cpufreq: sun50i: Fix build warning around snprint()

[PATCH] cpufreq: sun50i: Fix build warning around snprint()

Posted by Viresh Kumar 1 year, 9 months ago

The Sun50i driver generates a warning with W=1:

warning: '%d' directive output may be truncated writing between 1 and 10 bytes into a region of size 2 [-Wformat-truncation=]

Fix it by allocating a big enough array to print an integer.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202404191715.LDwMm2gP-lkp@intel.com/
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/sun50i-cpufreq-nvmem.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/cpufreq/sun50i-cpufreq-nvmem.c b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
index 30e5c337611c..cd50cea16a87 100644
--- a/drivers/cpufreq/sun50i-cpufreq-nvmem.c
+++ b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
@@ -19,8 +19,6 @@
 #include <linux/pm_opp.h>
 #include <linux/slab.h>
 
-#define MAX_NAME_LEN	7
-
 #define NVMEM_MASK	0x7
 #define NVMEM_SHIFT	5
 
@@ -208,7 +206,7 @@ static int sun50i_cpufreq_get_efuse(void)
 static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
 {
 	int *opp_tokens;
-	char name[MAX_NAME_LEN];
+	char name[] = "speedXXXXXXXXXXX"; /* Integers can take 11 chars max */
 	unsigned int cpu, supported_hw;
 	struct dev_pm_opp_config config = {};
 	int speed;
@@ -235,7 +233,7 @@ static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
 		config.supported_hw_count = 1;
 	}
 
-	snprintf(name, MAX_NAME_LEN, "speed%d", speed);
+	snprintf(name, sizeof(name), "speed%d", speed);
 	config.prop_name = name;
 
 	for_each_possible_cpu(cpu) {
-- 
2.31.1.272.g89b43f80a514

Re: [PATCH] cpufreq: sun50i: Fix build warning around snprint()

Posted by Julian Calaby 1 year, 9 months ago

Hi Viresh,

On Mon, Apr 22, 2024 at 1:31 PM Viresh Kumar <viresh.kumar@linaro.org> wrote:
>
> The Sun50i driver generates a warning with W=1:
>
> warning: '%d' directive output may be truncated writing between 1 and 10 bytes into a region of size 2 [-Wformat-truncation=]
>
> Fix it by allocating a big enough array to print an integer.
>
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202404191715.LDwMm2gP-lkp@intel.com/
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>

This'll absolutely fix the issue here, so:

Reviewed-by: Julian Calaby <julian.calaby@gmail.com>

However:

> ---
>  drivers/cpufreq/sun50i-cpufreq-nvmem.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/cpufreq/sun50i-cpufreq-nvmem.c b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> index 30e5c337611c..cd50cea16a87 100644
> --- a/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> +++ b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> @@ -208,7 +206,7 @@ static int sun50i_cpufreq_get_efuse(void)
>  static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
>  {
>         int *opp_tokens;
> -       char name[MAX_NAME_LEN];
> +       char name[] = "speedXXXXXXXXXXX"; /* Integers can take 11 chars max */

Would it make sense to just set a static length for the string here,
say 17-20 characters and add a comment explaining the number, say: /*
"speed" + 11 chars for the int */

The string constant, while it'll probably be optimised away, seems
weird and wasteful.

Thanks,

-- 
Julian Calaby

Email: julian.calaby@gmail.com
Profile: http://www.google.com/profiles/julian.calaby/

Re: [PATCH] cpufreq: sun50i: Fix build warning around snprint()

Posted by Viresh Kumar 1 year, 9 months ago

On 23-04-24, 11:38, Julian Calaby wrote:
> On Mon, Apr 22, 2024 at 1:31 PM Viresh Kumar <viresh.kumar@linaro.org> wrote:
> > diff --git a/drivers/cpufreq/sun50i-cpufreq-nvmem.c b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> > index 30e5c337611c..cd50cea16a87 100644
> > --- a/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> > +++ b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> > @@ -208,7 +206,7 @@ static int sun50i_cpufreq_get_efuse(void)
> >  static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
> >  {
> >         int *opp_tokens;
> > -       char name[MAX_NAME_LEN];
> > +       char name[] = "speedXXXXXXXXXXX"; /* Integers can take 11 chars max */
> 
> Would it make sense to just set a static length for the string here,
> say 17-20 characters and add a comment explaining the number, say: /*
> "speed" + 11 chars for the int */
> 
> The string constant, while it'll probably be optimised away, seems
> weird and wasteful.

The counting goes wrong (I have done it in the past) sometimes and so
I like to explicitly reserve space like this, it also makes it look
cleaner, i.e. how the eventual string will be named.

