Received-SPF: pass (google.com: domain of linux-kernel+bounces-186543-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Wed, 22 May 2024 18:13:57 +0100
From: Matthew Wilcox <willy@infradead.org>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: kernel test robot <oliver.sang@intel.com>,
	Yafang Shao <laoar.shao@gmail.com>, oe-lkp@lists.linux.dev,
	lkp@intel.com, Al Viro <viro@zeniv.linux.org.uk>,
	Christian Brauner <brauner@kernel.org>, Jan Kara <jack@suse.cz>,
	Waiman Long <longman@redhat.com>, Wangkai <wangkai86@huawei.com>,
	Colin Walters <walters@verbum.org>, linux-fsdevel@vger.kernel.org,
	ying.huang@intel.com, feng.tang@intel.com, fengwei.yin@intel.com,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] vfs: Delete the associated dentry when deleting a file
Message-ID: <Zk4n1eXLXkbKWFs2@casper.infradead.org>
References: <20240515091727.22034-1-laoar.shao@gmail.com>
 <202405221518.ecea2810-oliver.sang@intel.com>
 <CAHk-=wg2jGRLWhT1-Od3A74Cr4cSM9H+UhOD46b3_-mAfyf1gw@mail.gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAHk-=wg2jGRLWhT1-Od3A74Cr4cSM9H+UhOD46b3_-mAfyf1gw@mail.gmail.com>

On Wed, May 22, 2024 at 09:00:03AM -0700, Linus Torvalds wrote:
> Of course, if you do billions of lookups of different files that do
> not exist in the same directory, I suspect you just have yourself to
> blame, so the "lots of negative lookups" load doesn't sound
> particularly realistic.

Oh no.  We have real customers that this hits and it's not even stupid.

Every time an "event" happens, they look up something like a hash in three
different directories (like $PATH or multiple -I flags to the compiler).
Order is important, so they can't just look it up in the directory that
it's most likely to exist in.  It usually fails to exist in directory A
and B, so we create dentries that say it doesn't.  And those dentries are
literally never referenced again.  Then directory C has the file they're
looking for (or it doesn't and it gets created because the process has
write access to C and not A or B).

plan9 handles this so much better because it has that union-mount stuff
instead of search paths.  So it creates one dentry that tells it which of
those directories it actually exists in.  But we're stuck with unix-style
search paths, so life is pain.