From: John Hubbard <jhubbard@nvidia.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Jens Axboe <axboe@kernel.dk>,
Alexander Viro <viro@zeniv.linux.org.uk>,
Miklos Szeredi <miklos@szeredi.hu>,
Christoph Hellwig <hch@infradead.org>,
"Darrick J . Wong" <djwong@kernel.org>,
Trond Myklebust <trond.myklebust@hammerspace.com>,
Anna Schumaker <anna@kernel.org>, Jan Kara <jack@suse.cz>,
David Hildenbrand <david@redhat.com>,
Logan Gunthorpe <logang@deltatee.com>,
<linux-block@vger.kernel.org>, <linux-fsdevel@vger.kernel.org>,
<linux-xfs@vger.kernel.org>, <linux-nfs@vger.kernel.org>,
<linux-mm@kvack.org>, LKML <linux-kernel@vger.kernel.org>,
John Hubbard <jhubbard@nvidia.com>
Subject: [PATCH v2 4/7] iov_iter: new iov_iter_pin_pages*() routines
Date: Tue, 30 Aug 2022 21:18:40 -0700 [thread overview]
Message-ID: <20220831041843.973026-5-jhubbard@nvidia.com> (raw)
In-Reply-To: <20220831041843.973026-1-jhubbard@nvidia.com>
Provide two new wrapper routines that are intended for user space pages
only:
iov_iter_pin_pages()
iov_iter_pin_pages_alloc()
Internally, these routines call pin_user_pages_fast(), instead of
get_user_pages_fast(), for user_backed_iter(i) and iov_iter_bvec(i)
cases.
As always, callers must use unpin_user_pages() or a suitable FOLL_PIN
variant, to release the pages, if they actually were acquired via
pin_user_pages_fast().
This is a prerequisite to converting bio/block layers over to use
pin_user_pages_fast().
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
---
include/linux/uio.h | 4 +++
lib/iov_iter.c | 86 +++++++++++++++++++++++++++++++++++++++++----
2 files changed, 84 insertions(+), 6 deletions(-)
diff --git a/include/linux/uio.h b/include/linux/uio.h
index 5896af36199c..e26908e443d1 100644
--- a/include/linux/uio.h
+++ b/include/linux/uio.h
@@ -251,6 +251,10 @@ ssize_t iov_iter_get_pages2(struct iov_iter *i, struct page **pages,
size_t maxsize, unsigned maxpages, size_t *start);
ssize_t iov_iter_get_pages_alloc2(struct iov_iter *i, struct page ***pages,
size_t maxsize, size_t *start);
+ssize_t iov_iter_pin_pages(struct iov_iter *i, struct page **pages,
+ size_t maxsize, unsigned int maxpages, size_t *start);
+ssize_t iov_iter_pin_pages_alloc(struct iov_iter *i, struct page ***pages,
+ size_t maxsize, size_t *start);
int iov_iter_npages(const struct iov_iter *i, int maxpages);
void iov_iter_restore(struct iov_iter *i, struct iov_iter_state *state);
diff --git a/lib/iov_iter.c b/lib/iov_iter.c
index 4b7fce72e3e5..c63ce0eadfcb 100644
--- a/lib/iov_iter.c
+++ b/lib/iov_iter.c
@@ -1425,9 +1425,31 @@ static struct page *first_bvec_segment(const struct iov_iter *i,
return page;
}
+enum pages_alloc_internal_flags {
+ USE_FOLL_GET,
+ MAYBE_USE_FOLL_PIN
+};
+
+/*
+ * Pins pages, either via get_page(), or via pin_user_page*(). The caller is
+ * responsible for tracking which pinning mechanism was used here, and releasing
+ * pages via the appropriate call: put_page() or unpin_user_page().
+ *
+ * The way to figure that out is:
+ *
+ * a) If how_to_pin == FOLL_GET, then this routine will always pin via
+ * get_page().
+ *
+ * b) If how_to_pin == MAYBE_USE_FOLL_PIN, then this routine will pin via
+ * pin_user_page*() for either user_backed_iter(i) cases, or
+ * iov_iter_is_bvec(i) cases. However, for the other cases (pipe,
+ * xarray), pages will be pinned via get_page().
