Unrolled build for rust-lang#126967

Rollup merge of rust-lang#126967 - alexcrichton:wasm32-wasip2-tier-2, r=Mark-Simulacrum

Promote the `wasm32-wasip2` target to Tier 2

This commit promotes the `wasm32-wasip2` Rust target to tier 2 as proposed in rust-lang/compiler-team#760. There are two major changes in this PR:

1. The `dist-various-2` container, which already produces the other WASI targets, now has an extra target added for `wasm32-wasip2`.
2. A new `wasm-component-ld` binary is added to all host toolchains when LLD is enabled. This is the linker used for the `wasm32-wasip2` target.

This new linker is added for all host toolchains to ensure that all host toolchains can produce the `wasm32-wasip2` target. This is similar to how `rust-lld` was originally included for all host toolchains to be able to produce WebAssembly output when the targets were first added. The new linker is developed [here][wasm-component-ld] and is pulled in via a crates.io-based dependency to the tree here.

[wasm-component-ld]: https://github.com/bytecodealliance/wasm-component-ld

Author: rust-timer

Cargo.lock

@@ -1682,6 +1682,7 @@ dependencies = [
  "compiler_builtins",
  "rustc-std-workspace-alloc",
  "rustc-std-workspace-core",
+ "serde",
 ]
 
 [[package]]
@@ -1891,6 +1892,12 @@ dependencies = [
  "syn 2.0.67",
 ]
 
+[[package]]
+name = "id-arena"
+version = "2.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "25a2bc672d1148e28034f176e01fffebb08b35768468cc954630da77a1449005"
+
 [[package]]
 name = "ident_case"
 version = "1.0.1"
@@ -2120,6 +2127,12 @@ version = "1.0.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "db13adb97ab515a3691f56e4dbab09283d0b86cb45abd991d8634a9d6f501760"
 
+[[package]]
+name = "lexopt"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "baff4b617f7df3d896f97fe922b64817f6cd9a756bb81d40f8883f2f66dcb401"
+
 [[package]]
 name = "libc"
 version = "0.2.155"
@@ -2634,7 +2647,7 @@ dependencies = [
  "indexmap",
  "memchr",
  "ruzstd 0.5.0",
- "wasmparser",
+ "wasmparser 0.118.2",
 ]
 
 [[package]]
@@ -3436,7 +3449,7 @@ dependencies = [
  "object 0.34.0",
  "regex",
  "similar",
- "wasmparser",
+ "wasmparser 0.118.2",
 ]
 
 [[package]]
@@ -3823,7 +3836,7 @@ dependencies = [
  "thin-vec",
  "thorin-dwp",
  "tracing",
- "wasm-encoder",
+ "wasm-encoder 0.200.0",
  "windows",
 ]
 
@@ -5267,6 +5280,15 @@ dependencies = [
  "color-eyre",
 ]
 
+[[package]]
+name = "spdx"
+version = "0.10.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "47317bbaf63785b53861e1ae2d11b80d6b624211d42cb20efcd210ee6f8a14bc"
+dependencies = [
+ "smallvec",
+]
+
 [[package]]
 name = "spdx-expression"
 version = "0.5.2"
@@ -6315,6 +6337,28 @@ version = "0.2.92"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "af190c94f2773fdb3729c55b007a722abb5384da03bc0986df4c289bf5567e96"
 
+[[package]]
+name = "wasm-component-ld"
+version = "0.5.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "314d932d5e84c9678751b85498b1482b2f32f185744e449d3ce0b1d400376dad"
+dependencies = [
+ "anyhow",
+ "clap",
+ "lexopt",
+ "tempfile",
+ "wasmparser 0.210.0",
+ "wat",
+ "wit-component",
+]
+
+[[package]]
+name = "wasm-component-ld-wrapper"
+version = "0.1.0"
+dependencies = [
+ "wasm-component-ld",
+]
+
 [[package]]
 name = "wasm-encoder"
 version = "0.200.0"
@@ -6324,6 +6368,40 @@ dependencies = [
  "leb128",
 ]
 
+[[package]]
+name = "wasm-encoder"
+version = "0.210.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e7e3764d9d6edabd8c9e16195e177be0d20f6ab942ad18af52860f12f82bc59a"
+dependencies = [
+ "leb128",
+]
+
+[[package]]
+name = "wasm-encoder"
+version = "0.211.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5e7d931a1120ef357f32b74547646b6fa68ea25e377772b72874b131a9ed70d4"
+dependencies = [
+ "leb128",
+]
+
+[[package]]
+name = "wasm-metadata"
+version = "0.210.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "012729d1294907fcb0866f08460ab95426a6d0b176a599619b84cac7653452b4"
+dependencies = [
+ "anyhow",
+ "indexmap",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "spdx",
+ "wasm-encoder 0.210.0",
+ "wasmparser 0.210.0",
+]
+
 [[package]]
 name = "wasmparser"
 version = "0.118.2"
@@ -6334,6 +6412,42 @@ dependencies = [
  "semver",
 ]
 
