Add the `--override-abi` option (#2329)

* Add the `--override-abi` option.

This option can be used from the CLI with the <abi>:<regex> syntax and
it overrides the ABI of a function if it matches <regex>.

Fixes #2257

8 files changed, 190 insertions, 76 deletions

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 560ce38b..baa5cb20 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -148,6 +148,8 @@
 # Unreleased
 
 ## Added
+ * new feature: `--override-abi` flag to override the ABI used by functions
+   matching a regular expression.
 
 ## Changed
 
diff --git a/bindgen-cli/options.rs b/bindgen-cli/options.rs
index 5c3960e9..c186cfd9 100644
--- a/bindgen-cli/options.rs
+++ b/bindgen-cli/options.rs
@@ -568,6 +568,12 @@ where
             Arg::new("merge-extern-blocks")
                 .long("merge-extern-blocks")
                 .help("Deduplicates extern blocks."),
+            Arg::new("override-abi")
+                .long("override-abi")
+                .help("Overrides the ABI of functions matching <regex>. The <override> value must be of the shape <abi>:<regex> where <abi> can be one of C, stdcall, fastcall, thiscall, aapcs or win64.")
+                .value_name("override")
+                .multiple_occurrences(true)
+                .number_of_values(1),
             Arg::new("V")
                 .long("version")
                 .help("Prints the version, and exits"),
@@ -1088,5 +1094,17 @@ where
         builder = builder.merge_extern_blocks(true);
     }
 
+    if let Some(abi_overrides) = matches.values_of("override-abi") {
+        for abi_override in abi_overrides {
+            let (regex, abi_str) = abi_override
+                .rsplit_once("=")
+                .expect("Invalid ABI override: Missing `=`");
+            let abi = abi_str
+                .parse()
+                .unwrap_or_else(|err| panic!("Invalid ABI override: {}", err));
+            builder = builder.override_abi(abi, regex);
+        }
+    }
+
     Ok((builder, output, verbose))
 }
diff --git a/bindgen-tests/tests/expectations/tests/abi-override.rs b/bindgen-tests/tests/expectations/tests/abi-override.rs
new file mode 100644
index 00000000..49693692
--- /dev/null
+++ b/bindgen-tests/tests/expectations/tests/abi-override.rs
@@ -0,0 +1,16 @@
+#![allow(
+    dead_code,
+    non_snake_case,
+    non_camel_case_types,
+    non_upper_case_globals
+)]
+
+extern "fastcall" {
+    pub fn foo();
+}
+extern "stdcall" {
+    pub fn bar();
+}
+extern "C" {
+    pub fn baz();
+}
diff --git a/bindgen-tests/tests/headers/abi-override.h b/bindgen-tests/tests/headers/abi-override.h
new file mode 100644
index 00000000..cee72cae
--- /dev/null
+++ b/bindgen-tests/tests/headers/abi-override.h
@@ -0,0 +1,5 @@
+// bindgen-flags: --override-abi=foo=fastcall --override-abi=bar=stdcall
+
+void foo();
+void bar();
+void baz();
diff --git a/bindgen/codegen/dyngen.rs b/bindgen/codegen/dyngen.rs
index 26cfe5cc..d02c51e4 100644
--- a/bindgen/codegen/dyngen.rs
+++ b/bindgen/codegen/dyngen.rs
@@ -1,5 +1,5 @@
 use crate::codegen;
-use crate::ir::function::Abi;
+use crate::ir::function::ClangAbi;
 use proc_macro2::Ident;
 
 /// Used to build the output tokens for dynamic bindings.
@@ -113,10 +113,10 @@ impl DynamicItems {
     }
 
