Rollup merge of #106639 - RalfJung:miri, r=RalfJung

update Miri

Notable PRs:
- rust-lang#2748
- rust-lang#2752

r? `@ghost`

cargo-miri/src/phases.rs

@@ -185,7 +185,7 @@ pub fn phase_cargo_miri(mut args: impl Iterator<Item = String>) {
     // explicitly do this even if RUSTC_STAGE is set, since for these builds we do *not* want the
     // bootstrap `rustc` thing in our way! Instead, we have MIRI_HOST_SYSROOT to use for host
     // builds.
-    cmd.env("RUSTC", &fs::canonicalize(find_miri()).unwrap());
+    cmd.env("RUSTC", fs::canonicalize(find_miri()).unwrap());
     cmd.env("MIRI_BE_RUSTC", "target"); // we better remember to *unset* this in the other phases!
 
     // Set rustdoc to us as well, so we can run doctests.

ci.sh

@@ -108,7 +108,8 @@ case $HOST_TARGET in
     MIRI_TEST_TARGET=i686-pc-windows-msvc run_tests
     MIRI_TEST_TARGET=x86_64-unknown-freebsd run_tests_minimal hello integer vec panic/panic concurrency/simple atomic data_race env/var
     MIRI_TEST_TARGET=aarch64-linux-android run_tests_minimal hello integer vec panic/panic
-    MIRI_TEST_TARGET=wasm32-wasi run_tests_minimal no_std integer
+    MIRI_TEST_TARGET=wasm32-wasi run_tests_minimal no_std integer strings
+    MIRI_TEST_TARGET=wasm32-unknown-unknown run_tests_minimal no_std integer strings
     MIRI_TEST_TARGET=thumbv7em-none-eabihf MIRI_NO_STD=1 run_tests_minimal no_std # no_std embedded architecture
     MIRI_TEST_TARGET=tests/avr.json MIRI_NO_STD=1 run_tests_minimal no_std # JSON target file
     ;;

rust-version

@@ -1 +1 @@
-4f4d0586ad20c66a16d547581ca379beafece93a
+c54c8cbac882e149e04a9e1f2d146fd548ae30ae

src/borrow_tracker/stacked_borrows/diagnostics.rs

@@ -88,11 +88,7 @@ impl fmt::Display for InvalidationCause {
         match self {
             InvalidationCause::Access(kind) => write!(f, "{kind}"),
             InvalidationCause::Retag(perm, kind) =>
-                if *kind == RetagCause::FnEntry {
-                    write!(f, "{perm:?} FnEntry retag")
-                } else {
-                    write!(f, "{perm:?} retag")
-                },
+                write!(f, "{perm:?} {retag}", retag = kind.summary()),
         }
     }
 }
@@ -193,7 +189,7 @@ struct RetagOp {
 #[derive(Debug, Clone, Copy, PartialEq)]
 pub enum RetagCause {
     Normal,
-    FnReturn,
+    FnReturnPlace,
     FnEntry,
     TwoPhase,
 }
@@ -495,8 +491,8 @@ impl RetagCause {
     fn summary(&self) -> String {
         match self {
             RetagCause::Normal => "retag",
-            RetagCause::FnEntry => "FnEntry retag",
-            RetagCause::FnReturn => "FnReturn retag",
+            RetagCause::FnEntry => "function-entry retag",
+            RetagCause::FnReturnPlace => "return-place retag",
             RetagCause::TwoPhase => "two-phase retag",
         }
         .to_string()

src/borrow_tracker/stacked_borrows/mod.rs

@@ -998,7 +998,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
             access: Some(AccessKind::Write),
             protector: Some(ProtectorKind::StrongProtector),
         };
-        let val = this.sb_retag_reference(&val, new_perm, RetagCause::FnReturn)?;
+        let val = this.sb_retag_reference(&val, new_perm, RetagCause::FnReturnPlace)?;
         // And use reborrowed pointer for return place.
         let return_place = this.ref_to_mplace(&val)?;
         this.frame_mut().return_place = return_place.into();

