Re: file metadata via fs API (was: [GIT PULL] Filesystem Information)

From: Al Viro <viro@zeniv.linux.org.uk>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: linux-fsdevel@vger.kernel.org,
	David Howells <dhowells@redhat.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Karel Zak <kzak@redhat.com>, Jeff Layton <jlayton@redhat.com>,
	Miklos Szeredi <mszeredi@redhat.com>,
	Nicolas Dichtel <nicolas.dichtel@6wind.com>,
	Christian Brauner <christian@brauner.io>,
	Lennart Poettering <lennart@poettering.net>,
	Linux API <linux-api@vger.kernel.org>,
	Ian Kent <raven@themaw.net>,
	LSM <linux-security-module@vger.kernel.org>,
	linux-kernel@vger.kernel.org
Subject: Re: file metadata via fs API (was: [GIT PULL] Filesystem Information)
Date: Tue, 11 Aug 2020 15:31:07 +0100	[thread overview]
Message-ID: <20200811143107.GI1236603@ZenIV.linux.org.uk> (raw)
In-Reply-To: <CAJfpegsNj55pTXe97qE_i-=zFwca1=2W_NqFdn=rHqc_AJjr8Q@mail.gmail.com>

On Tue, Aug 11, 2020 at 04:22:19PM +0200, Miklos Szeredi wrote:
> On Tue, Aug 11, 2020 at 4:08 PM Al Viro <viro@zeniv.linux.org.uk> wrote:
> >
> > On Tue, Aug 11, 2020 at 03:54:19PM +0200, Miklos Szeredi wrote:
> > > On Wed, Aug 05, 2020 at 10:24:23AM +0200, Miklos Szeredi wrote:
> > > > On Tue, Aug 4, 2020 at 4:36 PM Miklos Szeredi <miklos@szeredi.hu> wrote:
> > > >
> > > > > I think we already lost that with the xattr API, that should have been
> > > > > done in a way that fits this philosophy.  But given that we  have "/"
> > > > > as the only special purpose char in filenames, and even repetitions
> > > > > are allowed, it's hard to think of a good way to do that.  Pity.
> > > >
> > > > One way this could be solved is to allow opting into an alternative
> > > > path resolution mode.
> > > >
> > > > E.g.
> > > >   openat(AT_FDCWD, "foo/bar//mnt/info", O_RDONLY | O_ALT);
> > >
> > > Proof of concept patch and test program below.
> > >
> > > Opted for triple slash in the hope that just maybe we could add a global
> > > /proc/sys/fs/resolve_alt knob to optionally turn on alternative (non-POSIX) path
> > > resolution without breaking too many things.  Will try that later...
> > >
> > > Comments?
> >
> > Hell, NO.  This is unspeakably tasteless.  And full of lovely corner cases wrt
> > symlink bodies, etc.
> 
> It's disabled inside symlink body resolution.
> 
> Rules are simple:
> 
>  - strip off trailing part after first instance of ///
>  - perform path lookup as normal
>  - resolve meta path after /// on result of normal lookup

... and interpolation of relative symlink body into the pathname does change
behaviour now, *including* the cases when said symlink body does not contain
that triple-X^Hslash garbage.  Wonderful...