Re: [RFC PATCH] fs: introduce mkdirat2 syscall for atomic mkdir

From: Al Viro <viro@zeniv.linux.org.uk>
To: Drew DeVault <sir@cmpwn.com>
Cc: linux-fsdevel@vger.kernel.org, linux-api@vger.kernel.org,
	Aleksa Sarai <cyphar@cyphar.com>
Subject: Re: [RFC PATCH] fs: introduce mkdirat2 syscall for atomic mkdir
Date: Sun, 28 Feb 2021 02:13:38 +0000	[thread overview]
Message-ID: <YDr8UihFQ3M469x8@zeniv-ca.linux.org.uk> (raw)
In-Reply-To: <20210228002500.11483-1-sir@cmpwn.com>

On Sat, Feb 27, 2021 at 07:25:00PM -0500, Drew DeVault wrote:

> This introduces mkdirat2, along with the requisite flag argument, which
> presently accepts the same flags as open - allowing the caller to
> specify, say, O_CLOEXEC - and leaving us room to expand the next time an
> unforeseeable addition to mkdir is called for. Otherwise, it behaves
> identically to mkdirat, but returns an open file descriptor for the new
> directory.

No to the ABI part; "on error it returns -E..., on success - 0 or
a non-negative number representing a file descriptor (zero also possible,
but unlikely)" is bloody awful as calling conventions go, especially
since the case when 0 happens to be a descriptor is not going to get
a lot of testing on the userland side.

Don't mix "return an error or descriptor" with "return an error or 0".
It's going to end up a regular source of userland bugs.