Map size_t to usize by default and check compatibility (fixes #1901, #1903)

This addresses the underlying issue identified in #1671, that size_t
(integer that can hold any object size) isn't guaranteed to match usize,
which is defined more like uintptr_t (integer that can hold any
pointer). However, on almost all platforms, this is true, and in fact
Rust already uses usize extensively in contexts where size_t would be
more appropriate, such as slice indexing. So, it's better for ergonomics
when interfacing with C code to map the C size_t type to usize. (See
also discussion in rust-lang/rust#65473 about how usize really should be
defined as size_t, not uintptr_t.)

The previous fix for #1671 removed the special case for size_t and
defaulted to binding it as a normal typedef.  This change effectively
reverts that and goes back to mapping size_t to usize (and ssize_t to
isize), but also ensures that if size_t is emitted, the typedef'd type
of size_t in fact is compatible with usize (defined by checking that the
size and alignment match the target pointer width). For (hypothetical)
platforms where this is not true, or for compatibility with the default
behavior of bindgen between 0.53 and this commit, onwards, you can
disable this mapping with --no-size_t-is-usize.

asd.fish

@@ -0,0 +1,31 @@
+for name in (rg -lF addr_of! tests/expectations | sd '.*/(.*).rs' '$1')
+    set path (fd --glob "$name.*" tests/headers)
+    if test -n "$path"
+
+        set flags (rg -F "// bindgen-flags" $path)
+        if test -n "$flags"
+            set minor (rg ".*\-\-rust\-target[ =]1.(\d+).*" $path -r '$1')
+            if test -n "$minor"
+                if test $minor -gt 47
+                    echo  $path needs to change the version from 1.$minor to 1.47
+                    sd -s "1.$minor" "1.47" $path
+                else
+                    echo $path uses version 1.$minor and that is fine
+                end
+            else
+                echo $path does not have the `--rust-target` flag
+                sd "// bindgen-flags: (.*)" '// bindgen-flags: --rust-target 1.47 $1' $path
+            end
+        else
+            echo $path does not have the flags at all
+            set contents (echo -e "// bindgen-flags: --rust-target 1.47\n"; cat $path)
+            rm $path
+            touch $path
+            for line in $contents
+                echo $line >> $path
+            end
+        end
+    else
+        echo $name headers not found
+    end
+end

src/codegen/mod.rs

@@ -836,9 +836,34 @@ impl CodeGenerator for Type {
                 }
 
                 // If this is a known named type, disallow generating anything
-                // for it too.
+                // for it too. If size_t -> usize conversions are enabled, we
+                // need to check that these conversions are permissible, but
+                // nothing needs to be generated, still.
                 let spelling = self.name().expect("Unnamed alias?");
                 if utils::type_from_named(ctx, spelling).is_some() {
+                    if let "size_t" | "ssize_t" = spelling {
+                        let layout = inner_item
+                            .kind()
+                            .expect_type()
+                            .layout(ctx)
+                            .expect("No layout?");
+                        assert_eq!(
+                            layout.size,
+                            ctx.target_pointer_size(),
+                            "Target platform requires `--no-size_t-is-usize`. The size of `{}` ({}) does not match the target pointer size ({})",
+                            spelling,
+                            layout.size,
+                            ctx.target_pointer_size(),
+                            );
+                        assert_eq!(
+                            layout.align,
+                            ctx.target_pointer_size(),
+                            "Target platform requires `--no-size_t-is-usize`. The alignment of `{}` ({}) does not match the target pointer size ({})",
+                            spelling,
+                            layout.align,
+                            ctx.target_pointer_size(),
+                        );
+                    }
                     return;
                 }
 

src/lib.rs

@@ -564,8 +564,8 @@ impl Builder {
             output_vector.push("--no-record-matches".into());
         }
 
