3 [![Actions build status][actions-badge]][actions-url]
5 [actions-badge]: https://github.com/rust-lang/miri/workflows/CI/badge.svg?branch=master
6 [actions-url]: https://github.com/rust-lang/miri/actions
8 An experimental interpreter for [Rust][rust]'s
9 [mid-level intermediate representation][mir] (MIR). It can run binaries and
10 test suites of cargo projects and detect certain classes of
11 [undefined behavior](https://doc.rust-lang.org/reference/behavior-considered-undefined.html),
14 * Out-of-bounds memory accesses and use-after-free
15 * Invalid use of uninitialized data
16 * Violation of intrinsic preconditions (an [`unreachable_unchecked`] being
17 reached, calling [`copy_nonoverlapping`] with overlapping ranges, ...)
18 * Not sufficiently aligned memory accesses and references
19 * Violation of *some* basic type invariants (a `bool` that is not 0 or 1, for example,
20 or an invalid enum discriminant)
21 * **Experimental**: Violations of the [Stacked Borrows] rules governing aliasing
24 On top of that, Miri will also tell you about memory leaks: when there is memory
25 still allocated at the end of the execution, and that memory is not reachable
26 from a global `static`, Miri will raise an error.
28 Miri has already discovered some [real-world bugs](#bugs-found-by-miri). If you
29 found a bug with Miri, we'd appreciate if you tell us and we'll add it to the
32 However, be aware that Miri will **not catch all cases of undefined behavior**
33 in your program, and cannot run all programs:
35 * There are still plenty of open questions around the basic invariants for some
36 types and when these invariants even have to hold. Miri tries to avoid false
37 positives here, so if you program runs fine in Miri right now that is by no
38 means a guarantee that it is UB-free when these questions get answered.
40 In particular, Miri does currently not check that integers/floats are
41 initialized or that references point to valid data.
42 * If the program relies on unspecified details of how data is laid out, it will
43 still run fine in Miri -- but might break (including causing UB) on different
44 compiler versions or different platforms.
45 * Program execution is non-deterministic when it depends, for example, on where
46 exactly in memory allocations end up, or on the exact interleaving of
47 concurrent threads. Miri tests one of many possible executions of your
48 program. You can alleviate this to some extend by running Miri with different
49 values for `-Zmiri-seed`, but that will still by far not explore all possible
51 * Miri runs the program as a platform-independent interpreter, so the program
52 has no access to most platform-specific APIs or FFI. A few APIs have been
53 implemented (such as printing to stdout) but most have not: for example, Miri
54 currently does not support SIMD or networking.
56 [rust]: https://www.rust-lang.org/
57 [mir]: https://github.com/rust-lang/rfcs/blob/master/text/1211-mir.md
58 [`unreachable_unchecked`]: https://doc.rust-lang.org/stable/std/hint/fn.unreachable_unchecked.html
59 [`copy_nonoverlapping`]: https://doc.rust-lang.org/stable/std/ptr/fn.copy_nonoverlapping.html
60 [Stacked Borrows]: https://github.com/rust-lang/unsafe-code-guidelines/blob/master/wip/stacked-borrows.md
65 Install Miri on Rust nightly via `rustup`:
68 rustup +nightly component add miri
71 If `rustup` says the `miri` component is unavailable, that's because not all
72 nightly releases come with all tools. Check out
73 [this website](https://rust-lang.github.io/rustup-components-history) to
74 determine a nightly version that comes with Miri and install that using
75 `rustup toolchain install nightly-YYYY-MM-DD`.
77 Now you can run your project in Miri:
79 1. Run `cargo clean` to eliminate any cached dependencies. Miri needs your
80 dependencies to be compiled the right way, that would not happen if they have
81 previously already been compiled.
82 2. To run all tests in your project through Miri, use `cargo miri test`.
83 3. If you have a binary project, you can run it through Miri using `cargo miri run`.
85 The first time you run Miri, it will perform some extra setup and install some
86 dependencies. It will ask you for confirmation before installing anything.
88 `cargo miri run/test` supports the exact same flags as `cargo run/test`. You
89 can pass arguments to Miri via `MIRIFLAGS`. For example,
90 `MIRIFLAGS="-Zmiri-disable-stacked-borrows" cargo miri run` runs the program
91 without checking the aliasing of references.
