2012-02-03 08:21:07

by Geunsik Lim

Subject: [PATCH] Fix potentially derefencing uninitialized 'r'.

struct memblock_region 'r' will not be initialized potentially
because of while statement's condition in __next_mem_pfn_range()function.
Initialize struct memblock_region data structure by default.

Signed-off-by: Geunsik Lim <[email protected]>
---
mm/memblock.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/memblock.c b/mm/memblock.c
index 77b5f22..867f5a2 100644
--- a/mm/memblock.c
+++ b/mm/memblock.c
@@ -671,7 +671,7 @@ void __init_memblock __next_mem_pfn_range(int *idx, int nid,
unsigned long *out_end_pfn, int *out_nid)
{
struct memblock_type *type = &memblock.memory;
- struct memblock_region *r;
+ struct memblock_region *r = &type->regions[*idx];

while (++*idx < type->cnt) {
r = &type->regions[*idx];
--
1.7.8.1

2012-02-17 22:28:09

by Andrew Morton

Subject: Re: [PATCH] Fix potentially derefencing uninitialized 'r'.

On Fri, 3 Feb 2012 17:20:56 +0900
Geunsik Lim <[email protected]> wrote:

> struct memblock_region 'r' will not be initialized potentially
> because of while statement's condition in __next_mem_pfn_range()function.
> Initialize struct memblock_region data structure by default.
>
> Signed-off-by: Geunsik Lim <[email protected]>
> ---
> mm/memblock.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/mm/memblock.c b/mm/memblock.c
> index 77b5f22..867f5a2 100644
> --- a/mm/memblock.c
> +++ b/mm/memblock.c
> @@ -671,7 +671,7 @@ void __init_memblock __next_mem_pfn_range(int *idx, int nid,
> unsigned long *out_end_pfn, int *out_nid)
> {
> struct memblock_type *type = &memblock.memory;
> - struct memblock_region *r;
> + struct memblock_region *r = &type->regions[*idx];
>
> while (++*idx < type->cnt) {
> r = &type->regions[*idx];

The following `if' test prevents any such dereference.

Maybe you saw a compilation warning (I didn't). If so,
unintialized_var() is one way of suppressing it.

A better way is to reorganise the code (nicely). Often that option
isn't available.

2012-02-20 07:57:38

by Geunsik Lim

Subject: Re: [PATCH] Fix potentially derefencing uninitialized 'r'.

On Sat, Feb 18, 2012 at 7:28 AM, Andrew Morton
<[email protected]> wrote:
> On Fri,  3 Feb 2012 17:20:56 +0900
> Geunsik Lim <[email protected]> wrote:
>
>> struct memblock_region 'r' will not be initialized potentially
>> because of while statement's condition in __next_mem_pfn_range()function.
>> Initialize struct memblock_region data structure by default.
>>
>> Signed-off-by: Geunsik Lim <[email protected]>
>> ---
>>  mm/memblock.c |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/mm/memblock.c b/mm/memblock.c
>> index 77b5f22..867f5a2 100644
>> --- a/mm/memblock.c
>> +++ b/mm/memblock.c
>> @@ -671,7 +671,7 @@ void __init_memblock __next_mem_pfn_range(int *idx, int nid,
>>                               unsigned long *out_end_pfn, int *out_nid)
>>  {
>>       struct memblock_type *type = &memblock.memory;
>> -     struct memblock_region *r;
>> +     struct memblock_region *r = &type->regions[*idx];
>>
>>       while (++*idx < type->cnt) {
>>               r = &type->regions[*idx];
>
> The following `if' test prevents any such dereference.
>
> Maybe you saw a compilation warning (I didn't).  If so,
> unintialized_var() is one way of suppressing it.
Yepp. This patch is for solving compilation warning as you commented.
>
> A better way is to reorganise the code (nicely).  Often that option
> isn't available.
I will post patch again after reorganizing the code with better way.
Thanks.
>
>



--
----
Best regards,
Geunsik Lim, Samsung Electronics
http://leemgs.fedorapeople.org
----
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/