Re: [PATCH v3 1/3] of: Move simple-framebuffer device handling from simplefb to of

[Hector Martin]

On 13/12/2021 17.44, Javier Martinez Canillas wrote:
> Hello Hector,
>
> On Sun, Dec 12, 2021 at 7:24 AM Hector Martin <marcan@xxxxxxxxx> wrote:
> > This code is required for both simplefb and simpledrm, so let's move it
> > into the OF core instead of having it as an ad-hoc initcall in the
> > drivers.
> >
> > Acked-by: Thomas Zimmermann <tzimmermann@xxxxxxx>
> > Signed-off-by: Hector Martin <marcan@xxxxxxxxx>
> > ---
> >   drivers/of/platform.c          |  4 ++++
> >   drivers/video/fbdev/simplefb.c | 21 +--------------------
> >   2 files changed, 5 insertions(+), 20 deletions(-)
>
> This is indeed a much better approach than what I suggested. I just
> have one comment.
>
> > diff --git a/drivers/of/platform.c b/drivers/of/platform.c
> > index b3faf89744aa..793350028906 100644
> > --- a/drivers/of/platform.c
> > +++ b/drivers/of/platform.c
> > @@ -540,6 +540,10 @@ static int __init of_platform_default_populate_init(void)
> >                  of_node_put(node);
> >          }
> >
> > +       node = of_get_compatible_child(of_chosen, "simple-framebuffer");
>
> You have to check if the node variable is NULL here.
>
> > +       of_platform_device_create(node, NULL, NULL);
>
> Otherwise this could lead to a NULL pointer dereference if debug
> output is enabled (the node->full_name is printed).

Where is it printed? I thought I might need a NULL check, but this code 
was suggested verbatim by Rob in v2 without the NULL check and digging 
through I found that the NULL codepath is safe.

of_platform_device_create calls of_platform_device_create_pdata 
directly, and:

static struct platform_device *of_platform_device_create_pdata(
                                        struct device_node *np,
                                        const char *bus_id,
                                        void *platform_data,
                                        struct device *parent)
{
        struct platform_device *dev;

        if (!of_device_is_available(np) ||
            of_node_test_and_set_flag(np, OF_POPULATED))
                return NULL;

of_device_is_available takes a global spinlock and then calls 
__of_device_is_available, and that does:

static bool __of_device_is_available(const struct device_node *device)
{
        const char *status;
        int statlen;

        if (!device)
                return false;

... so I don't see how this can do anything but immediately return false 
if node is NULL.

--
Hector Martin (marcan@xxxxxxxxx)
Public Key: https://mrcn.st/pub