Re: [LTP] Posix test case mmap/31-1 fails on Linux/eglibc

[Wanlong G. <gao...@cn...>]

On 01/04/2012 10:50 AM, Yasuaki Ishimatsu wrote:

> Hi Gao,
> 
> 2012/01/04 11:05, Wanlong Gao wrote:
>> On 01/04/2012 09:37 AM, Yasuaki Ishimatsu wrote:
>>
>>> Hi Kai,
>>>
>>> 2011/12/28 19:55, Kang Kai wrote:
>>>> Hi all,
>>>>
>>>> I run mmap/31-1 failed with ENOMEM on linux, with error message:
>>>> off: fffffffffffff000, len: fffffffffffff000
>>>> Test FAIL: expect EOVERFLOW but get other error: Cannot allocate memory
>>>>
>>>> But compare to mmap/24-2, it mmap() the same length but pass. Its output:
>>>> available memory: 18446744073709551615
>>>> addr: 7f78c54fa000, len: ffffffffffffffff
>>>> addr: 7f78c54fa000, len: fffffffffffff000
>>>> Test Pass: mmap/24-2.c Get ENOMEM: Cannot allocate memory
>>>>
>>>> So I suppose that mmap/31-1 may need update, but I don't know how to
>>>> make mmap() overflow as expect.
>>>> Any suggestion is appreciated.
>>>
>>> I don't think mmap/31-1 need update since it is Linux Kernel's problem,
>>> not the test's problem. The problem is being discussed on LKML.
>>>
>>> https://lkml.org/lkml/2011/12/22/85
>>
>>
>> But it seems that the patch was not acceptable.
> 
> As you say, it has not been accepted yet.
> But I think Linux will be fixed to return EOVERFLOW in mmap/31-1's test
> in the near future.


Yeah, agree, so we need not update this test case now and wait for a while
to let the kernel bug to be fixed.

> 
> Regards,
> Yasuaki Ishimatsu
> 
>>
>> Thanks
>> -Wanlong Gao
>>
>>>
>>> Regards,
>>> Yasuaki Ishimatsu
>>>
>>>>
>>>> Regards,
>>>> Kai
>>
>>
>>
>>
> 
>