[ruby-core:111844] [Ruby master Feature#19347] Add Dir.fchdir

Issue #19347 has been updated by kjtsanaktsidis (KJ Tsanaktsidis).


I have definitely wanted this before FWIW. The main use-case I've had for this feature is writing code to look at things in the `/proc` filesystem on Linux. If a process exits, its pid can be re-used, and so the same path in `/proc` can now refer to a totally different process!

To operate on `/proc` safely, you need to open the the `/proc/#{pid}` directory first, and use `openat(2)` to open individual files from within that directory. If the process exits, the directory suddenly becomes empty, so your operations will begin failing with `ENOENT`. Even if a new process is started with the same pid, the entry in `/proc` has a different inode, so you will never accidentally read a different process's data.

----------------------------------------
Feature #19347: Add Dir.fchdir
https://bugs.ruby-lang.org/issues/19347#change-101255

* Author: jeremyevans0 (Jeremy Evans)
* Status: Open
* Priority: Normal
----------------------------------------
Recently, I was working on a program that passes file descriptors over UNIX sockets (using `send_io`/`recv_io`).  For file/socket/device descriptors, this works fine using `File#reopen`.  However, I found that while Ruby supports `Dir#fileno` to return a directory file descriptor, it cannot use that file descriptor when changing directories.

I worked around this in my program by writing the directory path over the UNIX socket, but this results in a TOCTOU vulnerability in certain cases.  Passing the directory file descriptor would be simpler and avoid the vulnerability.

I've submitted a pull request to implement this method: https://github.com/ruby/ruby/pull/7135



-- 
https://bugs.ruby-lang.org/
 ______________________________________________
 ruby-core mailing list -- [e-mail]
 To unsubscribe send an email to [e-mail]
 ruby-core info -- https://ml.ruby-lang.org/mailman3/postorius/lists/ruby-core.ml.ruby-lang.org/