src/helpers.rs

@@ -943,7 +943,16 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         link_name: Symbol,
     ) -> InterpResult<'tcx, ()> {
         self.check_abi(abi, exp_abi)?;
-        if let Some((body, _)) = self.eval_context_mut().lookup_exported_symbol(link_name)? {
+        if let Some((body, instance)) = self.eval_context_mut().lookup_exported_symbol(link_name)? {
+            // If compiler-builtins is providing the symbol, then don't treat it as a clash.
+            // We'll use our built-in implementation in `emulate_foreign_item_by_name` for increased
+            // performance. Note that this means we won't catch any undefined behavior in
+            // compiler-builtins when running other crates, but Miri can still be run on
+            // compiler-builtins itself (or any crate that uses it as a normal dependency)
+            if self.eval_context_ref().tcx.is_compiler_builtins(instance.def_id().krate) {
+                return Ok(());
+            }
+
             throw_machine_stop!(TerminationInfo::SymbolShimClashing {
                 link_name,
                 span: body.span.data(),

src/range_map.rs

@@ -219,7 +219,6 @@ mod tests {
     /// Query the map at every offset in the range and collect the results.
     fn to_vec<T: Copy>(map: &RangeMap<T>, offset: u64, len: u64) -> Vec<T> {
         (offset..offset + len)
-            .into_iter()
             .map(|i| {
                 map.iter(Size::from_bytes(i), Size::from_bytes(1)).next().map(|(_, &t)| t).unwrap()
             })

src/shims/env.rs

@@ -166,7 +166,9 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
                 // `buf_size` represents the size in characters.
                 let buf_size = u64::from(this.read_scalar(size_op)?.to_u32()?);
                 Scalar::from_u32(windows_check_buffer_size(
-                    this.write_os_str_to_wide_str(&var, buf_ptr, buf_size)?,
+                    this.write_os_str_to_wide_str(
+                        &var, buf_ptr, buf_size, /*truncate*/ false,
+                    )?,
                 ))
             }
             None => {
@@ -366,7 +368,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         match env::current_dir() {
             Ok(cwd) =>
                 return Ok(Scalar::from_u32(windows_check_buffer_size(
-                    this.write_path_to_wide_str(&cwd, buf, size)?,
+                    this.write_path_to_wide_str(&cwd, buf, size, /*truncate*/ false)?,
                 ))),
             Err(e) => this.set_last_error_from_io_error(e.kind())?,
         }

src/shims/os_str.rs

@@ -101,17 +101,23 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         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.`
+    /// Helper function to write an OsStr as a 0x0000-terminated u16-sequence, which is what the
+    /// Windows APIs usually handle.
+    ///
+    /// If `truncate == false` (the usual mode of operation), 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.`
+    ///
+    /// If `truncate == true`, then in case `size` is not large enough it *will* write the first
+    /// `size.saturating_sub(1)` many items, followed by a null terminator (if `size > 0`).
     fn write_os_str_to_wide_str(
         &mut self,
         os_str: &OsStr,
         ptr: Pointer<Option<Provenance>>,
         size: u64,
+        truncate: bool,
     ) -> InterpResult<'tcx, (bool, u64)> {
         #[cfg(windows)]
         fn os_str_to_u16vec<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, Vec<u16>> {
@@ -129,7 +135,15 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         }
 
         let u16_vec = os_str_to_u16vec(os_str)?;
-        self.eval_context_mut().write_wide_str(&u16_vec, ptr, size)
+        let (written, size_needed) = self.eval_context_mut().write_wide_str(&u16_vec, ptr, size)?;
+        if truncate && !written && size > 0 {
+            // Write the truncated part that fits.
+            let truncated_data = &u16_vec[..size.saturating_sub(1).try_into().unwrap()];
+            let (written, written_len) =
+                self.eval_context_mut().write_wide_str(truncated_data, ptr, size)?;
+            assert!(written && written_len == size);
+        }
+        Ok((written, size_needed))
     }
 
     /// Allocate enough memory to store the given `OsStr` as a null-terminated sequence of bytes.
@@ -143,7 +157,8 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
 
