Skip site navigation (1)Skip section navigation (2) 
Date:      Tue, 26 Feb 2008 11:51:43 -0800
From:      ithilgore <ithilgore.fbsd@gmail.com>
To:        Giorgos Keramidas <keramida@ceid.upatras.gr>
Cc:        freebsd-net@freebsd.org
Subject:   Re: question about change in inet_ntoa.c
Message-ID:  <47C46DCF.6050202@gmail.com>
In-Reply-To: <20080226040438.GA2676@kobe.laptop>
References:  <47BFF17B.5080205@gmail.com> <47BFF74E.4010608@gmail.com> <20080226040438.GA2676@kobe.laptop>
next in thread | previous in thread | raw e-mail | index | archive | help 
Giorgos Keramidas wrote:
> On 2008-02-23 02:37, ithilgore <ithilgore.fbsd@gmail.com> wrote:
>   
>> ithilgore wrote:
>>     
>>> I was looking at the differences between some old FreeBSD code
>>> and the one of 7.0-RC1 and was wondering about a change
>>> in inet_ntoa.c
>>>
>>> /***** 7.0-RC1 **************/
>>>
>>>    sprintf(buf, "%d.%d.%d.%d",
>>>        ucp[0] & 0xff,
>>>        ucp[1] & 0xff,
>>>        ucp[2] & 0xff,
>>>        ucp[3] & 0xff);
>>>
>>>
>>> /****** 4.11-RELEASE ***********/
>>>
>>>
>>> static const char fmt[] = "%u.%u.%u%u";
>>> if ((size_t)snprintf(dst, size, fmt, src[0], src[1], src[2], src[3])
>>>    >= size) {
>>> ....
>>> ....
>>>
>>> Was there a specific purpose of changing the more easy and simple way
>>> of %u  instead of the combination of %d and and-ing with 0xff  ??
>>> It essentially gives the same result but increases overhead (i think) in 
>>> the more
>>> recent version.
>>>       
>> I just noticed I made a mistake. The second code is libc's version of
>> inet_ntoa.  But the question still counts. Why not use the plain
>> simpler version of libc ?
>>     
>
> I don't see ucp[] in RELENG_6, RELENG_7 or CURRENT.  Where did you get
> the version shown as `7.0-RC1' above?
>
>   

I got the source code from the ftp.freebsd.org and I just downloaded 
7.0-RC3 to be certain.
Both in 7.0-RC1 and in 7.0-RC3 in src/sys/libkern/ is the following code 
of inet_ntoa.c :

#include <sys/cdefs.h>
__FBSDID("$FreeBSD: src/sys/libkern/inet_ntoa.c,v 1.6 2005/01/07 
00:24:32 imp Exp $");

#include <sys/param.h>
#include <sys/systm.h>

#include <netinet/in.h>

char *
inet_ntoa(struct in_addr ina)
{
    static char buf[4*sizeof "123"];
    unsigned char *ucp = (unsigned char *)&ina;

    sprintf(buf, "%d.%d.%d.%d",
        ucp[0] & 0xff,
        ucp[1] & 0xff,
        ucp[2] & 0xff,
        ucp[3] & 0xff);
    return buf;
}

.....followed by the reentrant version of inet_ntoa : inet_ntoa_r


On the other hand, in version 4.11 RELEASE in 
/usr/src/lib/libc/net/inet_ntoa.c &
inet_ntop.c (actually it is inet_ntop.c code but with the same 
functionality) which is
called by inet_ntoa there is :

static const char *
inet_ntop4(src, dst, size)
    const u_char *src;
    char *dst;
    size_t size;
{
    static const char fmt[] = "%u.%u.%u.%u";

    if ((size_t)snprintf(dst, size, fmt, src[0], src[1], src[2], src[3])
        >= size) {
        errno = ENOSPC;
        return (NULL);
    }
    return (dst);
}

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?47C46DCF.6050202>