wasm_interp: support WASI argc/argv, and writing to stdout/stderr

crates/wasm_interp/src/instance.rs

@@ -145,15 +145,12 @@ impl<'a, I: ImportDispatcher> Instance<'a, I> {
         self.call_export_help(module, arg_type_bytes)
     }
 
-    pub fn call_export_from_cli<'arg, A>(
+    pub fn call_export_from_cli(
         &mut self,
         module: &WasmModule<'a>,
         fn_name: &str,
-        arg_strings: A,
+        arg_strings: &'a [&'a String],
-    ) -> Result<Option<Value>, String>
+    ) -> Result<Option<Value>, String> {
-    where
-        A: IntoIterator<Item = &'arg str>,
-    {
         let arg_type_bytes = self.prepare_to_call_export(module, fn_name)?;
 
         for (value_str, type_byte) in arg_strings.into_iter().zip(arg_type_bytes.iter().copied()) {
@@ -246,7 +243,6 @@ impl<'a, I: ImportDispatcher> Instance<'a, I> {
                 // We just popped the stack frame for the entry function. Terminate the program.
                 Action::Break
             } else {
-                dbg!(return_addr, block_depth);
                 self.program_counter = return_addr as usize;
                 self.outermost_block = block_depth;
                 Action::Continue
@@ -374,9 +370,13 @@ impl<'a, I: ImportDispatcher> Instance<'a, I> {
             if let Some(return_val) = optional_return_val {
                 self.value_stack.push(return_val);
             }
+            if let Some(debug_string) = self.debug_string.as_mut() {
+                std::write!(debug_string, " {}.{}", import.module, import.name).unwrap();
+            }
         } else {
             let return_addr = self.program_counter as u32;
-            self.program_counter = module.code.function_offsets[fn_index] as usize;
+            let internal_fn_index = fn_index - n_import_fns;
+            self.program_counter = module.code.function_offsets[internal_fn_index] as usize;
 
             let return_block_depth = self.outermost_block;
             self.outermost_block = self.block_loop_addrs.len() as u32;
@@ -407,6 +407,7 @@ impl<'a, I: ImportDispatcher> Instance<'a, I> {
         }
 
         let mut action = Action::Continue;
+        let mut implicit_return = false;
 
         match op_code {
             UNREACHABLE => {
@@ -461,6 +462,7 @@ impl<'a, I: ImportDispatcher> Instance<'a, I> {
                 if self.block_loop_addrs.len() == self.outermost_block as usize {
                     // implicit RETURN at end of function
                     action = self.do_return();
+                    implicit_return = true;
                 } else {
                     self.block_loop_addrs.pop().unwrap();
                 }
@@ -1457,6 +1459,18 @@ impl<'a, I: ImportDispatcher> Instance<'a, I> {
             let base = self.call_stack.value_stack_base();
             let slice = self.value_stack.get_slice(base as usize);
             eprintln!("{:#07x} {:17} {:?}", file_offset, debug_string, slice);
+            if op_code == RETURN || (op_code == END && implicit_return) {
+                let next_code_section_index = module
+                    .code
+                    .function_offsets
+                    .iter()
+                    .position(|o| *o as usize > self.program_counter)
+                    .unwrap_or(module.code.function_offsets.len());
+                let fn_index = module.import.imports.len() + next_code_section_index - 1;
+                eprintln!("returning to function {}\n", fn_index);
+            } else if op_code == CALL || op_code == CALLINDIRECT {
+                eprintln!("");
+            }
         }
 
         action

crates/wasm_interp/src/lib.rs

@@ -9,6 +9,7 @@ pub use instance::Instance;
 
 use roc_wasm_module::Value;
 use value_stack::ValueStack;
+use wasi::WasiDispatcher;
 
 pub trait ImportDispatcher {
     /// Dispatch a call from WebAssembly to your own code, based on module and function name.
@@ -21,11 +22,23 @@ pub trait ImportDispatcher {
     ) -> Option<Value>;
 }
 
-pub const DEFAULT_IMPORTS: DefaultImportDispatcher = DefaultImportDispatcher {};
+pub const DEFAULT_IMPORTS: DefaultImportDispatcher = DefaultImportDispatcher {
+    wasi: WasiDispatcher { args: &[] },
+};
 
-pub struct DefaultImportDispatcher {}
+pub struct DefaultImportDispatcher<'a> {
+    wasi: WasiDispatcher<'a>,
+}
 
