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