Re: [PATCH v4] Add /proc/PID/smaps support for DAX

[Dave Hansen]

I'm honestly not understanding what problem this solves.  Could you,
perhaps, do a before and after of smaps with and without this patch?

> +/* page structure behind DAX mappings is NOT compound page
> + * when it's a huge page mappings, so introduce new API to
> + * account for both PMD and PUD mapping.
> + */

Why do they need to be compound?  Why don't we just make them compound
instead of adding all this code which is *just* for DAX?

> +static void smaps_account_dax_huge(struct mem_size_stats *mss,
> +			struct page *page, unsigned long size, bool young, bool dirty)
> +{
> +	int mapcount = page_mapcount(page);
> +
> +	if (PageAnon(page)) {
> +		mss->anonymous += size;
> +		if (!PageSwapBacked(page) && !dirty && !PageDirty(page))
> +			mss->lazyfree += size;
> +	}

How can you have DAX anonymous huge pages?

> +	mss->resident += size;
> +	/* Accumulate the size in pages that have been accessed. */
> +	if (young || page_is_young(page) || PageReferenced(page))
> +		mss->referenced += size;

Isn't this just a copy'n'paste of smaps_account() code?

> +	/*
> +	 * page_count(page) == 1 guarantees the page is mapped exactly once.
> +	 * If any subpage of the compound page mapped with PTE it would elevate
> +	 * page_count().
> +	 */
> +	if (page_count(page) == 1) {
> +		if (dirty || PageDirty(page))
> +			mss->private_dirty += size;
> +		else
> +			mss->private_clean += size;
> +		mss->pss += (u64)size << PSS_SHIFT;
> +		return;
> +	}

PSS makes *zero* sense for DAX.  The "memory" is used whether the
mapping exists or not.

Also, the idea of "private" doesn't really make sense here.

> +	if (mapcount >= 2) {
> +		if (dirty || PageDirty(page))
> +			mss->shared_dirty += size;
> +		else
> +			mss->shared_clean += size;
> +		mss->pss += (size << PSS_SHIFT) / mapcount;
> +	} else {
> +		if (dirty || PageDirty(page))
> +			mss->private_dirty += size;
> +		else
> +			mss->private_clean += size;
> +		mss->pss += size << PSS_SHIFT;
> +	}
> +}
> +
>  #ifdef CONFIG_SHMEM
>  static int smaps_pte_hole(unsigned long addr, unsigned long end,
>  		struct mm_walk *walk)
> @@ -528,7 +577,16 @@ static void smaps_pte_entry(pte_t *pte, unsigned long addr,
>  	struct page *page = NULL;
>  
>  	if (pte_present(*pte)) {
> -		page = vm_normal_page(vma, addr, *pte);
> +		if (!vma_is_dax(vma))
> +			page = vm_normal_page(vma, addr, *pte);
> +		else if (pte_devmap(*pte)) {
> +			struct dev_pagemap *pgmap;
> +
> +			pgmap = get_dev_pagemap(pte_pfn(*pte), NULL);
> +			if (!pgmap)
> +				return;
> +			page = pte_page(*pte);
> +		}
>  	} else if (is_swap_pte(*pte)) {
>  		swp_entry_t swpent = pte_to_swp_entry(*pte);
>  
> @@ -579,7 +637,19 @@ static void smaps_pmd_entry(pmd_t *pmd, unsigned long addr,
>  	struct page *page;
>  
>  	/* FOLL_DUMP will return -EFAULT on huge zero page */
> -	page = follow_trans_huge_pmd(vma, addr, pmd, FOLL_DUMP);
> +	if (!vma_is_dax(vma))
> +		page = follow_trans_huge_pmd(vma, addr, pmd, FOLL_DUMP);
> +	else if (pmd_devmap(*pmd)) {
> +		struct dev_pagemap *pgmap;
> +
> +		pgmap = get_dev_pagemap(pmd_pfn(*pmd), NULL);
> +		if (!pgmap)
> +			return;
> +		page = pmd_page(*pmd);
> +		smaps_account_dax_huge(mss, page, PMD_SIZE, pmd_young(*pmd),
> +			pmd_dirty(*pmd));
> +		return;
> +	}
>  	if (IS_ERR_OR_NULL(page))
>  		return;
>  	if (PageAnon(page))
> 

There's a fair amount of copying and pasting going on here.  There is,
again, a bunch of specialized DAX code.  Isn't there a way to do this
more generically?