93 When compiling code via `cargo miri`, the `cfg(miri)` config flag is set. You
94 can use this to ignore test cases that fail under Miri because they do things
95 Miri does not support:
99 #[cfg_attr(miri, ignore)]
100 fn does_not_work_on_miri() {
101 std::thread::spawn(|| println!("Hello Thread!"))
107 There is no way to list all the infinite things Miri cannot do, but the
108 interpreter will explicitly tell you when it finds something unsupported:
111 error: unsupported operation: can't call foreign function: bind
113 = help: this is likely not a bug in the program; it indicates that the program \
114 performed an operation that the interpreter does not support
117 ### Cross-interpretation: running for different targets
119 Miri can not only run a binary or test suite for your host target, it can also
120 perform cross-interpretation for arbitrary foreign targets: `cargo miri run
121 --target x86_64-unknown-linux-gnu` will run your program as if it was a Linux
122 program, no matter your host OS. This is particularly useful if you are using
123 Windows, as the Linux target is much better supported than Windows targets.
125 You can also use this to test platforms with different properties than your host
126 platform. For example `cargo miri test --target mips64-unknown-linux-gnuabi64`
127 will run your test suite on a big-endian target, which is useful for testing
128 endian-sensitive code.
130 ### Running Miri on CI
132 To run Miri on CI, make sure that you handle the case where the latest nightly
133 does not ship the Miri component because it currently does not build. For
134 example, you can use the following snippet to always test with the latest
135 nightly that *does* come with Miri:
138 MIRI_NIGHTLY=nightly-$(curl -s https://rust-lang.github.io/rustup-components-history/x86_64-unknown-linux-gnu/miri)
139 echo "Installing latest nightly with Miri: $MIRI_NIGHTLY"
140 rustup set profile minimal
141 rustup default "$MIRI_NIGHTLY"
142 rustup component add miri
149 When using the above instructions, you may encounter a number of confusing compiler
152 ### "note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace"
154 You may see this when trying to get Miri to display a backtrace. By default, Miri
155 doesn't expose any environment to the program, so running
156 `RUST_BACKTRACE=1 cargo miri test` will not do what you expect.
158 To get a backtrace, you need to disable isolation
159 [using `-Zmiri-disable-isolation`](#miri-flags):
162 RUST_BACKTRACE=1 MIRIFLAGS="-Zmiri-disable-isolation" cargo miri test
165 #### "found possibly newer version of crate `std` which `<dependency>` depends on"
167 Your build directory may contain artifacts from an earlier build that have/have
168 not been built for Miri. Run `cargo clean` before switching from non-Miri to
169 Miri builds and vice-versa.
171 #### "found crate `std` compiled by an incompatible version of rustc"
173 You may be running `cargo miri` with a different compiler version than the one
174 used to build the custom libstd that Miri uses, and Miri failed to detect that.
175 Try deleting `~/.cache/miri`.
177 #### "no mir for `std::rt::lang_start_internal`"
179 This means the sysroot you are using was not compiled with Miri in mind. This
180 should never happen when you use `cargo miri` because that takes care of setting
181 up the sysroot. If you are using `miri` (the Miri driver) directly, see the
182 [contributors' guide](CONTRIBUTING.md) for how to use `./miri` to best do that.
185 ## Miri `-Z` flags and environment variables
186 [miri-flags]: #miri--z-flags-and-environment-variables
188 Miri adds its own set of `-Z` flags, which are usually set via the `MIRIFLAGS`
189 environment variable:
191 * `-Zmiri-disable-alignment-check` disables checking pointer alignment, so you
192 can focus on other failures, but it means Miri can miss bugs in your program.
193 Using this flag is **unsound**.
194 * `-Zmiri-disable-stacked-borrows` disables checking the experimental
195 [Stacked Borrows] aliasing rules. This can make Miri run faster, but it also
196 means no aliasing violations will be detected. Using this flag is **unsound**
197 (but the affected soundness rules are experimental).
198 * `-Zmiri-disable-validation` disables enforcing validity invariants, which are
199 enforced by default. This is mostly useful to focus on other failures (such
200 as out-of-bounds accesses) first. Setting this flag means Miri can miss bugs
201 in your program. However, this can also help to make Miri run faster. Using
202 this flag is **unsound**.
203 * `-Zmiri-disable-isolation` disables host isolation. As a consequence,
204 the program has access to host resources such as environment variables, file
205 systems, and randomness.
