Stop probing for statx unless necessary
As is the current toy program:
fn main() -> std::io::Result<()> {
use std::fs;
let metadata = fs::metadata("foo.txt")?;
assert!(!metadata.is_dir());
Ok(())
}
... observed under strace will issue:
[snip]
statx(0, NULL, AT_STATX_SYNC_AS_STAT, STATX_ALL, NULL) = -1 EFAULT (Bad address) statx(AT_FDCWD, "foo.txt", AT_STATX_SYNC_AS_STAT, STATX_ALL, {stx_mask=STATX_ALL|STATX_MNT_ID, stx_attributes=0, stx_mode=S_IFREG|0644, stx_size=0, ...}) = 0
While statx is not necessarily always present, checking for it can be delayed to the first error condition. Said condition may very well never happen, in which case the check got avoided altogether.
Note this is still suboptimal as there still will be programs issuing it, but bulk of the problem is removed.
Tested by forbidding the syscall for the binary and observing it correctly falls back to newfstatat.
While here tidy up the commentary, in particular by denoting some problems with the current approach.
Harden the pre-tyctxt query system against accidental recomputation
While the current compiler has no issues where we `take` and then compute the query again, in https://github.com/rust-lang/rust/pull/105462 I accidentally introduced such a case.
I also took the opportunity to remove `peek_mut`, which is only ever used for `global_tcx` to then invoke `enter`. I added an `enter` method directly on the query.
The test relied on Error::last_os_error() coming from the stat call on
the passed file, but there is no guarantee this will be the case.
Instead extract errno from the error returned by the routine.
Patch de facto written by joboet.
Co-authored-by: joboet <jonasboettiger@icloud.com>
When this happens, we ignore the symbol from `compiler_builtins`
in favor of Miri's builtin support.
This allows Miri to target platforms like wasm32-unknown-unknown,
where functions like `memcmp` are provided by `compiler_builtins`.
Data race spans
Fixes https://github.com/rust-lang/miri/issues/2205
This adds output to data race errors very similar to the spans we emit for Stacked Borrows errors. For example, from our test suite:
```
help: The Atomic Load on thread `<unnamed>` is here
--> tests/fail/data_race/atomic_read_na_write_race1.rs:23:13
|
23 | ... (&*c.0).load(Ordering::SeqCst) //~ ERROR: Data race detected between Atomic Load on thread `<unnamed>` and Write o...
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
help: The Write on thread `<unnamed>` is here
--> tests/fail/data_race/atomic_read_na_write_race1.rs:19:13
|
19 | *(c.0 as *mut usize) = 32;
| ^^^^^^^^^^^^^^^^^^^^^^^^^```
```
Because of https://github.com/rust-lang/miri/pull/2647 this comes without a perf regression, according to our benchmarks.
Retag as FnEntry on `drop_in_place`
This commit changes the mir drop shim to always retag its argument as if it were a `&mut`.
cc rust-lang/unsafe-code-guidelines#373
