Re: mmap syscall problem

[Michal Simek <michal.simek@petalogix.com>]

Arnd Bergmann wrote:
> On Monday 06 July 2009, Michal Simek wrote:
>   
>>>       *p1 = 0xaa; *p2 = 0x55;
>>>   
>>>       
>> I closed fd too.
>>     close(fd);
>>     
>>>       return *p1; /* returns 0xaa if broken, 0x55 if correct */
>>> }
>>>       
>
> If you close the fd between the assignment and reading
> from the pointer again, the test case becomes invalid because of
> timing. Closing the fd before the '*p1 = 0xaa' should be fine,
> but unnecessary.
>
> I also realized that you might need to mark the pointers
> as 'volatile' so that the compiler has to do the operations
> in order.
>
>   
The result are the same. I tried to run mmap test which are in LTP and
only first one failed.

Do you have any ideas what buffer could caused that problems?
>>>   
>>>       
>> # ls -la existing-4k-file
>> -rw-rw-r--    1 monstr   monstr       4096 Jul  6  2009 existing-4k-file
>>
>> # ./test-arnd
>> # echo $?
>> 85
>>     
>
> Ok, so inside a single task, this does not happen.
>   
I print output in that task and there is 0x55 too.
>   
>> # dd if=existing-4k-file of=/dev/console count=1 2>/dev/null
>> U#
>>
>> in file is first char U (0x55) which is IMO correct.
>>     
>
> Right, though that was not part of the test, I'd expect this in
> the file even if the return value was broken.
>   
Not in case that mmap is completely broken. This test confirm that mmap
map that fd and is possible to write there.

Thanks,
Michal


> 	Arnd <><
>   


-- 
Michal Simek, Ing. (M.Eng)
PetaLogix - Linux Solutions for a Reconfigurable World
w: www.petalogix.com p: +61-7-30090663,+42-0-721842854 f: +61-7-30090663