copy, Debug, Default, Eq, PartialEq)]
+    pub struct NewFdFlags: u8 {
+        const POSITIONED = 1;
+    }
+}
+
+impl Cqe {
+    pub fn extra(&self) -> u32 {
+        u32::from_ne_bytes([self.extra_raw[0], self.extra_raw[1], self.extra_raw[2], 0])
+    }
+}
+
+#[repr(u8)]
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub enum CqeOpcode {
+    RespondRegular,
+    RespondWithFd,
+    SendFevent, // no tag
+    ObtainFd,
+    RespondWithMultipleFds,
+    /// [`SchemeAsync::on_close`] and [`SchemeSync::on_close`] are only called when the last file
+    /// descriptor referring to the file description is closed. To implement traditional POSIX
+    /// advisory file locking, [`CqeOpcode::RespondAndNotifyOnDetach`] is used to notify the scheme
+    /// by sending a [`RequestKind::OnDetach`] request the next time the file description is
+    /// "detached" from a file descriptor. Not done by default to avoid unnecessary IPC.
+    RespondAndNotifyOnDetach,
+    // TODO: ProvideMmap
+}
+
+impl CqeOpcode {
+    pub fn try_from_raw(raw: u8) -> Option<Self> {
+        // TODO: Use a library where this match can be automated.
+        Some(match raw {
+            0 => Self::RespondRegular,
+            1 => Self::RespondWithFd,
+            2 => Self::SendFevent,
+            3 => Self::ObtainFd,
+            4 => Self::RespondWithMultipleFds,
+            5 => Self::RespondAndNotifyOnDetach,
+            _ => return None,
+        })
+    }
+}
+
+/// SqeOpcode
+#[repr(u8)]
+#[non_exhaustive]
+#[derive(Clone, Copy, Debug)]
+pub enum Opcode {
+    Close = 3,   // fd
+    Dup = 4,     // old fd, buf_ptr, buf_len
+    Read = 5,    // fd, buf_ptr, buf_len, TODO offset, TODO flags, _
+    Write = 6,   // fd, buf_ptr, buf_len, TODO offset, TODO flags)
+    Fsize = 7,   // fd
+    Fchmod = 8,  // fd, new mode
+    Fchown = 9,  // fd, new uid, new gid
+    Fcntl = 10,  // fd, cmd, arg
+    Fevent = 11, // fd, requested mask
+    Sendfd = 12,
+    Fpath = 13, // fd, buf_ptr, buf_len
+    Frename = 14,
+    Fstat = 15,     // fd, buf_ptr, buf_len
+    Fstatvfs = 16,  // fd, buf_ptr, buf_len
+    Fsync = 17,     // fd
+    Ftruncate = 18, // fd, new len
+    Futimens = 19,  // fd, times_buf, times_len
+
+    MmapPrep = 20,
+    RequestMmap = 21,
+    Mremap = 22,
+    Munmap = 23,
+    Msync = 24, // TODO
+
+    Cancel = 25, // @tag
+
+    Getdents = 26,
+    CloseMsg = 27,
+    Call = 28,
+
+    OpenAt = 29, // fd, buf_ptr, buf_len, flags
+    Flink = 30,
+    Recvfd = 31,
+
+    UnlinkAt = 32, // fd, path_ptr, path_len (utf8), flags
+    StdFsCall = 33,
+
+    Detach = 34,
+}
+
+impl Opcode {
+    pub fn try_from_raw(raw: u8) -> Option<Self> {
+        use Opcode::*;
+
+        // TODO: Use a library where this match can be automated.
