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:   Fri, 22 Jul 2022 09:08:32 -0700
From:   Luis Chamberlain <mcgrof@kernel.org>
To:     Muchun Song <songmuchun@bytedance.com>
Cc:     Zhang Yuchen <zhangyuchen.lcr@bytedance.com>,
        "Eric W. Biederman" <ebiederm@xmission.com>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        David Howells <dhowells@redhat.com>,
        Deepa Dinamani <deepa.kernel@gmail.com>,
        Christoph Hellwig <hch@lst.de>, linux-api@vger.kernel.org,
        Kees Cook <keescook@chromium.org>,
        Iurii Zaikin <yzaikin@google.com>,
        LKML <linux-kernel@vger.kernel.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>
Subject: Re: [RFC] proc: fix create timestamp of files in proc
Message-ID: <YtrLgCttVa7IP7B6@bombadil.infradead.org>
References: <20220721081617.36103-1-zhangyuchen.lcr@bytedance.com>
 <Ytl772fRS74eIneC@bombadil.infradead.org>
 <CAMZfGtXjK2BgpwTOGsWdKs9-3i0X9ohdbXJk=0DAmpEKUymS5w@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAMZfGtXjK2BgpwTOGsWdKs9-3i0X9ohdbXJk=0DAmpEKUymS5w@mail.gmail.com>
Sender: Luis Chamberlain <mcgrof@infradead.org>
Precedence: bulk

On Fri, Jul 22, 2022 at 11:43:49AM +0800, Muchun Song wrote:
> On Fri, Jul 22, 2022 at 12:16 AM Luis Chamberlain <mcgrof@kernel.org> wrote:
> > On Thu, Jul 21, 2022 at 04:16:17PM +0800, Zhang Yuchen wrote:
> > > The file timestamp in procfs is the timestamp when the inode was
> > > created. If the inode of a file in procfs is reclaimed, the inode
> > > will be recreated when it is opened again, and the timestamp will
> > > be changed to the time when it was recreated.
> >
> > The commit log above starts off with a report of the directory
> > of a PID. When does the directory of a PID change dates when its
> > respective start_time does not? When does this reclaim happen exactly?
> > Under what situation?
> 
> IMHO, when the system is under memory pressure, then the proc
> inode can be reclaimed, it is also true when we `echo 3 >
> /proc/sys/vm/drop_caches`. After this operation, the proc inode's
> timestamp is changed.

Good point.

> Maybe the users think the timestamp of /proc/$pid directory is equal to
> the start_time of a process, I think it is because of a comment of
> shortage about the meaning of the timestamp of /proc files.

I'll send a documentation enhancement for this. Thanks for helping with
improving the quality of your peer's patches / feedback in the future!

  Luis