duplicated_attributes false positive: cfg expressions

[dtolnay]

Lint Name

duplicated_attributes

Reproducer

#[cfg(any(
    all(
        target_os = "linux",
        any(
            target_arch = "aarch64",
            target_arch = "arm",
            target_arch = "hexagon",
            target_arch = "powerpc",
            target_arch = "powerpc64",
            target_arch = "s390x",
            target_arch = "riscv64",
            target_arch = "riscv32"
        )
    ),
    all(
        target_os = "android",
        any(target_arch = "aarch64", target_arch = "arm")
    ),
))]
pub type c_char = u8;

$ cargo clippy
warning: duplicated attribute
  --> src/main.rs:17:13
   |
17 |         any(target_arch = "aarch64", target_arch = "arm")
   |             ^^^^^^^^^^^^^^^^^^^^^^^
   |
note: first defined here
  --> src/main.rs:5:13
   |
5  |             target_arch = "aarch64",
   |             ^^^^^^^^^^^^^^^^^^^^^^^
help: remove this attribute
  --> src/main.rs:17:13
   |
17 |         any(target_arch = "aarch64", target_arch = "arm")
   |             ^^^^^^^^^^^^^^^^^^^^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes
   = note: `#[warn(clippy::duplicated_attributes)]` on by default

warning: duplicated attribute
  --> src/main.rs:17:38
   |
17 |         any(target_arch = "aarch64", target_arch = "arm")
   |                                      ^^^^^^^^^^^^^^^^^^^
   |
note: first defined here
  --> src/main.rs:6:13
   |
6  |             target_arch = "arm",
   |             ^^^^^^^^^^^^^^^^^^^
help: remove this attribute
  --> src/main.rs:17:38
   |
17 |         any(target_arch = "aarch64", target_arch = "arm")
   |                                      ^^^^^^^^^^^^^^^^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes

Clippy just says "remove this attribute" but doing so would violate the intended meaning of the program.

The case shown as an example of the intended purpose of this lint in https://rust-lang.github.io/rust-clippy/master/index.html#/duplicated_attributes (#[allow(dead_code)] #[allow(dead_code)]) seems like a quite different thing than this one.

Version

rustc 1.79.0-nightly (85e449a32 2024-03-22)
binary: rustc
commit-hash: 85e449a3237e82c9ade8936a82bd4fc64cfe1057
commit-date: 2024-03-22
host: x86_64-unknown-linux-gnu
release: 1.79.0-nightly
LLVM version: 18.1.2

Additional Labels

No response

[dtolnay]

@GuillaumeGomez @blyxyas (the lint is new from #12378)

[GuillaumeGomez]

You're right, it should not check cfg like this.

[SteveLauC]

Here is another false positive case found in Nix:

https://github.com/nix-rust/nix/blob/01cd697b82087730b00bafa60e25fd53b025cfdd/src/sys/statfs.rs#L77-L79

#[cfg(any(
    target_os = "freebsd",
    target_os = "android",
    all(target_os = "linux", target_arch = "s390x"),
    all(target_os = "linux", target_env = "musl"),
    all(target_os = "linux", target_env = "ohos"),
    all(
        target_os = "linux",
        not(any(target_arch = "s390x", target_env = "musl"))
    ),
))]
#[derive(Eq, Copy, Clone, PartialEq, Debug)]
pub struct FsType(pub fs_type_t);

Clippy suggests me to remove these "duplicate" linux attributes even though that they are actually not the same thing.

Full log here: https://github.com/nix-rust/nix/blob/01cd697b82087730b00bafa60e25fd53b025cfdd/src/sys/statfs.rs#L77-L79

[tamird]

There's also a false positive when using https://crates.io/crates/test-case.

#[test_case("pointer", false, None, 42)]
#[test_case("pointer", true, None, 21)]
#[test_case("struct_flavors", false, None, 1)]
#[test_case("struct_flavors", true, None, 2)]
fn relocation_tests(
    program: &str,
    with_relocations: bool,
    required_kernel_version: Option<(KernelVersion, &str)>,
    expected: u64,
) {
  ...
}

help: remove this attribute
  --> test/integration-test/src/tests/btf_relocations.rs:23:30
   |
23 | #[test_case("pointer", true, None, 21)]
   |                              ^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes

warning: duplicated attribute
  --> test/integration-test/src/tests/btf_relocations.rs:24:38
   |
24 | #[test_case("struct_flavors", false, None, 1)]
   |                                      ^^^^
   |
note: first defined here
  --> test/integration-test/src/tests/btf_relocations.rs:12:40
   |
12 | #[test_case("enum_unsigned_32", false, None, 0xAAAAAAAA)]
   |                                        ^^^^
help: remove this attribute
  --> test/integration-test/src/tests/btf_relocations.rs:24:38
   |
24 | #[test_case("struct_flavors", false, None, 1)]
   |                                      ^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes

warning: duplicated attribute
  --> test/integration-test/src/tests/btf_relocations.rs:25:37
   |
25 | #[test_case("struct_flavors", true, None, 2)]
   |                                     ^^^^
   |
note: first defined here
  --> test/integration-test/src/tests/btf_relocations.rs:12:40
   |
12 | #[test_case("enum_unsigned_32", false, None, 0xAAAAAAAA)]
   |                                        ^^^^
help: remove this attribute
  --> test/integration-test/src/tests/btf_relocations.rs:25:37
   |
25 | #[test_case("struct_flavors", true, None, 2)]
   |                                     ^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes

warning: `integration-test` (lib test) generated 9 warnings

[GuillaumeGomez]

Normally #12555 should fix both your cases since it's a late pass expansion, but please double-check @tamird once it's merged and the new clippy version published.