+        Some(match raw {
+            3 => Close,
+            4 => Dup,
+            5 => Read,
+            6 => Write,
+            7 => Fsize,
+            8 => Fchmod,
+            9 => Fchown,
+            10 => Fcntl,
+            11 => Fevent,
+            12 => Sendfd,
+            13 => Fpath,
+            14 => Frename,
+            15 => Fstat,
+            16 => Fstatvfs,
+            17 => Fsync,
+            18 => Ftruncate,
+            19 => Futimens,
+
+            20 => MmapPrep,
+            21 => RequestMmap,
+            22 => Mremap,
+            23 => Munmap,
+            24 => Msync,
+
+            25 => Cancel,
+            26 => Getdents,
+            27 => CloseMsg,
+            28 => Call,
+
+            29 => OpenAt,
+            30 => Flink,
+            31 => Recvfd,
+
+            32 => UnlinkAt,
+            33 => StdFsCall,
+            34 => Detach,
+
+            _ => return None,
+        })
+    }
+}

src/sigabi.rs

@@ -0,0 +1,340 @@
+use core::sync::atomic::{AtomicUsize, Ordering};
+
+/// Signal runtime struct for the entire process
+#[derive(Debug)]
+#[repr(C, align(4096))]
+pub struct SigProcControl {
+    pub pending: AtomicU64,
+    pub actions: [RawAction; 64],
+    pub sender_infos: [AtomicU64; 32],
+    //pub queue: [RealtimeSig; 32], TODO
+    // qhead, qtail TODO
+}
+/*#[derive(Debug)]
+#[repr(transparent)]
+pub struct RealtimeSig {
+    pub arg: NonatomicUsize,
+}*/
+#[derive(Debug, Default)]
+#[repr(C, align(16))]
+pub struct RawAction {
+    /// Only two MSBs are interesting for the kernel. If bit 63 is set, signal is ignored. If bit
+    /// 62 is set and the signal is SIGTSTP/SIGTTIN/SIGTTOU, it's equivalent to the action of
+    /// Stop.
+    pub first: AtomicU64,
+    /// Completely ignored by the kernel, but exists so userspace can (when 16-byte atomics exist)
+    /// atomically set both the handler, sigaction flags, and sigaction mask.
+    pub user_data: AtomicU64,
+}
+
+/// Signal runtime struct for a thread
+#[derive(Debug, Default)]
+#[repr(C)]
+pub struct Sigcontrol {
+    // composed of [lo "pending" | lo "unmasked", hi "pending" | hi "unmasked"]
+    pub word: [AtomicU64; 2],
+
+    // lo = sender pid, hi = sender ruid
+    pub sender_infos: [AtomicU64; 32],
+
+    pub control_flags: SigatomicUsize,
+
+    pub saved_ip: NonatomicUsize,          // rip/eip/pc
+    pub saved_archdep_reg: NonatomicUsize, // rflags(x64)/eflags(x86)/x0(aarch64)/t0(riscv64)
+}
+#[derive(Clone, Copy, Debug)]
+pub struct SenderInfo {
+    pub pid: u32,
+    pub ruid: u32,
+}
+impl SenderInfo {
+    #[inline]
+    pub fn raw(self) -> u64 {
+        u64::from(self.pid) | (u64::from(self.ruid) << 32)
+    }
+    #[inline]
+    pub const fn from_raw(raw: u64) -> Self {
+        Self {
+            pid: raw as u32,
+            ruid: (raw >> 32) as u32,
+        }
+    }
+}
+
+impl Sigcontrol {
+    pub fn currently_pending_unblocked(&self, proc: &SigProcControl) -> u64 {
+        let proc_pending = proc.pending.load(Ordering::Relaxed);
+        let [w0, w1] = core::array::from_fn(|i| {
+            let w = self.word[i].load(Ordering::Relaxed);
+            ((w | (proc_pending >> (i * 32))) & 0xffff_ffff) & (w >> 32)
+        });
+        //core::sync::atomic::fence(Ordering::Acquire);
+        w0 | (w1 << 32)
+    }
+    pub fn set_allowset(&self, new_allowset: u64) -> u64 {
+        //core::sync::atomic::fence(Ordering::Release);
+        let [w0, w1] = self.word.each_ref().map(|w| w.load(Ordering::Relaxed));
+        let old_a0 = w0 & 0xffff_ffff_0000_0000;
+        let old_a1 = w1 & 0xffff_ffff_0000_0000;
+        let new_a0 = (new_allowset & 0xffff_ffff) << 32;
+        let new_a1 = new_allowset & 0xffff_ffff_0000_0000;
+
+        let prev_w0 = self.word[0].fetch_add(new_a0.wrapping_sub(old_a0), Ordering::Relaxed);
+        let prev_w1 = self.word[0].fetch_add(new_a1.wrapping_sub(old_a1), Ordering::Relaxed);
