[PATCH] Fix uninitialized local variable "covered" in i386 acpi-cpufreq driver

[PATCH] Fix uninitialized local variable "covered" in i386 acpi-cpufreq driver

[Fenghua Yu]

The local variable "covered" is used without initialization in i386 acpi-cpufreq
driver. The initial value of covered should be 0. The bug will cause memory leak
when hit. The following patch fixes this bug.

Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>

---

 arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
index 6f846be..bfb4959 100644
--- a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -511,7 +511,7 @@ acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu)
 static int acpi_cpufreq_early_init(void)
 {
 	struct acpi_processor_performance *data;
-	cpumask_t covered;
+	cpumask_t covered=0;
 	unsigned int i, j;
 
 	dprintk("acpi_cpufreq_early_init\n");

Re:[PATCH] Fix uninitialized local variable "covered" in i386 acpi-cpufreq driver

[Fenghua Yu]

Updated patch:
Use CPU_MASK_NONE for initil value.

Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>

---

 arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
index 6f846be..14fc5b8 100644
--- a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -511,7 +511,7 @@ acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu)
 static int acpi_cpufreq_early_init(void)
 {
 	struct acpi_processor_performance *data;
-	cpumask_t covered;
+	cpumask_t covered = CPU_MASK_NONE;
 	unsigned int i, j;
 
 	dprintk("acpi_cpufreq_early_init\n");

Re: [PATCH] Fix uninitialized local variable "covered" in i386 acpi-cpufreq driver

[Andrew Morton]

On Thu, 26 Jul 2007 16:46:30 -0700 Fenghua Yu <fenghua.yu@intel.com> wrote:

> The local variable "covered" is used without initialization in i386 acpi-cpufreq
> driver. The initial value of covered should be 0. The bug will cause memory leak
> when hit. The following patch fixes this bug.
> 
> Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
> 
> ---
> 
>  arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c |    2 +-
>  1 files changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
> index 6f846be..bfb4959 100644
> --- a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
> +++ b/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
> @@ -511,7 +511,7 @@ acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu)
>  static int acpi_cpufreq_early_init(void)
>  {
>  	struct acpi_processor_performance *data;
> -	cpumask_t covered;
> +	cpumask_t covered=0;
>  	unsigned int i, j;
>  
>  	dprintk("acpi_cpufreq_early_init\n");

- please put spaces around "="

- that should have been CPU_MASK_NONE

- that code's way overengineered.  This:

--- a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c~a
+++ a/arch/i386/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -511,7 +511,6 @@ acpi_cpufreq_guess_freq(struct acpi_cpuf
 static int acpi_cpufreq_early_init(void)
 {
 	struct acpi_processor_performance *data;
-	cpumask_t covered;
 	unsigned int i, j;
 
 	dprintk("acpi_cpufreq_early_init\n");
@@ -520,14 +519,13 @@ static int acpi_cpufreq_early_init(void)
 		data = kzalloc(sizeof(struct acpi_processor_performance),
 			       GFP_KERNEL);
 		if (!data) {
-			for_each_cpu_mask(j, covered) {
+			for_each_possible_cpu(j) {
 				kfree(acpi_perf_data[j]);
 				acpi_perf_data[j] = NULL;
 			}
 			return -ENOMEM;
 		}
 		acpi_perf_data[i] = data;
-		cpu_set(i, covered);
 	}
 
 	/* Do initialization in ACPI core */
_

should do the trick (please check it)

- what we have here is an open-coded alloc_percpu().  Hows about
  converting it to alloc_percpu()?

- that function should have been be __init.


How's that for a one-liner? ;)

Re: [PATCH] Fix uninitialized local variable "covered" in i386 acpi-cpufreq driver

[Andrew Morton]

On Thu, 26 Jul 2007 17:12:22 -0700 Andrew Morton <akpm@linux-foundation.org> wrote:

> How's that for a one-liner? ;)

- the return value from acpi_cpufreq_early_init() gets ignored,
  so the module will still load, but won't work.

- Once that is fixed, the test for !acpi_perf_data[cpu] in
  acpi_cpufreq_cpu_init() can be removed.

RE: [PATCH] Fix uninitialized local variable "covered" in i386 acpi-cpufreq driver

[Yu, Fenghua]

>> How's that for a one-liner? ;)

>- the return value from acpi_cpufreq_early_init() gets ignored,
>  so the module will still load, but won't work.

>- Once that is fixed, the test for !acpi_perf_data[cpu] in
>  acpi_cpufreq_cpu_init() can be removed.

We will modify code in those places.