[Beignet] pow(n), erf(c), tgamma give wrong results

[Zhigang Gong]

On Wed, Nov 05, 2014 at 08:27:46AM +0000, Rebecca N. Palmer wrote:
> (Referring to https://bugs.debian.org/cgi-bin/bugreport.cgi?msg=5;filename=Fix-pow-erf-tgamma.patch;att=3;bug=768090
> )
> 
> >And it triggers some other bugs related to constant expression
> >handling when run all unit test cases.
> I didn't see that when I ran the tests in 0.9.3 on IvyBridge M GT2
> (only builtin_pow and builtin_tgamma failed, both due to the
> absolute error checking you've already fixed) but your patch says
> the problem does exist in 0.9.x: how do I test for this bug?  Debian
> testing is still on 0.8, and can't change to 0.9 now due to freeze
> rules.

I'm using LLVM 3.5. Maybe you are using a different LLVM version.
If you can try llvm/clang 3.5, it should be triggerred.

> 
> >As to the license issue, we will fix it to a consistent version.
> >I think change all statement to LGPL2.1+ should be ok.
> The lack of a Signed-off-by line was solely due to the license issue
> (I don't know if there even is a difference between 2.0 and 2.1
> other than the name change from Library to Lesser GPL), so given the
> above:

I haven't found any difference indication between LGPL v2.1 and v2.0.
And given the fact that our COPYING is already v2.1, and all LGPL v2.0
stuff could be safely change to LGPLv2.1 according to the license,
so I will send another patch to make all things consistent to LGPL v2.1.

Thanks for your contribution. 

> 
> Signed-off-by: Rebecca Palmer <rebecca_palmer at zoho.com>
>