PROBLEM: getpid() returning same value as getppid()

PROBLEM: getpid() returning same value as getppid()

[Sébastien Paumier]

[-- Attachment #1: Type: text/plain, Size: 311 bytes --]

Hi,
here is a bug that occurs on my kernel 2.6.31-21, maybe with older ones. 
  If a C program contains a function with the constructor attribute that 
calls getpid(), then, a call to syscall(SYS_fork) produces a son that 
obtains the same value calling getpid() or getppid().

Best regards,
Sébastien Paumier

[-- Attachment #2: test.c --]
[-- Type: text/x-csrc, Size: 510 bytes --]

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/syscall.h>
#include <sys/types.h>

void foo(void) __attribute__ ((constructor));

void foo(void) {
	/* Removing this call to getpid() removes the bug */
	getpid();
}


int main(void){
	int i=syscall(SYS_fork);
	if (0==i) 
		/* Big problem: the son prints the same value for getpid() and getppid() */
		printf("__son : %d   father : %d__\n", getpid(), getppid());
	else
		printf("<<father : %d   son : %d>>\n", getpid(), i);
	return 0;
}

Re: PROBLEM: getpid() returning same value as getppid()

[Nick Bowler]

On 16:45 Wed 26 May     , Sébastien Paumier wrote:
> Hi,
> here is a bug that occurs on my kernel 2.6.31-21, maybe with older ones. 
>   If a C program contains a function with the constructor attribute that 
> calls getpid(), then, a call to syscall(SYS_fork) produces a son that 
> obtains the same value calling getpid() or getppid().

The GNU C library caches the results of getpid, which is probably the
cause of your grief.  This cache relies on the glibc wrappers for fork
and friends, which you have bypassed by using syscall directly.

-- 
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)

Re: PROBLEM: getpid() returning same value as getppid()

[Michal Hocko]

On Wed 26-05-10 11:16:51, Nick Bowler wrote:
> On 16:45 Wed 26 May     , S?bastien Paumier wrote:
> > Hi,
> > here is a bug that occurs on my kernel 2.6.31-21, maybe with older ones. 
> >   If a C program contains a function with the constructor attribute that 
> > calls getpid(), then, a call to syscall(SYS_fork) produces a son that 
> > obtains the same value calling getpid() or getppid().
> 
> The GNU C library caches the results of getpid, which is probably the
> cause of your grief.  This cache relies on the glibc wrappers for fork
> and friends, which you have bypassed by using syscall directly.

man page for getpid states that explicitly:
NOTES
       Since  glibc  version  2.3.4, the glibc wrapper function for
       getpid() caches PIDs, so as to avoid additional system calls when
       a process calls getpid() repeatedly.  Normally this caching is
       invisible, but its correct operation relies on support in the
       wrapper  functions for  fork(2), vfork(2),  and  clone(2):
       if an application bypasses the glibc wrappers for these system
       calls by using syscall(2), then a call to getpid() in the child
       will return the wrong value (to be precise: it will return the
       PID of the parent process).  See also clone(2) for  discussion
       of  a case where getpid() may return the wrong value even when
       invoking clone(2) via the glibc wrapper function.

> 
> -- 
> Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

-- 
Michal Hocko
L3 team 
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic