add mapdir for emscripten; implement getdents, etc.

Cargo.lock

@@ -2336,6 +2336,7 @@ version = "0.4.2"
 dependencies = [
  "byteorder 1.3.1 (registry+https://github.com/rust-lang/crates.io-index)",
  "glob 0.2.11 (registry+https://github.com/rust-lang/crates.io-index)",
+ "hashbrown 0.1.8 (registry+https://github.com/rust-lang/crates.io-index)",
  "lazy_static 1.3.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "libc 0.2.54 (registry+https://github.com/rust-lang/crates.io-index)",
  "rand 0.6.5 (registry+https://github.com/rust-lang/crates.io-index)",

lib/emscripten/Cargo.toml

@@ -9,14 +9,15 @@ edition = "2018"
 build = "build/mod.rs"
 
 [dependencies]
-wasmer-runtime-core = { path = "../runtime-core", version = "0.4.2" }
+byteorder = "1"
+hashbrown = "0.1"
 lazy_static = "1.2.0"
 libc = "0.2.49"
-byteorder = "1"
 time = "0.1.41"
 wasmer-clif-backend = { path = "../clif-backend", version = "0.4.2" }
-wasmer-singlepass-backend = { path = "../singlepass-backend", version = "0.4.2", optional = true }
 wasmer-llvm-backend = { path = "../llvm-backend", version = "0.4.2", optional = true }
+wasmer-runtime-core = { path = "../runtime-core", version = "0.4.2" }
+wasmer-singlepass-backend = { path = "../singlepass-backend", version = "0.4.2", optional = true }
 
 [target.'cfg(windows)'.dependencies]
 rand = "0.6"

lib/emscripten/src/lib.rs

@@ -3,8 +3,10 @@
 #[macro_use]
 extern crate wasmer_runtime_core;
 
+use hashbrown::HashMap;
 use lazy_static::lazy_static;
 use std::cell::UnsafeCell;
+use std::path::PathBuf;
 use std::{f64, ffi::c_void};
 use wasmer_runtime_core::{
     error::CallResult,
@@ -141,10 +143,14 @@ pub struct EmscriptenData<'a> {
     pub stack_save: Option<Func<'a, (), i32>>,
     pub stack_restore: Option<Func<'a, (i32)>>,
     pub set_threw: Option<Func<'a, (i32, i32)>>,
+    pub mapped_dirs: HashMap<String, PathBuf>,
 }
 
 impl<'a> EmscriptenData<'a> {
-    pub fn new(instance: &'a mut Instance) -> EmscriptenData<'a> {
+    pub fn new(
+        instance: &'a mut Instance,
+        mapped_dirs: HashMap<String, PathBuf>,
+    ) -> EmscriptenData<'a> {
         let malloc = instance.func("_malloc").unwrap();
         let free = instance.func("_free").unwrap();
         let memalign = instance.func("_memalign").ok();
@@ -272,6 +278,7 @@ impl<'a> EmscriptenData<'a> {
             stack_save,
             stack_restore,
             set_threw,
+            mapped_dirs,
         }
     }
 }
@@ -282,8 +289,9 @@ pub fn run_emscripten_instance(
     path: &str,
     args: Vec<&str>,
     entrypoint: Option<String>,
+    mapped_dirs: Vec<(String, PathBuf)>,
 ) -> CallResult<()> {
-    let mut data = EmscriptenData::new(instance);
+    let mut data = EmscriptenData::new(instance, mapped_dirs.into_iter().collect());
     let data_ptr = &mut data as *mut _ as *mut c_void;
     instance.context_mut().data = data_ptr;
 

lib/emscripten/src/syscalls/mod.rs

@@ -10,7 +10,8 @@ pub use self::unix::*;
 #[cfg(windows)]
 pub use self::windows::*;
 
