Merge branch 'wasix-core-changes' into wasix

.github/workflows/benchmark.yaml

@@ -24,7 +24,7 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
-          toolchain: 1.63
+          toolchain: 1.64
       - name: Configure cargo data directory
         # After this point, all cargo registry and crate data is stored in
         # $GITHUB_WORKSPACE/.cargo_home. This allows us to cache only the files

.github/workflows/build.yml

@@ -100,7 +100,7 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
-          toolchain: 1.63
+          toolchain: 1.64
           target: ${{ matrix.target }}
       - uses: Swatinem/rust-cache@v1
         if: matrix.use_sccache != true

.github/workflows/cloudcompiler.yaml

@@ -21,7 +21,7 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
-          toolchain: 1.63
+          toolchain: 1.64
           target: ${{ matrix.target }}
       - name: Install wasm32-wasi target
         shell: bash

.github/workflows/documentation.yaml

@@ -16,7 +16,7 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
-          toolchain: 1.63
+          toolchain: 1.64
       - name: Install LLVM
         shell: bash
         run: |

.github/workflows/test-js.yaml

@@ -33,7 +33,7 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
-          toolchain: 1.63
+          toolchain: 1.64
 
       - name: Install NodeJS
         uses: actions/setup-node@v2

.github/workflows/test-sys.yaml

@@ -113,7 +113,7 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
         with:
-          toolchain: 1.63
+          toolchain: 1.64
           target: ${{ matrix.target }}
       - uses: Swatinem/rust-cache@v1
         if: matrix.use_sccache != true

bors.toml

@@ -1,13 +0,0 @@
-status = [
-    #"Audit",
-    "Code lint",
-    #"Test on linux-x64",
-    # "Test on linux-musl-x64",
-    # "Test on linux-aarch64",
-    #"Test on macos-arm64",
-    #"Test on macos-x64",
-    #"Test on windows-x64",
-]
-required_approvals = 0
-timeout_sec = 7200
-delete_merged_branches = true

lib/api/src/js/export.rs

@@ -54,7 +54,7 @@ impl VMMemory {
     }
 
     /// Copies this memory to a new memory
-    pub fn fork(&self) -> Result<VMMemory, wasmer_types::MemoryError> {
+    pub fn duplicate(&self) -> Result<VMMemory, wasmer_types::MemoryError> {
         let new_memory = crate::Memory::new_internal(self.ty.clone())?;
 
         #[cfg(feature = "tracing")]

lib/api/src/js/externals/memory.rs

@@ -265,9 +265,9 @@ impl Memory {
     }
 
     /// Copies this memory to a new memory
-    pub fn fork(&mut self, store: &impl AsStoreRef) -> Result<VMMemory, MemoryError> {
+    pub fn duplicate(&mut self, store: &impl AsStoreRef) -> Result<VMMemory, MemoryError> {
         let mem = self.handle.get(store.as_store_ref().objects());
-        mem.fork()
+        mem.duplicate()
     }
 }
 

lib/api/src/sys/tunables.rs

@@ -117,10 +117,12 @@ mod tests {
                 .unwrap()
             }
         }
+
         fn try_clone(&self) -> Option<Box<dyn LinearMemory + 'static>> {
             None
         }
-        fn fork(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
+
+        fn duplicate(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
             let mem = self.mem.clone();
             Ok(Box::new(Self {
                 memory_definition: Some(UnsafeCell::new(VMMemoryDefinition {

lib/emscripten/src/memory.rs

@@ -110,8 +110,6 @@ pub fn sbrk(mut ctx: FunctionEnvMut<EmEnv>, increment: i32) -> i32 {
         increment,
         total_memory
     );
-    let _ = dynamictop_ptr;
-
     if increment > 0 && new_dynamic_top < old_dynamic_top || new_dynamic_top < 0 {
         abort_on_cannot_grow_memory_old(ctx);
         return -1;

lib/vm/src/memory.rs

@@ -119,11 +119,11 @@ impl WasmMmap {
 
     /// Copies the memory
     /// (in this case it performs a copy-on-write to save memory)
-    pub fn fork(&mut self) -> Result<Self, MemoryError> {
+    pub fn duplicate(&mut self) -> Result<Self, MemoryError> {
         let mem_length = self.size.bytes().0;
         let mut alloc = self
             .alloc
-            .fork(Some(mem_length))
+            .duplicate(Some(mem_length))
             .map_err(MemoryError::Generic)?;
         let base_ptr = alloc.as_mut_ptr();
         Ok(Self {
@@ -289,9 +289,9 @@ impl VMOwnedMemory {
     }
 
