[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YYKE/ohWPf7jUBM/@casper.infradead.org>
Date: Wed, 3 Nov 2021 12:47:58 +0000
From: Matthew Wilcox <willy@...radead.org>
To: "Darrick J. Wong" <djwong@...nel.org>
Cc: linux-xfs@...r.kernel.org, linux-fsdevel@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-block@...r.kernel.org,
Jens Axboe <axboe@...nel.dk>,
Christoph Hellwig <hch@...radead.org>
Subject: Re: [PATCH 15/21] iomap: Convert iomap_write_begin and
iomap_write_end to folios
On Wed, Nov 03, 2021 at 03:15:13AM +0000, Matthew Wilcox wrote:
> On Tue, Nov 02, 2021 at 04:22:15PM -0700, Darrick J. Wong wrote:
> > > + page = folio_file_page(folio, pos >> PAGE_SHIFT);
> >
> > Isn't this only needed in the BUFFER_HEAD case?
>
> Good catch. Want me to fold this in?
>
> @@ -632,12 +631,12 @@ static int iomap_write_begin(const struct iomap_iter *iter, loff_t pos,
> goto out_no_page;
> }
>
> - page = folio_file_page(folio, pos >> PAGE_SHIFT);
> if (srcmap->type == IOMAP_INLINE)
> status = iomap_write_begin_inline(iter, folio);
> - else if (srcmap->flags & IOMAP_F_BUFFER_HEAD)
> + else if (srcmap->flags & IOMAP_F_BUFFER_HEAD) {
> + struct page *page = folio_file_page(folio, pos >> PAGE_SHIFT);
> status = __block_write_begin_int(page, pos, len, NULL, srcmap);
On second thoughts, this is silly. __block_write_begin_int() doesn't
want the precise page (because it constructs buffer_heads and attaches
them to the passed-in page). I should just pass &folio->page here.
And __block_write_begin_int() should be converted to take a folio
at some point.
Powered by blists - more mailing lists