target-ppc: convert SPE FP ops to TCG

Including a few bug fixes:
- Don't clear high part for instruction with 32-bit destination
- Fix efscmp* and etstcmp* return value

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5783 c046a42c-6fe2-441c-8c8c-71466251a162

5 files changed, 681 insertions, 1007 deletions

diff --git a/target-ppc/helper.h b/target-ppc/helper.h
index 6178a56..3fe5681 100644
--- a/target-ppc/helper.h
+++ b/target-ppc/helper.h
@@ -60,10 +60,77 @@ DEF_HELPER_3(fnmsub, i64, i64, i64, i64)
 DEF_HELPER_1(fabs, i64, i64)
 DEF_HELPER_1(fnabs, i64, i64)
 DEF_HELPER_1(fneg, i64, i64)
-DEF_HELPER_1(fsqrt, i64, i64);
-DEF_HELPER_1(fre, i64, i64);
-DEF_HELPER_1(fres, i64, i64);
-DEF_HELPER_1(frsqrte, i64, i64);
+DEF_HELPER_1(fsqrt, i64, i64)
+DEF_HELPER_1(fre, i64, i64)
+DEF_HELPER_1(fres, i64, i64)
+DEF_HELPER_1(frsqrte, i64, i64)
 DEF_HELPER_3(fsel, i64, i64, i64, i64)
 
+DEF_HELPER_1(efscfsi, i32, i32)
+DEF_HELPER_1(efscfui, i32, i32)
+DEF_HELPER_1(efscfuf, i32, i32)
+DEF_HELPER_1(efscfsf, i32, i32)
+DEF_HELPER_1(efsctsi, i32, i32)
+DEF_HELPER_1(efsctui, i32, i32)
+DEF_HELPER_1(efsctsiz, i32, i32)
+DEF_HELPER_1(efsctuiz, i32, i32)
+DEF_HELPER_1(efsctsf, i32, i32)
+DEF_HELPER_1(efsctuf, i32, i32)
+DEF_HELPER_1(evfscfsi, i64, i64)
+DEF_HELPER_1(evfscfui, i64, i64)
+DEF_HELPER_1(evfscfuf, i64, i64)
+DEF_HELPER_1(evfscfsf, i64, i64)
+DEF_HELPER_1(evfsctsi, i64, i64)
+DEF_HELPER_1(evfsctui, i64, i64)
+DEF_HELPER_1(evfsctsiz, i64, i64)
+DEF_HELPER_1(evfsctuiz, i64, i64)
+DEF_HELPER_1(evfsctsf, i64, i64)
+DEF_HELPER_1(evfsctuf, i64, i64)
+DEF_HELPER_2(efsadd, i32, i32, i32)
+DEF_HELPER_2(efssub, i32, i32, i32)
+DEF_HELPER_2(efsmul, i32, i32, i32)
+DEF_HELPER_2(efsdiv, i32, i32, i32)
+DEF_HELPER_2(evfsadd, i64, i64, i64)
+DEF_HELPER_2(evfssub, i64, i64, i64)
+DEF_HELPER_2(evfsmul, i64, i64, i64)
+DEF_HELPER_2(evfsdiv, i64, i64, i64)
+DEF_HELPER_2(efststlt, i32, i32, i32)
+DEF_HELPER_2(efststgt, i32, i32, i32)
+DEF_HELPER_2(efststeq, i32, i32, i32)
+DEF_HELPER_2(efscmplt, i32, i32, i32)
+DEF_HELPER_2(efscmpgt, i32, i32, i32)
+DEF_HELPER_2(efscmpeq, i32, i32, i32)
+DEF_HELPER_2(evfststlt, i32, i64, i64)
+DEF_HELPER_2(evfststgt, i32, i64, i64)
+DEF_HELPER_2(evfststeq, i32, i64, i64)
+DEF_HELPER_2(evfscmplt, i32, i64, i64)
+DEF_HELPER_2(evfscmpgt, i32, i64, i64)
+DEF_HELPER_2(evfscmpeq, i32, i64, i64)
+DEF_HELPER_1(efdcfsi, i64, i32)
+DEF_HELPER_1(efdcfsid, i64, i64)
+DEF_HELPER_1(efdcfui, i64, i32)
+DEF_HELPER_1(efdcfuid, i64, i64)
+DEF_HELPER_1(efdctsi, i32, i64)
+DEF_HELPER_1(efdctui, i32, i64)
+DEF_HELPER_1(efdctsiz, i32, i64)
+DEF_HELPER_1(efdctsidz, i64, i64)
+DEF_HELPER_1(efdctuiz, i32, i64)
+DEF_HELPER_1(efdctuidz, i64, i64)
+DEF_HELPER_1(efdcfsf, i64, i32)
+DEF_HELPER_1(efdcfuf, i64, i32)
+DEF_HELPER_1(efdctsf, i32, i64)
+DEF_HELPER_1(efdctuf, i32, i64)
+DEF_HELPER_1(efscfd, i32, i64)
+DEF_HELPER_1(efdcfs, i64, i32)
+DEF_HELPER_2(efdadd, i64, i64, i64)
+DEF_HELPER_2(efdsub, i64, i64, i64)
+DEF_HELPER_2(efdmul, i64, i64, i64)
+DEF_HELPER_2(efddiv, i64, i64, i64)
+DEF_HELPER_2(efdtstlt, i32, i64, i64)
+DEF_HELPER_2(efdtstgt, i32, i64, i64)
+DEF_HELPER_2(efdtsteq, i32, i64, i64)
+DEF_HELPER_2(efdcmplt, i32, i64, i64)
+DEF_HELPER_2(efdcmpgt, i32, i64, i64)
+DEF_HELPER_2(efdcmpeq, i32, i64, i64)
+
 #include "def-helper.h"
diff --git a/target-ppc/op.c b/target-ppc/op.c
index 8b3dbca..ee2f36e 100644
--- a/target-ppc/op.c
+++ b/target-ppc/op.c
@@ -960,468 +960,4 @@ void OPPROTO op_srli32_T1_64 (void)
     RETURN();
 }
 
