Re: [PATCH 1/4] android/pan: Register Network Access Point

[Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>]

Hi Johan,

On 01/07/2014 12:18 PM, Johan Hedberg wrote:
> Hi Ravi,
>
> On Tue, Jan 07, 2014, Ravi kumar Veeramally wrote:
>>>> @@ -49,7 +54,7 @@
>>>>   static bdaddr_t adapter_addr;
>>>>   GSList *devices = NULL;
>>>>   uint8_t local_role = HAL_PAN_ROLE_NONE;
>>>> -static uint32_t record_id = 0;
>>>> +char bridge[5] = "bnep\0";
>>> This last line raises several questions. Firstly, C-strings have an
>>> implicit nul-character at the end so no need to explicitly try to add
>>> one there. You also don't need to have an explicit size between the
>>> square brackets since this is automatically calculated if you do
>>> initialization upon declaration. Why isn't this static? Why isn't it
>>> const? Would a simple define make more sense instead of a dedicated
>>> variable?
>>   I tried these but there are some warnings.
>>
>> #define BNEP_BRIDGE   "bnep"
>> static char bridge[] = "bnep";
>> static char *bridge = "bnep";
>>
>> ==10198== Warning: noted but unhandled ioctl 0x89a1 with no
>> size/direction hints
>> ==10198==    This could cause spurious value errors to appear.
>> ==10198==    See README_MISSING_SYSCALL_OR_IOCTL for guidance on
>> writing a proper wrapper.
> Is this from valgrind?
    Yes.
> If so, then I think it's fine to ignore it as it
> simply doesn't know the details of all ioctls. We get this kind of
> stuff for Bluetooth specific ioctls too. I.e. go with whatever is the
> simplest (probably the define).
  Ok, I will change to define and send you v2.

Regards,
Ravi.