-        if self.options.size_t_is_usize {
-            output_vector.push("--size_t-is-usize".into());
+        if !self.options.size_t_is_usize {
+            output_vector.push("--no-size_t-is-usize".into());
         }
 
         if !self.options.rustfmt_bindings {
@@ -2253,7 +2253,7 @@ impl Default for BindgenOptions {
             time_phases: false,
             record_matches: true,
             rustfmt_bindings: true,
-            size_t_is_usize: false,
+            size_t_is_usize: true,
             rustfmt_configuration_file: None,
             no_partialeq_types: Default::default(),
             no_copy_types: Default::default(),

src/options.rs

@@ -465,7 +465,12 @@ where
                 ),
             Arg::new("size_t-is-usize")
                 .long("size_t-is-usize")
-                .help("Translate size_t to usize."),
+                .help("Ignored - this is enabled by default.")
+                .hidden(true),
+            Arg::with_name("no-size_t-is-usize")
+                .long("no-size_t-is-usize")
+                .help("Do not bind size_t as usize (useful on platforms \
+                       where those types are incompatible)."),
             Arg::new("no-rustfmt-bindings")
                 .long("no-rustfmt-bindings")
                 .help("Do not format the generated bindings with rustfmt."),
@@ -975,8 +980,8 @@ where
         builder = builder.record_matches(false);
     }
 
-    if matches.is_present("size_t-is-usize") {
-        builder = builder.size_t_is_usize(true);
+    if matches.is_present("no-size_t-is-usize") {
+        builder = builder.size_t_is_usize(false);
     }
 
     let no_rustfmt_bindings = matches.is_present("no-rustfmt-bindings");

tests/expectations/tests/blocks-signature.rs

@@ -7,7 +7,6 @@
 #![cfg(target_os = "macos")]
 
 extern crate block;
-pub type size_t = ::std::os::raw::c_ulonglong;
 extern "C" {
     #[link_name = "\u{1}_Z8atexit_bU13block_pointerFvvE"]
     pub fn atexit_b(arg1: _bindgen_ty_id_33);
@@ -82,16 +81,11 @@ impl Default for contains_block_pointers {
 }
 pub type _bindgen_ty_id_33 = *const ::block::Block<(), ()>;
 pub type _bindgen_ty_id_40 = *const ::block::Block<
-    (
-        dispatch_data_t,
-        size_t,
-        *const ::std::os::raw::c_void,
-        size_t,
-    ),
+    (dispatch_data_t, usize, *const ::std::os::raw::c_void, usize),
     bool,
 >;
-pub type _bindgen_ty_id_50 = *const ::block::Block<(size_t,), ()>;
-pub type _bindgen_ty_id_56 = *const ::block::Block<(size_t,), ()>;
+pub type _bindgen_ty_id_50 = *const ::block::Block<(usize,), ()>;
+pub type _bindgen_ty_id_56 = *const ::block::Block<(usize,), ()>;
 pub type contains_block_pointers__bindgen_ty_id_61 =
     *const ::block::Block<(::std::os::raw::c_int,), ()>;
 pub type _bindgen_ty_id_68 =

tests/expectations/tests/blocks.rs

@@ -6,7 +6,6 @@
 )]
 #![cfg(target_os = "macos")]
 
-pub type size_t = ::std::os::raw::c_ulonglong;
 extern "C" {
     #[link_name = "\u{1}_Z8atexit_bU13block_pointerFvvE"]
     pub fn atexit_b(arg1: *mut ::std::os::raw::c_void);

tests/expectations/tests/issue-1498.rs

@@ -5,7 +5,6 @@
     non_upper_case_globals
 )]
 
-pub type size_t = u64;
 #[repr(C, packed)]
 #[derive(Copy, Clone)]
 pub struct rte_memseg {
@@ -13,7 +12,7 @@ pub struct rte_memseg {
     pub phys_addr: u64,
     pub __bindgen_anon_1: rte_memseg__bindgen_ty_1,
     ///< Length of the segment.
-    pub len: size_t,
+    pub len: usize,
     ///< The pagesize of underlying memory
     pub hugepage_sz: u64,
     ///< NUMA socket ID.