206 * `-Zmiri-env-exclude=<var>` keeps the `var` environment variable isolated from
207 the host so that it cannot be accessed by the program. Can be used multiple
208 times to exclude several variables. On Windows, the `TERM` environment
209 variable is excluded by default.
210 * `-Zmiri-ignore-leaks` disables the memory leak checker.
211 * `-Zmiri-seed=<hex>` configures the seed of the RNG that Miri uses to resolve
212 non-determinism. This RNG is used to pick base addresses for allocations.
213 When isolation is enabled (the default), this is also used to emulate system
214 entropy. The default seed is 0. **NOTE**: This entropy is not good enough
215 for cryptographic use! Do not generate secret keys in Miri or perform other
216 kinds of cryptographic operations that rely on proper random numbers.
217 * `-Zmiri-symbolic-alignment-check` makes the alignment check more strict. By
218 default, alignment is checked by casting the pointer to an integer, and making
219 sure that is a multiple of the alignment. This can lead to cases where a
220 program passes the alignment check by pure chance, because things "happened to
221 be" sufficiently aligned -- there is no UB in this execution but there would
222 be UB in others. To avoid such cases, the symbolic alignment check only takes
223 into account the requested alignment of the relevant allocation, and the
224 offset into that allocation. This avoids missing such bugs, but it also
225 incurs some false positives when the code does manual integer arithmetic to
226 ensure alignment. (The standard library `align_to` method works fine in both
227 modes; under symbolic alignment it only fills the middle slice when the
228 allocation guarantees sufficient alignment.)
229 * `-Zmiri-track-alloc-id=<id>` shows a backtrace when the given allocation is
230 being allocated or freed. This helps in debugging memory leaks and
232 * `-Zmiri-track-call-id=<id>` shows a backtrace when the given call id is
233 assigned to a stack frame. This helps in debugging UB related to Stacked
234 Borrows "protectors".
235 * `-Zmiri-track-pointer-tag=<tag>` shows a backtrace when the given pointer tag
236 is popped from a borrow stack (which is where the tag becomes invalid and any
237 future use of it will error). This helps you in finding out why UB is
238 happening and where in your code would be a good place to look for it.
239 * `-Zmiri-track-raw-pointers` makes Stacked Borrows track a pointer tag even for
240 raw pointers. This can make valid code fail to pass the checks, but also can
241 help identify latent aliasing issues in code that Miri accepts by default. You
242 can recognize false positives by "<untagged>" occurring in the message -- this
243 indicates a pointer that was cast from an integer, so Miri was unable to track
244 this pointer. Make sure to use a non-Windows target with this flag, as the
245 Windows runtime makes use of integer-pointer casts.
247 Some native rustc `-Z` flags are also very relevant for Miri:
249 * `-Zmir-opt-level` controls how many MIR optimizations are performed. Miri
250 overrides the default to be `0`; be advised that using any higher level can
251 make Miri miss bugs in your program because they got optimized away.
252 * `-Zalways-encode-mir` makes rustc dump MIR even for completely monomorphic
253 functions. This is needed so that Miri can execute such functions, so Miri
254 sets this flag per default.
255 * `-Zmir-emit-retag` controls whether `Retag` statements are emitted. Miri
256 enables this per default because it is needed for [Stacked Borrows].
258 Moreover, Miri recognizes some environment variables:
260 * `MIRI_LOG`, `MIRI_BACKTRACE` control logging and backtrace printing during
261 Miri executions, also [see "Testing the Miri driver" in `CONTRIBUTING.md`][testing-miri].
262 * `MIRIFLAGS` (recognized by `cargo miri` and the test suite) defines extra
263 flags to be passed to Miri.
264 * `MIRI_SYSROOT` (recognized by `cargo miri` and the test suite)
265 indicates the sysroot to use. To do the same thing with `miri`
266 directly, use the `--sysroot` flag.
267 * `MIRI_TEST_TARGET` (recognized by the test suite) indicates which target
268 architecture to test against. `miri` and `cargo miri` accept the `--target`
269 flag for the same purpose.
271 The following environment variables are internal, but used to communicate between
272 different Miri binaries, and as such worth documenting:
274 * `MIRI_BE_RUSTC` when set to any value tells the Miri driver to actually not
275 interpret the code but compile it like rustc would. This is useful to be sure
276 that the compiled `rlib`s are compatible with Miri.
