cp: fix recursive socket file copy #8463
Conversation
|
GNU testsuite comparison: |
|
Thanks! MacOS and FreeBSD are unhappy ,-( |
drinkcat
left a comment
drinkcat
left a comment
There was a problem hiding this comment.
Forgot to press submit review...
| #[cfg(unix)] | ||
| pub fn make_socket(path: &Path) -> std::io::Result<()> { | ||
| let name = CString::new(path.to_str().unwrap()).unwrap(); | ||
| let err = unsafe { mknod(name.as_ptr(), S_IFSOCK, 0) }; |
There was a problem hiding this comment.
Should you still set permissions? e.g. S_IFSOCK | 0o666?
There was a problem hiding this comment.
I saw permissions are set from the source file after file creation. So I thought I could skip them here or did I miss something?
There was a problem hiding this comment.
It's tricky, it depends on the exact set of flags passed to cp I believe.
| log_info("mksocket", &full_path); | ||
| unsafe { | ||
| let socket_name: CString = CString::new(full_path).expect("CString creation failed."); | ||
| libc::mknod(socket_name.as_ptr(), libc::S_IFSOCK, 0); |
There was a problem hiding this comment.
Same question about permissions.
|
The issue seems to be that mknod requires super user privileges in freebsd/macos. I will open a new pull request with necessary changes. |
You can just add commits here. No need to create a new PR. Also, it would probably be acceptable to skip those tests on FreeBSD/MacOS if needed. |
|
My bad :( I already opened a pull request #8478 . Had to change the implementation. |
2 participants
Fixes #8378