Received-SPF: pass (google.com: domain of linux-kernel+bounces-76562-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <2uvhm6gweyl7iyyp2xpfryvcu2g3padagaeqcbiavjyiis6prl@yjm725bizncq>
 <CAJfpeguBzbhdcknLG4CjFr12_PdGo460FSRONzsYBKmT9uaSMA@mail.gmail.com>
 <20240221210811.GA1161565@perftesting> <CAJfpegucM5R_pi_EeDkg9yPNTj_esWYrFd6vG178_asram0=Ew@mail.gmail.com>
 <w534uujga5pqcbhbc5wad7bdt5lchxu6gcmwvkg6tdnkhnkujs@wjqrhv5uqxyx>
 <20240222110138.ckai4sxiin3a74ku@quack3> <u4btyvpsohnf77r5rm43tz426u3advjc4goea4obt2wtv6xyog@7bukhgoyumed>
 <20240222114417.wpcdkgsed7wklv3h@quack3> <2tsfxaf2blhcxlkfcagfavz3mnuga3qsjgpytbstvykmcq2prj@icc7vub55i3p>
In-Reply-To: <2tsfxaf2blhcxlkfcagfavz3mnuga3qsjgpytbstvykmcq2prj@icc7vub55i3p>
From: Miklos Szeredi <miklos@szeredi.hu>
Date: Thu, 22 Feb 2024 14:10:40 +0100
Message-ID: <CAJfpegsinKngjHmZPxXvGULusEw359FhBkVh4A9O_+pqccjUzw@mail.gmail.com>
Subject: Re: [Lsf-pc] [LSF TOPIC] statx extensions for subvol/snapshot
 filesystems & more
To: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Jan Kara <jack@suse.cz>, Josef Bacik <josef@toxicpanda.com>, linux-kernel@vger.kernel.org, 
	linux-bcachefs@vger.kernel.org, linux-fsdevel@vger.kernel.org, 
	lsf-pc@lists.linux-foundation.org, linux-btrfs@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"

On Thu, 22 Feb 2024 at 12:55, Kent Overstreet <kent.overstreet@linux.dev> wrote:
>
> On Thu, Feb 22, 2024 at 12:44:17PM +0100, Jan Kara wrote:
> > On Thu 22-02-24 06:27:14, Kent Overstreet wrote:

> > > My intent is to make a real effort towards getting better interfaces
> > > going, prod those maintainers, _then_ look at adding those hacks (that
> > > will necessarily be short term solutions since 64 bits is already
> > > looking cramped).
> >
> > OK, fine by me :) So one thing is still not quite clear to me - how do you
> > expect the INO_NOT_UNIQUE flag to be used by these apps? Do you expect them
> > to use st_dev + st_ino by default and fall back to fsid + fhandle only when
> > INO_NOT_UNIQUE is set?
>
> Shouldn't matter. If they care about performance and they're in some
> strange situation where the syscal overhead matters,

If it's expensive, then just make the overhead smaller (by adding fh
and uuid to statx(2), for example).

Using st_ino is also racy in some filesystems, due to the fact that
the ino can be reused.  If userspace is converted, it should be
converted properly, there's just no excuse to add conditional code
like that, which makes things more complex and less reliable.

Thanks,
Miklos