Merge pull request #4651 from RalfJung/user-relevant-crates

add -Zmiri-user-relevant-crates

src/tools/miri/README.md

@@ -373,6 +373,12 @@ environment variable. We first document the most relevant and most commonly used
   ensure alignment.  (The standard library `align_to` method works fine in both modes; under
   symbolic alignment it only fills the middle slice when the allocation guarantees sufficient
   alignment.)
+* `-Zmiri-user-relevant-crates=<crate>,<crate>,...` extends the list of crates that Miri considers
+  "user-relevant". This affects the rendering of backtraces (for user-relevant crates, Miri shows
+  not just the function name but the actual code) and it affects the spans collected for data races
+  and aliasing violations (where Miri will show the span of the topmost non-`#[track_caller]` frame
+  in a user-relevant crate). When using `cargo miri`, the crates in the local workspace are always
+  considered user-relevant.
 
 The remaining flags are for advanced use only, and more likely to change or be removed.
 Some of these are **unsound**, which means they can lead

src/tools/miri/src/bin/miri.rs

@@ -710,6 +710,8 @@ fn main() {
                 fatal_error!("-Zmiri-force-page-size requires a power of 2: {page_size}");
             };
             miri_config.page_size = Some(page_size);
+        } else if let Some(param) = arg.strip_prefix("-Zmiri-user-relevant-crates=") {
+            miri_config.user_relevant_crates.extend(param.split(',').map(|s| s.to_owned()));
         } else {
             // Forward to rustc.
             rustc_args.push(arg);

src/tools/miri/src/eval.rs

@@ -115,6 +115,8 @@ pub struct MiriConfig {
     pub float_rounding_error: FloatRoundingErrorMode,
     /// Whether Miri artifically introduces short reads/writes on file descriptors.
     pub short_fd_operations: bool,
+    /// A list of crates that are considered user-relevant.
+    pub user_relevant_crates: Vec<String>,
 }
 
 impl Default for MiriConfig {
@@ -158,6 +160,7 @@ impl Default for MiriConfig {
             float_nondet: true,
             float_rounding_error: FloatRoundingErrorMode::Random,
             short_fd_operations: true,
+            user_relevant_crates: vec![],
         }
     }
 }

src/tools/miri/src/helpers.rs

@@ -1091,7 +1091,7 @@ impl<'tcx> MiriMachine<'tcx> {
     /// This is the source of truth for the `is_user_relevant` flag in our `FrameExtra`.
     pub fn is_user_relevant(&self, frame: &Frame<'tcx, Provenance>) -> bool {
         let def_id = frame.instance().def_id();
-        (def_id.is_local() || self.local_crates.contains(&def_id.krate))
+        (def_id.is_local() || self.user_relevant_crates.contains(&def_id.krate))
             && !frame.instance().def.requires_caller_location(self.tcx)
     }
 }
@@ -1102,25 +1102,6 @@ pub fn isolation_abort_error<'tcx>(name: &str) -> InterpResult<'tcx> {
     )))
 }
 
-/// Retrieve the list of local crates that should have been passed by cargo-miri in
-/// MIRI_LOCAL_CRATES and turn them into `CrateNum`s.
-pub fn get_local_crates(tcx: TyCtxt<'_>) -> Vec<CrateNum> {
-    // Convert the local crate names from the passed-in config into CrateNums so that they can
-    // be looked up quickly during execution
-    let local_crate_names = std::env::var("MIRI_LOCAL_CRATES")
-        .map(|crates| crates.split(',').map(|krate| krate.to_string()).collect::<Vec<_>>())
-        .unwrap_or_default();
-    let mut local_crates = Vec::new();
-    for &crate_num in tcx.crates(()) {
-        let name = tcx.crate_name(crate_num);
-        let name = name.as_str();
-        if local_crate_names.iter().any(|local_name| local_name == name) {
-            local_crates.push(crate_num);
-        }
-    }
-    local_crates
-}
-
 pub(crate) fn bool_to_simd_element(b: bool, size: Size) -> Scalar {
     // SIMD uses all-1 as pattern for "true". In two's complement,
     // -1 has all its bits set to one and `from_int` will truncate or

src/tools/miri/src/machine.rs

@@ -580,8 +580,8 @@ pub struct MiriMachine<'tcx> {
     /// Equivalent setting as RUST_BACKTRACE on encountering an error.
     pub(crate) backtrace_style: BacktraceStyle,
 
