Re: [PATCH] mm/userfaultfd: Support operation on multiple VMAs

[David Hildenbrand <david@redhat.com>]

On 13.02.23 16:04, Muhammad Usama Anjum wrote:
> Hi David,
> 
> Thank you for quick review!
> 
> On 2/13/23 4:44 PM, David Hildenbrand wrote:
>> On 13.02.23 11:43, Muhammad Usama Anjum wrote:
>>> mwriteprotect_range() errors out if [start, end) doesn't fall in one
>>> VMA. We are facing a use case where multiple VMAs are present in one
>>> range of interest. For example, the following pseudocode reproduces the
>>> error which we are trying to fix:
>>>
>>> - Allocate memory of size 16 pages with PROT_NONE with mmap
>>> - Register userfaultfd
>>> - Change protection of the first half (1 to 8 pages) of memory to
>>>     PROT_READ | PROT_WRITE. This breaks the memory area in two VMAs.
>>> - Now UFFDIO_WRITEPROTECT_MODE_WP on the whole memory of 16 pages errors
>>>     out.
>>>
>>> This is a simple use case where user may or may not know if the memory
>>> area has been divided into multiple VMAs.
>>>
>>> Reported-by: Paul Gofman <pgofman@codeweavers.com>
>>> Signed-off-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
>>> ---
>>>    mm/userfaultfd.c | 36 +++++++++++++++++++-----------------
>>>    1 file changed, 19 insertions(+), 17 deletions(-)
>>>
>>> diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
>>> index 65ad172add27..46e0a014af68 100644
>>> --- a/mm/userfaultfd.c
>>> +++ b/mm/userfaultfd.c
>>> @@ -738,9 +738,11 @@ int mwriteprotect_range(struct mm_struct *dst_mm,
>>> unsigned long start,
>>>                unsigned long len, bool enable_wp,
>>>                atomic_t *mmap_changing)
>>>    {
>>> +    unsigned long end = start + len;
>>>        struct vm_area_struct *dst_vma;
>>>        unsigned long page_mask;
>>>        int err;
>>> +    VMA_ITERATOR(vmi, dst_mm, start);
>>>          /*
>>>         * Sanitize the command parameters:
>>> @@ -762,26 +764,26 @@ int mwriteprotect_range(struct mm_struct *dst_mm,
>>> unsigned long start,
>>>        if (mmap_changing && atomic_read(mmap_changing))
>>>            goto out_unlock;
>>>    -    err = -ENOENT;
>>> -    dst_vma = find_dst_vma(dst_mm, start, len);
>>> -
>>> -    if (!dst_vma)
>>> -        goto out_unlock;
>>> -    if (!userfaultfd_wp(dst_vma))
>>> -        goto out_unlock;
>>> -    if (!vma_can_userfault(dst_vma, dst_vma->vm_flags))
>>> -        goto out_unlock;
>>> +    for_each_vma_range(vmi, dst_vma, end) {
>>> +        err = -ENOENT;
>>>    -    if (is_vm_hugetlb_page(dst_vma)) {
>>> -        err = -EINVAL;
>>> -        page_mask = vma_kernel_pagesize(dst_vma) - 1;
>>> -        if ((start & page_mask) || (len & page_mask))
>>> -            goto out_unlock;
>>> -    }
>>> +        if (!dst_vma->vm_userfaultfd_ctx.ctx)
>>> +            break;
>>> +        if (!userfaultfd_wp(dst_vma))
>>> +            break;
>>> +        if (!vma_can_userfault(dst_vma, dst_vma->vm_flags))
>>> +            break;
>>>    -    uffd_wp_range(dst_mm, dst_vma, start, len, enable_wp);
>>> +        if (is_vm_hugetlb_page(dst_vma)) {
>>> +            err = -EINVAL;
>>> +            page_mask = vma_kernel_pagesize(dst_vma) - 1;
>>> +            if ((start & page_mask) || (len & page_mask))
>>> +                break;
>>> +        }
>>>    -    err = 0;
>>> +        uffd_wp_range(dst_mm, dst_vma, start, len, enable_wp);
>>
>> I suspect you should be adjusting the range to only cover that specific VMA
>> here.
> Sorry, you are right. I don't know why it is still working with the
> blunder. Will send a v2.

Maybe worth adding some sanity checks (VM_WARN_ONCE()) in there (e.g., 
change_protection()) to catch that.

-- 
Thanks,

David / dhildenb