-void OPPROTO op_evfssub (void)
-{
-    do_evfssub();
-    RETURN();
-}
-
-void OPPROTO op_evfsadd (void)
-{
-    do_evfsadd();
-    RETURN();
-}
-
-void OPPROTO op_evfsnabs (void)
-{
-    do_evfsnabs();
-    RETURN();
-}
-
-void OPPROTO op_evfsabs (void)
-{
-    do_evfsabs();
-    RETURN();
-}
-
-void OPPROTO op_evfsneg (void)
-{
-    do_evfsneg();
-    RETURN();
-}
-
-void OPPROTO op_evfsdiv (void)
-{
-    do_evfsdiv();
-    RETURN();
-}
-
-void OPPROTO op_evfsmul (void)
-{
-    do_evfsmul();
-    RETURN();
-}
-
-void OPPROTO op_evfscmplt (void)
-{
-    do_evfscmplt();
-    RETURN();
-}
-
-void OPPROTO op_evfscmpgt (void)
-{
-    do_evfscmpgt();
-    RETURN();
-}
-
-void OPPROTO op_evfscmpeq (void)
-{
-    do_evfscmpeq();
-    RETURN();
-}
-
-void OPPROTO op_evfscfsi (void)
-{
-    do_evfscfsi();
-    RETURN();
-}
-
-void OPPROTO op_evfscfui (void)
-{
-    do_evfscfui();
-    RETURN();
-}
-
-void OPPROTO op_evfscfsf (void)
-{
-    do_evfscfsf();
-    RETURN();
-}
-
-void OPPROTO op_evfscfuf (void)
-{
-    do_evfscfuf();
-    RETURN();
-}
-
-void OPPROTO op_evfsctsi (void)
-{
-    do_evfsctsi();
-    RETURN();
-}
-
-void OPPROTO op_evfsctui (void)
-{
-    do_evfsctui();
-    RETURN();
-}
-
-void OPPROTO op_evfsctsf (void)
-{
-    do_evfsctsf();
-    RETURN();
-}
-
-void OPPROTO op_evfsctuf (void)
-{
-    do_evfsctuf();
-    RETURN();
-}
-
-void OPPROTO op_evfsctuiz (void)
-{
-    do_evfsctuiz();
-    RETURN();
-}
-
-void OPPROTO op_evfsctsiz (void)
-{
-    do_evfsctsiz();
-    RETURN();
-}
-
-void OPPROTO op_evfststlt (void)
-{
-    do_evfststlt();
-    RETURN();
-}
-
-void OPPROTO op_evfststgt (void)
-{
-    do_evfststgt();
-    RETURN();
-}
-
-void OPPROTO op_evfststeq (void)
-{
-    do_evfststeq();
-    RETURN();
-}
-
-void OPPROTO op_efssub (void)
-{
-    T0_64 = _do_efssub(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efsadd (void)
-{
-    T0_64 = _do_efsadd(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efsnabs (void)
-{
-    T0_64 = _do_efsnabs(T0_64);
-    RETURN();
-}
-
-void OPPROTO op_efsabs (void)
-{
-    T0_64 = _do_efsabs(T0_64);
-    RETURN();
-}
-
-void OPPROTO op_efsneg (void)
-{
-    T0_64 = _do_efsneg(T0_64);
-    RETURN();
-}
-
-void OPPROTO op_efsdiv (void)
-{
-    T0_64 = _do_efsdiv(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efsmul (void)
-{
-    T0_64 = _do_efsmul(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efscmplt (void)
-{
-    do_efscmplt();
-    RETURN();
-}
-
-void OPPROTO op_efscmpgt (void)
-{
-    do_efscmpgt();
-    RETURN();
-}
-
-void OPPROTO op_efscfd (void)
-{
-    do_efscfd();
-    RETURN();
-}
-
-void OPPROTO op_efscmpeq (void)
-{
-    do_efscmpeq();
-    RETURN();
-}
-
-void OPPROTO op_efscfsi (void)
-{
-    do_efscfsi();
-    RETURN();
-}
-
-void OPPROTO op_efscfui (void)
-{
-    do_efscfui();
-    RETURN();
-}
-
-void OPPROTO op_efscfsf (void)
-{
-    do_efscfsf();
-    RETURN();
-}
-
-void OPPROTO op_efscfuf (void)
-{
-    do_efscfuf();
-    RETURN();
-}
-
-void OPPROTO op_efsctsi (void)
-{
-    do_efsctsi();
-    RETURN();
-}
-
-void OPPROTO op_efsctui (void)
-{
-    do_efsctui();
-    RETURN();
-}
-
-void OPPROTO op_efsctsf (void)
-{
-    do_efsctsf();
-    RETURN();
-}
-
-void OPPROTO op_efsctuf (void)
-{
-    do_efsctuf();
-    RETURN();
-}
-
-void OPPROTO op_efsctsiz (void)
-{
-    do_efsctsiz();
-    RETURN();
-}
-
-void OPPROTO op_efsctuiz (void)
-{
-    do_efsctuiz();
-    RETURN();
-}
-
-void OPPROTO op_efststlt (void)
-{
-    T0 = _do_efststlt(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efststgt (void)
-{
-    T0 = _do_efststgt(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efststeq (void)
-{
-    T0 = _do_efststeq(T0_64, T1_64);
-    RETURN();
-}
 
