🚀 go-pugleaf

On 7/7/2018 8:16 PM, Joe Pfeiffer wrote:
> Charles Richmond <nume...@aquaporin4.com> writes:
> 
>> On 7/7/2018 6:16 PM, Joe Pfeiffer wrote:
>>> Charles Richmond <nume...@aquaporin4.com> writes:
>>>
>>>> On 7/6/2018 8:00 PM, Dan Espen wrote:
>>>>> Charles Richmond <nume...@aquaporin4.com> writes:
>>>>>>>
>>>>>> Can an unsuccessful exec() call return???  And where would it return to???
>>>>>
>>>>> The answer is in the man page:
>>>>>
>>>>> RETURN VALUES
>>>>>      If a function in the exec family returns to the calling process image,
>>>>>      an error has occurred; the return value is -1 and errno is set to
>>>>>      indicate the error.
>>>>>      
>>>>
>>>> The calling process for exec() must be the child process.  ISTM the
>>>> child process must signal the parent process and abort... if exec()
>>>> fails.
>>>
>>> There is no requirement that there even be a child process.  There is no
>>> reason a process can't replace its own running image using an exec().
>>>
>>
>> Yes, of course... isn't that what any process hopes to achieve by
>> calling exec()???  I assume a process could even re-load its *own*
>> image using exec().
> 
> You're the one who said the child has to call exec(); I was pointing out
> that there's no requirement that there even *be* a child.
> 
>>> If it's a normal fork&exec, the child process can just exit() (presumably
>>> with an error return code) and the parent will find out about it in any
>>> of the ways already available (including, of course, SIGCHLD).
>>>
>>
>> But if tye parent process exits before killing the child... you could
>> end up with a "zombie" process.  ISTM that SVR4 Unix had system calls
>> that would allow the child to exit (and die) without parental
>> approval...
> 
> You're talking about an orphan.  A process is a zombie when it has
> terminated and its parent has not yet reaped it.  An orphan process is
> reparented; init becomes its parent.  I can't think off hand of a call
> that would assign a child to some other parent than that.
> 
> 

In the first edition of _Advanced Programming in the UNIX Environment-, 
by W. Richard Stevens (RIP)...   pages 280-281:

"If a process specifically sets its disposition to SIG_IGN, children of 
the calling process will *not* generate zombie processes. ...  Instead, 
on termination, the status of these child processes is just discarded."

In the above case, the parent does *not* have to "reap" the child process.

-- 
numerist at aquaporin4 dot com