Re: open()ing a directory without O_DIRECTORY

[Aymeric Vincent <aymericvincent%free.fr@localhost>]

Christos Zoulas <christos%zoulas.com@localhost> writes:

> No I don't need to do anything special I can open the device file as root.

You will still be able to do that for directories even if a normal user
can't read() a directory anymore.

> And this is not the reason that
> directory operations moved to the kernel. It is the same reason that
> mkdir(2) became a system call: atomicity/safety.

The reason is that otherwise you could create cycles in the tree
structure, point to inexistant inodes, and so on. Coherency has to be
maintained in the metadata.

>> That's easily fixed in getdents() if it's not already the case.
>
> getdents(2) has nothing to do with open(2). It takes an fd.

Well, in order for my suggestion to work, read() would have to fail on a
fd opened with O_DIRECTORY, and getdents() would become the only
checkpoint for this data.

> You are worried then about data disclosure, and instead of fixing it at
> the place where it happens (the filesystem code should zero the
> dirent data when it removes the last link to one), you just want to hide it.

Agreed. The same as when you unlink a file, it doesn't destroy its
contents, just renders it inaccessible to normal users. My approach
forbids disclosure to standard users in a filesystem independent way,
whereas cleaning upon unlink() requires support in each filesystem. Of
course I'm not opposed to the latter either.

 Aymeric