Support defining C compatible variadic functions

Add support for defining C compatible variadic functions in unsafe rust
with extern "C".

Author: dlrobertson

src/librustc/hir/intravisit.rs

@@ -617,6 +617,9 @@ pub fn walk_ty<'v, V: Visitor<'v>>(visitor: &mut V, typ: &'v Ty) {
         TyKind::Typeof(ref expression) => {
             visitor.visit_anon_const(expression)
         }
+        TyKind::CVarArgs(ref lt) => {
+            visitor.visit_lifetime(lt)
+        }
         TyKind::Infer | TyKind::Err => {}
     }
 }

src/librustc/hir/lowering.rs

@@ -1345,6 +1345,12 @@ impl<'a> LoweringContext<'a> {
                 }
             }
             TyKind::Mac(_) => panic!("TyMac should have been expanded by now."),
+            TyKind::CVarArgs => {
+                // Create the implicit lifetime of the "spoofed" `VaList`.
+                let span = self.sess.source_map().next_point(t.span.shrink_to_lo());
+                let lt = self.new_implicit_lifetime(span);
+                hir::TyKind::CVarArgs(lt)
+            },
         };
 
         let LoweredNodeId { node_id: _, hir_id } = self.lower_node_id(t.id);

src/librustc/hir/mod.rs

@@ -1829,6 +1829,9 @@ pub enum TyKind {
     Infer,
     /// Placeholder for a type that has failed to be defined.
     Err,
+    /// Placeholder for C-variadic arguments. We "spoof" the `VaList` created
+    /// from the variadic arguments. This type is only valid up to typeck.
+    CVarArgs(Lifetime),
 }
 
 #[derive(Clone, RustcEncodable, RustcDecodable, Debug)]

src/librustc/hir/print.rs

@@ -434,6 +434,9 @@ impl<'a> State<'a> {
                 self.s.word("/*ERROR*/")?;
                 self.pclose()?;
             }
+            hir::TyKind::CVarArgs(_) => {
+                self.s.word("...")?;
+            }
         }
         self.end()
     }

src/librustc/ich/impls_hir.rs

@@ -361,7 +361,8 @@ impl_stable_hash_for!(enum hir::TyKind {
     TraitObject(trait_refs, lifetime),
     Typeof(body_id),
     Err,
-    Infer
+    Infer,
+    CVarArgs(lt),
 });
 
 impl_stable_hash_for!(struct hir::FnDecl {

src/librustc/middle/resolve_lifetime.rs

@@ -764,6 +764,13 @@ impl<'a, 'tcx> Visitor<'tcx> for LifetimeContext<'a, 'tcx> {
                     });
                 }
             }
+            hir::TyKind::CVarArgs(ref lt) => {
+                // Resolve the generated lifetime for the C-variadic arguments.
+                // The lifetime is generated in AST -> HIR lowering.
+                if lt.name.is_elided() {
+                    self.resolve_elided_lifetimes(vec![lt])
+                }
+            }
             _ => intravisit::walk_ty(self, ty),
         }
     }
@@ -2225,18 +2232,22 @@ impl<'a, 'tcx> LifetimeContext<'a, 'tcx> {
                 if let hir::TyKind::BareFn(_) = ty.node {
                     self.outer_index.shift_in(1);
                 }
-                if let hir::TyKind::TraitObject(ref bounds, ref lifetime) = ty.node {
-                    for bound in bounds {
-                        self.visit_poly_trait_ref(bound, hir::TraitBoundModifier::None);
-                    }
+                match ty.node {
+                    hir::TyKind::TraitObject(ref bounds, ref lifetime) => {
+                        for bound in bounds {
+                            self.visit_poly_trait_ref(bound, hir::TraitBoundModifier::None);
+                        }
 
-                    // Stay on the safe side and don't include the object
-                    // lifetime default (which may not end up being used).
-                    if !lifetime.is_elided() {
-                        self.visit_lifetime(lifetime);
+                        // Stay on the safe side and don't include the object
+                        // lifetime default (which may not end up being used).
+                        if !lifetime.is_elided() {
+                            self.visit_lifetime(lifetime);
+                        }
+                    }
+                    hir::TyKind::CVarArgs(_) => {}
+                    _ => {
+                        intravisit::walk_ty(self, ty);
                     }
-                } else {
-                    intravisit::walk_ty(self, ty);
                 }
                 if let hir::TyKind::BareFn(_) = ty.node {
                     self.outer_index.shift_out(1);

