[PATCH v2 0/2] fix a possible NULL dereference

[PATCH v2 0/2] fix a possible NULL dereference

[LABBE Corentin]

Hello

The main goal of this patch series is to fix a possible NULL dereference.
Even if the probability of this case is very low, fixing it made
static analyzers happy.
In the same time it permits to remove a "cast that drop const qualifiers.

Regards

Changes since v1
- Use of_device_get_match_data
- Add the missing patch for constify atmel_nand_caps structures

LABBE Corentin (2):
  mtd: nand: atmel_nand: constify atmel_nand_caps structures
  mtd: nand: atmel_nand: fix a possible NULL dereference

 drivers/mtd/nand/atmel_nand.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

-- 
2.4.10

[PATCH v2 1/2] mtd: nand: atmel_nand: constify atmel_nand_caps structures

[LABBE Corentin]

All atmel_nand_caps are never modified, consitify them.

Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
---
 drivers/mtd/nand/atmel_nand.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
index 583cdd9..475c938 100644
--- a/drivers/mtd/nand/atmel_nand.c
+++ b/drivers/mtd/nand/atmel_nand.c
@@ -128,7 +128,7 @@ struct atmel_nand_host {
 
 	struct atmel_nfc	*nfc;
 
-	struct atmel_nand_caps	*caps;
+	const struct atmel_nand_caps	*caps;
 	bool			has_pmecc;
 	u8			pmecc_corr_cap;
 	u16			pmecc_sector_size;
@@ -2304,11 +2304,11 @@ static int atmel_nand_remove(struct platform_device *pdev)
 	return 0;
 }
 
-static struct atmel_nand_caps at91rm9200_caps = {
+static const struct atmel_nand_caps at91rm9200_caps = {
 	.pmecc_correct_erase_page = false,
 };
 
-static struct atmel_nand_caps sama5d4_caps = {
+static const struct atmel_nand_caps sama5d4_caps = {
 	.pmecc_correct_erase_page = true,
 };
 
-- 
2.4.10

Re: [PATCH v2 1/2] mtd: nand: atmel_nand: constify atmel_nand_caps structures

[Josh Wu]

Hi, Corentin

Thanks for the patch.

On 11/20/2015 3:45 PM, LABBE Corentin wrote:
> All atmel_nand_caps are never modified, consitify them.
>
> Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>

Acked-by: Josh Wu <josh.wu@atmel.com>

Best Regards,
Josh Wu
> ---
>   drivers/mtd/nand/atmel_nand.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
> index 583cdd9..475c938 100644
> --- a/drivers/mtd/nand/atmel_nand.c
> +++ b/drivers/mtd/nand/atmel_nand.c
> @@ -128,7 +128,7 @@ struct atmel_nand_host {
>   
>   	struct atmel_nfc	*nfc;
>   
> -	struct atmel_nand_caps	*caps;
> +	const struct atmel_nand_caps	*caps;
>   	bool			has_pmecc;
>   	u8			pmecc_corr_cap;
>   	u16			pmecc_sector_size;
> @@ -2304,11 +2304,11 @@ static int atmel_nand_remove(struct platform_device *pdev)
>   	return 0;
>   }
>   
> -static struct atmel_nand_caps at91rm9200_caps = {
> +static const struct atmel_nand_caps at91rm9200_caps = {
>   	.pmecc_correct_erase_page = false,
>   };
>   
> -static struct atmel_nand_caps sama5d4_caps = {
> +static const struct atmel_nand_caps sama5d4_caps = {
>   	.pmecc_correct_erase_page = true,
>   };
>   

Re: [PATCH v2 1/2] mtd: nand: atmel_nand: constify atmel_nand_caps structures

[Brian Norris]

On Fri, Nov 20, 2015 at 04:28:10PM +0800, Josh Wu wrote:
> Hi, Corentin
> 
> Thanks for the patch.
> 
> On 11/20/2015 3:45 PM, LABBE Corentin wrote:
> >All atmel_nand_caps are never modified, consitify them.
> >
> >Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
> 
> Acked-by: Josh Wu <josh.wu@atmel.com>

Pushed patch 1 to l2-mtd.git

[PATCH v2 2/2] mtd: nand: atmel_nand: fix a possible NULL dereference

[LABBE Corentin]

of_match_device could return NULL, and so cause a NULL pointer
dereference later.

Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
---
 drivers/mtd/nand/atmel_nand.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
index 475c938..7902967 100644
--- a/drivers/mtd/nand/atmel_nand.c
+++ b/drivers/mtd/nand/atmel_nand.c
@@ -1496,8 +1496,9 @@ static int atmel_of_init_port(struct atmel_nand_host *host,
 	struct atmel_nand_data *board = &host->board;
 	enum of_gpio_flags flags = 0;
 
-	host->caps = (struct atmel_nand_caps *)
-		of_match_device(atmel_nand_dt_ids, host->dev)->data;
+	host->caps = of_device_get_match_data(host->dev);
+	if (!host->caps)
+		return 1;
 
 	if (of_property_read_u32(np, "atmel,nand-addr-offset", &val) == 0) {
 		if (val >= 32) {
-- 
2.4.10