-impl ImportDispatcher for DefaultImportDispatcher {
+impl<'a> DefaultImportDispatcher<'a> {
+    pub fn new(args: &'a [&'a String]) -> Self {
+        DefaultImportDispatcher {
+            wasi: WasiDispatcher { args },
+        }
+    }
+}
+
+impl<'a> ImportDispatcher for DefaultImportDispatcher<'a> {
     fn dispatch(
         &mut self,
         module_name: &str,
@@ -34,7 +47,7 @@ impl ImportDispatcher for DefaultImportDispatcher {
         memory: &mut [u8],
     ) -> Option<Value> {
         if module_name == wasi::MODULE_NAME {
-            wasi::dispatch(function_name, arguments, memory)
+            self.wasi.dispatch(function_name, arguments, memory)
         } else {
             panic!(
                 "DefaultImportDispatcher does not implement {}.{}",

crates/wasm_interp/src/main.rs

@@ -1,4 +1,4 @@
-use bumpalo::Bump;
+use bumpalo::{collections::Vec, Bump};
 use clap::ArgAction;
 use clap::{Arg, Command};
 use std::ffi::OsString;
@@ -6,7 +6,7 @@ use std::fs;
 use std::io;
 use std::process;
 
-use roc_wasm_interp::{Instance, DEFAULT_IMPORTS};
+use roc_wasm_interp::{DefaultImportDispatcher, Instance};
 use roc_wasm_module::WasmModule;
 
 pub const FLAG_FUNCTION: &str = "function";
@@ -61,10 +61,7 @@ fn main() -> io::Result<()> {
     let start_fn_name = matches.get_one::<String>(FLAG_FUNCTION).unwrap();
     let is_debug_mode = matches.get_flag(FLAG_DEBUG);
     let is_hex_format = matches.get_flag(FLAG_HEX);
-    let start_arg_strings = matches
+    let start_arg_strings = matches.get_many::<String>(ARGS_FOR_APP).unwrap_or_default();
-        .get_many::<String>(ARGS_FOR_APP)
-        .unwrap_or_default()
-        .map(|s| s.as_str());
 
     // Load the WebAssembly binary file
 
@@ -87,15 +84,17 @@ fn main() -> io::Result<()> {
 
     // Create an execution instance
 
-    let mut inst = Instance::for_module(&arena, &module, DEFAULT_IMPORTS, is_debug_mode)
+    let args = Vec::from_iter_in(start_arg_strings, &arena);
-        .unwrap_or_else(|e| {
+    let dispatcher = DefaultImportDispatcher::new(&args);
+    let mut inst =
+        Instance::for_module(&arena, &module, dispatcher, is_debug_mode).unwrap_or_else(|e| {
             eprintln!("{}", e);
             process::exit(2);
         });
 
     // Run
 
-    let result = inst.call_export_from_cli(&module, start_fn_name, start_arg_strings);
+    let result = inst.call_export_from_cli(&module, start_fn_name, &args);
 
     // Print out return value, if any
 

crates/wasm_interp/src/wasi.rs

@@ -1,54 +1,345 @@
 use roc_wasm_module::Value;
+use std::io::{self, Write};
 
 pub const MODULE_NAME: &str = "wasi_snapshot_preview1";
 
