Update wasmparser to 0.59 (#172)

* Update how we parse modules at a top-level
* Lean on wasmparser for module structure validation
* Sync with wasmparser's implemented opcodes and such

Cargo.toml

@@ -29,7 +29,7 @@ leb128 = "0.2.4"
 log = "0.4.8"
 rayon = { version = "1.1.0", optional = true }
 walrus-macro = { path = './crates/macro', version = '=0.17.0' }
-wasmparser = "0.55.0"
+wasmparser = "0.59.0"
 
 [features]
 parallel = ['rayon', 'id-arena/rayon']

crates/fuzz-utils/Cargo.toml

@@ -10,7 +10,7 @@ anyhow = "1.0"
 env_logger = "0.7.0"
 rand = { version = "0.7.0", features = ['small_rng'] }
 tempfile = "3.1.0"
-wasmparser = "0.55"
+wasmparser = "0.59"
 wat = "1.0"
 
 [dependencies.walrus]

crates/fuzz-utils/src/lib.rs

@@ -417,7 +417,7 @@ impl TestCaseGenerator for WasmOptTtf {
 
             // Only generate programs that wat2wasm can handle.
             if let Ok(bytes) = wat::parse_bytes(&wat) {
-                if wasmparser::validate(&bytes, None).is_ok() {
+                if wasmparser::validate(&bytes).is_ok() {
                     return String::from_utf8(wat).unwrap();
                 }
             }

crates/tests/tests/round_trip/anyref1.wat

@@ -1,6 +1,6 @@
 (module
-  (import "x" "y" (func (param anyref)))
-  (func (export "a") (param anyref)
+  (import "x" "y" (func (param externref)))
+  (func (export "a") (param externref)
     local.get 0
     call 0))
 

crates/tests/tests/round_trip/anyref2.wat

@@ -1,6 +1,6 @@
 (module
-  (table 1 anyref)
-  (func (export "a") (param i32) (result anyref)
+  (table 1 externref)
+  (func (export "a") (param i32) (result externref)
     local.get 0
     table.get 0))
 

crates/tests/tests/round_trip/anyref3.wat

@@ -1,8 +1,8 @@
 (module
   (type (func))
-  (table 1 anyfunc)
-  (table 1 anyref)
-  (func (export "a") (param i32) (result anyref)
+  (table 1 funcref)
+  (table 1 externref)
+  (func (export "a") (param i32) (result externref)
     local.get 0
     call_indirect (type 0)
     local.get 0

src/ir/mod.rs

@@ -527,12 +527,8 @@ pub enum Instr {
         ty: ValType,
     },
 
-    /// `ref.is_null $ty`
-    RefIsNull {
-        /// The type of value we're testing for null
-        #[walrus(skip_visit)]
-        ty: ValType,
-    },
+    /// `ref.is_null`
+    RefIsNull {},
 
     /// `ref.func`
     RefFunc {
@@ -946,14 +942,17 @@ pub enum UnaryOp {
     I8x16Neg,
     I8x16AnyTrue,
     I8x16AllTrue,
+    I8x16Bitmask,
     I16x8Abs,
     I16x8Neg,
     I16x8AnyTrue,
     I16x8AllTrue,
+    I16x8Bitmask,
     I32x4Abs,
     I32x4Neg,
     I32x4AnyTrue,
     I32x4AllTrue,
+    I32x4Bitmask,
     I64x2Neg,
 
     F32x4Abs,

src/module/data.rs

@@ -185,20 +185,15 @@ impl Module {
         &mut self,
         section: wasmparser::DataSectionReader,
         ids: &IndicesToIds,
-        data_count: Option<u32>,
     ) -> Result<()> {
         log::debug!("parse data section");
-        if let Some(count) = data_count {
-            if count != section.get_count() {
-                bail!("data count section mismatches actual data section");
-            }
-        }
+        let preallocated = self.data.arena.len() > 0;
         for (i, segment) in section.into_iter().enumerate() {
             let segment = segment?;
 
             // If we had the `DataCount` section, then we already pre-allocated
             // a data segment. Otherwise, allocate one now.
-            let id = if data_count.is_some() {
+            let id = if preallocated {
                 ids.get_data(i as u32)?
             } else {
                 self.data.arena.alloc_with_id(|id| Data {

src/module/exports.rs

@@ -142,6 +142,9 @@ impl Module {
                 Table => ExportItem::Table(ids.get_table(entry.index)?),
                 Memory => ExportItem::Memory(ids.get_memory(entry.index)?),
                 Global => ExportItem::Global(ids.get_global(entry.index)?),
+                Type | Module | Instance => {
+                    unimplemented!("module linking not supported");
+                }
             };
             self.exports.arena.alloc_with_id(|id| Export {
                 id,

src/module/functions/local_function/emit.rs

@@ -498,11 +498,13 @@ impl<'instr> Visitor<'instr> for Emit<'_, '_> {
                     I8x16Neg => self.simd(0x61),
                     I8x16AnyTrue => self.simd(0x62),
                     I8x16AllTrue => self.simd(0x63),
+                    I8x16Bitmask => self.simd(0x64),
 
                     I16x8Abs => self.simd(0x80),
                     I16x8Neg => self.simd(0x81),
                     I16x8AnyTrue => self.simd(0x82),
                     I16x8AllTrue => self.simd(0x83),
+                    I16x8Bitmask => self.simd(0x84),
                     I16x8WidenLowI8x16S => self.simd(0x87),
                     I16x8WidenHighI8x16S => self.simd(0x88),
                     I16x8WidenLowI8x16U => self.simd(0x89),
@@ -512,6 +514,7 @@ impl<'instr> Visitor<'instr> for Emit<'_, '_> {
                     I32x4Neg => self.simd(0xa1),
                     I32x4AnyTrue => self.simd(0xa2),
                     I32x4AllTrue => self.simd(0xa3),
+                    I32x4Bitmask => self.simd(0xa4),
                     I32x4WidenLowI16x8S => self.simd(0xa7),
                     I32x4WidenHighI16x8S => self.simd(0xa8),
                     I32x4WidenLowI16x8U => self.simd(0xa9),
@@ -802,9 +805,8 @@ impl<'instr> Visitor<'instr> for Emit<'_, '_> {
                 self.encoder.byte(0xd0);
                 e.ty.emit(self.encoder);
             }
-            RefIsNull(e) => {
+            RefIsNull(_e) => {
                 self.encoder.byte(0xd1);
-                e.ty.emit(self.encoder);
             }
             RefFunc(e) => {
                 self.encoder.byte(0xd2);

src/module/functions/local_function/mod.rs

@@ -1198,10 +1198,9 @@ fn validate_instruction<'context>(
             ctx.alloc_instr(RefNull { ty }, loc);
             ctx.push_operand(Some(ty));
         }
-        Operator::RefIsNull { ty } => {
-            let ty = ValType::parse(&ty)?;
-            ctx.pop_operand_expected(Some(ty))?;
-            ctx.alloc_instr(RefIsNull { ty }, loc);
+        Operator::RefIsNull => {
+            // ctx.pop_operand_expected(Some(ty))?;
+            ctx.alloc_instr(RefIsNull {}, loc);
             ctx.push_operand(Some(I32));
         }
         Operator::RefFunc { function_index } => {
@@ -1462,6 +1461,10 @@ fn validate_instruction<'context>(
         Operator::I32x4MaxS => binop(ctx, V128, BinaryOp::I32x4MaxS)?,
         Operator::I32x4MaxU => binop(ctx, V128, BinaryOp::I32x4MaxU)?,
 
+        Operator::I8x16Bitmask => one_op(ctx, V128, I32, UnaryOp::I8x16Bitmask)?,
+        Operator::I16x8Bitmask => one_op(ctx, V128, I32, UnaryOp::I16x8Bitmask)?,
+        Operator::I32x4Bitmask => one_op(ctx, V128, I32, UnaryOp::I32x4Bitmask)?,
+
         Operator::TableCopy {
             src_table,
             dst_table,
@@ -1497,6 +1500,10 @@ fn validate_instruction<'context>(
             let elem = ctx.indices.get_element(segment)?;
             ctx.alloc_instr(ElemDrop { elem }, loc);
         }
+
+        Operator::ReturnCall { .. } | Operator::ReturnCallIndirect { .. } => {
+            unimplemented!("not supported");
+        }
     }
     Ok(())
 }

src/module/functions/mod.rs

@@ -12,8 +12,8 @@ use crate::parse::IndicesToIds;
 use crate::tombstone_arena::{Id, Tombstone, TombstoneArena};
 use crate::ty::TypeId;
 use crate::ty::ValType;
-use anyhow::bail;
 use std::cmp;
+use wasmparser::{FuncValidator, FunctionBody, OperatorsReader};
 
