https://gcc.gnu.org/g:2f541cae69f67d9e61b6bafa6f4847fa27fdaf05
commit 2f541cae69f67d9e61b6bafa6f4847fa27fdaf05
Author: Michael Meissner <meiss...@linux.ibm.com>
Date: Tue Jun 10 18:24:33 2025 -0400
PR target/117251: Add PowerPC XXEVAL support to speed up SHA3 calculations
2025-06-10 Michael Meissner <meiss...@linux.ibm.com>
gcc/testsuite/
PR target/117251
* gcc.target/powerpc/p10-vector-fused-1.c: New test.
* gcc.target/powerpc/p10-vector-fused-2.c: Likewise.
Diff:
---
.../gcc.target/powerpc/p10-vector-fused-1.c | 409 +++++++++
.../gcc.target/powerpc/p10-vector-fused-2.c | 936 +++++++++++++++++++++
2 files changed, 1345 insertions(+)
diff --git a/gcc/testsuite/gcc.target/powerpc/p10-vector-fused-1.c
b/gcc/testsuite/gcc.target/powerpc/p10-vector-fused-1.c
new file mode 100644
index 000000000000..28e0874b3454
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/p10-vector-fused-1.c
@@ -0,0 +1,409 @@
+/* { dg-do run } */
+/* { dg-require-effective-target power10_hw } */
+/* { dg-options "-mdejagnu-cpu=power10 -O2" } */
+
+/* Generate and check most of the vector logical instruction combinations that
+ may or may not generate xxeval to do a fused operation on power10. */
+
+#include <stddef.h>
+#include <stdlib.h>
+#include <altivec.h>
+
+#ifdef DEBUG
+#include <stdio.h>
+
+static int errors = 0;
+static int tests = 0;
+#endif
+
+typedef vector unsigned int vector_t;
+typedef unsigned int scalar_t;
+
+/* Vector logical functions. */
+static inline vector_t
+vector_and (vector_t x, vector_t y)
+{
+ return x & y;
+}
+
+static inline vector_t
+vector_or (vector_t x, vector_t y)
+{
+ return x | y;
+}
+
+static inline vector_t
+vector_xor (vector_t x, vector_t y)
+{
+ return x ^ y;
+}
+
+static inline vector_t
+vector_andc (vector_t x, vector_t y)
+{
+ return x & ~y;
+}
+
+static inline vector_t
+vector_orc (vector_t x, vector_t y)
+{
+ return x | ~y;
+}
+
+static inline vector_t
+vector_nand (vector_t x, vector_t y)
+{
+ return ~(x & y);
+}
+
+static inline vector_t
+vector_nor (vector_t x, vector_t y)
+{
+ return ~(x | y);
+}
+
+static inline vector_t
+vector_eqv (vector_t x, vector_t y)
+{
+ return ~(x ^ y);
+}
+
+/* Scalar logical functions. */
+static inline scalar_t
+scalar_and (scalar_t x, scalar_t y)
+{
+ return x & y;
+}
+
+static inline scalar_t
+scalar_or (scalar_t x, scalar_t y)
+{
+ return x | y;
+}
+
+static inline scalar_t
+scalar_xor (scalar_t x, scalar_t y)
+{
+ return x ^ y;
+}
+
+static inline scalar_t
+scalar_andc (scalar_t x, scalar_t y)
+{
+ return x & ~y;
+}
+
+static inline scalar_t
+scalar_orc (scalar_t x, scalar_t y)
+{
+ return x | ~y;
+}
+
+static inline scalar_t
+scalar_nand (scalar_t x, scalar_t y)
+{
+ return ~(x & y);
+}
+
+static inline scalar_t
+scalar_nor (scalar_t x, scalar_t y)
+{
+ return ~(x | y);
+}
+
+static inline scalar_t
+scalar_eqv (scalar_t x, scalar_t y)
+{
+ return ~(x ^ y);
+}
+
+
+/*
+ * Generate one function for each combination that we are checking. Do 4
+ * operations:
+ *
+ * Use FPR regs that should generate either XXEVAL or XXL* insns;
+ * Use Altivec registers than may generated fused V* insns;
+ * Use VSX registers, insure fusing it not done via asm; (and)
+ * Use GPR registers on scalar operations.
