jonaskruckenberg / k23

Experimental WASM Microkernel Operating System

k23

Experimental WASM Microkernel Operating System

Manual k23 Discord

About

k23 is an active research project exploring a more secure, modular, and easy to develop for operating system by using WebAssembly as the primary execution environment.

Contributing

I believe OS development should be fun, easy, and approachable. If you would like to hack on k23, fork it for your own experiments or just hang out and philosophize about computers. Be my guest! You can join our small, but growing community of likeminded, awesome people that all believe better computer system are possible!

k23's Issues

WASM: Relaxed dead code validation

This is the tracking issue for the Relaxed dead code validation proposal.

WASM: Branch Hinting

This is the tracking issue for the Branch Hinting proposal.

WASM: Multiple memories

This is the tracking issue for the Multiple memories proposal.

WASM: Relaxed SIMD

This is the tracking issue for the Relaxed SIMD proposal.

WASM: Multi-value

This is the tracking issue for the Multi-value proposal.

WASM: Numeric Values in WAT Data Segments

This is the tracking issue for the Numeric Values in WAT Data Segments proposal.

WASM: Custom Page Sizes

This is the tracking issue for the Custom Page Sizes proposal.

WASM: Extended Constant Expressions

This is the tracking issue for the Extended Constant Expressions proposal.

WASM: Compilation Hints

This is the tracking issue for the Compilation Hints proposal.

WASM: Memory64

This is the tracking issue for the Memory64 proposal.

Open Questions

• Bounds checks?
• Multiple Memories?

WASM: Exception handling

This is the tracking issue for the Exception handling proposal.

feat: allow filtering log messages

Currently all log messages are emitted regardless of level or origin. As cranelift and other dependencies produce lots of messages it would be very helpful to filter which messages are emitted.

Having a similar system to what env_logger does would be very helpful, env vars could be set at compile time or better yet passed to the kernel by the loader.

For reference the current log implementation for the kernel and loader respectively can be found here: https://github.com/JonasKruckenberg/k23/blob/main/kernel/src/logger.rs https://github.com/JonasKruckenberg/k23/blob/main/loader/src/logger.rs

PCI support

The pci_types crate is likely useful for this. It is used by Redox OS too.

WASM: Custom Annotation Syntax in the Text Format

This is the tracking issue for the Custom Annotation Syntax in the Text Format proposal.

Dependency Dashboard

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Open

These updates have all been created already. Click a checkbox below to force a retry/rebase of any.

• chore(deps): update rust crate clap to v4.5.15
• chore(deps): update rust crate syn to v2.0.73
• fix(deps): update rust crate serde to v1.0.206
• chore(deps): update rust crate wasmparser to 0.215.0
• Click on this checkbox to rebase all open PRs at once

Detected dependencies

cargo

Cargo.toml

• onlyerror 0.1.4
• log 0.4.21
• bitflags 2.6.0
• cfg-if 1.0.0
• linked_list_allocator 0.10.5
• arrayvec 0.7.4
• ed25519-dalek 2.1.1
• object 0.36.0
• lz4_flex 0.11
• lock_api 0.4.12
• unicode-width 0.1.9
• memchr 2.4.1
• bumpalo 3.14.0
• rustc-demangle 0.1.24
• postcard 1.0.8
• talc 4.4.1
• hashbrown 0.14.5
• wasmparser 0.214.0
• target-lexicon 0.12.14
• serde 1.0.203
• rand_core 0.6.4
• clap 4.5.7
• env_logger 0.11.3
• anstyle 1.0.7
• cargo_metadata 0.18.1
• proc-macro2 1
• quote 1
• syn 2
• toml 0.8.14

build/bootimg-runner/Cargo.toml

• serde 1.0.203
• toml 0.8.14

kernel/Cargo.toml

libs/dtb-parser/Cargo.toml

libs/fxhash/Cargo.toml

libs/gimli/Cargo.toml

• indexmap 2.2.6
• test-assembler 0.1.6

libs/kstd/Cargo.toml

libs/ktest/Cargo.toml

libs/ktest/macros/Cargo.toml

libs/uart-16550/Cargo.toml

libs/vmm/Cargo.toml

libs/wasm-encoder/Cargo.toml

libs/wast/Cargo.toml

libs/wat/Cargo.toml

loader/Cargo.toml

loader/api/Cargo.toml

loader/api/macros/Cargo.toml

github-actions

.github/workflows/ci.yml

• actions/checkout v4
• Swatinem/rust-cache v2
• extractions/setup-just v2
• actions/checkout v4
• Swatinem/rust-cache v2
• extractions/setup-just v2
• actions/checkout v4
• EmbarkStudios/cargo-deny-action v2
• actions/checkout v4
• Swatinem/rust-cache v2
• extractions/setup-just v2

.github/workflows/manual.yml

• actions/checkout v4
• actions/configure-pages v5
• actions/upload-pages-artifact v3
• actions/deploy-pages v4

.github/workflows/qemu.yml

• actions/upload-artifact v4

.github/workflows/spelling.yml

• actions/checkout v4

---

• Check this box to trigger a request for Renovate to run again on this repository

WASM: Extended Name Section

This is the tracking issue for the Extended Name Section proposal.

