This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: resources.c patch updated....
- To: Jan Hubicka <hubicka at atrey dot karlin dot mff dot cuni dot cz>
- Subject: Re: resources.c patch updated....
- From: Richard Henderson <rth at cygnus dot com>
- Date: Wed, 3 Nov 1999 09:53:09 -0800
- Cc: egcs-patches at egcs dot cygnus dot com
- References: <19991103142412.19578@atrey.karlin.mff.cuni.cz>
On Wed, Nov 03, 1999 at 02:24:12PM +0100, Jan Hubicka wrote:
> Even with recent roolback in resource.c, this part of patch is usefull
> for peephole2. I really need peephole2 reliable for my reg-stack changes.
I'm thinking of tackling this from a different direction.
For one, resource.c does far too much work. What we want
is to begin with global_live_at_end and back up over the
insns one by one as we match for peepholes.
To convince mark_target_regs_live, by way of find_free_reg
not to do too much work, we can interface with the hash
table that it sets up. I.e. do our own liveness calculation
and then inject the results into resource.c's tables.
I'm undecided how much liveness calculation infrastructure
could be reused. Certainly it seems like we've got several
copies of that running around...
r~