Merge pull request rust-lang#4067 from CraftSpider/windows-metadata

Support getting file metadata on Windows

Author: RalfJung

src/tools/miri/Cargo.lock

@@ -538,6 +538,7 @@ name = "miri"
 version = "0.1.0"
 dependencies = [
  "aes",
+ "bitflags",
  "chrono",
  "chrono-tz",
  "colored",

src/tools/miri/Cargo.toml

@@ -26,6 +26,7 @@ measureme = "12"
 chrono = { version = "0.4.38", default-features = false }
 chrono-tz = "0.10"
 directories = "6"
+bitflags = "2.6"
 
 # Copied from `compiler/rustc/Cargo.toml`.
 # But only for some targets, it fails for others. Rustc configures this in its CI, but we can't

src/tools/miri/src/shims/files.rs

@@ -1,6 +1,7 @@
 use std::any::Any;
 use std::collections::BTreeMap;
-use std::io::{IsTerminal, SeekFrom, Write};
+use std::fs::{File, Metadata};
+use std::io::{IsTerminal, Seek, SeekFrom, Write};
 use std::marker::CoercePointee;
 use std::ops::Deref;
 use std::rc::{Rc, Weak};
@@ -192,7 +193,7 @@ pub trait FileDescription: std::fmt::Debug + FileDescriptionExt {
         false
     }
 
-    fn as_unix(&self) -> &dyn UnixFileDescription {
+    fn as_unix<'tcx>(&self, _ecx: &MiriInterpCx<'tcx>) -> &dyn UnixFileDescription {
         panic!("Not a unix file descriptor: {}", self.name());
     }
 }
@@ -278,6 +279,97 @@ impl FileDescription for io::Stderr {
     }
 }
 
+#[derive(Debug)]
+pub struct FileHandle {
+    pub(crate) file: File,
+    pub(crate) writable: bool,
+}
+
+impl FileDescription for FileHandle {
+    fn name(&self) -> &'static str {
+        "file"
+    }
+
+    fn read<'tcx>(
+        self: FileDescriptionRef<Self>,
+        communicate_allowed: bool,
+        ptr: Pointer,
+        len: usize,
+        ecx: &mut MiriInterpCx<'tcx>,
+        finish: DynMachineCallback<'tcx, Result<usize, IoError>>,
+    ) -> InterpResult<'tcx> {
+        assert!(communicate_allowed, "isolation should have prevented even opening a file");
+
+        let result = ecx.read_from_host(&self.file, len, ptr)?;
+        finish.call(ecx, result)
+    }
+
+    fn write<'tcx>(
+        self: FileDescriptionRef<Self>,
+        communicate_allowed: bool,
+        ptr: Pointer,
+        len: usize,
+        ecx: &mut MiriInterpCx<'tcx>,
+        finish: DynMachineCallback<'tcx, Result<usize, IoError>>,
+    ) -> InterpResult<'tcx> {
+        assert!(communicate_allowed, "isolation should have prevented even opening a file");
+
+        let result = ecx.write_to_host(&self.file, len, ptr)?;
+        finish.call(ecx, result)
+    }
+
+    fn seek<'tcx>(
+        &self,
+        communicate_allowed: bool,
+        offset: SeekFrom,
+    ) -> InterpResult<'tcx, io::Result<u64>> {
+        assert!(communicate_allowed, "isolation should have prevented even opening a file");
+        interp_ok((&mut &self.file).seek(offset))
+    }
+
+    fn close<'tcx>(
+        self,
+        communicate_allowed: bool,
+        _ecx: &mut MiriInterpCx<'tcx>,
+    ) -> InterpResult<'tcx, io::Result<()>> {
+        assert!(communicate_allowed, "isolation should have prevented even opening a file");
+        // We sync the file if it was opened in a mode different than read-only.
+        if self.writable {
+            // `File::sync_all` does the checks that are done when closing a file. We do this to
+            // to handle possible errors correctly.
+            let result = self.file.sync_all();
+            // Now we actually close the file and return the result.
+            drop(self.file);
+            interp_ok(result)
+        } else {
+            // We drop the file, this closes it but ignores any errors
+            // produced when closing it. This is done because
+            // `File::sync_all` cannot be done over files like
+            // `/dev/urandom` which are read-only. Check
+            // https://github.com/rust-lang/miri/issues/999#issuecomment-568920439
+            // for a deeper discussion.
+            drop(self.file);
+            interp_ok(Ok(()))
+        }
+    }
+
+    fn metadata<'tcx>(&self) -> InterpResult<'tcx, io::Result<Metadata>> {
+        interp_ok(self.file.metadata())
+    }
+
+    fn is_tty(&self, communicate_allowed: bool) -> bool {
+        communicate_allowed && self.file.is_terminal()
+    }
+
+    fn as_unix<'tcx>(&self, ecx: &MiriInterpCx<'tcx>) -> &dyn UnixFileDescription {
+        assert!(
+            ecx.target_os_is_unix(),
+            "unix file operations are only available for unix targets"
+        );
+        self
+    }
+}
+
 /// Like /dev/null
 #[derive(Debug)]
 pub struct NullOutput;
@@ -300,10 +392,13 @@ impl FileDescription for NullOutput {
     }
 }
 