277 When set while running `cargo-miri`, it indicates that we are part of a sysroot
278 build (for which some crates need special treatment).
279 * `MIRI_CWD` when set to any value tells the Miri driver to change to the given
280 directory after loading all the source files, but before commencing
281 interpretation. This is useful if the interpreted program wants a different
282 working directory at run-time than at build-time.
284 [testing-miri]: CONTRIBUTING.md#testing-the-miri-driver
286 ## Miri `extern` functions
288 Miri provides some `extern` functions that programs can import to access
289 Miri-specific functionality:
294 /// Miri-provided extern function to mark the block `ptr` points to as a "root"
295 /// for some static memory. This memory and everything reachable by it is not
296 /// considered leaking even if it still exists when the program terminates.
298 /// `ptr` has to point to the beginning of an allocated block.
299 fn miri_static_root(ptr: *const u8);
301 /// Miri-provided extern function to obtain a backtrace of the current call stack.
302 /// This returns a boxed slice of pointers - each pointer is an opaque value
303 /// that is only useful when passed to `miri_resolve_frame`
304 /// The `flags` argument must be `0`.
305 fn miri_get_backtrace(flags: u64) -> Box<[*mut ()]>;
307 /// Miri-provided extern function to resolve a frame pointer obtained
308 /// from `miri_get_backtrace`. The `flags` argument must be `0`,
309 /// and `MiriFrame` should be declared as follows:
313 /// struct MiriFrame {
314 /// // The name of the function being executed, encoded in UTF-8
316 /// // The filename of the function being executed, encoded in UTF-8
317 /// filename: Box<[u8]>,
318 /// // The line number currently being executed in `filename`, starting from '1'.
320 /// // The column number currently being executed in `filename`, starting from '1'.
322 /// // The function pointer to the function currently being executed.
323 /// // This can be compared against function pointers obtained by
324 /// // casting a function (e.g. `my_fn as *mut ()`)
329 /// The fields must be declared in exactly the same order as they appear in `MiriFrame` above.
330 /// This function can be called on any thread (not just the one which obtained `frame`).
331 fn miri_resolve_frame(frame: *mut (), flags: u64) -> MiriFrame;
333 /// Miri-provided extern function to begin unwinding with the given payload.
335 /// This is internal and unstable and should not be used; we give it here
336 /// just to be complete.
337 fn miri_start_panic(payload: *mut u8) -> !;
341 ## Contributing and getting help
343 If you want to contribute to Miri, great! Please check out our
344 [contribution guide](CONTRIBUTING.md).
346 For help with running Miri, you can open an issue here on
347 GitHub or contact us (`oli-obk` and `RalfJ`) on the [Rust Zulip].
349 [Rust Zulip]: https://rust-lang.zulipchat.com
353 This project began as part of an undergraduate research course in 2015 by
354 @solson at the [University of Saskatchewan][usask]. There are [slides] and a
355 [report] available from that project. In 2016, @oli-obk joined to prepare miri
356 for eventually being used as const evaluator in the Rust compiler itself
357 (basically, for `const` and `static` stuff), replacing the old evaluator that
358 worked directly on the AST. In 2017, @RalfJung did an internship with Mozilla
359 and began developing miri towards a tool for detecting undefined behavior, and
360 also using miri as a way to explore the consequences of various possible
361 definitions for undefined behavior in Rust. @oli-obk's move of the miri engine
362 into the compiler finally came to completion in early 2018. Meanwhile, later
363 that year, @RalfJung did a second internship, developing miri further with
364 support for checking basic type invariants and verifying that references are
365 used according to their aliasing restrictions.
