Subject: Re: fault address in signal handler
To: John R Towler <towl0008@tc.umn.edu>
From: Andrey Petrov <petrov@netbsd.org>
List: tech-kern
Date: 08/09/2002 11:31:30
On Sun, Aug 04, 2002 at 05:53:59PM -0500, John R Towler wrote:
>
> So,
>
> There was some mention of what might need to be done in order
> to add SA_SIGINFO to the src tree (perhaps this is the wrong technical
> term, there is a test for this in the new swi-prolog configure. The
> siginfo struct in OpenBSD contains a pointer to the information.) My
> interest is directed towards facilities which will help with Lisps and
> Prologs, where there is some use for the page address of a page fault
> apparently. What, in addition to material on the uvm system, do I
> need to begin reading in the src tree and affiliated literature in
> order to begin to understand the issues involved and be in a position
> to contribute rather than merely watching and waiting. Thanks in
> advance.
>
It seems that uvm system in not involved in this(well, it does but only
to say if faulting address is valid or not), signal is generated by
arch dependant trap handler. So I would say you need to read(modify)
kern_sig.c for trapsignal and it's users in ARCH/ARCH/trap.c.
Andrey