[Mesa-dev] [PATCH 2/5] util/list: fix member access within null pointer

[Nicolai Hähnle]

On 10.02.2017 11:46, Emil Velikov wrote:
> On 10 February 2017 at 09:26, Nicolai Hähnle <nhaehnle at gmail.com> wrote:
>> On 07.02.2017 19:34, Bartosz Tomczyk wrote:
>>>
>>> ---
>>>  configure.ac    | 3 +++
>>>  src/util/list.h | 9 +++++++++
>>>  2 files changed, 12 insertions(+)
>>>
>>> diff --git a/configure.ac b/configure.ac
>>> index a6ceee95a3..87f635c1a3 100644
>>> --- a/configure.ac
>>> +++ b/configure.ac
>>> @@ -415,6 +415,9 @@ AC_C_BIGENDIAN(
>>>     little_endian=no
>>>  )
>>>
>>> +dnl Chek for typeof support
>>> +AC_C_TYPEOF()
>>> +
>>>  dnl Check for POWER8 Architecture
>>>  PWR8_CFLAGS="-mpower8-vector"
>>>  have_pwr8_intrinsics=no
>>> diff --git a/src/util/list.h b/src/util/list.h
>>> index 07eb9f3e6a..eaaac43fec 100644
>>> --- a/src/util/list.h
>>> +++ b/src/util/list.h
>>> @@ -171,10 +171,19 @@ static inline void list_validate(struct list_head
>>> *list)
>>>   * 'sample' MUST be initialized, or else the result is undefined!
>>>   */
>>>  #ifndef container_of
>>> +#if HAVE_TYPEOF
>>> +#define container_of_typed(ptr, type, member) ({                       \
>>> +       const __typeof__( ((type *)0)->member ) *__mptr = (ptr);        \
>>> +       (type *)( (char *)__mptr - offsetof(type,member) );})
>>
>>
>> I believe this syntax is a GCC extension. Do all compilers with typeof also
>> support that extension? This is something to be checked especially on
>> Windows. Apart from that, the change looks good to me.
>>
> HAVE_TYPEOF will only be set if the compiler supports the typeof
> extension, so we should be safe ;-)

I was thinking about the ({ ... }) statement expression syntax.

Cheers,
Nicolai

>
> For the patch
> Acked-by: Emil Velikov <emil.velikov at collabora.com>
>
> -Emil
>