tests/expectations/tests/jsval_layout_opaque.rs

@@ -94,7 +94,6 @@ where
 pub const JSVAL_TAG_SHIFT: u32 = 47;
 pub const JSVAL_PAYLOAD_MASK: u64 = 140737488355327;
 pub const JSVAL_TAG_MASK: i64 = -140737488355328;
-pub type size_t = ::std::os::raw::c_ulonglong;
 #[repr(u8)]
 #[derive(Debug, Copy, Clone, Hash, PartialEq, Eq)]
 pub enum JSValueType {
@@ -186,7 +185,7 @@ pub union jsval_layout {
     pub s: jsval_layout__bindgen_ty_2,
     pub asDouble: f64,
     pub asPtr: *mut ::std::os::raw::c_void,
-    pub asWord: size_t,
+    pub asWord: usize,
     pub asUIntPtr: usize,
 }
 #[repr(C)]

tests/expectations/tests/jsval_layout_opaque_1_0.rs

@@ -137,7 +137,6 @@ impl<T> ::std::cmp::Eq for __BindgenUnionField<T> {}
 pub const JSVAL_TAG_SHIFT: u32 = 47;
 pub const JSVAL_PAYLOAD_MASK: u64 = 140737488355327;
 pub const JSVAL_TAG_MASK: i64 = -140737488355328;
-pub type size_t = ::std::os::raw::c_ulonglong;
 #[repr(u8)]
 #[derive(Debug, Copy, Clone, Hash, PartialEq, Eq)]
 pub enum JSValueType {
@@ -229,7 +228,7 @@ pub struct jsval_layout {
     pub s: __BindgenUnionField<jsval_layout__bindgen_ty_2>,
     pub asDouble: __BindgenUnionField<f64>,
     pub asPtr: __BindgenUnionField<*mut ::std::os::raw::c_void>,
-    pub asWord: __BindgenUnionField<size_t>,
+    pub asWord: __BindgenUnionField<usize>,
     pub asUIntPtr: __BindgenUnionField<usize>,
     pub bindgen_union_field: u64,
 }

tests/expectations/tests/layout_array.rs

@@ -9,7 +9,6 @@ pub const RTE_CACHE_LINE_SIZE: u32 = 64;
 pub const RTE_MEMPOOL_OPS_NAMESIZE: u32 = 32;
 pub const RTE_MEMPOOL_MAX_OPS_IDX: u32 = 16;
 pub const RTE_HEAP_NUM_FREELISTS: u32 = 13;
-pub type size_t = ::std::os::raw::c_longlong;
 #[repr(C)]
 #[derive(Debug, Copy, Clone)]
 pub struct rte_mempool {
@@ -279,7 +278,7 @@ pub struct malloc_heap {
     pub lock: rte_spinlock_t,
     pub free_head: [malloc_heap__bindgen_ty_1; 13usize],
     pub alloc_count: ::std::os::raw::c_uint,
-    pub total_size: size_t,
+    pub total_size: usize,
 }
 #[repr(C)]
 #[derive(Debug, Copy, Clone, Hash, PartialEq, Eq)]

tests/expectations/tests/msvc-no-usr.rs

@@ -5,11 +5,10 @@
     non_upper_case_globals
 )]
 
-pub type size_t = ::std::os::raw::c_ulonglong;
 #[repr(C)]
 #[derive(Debug, Default, Copy, Clone)]
 pub struct A {
-    pub foo: size_t,
+    pub foo: usize,
 }
 #[test]
 fn bindgen_test_layout_A() {

tests/expectations/tests/no_size_t_is_usize.rs

@@ -5,11 +5,13 @@
     non_upper_case_globals
 )]
 
+pub type size_t = ::std::os::raw::c_ulong;
+pub type ssize_t = ::std::os::raw::c_long;
 #[repr(C)]
 #[derive(Debug, Copy, Clone)]
 pub struct A {
-    pub len: usize,
-    pub offset: isize,
+    pub len: size_t,
+    pub offset: ssize_t,
     pub next: *mut A,
 }
 #[test]

tests/expectations/tests/nsBaseHashtable.rs

@@ -5,7 +5,6 @@
     non_upper_case_globals
 )]
 
-pub type size_t = ::std::os::raw::c_ulonglong;
 #[repr(C)]
 #[derive(Debug, Default, Copy, Clone)]
 pub struct nsBaseHashtableET {

tests/headers/no_size_t_is_usize.h

@@ -1,4 +1,4 @@
-// bindgen-flags: --size_t-is-usize
+// bindgen-flags: --no-size_t-is-usize
 
 typedef unsigned long size_t;
 typedef long ssize_t;