-    /// Crates which are considered local for the purposes of error reporting.
-    pub(crate) local_crates: Vec<CrateNum>,
+    /// Crates which are considered user-relevant for the purposes of error reporting.
+    pub(crate) user_relevant_crates: Vec<CrateNum>,
 
     /// Mapping extern static names to their pointer.
     extern_statics: FxHashMap<Symbol, StrictPointer>,
@@ -684,7 +684,7 @@ impl<'tcx> MiriMachine<'tcx> {
         genmc_ctx: Option<Rc<GenmcCtx>>,
     ) -> Self {
         let tcx = layout_cx.tcx();
-        let local_crates = helpers::get_local_crates(tcx);
+        let user_relevant_crates = Self::get_user_relevant_crates(tcx, config);
         let layouts =
             PrimitiveLayouts::new(layout_cx).expect("Couldn't get layouts of primitive types");
         let profiler = config.measureme_out.as_ref().map(|out| {
@@ -774,7 +774,7 @@ impl<'tcx> MiriMachine<'tcx> {
             string_cache: Default::default(),
             exported_symbols_cache: FxHashMap::default(),
             backtrace_style: config.backtrace_style,
-            local_crates,
+            user_relevant_crates,
             extern_statics: FxHashMap::default(),
             rng: RefCell::new(rng),
             allocator: (!config.native_lib.is_empty())
@@ -865,6 +865,29 @@ impl<'tcx> MiriMachine<'tcx> {
         symbols
     }
 
+    /// Retrieve the list of user-relevant crates based on MIRI_LOCAL_CRATES as set by cargo-miri,
+    /// and extra crates set in the config.
+    fn get_user_relevant_crates(tcx: TyCtxt<'_>, config: &MiriConfig) -> Vec<CrateNum> {
+        // Convert the local crate names from the passed-in config into CrateNums so that they can
+        // be looked up quickly during execution
+        let local_crate_names = std::env::var("MIRI_LOCAL_CRATES")
+            .map(|crates| crates.split(',').map(|krate| krate.to_string()).collect::<Vec<_>>())
+            .unwrap_or_default();
+        let mut local_crates = Vec::new();
+        for &crate_num in tcx.crates(()) {
+            let name = tcx.crate_name(crate_num);
+            let name = name.as_str();
+            if local_crate_names
+                .iter()
+                .chain(&config.user_relevant_crates)
+                .any(|local_name| local_name == name)
+            {
+                local_crates.push(crate_num);
+            }
+        }
+        local_crates
+    }
+
     pub(crate) fn late_init(
         ecx: &mut MiriInterpCx<'tcx>,
         config: &MiriConfig,
@@ -889,7 +912,7 @@ impl<'tcx> MiriMachine<'tcx> {
     /// Check whether the stack frame that this `FrameInfo` refers to is part of a local crate.
     pub(crate) fn is_local(&self, frame: &FrameInfo<'_>) -> bool {
         let def_id = frame.instance.def_id();
-        def_id.is_local() || self.local_crates.contains(&def_id.krate)
+        def_id.is_local() || self.user_relevant_crates.contains(&def_id.krate)
     }
 
     /// Called when the interpreter is going to shut down abnormally, such as due to a Ctrl-C.
@@ -1006,7 +1029,7 @@ impl VisitProvenance for MiriMachine<'_> {
             string_cache: _,
             exported_symbols_cache: _,
             backtrace_style: _,
-            local_crates: _,
+            user_relevant_crates: _,
             rng: _,
             allocator: _,
             tracked_alloc_ids: _,