Re: For review: pidfd_send_signal(2) manual page

[Michael Kerrisk (man-pages)]

Hello Christian,

On 9/25/19 3:53 PM, Christian Brauner wrote:
> On Wed, Sep 25, 2019 at 03:46:26PM +0200, Michael Kerrisk (man-pages) wrote:
>> On 9/24/19 11:53 PM, Christian Brauner wrote:
>>> On Tue, Sep 24, 2019 at 11:00:03PM +0200, Michael Kerrisk (man-pages) wrote:
>>>> Hello Christian,
>>>>
>>>>>>> If you're the parent of the process you can do this without CLONE_PIDFD:
>>>>>>> pid = fork();
>>>>>>> pidfd = pidfd_open();
>>>>>>> ret = pidfd_send_signal(pidfd, 0, NULL, 0);
>>>>>>> if (ret < 0 && errno == ESRCH)
>>>>>>> 	/* pidfd refers to another, recycled process */
>>>>>>
>>>>>> Although there is still the race between the fork() and the
>>>>>> pidfd_open(), right?
>>>>>
>>>>> Actually no and my code is even too complex.
>>>>> If you are the parent, and this is really a sequence that obeys the
>>>>> ordering pidfd_open() before waiting:
>>>>>
>>>>> pid = fork();
>>>>> if (pid == 0)
>>>>> 	exit(EXIT_SUCCESS);
>>>>> pidfd = pidfd_open(pid, 0);
>>>>> waitid(pid, ...);
>>>>>
>>>>> Then you are guaranteed that pidfd will refer to pid. No recycling can
>>>>> happen since the process has not been waited upon yet (That is,
>>>>
>>>> D'oh! Yes, of course. 
>>>>
>>>>> excluding special cases such as where you have a mainloop where a
>>>>> callback reacts to a SIGCHLD event and waits on the child behind your
>>>>> back and your next callback in the mainloop calls pidfd_open() while the
>>>>> pid has been recycled etc.).
>>>>> A race could only appear in sequences where waiting happens before
>>>>> pidfd_open():
>>>>>
>>>>> pid = fork();
>>>>> if (pid == 0)
>>>>> 	exit(EXIT_SUCCESS);
>>>>> waitid(pid, ...);
>>>>> pidfd = pidfd_open(pid, 0);
>>>>>
>>>>> which honestly simply doesn't make any sense. So if you're the parent
>>>>> and you combine fork() + pidfd_open() correctly things should be fine
>>>>> without even having to verify via pidfd_send_signal() (I missed that in
>>>>> my first mail.).
>>>>
>>>> Thanks for the additional detail.
>>>
>>> You're very welcome.
>>>
>>>>
>>>> I added the following to the pidfd_open() page, to
>>>> prevent people making the same thinko as me:
>>>>
>>>>        The following code sequence can be used to obtain a file  descripâ
>>>>        tor for the child of fork(2):
>>>>
>>>>            pid = fork();
>>>>            if (pid > 0) {     /* If parent */
>>>>                pidfd = pidfd_open(pid, 0);
>>>>                ...
>>>>            }
>>>>
>>>>        Even  if  the  child process has already terminated by the time of
>>>>        the pidfd_open() call, the returned file descriptor is  guaranteed
>>>>        to refer to the child because the parent has not yet waited on the
>>>>        child (and therefore, the child's ID has not been recycled).
>>>
>>> Thanks! I'm fine with the example. The code illustrates the basics. If
>>> you want to go overboard, you can mention my callback example and put my
>>> SIG_IGN code snippet from my earlier mails (cf. [1] and [2]) in there.
>>> But imho, that'll complicate the manpage and I'm not sure it's worth it.
>>
>> I agree that we should not complicate this discussion with more code,
>> but how about we refine the text as follows:
>>
>>        The following code sequence can be used to obtain a file  descripâ
>>        tor for the child of fork(2):
>>
>>            pid = fork();
>>            if (pid > 0) {     /* If parent */
>>                pidfd = pidfd_open(pid, 0);
>>                ...
>>            }
>>
>>        Even  if  the  child  has  already  terminated  by the time of the
>>        pidfd_open() call, its PID will not have  been  recycled  and  the
>>        returned  file  descriptor  will  refer  to  the  resulting zombie
>>        process.  Note, however, that this is guaranteed only if the  folâ
>>        lowing conditions hold true:
>>
>>        *  the  disposition  of  SIGCHLD  has  not  been explicitly set to
>>           SIG_IGN (see sigaction(2)); and
> 
> Ugh, I forgot a third one. There's also SA_NOCLDWAIT. When set and
> the SIGCHLD handler is set to SIG_DFL then no zombie processes are
> created and no SIGCHLD signal is sent. When an explicit handler for
> SIGCHLD is set then a SIGCHLD signal is generated but the process will
> still not be turned into a zombie...

Oh, yes. I added:

       *  the SA_NOCLDSTOP flag was not specified  while  establishing  a
          handler  for  SIGCHLD  or while setting the disposition of that
          signal to SIG_DFL (see sigaction(2));

>>        *  the zombie process was not  reaped  elsewhere  in  the  program
>>           (e.g.,  either  by an asynchronously executed signal handler or
>>           by wait(2) or similar in another thread).
>>
>>        If these conditions don't hold true, then the child process should
> 
> "If any of these conditions does not hold, the child process..."
> 
> That might be clearer. But I leave the call on that to you. :)

Yep, your wording is better. Fixed.

Thanks,

Michael

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/