-void OPPROTO op_efdsub (void)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = T0_64;
-    u2.ll = T1_64;
-    u1.d = float64_sub(u1.d, u2.d, &env->spe_status);
-    T0_64 = u1.ll;
-    RETURN();
-}
-
-void OPPROTO op_efdadd (void)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = T0_64;
-    u2.ll = T1_64;
-    u1.d = float64_add(u1.d, u2.d, &env->spe_status);
-    T0_64 = u1.ll;
-    RETURN();
-}
-
-void OPPROTO op_efdcfsid (void)
-{
-    do_efdcfsi();
-    RETURN();
-}
-
-void OPPROTO op_efdcfuid (void)
-{
-    do_efdcfui();
-    RETURN();
-}
-
-void OPPROTO op_efdnabs (void)
-{
-    T0_64 |= 0x8000000000000000ULL;
-    RETURN();
-}
-
-void OPPROTO op_efdabs (void)
-{
-    T0_64 &= ~0x8000000000000000ULL;
-    RETURN();
-}
-
-void OPPROTO op_efdneg (void)
-{
-    T0_64 ^= 0x8000000000000000ULL;
-    RETURN();
-}
-
-void OPPROTO op_efddiv (void)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = T0_64;
-    u2.ll = T1_64;
-    u1.d = float64_div(u1.d, u2.d, &env->spe_status);
-    T0_64 = u1.ll;
-    RETURN();
-}
-
-void OPPROTO op_efdmul (void)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = T0_64;
-    u2.ll = T1_64;
-    u1.d = float64_mul(u1.d, u2.d, &env->spe_status);
-    T0_64 = u1.ll;
-    RETURN();
-}
-
-void OPPROTO op_efdctsidz (void)
-{
-    do_efdctsiz();
-    RETURN();
-}
-
-void OPPROTO op_efdctuidz (void)
-{
-    do_efdctuiz();
-    RETURN();
-}
-
-void OPPROTO op_efdcmplt (void)
-{
-    do_efdcmplt();
-    RETURN();
-}
-
-void OPPROTO op_efdcmpgt (void)
-{
-    do_efdcmpgt();
-    RETURN();
-}
-
-void OPPROTO op_efdcfs (void)
-{
-    do_efdcfs();
-    RETURN();
-}
-
-void OPPROTO op_efdcmpeq (void)
-{
-    do_efdcmpeq();
-    RETURN();
-}
-
-void OPPROTO op_efdcfsi (void)
-{
-    do_efdcfsi();
-    RETURN();
-}
-
-void OPPROTO op_efdcfui (void)
-{
-    do_efdcfui();
-    RETURN();
-}
-
-void OPPROTO op_efdcfsf (void)
-{
-    do_efdcfsf();
-    RETURN();
-}
-
-void OPPROTO op_efdcfuf (void)
-{
-    do_efdcfuf();
-    RETURN();
-}
-
-void OPPROTO op_efdctsi (void)
-{
-    do_efdctsi();
-    RETURN();
-}
-
-void OPPROTO op_efdctui (void)
-{
-    do_efdctui();
-    RETURN();
-}
-
-void OPPROTO op_efdctsf (void)
-{
-    do_efdctsf();
-    RETURN();
-}
-
-void OPPROTO op_efdctuf (void)
-{
-    do_efdctuf();
-    RETURN();
-}
-
-void OPPROTO op_efdctuiz (void)
-{
-    do_efdctuiz();
-    RETURN();
-}
-
-void OPPROTO op_efdctsiz (void)
-{
-    do_efdctsiz();
-    RETURN();
-}
-
-void OPPROTO op_efdtstlt (void)
-{
-    T0 = _do_efdtstlt(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efdtstgt (void)
-{
-    T0 = _do_efdtstgt(T0_64, T1_64);
-    RETURN();
-}
-
-void OPPROTO op_efdtsteq (void)
-{
-    T0 = _do_efdtsteq(T0_64, T1_64);
-    RETURN();
-}
diff --git a/target-ppc/op_helper.c b/target-ppc/op_helper.c
index 4f7521b..6ea3ba3 100644
--- a/target-ppc/op_helper.c
+++ b/target-ppc/op_helper.c
@@ -1757,6 +1757,7 @@ void do_440_dlmzb (void)
     T0 = i;
 }
 
+/*****************************************************************************/
 /* SPE extension helpers */
 /* Use a table to make this quicker */
 static uint8_t hbrev[16] = {
@@ -1800,36 +1801,8 @@ uint32_t helper_cntlzw32 (uint32_t val)
     return clz32(val);
 }
 
-#define DO_SPE_OP1(name)                                                      \
-void do_ev##name (void)                                                       \
-{                                                                             \
-    T0_64 = ((uint64_t)_do_e##name(T0_64 >> 32) << 32) |                      \
-        (uint64_t)_do_e##name(T0_64);                                         \
-}
-
-#define DO_SPE_OP2(name)                                                      \
-void do_ev##name (void)                                                       \
-{                                                                             \
-    T0_64 = ((uint64_t)_do_e##name(T0_64 >> 32, T1_64 >> 32) << 32) |         \
-        (uint64_t)_do_e##name(T0_64, T1_64);                                  \
-}
-
-/* Fixed-point vector comparisons */
-#define DO_SPE_CMP(name)                                                      \
-void do_ev##name (void)                                                       \
-{                                                                             \
-    T0 = _do_evcmp_merge((uint64_t)_do_e##name(T0_64 >> 32,                   \
-                                               T1_64 >> 32) << 32,            \
-                         _do_e##name(T0_64, T1_64));                          \
-}
-
-static always_inline uint32_t _do_evcmp_merge (int t0, int t1)
-{
-    return (t0 << 3) | (t1 << 2) | ((t0 | t1) << 1) | (t0 & t1);
-}
-
-/* Single precision floating-point conversions from/to integer */
-static always_inline uint32_t _do_efscfsi (int32_t val)
+/* Single-precision floating-point conversions */
+static always_inline uint32_t efscfsi (uint32_t val)
 {
     CPU_FloatU u;
 
@@ -1838,7 +1811,7 @@ static always_inline uint32_t _do_efscfsi (int32_t val)
     return u.l;
 }
 
-static always_inline uint32_t _do_efscfui (uint32_t val)
+static always_inline uint32_t efscfui (uint32_t val)
 {
     CPU_FloatU u;
 
@@ -1847,7 +1820,7 @@ static always_inline uint32_t _do_efscfui (uint32_t val)
     return u.l;
 }
 
-static always_inline int32_t _do_efsctsi (uint32_t val)
+static always_inline int32_t efsctsi (uint32_t val)
 {
     CPU_FloatU u;
 
@@ -1859,7 +1832,7 @@ static always_inline int32_t _do_efsctsi (uint32_t val)
     return float32_to_int32(u.f, &env->spe_status);
 }
 
-static always_inline uint32_t _do_efsctui (uint32_t val)
+static always_inline uint32_t efsctui (uint32_t val)
 {
     CPU_FloatU u;
 
@@ -1871,7 +1844,7 @@ static always_inline uint32_t _do_efsctui (uint32_t val)
     return float32_to_uint32(u.f, &env->spe_status);
 }
 
