[gcc(refs/users/meissner/heads/work252-float)] Add conversions between _Float16 and float/double.

https://gcc.gnu.org/g:d8c4b1bba6cb245a217cbbc4d027666656754c6b

commit d8c4b1bba6cb245a217cbbc4d027666656754c6b
Author: Michael Meissner <[email protected]>
Date:   Fri Jul 24 00:54:07 2026 -0400

    Add conversions between _Float16 and float/double.
    
    This patch adds support to generate xscvhpdp and xscvdphp on Power9 systems and
    later, to convert between _Float16 and float scalar values.
    
    I have committed all of the patches in my backlog (dense math registers, other
    -mcpu=future instructions, random bug fixes, support for _Float16 and
    __bfloat16, and optimizations for vector logical operations on power10/power11)
    into the IBM vendor branch:
    
            vendors/ibm/gcc-17-future
    
    2026-07-24  Michael Meissner  <[email protected]>
    
    gcc/
    
            * config/rs6000/float16.md (FP16_HW): New mode iterator.
            (extendhf<mode>2): Add support converting between HFmode and
            SFmode/DFmoded if we are on power9 or later.
            (trunc<mode>hf2): Likewise.
            * config/rs6000/rs6000-c.cc (rs6000_target_modify_macros): Define
            __FLOAT16_HW__ if we have hardware support for _Float16.
            * config/rs6000/rs6000.cc (rs6000_init_hard_regno_mode_ok): Mark that we
            use VSX arithmetic support for V8HFmode if we are a power9 or later.

Diff:
---
 gcc/config/rs6000/float16.md  | 22 ++++++++++++++++++++++
 gcc/config/rs6000/rs6000-c.cc |  3 +++
 gcc/config/rs6000/rs6000.cc   |  3 +++
 3 files changed, 28 insertions(+)

diff --git a/gcc/config/rs6000/float16.md b/gcc/config/rs6000/float16.md
index 967a6f03612c..7c66c527e8d8 100644
--- a/gcc/config/rs6000/float16.md
+++ b/gcc/config/rs6000/float16.md
@@ -26,6 +26,10 @@
 (define_mode_iterator VS_FP16	[BF HF V8BF V8HF])
 (define_mode_iterator VFP16	[V8BF V8HF])
 
+;; Mode iterator for 16-bit floating point modes on machines with
+;; hardware support both as a scalar and as a vector.
+(define_mode_iterator FP16_HW [(HF "TARGET_FLOAT16_HW")])
+
 ;; Mode attribute giving the vector mode for a 16-bit floating point
 ;; scalar in both upper and lower case.
 (define_mode_attr FP16_VECTOR8 [(BF "V8BF")
@@ -157,3 +161,21 @@
 }
   [(set_attr "type" "veclogical,vecperm")
    (set_attr "prefixed" "*,yes")])
+
+;; Convert IEEE 16-bit floating point to/from other floating point modes.
+
+(define_insn "extendhf<mode>2"
+  [(set (match_operand:SFDF 0 "vsx_register_operand" "=wa")
+	(float_extend:SFDF
+	 (match_operand:HF 1 "vsx_register_operand" "wa")))]
+  "TARGET_FLOAT16_HW"
+  "xscvhpdp %x0,%x1"
+  [(set_attr "type" "fpsimple")])
+
+(define_insn "trunc<mode>hf2"
+  [(set (match_operand:HF 0 "vsx_register_operand" "=wa")
+	(float_truncate:HF
+	 (match_operand:SFDF 1 "vsx_register_operand" "wa")))]
+  "TARGET_FLOAT16_HW"
+  "xscvdphp %x0,%x1"
+  [(set_attr "type" "fpsimple")])
diff --git a/gcc/config/rs6000/rs6000-c.cc b/gcc/config/rs6000/rs6000-c.cc
index ca3ac7b25810..e8eecabefce8 100644
--- a/gcc/config/rs6000/rs6000-c.cc
+++ b/gcc/config/rs6000/rs6000-c.cc
@@ -588,6 +588,9 @@ rs6000_target_modify_macros (bool define_p, HOST_WIDE_INT flags)
     {
       rs6000_define_or_undefine_macro (define_p, "__FLOAT16__");
       rs6000_define_or_undefine_macro (define_p, "__BFLOAT16__");
+
+      if ((flags & OPTION_MASK_P9_VECTOR) != 0)
+	rs6000_define_or_undefine_macro (define_p, "__FLOAT16_HW__");
     }
   /* Tell the user if we are targeting CELL.  */
   if (rs6000_cpu == PROCESSOR_CELL)
diff --git a/gcc/config/rs6000/rs6000.cc b/gcc/config/rs6000/rs6000.cc
index 1411ad3388a0..d3f4b8c5002e 100644
--- a/gcc/config/rs6000/rs6000.cc
+++ b/gcc/config/rs6000/rs6000.cc
@@ -3000,6 +3000,9 @@ rs6000_init_hard_regno_mode_ok (bool global_init_p)
       rs6000_vector_mem[BFmode] = VECTOR_VSX;
       rs6000_vector_align[HFmode] = 16;
       rs6000_vector_align[BFmode] = 16;
+
+      if (TARGET_P9_VECTOR)
+	rs6000_vector_unit[V8HFmode] = VECTOR_VSX;
     }
 
   /* DFmode, see if we want to use the VSX unit.  Memory is handled