+/// Internal type of a file-descriptor - this is what [`FdTable`] expects
+pub type FdNum = i32;
+
 /// The file descriptor table
 #[derive(Debug)]
 pub struct FdTable {
-    pub fds: BTreeMap<i32, DynFileDescriptionRef>,
+    pub fds: BTreeMap<FdNum, DynFileDescriptionRef>,
     /// Unique identifier for file description, used to differentiate between various file description.
     next_file_description_id: FdId,
 }
@@ -339,21 +434,21 @@ impl FdTable {
     }
 
     /// Insert a new file description to the FdTable.
-    pub fn insert_new(&mut self, fd: impl FileDescription) -> i32 {
+    pub fn insert_new(&mut self, fd: impl FileDescription) -> FdNum {
         let fd_ref = self.new_ref(fd);
         self.insert(fd_ref)
     }
 
-    pub fn insert(&mut self, fd_ref: DynFileDescriptionRef) -> i32 {
+    pub fn insert(&mut self, fd_ref: DynFileDescriptionRef) -> FdNum {
         self.insert_with_min_num(fd_ref, 0)
     }
 
     /// Insert a file description, giving it a file descriptor that is at least `min_fd_num`.
     pub fn insert_with_min_num(
         &mut self,
         file_handle: DynFileDescriptionRef,
-        min_fd_num: i32,
-    ) -> i32 {
+        min_fd_num: FdNum,
+    ) -> FdNum {
         // Find the lowest unused FD, starting from min_fd. If the first such unused FD is in
         // between used FDs, the find_map combinator will return it. If the first such unused FD
         // is after all other used FDs, the find_map combinator will return None, and we will use
@@ -379,16 +474,16 @@ impl FdTable {
         new_fd_num
     }
 
-    pub fn get(&self, fd_num: i32) -> Option<DynFileDescriptionRef> {
+    pub fn get(&self, fd_num: FdNum) -> Option<DynFileDescriptionRef> {
         let fd = self.fds.get(&fd_num)?;
         Some(fd.clone())
     }
 
-    pub fn remove(&mut self, fd_num: i32) -> Option<DynFileDescriptionRef> {
+    pub fn remove(&mut self, fd_num: FdNum) -> Option<DynFileDescriptionRef> {
         self.fds.remove(&fd_num)
     }
 
-    pub fn is_fd_num(&self, fd_num: i32) -> bool {
+    pub fn is_fd_num(&self, fd_num: FdNum) -> bool {
         self.fds.contains_key(&fd_num)
     }
 }

src/tools/miri/src/shims/time.rs

@@ -219,16 +219,8 @@ pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
 
         let filetime = this.deref_pointer_as(LPFILETIME_op, this.windows_ty_layout("FILETIME"))?;
 
