linux-kernel - Re: [f2fs-dev] [PATCH 2/2 v2] f2fs: avoid an infinite loop in f2fs_sync_dirty

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <YhkpxevY0ZGvZ23I@google.com>
Date:   Fri, 25 Feb 2022 11:11:01 -0800
From:   Jaegeuk Kim <jaegeuk@...nel.org>
To:     Chao Yu <chao@...nel.org>
Cc:     linux-kernel@...r.kernel.org,
        linux-f2fs-devel@...ts.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH 2/2 v2] f2fs: avoid an infinite loop in
 f2fs_sync_dirty_inodes

On 02/25, Chao Yu wrote:
> On 2022/2/15 7:27, Jaegeuk Kim wrote:
> > If one read IO is always failing, we can fall into an infinite loop in
> > f2fs_sync_dirty_inodes. This happens during xfstests/generic/475.
> > 
> > [  142.803335] Buffer I/O error on dev dm-1, logical block 8388592, async page read
> > ...
> > [  382.887210]  submit_bio_noacct+0xdd/0x2a0
> > [  382.887213]  submit_bio+0x80/0x110
> > [  382.887223]  __submit_bio+0x4d/0x300 [f2fs]
> > [  382.887282]  f2fs_submit_page_bio+0x125/0x200 [f2fs]
> > [  382.887299]  __get_meta_page+0xc9/0x280 [f2fs]
> > [  382.887315]  f2fs_get_meta_page+0x13/0x20 [f2fs]
> > [  382.887331]  f2fs_get_node_info+0x317/0x3c0 [f2fs]
> > [  382.887350]  f2fs_do_write_data_page+0x327/0x6f0 [f2fs]
> > [  382.887367]  f2fs_write_single_data_page+0x5b7/0x960 [f2fs]
> > [  382.887386]  f2fs_write_cache_pages+0x302/0x890 [f2fs]
> > [  382.887405]  ? preempt_count_add+0x7a/0xc0
> > [  382.887408]  f2fs_write_data_pages+0xfd/0x320 [f2fs]
> > [  382.887425]  ? _raw_spin_unlock+0x1a/0x30
> > [  382.887428]  do_writepages+0xd3/0x1d0
> > [  382.887432]  filemap_fdatawrite_wbc+0x69/0x90
> > [  382.887434]  filemap_fdatawrite+0x50/0x70
> > [  382.887437]  f2fs_sync_dirty_inodes+0xa4/0x270 [f2fs]
> > [  382.887453]  f2fs_write_checkpoint+0x189/0x1640 [f2fs]
> > [  382.887469]  ? schedule_timeout+0x114/0x150
> > [  382.887471]  ? ttwu_do_activate+0x6d/0xb0
> > [  382.887473]  ? preempt_count_add+0x7a/0xc0
> > [  382.887476]  kill_f2fs_super+0xca/0x100 [f2fs]
> > [  382.887491]  deactivate_locked_super+0x35/0xa0
> > [  382.887494]  deactivate_super+0x40/0x50
> > [  382.887497]  cleanup_mnt+0x139/0x190
> > [  382.887499]  __cleanup_mnt+0x12/0x20
> > [  382.887501]  task_work_run+0x64/0xa0
> > [  382.887505]  exit_to_user_mode_prepare+0x1b7/0x1c0
> > [  382.887508]  syscall_exit_to_user_mode+0x27/0x50
> > [  382.887510]  do_syscall_64+0x48/0xc0
> > [  382.887513]  entry_SYSCALL_64_after_hwframe+0x44/0xae
> > 
> > Signed-off-by: Jaegeuk Kim <jaegeuk@...nel.org>
> > ---
> >   Change log from v1:
> >    - fix a regression to report EIO too early
> > 
> >   fs/f2fs/checkpoint.c | 13 ++++++++-----
> >   fs/f2fs/f2fs.h       |  3 +++
> >   2 files changed, 11 insertions(+), 5 deletions(-)
> > 
> > diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
> > index 203a1577942d..56c81c68ef71 100644
> > --- a/fs/f2fs/checkpoint.c
> > +++ b/fs/f2fs/checkpoint.c
> > @@ -1059,13 +1059,13 @@ int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type)
> >   	struct inode *inode;
> >   	struct f2fs_inode_info *fi;
> >   	bool is_dir = (type == DIR_INODE);
> > -	unsigned long ino = 0;
> > +	unsigned long ino = 0, retry_count = DEFAULT_RETRY_SYNC_DIR_COUNT;
> >   	trace_f2fs_sync_dirty_inodes_enter(sbi->sb, is_dir,
> >   				get_pages(sbi, is_dir ?
> >   				F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA));
> >   retry:
> > -	if (unlikely(f2fs_cp_error(sbi))) {
> > +	if (unlikely(f2fs_cp_error(sbi) || (is_dir && !retry_count))) {
> >   		trace_f2fs_sync_dirty_inodes_exit(sbi->sb, is_dir,
> >   				get_pages(sbi, is_dir ?
> >   				F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA));
> > @@ -1096,10 +1096,13 @@ int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type)
> >   		iput(inode);
> >   		/* We need to give cpu to another writers. */
> > -		if (ino == cur_ino)
> > -			cond_resched();
> > -		else
> > +		if (ino == cur_ino) {
> > +			retry_count--;
> > +			io_schedule_timeout(DEFAULT_IO_TIMEOUT);
> > +		} else {
> > +			retry_count = DEFAULT_RETRY_SYNC_DIR_COUNT;
> >   			ino = cur_ino;
> > +		}
> >   	} else {
> >   		/*
> >   		 * We should submit bio, since it exists several
> > diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> > index c9515c3c54fd..f40ef7b61965 100644
> > --- a/fs/f2fs/f2fs.h
> > +++ b/fs/f2fs/f2fs.h
> > @@ -577,6 +577,9 @@ enum {
> >   /* maximum retry quota flush count */
> >   #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT		8
> > +/* maximum retry sync dirty inodes */
> > +#define DEFAULT_RETRY_SYNC_DIR_COUNT	3000
> 
> 3000 * 20ms/round = 60sec
> 
> How about just trying 5 or 10 sec?

It seems this causes another EIO issue in other test. Let me drop this for now.

> 
> Thanks,
> 
> > +
> >   #define F2FS_LINK_MAX	0xffffffff	/* maximum link count per file */
> >   #define MAX_DIR_RA_PAGES	4	/* maximum ra pages of dir */