linux-kernel - Re: [PATCH] f2fs: fix to determine start_cp_addr by sbi->cur_cp

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:   Mon, 12 Dec 2016 10:56:14 -0800
From:   Jaegeuk Kim <jaegeuk@...nel.org>
To:     Chao Yu <yuchao0@...wei.com>
Cc:     linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        linux-f2fs-devel@...ts.sourceforge.net, stable@...r.kernel.org
Subject: Re: [PATCH] f2fs: fix to determine start_cp_addr by sbi->cur_cp_pack

On 12/10, Chao Yu wrote:
> Hi Jaegeuk,
> 
> Let me try to understand this, in some cases, we can write a checkpoint pack
> which has wrong cp_ver, like in 1st cp pack which has even version number or 2nd
> cp pack which has odd version number, so if we load that kind of cp pack during
> fill_super, we may load wrong summary data from another older cp pack which cp
> version pointed to, is that right?

Yup. Especially we can use __start_cp_addr() when finding summary blocks. So, I
suspect it was able to cause missing summary blocks.

Thanks,

> 
> Thanks,
> 
> On 2016/11/25 10:32, Jaegeuk Kim wrote:
> > We don't guarantee cp_addr is fixed by cp_version.
> > This is to sync with f2fs-tools.
> > 
> > Cc: stable@...r.kernel.org
> > Signed-off-by: Jaegeuk Kim <jaegeuk@...nel.org>
> > ---
> >  fs/f2fs/checkpoint.c |  5 +++++
> >  fs/f2fs/f2fs.h       | 14 +++-----------
> >  2 files changed, 8 insertions(+), 11 deletions(-)
> > 
> > diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
> > index 54cc6a9..bf2f44c 100644
> > --- a/fs/f2fs/checkpoint.c
> > +++ b/fs/f2fs/checkpoint.c
> > @@ -789,6 +789,11 @@ int get_valid_checkpoint(struct f2fs_sb_info *sbi)
> >  		f2fs_put_page(cur_page, 1);
> >  	}
> >  done:
> > +	if (cur_page == cp1)
> > +		sbi->cur_cp_pack = 1;
> > +	else
> > +		sbi->cur_cp_pack = 2;
> > +
> >  	f2fs_put_page(cp1, 1);
> >  	f2fs_put_page(cp2, 1);
> >  	return 0;
> > diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> > index 62383d2..e22e7e1 100644
> > --- a/fs/f2fs/f2fs.h
> > +++ b/fs/f2fs/f2fs.h
> > @@ -792,6 +792,7 @@ struct f2fs_sb_info {
> >  
> >  	/* for checkpoint */
> >  	struct f2fs_checkpoint *ckpt;		/* raw checkpoint pointer */
> > +	int cur_cp_pack;			/* remain current cp pack */
> >  	spinlock_t cp_lock;			/* for flag in ckpt */
> >  	struct inode *meta_inode;		/* cache meta blocks */
> >  	struct mutex cp_mutex;			/* checkpoint procedure lock */
> > @@ -1352,19 +1353,10 @@ static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
> >  
> >  static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
> >  {
> > -	block_t start_addr;
> > -	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
> > -	unsigned long long ckpt_version = cur_cp_version(ckpt);
> > -
> > -	start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
> > +	block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
> >  
> > -	/*
> > -	 * odd numbered checkpoint should at cp segment 0
> > -	 * and even segment must be at cp segment 1
> > -	 */
> > -	if (!(ckpt_version & 1))
> > +	if (sbi->cur_cp_pack == 2)
> >  		start_addr += sbi->blocks_per_seg;
> > -
> >  	return start_addr;
> >  }
> >  
> >