-        let NANOS_PER_SEC = this.eval_windows_u64("time", "NANOS_PER_SEC");
-        let INTERVALS_PER_SEC = this.eval_windows_u64("time", "INTERVALS_PER_SEC");
-        let INTERVALS_TO_UNIX_EPOCH = this.eval_windows_u64("time", "INTERVALS_TO_UNIX_EPOCH");
-        let NANOS_PER_INTERVAL = NANOS_PER_SEC / INTERVALS_PER_SEC;
-        let SECONDS_TO_UNIX_EPOCH = INTERVALS_TO_UNIX_EPOCH / INTERVALS_PER_SEC;
-
-        let duration = system_time_to_duration(&SystemTime::now())?
-            + Duration::from_secs(SECONDS_TO_UNIX_EPOCH);
-        let duration_ticks = u64::try_from(duration.as_nanos() / u128::from(NANOS_PER_INTERVAL))
-            .map_err(|_| err_unsup_format!("programs running more than 2^64 Windows ticks after the Windows epoch are not supported"))?;
+        let duration = this.system_time_since_windows_epoch(&SystemTime::now())?;
+        let duration_ticks = this.windows_ticks_for(duration)?;
 
         let dwLowDateTime = u32::try_from(duration_ticks & 0x00000000FFFFFFFF).unwrap();
         let dwHighDateTime = u32::try_from((duration_ticks & 0xFFFFFFFF00000000) >> 32).unwrap();
@@ -281,6 +273,30 @@ pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
         interp_ok(Scalar::from_i32(-1)) // Return non-zero on success
     }
 
+    #[allow(non_snake_case, clippy::arithmetic_side_effects)]
+    fn system_time_since_windows_epoch(&self, time: &SystemTime) -> InterpResult<'tcx, Duration> {
+        let this = self.eval_context_ref();
+
+        let INTERVALS_PER_SEC = this.eval_windows_u64("time", "INTERVALS_PER_SEC");
+        let INTERVALS_TO_UNIX_EPOCH = this.eval_windows_u64("time", "INTERVALS_TO_UNIX_EPOCH");
+        let SECONDS_TO_UNIX_EPOCH = INTERVALS_TO_UNIX_EPOCH / INTERVALS_PER_SEC;
+
+        interp_ok(system_time_to_duration(time)? + Duration::from_secs(SECONDS_TO_UNIX_EPOCH))
+    }
+
+    #[allow(non_snake_case, clippy::arithmetic_side_effects)]
+    fn windows_ticks_for(&self, duration: Duration) -> InterpResult<'tcx, u64> {
+        let this = self.eval_context_ref();
+
+        let NANOS_PER_SEC = this.eval_windows_u64("time", "NANOS_PER_SEC");
+        let INTERVALS_PER_SEC = this.eval_windows_u64("time", "INTERVALS_PER_SEC");
+        let NANOS_PER_INTERVAL = NANOS_PER_SEC / INTERVALS_PER_SEC;
+
+        let ticks = u64::try_from(duration.as_nanos() / u128::from(NANOS_PER_INTERVAL))
+            .map_err(|_| err_unsup_format!("programs running more than 2^64 Windows ticks after the Windows epoch are not supported"))?;
+        interp_ok(ticks)
+    }
+
     fn mach_absolute_time(&self) -> InterpResult<'tcx, Scalar> {
         let this = self.eval_context_ref();
 

src/tools/miri/src/shims/unix/fd.rs

@@ -121,7 +121,7 @@ pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
             throw_unsup_format!("unsupported flags {:#x}", op);
         };
 
-        let result = fd.as_unix().flock(this.machine.communicate(), parsed_op)?;
+        let result = fd.as_unix(this).flock(this.machine.communicate(), parsed_op)?;
         // return `0` if flock is successful
         let result = result.map(|()| 0i32);
         interp_ok(Scalar::from_i32(this.try_unwrap_io_result(result)?))
@@ -273,7 +273,7 @@ pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
                 let Ok(offset) = u64::try_from(offset) else {
                     return this.set_last_error_and_return(LibcError("EINVAL"), dest);
                 };
-                fd.as_unix().pread(communicate, offset, buf, count, this, finish)?
+                fd.as_unix(this).pread(communicate, offset, buf, count, this, finish)?
             }
         };
         interp_ok(())
@@ -333,7 +333,7 @@ pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
                 let Ok(offset) = u64::try_from(offset) else {
                     return this.set_last_error_and_return(LibcError("EINVAL"), dest);
                 };
-                fd.as_unix().pwrite(communicate, buf, count, offset, this, finish)?
+                fd.as_unix(this).pwrite(communicate, buf, count, offset, this, finish)?
             }
         };
         interp_ok(())

src/tools/miri/src/shims/unix/fs.rs

@@ -2,10 +2,9 @@
 
 use std::borrow::Cow;
 use std::fs::{
-    DirBuilder, File, FileType, Metadata, OpenOptions, ReadDir, read_dir, remove_dir, remove_file,
-    rename,
+    DirBuilder, File, FileType, OpenOptions, ReadDir, read_dir, remove_dir, remove_file, rename,
 };
-use std::io::{self, ErrorKind, IsTerminal, Read, Seek, SeekFrom, Write};
+use std::io::{self, ErrorKind, Read, Seek, SeekFrom, Write};
 use std::path::{Path, PathBuf};
 use std::time::SystemTime;
 
