Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
MIME-Version: 1.0
References: <20220322192712.709170-1-mszeredi@redhat.com> <20220323114215.pfrxy2b6vsvqig6a@wittgenstein>
 <CAJfpegsCKEx41KA1S2QJ9gX9BEBG4_d8igA0DT66GFH2ZanspA@mail.gmail.com>
 <YjudB7XARLlRtBiR@mit.edu> <CAJfpegtiRx6jRFUuPeXDxwJpBhYn0ekKkwYbGowUehGZkqVmAw@mail.gmail.com>
 <20220325084646.7g6oto2ce3vou54x@ws.net.home>
In-Reply-To: <20220325084646.7g6oto2ce3vou54x@ws.net.home>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 25 Mar 2022 11:40:15 -0700
Message-ID: <CAHk-=wibqjNiLgnsL2LcTDKqnKyQ17rhHJv-qkHY1w93LJuXjA@mail.gmail.com>
Subject: Re: [RFC PATCH] getvalues(2) prototype
To:     Karel Zak <kzak@redhat.com>
Cc:     Miklos Szeredi <miklos@szeredi.hu>,
        "Theodore Ts'o" <tytso@mit.edu>,
        Christian Brauner <brauner@kernel.org>,
        Miklos Szeredi <mszeredi@redhat.com>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux API <linux-api@vger.kernel.org>,
        linux-man <linux-man@vger.kernel.org>,
        LSM <linux-security-module@vger.kernel.org>,
        Ian Kent <raven@themaw.net>,
        David Howells <dhowells@redhat.com>,
        Al Viro <viro@zeniv.linux.org.uk>,
        Christian Brauner <christian@brauner.io>,
        Amir Goldstein <amir73il@gmail.com>,
        James Bottomley <James.Bottomley@hansenpartnership.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Mar 25, 2022 at 1:46 AM Karel Zak <kzak@redhat.com> wrote:
>
> I can imagine something like getvalues(2) in lsblk (based on /sys) or
> in lsfd (based on /proc; lsof replacement).

I really would be very hesitant to add new interfaces for completely
specialty purposes.

As others have mentioned, this has been tried for much more
fundamental reasons (that whole "open-and-read" thing), and it hasn't
been an obvious improvement.

It *could* possibly be an improvement if it would allow us to take
advantage of special server-side operations in a networked filesystem
(ie like the whole "copy_file_range" kinds of interfaces that allow
server-side things) where you need to transfer less data, or need
fewer back-and-forth operations.

And even that is clearly questionable, with some of those network file
interfaces basically not having been used in real world situations in
the past, so..

(Also, with "copy_file_range" we not only had others actively doing
it, but the wins were "several orders of manitude", so even if it was
fairly rare, it was _so_ big that it was worth doing anyway).

With the "open-and-read" thing, the wins aren't that enormous.

And getvalues() isn't even that. It's literally a speciality interface
for a very special thing. Yes, I'm sure it avoids several system
calls. Yes, I'm sure it avoids parsing strings etc. But I really don't
think this is something we want to do unless people can show enormous
and real-world examples of where it makes such a huge difference that
we absolutely have to do it.

             Linus