Re: bmake not killing childrens

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:
:> >     I've fixed the SIG_IGN problem in our make.  Someone needs to submit
:> >     a similar fix in pkgsrc's bmake.
:> 
:> I will handle this. Thanks!
:
:Sorry I will not handle this. At least not this week. NetBSD's make has 
:diverged a lot more due to recent changes.

    Oh, it won't patch directly, but if you and I are looking at the same
    NetBSD code: ( /archive/NetBSD-current/src/usr.bin/make on LEAF ),
    then the basic issue is the same.  They are using signal() calls to
    test the current signal handler, but these also install SIG_IGN
    temporarily until the conditional reinstalls the actual handler.

    So the fix is the same.  It has to be made in four places in compat.c
    and one place in job.c, along with the wrapper procedure I added to
    poll the signal handler.

    It is a very small window, but it can be hit in a heavily-forking build.
    I think the DragonFly bug causes more reports of the problem then the
    bug in make, but both have to be fixed.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>