Fesetenv does not change fpu rounding mode. This bug already described
in https://sourceforge.net/p/mingw-w64/bugs/541/
--
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based dev
FWIW, I agree that this code does not look right.
The purpose of fesetenv(const fenv_t * envp) is to change the settings
of the floating point environment as specified by the parameter.
However, instead of using the values from the parameter, the current
code *overwrites* the passed in setting
Ahh, the wonders of google. I believe I have found the problem kai's
change was originally intended to fix:
http://mingw-w64-public.narkive.com/jr5jMaJ2/quadmath-h-s-expq-crashes-at-runtime
And, I believe I have found what the actual problem was. I don't think
what kai did was the right fi
On 12/21/2016 07:29 AM, David Wohlferd wrote:
> Ahh, the wonders of google. I believe I have found the problem kai's
> change was originally intended to fix:
> http://mingw-w64-public.narkive.com/jr5jMaJ2/quadmath-h-s-expq-crashes-at-runtime
>
>
> And, I believe I have found what the actual prob
On 12/21/2016 4:27 AM, JonY wrote:
> On 12/21/2016 07:29 AM, David Wohlferd wrote:
>> Comments? Zidane, can you try this?
> David, can you also test against the gfortran testsuites?
Since mingw-w64 doesn't have any gfortran testsuites, I assume you want
me to run gcc's. Running "make -k check-f
On 12/22/2016 6:52 AM, David Wohlferd wrote:
> On 12/21/2016 4:27 AM, JonY wrote:
>> On 12/21/2016 07:29 AM, David Wohlferd wrote:
>>> Comments? Zidane, can you try this?
>> David, can you also test against the gfortran testsuites?
>
> Since mingw-w64 doesn't have any gfortran testsuites, I assum
On 12/22/2016 11:52 AM, David Wohlferd wrote:
> On 12/21/2016 4:27 AM, JonY wrote:
>> On 12/21/2016 07:29 AM, David Wohlferd wrote:
>>> Comments? Zidane, can you try this?
>> David, can you also test against the gfortran testsuites?
>
> Since mingw-w64 doesn't have any gfortran testsuites, I assu
OK, clear to apply to master, thanks for testing!
When I went to push, I had a problem with git that took me a while to
sort out. While I was doing that, I realized that this patch wasn't
quite right (so I never push-ed it). This x87 stuff is just weird.
According to 7.6.4.2 in c99, feho
Attached is an alternate fix. This does everything the previous patch
does, plus a bit more.
On the plus side it doesn't try to cram the SSE settings into the x87
settings. This not only makes things cleaner to code from the library
side, it makes things easier for users who actually want to