linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: NeilBrown <neilb@suse.de>
To: Trond Myklebust <trond.myklebust@hammerspace.com>,
	Anna Schumaker <anna.schumaker@netapp.com>,
	Chuck Lever <chuck.lever@oracle.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Mel Gorman <mgorman@suse.de>
Cc: linux-nfs@vger.kernel.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: [PATCH 11/13] NFS: swap-out must always use STABLE writes.
Date: Tue, 16 Nov 2021 13:44:04 +1100	[thread overview]
Message-ID: <163703064457.25805.5324231084334985723.stgit@noble.brown> (raw)
In-Reply-To: <163702956672.25805.16457749992977493579.stgit@noble.brown>

The commit handling code is not safe against memory-pressure deadlocks
when writing to swap.  In particular, nfs_commitdata_alloc() blocks
indefinitely waiting for memory, and this can consume all available
workqueue threads.

swap-out most likely uses STABLE writes anyway as COND_STABLE indicates
that a stable write should be used if the write fits in a single
request, and it normally does.  However if we ever swap with a small
wsize, or gather unusually large numbers of pages for a single write,
this might change.

For safety, make it explicit in the code that direct writes use for swap
must always use FLUSH_COND_STABLE.

Signed-off-by: NeilBrown <neilb@suse.de>
---
 fs/nfs/direct.c        |   12 +++++++-----
 fs/nfs/file.c          |    2 +-
 include/linux/nfs_fs.h |    3 ++-
 3 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 1e80d243ba25..8d3b12402725 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -173,7 +173,7 @@ ssize_t nfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 
 	if (iov_iter_rw(iter) == READ)
 		return nfs_file_direct_read(iocb, iter);
-	return nfs_file_direct_write(iocb, iter);
+	return nfs_file_direct_write(iocb, iter, FLUSH_STABLE);
 }
 
 static void nfs_direct_release_pages(struct page **pages, unsigned int npages)
@@ -789,7 +789,7 @@ static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops = {
  */
 static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
 					       struct iov_iter *iter,
-					       loff_t pos)
+					       loff_t pos, int ioflags)
 {
 	struct nfs_pageio_descriptor desc;
 	struct inode *inode = dreq->inode;
@@ -797,7 +797,7 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
 	size_t requested_bytes = 0;
 	size_t wsize = max_t(size_t, NFS_SERVER(inode)->wsize, PAGE_SIZE);
 
-	nfs_pageio_init_write(&desc, inode, FLUSH_COND_STABLE, false,
+	nfs_pageio_init_write(&desc, inode, ioflags, false,
 			      &nfs_direct_write_completion_ops);
 	desc.pg_dreq = dreq;
 	get_dreq(dreq);
@@ -875,6 +875,7 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
  * nfs_file_direct_write - file direct write operation for NFS files
  * @iocb: target I/O control block
  * @iter: vector of user buffers from which to write data
+ * @ioflags: flags for nfs_pageio_init_write()
  *
  * We use this function for direct writes instead of calling
  * generic_file_aio_write() in order to avoid taking the inode
@@ -891,7 +892,8 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
  * Note that O_APPEND is not supported for NFS direct writes, as there
  * is no atomic O_APPEND write facility in the NFS protocol.
  */
-ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
+ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter,
+			      int ioflags)
 {
 	ssize_t result, requested;
 	size_t count;
@@ -935,7 +937,7 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
 
 	nfs_start_io_direct(inode);
 
-	requested = nfs_direct_write_schedule_iovec(dreq, iter, pos);
+	requested = nfs_direct_write_schedule_iovec(dreq, iter, pos, ioflags);
 
 	if (mapping->nrpages) {
 		invalidate_inode_pages2_range(mapping,
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 59c271f42ea5..878a6a510a5e 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -630,7 +630,7 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 		result = generic_write_checks(iocb, from);
 		if (result <= 0)
 			return result;
-		return nfs_file_direct_write(iocb, from);
+		return nfs_file_direct_write(iocb, from, FLUSH_COND_STABLE);
 	}
 
 	dprintk("NFS: write(%pD2, %zu@%Ld)\n",
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index 5a605e51f4b1..ca312aea6bec 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -509,7 +509,8 @@ extern ssize_t nfs_direct_IO(struct kiocb *, struct iov_iter *);
 extern ssize_t nfs_file_direct_read(struct kiocb *iocb,
 			struct iov_iter *iter);
 extern ssize_t nfs_file_direct_write(struct kiocb *iocb,
-			struct iov_iter *iter);
+				     struct iov_iter *iter,
+				     int ioflags);
 
 /*
  * linux/fs/nfs/dir.c



  parent reply	other threads:[~2021-11-16  2:50 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-16  2:44 [PATCH 00/13] Repair SWAP-over-NFS NeilBrown
2021-11-16  2:44 ` [PATCH 09/13] SUNRPC: improve 'swap' handling: scheduling and PF_MEMALLOC NeilBrown
2021-11-16  2:44 ` NeilBrown [this message]
2021-11-16  2:44 ` [PATCH 02/13] NFS: do not take i_rwsem for swap IO NeilBrown
2021-11-16  7:52   ` Christoph Hellwig
2021-11-16 21:50     ` NeilBrown
2021-11-17  5:49       ` Christoph Hellwig
2021-11-16  2:44 ` [PATCH 08/13] NFS: discard NFS_RPC_SWAPFLAGS and RPC_TASK_ROOTCREDS NeilBrown
2021-11-16  2:44 ` [PATCH 01/13] NFS: move generic_write_checks() call from nfs_file_direct_write() to nfs_file_write() NeilBrown
2021-11-16  2:44 ` [PATCH 07/13] SUNRPC: remove scheduling boost for "SWAPPER" tasks NeilBrown
2021-11-16  2:44 ` [PATCH 10/13] NFSv4: keep state manager thread active if swap is enabled NeilBrown
2021-11-16  2:44 ` [PATCH 05/13] SUNRPC/auth: async tasks mustn't block waiting for memory NeilBrown
2021-11-16  2:44 ` [PATCH 04/13] SUNRPC/call_alloc: " NeilBrown
2021-11-16  2:44 ` [PATCH 13/13] MM: use AIO for DIO writes to swap NeilBrown
2021-11-16  2:44 ` [PATCH 06/13] SUNRPC/xprt: async tasks mustn't block waiting for memory NeilBrown
2021-11-16  2:44 ` [PATCH 03/13] MM: reclaim mustn't enter FS for swap-over-NFS NeilBrown
2021-11-16  8:32   ` Christoph Hellwig
2021-11-16 21:35     ` NeilBrown
2021-11-17  5:50       ` Christoph Hellwig
2021-11-18  1:43   ` kernel test robot
2021-11-16  2:44 ` [PATCH 12/13] MM: use AIO/DIO for reads from SWP_FS_OPS swap-space NeilBrown
2021-11-16  8:31   ` Christoph Hellwig
2021-11-16 21:46     ` NeilBrown
2021-11-16  3:29 ` [PATCH 00/13] Repair SWAP-over-NFS Matthew Wilcox
2021-11-16  3:55   ` NeilBrown

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=163703064457.25805.5324231084334985723.stgit@noble.brown \
    --to=neilb@suse.de \
    --cc=akpm@linux-foundation.org \
    --cc=anna.schumaker@netapp.com \
    --cc=chuck.lever@oracle.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-nfs@vger.kernel.org \
    --cc=mgorman@suse.de \
    --cc=trond.myklebust@hammerspace.com \
    /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).