From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
stable@vger.kernel.org, Tejun Heo <tj@kernel.org>,
Jiufei Xue <xuejiufei@gmail.com>, Jan Kara <jack@suse.cz>,
Jens Axboe <axboe@kernel.dk>
Subject: [PATCH 4.9 39/44] writeback: synchronize sync(2) against cgroup writeback membership switches
Date: Mon, 20 May 2019 14:14:28 +0200 [thread overview]
Message-ID: <20190520115235.564183757@linuxfoundation.org> (raw)
In-Reply-To: <20190520115230.720347034@linuxfoundation.org>
From: Tejun Heo <tj@kernel.org>
commit 7fc5854f8c6efae9e7624970ab49a1eac2faefb1 upstream.
sync_inodes_sb() can race against cgwb (cgroup writeback) membership
switches and fail to writeback some inodes. For example, if an inode
switches to another wb while sync_inodes_sb() is in progress, the new
wb might not be visible to bdi_split_work_to_wbs() at all or the inode
might jump from a wb which hasn't issued writebacks yet to one which
already has.
This patch adds backing_dev_info->wb_switch_rwsem to synchronize cgwb
switch path against sync_inodes_sb() so that sync_inodes_sb() is
guaranteed to see all the target wbs and inodes can't jump wbs to
escape syncing.
v2: Fixed misplaced rwsem init. Spotted by Jiufei.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Jiufei Xue <xuejiufei@gmail.com>
Link: http://lkml.kernel.org/r/dc694ae2-f07f-61e1-7097-7c8411cee12d@gmail.com
Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
fs/fs-writeback.c | 40 +++++++++++++++++++++++++++++++++++++--
include/linux/backing-dev-defs.h | 1
mm/backing-dev.c | 1
3 files changed, 40 insertions(+), 2 deletions(-)
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -331,11 +331,22 @@ struct inode_switch_wbs_context {
struct work_struct work;
};
+static void bdi_down_write_wb_switch_rwsem(struct backing_dev_info *bdi)
+{
+ down_write(&bdi->wb_switch_rwsem);
+}
+
+static void bdi_up_write_wb_switch_rwsem(struct backing_dev_info *bdi)
+{
+ up_write(&bdi->wb_switch_rwsem);
+}
+
static void inode_switch_wbs_work_fn(struct work_struct *work)
{
struct inode_switch_wbs_context *isw =
container_of(work, struct inode_switch_wbs_context, work);
struct inode *inode = isw->inode;
+ struct backing_dev_info *bdi = inode_to_bdi(inode);
struct address_space *mapping = inode->i_mapping;
struct bdi_writeback *old_wb = inode->i_wb;
struct bdi_writeback *new_wb = isw->new_wb;
@@ -344,6 +355,12 @@ static void inode_switch_wbs_work_fn(str
void **slot;
/*
+ * If @inode switches cgwb membership while sync_inodes_sb() is
+ * being issued, sync_inodes_sb() might miss it. Synchronize.
+ */
+ down_read(&bdi->wb_switch_rwsem);
+
+ /*
* By the time control reaches here, RCU grace period has passed
* since I_WB_SWITCH assertion and all wb stat update transactions
* between unlocked_inode_to_wb_begin/end() are guaranteed to be
@@ -435,6 +452,8 @@ skip_switch:
spin_unlock(&new_wb->list_lock);
spin_unlock(&old_wb->list_lock);
+ up_read(&bdi->wb_switch_rwsem);
+
if (switched) {
wb_wakeup(new_wb);
wb_put(old_wb);
@@ -475,9 +494,18 @@ static void inode_switch_wbs(struct inod
if (inode->i_state & I_WB_SWITCH)
return;
+ /*
+ * Avoid starting new switches while sync_inodes_sb() is in
+ * progress. Otherwise, if the down_write protected issue path
+ * blocks heavily, we might end up starting a large number of
+ * switches which will block on the rwsem.
+ */
+ if (!down_read_trylock(&bdi->wb_switch_rwsem))
+ return;
+
isw = kzalloc(sizeof(*isw), GFP_ATOMIC);
if (!isw)
- return;
+ goto out_unlock;
/* find and pin the new wb */
rcu_read_lock();
@@ -511,12 +539,14 @@ static void inode_switch_wbs(struct inod
* Let's continue after I_WB_SWITCH is guaranteed to be visible.
