From: Christoph Hellwig <hch@lst.de>
To: Matthew Wilcox <willy@infradead.org>
Cc: Christoph Hellwig <hch@lst.de>,
Andrew Morton <akpm@linux-foundation.org>,
Kent Overstreet <kent.overstreet@gmail.com>,
linux-mm@kvack.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH 05/13] mm: simplify generic_file_buffered_read_no_cached_page
Date: Sun, 1 Nov 2020 11:29:08 +0100 [thread overview]
Message-ID: <20201101102908.GB26447@lst.de> (raw)
In-Reply-To: <20201031162813.GS27442@casper.infradead.org>
On Sat, Oct 31, 2020 at 04:28:13PM +0000, Matthew Wilcox wrote:
> On Sat, Oct 31, 2020 at 09:59:56AM +0100, Christoph Hellwig wrote:
> > +static int filemap_new_page(struct kiocb *iocb, struct iov_iter *iter,
> > + struct page **page)
>
> I don't like this calling convention. It's too easy to get it wrong,
> as you demonstrated. I preferred the way Kent had it with returning
> an ERR_PTR.
I guess for this function we can do that, here is the untested patch
on top of my series to show what we'd end up with:
diff --git a/mm/filemap.c b/mm/filemap.c
index b45f0bafdbaebf..1ac1fcd0067bf1 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2281,38 +2281,40 @@ static int filemap_make_page_uptodate(struct kiocb *iocb, struct iov_iter *iter,
return 0;
}
-static int filemap_new_page(struct kiocb *iocb, struct iov_iter *iter,
- struct page **page)
+static struct page *filemap_new_page(struct kiocb *iocb, struct iov_iter *iter)
{
struct address_space *mapping = iocb->ki_filp->f_mapping;
gfp_t gfp = mapping_gfp_constraint(mapping, GFP_KERNEL);
pgoff_t index = iocb->ki_pos >> PAGE_SHIFT;
+ struct page *page;
int error;
if (iocb->ki_flags & IOCB_NOIO)
- return -EAGAIN;
+ return ERR_PTR(-EAGAIN);
- *page = page_cache_alloc(mapping);
+ page = page_cache_alloc(mapping);
if (!page)
- return -ENOMEM;
- error = add_to_page_cache_lru(*page, mapping, index, gfp);
+ return ERR_PTR(-ENOMEM);
+ error = add_to_page_cache_lru(page, mapping, index, gfp);
if (error)
goto put_page;
- error = filemap_readpage(iocb, *page);
+ error = filemap_readpage(iocb, page);
if (error)
goto put_page;
- if (PageReadahead(*page)) {
+ if (PageReadahead(page)) {
error = -EAGAIN;
if (iocb->ki_flags & IOCB_NOIO)
goto put_page;
page_cache_async_readahead(mapping, &iocb->ki_filp->f_ra,
- iocb->ki_filp, *page, index,
+ iocb->ki_filp, page, index,
(iter->count + PAGE_SIZE - 1) >> PAGE_SHIFT);
}
return 0;
put_page:
- put_page(*page);
- return error;
+ put_page(page);
+ if (error == -EEXIST || error == AOP_TRUNCATED_PAGE)
+ return NULL;
+ return ERR_PTR(error);
}
static int filemap_find_get_pages(struct kiocb *iocb, struct iov_iter *iter,
@@ -2347,28 +2349,30 @@ static int filemap_read_pages(struct kiocb *iocb, struct iov_iter *iter,
return -EINTR;
nr_pages = filemap_find_get_pages(iocb, iter, pages, nr);
- if (nr_pages) {
- for (i = 0; i < nr_pages; i++) {
- err = filemap_make_page_uptodate(iocb, iter, pages[i],
- index + i, i == 0);
- if (err) {
- for (j = i; j < nr_pages; j++)
- put_page(pages[j]);
- nr_pages = i;
- break;
- }
+ if (!nr_pages) {
+ pages[0] = filemap_new_page(iocb, iter);
+ if (pages[0] == NULL)
