Re: Allowing signal handlers to modify SE and BE

Gabriel Paubert [off-list ref] writes:

On Tue, 17 Oct 2000, Dan Malek wrote:

quoted

quoted

Is this really the case?  I noticed in the x86 version that setting
the equivalent of the SE bit is allowed, how does the x86 get away
with this while the PPC can't?

These bits are optionally supported by processors.  As I recall,
the 601 doesn't but I don't know of any others.  You may find a
processor where they don't have any effect.  I can't think of any
other reason.

Slight correction: the 601 does have the SE bit but not the BE bit. The
other important difference of the 601 is that it handles debug exceptions
differently (different vector etc...).

So at least the SE bit is present on all processors.

Yes, I believe the SE bit is required, but I don't have my PPC manual
with me right now.  But that shouldn't hurt allowing the bits in.

I was actually kind of looking for the person that originally put the
comment in.  I was thinking that maybe I was missing something and
this would cause a problem.  I haven't seen any, but you never know.
If it caused problems for others, that would be a good reason to not
put it into the mainstream kernel code :-).

-Corey

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/