2010-11-11 15:04:02

by Fabio Estevam

Subject: [PATCH] mfd: check for NULL platform data

Avoid kernel crash if platform data is NULL.

Signed-off-by: Fabio Estevam <[email protected]>
---
drivers/mfd/mc13xxx-core.c | 3 +++
1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/drivers/mfd/mc13xxx-core.c b/drivers/mfd/mc13xxx-core.c
index a2ac2ed..b4d6bb1 100644
--- a/drivers/mfd/mc13xxx-core.c
+++ b/drivers/mfd/mc13xxx-core.c
@@ -757,6 +757,9 @@ err_revision:

mc13xxx_unlock(mc13xxx);

+ if (pdata == NULL)
+ return 0;
+
if (pdata->flags & MC13XXX_USE_ADC)
mc13xxx_add_subdevice(mc13xxx, "%s-adc");

--
1.6.0.4

2010-11-11 15:44:10

by Uwe Kleine-König

Subject: Re: [PATCH] mfd: check for NULL platform data

Hi Fabio,

On Thu, Nov 11, 2010 at 06:58:14AM -0800, Fabio Estevam wrote:
> Avoid kernel crash if platform data is NULL.
>
> Signed-off-by: Fabio Estevam <[email protected]>
> ---
> drivers/mfd/mc13xxx-core.c | 3 +++
> 1 files changed, 3 insertions(+), 0 deletions(-)
>
> diff --git a/drivers/mfd/mc13xxx-core.c b/drivers/mfd/mc13xxx-core.c
> index a2ac2ed..b4d6bb1 100644
> --- a/drivers/mfd/mc13xxx-core.c
> +++ b/drivers/mfd/mc13xxx-core.c
> @@ -757,6 +757,9 @@ err_revision:
>
> mc13xxx_unlock(mc13xxx);
>
> + if (pdata == NULL)
> + return 0;
> +
I'm not sure it's really needed to catch this error. Not passing pdata
isn't sensible and then maybe failing with a bang is better than
handling silently.

And if you want to break probing, do you really want to return 0, i.e.
let the binding succeed? IMHO (if you really want to handle pdata ==
NULL) you should fail before allocating the private data with
-ESOMETHINGSENSIBLE.

Thanks,
Uwe

--
Pengutronix e.K. | Uwe Kleine-K?nig |
Industrial Linux Solutions | http://www.pengutronix.de/ |

2010-11-26 10:23:44

by Samuel Ortiz

Subject: Re: [PATCH] mfd: check for NULL platform data

Hi Uwe,

On Thu, Nov 11, 2010 at 04:44:07PM +0100, Uwe Kleine-K?nig wrote:
> Hi Fabio,
>
> On Thu, Nov 11, 2010 at 06:58:14AM -0800, Fabio Estevam wrote:
> > Avoid kernel crash if platform data is NULL.
> >
> > Signed-off-by: Fabio Estevam <[email protected]>
> > ---
> > drivers/mfd/mc13xxx-core.c | 3 +++
> > 1 files changed, 3 insertions(+), 0 deletions(-)
> >
> > diff --git a/drivers/mfd/mc13xxx-core.c b/drivers/mfd/mc13xxx-core.c
> > index a2ac2ed..b4d6bb1 100644
> > --- a/drivers/mfd/mc13xxx-core.c
> > +++ b/drivers/mfd/mc13xxx-core.c
> > @@ -757,6 +757,9 @@ err_revision:
> >
> > mc13xxx_unlock(mc13xxx);
> >
> > + if (pdata == NULL)
> > + return 0;
> > +
> I'm not sure it's really needed to catch this error. Not passing pdata
> isn't sensible and then maybe failing with a bang is better than
> handling silently.
>
> And if you want to break probing, do you really want to return 0, i.e.
> let the binding succeed? IMHO (if you really want to handle pdata ==
> NULL) you should fail before allocating the private data with
> -ESOMETHINGSENSIBLE.
Agreed. Besides failing silently, Fabio's patch leaks the mc13xxx pointer and
doesn't free the requested IRQ line.
So NAK for now, although I wouldn't be against checking for a NULL pdata at
the very begining of the probe routine.

Cheers,
Samuel.

--
Intel Open Source Technology Centre
http://oss.intel.com/