Re: popen() fails while system() works

[Jason Tishler <jason at tishler dot net>]

On Tue, Jan 11, 2005 at 02:21:22PM +0100, Rainer Hochreiter wrote:
> > On Mon, Jan 10, 2005 at 08:56:58AM +0100, Rainer Hochreiter wrote:
> >> but what's the real reason why popen() results in the address
> >> conflict and system() does not?!
> >
> > AFAICT, popen() calls fork() and system() does not.
> 
> this is not true - both functions call vfork()!
> 
> cygwin-1.5.12-1/newlib/libc/stdlib/system.c:165 if ((pid = vfork ()) == 0)

AFAICT, Cygwin's system() is implemented in:

    winsup/cygwin/syscalls.cc

> system() calls execve() while popen() calls execl() to start execution
> of another file!

Nevertheless, the above is not quite right.  fork() is called, but the
call occurs in sh *not* the original application.  See the attached
testcase, st.c, and the following strace snippets:

    $ strace st | egrep '(Program name| fork):'
    Program name: c:\home\jtishler\src\c++\st.exe (2788)
    Program name: C:\cygwin\bin\sh.exe (928)
      227   17561 [main] sh 928 fork: entering
    Program name: C:\cygwin\bin\sh.exe (4948)
      199   42119 [main] sh 928 fork: 4948 = fork()
      151   43337 [main] sh 4948 fork: 0 = fork()
    Program name: C:\cygwin\bin\id.exe (4948)

This is why system() does not trigger the base address conflict problem
and popen() does.

Jason

-- 
PGP/GPG Key: http://www.tishler.net/jason/pubkey.asc or key servers
Fingerprint: 7A73 1405 7F2B E669 C19D  8784 1AFD E4CC ECF4 8EF6

Attachment: st.c
Description: Text document

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/