Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 18 May 2023 16:00:57 +0200
From:   Christian Brauner <brauner@kernel.org>
To:     Karel Zak <kzak@redhat.com>
Cc:     linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        util-linux@vger.kernel.org
Subject: Re: [ANNOUNCE] util-linux v2.39
Message-ID: <20230518-gejagt-vervollkommnen-01451a4325b9@brauner>
References: <20230517112242.3rubpxvxhzsc4kt2@ws.net.home>
 <20230517-mahnmal-setzen-37937c35cf78@brauner>
 <20230518102316.f6s6v6xxnicx646r@ws.net.home>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20230518102316.f6s6v6xxnicx646r@ws.net.home>
Precedence: bulk

On Thu, May 18, 2023 at 12:23:16PM +0200, Karel Zak wrote:
> On Wed, May 17, 2023 at 03:48:54PM +0200, Christian Brauner wrote:
> > This is a very exciting release! There's good reason for us to be happy
> > imho. This is the first release of util-linux with comprehensive support
> > for the new mount api which is very exciting.
> 
> We will see how many things in libmount and kernel are not ready ;-)

Yeah, I think we will indeed...
So, I think we need to port overlayfs to the new mount api because of
https://github.com/util-linux/util-linux/issues/1992#issuecomment-1486475153

> 
> > A part of that is of course the support for idmapped mounts and the
> > ability to recursively change mount properties, i.e., idempotently
> > change the mount properties of a whole mount tree.
> > 
> > It's also great to see support for disk sequence numbers via the
> > BLKGETDISKSEQ ioctl and the port to util-linux to rely on
> 
> BLKGETDISKSEQ is supported in the blockdev command only.
> 
> Lennart has also idea to support it in libmount to verify devices
> before the filesystem is attached to VFS. 
> 
> https://github.com/util-linux/util-linux/issues/1786
> 
> That's something we can work on in the next release.

Yeah, I remember discussing that. Though that doesn't eliminate all
races as we discussed on the thread and I plan to implement what I said in
https://github.com/util-linux/util-linux/issues/1786#issuecomment-1410515391
rather soon and talked about at LSFMM last week.

> 
> > statx(AT_STATX_DONT_SYNC|AT_NO_AUTOMOUNT) to avoid tripping over
> > automounts or hung network filesystems as we just recently discussed
> > this!
> > 
> > Thanks for working on this and hopefully we can add the missing pieces
> > of the new mount api in the coming months!
> 
> I would like to make the v2.40 development cycle shorter. The v2.39
> cycle was excessively long and large.

Yeah, but that was kinda expected given the switch to the new mount api.
I mean, after I did the initial support to get idmapped mounts working
in there you still had to port all the rest of libmount...