Merge pull request #37 from wasmerio/feature/pass-arguments-to-wasm-application

Pass arguments to wasm application

src/apis/emscripten/mod.rs

@@ -19,14 +19,14 @@ mod utils;
 mod varargs;
 
 pub use self::storage::{align_memory, static_alloc};
-pub use self::utils::is_emscripten_module;
+pub use self::utils::{is_emscripten_module, allocate_on_stack, allocate_cstr_on_stack};
 
 // TODO: Magic number - how is this calculated?
 const TOTAL_STACK: u32 = 5242880;
-// TODO: Magic number stolen from the generated JS - how is this calculated?
+// TODO: Magic number - how is this calculated?
 const DYNAMICTOP_PTR_DIFF: u32 = 1088;
-
-const STATIC_BUMP: u32 = 215536; // TODO: make this variable
+// TODO: make this variable
+const STATIC_BUMP: u32 = 215536;
 
 fn stacktop(static_bump: u32) -> u32 {
     align_memory(dynamictop_ptr(static_bump) + 4)
@@ -54,10 +54,18 @@ pub fn emscripten_set_up_memory(memory: &mut LinearMemory) {
     let dynamictop_ptr = dynamictop_ptr(STATIC_BUMP) as usize;
     let dynamictop_ptr_offset = dynamictop_ptr + mem::size_of::<u32>();
 
+    // println!("value = {:?}");
+
     // We avoid failures of setting the u32 in our memory if it's out of bounds
     if dynamictop_ptr_offset > memory.len() {
-        return;
+        return; // TODO: We should panic instead?
     }
+
+    // debug!("###### dynamic_base = {:?}", dynamic_base(STATIC_BUMP));
+    // debug!("###### dynamictop_ptr = {:?}", dynamictop_ptr);
+    // debug!("###### dynamictop_ptr_offset = {:?}", dynamictop_ptr_offset);
+
+
     let mem = &mut memory[dynamictop_ptr..dynamictop_ptr_offset];
     LittleEndian::write_u32(mem, dynamic_base(STATIC_BUMP));
 }
@@ -74,23 +82,7 @@ macro_rules! mock_external {
 
 pub fn generate_emscripten_env<'a, 'b>() -> ImportObject<&'a str, &'b str> {
     let mut import_object = ImportObject::new();
-    // Global
-    import_object.set(
-        "env",
-        "global1",
-        ImportValue::Global(24), // TODO
-    );
-    import_object.set(
-        "env",
-        "global2",
-        ImportValue::Global(50), // TODO
-    );
-    import_object.set(
-        "env",
-        "global3",
-        ImportValue::Global(67), // TODO
-    );
-
+    // Globals
     import_object.set(
         "env",
         "STACKTOP",
@@ -107,7 +99,6 @@ pub fn generate_emscripten_env<'a, 'b>() -> ImportObject<&'a str, &'b str> {
         ImportValue::Global(dynamictop_ptr(STATIC_BUMP) as _),
     );
     import_object.set("env", "tableBase", ImportValue::Global(0));
-
     // Print functions
     import_object.set("env", "printf", ImportValue::Func(io::printf as _));
     import_object.set("env", "putchar", ImportValue::Func(io::putchar as _));

src/apis/emscripten/syscalls.rs

@@ -524,7 +524,7 @@ pub extern "C" fn ___syscall192(
         "=> addr: {}, len: {}, prot: {}, flags: {}, fd: {}, off: {}",
         addr, len, prot, flags, fd, off
     );
-    
+
     let (memalign, memset) = {
         let emscripten_data = &instance.emscripten_data.as_ref().unwrap();
         (emscripten_data.memalign, emscripten_data.memset)
@@ -829,7 +829,7 @@ pub extern "C" fn ___syscall63(
     unsafe { dup2(src, dst) }
 }
 
-// newselect
+// select
 pub extern "C" fn ___syscall142(
     _which: c_int,
     mut varargs: VarArgs,
@@ -848,7 +848,7 @@ pub extern "C" fn ___syscall142(
 
     let readfds_ptr = instance.memory_offset_addr(0, readfds as _) as _;
     let writefds_ptr = instance.memory_offset_addr(0, writefds as _) as _;
-    
+
     unsafe { select(nfds, readfds_ptr, writefds_ptr, 0 as _, 0 as _) }
 }
 

src/apis/emscripten/utils.rs

@@ -2,9 +2,10 @@ use byteorder::{ByteOrder, LittleEndian};
 use crate::webassembly::module::Module;
 use crate::webassembly::Instance;
 use libc::stat;
-use std::ffi::CStr;
+use std::ffi::{CStr, CString};
 use std::os::raw::c_char;
 use std::slice;
+use std::mem::size_of;
 
