Implement a readdir64() shim for Linux

Partial fix for #1966.

Author: tavianator

src/shims/posix/fs.rs

@@ -16,6 +16,7 @@ use rustc_target::abi::{Align, Size};
 
 use crate::*;
 use helpers::{check_arg_count, immty_from_int_checked, immty_from_uint_checked};
+use shims::os_str::os_str_to_bytes;
 use shims::time::system_time_to_duration;
 
 #[derive(Debug)]
@@ -421,6 +422,24 @@ trait EvalContextExtPrivate<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, '
     }
 }
 
+/// An open directory, tracked by DirHandler.
+#[derive(Debug)]
+pub struct OpenDir {
+    /// The directory reader on the host.
+    read_dir: ReadDir,
+    /// The most recent entry returned by readdir()
+    entry: Pointer<Option<Tag>>,
+}
+
+impl OpenDir {
+    fn new(read_dir: ReadDir) -> Self {
+        Self {
+            read_dir,
+            entry: Pointer::null(),
+        }
+    }
+}
+
 #[derive(Debug)]
 pub struct DirHandler {
     /// Directory iterators used to emulate libc "directory streams", as used in opendir, readdir,
@@ -432,7 +451,7 @@ pub struct DirHandler {
     /// the corresponding ReadDir iterator from this map, and information from the next
     /// directory entry is returned. When closedir is called, the ReadDir iterator is removed from
     /// the map.
-    streams: FxHashMap<u64, ReadDir>,
+    streams: FxHashMap<u64, OpenDir>,
     /// ID number to be used by the next call to opendir
     next_id: u64,
 }
@@ -441,7 +460,7 @@ impl DirHandler {
     fn insert_new(&mut self, read_dir: ReadDir) -> u64 {
         let id = self.next_id;
         self.next_id += 1;
-        self.streams.try_insert(id, read_dir).unwrap();
+        self.streams.try_insert(id, OpenDir::new(read_dir)).unwrap();
         id
     }
 }
@@ -1196,6 +1215,85 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
         }
     }
 
+    fn linux_readdir64(&mut self, dirp_op: &OpTy<'tcx, Tag>) -> InterpResult<'tcx, Scalar<Tag>> {
+        let this = self.eval_context_mut();
+
+        this.assert_target_os("linux", "readdir64");
+
+        let dirp = this.read_scalar(dirp_op)?.to_machine_usize(this)?;
+
+        // Reject if isolation is enabled.
+        if let IsolatedOp::Reject(reject_with) = this.machine.isolated_op {
+            this.reject_in_isolation("`readdir`", reject_with)?;
+            let eacc = this.eval_libc("EACCES")?;
+            this.set_last_error(eacc)?;
+            return Ok(Scalar::null_ptr(this));
+        }
+
+        let open_dir = this.machine.dir_handler.streams.get_mut(&dirp).ok_or_else(|| {
+            err_unsup_format!("the DIR pointer passed to readdir64 did not come from opendir")
+        })?;
+
+        let entry = match open_dir.read_dir.next() {
+            Some(Ok(dir_entry)) => {
+                let ino64_t_layout = this.libc_ty_layout("ino64_t")?;
+                let off64_t_layout = this.libc_ty_layout("off64_t")?;
+                let c_ushort_layout = this.libc_ty_layout("c_ushort")?;
+                let c_uchar_layout = this.libc_ty_layout("c_uchar")?;
+
+                // If the host is a Unix system, fill in the inode number with its real value.
+                // If not, use 0 as a fallback value.
+                #[cfg(unix)]
+                let ino = std::os::unix::fs::DirEntryExt::ino(&dir_entry);
+                #[cfg(not(unix))]
+                let ino = 0u64;
+
+                let file_type = this.file_type_to_d_type(dir_entry.file_type())?;
+
+                let imms = [
+                    immty_from_uint_checked(ino, ino64_t_layout)?, // d_ino
+                    immty_from_uint_checked(0u128, off64_t_layout)?, // d_off
+                    immty_from_uint_checked(0u128, c_ushort_layout)?, // d_reclen
+                    immty_from_int_checked(file_type, c_uchar_layout)?, // d_type
+                ];
+                let name = dir_entry.file_name();
+                let name_bytes = os_str_to_bytes(&name)?;
+                let name_offset = imms.iter()
+                    .map(|imm| imm.layout.size.bytes())
+                    .sum::<u64>();
+                let size = name_offset
+                    .checked_add(u64::try_from(name_bytes.len()).unwrap())
+                    .unwrap()
+                    .checked_add(1)
+                    .unwrap();
+
+                let entry = this.malloc(size, /*zero_init:*/ false, MiriMemoryKind::C)?;
+                let entry_layout = this.layout_of(this.tcx.mk_array(this.tcx.types.u8, size))?;
+                let entry_place = MPlaceTy::from_aligned_ptr(entry, entry_layout);
+                this.write_packed_immediates(&entry_place, &imms)?;
+
+                let name_ptr = entry.offset(Size::from_bytes(name_offset), this)?;
+                this.memory.write_bytes(name_ptr, name_bytes.iter().copied().chain(std::iter::once(0)))?;
+
+                entry
+            },
+            None => {
+                // end of stream: return NULL
+                Pointer::null()
+            }
+            Some(Err(e)) => {
+                this.set_last_error_from_io_error(e.kind())?;
+                Pointer::null()
+            }
+        };
+
+        let open_dir = this.machine.dir_handler.streams.get_mut(&dirp).unwrap();
+        let old_entry = std::mem::replace(&mut open_dir.entry, entry);
+        this.free(old_entry, MiriMemoryKind::C)?;
+
+        Ok(Scalar::from_maybe_pointer(entry, this))
+    }
+
     fn linux_readdir64_r(
         &mut self,
         dirp_op: &OpTy<'tcx, Tag>,
@@ -1215,10 +1313,10 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
             return this.handle_not_found();
         }
 