*/
call_rcu(&isw->rcu_head, inode_switch_wbs_rcu_fn);
- return;
+ goto out_unlock;
out_free:
if (isw->new_wb)
wb_put(isw->new_wb);
kfree(isw);
+out_unlock:
+ up_read(&bdi->wb_switch_rwsem);
}
/**
@@ -894,6 +924,9 @@ fs_initcall(cgroup_writeback_init);
#else /* CONFIG_CGROUP_WRITEBACK */
+static void bdi_down_write_wb_switch_rwsem(struct backing_dev_info *bdi) { }
+static void bdi_up_write_wb_switch_rwsem(struct backing_dev_info *bdi) { }
+
static struct bdi_writeback *
locked_inode_to_wb_and_lock_list(struct inode *inode)
__releases(&inode->i_lock)
@@ -2408,8 +2441,11 @@ void sync_inodes_sb(struct super_block *
return;
WARN_ON(!rwsem_is_locked(&sb->s_umount));
+ /* protect against inode wb switch, see inode_switch_wbs_work_fn() */
+ bdi_down_write_wb_switch_rwsem(bdi);
bdi_split_work_to_wbs(bdi, &work, false);
wb_wait_for_completion(bdi, &done);
+ bdi_up_write_wb_switch_rwsem(bdi);
wait_sb_inodes(sb);
}
--- a/include/linux/backing-dev-defs.h
+++ b/include/linux/backing-dev-defs.h
@@ -157,6 +157,7 @@ struct backing_dev_info {
struct radix_tree_root cgwb_tree; /* radix tree of active cgroup wbs */
struct rb_root cgwb_congested_tree; /* their congested states */
atomic_t usage_cnt; /* counts both cgwbs and cgwb_contested's */
+ struct rw_semaphore wb_switch_rwsem; /* no cgwb switch while syncing */
#else
struct bdi_writeback_congested *wb_congested;
#endif
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c
@@ -669,6 +669,7 @@ static int cgwb_bdi_init(struct backing_
INIT_RADIX_TREE(&bdi->cgwb_tree, GFP_ATOMIC);
bdi->cgwb_congested_tree = RB_ROOT;
atomic_set(&bdi->usage_cnt, 1);
+ init_rwsem(&bdi->wb_switch_rwsem);
ret = wb_init(&bdi->wb, bdi, 1, GFP_KERNEL);
if (!ret) {
next prev parent reply other threads:[~2019-05-20 12:53 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-20 12:13 [PATCH 4.9 00/44] 4.9.178-stable review Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 01/44] net: core: another layer of lists, around PF_MEMALLOC skb handling Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 02/44] locking/rwsem: Prevent decrement of reader count before increment Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 03/44] PCI: hv: Fix a memory leak in hv_eject_device_work() Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 04/44] x86/speculation/mds: Revert CPU buffer clear on double fault exit Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 05/44] x86/speculation/mds: Improve CPU buffer clear documentation Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 06/44] objtool: Fix function fallthrough detection Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 07/44] ARM: exynos: Fix a leaked reference by adding missing of_node_put Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 08/44] power: supply: axp288_charger: Fix unchecked return value Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 09/44] arm64: compat: Reduce address limit Greg Kroah-Hartman
2019-05-20 12:13 ` [PATCH 4.9 10/44] arm64: Clear OSDLR_EL1 on CPU boot Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 11/44] sched/x86: Save [ER]FLAGS on context switch Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 12/44] crypto: chacha20poly1305 - set cra_name correctly Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 13/44] crypto: vmx - fix copy-paste error in CTR mode Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 14/44] crypto: crct10dif-generic - fix use via crypto_shash_digest() Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 15/44] crypto: x86/crct10dif-pcl " Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 16/44] ALSA: usb-audio: Fix a memory leak bug Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 17/44] ALSA: hda/hdmi - Read the pin sense from register when repolling Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 18/44] ALSA: hda/hdmi - Consider eld_valid when reporting jack event Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 19/44] ALSA: hda/realtek - EAPD turn on later Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 20/44] ASoC: max98090: Fix restore of DAPM Muxes Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 21/44] ASoC: RT5677-SPI: Disable 16Bit SPI Transfers Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 22/44] mm/mincore.c: make mincore() more conservative Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 23/44] ocfs2: fix ocfs2 read inode data panic in ocfs2_iget Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 24/44] mfd: da9063: Fix OTP control register names to match datasheets for DA9063/63L Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 25/44] mfd: max77620: Fix swapped FPS_PERIOD_MAX_US values Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 26/44] tty/vt: fix write/write race in ioctl(KDSKBSENT) handler Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 27/44] jbd2: check superblock mapped prior to committing Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 28/44] ext4: actually request zeroing of inode table after grow Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 29/44] ext4: fix ext4_show_options for file systems w/o journal Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 30/44] Btrfs: do not start a transaction at iterate_extent_inodes() Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 31/44] bcache: fix a race between cache register and cacheset unregister Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 32/44] bcache: never set KEY_PTRS of journal key to 0 in journal_reclaim() Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 33/44] ipmi:ssif: compare block number correctly for multi-part return messages Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 34/44] crypto: gcm - Fix error return code in crypto_gcm_create_common() Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 35/44] crypto: gcm - fix incompatibility between "gcm" and "gcm_base" Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 36/44] crypto: salsa20 - dont access already-freed walk.iv Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 37/44] crypto: arm/aes-neonbs " Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 38/44] fib_rules: fix error in backport of e9919a24d302 ("fib_rules: return 0...") Greg Kroah-Hartman
2019-05-20 12:14 ` Greg Kroah-Hartman [this message]
2019-05-20 12:14 ` [PATCH 4.9 40/44] fs/writeback.c: use rcu_barrier() to wait for inflight wb switches going into workqueue when umount Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 41/44] ext4: zero out the unused memory region in the extent tree block Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 42/44] ext4: fix data corruption caused by overlapping unaligned and aligned IO Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 43/44] ALSA: hda/realtek - Fix for Lenovo B50-70 inverted internal microphone bug Greg Kroah-Hartman
2019-05-20 12:14 ` [PATCH 4.9 44/44] KVM: x86: Skip EFER vs. guest CPUID checks for host-initiated writes Greg Kroah-Hartman
2019-05-20 19:28 ` [PATCH 4.9 00/44] 4.9.178-stable review kernelci.org bot
2019-05-21 8:51 ` Jon Hunter
2019-05-21 10:34 ` Naresh Kamboju
2019-05-21 16:46 ` Greg Kroah-Hartman
2019-05-21 21:39 ` shuah
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=20190520115235.564183757@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=axboe@kernel.dk \
--cc=jack@suse.cz \
--cc=linux-kernel@vger.kernel.org \
--cc=stable@vger.kernel.org \
--cc=tj@kernel.org \
--cc=xuejiufei@gmail.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).