feat(rust) add native link option

Author: BjornTheProgrammer

crates/core/src/lib.rs

@@ -14,6 +14,7 @@ mod path;
 pub use path::name_package_module;
 mod async_;
 pub use async_::AsyncFilterSet;
+pub mod symbol_name;
 
 #[derive(Default, Copy, Clone, PartialEq, Eq, Debug)]
 pub enum Direction {

crates/cpp/src/symbol_name.rs crates/core/src/symbol_name.rscrates/cpp/src/symbol_name.rs renamed to crates/core/src/symbol_name.rs

@@ -1,4 +1,4 @@
-use wit_bindgen_core::abi;
+use crate::abi;
 
 fn hexdigit(v: u32) -> char {
     if v < 10 {

crates/cpp/src/lib.rs

@@ -9,12 +9,13 @@ use std::{
     process::{Command, Stdio},
     str::FromStr,
 };
-use symbol_name::{make_external_component, make_external_symbol};
 use wit_bindgen_c::to_c_ident;
 use wit_bindgen_core::{
     Files, InterfaceGenerator, Source, Types, WorldGenerator,
     abi::{self, AbiVariant, Bindgen, Bitcast, LiftLower, WasmSignature, WasmType},
-    name_package_module, uwrite, uwriteln,
+    name_package_module,
+    symbol_name::{make_external_component, make_external_symbol},
+    uwrite, uwriteln,
     wit_parser::{
         Alignment, ArchitectureSize, Docs, Function, FunctionKind, Handle, Int, InterfaceId, Param,
         Resolve, SizeAlign, Stability, Type, TypeDef, TypeDefKind, TypeId, TypeOwner, WorldId,
@@ -24,7 +25,6 @@ use wit_bindgen_core::{
 use wit_parser::TypeIdVisitor;
 
 // mod wamr;
-mod symbol_name;
 
 pub const RESOURCE_IMPORT_BASE_CLASS_NAME: &str = "ResourceImportBase";
 pub const RESOURCE_EXPORT_BASE_CLASS_NAME: &str = "ResourceExportBase";

crates/guest-rust/macro/src/lib.rs

@@ -175,6 +175,9 @@ impl Parse for Config {
                     Opt::MergeStructurallyEqualTypes(enable) => {
                         opts.merge_structurally_equal_types = Some(Some(enable.value()))
                     }
+                    Opt::LinkNativeSymbols(enable) => {
+                        opts.link_native_symbols = enable.value();
+                    }
                 }
             }
         } else {
@@ -332,6 +335,7 @@ mod kw {
     syn::custom_keyword!(debug);
     syn::custom_keyword!(enable_method_chaining);
     syn::custom_keyword!(merge_structurally_equal_types);
+    syn::custom_keyword!(link_native_symbols);
 }
 
 #[derive(Clone)]
@@ -416,6 +420,7 @@ enum Opt {
     Debug(syn::LitBool),
     EnableMethodChaining(syn::LitBool),
     MergeStructurallyEqualTypes(syn::LitBool),
+    LinkNativeSymbols(syn::LitBool),
 }
 
 impl Parse for Opt {
@@ -623,6 +628,10 @@ impl Parse for Opt {
             input.parse::<kw::merge_structurally_equal_types>()?;
             input.parse::<Token![:]>()?;
             Ok(Opt::MergeStructurallyEqualTypes(input.parse()?))
+        } else if l.peek(kw::link_native_symbols) {
+            input.parse::<kw::link_native_symbols>()?;
+            input.parse::<Token![:]>()?;
+            Ok(Opt::LinkNativeSymbols(input.parse()?))
         } else {
             Err(l.error())
         }

crates/guest-rust/src/lib.rs

@@ -891,6 +891,24 @@ extern crate std;
 ///     // structurally equal, which is useful when import and export the same
 ///     // interface.
 ///     merge_structurally_equal_types: true,
+///
+///     // Make the same generated bindings usable on a native (non-wasm)
+///     // target as well as on wasm32.
+///     //
+///     // Imports normally compile to `unreachable!()` off wasm32. With this
+///     // enabled each one instead calls through a function pointer that a host
+///     // installs at load time via a generated
+///     // `__wit_bindgen_register_*` symbol, and exports additionally get a
+///     // native symbol whose name encodes the characters a linker cannot
+///     // accept. Both targets still build from one source.
+///     //
+///     // The registration symbols are prefixed with a hex-encoded
+///     // `<package>/<world>` so that two `generate!` invocations in one crate
+///     // don't collide. Binding the *same* world twice in one linkage unit
+///     // still does; use `type_section_suffix` to tell them apart. See
+///     // `wit_bindgen_rust::Opts::link_native_symbols` for the full list of
+///     // symbols a host can expect.
+///     link_native_symbols: true,
 /// });
 /// ```
 ///

crates/guest-rust/src/rt/mod.rs

@@ -153,7 +153,7 @@ pub fn maybe_link_cabi_realloc() {
 /// `cabi_realloc` module above. It's otherwise never explicitly called.
 ///
 /// For more information about this see `./ci/rebuild-libwit-bindgen-cabi.sh`.
-#[cfg(any(target_env = "p1", target_env = ""))]
+#[cfg(any(target_env = "p1", target_env = "", not(target_arch = "wasm32")))]
 pub unsafe fn cabi_realloc(
     old_ptr: *mut u8,
     old_len: usize,

crates/rust/src/bindgen.rs

@@ -67,6 +67,7 @@ impl<'a, 'b> FunctionBindgen<'a, 'b> {
             &rust_name,
             params,
             results,
+            self.r#gen.r#gen.native_symbols(),
         ));
         rust_name
     }

crates/rust/src/interface.rs

@@ -12,7 +12,7 @@ use std::fmt::Write as _;
 use std::mem;
 use wit_bindgen_core::abi::{self, AbiVariant, LiftLower};
 use wit_bindgen_core::{
-    AnonymousTypeGenerator, Source, TypeInfo, dealias, uwrite, uwriteln, wit_parser::*,
+    AnonymousTypeGenerator, Source, TypeInfo, dealias, symbol_name, uwrite, uwriteln, wit_parser::*,
 };
 
 pub struct InterfaceGenerator<'a> {
@@ -212,13 +212,15 @@ impl<'i> InterfaceGenerator<'i> {
                 "new",
                 &[abi::WasmType::Pointer],
                 &[abi::WasmType::I32],
+                self.r#gen.native_symbols(),
             );
             let import_rep = crate::declare_import(
                 &wasm_import_module,
                 &format!("[resource-rep]{resource_name}"),
                 "rep",
                 &[abi::WasmType::I32],
                 &[abi::WasmType::Pointer],
+                self.r#gen.native_symbols(),
             );
             uwriteln!(
                 self.src,
@@ -347,7 +349,6 @@ macro_rules! {macro_name} {{
             };
             self.generate_raw_cabi_export(func, &ty, "$($path_to_types)*", async_);
         }
-        let export_prefix = self.r#gen.opts.export_prefix.as_deref().unwrap_or("");
         for name in resources_to_drop {
             let module = match self.identifier {
                 Identifier::Interface(_, key) => self.resolve.name_world_key(key),
@@ -356,23 +357,25 @@ macro_rules! {macro_name} {{
                 }
             };
             let camel = name.to_upper_camel_case();
-            uwriteln!(
-                self.src,
-                r#"
-                const _: () = {{
-                    #[doc(hidden)]
-                    #[unsafe(export_name = "{export_prefix}{module}#[dtor]{name}")]
-                    #[allow(non_snake_case)]
-                    unsafe extern "C" fn dtor(rep: *mut u8) {{
-                        unsafe {{
-                            $($path_to_types)*::{camel}::dtor::<
-                                <$ty as $($path_to_types)*::Guest>::{camel}
-                            >(rep)
+            for (cfg, symbol) in self.core_export_symbols(&format!("{module}#[dtor]{name}")) {
+                uwriteln!(
+                    self.src,
+                    r#"
+                    const _: () = {{
+                        #[doc(hidden)]
+                        {cfg}#[unsafe(export_name = "{symbol}")]
+                        #[allow(non_snake_case)]
+                        unsafe extern "C" fn dtor(rep: *mut u8) {{
+                            unsafe {{
+                                $($path_to_types)*::{camel}::dtor::<
+                                    <$ty as $($path_to_types)*::Guest>::{camel}
+                                >(rep)
+                            }}
                         }}
-                    }}
-                }};
-                "#
-            );
+                    }};
+                    "#
+                );
+            }
         }
         uwriteln!(self.src, "}};);");
         uwriteln!(self.src, "}}");
@@ -1019,6 +1022,7 @@ fn abi_layout(&mut self) -> ::core::alloc::Layout {{
             "call",
             &sig.params,
             &sig.results,
+            self.r#gen.native_symbols(),
         );
         let mut args = String::new();
         for i in 0..params_lower.len() {
@@ -1281,60 +1285,93 @@ unsafe fn call_import(&mut self, _params: Self::ParamsLower, _results: *mut u8)
             Identifier::World(_) => None,
             Identifier::StreamOrFuturePayload => unreachable!(),
         };
-        let export_prefix = self.r#gen.opts.export_prefix.as_deref().unwrap_or("");
         let export_name = func.legacy_core_export_name(wasm_module_export_name.as_deref());
         let export_name = if async_ {
             format!("[async-lift]{export_name}")
         } else {
             export_name.to_string()
         };
-        uwrite!(
-            self.src,
-            "\
-                #[unsafe(export_name = \"{export_prefix}{export_name}\")]
-                unsafe extern \"C\" fn export_{name_snake}\
-",
-        );
 