 /// We check if a provided module is an Emscripten generated one
 pub fn is_emscripten_module(module: &Module) -> bool {
@@ -18,16 +19,39 @@ pub fn is_emscripten_module(module: &Module) -> bool {
 
 pub unsafe fn copy_cstr_into_wasm(instance: &mut Instance, cstr: *const c_char) -> u32 {
     let s = CStr::from_ptr(cstr).to_str().unwrap();
-    let space_offset = (instance.emscripten_data.as_ref().unwrap().malloc)(s.len() as _, instance);
+    let cstr_len = s.len();
+    let space_offset = (instance.emscripten_data.as_ref().unwrap().malloc)((cstr_len as i32) + 1, instance);
     let raw_memory = instance.memory_offset_addr(0, space_offset as _) as *mut u8;
-    let slice = slice::from_raw_parts_mut(raw_memory, s.len());
+    let slice = slice::from_raw_parts_mut(raw_memory, cstr_len);
 
     for (byte, loc) in s.bytes().zip(slice.iter_mut()) {
         *loc = byte;
     }
+
+    *raw_memory.add(cstr_len) = 0;
+
     space_offset
 }
 
+pub unsafe fn allocate_on_stack<'a, T: Copy>(count: u32, instance: &'a Instance) -> (u32, &'a mut [T]) {
+    let offset = (instance.emscripten_data.as_ref().unwrap().stack_alloc)(count * (size_of::<T>() as u32), instance);
+    let addr = instance.memory_offset_addr(0, offset as _) as *mut T;
+    let slice = slice::from_raw_parts_mut(addr, count as usize);
+
+    (offset, slice)
+}
+
+pub unsafe fn allocate_cstr_on_stack<'a>(s: &str, instance: &'a Instance) -> (u32, &'a [u8]) {
+    let (offset, slice) = allocate_on_stack((s.len() + 1) as u32, instance);
+
+    use std::iter;
+    for (byte, loc) in s.bytes().chain(iter::once(0)).zip(slice.iter_mut()) {
+        *loc = byte;
+    }
+
+    (offset, slice)
+}
+
 pub unsafe fn copy_terminated_array_of_cstrs(
     _instance: &mut Instance,
     cstrs: *mut *mut c_char,

src/bin/wasmer.rs

@@ -7,6 +7,7 @@ use std::io::Read;
 use std::path::PathBuf;
 use std::process::exit;
 
+use apis::emscripten::{allocate_on_stack, allocate_cstr_on_stack};
 use structopt::StructOpt;
 
 use wasmer::*;
@@ -28,11 +29,17 @@ enum CLIOptions {
 struct Run {
     #[structopt(short = "d", long = "debug")]
     debug: bool,
+
     /// Input file
     #[structopt(parse(from_os_str))]
     path: PathBuf,
+
+    /// Application arguments
+    #[structopt(name = "--", raw(multiple="true"))]
+    args: Vec<String>,
 }
 
+
 /// Read the contents of a file
 fn read_file_contents(path: &PathBuf) -> Result<Vec<u8>, io::Error> {
     let mut buffer: Vec<u8> = Vec::new();
@@ -42,26 +49,31 @@ fn read_file_contents(path: &PathBuf) -> Result<Vec<u8>, io::Error> {
 }
 
 /// Execute a WASM/WAT file
-fn execute_wasm(wasm_path: PathBuf) -> Result<(), String> {
-    let mut wasm_binary: Vec<u8> = read_file_contents(&wasm_path).map_err(|err| {
+fn execute_wasm(options: &Run) -> Result<(), String> {
+    let wasm_path = &options.path;
+
+    let mut wasm_binary: Vec<u8> = read_file_contents(wasm_path).map_err(|err| {
         format!(
             "Can't read the file {}: {}",
             wasm_path.as_os_str().to_string_lossy(),
             err
         )
     })?;
+
     if !webassembly::utils::is_wasm_binary(&wasm_binary) {
         wasm_binary = wabt::wat2wasm(wasm_binary)
             .map_err(|err| format!("Can't convert from wast to wasm: {:?}", err))?;
     }
+
     // TODO: We should instantiate after compilation, so we provide the
     // emscripten environment conditionally based on the module
     let import_object = apis::generate_emscripten_env();
-    let webassembly::ResultObject { module, instance } =
+    let webassembly::ResultObject { module, mut instance } =
         webassembly::instantiate(wasm_binary, import_object)
             .map_err(|err| format!("Can't instantiate the WebAssembly module: {}", err))?;
 
