Closed
Description
On a 32-bit target, Miri says this program encounters UB:
fn main() {
for _ in 0..4 {
let a = [0u8; 1024 * 1024 * 1024];
drop(&a[..]);
}
}error: Undefined Behavior: overflowing in-bounds pointer arithmetic
--> src/main.rs:4:15
|
4 | drop(&a[..]);
| ^ overflowing in-bounds pointer arithmetic
|
= help: this indicates a bug in the program: it performed an invalid operation, and caused Undefined Behavior
= help: see https://doc.rust-lang.org/nightly/reference/behavior-considered-undefined.html for further information
= note: BACKTRACE:
= note: inside `main` at src/main.rs:4:15: 4:16
The program above is a bit slow to execute but it gets the job done without any unsafe. Using Vec::with_capacity is faster.
I feel like this shouldn't be possible? Or at least we shouldn't report UB?
The Vec version is this:
fn main() {
for _ in 0..4 {
let mut a: Vec<u8> = Vec::with_capacity(1024 * 1024 * 1024);
drop(a.spare_capacity_mut().as_ptr());
}
}error: Undefined Behavior: overflowing in-bounds pointer arithmetic
--> src/main.rs:4:14
|
4 | drop(a.spare_capacity_mut().as_ptr());
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ overflowing in-bounds pointer arithmetic
|
= help: this indicates a bug in the program: it performed an invalid operation, and caused Undefined Behavior
= help: see https://doc.rust-lang.org/nightly/reference/behavior-considered-undefined.html for further information
= note: BACKTRACE:
= note: inside `main` at src/main.rs:4:14: 4:45
In this case I think it's much more clear that the last allocation should fail
Metadata
Metadata
Assignees
Labels
No labels