-static always_inline int32_t _do_efsctsiz (uint32_t val)
+static always_inline uint32_t efsctsiz (uint32_t val)
 {
     CPU_FloatU u;
 
@@ -1883,7 +1856,7 @@ static always_inline int32_t _do_efsctsiz (uint32_t val)
     return float32_to_int32_round_to_zero(u.f, &env->spe_status);
 }
 
-static always_inline uint32_t _do_efsctuiz (uint32_t val)
+static always_inline uint32_t efsctuiz (uint32_t val)
 {
     CPU_FloatU u;
 
@@ -1895,38 +1868,7 @@ static always_inline uint32_t _do_efsctuiz (uint32_t val)
     return float32_to_uint32_round_to_zero(u.f, &env->spe_status);
 }
 
-void do_efscfsi (void)
-{
-    T0_64 = _do_efscfsi(T0_64);
-}
-
-void do_efscfui (void)
-{
-    T0_64 = _do_efscfui(T0_64);
-}
-
-void do_efsctsi (void)
-{
-    T0_64 = _do_efsctsi(T0_64);
-}
-
-void do_efsctui (void)
-{
-    T0_64 = _do_efsctui(T0_64);
-}
-
-void do_efsctsiz (void)
-{
-    T0_64 = _do_efsctsiz(T0_64);
-}
-
-void do_efsctuiz (void)
-{
-    T0_64 = _do_efsctuiz(T0_64);
-}
-
-/* Single precision floating-point conversion to/from fractional */
-static always_inline uint32_t _do_efscfsf (uint32_t val)
+static always_inline uint32_t efscfsf (uint32_t val)
 {
     CPU_FloatU u;
     float32 tmp;
@@ -1938,7 +1880,7 @@ static always_inline uint32_t _do_efscfsf (uint32_t val)
     return u.l;
 }
 
-static always_inline uint32_t _do_efscfuf (uint32_t val)
+static always_inline uint32_t efscfuf (uint32_t val)
 {
     CPU_FloatU u;
     float32 tmp;
@@ -1950,7 +1892,7 @@ static always_inline uint32_t _do_efscfuf (uint32_t val)
     return u.l;
 }
 
-static always_inline int32_t _do_efsctsf (uint32_t val)
+static always_inline uint32_t efsctsf (uint32_t val)
 {
     CPU_FloatU u;
     float32 tmp;
@@ -1965,7 +1907,7 @@ static always_inline int32_t _do_efsctsf (uint32_t val)
     return float32_to_int32(u.f, &env->spe_status);
 }
 
-static always_inline uint32_t _do_efsctuf (uint32_t val)
+static always_inline uint32_t efsctuf (uint32_t val)
 {
     CPU_FloatU u;
     float32 tmp;
@@ -1980,102 +1922,220 @@ static always_inline uint32_t _do_efsctuf (uint32_t val)
     return float32_to_uint32(u.f, &env->spe_status);
 }
 
-static always_inline int32_t _do_efsctsfz (uint32_t val)
-{
-    CPU_FloatU u;
-    float32 tmp;
-
-    u.l = val;
-    /* NaN are not treated the same way IEEE 754 does */
-    if (unlikely(isnan(u.f)))
-        return 0;
-    tmp = uint64_to_float32(1ULL << 32, &env->spe_status);
-    u.f = float32_mul(u.f, tmp, &env->spe_status);
-
-    return float32_to_int32_round_to_zero(u.f, &env->spe_status);
+#define HELPER_SPE_SINGLE_CONV(name)                                          \
+uint32_t helper_e##name (uint32_t val)                                        \
+{                                                                             \
+    return e##name(val);                                                      \
+}
+/* efscfsi */
+HELPER_SPE_SINGLE_CONV(fscfsi);
+/* efscfui */
+HELPER_SPE_SINGLE_CONV(fscfui);
+/* efscfuf */
+HELPER_SPE_SINGLE_CONV(fscfuf);
+/* efscfsf */
+HELPER_SPE_SINGLE_CONV(fscfsf);
+/* efsctsi */
+HELPER_SPE_SINGLE_CONV(fsctsi);
+/* efsctui */
+HELPER_SPE_SINGLE_CONV(fsctui);
+/* efsctsiz */
+HELPER_SPE_SINGLE_CONV(fsctsiz);
+/* efsctuiz */
+HELPER_SPE_SINGLE_CONV(fsctuiz);
+/* efsctsf */
+HELPER_SPE_SINGLE_CONV(fsctsf);
+/* efsctuf */
+HELPER_SPE_SINGLE_CONV(fsctuf);
+
+#define HELPER_SPE_VECTOR_CONV(name)                                          \
+uint64_t helper_ev##name (uint64_t val)                                       \
+{                                                                             \
+    return ((uint64_t)e##name(val >> 32) << 32) |                             \
+            (uint64_t)e##name(val);                                           \
 }
+/* evfscfsi */
+HELPER_SPE_VECTOR_CONV(fscfsi);
+/* evfscfui */
+HELPER_SPE_VECTOR_CONV(fscfui);
+/* evfscfuf */
+HELPER_SPE_VECTOR_CONV(fscfuf);
+/* evfscfsf */
+HELPER_SPE_VECTOR_CONV(fscfsf);
+/* evfsctsi */
+HELPER_SPE_VECTOR_CONV(fsctsi);
+/* evfsctui */
+HELPER_SPE_VECTOR_CONV(fsctui);
+/* evfsctsiz */
+HELPER_SPE_VECTOR_CONV(fsctsiz);
+/* evfsctuiz */
+HELPER_SPE_VECTOR_CONV(fsctuiz);
+/* evfsctsf */
+HELPER_SPE_VECTOR_CONV(fsctsf);
+/* evfsctuf */
+HELPER_SPE_VECTOR_CONV(fsctuf);
 
-static always_inline uint32_t _do_efsctufz (uint32_t val)
+/* Single-precision floating-point arithmetic */
+static always_inline uint32_t efsadd (uint32_t op1, uint32_t op2)
 {
-    CPU_FloatU u;
-    float32 tmp;
-
-    u.l = val;
-    /* NaN are not treated the same way IEEE 754 does */
-    if (unlikely(isnan(u.f)))
-        return 0;
-    tmp = uint64_to_float32(1ULL << 32, &env->spe_status);
-    u.f = float32_mul(u.f, tmp, &env->spe_status);
-
-    return float32_to_uint32_round_to_zero(u.f, &env->spe_status);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    u1.f = float32_add(u1.f, u2.f, &env->spe_status);
+    return u1.l;
 }
 
