support 64-bit file offsets on systems with 32-bit off_t

Adds large file support (64-bit file positions) to the existing Nix
API when targetting systems that have separate 32-bit and 64-bit
APIs.

On many platforms that support 64-bit file positions, this support is
present in the standard I/O functions. On others, there have
historically been two APIs: the standard API, which limits file size to
2**31-1 bytes, and a separate API that suffixes function names with "64"
and supports 64-bit file sizes. This "transitional interface for 64-bit
file offsets" is not present on every platform. As a result, using the
*64 API will make programs non-portable, whereas using the standard API
will result in programs that cannot handle large file offsets on some
systems, even if those systems actually do have a way to handle
large file offsets.

This change enables 64-bit file offsets on the following platforms
where the standard API only provides 32-bit offsets:

 - 32-bit Linux with glibc.
 - 32-bit Android.

To support large files with a consistent API across platforms, this
change makes Nix functions call the 64-bit capable equivalents where
necessary. Other C libraries may not provide the *64 API, so we continue
to call the standard functions on those.

Broadly, the change consists of 5 parts:

 1. Uses of libc::off_t have are replaced by i64. This provides a
    consistent API across platforms and allows 64-bit offsets to
    be used even when libc::off_t is 32-bit. This is similar to
    std::fs, which uses u64 for file positions. Nix uses i64
    because off_t is a signed type. Into<i64> is used where appropriate
    so that existing code that uses libc::off_t will continue to
    work without changes.

 2. A largefile_fn macro that is used to select the large-file-capable
    version of a function. E.g. largefile_fn![pwrite] is equivalent
    to libc::pwrite64 on glibc/Linux and plain libc::pwrite on
    systems that don't have or need libc::pwrite64.

 3. A new require_largefile macro that is used to skip tests that
    require libc::off_t to be larger than 32 bits.

 4. Changes to fallocate, ftruncate, lseek, mmap, open, openat,
    posix_fadvise, posix_fallocate, pread, preadv, pwrite, pwritev,
    sendfile, and truncate, making them support large files.

 5. A set of test_*_largefile tests to verify that each of the
    previously mentioned functions now works with files whose size
    requires more than 32 bits to represent.

A few functions are still limited to 32-bit file sizes after this
change. This includes the aio* functions, the *stat* functions, and
mkstemp(). Adding large file support to those requires a bit more
work than simply calling a different function and is therefore left
for later.

Author: inglorion

src/fcntl.rs

@@ -2,7 +2,7 @@
 use crate::errno::Errno;
 #[cfg(all(target_os = "freebsd", target_arch = "x86_64"))]
 use core::slice;
-use libc::{c_int, c_uint, size_t, ssize_t};
+use libc::{self, c_int, c_uint, size_t, ssize_t};
 #[cfg(any(
     target_os = "netbsd",
     apple_targets,
@@ -163,8 +163,12 @@ libc_bitflags!(
                   all(target_os = "linux", not(target_env = "musl"), not(target_env = "ohos")),
                   target_os = "redox"))]
         O_FSYNC;
-        /// Allow files whose sizes can't be represented in an `off_t` to be opened.
+        /// On 32-bit Linux, O_LARGEFILE allows the use of file
+        /// offsets greater than 32 bits. Nix accepts the flag for
+        /// compatibility, but always opens files in large file mode
+        /// even if it isn't specified.
         #[cfg(linux_android)]
+        #[cfg_attr(docsrs, doc(cfg(all())))]
         O_LARGEFILE;
         /// Do not update the file last access time during `read(2)`s.
         #[cfg(linux_android)]
@@ -249,7 +253,7 @@ pub fn open<P: ?Sized + NixPath>(
     use std::os::fd::FromRawFd;
 
     let fd = path.with_nix_path(|cstr| unsafe {
-        libc::open(cstr.as_ptr(), oflag.bits(), mode.bits() as c_uint)
+        largefile_fn![open](cstr.as_ptr(), oflag.bits(), mode.bits() as c_uint)
     })?;
     Errno::result(fd)?;
 
@@ -280,7 +284,12 @@ pub fn openat<P: ?Sized + NixPath, Fd: std::os::fd::AsFd>(
     use std::os::fd::FromRawFd;
 
     let fd = path.with_nix_path(|cstr| unsafe {
-        libc::openat(dirfd.as_fd().as_raw_fd(), cstr.as_ptr(), oflag.bits(), mode.bits() as c_uint)
+         largefile_fn![openat](
+             dirfd.as_fd().as_raw_fd(),
+             cstr.as_ptr(),
+             oflag.bits(),
+             mode.bits() as c_uint,
+         )
     })?;
     Errno::result(fd)?;
 