@@ -14,98 +13,11 @@ use rustc_data_structures::fx::FxHashMap;
 
 use self::shims::time::system_time_to_duration;
 use crate::helpers::check_min_vararg_count;
-use crate::shims::files::{EvalContextExt as _, FileDescription, FileDescriptionRef};
+use crate::shims::files::FileHandle;
 use crate::shims::os_str::bytes_to_os_str;
 use crate::shims::unix::fd::{FlockOp, UnixFileDescription};
 use crate::*;
 
-#[derive(Debug)]
-struct FileHandle {
-    file: File,
-    writable: bool,
-}
-
-impl FileDescription for FileHandle {
-    fn name(&self) -> &'static str {
-        "file"
-    }
-
-    fn read<'tcx>(
-        self: FileDescriptionRef<Self>,
-        communicate_allowed: bool,
-        ptr: Pointer,
-        len: usize,
-        ecx: &mut MiriInterpCx<'tcx>,
-        finish: DynMachineCallback<'tcx, Result<usize, IoError>>,
-    ) -> InterpResult<'tcx> {
-        assert!(communicate_allowed, "isolation should have prevented even opening a file");
-
-        let result = ecx.read_from_host(&self.file, len, ptr)?;
-        finish.call(ecx, result)
-    }
-
-    fn write<'tcx>(
-        self: FileDescriptionRef<Self>,
-        communicate_allowed: bool,
-        ptr: Pointer,
-        len: usize,
-        ecx: &mut MiriInterpCx<'tcx>,
-        finish: DynMachineCallback<'tcx, Result<usize, IoError>>,
-    ) -> InterpResult<'tcx> {
-        assert!(communicate_allowed, "isolation should have prevented even opening a file");
-
-        let result = ecx.write_to_host(&self.file, len, ptr)?;
-        finish.call(ecx, result)
-    }
-
-    fn seek<'tcx>(
-        &self,
-        communicate_allowed: bool,
-        offset: SeekFrom,
-    ) -> InterpResult<'tcx, io::Result<u64>> {
-        assert!(communicate_allowed, "isolation should have prevented even opening a file");
-        interp_ok((&mut &self.file).seek(offset))
-    }
-
-    fn close<'tcx>(
-        self,
-        communicate_allowed: bool,
-        _ecx: &mut MiriInterpCx<'tcx>,
-    ) -> InterpResult<'tcx, io::Result<()>> {
-        assert!(communicate_allowed, "isolation should have prevented even opening a file");
-        // We sync the file if it was opened in a mode different than read-only.
-        if self.writable {
-            // `File::sync_all` does the checks that are done when closing a file. We do this to
-            // to handle possible errors correctly.
-            let result = self.file.sync_all();
-            // Now we actually close the file and return the result.
-            drop(self.file);
-            interp_ok(result)
-        } else {
-            // We drop the file, this closes it but ignores any errors
-            // produced when closing it. This is done because
-            // `File::sync_all` cannot be done over files like
-            // `/dev/urandom` which are read-only. Check
-            // https://github.com/rust-lang/miri/issues/999#issuecomment-568920439
-            // for a deeper discussion.
-            drop(self.file);
-            interp_ok(Ok(()))
-        }
-    }
-
-    fn metadata<'tcx>(&self) -> InterpResult<'tcx, io::Result<Metadata>> {
-        interp_ok(self.file.metadata())
-    }
-
-    fn is_tty(&self, communicate_allowed: bool) -> bool {
-        communicate_allowed && self.file.is_terminal()
-    }
-
-    fn as_unix(&self) -> &dyn UnixFileDescription {
-        self
-    }
-}
-
 impl UnixFileDescription for FileHandle {
     fn pread<'tcx>(
         &self,