PowerPC64 support for operators with overflow check

[gustavotemple]

This adds fast architecture-specific implementations of
the following functions for the ppc64:

• fast_increment_function
• fast_decrement_function

[remicollet]

Test against master on

Test on POWER8 (architected), altivec supported
Linux foo 3.10.0-123.el7.ppc64 #1 SMP Mon May 5 11:18:37 EDT 2014 ppc64 ppc64 ppc64 GNU/Linux

5 new failed tests

$ cat tests/lang/operators/postdec_basiclong_64bit.diff
004+ int(-4332462841530417152)
004- float(-9.2233720368548E+18)

$ cat tests/lang/operators/postinc_basiclong_64bit.diff
002+ int(4890909195324358656)
002- float(9.2233720368548E+18)

$ cat tests/lang/operators/predec_basiclong_64bit.diff
004+ int(-4332462841530417152)
004- float(-9.2233720368548E+18)

$ cat tests/lang/operators/preinc_basiclong_64bit.diff
002+ int(4890909195324358656)
002- float(9.2233720368548E+18)

$ cat Zend/tests/decrement_001_64bit.diff
023+ int(-4332462841530417152)
024- float(-9.2233720368548E+18)

[gustavotemple]

@remicollet, sorry, I tried, but I couldn't reproduce the error.

I have two environments:

POWER8 
Linux debian-le-1-2 3.13.4 #2 SMP Fri Apr 11 15:34:54 UTC 2014 ppc64le GNU/Linux

POWER7+
Linux debian-le-1-2 3.13.4 #2 SMP Fri Apr 11 15:34:54 UTC 2014 ppc64le GNU/Linux

Do you have some tip?

Did you try build with --disable-all?

Thanks.

[Tyrael]

maybe @remicollet is testing on a bigendian platform?

[remicollet]

Yes, I tested on a POWER8 bigendian.

[gustavotemple]

@Tyrael, @remicollet, fixed, working well in LE and BE.

[remicollet]

Tested on (bigendian)

POWER8 (architected), altivec supported
Linux foo 3.10.0-216.el7.ppc64 #1 SMP Tue Dec 9 15:02:02 EST 2014 ppc64 ppc64 ppc64 GNU/Linux

No regression detected,
Will merged it.

[php-pulls]

Comment on behalf of remi at php.net:

merged