Re: [LTP] [PATCH v2]extend the -I option value

[Garrett C. <yan...@gm...>]

On Nov 25, 2010, at 10:12 PM, Zhouping Liu wrote:

> 
> ----- "Garrett Cooper" <yan...@gm...> wrote:
> 
>> From: "Garrett Cooper" <yan...@gm...>
>> To: "Zhouping Liu" <zl...@re...>
>> Cc: ltp...@li...
>> Sent: Friday, November 26, 2010 4:41:24 AM GMT +08:00 Beijing / Chongqing / Hong Kong / Urumqi
>> Subject: Re: [LTP] [PATCH v2]extend the -I option value
>> 
>> On Thu, Nov 25, 2010 at 1:23 AM, Zhouping Liu <zl...@re...>
>> wrote:
>>> Hi,all
>>> I found it just support a small value for -I option.
>>> when I run '#./hugemmap04 -I 3600 -H /var/hugetlbfs',the test
>> returned
>>> immediately.
>>> the -I option details :
>>> -I x Execute test for x seconds
>>> And if you give a much bigger value to -I option such as 3600 or
>> larger, you
>>> will find the -I option disabled,
>>> so I changed the lib/parse_opts.c file, and just modified the
>> stop_time
>>> variable type from 'int' to 'double',
>>> after changed, it can support a very bigger value for -I option.
>> 
>>    Or maybe this should be time_t or long?
>>    Honestly, this could potentially introduce a lot of headache if
>> done improperly in lib/parse_opts.c (especially if you use a double
>> precisioned time amount as this could seriously impact architectures
>> without FPUs).
> yeah,you're right, then I think it should change long long type,as there
> is no apparent difference between int type and long type.(the long long
> type can arrive at the maximum:9223372036854775807,and it can support a much bigger value.)

Are you so sure about that statement?

(amd64 FreeBSD CURRENT)
$ ~/test_int_and_long 
(I) 4294967295, (L) 18446744073709551615

(i386 Fedora 13)
$ ~/test_int_and_long
(I) 4294967295, (L) 4294967295

$ cat test_int_and_long.c 
#include <limits.h>
#include <stdio.h>

int
main(void)
{
	printf("(I) %u, (L) %lu\n", UINT_MAX, ULONG_MAX);
	return (0);
}

Longs are 32-bit integers on 32-bit architectures and 64-bit integers on 64-bit architectures...

> ( but I also have a question: why the STD_LOOP_DURATION (defined in lib/parse_opts.c) use float type? )

Don't know... I'll have to look.
Thanks,
-Garrett