On Fri, Jul 28, 2017 at 04:08:50PM -0500, Segher Boessenkool wrote:
> 
> "emit" is not a good name for this: that is generally used for something
> that does emit_insn, i.e. put an insn in the instruction stream.  This
> function returns a string a define_insn can return.  For the rl* insns
> I called the similar functions rs6000_insn_for_*, maybe something like
> that is better here?

...

> I think calling this with the rtx elementN args makes this only more
> complicated (the function comment doesn't say what they are or what
> NULL means, btw).

...

> In this and the other testcase, should you test no other insns at all
> are generated?

Here are the revised patches.  I tested on a little endian power8 system and a
big endian power7 system.  Are these patches ok for the trunk?

[gcc]
2017-08-02  Michael Meissner  <meissner@linux.vnet.ibm.com>

	PR target/81593
	* config/rs6000/rs6000-protos.h (rs6000_output_xxpermdi): New
	declaration.
	* config/rs6000/rs6000.c (rs6000_output_xxpermdi): New function to
	emit XXPERMDI accessing either double word in either vector
	register inputs.
	* config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D iterator):
	Rewrite VEC_CONCAT insn to call rs6000_output_xxpermdi.  Simplify
	the constraints with the removal of the -mupper-regs-* switches.
	(vsx_concat_<mode>_1): New combiner insns to optimize CONCATs
	where either register might have come from VEC_SELECT.
	(vsx_concat_<mode>_2): Likewise.
	(vsx_concat_<mode>_3): Likewise.
	(vsx_set_<mode>, VSX_D iterator): Rewrite insn to generate a
	VEC_CONCAT rather than use an UNSPEC to specify the option.

[gcc/testsuite]
2017-08-02  Michael Meissner  <meissner@linux.vnet.ibm.com>

	PR target/81593
	* gcc.target/powerpc/vsx-extract-6.c: New test.
	* gcc.target/powerpc/vsx-extract-7.c: Likewise.


-- 
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797