Auto merge of #3280 - rust-lang:rustup-2024-01-26, r=RalfJung

Automatic Rustup

Author: bors

rust-version

@@ -1 +1 @@
-f6ee4bf3847277d6d6e2007ff664f8ea0895b11b
+dd2559e08e1530806740931037d6bb83ef956161

src/machine.rs

@@ -3,12 +3,14 @@
 
 use std::borrow::Cow;
 use std::cell::{Cell, RefCell};
+use std::collections::hash_map::Entry;
 use std::fmt;
 use std::path::Path;
 use std::process;
 
 use either::Either;
 use rand::rngs::StdRng;
+use rand::Rng;
 use rand::SeedableRng;
 
 use rustc_ast::ast::Mutability;
@@ -45,6 +47,11 @@ pub const SIGRTMIN: i32 = 34;
 /// `SIGRTMAX` - `SIGRTMIN` >= 8 (which is the value of `_POSIX_RTSIG_MAX`)
 pub const SIGRTMAX: i32 = 42;
 
+/// Each const has multiple addresses, but only this many. Since const allocations are never
+/// deallocated, choosing a new [`AllocId`] and thus base address for each evaluation would
+/// produce unbounded memory usage.
+const ADDRS_PER_CONST: usize = 16;
+
 /// Extra data stored with each stack frame
 pub struct FrameExtra<'tcx> {
     /// Extra data for the Borrow Tracker.
@@ -65,12 +72,19 @@ pub struct FrameExtra<'tcx> {
     /// optimization.
     /// This is used by `MiriMachine::current_span` and `MiriMachine::caller_span`
     pub is_user_relevant: bool,
+
+    /// We have a cache for the mapping from [`mir::Const`] to resulting [`AllocId`].
+    /// However, we don't want all frames to always get the same result, so we insert
+    /// an additional bit of "salt" into the cache key. This salt is fixed per-frame
+    /// so that within a call, a const will have a stable address.
+    salt: usize,
 }
 
 impl<'tcx> std::fmt::Debug for FrameExtra<'tcx> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         // Omitting `timing`, it does not support `Debug`.
-        let FrameExtra { borrow_tracker, catch_unwind, timing: _, is_user_relevant: _ } = self;
+        let FrameExtra { borrow_tracker, catch_unwind, timing: _, is_user_relevant: _, salt: _ } =
+            self;
         f.debug_struct("FrameData")
             .field("borrow_tracker", borrow_tracker)
             .field("catch_unwind", catch_unwind)
@@ -80,7 +94,8 @@ impl<'tcx> std::fmt::Debug for FrameExtra<'tcx> {
 
 impl VisitProvenance for FrameExtra<'_> {
     fn visit_provenance(&self, visit: &mut VisitWith<'_>) {
-        let FrameExtra { catch_unwind, borrow_tracker, timing: _, is_user_relevant: _ } = self;
+        let FrameExtra { catch_unwind, borrow_tracker, timing: _, is_user_relevant: _, salt: _ } =
+            self;
 
         catch_unwind.visit_provenance(visit);
         borrow_tracker.visit_provenance(visit);
@@ -552,6 +567,11 @@ pub struct MiriMachine<'mir, 'tcx> {
     /// The spans we will use to report where an allocation was created and deallocated in
     /// diagnostics.
     pub(crate) allocation_spans: RefCell<FxHashMap<AllocId, (Span, Option<Span>)>>,
+
+    /// Maps MIR consts to their evaluated result. We combine the const with a "salt" (`usize`)
+    /// that is fixed per stack frame; this lets us have sometimes different results for the
+    /// same const while ensuring consistent results within a single call.
+    const_cache: RefCell<FxHashMap<(mir::Const<'tcx>, usize), OpTy<'tcx, Provenance>>>,
 }
 
 impl<'mir, 'tcx> MiriMachine<'mir, 'tcx> {
@@ -677,6 +697,7 @@ impl<'mir, 'tcx> MiriMachine<'mir, 'tcx> {
             stack_size,
             collect_leak_backtraces: config.collect_leak_backtraces,
             allocation_spans: RefCell::new(FxHashMap::default()),
+            const_cache: RefCell::new(FxHashMap::default()),
         }
     }
 
@@ -788,6 +809,7 @@ impl VisitProvenance for MiriMachine<'_, '_> {
             stack_size: _,
             collect_leak_backtraces: _,
             allocation_spans: _,
+            const_cache: _,
         } = self;
 
         threads.visit_provenance(visit);
@@ -1345,6 +1367,7 @@ impl<'mir, 'tcx> Machine<'mir, 'tcx> for MiriMachine<'mir, 'tcx> {
             catch_unwind: None,
             timing,
             is_user_relevant: ecx.machine.is_user_relevant(&frame),
+            salt: ecx.machine.rng.borrow_mut().gen::<usize>() % ADDRS_PER_CONST,
         };
 
