Auto merge of #58408 - alexcrichton:update-llvm, r=michaelwoerister

rustc: Update LLVM, remove dead wasm code

This commit updates the LLVM branch to the rebased version of the
upstream release/8.x branch. This includes a wasm patch which means that
the `rewrite_imports` pass in rustc is no longer needed (yay!) and we
can instead rely on `wasm-import-module`, an attribute we're already
emitting, to take care of all the work.

Author: bors

config.toml.example

@@ -104,6 +104,8 @@
 # The value specified here will be passed as `-DLLVM_USE_LINKER` to CMake.
 #use-linker = "lld"
 
+# Whether or not to specify `-DLLVM_TEMPORARILY_ALLOW_OLD_TOOLCHAIN=YES`
+#allow-old-toolchain = false
 
 # =============================================================================
 # General build configuration options

src/bootstrap/config.rs

@@ -78,6 +78,7 @@ pub struct Config {
     pub llvm_link_jobs: Option<u32>,
     pub llvm_version_suffix: Option<String>,
     pub llvm_use_linker: Option<String>,
+    pub llvm_allow_old_toolchain: Option<bool>,
 
     pub lld_enabled: bool,
     pub lldb_enabled: bool,
@@ -263,6 +264,7 @@ struct Llvm {
     ldflags: Option<String>,
     use_libcxx: Option<bool>,
     use_linker: Option<String>,
+    allow_old_toolchain: Option<bool>,
 }
 
 #[derive(Deserialize, Default, Clone)]
@@ -530,6 +532,7 @@ impl Config {
             config.llvm_ldflags = llvm.ldflags.clone();
             set(&mut config.llvm_use_libcxx, llvm.use_libcxx);
             config.llvm_use_linker = llvm.use_linker.clone();
+            config.llvm_allow_old_toolchain = llvm.allow_old_toolchain.clone();
         }
 
         if let Some(ref rust) = toml.rust {

src/bootstrap/native.rs

@@ -238,6 +238,10 @@ impl Step for Llvm {
             cfg.define("LLVM_USE_LINKER", linker);
         }
 
+        if let Some(true) = builder.config.llvm_allow_old_toolchain {
+            cfg.define("LLVM_TEMPORARILY_ALLOW_OLD_TOOLCHAIN", "YES");
+        }
+
         if let Some(ref python) = builder.config.python {
             cfg.define("PYTHON_EXECUTABLE", python);
         }

src/ci/docker/dist-x86_64-netbsd/Dockerfile

@@ -33,5 +33,6 @@ ENV \
 
 ENV HOSTS=x86_64-unknown-netbsd
 
-ENV RUST_CONFIGURE_ARGS --enable-extended --disable-docs
+ENV RUST_CONFIGURE_ARGS --enable-extended --disable-docs \
+  --set llvm.allow-old-toolchain
 ENV SCRIPT python2.7 ../x.py dist --host $HOSTS --target $HOSTS

src/librustc_codegen_llvm/back/link.rs

@@ -698,7 +698,6 @@ fn link_natively(sess: &Session,
     }
 
     if sess.opts.target_triple.triple() == "wasm32-unknown-unknown" {
-        wasm::rewrite_imports(&out_filename, &codegen_results.crate_info.wasm_imports);
         wasm::add_producer_section(
             &out_filename,
             &sess.edition().to_string(),

src/librustc_codegen_llvm/back/wasm.rs

@@ -2,116 +2,11 @@ use std::fs;
 use std::path::Path;
 use std::str;
 
-use rustc_data_structures::fx::FxHashMap;
 use serialize::leb128;
 
 // https://webassembly.github.io/spec/core/binary/modules.html#binary-importsec
-const WASM_IMPORT_SECTION_ID: u8 = 2;
 const WASM_CUSTOM_SECTION_ID: u8 = 0;
 
