From: "Pankaj Raghav (Samsung)" <kernel@pankajraghav.com>
To: Dave Chinner <david@fromorbit.com>
Cc: linux-xfs@vger.kernel.org
Subject: Re: [PATCH 3/9] xfs: convert buffer cache to use high order folios
Date: Mon, 25 Mar 2024 12:17:55 +0100 [thread overview]
Message-ID: <p73usozirhag4kembg43hlfaqwyn2oyjzu3au5p4aw3tdf3cj5@2icixk7z7ltz> (raw)
In-Reply-To: <Zf4Ag349VrT/kv8n@dread.disaster.area>
On Sat, Mar 23, 2024 at 09:04:51AM +1100, Dave Chinner wrote:
> On Fri, Mar 22, 2024 at 09:02:31AM +0100, Pankaj Raghav (Samsung) wrote:
> > > * Bulk filling of pages can take multiple calls. Not filling the entire
> > > @@ -426,7 +484,7 @@ _xfs_buf_map_folios(
> > > {
> > > ASSERT(bp->b_flags & _XBF_FOLIOS);
> > > if (bp->b_folio_count == 1) {
> > > - /* A single page buffer is always mappable */
> > > + /* A single folio buffer is always mappable */
> > > bp->b_addr = folio_address(bp->b_folios[0]);
> > > } else if (flags & XBF_UNMAPPED) {
> > > bp->b_addr = NULL;
> > > @@ -1525,20 +1583,28 @@ xfs_buf_ioapply_map(
> > > int *count,
> > > blk_opf_t op)
> > > {
> > > - int page_index;
> > > - unsigned int total_nr_pages = bp->b_folio_count;
> > > - int nr_pages;
> > > + int folio_index;
> > > + unsigned int total_nr_folios = bp->b_folio_count;
> > > + int nr_folios;
> > > struct bio *bio;
> > > sector_t sector = bp->b_maps[map].bm_bn;
> > > int size;
> > > int offset;
> > >
> > > - /* skip the pages in the buffer before the start offset */
> > > - page_index = 0;
> > > + /*
> > > + * If the start offset if larger than a single page, we need to be
> > > + * careful. We might have a high order folio, in which case the indexing
> > > + * is from the start of the buffer. However, if we have more than one
> > > + * folio single page folio in the buffer, we need to skip the folios in
> > s/folio single page folio/single page folio/
> >
> > > + * the buffer before the start offset.
> > > + */
> > > + folio_index = 0;
> > > offset = *buf_offset;
> > > - while (offset >= PAGE_SIZE) {
> > > - page_index++;
> > > - offset -= PAGE_SIZE;
> > > + if (bp->b_folio_count > 1) {
> > > + while (offset >= PAGE_SIZE) {
> > > + folio_index++;
> > > + offset -= PAGE_SIZE;
> >
> > Can this be:
> > folio_index = offset >> PAGE_SHIFT;
> > offset = offset_in_page(offset);
> >
> > instead of a loop?
>
> It could, but I'm not going to change it or even bother to fix the
> comment as this code goes away later in the patch set.
>
> See "[PATCH 7/9] xfs: walk b_addr for buffer I/O" later in the
> series - once we can rely on bp->b_addr always being set for buffers
> we convert this code to a simpler and more efficient folio based
> iteration that is not reliant on pages or PAGE_SIZE at all.
>
Ah, I missed seeing the whole series before replying to individual patches.
Nvm.
Thanks for the pointer on patch 7, and I agree that it does not make
sense to change this patch as it goes away later.
--
Pankaj
next prev parent reply other threads:[~2024-03-25 11:23 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-03-18 22:45 [PATCH v2 0/9] xfs: use large folios for buffers Dave Chinner
2024-03-18 22:45 ` [PATCH 1/9] xfs: unmapped buffer item size straddling mismatch Dave Chinner
2024-03-18 22:45 ` [PATCH 2/9] xfs: use folios in the buffer cache Dave Chinner
2024-03-19 6:38 ` Christoph Hellwig
2024-03-19 6:52 ` Dave Chinner
2024-03-19 6:53 ` Christoph Hellwig
2024-03-19 21:42 ` Dave Chinner
2024-03-19 21:42 ` Dave Chinner
2024-03-19 17:15 ` Darrick J. Wong
2024-03-18 22:45 ` [PATCH 3/9] xfs: convert buffer cache to use high order folios Dave Chinner
2024-03-19 6:55 ` Christoph Hellwig
2024-03-19 17:29 ` Darrick J. Wong
2024-03-19 21:32 ` Christoph Hellwig
2024-03-19 21:38 ` Darrick J. Wong
2024-03-19 21:41 ` Christoph Hellwig
2024-03-19 22:23 ` Dave Chinner
2024-03-21 2:12 ` Darrick J. Wong
2024-03-21 2:40 ` Darrick J. Wong
2024-03-21 21:28 ` Christoph Hellwig
2024-03-21 21:39 ` Darrick J. Wong
2024-03-21 22:02 ` Christoph Hellwig
2024-03-19 21:55 ` Dave Chinner
2024-03-22 8:02 ` Pankaj Raghav (Samsung)
2024-03-22 22:04 ` Dave Chinner
2024-03-25 11:17 ` Pankaj Raghav (Samsung) [this message]
2024-03-18 22:45 ` [PATCH 4/9] xfs: kill XBF_UNMAPPED Dave Chinner
2024-03-19 17:30 ` Darrick J. Wong
2024-03-19 23:36 ` Dave Chinner
2024-03-18 22:45 ` [PATCH 5/9] xfs: buffer items don't straddle pages anymore Dave Chinner
2024-03-19 6:56 ` Christoph Hellwig
2024-03-19 17:31 ` Darrick J. Wong
2024-03-18 22:45 ` [PATCH 6/9] xfs: map buffers in xfs_buf_alloc_folios Dave Chinner
2024-03-19 17:34 ` Darrick J. Wong
2024-03-19 21:32 ` Christoph Hellwig
2024-03-19 21:39 ` Darrick J. Wong
2024-03-19 21:41 ` Christoph Hellwig
2024-03-18 22:45 ` [PATCH 7/9] xfs: walk b_addr for buffer I/O Dave Chinner
2024-03-19 17:42 ` Darrick J. Wong
2024-03-19 21:33 ` Christoph Hellwig
2024-03-18 22:45 ` [PATCH 8/9] xfs: use vmalloc for multi-folio buffers Dave Chinner
2024-03-19 17:48 ` Darrick J. Wong
2024-03-20 0:20 ` Dave Chinner
2024-03-18 22:46 ` [PATCH 9/9] xfs: rename bp->b_folio_count Dave Chinner
2024-03-19 7:37 ` Christoph Hellwig
2024-03-19 23:59 ` Dave Chinner
2024-03-19 0:24 ` [PATCH v2 0/9] xfs: use large folios for buffers Christoph Hellwig
2024-03-19 0:44 ` Dave Chinner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=p73usozirhag4kembg43hlfaqwyn2oyjzu3au5p4aw3tdf3cj5@2icixk7z7ltz \
--to=kernel@pankajraghav.com \
--cc=david@fromorbit.com \
--cc=linux-xfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).