-void do_efscfsf (void)
+static always_inline uint32_t efssub (uint32_t op1, uint32_t op2)
 {
-    T0_64 = _do_efscfsf(T0_64);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    u1.f = float32_sub(u1.f, u2.f, &env->spe_status);
+    return u1.l;
 }
 
-void do_efscfuf (void)
+static always_inline uint32_t efsmul (uint32_t op1, uint32_t op2)
 {
-    T0_64 = _do_efscfuf(T0_64);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    u1.f = float32_mul(u1.f, u2.f, &env->spe_status);
+    return u1.l;
 }
 
-void do_efsctsf (void)
+static always_inline uint32_t efsdiv (uint32_t op1, uint32_t op2)
 {
-    T0_64 = _do_efsctsf(T0_64);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    u1.f = float32_div(u1.f, u2.f, &env->spe_status);
+    return u1.l;
 }
 
-void do_efsctuf (void)
+#define HELPER_SPE_SINGLE_ARITH(name)                                         \
+uint32_t helper_e##name (uint32_t op1, uint32_t op2)                          \
+{                                                                             \
+    return e##name(op1, op2);                                                 \
+}
+/* efsadd */
+HELPER_SPE_SINGLE_ARITH(fsadd);
+/* efssub */
+HELPER_SPE_SINGLE_ARITH(fssub);
+/* efsmul */
+HELPER_SPE_SINGLE_ARITH(fsmul);
+/* efsdiv */
+HELPER_SPE_SINGLE_ARITH(fsdiv);
+
+#define HELPER_SPE_VECTOR_ARITH(name)                                         \
+uint64_t helper_ev##name (uint64_t op1, uint64_t op2)                         \
+{                                                                             \
+    return ((uint64_t)e##name(op1 >> 32, op2 >> 32) << 32) |                  \
+            (uint64_t)e##name(op1, op2);                                      \
+}
+/* evfsadd */
+HELPER_SPE_VECTOR_ARITH(fsadd);
+/* evfssub */
+HELPER_SPE_VECTOR_ARITH(fssub);
+/* evfsmul */
+HELPER_SPE_VECTOR_ARITH(fsmul);
+/* evfsdiv */
+HELPER_SPE_VECTOR_ARITH(fsdiv);
+
+/* Single-precision floating-point comparisons */
+static always_inline uint32_t efststlt (uint32_t op1, uint32_t op2)
 {
-    T0_64 = _do_efsctuf(T0_64);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    return float32_lt(u1.f, u2.f, &env->spe_status) ? 4 : 0;
 }
 
-void do_efsctsfz (void)
+static always_inline uint32_t efststgt (uint32_t op1, uint32_t op2)
 {
-    T0_64 = _do_efsctsfz(T0_64);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    return float32_le(u1.f, u2.f, &env->spe_status) ? 0 : 4;
 }
 
-void do_efsctufz (void)
+static always_inline uint32_t efststeq (uint32_t op1, uint32_t op2)
 {
-    T0_64 = _do_efsctufz(T0_64);
+    CPU_FloatU u1, u2;
+    u1.l = op1;
+    u2.l = op2;
+    return float32_eq(u1.f, u2.f, &env->spe_status) ? 4 : 0;
 }
 
-/* Double precision floating point helpers */
-static always_inline int _do_efdcmplt (uint64_t op1, uint64_t op2)
+static always_inline uint32_t efscmplt (uint32_t op1, uint32_t op2)
 {
     /* XXX: TODO: test special values (NaN, infinites, ...) */
-    return _do_efdtstlt(op1, op2);
+    return efststlt(op1, op2);
 }
 
-static always_inline int _do_efdcmpgt (uint64_t op1, uint64_t op2)
+static always_inline uint32_t efscmpgt (uint32_t op1, uint32_t op2)
 {
     /* XXX: TODO: test special values (NaN, infinites, ...) */
-    return _do_efdtstgt(op1, op2);
+    return efststgt(op1, op2);
 }
 
-static always_inline int _do_efdcmpeq (uint64_t op1, uint64_t op2)
+static always_inline uint32_t efscmpeq (uint32_t op1, uint32_t op2)
 {
     /* XXX: TODO: test special values (NaN, infinites, ...) */
-    return _do_efdtsteq(op1, op2);
+    return efststeq(op1, op2);
 }
 
-void do_efdcmplt (void)
+#define HELPER_SINGLE_SPE_CMP(name)                                           \
+uint32_t helper_e##name (uint32_t op1, uint32_t op2)                          \
+{                                                                             \
+    return e##name(op1, op2) << 2;                                            \
+}
+/* efststlt */
+HELPER_SINGLE_SPE_CMP(fststlt);
+/* efststgt */
+HELPER_SINGLE_SPE_CMP(fststgt);
+/* efststeq */
+HELPER_SINGLE_SPE_CMP(fststeq);
+/* efscmplt */
+HELPER_SINGLE_SPE_CMP(fscmplt);
+/* efscmpgt */
+HELPER_SINGLE_SPE_CMP(fscmpgt);
+/* efscmpeq */
+HELPER_SINGLE_SPE_CMP(fscmpeq);
+
+static always_inline uint32_t evcmp_merge (int t0, int t1)
 {
-    T0 = _do_efdcmplt(T0_64, T1_64);
+    return (t0 << 3) | (t1 << 2) | ((t0 | t1) << 1) | (t0 & t1);
 }
 
-void do_efdcmpgt (void)
-{
-    T0 = _do_efdcmpgt(T0_64, T1_64);
+#define HELPER_VECTOR_SPE_CMP(name)                                           \
+uint32_t helper_ev##name (uint64_t op1, uint64_t op2)                         \
+{                                                                             \
+    return evcmp_merge(e##name(op1 >> 32, op2 >> 32), e##name(op1, op2));     \
 }
+/* evfststlt */
+HELPER_VECTOR_SPE_CMP(fststlt);
+/* evfststgt */
+HELPER_VECTOR_SPE_CMP(fststgt);
+/* evfststeq */
+HELPER_VECTOR_SPE_CMP(fststeq);
+/* evfscmplt */
+HELPER_VECTOR_SPE_CMP(fscmplt);
+/* evfscmpgt */
+HELPER_VECTOR_SPE_CMP(fscmpgt);
+/* evfscmpeq */
+HELPER_VECTOR_SPE_CMP(fscmpeq);
 
