Re: [LTP] [PATCH 1/1] madvise03: let the system select an address for shmat()

[Simon Xu <xu....@or...>]

On 2012/8/3 14:38, Jan Stancek wrote:
> ----- Original Message -----
>> From: "Wanlong Gao" <gao...@cn...>
>> To: "Simon Xu" <xu....@or...>
>> Cc: ltp...@li...
>> Sent: Friday, 3 August, 2012 5:46:31 AM
>> Subject: Re: [LTP] [PATCH 1/1] madvise03: let the system select an address for shmat()
>>
>> On 08/03/2012 12:53 AM, Simon Xu wrote:
>>> In certain occasions, 0x22000000 is invalid for shmat() and
>>> madvise03 fails:
>>>
>>> madvise03 1 TBROK : shmat error
>>> madvise03 2 TBROK : Remaining cases broken
>>>
>>> Use `NULL' for shmat() to let the system attach the shared memory
>>> segment
>>> at the first available address.
>>>
>>> Signed-off-by: Simon Xu <xu....@or...>
>>> ---
>>>   testcases/kernel/syscalls/madvise/madvise03.c | 6 +++---
>>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/testcases/kernel/syscalls/madvise/madvise03.c
>>> b/testcases/kernel/syscalls/madvise/madvise03.c
>>> index 2e696e8..fd0bd97 100644
>>> --- a/testcases/kernel/syscalls/madvise/madvise03.c
>>> +++ b/testcases/kernel/syscalls/madvise/madvise03.c
>>> @@ -154,13 +154,13 @@ int main(int argc, char *argv[])
>>>   		if (shmid1 == -1)
>>>   			tst_brkm(TBROK, cleanup, "shmget failed");
>>>   
>>> -		/* Attach shared memory segment to 0x22000000 address */
>>> -		addr1 = shmat(shmid1, (void *)0x22000000, 0);
>>> +		/* Attach shared memory segment to an address selected by the
>>> system */
>>> +		addr1 = shmat(shmid1, NULL, 0);
>>>   		if (addr1 == (void *) -1)
>>>   			tst_brkm(TBROK, cleanup, "shmat error");
>>>   
>>>   		/* (1) Test case for MADV_REMOVE */
>>> -		TEST(madvise((void *)0x22000000, 4096, MADV_REMOVE));
>>> +		TEST(madvise((void *)addr1, 4096, MADV_REMOVE));
>> (void *) is not needed now ?
> That shouldn't be needed for NULL.

addr1 is the return value of shmat(), whose return value type is void 
*.  Yes, `(void *)' is no longer needed.  Thanks for pointing this out.


Thanks
Simon

>> otherwise, looks OK to me.
> +1, this looks better than hardcoding some random address.
>
>> Thanks,
>> Wanlong Gao
>>
>>>   		check_and_print("MADV_REMOVE");
>>>   
>>>   		/* (2) Test case for MADV_DONTFORK */