@@ -1303,23 +1312,26 @@ feature! {
 /// file referred to by fd.
 #[cfg(target_os = "linux")]
 #[cfg(feature = "fs")]
-pub fn fallocate<Fd: std::os::fd::AsFd>(
+pub fn fallocate<Fd: std::os::fd::AsFd, Off: Into<i64>>(
     fd: Fd,
     mode: FallocateFlags,
-    offset: libc::off_t,
-    len: libc::off_t,
+    offset: Off,
+    len: Off,
 ) -> Result<()> {
     use std::os::fd::AsRawFd;
 
-    let res = unsafe { libc::fallocate(fd.as_fd().as_raw_fd(), mode.bits(), offset, len) };
+    let res = unsafe {
+        largefile_fn![fallocate](
+            fd.as_fd().as_raw_fd(), mode.bits(), offset.into(), len.into())
+    };
     Errno::result(res).map(drop)
 }
 
 /// Argument to [`fspacectl`] describing the range to zero.  The first member is
 /// the file offset, and the second is the length of the region.
 #[cfg(any(target_os = "freebsd"))]
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
-pub struct SpacectlRange(pub libc::off_t, pub libc::off_t);
+pub struct SpacectlRange(pub i64, pub i64);
 
 #[cfg(any(target_os = "freebsd"))]
 impl SpacectlRange {
@@ -1334,13 +1346,13 @@ impl SpacectlRange {
 
     /// Remaining length of the range
     #[inline]
-    pub fn len(&self) -> libc::off_t {
+    pub fn len(&self) -> i64 {
         self.1
     }
 
     /// Next file offset to operate on
     #[inline]
-    pub fn offset(&self) -> libc::off_t {
+    pub fn offset(&self) -> i64 {
         self.0
     }
 }
@@ -1437,16 +1449,16 @@ pub fn fspacectl<Fd: std::os::fd::AsFd>(fd: Fd, range: SpacectlRange) -> Result<
 /// ```
 #[cfg(target_os = "freebsd")]
 #[inline] // Delays codegen, preventing linker errors with dylibs and --no-allow-shlib-undefined
-pub fn fspacectl_all<Fd: std::os::fd::AsFd>(
+pub fn fspacectl_all<Fd: std::os::fd::AsFd, Off: Into<i64>>(
     fd: Fd,
-    offset: libc::off_t,
-    len: libc::off_t,
+    offset: Off,
+    len: Off,
 ) -> Result<()> {
     use std::os::fd::AsRawFd;
 
     let mut rqsr = libc::spacectl_range {
-        r_offset: offset,
-        r_len: len,
+        r_offset: offset.into(),
+        r_len: len.into(),
     };
     while rqsr.r_len > 0 {
         let res = unsafe {
@@ -1505,15 +1517,22 @@ mod posix_fadvise {
     ///
     /// # See Also
     /// * [`posix_fadvise`](https://pubs.opengroup.org/onlinepubs/9699919799/functions/posix_fadvise.html)
-    pub fn posix_fadvise<Fd: std::os::fd::AsFd>(
+    pub fn posix_fadvise<Fd: std::os::fd::AsFd, Off: Into<i64>>(
         fd: Fd,
-        offset: libc::off_t,
-        len: libc::off_t,
+        offset: Off,
+        len: Off,
         advice: PosixFadviseAdvice,
     ) -> Result<()> {
         use std::os::fd::AsRawFd;
 
-        let res = unsafe { libc::posix_fadvise(fd.as_fd().as_raw_fd(), offset, len, advice as libc::c_int) };
+        let res = unsafe {
+            largefile_fn![posix_fadvise](
+                fd.as_fd().as_raw_fd(),
+                offset.into(),
+                len.into(),
+                advice as libc::c_int,
+            )
+        };
 
         if res == 0 {
             Ok(())
@@ -1535,14 +1554,20 @@ mod posix_fadvise {
     target_os = "fuchsia",
     target_os = "wasi",
 ))]
-pub fn posix_fallocate<Fd: std::os::fd::AsFd>(
+pub fn posix_fallocate<Fd: std::os::fd::AsFd, Off: Into<i64>>(
     fd: Fd,
-    offset: libc::off_t,
-    len: libc::off_t,
+    offset: Off,
+    len: Off,
 ) -> Result<()> {
     use std::os::fd::AsRawFd;
 
