Re: [PATCH 2.5.73] Signal stack fixes #1 introduce PF_SS_ACTIVE

[Jörn Engel]

On Mon, 7 July 2003 21:33:45 +1000, Paul Mackerras wrote:
> 
> It just occurred to me that the simplest and best fix for the specific
> problem you mention is for you to set the SA_ONESHOT flag when you
> install the SIGSEGV handler.  That way, if you get another
> segmentation violation while you are already in the SIGSEGV handler,
> it will just dump core straight away.

That makes sense.  One would still have to re-enable the signal
handler inside the signal handler, but if htat is the last code before
function return, we should be safe.

Great idea, thanks!

Jörn

-- 
A defeated army first battles and then seeks victory.
-- Sun Tzu