-- 
viresh

Re: [PATCH] cpufreq: sun50i: Fix build warning around snprint()

Posted by Julian Calaby 1 year, 9 months ago

Hi Viresh,

On Tue, Apr 23, 2024 at 4:12 PM Viresh Kumar <viresh.kumar@linaro.org> wrote:
>
> On 23-04-24, 11:38, Julian Calaby wrote:
> > On Mon, Apr 22, 2024 at 1:31 PM Viresh Kumar <viresh.kumar@linaro.org> wrote:
> > > diff --git a/drivers/cpufreq/sun50i-cpufreq-nvmem.c b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> > > index 30e5c337611c..cd50cea16a87 100644
> > > --- a/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> > > +++ b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> > > @@ -208,7 +206,7 @@ static int sun50i_cpufreq_get_efuse(void)
> > >  static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
> > >  {
> > >         int *opp_tokens;
> > > -       char name[MAX_NAME_LEN];
> > > +       char name[] = "speedXXXXXXXXXXX"; /* Integers can take 11 chars max */
> >
> > Would it make sense to just set a static length for the string here,
> > say 17-20 characters and add a comment explaining the number, say: /*
> > "speed" + 11 chars for the int */
> >
> > The string constant, while it'll probably be optimised away, seems
> > weird and wasteful.
>
> The counting goes wrong (I have done it in the past) sometimes and so
> I like to explicitly reserve space like this, it also makes it look
> cleaner, i.e. how the eventual string will be named.

I completely agree - ultimately it's whatever works and either way
works equally well.

Thanks,

-- 
Julian Calaby

Email: julian.calaby@gmail.com
Profile: http://www.google.com/profiles/julian.calaby/

Re: [PATCH] cpufreq: sun50i: Fix build warning around snprint()

Posted by Andre Przywara 1 year, 9 months ago

On Mon, 22 Apr 2024 09:01:09 +0530
Viresh Kumar <viresh.kumar@linaro.org> wrote:

Hi,

> The Sun50i driver generates a warning with W=1:
> 
> warning: '%d' directive output may be truncated writing between 1 and 10 bytes into a region of size 2 [-Wformat-truncation=]
> 
> Fix it by allocating a big enough array to print an integer.
> 
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202404191715.LDwMm2gP-lkp@intel.com/
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>

For the records: as it stands right now, "speed" will always be smaller
than 10 at the moment, but it's indeed better to play safe here.
So the fix makes sense to me and fixes the issue:

Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Tested-by: Andre Przywara <andre.przywara@arm.com>

Cheers,
Andre

> ---
>  drivers/cpufreq/sun50i-cpufreq-nvmem.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/cpufreq/sun50i-cpufreq-nvmem.c b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> index 30e5c337611c..cd50cea16a87 100644
> --- a/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> +++ b/drivers/cpufreq/sun50i-cpufreq-nvmem.c
> @@ -19,8 +19,6 @@
>  #include <linux/pm_opp.h>
>  #include <linux/slab.h>
>  
> -#define MAX_NAME_LEN	7
> -
>  #define NVMEM_MASK	0x7
>  #define NVMEM_SHIFT	5
>  
> @@ -208,7 +206,7 @@ static int sun50i_cpufreq_get_efuse(void)
>  static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
>  {
>  	int *opp_tokens;
> -	char name[MAX_NAME_LEN];
> +	char name[] = "speedXXXXXXXXXXX"; /* Integers can take 11 chars max */
>  	unsigned int cpu, supported_hw;
>  	struct dev_pm_opp_config config = {};
>  	int speed;
> @@ -235,7 +233,7 @@ static int sun50i_cpufreq_nvmem_probe(struct platform_device *pdev)
>  		config.supported_hw_count = 1;
>  	}
>  
> -	snprintf(name, MAX_NAME_LEN, "speed%d", speed);
> +	snprintf(name, sizeof(name), "speed%d", speed);
>  	config.prop_name = name;
>  
>  	for_each_possible_cpu(cpu) {

Re: [PATCH] cpufreq: sun50i: Fix build warning around snprint()

Posted by Chen-Yu Tsai 1 year, 9 months ago

On Sun, Apr 21, 2024 at 8:31 PM Viresh Kumar <viresh.kumar@linaro.org> wrote:
>
> The Sun50i driver generates a warning with W=1:
>
> warning: '%d' directive output may be truncated writing between 1 and 10 bytes into a region of size 2 [-Wformat-truncation=]
>
> Fix it by allocating a big enough array to print an integer.
>
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202404191715.LDwMm2gP-lkp@intel.com/
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>

Acked-by: Chen-Yu Tsai <wens@csie.org>