-const WASM_EXTERNAL_KIND_FUNCTION: u8 = 0;
-const WASM_EXTERNAL_KIND_TABLE: u8 = 1;
-const WASM_EXTERNAL_KIND_MEMORY: u8 = 2;
-const WASM_EXTERNAL_KIND_GLOBAL: u8 = 3;
-
-/// Rewrite the module imports are listed from in a wasm module given the field
-/// name to module name mapping in `import_map`.
-///
-/// LLVM 6 which we're using right now doesn't have the ability to configure the
-/// module a wasm symbol is import from. Rather all imported symbols come from
-/// the bland `"env"` module unconditionally. Furthermore we'd *also* need
-/// support in LLD for preserving these import modules, which it unfortunately
-/// currently does not.
-///
-/// This function is intended as a hack for now where we manually rewrite the
-/// wasm output by LLVM to have the correct import modules listed. The
-/// `#[link(wasm_import_module = "...")]` attribute in Rust translates to the
-/// module that each symbol is imported from, so here we manually go through the
-/// wasm file, decode it, rewrite imports, and then rewrite the wasm module.
-///
-/// Support for this was added to LLVM in
-/// https://github.com/llvm-mirror/llvm/commit/0f32e1365, although support still
-/// needs to be added, tracked at https://bugs.llvm.org/show_bug.cgi?id=37168
-pub fn rewrite_imports(path: &Path, import_map: &FxHashMap<String, String>) {
-    if import_map.is_empty() {
-        return
-    }
-
-    let wasm = fs::read(path).expect("failed to read wasm output");
-    let mut ret = WasmEncoder::new();
-    ret.data.extend(&wasm[..8]);
-
-    // skip the 8 byte wasm/version header
-    for (id, raw) in WasmSections(WasmDecoder::new(&wasm[8..])) {
-        ret.byte(id);
-        if id == WASM_IMPORT_SECTION_ID {
-            info!("rewriting import section");
-            let data = rewrite_import_section(
-                &mut WasmDecoder::new(raw),
-                import_map,
-            );
-            ret.bytes(&data);
-        } else {
-            info!("carry forward section {}, {} bytes long", id, raw.len());
-            ret.bytes(raw);
-        }
-    }
-
-    fs::write(path, &ret.data).expect("failed to write wasm output");
-
-    fn rewrite_import_section(
-        wasm: &mut WasmDecoder<'_>,
-        import_map: &FxHashMap<String, String>,
-    )
-        -> Vec<u8>
-    {
-        let mut dst = WasmEncoder::new();
-        let n = wasm.u32();
-        dst.u32(n);
-        info!("rewriting {} imports", n);
-        for _ in 0..n {
-            rewrite_import_entry(wasm, &mut dst, import_map);
-        }
-        return dst.data
-    }
-
-    fn rewrite_import_entry(wasm: &mut WasmDecoder<'_>,
-                            dst: &mut WasmEncoder,
-                            import_map: &FxHashMap<String, String>) {
-        // More info about the binary format here is available at:
-        // https://webassembly.github.io/spec/core/binary/modules.html#import-section
-        //
-        // Note that you can also find the whole point of existence of this
-        // function here, where we map the `module` name to a different one if
-        // we've got one listed.
-        let module = wasm.str();
-        let field = wasm.str();
-        let new_module = if module == "env" {
-            import_map.get(field).map(|s| &**s).unwrap_or(module)
-        } else {
-            module
-        };
-        info!("import rewrite ({} => {}) / {}", module, new_module, field);
-        dst.str(new_module);
-        dst.str(field);
-        let kind = wasm.byte();
-        dst.byte(kind);
-        match kind {
-            WASM_EXTERNAL_KIND_FUNCTION => dst.u32(wasm.u32()),
-            WASM_EXTERNAL_KIND_TABLE => {
-                dst.byte(wasm.byte()); // element_type
-                dst.limits(wasm.limits());
-            }
-            WASM_EXTERNAL_KIND_MEMORY => dst.limits(wasm.limits()),
-            WASM_EXTERNAL_KIND_GLOBAL => {
-                dst.byte(wasm.byte()); // content_type
-                dst.bool(wasm.bool()); // mutable
-            }
-            b => panic!("unknown kind: {}", b),
-        }
-    }
-}
-
 /// Adds or augment the existing `producers` section to encode information about
 /// the Rust compiler used to produce the wasm file.
 pub fn add_producer_section(
@@ -266,15 +161,6 @@ impl<'a> WasmDecoder<'a> {
         let len = self.u32();
         str::from_utf8(self.skip(len as usize)).unwrap()
     }
-
-    fn bool(&mut self) -> bool {
-        self.byte() == 1
-    }
-
-    fn limits(&mut self) -> (u32, Option<u32>) {
-        let has_max = self.bool();
-        (self.u32(), if has_max { Some(self.u32()) } else { None })
-    }
 }
 
