This is the mail archive of the
mailing list for the GCC project.
Re: RFC patch: invariant addresses too cheap
On 10/31/2009 02:45 PM, Richard Sandiford wrote:
I prefer the abstract address costs we have now, and in that situation,
Michael's original suggestion of having a target-defined threshold
macro seems more natural. I'm not sure why it got shot down.
I guess it was just "this patch is ready".
I said I considered it premature, but this does not mean I would oppose
it in any way, especially after you brought up an argument in favor of
the target hook.
However, now I'm thinking this: if the address_cost is just meant to be
compared *except for this particular occasion*, why not make "<= 0"
cheap? This would of course need fixing the ix86 backend too.