     #[allow(clippy::too_many_arguments)]
-    pub fn push(
+    pub(crate) fn push(
         &mut self,
         ident: Ident,
-        abi: Abi,
+        abi: ClangAbi,
         is_variadic: bool,
         is_required: bool,
         args: Vec<proc_macro2::TokenStream>,
diff --git a/bindgen/codegen/mod.rs b/bindgen/codegen/mod.rs
index a59265ab..4e90102d 100644
--- a/bindgen/codegen/mod.rs
+++ b/bindgen/codegen/mod.rs
@@ -32,7 +32,9 @@ use crate::ir::derive::{
 };
 use crate::ir::dot;
 use crate::ir::enum_ty::{Enum, EnumVariant, EnumVariantValue};
-use crate::ir::function::{Abi, Function, FunctionKind, FunctionSig, Linkage};
+use crate::ir::function::{
+    Abi, ClangAbi, Function, FunctionKind, FunctionSig, Linkage,
+};
 use crate::ir::int::IntKind;
 use crate::ir::item::{IsOpaque, Item, ItemCanonicalName, ItemCanonicalPath};
 use crate::ir::item_kind::ItemKind;
@@ -2474,9 +2476,13 @@ impl MethodCodegen for Method {
             _ => panic!("How in the world?"),
         };
 
-        let supported_abi = match signature.abi() {
-            Abi::ThisCall => ctx.options().rust_features().thiscall_abi,
-            Abi::Vectorcall => ctx.options().rust_features().vectorcall_abi,
+        let supported_abi = match signature.abi(ctx, Some(&*name)) {
+            ClangAbi::Known(Abi::ThisCall) => {
+                ctx.options().rust_features().thiscall_abi
+            }
+            ClangAbi::Known(Abi::Vectorcall) => {
+                ctx.options().rust_features().vectorcall_abi
+            }
             _ => true,
         };
 
@@ -3988,14 +3994,16 @@ impl TryToRustTy for FunctionSig {
         // TODO: we might want to consider ignoring the reference return value.
         let ret = utils::fnsig_return_ty(ctx, self);
         let arguments = utils::fnsig_arguments(ctx, self);
-        let abi = self.abi();
+        let abi = self.abi(ctx, None);
 
         match abi {
-            Abi::ThisCall if !ctx.options().rust_features().thiscall_abi => {
+            ClangAbi::Known(Abi::ThisCall)
+                if !ctx.options().rust_features().thiscall_abi =>
+            {
                 warn!("Skipping function with thiscall ABI that isn't supported by the configured Rust target");
                 Ok(proc_macro2::TokenStream::new())
             }
-            Abi::Vectorcall
+            ClangAbi::Known(Abi::Vectorcall)
                 if !ctx.options().rust_features().vectorcall_abi =>
             {
                 warn!("Skipping function with vectorcall ABI that isn't supported by the configured Rust target");
@@ -4099,22 +4107,24 @@ impl CodeGenerator for Function {
             attributes.push(attributes::doc(comment));
         }
 
-        let abi = match signature.abi() {
-            Abi::ThisCall if !ctx.options().rust_features().thiscall_abi => {
+        let abi = match signature.abi(ctx, Some(name)) {
+            ClangAbi::Known(Abi::ThisCall)
+                if !ctx.options().rust_features().thiscall_abi =>
+            {
                 warn!("Skipping function with thiscall ABI that isn't supported by the configured Rust target");
                 return None;
             }
-            Abi::Vectorcall
+            ClangAbi::Known(Abi::Vectorcall)
                 if !ctx.options().rust_features().vectorcall_abi =>
             {
                 warn!("Skipping function with vectorcall ABI that isn't supported by the configured Rust target");
                 return None;
             }
-            Abi::Win64 if signature.is_variadic() => {
+            ClangAbi::Known(Abi::Win64) if signature.is_variadic() => {
                 warn!("Skipping variadic function with Win64 ABI that isn't supported");
                 return None;
             }
-            Abi::Unknown(unknown_abi) => {
+            ClangAbi::Unknown(unknown_abi) => {
                 panic!(
                     "Invalid or unknown abi {:?} for function {:?} ({:?})",
                     unknown_abi, canonical_name, self
@@ -4512,7 +4522,7 @@ pub(crate) fn codegen(
 pub mod utils {
     use super::{error, ToRustTyOrOpaque};
     use crate::ir::context::BindgenContext;
-    use crate::ir::function::{Abi, FunctionSig};
+    use crate::ir::function::{Abi, ClangAbi, FunctionSig};
     use crate::ir::item::{Item, ItemCanonicalPath};
     use crate::ir::ty::TypeKind;
     use proc_macro2;
@@ -4973,10 +4983,10 @@ pub mod utils {
     // Returns true if `canonical_name` will end up as `mangled_name` at the
     // machine code level, i.e. after LLVM has applied any target specific
     // mangling.
-    pub fn names_will_be_identical_after_mangling(
+    pub(crate) fn names_will_be_identical_after_mangling(
         canonical_name: &str,
         mangled_name: &str,
-        call_conv: Option<Abi>,
+        call_conv: Option<ClangAbi>,
     ) -> bool {
         // If the mangled name and the canonical name are the same then no
         // mangling can have happened between the two versions.
@@ -4989,13 +4999,13 @@ pub mod utils {
         let mangled_name = mangled_name.as_bytes();
 