     /// Copies this memory to a new memory
-    pub fn fork(&mut self) -> Result<Self, MemoryError> {
+    pub fn duplicate(&mut self) -> Result<Self, MemoryError> {
         Ok(Self {
-            mmap: self.mmap.fork()?,
+            mmap: self.mmap.duplicate()?,
             config: self.config.clone(),
         })
     }
@@ -333,8 +333,8 @@ impl LinearMemory for VMOwnedMemory {
     }
 
     /// Copies this memory to a new memory
-    fn fork(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
-        let forked = Self::fork(self)?;
+    fn duplicate(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
+        let forked = Self::duplicate(self)?;
         Ok(Box::new(forked))
     }
 }
@@ -376,10 +376,10 @@ impl VMSharedMemory {
     }
 
     /// Copies this memory to a new memory
-    pub fn fork(&mut self) -> Result<Self, MemoryError> {
+    pub fn duplicate(&mut self) -> Result<Self, MemoryError> {
         let mut guard = self.mmap.write().unwrap();
         Ok(Self {
-            mmap: Arc::new(RwLock::new(guard.fork()?)),
+            mmap: Arc::new(RwLock::new(guard.duplicate()?)),
             config: self.config.clone(),
         })
     }
@@ -427,8 +427,8 @@ impl LinearMemory for VMSharedMemory {
     }
 
     /// Copies this memory to a new memory
-    fn fork(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
-        let forked = Self::fork(self)?;
+    fn duplicate(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
+        let forked = Self::duplicate(self)?;
         Ok(Box::new(forked))
     }
 }
@@ -495,8 +495,8 @@ impl LinearMemory for VMMemory {
     }
 
     /// Copies this memory to a new memory
-    fn fork(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
-        self.0.fork()
+    fn duplicate(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
+        self.0.duplicate()
     }
 }
 
@@ -558,8 +558,8 @@ impl VMMemory {
     }
 
     /// Copies this memory to a new memory
-    pub fn fork(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
-        LinearMemory::fork(self)
+    pub fn duplicate(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError> {
+        LinearMemory::duplicate(self)
     }
 }
 
@@ -615,5 +615,5 @@ where
     }
 
     /// Copies this memory to a new memory
-    fn fork(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError>;
+    fn duplicate(&mut self) -> Result<Box<dyn LinearMemory + 'static>, MemoryError>;
 }

lib/vm/src/mmap.rs

@@ -317,8 +317,10 @@ impl Mmap {
 
     /// Copies the memory to a new swap file (using copy-on-write if available)
     #[cfg(not(target_os = "windows"))]
-    pub fn fork(&mut self, hint_used: Option<usize>) -> Result<Self, String> {
+    pub fn duplicate(&mut self, hint_used: Option<usize>) -> Result<Self, String> {
         // Empty memory is an edge case
+
+        use std::os::unix::prelude::FromRawFd;
         if self.len == 0 {
             return Ok(Self::new());
         }
@@ -355,28 +357,11 @@ impl Mmap {
                 };
 
                 // The shallow copy failed so we have to do it the hard way
-                let mut off_in: libc::off_t = 0;
-                let mut off_out: libc::off_t = 0;
 
-                cfg_if::cfg_if! {
-                    if #[cfg(not(any(target_env = "musl", target_vendor = "apple")))]
-                    {
-                        let ret = libc::copy_file_range(self.fd.0, &mut off_in, fd.0, &mut off_out, len, 0);
-                    } else {
-                        // TODO: don't use as casts...
-                        let ret = match copy_file_range_impl(self.fd.0, off_in as u64, fd.0, off_out as u64, len) {
-                            Ok(_) => 0,
-                            Err(_err) => -1,
-                        };
-                    }
-                }
-
-                if ret < 0 {
-                    return Err(format!(
-                        "failed to copy temporary file data - {}",
-                        io::Error::last_os_error()
-                    ));
-                }
+                let mut source = std::fs::File::from_raw_fd(self.fd.0);
+                let mut out = std::fs::File::from_raw_fd(fd.0);
+                copy_file_range(&mut source, 0, &mut out, 0, len)
+                    .map_err(|err| format!("Could not copy memory: {err}"))?;
 
