Re: [PATCH 2/2] debugfs: only allow root access to debugging interfaces

[Greg KH <greg@kroah.com>]

On Fri, Feb 25, 2011 at 12:40:43PM -0800, Hugh Dickins wrote:
> On Fri, 25 Feb 2011, Greg KH wrote:
> > On Tue, Feb 22, 2011 at 11:29:15PM +0300, Dan Carpenter wrote:
> > > On Tue, Feb 22, 2011 at 11:33:14AM -0800, Greg KH wrote:
> > > > Again, let's fix the real problems here, world-writable debugfs files.
> > > 
> > > We could just ban them?
> > > 
> > > diff --git a/fs/debugfs/inode.c b/fs/debugfs/inode.c
> > > index e7a7a2f..03ae095 100644
> > > --- a/fs/debugfs/inode.c
> > > +++ b/fs/debugfs/inode.c
> > > @@ -223,6 +223,9 @@ struct dentry *debugfs_create_file(const char *name, mode_t mode,
> > >  
> > >  	pr_debug("debugfs: creating file '%s'\n",name);
> > >  
> > > +	/* don't allow world writable files */
> > > +	mode &= ~S_IWOTH;
> > > +
> > >  	error = simple_pin_fs(&debug_fs_type, &debugfs_mount,
> > >  			      &debugfs_mount_count);
> > >  	if (error)
> > > 
> > 
> > I have no objection to this patch, care to resend it with a
> > signed-off-by: so that I can apply it?
> 
> That's funny, I wrote something in debugfs a few months ago which was
> very deliberately rw--w--w-: a kind of circular trace buffer into which
> any userspace could easily echo its progress for debugging.
> 
> Probably won't be upstreamed in that form, nor use debugfs if it is.
> But I mention it as an example of why any such limitation on debugfs
> seems inappropriate to me.

Ok, fair enough, then I'll not add this patch, as you just proved it is
a valuable thing to have world writable debugfs files.

thanks,

greg k-h