-    let res = unsafe { libc::posix_fallocate(fd.as_fd().as_raw_fd(), offset, len) };
+    let res = unsafe {
+        largefile_fn![posix_fallocate](
+            fd.as_fd().as_raw_fd(),
+            offset.into(),
+            len.into(),
+        )
+    };
     match Errno::result(res) {
         Err(err) => Err(err),
         Ok(0) => Ok(()),

src/macros.rs

@@ -1,3 +1,5 @@
+use cfg_if::cfg_if;
+
 // Thanks to Tokio for this macro
 macro_rules! feature {
     (
@@ -329,3 +331,48 @@ macro_rules! libc_enum {
         }
     };
 }
+
+cfg_if! {
+    if #[cfg(any(
+        all(target_os = "linux", target_env = "gnu"),
+        target_os = "android",
+    ))] {
+        /// Function variant that supports large file positions.
+        ///
+        /// On some platforms, the standard I/O functions support a limited
+        /// range of file positions, and there is an alternate set of
+        /// functions that support larger file positions. This macro takes
+        /// the identifier of a standard I/O function and returns the
+        /// identifier of the corresponding I/O function with large file
+        /// support.
+        #[allow(unused_macro_rules)]
+        macro_rules! largefile_fn {
+            [fallocate] => (libc::fallocate64);
+            [ftruncate] => (libc::ftruncate64);
+            [lseek] => (libc::lseek64);
+            [mmap] => (libc::mmap64);
+            [open] => (libc::open64);
+            [openat] => (libc::openat64);
+            [posix_fadvise] => (libc::posix_fadvise64);
+            [posix_fallocate] => (libc::posix_fallocate64);
+            [pread] => (libc::pread64);
+            [preadv] => (libc::preadv64);
+            [pwrite] => (libc::pwrite64);
+            [pwritev] => (libc::pwritev64);
+            [sendfile] => (libc::sendfile64);
+            [truncate] => (libc::truncate64);
+        }
+    } else {
+        /// Function variant that supports large file positions.
+        ///
+        /// On some platforms, the standard I/O functions support a limited
+        /// range of file positions, and there is an alternate set of
+        /// functions that support larger file positions. This macro takes
+        /// the identifier of a standard I/O function and returns the
+        /// identifier of the corresponding I/O function with large file
+        /// support.
+        macro_rules! largefile_fn {
+            [$id:ident] => (libc::$id);
+        }
+    }
+}

src/sys/mman.rs

@@ -7,7 +7,7 @@ use crate::Result;
 #[cfg(not(target_os = "android"))]
 #[cfg(feature = "fs")]
 use crate::{fcntl::OFlag, sys::stat::Mode};
-use libc::{self, c_int, c_void, off_t, size_t};
+use libc::{self, c_int, c_void, size_t};
 use std::ptr::NonNull;
 use std::{
     num::NonZeroUsize,
@@ -392,19 +392,20 @@ pub fn munlockall() -> Result<()> {
 /// See the [`mmap(2)`] man page for detailed requirements.
 ///
 /// [`mmap(2)`]: https://man7.org/linux/man-pages/man2/mmap.2.html
-pub unsafe fn mmap<F: AsFd>(
+pub unsafe fn mmap<F: AsFd, Off: Into<i64>>(
     addr: Option<NonZeroUsize>,
     length: NonZeroUsize,
     prot: ProtFlags,
     flags: MapFlags,
     f: F,
-    offset: off_t,
+    offset: Off,
 ) -> Result<NonNull<c_void>> {
     let ptr = addr.map_or(std::ptr::null_mut(), |a| a.get() as *mut c_void);
 
     let fd = f.as_fd().as_raw_fd();
     let ret = unsafe {
-        libc::mmap(ptr, length.into(), prot.bits(), flags.bits(), fd, offset)
+        largefile_fn![mmap](
+            ptr, length.into(), prot.bits(), flags.bits(), fd, offset.into())
     };
 
     if ret == libc::MAP_FAILED {

src/sys/sendfile.rs

@@ -4,7 +4,7 @@ use cfg_if::cfg_if;
 use std::os::unix::io::{AsFd, AsRawFd};
 use std::ptr;
 
-use libc::{self, off_t};
+use libc;
 
 use crate::errno::Errno;
 use crate::Result;
@@ -26,14 +26,14 @@ use crate::Result;
 pub fn sendfile<F1: AsFd, F2: AsFd>(
     out_fd: F1,
     in_fd: F2,
-    offset: Option<&mut off_t>,
+    offset: Option<&mut i64>,
     count: usize,
 ) -> Result<usize> {
     let offset = offset
         .map(|offset| offset as *mut _)
         .unwrap_or(ptr::null_mut());
     let ret = unsafe {
-        libc::sendfile(
+        largefile_fn![sendfile](
             out_fd.as_fd().as_raw_fd(),
             in_fd.as_fd().as_raw_fd(),
             offset,
@@ -133,20 +133,20 @@ cfg_if! {
         impl<'fd> SendfileVec<'fd> {
             /// initialises SendfileVec to send data directly from the process's address space
             /// same in C with sfv_fd set to SFV_FD_SELF.
-            pub fn newself(
-                off: off_t,
+            pub fn newself<Off: Into<i64>>(
+                off: Off,
                 len: usize
             ) -> Self {
-                Self{raw: libc::sendfilevec_t{sfv_fd: libc::SFV_FD_SELF, sfv_flag: 0, sfv_off: off, sfv_len: len}, phantom: PhantomData}
+                Self{raw: libc::sendfilevec_t{sfv_fd: libc::SFV_FD_SELF, sfv_flag: 0, sfv_off: off.into(), sfv_len: len}, phantom: PhantomData}
             }
 
