Re: [PATCH v6 16/24] mm: Protect mm_rb tree with a rwlock

[Matthew Wilcox]

On Fri, Jan 12, 2018 at 06:26:00PM +0100, Laurent Dufour wrote:
> -static void __vma_rb_erase(struct vm_area_struct *vma, struct rb_root *root)
> +static void __vma_rb_erase(struct vm_area_struct *vma, struct mm_struct *mm)
>  {
> +	struct rb_root *root = &mm->mm_rb;
>  	/*
>  	 * Note rb_erase_augmented is a fairly large inline function,
>  	 * so make sure we instantiate it only once with our desired
>  	 * augmented rbtree callbacks.
>  	 */
> +#ifdef CONFIG_SPF
> +	write_lock(&mm->mm_rb_lock);
> +#endif
>  	rb_erase_augmented(&vma->vm_rb, root, &vma_gap_callbacks);
> +#ifdef CONFIG_SPF
> +	write_unlock(&mm->mm_rb_lock); /* wmb */
> +#endif

I can't say I love this.  Have you considered:

#ifdef CONFIG_SPF
#define vma_rb_write_lock(mm)	write_lock(&mm->mm_rb_lock)
#define vma_rb_write_unlock(mm)	write_unlock(&mm->mm_rb_lock)
#else
#define vma_rb_write_lock(mm)	do { } while (0)
#define vma_rb_write_unlock(mm)	do { } while (0)
#endif

Also, SPF is kind of uninformative.  CONFIG_MM_SPF might be better?
Or perhaps even CONFIG_SPECULATIVE_PAGE_FAULT, just to make it really
painful to do these one-liner ifdefs that make the code so hard to read.