[PATCH] Fix canonicalization of addresses
Wed Jan 7 10:25:00 GMT 2009
> Ok, I guess I can agree with that one (it makes it more obvious
> what happens). I'll bootstrap & test & apply this.
Thanks. Please remove the obsolete comment in extract_muldiv, see my patch.
What about the canonicalization of offsets just before RTL expansion? Your
2005 patch has de factor changed it and this can be harmful too. I have an
Ada testcase on PowerPC with a lot of consecutive, perturbated multi-dim
and so on for which it's a performance killer at -O2 because RTL CSE is
seriously hampered and cannot factorize all the (base + index) expressions.
The factorization is *optimal* with GCC 4.1, all the accesses use the same
(base + index) expression, only the displacements vary.
> Is there a bugzilla I can reference for the problem?
More information about the Gcc-patches