structure size

structure size

[Mohammed Khalid Ansari]

Hi,

why the sizeof sturcture is alway greater than the collective sizes of all 
the elements in it.

eg
struct node {
	int	a;
	char	c;
	int	b;
};

int main()
{
	printf ("%d\n", sizeof (struct node));
	return 0;
}


it prints 12 in my case.

where is the catch.

-- 

**************************************************************************

Mohammed Khalid Ansari                    Tel (res) : 0091-022-3051360
Assistant Manager II                          (off) : 0091-022-2024641
National Centre for Software Technology   Fax       : 0091-022-2049573 
8th flr,Air India Build. Nariman Point,   E-Mail    : khalid@ncst.ernet.in 	
Mumbai 400021.

Homepage : http://soochak.ncst.ernet.in/~khalid			  	  

**************************************************************************

Re: structure size

[Helmut Djurkin]

hi,
it's the alignment.

with gcc and this:
struct node {
	/* member here */
} __attribute__((packed));

you will get the your "collective sizes".


Mohammed Khalid Ansari wrote:
> Hi,
> 
> why the sizeof sturcture is alway greater than the collective sizes of all 
> the elements in it.
> 
> eg
> struct node {
> 	int	a;
> 	char	c;
> 	int	b;
> };
> 
> int main()
> {
> 	printf ("%d\n", sizeof (struct node));
> 	return 0;
> }
> 
> 
> it prints 12 in my case.
> 
> where is the catch.
> 

Re: (F)structure size

[Steven Smith]

[-- Attachment #1: Type: text/plain, Size: 839 bytes --]

> why the sizeof sturcture is alway greater than the collective sizes of all 
> the elements in it.
> 
> eg
> struct node {
> 	int	a;
> 	char	c;
> 	int	b;
> };
It is not, in general, possible to (portably) estimate the size of a
structure just by looking at its fields, as the compiler is free to add
padding if that allows better code.

In particular, most systems can only efficiently access memory
locations if the address of the location is a multiple of its size
(i.e. if its naturally aligned).  This means that if the structure is
to be stored in an array, its size must be a multiple of the size of
the largest single member, so that every element has proper alignment.
In this case, the largest member is an int, and so the whole
structure's size must be a multiple of four.

Steven Smith,
sos22@cam.ac.uk.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: structure size

[Mohammed Khalid Ansari]

On Wed, 5 Mar 2003 hbajaj@hss.hns.com wrote:

> 
> 
> the structure size is always multiple of the size of element with max size in
> that structure.
> Suppose
> struct {
> int a;
> char b;
> }
> then size of struct will be multiple of the size of int i.e 4 on solaris.So size
> will be 8 bytes as sizeof(a)+sizeof(b) = 5 so to make it multiple of 4 it pads
> it with 3 bytes.

Why it is so? any specific reason!

> if we have
> struct {
> short int a;
> char b;
> }
> then size of struct will be multiple of the size of int i.e 2 on solaris.
> 
> Rgds
> Harvinder Bajaj
> 
> 
> 
> 
> Mohammed Khalid Ansari <khalid@ncst.ernet.in> on 05-03-2003 12:32:01 PM
> 
> To:   linux c programming mailing list <linux-c-programming@vger.kernel.org>
> cc:    (bcc: Harvinder Bajaj/HSS)
> 
> Subject:  structure size
> 
> 
> 
> 
> 
> Hi,
> 
> why the sizeof sturcture is alway greater than the collective sizes of all
> the elements in it.
> 
> eg
> struct node {
>      int  a;
>      char c;
>      int  b;
> };
> 
> int main()
> {
>      printf ("%d\n", sizeof (struct node));
>      return 0;
> }
> 
> 
> it prints 12 in my case.
> 
> where is the catch.
> 
> --
> 
> **************************************************************************
> 
> Mohammed Khalid Ansari                    Tel (res) : 0091-022-3051360
> Assistant Manager II                          (off) : 0091-022-2024641
> National Centre for Software Technology   Fax       : 0091-022-2049573
> 8th flr,Air India Build. Nariman Point,   E-Mail    : khalid@ncst.ernet.in
> Mumbai 400021.
> 
> Homepage : http://soochak.ncst.ernet.in/~khalid
> 
> **************************************************************************
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-c-programming"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 
> 
> 

getlogin(), prints ((NULL)) when printing multiple times in a row.

[J.]

Hello,

I have ran into undefined behaviour. Not from my girl-friend this time,
but from my C program. (It's not that time of month, yet... :()

But, Everytime when I execute the following C code (bottom) from within a
shell loop as in:

while : ; do ./program ; done

it prints: ((NULL))

But when I excute it in the normal way:
./program
It prints perfectly ok: myusername

C Code.
#include <stdio.h>
#include <unistd.h>
#include <sys/types.h>

int main(void) {
 char *username = NULL;

 username = getlogin();
 /* for(;;) */
  printf("%s\n", username);

 return 0;
}

When I call the printf from a for loop in the native C code it also
starts to print ((NULL)) after a couple of loops.

I am writing a network program that gives a lot of status output when
working verbose. It doesn't look very good ((NULL)) everytime when the
output is above average.