     if apis::emscripten::is_emscripten_module(&module) {
+
         // Emscripten __ATINIT__
         if let Some(&webassembly::Export::Function(environ_constructor_index)) = module.info.exports.get("___emscripten_environ_constructor") {
             debug!("emscripten::___emscripten_environ_constructor");
@@ -69,14 +81,19 @@ fn execute_wasm(wasm_path: PathBuf) -> Result<(), String> {
                 get_instance_function!(instance, environ_constructor_index);
             call_protected!(___emscripten_environ_constructor(&instance)).map_err(|err| format!("{}", err))?;
         };
+
         // TODO: We also need to handle TTY.init() and SOCKFS.root = FS.mount(SOCKFS, {}, null)
         let func_index = match module.info.exports.get("_main") {
             Some(&webassembly::Export::Function(index)) => index,
             _ => panic!("_main emscripten function not found"),
         };
+
         let main: extern "C" fn(u32, u32, &webassembly::Instance) =
             get_instance_function!(instance, func_index);
-        return call_protected!(main(0, 0, &instance)).map_err(|err| format!("{}", err));
+
+        let (argc, argv) = store_module_arguments(options, &mut instance);
+
+        return call_protected!(main(argc, argv, &instance)).map_err(|err| format!("{}", err));
         // TODO: We should implement emscripten __ATEXIT__
     } else {
         let func_index =
@@ -93,7 +110,7 @@ fn execute_wasm(wasm_path: PathBuf) -> Result<(), String> {
 }
 
 fn run(options: Run) {
-    match execute_wasm(options.path.clone()) {
+    match execute_wasm(&options) {
         Ok(()) => {}
         Err(message) => {
             // let name = options.path.as_os_str().to_string_lossy();
@@ -110,3 +127,54 @@ fn main() {
         CLIOptions::SelfUpdate => update::self_update(),
     }
 }
+
+fn store_module_arguments(options: &Run, instance: &mut webassembly::Instance) -> (u32, u32) {
+    let argc = options.args.len() + 1;
+
+    let (argv_offset, argv_slice): (_, &mut [u32]) = unsafe { allocate_on_stack(((argc + 1) * 4) as u32, instance) };
+    assert!(argv_slice.len() >= 1);
+
+    argv_slice[0] = unsafe { allocate_cstr_on_stack(options.path.to_str().unwrap(), instance).0 };
+
+    for (slot, arg) in argv_slice[1..argc].iter_mut().zip(options.args.iter()) {
+        *slot = unsafe { allocate_cstr_on_stack(&arg, instance).0 };
+    }
+
+    argv_slice[argc] = 0;
+
+    (argc as u32, argv_offset)
+}
+
+// fn get_module_arguments(options: &Run, instance: &mut webassembly::Instance) -> (u32, u32) {
+//     // Application Arguments
+//     let mut arg_values: Vec<String> = Vec::new();
+//     let mut arg_addrs: Vec<*const u8> = Vec::new();
+//     let arg_length = options.args.len() + 1;
+
+//     arg_values.reserve_exact(arg_length);
+//     arg_addrs.reserve_exact(arg_length);
+
+//     // Push name of wasm file
+//     arg_values.push(format!("{}\0", options.path.to_str().unwrap()));
+//     arg_addrs.push(arg_values[0].as_ptr());
+
+//     // Push additional arguments
+//     for (i, arg) in options.args.iter().enumerate() {
+//         arg_values.push(format!("{}\0", arg));
+//         arg_addrs.push(arg_values[i + 1].as_ptr());
+//     }
+
+//     // Get argument count and pointer to addresses
+//     let argv = arg_addrs.as_ptr() as *mut *mut i8;
+//     let argc = arg_length as u32;
+
+//     // Copy the the arguments into the wasm memory and get offset
+//     let argv_offset =  unsafe {
+//         copy_cstr_array_into_wasm(argc, argv, instance)
+//     };
+
+//     debug!("argc = {:?}", argc);
+//     debug!("argv = {:?}", arg_addrs);
+
+//     (argc, argv_offset)
+// }

