In hx83102_get_modes(), the return value of drm_mode_duplicate()
is assigned to mode, which will lead to a possible NULL pointer
dereference on failure of drm_mode_duplicate(). Even though a
small allocation failing is basically impossible, kernel policy
is still to check for NULL so add the check.

Fixes: 0ef94554dc40 ("drm/panel: himax-hx83102: Break out as separate driver")
Signed-off-by: Charles Han <hanchunchao@inspur.com>
---
 drivers/gpu/drm/panel/panel-himax-hx83102.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/panel/panel-himax-hx83102.c b/drivers/gpu/drm/panel/panel-himax-hx83102.c
index 6e4b7e4644ce..e67555323d3b 100644
--- a/drivers/gpu/drm/panel/panel-himax-hx83102.c
+++ b/drivers/gpu/drm/panel/panel-himax-hx83102.c
@@ -565,6 +565,8 @@ static int hx83102_get_modes(struct drm_panel *panel,
 	struct drm_display_mode *mode;
 
 	mode = drm_mode_duplicate(connector->dev, m);
+	if (!mode)
+		return -EINVAL;
 
 	mode->type = DRM_MODE_TYPE_DRIVER | DRM_MODE_TYPE_PREFERRED;
 	drm_mode_set_name(mode);
-- 
2.31.1

Hi,

On Fri, Aug 23, 2024 at 9:35 AM Charles Han <hanchunchao@inspur.com> wrote:
>
> In hx83102_get_modes(), the return value of drm_mode_duplicate()
> is assigned to mode, which will lead to a possible NULL pointer
> dereference on failure of drm_mode_duplicate(). Even though a
> small allocation failing is basically impossible, kernel policy
> is still to check for NULL so add the check.
>
> Fixes: 0ef94554dc40 ("drm/panel: himax-hx83102: Break out as separate driver")
> Signed-off-by: Charles Han <hanchunchao@inspur.com>
> ---
>  drivers/gpu/drm/panel/panel-himax-hx83102.c | 2 ++
>  1 file changed, 2 insertions(+)

FWIW, this looks to be v4 of your patch, right? The subject line
should include a version number and you should be providing version
history "after the cut" in your patch. Tools like "b4" and "patman"
can help you get this correct [1]. If you plan to continue posting
patches you'll need to start getting this right. The next version of
your patch would be v5.

[1] https://sched.co/1aBGS

I see:

v1: https://lore.kernel.org/r/20240821095039.15282-1-hanchunchao@inspur.com
v2: https://lore.kernel.org/r/20240822093442.4262-1-hanchunchao@inspur.com
v3: https://lore.kernel.org/r/20240823083657.7100-1-hanchunchao@inspur.com


> diff --git a/drivers/gpu/drm/panel/panel-himax-hx83102.c b/drivers/gpu/drm/panel/panel-himax-hx83102.c
> index 6e4b7e4644ce..e67555323d3b 100644
> --- a/drivers/gpu/drm/panel/panel-himax-hx83102.c
> +++ b/drivers/gpu/drm/panel/panel-himax-hx83102.c
> @@ -565,6 +565,8 @@ static int hx83102_get_modes(struct drm_panel *panel,
>         struct drm_display_mode *mode;
>
>         mode = drm_mode_duplicate(connector->dev, m);
> +       if (!mode)
> +               return -EINVAL;

I would have returned -ENOMEM since drm_mode_duplicate() is defined to
allocate memory copy the mode (like strdup does for strings) and it
should be clear that the only failure case is failure to allocate
memory. Other callers convert a NULL return as -ENOMEM.

-Doug

Hi,

On Fri, Aug 23, 2024 at 9:49 AM Doug Anderson <dianders@chromium.org> wrote:
>
> Hi,
>
> On Fri, Aug 23, 2024 at 9:35 AM Charles Han <hanchunchao@inspur.com> wrote:
> >
> > In hx83102_get_modes(), the return value of drm_mode_duplicate()
> > is assigned to mode, which will lead to a possible NULL pointer
> > dereference on failure of drm_mode_duplicate(). Even though a
> > small allocation failing is basically impossible, kernel policy
> > is still to check for NULL so add the check.
> >
> > Fixes: 0ef94554dc40 ("drm/panel: himax-hx83102: Break out as separate driver")
> > Signed-off-by: Charles Han <hanchunchao@inspur.com>
> > ---
> >  drivers/gpu/drm/panel/panel-himax-hx83102.c | 2 ++
> >  1 file changed, 2 insertions(+)
>
> FWIW, this looks to be v4 of your patch, right? The subject line
> should include a version number and you should be providing version
> history "after the cut" in your patch. Tools like "b4" and "patman"
> can help you get this correct [1]. If you plan to continue posting
> patches you'll need to start getting this right. The next version of
> your patch would be v5.
>
> [1] https://sched.co/1aBGS
>
> I see:
>
> v1: https://lore.kernel.org/r/20240821095039.15282-1-hanchunchao@inspur.com
> v2: https://lore.kernel.org/r/20240822093442.4262-1-hanchunchao@inspur.com
> v3: https://lore.kernel.org/r/20240823083657.7100-1-hanchunchao@inspur.com
>
>
> > diff --git a/drivers/gpu/drm/panel/panel-himax-hx83102.c b/drivers/gpu/drm/panel/panel-himax-hx83102.c
> > index 6e4b7e4644ce..e67555323d3b 100644
> > --- a/drivers/gpu/drm/panel/panel-himax-hx83102.c
> > +++ b/drivers/gpu/drm/panel/panel-himax-hx83102.c
> > @@ -565,6 +565,8 @@ static int hx83102_get_modes(struct drm_panel *panel,
> >         struct drm_display_mode *mode;
> >
> >         mode = drm_mode_duplicate(connector->dev, m);
> > +       if (!mode)
> > +               return -EINVAL;
>
> I would have returned -ENOMEM since drm_mode_duplicate() is defined to
> allocate memory copy the mode (like strdup does for strings) and it
> should be clear that the only failure case is failure to allocate
> memory. Other callers convert a NULL return as -ENOMEM.

FWIW: if you spin v5 of this patch and have it return -ENOMEM then I'm
happy to apply it.

-Doug