Fix bugs where unique_range became invalid
And also expand the cache integrity checks to cover this case.
I'm going to run this over all the ICEs I've gotten out of Miri recently, could be a bit.
Fixes https://github.com/rust-lang/miri/issues/2389
fix comparing wide raw pointers
Fixes https://github.com/rust-lang/rust/issues/96169
However I am not sure if these are the correct semantics. I'll wait for confirmation in that issue.
Enable permissive provenance by default
This completes the plan laid out in https://github.com/rust-lang/miri/issues/2133:
- We use permissive provenance with wildcard pointers by default.
- We print a warning on int2ptr casts. `-Zmiri-permissive-provenance` suppresses the warning; `-Zmiri-strict-provenance` turns it into a hard error.
- Raw pointer tagging is now always enabled, so we remove the `-Zmiri-tag-raw-pointers` flag and the code for untagged pointers. (Passing the flag still works, for compatibility -- but we just ignore it, with a warning.)
We also fix an intptrcast issue:
- Only live allocations are considered when computing the AllocId from an address.
So, finally, Miri has a good story for ptr2int2ptr roundtrips *and* no weird false negatives when doing raw pointer stuff with Stacked Borrows. :-) 🎉 Thanks a lot to everyone who helped with this, in particular `@carbotaniuman` who convinced me this is even possible.
Fixes https://github.com/rust-lang/miri/issues/2133
Fixes https://github.com/rust-lang/miri/issues/1866
Fixes https://github.com/rust-lang/miri/issues/1993
test that futexes induce appropriate synchronization
This fails when I remove the `validate_lock_acquire` or `validate_lock_release` from `futex_wake`. So finally we got those code paths actually covered in tests. :)
Format tests with rustfmt (288-299 of 299)
Extracted from #2097.
I'll make a separate PR to enable checking the `tests` directory's formatting in CI. I'll need to rebase that after both this and #2254 have landed, and if any new non-rustfmt-formatted files appear in the meantime, we can include formatting those in the same PR that enables the CI.
Format tests with rustfmt (151-200 of 300)
Extracted from #2097.
This PR is still only doing the easy cases with no comments involved.
In the next PRs after this, I'll start grouping by common comment patterns, e.g. all the cases resembling https://github.com/rust-lang/miri/pull/2097#discussion_r862436672 together in one PR.
Format tests with rustfmt (51-100 of 300)
Extracted from #2097.
Like #2244, this is intended to be "easy" cases which don't involve comments in the vicinity.
Five of the files being skipped here are because rustfmt is buggy (see
the error messages below). The others have clearly preferable manual
formatting.
error[internal]: left behind trailing whitespace
--> tests/fail/validity/transmute_through_ptr.rs:18:18:1
|
18 |
| ^^^^
|
warning: rustfmt has failed to format. See previous 1 errors.
error[internal]: left behind trailing whitespace
--> tests/fail/stacked_borrows/illegal_read2.rs:10:10:1
|
10 |
| ^^^^
|
warning: rustfmt has failed to format. See previous 1 errors.
error[internal]: left behind trailing whitespace
--> tests/fail/stacked_borrows/illegal_read5.rs:15:15:1
|
15 |
| ^^^^
|
warning: rustfmt has failed to format. See previous 1 errors.
error[internal]: left behind trailing whitespace
--> tests/fail/stacked_borrows/illegal_read1.rs:10:10:1
|
10 |
| ^^^^
|
warning: rustfmt has failed to format. See previous 1 errors.
error[internal]: left behind trailing whitespace
--> tests/fail/erroneous_const2.rs:9:9:1
|
9 |
| ^^^^
|
warning: rustfmt has failed to format. See previous 1 errors.
Prevent futex_wait from actually waiting if a concurrent waker was executed before us
Fixes#2223
Two SC fences were placed in `futex_wake` (after the caller has changed `addr`), and in `futex_wait` (before we read `addr`). This guarantees that `futex_wait` sees the value written to `addr` before the last `futex_wake` call, should one exists, and avoid going into sleep with no one else to wake us up.
ada7b72a87/src/concurrency/weak_memory.rs (L324-L326)
Earlier I proposed to use `fetch_add(0)` to read the latest value in MO, though this isn't the proper way to do it and breaks aliasing: syscall caller may pass in a `*const` from a `&` and Miri complains about write to a `SharedReadOnly` location, causing this test to fail.
ada7b72a87/tests/pass/concurrency/linux-futex.rs (L56-L68)
make Miri's scheduler proper round-robin
When thread N blocks or yields, we activate thread N+1 next, rather than always activating thread 0. This should guarantee that as long as all threads regularly yield, each thread eventually takes a step again.
Fixes the "multiple loops that yield playing ping-pong" part of https://github.com/rust-lang/miri/issues/1388.
`@cbeuw` I hope this doesn't screw up the scheduler-dependent tests you are adding in your PR.
