Unimplement fmt::Debug for c_void

[Jules-Bertholet]

Taking a shared reference to c_void is never correct, and this impl should never have been added.

This is obviously a breaking change, so needs crater and FCP.

@rustbot label T-libs-api needs-fcp A-FFI

See also #159986, #159935

[rustbot]

r? @Darksonn

rustbot has assigned @Darksonn.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: libs
• libs expanded to 12 candidates
• Random selection from 7 candidates

[Mark-Simulacrum]

@bors try

[rust-bors]

☀️ Try build successful (CI)
Build commit: 82b0460 (82b04601f285a69721a2b7d1db460b48a7edeef4)
Base parent: 26ae60a (26ae60a9eeb20b4935be49d7a931a650fa1d2923)

[theemathas]

@craterbot check

[craterbot]

👌 Experiment pr-160115 created and queued.
🤖 Automatically detected try build 82b0460
⚠️ Try build based on commit 03942bc, but latest commit is 588c33e. Did you forget to make a new try build?
🔍 You can check out the queue and this experiment's details.

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[clarfonthey]

Worth also commenting that there's a blanket Debug impl for *const T and *mut T, and so, this would not affect anything reasonable that uses c_void from deriving Debug.

(Regardless, it would be unsound to actually access data inside these pointers without verifying it's valid, but I could imagine some scenario where someone thought those only implemented Debug when T: Debug, for some weird reason. Those weird reasons don't exist.)

[craterbot]

🚧 Experiment pr-160115 is now running

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[asquared31415]

Something of note is that there exists impl Debug for ! which is similarly invalid to obtain. However, I think that this PR is useful because it's significantly easier to accidentally obtain a c_void in ways that are not protected as well as !. In particular, using Debug on a struct with &c_void is relatively common in the ecosystem, but &! less so because there's no common expectation that it's valid as a pointee.

[Jules-Bertholet]

impl Debug for ! is also useful in dead code (e.g unwrapping a Result<_, !>); the impl for c_void is not.

[clarfonthey]

Yeah, I would expect impl Debug for ! to mostly cover cases like Result<T, !> being Debug, whereas I genuinely can't imagine a case where this would happen with c_void. With how infectious the unsafety from FFI can be, if you're doing type maths enough to the point where &c_void shows up, I would say you're probably doing something wrong.

But either way, crater should let us know how people are using this.

[craterbot]

🎉 Experiment pr-160115 is completed!
📊 674 regressed and 1 fixed (1052665 total)
📊 5754 spurious results on the retry-regressed-list.txt, consider a retry¹ if this is a significant amount.
📰 Open the summary report.

⚠️ If you notice any spurious failure please add them to the denylist!
ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

Footnotes

1. re-run the experiment with crates=https://crater-reports.s3.amazonaws.com/pr-160115/retry-regressed-list.txt ↩

[theemathas]

Most of the regressions (608 of them) are from the lz4-sys crate, here.

#[derive(Debug)]
#[repr(C)]
pub struct LZ4StreamEncode(c_void);

#[derive(Debug)]
#[repr(C)]
pub struct LZ4StreamDecode(c_void);

These structs are only ever used via a raw pointer, and the derives seem extraneous.

[rustbot]

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.