Re: [PATCH][BZ 14561] Testcase

["Carlos O'Donell" <carlos at redhat dot com>]

On 06/03/2013 04:41 PM, OndÅej BÃlka wrote:
> On Thu, May 23, 2013 at 06:10:59PM -0400, Carlos O'Donell wrote:
>> On 05/20/2013 03:06 PM, OndÅej BÃlka wrote:
>>> Hi, 
>>> Functions rand calls random which causes seed to be shared between them.
>>>
>>> This patch causes rand and random use separate seeds.
>>>
>>> Comments?
>>>
>>> Ondra
>>>
>>> 	* stdlib/rand.c: Include stdlib/random.c
>>> 	* stdlib/random.c: Separate rand and random.
>>
>> This is user visible change and needs a BZ#.
>>
> Its 14561.
> 
>> This also needs a test case similar to what Rich posted
>> with a comment indicating the rationale from the standards
>> perspective.
>>
>> Repost with test, BZ#, and fixes below.
>>
> As testcases can be written independently I decided to post testcase
> separetely
> 
> Ondra
> 
> 	* bug-srand-srandom-dependency.c: New file.
> 	* Makefile (tests) : Add bug-srand-srandom-dependency.

This is a good first pass, but we need to make this easier to debug
when it fails and when we switch to PASS/FAIL test we should be able
to explain which ordering failed.

> ---
>  stdlib/Makefile                       |    3 +-
>  stdlib/bug-srand-srandom-dependency.c |   65 +++++++++++++++++++++++++++++++++
>  2 files changed, 67 insertions(+), 1 deletion(-)
>  create mode 100644 stdlib/bug-srand-srandom-dependency.c
> 
> diff --git a/stdlib/Makefile b/stdlib/Makefile
> index 17d80e0..d635371 100644
> --- a/stdlib/Makefile
> +++ b/stdlib/Makefile
> @@ -71,7 +71,8 @@ tests		:= tst-strtol tst-strtod testmb testrand testsort testdiv   \
>  		   tst-makecontext2 tst-strtod6 tst-unsetenv1		    \
>  		   tst-makecontext3 bug-getcontext bug-fmtmsg1		    \
>  		   tst-secure-getenv tst-strtod-overflow tst-strtod-round   \
> -		   tst-tininess tst-strtod-underflow tst-tls-atexit
> +		   tst-tininess tst-strtod-underflow tst-tls-atexit         \
> +		   bug-srand-srandom-dependency
>  tests-static	:= tst-secure-getenv
>  
>  modules-names	= tst-tls-atexit-lib
> diff --git a/stdlib/bug-srand-srandom-dependency.c b/stdlib/bug-srand-srandom-dependency.c
> new file mode 100644
> index 0000000..42ed02f
> --- /dev/null
> +++ b/stdlib/bug-srand-srandom-dependency.c
> @@ -0,0 +1,65 @@
> +/* Test that srand and srandom do not interact.
> +   Copyright (C) 2013 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <http://www.gnu.org/licenses/>.  */
> +
> +
> +
> +#include <stdlib.h>
> +
> +int
> +do_test (int argc, char **argv)
> +{
> +  int rand1, rand2;
> +  int random1, random2;
> +  srand (43);
> +  rand1 = rand ();
> +  rand2 = rand ();
> +  srandom (42);
> +  random1 = random ();
> +  random2 = random ();
> +   
> +  /* We enumerate all combinations how rand and random
> +     can be interleaved and test independency.  */

This comment needs to clearly explain the way in which the
interleaving is done and how many patterns are generated.

> +  for (int com = 0; com < 64; com++)
> +    {
> +      int randstate = 0, randomstate = 0;
> +      for (int i = 0; i < 6; i++)
> +	if (com & (1 << i))
> +	  {
> +	    if (randstate == 0)
> +	      srand (43);
> +	    if (randstate == 1 && rand1 != rand ())
> +	      return 1;

Each of the `return 1'; needs to be able to print the ordering
which lead to the failure (since it might be intermittent).

If that's too difficult to do then I suggest manually (using
a script) unrolling this loop into a static ordering that can
use distinct messages for each failure state.

Feel free to compact the error message printout to something
like: "FAIL: Order was rRr.\n"

> +	    if (randstate == 2 && rand2 != rand ())
> +	      return 1;
> +	    randstate++;
> +	  }
> +	else
> +	  {
> +	    if (randomstate == 0)
> +	      srandom (42);
> +	    if (randomstate == 1 && random1 != random ())
> +	      return 1;
> +	    if (randomstate == 2 && random2 != random ())
> +	      return 1;
> +	    randomstate++;
> +	  }
> +    }
> +  return 0;
> +}
> +
> +#include "../test-skeleton.c"
> 

Cheers,
Carlos.