-pub fn dispatch(function_name: &str, arguments: &[Value], _memory: &mut [u8]) -> Option<Value> {
+pub struct WasiDispatcher<'a> {
-    match function_name {
+    pub args: &'a [&'a String],
-        "args_get" => todo!("WASI {}({:?})", function_name, arguments),
+}
-        "args_sizes_get" => todo!("WASI {}({:?})", function_name, arguments),
+
-        "environ_get" => todo!("WASI {}({:?})", function_name, arguments),
+/// Implementation of WASI syscalls
-        "environ_sizes_get" => todo!("WASI {}({:?})", function_name, arguments),
+/// References for other engines:
-        "clock_res_get" => todo!("WASI {}({:?})", function_name, arguments),
+/// https://github.com/wasmerio/wasmer/blob/ef8d2f651ed29b4b06fdc2070eb8189922c54d82/lib/wasi/src/syscalls/mod.rs
-        "clock_time_get" => todo!("WASI {}({:?})", function_name, arguments),
+/// https://github.com/wasm3/wasm3/blob/045040a97345e636b8be4f3086e6db59cdcc785f/source/extra/wasi_core.h
-        "fd_advise" => todo!("WASI {}({:?})", function_name, arguments),
+impl<'a> WasiDispatcher<'a> {
-        "fd_allocate" => todo!("WASI {}({:?})", function_name, arguments),
+    pub fn new(args: &'a [&'a String]) -> Self {
-        "fd_close" => todo!("WASI {}({:?})", function_name, arguments),
+        WasiDispatcher { args }
-        "fd_datasync" => todo!("WASI {}({:?})", function_name, arguments),
+    }
-        "fd_fdstat_get" => todo!("WASI {}({:?})", function_name, arguments),
+
-        "fd_fdstat_set_flags" => todo!("WASI {}({:?})", function_name, arguments),
+    pub fn dispatch(
-        "fd_fdstat_set_rights" => todo!("WASI {}({:?})", function_name, arguments),
+        &mut self,
-        "fd_filestat_get" => todo!("WASI {}({:?})", function_name, arguments),
+        function_name: &str,
-        "fd_filestat_set_size" => todo!("WASI {}({:?})", function_name, arguments),
+        arguments: &[Value],
-        "fd_filestat_set_times" => todo!("WASI {}({:?})", function_name, arguments),
+        memory: &mut [u8],
-        "fd_pread" => todo!("WASI {}({:?})", function_name, arguments),
+    ) -> Option<Value> {
-        "fd_prestat_get" => todo!("WASI {}({:?})", function_name, arguments),
+        let success_code = Some(Value::I32(Errno::Success as i32));
-        "fd_prestat_dir_name" => todo!("WASI {}({:?})", function_name, arguments),
+
-        "fd_pwrite" => todo!("WASI {}({:?})", function_name, arguments),
+        match function_name {
-        "fd_read" => todo!("WASI {}({:?})", function_name, arguments),
+            "args_get" => {
-        "fd_readdir" => todo!("WASI {}({:?})", function_name, arguments),
+                // uint8_t ** argv,
-        "fd_renumber" => todo!("WASI {}({:?})", function_name, arguments),
+                let mut ptr_ptr_argv = arguments[0].unwrap_i32() as usize;
-        "fd_seek" => todo!("WASI {}({:?})", function_name, arguments),
+                // uint8_t * argv_buf
-        "fd_sync" => todo!("WASI {}({:?})", function_name, arguments),
+                let mut ptr_argv_buf = arguments[1].unwrap_i32() as usize;
-        "fd_tell" => todo!("WASI {}({:?})", function_name, arguments),
+
-        "fd_write" => todo!("WASI {}({:?})", function_name, arguments),
+                for arg in self.args {
-        "path_create_directory" => todo!("WASI {}({:?})", function_name, arguments),
+                    write_u32(memory, ptr_ptr_argv, ptr_argv_buf as u32);
-        "path_filestat_get" => todo!("WASI {}({:?})", function_name, arguments),
+                    let bytes_target = &mut memory[ptr_argv_buf..][..arg.len()];
-        "path_filestat_set_times" => todo!("WASI {}({:?})", function_name, arguments),
+                    bytes_target.copy_from_slice(arg.as_bytes());
-        "path_link" => todo!("WASI {}({:?})", function_name, arguments),
+                    memory[ptr_argv_buf + arg.len()] = 0; // C string zero termination
-        "path_open" => todo!("WASI {}({:?})", function_name, arguments),
+                    ptr_argv_buf += arg.len() + 1;
-        "path_readlink" => todo!("WASI {}({:?})", function_name, arguments),
+                    ptr_ptr_argv += 4;
-        "path_remove_directory" => todo!("WASI {}({:?})", function_name, arguments),
+                }
-        "path_rename" => todo!("WASI {}({:?})", function_name, arguments),
+
-        "path_symlink" => todo!("WASI {}({:?})", function_name, arguments),
+                success_code
-        "path_unlink_file" => todo!("WASI {}({:?})", function_name, arguments),
+            }
-        "poll_oneoff" => todo!("WASI {}({:?})", function_name, arguments),
+            "args_sizes_get" => {
-        "proc_exit" => todo!("WASI {}({:?})", function_name, arguments),
+                // (i32, i32) -> i32
-        "proc_raise" => todo!("WASI {}({:?})", function_name, arguments),
+
-        "sched_yield" => todo!("WASI {}({:?})", function_name, arguments),
+                // number of string arguments
-        "random_get" => todo!("WASI {}({:?})", function_name, arguments),
+                let ptr_argc = arguments[0].unwrap_i32() as usize;
-        "sock_recv" => todo!("WASI {}({:?})", function_name, arguments),
+                // size of string arguments buffer
-        "sock_send" => todo!("WASI {}({:?})", function_name, arguments),