-use super::utils::copy_stat_into_wasm;
+use crate::utils::{copy_stat_into_wasm, get_cstr_path, get_current_directory};
+
 use super::varargs::VarArgs;
 use byteorder::{ByteOrder, LittleEndian};
 /// NOTE: TODO: These syscalls only support wasm_32 for now because they assume offsets are u32
@@ -94,13 +95,17 @@ pub fn ___syscall6(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int
 pub fn ___syscall12(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int {
     debug!("emscripten::___syscall12 (chdir) {}", _which);
     let path_addr: i32 = varargs.get(ctx);
-    unsafe {
     let path_ptr = emscripten_memory_pointer!(ctx.memory(0), path_addr) as *const i8;
-        let _path = std::ffi::CStr::from_ptr(path_ptr);
+    let real_path = get_cstr_path(ctx, path_ptr)
-        let ret = chdir(path_ptr);
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
-        debug!("=> path: {:?}, ret: {}", _path, ret);
+        .unwrap_or(path_ptr);
+    let ret = unsafe { chdir(real_path) };
+    debug!(
+        "=> path: {:?}, ret: {}",
+        unsafe { std::ffi::CStr::from_ptr(real_path) },
+        ret
+    );
     ret
-    }
 }
 
 pub fn ___syscall10(_ctx: &mut Ctx, _one: i32, _two: i32) -> i32 {
@@ -126,11 +131,17 @@ pub fn ___syscall38(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> i32 {
     let new_path_addr: u32 = varargs.get(ctx);
     let old_path = emscripten_memory_pointer!(ctx.memory(0), old_path_addr) as *const i8;
     let new_path = emscripten_memory_pointer!(ctx.memory(0), new_path_addr) as *const i8;
-    let result = unsafe { rename(old_path, new_path) };
+    let real_old_path = get_cstr_path(ctx, old_path)
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(old_path);
+    let real_new_path = get_cstr_path(ctx, new_path)
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(new_path);
+    let result = unsafe { rename(real_old_path, real_new_path) };
     debug!(
         "=> old_path: {}, new_path: {}, result: {}",
-        unsafe { std::ffi::CStr::from_ptr(old_path).to_str().unwrap() },
+        unsafe { std::ffi::CStr::from_ptr(real_old_path).to_str().unwrap() },
-        unsafe { std::ffi::CStr::from_ptr(new_path).to_str().unwrap() },
+        unsafe { std::ffi::CStr::from_ptr(real_new_path).to_str().unwrap() },
         result
     );
     result
@@ -141,7 +152,10 @@ pub fn ___syscall40(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int
     debug!("emscripten::___syscall40 (rmdir)");
     let pathname: u32 = varargs.get(ctx);
     let pathname_addr = emscripten_memory_pointer!(ctx.memory(0), pathname) as *const i8;
-    unsafe { rmdir(pathname_addr) }
+    let real_path = get_cstr_path(ctx, pathname_addr)
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(pathname_addr);
+    unsafe { rmdir(real_path) }
 }
 
 // pipe
@@ -224,10 +238,9 @@ pub fn ___syscall110(_ctx: &mut Ctx, _one: i32, _two: i32) -> i32 {
 // getcwd
 pub fn ___syscall183(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> i32 {
     debug!("emscripten::___syscall183");
-    use std::env;
     let buf_offset: c_int = varargs.get(ctx);
     let _size: c_int = varargs.get(ctx);
-    let path = env::current_dir();
+    let path = get_current_directory(ctx);
     let path_string = path.unwrap().display().to_string();
     let len = path_string.len();
     unsafe {
@@ -401,15 +414,18 @@ pub fn ___syscall195(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_in
     let buf: u32 = varargs.get(ctx);
 
     let pathname_addr = emscripten_memory_pointer!(ctx.memory(0), pathname) as *const i8;
+    let real_path = get_cstr_path(ctx, pathname_addr)
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(pathname_addr);
 
     unsafe {
         let mut _stat: stat = std::mem::zeroed();
-        let ret = stat(pathname_addr, &mut _stat);
+        let ret = stat(real_path, &mut _stat);
         debug!(
             "=> pathname: {}, buf: {}, path: {} = {}\nlast os error: {}",
             pathname,
             buf,
-            std::ffi::CStr::from_ptr(pathname_addr).to_str().unwrap(),
+            std::ffi::CStr::from_ptr(real_path).to_str().unwrap(),
             ret,
             Error::last_os_error()
         );
@@ -440,11 +456,6 @@ pub fn ___syscall197(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_in
     0
 }
 