                 #[cfg(feature = "tracing")]
                 trace!("memory copy finished (size={})", len);
@@ -410,7 +395,7 @@ impl Mmap {
 
     /// Copies the memory to a new swap file (using copy-on-write if available)
     #[cfg(target_os = "windows")]
-    pub fn fork(&mut self, hint_used: Option<usize>) -> Result<Self, String> {
+    pub fn duplicate(&mut self, hint_used: Option<usize>) -> Result<Self, String> {
         // Create a new memory which we will copy to
         let new_mmap = Self::with_at_least(self.len)?;
 
@@ -436,66 +421,6 @@ impl Mmap {
     }
 }
 
-/// Rust implementation of libc::copy_file_range.
-///
-/// Needed because that function is not available on all platforms.
-// TODO: better implementation! (this is very quick, low effort)
-// TODO: this function needs tests!
-#[cfg(target_family = "unix")]
-#[allow(dead_code)]
-fn copy_file_range_impl(
-    source_fd: i32,
-    source_offset: u64,
-    out_fd: i32,
-    out_offset: u64,
-    len: usize,
-) -> Result<(), std::io::Error> {
-    use std::{
-        io::{Seek, SeekFrom},
-        os::unix::io::FromRawFd,
-    };
-
-    let mut f1 = unsafe { std::fs::File::from_raw_fd(source_fd) };
-    let f1_original_pos = f1.stream_position()?;
-
-    // TODO: don't cast with as
-    f1.seek(SeekFrom::Start(source_offset))?;
-
-    let mut f2 = unsafe { std::fs::File::from_raw_fd(out_fd) };
-    let f2_original_pos = f2.stream_position()?;
-    f2.seek(SeekFrom::Start(out_offset))?;
-
-    let mut reader = std::io::BufReader::new(f1);
-    let mut writer = std::io::BufWriter::new(f2);
-
-    let mut buffer = vec![0u8; 4096];
-
-    let mut offset = 0;
-    let end = len.saturating_sub(buffer.len());
-    while offset < end {
-        let read = reader.read(&mut buffer)?;
-        writer.write_all(&buffer)?;
-        offset += read;
-    }
-    // Need to read the last chunk.
-    let remaining = len - offset;
-    if remaining > 0 {
-        reader.read_exact(&mut buffer[0..remaining])?;
-        writer.write_all(&buffer[0..remaining])?;
-    }
-
-    writer.flush()?;
-
-    // Restore files to original position.
-    let mut f1 = reader.into_inner();
-    f1.seek(SeekFrom::Start(f1_original_pos))?;
-
-    let mut f2 = writer.into_inner()?;
-    f2.seek(SeekFrom::Start(f2_original_pos))?;
-
-    Ok(())
-}
-
 impl Drop for Mmap {
     #[cfg(not(target_os = "windows"))]
     fn drop(&mut self) {
@@ -522,6 +447,54 @@ fn _assert() {
     _assert_send_sync::<Mmap>();
 }
 