         Ok(frame.with_extra(extra))
@@ -1450,4 +1473,31 @@ impl<'mir, 'tcx> Machine<'mir, 'tcx> for MiriMachine<'mir, 'tcx> {
         ecx.machine.allocation_spans.borrow_mut().insert(alloc_id, (span, None));
         Ok(())
     }
+
+    fn eval_mir_constant<F>(
+        ecx: &InterpCx<'mir, 'tcx, Self>,
+        val: mir::Const<'tcx>,
+        span: Option<Span>,
+        layout: Option<TyAndLayout<'tcx>>,
+        eval: F,
+    ) -> InterpResult<'tcx, OpTy<'tcx, Self::Provenance>>
+    where
+        F: Fn(
+            &InterpCx<'mir, 'tcx, Self>,
+            mir::Const<'tcx>,
+            Option<Span>,
+            Option<TyAndLayout<'tcx>>,
+        ) -> InterpResult<'tcx, OpTy<'tcx, Self::Provenance>>,
+    {
+        let frame = ecx.active_thread_stack().last().unwrap();
+        let mut cache = ecx.machine.const_cache.borrow_mut();
+        match cache.entry((val, frame.extra.salt)) {
+            Entry::Vacant(ve) => {
+                let op = eval(ecx, val, span, layout)?;
+                ve.insert(op.clone());
+                Ok(op)
+            }
+            Entry::Occupied(oe) => Ok(oe.get().clone()),
+        }
+    }
 }

src/shims/intrinsics/mod.rs

@@ -5,6 +5,7 @@ use std::iter;
 
 use log::trace;
 
+use rand::Rng;
 use rustc_apfloat::{Float, Round};
 use rustc_middle::ty::layout::LayoutOf;
 use rustc_middle::{
@@ -141,6 +142,18 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriInterpCxExt<'mir, 'tcx> {
                 this.write_pointer(Pointer::new(ptr.provenance, masked_addr), dest)?;
             }
 
+            // We want to return either `true` or `false` at random, or else something like
+            // ```
+            // if !is_val_statically_known(0) { unreachable_unchecked(); }
+            // ```
+            // Would not be considered UB, or the other way around (`is_val_statically_known(0)`).
+            "is_val_statically_known" => {
+                let [arg] = check_arg_count(args)?;
+                this.validate_operand(arg)?;
+                let branch: bool = this.machine.rng.get_mut().gen();
+                this.write_scalar(Scalar::from_bool(branch), dest)?;
+            }
+
             // Floating-point operations
             "fabsf32" => {
                 let [f] = check_arg_count(args)?;

tests/pass/const-addrs.rs

@@ -0,0 +1,38 @@
+// The const fn interpreter creates a new AllocId every time it evaluates any const.
+// If we do that in Miri, repeatedly evaluating a const causes unbounded memory use
+// we need to keep track of the base address for that AllocId, and the allocation is never
+// deallocated.
+// In Miri we explicitly store previously-assigned AllocIds for each const and ensure
+// that we only hand out a finite number of AllocIds per const.
+// MIR inlining will put every evaluation of the const we're repeatedly evaluting into the same
+// stack frame, breaking this test.
+//@compile-flags: -Zinline-mir=no
+#![feature(strict_provenance)]
+
+const EVALS: usize = 256;
+
+use std::collections::HashSet;
+fn main() {
+    let mut addrs = HashSet::new();
+    for _ in 0..EVALS {
+        addrs.insert(const_addr());
+    }
+    // Check that the const allocation has multiple base addresses
+    assert!(addrs.len() > 1);
+    // But also that we get a limited number of unique base addresses
+    assert!(addrs.len() < EVALS);
+
+    // Check that within a call we always produce the same address
+    let mut prev = 0;
+    for iter in 0..EVALS {
+        let addr = "test".as_bytes().as_ptr().addr();
+        if iter > 0 {
+            assert_eq!(prev, addr);
+        }
+        prev = addr;
+    }
+}
+
+fn const_addr() -> usize {
+    "test".as_bytes().as_ptr().addr()
+}

tests/pass/intrinsics.rs

@@ -33,6 +33,21 @@ fn main() {
     assert_eq!(intrinsics::likely(false), false);
     assert_eq!(intrinsics::unlikely(true), true);
 
+    let mut saw_true = false;
+    let mut saw_false = false;
+
+    for _ in 0..50 {
+        if unsafe { intrinsics::is_val_statically_known(0) } {
+            saw_true = true;
+        } else {
+            saw_false = true;
+        }
+    }
+    assert!(
+        saw_true && saw_false,
+        "`is_val_statically_known` failed to return both true and false. Congrats, you won the lottery!"
+    );
+
     intrinsics::forget(Bomb);
 
     let _v = intrinsics::discriminant_value(&Some(()));