Re: [PATCH v3 23/25] mm/memremap_pages: Initialize all ZONE_DEVICE pages to start at refcount 0

From: Jason Gunthorpe <jgg@nvidia.com>
To: Dan Williams <dan.j.williams@intel.com>
Cc: linux-mm@kvack.org, Matthew Wilcox <willy@infradead.org>,
	Jan Kara <jack@suse.cz>, "Darrick J. Wong" <djwong@kernel.org>,
	Christoph Hellwig <hch@lst.de>,
	John Hubbard <jhubbard@nvidia.com>,
	Alistair Popple <apopple@nvidia.com>,
	david@fromorbit.com, nvdimm@lists.linux.dev,
	akpm@linux-foundation.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH v3 23/25] mm/memremap_pages: Initialize all ZONE_DEVICE pages to start at refcount 0
Date: Mon, 17 Oct 2022 16:48:01 -0300	[thread overview]
Message-ID: <Y02xceDlo8hfrPm/@nvidia.com> (raw)
In-Reply-To: <166579195218.2236710.8731183545033177929.stgit@dwillia2-xfh.jf.intel.com>

On Fri, Oct 14, 2022 at 04:59:12PM -0700, Dan Williams wrote:
> diff --git a/mm/memremap.c b/mm/memremap.c
> index c46e700f5245..368ff41c560b 100644
> --- a/mm/memremap.c
> +++ b/mm/memremap.c
> @@ -469,8 +469,10 @@ EXPORT_SYMBOL_GPL(get_dev_pagemap);
>  
>  void free_zone_device_page(struct page *page)
>  {
> -	if (WARN_ON_ONCE(!page->pgmap->ops || !page->pgmap->ops->page_free))
> -		return;
> +	struct dev_pagemap *pgmap = page->pgmap;
> +
> +	/* wake filesystem 'break dax layouts' waiters */
> +	wake_up_var(page);

Shouldn't this be in the DAX page_free() op?

> +/*
> + * A symmetric helper to undo the page references acquired by
> + * pgmap_request_folios(), but the caller can also just arrange
> + * folio_put() on all the folios it acquired previously for the same
> + * effect.
> + */
> +void pgmap_release_folios(struct folio *folio, int nr_folios)
>  {
>  	struct folio *iter;
>  	int i;
>  
> -	for (iter = folio, i = 0; i < nr_folios; iter = folio_next(iter), i++) {
> -		if (!put_devmap_managed_page(&iter->page))
> -			folio_put(iter);
> -		if (!folio_ref_count(iter))
> -			put_dev_pagemap(pgmap);
> -	}
> +	for (iter = folio, i = 0; i < nr_folios; iter = folio_next(folio), i++)
> +		folio_put(iter);
>  }

Oh, so now this half makes more sense as an API, but It seems like it
is not named right, if folio-multiput is useful shouldn't it be a
folio_put_many() or something?

Jason