+[[package]]
+name = "wasmparser"
+version = "0.210.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7bbcd21e7581619d9f6ca00f8c4f08f1cacfe58bf63f83af57cd0476f1026f5"
+dependencies = [
+ "ahash",
+ "bitflags 2.5.0",
+ "hashbrown",
+ "indexmap",
+ "semver",
+ "serde",
+]
+
+[[package]]
+name = "wast"
+version = "211.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b25506dd82d00da6b14a87436b3d52b1d264083fa79cdb72a0d1b04a8595ccaa"
+dependencies = [
+ "bumpalo",
+ "leb128",
+ "memchr",
+ "unicode-width",
+ "wasm-encoder 0.211.1",
+]
+
+[[package]]
+name = "wat"
+version = "1.211.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "eb716ca6c86eecac2d82541ffc39860118fc0af9309c4f2670637bea2e1bdd7d"
+dependencies = [
+ "wast",
+]
+
 [[package]]
 name = "winapi"
 version = "0.3.9"
@@ -6561,6 +6675,43 @@ dependencies = [
  "memchr",
 ]
 
+[[package]]
+name = "wit-component"
+version = "0.210.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a450bdb5d032acf1fa0865451fa0c6f50e62f2d31eaa8dba967c2e2d068694a4"
+dependencies = [
+ "anyhow",
+ "bitflags 2.5.0",
+ "indexmap",
+ "log",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "wasm-encoder 0.210.0",
+ "wasm-metadata",
+ "wasmparser 0.210.0",
+ "wit-parser",
+]
+
+[[package]]
+name = "wit-parser"
+version = "0.210.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "60a965cbd439af19a4b44a54a97ab8957d86f02d01320efc9e31c1d3605c6710"
+dependencies = [
+ "anyhow",
+ "id-arena",
+ "indexmap",
+ "log",
+ "semver",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "unicode-xid",
+ "wasmparser 0.210.0",
+]
+
 [[package]]
 name = "writeable"
 version = "0.5.5"

Cargo.toml

@@ -45,6 +45,7 @@ members = [
   "src/tools/opt-dist",
   "src/tools/coverage-dump",
   "src/tools/rustc-perf-wrapper",
+  "src/tools/wasm-component-ld",
 ]
 
 exclude = [
@@ -104,6 +105,9 @@ rustc-demangle.debug = 0
 [profile.release.package.lld-wrapper]
 debug = 0
 strip = true
+[profile.release.package.wasm-component-ld-wrapper]
+debug = 0
+strip = true
 
 [patch.crates-io]
 # See comments in `library/rustc-std-workspace-core/README.md` for what's going on

src/bootstrap/src/core/build_steps/compile.rs

@@ -1820,6 +1820,21 @@ impl Step for Assemble {
                     &self_contained_lld_dir.join(exe(name, target_compiler.host)),
                 );
             }
+
+            // In addition to `rust-lld` also install `wasm-component-ld` when
+            // LLD is enabled. This is a relatively small binary that primarily
+            // delegates to the `rust-lld` binary for linking and then runs
+            // logic to create the final binary. This is used by the
+            // `wasm32-wasip2` target of Rust.
+            let wasm_component_ld_exe =
+                builder.ensure(crate::core::build_steps::tool::WasmComponentLd {
+                    compiler: build_compiler,
+                    target: target_compiler.host,
+                });
+            builder.copy_link(
+                &wasm_component_ld_exe,
+                &libdir_bin.join(wasm_component_ld_exe.file_name().unwrap()),
+            );
         }
 
         if builder.config.llvm_enabled(target_compiler.host) {

src/bootstrap/src/core/build_steps/tool.rs

@@ -337,6 +337,7 @@ bootstrap_tool!(
     RustdocGUITest, "src/tools/rustdoc-gui-test", "rustdoc-gui-test", is_unstable_tool = true, allow_features = "test";
     CoverageDump, "src/tools/coverage-dump", "coverage-dump";
     RustcPerfWrapper, "src/tools/rustc-perf-wrapper", "rustc-perf-wrapper";
+    WasmComponentLd, "src/tools/wasm-component-ld", "wasm-component-ld", is_unstable_tool = true, allow_features = "min_specialization";
 );
 
 #[derive(Debug, Clone, Hash, PartialEq, Eq)]

src/ci/docker/host-x86_64/dist-various-2/Dockerfile

@@ -112,6 +112,7 @@ ENV TARGETS=$TARGETS,wasm32-unknown-unknown
 ENV TARGETS=$TARGETS,wasm32-wasi
 ENV TARGETS=$TARGETS,wasm32-wasip1
 ENV TARGETS=$TARGETS,wasm32-wasip1-threads
+ENV TARGETS=$TARGETS,wasm32-wasip2
 ENV TARGETS=$TARGETS,sparcv9-sun-solaris
 ENV TARGETS=$TARGETS,x86_64-pc-solaris
 ENV TARGETS=$TARGETS,x86_64-unknown-linux-gnux32

