On Fri, 22 Jun 2018, Dmitry Torokhov wrote:
> On Fri, Jun 22, 2018 at 01:42:28PM +0200, Matthias Brugger wrote:
> > The drivers gets probed from a mfd devices. So the driver runs
> > probe although no DT node exists. This leads to a NULL pointer
> > dereference in the probe function. Check if a node exists and
> > error out in case none is present.
>
> Hmm, if a cell specifies of_compatible and no such node is present in DT
> maybe we should bail out in mfd_add_device() instead of pushing the
> check into individual drivers? Something like:
>
> diff --git a/drivers/mfd/mfd-core.c b/drivers/mfd/mfd-core.c
> index 94e3f32ce935..36a5fd98000d 100644
> --- a/drivers/mfd/mfd-core.c
> +++ b/drivers/mfd/mfd-core.c
> @@ -137,6 +137,18 @@ static inline void mfd_acpi_add_device(const struct
> mfd_cell *cell,
> }
> #endif
>
> +static struct device_node *mfd_get_node_for_cell(struct device *parent,
> + const struct mfd_cell *cell)
> +{
> + struct device_node *np;
> +
> + for_each_child_of_node(parent->of_node, np)
> + if (of_device_is_compatible(np, cell->of_compatible))
> + return np;
> +
> + return NULL;
> +}
> +
> static int mfd_add_device(struct device *parent, int id,
> const struct mfd_cell *cell, atomic_t *usage_count,
> struct resource *mem_base,
> @@ -144,7 +156,6 @@ static int mfd_add_device(struct device *parent, int id,
> {
> struct resource *res;
> struct platform_device *pdev;
> - struct device_node *np = NULL;
> int ret = -ENOMEM;
> int platform_id;
> int r;
> @@ -176,11 +187,11 @@ static int mfd_add_device(struct device *parent, int id,
> goto fail_res;
>
> if (parent->of_node && cell->of_compatible) {
> - for_each_child_of_node(parent->of_node, np) {
> - if (of_device_is_compatible(np, cell->of_compatible)) {
> - pdev->dev.of_node = np;
> - break;
> - }
> + pdev->dev.of_node = mfd_get_node_for_cell(parent, cell);
> + if (!pdev->dev.of_node) {
> + dev_err(parent, "unable to locate node for %s (%s)\n",
> + cell->name, cell->of_compatible);
> + goto fail_res;
> }
> }
>
> Lee?
This change results the wrong semantics. MFD overrides Device Tree,
such that if a device is supplied and registered using
mfd_add_device(s) it should be probed. The only reason we provide a
pointer to the of_node is for the child device to pull properties
from.
If a device should only be probed if an OF node exists, then it should
not be registered at the MFD level.
> > Fixes: 3e9f0b3e2b27 ("input: Add MediaTek PMIC keys support")
> > Signed-off-by: Matthias Brugger <[email protected]>
> > ---
> > drivers/input/keyboard/mtk-pmic-keys.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/input/keyboard/mtk-pmic-keys.c
> > b/drivers/input/keyboard/mtk-pmic-keys.c
> > index 02c67a1749fc..388043e1939c 100644
> > --- a/drivers/input/keyboard/mtk-pmic-keys.c
> > +++ b/drivers/input/keyboard/mtk-pmic-keys.c
> > @@ -257,6 +257,9 @@ static int mtk_pmic_keys_probe(struct platform_device
> > *pdev)
> > const struct of_device_id *of_id =
> > of_match_device(of_mtk_pmic_keys_match_tbl, &pdev->dev);
> >
> > + if (of_id == NULL)
> > + return -ENODEV;
> > +
> > keys = devm_kzalloc(&pdev->dev, sizeof(*keys), GFP_KERNEL);
> > if (!keys)
> > return -ENOMEM;
>
> Thanks.
>
--
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
