Commit Graph

6972 Commits

Author SHA1 Message Date
Ralf Jung
d39b683053 dont rerun build script unnecessarily 2022-08-27 16:08:05 -04:00
bors
12907ab4d7 Auto merge of #2514 - RalfJung:dont-compare, r=RalfJung
ensure we don't compare provenance

Comparing provenance is meaningless, since `Wildcard` might be any provenance.
2022-08-27 19:56:35 +00:00
Ralf Jung
b4eff16e0c ensure we don't compare provenance 2022-08-27 15:56:00 -04:00
Andy Wang
6dea99ec71
Supress clippy error 2022-08-27 15:24:59 +01:00
Andy Wang
0f9e009987
Fix C++20 SC access unsoundness 2022-08-27 15:24:57 +01:00
Andy Wang
a2467c9b2a
Add C++20 SC access test 2022-08-27 15:24:25 +01:00
Andy Wang
01dffe0575
Remove useless store buffer search logging 2022-08-27 15:24:25 +01:00
Andy Wang
f4ba8b1160
Improve SC comments 2022-08-27 15:24:24 +01:00
bors
bb8212484f Auto merge of #2455 - RalfJung:scalar-always-init, r=RalfJung
adjust for earlier init checking in the core engine

Miri side of https://github.com/rust-lang/rust/pull/100043
2022-08-27 13:17:31 +00:00
Ralf Jung
df19b856ce rustup 2022-08-27 08:57:06 -04:00
Ralf Jung
3a2252b7b3 adjust for earlier init checking in the core engine 2022-08-27 08:51:28 -04:00
bors
101c4f2e38 Auto merge of #2511 - RalfJung:extern-so, r=RalfJung
some extern-so cleanup and fixes
2022-08-26 21:44:32 +00:00
Ralf Jung
235036fcb3 nicer errors 2022-08-26 17:43:04 -04:00
Ralf Jung
82802337a8 rename test to match usual naming conventions
what's a "fcts"?
2022-08-26 17:40:26 -04:00
Ralf Jung
4e017b54ad fix host/target check for extern-so 2022-08-26 17:38:59 -04:00
bors
4ae68749f7 Auto merge of #2510 - oli-obk:merge_conflict, r=oli-obk
Fix merge conflict

found in https://github.com/rust-lang/miri/pull/2363#discussion_r956100939
2022-08-26 14:38:14 +00:00
Oli Scherer
fa1e51ae7d Fix merge conflict 2022-08-26 14:35:40 +00:00
bors
515038ed7d Auto merge of #2509 - RalfJung:env-data-race, r=RalfJung
fix data race error during env var cleanup

Fixes https://github.com/rust-lang/miri/issues/2508
2022-08-26 13:33:40 +00:00
Ralf Jung
10a1a59c4b fix data race error during env var cleanup 2022-08-26 09:32:30 -04:00
bors
64185014af Auto merge of #2363 - emarteca:int-function-args-returns, r=oli-obk
Adding support for external C functions that have integer (or empty) args and/or returns

Starts addressing `@https://github.com/rust-lang/miri/issues/11`

### Implementation
Adding support for calling external C functions that have any number of integer arguments (types of integers: `i8`, `i16`, `i32`, `i64`, `u8`, `u16`, `u32`, `u64`) and an integer return type (or `void`).
As suggested in `@https://github.com/rust-lang/miri/issues/11,` the [`libffi` crate](https://docs.rs/libffi/latest/libffi/index.html) is used to dispatch the calls to external C functions.

#### Modifications
Main modifications are to:
* [helper.rs](https://github.com/emarteca/miri/blob/int-function-args-returns/src/helpers.rs) : adding a function `call_and_add_external_c_fct_to_context` to read the code pointer to the external C function, dispatch the call, and save the return in MIRI internal memory. Handles all conversions between MIRI and C values (using some macros, also defined in this file).
* [foreign_items.rs](https://github.com/emarteca/miri/blob/int-function-args-returns/src/shims/foreign_items.rs) : handles the calling of `call_and_add_external_c_fct_to_context` in [helper.rs](https://github.com/emarteca/miri/blob/int-function-args-returns/src/helpers.rs) when a foreign item is encountered. Also adds some structs to model C representations of arguments, and the signature of the external C call.