src/librustc/ty/sty.rs

@@ -977,9 +977,9 @@ impl<'tcx> PolyGenSig<'tcx> {
 /// Signature of a function type, which I have arbitrarily
 /// decided to use to refer to the input/output types.
 ///
-/// - `inputs` is the list of arguments and their modes.
-/// - `output` is the return type.
-/// - `variadic` indicates whether this is a variadic function. (only true for foreign fns)
+/// - `inputs`: is the list of arguments and their modes.
+/// - `output`: is the return type.
+/// - `variadic`: indicates whether this is a C-variadic function.
 #[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, RustcEncodable, RustcDecodable)]
 pub struct FnSig<'tcx> {
     pub inputs_and_output: &'tcx List<Ty<'tcx>>,

src/librustc_codegen_llvm/abi.rs

@@ -258,7 +258,7 @@ impl ArgTypeExt<'ll, 'tcx> for ArgType<'tcx, Ty<'tcx>> {
             val
         };
         match self.mode {
-            PassMode::Ignore => {},
+            PassMode::Ignore(_) => {}
             PassMode::Pair(..) => {
                 OperandValue::Pair(next(), next()).store(bx, dst);
             }
@@ -507,6 +507,14 @@ impl<'tcx> FnTypeExt<'tcx> for FnType<'tcx, Ty<'tcx>> {
             }
         };
 
+        // Store the index of the last argument. This is useful for working with
+        // C-compatible variadic arguments.
+        let last_arg_idx = if sig.inputs().is_empty() {
+            None
+        } else {
+            Some(sig.inputs().len() - 1)
+        };
+
         let arg_of = |ty: Ty<'tcx>, arg_idx: Option<usize>| {
             let is_return = arg_idx.is_none();
             let mut arg = mk_arg_type(ty, arg_idx);
@@ -516,7 +524,30 @@ impl<'tcx> FnTypeExt<'tcx> for FnType<'tcx, Ty<'tcx>> {
                 // The same is true for s390x-unknown-linux-gnu
                 // and sparc64-unknown-linux-gnu.
                 if is_return || rust_abi || (!win_x64_gnu && !linux_s390x && !linux_sparc64) {
-                    arg.mode = PassMode::Ignore;
+                    arg.mode = PassMode::Ignore(IgnoreMode::Zst);
+                }
+            }
+
+            // If this is a C-variadic function, this is not the return value,
+            // and there is one or more fixed arguments; ensure that the `VaList`
+            // is ignored as an argument.
+            if sig.variadic {
+                match (last_arg_idx, arg_idx) {
+                    (Some(last_idx), Some(cur_idx)) if last_idx == cur_idx => {
+                        let va_list_did = match cx.tcx.lang_items().va_list() {
+                            Some(did) => did,
+                            None => bug!("`va_list` lang item required for C-variadic functions"),
+                        };
+                        match ty.sty {
+                            ty::Adt(def, _) if def.did == va_list_did => {
+                                // This is the "spoofed" `VaList`. Set the arguments mode
+                                // so that it will be ignored.
+                                arg.mode = PassMode::Ignore(IgnoreMode::CVarArgs);
+                            },
+                            _ => (),
+                        }
+                    }
+                    _ => {}
                 }
             }
 
@@ -646,7 +677,9 @@ impl<'tcx> FnTypeExt<'tcx> for FnType<'tcx, Ty<'tcx>> {
         );
 
         let llreturn_ty = match self.ret.mode {
-            PassMode::Ignore => cx.type_void(),
+            PassMode::Ignore(IgnoreMode::Zst) => cx.type_void(),
+            PassMode::Ignore(IgnoreMode::CVarArgs) =>
+                bug!("`va_list` should never be a return type"),
             PassMode::Direct(_) | PassMode::Pair(..) => {
                 self.ret.layout.immediate_llvm_type(cx)
             }
@@ -664,7 +697,7 @@ impl<'tcx> FnTypeExt<'tcx> for FnType<'tcx, Ty<'tcx>> {
             }
 
             let llarg_ty = match arg.mode {
-                PassMode::Ignore => continue,
+                PassMode::Ignore(_) => continue,
                 PassMode::Direct(_) => arg.layout.immediate_llvm_type(cx),
                 PassMode::Pair(..) => {
                     llargument_tys.push(arg.layout.scalar_pair_element_llvm_type(cx, 0, true));
@@ -733,7 +766,7 @@ impl<'tcx> FnTypeExt<'tcx> for FnType<'tcx, Ty<'tcx>> {
                 apply(&ArgAttributes::new());
             }
             match arg.mode {
-                PassMode::Ignore => {}
+                PassMode::Ignore(_) => {}
                 PassMode::Direct(ref attrs) |
                 PassMode::Indirect(ref attrs, None) => apply(attrs),
                 PassMode::Indirect(ref attrs, Some(ref extra_attrs)) => {
@@ -780,7 +813,7 @@ impl<'tcx> FnTypeExt<'tcx> for FnType<'tcx, Ty<'tcx>> {
                 apply(&ArgAttributes::new());
             }
             match arg.mode {
-                PassMode::Ignore => {}
+                PassMode::Ignore(_) => {}
                 PassMode::Direct(ref attrs) |
                 PassMode::Indirect(ref attrs, None) => apply(attrs),
                 PassMode::Indirect(ref attrs, Some(ref extra_attrs)) => {