+                let ptr_argv_buf_size = arguments[1].unwrap_i32() as usize;
-        "sock_shutdown" => todo!("WASI {}({:?})", function_name, arguments),
+
-        _ => panic!("Unknown WASI function {}({:?})", function_name, arguments),
+                let argc = self.args.len() as u32;
+                write_u32(memory, ptr_argc, argc);
+
+                let argv_buf_size: u32 = self.args.iter().map(|a| 1 + a.len() as u32).sum();
+                write_u32(memory, ptr_argv_buf_size, argv_buf_size);
+
+                success_code
+            }
+            "environ_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "environ_sizes_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "clock_res_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "clock_time_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_advise" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_allocate" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_close" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_datasync" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_fdstat_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_fdstat_set_flags" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_fdstat_set_rights" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_filestat_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_filestat_set_size" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_filestat_set_times" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_pread" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_prestat_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_prestat_dir_name" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_pwrite" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_read" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_readdir" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_renumber" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_seek" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_sync" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_tell" => todo!("WASI {}({:?})", function_name, arguments),
+            "fd_write" => {
+                // file descriptor
+                let fd = arguments[0].unwrap_i32();
+                // Array of IO vectors
+                let ptr_iovs = arguments[1].unwrap_i32() as usize;
+                // Length of array
+                let iovs_len = arguments[2].unwrap_i32();
+                // Out param: number of bytes written
+                let ptr_nwritten = arguments[3].unwrap_i32() as usize;
+
+                let mut write_lock = match fd {
+                    1 => Ok(io::stdout().lock()),
+                    2 => Err(io::stderr().lock()),
+                    _ => return Some(Value::I32(Errno::Inval as i32)),
+                };
+
+                let mut n_written: i32 = 0;
+                let mut negative_length_count = 0;
+                for _ in 0..iovs_len {
+                    // https://man7.org/linux/man-pages/man2/readv.2.html
+                    // struct iovec {
+                    //     void  *iov_base;    /* Starting address */
+                    //     size_t iov_len;     /* Number of bytes to transfer */
+                    // };
+                    let iov_base = read_u32(memory, ptr_iovs) as usize;
+                    let iov_len = read_i32(memory, ptr_iovs + 4);
+                    if iov_len < 0 {
+                        // I found negative-length iov's when I implemented this in JS for the web REPL (see wasi.js)
+                        // I'm not sure why, but this solution worked, and it's the same WASI libc - there's only one.
+                        n_written += iov_len;
+                        negative_length_count += 1;
+                        continue;
+                    }
+                    let bytes = &memory[iov_base..][..iov_len as usize];
+
+                    match &mut write_lock {
+                        Ok(stdout) => {
+                            n_written += stdout.write(bytes).unwrap() as i32;
+                        }
+                        Err(stderr) => {
+                            n_written += stderr.write(bytes).unwrap() as i32;
+                        }
+                    }
+                }
+
+                write_i32(memory, ptr_nwritten, n_written);
+                if negative_length_count > 0 {
+                    // Let's see if we ever get this message. If not, we can remove this negative-length stuff.
+                    eprintln!(
+                        "WASI DEV INFO: found {} negative-length iovecs.",
+                        negative_length_count
+                    );
+                }
+
+                success_code
+            }
+            "path_create_directory" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_filestat_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_filestat_set_times" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_link" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_open" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_readlink" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_remove_directory" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_rename" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_symlink" => todo!("WASI {}({:?})", function_name, arguments),