### Testing
Adds tests for the following external functions which are now supported:
* [int tests](https://github.com/emarteca/miri/blob/int-function-args-returns/tests/pass/external_C/int_c_tests.rs):
     - adds 2 to a provided int (no type of int specified, so autocasts)
     - takes the sum of its 12 arguments (tests stack spill)
     - adds 3 to a 16 bit int
     - adds an `i16` to an `i64`
     - returns -10 as an unsigned int
* [void tests](https://github.com/emarteca/miri/blob/int-function-args-returns/tests/pass/external_C/print_from_c.rs)
     - void function that prints from C

### Code review
The code in this PR was reviewed by `@maurer` on [another fork](https://github.com/maurer/miri/pull/1) -- thanks!
2022-08-26 08:51:11 +00:00
Ellen Arteca
88a7882615 C FFI support for functions with int args and returns 2022-08-26 00:53:23 +00:00
bors
d5853bc724 Auto merge of #2449 - oli-obk:ui_test_subtree_sync, r=RalfJung
Use ui_test from crates.io instead of having it in-tree

I have moved a copy of the `ui_test` crate into [a separate repo](https://github.com/oli-obk/ui_test) to facilitate the further non-miri development of it. I will keep syncing until we have reached a point where we don't touch it anymore for miri. At that point we can remove the in-tree version and do further development out of tree.
2022-08-25 15:35:16 +00:00
Oli Scherer
fb071a14bd Use ui_test from crates.io instead of having it in-tree 2022-08-25 15:34:57 +00:00
bors
d09542977c Auto merge of #2448 - oli-obk:test_crates_on_their_own, r=RalfJung
Run `pass` tests without building dependencies first

fixes https://github.com/rust-lang/miri/issues/2442

One thing that I'm wondering is if we should do away with running each folder individually and make `ui_test` support running all tests in parallel (so building deps becomes a thread, and all tests needing deps are blocked on it)
2022-08-25 12:21:07 +00:00
Oli Scherer
6f8885e60f Run pass tests without building dependencies first 2022-08-25 08:03:50 +00:00
bors
9f99aa9d0d Auto merge of #2505 - RalfJung:ui-test, r=oli-obk
cope with rustc aborting due to a signal

`status.code().unwrap()` will panic when the process is aborted due to a signal.  Let's not do that.
2022-08-23 09:43:52 +00:00
Ralf Jung
1914b615ff cope with rustc aborting due to a signal 2022-08-22 21:23:02 -04:00
bors
53138c6fb7 Auto merge of #2504 - Hiroki6:move-thread-to-concurrency-module, r=RalfJung
move thread.rs into concurrency

Follow-up of https://github.com/rust-lang/miri/pull/2500 and closes https://github.com/rust-lang/miri/issues/2199

I moved the `thread.rs` into the `concurrency` module.
2022-08-22 18:49:07 +00:00
Ralf Jung
2107cbbe2f reorganize imports a bit 2022-08-22 14:48:47 -04:00
Hiroki6
5259fb9bb0 move thread.rs into concurrency 2022-08-22 19:13:39 +02:00
bors
e0f0e1fce7 Auto merge of #2503 - RalfJung:tls-dtor-order, r=RalfJung
notes on TLS dtor order

Fixes https://github.com/rust-lang/miri/issues/2486
2022-08-22 16:21:11 +00:00
Ralf Jung
afacf62cf0 notes on TLS dtor order 2022-08-22 12:20:53 -04:00
bors
a109994f22 Auto merge of #2481 - RalfJung:shim-pattern, r=oli-obk
document general shim pattern

r? `@oli-obk`
2022-08-22 14:44:53 +00:00
Ralf Jung
daaa81fc5e document general shim pattern 2022-08-22 10:16:49 -04:00
bors
6e306f9483 Auto merge of #2441 - RalfJung:arithmetic, r=oli-obk
pass clippy::integer_arithmetic in our shims

`@oli-obk` [raised some concerns](https://github.com/rust-lang/miri/pull/2422#discussion_r928220546) about this one. I still think it is the right call, since I don't see a good way to enable overflow checks for our official release builds. I'm open to suggestions though!