-void do_efdcmpeq (void)
+/* Double-precision floating-point conversion */
+uint64_t helper_efdcfsi (uint32_t val)
 {
-    T0 = _do_efdcmpeq(T0_64, T1_64);
+    CPU_DoubleU u;
+
+    u.d = int32_to_float64(val, &env->spe_status);
+
+    return u.ll;
 }
 
-/* Double precision floating-point conversion to/from integer */
-static always_inline uint64_t _do_efdcfsi (int64_t val)
+uint64_t helper_efdcfsid (uint64_t val)
 {
     CPU_DoubleU u;
 
@@ -2084,7 +2144,16 @@ static always_inline uint64_t _do_efdcfsi (int64_t val)
     return u.ll;
 }
 
-static always_inline uint64_t _do_efdcfui (uint64_t val)
+uint64_t helper_efdcfui (uint32_t val)
+{
+    CPU_DoubleU u;
+
+    u.d = uint32_to_float64(val, &env->spe_status);
+
+    return u.ll;
+}
+
+uint64_t helper_efdcfuid (uint64_t val)
 {
     CPU_DoubleU u;
 
@@ -2093,7 +2162,7 @@ static always_inline uint64_t _do_efdcfui (uint64_t val)
     return u.ll;
 }
 
-static always_inline int64_t _do_efdctsi (uint64_t val)
+uint32_t helper_efdctsi (uint64_t val)
 {
     CPU_DoubleU u;
 
@@ -2102,10 +2171,10 @@ static always_inline int64_t _do_efdctsi (uint64_t val)
     if (unlikely(isnan(u.d)))
         return 0;
 
-    return float64_to_int64(u.d, &env->spe_status);
+    return float64_to_int32(u.d, &env->spe_status);
 }
 
-static always_inline uint64_t _do_efdctui (uint64_t val)
+uint32_t helper_efdctui (uint64_t val)
 {
     CPU_DoubleU u;
 
@@ -2114,10 +2183,10 @@ static always_inline uint64_t _do_efdctui (uint64_t val)
     if (unlikely(isnan(u.d)))
         return 0;
 
-    return float64_to_uint64(u.d, &env->spe_status);
+    return float64_to_uint32(u.d, &env->spe_status);
 }
 
-static always_inline int64_t _do_efdctsiz (uint64_t val)
+uint32_t helper_efdctsiz (uint64_t val)
 {
     CPU_DoubleU u;
 
@@ -2126,10 +2195,10 @@ static always_inline int64_t _do_efdctsiz (uint64_t val)
     if (unlikely(isnan(u.d)))
         return 0;
 
-    return float64_to_int64_round_to_zero(u.d, &env->spe_status);
+    return float64_to_int32_round_to_zero(u.d, &env->spe_status);
 }
 
-static always_inline uint64_t _do_efdctuiz (uint64_t val)
+uint64_t helper_efdctsidz (uint64_t val)
 {
     CPU_DoubleU u;
 
@@ -2138,41 +2207,34 @@ static always_inline uint64_t _do_efdctuiz (uint64_t val)
     if (unlikely(isnan(u.d)))
         return 0;
 
-    return float64_to_uint64_round_to_zero(u.d, &env->spe_status);
+    return float64_to_int64_round_to_zero(u.d, &env->spe_status);
 }
 
-void do_efdcfsi (void)
+uint32_t helper_efdctuiz (uint64_t val)
 {
-    T0_64 = _do_efdcfsi(T0_64);
-}
+    CPU_DoubleU u;
 
-void do_efdcfui (void)
-{
-    T0_64 = _do_efdcfui(T0_64);
-}
+    u.ll = val;
+    /* NaN are not treated the same way IEEE 754 does */
+    if (unlikely(isnan(u.d)))
+        return 0;
 
-void do_efdctsi (void)
-{
-    T0_64 = _do_efdctsi(T0_64);
+    return float64_to_uint32_round_to_zero(u.d, &env->spe_status);
 }
 
-void do_efdctui (void)
+uint64_t helper_efdctuidz (uint64_t val)
 {
-    T0_64 = _do_efdctui(T0_64);
-}
+    CPU_DoubleU u;
 
-void do_efdctsiz (void)
-{
-    T0_64 = _do_efdctsiz(T0_64);
-}
+    u.ll = val;
+    /* NaN are not treated the same way IEEE 754 does */
+    if (unlikely(isnan(u.d)))
+        return 0;
 
-void do_efdctuiz (void)
-{
-    T0_64 = _do_efdctuiz(T0_64);
+    return float64_to_uint64_round_to_zero(u.d, &env->spe_status);
 }
 
-/* Double precision floating-point conversion to/from fractional */
-static always_inline uint64_t _do_efdcfsf (int64_t val)
+uint64_t helper_efdcfsf (uint32_t val)
 {
     CPU_DoubleU u;
     float64 tmp;
@@ -2184,7 +2246,7 @@ static always_inline uint64_t _do_efdcfsf (int64_t val)
     return u.ll;
 }
 
-static always_inline uint64_t _do_efdcfuf (uint64_t val)
+uint64_t helper_efdcfuf (uint32_t val)
 {
     CPU_DoubleU u;
     float64 tmp;
@@ -2196,7 +2258,7 @@ static always_inline uint64_t _do_efdcfuf (uint64_t val)
     return u.ll;
 }
 
-static always_inline int64_t _do_efdctsf (uint64_t val)
+uint32_t helper_efdctsf (uint64_t val)
 {
     CPU_DoubleU u;
     float64 tmp;
@@ -2211,7 +2273,7 @@ static always_inline int64_t _do_efdctsf (uint64_t val)
     return float64_to_int32(u.d, &env->spe_status);
 }
 
