feat(plugin): Enable validation (#7250)

crates/swc_atoms/Cargo.toml

@@ -24,6 +24,7 @@ bytecheck = { version = "0.6.9", optional = true }
 once_cell = "1"
 rkyv = { package = "rkyv", version = "=0.7.40", optional = true, features = [
   "strict",
+  "validation",
 ] }
 # This is to avoid cargo version selection conflict between rkyv=0.7.40 and other versions, as it is strictly pinned
 # cannot be merged.

crates/swc_common/Cargo.toml

@@ -49,18 +49,21 @@ rkyv-impl = ["__rkyv", "rkyv", "swc_atoms/rkyv-impl"]
 rkyv-bytecheck-impl = ["__rkyv", "rkyv-latest", "swc_atoms/rkyv-bytecheck-impl"]
 
 [dependencies]
-ahash                 = "0.7.4"
-anyhow                = { version = "1.0.45", optional = true }
-arbitrary             = { version = "1", optional = true, features = ["derive"] }
-atty                  = { version = "0.2", optional = true }
-bytecheck             = { version = "0.6.9", optional = true }
-cfg-if                = "1.0.0"
-either                = "1.5"
+ahash = "0.7.4"
+anyhow = { version = "1.0.45", optional = true }
+arbitrary = { version = "1", optional = true, features = ["derive"] }
+atty = { version = "0.2", optional = true }
+bytecheck = { version = "0.6.9", optional = true }
+cfg-if = "1.0.0"
+either = "1.5"
 new_debug_unreachable = "1.0.4"
-num-bigint            = "0.4"
-once_cell             = "1.10.0"
-parking_lot           = { version = "0.12.0", optional = true }
-rkyv                  = { version = "=0.7.40", optional = true, features = ["strict"] }
+num-bigint = "0.4"
+once_cell = "1.10.0"
+parking_lot = { version = "0.12.0", optional = true }
+rkyv = { version = "=0.7.40", optional = true, features = [
+  "strict",
+  "validation",
+] }
 # This is to avoid cargo version selection conflict between rkyv=0.7.40 and other versions, as it is strictly pinned
 # cannot be merged.
 rkyv-latest   = { package = "rkyv-test", version = "=0.7.38-test.2", optional = true }

crates/swc_common/src/plugin/serialized.rs

@@ -140,12 +140,8 @@ where
 /// SharedSerializeRegistry which cannot be Infallible. Internally this does
 /// not call deserialize with Infallible deserializer, use
 /// SharedDeserializeMap instead.
-///
-/// # Safety
-/// This is unsafe by construting bytes slice from raw ptr also deserialize
-/// it without slice bound check.
 #[tracing::instrument(level = "info", skip_all)]
-pub unsafe fn deserialize_from_ptr_into_fallible<W>(
+pub fn deserialize_from_ptr_into_fallible<W>(
     raw_allocated_ptr: *const u8,
     raw_allocated_ptr_len: u32,
 ) -> Result<W, Error>
@@ -156,8 +152,7 @@ where
     let serialized =
         PluginSerializedBytes::from_raw_ptr(raw_allocated_ptr, raw_allocated_ptr_len as usize);
 
-    rkyv::from_bytes_unchecked(&serialized.field)
-        .map_err(|_err| Error::msg("Failed to deserialize given ptr"))
+    serialized.deserialize()
 }
 
 /*

crates/swc_css_ast/Cargo.toml

@@ -20,9 +20,12 @@ serde-impl          = ["serde"]
 
 [dependencies]
 bytecheck = { version = "0.6.9", optional = true }
-is-macro  = "0.2.0"
-rkyv      = { version = "=0.7.40", optional = true, features = ["strict"] }
-serde     = { version = "1.0.127", features = ["derive"], optional = true }
+is-macro = "0.2.0"
+rkyv = { version = "=0.7.40", optional = true, features = [
+  "strict",
+  "validation",
+] }
+serde = { version = "1.0.127", features = ["derive"], optional = true }
 
 string_enum = { version = "0.4.0", path = "../string_enum/" }
 swc_atoms   = { version = "0.5.0", path = "../swc_atoms" }

crates/swc_ecma_ast/Cargo.toml

@@ -38,6 +38,7 @@ is-macro = "0.2.1"
 num-bigint = { version = "0.4", features = ["serde"] }
 rkyv = { package = "rkyv", version = "=0.7.40", optional = true, features = [
   "strict",
+  "validation",
 ] }
 # This is to avoid cargo version selection conflict between rkyv=0.7.40 and other versions, as it is strictly pinned
 # cannot be merged.