Fixes https://github.com/rust-lang/miri/issues/1236
2022-08-22 14:11:46 +00:00
Ralf Jung
8497fd4906 pass clippy::integer_arithmetic in our shims 2022-08-22 10:11:16 -04:00
bors
7b071620cb Auto merge of #2500 - Hiroki6:make-vector-clock-private, r=RalfJung
move vector_clock and sync into concurrency & make vector_clock private

Closes https://github.com/rust-lang/miri/issues/2199

The thread.rs didn't have to be moved to build.
If it makes more sense to move the thread.rs as well, please let me know. I can move it as well.
2022-08-21 23:12:41 +00:00
Hiroki6
8ee6849012 rustfmt 2022-08-21 23:44:43 +02:00
Hiroki6
b073fe2537 move vector_clock and sync into concurrency & make vector_clock private
move thread it back
2022-08-21 23:35:29 +02:00
bors
ab88e64b15 Auto merge of #2495 - RalfJung:ra, r=RalfJung
add ./miri cargo for RA to invoke

RA expects a check command to check *individual workspaces*, whereas `./miri` is designed to check/build/test all parts of Miri. So add a new `./miri cargo` that performs just a single cargo invocation, but with the right env vars so that the build cache can be shared with `./miri check`.
2022-08-20 13:30:34 +00:00
Ralf Jung
0c3ad68a5a add ./miri cargo for RA to invoke 2022-08-19 23:46:17 -04:00
bors
1a03e30b79 Auto merge of #2473 - RalfJung:dyn-upcast-nop, r=RalfJung
allow NOP-casts with mismatching vtables

The Miri side of https://github.com/rust-lang/rust/pull/100208.
2022-08-20 03:33:42 +00:00
Ralf Jung
cf04c1f1ce rustup 2022-08-19 23:33:15 -04:00
Ralf Jung
53037a7c09 allow NOP-casts with mismatching vtables 2022-08-19 23:33:15 -04:00
bors
09118da54f Auto merge of #2454 - saethlin:diagnostics-cleanup, r=RalfJung
Improve information sharing across SB diagnostics

Previous Stacked Borrows diagnostics were missing a lot of information about the state of the interpreter, and it was difficult to add additional state because it was threaded through all the intervening function signatures.

This change factors a lot of the arguments which used to be passed individually to many stacked borrows functions into a single `DiagnosticCx`, which is built in `Stacks::for_each`, and since it wraps a handle to `AllocHistory`, we can now handle more nuanced things like heterogeneous borrow of `!Freeze` types.
2022-08-18 20:37:50 +00:00
Ralf Jung
7397c8e9cf re-bless after rebase 2022-08-18 16:36:53 -04:00
Ben Kimock
15a4f0a9e0 some CurrentSpan cleanup 2022-08-18 16:36:53 -04:00
Ben Kimock
17fc52a06d Clean up diff churn a bit, adjust comments 2022-08-18 15:01:05 -04:00
Ben Kimock
14e72e7ffa Improve information sharing across SB diagnostics
Previous Stacked Borrows diagnostics were missing a lot of information
about the state of the interpreter, and it was difficult to add
additional state because it was threaded through all the intervening
function signatures.

This change factors a lot of the arguments which used to be passed
individually to many stacked borrows functions into a single
`DiagnosticCx`, which is built in `Stacks::for_each`, and since it
wraps a handle to `AllocHistory`, we can now handle more nuanced
things like heterogeneous borrow of `!Freeze` types.
2022-08-18 15:01:03 -04:00
bors
46da748502 Auto merge of #2231 - DrMeepster:winfred, r=RalfJung
Windows thread support: Part 1

This PR adds support for threads on Windows.
2022-08-18 15:26:07 +00:00