make unix path handling on Windows hosts preserve absoluteness

Author: RalfJung

src/shims/os_str.rs

@@ -174,7 +174,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         let this = self.eval_context_ref();
         let os_str = this.read_os_str_from_c_str(ptr)?;
 
-        Ok(match this.convert_path_separator(Cow::Borrowed(os_str), PathConversion::TargetToHost) {
+        Ok(match this.convert_path(Cow::Borrowed(os_str), PathConversion::TargetToHost) {
             Cow::Borrowed(x) => Cow::Borrowed(Path::new(x)),
             Cow::Owned(y) => Cow::Owned(PathBuf::from(y)),
         })
@@ -188,10 +188,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         let this = self.eval_context_ref();
         let os_str = this.read_os_str_from_wide_str(ptr)?;
 
-        Ok(this
-            .convert_path_separator(Cow::Owned(os_str), PathConversion::TargetToHost)
-            .into_owned()
-            .into())
+        Ok(this.convert_path(Cow::Owned(os_str), PathConversion::TargetToHost).into_owned().into())
     }
 
     /// Write a Path to the machine memory (as a null-terminated sequence of bytes),
@@ -203,8 +200,8 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         size: u64,
     ) -> InterpResult<'tcx, (bool, u64)> {
         let this = self.eval_context_mut();
-        let os_str = this
-            .convert_path_separator(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
+        let os_str =
+            this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
         this.write_os_str_to_c_str(&os_str, ptr, size)
     }
 
@@ -217,8 +214,8 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         size: u64,
     ) -> InterpResult<'tcx, (bool, u64)> {
         let this = self.eval_context_mut();
-        let os_str = this
-            .convert_path_separator(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
+        let os_str =
+            this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
         this.write_os_str_to_wide_str(&os_str, ptr, size)
     }
 
@@ -230,18 +227,19 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         memkind: MemoryKind<MiriMemoryKind>,
     ) -> InterpResult<'tcx, Pointer<Option<Provenance>>> {
         let this = self.eval_context_mut();
-        let os_str = this
-            .convert_path_separator(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
+        let os_str =
+            this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
         this.alloc_os_str_as_c_str(&os_str, memkind)
     }
 
-    fn convert_path_separator<'a>(
+    fn convert_path<'a>(
         &self,
         os_str: Cow<'a, OsStr>,
         direction: PathConversion,
     ) -> Cow<'a, OsStr> {
         let this = self.eval_context_ref();
         let target_os = &this.tcx.sess.target.os;
+
         #[cfg(windows)]
         return if target_os == "windows" {
             // Windows-on-Windows, all fine.
@@ -252,10 +250,34 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
                 PathConversion::HostToTarget => ('\\', '/'),
                 PathConversion::TargetToHost => ('/', '\\'),
             };
-            let converted = os_str
+            let mut converted = os_str
                 .encode_wide()
                 .map(|wchar| if wchar == from as u16 { to as u16 } else { wchar })
                 .collect::<Vec<_>>();
+            // We also have to adjust the prefix to keep absolute paths absolute.
+            match direction {
+                PathConversion::HostToTarget => {
+                    // If this is an absolute Windows path that does not start with a separator
+                    // (`C:/...` after separator conversion), we need to make it start with a
+                    // separator so that Unix targets consider it absolute.
+                    if converted.get(1).copied() == Some(':' as u16)
+                        && converted.get(2).copied() == Some('/' as u16)
+                    {
+                        // We add a `/$miri-absolute-path` at the beginning, to store the absolute
+                        // Windows path in something that looks like an absolute Unix path.
+                        converted.splice(0..0, b"/$miri-absolute-path".map(|c| c as u16));
+                    }
+                }
+                PathConversion::TargetToHost => {
+                    // We have to un-do the `\$miri-absolute-path` prefix, if it is present.
+                    let prefix = b"\\$miri-absolute-path";
+                    if converted.len() >= prefix.len()
+                        && converted.iter().zip(prefix.iter()).all(|(&c, &p)| c == p as u16)
+                    {
+                        converted.splice(0..prefix.len(), std::iter::empty());
+                    }
+                }
+            }
             Cow::Owned(OsString::from_wide(&converted))
         };
         #[cfg(unix)]
@@ -270,6 +292,8 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
                 .iter()
                 .map(|&wchar| if wchar == from as u8 { to as u8 } else { wchar })
                 .collect::<Vec<_>>();
+            // TODO: Once we actually support file system things on Windows targets, we'll probably
+            // have to also do something clever for absolute path preservation here, like above.
             Cow::Owned(OsString::from_vec(converted))
         } else {
             // Unix-on-Unix, all is fine.

src/shims/unix/fs.rs

@@ -1667,7 +1667,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
                 // 'readlink' truncates the resolved path if the provided buffer is not large
                 // enough, and does *not* add a null terminator. That means we cannot use the usual
                 // `write_path_to_c_str` and have to re-implement parts of it ourselves.
-                let resolved = this.convert_path_separator(
+                let resolved = this.convert_path(
                     Cow::Borrowed(resolved.as_ref()),
                     crate::shims::os_str::PathConversion::HostToTarget,
                 );

tests/pass/shims/fs.rs

@@ -15,6 +15,7 @@ use std::io::{Error, ErrorKind, IsTerminal, Read, Result, Seek, SeekFrom, Write}
 use std::path::{Path, PathBuf};
 
 fn main() {
+    test_path_conversion();
     test_file();
     test_file_clone();
     test_file_create_new();
@@ -28,7 +29,6 @@ fn main() {
     test_directory();
     test_canonicalize();
     test_from_raw_os_error();
-    test_path_conversion();
 }
 
 fn tmp() -> PathBuf {