wasi.rs

  1use std::{
  2    collections::HashMap, fs::File, future::Future, marker::PhantomData, path::Path, pin::Pin,
  3};
  4
  5use anyhow::{anyhow, Error};
  6use serde::{de::DeserializeOwned, Serialize};
  7
  8use wasi_common::{dir, file};
  9use wasmtime::{
 10    AsContext, AsContextMut, Caller, Config, Engine, Extern, Instance, Linker, Module, Store,
 11    StoreContext, StoreContextMut, Trap, TypedFunc,
 12};
 13use wasmtime::{IntoFunc, Memory};
 14use wasmtime_wasi::{Dir, WasiCtx, WasiCtxBuilder};
 15
 16pub struct WasiResource(u32);
 17
 18pub struct WasiFn<A: Serialize, R: DeserializeOwned> {
 19    function: TypedFunc<(u32, u32), u32>,
 20    _function_type: PhantomData<fn(A) -> R>,
 21}
 22
 23impl<A: Serialize, R: DeserializeOwned> Copy for WasiFn<A, R> {}
 24
 25impl<A: Serialize, R: DeserializeOwned> Clone for WasiFn<A, R> {
 26    fn clone(&self) -> Self {
 27        Self {
 28            function: self.function,
 29            _function_type: PhantomData,
 30        }
 31    }
 32}
 33
 34// impl<A: Serialize, R: DeserializeOwned> WasiFn<A, R> {
 35//     #[inline(always)]
 36//     pub async fn call(&self, runtime: &mut Wasi, arg: A) -> Result<R, Error> {
 37//         runtime.call(self, arg).await
 38//     }
 39// }
 40
 41// type signature derived from:
 42// https://docs.rs/wasmtime/latest/wasmtime/struct.Linker.html#method.func_wrap2_async
 43// macro_rules! dynHostFunction {
 44//     () => {
 45//         Box<
 46//             dyn for<'a> Fn(Caller<'a, WasiCtx>, u32, u32)
 47//                 -> Box<dyn Future<Output = u32> + Send + 'a>
 48//                     + Send
 49//                     + Sync
 50//                     + 'static
 51//         >
 52//     };
 53// }
 54
 55// macro_rules! implHostFunction {
 56//     () => {
 57//         impl for<'a> Fn(Caller<'a, WasiCtx>, u32, u32)
 58//             -> Box<dyn Future<Output = u32> + Send + 'a>
 59//                 + Send
 60//                 + Sync
 61//                 + 'static
 62//     };
 63// }
 64
 65pub struct WasiPluginBuilder {
 66    wasi_ctx: WasiCtx,
 67    engine: Engine,
 68    linker: Linker<WasiCtxAlloc>,
 69}
 70
 71impl WasiPluginBuilder {
 72    pub fn new(wasi_ctx: WasiCtx) -> Result<Self, Error> {
 73        let mut config = Config::default();
 74        config.async_support(true);
 75        let engine = Engine::new(&config)?;
 76        let mut linker = Linker::new(&engine);
 77
 78        Ok(WasiPluginBuilder {
 79            // host_functions: HashMap::new(),
 80            wasi_ctx,
 81            engine,
 82            linker,
 83        })
 84    }
 85
 86    pub fn new_with_default_ctx() -> Result<Self, Error> {
 87        let wasi_ctx = WasiCtxBuilder::new()
 88            .inherit_stdin()
 89            .inherit_stderr()
 90            .build();
 91        Self::new(wasi_ctx)
 92    }
 93
 94    pub fn host_function<A: DeserializeOwned, R: Serialize>(
 95        mut self,
 96        name: &str,
 97        function: impl Fn(A) -> R + Send + Sync + 'static,
 98    ) -> Result<Self, Error> {
 99        self.linker.func_wrap(
100            "env",
101            name,
102            move |mut caller: Caller<'_, WasiCtxAlloc>, ptr: u32, len: u32| {
103                let mut plugin_memory = match caller.get_export("memory") {
104                    Some(Extern::Memory(mem)) => mem,
105                    _ => return Err(Trap::new("Could not grab slice of plugin memory")),
106                };
107                let args = Wasi::deserialize_from_buffer(&mut plugin_memory, &caller, ptr, len)?;
108
109                let result = function(args);
110                let buffer = Wasi::serialize_to_buffer(
111                    caller.data().alloc_buffer(),
112                    &mut plugin_memory,
113                    &mut caller,
114                    result,
115                )
116                .await;
117
118                Ok(7u32)
119            },
120        )?;
121        Ok(self)
122    }
123
124    pub async fn init<T: AsRef<[u8]>>(self, module: T) -> Result<Wasi, Error> {
125        Wasi::init(module.as_ref().to_vec(), self).await
126    }
127}
128
129// // TODO: remove
130// /// Represents a to-be-initialized plugin.
