Re: [patch] mmap: feed back correct prev vma when finding vma

[Hugh Dickins]

On Thu, 9 Aug 2012, Hillf Danton wrote:
> After walking rb tree, if vma is determined, prev vma has to be determined
> based on vma; and rb_prev should be considered only if no vma determined.

Why?  Because you think more code is better code?  I disagree.

If you have seen a bug here, please tell how to reproduce it.

I have not heard of a bug here: I think you're saying, if the rbtree
were inconsistent with the vma list, then you think it would be a good
idea to believe the vma list instead of the rbtree where there's a choice.

But the rbtree had better not be inconsistent with the vma list.

Hugh

> 
> Signed-off-by: Hillf Danton <dhillf@xxxxxxxxx>
> ---
> 
> --- a/mm/mmap.c	Fri Aug  3 07:38:10 2012
> +++ b/mm/mmap.c	Mon Aug  6 20:10:18 2012
> @@ -385,9 +385,13 @@ find_vma_prepare(struct mm_struct *mm, u
>  		}
>  	}
> 
> -	*pprev = NULL;
> -	if (rb_prev)
> -		*pprev = rb_entry(rb_prev, struct vm_area_struct, vm_rb);
> +	if (vma) {
> +		*pprev = vma->vm_prev;
> +	} else {
> +		*pprev = NULL;
> +		if (rb_prev)
> +			*pprev = rb_entry(rb_prev, struct vm_area_struct, vm_rb);
> +	}
>  	*rb_link = __rb_link;
>  	*rb_parent = __rb_parent;
>  	return vma;
> --
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/