A preliminary implementation (temporarily disabled) can be found here

k23/kernel/src/runtime/codegen/module_env.rs

Lines 444 to 527 in 2844607

	// fn parse_name_section(&mut self, reader: NameSectionReader<'wasm>) -> WasmResult<()> {
	// for subsection in reader {
	// match subsection? {
	// Name::Module { name, .. } => {
	// self.result.module.debug_info.names.module_name = Some(name);
	// }
	// Name::Function(names) => {
	// for name in names {
	// let name = name?;
	//
	// // Skip this naming if it's naming a function that
	// // doesn't actually exist.
	// if (name.index as usize) >= self.result.module.functions.len() {
	// continue;
	// }
	//
	// self.result
	// .module
	// .debug_info
	// .names
	// .func_names
	// .insert(FuncIndex::from_u32(name.index), name.name);
	// }
	// }
	// Name::Local(names) => {
	// for naming in names {
	// let name = naming?;
	// let mut result = HashMap::default();
	//
	// for name in name.names {
	// let name = name?;
	//
	// // Skip this naming if it's naming a function that
	// // doesn't actually exist.
	// if (name.index as usize) >= self.result.module.functions.len() {
	// continue;
	// }
	//
	// result.insert(name.index, name.name);
	// }
	//
	// self.result
	// .module
	// .debug_info
	// .names
	// .locals_names
	// .insert(FuncIndex::from_u32(name.index), result);
	// }
	// }
	// Name::Global(names) => {
	// for name in names {
	// let name = name?;
	// self.result
	// .module
	// .debug_info
	// .names
	// .global_names
	// .insert(GlobalIndex::from_u32(name.index), name.name);
	// }
	// }
	// Name::Data(names) => {
	// for name in names {
	// let name = name?;
	// self.result
	// .module
	// .debug_info
	// .names
	// .data_names
	// .insert(DataIndex::from_u32(name.index), name.name);
	// }
	// }
	// Name::Label(_) => log::debug!("unused name subsection label"),
	// Name::Type(_) => log::debug!("unused name subsection type"),
	// Name::Table(_) => log::debug!("unused name subsection table"),
	// Name::Memory(_) => log::debug!("unused name subsection memory"),
	// Name::Element(_) => log::debug!("unused name subsection element"),
	// Name::Field(_) => log::debug!("unused name subsection field"),
	// Name::Tag(_) => log::debug!("unused name subsection tag"),
	// Name::Unknown { .. } => {}
	// }
	// }
	//
	// Ok(())
	// }

Fault on invalid write

Hey, thanks for putting this up! I just tried running it, but it seems to be getting stuck on an unhandled exception:

❯ just run-riscv64
...
riscv_cpu_do_interrupt: hart:0, async:0, cause:0000000000000007, epc:0x000000008020aa2a, tval:0x0001000380ae6000, desc=fault_store
Invalid write at addr 0x1000380AE6000, size 4, region '(null)', reason: rejected
riscv_cpu_do_interrupt: hart:0, async:0, cause:0000000000000007, epc:0x000000008020aa2a, tval:0x0001000380ae6000, desc=fault_store
Invalid write at addr 0x1000380AE6000, size 4, region '(null)', reason: rejected
...

This is with

❯ cargo --version
cargo 1.81.0-nightly (4ed7bee47 2024-06-25)

❯ qemu-system-riscv64 --version
QEMU emulator version 6.2.0 (Debian 1:6.2+dfsg-2ubuntu6.21)

Looking at the binaries, the faulting address seems to be here:

❯ riscv64-unknown-elf-objdump -d target/riscv64gc-unknown-none-elf/debug/loader

000000008020aa1c <_ZN6loader4arch7riscv649fillstack17hca16c52fb1691861E>:
    8020aa1c:   2b383337                lui     t1,0x2b383
    8020aa20:   030a                    slli    t1,t1,0x2
    8020aa22:   ace30313                addi    t1,t1,-1330 # 2b382ace <$d.168+0x2b20fb28>
    8020aa26:   405102b3                sub     t0,sp,t0
    8020aa2a:   0062a023                sw      t1,0(t0) # 20000 <$d.18+0x2c5>
    8020aa2e:   02a1                    addi    t0,t0,8
    8020aa30:   fe22ede3                bltu    t0,sp,8020aa2a <_ZN6loader4arch7riscv649fillstack17hca16c52fb1691861E+0xe>
    8020aa34:   8082                    ret
        ...

By the way, how is enabling of logging supposed to work? Just by setting the RUST_LOG env variable?

WASM: Typed Function References

This is the tracking issue for the Typed Function References proposal.

WASM: Instrument and Tracing Technology

This is the tracking issue for the Instrument and Tracing Technology proposal as well as general guest instrumentation features

The corresponding proposal is not very well fleshed out, but the need for tracing and instrumentation features that allow developers and developer tools to trace back generated machine code to source code, measure execution times etc.

WASM: Threads

This is the tracking issue for the Threads proposal.

WASM: Reference types

This is the tracking issue for the Reference types proposal.

WASM: Constant Time

This is the tracking issue for the Constant Time proposal.

WASM: Garbage collection

This is the tracking issue for the Garbage collection proposal.

WASM: Tail call

This is the tracking issue for the Tail call proposal.

refactor(vmm): Allow specifying `vmm::Mode` at runtime

Currently the chosen memory mode for the system is set at compile-time (by all structs being generic over M: vmm::Mode).

Ideally however, we would be able to chose the "right" memory mode based on the CPU feature set. This probably requires removing the generic, making vmm::Mode object-safe and passing it as a &mut dyn vmm::Mode throughout (or a static MODE variable similar to how logs logger works?).

Support for declaring a chosen memory mode is already implemented in the loader_api::LoaderConfig struct, just not utilised yet.

feat: test harness

We need a proper test harness setup.

WASM: Fixed-width SIMD

This is the tracking issue for the Fixed-width SIMD proposal.