+            "path_unlink_file" => todo!("WASI {}({:?})", function_name, arguments),
+            "poll_oneoff" => todo!("WASI {}({:?})", function_name, arguments),
+            "proc_exit" => todo!("WASI {}({:?})", function_name, arguments),
+            "proc_raise" => todo!("WASI {}({:?})", function_name, arguments),
+            "sched_yield" => todo!("WASI {}({:?})", function_name, arguments),
+            "random_get" => todo!("WASI {}({:?})", function_name, arguments),
+            "sock_recv" => todo!("WASI {}({:?})", function_name, arguments),
+            "sock_send" => todo!("WASI {}({:?})", function_name, arguments),
+            "sock_shutdown" => todo!("WASI {}({:?})", function_name, arguments),
+            _ => panic!("Unknown WASI function {}({:?})", function_name, arguments),
+        }
     }
 }
+
+fn read_u32(memory: &mut [u8], addr: usize) -> u32 {
+    let mut bytes = [0; 4];
+    bytes.copy_from_slice(&memory[addr..][..4]);
+    u32::from_le_bytes(bytes)
+}
+
+fn read_i32(memory: &mut [u8], addr: usize) -> i32 {
+    let mut bytes = [0; 4];
+    bytes.copy_from_slice(&memory[addr..][..4]);
+    i32::from_le_bytes(bytes)
+}
+
+fn write_u32(memory: &mut [u8], addr: usize, value: u32) {
+    memory[addr..][..4].copy_from_slice(&value.to_le_bytes());
+}
+
+fn write_i32(memory: &mut [u8], addr: usize, value: i32) {
+    memory[addr..][..4].copy_from_slice(&value.to_le_bytes());
+}
+
+/// Error codes returned by functions.
+/// Not all of these error codes are returned by the functions provided by this
+/// API; some are used in higher-level library layers, and others are provided
+/// merely for alignment with POSIX.
+#[repr(u8)]
+#[derive(Clone, Copy, PartialEq, Eq)]
+pub enum Errno {
+    /// No error occurred. System call completed successfully.
+    Success,
+    /// Argument list too long.
+    Toobig,
+    /// Permission denied.
+    Access,
+    /// Address in use.
+    Addrinuse,
+    /// Address not available.
+    Addrnotavail,
+    /// Address family not supported.
+    Afnosupport,
+    /// Resource unavailable, or operation would block.
+    Again,
+    /// Connection already in progress.
+    Already,
+    /// Bad file descriptor.
+    Badf,
+    /// Bad message.
+    Badmsg,
+    /// Device or resource busy.
+    Busy,
+    /// Operation canceled.
+    Canceled,
+    /// No child processes.
+    Child,
+    /// Connection aborted.
+    Connaborted,
+    /// Connection refused.
+    Connrefused,
+    /// Connection reset.
+    Connreset,
+    /// Resource deadlock would occur.
+    Deadlk,
+    /// Destination address required.
+    Destaddrreq,
+    /// Mathematics argument out of domain of function.
+    Dom,
+    /// Reserved.
+    Dquot,
+    /// File exists.
+    Exist,
+    /// Bad address.
+    Fault,
+    /// File too large.
+    Fbig,
+    /// Host is unreachable.
+    Hostunreach,
+    /// Identifier removed.
+    Idrm,
+    /// Illegal byte sequence.
+    Ilseq,
+    /// Operation in progress.
+    Inprogress,
+    /// Interrupted function.
+    Intr,
+    /// Invalid argument.
+    Inval,
+    /// I/O error.
+    Io,
+    /// Socket is connected.
+    Isconn,
+    /// Is a directory.
+    Isdir,
+    /// Too many levels of symbolic links.
+    Loop,
+    /// File descriptor value too large.
+    Mfile,
+    /// Too many links.
+    Mlink,
+    /// Message too large.
+    Msgsize,
+    /// Reserved.
+    Multihop,
+    /// Filename too long.
+    Nametoolong,
+    /// Network is down.
+    Netdown,
+    /// Connection aborted by network.
+    Netreset,
+    /// Network unreachable.
+    Netunreach,
+    /// Too many files open in system.
+    Nfile,
+    /// No buffer space available.
+    Nobufs,
+    /// No such device.
+    Nodev,
+    /// No such file or directory.
+    Noent,
+    /// Executable file format error.
+    Noexec,
+    /// No locks available.
+    Nolck,
+    /// Reserved.
+    Nolink,
+    /// Not enough space.
+    Nomem,
+    /// No message of the desired type.
+    Nomsg,
+    /// Protocol not available.
+    Noprotoopt,
+    /// No space left on device.
+    Nospc,
+    /// Function not supported.
+    Nosys,
+    /// The socket is not connected.
+    Notconn,
+    /// Not a directory or a symbolic link to a directory.
+    Notdir,
+    /// Directory not empty.
+    Notempty,
+    /// State not recoverable.
+    Notrecoverable,
+    /// Not a socket.
+    Notsock,
+    /// Not supported, or operation not supported on socket.
+    Notsup,
+    /// Inappropriate I/O control operation.
+    Notty,
+    /// No such device or address.
+    Nxio,
+    /// Value too large to be stored in data type.
+    Overflow,
+    /// Previous owner died.
+    Ownerdead,
+    /// Operation not permitted.
+    Perm,
+    /// Broken pipe.
+    Pipe,
+    /// Protocol error.
+    Proto,
+    /// Protocol not supported.
+    Protonosupport,
+    /// Protocol wrong type for socket.
+    Prototype,
+    /// Result too large.
+    Range,
+    /// Read-only file system.
+    Rofs,
+    /// Invalid seek.
+    Spipe,
+    /// No such process.
+    Srch,
+    /// Reserved.
+    Stale,
+    /// Connection timed out.
+    Timedout,
+    /// Text file busy.
+    Txtbsy,
+    /// Cross-device link.
+    Xdev,
+    /// Extension: Capabilities insufficient.
+    Notcapable,
+}