         let (mangling_prefix, expect_suffix) = match call_conv {
-            Some(Abi::C) |
+            Some(ClangAbi::Known(Abi::C)) |
             // None is the case for global variables
             None => {
                 (b'_', false)
             }
-            Some(Abi::Stdcall) => (b'_', true),
-            Some(Abi::Fastcall) => (b'@', true),
+            Some(ClangAbi::Known(Abi::Stdcall)) => (b'_', true),
+            Some(ClangAbi::Known(Abi::Fastcall)) => (b'@', true),
 
             // This is something we don't recognize, stay on the safe side
             // by emitting the `#[link_name]` attribute
diff --git a/bindgen/ir/function.rs b/bindgen/ir/function.rs
index f488b384..8a70d60a 100644
--- a/bindgen/ir/function.rs
+++ b/bindgen/ir/function.rs
@@ -15,6 +15,7 @@ use proc_macro2;
 use quote;
 use quote::TokenStreamExt;
 use std::io;
+use std::str::FromStr;
 
 const RUST_DERIVE_FUNPTR_LIMIT: usize = 12;
 
@@ -170,8 +171,8 @@ impl DotAttributes for Function {
     }
 }
 
-/// An ABI extracted from a clang cursor.
-#[derive(Debug, Copy, Clone)]
+/// A valid rust ABI.
+#[derive(Debug, Copy, Clone, Hash, Eq, PartialEq)]
 pub enum Abi {
     /// The default C ABI.
     C,
@@ -187,32 +188,72 @@ pub enum Abi {
     Aapcs,
     /// The "win64" ABI.
     Win64,
+}
+
+impl FromStr for Abi {
+    type Err = String;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        match s {
+            "C" => Ok(Self::C),
+            "stdcall" => Ok(Self::Stdcall),
+            "fastcall" => Ok(Self::Fastcall),
+            "thiscall" => Ok(Self::ThisCall),
+            "vectorcall" => Ok(Self::Vectorcall),
+            "aapcs" => Ok(Self::Aapcs),
+            "win64" => Ok(Self::Win64),
+            _ => Err(format!("Invalid or unknown ABI {:?}", s)),
+        }
+    }
+}
+
+impl std::fmt::Display for Abi {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        let s = match *self {
+            Self::C => "C",
+            Self::Stdcall => "stdcall",
+            Self::Fastcall => "fastcall",
+            Self::ThisCall => "thiscall",
+            Self::Vectorcall => "vectorcall",
+            Self::Aapcs => "aapcs",
+            Self::Win64 => "win64",
+        };
+
+        s.fmt(f)
+    }
+}
+
+impl quote::ToTokens for Abi {
+    fn to_tokens(&self, tokens: &mut proc_macro2::TokenStream) {
+        let abi = self.to_string();
+        tokens.append_all(quote! { #abi });
+    }
+}
+
+/// An ABI extracted from a clang cursor.
+#[derive(Debug, Copy, Clone)]
+pub(crate) enum ClangAbi {
+    Known(Abi),
     /// An unknown or invalid ABI.
     Unknown(CXCallingConv),
 }
 