Re: [PATCH v2 2/2] mtd: nand: atmel_nand: fix a possible NULL dereference

[Josh Wu]

Hi, Corentin

Thanks for the patch. It looks fine to me, just one nitpick in below:

On 11/20/2015 3:45 PM, LABBE Corentin wrote:
> of_match_device could return NULL, and so cause a NULL pointer
> dereference later.
>
> Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
> ---
>   drivers/mtd/nand/atmel_nand.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
> index 475c938..7902967 100644
> --- a/drivers/mtd/nand/atmel_nand.c
> +++ b/drivers/mtd/nand/atmel_nand.c
> @@ -1496,8 +1496,9 @@ static int atmel_of_init_port(struct atmel_nand_host *host,
>   	struct atmel_nand_data *board = &host->board;
>   	enum of_gpio_flags flags = 0;
>   
> -	host->caps = (struct atmel_nand_caps *)
> -		of_match_device(atmel_nand_dt_ids, host->dev)->data;
> +	host->caps = of_device_get_match_data(host->dev);
> +	if (!host->caps)
> +		return 1;

it's better to use -EINVAL here.

Best Regards,
Josh Wu

>   
>   	if (of_property_read_u32(np, "atmel,nand-addr-offset", &val) == 0) {
>   		if (val >= 32) {

Re: [PATCH v2 2/2] mtd: nand: atmel_nand: fix a possible NULL dereference

[LABBE Corentin]

On Fri, Nov 20, 2015 at 04:33:14PM +0800, Josh Wu wrote:
> Hi, Corentin
> 
> Thanks for the patch. It looks fine to me, just one nitpick in below:
> 
> On 11/20/2015 3:45 PM, LABBE Corentin wrote:
> > of_match_device could return NULL, and so cause a NULL pointer
> > dereference later.
> >
> > Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
> > ---
> >   drivers/mtd/nand/atmel_nand.c | 5 +++--
> >   1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
> > index 475c938..7902967 100644
> > --- a/drivers/mtd/nand/atmel_nand.c
> > +++ b/drivers/mtd/nand/atmel_nand.c
> > @@ -1496,8 +1496,9 @@ static int atmel_of_init_port(struct atmel_nand_host *host,
> >   	struct atmel_nand_data *board = &host->board;
> >   	enum of_gpio_flags flags = 0;
> >   
> > -	host->caps = (struct atmel_nand_caps *)
> > -		of_match_device(atmel_nand_dt_ids, host->dev)->data;
> > +	host->caps = of_device_get_match_data(host->dev);
> > +	if (!host->caps)
> > +		return 1;
> 
> it's better to use -EINVAL here.
> 

Hello

I do that Uwe Kleine-König said to me to do in others thread:
https://lkml.org/lkml/2015/11/12/70 and https://lkml.org/lkml/2015/11/16/211

Regards

Re: [PATCH v2 2/2] mtd: nand: atmel_nand: fix a possible NULL dereference

[Josh Wu]

Hi, Corentin

On 11/24/2015 9:12 PM, LABBE Corentin wrote:
> On Fri, Nov 20, 2015 at 04:33:14PM +0800, Josh Wu wrote:
>> Hi, Corentin
>>
>> Thanks for the patch. It looks fine to me, just one nitpick in below:
>>
>> On 11/20/2015 3:45 PM, LABBE Corentin wrote:
>>> of_match_device could return NULL, and so cause a NULL pointer
>>> dereference later.
>>>
>>> Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
>>> ---
>>>    drivers/mtd/nand/atmel_nand.c | 5 +++--
>>>    1 file changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
>>> index 475c938..7902967 100644
>>> --- a/drivers/mtd/nand/atmel_nand.c
>>> +++ b/drivers/mtd/nand/atmel_nand.c
>>> @@ -1496,8 +1496,9 @@ static int atmel_of_init_port(struct atmel_nand_host *host,
>>>    	struct atmel_nand_data *board = &host->board;
>>>    	enum of_gpio_flags flags = 0;
>>>    
>>> -	host->caps = (struct atmel_nand_caps *)
>>> -		of_match_device(atmel_nand_dt_ids, host->dev)->data;
>>> +	host->caps = of_device_get_match_data(host->dev);
>>> +	if (!host->caps)
>>> +		return 1;
>> it's better to use -EINVAL here.
>>
> Hello
>
> I do that Uwe Kleine-König said to me to do in others thread:
> https://lkml.org/lkml/2015/11/12/70 and https://lkml.org/lkml/2015/11/16/211

Thank you for the information.
I checked the imx serial driver(drivers/tty/serial/imx.c) in 
serial_imx_probe():

     ...
     ret = serial_imx_probe_dt(sport, pdev);
     if (ret > 0)     --->  So here is why you need to return 1 in 
serial_imx_probe_dt().
         serial_imx_probe_pdata(sport, pdev);
     else if (ret < 0)
         return ret;

There has a branch to check whether ret is > 0 or < 0. So that's why you 
need to return 1 in imx serial driver.

But in atmel_nand driver, we don't have such code for that. The 
atmel_of_init_port() is only used for dt probe. So if dt is not matched, 
just return an error.

>
> Regards
>
Best Regards,
Josh Wu