Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <91e18a2f-c93d-00b8-7c1b-6d8493c3b2d5@redhat.com>
Date:   Tue, 9 Aug 2022 22:07:44 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.11.0
Subject: Re: [PATCH v1] mm/gup: fix FOLL_FORCE COW security issue and remove
 FOLL_COW
Content-Language: en-US
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        stable@vger.kernel.org, Andrew Morton <akpm@linux-foundation.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Axel Rasmussen <axelrasmussen@google.com>,
        Peter Xu <peterx@redhat.com>, Hugh Dickins <hughd@google.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Matthew Wilcox <willy@infradead.org>,
        Vlastimil Babka <vbabka@suse.cz>,
        John Hubbard <jhubbard@nvidia.com>,
        Jason Gunthorpe <jgg@nvidia.com>
References: <20220808073232.8808-1-david@redhat.com>
 <CAHk-=wgsDOz5MfYYS9mE7PvFn4kLhTFdBwXvN6HCEsw1kvJnRQ@mail.gmail.com>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
In-Reply-To: <CAHk-=wgsDOz5MfYYS9mE7PvFn4kLhTFdBwXvN6HCEsw1kvJnRQ@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 09.08.22 22:00, Linus Torvalds wrote:
> On Mon, Aug 8, 2022 at 12:32 AM David Hildenbrand <david@redhat.com> wrote:
>>
> 
> So I've read through the patch several times, and it seems fine, but
> this function (and the pmd version of it) just read oddly to me.
> 
>> +static inline bool can_follow_write_pte(pte_t pte, struct page *page,
>> +                                       struct vm_area_struct *vma,
>> +                                       unsigned int flags)
>> +{
>> +       if (pte_write(pte))
>> +               return true;
>> +       if (!(flags & FOLL_FORCE))
>> +               return false;
>> +
>> +       /*
>> +        * See check_vma_flags(): only COW mappings need that special
>> +        * "force" handling when they lack VM_WRITE.
>> +        */
>> +       if (vma->vm_flags & VM_WRITE)
>> +               return false;
>> +       VM_BUG_ON(!is_cow_mapping(vma->vm_flags));
> 
> So apart from the VM_BUG_ON(), this code just looks really strange -
> even despite the comment. Just conceptually, the whole "if it's
> writable, return that you cannot follow it for a write" just looks so
> very very strange.
> 
> That doesn't make the code _wrong_, but considering how many times
> this has had subtle bugs, let's not write code that looks strange.
> 
> So I would suggest that to protect against future bugs, we try to make
> it be fairly clear and straightforward, and maybe even a bit overly
> protective.
> 
> For example, let's kill the "shared mapping that you don't have write
> permissions to" very explicitly and without any subtle code at all.
> The vm_flags tests are cheap and easy, and we could very easily just
> add some core ones to make any mistakes much less critical.
> 
> Now, making that 'is_cow_mapping()' check explicit at the very top of
> this would already go a long way:
> 
>         /* FOLL_FORCE for writability only affects COW mappings */
>         if (!is_cow_mapping(vma->vm_flags))
>                 return false;
> 
> but I'd actually go even further: in this case that "is_cow_mapping()"
> helper to some degree actually hides what is going on.
> 
> So I'd actually prefer for that function to be written something like
> 
>         /* If the pte is writable, we can write to the page */
>         if (pte_write(pte))
>                 return true;
> 
>         /* Maybe FOLL_FORCE is set to override it? */
>         if (flags & FOLL_FORCE)
>                 return false;
> 
>         /* But FOLL_FORCE has no effect on shared mappings */
>         if (vma->vm_flags & MAP_SHARED)
>                 return false;

I'd actually rather check for MAP_MAYSHARE here, which is even stronger.
Thoughts?

-- 
Thanks,

David / dhildenb