Don't eliminate insns that may trap

[Jan Hubicka]

> Richard Henderson writes:
>  > On Mon, Sep 30, 2002 at 06:46:07PM +0100, Andrew Haley wrote:
>  > > 	* flow.c (insn_dead_p): When using non-call-exceptions, don't 
>  > > 	eliminate insns that may trap.
>  > > 	* cse.c (insn_live_p): Likewise.
>  > 
>  > I'm not really fond of this, but I guess it's the
>  > best we can do
> 
> This seems to be the response to many of my patches...  :-)

Perhaps this should be controlled by separate flag, like
-fhonnor-exceptions or something like that.
Except for removing insns that may trap we also need to disable some
algebraic simplifications (like noticing that x*0 can be elliminated)
and friends.

Honza
> 
>  > for 3.3.  We really need a global optimizer that can prove that
>  > memory ops shadowed by a preceeding trapping memory op can't trap.
> 
> Also, may_trap_p needs to be rather smarter than it is -- I have
> noticed that it returns true for far more RTXes than it should.  With
> Java, the front end has a very good idea which memory references might
> trap.
> 
> Andrew.