Re: [LTP] mmapstress03: errno = 0: really large mmap didn't fail

[CAI Q. <ca...@cc...>]

Hi,

--- Dmitry Guryanov <dgu...@pa...> wrote:

> Hi.
> 
> I'm run mmapstress03 test on rhel-5, i386 with default kernel
> 2.6.18-53.el5 and 
> got follwing fail:
> 
> ./mmapstress03: errno = 0: really large mmap didn't fail
> mmapstress03    1  FAIL  :  Test failed
> 

I have seen this test has failed on all architectures IA-32, x86-64,
s390x, IA-64 and PPC64. Sometimes, it failed like the following on an
x86-64 machine,

mmapstress03: errno = 12: couldn't brk back over holes
mmapstress03    1  FAIL  :  Test failed

Looks like this is a newly added test, as well as mmapstress04,
mmapstress07, and mmapstress08 which also give failures on different
architectures. May I suggest that we do some more testing before adding
new tests to LTP default run in the future? If their purposes are to
see if the kernel crash or not, and don't care about the correctness,
is it possible to change those failures from FAIL to INFO, or
differentiate them from tests for correctness. For example, move them
to "crashme" test run?

Cai Qian 

> machine has 2Gb of RAM and 2Gb swap.
> 
> fail is in the following lines:
> 
>         if ((x = mmap((caddr_t)(INT_MAX - INT_MAX%pagesize),
>                 (size_t)(INT_MAX - INT_MAX%pagesize),
>                 /*PROT_READ|PROT_WRITE,
> MAP_ANONYMOUS|MAP_FIXED|MAP_SHARED, 0, 
> 0) 12/31/02 */
>                 PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_SHARED, 0,
> 0))
>                 != (caddr_t)-1)
>         {
>                 printf("%p %p\n", x, INT_MAX - INT_MAX%pagesize);
>                 ERROR("really large mmap didn't fail");
>                 anyfail();
>         }
> 
> mmap can give 2Gb of RAM to process, but maps them not to 'INT_MAX - 
> INT_MAX%pagesize)' address, but to smaller one, because there is no
> MAP_FIXED 
> in flags.
> 
> So the question is what should do this check and why someone has
> commented out 
> string with MAP_FIXED ?
> 
> -- 
> Dmitry Guryanov
> 
>
-------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's
> challenge
> Build the coolest Linux based applications with Moblin SDK & win
> great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the
> world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> Ltp-list mailing list
> Ltp...@li...
> https://lists.sourceforge.net/lists/listinfo/ltp-list
>