+        //core::sync::atomic::fence(Ordering::Acquire);
+        let up0 = prev_w0 & (prev_w0 >> 32);
+        let up1 = prev_w1 & (prev_w1 >> 32);
+
+        up0 | (up1 << 32)
+    }
+}
+
+#[derive(Debug, Default)]
+#[repr(transparent)]
+pub struct SigatomicUsize(AtomicUsize);
+
+impl SigatomicUsize {
+    #[inline]
+    pub fn load(&self, ordering: Ordering) -> usize {
+        let value = self.0.load(Ordering::Relaxed);
+        if ordering != Ordering::Relaxed {
+            core::sync::atomic::compiler_fence(ordering);
+        }
+        value
+    }
+    #[inline]
+    pub fn store(&self, value: usize, ordering: Ordering) {
+        if ordering != Ordering::Relaxed {
+            core::sync::atomic::compiler_fence(ordering);
+        }
+        self.0.store(value, Ordering::Relaxed);
+    }
+}
+#[derive(Debug, Default)]
+#[repr(transparent)]
+pub struct NonatomicUsize(AtomicUsize);
+
+impl NonatomicUsize {
+    #[inline]
+    pub const fn new(a: usize) -> Self {
+        Self(AtomicUsize::new(a))
+    }
+
+    #[inline]
+    pub fn get(&self) -> usize {
+        self.0.load(Ordering::Relaxed)
+    }
+    #[inline]
+    pub fn set(&self, value: usize) {
+        self.0.store(value, Ordering::Relaxed);
+    }
+}
+
+pub fn sig_bit(sig: usize) -> u64 {
+    1 << (sig - 1)
+}
+
+// TODO: Move to redox_rt?
+impl SigProcControl {
+    /// Checks if `sig` should be ignored based on the current action flags.
+    ///
+    /// * `sig` - The signal to check (e.g. `SIGCHLD`).
+    ///
+    /// * `stop_or_continue` - Whether the signal is generated because a child
+    /// process stopped (`SIGSTOP`, `SIGTSTP`) or continued (`SIGCONT`). If
+    /// `true` and `sig` is `SIGCHLD`, the signal shall not be delivered if the
+    /// `SA_NOCLDSTOP` flag is set for `SIGCHLD`.
+    pub fn signal_will_ign(&self, sig: usize, stop_or_continue: bool) -> bool {
+        let flags = self.actions[sig - 1].first.load(Ordering::Relaxed);
+        let will_ign = flags & (1 << 63) != 0;
+        let sig_specific = flags & (1 << 62) != 0; // SA_NOCLDSTOP if sig == SIGCHLD
+
+        will_ign || (sig == SIGCHLD && stop_or_continue && sig_specific)
+    }
+    // TODO: Move to redox_rt?
+    pub fn signal_will_stop(&self, sig: usize) -> bool {
+        use crate::flag::*;
+        matches!(sig, SIGTSTP | SIGTTIN | SIGTTOU)
+            && self.actions[sig - 1].first.load(Ordering::Relaxed) & (1 << 62) != 0
+    }
+}
+
+#[cfg(not(target_arch = "x86"))]
+pub use core::sync::atomic::AtomicU64;
+
+use crate::SIGCHLD;
+
+#[cfg(target_arch = "x86")]
+pub use self::atomic::AtomicU64;
+
+#[cfg(target_arch = "x86")]
+mod atomic {
+    use core::{cell::UnsafeCell, sync::atomic::Ordering};
+
+    #[derive(Debug, Default)]
+    pub struct AtomicU64(UnsafeCell<u64>);
+
+    unsafe impl Send for AtomicU64 {}
+    unsafe impl Sync for AtomicU64 {}
+
+    impl AtomicU64 {
+        pub const fn new(inner: u64) -> Self {
+            Self(UnsafeCell::new(inner))
+        }
+        pub fn compare_exchange(
+            &self,
+            old: u64,
+            new: u64,
+            _success: Ordering,
+            _failure: Ordering,
+        ) -> Result<u64, u64> {
+            let old_hi = (old >> 32) as u32;
+            let old_lo = old as u32;
+            let new_hi = (new >> 32) as u32;
+            let new_lo = new as u32;
+            let mut out_hi;
+            let mut out_lo;
+
+            unsafe {
+                core::arch::asm!("lock cmpxchg8b [{}]", in(reg) self.0.get(), inout("edx") old_hi => out_hi, inout("eax") old_lo => out_lo, in("ecx") new_hi, in("ebx") new_lo);
+            }
+
+            if old_hi == out_hi && old_lo == out_lo {
+                Ok(old)
+            } else {
+                Err(u64::from(out_lo) | (u64::from(out_hi) << 32))
+            }
+        }
+        pub fn load(&self, ordering: Ordering) -> u64 {
+            match self.compare_exchange(0, 0, ordering, ordering) {
+                Ok(new) => new,