131// /// Please use [`WasiPluginBuilder`], don't use this directly.
132// pub struct WasiPlugin {
133//     pub module: Vec<u8>,
134//     pub wasi_ctx: WasiCtx,
135//     pub host_functions:
136//         HashMap<String, Box<dyn Fn(&str, &mut Linker<WasiCtx>) -> Result<(), Error>>>,
137// }
138
139#[derive(Copy, Clone)]
140struct WasiAlloc {
141    alloc_buffer: TypedFunc<u32, u32>,
142    free_buffer: TypedFunc<u32, u32>,
143}
144
145struct WasiCtxAlloc {
146    wasi_ctx: WasiCtx,
147    alloc: Option<WasiAlloc>,
148}
149
150impl WasiCtxAlloc {
151    fn alloc_buffer(&self) -> TypedFunc<u32, u32> {
152        self.alloc
153            .expect("allocator has been not initialized, cannot allocate buffer!")
154            .alloc_buffer
155    }
156
157    fn free_buffer(&self) -> TypedFunc<u32, u32> {
158        self.alloc
159            .expect("allocator has been not initialized, cannot free buffer!")
160            .free_buffer
161    }
162
163    fn init_alloc(&mut self, alloc: WasiAlloc) {
164        self.alloc = Some(alloc)
165    }
166}
167
168pub struct Wasi {
169    engine: Engine,
170    module: Module,
171    store: Store<WasiCtxAlloc>,
172    instance: Instance,
173}
174
175impl Wasi {
176    pub fn dump_memory(data: &[u8]) {
177        for (i, byte) in data.iter().enumerate() {
178            if i % 32 == 0 {
179                println!();
180            }
181            if i % 4 == 0 {
182                print!("|");
183            }
184            if *byte == 0 {
185                print!("__")
186            } else {
187                print!("{:02x}", byte);
188            }
189        }
190        println!();
191    }
192}
193
194impl Wasi {
195    async fn init(module: Vec<u8>, plugin: WasiPluginBuilder) -> Result<Self, Error> {
196        // initialize the WebAssembly System Interface context
197        let engine = plugin.engine;
198        let mut linker = plugin.linker;
199        wasmtime_wasi::add_to_linker(&mut linker, |s| &mut s.wasi_ctx)?;
200
201        // create a store, note that we can't initialize the allocator,
202        // because we can't grab the functions until initialized.
203        let mut store: Store<WasiCtxAlloc> = Store::new(
204            &engine,
205            WasiCtxAlloc {
206                wasi_ctx: plugin.wasi_ctx,
207                alloc: None,
208            },
209        );
210        let module = Module::new(&engine, module)?;
211
212        // load the provided module into the asynchronous runtime
213        linker.module_async(&mut store, "", &module).await?;
214        let instance = linker.instantiate_async(&mut store, &module).await?;
215
216        // now that the module is initialized,
217        // we can initialize the store's allocator
218        let alloc_buffer = instance.get_typed_func(&mut store, "__alloc_buffer")?;
219        let free_buffer = instance.get_typed_func(&mut store, "__free_buffer")?;
220        store.data_mut().init_alloc(WasiAlloc {
221            alloc_buffer,
222            free_buffer,
223        });
224
225        Ok(Wasi {
226            engine,
227            module,
228            store,
229            instance,
230        })
231    }
232
233    /// Attaches a file or directory the the given system path to the runtime.
234    /// Note that the resource must be freed by calling `remove_resource` afterwards.
