Re: Patch: IA64 inline integer division

[Richard Henderson <rth at redhat dot com>]

On Wed, Mar 10, 2004 at 01:26:42PM -0800, Zack Weinberg wrote:
> I worry that GCC will try to do math on HFmode and horrible things
> will happen.

Given that there won't be any HFmode arithmetic patterns,
I don't see how that could happen.  At minimum, optabs.c
will try to promote to SFmode for that, even thought that
might not necessarily be correct (HFmode will have a 
larger exponent than SFmode).

I think all the mode narrowing stuff happens at the tree
level though, so we should be safe.


r~