         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);
+        let (written, _) = self.write_os_str_to_c_str(os_str, arg_place.ptr, size).unwrap();
+        assert!(written);
         Ok(arg_place.ptr)
     }
 
@@ -158,7 +173,9 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
 
         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);
+        let (written, _) =
+            self.write_os_str_to_wide_str(os_str, arg_place.ptr, size, /*truncate*/ false).unwrap();
+        assert!(written);
         Ok(arg_place.ptr)
     }
 
@@ -212,11 +229,12 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         path: &Path,
         ptr: Pointer<Option<Provenance>>,
         size: u64,
+        truncate: bool,
     ) -> InterpResult<'tcx, (bool, u64)> {
         let this = self.eval_context_mut();
         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)
+        this.write_os_str_to_wide_str(&os_str, ptr, size, truncate)
     }
 
     /// Allocate enough memory to store a Path as a null-terminated sequence of bytes,
@@ -232,6 +250,19 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
         this.alloc_os_str_as_c_str(&os_str, memkind)
     }
 
+    /// Allocate enough memory to store a Path as a null-terminated sequence of `u16`s,
+    /// adjusting path separators if needed.
+    fn alloc_path_as_wide_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(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
+        this.alloc_os_str_as_wide_str(&os_str, memkind)
+    }
+
     #[allow(clippy::get_first)]
     fn convert_path<'a>(
         &self,

src/shims/windows/foreign_items.rs

@@ -381,6 +381,46 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
 
                 this.write_scalar(Scalar::from_u32(1), dest)?;
             }
+            "GetModuleFileNameW" => {
+                let [handle, filename, size] =
+                    this.check_shim(abi, Abi::System { unwind: false }, link_name, args)?;
+                this.check_no_isolation("`GetModuleFileNameW`")?;
+
+                let handle = this.read_machine_usize(handle)?;
+                let filename = this.read_pointer(filename)?;
+                let size = this.read_scalar(size)?.to_u32()?;
+
+                if handle != 0 {
+                    throw_unsup_format!("`GetModuleFileNameW` only supports the NULL handle");
+                }
+
+                // Using the host current_exe is a bit off, but consistent with Linux
+                // (where stdlib reads /proc/self/exe).
+                // Unfortunately this Windows function has a crazy behavior so we can't just use
+                // `write_path_to_wide_str`...
+                let path = std::env::current_exe().unwrap();
+                let (all_written, size_needed) = this.write_path_to_wide_str(
+                    &path,
+                    filename,
+                    size.into(),
+                    /*truncate*/ true,
+                )?;
+
+                if all_written {
+                    // If the function succeeds, the return value is the length of the string that
+                    // is copied to the buffer, in characters, not including the terminating null
+                    // character.
+                    this.write_int(size_needed.checked_sub(1).unwrap(), dest)?;
+                } else {
+                    // If the buffer is too small to hold the module name, the string is truncated
+                    // to nSize characters including the terminating null character, the function
+                    // returns nSize, and the function sets the last error to
+                    // ERROR_INSUFFICIENT_BUFFER.
+                    this.write_int(size, dest)?;
+                    let insufficient_buffer = this.eval_windows("c", "ERROR_INSUFFICIENT_BUFFER");
+                    this.set_last_error(insufficient_buffer)?;
+                }
+            }
 
             // Threading
             "CreateThread" => {

test_dependencies/Cargo.toml

@@ -13,11 +13,11 @@ libc = "0.2"
 num_cpus = "1.10.1"
 
 getrandom_1 = { package = "getrandom", version = "0.1" }
-getrandom = { version = "0.2" }
+getrandom = { version = "0.2", features = ["js"] }
 rand = { version = "0.8", features = ["small_rng"] }
 
 [target.'cfg(not(any(target_arch = "wasm32", target_arch = "wasm64")))'.dependencies]
 page_size = "0.5"
-tokio = { version = "1.0", features = ["full"] }
+tokio = { version = "1.23", features = ["full"] }
 
 [workspace]