src/webassembly/instance.rs

@@ -70,10 +70,11 @@ fn get_function_addr(
 }
 
 pub struct EmscriptenData {
-    pub malloc: extern "C" fn(i32, &mut Instance) -> u32,
+    pub malloc: extern "C" fn(i32, &Instance) -> u32,
     pub free: extern "C" fn(i32, &mut Instance),
     pub memalign: extern "C" fn (u32, u32, &mut Instance) -> u32,
     pub memset: extern "C" fn(u32, i32, u32, &mut Instance) -> u32,
+    pub stack_alloc: extern "C" fn (u32, &Instance) -> u32,
 }
 
 impl fmt::Debug for EmscriptenData {
@@ -265,7 +266,7 @@ impl Instance {
                     func
                     // unimplemented!()
                 }).collect();
-            
+
             if let Some(ref progress_bar) = progress_bar_option {
                 progress_bar.set_style(ProgressStyle::default_bar()
                     .template(&format!("{} {{msg}}", style("[{elapsed_precise}]").bold().dim())));
@@ -550,21 +551,44 @@ impl Instance {
                 let free_export = module.info.exports.get("_free");
                 let memalign_export = module.info.exports.get("_memalign");
                 let memset_export = module.info.exports.get("_memset");
+                let stack_alloc_export = module.info.exports.get("stackAlloc");
+
+                let mut malloc_addr = 0 as *const u8;
+                let mut free_addr = 0 as *const u8;
+                let mut memalign_addr = 0 as *const u8;
+                let mut memset_addr = 0 as *const u8;
+                let mut stack_alloc_addr = 0 as _;
+
+                if malloc_export.is_none() && free_export.is_none() && memalign_export.is_none() && memset_export.is_none() {
+                    None
+                } else {
+                    if let Some(Export::Function(malloc_index)) = malloc_export {
+                        malloc_addr = get_function_addr(&malloc_index, &import_functions, &functions);
+                    }
+
+                    if let Some(Export::Function(free_index)) = free_export {
+                        free_addr = get_function_addr(&free_index, &import_functions, &functions);
+                    }
+
+                    if let Some(Export::Function(memalign_index)) = memalign_export {
+                        memalign_addr = get_function_addr(&memalign_index, &import_functions, &functions);
+                    }
+
+                    if let Some(Export::Function(memset_index)) = memset_export {
+                        memset_addr = get_function_addr(&memset_index, &import_functions, &functions);
+                    }
+
+                    if let Some(Export::Function(stack_alloc_index)) = stack_alloc_export {
+                        stack_alloc_addr = get_function_addr(&stack_alloc_index, &import_functions, &functions);
+                    }
 
-                if let (Some(Export::Function(malloc_index)), Some(Export::Function(free_index)), Some(Export::Function(memalign_index)), Some(Export::Function(memset_index))) = (malloc_export, free_export, memalign_export, memset_export) {
-                    let malloc_addr = get_function_addr(&malloc_index, &import_functions, &functions);
-                    let free_addr = get_function_addr(&free_index, &import_functions, &functions);
-                    let memalign_addr = get_function_addr(&memalign_index, &import_functions, &functions);
-                    let memset_addr = get_function_addr(&memset_index, &import_functions, &functions);
-                    
                     Some(EmscriptenData {
                         malloc: mem::transmute(malloc_addr),
                         free: mem::transmute(free_addr),
                         memalign: mem::transmute(memalign_addr),
                         memset: mem::transmute(memset_addr),
+                        stack_alloc: mem::transmute(stack_alloc_addr),
                     })
-                } else {
-                    None
                 }
             }
         } else {

src/webassembly/module.rs

@@ -9,7 +9,7 @@ use cranelift_codegen::cursor::FuncCursor;
 use cranelift_codegen::ir::immediates::{Imm64, Offset32};
 use cranelift_codegen::ir::types::*;
 use cranelift_codegen::ir::{
-    self, AbiParam, ArgumentPurpose, ExtFuncData, ExternalName, FuncRef, InstBuilder, Signature,
+    self, AbiParam, ArgumentPurpose, ExtFuncData, ExternalName, FuncRef, InstBuilder, Signature, TrapCode,
 };
 use cranelift_codegen::isa::{CallConv, TargetFrontendConfig};
 use cranelift_entity::{EntityRef, PrimaryMap};
@@ -532,6 +532,8 @@ impl<'environment> FuncEnvironmentTrait for FuncEnvironment<'environment> {
         mflags.set_aligned();
         let func_ptr = pos.ins().load(ptr, mflags, entry_addr, 0);
 
+        pos.ins().trapz(func_ptr, TrapCode::IndirectCallToNull);
+
         // Build a value list for the indirect call instruction containing the callee, call_args,
         // and the vmctx parameter.
         let mut args = ir::ValueList::default();