RFR(M): 8213528: fp registers should not need to be saved around a CallLeafNoFP
Vladimir Kozlov
vladimir.kozlov at oracle.com
Thu Nov 8 19:07:09 UTC 2018
The we don't need CallLeafNoFPNode, IdealKit::make_leaf_call_no_fp() and NO_FP value. Nobody use it
after your change. Or I am missing something?
Vladimir
On 11/8/18 3:01 AM, Roland Westrelin wrote:
>
> http://cr.openjdk.java.net/~roland/8213528/webrev.00/
>
> Currently CallLeafNoFP and CallLeaf don't differ in any way. My
> understanding is that CallLeafNoFP should give the guarantee that it
> leaves fp registers untouched and so fp registers shouldn't be saved
> around a CallLeafNoFP. That's what this change implements. To be on the
> safe side, all CallLeafNoFP calls are turned into CallLeaf and existing
> behavior is preserved. That's another change motivated by Shenandoah
> where we have a use for this.
>
> Roland.
>
More information about the hotspot-compiler-dev
mailing list