-impl Abi {
+impl ClangAbi {
     /// Returns whether this Abi is known or not.
     fn is_unknown(&self) -> bool {
-        matches!(*self, Abi::Unknown(..))
+        matches!(*self, ClangAbi::Unknown(..))
     }
 }
 
-impl quote::ToTokens for Abi {
+impl quote::ToTokens for ClangAbi {
     fn to_tokens(&self, tokens: &mut proc_macro2::TokenStream) {
-        tokens.append_all(match *self {
-            Abi::C => quote! { "C" },
-            Abi::Stdcall => quote! { "stdcall" },
-            Abi::Fastcall => quote! { "fastcall" },
-            Abi::ThisCall => quote! { "thiscall" },
-            Abi::Vectorcall => quote! { "vectorcall" },
-            Abi::Aapcs => quote! { "aapcs" },
-            Abi::Win64 => quote! { "win64" },
-            Abi::Unknown(cc) => panic!(
+        match *self {
+            Self::Known(abi) => abi.to_tokens(tokens),
+            Self::Unknown(cc) => panic!(
                 "Cannot turn unknown calling convention to tokens: {:?}",
                 cc
             ),
-        });
+        }
     }
 }
 
@@ -234,21 +275,21 @@ pub struct FunctionSig {
     must_use: bool,
 
     /// The ABI of this function.
-    abi: Abi,
+    abi: ClangAbi,
 }
 
-fn get_abi(cc: CXCallingConv) -> Abi {
+fn get_abi(cc: CXCallingConv) -> ClangAbi {
     use clang_sys::*;
     match cc {
-        CXCallingConv_Default => Abi::C,
-        CXCallingConv_C => Abi::C,
-        CXCallingConv_X86StdCall => Abi::Stdcall,
-        CXCallingConv_X86FastCall => Abi::Fastcall,
-        CXCallingConv_X86ThisCall => Abi::ThisCall,
-        CXCallingConv_X86VectorCall => Abi::Vectorcall,
-        CXCallingConv_AAPCS => Abi::Aapcs,
-        CXCallingConv_X86_64Win64 => Abi::Win64,
-        other => Abi::Unknown(other),
+        CXCallingConv_Default => ClangAbi::Known(Abi::C),
+        CXCallingConv_C => ClangAbi::Known(Abi::C),
+        CXCallingConv_X86StdCall => ClangAbi::Known(Abi::Stdcall),
+        CXCallingConv_X86FastCall => ClangAbi::Known(Abi::Fastcall),
+        CXCallingConv_X86ThisCall => ClangAbi::Known(Abi::ThisCall),
+        CXCallingConv_X86VectorCall => ClangAbi::Known(Abi::Vectorcall),
+        CXCallingConv_AAPCS => ClangAbi::Known(Abi::Aapcs),
+        CXCallingConv_X86_64Win64 => ClangAbi::Known(Abi::Win64),
+        other => ClangAbi::Unknown(other),
     }
 }
 
@@ -354,25 +395,6 @@ fn args_from_ty_and_cursor(
 }
 
 impl FunctionSig {
-    /// Construct a new function signature.
-    pub fn new(
-        return_type: TypeId,
-        argument_types: Vec<(Option<String>, TypeId)>,
-        is_variadic: bool,
-        is_divergent: bool,
-        must_use: bool,
-        abi: Abi,
-    ) -> Self {
-        FunctionSig {
-            return_type,
-            argument_types,
-            is_variadic,
-            is_divergent,
-            must_use,
-            abi,
-        }
-    }
-
     /// Construct a new function signature from the given Clang type.
     pub fn from_ty(
         ty: &clang::Type,
@@ -540,20 +562,21 @@ impl FunctionSig {
                 call_conv = cursor_call_conv;
             }
         }
+
         let abi = get_abi(call_conv);
 
         if abi.is_unknown() {
             warn!("Unknown calling convention: {:?}", call_conv);
         }
 
