Stream: git-wasmtime

Topic: wasmtime / issue #14581 MPK: Segfault running an `Engine`...

Wasmtime GitHub notifications bot (Oct 06 2026 at 19:08):

alexcrichton opened issue #14581:

This input:

use std::sync::mpsc;
use wasmtime::*;

const WAT: &str = r#"(module
    (memory (export "m") 1)
    (data (i32.const 0) "hello")
    (func (export "f") (i32.store (i32.const 100) (i32.const 1))))"#;

fn engine() -> Result<Engine> {
    let mut pool = PoolingAllocationConfig::default();
    pool.memory_protection_keys(Enabled::Yes);
    pool.total_memories(32);
    pool.linear_memory_keep_resident(65536);
    let mut config = Config::new();
    config.allocation_strategy(InstanceAllocationStrategy::Pooling(pool));
    Engine::new(&config)
}

fn main() -> Result<()> {
    let (tx, rx) = mpsc::channel::<Store<()>>();
    let pre = std::thread::spawn(move || {
        let store = rx.recv().unwrap();
        eprintln!("dropping store on other thread...");
        drop(store);
        eprintln!("dropped");
    });
    let engine = engine()?;
    let module = Module::new(&engine, WAT)?;
    let mut store = Store::new(&engine, ());
    let instance = Instance::new(&mut store, &module, &[])?;
    instance
        .get_typed_func::<(), ()>(&mut store, "f")?
        .call(&mut store, ())?;
    tx.send(store).unwrap();
    pre.join().unwrap();

    Ok(())
}

fails with:

$ cargo run
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.04s
     Running `target/debug/mpk-thread`
dropping store on other thread...
zsh: segmentation fault (core dumped)  cargo run

and this input:

use std::sync::mpsc;
use wasmtime::*;

const WAT: &str = r#"(module
    (memory (export "m") 1)
    (data (i32.const 0) "hello")
    (func (export "f") (i32.store (i32.const 100) (i32.const 1))))"#;

fn engine() -> Result<Engine> {
    let mut pool = PoolingAllocationConfig::default();
    pool.memory_protection_keys(Enabled::Yes);
    pool.total_memories(32);
    pool.linear_memory_keep_resident(65536);
    let mut config = Config::new();
    config.allocation_strategy(InstanceAllocationStrategy::Pooling(pool));
    Engine::new(&config)
}

fn main() -> Result<()> {
    let (tx, rx) = mpsc::channel::<Engine>();
    let worker = std::thread::spawn(move || -> Result<()> {
        let engine = rx.recv().unwrap();
        let module = Module::new(&engine, WAT)?;
        let mut store = Store::new(&engine, ());
        let instance = Instance::new(&mut store, &module, &[])?;
        let mem = instance.get_memory(&mut store, "m").unwrap();
        eprintln!("reading memory from pre-existing thread...");
        eprintln!("data = {:?}", &mem.data(&store)[..5]);
        Ok(())
    });
    tx.send(engine()?).unwrap();
    worker.join().unwrap()?;
    Ok(())
}

fails with:

$ cargo run
   Compiling mpk-thread v0.1.0 (/home/alex/code/wasmtime2/reports/001-mpk-pkru-preexisting-thread-segv)
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.39s
     Running `target/debug/mpk-thread`
reading memory from pre-existing thread...
data = [zsh: segmentation fault (core dumped)  cargo run

Wasmtime GitHub notifications bot (Oct 06 2026 at 19:09):

alexcrichton added the wasmtime:pooling-allocator label to Issue #14581.

Wasmtime GitHub notifications bot (Oct 08 2026 at 16:50):

fitzgen commented on issue #14581:

cc @jlb6740 @rahulchaphalkar

Wasmtime GitHub notifications bot (Oct 08 2026 at 16:50):

fitzgen added the bug label to Issue #14581.

Last updated: Oct 11 2026 at 04:10 UTC