This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Analysis of Brad's GCSE problem with computed gotos
- To: Brad Lucier <lucier at math dot purdue dot edu>
- Subject: Re: Analysis of Brad's GCSE problem with computed gotos
- From: Richard Henderson <rth at redhat dot com>
- Date: Tue, 5 Dec 2000 14:06:55 -0800
- Cc: gcc at gcc dot gnu dot org
- References: <200012052158.eB5LwCm09403@polya.math.purdue.edu>
On Tue, Dec 05, 2000 at 04:58:12PM -0500, Brad Lucier wrote:
> Tony Hosking explained to me yesterday that to prevent incorrect code
> generation one needs to make either a precise flow analysis (which may
> be effectively impossible with abnormal critical edges), which allows
> the most optimization, or an imprecise but conservative flow analysis,
> which allows correct but imperfect optimization.
Um, no. One must have precise flow analysis in order to do these
sorts of optimizations _at all_.
But just because you have an exact flow graph doesn't mean that
you can necessarily modify it arbitrarily. And that is our
problem with abnormal edges -- they cannot be split.
r~