[Qemu-devel] [PATCH 08/14] Add vmaddfp and vnmsubfp instructions

[Nathan Froyd <froydnj@codesourcery.com>]

Do the computation in higher precision and check for invalid operation
conditions.

Signed-off-by: Nathan Froyd <froydnj@codesourcery.com>
---
 target-ppc/helper.h    |    2 +
 target-ppc/op_helper.c |   54 ++++++++++++++++++++++++++++++++++++++++++++++++
 target-ppc/translate.c |    1 +
 3 files changed, 57 insertions(+), 0 deletions(-)

diff --git a/target-ppc/helper.h b/target-ppc/helper.h
index a5d1972..6f5d6a6 100644
--- a/target-ppc/helper.h
+++ b/target-ppc/helper.h
@@ -236,6 +236,8 @@ DEF_HELPER_3(vaddfp, void, avr, avr, avr)
 DEF_HELPER_3(vsubfp, void, avr, avr, avr)
 DEF_HELPER_3(vmaxfp, void, avr, avr, avr)
 DEF_HELPER_3(vminfp, void, avr, avr, avr)
+DEF_HELPER_4(vmaddfp, void, avr, avr, avr, avr)
+DEF_HELPER_4(vnmsubfp, void, avr, avr, avr, avr)
 
 DEF_HELPER_1(efscfsi, i32, i32)
 DEF_HELPER_1(efscfui, i32, i32)
diff --git a/target-ppc/op_helper.c b/target-ppc/op_helper.c
index d952113..532a08a 100644
--- a/target-ppc/op_helper.c
+++ b/target-ppc/op_helper.c
@@ -2223,6 +2223,29 @@ VCMP(gtsw, >, s32)
 #undef VCMP_DO
 #undef VCMP
 
+void helper_vmaddfp (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
+{
+    int i;
+    for (i = 0; i < ARRAY_SIZE(r->f); i++) {
+        HANDLE_NAN3(r->f[i], a->f[i], b->f[i], c->f[i]) {
+            /* Need to do the computation in higher precision and round
+             * once at the end.  */
+            float64 af, cf;
+            af = float32_to_float64(a->f[i], &env->vec_status);
+            cf = float32_to_float64(c->f[i], &env->vec_status);
+            if ((float64_is_infinity(af) && float64_is_zero(cf)) ||
+                (float64_is_zero(af) && float64_is_infinity(cf))) {
+                r->u32[i] = 0x7fc00000;
+            } else {
+                float64 t = float64_mul(af, cf, &env->vec_status);
+                float64 bf = float32_to_float64(b->f[i], &env->vec_status);
+                t = float64_add(t, bf, &env->vec_status);
+                r->f[i] = float64_to_float32(t, &env->vec_status);
+            }
+        }
+    }
+}
+
 void helper_vmhaddshs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
 {
     int sat = 0;
@@ -2460,6 +2483,37 @@ VMUL(uh, u16, u32)
 #undef VMUL_DO
 #undef VMUL
 
+void helper_vnmsubfp (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
+{
+    int i;
+    for (i = 0; i < ARRAY_SIZE(r->f); i++) {
+        HANDLE_NAN3(r->f[i], a->f[i], b->f[i], c->f[i]) {
+            /* Need to do the computation is higher precision and round
+             * once at the end.  */
+            float64 af, cf;
+            af = float32_to_float64(a->f[i], &env->vec_status);
+            cf = float32_to_float64(c->f[i], &env->vec_status);
+            if ((float64_is_infinity(af) && float64_is_zero(cf)) ||
+                (float64_is_zero(af) && float64_is_infinity(cf))) {
+                r->u32[i] = 0x7fc00000;
+            } else {
+                float64 t = float64_mul(af, cf, &env->vec_status);
+                float64 bf = float32_to_float64(b->f[i], &env->vec_status);
+                if (unlikely(float64_is_infinity(t) &&
+                             float64_is_infinity(bf) &&
+                             float64_is_neg(t) == float64_is_neg(bf))) {
+                    /* Magnitude subtraction of infinities */
+                    r->u32[i] = 0x7fc00000;
+                } else {
+                    t = float64_sub(t, bf, &env->vec_status);
+                    t = float64_chs(t);
+                    r->f[i] = float64_to_float32(t, &env->vec_status);
+                }
+            }
+        }
+    }
+}
+
 void helper_vperm (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
 {
     ppc_avr_t result;
diff --git a/target-ppc/translate.c b/target-ppc/translate.c
index 6ff46e0..8b67333 100644
--- a/target-ppc/translate.c
+++ b/target-ppc/translate.c
@@ -6577,6 +6577,7 @@ GEN_VAFORM_PAIRED(vmsumubm, vmsummbm, 18)
 GEN_VAFORM_PAIRED(vmsumuhm, vmsumuhs, 19)
 GEN_VAFORM_PAIRED(vmsumshm, vmsumshs, 20)
 GEN_VAFORM_PAIRED(vsel, vperm, 21)
+GEN_VAFORM_PAIRED(vmaddfp, vnmsubfp, 23)
 
 /***                           SPE extension                               ***/
 /* Register moves */
-- 
1.6.0.5