 struct WasmEncoder {
@@ -302,16 +188,4 @@ impl WasmEncoder {
     fn str(&mut self, val: &str) {
         self.bytes(val.as_bytes())
     }
-
-    fn bool(&mut self, b: bool) {
-        self.byte(b as u8);
-    }
-
-    fn limits(&mut self, limits: (u32, Option<u32>)) {
-        self.bool(limits.1.is_some());
-        self.u32(limits.0);
-        if let Some(c) = limits.1 {
-            self.u32(c);
-        }
-    }
 }

src/librustc_codegen_ssa/base.rs

@@ -16,7 +16,7 @@
 use crate::{ModuleCodegen, ModuleKind, CachedModuleCodegen};
 
 use rustc::dep_graph::cgu_reuse_tracker::CguReuse;
-use rustc::hir::def_id::{CrateNum, DefId, LOCAL_CRATE};
+use rustc::hir::def_id::{DefId, LOCAL_CRATE};
 use rustc::middle::lang_items::StartFnLangItem;
 use rustc::middle::weak_lang_items;
 use rustc::mir::mono::{Stats, CodegenUnitNameBuilder};
@@ -816,21 +816,11 @@ impl CrateInfo {
             used_crates_dynamic: cstore::used_crates(tcx, LinkagePreference::RequireDynamic),
             used_crates_static: cstore::used_crates(tcx, LinkagePreference::RequireStatic),
             used_crate_source: Default::default(),
-            wasm_imports: Default::default(),
             lang_item_to_crate: Default::default(),
             missing_lang_items: Default::default(),
         };
         let lang_items = tcx.lang_items();
 
-        let load_wasm_items = tcx.sess.crate_types.borrow()
-            .iter()
-            .any(|c| *c != config::CrateType::Rlib) &&
-            tcx.sess.opts.target_triple.triple() == "wasm32-unknown-unknown";
-
-        if load_wasm_items {
-            info.load_wasm_imports(tcx, LOCAL_CRATE);
-        }
-
         let crates = tcx.crates();
 
         let n_crates = crates.len();
@@ -858,9 +848,6 @@ impl CrateInfo {
             if tcx.is_no_builtins(cnum) {
                 info.is_no_builtins.insert(cnum);
             }
-            if load_wasm_items {
-                info.load_wasm_imports(tcx, cnum);
-            }
             let missing = tcx.missing_lang_items(cnum);
             for &item in missing.iter() {
                 if let Ok(id) = lang_items.require(item) {
@@ -879,15 +866,6 @@ impl CrateInfo {
 
         return info
     }
-
-    fn load_wasm_imports(&mut self, tcx: TyCtxt<'_, '_, '_>, cnum: CrateNum) {
-        self.wasm_imports.extend(tcx.wasm_import_module_map(cnum).iter().map(|(&id, module)| {
-            let instance = Instance::mono(tcx, id);
-            let import_name = tcx.symbol_name(instance);
-
-            (import_name.to_string(), module.clone())
-        }));
-    }
 }
 
 fn is_codegened_item(tcx: TyCtxt<'_, '_, '_>, id: DefId) -> bool {

src/librustc_codegen_ssa/lib.rs

@@ -138,7 +138,6 @@ pub struct CrateInfo {
     pub used_crate_source: FxHashMap<CrateNum, Lrc<CrateSource>>,
     pub used_crates_static: Vec<(CrateNum, LibSource)>,
     pub used_crates_dynamic: Vec<(CrateNum, LibSource)>,
-    pub wasm_imports: FxHashMap<String, String>,
     pub lang_item_to_crate: FxHashMap<LangItem, CrateNum>,
     pub missing_lang_items: FxHashMap<CrateNum, Vec<LangItem>>,
 }