Anyone ?

Thnkx 

J.

Re: getlogin(), prints ((NULL)) when printing multiple times in a row.

[Glynn Clements]

J. wrote:

> But, Everytime when I execute the following C code (bottom) from within a
> shell loop as in:
> 
> while : ; do ./program ; done
> 
> it prints: ((NULL))
> 
> But when I excute it in the normal way:
> ./program
> It prints perfectly ok: myusername
> 
> C Code.
> #include <stdio.h>
> #include <unistd.h>
> #include <sys/types.h>
> 
> int main(void) {
>  char *username = NULL;
> 
>  username = getlogin();
>  /* for(;;) */
>   printf("%s\n", username);
> 
>  return 0;
> }
> 
> When I call the printf from a for loop in the native C code it also
> starts to print ((NULL)) after a couple of loops.

Odd; for me, it's consistent: it either always prints "glynn" or
always prints "(null)".

But is getlogin() (i.e. the name from the utmp entry for the
controlling tty) really the information that you want? Or would you be
better off with:

	struct passwd *pw = getpwuid(getuid());
	pw ? pw->pw_name : "<none>"

?

Bear in mind that not all processes have a controlling terminal, and
even for those that do, there may not be an associated utmp entry.

-- 
Glynn Clements <glynn.clements@virgin.net>

Re: getlogin(), prints ((NULL)) solved.. thnkx.. glen.

[J.]

On Wed, 12 Mar 2003, Glynn Clements wrote:

> 
> Jeroen Reynders. wrote:
> 
> > But, Everytime when I execute the following C code (bottom) from within a
> > shell loop as in:
> > 
> > while : ; do ./program ; done
> > 
> > it prints: ((NULL))
> > 
> > But when I excute it in the normal way:
> > ./program
> > It prints perfectly ok: myusername
> > 
> > C Code.
> > #include <stdio.h>
> > #include <unistd.h>
> > #include <sys/types.h>
> > 
> > int main(void) {
> >  char *username = NULL;
> > 
> >  username = getlogin();
> >  /* for(;;) */
> >   printf("%s\n", username);
> > 
> >  return 0;
> > }
> > 
> > When I call the printf from a for loop in the native C code it also
> > starts to print ((NULL)) after a couple of loops.
> 
> Odd; for me, it's consistent: it either always prints "glynn" or
> always prints "(null)".

Cool thankx for trying...

> But is getlogin() (i.e. the name from the utmp entry for the
> controlling tty) really the information that you want? Or would you be
> better off with:
> 
> 	struct passwd *pw = getpwuid(getuid());
> 	pw ? pw->pw_name : "<none>"
> 
> ?
> 
> Bear in mind that not all processes have a controlling terminal, and
> even for those that do, there may not be an associated utmp entry.

I havent thought of that one, it's indeed a good one to remember...

I have searched all the documentation I could find this evening and
eventually going back to the getlogin() manual page. It seems that I have
overlooked the `BUGS' section in the end of the manual page.

BUGS
    Unfortunately, it is often rather easy to fool getlogin().  Sometimes
    it does not work at all, because some program messed up the utmp
    file. Often, it gives only the first 8 characters of the login
    name. The user currently logged in on the controlling tty of our
    program need not be the user who started it.  Avoid getlogin() for
    security-related purposes.

Hmmm... Again a type of lesson I am not going to forgett easily ... :)

Anyway I solved the problem the way you suggested, the following works
fine on my system.

#include <pwd.h>
#include <stdio.h>
#include <unistd.h>
#include <sys/types.h>

int main(void) {
 uid_t uid;
 struct passwd *pw;

 uid = getuid();

 pw = getpwuid(uid);
 printf("my name is: %s\n", pw->pw_name);

 return 0;
}

Thank you for helping me out. 

> -- 
> Glynn Clements <glynn.clements@virgin.net>

Jeroen Reynders

strange SIGSEGV

[Alexi Jordanov]

Hello to all,

I have a strange problem with listening for file change notification.
Everything is fine in my handler fuction till the moment when I try to send
this information through pipes:

This works fine:
===========
static void handler(int sig, siginfo_t *si, void *data)
{
 fprintf(stderr, "Got an event: %d.\n", si->si_fd);
}

This cause SIGSEGV:
===============
static void handler(int sig, siginfo_t *si, void *data)
{
 fprintf(stderr, "Got an event: %d.\n", si->si_fd);
 if (write(event_pipe[1], &si->si_fd, sizeof(int)) != sizeof(int))
 {
  fprintf(stderr, "Write failed (handler) - %s.\n", errstr);
  fflush(stderr);
 }
}

I try to start debugger to look for the problem, but all that I was found
was:
==================================================
Program received signal SIGSEGV, Segmentation fault.
0x00000004 in ?? ()
(gdb) where
#0  0x00000004 in ?? ()
#1  0x90c3c9ec in ?? ()
Cannot access memory at address 0x5d8bc031
(gdb)

The problem appears when I try to watch simultaneously two times one and the
same directory. But I don't think that this is 100% true, because I don't
have problem with other directories.

Can somebody advise me how to deal with this problem?!

Regards, Alex