Gate WebAssembly validation behind a new 'validate' feature (#1532)

* Gate WebAssembly validation behind a new 'validate' feature

This commit moves all of the validation logic of `wasmparser` behind a
new Cargo features named `validate`. This cuts down the dependency tree
for users who only need to parse wasm files. For example `wasmprinter`
doesn't need to validate, it only needs to parse.

Closes #1528

* Fix wasm-smith build and tests

Author: alexcrichton

.github/workflows/main.yml

@@ -190,6 +190,7 @@ jobs:
       - run: cargo check --no-default-features -p wasmparser
       - run: cargo check --no-default-features -p wasmparser --target x86_64-unknown-none
       - run: cargo check --no-default-features -p wasmparser --features std
+      - run: cargo check --no-default-features -p wasmparser --features validate
       - run: |
           if cargo tree -p wasm-smith --no-default-features -e no-dev | grep wasmparser; then
             echo wasm-smith without default features should not depend on wasmparser

Cargo.toml

@@ -68,7 +68,7 @@ wasm-metadata = { version = "0.206.0", path = "crates/wasm-metadata" }
 wasm-mutate = { version = "0.206.0", path = "crates/wasm-mutate" }
 wasm-shrink = { version = "0.206.0", path = "crates/wasm-shrink" }
 wasm-smith = { version = "0.206.0", path = "crates/wasm-smith" }
-wasmparser = { version = "0.206.0", path = "crates/wasmparser" }
+wasmparser = { version = "0.206.0", path = "crates/wasmparser", default-features = false, features = ['std'] }
 wasmprinter = { version = "0.206.0", path = "crates/wasmprinter" }
 wast = { version = "206.0.0", path = "crates/wast" }
 wat = { version = "1.206.0", path = "crates/wat" }
@@ -87,7 +87,7 @@ wat = { workspace = true }
 termcolor = "1.2.0"
 
 # Dependencies of `validate`
-wasmparser = { workspace = true, optional = true }
+wasmparser = { workspace = true, optional = true, features = ['validate'] }
 rayon = { workspace = true, optional = true }
 
 # Dependencies of `print`

crates/wasm-compose/Cargo.toml

@@ -16,7 +16,7 @@ workspace = true
 [dependencies]
 wat = { workspace = true }
 wasm-encoder = { workspace = true, features = ['wasmparser'] }
-wasmparser = { workspace = true }
+wasmparser = { workspace = true, features = ['validate'] }
 indexmap = { workspace = true, features = ["serde"] }
 anyhow = { workspace = true }
 serde = { workspace = true }

crates/wasm-shrink/Cargo.toml

@@ -20,7 +20,7 @@ log = { workspace = true }
 rand = { workspace = true }
 clap = { workspace = true, optional = true }
 wasm-mutate = { workspace = true }
-wasmparser = { workspace = true }
+wasmparser = { workspace = true, features = ['validate'] }
 
 [dev-dependencies]
 env_logger = { workspace = true }

crates/wasm-smith/Cargo.toml

@@ -28,7 +28,7 @@ leb128 = { workspace = true }
 serde = { workspace = true, optional = true }
 serde_derive = { workspace = true, optional = true }
 wasm-encoder = { workspace = true }
-wasmparser = { workspace = true, optional = true }
+wasmparser = { workspace = true, optional = true, features = ['validate'] }
 wat = { workspace = true, optional = true }
 
 [dev-dependencies]

crates/wasmparser/Cargo.toml

@@ -17,10 +17,10 @@ workspace = true
 
 [dependencies]
 bitflags = "2.4.1"
-indexmap = { workspace = true }
-semver = { workspace = true }
-hashbrown = { workspace = true }
-ahash = { workspace = true }
+indexmap = { workspace = true, optional = true }
+semver = { workspace = true, optional = true }
+hashbrown = { workspace = true, optional = true }
+ahash = { workspace = true, optional = true }
 
 [dev-dependencies]
 anyhow = { workspace = true }
@@ -38,5 +38,19 @@ name = "benchmark"
 harness = false
 
 [features]
-default = ['std']
+default = ['std', 'validate']
+
+# A feature which enables implementations of `std::error::Error` as appropriate
+# along with other convenience APIs. This additionally uses the standard
+# library's source of randomness for seeding hash maps.
 std = ['indexmap/std']
+
+# A feature that enables validating WebAssembly files. This is enabled by
+# default but not required if you're only parsing a file, for example, as
+# opposed to validating all of its contents.
+validate = [
+  'dep:indexmap',
+  'dep:semver',
+  'dep:hashbrown',
+  'dep:ahash',
+]

crates/wasmparser/src/binary_reader.rs

@@ -93,6 +93,7 @@ impl BinaryReaderError {
         self.inner.offset
     }
 
+    #[cfg(feature = "validate")]
     pub(crate) fn add_context(&mut self, mut context: String) {
         context.push_str("\n");
         self.inner.message.insert_str(0, &context);

crates/wasmparser/src/lib.rs

@@ -28,6 +28,7 @@
 
 #![deny(missing_docs)]
 #![no_std]
+#![cfg_attr(docsrs, feature(doc_auto_cfg))]
 
 extern crate alloc;
 #[cfg(feature = "std")]
@@ -48,6 +49,7 @@ mod prelude {
     pub use alloc::vec;
     pub use alloc::vec::Vec;
 
+    #[cfg(feature = "validate")]
     pub use crate::map::{HashMap, HashSet, IndexMap, IndexSet};
 }
 
@@ -784,14 +786,20 @@ macro_rules! bail {
 pub use crate::binary_reader::{BinaryReader, BinaryReaderError, Result};
 pub use crate::parser::*;
 pub use crate::readers::*;
-pub use crate::resources::*;
-pub use crate::validator::*;
 
 mod binary_reader;
 mod limits;
 mod parser;
 mod readers;
+
+#[cfg(feature = "validate")]
 mod resources;
+#[cfg(feature = "validate")]
 mod validator;
+#[cfg(feature = "validate")]
+pub use crate::resources::*;
+#[cfg(feature = "validate")]
+pub use crate::validator::*;
 
+#[cfg(feature = "validate")]
 pub mod map;

crates/wasmparser/src/limits.rs

@@ -13,6 +13,8 @@
  * limitations under the License.
  */
 
+#![cfg_attr(not(feature = "validate"), allow(dead_code))]
+
 // The following limits are imposed by wasmparser on WebAssembly modules.
 // The limits are agreed upon with other engines for consistency.
 pub const MAX_WASM_TYPES: usize = 1_000_000;

crates/wasmparser/src/readers/component/types.rs

@@ -194,6 +194,7 @@ impl PrimitiveValType {
         })
     }
 
+    #[cfg(feature = "validate")]
     pub(crate) fn contains_ptr(&self) -> bool {
         matches!(self, Self::String)
     }