refactor: enable clippy unsafe blocks related lints

[TheBestTvarynka]

Checks for unsafe blocks and impls without a // SAFETY: comment explaining why the unsafe operations performed inside the block are safe.

https://rust-lang.github.io/rust-clippy/master/index.html#undocumented_unsafe_blocks

Checks for unsafe blocks that contain more than one unsafe operation.

https://rust-lang.github.io/rust-clippy/master/index.html#multiple_unsafe_ops_per_block

[CBenoit]

Thanks!

[CBenoit]

You need to verify on other platforms too, but I enabled auto-merge!

[TheBestTvarynka]

Honestly, I do not know why this commit fixed the error reported by Miri (see the last failed CI pipeline): 6b30a7a.

I wrote a similar example in the playground, and it works with the previous approach: https://play.rust-lang.org/?version=stable&mode=debug&edition=2024&gist=3912329ddc1fc332b47c22ce68d33fdd.

Maybe it's a bug in the stacked borrows implementation (they are still experimental). But then, why did it not happen before? 🤔🤔🤔

[CBenoit]

I’m a bit surprised by this because obtaining a pointer is generally not unsafe 🤔

[CBenoit]

Honestly, I do not know why this commit fixed the error reported by Miri (see the last failed CI pipeline): 6b30a7a.

I wrote a similar example in the playground, and it works with the previous approach: https://play.rust-lang.org/?version=stable&mode=debug&edition=2024&gist=3912329ddc1fc332b47c22ce68d33fdd.

Maybe it's a bug in the stacked borrows implementation (they are still experimental). But then, why did it not happen before? 🤔🤔🤔

Very good question! It’s nice that you wrote a reproducer, unfortunate that it didn’t help isolate the issue 😕
Since miri is performing dynamic analysis (the program is actually running), any chance you didn’t reproduce some specific memory state?