src/librustc_codegen_llvm/intrinsic.rs

@@ -136,22 +136,18 @@ impl IntrinsicCallMethods<'tcx> for Builder<'a, 'll, 'tcx> {
                 let tp_ty = substs.type_at(0);
                 self.const_usize(self.size_of(tp_ty).bytes())
             }
-            func @ "va_start" | func @ "va_end" => {
-                let va_list = match (tcx.lang_items().va_list(), &result.layout.ty.sty) {
-                    (Some(did), ty::Adt(def, _)) if def.did == did => args[0].immediate(),
-                    (Some(_), _) => self.load(args[0].immediate(),
-                                              tcx.data_layout.pointer_align.abi),
-                    (None, _) => bug!("va_list language item must be defined")
-                };
-                let intrinsic = self.cx().get_intrinsic(&format!("llvm.{}", func));
-                self.call(intrinsic, &[va_list], None)
+            "va_start" => {
+                self.va_start(args[0].immediate())
+            }
+            "va_end" => {
+                self.va_end(args[0].immediate())
             }
             "va_copy" => {
                 let va_list = match (tcx.lang_items().va_list(), &result.layout.ty.sty) {
                     (Some(did), ty::Adt(def, _)) if def.did == did => args[0].immediate(),
                     (Some(_), _)  => self.load(args[0].immediate(),
                                                tcx.data_layout.pointer_align.abi),
-                    (None, _) => bug!("va_list language item must be defined")
+                    (None, _) => bug!("`va_list` language item must be defined")
                 };
                 let intrinsic = self.cx().get_intrinsic(&("llvm.va_copy"));
                 self.call(intrinsic, &[llresult, va_list], None);
@@ -722,6 +718,41 @@ impl IntrinsicCallMethods<'tcx> for Builder<'a, 'll, 'tcx> {
         let expect = self.get_intrinsic(&"llvm.expect.i1");
         self.call(expect, &[cond, self.const_bool(expected)], None)
     }
+
+    fn va_start(&mut self, list: &'ll Value) -> &'ll Value {
+        let target = &self.cx.tcx.sess.target.target;
+        let arch = &target.arch;
+        // A pointer to the architecture specific structure is passed to this
+        // function. For pointer variants (i686, RISC-V, Windows, etc), we
+        // should do do nothing, as the address to the pointer is needed. For
+        // architectures with a architecture specific structure (`Aarch64`,
+        // `X86_64`, etc), this function should load the structure from the
+        // address provided.
+        let va_list = match &**arch {
+            _ if target.options.is_like_windows => list,
+            "aarch64" if target.target_os == "ios" => list,
+            "aarch64" | "x86_64" | "powerpc" =>
+                self.load(list, self.tcx().data_layout.pointer_align.abi),
+            _ => list,
+        };
+        let intrinsic = self.cx().get_intrinsic("llvm.va_start");
+        self.call(intrinsic, &[va_list], None)
+    }
+
+    fn va_end(&mut self, list: &'ll Value) -> &'ll Value {
+        let target = &self.cx.tcx.sess.target.target;
+        let arch = &target.arch;
+        // See the comment in `va_start` for the purpose of the following.
+        let va_list = match &**arch {
+            _ if target.options.is_like_windows => list,
+            "aarch64" if target.target_os == "ios" => list,
+            "aarch64" | "x86_64" | "powerpc" =>
+                self.load(list, self.tcx().data_layout.pointer_align.abi),
+            _ => list,
+        };
+        let intrinsic = self.cx().get_intrinsic("llvm.va_end");
+        self.call(intrinsic, &[va_list], None)
+    }
 }
 
 fn copy_intrinsic(

src/librustc_codegen_llvm/mono_item.rs

@@ -36,10 +36,10 @@ impl PreDefineMethods<'tcx> for CodegenCx<'ll, 'tcx> {
     }
 
     fn predefine_fn(&self,
-                              instance: Instance<'tcx>,
-                              linkage: Linkage,
-                              visibility: Visibility,
-                              symbol_name: &str) {
+                    instance: Instance<'tcx>,
+                    linkage: Linkage,
+                    visibility: Visibility,
+                    symbol_name: &str) {
         assert!(!instance.substs.needs_infer() &&
                 !instance.substs.has_param_types());