+ goto retry;
+ if (IS_ERR(pages[0]))
+ return PTR_ERR(pages[0]);
+ return 1;
+ }
+
+ for (i = 0; i < nr_pages; i++) {
+ err = filemap_make_page_uptodate(iocb, iter, pages[i],
+ index + i, i == 0);
+ if (err) {
+ for (j = i; j < nr_pages; j++)
+ put_page(pages[j]);
+ if (likely(i > 0))
+ return i;
+ if (err == AOP_TRUNCATED_PAGE)
+ goto retry;
+ return err;
}
- } else {
- err = filemap_new_page(iocb, iter, &pages[0]);
- if (!err)
- nr_pages = 1;
}
- if (likely(nr_pages))
- return nr_pages;
- if (err == -EEXIST || err == AOP_TRUNCATED_PAGE)
- goto retry;
- return err;
+ return nr_pages;
}
/**
next prev parent reply other threads:[~2020-11-01 10:29 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-31 8:59 clean up the generic pagecache read helpers Christoph Hellwig
2020-10-31 8:59 ` [PATCH 01/13] mm: simplify generic_file_buffered_read_readpage Christoph Hellwig
2020-10-31 8:59 ` [PATCH 02/13] mm: simplify generic_file_buffered_read_pagenotuptodate Christoph Hellwig
2020-10-31 8:59 ` [PATCH 03/13] mm: lift the nowait checks into generic_file_buffered_read_pagenotuptodate Christoph Hellwig
2020-10-31 8:59 ` [PATCH 04/13] mm: handle readahead in generic_file_buffered_read_pagenotuptodate Christoph Hellwig
2020-10-31 17:06 ` Matthew Wilcox
2020-11-01 10:31 ` Christoph Hellwig
2020-11-01 10:49 ` Matthew Wilcox
2020-11-01 10:51 ` Christoph Hellwig
2020-11-01 10:51 ` Christoph Hellwig
2020-11-01 11:04 ` Matthew Wilcox
2020-11-01 11:52 ` Christoph Hellwig
2020-11-01 14:55 ` Matthew Wilcox
2020-11-02 8:18 ` Christoph Hellwig
2020-10-31 8:59 ` [PATCH 05/13] mm: simplify generic_file_buffered_read_no_cached_page Christoph Hellwig
2020-10-31 16:20 ` Matthew Wilcox
2020-10-31 16:28 ` Matthew Wilcox
2020-11-01 10:29 ` Christoph Hellwig [this message]
2020-10-31 8:59 ` [PATCH 06/13] mm: factor out a filemap_find_get_pages helper Christoph Hellwig
2020-10-31 8:59 ` [PATCH 07/13] mm: refactor generic_file_buffered_read_get_pages Christoph Hellwig
2020-11-01 11:18 ` Matthew Wilcox
2020-10-31 8:59 ` [PATCH 08/13] mm: move putting the page on error out of filemap_readpage Christoph Hellwig
2020-10-31 9:00 ` [PATCH 09/13] mm: move putting the page on error out of filemap_make_page_uptodate Christoph Hellwig
2020-10-31 9:00 ` [PATCH 10/13] mm: open code readahead in filemap_new_page Christoph Hellwig
2020-11-01 11:20 ` Matthew Wilcox
2020-10-31 9:00 ` [PATCH 11/13] mm: streamline the partially uptodate checks in filemap_make_page_uptodate Christoph Hellwig
2020-11-01 11:23 ` Matthew Wilcox
2020-10-31 9:00 ` [PATCH 12/13] mm: rename generic_file_buffered_read to filemap_read Christoph Hellwig
2020-10-31 9:00 ` [PATCH 13/13] mm: simplify generic_file_read_iter Christoph Hellwig
2020-10-31 15:42 ` clean up the generic pagecache read helpers Matthew Wilcox
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=20201101102908.GB26447@lst.de \
--to=hch@lst.de \
--cc=akpm@linux-foundation.org \
--cc=kent.overstreet@gmail.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=willy@infradead.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 NNTP newsgroup(s).