+ */
static ssize_t __iov_iter_get_pages_alloc(struct iov_iter *i,
struct page ***pages, size_t maxsize,
- unsigned int maxpages, size_t *start)
+ unsigned int maxpages, size_t *start,
+ enum pages_alloc_internal_flags how_to_pin)
+
{
unsigned int n;
@@ -1454,7 +1476,12 @@ static ssize_t __iov_iter_get_pages_alloc(struct iov_iter *i,
n = want_pages_array(pages, maxsize, *start, maxpages);
if (!n)
return -ENOMEM;
- res = get_user_pages_fast(addr, n, gup_flags, *pages);
+
+ if (how_to_pin == MAYBE_USE_FOLL_PIN)
+ res = pin_user_pages_fast(addr, n, gup_flags, *pages);
+ else
+ res = get_user_pages_fast(addr, n, gup_flags, *pages);
+
if (unlikely(res <= 0))
return res;
maxsize = min_t(size_t, maxsize, res * PAGE_SIZE - *start);
@@ -1470,8 +1497,13 @@ static ssize_t __iov_iter_get_pages_alloc(struct iov_iter *i,
if (!n)
return -ENOMEM;
p = *pages;
- for (int k = 0; k < n; k++)
- get_page(p[k] = page + k);
+ for (int k = 0; k < n; k++) {
+ p[k] = page + k;
+ if (how_to_pin == MAYBE_USE_FOLL_PIN)
+ pin_user_page(p[k]);
+ else
+ get_page(p[k]);
+ }
maxsize = min_t(size_t, maxsize, n * PAGE_SIZE - *start);
i->count -= maxsize;
i->iov_offset += maxsize;
@@ -1497,10 +1529,29 @@ ssize_t iov_iter_get_pages2(struct iov_iter *i,
return 0;
BUG_ON(!pages);
- return __iov_iter_get_pages_alloc(i, &pages, maxsize, maxpages, start);
+ return __iov_iter_get_pages_alloc(i, &pages, maxsize, maxpages, start,
+ USE_FOLL_GET);
}
EXPORT_SYMBOL(iov_iter_get_pages2);
+/*
+ * A FOLL_PIN variant that calls pin_user_pages_fast() instead of
+ * get_user_pages_fast().
+ */
+ssize_t iov_iter_pin_pages(struct iov_iter *i,
+ struct page **pages, size_t maxsize, unsigned int maxpages,
+ size_t *start)
+{
+ if (!maxpages)
+ return 0;
+ if (WARN_ON_ONCE(!pages))
+ return -EINVAL;
+
+ return __iov_iter_get_pages_alloc(i, &pages, maxsize, maxpages, start,
+ MAYBE_USE_FOLL_PIN);
+}
+EXPORT_SYMBOL(iov_iter_pin_pages);
+
ssize_t iov_iter_get_pages_alloc2(struct iov_iter *i,
struct page ***pages, size_t maxsize,
size_t *start)
@@ -1509,7 +1560,8 @@ ssize_t iov_iter_get_pages_alloc2(struct iov_iter *i,
*pages = NULL;
- len = __iov_iter_get_pages_alloc(i, pages, maxsize, ~0U, start);
+ len = __iov_iter_get_pages_alloc(i, pages, maxsize, ~0U, start,
+ USE_FOLL_GET);
if (len <= 0) {
kvfree(*pages);
*pages = NULL;
@@ -1518,6 +1570,28 @@ ssize_t iov_iter_get_pages_alloc2(struct iov_iter *i,
}
EXPORT_SYMBOL(iov_iter_get_pages_alloc2);
+/*
+ * A FOLL_PIN variant that calls pin_user_pages_fast() instead of
+ * get_user_pages_fast().