crates/swc_html_ast/Cargo.toml

@@ -22,9 +22,12 @@ serde-impl          = ["serde"]
 
 [dependencies]
 bytecheck = { version = "0.6.9", optional = true }
-is-macro  = "0.2.0"
-rkyv      = { version = "=0.7.40", optional = true, features = ["strict"] }
-serde     = { version = "1.0.127", features = ["derive"], optional = true }
+is-macro = "0.2.0"
+rkyv = { version = "=0.7.40", optional = true, features = [
+  "strict",
+  "validation",
+] }
+serde = { version = "1.0.127", features = ["derive"], optional = true }
 
 string_enum = { version = "0.4.0", path = "../string_enum/" }
 swc_atoms   = { version = "0.5.0", path = "../swc_atoms" }

crates/swc_plugin_proxy/Cargo.toml

@@ -39,6 +39,7 @@ plugin-mode = ["__plugin_mode", "swc_common/plugin-base", "rkyv-impl"]
 
 rkyv = { package = "rkyv", version = "=0.7.40", optional = true, features = [
   "strict",
+  "validation",
 ] }
 # This is to avoid cargo version selection conflict between rkyv=0.7.40 and other versions, as it is strictly pinned
 # cannot be merged.

crates/swc_plugin_proxy/src/memory_interop/mod.rs

@@ -1,6 +1,4 @@
 mod read_returned_result_from_host;
-pub use read_returned_result_from_host::AllocatedBytesPtr;
 #[cfg(all(feature = "__rkyv", feature = "__plugin_mode", target_arch = "wasm32"))]
-pub(crate) use read_returned_result_from_host::{
-    read_returned_result_from_host, read_returned_result_from_host_fallible,
-};
+pub(crate) use read_returned_result_from_host::read_returned_result_from_host;
+pub use read_returned_result_from_host::AllocatedBytesPtr;

crates/swc_plugin_proxy/src/memory_interop/read_returned_result_from_host.rs

@@ -2,9 +2,7 @@
 use rkyv_latest as rkyv;
 #[cfg_attr(not(target_arch = "wasm32"), allow(unused))]
 #[cfg(any(feature = "__plugin_rt", feature = "__plugin_mode"))]
-use swc_common::plugin::serialized::{
-    deserialize_from_ptr, deserialize_from_ptr_into_fallible, PluginSerializedBytes,
-};
+use swc_common::plugin::serialized::{deserialize_from_ptr, PluginSerializedBytes};
 
 /// A struct to exchange allocated data between memory spaces.
 #[cfg_attr(
@@ -89,58 +87,3 @@ where
         .expect("Returned value should be serializable")
     })
 }
-
-#[cfg(not(feature = "__rkyv"))]
-pub fn read_returned_result_from_host_fallible<F, R>(f: F) -> Option<R> {
-    unimplemented!("Plugin proxy does not work without serialization support")
-}
-
-/// Performs deserialization to the actual return value type from returned ptr.
-///
-/// This behaves same as read_returned_result_from_host, the only difference is
-/// this is for the `Fallible` struct to deserialize. If a struct contains
-/// shared pointers like Arc, Rc rkyv requires trait bounds to the
-/// SharedSerializeRegistry which cannot be infallible.
-#[cfg(all(feature = "__rkyv", target_arch = "wasm32"))]
-#[tracing::instrument(level = "info", skip_all)]
-pub fn read_returned_result_from_host_fallible<F, R>(f: F) -> Option<R>
-where
-    F: FnOnce(u32) -> u32,
-    R: rkyv::Archive,
-    R::Archived: rkyv::Deserialize<R, rkyv::de::deserializers::SharedDeserializeMap>,
-{
-    // Allocate AllocatedBytesPtr to get return value from the host
-    let allocated_bytes_ptr = AllocatedBytesPtr(0, 0);
-    let serialized_allocated_bytes_ptr = PluginSerializedBytes::try_serialize(&allocated_bytes_ptr)
-        .expect("Should able to serialize AllocatedBytesPtr");
-    let (serialized_allocated_bytes_raw_ptr, serialized_allocated_bytes_raw_ptr_size) =
-        serialized_allocated_bytes_ptr.as_ptr();
-
-    let ret = f(serialized_allocated_bytes_raw_ptr as _);
-
-    // Host fn call completes: by contract in host proxy, if return value is 0
-    // we know there's no value to read. Otherwise, we know host filled in
-    // AllocatedBytesPtr to the pointer for the actual value for the
-    // results.
-    if ret == 0 {
-        return None;
-    }
-
-    // Now reconstruct AllocatedBytesPtr to reveal ptr to the allocated bytes
-    let allocated_returned_value_ptr: AllocatedBytesPtr = unsafe {
-        deserialize_from_ptr(
-            serialized_allocated_bytes_raw_ptr,
-            serialized_allocated_bytes_raw_ptr_size as u32,
-        )
-        .expect("Should able to deserialize AllocatedBytesPtr")
-    };
-
-    // Using AllocatedBytesPtr's value, reconstruct actual return value
-    Some(unsafe {
-        deserialize_from_ptr_into_fallible(
-            allocated_returned_value_ptr.0 as _,
-            allocated_returned_value_ptr.1,
-        )
-        .expect("Returned value should be serializable")
-    })
-}