-        let dir_iter = this.machine.dir_handler.streams.get_mut(&dirp).ok_or_else(|| {
+        let open_dir = this.machine.dir_handler.streams.get_mut(&dirp).ok_or_else(|| {
             err_unsup_format!("the DIR pointer passed to readdir64_r did not come from opendir")
         })?;
-        match dir_iter.next() {
+        match open_dir.read_dir.next() {
             Some(Ok(dir_entry)) => {
                 // Write into entry, write pointer to result, return 0 on success.
                 // The name is written with write_os_str_to_c_str, while the rest of the
@@ -1314,10 +1412,10 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
             return this.handle_not_found();
         }
 
-        let dir_iter = this.machine.dir_handler.streams.get_mut(&dirp).ok_or_else(|| {
+        let open_dir = this.machine.dir_handler.streams.get_mut(&dirp).ok_or_else(|| {
             err_unsup_format!("the DIR pointer passed to readdir_r did not come from opendir")
         })?;
-        match dir_iter.next() {
+        match open_dir.read_dir.next() {
             Some(Ok(dir_entry)) => {
                 // Write into entry, write pointer to result, return 0 on success.
                 // The name is written with write_os_str_to_c_str, while the rest of the
@@ -1408,8 +1506,9 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
             return this.handle_not_found();
         }
 
-        if let Some(dir_iter) = this.machine.dir_handler.streams.remove(&dirp) {
-            drop(dir_iter);
+        if let Some(open_dir) = this.machine.dir_handler.streams.remove(&dirp) {
+            this.free(open_dir.entry, MiriMemoryKind::C)?;
+            drop(open_dir);
             Ok(0)
         } else {
             this.handle_not_found()

src/shims/posix/linux/foreign_items.rs

@@ -43,6 +43,11 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
                 let result = this.opendir(name)?;
                 this.write_scalar(result, dest)?;
             }
+            "readdir64" => {
+                let &[ref dirp] = this.check_shim(abi, Abi::C { unwind: false }, link_name, args)?;
+                let result = this.linux_readdir64(dirp)?;
+                this.write_scalar(result, dest)?;
+            }
             "readdir64_r" => {
                 let &[ref dirp, ref entry, ref result] =
                     this.check_shim(abi, Abi::C { unwind: false }, link_name, args)?;

tests/run-pass/fs.rs

@@ -331,19 +331,17 @@ fn test_directory() {
     let path_2 = dir_path.join("test_file_2");
     drop(File::create(&path_2).unwrap());
     // Test that the files are present inside the directory
-    /* FIXME(1966) disabled due to missing readdir support
     let dir_iter = read_dir(&dir_path).unwrap();
     let mut file_names = dir_iter.map(|e| e.unwrap().file_name()).collect::<Vec<_>>();
     file_names.sort_unstable();
-    assert_eq!(file_names, vec!["test_file_1", "test_file_2"]); */
+    assert_eq!(file_names, vec!["test_file_1", "test_file_2"]);
     // Clean up the files in the directory
     remove_file(&path_1).unwrap();
     remove_file(&path_2).unwrap();
     // Now there should be nothing left in the directory.
-    /* FIXME(1966) disabled due to missing readdir support
     dir_iter = read_dir(&dir_path).unwrap();
     let file_names = dir_iter.map(|e| e.unwrap().file_name()).collect::<Vec<_>>();
-    assert!(file_names.is_empty());*/
+    assert!(file_names.is_empty());
 
     // Deleting the directory should succeed.
     remove_dir(&dir_path).unwrap();