             /// initialises SendfileVec to send data from `fd`.
-            pub fn new(
+            pub fn new<Off: Into<i64>>(
                 fd: BorrowedFd<'fd>,
-                off: off_t,
+                off: Off,
                 len: usize
             ) -> SendfileVec<'fd> {
-                Self{raw: libc::sendfilevec_t{sfv_fd: fd.as_raw_fd(), sfv_flag: 0, sfv_off:off, sfv_len: len}, phantom: PhantomData}
+                Self{raw: libc::sendfilevec_t{sfv_fd: fd.as_raw_fd(), sfv_flag: 0, sfv_off:off.into(), sfv_len: len}, phantom: PhantomData}
             }
         }
 
@@ -208,19 +208,19 @@ cfg_if! {
         pub fn sendfile<F1: AsFd, F2: AsFd>(
             in_fd: F1,
             out_sock: F2,
-            offset: off_t,
+            offset: i64,
             count: Option<usize>,
             headers: Option<&[&[u8]]>,
             trailers: Option<&[&[u8]]>,
             flags: SfFlags,
             readahead: u16
-        ) -> (Result<()>, off_t) {
+        ) -> (Result<()>, i64) {
             // Readahead goes in upper 16 bits
             // Flags goes in lower 16 bits
             // see `man 2 sendfile`
             let ra32 = u32::from(readahead);
             let flags: u32 = (ra32 << 16) | (flags.bits() as u32);
-            let mut bytes_sent: off_t = 0;
+            let mut bytes_sent: i64 = 0;
             let hdtr = headers.or(trailers).map(|_| SendfileHeaderTrailer::new(headers, trailers));
             let hdtr_ptr = hdtr.as_ref().map_or(ptr::null(), |s| &s.raw as *const libc::sf_hdtr);
             let return_code = unsafe {
@@ -229,7 +229,7 @@ cfg_if! {
                                offset,
                                count.unwrap_or(0),
                                hdtr_ptr as *mut libc::sf_hdtr,
-                               &mut bytes_sent as *mut off_t,
+                               &mut bytes_sent as *mut i64,
                                flags as c_int)
             };
             (Errno::result(return_code).and(Ok(())), bytes_sent)
@@ -258,12 +258,12 @@ cfg_if! {
         pub fn sendfile<F1: AsFd, F2: AsFd>(
             in_fd: F1,
             out_sock: F2,
-            offset: off_t,
+            offset: i64,
             count: Option<usize>,
             headers: Option<&[&[u8]]>,
             trailers: Option<&[&[u8]]>,
-        ) -> (Result<()>, off_t) {
-            let mut bytes_sent: off_t = 0;
+        ) -> (Result<()>, i64) {
+            let mut bytes_sent: i64 = 0;
             let hdtr = headers.or(trailers).map(|_| SendfileHeaderTrailer::new(headers, trailers));
             let hdtr_ptr = hdtr.as_ref().map_or(ptr::null(), |s| &s.raw as *const libc::sf_hdtr);
             let return_code = unsafe {
@@ -272,7 +272,7 @@ cfg_if! {
                                offset,
                                count.unwrap_or(0),
                                hdtr_ptr as *mut libc::sf_hdtr,
-                               &mut bytes_sent as *mut off_t,
+                               &mut bytes_sent as *mut i64,
                                0)
             };
             (Errno::result(return_code).and(Ok(())), bytes_sent)
@@ -304,19 +304,19 @@ cfg_if! {
         pub fn sendfile<F1: AsFd, F2: AsFd>(
             in_fd: F1,
             out_sock: F2,
-            offset: off_t,
-            count: Option<off_t>,
+            offset: i64,
+            count: Option<i64>,
             headers: Option<&[&[u8]]>,
             trailers: Option<&[&[u8]]>
-        ) -> (Result<()>, off_t) {
+        ) -> (Result<()>, i64) {
             let mut len = count.unwrap_or(0);
             let hdtr = headers.or(trailers).map(|_| SendfileHeaderTrailer::new(headers, trailers));
             let hdtr_ptr = hdtr.as_ref().map_or(ptr::null(), |s| &s.raw as *const libc::sf_hdtr);
             let return_code = unsafe {
                 libc::sendfile(in_fd.as_fd().as_raw_fd(),
                                out_sock.as_fd().as_raw_fd(),
                                offset,
-                               &mut len as *mut off_t,
+                               &mut len as *mut i64,
                                hdtr_ptr as *mut libc::sf_hdtr,
                                0)
             };