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:   Sat, 1 May 2021 14:19:40 -0700
From:   Michel Lespinasse <michel@lespinasse.org>
To:     Theodore Ts'o <tytso@mit.edu>
Cc:     Michel Lespinasse <michel@lespinasse.org>,
        Linux-MM <linux-mm@kvack.org>,
        Linux-Kernel <linux-kernel@vger.kernel.org>,
        Laurent Dufour <ldufour@linux.ibm.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Michal Hocko <mhocko@suse.com>,
        Matthew Wilcox <willy@infradead.org>,
        Rik van Riel <riel@surriel.com>,
        Paul McKenney <paulmck@kernel.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Suren Baghdasaryan <surenb@google.com>,
        Joel Fernandes <joelaf@google.com>,
        Andy Lutomirski <luto@kernel.org>
Subject: Re: [PATCH 00/29] Speculative page faults (anon vmas only)
Message-ID: <20210501211940.GA11658@lespinasse.org>
References: <20210430195232.30491-1-michel@lespinasse.org>
 <YI2yZ+10MGrlFDBg@mit.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YI2yZ+10MGrlFDBg@mit.edu>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

Hi Ted,

On Sat, May 01, 2021 at 03:56:23PM -0400, Theodore Ts'o wrote:
> I tried running xfstests against the spf branch, and I've noticed it's
> causing regression for generic/619.  It's failing due to a umount
> failure due to a busy mount point:
> 
> QA output created by 619
> umount: /vdc: target is busy.
> 
> I haven't had a chance to investigate, but I thought I should let you know.

Thanks for the report. I think the issue is likely caused by commit
06adfeb8150d "mm: rcu safe vma->vm_file freeing", which will defer
fput on mapped files for one rcu grace period. I expect adding
synchronize_rcu to the proper place (I'm not sure exactly where though)
when unmounting file systems would be a workable fix.

Note though - at this point I am only submitting the anon part of the
patchset for inclusion. That is, the v5.12-spf-anon branch, rather
than the v5.12-spf branch which has the additional / less mature
changes for file mapped vma faults.

--
Michel "walken" Lespinasse