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;
MIME-Version: 1.0
References: <20211019134204.3382645-1-agruenba@redhat.com> <CAHk-=wh0_3y5s7-G74U0Pcjm7Y_yHB608NYrQSvgogVNBxsWSQ@mail.gmail.com>
 <YXBFqD9WVuU8awIv@arm.com> <CAHk-=wgv=KPZBJGnx_O5-7hhST8CL9BN4wJwtVuycjhv_1MmvQ@mail.gmail.com>
 <YXCbv5gdfEEtAYo8@arm.com>
In-Reply-To: <YXCbv5gdfEEtAYo8@arm.com>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Wed, 20 Oct 2021 20:19:40 -1000
Message-ID: <CAHk-=wgP058PNY8eoWW=5uRMox-PuesDMrLsrCWPS+xXhzbQxQ@mail.gmail.com>
Subject: Re: [PATCH v8 00/17] gfs2: Fix mmap + page fault deadlocks
To:     Catalin Marinas <catalin.marinas@arm.com>
Cc:     Andreas Gruenbacher <agruenba@redhat.com>,
        Paul Mackerras <paulus@ozlabs.org>,
        Alexander Viro <viro@zeniv.linux.org.uk>,
        Christoph Hellwig <hch@infradead.org>,
        "Darrick J. Wong" <djwong@kernel.org>, Jan Kara <jack@suse.cz>,
        Matthew Wilcox <willy@infradead.org>,
        cluster-devel <cluster-devel@redhat.com>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        ocfs2-devel@oss.oracle.com, kvm-ppc@vger.kernel.org,
        linux-btrfs <linux-btrfs@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Oct 20, 2021 at 12:44 PM Catalin Marinas
<catalin.marinas@arm.com> wrote:
>
> However, with MTE doing both get_user() every 16 bytes and
> gup can get pretty expensive.

So I really think that anything that is performance-critical had
better only do the "fault_in_write()" code path in the cold error path
where you took a page fault.

IOW, the loop structure should be


     repeat:
                take_locks();
                pagefault_disable();
                ret = do_things_with_locks();
                pagefault_enable();
                release_locks();

                // Failure path?
                if (unlikely(ret == -EFAULT)) {
                        if (fault_in_writable(..))
                                goto repeat;
                        return -EFAULT;
                }

rather than have it be some kind of "first do fault_in_writable(),
then do the work" model.

So I wouldn't worry too much about the performance concerns. It simply
shouldn't be a common or hot path.

And yes, I've seen code that does that "fault_in_xyz()" before the
critical operation that cannot take page faults, and does it
unconditionally.

But then it isn't the "fault_in_xyz()" that should be blamed if it is
slow, but the caller that does things the wrong way around.

            Linus