Re: [PATCH] ext4: take i_mutex in ext4_symlink to eliminate a warning from ext4_truncate

[Zheng Liu <gnehzuil.liu@gmail.com>]

On Wed, Mar 27, 2013 at 09:51:55AM -0400, Theodore Ts'o wrote:
> Ah, now I see.  Thanks for sending the stack trace.  On the failure
> path, we're calling the inline function ext4_truncate_filaed_write()
> and this is calling ext4_truncate().
> 
> But I'm now wondering if we need to take the i_data_sem mutex in
> ext4_truncate_failed_write().
> 
> Otherwise, couldn't we end up with problems where a failed write calls
> ext4_truncate() without i_data_sem(), and that races with something
> else --- say, a punch or truncate call to that same inode?

I don't think we need to take i_mutex lock honestly.  In ext4_symlink
when we call __page_symlink() the new inode doesn't access yet.  So no
one can do a punching hole or truncation to this inode.  But I also
think we need to add WARN_ON in ext4_truncate because i_mutex lock is
used to serialize truncate/punch hole and buffered io.

Regards,
                                                - Zheng