-        let params = self.print_export_sig(func, async_);
-        self.push_str(" {\n");
-        uwriteln!(
-            self.src,
-            "unsafe {{ {path_to_self}::_export_{name_snake}_cabi::<{ty}>({}) }}",
-            params.join(", ")
-        );
-        self.push_str("}\n");
-
-        let export_prefix = self.r#gen.opts.export_prefix.as_deref().unwrap_or("");
-        if async_ {
+        for (cfg, symbol) in self.core_export_symbols(&export_name) {
             uwrite!(
                 self.src,
                 "\
-                    #[unsafe(export_name = \"{export_prefix}[callback]{export_name}\")]
-                    unsafe extern \"C\" fn _callback_{name_snake}(event0: u32, event1: u32, event2: u32) -> u32 {{
-                        unsafe {{
-                            {path_to_self}::__callback_{name_snake}(event0, event1, event2)
-                        }}
-                    }}
-                "
-            );
-        } else if abi::guest_export_needs_post_return(self.resolve, func) {
-            uwrite!(
-                self.src,
-                "\
-                    #[unsafe(export_name = \"{export_prefix}cabi_post_{export_name}\")]
-                    unsafe extern \"C\" fn _post_return_{name_snake}\
-"
+                    {cfg}#[unsafe(export_name = \"{symbol}\")]
+                    unsafe extern \"C\" fn export_{name_snake}\
+",
             );
-            let params = self.print_post_return_sig(func);
-            self.src.push_str("{\n");
+            let params = self.print_export_sig(func, async_);
+            self.push_str(" {\n");
             uwriteln!(
                 self.src,
-                "unsafe {{ {path_to_self}::__post_return_{name_snake}::<{ty}>({}) }}",
+                "unsafe {{ {path_to_self}::_export_{name_snake}_cabi::<{ty}>({}) }}",
                 params.join(", ")
             );
-            self.src.push_str("}\n");
+            self.push_str("}\n");
+        }
+
+        if async_ {
+            for (cfg, symbol) in self.core_export_symbols(&format!("[callback]{export_name}")) {
+                uwrite!(
+                    self.src,
+                    "\
+                        {cfg}#[unsafe(export_name = \"{symbol}\")]
+                        unsafe extern \"C\" fn _callback_{name_snake}(event0: u32, event1: u32, event2: u32) -> u32 {{
+                            unsafe {{
+                                {path_to_self}::__callback_{name_snake}(event0, event1, event2)
+                            }}
+                        }}
+                    "
+                );
+            }
+        } else if abi::guest_export_needs_post_return(self.resolve, func) {
+            for (cfg, symbol) in self.core_export_symbols(&format!("cabi_post_{export_name}")) {
+                uwrite!(
+                    self.src,
+                    "\
+                        {cfg}#[unsafe(export_name = \"{symbol}\")]
+                        unsafe extern \"C\" fn _post_return_{name_snake}\
+"
+                );
+                let params = self.print_post_return_sig(func);
+                self.src.push_str("{\n");
+                uwriteln!(
+                    self.src,
+                    "unsafe {{ {path_to_self}::__post_return_{name_snake}::<{ty}>({}) }}",
+                    params.join(", ")
+                );
+                self.src.push_str("}\n");
+            }
+        }
+    }
+
+    /// Returns each copy of a core export named `export_name` that needs to be
+    /// emitted, as `(cfg, symbol)`: the `cfg` attribute to gate the copy with
+    /// and the symbol to export it as.
+    ///
+    /// Normally there's just one copy: the canonical ABI name with no `cfg`.
+    /// With `link_native_symbols` enabled a second, hex-encoded copy is emitted
+    /// for native targets as well, because native linkers reject the `:`, `/`,
+    /// `#`, `[` and `]` characters that canonical names contain. Names that
+    /// survive encoding unchanged (`$root` exports, for instance) are emitted
+    /// once with no `cfg` rather than twice.
+    fn core_export_symbols(&self, export_name: &str) -> Vec<(&'static str, String)> {
+        let prefix = self.r#gen.opts.export_prefix.as_deref().unwrap_or("");
+        let wasm = format!("{prefix}{export_name}");
+        if self.r#gen.native_symbols().is_none() {
+            return vec![("", wasm)];
+        }
+        let native = format!(
+            "{prefix}{}",
+            symbol_name::make_external_component(export_name)
+        );
+        if native == wasm {
+            return vec![("", wasm)];
         }
+        vec![
+            ("#[cfg(target_arch = \"wasm32\")]\n", wasm),
+            ("#[cfg(not(target_arch = \"wasm32\"))]\n", native),
+        ]
     }
 
     fn print_export_sig(&mut self, func: &Function, async_: bool) -> Vec<String> {
@@ -2952,6 +2989,7 @@ impl<'a> {camel}Borrow<'a>{{
             "drop",
             &[abi::WasmType::I32],
             &[],
+            self.r#gen.native_symbols(),
         );
         uwriteln!(
             self.src,