367 [usask]: https://www.usask.ca/
368 [slides]: https://solson.me/miri-slides.pdf
369 [report]: https://solson.me/miri-report.pdf
371 ## Bugs found by Miri
373 Miri has already found a number of bugs in the Rust standard library and beyond, which we collect here.
377 * [`Debug for vec_deque::Iter` accessing uninitialized memory](https://github.com/rust-lang/rust/issues/53566)
378 * [`Vec::into_iter` doing an unaligned ZST read](https://github.com/rust-lang/rust/pull/53804)
379 * [`From<&[T]> for Rc` creating a not sufficiently aligned reference](https://github.com/rust-lang/rust/issues/54908)
380 * [`BTreeMap` creating a shared reference pointing to a too small allocation](https://github.com/rust-lang/rust/issues/54957)
381 * [`Vec::append` creating a dangling reference](https://github.com/rust-lang/rust/pull/61082)
382 * [Futures turning a shared reference into a mutable one](https://github.com/rust-lang/rust/pull/56319)
383 * [`str` turning a shared reference into a mutable one](https://github.com/rust-lang/rust/pull/58200)
384 * [`rand` performing unaligned reads](https://github.com/rust-random/rand/issues/779)
385 * [The Unix allocator calling `posix_memalign` in an invalid way](https://github.com/rust-lang/rust/issues/62251)
386 * [`getrandom` calling the `getrandom` syscall in an invalid way](https://github.com/rust-random/getrandom/pull/73)
387 * [`Vec`](https://github.com/rust-lang/rust/issues/69770) and [`BTreeMap`](https://github.com/rust-lang/rust/issues/69769) leaking memory under some (panicky) conditions
388 * [`beef` leaking memory](https://github.com/maciejhirsz/beef/issues/12)
389 * [`EbrCell` using uninitialized memory incorrectly](https://github.com/Firstyear/concread/commit/b15be53b6ec076acb295a5c0483cdb4bf9be838f#diff-6282b2fc8e98bd089a1f0c86f648157cR229)
390 * [TiKV performing an unaligned pointer access](https://github.com/tikv/tikv/issues/7613)
391 * [`servo_arc` creating a dangling shared reference](https://github.com/servo/servo/issues/26357)
392 * [TiKV constructing out-of-bounds pointers (and overlapping mutable references)](https://github.com/tikv/tikv/pull/7751)
393 * [`encoding_rs` doing out-of-bounds pointer arithmetic](https://github.com/hsivonen/encoding_rs/pull/53)
394 * [TiKV using `Vec::from_raw_parts` incorrectly](https://github.com/tikv/agatedb/pull/24)
396 Violations of [Stacked Borrows] found that are likely bugs (but Stacked Borrows is currently just an experiment):
398 * [`VecDeque::drain` creating overlapping mutable references](https://github.com/rust-lang/rust/pull/56161)
399 * Various `BTreeMap` problems
400 * [`BTreeMap` iterators creating mutable references that overlap with shared references](https://github.com/rust-lang/rust/pull/58431)
401 * [`BTreeMap::iter_mut` creating overlapping mutable references](https://github.com/rust-lang/rust/issues/73915)
402 * [`BTreeMap` node insertion using raw pointers outside their valid memory area](https://github.com/rust-lang/rust/issues/78477)
403 * [`LinkedList` cursor insertion creating overlapping mutable references](https://github.com/rust-lang/rust/pull/60072)
404 * [`Vec::push` invalidating existing references into the vector](https://github.com/rust-lang/rust/issues/60847)
405 * [`align_to_mut` violating uniqueness of mutable references](https://github.com/rust-lang/rust/issues/68549)
406 * [`sized-chunks` creating aliasing mutable references](https://github.com/bodil/sized-chunks/issues/8)
407 * [`String::push_str` invalidating existing references into the string](https://github.com/rust-lang/rust/issues/70301)
408 * [`ryu` using raw pointers outside their valid memory area](https://github.com/dtolnay/ryu/issues/24)
409 * [ink! creating overlapping mutable references](https://github.com/rust-lang/miri/issues/1364)
410 * [TiKV creating overlapping mutable reference and raw pointer](https://github.com/tikv/tikv/pull/7709)
411 * [Windows `Env` iterator using a raw pointer outside its valid memory area](https://github.com/rust-lang/rust/pull/70479)
412 * [`VecDeque::iter_mut` creating overlapping mutable references](https://github.com/rust-lang/rust/issues/74029)
413 * [Various standard library aliasing issues involving raw pointers](https://github.com/rust-lang/rust/pull/78602)
417 Licensed under either of
419 * Apache License, Version 2.0 ([LICENSE-APACHE](LICENSE-APACHE) or
420 http://www.apache.org/licenses/LICENSE-2.0)
421 * MIT license ([LICENSE-MIT](LICENSE-MIT) or
422 http://opensource.org/licenses/MIT)
428 Unless you explicitly state otherwise, any contribution intentionally submitted
429 for inclusion in the work by you shall be dual licensed as above, without any
430 additional terms or conditions.