Re: [PATCH] mm/mmap.c: rb_parent is not necessary in __vma_link_list

[Matthew Wilcox]

On Tue, Aug 13, 2019 at 11:26:56AM +0800, Wei Yang wrote:
> Now we use rb_parent to get next, while this is not necessary.
> 
> When prev is NULL, this means vma should be the first element in the
> list. Then next should be current first one (mm->mmap), no matter
> whether we have parent or not.
> 
> After removing it, the code shows the beauty of symmetry.

Uhh ... did you test this?

> @@ -273,12 +273,8 @@ void __vma_link_list(struct mm_struct *mm, struct 
> vm_area_struct *vma,
>               next = prev->vm_next;
>               prev->vm_next = vma;
>       } else {
> +             next = mm->mmap;
>               mm->mmap = vma;
> -             if (rb_parent)
> -                     next = rb_entry(rb_parent,
> -                                     struct vm_area_struct, vm_rb);
> -             else
> -                     next = NULL;
>       }

The full context is:

        if (prev) {
                next = prev->vm_next;
                prev->vm_next = vma;
        } else {
                mm->mmap = vma;
                if (rb_parent)
                        next = rb_entry(rb_parent,
                                        struct vm_area_struct, vm_rb);
                else
                        next = NULL;
        }

Let's imagine we have a small tree with three ranges in it.

A: 5-7
B: 8-10
C: 11-13

I would imagine an rbtree for this case has B at the top with A
to its left and B to its right.

Now we're going to add range D at 3-4.  'next' should clearly be range A.
It will have NULL prev.  Your code is going to make 'B' next, not A.
Right?