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:   Mon, 14 Sep 2020 16:27:24 +0200
From:   Oleg Nesterov <oleg@redhat.com>
To:     Peter Xu <peterx@redhat.com>
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        "Maya B . Gokhale" <gokhale2@llnl.gov>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Yang Shi <yang.shi@linux.alibaba.com>,
        Marty Mcfadden <mcfadden8@llnl.gov>,
        Kirill Shutemov <kirill@shutemov.name>,
        Jann Horn <jannh@google.com>, Jan Kara <jack@suse.cz>,
        Kirill Tkhai <ktkhai@virtuozzo.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Christoph Hellwig <hch@lst.de>,
        Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH 3/4] mm/gup: Remove enfornced COW mechanism
Message-ID: <20200914142722.GA21906@redhat.com>
References: <20200821234958.7896-1-peterx@redhat.com>
 <20200821234958.7896-4-peterx@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200821234958.7896-4-peterx@redhat.com>
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 08/21, Peter Xu wrote:
>
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -381,22 +381,13 @@ static int follow_pfn_pte(struct vm_area_struct *vma, unsigned long address,
>  }
>  
>  /*
> - * FOLL_FORCE or a forced COW break can write even to unwritable pte's,
> - * but only after we've gone through a COW cycle and they are dirty.
> + * FOLL_FORCE can write to even unwritable pte's, but only
> + * after we've gone through a COW cycle and they are dirty.
>   */
>  static inline bool can_follow_write_pte(pte_t pte, unsigned int flags)
>  {
> -	return pte_write(pte) || ((flags & FOLL_COW) && pte_dirty(pte));
> -}
> -
> -/*
> - * A (separate) COW fault might break the page the other way and
> - * get_user_pages() would return the page from what is now the wrong
> - * VM. So we need to force a COW break at GUP time even for reads.
> - */
> -static inline bool should_force_cow_break(struct vm_area_struct *vma, unsigned int flags)
> -{
> -	return is_cow_mapping(vma->vm_flags) && (flags & (FOLL_GET | FOLL_PIN));
> +	return pte_write(pte) ||
> +		((flags & FOLL_FORCE) && (flags & FOLL_COW) && pte_dirty(pte));

Do we really need to add the FOLL_FORCE check back?

Afaics, FOLL_COW is only possible if FOLL_FORCE was set.

Oleg.