235    pub fn attach_path<T: AsRef<Path>>(&mut self, path: T) -> Result<WasiResource, Error> {
236        // grab the WASI context
237        let ctx = self.store.data_mut();
238
239        // open the file we want, and convert it into the right type
240        // this is a footgun and a half
241        let file = File::open(&path).unwrap();
242        let dir = Dir::from_std_file(file);
243        let dir = Box::new(wasmtime_wasi::dir::Dir::from_cap_std(dir));
244
245        // grab an empty file descriptor, specify capabilities
246        let fd = ctx.wasi_ctx.table().push(Box::new(()))?;
247        let caps = dir::DirCaps::all();
248        let file_caps = file::FileCaps::all();
249
250        // insert the directory at the given fd,
251        // return a handle to the resource
252        ctx.wasi_ctx
253            .insert_dir(fd, dir, caps, file_caps, path.as_ref().to_path_buf());
254        Ok(WasiResource(fd))
255    }
256
257    /// Returns `true` if the resource existed and was removed.
258    pub fn remove_resource(&mut self, resource: WasiResource) -> Result<(), Error> {
259        self.store
260            .data_mut()
261            .wasi_ctx
262            .table()
263            .delete(resource.0)
264            .ok_or_else(|| anyhow!("Resource did not exist, but a valid handle was passed in"))?;
265        Ok(())
266    }
267
268    // pub fn with_resource<T>(
269    //     &mut self,
270    //     resource: WasiResource,
271    //     callback: fn(&mut Self) -> Result<T, Error>,
272    // ) -> Result<T, Error> {
273    //     let result = callback(self);
274    //     self.remove_resource(resource)?;
275    //     return result;
276    // }
277
278    // So this call function is kinda a dance, I figured it'd be a good idea to document it.
279    // the high level is we take a serde type, serialize it to a byte array,
280    // (we're doing this using bincode for now)
281    // then toss that byte array into webassembly.
282    // webassembly grabs that byte array, does some magic,
283    // and serializes the result into yet another byte array.
284    // we then grab *that* result byte array and deserialize it into a result.
285    //
286    // phew...
287    //
288    // now the problem is, webassambly doesn't support buffers.
289    // only really like i32s, that's it (yeah, it's sad. Not even unsigned!)
290    // (ok, I'm exaggerating a bit).
291    //
292    // the Wasm function that this calls must have a very specific signature:
293    //
294    // fn(pointer to byte array: i32, length of byte array: i32)
295    //     -> pointer to (
296    //            pointer to byte_array: i32,
297    //            length of byte array: i32,
298    //     ): i32
299    //
300    // This pair `(pointer to byte array, length of byte array)` is called a `Buffer`
301    // and can be found in the cargo_test plugin.
302    //
303    // so on the wasm side, we grab the two parameters to the function,
304    // stuff them into a `Buffer`,
305    // and then pray to the `unsafe` Rust gods above that a valid byte array pops out.
306    //
307    // On the flip side, when returning from a wasm function,
308    // we convert whatever serialized result we get into byte array,
309    // which we stuff into a Buffer and allocate on the heap,
310    // which pointer to we then return.
311    // Note the double indirection!
312    //
313    // So when returning from a function, we actually leak memory *twice*:
314    //
315    // 1) once when we leak the byte array
316    // 2) again when we leak the allocated `Buffer`
317    //
318    // This isn't a problem because Wasm stops executing after the function returns,
319    // so the heap is still valid for our inspection when we want to pull things out.
320
321    /// Takes an item, allocates a buffer, serializes the argument to that buffer,
322    /// and returns a (ptr, len) pair to that buffer.
323    async fn serialize_to_buffer<A: Serialize>(
324        alloc_buffer: TypedFunc<u32, u32>,
325        plugin_memory: &mut Memory,
326        mut store: impl AsContextMut<Data = WasiCtxAlloc>,
327        item: A,
328    ) -> Result<(u32, u32), Error> {
329        // serialize the argument using bincode
330        let item = bincode::serialize(&item)?;
331        let buffer_len = item.len() as u32;
332
333        // allocate a buffer and write the argument to that buffer
334        let buffer_ptr = alloc_buffer.call_async(&mut store, buffer_len).await?;
335        plugin_memory.write(&mut store, buffer_ptr as usize, &item)?;
336        Ok((buffer_ptr, buffer_len))
337    }
338
339    /// Takes `ptr to a `(ptr, len)` pair, and returns `(ptr, len)`.
