RFR: 8295159: DSO created with -ffast-math breaks Java floating-point arithmetic [v15]

[Andrew Haley]

On 10/27/23 13:43, Thomas Stuefe wrote:
> On Fri, 27 Oct 2023 11:49:16 GMT, Andrew Haley <aph at openjdk.org> wrote:
> 
>>> One more thought, it would be good to add the FTZ_mode_enabled check to `os::run_periodic_checks()`.
>>> We already do signal handler checks there, and it is the right place to check for "global things third party native code may mess up". It runs when one uses `-XX:CheckJNICalls`. If a native library messes with fenv, one will get a delayed assertion, with a hs-err file that lists all the shared objects.
>>
>> That's a terrific idea, but maybe it'd want a CSR?
> 
> Maybe. CheckJNICalls is a product switch and its behavior has not changed since initial check-in AFAICS.
> 
> Since CSR-freeze is close for JDK 22, maybe this is better as a separate RFE. If you want, I can take care of it.

I think so. This patch has already suffered greatly from "Wouldn't it be nice if..."