[tamird]

@GuillaumeGomez it'd be helpful if you pinged this issue when a new nightly is out with this fix.

[GuillaumeGomez]

I can ask @flip1995 for that when he does it. :)

[flip1995]

2024-04-05

[tamird]

Looks like #12555 did not fix this issue. Tested with rustc 1.79.0-nightly (aa1c45908 2024-04-06) which includes #12555 as rust-lang/rust@e3f3a4b.

[GuillaumeGomez]

Just tested locally and can't reproduce it. Just in case, I opened #12646 to ensure that attributes generated by proc-macros don't emit a warning.

[tamird]

I'm able to reproduce with yesterday's (2024-04-07) nightly as well. If you want, you can revert aya-rs/aya@b552c83 locally and then run cargo +nightly clippy --all-targets --workspace --target x86_64-unknown-linux-gnu -- --deny warnings from the root of the repository.

[GuillaumeGomez]

I'm starting to wonder if proc-macros don't need special treatment here...

[blyxyas]

I'll do some testing =^w^=

[blyxyas]

I am also having these issues. It seems like the current master still lints on it. You can try it by cloning the aya repo, pulling master and doing the following command:

cargo dev lint ../aya -- --all-targets --workspace --target x86_64-unknown-linux-gnu

[GuillaumeGomez]

Ok so that confirms that some tokens are not marked as "from expansion" in proc-macros. Gonna investigate a bit more.

[la10736]

Also on rstest we have the same issue : la10736/rstest#238

warning: duplicated attribute
  --> src/main.rs:13:14
   |
13 | #[case(None, None)] // warns
   |              ^^^^
   |
note: first defined here
  --> src/main.rs:13:8
   |
13 | #[case(None, None)] // warns
   |        ^^^^
help: remove this attribute
  --> src/main.rs:13:14
   |
13 | #[case(None, None)] // warns
   |              ^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes
   = note: `#[warn(clippy::duplicated_attributes)]` on by default

warning: duplicated attribute
  --> src/main.rs:14:17
   |
14 | #[case(Some(1), None)] // warns
   |                 ^^^^
   |
note: first defined here
  --> src/main.rs:13:8
   |
13 | #[case(None, None)] // warns
   |        ^^^^
help: remove this attribute
  --> src/main.rs:14:17
   |
14 | #[case(Some(1), None)] // warns
   |                 ^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes

warning: duplicated attribute
  --> src/main.rs:15:8
   |
15 | #[case(None, Some(1.0))] // warns
   |        ^^^^
   |
note: first defined here
  --> src/main.rs:13:8
   |
13 | #[case(None, None)] // warns
   |        ^^^^
help: remove this attribute
  --> src/main.rs:15:8
   |
15 | #[case(None, Some(1.0))] // warns
   |        ^^^^
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#duplicated_attributes

warning: `rstest_238` (bin "rstest_238") generated 3 warnings
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 1.38s
mdamico@miklap:~/dev_random/rstest_238$ cargo +nightly clippy --version
clippy 0.1.79 (ab5bda1 2024-04-08)

[GuillaumeGomez]

Yes got it. Still trying to write a minimal code to reproduce the bug.

[blyxyas]

I tested #12646, and this should be fixed when that hits nightly :)

[la10736]

@GuillaumeGomez Still have the issue also on the last nightly

mdamico@miklap:~/dev_random/rstest_238$ cargo +nightly clippy --version
clippy 0.1.79 (0bf471f 2024-04-13)

Clippy behaviour is still the same of #12537 (comment)

[la10736]

@tamird I guess that also for test_case the issue is still unresolved with clippy 0.1.79 (0bf471f 2024-04-13).

[GuillaumeGomez]

Then please provide a minimal code I can use to reproduce the bug and fix it (and also use as regression test).

[y21]

Worth noting that the last change here (#12646) is, again, not on nightly yet, since the clippy subtree is only updated every 2 weeks. So it's entirely possible that it is fixed, just not on nightly yet and so we can't do anything but wait. It'd be useful to also run master clippy on the crate (like here) and see what happens there.

[blyxyas]

I second y21's comment. If you're open to some testing, clone the repo (or pull master) and test it with a similar command. If that doesn't work, please report back! ❤️

[la10736]

@GuillaumeGomez I tested with clippy 0.1.79 (7f2fc33 2024-04-22) and now it's works!!!

Thanks!!!

[GuillaumeGomez]

Perfect. :)

Closing then!