crates/swc_plugin_proxy/src/source_map/plugin_source_map_proxy.rs

@@ -14,7 +14,7 @@ use swc_ecma_ast::SourceMapperExt;
 use swc_trace_macro::swc_trace;
 
 #[cfg(all(feature = "__rkyv", feature = "__plugin_mode", target_arch = "wasm32"))]
-use crate::memory_interop::read_returned_result_from_host_fallible;
+use crate::memory_interop::read_returned_result_from_host;
 
 #[cfg(target_arch = "wasm32")]
 extern "C" {
@@ -84,7 +84,7 @@ impl PluginSourceMapProxy {
         #[cfg(target_arch = "wasm32")]
         {
             let src: Result<String, Box<SpanSnippetError>> =
-                read_returned_result_from_host_fallible(|serialized_ptr| unsafe {
+                read_returned_result_from_host(|serialized_ptr| unsafe {
                     __span_to_source_proxy(sp.lo.0, sp.hi.0, sp.ctxt.as_u32(), serialized_ptr)
                 })
                 .expect("Host should return source code");
@@ -111,15 +111,14 @@ impl SourceMapper for PluginSourceMapProxy {
             } else {
                 0
             };
-            let partial_loc: PartialLoc =
-                read_returned_result_from_host_fallible(|serialized_ptr| unsafe {
-                    __lookup_char_pos_source_map_proxy(
-                        pos.0,
-                        should_request_source_file,
-                        serialized_ptr,
-                    )
-                })
-                .expect("Host should return PartialLoc");
+            let partial_loc: PartialLoc = read_returned_result_from_host(|serialized_ptr| unsafe {
+                __lookup_char_pos_source_map_proxy(
+                    pos.0,
+                    should_request_source_file,
+                    serialized_ptr,
+                )
+            })
+            .expect("Host should return PartialLoc");
 
             if self.source_file.get().is_none() {
                 if let Some(source_file) = partial_loc.source_file {
@@ -154,7 +153,7 @@ impl SourceMapper for PluginSourceMapProxy {
                 0
             };
             let partial_files: PartialFileLinesResult =
-                read_returned_result_from_host_fallible(|serialized_ptr| unsafe {
+                read_returned_result_from_host(|serialized_ptr| unsafe {
                     __span_to_lines_proxy(
                         sp.lo.0,
                         sp.hi.0,
@@ -191,7 +190,7 @@ impl SourceMapper for PluginSourceMapProxy {
 
     fn span_to_string(&self, sp: Span) -> String {
         #[cfg(target_arch = "wasm32")]
-        return read_returned_result_from_host_fallible(|serialized_ptr| unsafe {
+        return read_returned_result_from_host(|serialized_ptr| unsafe {
             __span_to_string_proxy(sp.lo.0, sp.hi.0, sp.ctxt.as_u32(), serialized_ptr)
         })
         .expect("Host should return String");
@@ -202,7 +201,7 @@ impl SourceMapper for PluginSourceMapProxy {
 
     fn span_to_filename(&self, sp: Span) -> FileName {
         #[cfg(target_arch = "wasm32")]
-        return read_returned_result_from_host_fallible(|serialized_ptr| unsafe {
+        return read_returned_result_from_host(|serialized_ptr| unsafe {
             __span_to_filename_proxy(sp.lo.0, sp.hi.0, sp.ctxt.as_u32(), serialized_ptr)
         })
         .expect("Host should return Filename");

crates/swc_plugin_runner/src/imported_fn/source_map.rs

@@ -46,12 +46,17 @@ pub fn lookup_char_pos_proxy(
     byte_pos: u32,
     should_include_source_file: i32,
     allocated_ret_ptr: u32,
-) -> i32 {
+) -> u32 {
     let memory = env.data().memory.clone();
     let memory = memory
         .as_ref()
         .expect("Memory instance should be available, check initialization");
 
