From: Michael Tokarev Subject: what happened with dccaf33fa37 "ext4: flush any pending end_io requests before DIO" for 3.0? Date: Thu, 01 Dec 2011 00:38:07 +0400 Message-ID: <4ED6942F.7070006@msgid.tls.msk.ru> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigBEA045EE1588DA5E4CC6B678" Cc: linux-ext4@vger.kernel.org To: stable@kernel.org Return-path: Received: from isrv.corpit.ru ([86.62.121.231]:35134 "EHLO isrv.corpit.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751017Ab1K3UiO (ORCPT ); Wed, 30 Nov 2011 15:38:14 -0500 Sender: linux-ext4-owner@vger.kernel.org List-ID: This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigBEA045EE1588DA5E4CC6B678 Content-Type: multipart/mixed; boundary="------------010800080901080905040500" This is a multi-part message in MIME format. --------------010800080901080905040500 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Hello. Back in August 2011, a commit has been tagged to be included into stable, this one: commit dccaf33fa37a1bc5d651baeb3bfeb6becb86597b Author: Jiaying Zhang Date: Fri Aug 19 19:13:32 2011 -0400 ext4: flush any pending end_io requests before DIO reads w/dioread_no= lock There is a race between ext4 buffer write and direct_IO read with dioread_nolock mount option enabled. The problem is that we clear PageWriteback flag during end_io time but will do uninitialized-to-initialized extent conversion later with dioread_nol= ock. If an O_direct read request comes in during this period, ext4 will re= turn zero instead of the recently written data. This patch checks whether there are any pending uninitialized-to-init= ialized extent conversion requests before doing O_direct read to close the ra= ce. Note that this is just a bandaid fix. The fundamental issue is that w= e clear PageWriteback flag before we really complete an IO, which is problem-prone. To fix the fundamental issue, we may need to implement= an extent tree cache that we can use to look up pending to-be-converted = extents. Signed-off-by: Jiaying Zhang Signed-off-by: "Theodore Ts'o" Cc: stable@kernel.org There was one more ext4 commit at that time, which made its way into stable but this one did not. I wonder if the reason for that was the fact that it needed a small "backport" for 3.0, since in 3.1+ the code has been moved into another file, and the context is slightly different. In that case, attached is the "backport" which we use with 3.0.x since that time. Thanks! /mjt --------------010800080901080905040500 Content-Type: text/x-diff; name="ext4-flush-any-pending-end_io-requests-before-DIO-reads-w-dioread_nolock-dccaf33fa3.diff" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename*0="ext4-flush-any-pending-end_io-requests-before-DIO-reads-w-di"; filename*1="oread_nolock-dccaf33fa3.diff" (backported to 3.0 by mjt) commit dccaf33fa37a1bc5d651baeb3bfeb6becb86597b Author: Jiaying Zhang Date: Fri Aug 19 19:13:32 2011 -0400 ext4: flush any pending end_io requests before DIO reads w/dioread_no= lock =20 There is a race between ext4 buffer write and direct_IO read with dioread_nolock mount option enabled. The problem is that we clear PageWriteback flag during end_io time but will do uninitialized-to-initialized extent conversion later with dioread_nol= ock. If an O_direct read request comes in during this period, ext4 will re= turn zero instead of the recently written data. =20 This patch checks whether there are any pending uninitialized-to-init= ialized extent conversion requests before doing O_direct read to close the ra= ce. Note that this is just a bandaid fix. The fundamental issue is that w= e clear PageWriteback flag before we really complete an IO, which is problem-prone. To fix the fundamental issue, we may need to implement= an extent tree cache that we can use to look up pending to-be-converted = extents. =20 Signed-off-by: Jiaying Zhang Signed-off-by: "Theodore Ts'o" Signed-off-by: Michael Tokarev Cc: stable@kernel.org diff --git a/fs/ext4/indirect.c b/fs/ext4/indirect.c index b8602cd..0962642 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3507,12 +3507,17 @@ ssize_t ext4_ind_direct_IO(int rw, struct kiocb *= iocb, } =20 retry: - if (rw =3D=3D READ && ext4_should_dioread_nolock(inode)) + if (rw =3D=3D READ && ext4_should_dioread_nolock(inode)) { + if (unlikely(!list_empty(&ei->i_completed_io_list))) { + mutex_lock(&inode->i_mutex); + ext4_flush_completed_IO(inode); + mutex_unlock(&inode->i_mutex); + } ret =3D __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov, offset, nr_segs, ext4_get_block, NULL, NULL, 0); - else { + } else { ret =3D blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov, offset, nr_segs, =20 --------------010800080901080905040500-- --------------enigBEA045EE1588DA5E4CC6B678 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iJwEAQECAAYFAk7WlDQACgkQUlPFrXTwyDiMMQQAj/6MV9F3cQW48yQMTYJp+j92 bwfEqMqS0y3SCE7MGqoVwn8+7wQAFqpFwGoGcKssAxpqm4tEzGNWC6KYHNKnwJal ar+ZHCxtjMm4/ZVL0ABU23YIpEWyOXZUDGQWV5zDCLyl8UQYxDW9sj6LAjzOXedt FKYnJUA7D5P3CR8nFqg= =rpBy -----END PGP SIGNATURE----- --------------enigBEA045EE1588DA5E4CC6B678--