1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
use std::borrow::Cow;
use std::ffi::{OsStr, OsString};
use std::path::{Path, PathBuf};

#[cfg(unix)]
use std::os::unix::ffi::{OsStrExt, OsStringExt};
#[cfg(windows)]
use std::os::windows::ffi::{OsStrExt, OsStringExt};

use rustc_middle::ty::layout::LayoutOf;

use crate::*;

/// Represent how path separator conversion should be done.
pub enum PathConversion {
    HostToTarget,
    TargetToHost,
}

#[cfg(unix)]
pub fn os_str_to_bytes<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, &[u8]> {
    Ok(os_str.as_bytes())
}

#[cfg(not(unix))]
pub fn os_str_to_bytes<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, &[u8]> {
    // On non-unix platforms the best we can do to transform bytes from/to OS strings is to do the
    // intermediate transformation into strings. Which invalidates non-utf8 paths that are actually
    // valid.
    os_str
        .to_str()
        .map(|s| s.as_bytes())
        .ok_or_else(|| err_unsup_format!("{:?} is not a valid utf-8 string", os_str).into())
}

#[cfg(unix)]
pub fn bytes_to_os_str<'tcx>(bytes: &[u8]) -> InterpResult<'tcx, &OsStr> {
    Ok(OsStr::from_bytes(bytes))
}
#[cfg(not(unix))]
pub fn bytes_to_os_str<'tcx>(bytes: &[u8]) -> InterpResult<'tcx, &OsStr> {
    let s = std::str::from_utf8(bytes)
        .map_err(|_| err_unsup_format!("{:?} is not a valid utf-8 string", bytes))?;
    Ok(OsStr::new(s))
}

impl<'mir, 'tcx: 'mir> EvalContextExt<'mir, 'tcx> for crate::MiriInterpCx<'mir, 'tcx> {}
pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
    /// Helper function to read an OsString from a null-terminated sequence of bytes, which is what
    /// the Unix APIs usually handle.
    fn read_os_str_from_c_str<'a>(
        &'a self,
        ptr: Pointer<Option<Provenance>>,
    ) -> InterpResult<'tcx, &'a OsStr>
    where
        'tcx: 'a,
        'mir: 'a,
    {
        let this = self.eval_context_ref();
        let bytes = this.read_c_str(ptr)?;
        bytes_to_os_str(bytes)
    }

    /// Helper function to read an OsString from a 0x0000-terminated sequence of u16,
    /// which is what the Windows APIs usually handle.
    fn read_os_str_from_wide_str<'a>(
        &'a self,
        ptr: Pointer<Option<Provenance>>,
    ) -> InterpResult<'tcx, OsString>
    where
        'tcx: 'a,
        'mir: 'a,
    {
        #[cfg(windows)]
        pub fn u16vec_to_osstring<'tcx>(u16_vec: Vec<u16>) -> InterpResult<'tcx, OsString> {
            Ok(OsString::from_wide(&u16_vec[..]))
        }
        #[cfg(not(windows))]
        pub fn u16vec_to_osstring<'tcx>(u16_vec: Vec<u16>) -> InterpResult<'tcx, OsString> {
            let s = String::from_utf16(&u16_vec[..])
                .map_err(|_| err_unsup_format!("{:?} is not a valid utf-16 string", u16_vec))?;
            Ok(s.into())
        }

        let u16_vec = self.eval_context_ref().read_wide_str(ptr)?;
        u16vec_to_osstring(u16_vec)
    }

    /// Helper function to write an OsStr as a null-terminated sequence of bytes, which is what
    /// the Unix APIs usually handle. This function returns `Ok((false, length))` without trying
    /// to write if `size` is not large enough to fit the contents of `os_string` plus a null
    /// terminator. It returns `Ok((true, length))` if the writing process was successful. The
    /// string length returned does include the null terminator.
    fn write_os_str_to_c_str(
        &mut self,
        os_str: &OsStr,
        ptr: Pointer<Option<Provenance>>,
        size: u64,
    ) -> InterpResult<'tcx, (bool, u64)> {
        let bytes = os_str_to_bytes(os_str)?;
        self.eval_context_mut().write_c_str(bytes, ptr, size)
    }

    /// Helper function to write an OsStr as a 0x0000-terminated u16-sequence, which is what
    /// the Windows APIs usually handle. This function returns `Ok((false, length))` without trying
    /// to write if `size` is not large enough to fit the contents of `os_string` plus a null
    /// terminator. It returns `Ok((true, length))` if the writing process was successful. The
    /// string length returned does include the null terminator. Length is measured in units of
    /// `u16.`
    fn write_os_str_to_wide_str(
        &mut self,
        os_str: &OsStr,
        ptr: Pointer<Option<Provenance>>,
        size: u64,
    ) -> InterpResult<'tcx, (bool, u64)> {
        #[cfg(windows)]
        fn os_str_to_u16vec<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, Vec<u16>> {
            Ok(os_str.encode_wide().collect())
        }
        #[cfg(not(windows))]
        fn os_str_to_u16vec<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, Vec<u16>> {
            // On non-Windows platforms the best we can do to transform Vec<u16> from/to OS strings is to do the
            // intermediate transformation into strings. Which invalidates non-utf8 paths that are actually
            // valid.
            os_str
                .to_str()
                .map(|s| s.encode_utf16().collect())
                .ok_or_else(|| err_unsup_format!("{:?} is not a valid utf-8 string", os_str).into())
        }

        let u16_vec = os_str_to_u16vec(os_str)?;
        self.eval_context_mut().write_wide_str(&u16_vec, ptr, size)
    }