+ */
+
+#ifdef DEBUG
+#define TRACE(INNER, OUTER) \
+ do { \
+ tests++; \
+ printf ("%s_%s\n", INNER, OUTER); \
+ fflush (stdout); \
+ } while (0) \
+
+#define FAILED(INNER, OUTER) \
+ do { \
+ errors++; \
+ printf ("%s_%s failed\n", INNER, OUTER); \
+ fflush (stdout); \
+ } while (0) \
+
+#else
+#define TRACE(INNER, OUTER)
+#define FAILED(INNER, OUTER) abort ()
+#endif
+
+#define FUSED_FUNC(INNER, OUTER) \
+static void \
+INNER ## _ ## OUTER (vector_t a, vector_t b, vector_t c) \
+{ \
+ vector_t f_a, f_b, f_c, f_r, f_t; \
+ vector_t v_a, v_b, v_c, v_r, v_t; \
+ vector_t w_a, w_b, w_c, w_r, w_t; \
+ scalar_t s_a, s_b, s_c, s_r, s_t; \
+ \
+ TRACE (#INNER, #OUTER); \
+ \
+ f_a = a; \
+ f_b = b; \
+ f_c = c; \
+ \
+ __asm__ (" # fpr regs: %x0,%x1,%x2 " #INNER "_" #OUTER \
+ : "+d" (f_a), \
+ "+d" (f_b), \
+ "+d" (f_c)); \
+ \
+ f_t = vector_ ## INNER (f_b, f_c); \
+ f_r = vector_ ## OUTER (f_a, f_t); \
+ \
+ __asm__ (" # fpr regs result: %x0 " #INNER "_" #OUTER
\
+ : "+d" (f_r)); \
+ \
+ v_a = a; \
+ v_b = b; \
+ v_c = c; \
+ \
+ __asm__ (" # altivec regs: %x0,%x1,%x2 " #INNER "_" #OUTER \
+ : "+v" (v_a), \
+ "+v" (v_b), \
+ "+v" (v_c)); \
+ \
+ v_t = vector_ ## INNER (v_b, v_c); \
+ v_r = vector_ ## OUTER (v_a, v_t); \
+ \
+ __asm__ (" # altivec regs result: %x0 " #INNER "_" #OUTER \
+ : "+v" (v_r)); \
+ \
+ w_a = a; \
+ w_b = b; \
+ w_c = c; \
+ \
+ __asm__ (" # vsx regs: %x0,%x1,%x2 " #INNER "_" #OUTER \
+ : "+wa" (w_a), \
+ "+wa" (w_b), \
+ "+wa" (w_c)); \
+ \
+ w_t = vector_ ## INNER (w_b, w_c); \
+ __asm__ ("nop # break vsx fusion reg %x0" : "+wa" (w_t)); \
+ w_r = vector_ ## OUTER (w_a, w_t); \
+ \
+ __asm__ (" # vsx regs result: %x0 " #INNER "_" #OUTER
\
+ : "+wa" (w_r)); \
+ \
+ s_a = a[0]; \
+ s_b = b[0]; \
+ s_c = c[0]; \
+ \
+ __asm__ (" # gpr regs: %0,%1,%2 " #INNER "_" #OUTER \
+ : "+r" (s_a), \
+ "+r" (s_b), \
+ "+r" (s_c)); \
+ \
+ s_t = scalar_ ## INNER (s_b, s_c); \
+ s_r = scalar_ ## OUTER (s_a, s_t); \
+ \
+ __asm__ (" # gpr regs result: %0 " #INNER "_" #OUTER \
+ : "+r" (s_r)); \
+ \
+ if (!vec_all_eq (w_r, f_r) \
+ || !vec_all_eq (w_r, v_r)
\
+ || s_r != w_r[0])
\
+ FAILED (#INNER, #OUTER); \
+ \
+ return; \
+}
+
+FUSED_FUNC (and, and)
+FUSED_FUNC (andc, and)
+FUSED_FUNC (eqv, and)
+FUSED_FUNC (nand, and)
+FUSED_FUNC (nor, and)
+FUSED_FUNC (or, and)
+FUSED_FUNC (orc, and)
+FUSED_FUNC (xor, and)
+
+FUSED_FUNC (and, andc)
+FUSED_FUNC (andc, andc)
+FUSED_FUNC (eqv, andc)
+FUSED_FUNC (nand, andc)
+FUSED_FUNC (nor, andc)
+FUSED_FUNC (or, andc)
+FUSED_FUNC (orc, andc)
+FUSED_FUNC (xor, andc)
+
+FUSED_FUNC (and, eqv)
+FUSED_FUNC (andc, eqv)
+FUSED_FUNC (eqv, eqv)
+FUSED_FUNC (nand, eqv)
+FUSED_FUNC (nor, eqv)
+FUSED_FUNC (or, eqv)
+FUSED_FUNC (orc, eqv)
+FUSED_FUNC (xor, eqv)
+
+FUSED_FUNC (and, nand)
+FUSED_FUNC (andc, nand)
+FUSED_FUNC (eqv, nand)
+FUSED_FUNC (nand, nand)
+FUSED_FUNC (nor, nand)
+FUSED_FUNC (or, nand)
+FUSED_FUNC (orc, nand)
+FUSED_FUNC (xor, nand)
+
+FUSED_FUNC (and, nor)
+FUSED_FUNC (andc, nor)
+FUSED_FUNC (eqv, nor)
+FUSED_FUNC (nand, nor)
+FUSED_FUNC (nor, nor)
+FUSED_FUNC (or, nor)
+FUSED_FUNC (orc, nor)
+FUSED_FUNC (xor, nor)
+
+FUSED_FUNC (and, or)
+FUSED_FUNC (andc, or)
+FUSED_FUNC (eqv, or)
+FUSED_FUNC (nand, or)
+FUSED_FUNC (nor, or)
+FUSED_FUNC (or, or)
+FUSED_FUNC (orc, or)
+FUSED_FUNC (xor, or)
+
+FUSED_FUNC (and, orc)
+FUSED_FUNC (andc, orc)
+FUSED_FUNC (eqv, orc)
+FUSED_FUNC (nand, orc)
+FUSED_FUNC (nor, orc)
+FUSED_FUNC (or, orc)
+FUSED_FUNC (orc, orc)
+FUSED_FUNC (xor, orc)
+
+FUSED_FUNC (and, xor)
+FUSED_FUNC (andc, xor)
+FUSED_FUNC (eqv, xor)
+FUSED_FUNC (nand, xor)
+FUSED_FUNC (nor, xor)
+FUSED_FUNC (or, xor)
+FUSED_FUNC (orc, xor)
+FUSED_FUNC (xor, xor)
+
+
+/* List of functions to check. */
+typedef void func_t (vector_t,
+ vector_t,
+ vector_t);
+
+typedef func_t *ptr_func_t;
+
+static ptr_func_t functions[] = {
+ and_and,
+ andc_and,
+ eqv_and,
+ nand_and,
+ nor_and,
+ or_and,
+ orc_and,
+ xor_and,
+
+ and_andc,
+ andc_andc,
+ eqv_andc,
+ nand_andc,
+ nor_andc,
+ or_andc,
+ orc_andc,
+ xor_andc,
+
+ and_eqv,
+ andc_eqv,
+ eqv_eqv,
+ nand_eqv,
+ nor_eqv,
+ or_eqv,
+ orc_eqv,
+ xor_eqv,
+
+ and_nand,
+ andc_nand,
+ eqv_nand,
+ nand_nand,
+ nor_nand,
+ or_nand,
+ orc_nand,
+ xor_nand,
+
+ and_nor,
+ andc_nor,
+ eqv_nor,
+ nand_nor,
+ nor_nor,
+ or_nor,
+ orc_nor,
+ xor_nor,
+
+ and_or,
+ andc_or,
+ eqv_or,
+ nand_or,
+ nor_or,
+ or_or,
+ orc_or,
+ xor_or,
+
+ and_orc,
+ andc_orc,
+ eqv_orc,
+ nand_orc,
+ nor_orc,
