Re: Want way to run background processes with SIGINT unignored

[Chet Ramey]

On 10/9/15 2:42 PM, Ian Jackson wrote:
> Hi.
> 
> I've been wrestling recently[1] with a bash script which invokes a
> number of subprocesses in parallel and collects the output.  The
> problem is that if you ^C the script, the subprocesses carry on
> running.  This is because of the standards-mandated resetting of
> SIGINT (and QUIT) to SIG_IGN in children.
> 
> Working around this in a race-free way with additional code in the
> script is very hard indeed.  I can't see how to do it without having
> the parent install an INT trap handler which synchronises with all the
> children, or something equally baroque.
> 
> 
> The reason for SIGINT being ignored is purely historical: back in the
> dawn of time, there was no job control.  If you interactively spawned
> a background process with &, you wouldn't want your attempts to ^C
> your foreground process to kill it.  This SIGINT-ignoring also applied
> to noninteractive shells and of course came to be relied on.  So it is
> too late to change the default :-/.
> 
> 
> However, it would be very easy for bash to provide an option (via `set
> -o' perhaps) to disable this behaviour.  That is, to allow SIGINT to
> be delivered normally to child processes.

I'm restricting non-standard options to `shopt' to avoid any possible
conflict with future posix changes.

> 
> With such an option, scripts which run on modern systems and which
> attempt to parallelise their work, would be able to arrange that ^C
> properly cleans up the whole process group, rather than leaving the
> background tasks running (doing needless work and perhaps causing
> lossage).

I'd be willing to look at a patch that implemented a new option to
enable this.  You only need one option; you only have two behavior
modes and you're introducing one new behavior.


>  2. In the child, reset SIGINT and SIGQUIT to the values found at
>     shell startup.  That is, uninstall trap handlers.  This is what
>     most ordinary scripts will want, because they don't want the trap
>     handler running in both parent and child.  It's the same as is
>     done for all other signals, and for all signals in non-background
>     subshells and subprocesses.

This is the behavior that any new option would toggle.  Some name like
`async_sig_ignore' or `async_sig_restore' would work.

Chet

-- 
``The lyf so short, the craft so long to lerne.'' - Chaucer
                 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, ITS, CWRU    chet@case.edu    http://cnswww.cns.cwru.edu/~chet/