-        Ok(Self::new(
-            ret,
-            args,
-            ty.is_variadic(),
+        Ok(FunctionSig {
+            return_type: ret,
+            argument_types: args,
+            is_variadic: ty.is_variadic(),
             is_divergent,
             must_use,
             abi,
-        ))
+        })
     }
 
     /// Get this function signature's return type.
@@ -567,8 +590,27 @@ impl FunctionSig {
     }
 
     /// Get this function signature's ABI.
-    pub fn abi(&self) -> Abi {
-        self.abi
+    pub(crate) fn abi(
+        &self,
+        ctx: &BindgenContext,
+        name: Option<&str>,
+    ) -> ClangAbi {
+        // FIXME (pvdrz): Try to do this check lazily instead. Maybe store the ABI inside `ctx`
+        // instead?.
+        if let Some(name) = name {
+            if let Some((abi, _)) = ctx
+                .options()
+                .abi_overrides
+                .iter()
+                .find(|(_, regex_set)| regex_set.matches(name))
+            {
+                ClangAbi::Known(*abi)
+            } else {
+                self.abi
+            }
+        } else {
+            self.abi
+        }
     }
 
     /// Is this function signature variadic?
@@ -598,7 +640,7 @@ impl FunctionSig {
             return false;
         }
 
-        matches!(self.abi, Abi::C | Abi::Unknown(..))
+        matches!(self.abi, ClangAbi::Known(Abi::C) | ClangAbi::Unknown(..))
     }
 
     pub(crate) fn is_divergent(&self) -> bool {
diff --git a/bindgen/lib.rs b/bindgen/lib.rs
index 5e04acfe..10410734 100644
--- a/bindgen/lib.rs
+++ b/bindgen/lib.rs
@@ -74,6 +74,7 @@ pub use crate::features::{
     RustTarget, LATEST_STABLE_RUST, RUST_TARGET_STRINGS,
 };
 use crate::ir::context::{BindgenContext, ItemId};
+pub use crate::ir::function::Abi;
 use crate::ir::item::Item;
 use crate::parse::{ClangItemParser, ParseError};
 use crate::regex_set::RegexSet;
@@ -364,6 +365,13 @@ impl Builder {
             }
         }
 
+        for (abi, set) in &self.options.abi_overrides {
+            for item in set.get_items() {
+                output_vector.push("--override-abi".to_owned());
+                output_vector.push(format!("{}={}", item, abi));
+            }
+        }
+
         if !self.options.layout_tests {
             output_vector.push("--no-layout-tests".into());
         }
@@ -1774,6 +1782,16 @@ impl Builder {
         self.options.c_naming = doit;
         self
     }
+
+    /// Override the ABI of a given function. Regular expressions are supported.
+    pub fn override_abi<T: Into<String>>(mut self, abi: Abi, arg: T) -> Self {
+        self.options
+            .abi_overrides
+            .entry(abi)
+            .or_default()
+            .insert(arg.into());
+        self
+    }
 }
 
 /// Configuration options for generated bindings.
@@ -2109,11 +2127,13 @@ struct BindgenOptions {
 
     /// Deduplicate `extern` blocks.
     merge_extern_blocks: bool,
+
+    abi_overrides: HashMap<Abi, RegexSet>,
 }
 
 impl BindgenOptions {
     fn build(&mut self) {
-        let mut regex_sets = [
+        let regex_sets = [
             &mut self.allowlisted_vars,
             &mut self.allowlisted_types,
             &mut self.allowlisted_functions,
@@ -2143,7 +2163,7 @@ impl BindgenOptions {
             &mut self.must_use_types,
         ];
         let record_matches = self.record_matches;
-        for regex_set in &mut regex_sets {
+        for regex_set in self.abi_overrides.values_mut().chain(regex_sets) {
             regex_set.build(record_matches);
         }
     }
@@ -2280,6 +2300,7 @@ impl Default for BindgenOptions {
             vtable_generation: false,
             sort_semantically: false,
             merge_extern_blocks: false,
+            abi_overrides: Default::default(),
         }
     }
 }