+                Err(new) => new,
+            }
+        }
+        pub fn store(&self, new: u64, ordering: Ordering) {
+            let mut old = 0;
+
+            loop {
+                match self.compare_exchange(old, new, ordering, Ordering::Relaxed) {
+                    Ok(_) => break,
+                    Err(new) => {
+                        old = new;
+                        core::hint::spin_loop();
+                    }
+                }
+            }
+        }
+        pub fn fetch_update(
+            &self,
+            set_order: Ordering,
+            fetch_order: Ordering,
+            mut f: impl FnMut(u64) -> Option<u64>,
+        ) -> Result<u64, u64> {
+            let mut old = self.load(fetch_order);
+
+            loop {
+                let new = f(old).ok_or(old)?;
+                match self.compare_exchange(old, new, set_order, Ordering::Relaxed) {
+                    Ok(_) => return Ok(new),
+                    Err(changed) => {
+                        old = changed;
+                        core::hint::spin_loop();
+                    }
+                }
+            }
+        }
+        pub fn fetch_or(&self, bits: u64, order: Ordering) -> u64 {
+            self.fetch_update(order, Ordering::Relaxed, |b| Some(b | bits))
+                .unwrap()
+        }
+        pub fn fetch_and(&self, bits: u64, order: Ordering) -> u64 {
+            self.fetch_update(order, Ordering::Relaxed, |b| Some(b & bits))
+                .unwrap()
+        }
+        pub fn fetch_add(&self, term: u64, order: Ordering) -> u64 {
+            self.fetch_update(order, Ordering::Relaxed, |b| Some(b.wrapping_add(term)))
+                .unwrap()
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::{
+        atomic::{AtomicU64, Ordering},
+        Arc,
+    };
+
+    #[cfg(not(loom))]
+    use std::{sync::Mutex, thread};
+    #[cfg(not(loom))]
+    fn model(f: impl FnOnce()) {
+        f()
+    }
+
+    #[cfg(loom)]
+    use loom::{model, sync::Mutex, thread};
+
+    use crate::{RawAction, SigProcControl, Sigcontrol};
+
+    struct FakeThread {
+        ctl: Sigcontrol,
+        pctl: SigProcControl,
+        ctxt: Mutex<()>,
+    }
+    impl Default for FakeThread {
+        fn default() -> Self {
+            Self {
+                ctl: Sigcontrol::default(),
+                pctl: SigProcControl {
+                    pending: AtomicU64::new(0),
+                    actions: core::array::from_fn(|_| RawAction::default()),
+                    sender_infos: Default::default(),
+                },
+                ctxt: Default::default(),
+            }
+        }
+    }
+
+    #[test]
+    fn singlethread_mask() {
+        model(|| {
+            let fake_thread = Arc::new(FakeThread::default());
+
+            let thread = {
+                let fake_thread = Arc::clone(&fake_thread);
+
+                thread::spawn(move || {
+                    fake_thread.ctl.set_allowset(!0);
+                    {
+                        let _g = fake_thread.ctxt.lock();
+                        if fake_thread
+                            .ctl
+                            .currently_pending_unblocked(&fake_thread.pctl)
+                            == 0
+                        {
+                            drop(_g);
+                            thread::park();
+                        }
+                    }
+                })
+            };
+
+            for sig in 1..=64 {
+                let _g = fake_thread.ctxt.lock();
+
+                let idx = sig - 1;
+                let bit = 1 << (idx % 32);
+
+                fake_thread.ctl.word[idx / 32].fetch_or(bit, Ordering::Relaxed);
+                let w = fake_thread.ctl.word[idx / 32].load(Ordering::Relaxed);
+
+                if w & (w >> 32) != 0 {
+                    thread.thread().unpark();
+                }
+            }
+
+            thread.join().unwrap();
+        });
+    }
+}