Message-ID: <c239e126-9b75-9a81-6142-a8e518e769e2@collabora.com>
Date:   Thu, 16 Feb 2023 14:48:51 +0500
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.7.1
Cc:     Muhammad Usama Anjum <usama.anjum@collabora.com>,
        kernel@collabora.com, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4 2/2] mm/userfaultfd: add VM_WARN_ONCE()
Content-Language: en-US
To:     David Hildenbrand <david@redhat.com>, peterx@redhat.com,
        Andrew Morton <akpm@linux-foundation.org>
References: <20230216091656.2045471-1-usama.anjum@collabora.com>
 <20230216091656.2045471-2-usama.anjum@collabora.com>
 <4d29a9ec-027d-06b9-8543-87d386e58c94@redhat.com>
From:   Muhammad Usama Anjum <usama.anjum@collabora.com>
In-Reply-To: <4d29a9ec-027d-06b9-8543-87d386e58c94@redhat.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 2/16/23 2:24 PM, David Hildenbrand wrote:
> On 16.02.23 10:16, Muhammad Usama Anjum wrote:
>> Add VM_WARN_ONCE() to uffd_wp_range() to detect range (start, len) abuse.
>>
>> Signed-off-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
>> ---
>>   mm/userfaultfd.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
>> index 77c5839e591c..d89ed44d2668 100644
>> --- a/mm/userfaultfd.c
>> +++ b/mm/userfaultfd.c
>> @@ -717,6 +717,8 @@ long uffd_wp_range(struct mm_struct *dst_mm, struct
>> vm_area_struct *dst_vma,
>>       struct mmu_gather tlb;
>>       long ret;
>>   +    VM_WARN_ONCE(start < dst_vma->vm_start || start + len >
>> dst_vma->vm_end,
>> +             "The address range exceeds VMA boundary.\n");
> 
> VM_WARN_ON_ONCE is sufficient (sorry for spelling out the wrong variant
> earlier).
Will do in the next version. Thanks.

> 
> These kinds of bugs are expected to be found early during testing, still it
> might make sense to implement a backup path
> 
> if (WARN_ON_ONCE(...))
>     return -EINVAL;
> 
> But we can't use VM_WARN_ON_ONCE here, so we can't compile it out anymore
> ... so I guess a simple VM_WARN_ON_ONCE() is sufficient.
> 

-- 
BR,
Muhammad Usama Anjum