This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree0ssa] Dead const/pure function removal take II


> > On Wed, 12 Nov 2003 14:08:57 +0100, Jan Hubicka <hubicka@ucw.cz> wrote:
> > 
> > > I've fixed it for !TREE_NOTHROW (I think it would make sense to clear
> > > NOTHROW flag on call expr when we know that his particular call will not
> > > trap in the future, so we should not re-set it.
> > 
> > *Setting* TREE_NOTHROW means the call won't trap.  And that's the case that
> > we're likely to discover by compiling the called function; the flag is
> > cleared by default.
> 
> Oops, thanks!  I somehow managed to get the double negation half-wrong.
> I am testing the obvious patch.  Any idea how to write testcase to check
> that the transformation is happening?
> (from now on I will try to write testcase for each transformation on
> tree-ssa and lets see how far I can get)

Here is an attempt.  I introduced C++ tree-ssa tests for that.
OK?

Honza
> 
> Honza
> > 
> > Jason

Attachment: tree-ssa.exp
Description: Text document

Attachment: chlg
Description: Text document

/* { dg-do compile } */
/* { dg-options "-O1 -fdump-tree-cfg" } */
double a;
void t()
{
	a=1;
}
void t1(void);
void abort(void);

void q()
{
	try {
		t();
	}
	catch (...) {abort();}
}
/* We shouldnotice nothrow attribute.  */
/* { dg-final { scan-tree-dump-times "exception" 0 "cfg"} } */

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]