Re: [PATCH v11 2/5] ext4: introduce ext4_put_ea_inode() for safe deferred iput

[Jan Kara <jack@suse.cz>]

On Mon 29-06-26 19:08:45, Yun Zhou wrote:
> Calling iput() on EA inodes while holding xattr_sem or a jbd2 handle
> can trigger write_inode_now() -> ext4_writepages() -> s_writepages_rwsem,
> creating a lock ordering issue during mount (!SB_ACTIVE).
> 
> Add ext4_put_ea_inode() which uses iput_if_not_last() as a fast path.
> If this is not the last reference, it is dropped immediately.  If this
> is the last reference, the inode is linked onto a per-sb lock-free llist
> via i_ea_iput_node (embedded in ext4_inode_info, sharing space with the
> unused xattr_sem of EA inodes via a union) and a delayed worker
> (1 jiffie) performs the final iput() in a clean context.  This avoids
> per-iput memory allocation.
> 
> Flush points are placed before quota shutdown (ext4_put_super and
> failed_mount9) and before freeing structures that eviction depends on
> (failed_mount_wq and failed_mount3a).  Initialization is placed before
> journal loading since fast commit replay may trigger evictions that call
> ext4_put_ea_inode().
> 
> Also moves init_rwsem(xattr_sem) from init_once to ext4_alloc_inode to
> handle slab object reuse after the union field has been overwritten.
> 
> Signed-off-by: Yun Zhou <yun.zhou@windriver.com>
> Suggested-by: Jan Kara <jack@suse.cz>

...

> +/*
> + * Release a VFS reference on an EA inode.  Must be used instead of iput()
> + * in any context where xattr_sem or a jbd2 handle is held.
> + *
> + * If this is not the last reference, drops it immediately via
> + * iput_if_not_last() with no further action needed.
> + *
> + * If this is the last reference, the inode is linked onto a per-sb
> + * llist via i_ea_iput_node (embedded in ext4_inode_info, sharing space
> + * with the unused xattr_sem) and a delayed worker performs the final
> + * iput() in a clean context.
> + *
> + * Note: while an inode is on s_ea_inode_to_free, the unconsumed i_count
> + * reference (still 1) keeps it in the inode cache, so any concurrent
> + * iget() bumps i_count to >= 2 and iput_if_not_last() will succeed.
> + * Nobody will add the inode a second time until ext4_ea_inode_work()
> + * drops that reference via iput().
> + */
> +void ext4_put_ea_inode(struct super_block *sb, struct inode *inode)

One small question here: Any reason why you explicitely pass sb here
instead of using inode->i_sb? EA inodes are guaranteed to be from the same
superblock... It would somewhat simplify the callers.

								Honza

-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR