Re: [PATCH] drm/mm: avoid possible null pointer dereference

[Eric Engestrom]

On Mon, May 23, 2016 at 02:46:59PM +0100, Chris Wilson wrote:
> On Mon, May 23, 2016 at 02:38:29PM +0100, Eric Engestrom wrote:
> > On Mon, May 23, 2016 at 02:02:26PM +0100, Chris Wilson wrote:
> > > On Mon, May 23, 2016 at 02:56:45PM +0200, Daniel Vetter wrote:
> > > > On Mon, May 23, 2016 at 11:27:14AM +0100, Eric Engestrom wrote:
> > > > > On Wed, May 18, 2016 at 10:17:19PM +0200, Heinrich Schuchardt wrote:
> > > > > > Do not dereference node before the check if node is NULL.
> > > > > > 
> > > > > > Signed-off-by: Heinrich Schuchardt <xypron.glpk@xxxxxx>
> > > > > > ---
> > > > > >  drivers/gpu/drm/drm_mm.c | 4 +++-
> > > > > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > > > > > 
> > > > > > diff --git a/drivers/gpu/drm/drm_mm.c b/drivers/gpu/drm/drm_mm.c
> > > > > > index 04de6fd..cb39f45 100644
> > > > > > --- a/drivers/gpu/drm/drm_mm.c
> > > > > > +++ b/drivers/gpu/drm/drm_mm.c
> > > > > > @@ -179,12 +179,14 @@ static void drm_mm_insert_helper(struct drm_mm_node *hole_node,
> > > > > >  int drm_mm_reserve_node(struct drm_mm *mm, struct drm_mm_node *node)
> > > > > >  {
> > > > > >  	struct drm_mm_node *hole;
> > > > > > -	u64 end = node->start + node->size;
> > > > > > +	u64 end;
> > > > > >  	u64 hole_start;
> > > > > >  	u64 hole_end;
> > > > > >  
> > > > > >  	BUG_ON(node == NULL);
> > > > > >  
> > > > > > +	end = node->start + node->size;
> > > > > > +
> > > > > >  	/* Find the relevant hole to add our node to */
> > > > > >  	drm_mm_for_each_hole(hole, mm, hole_start, hole_end) {
> > > > > >  		if (hole_start > node->start || hole_end < end)
> > > > > > -- 
> > > > > > 2.1.4
> > > > > 
> > > > > Reviewed-by: Eric Engestrom <eric.engestrom@xxxxxxxxxx>
> > > 
> > > Remove useless check instead?
> > > -Chris
> > 
> > I tend to prefer erring on the side of caution and have (too) many checks,
> > especially simple ones like avoiding null dereferences.
> 
> BUG(node == NULL) is no more informative than hitting a GFP with *node.

Good point.

> 
> > That said, BUG() might be too extreme. I'm not all that familiar with
> > the code, but it doesn't seem like it would be an unrecoverable failure.
> > WARN() instead?
> 
> It's a programming error, just as would be passing in mm == NULL. Mark up
> the function as requiring non-NULL parameters.
> -Chris

Using `__attribute__((nonnull))`? I approve of that idea, as I find
those attributes extremely useful and try to get them used as much as
possible at $DAYJOB.

I have noticed however that this particular attribute isn't used much in
the kernel (4 times that I could find with a simple grep), to the point
of not even having a `__nonnull`. Any reason you know of?

Cheers,
  Eric