Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 8 May 2020 07:56:59 -0700
From:   Matthew Wilcox <willy@infradead.org>
To:     Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Cc:     linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        linux-mm@kvack.org, Alexander Viro <viro@zeniv.linux.org.uk>,
        Waiman Long <longman@redhat.com>
Subject: Re: [PATCH RFC 8/8] dcache: prevent flooding with negative dentries
Message-ID: <20200508145659.GQ16070@bombadil.infradead.org>
References: <158893941613.200862.4094521350329937435.stgit@buzz>
 <158894061332.200862.9812452563558764287.stgit@buzz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <158894061332.200862.9812452563558764287.stgit@buzz>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, May 08, 2020 at 03:23:33PM +0300, Konstantin Khlebnikov wrote:
> This patch implements heuristic which detects such scenarios and prevents
> unbounded growth of completely unneeded negative dentries. It keeps up to
> three latest negative dentry in each bucket unless they were referenced.
> 
> At first dput of negative dentry when it swept to the tail of siblings
> we'll also clear it's reference flag and look at next dentries in chain.
> Then kill third in series of negative, unused and unreferenced denries.
> 
> This way each hash bucket will preserve three negative dentry to let them
> get reference and survive. Adding positive or used dentry into hash chain
> also protects few recent negative dentries. In result total size of dcache
> asymptotically limited by count of buckets and positive or used dentries.
> 
> This heuristic isn't bulletproof and solves only most practical case.
> It's easy to deceive: just touch same random name twice.

I'm not sure if that's "easy to deceive" ... My concern with limiting
negative dentries is something like a kernel compilation where there
are many (11 for mm/mmap.c, 9 in general) and there will be a lot of
places where <linux/fs.h> does not exist

-isystem /usr/lib/gcc/x86_64-linux-gnu/9/include
-I../arch/x86/include
-I./arch/x86/include/generated
-I../include
-I./include
-I../arch/x86/include/uapi
-I./arch/x86/include/generated/uapi
-I../include/uapi
-I./include/generated/uapi
-I ../mm
-I ./mm

So it'd be good to know that kernel compilation times are unaffected by
this patch.