+ */
+ssize_t iov_iter_pin_pages_alloc(struct iov_iter *i,
+ struct page ***pages, size_t maxsize,
+ size_t *start)
+{
+ ssize_t len;
+
+ *pages = NULL;
+
+ len = __iov_iter_get_pages_alloc(i, pages, maxsize, ~0U, start,
+ MAYBE_USE_FOLL_PIN);
+ if (len <= 0) {
+ kvfree(*pages);
+ *pages = NULL;
+ }
+ return len;
+}
+EXPORT_SYMBOL(iov_iter_pin_pages_alloc);
+
size_t csum_and_copy_from_iter(void *addr, size_t bytes, __wsum *csum,
struct iov_iter *i)
{
--
2.37.2
next prev parent reply other threads:[~2022-08-31 4:19 UTC|newest]
Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-31 4:18 [PATCH v2 0/7] convert most filesystems to pin_user_pages_fast() John Hubbard
2022-08-31 4:18 ` [PATCH v2 1/7] mm: change release_pages() to use unsigned long for npages John Hubbard
2022-08-31 4:18 ` [PATCH v2 2/7] mm/gup: introduce pin_user_page() John Hubbard
2022-09-06 6:37 ` Christoph Hellwig
2022-09-06 7:12 ` John Hubbard
2022-08-31 4:18 ` [PATCH v2 3/7] block: add dio_w_*() wrappers for pin, unpin user pages John Hubbard
2022-08-31 4:18 ` John Hubbard [this message]
2022-09-01 0:42 ` [PATCH v2 4/7] iov_iter: new iov_iter_pin_pages*() routines Al Viro
2022-09-01 1:48 ` John Hubbard
2022-09-06 6:47 ` Christoph Hellwig
2022-09-06 7:44 ` John Hubbard
2022-09-06 7:48 ` Christoph Hellwig
2022-09-06 7:58 ` John Hubbard
2022-09-07 8:50 ` Christoph Hellwig
2022-09-06 10:21 ` Jan Kara
2022-09-07 8:45 ` Christoph Hellwig
2022-09-14 3:51 ` Al Viro
2022-09-14 14:52 ` Jan Kara
2022-09-14 16:42 ` Al Viro
2022-09-15 8:16 ` Jan Kara
2022-09-16 1:55 ` Al Viro
2022-09-20 5:02 ` Al Viro
2022-09-22 14:36 ` Christoph Hellwig
2022-09-22 14:43 ` David Hildenbrand
2022-09-22 14:45 ` Christoph Hellwig
2022-09-22 2:22 ` Al Viro
2022-09-22 6:09 ` John Hubbard
2022-09-22 11:29 ` Jan Kara
2022-09-23 3:19 ` Al Viro
2022-09-23 4:05 ` John Hubbard
2022-09-23 8:39 ` Christoph Hellwig
2022-09-23 12:22 ` Jan Kara
2022-09-23 4:34 ` John Hubbard
2022-09-22 14:38 ` Christoph Hellwig
2022-09-23 4:22 ` Al Viro
2022-09-23 8:44 ` Christoph Hellwig
2022-09-23 16:13 ` Al Viro
2022-09-26 15:53 ` Christoph Hellwig
2022-09-26 19:55 ` Al Viro
2022-09-22 14:31 ` Christoph Hellwig
2022-09-22 14:36 ` Al Viro
2022-08-31 4:18 ` [PATCH v2 5/7] block, bio, fs: convert most filesystems to pin_user_pages_fast() John Hubbard
2022-09-06 6:48 ` Christoph Hellwig
2022-09-06 7:15 ` John Hubbard
2022-08-31 4:18 ` [PATCH v2 6/7] NFS: direct-io: convert to FOLL_PIN pages John Hubbard
2022-09-06 6:49 ` Christoph Hellwig
2022-09-06 7:16 ` John Hubbard
2022-08-31 4:18 ` [PATCH v2 7/7] fuse: convert direct IO paths to use FOLL_PIN John Hubbard
2022-08-31 10:37 ` Miklos Szeredi
2022-09-01 1:33 ` John Hubbard
2022-09-06 6:36 ` [PATCH v2 0/7] convert most filesystems to pin_user_pages_fast() Christoph Hellwig
2022-09-06 7:10 ` John Hubbard
2022-09-06 7:22 ` Christoph Hellwig
2022-09-06 7:37 ` John Hubbard
2022-09-06 7:46 ` Christoph Hellwig
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=20220831041843.973026-5-jhubbard@nvidia.com \
--to=jhubbard@nvidia.com \
--cc=akpm@linux-foundation.org \
--cc=anna@kernel.org \
--cc=axboe@kernel.dk \
--cc=david@redhat.com \
--cc=djwong@kernel.org \
--cc=hch@infradead.org \
--cc=jack@suse.cz \
--cc=linux-block@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-nfs@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=logang@deltatee.com \
--cc=miklos@szeredi.hu \
--cc=trond.myklebust@hammerspace.com \
--cc=viro@zeniv.linux.org.uk \
/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).