340    fn deref_buffer(
341        plugin_memory: &mut Memory,
342        store: impl AsContext<Data = WasiCtxAlloc>,
343        buffer: u32,
344    ) -> Result<(u32, u32), Error> {
345        // create a buffer to read the (ptr, length) pair into
346        // this is a total of 4 + 4 = 8 bytes.
347        let raw_buffer = &mut [0; 8];
348        plugin_memory.read(store, buffer as usize, raw_buffer)?;
349
350        // use these bytes (wasm stores things little-endian)
351        // to get a pointer to the buffer and its length
352        let b = raw_buffer;
353        let buffer_ptr = u32::from_le_bytes([b[0], b[1], b[2], b[3]]);
354        let buffer_len = u32::from_le_bytes([b[4], b[5], b[6], b[7]]);
355
356        return Ok((buffer_ptr, buffer_len));
357    }
358
359    /// Takes a `(ptr, len)` pair and returns the corresponding deserialized buffer.
360    fn deserialize_from_buffer<R: DeserializeOwned>(
361        plugin_memory: &mut Memory,
362        store: impl AsContext<Data = WasiCtxAlloc>,
363        buffer_ptr: u32,
364        buffer_len: u32,
365    ) -> Result<R, Error> {
366        let buffer_ptr = buffer_ptr as usize;
367        let buffer_len = buffer_len as usize;
368        let buffer_end = buffer_ptr + buffer_len;
369
370        // read the buffer at this point into a byte array
371        // deserialize the byte array into the provided serde type
372        let result = &plugin_memory.data(store.as_context())[buffer_ptr..buffer_end];
373        let result = bincode::deserialize(result)?;
374
375        // TODO: this is handled wasm-side
376        // // deallocate the argument buffer
377        // self.free_buffer.call(&mut self.store, arg_buffer);
378
379        Ok(result)
380    }
381
382    pub fn function<A: Serialize, R: DeserializeOwned, T: AsRef<str>>(
383        &mut self,
384        name: T,
385    ) -> Result<WasiFn<A, R>, Error> {
386        let fun_name = format!("__{}", name.as_ref());
387        let fun = self
388            .instance
389            .get_typed_func::<(u32, u32), u32, _>(&mut self.store, &fun_name)?;
390        Ok(WasiFn {
391            function: fun,
392            _function_type: PhantomData,
393        })
394    }
395
396    // TODO: dont' use as for conversions
397    pub async fn call<A: Serialize, R: DeserializeOwned>(
398        &mut self,
399        handle: &WasiFn<A, R>,
400        arg: A,
401    ) -> Result<R, Error> {
402        // dbg!(&handle.name);
403        // dbg!(serde_json::to_string(&arg)).unwrap();
404
405        let mut plugin_memory = self
406            .instance
407            .get_memory(&mut self.store, "memory")
408            .ok_or_else(|| anyhow!("Could not grab slice of plugin memory"))?;
409
410        // write the argument to linear memory
411        // this returns a (ptr, lentgh) pair
412        let arg_buffer = Self::serialize_to_buffer(
413            self.store.data().alloc_buffer(),
414            &mut plugin_memory,
415            &mut self.store,
416            arg,
417        )
418        .await?;
419
420        // call the function, passing in the buffer and its length
421        // this returns a ptr to a (ptr, lentgh) pair
422        let result_buffer = handle
423            .function
424            .call_async(&mut self.store, arg_buffer)
425            .await?;
426        let (result_buffer_ptr, result_buffer_len) =
427            Self::deref_buffer(&mut plugin_memory, &mut self.store, result_buffer)?;
428
429        Self::deserialize_from_buffer(
430            &mut plugin_memory,
431            &mut self.store,
432            result_buffer_ptr,
433            result_buffer_len,
434        )
435    }
436}