Merge #1284

1284: feat(interface-types) Implement string and memory instructions r=Hywan a=Hywan

Previously, we had `read-utf8` and `write-utf8` instructions. The latest version of the standard proposes respectively `memory-to-string` and `string-to-memory`. The semantics is close though.

Co-authored-by: Ivan Enderlin <ivan.enderlin@hoa-project.net>

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## **[Unreleased]**
 
+- [#1284](https://github.com/wasmerio/wasmer/pull/1284) Implement string and memory instructions in `wasmer-interface-types`
 - [#1272](https://github.com/wasmerio/wasmer/pull/1272) Fix off-by-one error bug when accessing memory with a `WasmPtr` that contains the last valid byte of memory. Also changes the behavior of `WasmPtr<T, Array>` with a length of 0 and `WasmPtr<T>` where `std::mem::size_of::<T>()` is 0 to always return `None`
 
 ## 0.15.0 - 2020-03-04

lib/interface-types/src/ast.rs

@@ -89,12 +89,12 @@ pub struct Export<'input> {
 
 /// Represents an adapter.
 #[derive(PartialEq, Debug)]
-pub struct Adapter<'input> {
+pub struct Adapter {
     /// The adapter function type.
     pub function_type: u32,
 
     /// The instructions.
-    pub instructions: Vec<Instruction<'input>>,
+    pub instructions: Vec<Instruction>,
 }
 
 /// Represents an implementation.
@@ -137,7 +137,7 @@ pub struct Interfaces<'input> {
     pub imports: Vec<Import<'input>>,
 
     /// All the adapters.
-    pub adapters: Vec<Adapter<'input>>,
+    pub adapters: Vec<Adapter>,
 
     /// All the exported functions.
     pub exports: Vec<Export<'input>>,

lib/interface-types/src/decoders/binary.rs

@@ -174,14 +174,14 @@ fn instruction<'input, E: ParseError<&'input [u8]>>(
             )
         }
 
-        0x03 => (input, Instruction::ReadUtf8),
+        0x03 => (input, Instruction::MemoryToString),
 
         0x04 => {
-            consume!((input, argument_0) = string(input)?);
+            consume!((input, argument_0) = uleb(input)?);
             (
                 input,
-                Instruction::WriteUtf8 {
-                    allocator_name: argument_0,
+                Instruction::StringToMemory {
+                    allocator_index: argument_0 as u32,
                 },
             )
         }
@@ -630,8 +630,8 @@ mod tests {
             0x2b, // list of 43 items
             0x00, 0x01, // ArgumentGet { index: 1 }
             0x01, 0x01, // CallCore { function_index: 1 }
-            0x03, // ReadUtf8
-            0x04, 0x03, 0x61, 0x62, 0x63, // WriteUtf8 { allocator_name: "abc" }
+            0x03, // MemoryToString
+            0x04, 0x01, // StringToMemory { allocator_index: 1 }
             0x07, // I32ToS8
             0x08, // I32ToS8X
             0x09, // I32ToU8
@@ -678,10 +678,8 @@ mod tests {
             vec![
                 Instruction::ArgumentGet { index: 1 },
                 Instruction::CallCore { function_index: 1 },
-                Instruction::ReadUtf8,
-                Instruction::WriteUtf8 {
-                    allocator_name: "abc",
-                },
+                Instruction::MemoryToString,
+                Instruction::StringToMemory { allocator_index: 1 },
                 Instruction::I32ToS8,
                 Instruction::I32ToS8X,
                 Instruction::I32ToU8,

lib/interface-types/src/decoders/wat.rs

@@ -28,8 +28,8 @@ mod keyword {
     // Instructions.
     custom_keyword!(argument_get = "arg.get");
     custom_keyword!(call_core = "call-core");
-    custom_keyword!(read_utf8 = "read-utf8");
-    custom_keyword!(write_utf8 = "write-utf8");
+    custom_keyword!(memory_to_string = "memory-to-string");
+    custom_keyword!(string_to_memory = "string-to-memory");
     custom_keyword!(i32_to_s8 = "i32-to-s8");
     custom_keyword!(i32_to_s8x = "i32-to-s8x");
     custom_keyword!(i32_to_u8 = "i32-to-u8");
@@ -137,7 +137,7 @@ impl Parse<'_> for InterfaceType {
     }
 }
 
