OpenSSH 2.4.0 patch call..

[Markus Friedl]

the reason for mysignal() is that it has a defined
semantics and what Stevens calls 'reliable signals'.
the handler are not deinstalled if a signal occurs.

-m

On Sun, Jan 07, 2001 at 02:19:46PM -0600, mouring at etoh.eviladmin.org wrote:
> On Sun, 7 Jan 2001, Kevin Steves wrote:
> 
> > On Thu, 4 Jan 2001 mouring at etoh.eviladmin.org wrote:
> > : On Fri, 5 Jan 2001, Kevin Steves wrote:
> > : > ok, i missed that one. is it feasable to move entirely to sigaction()
> > : > and emulate with sigvec() if needed?
> > : > 
> > : Don't see a problem with that.  I believe that is what Markus was
> > : suggesting with his 'mysignal()'.
> > 
> > mysignal() was a wrapper for sigaction()/signal().  signal was used if
> > there was no sigaction.  i was thinking of using sigaction directly,
> > then use sigaction/sigvec emulation if needed.  will that work?
> > 
> As far as I know it should work fine.  Don't know how many platforms lack
> sigaction() and would require sigvec().  I know NeXT lacks it.  
> 
> I threw together a quick patch that replaced all signal() w/ mysignal(),
> and I did not see any ill effects under Linux.  However, I did not push it
> very hard either (due to time, and lack of any indepth knowledge of
> signal() vs sigaction()).
> 
> - Ben
> 
>