Re: i386-stub.c patch, remove exceptionHook cruft.

[jtc@redback.com (J.T. Conklin)]

>>>>> "Stan" == Stan Shebs <shebs@cygnus.com> writes:
Stan> Unlike the previous patch, this gives me a slight pang of "well,
Stan> can't we just finish implementing it?".  But someone will
Stan> probably have to do it from scratch anyway, the old code more
Stan> likely to be misleading than helpful, so out it comes.

The ExceptionHook in the m68k-stub is intended to provide a mechanism
for chaining exceptions.  For example, when a exception is raised the
stub's exception handler (handle_exception) would get control first.
When the user steps or continues, the original exception handler is
called.

Although this can be a useful feature, IHMO a stub shouldn't impose
it's own model of exception handling on a target.  I've used targets
with all sorts of different exception mechanisms.  It is convenient
that the sample stubs come with code that saves an exception frame,
calls handle_exception(), and restores processor state; but it should
should not be unnecessarily difficult to remove/replace all that with
code specific to the target.

After examining the m68k stub a bit more, I believe that all the code
that attempts to find the corresponding exception frame on the stack,
etc. probably belongs in _returnFromException() so that the exception
handler itself is fairly generic.

But that's a bit more than I want to tackle at the moment.

        --jtc

-- 
J.T. Conklin
RedBack Networks