Subject: Re: core dump filename format
To: Chris G. Demetriou <email@example.com>
From: Christos Zoulas <firstname.lastname@example.org>
Date: 09/09/1999 22:39:18
On Sep 9, 7:15pm, email@example.com (Chris G. Demetriou) wrote:
-- Subject: Re: core dump filename format
| firstname.lastname@example.org (Christos Zoulas) writes:
| > >attribute setting, but i forget what it is. I'd guess solaris might
| > >do something similar for sparc.
| > It uses a trap actually T_FIXALIGN :-)
| that's the trap to signal an alignment fault, though, right? (so i'd
| guess from the NetBSD/sparc code...)
No, T_ALIGN is the trap you get on unaligned access. T_FIXALIGN is the
trap that turns it on (I don't think that there is a way to turn it off
from a process, but maybe I have implemented it incorrectly).
| How does it communicate that to user-land, and/or let user-land
| disable unaligned fault handling? (i used a few 'apropos' cmds
| on a solaris box i have access to, and got nothing that seemed
Well, you link with a special crt0.o that issues the trap instruction.
It is handled by a compiler flag -missalign or something; I forget.
Since we have that ability already in the kernel for the sparc it
might be useful to have a way to use it from userland, but we currently
don't (short of writing a small assembly stub...)