+/// Copy a range of a file to another file.
+// We could also use libc::copy_file_range on some systems, but it's
+// hard to do this because it is not available on many libc implementations.
+// (not on Mac OS, musl, ...)
+#[cfg(target_family = "unix")]
+fn copy_file_range(
+    source: &mut std::fs::File,
+    source_offset: u64,
+    out: &mut std::fs::File,
+    out_offset: u64,
+    len: usize,
+) -> Result<(), std::io::Error> {
+    use std::io::{Seek, SeekFrom};
+
+    let source_original_pos = source.stream_position()?;
+    source.seek(SeekFrom::Start(source_offset))?;
+
+    // TODO: don't cast with as
+
+    let out_original_pos = out.stream_position()?;
+    out.seek(SeekFrom::Start(out_offset))?;
+
+    // TODO: don't do this horrible "triple buffering" below".
+    // let mut reader = std::io::BufReader::new(source);
+
+    // TODO: larger buffer?
+    let mut buffer = vec![0u8; 4096];
+
+    let mut to_read = len;
+    while to_read > 0 {
+        let chunk_size = std::cmp::min(to_read, buffer.len());
+        let read = source.read(&mut buffer[0..chunk_size])?;
+        out.write_all(&buffer[0..read])?;
+        to_read -= read;
+    }
+
+    // Need to read the last chunk.
+    out.flush()?;
+
+    // Restore files to original position.
+    source.seek(SeekFrom::Start(source_original_pos))?;
+    out.flush()?;
+    out.sync_data()?;
+    out.seek(SeekFrom::Start(out_original_pos))?;
+
+    Ok(())
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -533,4 +506,59 @@ mod tests {
         assert_eq!(round_up_to_page_size(4096, 4096), 4096);
         assert_eq!(round_up_to_page_size(4097, 4096), 8192);
     }
+
+    #[cfg(target_family = "unix")]
+    #[test]
+    fn test_copy_file_range() -> Result<(), std::io::Error> {
+        // I know tempfile:: exists, but this doesn't bring in an extra
+        // dependency.
+
+        use std::{fs::OpenOptions, io::Seek};
+
+        let dir = std::env::temp_dir().join("wasmer/copy_file_range");
+        if dir.is_dir() {
+            std::fs::remove_dir_all(&dir).unwrap()
+        }
+        std::fs::create_dir_all(&dir).unwrap();
+
+        let pa = dir.join("a");
+        let pb = dir.join("b");
+
+        let data: Vec<u8> = (0..100).collect();
+        let mut a = OpenOptions::new()
+            .read(true)
+            .write(true)
+            .create_new(true)
+            .open(&pa)
+            .unwrap();
+        a.write_all(&data).unwrap();
+
+        let datb: Vec<u8> = (100..200).collect();
+        let mut b = OpenOptions::new()
+            .read(true)
+            .write(true)
+            .create_new(true)
+            .open(&pb)
+            .unwrap();
+        b.write_all(&datb).unwrap();
+
+        a.seek(io::SeekFrom::Start(30)).unwrap();
+        b.seek(io::SeekFrom::Start(99)).unwrap();
+        copy_file_range(&mut a, 10, &mut b, 40, 15).unwrap();
+
+        assert_eq!(a.stream_position().unwrap(), 30);
+        assert_eq!(b.stream_position().unwrap(), 99);
+
+        b.seek(io::SeekFrom::Start(0)).unwrap();
+        let mut out = Vec::new();
+        let len = b.read_to_end(&mut out).unwrap();
+        assert_eq!(len, 100);
+        assert_eq!(out[0..40], datb[0..40]);
+        assert_eq!(out[40..55], data[10..25]);
+        assert_eq!(out[55..100], datb[55..100]);
+
+        // TODO: needs more variant tests, but this is enough for now.
+
+        Ok(())
+    }
 }

lib/vm/src/store.rs

@@ -1,3 +1,4 @@
+use core::slice::Iter;
 use std::{
     cell::UnsafeCell,
     fmt,
@@ -9,9 +10,9 @@ use std::{
 
 use wasmer_types::StoreSnapshot;
 
-use crate::VMExternObj;
-
-use crate::{InstanceHandle, VMFunction, VMFunctionEnvironment, VMGlobal, VMMemory, VMTable};
+use crate::{
+    InstanceHandle, VMExternObj, VMFunction, VMFunctionEnvironment, VMGlobal, VMMemory, VMTable,
+};
 
 /// Unique ID to identify a context.
 ///
@@ -104,6 +105,21 @@ impl StoreObjects {
         }
     }
 
+    /// Return an immutable iterator over all globals
+    pub fn iter_globals(&self) -> Iter<VMGlobal> {
+        self.globals.iter()
+    }
+
+    /// Set a global, at index idx. Will panic if idx is out of range
+    /// Safety: the caller should check taht the raw value is compatible
+    /// with destination VMGlobal type
+    pub fn set_global_unchecked(&self, idx: usize, val: u128) {
+        assert!(idx < self.globals.len());
+        unsafe {
+            self.globals[idx].vmglobal().as_mut().val.u128 = val;
+        }
+    }
+
     /// Serializes the mutable things into a snapshot
     pub fn save_snapshot(&self) -> StoreSnapshot {
         let mut ret = StoreSnapshot::default();