-impl<'a> Parse<'a> for Instruction<'a> {
+impl<'a> Parse<'a> for Instruction {
     #[allow(clippy::cognitive_complexity)]
     fn parse(parser: Parser<'a>) -> Result<Self> {
         let mut lookahead = parser.lookahead1();
@@ -154,15 +154,15 @@ impl<'a> Parse<'a> for Instruction<'a> {
             Ok(Instruction::CallCore {
                 function_index: parser.parse::<u64>()? as usize,
             })
-        } else if lookahead.peek::<keyword::read_utf8>() {
-            parser.parse::<keyword::read_utf8>()?;
+        } else if lookahead.peek::<keyword::memory_to_string>() {
+            parser.parse::<keyword::memory_to_string>()?;
 
-            Ok(Instruction::ReadUtf8)
-        } else if lookahead.peek::<keyword::write_utf8>() {
-            parser.parse::<keyword::write_utf8>()?;
+            Ok(Instruction::MemoryToString)
+        } else if lookahead.peek::<keyword::string_to_memory>() {
+            parser.parse::<keyword::string_to_memory>()?;
 
-            Ok(Instruction::WriteUtf8 {
-                allocator_name: parser.parse()?,
+            Ok(Instruction::StringToMemory {
+                allocator_index: parser.parse()?,
             })
         } else if lookahead.peek::<keyword::i32_to_s8>() {
             parser.parse::<keyword::i32_to_s8>()?;
@@ -392,7 +392,7 @@ impl Parse<'_> for FunctionType {
 enum Interface<'a> {
     Type(Type),
     Import(Import<'a>),
-    Adapter(Adapter<'a>),
+    Adapter(Adapter),
     Export(Export<'a>),
     Implementation(Implementation),
 }
@@ -520,7 +520,7 @@ impl<'a> Parse<'a> for Implementation {
     }
 }
 
-impl<'a> Parse<'a> for Adapter<'a> {
+impl<'a> Parse<'a> for Adapter {
     fn parse(parser: Parser<'a>) -> Result<Self> {
         parser.parse::<keyword::func>()?;
 
@@ -667,8 +667,8 @@ mod tests {
         let inputs = vec![
             "arg.get 7",
             "call-core 7",
-            "read-utf8",
-            r#"write-utf8 "foo""#,
+            "memory-to-string",
+            "string-to-memory 42",
             "i32-to-s8",
             "i32-to-s8x",
             "i32-to-u8",
@@ -712,9 +712,9 @@ mod tests {
         let outputs = vec![
             Instruction::ArgumentGet { index: 7 },
             Instruction::CallCore { function_index: 7 },
-            Instruction::ReadUtf8,
-            Instruction::WriteUtf8 {
-                allocator_name: "foo",
+            Instruction::MemoryToString,
+            Instruction::StringToMemory {
+                allocator_index: 42,
             },
             Instruction::I32ToS8,
             Instruction::I32ToS8X,

lib/interface-types/src/encoders/binary.rs

@@ -162,7 +162,7 @@ where
 /// Encode an `Adapter` into bytes.
 ///
 /// Decoder is in `decoders::binary::adapters`.
-impl<W> ToBytes<W> for Adapter<'_>
+impl<W> ToBytes<W> for Adapter
 where
     W: Write,
 {
@@ -244,7 +244,7 @@ where
 /// Encode an `Instruction` into bytes.
 ///
 /// Decoder is `decoders::binary::instruction`.
-impl<W> ToBytes<W> for Instruction<'_>
+impl<W> ToBytes<W> for Instruction
 where
     W: Write,
 {
@@ -260,11 +260,11 @@ where
                 (*function_index as u64).to_bytes(writer)?;
             }
 
-            Instruction::ReadUtf8 => 0x03_u8.to_bytes(writer)?,
+            Instruction::MemoryToString => 0x03_u8.to_bytes(writer)?,
 
-            Instruction::WriteUtf8 { allocator_name } => {
+            Instruction::StringToMemory { allocator_index } => {
                 0x04_u8.to_bytes(writer)?;
-                allocator_name.to_bytes(writer)?;
+                (*allocator_index as u64).to_bytes(writer)?;
             }
 
