Re: [PATCH 1/1] vfs: umount_begin BKL pushdown v2

[Christoph Hellwig <hch@infradead.org>]

On Fri, Apr 24, 2009 at 08:15:36AM +0100, Al Viro wrote:
> On Fri, Apr 24, 2009 at 08:13:12AM +0100, Al Viro wrote:
> > On Fri, Apr 24, 2009 at 09:06:53AM +0200, Alessio Igor Bogani wrote:
> > >  static void cifs_umount_begin(struct super_block *sb)
> > >  {
> > > -	struct cifs_sb_info *cifs_sb = CIFS_SB(sb);
> > > +	struct cifs_sb_info *cifs_sb;
> > >  	struct cifsTconInfo *tcon;
> > >  
> > > -	if (cifs_sb == NULL)
> > > +	lock_kernel();
> > > +	cifs_sb = CIFS_SB(sb);
> > > +
> > > +	if (cifs_sb == NULL) {
> > > +		unlock_kernel();
> > >  		return;
> > > +	}
> > >  
> > >  	tcon = cifs_sb->tcon;
> > > -	if (tcon == NULL)
> > > +	if (tcon == NULL) {
> > > +		unlock_kernel();
> > >  		return;
> > > +	}
> > 
> > AFAICS, both CIFS_SB(sb) and ->tcon are assign-once, so lock_kernel() should
> > really go here (if it can't be removed completely, of course, but that's up
> > to CIFS folks).  Applied with such modification.
> 
> PS: I suspect that checks for NULL are actually "what if kernel memory got
> corrupted", but I'm too lazy to verify that at the moment; again, up to
> CIFS folks.

NULL checks are superflous here, but that should be a separate patch.