Re: [PATCH] f2fs: fix to do sanity check for inline inode

[Jaegeuk Kim]

On 05/05, Chao Yu wrote:
> On 2022/5/5 5:28, Jaegeuk Kim wrote:
> > On 04/28, Chao Yu wrote:
> > > As Yanming reported in bugzilla:
> > > 
> > > https://bugzilla.kernel.org/show_bug.cgi?id=215895
> > > 
> > > I have encountered a bug in F2FS file system in kernel v5.17.
> > > 
> > > The kernel message is shown below:
> > > 
> > > kernel BUG at fs/inode.c:611!
> > > Call Trace:
> > >   evict+0x282/0x4e0
> > >   __dentry_kill+0x2b2/0x4d0
> > >   dput+0x2dd/0x720
> > >   do_renameat2+0x596/0x970
> > >   __x64_sys_rename+0x78/0x90
> > >   do_syscall_64+0x3b/0x90
> > > 
> > > The root cause is: fuzzed inode has both inline_data flag and encrypted
> > > flag, so after it was deleted by rename(), during f2fs_evict_inode(),
> > > it will cause inline data conversion due to flags confilction, then
> > > page cache will be polluted and trigger panic in clear_inode().
> > > 
> > > This patch tries to fix the issue by do more sanity checks for inline
> > > data inode in sanity_check_inode().
> > > 
> > > Cc: stable@xxxxxxxxxxxxxxx
> > > Reported-by: Ming Yan <yanming@xxxxxxxxxx>
> > > Signed-off-by: Chao Yu <chao.yu@xxxxxxxx>
> > > ---
> > >   fs/f2fs/f2fs.h  | 7 +++++++
> > >   fs/f2fs/inode.c | 3 +--
> > >   2 files changed, 8 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> > > index 27aa93caec06..64c511b498cc 100644
> > > --- a/fs/f2fs/f2fs.h
> > > +++ b/fs/f2fs/f2fs.h
> > > @@ -4173,6 +4173,13 @@ static inline void f2fs_set_encrypted_inode(struct inode *inode)
> > >    */
> > >   static inline bool f2fs_post_read_required(struct inode *inode)
> > >   {
> > > +	/*
> > > +	 * used by sanity_check_inode(), when disk layout fields has not
> > > +	 * been synchronized to inmem fields.
> > > +	 */
> > > +	if (file_is_encrypt(inode) || file_is_verity(inode) ||
> > > +			F2FS_I(inode)->i_flags & F2FS_COMPR_FL)
> > > +		return true;
> > >   	return f2fs_encrypted_file(inode) || fsverity_active(inode) ||
> > >   		f2fs_compressed_file(inode);
> > >   }
> > > diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c
> > > index 83639238a1fe..234b8ed02644 100644
> > > --- a/fs/f2fs/inode.c
> > > +++ b/fs/f2fs/inode.c
> > > @@ -276,8 +276,7 @@ static bool sanity_check_inode(struct inode *inode, struct page *node_page)
> > >   		}
> > >   	}
> > > -	if (f2fs_has_inline_data(inode) &&
> > > -			(!S_ISREG(inode->i_mode) && !S_ISLNK(inode->i_mode))) {
> > > +	if (f2fs_has_inline_data(inode) && !f2fs_may_inline_data(inode)) {
> > 
> > It seems f2fs_may_inline_data() is breaking the atomic write case. Please fix.
> 
> sanity_check_inode() change only affect f2fs_iget(), during inode initialization,
> file should not be set as atomic one, right?
> 
> I didn't see any failure during 'f2fs_io write atomic_write' testcase... could you
> please provide me detail of the testcase?

I just applied this into my device and was getting lots of the below error
messages resulting in open failures of database files.

> 
> Thanks,
> 
> > 
> > >   		set_sbi_flag(sbi, SBI_NEED_FSCK);
> > >   		f2fs_warn(sbi, "%s: inode (ino=%lx, mode=%u) should not have inline_data, run fsck to fix",
> > >   			  __func__, inode->i_ino, inode->i_mode);
> > > -- 
> > > 2.25.1