             Instruction::I32ToS8 => 0x07_u8.to_bytes(writer)?,
@@ -550,10 +550,8 @@ mod tests {
             vec![
                 Instruction::ArgumentGet { index: 1 },
                 Instruction::CallCore { function_index: 1 },
-                Instruction::ReadUtf8,
-                Instruction::WriteUtf8 {
-                    allocator_name: "abc",
-                },
+                Instruction::MemoryToString,
+                Instruction::StringToMemory { allocator_index: 1 },
                 Instruction::I32ToS8,
                 Instruction::I32ToS8X,
                 Instruction::I32ToU8,
@@ -598,8 +596,8 @@ mod tests {
                 0x2b, // list of 43 items
                 0x00, 0x01, // ArgumentGet { index: 1 }
                 0x01, 0x01, // CallCore { function_index: 1 }
-                0x03, // ReadUtf8
-                0x04, 0x03, 0x61, 0x62, 0x63, // WriteUtf8 { allocator_name: "abc" }
+                0x03, // MemoryToString
+                0x04, 0x01, // StringToMemory { allocator_index: 1 }
                 0x07, // I32ToS8
                 0x08, // I32ToS8X
                 0x09, // I32ToU8

lib/interface-types/src/encoders/wat.rs

@@ -80,14 +80,14 @@ impl ToString for &InterfaceType {
 }
 
 /// Encode an `Instruction` into a string.
-impl<'input> ToString for &Instruction<'input> {
+impl ToString for &Instruction {
     fn to_string(&self) -> String {
         match self {
             Instruction::ArgumentGet { index } => format!("arg.get {}", index),
             Instruction::CallCore { function_index } => format!("call-core {}", function_index),
-            Instruction::ReadUtf8 => "read-utf8".into(),
-            Instruction::WriteUtf8 { allocator_name } => {
-                format!(r#"write-utf8 "{}""#, allocator_name)
+            Instruction::MemoryToString => "memory-to-string".into(),
+            Instruction::StringToMemory { allocator_index } => {
+                format!(r#"string-to-memory {}"#, allocator_index)
             }
             Instruction::I32ToS8 => "i32-to-s8".into(),
             Instruction::I32ToS8X => "i32-to-s8x".into(),
@@ -194,7 +194,7 @@ impl<'input> ToString for &Import<'input> {
 }
 
 /// Encode an `Adapter` into a string.
-impl<'input> ToString for &Adapter<'input> {
+impl ToString for &Adapter {
     fn to_string(&self) -> String {
         format!(
             r#"(@interface func (type {function_type}){instructions})"#,
@@ -361,9 +361,9 @@ mod tests {
         let inputs: Vec<String> = vec![
             (&Instruction::ArgumentGet { index: 7 }).to_string(),
             (&Instruction::CallCore { function_index: 7 }).to_string(),
-            (&Instruction::ReadUtf8).to_string(),
-            (&Instruction::WriteUtf8 {
-                allocator_name: "foo",
+            (&Instruction::MemoryToString).to_string(),
+            (&Instruction::StringToMemory {
+                allocator_index: 42,
             })
                 .to_string(),
             (&Instruction::I32ToS8).to_string(),
@@ -409,8 +409,8 @@ mod tests {
         let outputs = vec![
             "arg.get 7",
             "call-core 7",
-            "read-utf8",
-            r#"write-utf8 "foo""#,
+            "memory-to-string",
+            "string-to-memory 42",
             "i32-to-s8",
             "i32-to-s8x",
             "i32-to-u8",

lib/interface-types/src/interpreter/instruction.rs

@@ -2,7 +2,7 @@
 
 /// Represents all the possible WIT instructions.
 #[derive(PartialEq, Debug)]
-pub enum Instruction<'input> {
+pub enum Instruction {
     /// The `arg.get` instruction.
     ArgumentGet {
         /// The argument index.
@@ -15,13 +15,13 @@ pub enum Instruction<'input> {
         function_index: usize,
     },
 
-    /// The `read-utf8` instruction.
-    ReadUtf8,
+    /// The `memory-to-string` instruction.
+    MemoryToString,
 