 #[cfg(feature = "parallel")]
 use rayon::prelude::*;
@@ -323,25 +323,19 @@ impl Module {
     /// Add the locally defined functions in the wasm module to this instance.
     pub(crate) fn parse_local_functions(
         &mut self,
-        section: wasmparser::CodeSectionReader,
-        function_section_count: u32,
+        functions: Vec<(FunctionBody<'_>, FuncValidator, OperatorsReader<'_>)>,
         indices: &mut IndicesToIds,
         on_instr_pos: Option<&(dyn Fn(&usize) -> InstrLocId + Sync + Send + 'static)>,
     ) -> Result<()> {
         log::debug!("parse code section");
-        let amt = section.get_count();
-        if amt != function_section_count {
-            bail!("code and function sections must have same number of entries")
-        }
-        let num_imports = self.funcs.arena.len() - (amt as usize);
+        let num_imports = self.funcs.arena.len() - functions.len();
 
         // First up serially create corresponding `LocalId` instances for all
         // functions as well as extract the operators parser for each function.
         // This is pretty tough to parallelize, but we can look into it later if
         // necessary and it's a bottleneck!
-        let mut bodies = Vec::with_capacity(amt as usize);
-        for (i, body) in section.into_iter().enumerate() {
-            let body = body?;
+        let mut bodies = Vec::with_capacity(functions.len());
+        for (i, (body, _validator, ops)) in functions.into_iter().enumerate() {
             let index = (num_imports + i) as u32;
             let id = indices.get_func(index)?;
             let ty = match self.funcs.arena[id].kind {
@@ -370,19 +364,7 @@ impl Module {
             let results = type_.results().to_vec();
             self.types.add_entry_ty(&results);
 
-            // WebAssembly local indices are 32 bits, so it's a validation error to
-            // have more than 2^32 locals. Sure enough there's a spec test for this!
-            let mut total = 0u32;
-            for local in body.get_locals_reader()? {
-                let (count, _) = local?;
-                total = match total.checked_add(count) {
-                    Some(n) => n,
-                    None => bail!("can't have more than 2^32 locals"),
-                };
-            }
-
-            // Now that we know we have a reasonable amount of locals, put them in
-            // our map.
+            // Next up comes all the locals of the function.
             for local in body.get_locals_reader()? {
                 let (count, ty) = local?;
                 let ty = ValType::parse(&ty)?;
@@ -396,8 +378,7 @@ impl Module {
                 }
             }
 
-            let body = body.get_operators_reader()?;
-            bodies.push((id, body, args, ty));
+            bodies.push((id, ops, args, ty));
         }
 
         // Wasm modules can often have a lot of functions and this operation can

src/module/imports.rs

@@ -117,14 +117,18 @@ impl Module {
             match entry.ty {
                 wasmparser::ImportSectionEntryType::Function(idx) => {
                     let ty = ids.get_type(idx)?;
-                    let id = self.add_import_func(entry.module, entry.field, ty);
+                    let id = self.add_import_func(
+                        entry.module,
+                        entry.field.expect("module linking not supported"),
+                        ty,
+                    );
                     ids.push_func(id.0);
                 }
                 wasmparser::ImportSectionEntryType::Table(t) => {
                     let ty = ValType::parse(&t.element_type)?;
                     let id = self.add_import_table(
                         entry.module,
-                        entry.field,
+                        entry.field.expect("module linking not supported"),
                         t.limits.initial,
                         t.limits.maximum,
                         ty,
@@ -134,7 +138,7 @@ impl Module {
                 wasmparser::ImportSectionEntryType::Memory(m) => {
                     let id = self.add_import_memory(
                         entry.module,
-                        entry.field,
+                        entry.field.expect("module linking not supported"),
                         m.shared,
                         m.limits.initial,
                         m.limits.maximum,
@@ -144,12 +148,16 @@ impl Module {
                 wasmparser::ImportSectionEntryType::Global(g) => {
                     let id = self.add_import_global(
                         entry.module,
-                        entry.field,
+                        entry.field.expect("module linking not supported"),
                         ValType::parse(&g.content_type)?,
                         g.mutable,
                     );
                     ids.push_global(id.0);
                 }
+                wasmparser::ImportSectionEntryType::Module(_)
+                | wasmparser::ImportSectionEntryType::Instance(_) => {
+                    unimplemented!("module linking not implemented");
+                }
             }
         }