+ or_orc,
+ orc_orc,
+ xor_orc,
+
+ and_xor,
+ andc_xor,
+ eqv_xor,
+ nand_xor,
+ nor_xor,
+ or_xor,
+ orc_xor,
+ xor_xor,
+};
+
+
+int
+main (void)
+{
+ scalar_t s_a = 0x0fu;
+ scalar_t s_b = 0xaau;
+ scalar_t s_c = 0xccu;
+
+ vector_t a = (vector_t) { s_a, s_a, ~s_a, ~s_a };
+ vector_t b = (vector_t) { s_b, ~s_b, s_b, ~s_b };
+ vector_t c = (vector_t) { s_c, ~s_c, ~s_c, s_c };
+
+ size_t i;
+
+ for (i = 0; i < sizeof (functions) / sizeof (functions[0]); i++)
+ functions[i] (a, b, c);
+
+#ifdef DEBUG
+ printf ("Done, %d tests, %d failures\n", tests, errors);
+ return errors;
+
+#else
+ return 0;
+#endif
+}
diff --git a/gcc/testsuite/gcc.target/powerpc/p10-vector-fused-2.c
b/gcc/testsuite/gcc.target/powerpc/p10-vector-fused-2.c
new file mode 100644
index 000000000000..f074622c9f67
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/p10-vector-fused-2.c
@@ -0,0 +1,936 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target power10_ok } */
+/* { dg-options "-mdejagnu-cpu=power10 -O2" } */
+
+/* Make sure all of the fusion cases that generate the xxeval instruction
+ actually generate it. */
+typedef vector unsigned int vector_t;
+
+static inline vector_t
+vector_and (vector_t x, vector_t y)
+{
+ return x & y;
+}
+
+static inline vector_t
+vector_or (vector_t x, vector_t y)
+{
+ return x | y;
+}
+
+static inline vector_t
+vector_xor (vector_t x, vector_t y)
+{
+ return x ^ y;
+}
+
+static inline vector_t
+vector_andc (vector_t x, vector_t y)
+{
+ return x & ~y;
+}
+
+static inline vector_t
+vector_orc (vector_t x, vector_t y)
+{
+ return x | ~y;
+}
+
+static inline vector_t
+vector_nand (vector_t x, vector_t y)
+{
+ return ~(x & y);
+}
+
+static inline vector_t
+vector_nor (vector_t x, vector_t y)
+{
+ return ~(x | y);
+}
+
+static inline vector_t
+vector_eqv (vector_t x, vector_t y)
+{
+ return ~(x ^ y);
+}
+
+void
+and_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,1. */
+ r = vector_and (a, vector_and (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+and_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,14. */
+ r = vector_andc (a, vector_and (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+and_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,31. */
+ r = vector_or (a, vector_and (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+and_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,239. */
+ r = vector_orc (a, vector_and (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+and_xor (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,30. */
+ r = vector_xor (a, vector_and (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,14. */
+ r = vector_andc (a, vector_and (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,11. */
+ r = vector_andc (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_eqv (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,210. */
+ r = vector_eqv (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_nand (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,239. */
+ r = vector_nand (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,47. */
+ r = vector_or (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,191. */
+ r = vector_orc (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+andc_xor (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,45. */
+ r = vector_xor (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+eqv_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,9. */
+ r = vector_and (a, vector_eqv (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+eqv_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,210. */
+ r = vector_eqv (a, vector_andc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+eqv_eqv (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,105. */
+ r = vector_eqv (a, vector_eqv (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+eqv_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,159. */
+ r = vector_or (a, vector_eqv (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+eqv_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,111. */
+ r = vector_orc (a, vector_eqv (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nand_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,14. */
+ r = vector_and (a, vector_nand (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nand_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,1. */
+ r = vector_andc (a, vector_nand (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nand_eqv (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,30. */
