On Fri, May 06, 2011 at 06:30:07PM +0200, Jakub Jelinek wrote:
> On Fri, May 06, 2011 at 12:21:24PM -0400, Michael Meissner wrote:
> > On Thu, Apr 21, 2011 at 03:02:10PM -0400, Michael Meissner wrote:
> > > In looking at some improvements to the powerpc, we wanted to change the 
> > > default
> > > for when a table jump is generated vs. a series of if statements.  Now, we
> > > could just add a powerpc specific TARGET_CASE_VALUES_THRESHOLD, but I 
> > > tend to
> > > think that these should be settable on all/most ports with --param.
> > > 
> > > At present, there are only two ports (avr and mn10300) that define their 
> > > own
> > > TARGET_CASE_VALUES_THRESHOLD hook.  My first patch does not remove the 
> > > target
> > > hook and modify the avr/mn10300 ports to use maybe_set_param_value, but 
> > > that
> > > can be done if desired.
> > > 
> > > The patch adds two --param values, one for when the port is using the 
> > > casesi
> > > insn, and the other when it uses the more primitive tablejump insn.
> > > 
> > > I have bootstrapped the compiler with this patch and run the test suite 
> > > with no
> > > regressions.  Is it ok to apply as is?  Should I modify the avr and 
> > > mn10300
> > > ports to use the parameters and do away with the target hook?  Or should 
> > > I do
> > > this just as a powerpc target hook?
> > 
> > I never got a response for this, and my earlier ping didn't seem to go out.
> > I'll check it in on Monday if there are no objections.
> 
> I think it is very weird to have two different params, if we need any such
> param, there should be just one and its default value should depend on
> HAVE_casesi.

The problem is the values in params.def must be constant, and can't depend on
switches.  I imagine we can have a single param that is normally 0, and if it
is non-zero use that value, otherwise fall back to (HAVE_casesi ? 4 : 5).  Or
we could set it in finish_options in opts.c.  Any preference?

-- 
Michael Meissner, IBM
5 Technology Place Drive, M/S 2757, Westford, MA 01886-3141, USA
meiss...@linux.vnet.ibm.com     fax +1 (978) 399-6899

Reply via email to