cp: fix device file creation with -r flag

[AnarchistHoneybun]

Fix cp -r to create device files instead of copying content when copying character and block devices.

Add device detection for character and block devices similar to existing FIFO handling. Implement make_char_device() and make_block_device() functions in uucore using mknod syscall to create device files with correct major/minor numbers.

Update copy logic to create device files when --recursive is used and --copy-contents is not specified.

Fixes #8679

• Add regression tests
• Come up with a refactor plan

[sylvestre]

it is starting to be too much, can we refactor this ?

[AnarchistHoneybun]

could you elaborate a little?

[sylvestre]

the function has too many parameters, we should have less

[AnarchistHoneybun]

will look into it

[AnarchistHoneybun]

the function has too many parameters, we should have less

Would something like this be acceptable:

#[derive(Debug, Clone)]
struct SourceFileType {
    is_symlink: bool,
    is_fifo: bool,
    is_socket: bool,
    is_char_device: bool,
    is_block_device: bool,
    #[cfg(unix)]
    is_stream: bool,
}

impl SourceFileType {
    fn from_metadata(metadata: &Metadata) -> Self {
        #[cfg(unix)]
        {
            Self {
                is_symlink: metadata.is_symlink(),
                is_fifo: metadata.file_type().is_fifo(),
                is_socket: metadata.file_type().is_socket(),
                is_char_device: metadata.file_type().is_char_device(),
                is_block_device: metadata.file_type().is_block_device(),
                is_stream: is_stream(metadata),
            }
        }
        #[cfg(not(unix))]
        {
            Self {
                is_symlink: metadata.is_symlink(),
                is_fifo: false,
                is_socket: false,
                is_char_device: false,
                is_block_device: false,
            }
        }
    }
    
    fn is_special_file(&self) -> bool {
        self.is_fifo || self.is_socket || self.is_char_device || self.is_block_device
    }
}
...

fn handle_copy_mode(
    source: &Path,
    dest: &Path,
    options: &Options,
    context: &str,
    source_metadata: &Metadata,
    symlinked_files: &mut HashSet<FileInformation>,
    source_in_command_line: bool,
    source_file_type: &SourceFileType,
) -> CopyResult<PerformedAction>

[bdrung]

Why not use an enum here? All those file types (symlink, fifo, socket, char, block) are distinct.

[AnarchistHoneybun]

hmm, what about stream? (apologise if these questions are stupid, I'm less familiar with the linux environment than I am with Rust, I thought of an enum at first but then stopped short imagining well if the first person to do it didn't make it an enum, maybe it's possible they can happen simultaneously)

[bdrung]

I checked the is_stream functions and it says:

file_type.is_fifo() || file_type.is_char_device() || file_type.is_block_device()

So the is_stream struct element could be dropped and converted into a check for matching the enum.

[codspeed-hq]

CodSpeed Performance Report

Merging #8693 will not alter performance

_{Comparing AnarchistHoneybun:bug/cp-device-file (8d3efa4) with main (5287738)}

Summary

✅ 45 untouched
⏩ 1 skipped¹

Footnotes

1. 1 benchmark was skipped, so the baseline result was used instead. If it was deleted from the codebase, click here and archive it to remove it from the performance reports. ↩

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/misc/tee (fails in this run but passes in the 'main' branch)

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/misc/tee (fails in this run but passes in the 'main' branch)
Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)