Re: [RFC 3/8] shmem: account for high order folios

From: Matthew Wilcox <willy@infradead.org>
To: Luis Chamberlain <mcgrof@kernel.org>
Cc: hughd@google.com, akpm@linux-foundation.org, brauner@kernel.org,
	djwong@kernel.org, p.raghav@samsung.com, da.gomez@samsung.com,
	a.manzanares@samsung.com, dave@stgolabs.net,
	yosryahmed@google.com, keescook@chromium.org, hare@suse.de,
	kbusch@kernel.org, patches@lists.linux.dev,
	linux-block@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC 3/8] shmem: account for high order folios
Date: Fri, 21 Apr 2023 23:46:58 +0100	[thread overview]
Message-ID: <ZEMSYtF3np7W6bIX@casper.infradead.org> (raw)
In-Reply-To: <20230421214400.2836131-4-mcgrof@kernel.org>

On Fri, Apr 21, 2023 at 02:43:55PM -0700, Luis Chamberlain wrote:
> -		if (xa_is_value(page))
> -			swapped++;
> +		if (xa_is_value(folio))
> +			swapped+=(folio_nr_pages(folio));

			swapped += folio_nr_pages(folio);

>  			if (xa_is_value(folio)) {
> +				long swaps_freed = 0;
>  				if (unfalloc)
>  					continue;
> -				nr_swaps_freed += !shmem_free_swap(mapping,
> -							indices[i], folio);
> +				swaps_freed = folio_nr_pages(folio);

Why initialise it to 0 when you're about to set it to folio_nr_pages()?

> +				if (!shmem_free_swap(mapping, indices[i], folio)) {
> +					if (swaps_freed > 1)
> +						pr_warn("swaps freed > 1 -- %lu\n", swaps_freed);

Debug code that escaped into this patch?

> -		info->swapped++;
> +		info->swapped+=folio_nr_pages(folio);

Same comment as earlier.

> -	info->alloced--;
> -	info->swapped--;
> +	info->alloced-=num_swap_pages;
> +	info->swapped-=num_swap_pages;

Spacing

> -	info->swapped--;
> +	info->swapped-= folio_nr_pages(folio);

Spacing.