-    /// The `write-utf8` instruction.
-    WriteUtf8 {
-        /// The allocator function name.
-        allocator_name: &'input str,
+    /// The `string-to-memory` instruction.
+    StringToMemory {
+        /// The allocator function index.
+        allocator_index: u32,
     },
 
     /// The `i32-to-s8,` instruction.

lib/interface-types/src/interpreter/instructions/read_utf8.rs → lib/interface-types/src/interpreter/instructions/memory_to_string.rs

@@ -2,7 +2,7 @@ use crate::interpreter::wasm::values::InterfaceValue;
 use std::{cell::Cell, convert::TryFrom};
 
 executable_instruction!(
-    read_utf8(instruction_name: String) -> _ {
+    memory_to_string(instruction_name: String) -> _ {
         move |runtime| -> _ {
             match runtime.stack.pop(2) {
                 Some(inputs) => match runtime.wasm_instance.memory(0) {
@@ -55,11 +55,11 @@ executable_instruction!(
 #[cfg(test)]
 mod tests {
     test_executable_instruction!(
-        test_read_utf8 =
+        test_memory_to_string =
             instructions: [
                 Instruction::ArgumentGet { index: 1 },
                 Instruction::ArgumentGet { index: 0 },
-                Instruction::ReadUtf8,
+                Instruction::MemoryToString,
             ],
             invocation_inputs: [
                 InterfaceValue::I32(13),
@@ -75,11 +75,11 @@ mod tests {
     );
 
     test_executable_instruction!(
-        test_read_utf8__read_out_of_memory =
+        test_memory_to_string__read_out_of_memory =
             instructions: [
                 Instruction::ArgumentGet { index: 1 },
                 Instruction::ArgumentGet { index: 0 },
-                Instruction::ReadUtf8,
+                Instruction::MemoryToString,
             ],
             invocation_inputs: [
                 InterfaceValue::I32(13),
@@ -91,15 +91,15 @@ mod tests {
                 memory: Memory::new("Hello!".as_bytes().iter().map(|u| Cell::new(*u)).collect()),
                 ..Default::default()
             },
-            error: r#"`read-utf8` failed because it has to read out of the memory bounds (index 13 > memory length 6)."#,
+            error: r#"`memory-to-string` failed because it has to read out of the memory bounds (index 13 > memory length 6)."#,
     );
 
     test_executable_instruction!(
-        test_read_utf8__invalid_encoding =
+        test_memory_to_string__invalid_encoding =
             instructions: [
                 Instruction::ArgumentGet { index: 1 },
                 Instruction::ArgumentGet { index: 0 },
-                Instruction::ReadUtf8,
+                Instruction::MemoryToString,
             ],
             invocation_inputs: [
                 InterfaceValue::I32(4),
@@ -111,21 +111,21 @@ mod tests {
                 memory: Memory::new(vec![0, 159, 146, 150].iter().map(|b| Cell::new(*b)).collect::<Vec<Cell<u8>>>()),
                 ..Default::default()
             },
-            error: r#"`read-utf8` failed because the read string isn't UTF-8 valid (invalid utf-8 sequence of 1 bytes from index 1)."#,
+            error: r#"`memory-to-string` failed because the read string isn't UTF-8 valid (invalid utf-8 sequence of 1 bytes from index 1)."#,
     );
 
     test_executable_instruction!(
-        test_read_utf8__stack_is_too_small =
+        test_memory_to_string__stack_is_too_small =
             instructions: [
                 Instruction::ArgumentGet { index: 0 },
-                Instruction::ReadUtf8,
-                //           ^^^^^^^^ `read-utf8` expects 2 values on the stack, only one is present.
+                Instruction::MemoryToString,
+                //           ^^^^^^^^^^^^^^ `memory-to-string` expects 2 values on the stack, only one is present.
             ],
             invocation_inputs: [
                 InterfaceValue::I32(13),
                 InterfaceValue::I32(0),
             ],
             instance: Instance::new(),
-            error: r#"`read-utf8` failed because there is not enough data on the stack (needs 2)."#,
+            error: r#"`memory-to-string` failed because there is not enough data on the stack (needs 2)."#,
     );
 }