Subject: Re: kern/5504: signal handler does not get called again after execve
To: Charles M. Hannum <mycroft@mit.edu>
From: Andrew Brown <twofsonet@graffiti.com>
List: netbsd-bugs
Date: 05/28/1998 22:18:10
>> The oddity of the current situation is that the signal handlers are
>> removed at exec time, but the signal delivery masks are left in place.
>> What use is it to have a delivery mask for a signal with no handler?
>
>So?  Likewise, a program should not be relying on what was already in
>the mask, so it doesn't matter.  The entire point of changing the
>behaviour was to kluge it so that broken software would work.

the point was not that a signal was masked (or blocked) when no
handler was in please, but that the execed process can subsequently
respond to it.  if it desires to do so.  if not, it can just reset the
mask.  it is merely a cost of doing an exec from within a signal
handler (that i was not aware of at the time, but would have been had
i read the exec manpage more closely and more recently).

>I say nay.  The potential benefit is minimal, and the potential cost
>is that we may break software which relies on the existing (decades
>old) behaviour.

i also say (we are talking about the proposed patch, correct?) nay.
the potential benefit is null.  programs that know this, are already
dealing with it (cf sendmail and, now, my program :).  programs that
don't are already broken.  this is yet another example of a "feature"
of which almost no one is aware.

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior@daemon.org             * "ah!  i see you have the internet
twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
warfare@graffiti.com      * "information is power -- share the wealth."