This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Thoughts on live-on-entry variables
- From: Joern Rennecke <joern dot rennecke at superh dot com>
- To: Diego Novillo <dnovillo at redhat dot com>
- Cc: "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Mon, 16 Jun 2003 18:47:41 +0100
- Subject: Re: [tree-ssa] Thoughts on live-on-entry variables
- Organization: SuperH UK Ltd.
> an uninitialized variable. This pass would run at the end of the
> optimization pipeline, which would minimize (eliminate?) false
> positives.
If you manage to get the optimizers to a point that they eliminate
false positives, you will cause quite a stir in computing science.
consider:
int
f (void (*fun)(void))
{
int u;
(*fun) ();
return u;
}
If f is linked statically with a caller that passes a pointer to a
function that doesn't terminate, no warning is issued.
If it is statically linked to a caller that passes a pointer to a
function that terminates, a warning about 'u' being used
uninitialized is issued. Hence, you have solved the halting problem.
--
--------------------------
SuperH (UK) Ltd.
2410 Aztec West / Almondsbury / BRISTOL / BS32 4QX
T:+44 1454 465658