-pub fn ___syscall220(_ctx: &mut Ctx, _one: i32, _two: i32) -> i32 {
-    debug!("emscripten::___syscall220");
-    -1
-}
-
 // fcntl64
 pub fn ___syscall221(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int {
     debug!("emscripten::___syscall221 (fcntl64) {}", _which);
@@ -457,7 +468,7 @@ pub fn ___syscall221(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_in
     // |FNONBLOCK - 0x04
     debug!("=> fd: {}, cmd: {}", _fd, cmd);
     match cmd {
-        2 => 0,
+        1 | 2 => 0,
         13 | 14 => 0, // pretend file locking worked
         _ => -1,
     }

lib/emscripten/src/syscalls/unix.rs

@@ -43,6 +43,7 @@ use libc::{
     pid_t,
     pread,
     pwrite,
+    readdir,
     // readv,
     recvfrom,
     recvmsg,
@@ -108,8 +109,11 @@ pub fn ___syscall5(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int
     let flags: i32 = varargs.get(ctx);
     let mode: u32 = varargs.get(ctx);
     let pathname_addr = emscripten_memory_pointer!(ctx.memory(0), pathname) as *const i8;
-    let _path_str = unsafe { std::ffi::CStr::from_ptr(pathname_addr).to_str().unwrap() };
+    let real_path = utils::get_cstr_path(ctx, pathname_addr)
-    let fd = unsafe { open(pathname_addr, flags, mode) };
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(pathname_addr);
+    let _path_str = unsafe { std::ffi::CStr::from_ptr(real_path).to_str().unwrap() };
+    let fd = unsafe { open(real_path, flags, mode) };
     debug!(
         "=> pathname: {}, flags: {}, mode: {} = fd: {}\npath: {}\nlast os error: {}",
         pathname,
@@ -160,11 +164,17 @@ pub fn ___syscall83(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int
     let path2_ptr: c_int = varargs.get(ctx);
     let path1 = emscripten_memory_pointer!(ctx.memory(0), path1_ptr) as *mut i8;
     let path2 = emscripten_memory_pointer!(ctx.memory(0), path2_ptr) as *mut i8;
-    let result = unsafe { symlink(path1, path2) };
+    let real_path1 = utils::get_cstr_path(ctx, path1)
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(path1);
+    let real_path2 = utils::get_cstr_path(ctx, path2)
+        .map(|cstr| cstr.as_c_str() as *const _ as *const i8)
+        .unwrap_or(path2);
+    let result = unsafe { symlink(real_path1, real_path2) };
     debug!(
         "=> path1: {}, path2: {}, result: {}",
-        unsafe { std::ffi::CStr::from_ptr(path1).to_str().unwrap() },
+        unsafe { std::ffi::CStr::from_ptr(real_path1).to_str().unwrap() },
-        unsafe { std::ffi::CStr::from_ptr(path2).to_str().unwrap() },
+        unsafe { std::ffi::CStr::from_ptr(real_path2).to_str().unwrap() },
         result,
     );
     result
@@ -796,6 +806,81 @@ pub fn ___syscall196(ctx: &mut Ctx, _which: i32, mut varargs: VarArgs) -> i32 {
     0
 }
 
+// getdents
+// dirent structure is
+// i64, i64, u16 (280), i8, [i8; 256]
+pub fn ___syscall220(ctx: &mut Ctx, _which: i32, mut varargs: VarArgs) -> i32 {
+    debug!("emscripten::___syscall220");
+    let fd: i32 = varargs.get(ctx);
+    let dirp_addr: i32 = varargs.get(ctx);
+    let count: u32 = varargs.get(ctx);
+
+    //let dir = dbg!(emscripten_memory_pointer!(ctx.memory(0), dbg!(fd)) as *mut libc::DIR);
+    let dirp = emscripten_memory_pointer!(ctx.memory(0), dirp_addr) as *mut u8;
+
+    let mut pos = 0;
+    // need to persist stream across calls?
+
+    let dir: *mut libc::DIR = unsafe { libc::fdopendir(fd) };
+
+    dbg!("Start loop");
+    while pos + 280 <= dbg!(count) as usize {
+        dbg!("Pre readdir");
+        let dirent = unsafe { readdir(dir) };
+        dbg!("post readdir");
+        if dirent.is_null() {
+            break;
+        }
+        dbg!("dirent is not null");
+        unsafe {
+            *(dirp.add(pos) as *mut u64) = dbg!((*dirent).d_ino);
+            #[cfg(not(target_os = "macos"))]
+            {
+                *(dirp.add(pos + 8) as *mut u64) = 280 //dbg!((*dirent).d_off);
+            }
+            #[cfg(target_os = "macos")]
+            {
+                *(dirp.add(pos + 8) as *mut u64) = if pos + 280 > count as usize {
+                    count.into()
+                } else {
+                    dbg!((*dirent).d_seekoff);
+                    pos as u64 + 56 //280
+                }; //;
+            }
+            dbg!((*dirent).d_namlen);
+            *(dirp.add(pos + 16) as *mut u16) = 280; //dbg!((*dirent).d_reclen);
+            *(dirp.add(pos + 18) as *mut u8) = dbg!((*dirent).d_type);
+            let upper_bound = std::cmp::min((*dirent).d_reclen, 255) as usize;
+            let mut i = 0;
+            while i < upper_bound {
+                *(dirp.add(pos + 19 + i) as *mut i8) = (*dirent).d_name[i];
+                //dbg!((*dirent).d_name[i] as u8 as char);
+                //dbg!((*dirent).d_name[i] as u8 as char);
+                i += 1;
+            }
+            *(dirp.add(pos + 19 + i) as *mut i8) = 0 as i8;
+        }
+        dbg!("dirent written to memory");
+        pos += 280;
+        /*unsafe {
+            eprintln!(
+                "{}",
+                std::ffi::CStr::from_bytes_with_nul_unchecked({
+                    let arr = *(dirent as *const u8 as *const [u8; 256]);
+                    &arr.to_vec()
+                        .into_iter()
+                        .map(|b| b as u8)
+                        .collect::<Vec<u8>>()[..20]
+                })
+                .to_str()
+                .unwrap()
+            );
+        }*/
+    }
+
+    dbg!(pos as i32)
+}
+
 /// fallocate
 pub fn ___syscall324(ctx: &mut Ctx, _which: c_int, mut varargs: VarArgs) -> c_int {
     debug!("emscripten::___syscall324 (fallocate) {}", _which);