src/tools/tidy/src/deps.rs

@@ -95,7 +95,12 @@ const EXCEPTIONS: ExceptionList = &[
     ("self_cell", "Apache-2.0"),                             // rustc (fluent translations)
     ("snap", "BSD-3-Clause"),                                // rustc
     ("wasm-encoder", "Apache-2.0 WITH LLVM-exception"),      // rustc
+    ("wasm-metadata", "Apache-2.0 WITH LLVM-exception"),     // rustc
     ("wasmparser", "Apache-2.0 WITH LLVM-exception"),        // rustc
+    ("wast", "Apache-2.0 WITH LLVM-exception"),              // rustc
+    ("wat", "Apache-2.0 WITH LLVM-exception"),               // rustc
+    ("wit-component", "Apache-2.0 WITH LLVM-exception"),     // rustc
+    ("wit-parser", "Apache-2.0 WITH LLVM-exception"),        // rustc
     // tidy-alphabetical-end
 ];
 

src/tools/wasm-component-ld/Cargo.toml

@@ -0,0 +1,13 @@
+# See the `README.md` in this directory for what this tool is.
+
+[package]
+name = "wasm-component-ld-wrapper"
+version = "0.1.0"
+edition = "2021"
+
+[[bin]]
+name = "wasm-component-ld"
+path = "src/main.rs"
+
+[dependencies]
+wasm-component-ld = "0.5.4"

src/tools/wasm-component-ld/README.md

@@ -0,0 +1,62 @@
+# `wasm-component-ld`
+
+This wrapper is a wrapper around the [`wasm-component-ld`] crates.io crate. That
+crate. That crate is itself a thin wrapper around two pieces:
+
+* `wasm-ld` - the LLVM-based linker distributed as part of LLD and packaged in
+  Rust as `rust-lld`.
+* [`wit-component`] - a Rust crate for creating a [WebAssembly Component] from a
+  core wasm module.
+
+This linker is used for Rust's `wasm32-wasip2` target to natively output a
+component instead of a core WebAssembly module, unlike other WebAssembly
+targets. If you're confused about any of this here's an FAQ-style explanation of
+what's going on here:
+
+* **What's a component?** - It's a proposal to the WebAssembly standard
+  primarily developed at this time by out-of-browser use cases of WebAssembly.
+  You can find high-level documentation [here][component docs].
+
+* **What's WASIp2?** - Not to be confused with WASIp1, WASIp0,
+  `wasi_snapshot_preview1`, or `wasi_unstable`, it's a version of WASI. Released
+  in January 2024 it's the first version of WASI defined in terms of the
+  component model.
+
+* **Why does this need its own linker?** - like any target that Rust has the
+  `wasm32-wasip2` target needs a linker. What makes this different from other
+  WebAssembly targets is that WASIp2 is defined at the component level, not core
+  WebAssembly level. This means that filesystem functions take a `string`
+  instead of `i32 i32`, for example. This means that the raw output of LLVM and
+  `wasm-ld`, a core WebAssembly module, is not suitable.
+
+* **Isn't writing a linker really hard?** - Generally, yes, but this linker
+  works by first asking `wasm-ld` to do all the hard work. It invokes `wasm-ld`
+  and then uses the output core WebAssembly module to create a component.
+
+* **How do you create a component from a core module?** - this is the purpose of
+  the [`wit-component`] crate, notably the `ComponentEncoder` type. This uses
+  component type information embedded in the core module and a general set of
+  conventions/guidelines with what the core module imports/exports. A component
+  is then hooked up to codify all of these conventions in a component itself.
+
+* **Why not require users to run `wit-component` themselves?** - while possible
+  it adds friction to the usage `wasm32-wasip2` target. More importantly though
+  the "module only" output of the `wasm32-wasip2` target is not ready right now.
+  The standard library still imports from `wasi_snapshot_preview1` and it will
+  take time to migrate all usage to WASIp2.
+
+* **What exactly does this linker do?** - the `wasm-component-ld` has the same
+  CLI interface and flags as `wasm-ld`, plus some more that are
+  component-specific. These flags are used to forward most flags to `wasm-ld` to
+  produce a core wasm module. After the core wasm module is produced the
+  `wit-component` crate will read custom sections in the final binary which
+  contain component type information. After merging all this type information
+  together a component is produced which wraps the core module.
+
+If you've got any other questions about this linker or its operation don't
+hesitate to reach out to the maintainers of the `wasm32-wasip2` target.
+
+[`wasm-component-ld`]: https://crates.io/crates/wasm-component-ld
+[`wit-component`]: https://crates.io/crates/wit-component
+[WebAssembly Component]: https://github.com/webassembly/component-model
+[component docs]: https://component-model.bytecodealliance.org/

src/tools/wasm-component-ld/src/main.rs

@@ -0,0 +1,9 @@
+// See the `README.md` in this directory for what this tool is.
+
+// The source for this crate lives at
+// https://github.com/bytecodealliance/wasm-component-ld and the binary is
+// independently used in other projects such as `wasi-sdk` so the `main`
+// function is just reexported here to delegate. A Cargo dependency is used to
+// facilitate version management in the Rust repository and work well with
+// vendored/offline builds.
+use wasm_component_ld::main;