+    let alloc_guest_memory = env.data().alloc_guest_memory.clone();
+    let alloc_guest_memory = alloc_guest_memory
+        .as_ref()
+        .expect("Alloc guest memory fn should be available, check initialization");
+
     let original_loc = (env.data().source_map.lock()).lookup_char_pos(BytePos(byte_pos));
     let ret = PartialLoc {
         source_file: if should_include_source_file == 0 {
@@ -67,18 +72,14 @@ pub fn lookup_char_pos_proxy(
     let serialized_loc_bytes =
         PluginSerializedBytes::try_serialize(&ret).expect("Should be serializable");
 
-    if let Some(alloc_guest_memory) = env.data().alloc_guest_memory.clone().as_ref() {
-        allocate_return_values_into_guest(
-            memory,
-            &mut env.as_store_mut(),
-            alloc_guest_memory,
-            allocated_ret_ptr,
-            &serialized_loc_bytes,
-        );
-        1
-    } else {
-        0
-    }
+    allocate_return_values_into_guest(
+        memory,
+        &mut env.as_store_mut(),
+        alloc_guest_memory,
+        allocated_ret_ptr,
+        &serialized_loc_bytes,
+    );
+    1
 }
 
 #[tracing::instrument(level = "info", skip_all)]

crates/swc_plugin_runner/src/memory_interop.rs

@@ -5,8 +5,9 @@ use wasmer::{Memory, MemoryView, StoreMut, TypedFunction, WasmPtr};
 #[tracing::instrument(level = "info", skip_all)]
 pub fn copy_bytes_into_host(memory: &MemoryView, bytes_ptr: i32, bytes_ptr_len: i32) -> Vec<u8> {
     let ptr: WasmPtr<u8> = WasmPtr::new(bytes_ptr as _);
-    let _derefed_ptr = ptr.deref(memory);
-    let values = ptr.slice(memory, bytes_ptr_len as u32).expect("xxx");
+    let values = ptr
+        .slice(memory, bytes_ptr_len as u32)
+        .expect("Should able to get a slice from memory view");
 
     values
         .read_to_vec()

crates/swc_plugin_runner/tests/fixture/swc_internal_plugin/src/lib.rs

@@ -1,5 +1,5 @@
 use swc_core::{
-    common::DUMMY_SP,
+    common::{SourceMapper, DUMMY_SP},
     ecma::{ast::*, atoms::*, visit::*},
     plugin::{
         errors::HANDLER,
@@ -9,7 +9,9 @@ use swc_core::{
     quote,
 };
 
-struct ConsoleOutputReplacer;
+struct ConsoleOutputReplacer {
+    metadata: TransformPluginProgramMetadata,
+}
 
 /// An example plugin replaces any `console.log(${text})` into
 /// `console.log('changed_via_plugin')`.
@@ -18,6 +20,10 @@ impl VisitMut for ConsoleOutputReplacer {
         if let Callee::Expr(expr) = &call.callee {
             if let Expr::Member(MemberExpr { obj, .. }) = &**expr {
                 if let Expr::Ident(ident) = &**obj {
+                    println!(
+                        "lookup_char_pos {:#?}",
+                        self.metadata.source_map.lookup_char_pos(ident.span.lo)
+                    );
                     if ident.sym == *"console" {
                         call.args[0].expr = Box::new(Expr::Lit(Lit::Str(Str {
                             span: DUMMY_SP,
@@ -101,5 +107,5 @@ pub fn process(program: Program, metadata: TransformPluginProgramMetadata) -> Pr
         panic!("Plugin config should be testValue");
     }
 
-    program.fold_with(&mut as_folder(ConsoleOutputReplacer))
+    program.fold_with(&mut as_folder(ConsoleOutputReplacer { metadata }))
 }