lib/emscripten/src/syscalls/windows.rs

@@ -251,6 +251,12 @@ pub fn ___syscall196(_ctx: &mut Ctx, _one: i32, _two: i32) -> i32 {
     -1
 }
 
+// getdents
+pub fn ___syscall220(_ctx: &mut Ctx, _one: i32, _two: i32) -> i32 {
+    debug!("emscripten::___syscall220");
+    -1
+}
+
 /// fchown
 pub fn ___syscall207(_ctx: &mut Ctx, _which: c_int, _varargs: VarArgs) -> c_int {
     debug!("emscripten::___syscall207 (fchown) {}", _which);

lib/emscripten/src/utils.rs

@@ -5,6 +5,7 @@ use libc::stat;
 use std::ffi::CStr;
 use std::mem::size_of;
 use std::os::raw::c_char;
+use std::path::PathBuf;
 use std::slice;
 use wasmer_runtime_core::memory::Memory;
 use wasmer_runtime_core::{
@@ -204,6 +205,37 @@ pub fn read_string_from_wasm(memory: &Memory, offset: u32) -> String {
     String::from_utf8_lossy(&v).to_owned().to_string()
 }
 
+/// This function trys to find an entry in mapdir
+/// translating paths into their correct value
+pub fn get_cstr_path(ctx: &mut Ctx, path: *const i8) -> Option<std::ffi::CString> {
+    let path_str = unsafe { std::ffi::CStr::from_ptr(path).to_str().unwrap() }.to_string();
+    if let Some(val) = get_emscripten_data(ctx).mapped_dirs.get(&path_str) {
+        std::ffi::CString::new(val.to_string_lossy().as_bytes()).ok()
+    } else {
+        None
+    }
+}
+
+/// gets the current directory
+/// handles mapdir logic
+pub fn get_current_directory(ctx: &mut Ctx) -> Option<PathBuf> {
+    if let Some(val) = get_emscripten_data(ctx).mapped_dirs.get(".") {
+        return Some(val.clone());
+    }
+    std::env::current_dir()
+        .map(|cwd| {
+            if let Some(val) = get_emscripten_data(ctx)
+                .mapped_dirs
+                .get(&cwd.to_string_lossy().to_string())
+            {
+                val.clone()
+            } else {
+                cwd
+            }
+        })
+        .ok()
+}
+
 #[cfg(test)]
 mod tests {
     use super::is_emscripten_module;

src/bin/wasmer.rs

@@ -458,6 +458,7 @@ fn execute_wasm(options: &Run) -> Result<(), String> {
             },
             options.args.iter().map(|arg| arg.as_str()).collect(),
             options.em_entrypoint.clone(),
+            mapped_dirs,
         )
         .map_err(|e| format!("{:?}", e))?;
     } else {