    /// Allocate enough memory to store the given `OsStr` as a null-terminated sequence of bytes.
    fn alloc_os_str_as_c_str(
        &mut self,
        os_str: &OsStr,
        memkind: MemoryKind<MiriMemoryKind>,
    ) -> InterpResult<'tcx, Pointer<Option<Provenance>>> {
        let size = u64::try_from(os_str.len()).unwrap().checked_add(1).unwrap(); // Make space for `0` terminator.
        let this = self.eval_context_mut();

        let arg_type = this.tcx.mk_array(this.tcx.types.u8, size);
        let arg_place = this.allocate(this.layout_of(arg_type).unwrap(), memkind)?;
        assert!(self.write_os_str_to_c_str(os_str, arg_place.ptr, size).unwrap().0);
        Ok(arg_place.ptr)
    }

    /// Allocate enough memory to store the given `OsStr` as a null-terminated sequence of `u16`.
    fn alloc_os_str_as_wide_str(
        &mut self,
        os_str: &OsStr,
        memkind: MemoryKind<MiriMemoryKind>,
    ) -> InterpResult<'tcx, Pointer<Option<Provenance>>> {
        let size = u64::try_from(os_str.len()).unwrap().checked_add(1).unwrap(); // Make space for `0x0000` terminator.
        let this = self.eval_context_mut();

        let arg_type = this.tcx.mk_array(this.tcx.types.u16, size);
        let arg_place = this.allocate(this.layout_of(arg_type).unwrap(), memkind)?;
        assert!(self.write_os_str_to_wide_str(os_str, arg_place.ptr, size).unwrap().0);
        Ok(arg_place.ptr)
    }

    /// Read a null-terminated sequence of bytes, and perform path separator conversion if needed.
    fn read_path_from_c_str<'a>(
        &'a self,
        ptr: Pointer<Option<Provenance>>,
    ) -> InterpResult<'tcx, Cow<'a, Path>>
    where
        'tcx: 'a,
        'mir: 'a,
    {
        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) {
            Cow::Borrowed(x) => Cow::Borrowed(Path::new(x)),
            Cow::Owned(y) => Cow::Owned(PathBuf::from(y)),
        })
    }

    /// Read a null-terminated sequence of `u16`s, and perform path separator conversion if needed.
    fn read_path_from_wide_str(
        &self,
        ptr: Pointer<Option<Provenance>>,
    ) -> InterpResult<'tcx, PathBuf> {
        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())
    }

    /// Write a Path to the machine memory (as a null-terminated sequence of bytes),
    /// adjusting path separators if needed.
    fn write_path_to_c_str(
        &mut self,
        path: &Path,
        ptr: Pointer<Option<Provenance>>,
        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);
        this.write_os_str_to_c_str(&os_str, ptr, size)
    }

    /// Write a Path to the machine memory (as a null-terminated sequence of `u16`s),
    /// adjusting path separators if needed.
    fn write_path_to_wide_str(
        &mut self,
        path: &Path,
        ptr: Pointer<Option<Provenance>>,
        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);
        this.write_os_str_to_wide_str(&os_str, ptr, size)
    }

    /// Allocate enough memory to store a Path as a null-terminated sequence of bytes,
    /// adjusting path separators if needed.
    fn alloc_path_as_c_str(
        &mut self,
        path: &Path,
        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);
        this.alloc_os_str_as_c_str(&os_str, memkind)
    }

    fn convert_path_separator<'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.
            os_str
        } else {
            // Unix target, Windows host.
            let (from, to) = match direction {
                PathConversion::HostToTarget => ('\\', '/'),
                PathConversion::TargetToHost => ('/', '\\'),
            };
            let converted = os_str
                .encode_wide()
                .map(|wchar| if wchar == from as u16 { to as u16 } else { wchar })
                .collect::<Vec<_>>();
            Cow::Owned(OsString::from_wide(&converted))
        };
        #[cfg(unix)]
        return if target_os == "windows" {
            // Windows target, Unix host.
            let (from, to) = match direction {
                PathConversion::HostToTarget => ('/', '\\'),
                PathConversion::TargetToHost => ('\\', '/'),
            };
            let converted = os_str
                .as_bytes()
                .iter()
                .map(|&wchar| if wchar == from as u8 { to as u8 } else { wchar })
                .collect::<Vec<_>>();
            Cow::Owned(OsString::from_vec(converted))
        } else {
            // Unix-on-Unix, all is fine.
            os_str
        };
    }
}