+ r = vector_eqv (a, vector_nand (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nand_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,2. */
+ r = vector_nor (a, vector_nand (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nand_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,31. */
+ r = vector_orc (a, vector_nand (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nor_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,8. */
+ r = vector_and (a, vector_nor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nor_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,7. */
+ r = vector_andc (a, vector_nor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nor_eqv (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,120. */
+ r = vector_eqv (a, vector_nor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nor_nand (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,191. */
+ r = vector_nand (a, vector_nor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nor_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,143. */
+ r = vector_or (a, vector_nor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+nor_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,127. */
+ r = vector_orc (a, vector_nor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+or_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,7. */
+ r = vector_and (a, vector_or (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+or_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,8. */
+ r = vector_andc (a, vector_or (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+or_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,127. */
+ r = vector_or (a, vector_or (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+or_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,143. */
+ r = vector_orc (a, vector_or (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+or_xor (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,120. */
+ r = vector_xor (a, vector_or (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,11. */
+ r = vector_and (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,2. */
+ r = vector_andc (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_eqv (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,75. */
+ r = vector_eqv (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_nor (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,8. */
+ r = vector_nor (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,191. */
+ r = vector_or (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,47. */
+ r = vector_orc (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+orc_xor (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,180. */
+ r = vector_xor (a, vector_orc (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+xor_and (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,6. */
+ r = vector_and (a, vector_xor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+xor_andc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,9. */
+ r = vector_andc (a, vector_xor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+xor_nand (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,249. */
+ r = vector_nand (a, vector_xor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+xor_or (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,111. */
+ r = vector_or (a, vector_xor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+xor_orc (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,159. */
+ r = vector_orc (a, vector_xor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+void
+xor_xor (vector_t *p_a, vector_t *p_b, vector_t *p_c, vector_t *p_r)
+{
+ vector_t a = *p_a;
+ vector_t b = *p_b;
+ vector_t c = *p_c;
+ vector_t r;
+
+ __asm__ (" # force fpr registers, %x0,%x1,%x2"
+ : "+d" (a), "+d" (b), "+d" (c));
+
+ /* xxeval r,a,b,c,105. */
+ r = vector_xor (a, vector_xor (b, c));
+
+ __asm__ (" # force fpr result, %x0" : "+d" (r));
+ *p_r = r;
+ return;
+}
+
+/* Make sure none of traditional logical instructions are generated. Skip
+ checking for xxlor in case the register allocator decides to add some vector
+ moves. */
+/* { dg-final { scan-assembler-not {\mv(and|or|xor|andc|orc|nand|nor|eqv)\M}
} } */
+/* { dg-final { scan-assembler-not {\mxxl(and|xor|andc|orc|nand|nor|eqv)\M}
} } */
+/* { dg-final { scan-assembler-times {\mxxeval\M} 46 } } */
