Hunting down an issue on the tracker

ExpHP · January 15, 2019, 12:38am

So... this is going to sound silly.

I'm trying to hunt down a specific github issue that I know I saw on the tracker---but that I never commented in or subscribed to---because seeing it will remind me of how to create a soundness bug.

Some months ago, I think when NLL was riding the nightly and beta trains into stable, an inconspicuous issue appeared on the tracker with a vague title in broken English---like, "Lifetime is not as helpful as they could be" or something like that. The poster was annoyed that rust did not catch a bug in their program. The last time I saw it, it was a week old and had 0 comments, and appeared to be doomed to forever remain that way.

If I recall, the OP's code sample compiled on nightly, but not on stable, so it was an example of something "newly allowed by NLL." The code sample used a &'static mut in some manner (I think as a member of a struct?) that seemingly ought to have made the borrow checker mad. Playing around with the sample more, I eventually came up with an example of true UB that was enabled by the change.

However, the UB was only possible if you possessed a &'static mut T. I kept trying but I couldn't come up with any way to create a &'static mut T without using unsafe... and then life moved on and I never posted my comment.

...then today, I suddenly remembered that Box::leak exists, but I can't remember what I did to create UB before! Seeing the issue would help jog my memory, but I can't seem to find it any more even with some really vague search queries.

Maybe the issue was renamed? The bug has probably already been found and fixed, but I just want some peace of mind...

RustyYato · January 15, 2019, 12:55am

Did you mean this

https://github.com/rust-lang/rust/issues/50162

This has been fixed, and I don't think using Box::leak would lead to a soundness bug here

RustyYato · January 15, 2019, 12:57am

I also found this one, but it is from 2017 and stranded here on the user forums

Here's the git tracking issue,

github.com/rust-lang/rust

borrowed referent of a `&T` sometimes incorrectly allowed

opened 12:45AM - 07 Jan 17 UTC

closed 12:15AM - 27 Sep 19 UTC

nikomatsakis

E-needs-test P-medium A-borrow-checker T-compiler I-unsound A-NLL C-bug NLL-fixed-by-NLL

@jorendorf asks on [the users forum](https://users.rust-lang.org/t/why-are-reads…-sometimes-tolerated-during-mut-borrows-and-sometimes-not/8774) about a curious discrepancy around fields. It seems that implicit borrows sometimes seem to get overlooked in the borrow checker. This seems like a kind of bad bug, though it's exact scope is unclear until we investigate a bit more. Here is a variant of @jorendorf's example which is pretty clearly wrong. Here, the `block` variable is mutably borrowed into `x`, so it should not be accessible via `let p`: ```rust #![allow(dead_code)] pub struct Block<'a> { current: &'a u8, unrelated: &'a u8, } fn bump<'a>(mut block: &mut Block<'a>) { let x = &mut block; let p: &'a u8 = &*block.current; // (use `x` and `p` so enabling NLL doesn't assign overly short lifetimes) drop(x); drop(p); } fn main() {} ``` I'm guessing that the problem has to do with the logic around borrowing the referent of an `&T` (in this case, we are borrowing `*block.current`). In particular, we deem that to be "safe" for the scope of `'a` because the data is independently guaranteed to be valid that long (this is reasonable). But we still need to validate that `block.current` can be (instantaneously) read. It seems we are not doing that. But this is all a hypothesis: I've not dug into the code to validate it.

ExpHP · January 15, 2019, 1:47pm

Thanks for looking! Unfortunately neither look like it. I think it might have involved a destructor?

It definitely was not posted by a regular.

(I also remember the second one; I posted in it!)

RustyYato · January 15, 2019, 9:53pm

Huh, I didn't notice that

Topic		Replies	Views
Blog post series: After NLL -- what's next for borrowing and lifetimes? announcements	121	9035	January 2, 2022
Borrow checker too strict. Will this be solved via NLL?	10	1214	January 12, 2023
Inconsistency between stacked borrows and my mental model	3	980	November 10, 2020
NLL seems to not work under HRTB help	2	478	February 20, 2023
Feedback on the new "borrowing" section	5	1413	January 12, 2023

Hunting down an issue on the tracker

Related Topics