Date: Sun, 28 Sep 2008 12:27:01 -0400
From: Christopher Faylor <cgf-use-the-mailinglist-please AT cygwin DOT com>
To: cygwin AT cygwin DOT com
Subject: Re: crash if fork(2) from another thread
Message-ID: <20080928162701.GB2219@ednor.casa.cgf.cx>
Reply-To: cygwin AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
References: <003301c92110$336cee20$0200a8c0 AT whiterabc2znlh> <48DF3317 DOT 5060300 AT sh DOT cvut DOT cz> <20080928152527 DOT GA2219 AT ednor DOT casa DOT cgf DOT cx> <48DFA78A DOT 3020506 AT sh DOT cvut DOT cz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <48DFA78A.3020506@sh.cvut.cz>
User-Agent: Mutt/1.5.16 (2007-06-09)
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Precedence: bulk
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com

On Sun, Sep 28, 2008 at 05:49:30PM +0200, V??clav Haisman wrote:
>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA256
>
>Christopher Faylor wrote, On 28.9.2008 17:25:
>> On Sun, Sep 28, 2008 at 09:32:39AM +0200, V??clav Haisman wrote:
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA256
>>>
>>> Hirokazu Yamamoto wrote, On 28.9.2008 4:16:
>>>> # I've post mail, but it didn't show up in
>>>> http://www.nabble.com/Cygwin-f12165.html.
>>>> # Maybe it was not good to attach a file. So try again...
>>>>
>>>> I'm not familiar with pthread & fork, but I think following code should not
>>>> crash. Is this expected behavior?
>>>>
>>>> #include <stdio.h>
>>>> #include <stdlib.h>
>>>> #include <pthread.h>
>>>> #include <assert.h>
>>>>
>>>> void *thread_func(void* args)
>>>> {
>>>>     int ret;
>>>>     pid_t pid;
>>>>     pthread_t thread;
>>>>
>>>>     puts("thread_func");
>>>>
>>>>     pid = fork();
>>>>
>>>>     assert(pid != (pid_t)-1);
>>>>
>>>>     if (pid != 0) /* parent process */
>>>>     {
>>>>         int status;
>>>>
>>>>         printf("parent process (child pid = %d)\n", pid);
>>>>
>>>>         waitpid(pid, &status, 0);
>>>>
>>>>         puts("parent process end");
>>>>     }
>>>>     else /* child process */
>>>>     {
>>>>         puts("child process"); /* crash here */
>>> I think this is not allowed in here.  You can only do async-signal-safe
>>> stuff in the child.  IO is AFAIK not in that category.  Basically, the
>>> only thing you can safely do in the child process is to call exec().
>> 
>> I don't see any reason why this shouldn't be allowed.  You should
>> be able to do anything you want in a child process.
>I do not think that is true. As per
><http://www.opengroup.org/onlinepubs/009695399/functions/fork.html>,
>RATIONALE, the penultimate paragraph, it is basically undefined behaviour.

I wrote the current version of cygwin's fork.

cgf

--
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/