-static always_inline uint64_t _do_efdctuf (uint64_t val)
+uint32_t helper_efdctuf (uint64_t val)
 {
     CPU_DoubleU u;
     float64 tmp;
@@ -2226,68 +2288,7 @@ static always_inline uint64_t _do_efdctuf (uint64_t val)
     return float64_to_uint32(u.d, &env->spe_status);
 }
 
-static always_inline int64_t _do_efdctsfz (uint64_t val)
-{
-    CPU_DoubleU u;
-    float64 tmp;
-
-    u.ll = val;
-    /* NaN are not treated the same way IEEE 754 does */
-    if (unlikely(isnan(u.d)))
-        return 0;
-    tmp = uint64_to_float64(1ULL << 32, &env->spe_status);
-    u.d = float64_mul(u.d, tmp, &env->spe_status);
-
-    return float64_to_int32_round_to_zero(u.d, &env->spe_status);
-}
-
-static always_inline uint64_t _do_efdctufz (uint64_t val)
-{
-    CPU_DoubleU u;
-    float64 tmp;
-
-    u.ll = val;
-    /* NaN are not treated the same way IEEE 754 does */
-    if (unlikely(isnan(u.d)))
-        return 0;
-    tmp = uint64_to_float64(1ULL << 32, &env->spe_status);
-    u.d = float64_mul(u.d, tmp, &env->spe_status);
-
-    return float64_to_uint32_round_to_zero(u.d, &env->spe_status);
-}
-
-void do_efdcfsf (void)
-{
-    T0_64 = _do_efdcfsf(T0_64);
-}
-
-void do_efdcfuf (void)
-{
-    T0_64 = _do_efdcfuf(T0_64);
-}
-
-void do_efdctsf (void)
-{
-    T0_64 = _do_efdctsf(T0_64);
-}
-
-void do_efdctuf (void)
-{
-    T0_64 = _do_efdctuf(T0_64);
-}
-
-void do_efdctsfz (void)
-{
-    T0_64 = _do_efdctsfz(T0_64);
-}
-
-void do_efdctufz (void)
-{
-    T0_64 = _do_efdctufz(T0_64);
-}
-
-/* Floating point conversion between single and double precision */
-static always_inline uint32_t _do_efscfd (uint64_t val)
+uint32_t helper_efscfd (uint64_t val)
 {
     CPU_DoubleU u1;
     CPU_FloatU u2;
@@ -2298,7 +2299,7 @@ static always_inline uint32_t _do_efscfd (uint64_t val)
     return u2.l;
 }
 
-static always_inline uint64_t _do_efdcfs (uint32_t val)
+uint64_t helper_efdcfs (uint32_t val)
 {
     CPU_DoubleU u2;
     CPU_FloatU u1;
@@ -2309,101 +2310,85 @@ static always_inline uint64_t _do_efdcfs (uint32_t val)
     return u2.ll;
 }
 
-void do_efscfd (void)
+/* Double precision fixed-point arithmetic */
+uint64_t helper_efdadd (uint64_t op1, uint64_t op2)
 {
-    T0_64 = _do_efscfd(T0_64);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    u1.d = float64_add(u1.d, u2.d, &env->spe_status);
+    return u1.ll;
 }
 
-void do_efdcfs (void)
+uint64_t helper_efdsub (uint64_t op1, uint64_t op2)
 {
-    T0_64 = _do_efdcfs(T0_64);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    u1.d = float64_sub(u1.d, u2.d, &env->spe_status);
+    return u1.ll;
 }
 
-/* Single precision fixed-point vector arithmetic */
-/* evfsabs */
-DO_SPE_OP1(fsabs);
-/* evfsnabs */
-DO_SPE_OP1(fsnabs);
-/* evfsneg */
-DO_SPE_OP1(fsneg);
-/* evfsadd */
-DO_SPE_OP2(fsadd);
-/* evfssub */
-DO_SPE_OP2(fssub);
-/* evfsmul */
-DO_SPE_OP2(fsmul);
-/* evfsdiv */
-DO_SPE_OP2(fsdiv);
-
-/* Single-precision floating-point comparisons */
-static always_inline int _do_efscmplt (uint32_t op1, uint32_t op2)
+uint64_t helper_efdmul (uint64_t op1, uint64_t op2)
 {
-    /* XXX: TODO: test special values (NaN, infinites, ...) */
-    return _do_efststlt(op1, op2);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    u1.d = float64_mul(u1.d, u2.d, &env->spe_status);
+    return u1.ll;
 }
 
-static always_inline int _do_efscmpgt (uint32_t op1, uint32_t op2)
+uint64_t helper_efddiv (uint64_t op1, uint64_t op2)
 {
-    /* XXX: TODO: test special values (NaN, infinites, ...) */
-    return _do_efststgt(op1, op2);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    u1.d = float64_div(u1.d, u2.d, &env->spe_status);
+    return u1.ll;
 }
 
-static always_inline int _do_efscmpeq (uint32_t op1, uint32_t op2)
+/* Double precision floating point helpers */
+uint32_t helper_efdtstlt (uint64_t op1, uint64_t op2)
 {
-    /* XXX: TODO: test special values (NaN, infinites, ...) */
-    return _do_efststeq(op1, op2);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    return float64_lt(u1.d, u2.d, &env->spe_status) ? 4 : 0;
 }
 
-void do_efscmplt (void)
+uint32_t helper_efdtstgt (uint64_t op1, uint64_t op2)
 {
-    T0 = _do_efscmplt(T0_64, T1_64);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    return float64_le(u1.d, u2.d, &env->spe_status) ? 0 : 4;
 }
 
-void do_efscmpgt (void)
+uint32_t helper_efdtsteq (uint64_t op1, uint64_t op2)
 {
-    T0 = _do_efscmpgt(T0_64, T1_64);
+    CPU_DoubleU u1, u2;
+    u1.ll = op1;
+    u2.ll = op2;
+    return float64_eq(u1.d, u2.d, &env->spe_status) ? 4 : 0;
 }
 
-void do_efscmpeq (void)
+uint32_t helper_efdcmplt (uint64_t op1, uint64_t op2)
 {
-    T0 = _do_efscmpeq(T0_64, T1_64);
+    /* XXX: TODO: test special values (NaN, infinites, ...) */
+    return helper_efdtstlt(op1, op2);
 }
 
-/* Single-precision floating-point vector comparisons */
-/* evfscmplt */
-DO_SPE_CMP(fscmplt);
-/* evfscmpgt */
-DO_SPE_CMP(fscmpgt);
-/* evfscmpeq */
-DO_SPE_CMP(fscmpeq);
-/* evfststlt */
-DO_SPE_CMP(fststlt);
-/* evfststgt */
-DO_SPE_CMP(fststgt);
-/* evfststeq */
-DO_SPE_CMP(fststeq);
+uint32_t helper_efdcmpgt (uint64_t op1, uint64_t op2)
+{
+    /* XXX: TODO: test special values (NaN, infinites, ...) */
+    return helper_efdtstgt(op1, op2);
+}
 
-/* Single-precision floating-point vector conversions */
-/* evfscfsi */
-DO_SPE_OP1(fscfsi);
-/* evfscfui */
-DO_SPE_OP1(fscfui);
-/* evfscfuf */
-DO_SPE_OP1(fscfuf);
-/* evfscfsf */
-DO_SPE_OP1(fscfsf);
-/* evfsctsi */
-DO_SPE_OP1(fsctsi);
-/* evfsctui */
-DO_SPE_OP1(fsctui);
-/* evfsctsiz */
-DO_SPE_OP1(fsctsiz);
-/* evfsctuiz */
-DO_SPE_OP1(fsctuiz);
-/* evfsctsf */
-DO_SPE_OP1(fsctsf);
-/* evfsctuf */
-DO_SPE_OP1(fsctuf);
+uint32_t helper_efdcmpeq (uint64_t op1, uint64_t op2)
+{
+    /* XXX: TODO: test special values (NaN, infinites, ...) */
+    return helper_efdtsteq(op1, op2);
+}
 
 /*****************************************************************************/
 /* Softmmu support */
diff --git a/target-ppc/op_helper.h b/target-ppc/op_helper.h
index 39b4cac..30f9ca1 100644
--- a/target-ppc/op_helper.h
+++ b/target-ppc/op_helper.h
@@ -125,155 +125,4 @@ void do_load_403_pb (int num);
 void do_store_403_pb (int num);
 #endif
 
-/* SPE extension helpers */
-/* Single precision floating-point helpers */
-void do_efscmplt (void);
-void do_efscmpgt (void);
-void do_efscmpeq (void);
-void do_efscfsf (void);
-void do_efscfuf (void);
-void do_efsctsf (void);
-void do_efsctuf (void);
-
-void do_efscfsi (void);
-void do_efscfui (void);
-void do_efsctsi (void);
-void do_efsctui (void);
-void do_efsctsiz (void);
-void do_efsctuiz (void);
-
-/* Double precision floating-point helpers */
-void do_efdcmplt (void);
-void do_efdcmpgt (void);
-void do_efdcmpeq (void);
-void do_efdcfsf (void);
-void do_efdcfuf (void);
-void do_efdctsf (void);
-void do_efdctuf (void);
-
-void do_efdcfsi (void);
-void do_efdcfui (void);
-void do_efdctsi (void);
-void do_efdctui (void);
-void do_efdctsiz (void);
-void do_efdctuiz (void);
-
-void do_efdcfs (void);
-void do_efscfd (void);
-
-/* Floating-point vector helpers */
-void do_evfsabs (void);
-void do_evfsnabs (void);
-void do_evfsneg (void);
-void do_evfsadd (void);
-void do_evfssub (void);
-void do_evfsmul (void);
-void do_evfsdiv (void);
-void do_evfscmplt (void);
-void do_evfscmpgt (void);
-void do_evfscmpeq (void);
-void do_evfststlt (void);
-void do_evfststgt (void);
-void do_evfststeq (void);
-void do_evfscfsi (void);
-void do_evfscfui (void);
-void do_evfscfsf (void);
-void do_evfscfuf (void);
-void do_evfsctsf (void);
-void do_evfsctuf (void);
-void do_evfsctsi (void);
-void do_evfsctui (void);
-void do_evfsctsiz (void);
-void do_evfsctuiz (void);
-
-/* SPE extension */
-/* Single precision floating-point helpers */
-static always_inline uint32_t _do_efsabs (uint32_t val)
-{
-    return val & ~0x80000000;
-}
-static always_inline uint32_t _do_efsnabs (uint32_t val)
-{
-    return val | 0x80000000;
-}
-static always_inline uint32_t _do_efsneg (uint32_t val)
-{
-    return val ^ 0x80000000;
-}
-static always_inline uint32_t _do_efsadd (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    u1.f = float32_add(u1.f, u2.f, &env->spe_status);
-    return u1.l;
-}
-static always_inline uint32_t _do_efssub (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    u1.f = float32_sub(u1.f, u2.f, &env->spe_status);
-    return u1.l;
-}
-static always_inline uint32_t _do_efsmul (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    u1.f = float32_mul(u1.f, u2.f, &env->spe_status);
-    return u1.l;
-}
-static always_inline uint32_t _do_efsdiv (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    u1.f = float32_div(u1.f, u2.f, &env->spe_status);
-    return u1.l;
-}
-
-static always_inline int _do_efststlt (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    return float32_lt(u1.f, u2.f, &env->spe_status) ? 4 : 0;
-}
-static always_inline int _do_efststgt (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    return float32_le(u1.f, u2.f, &env->spe_status) ? 0 : 4;
-}
-static always_inline int _do_efststeq (uint32_t op1, uint32_t op2)
-{
-    CPU_FloatU u1, u2;
-    u1.l = op1;
-    u2.l = op2;
-    return float32_eq(u1.f, u2.f, &env->spe_status) ? 4 : 0;
-}
-/* Double precision floating-point helpers */
-static always_inline int _do_efdtstlt (uint64_t op1, uint64_t op2)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = op1;
-    u2.ll = op2;
-    return float64_lt(u1.d, u2.d, &env->spe_status) ? 4 : 0;
-}
-static always_inline int _do_efdtstgt (uint64_t op1, uint64_t op2)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = op1;
-    u2.ll = op2;
-    return float64_le(u1.d, u2.d, &env->spe_status) ? 0 : 4;
-}
-static always_inline int _do_efdtsteq (uint64_t op1, uint64_t op2)
-{
-    CPU_DoubleU u1, u2;
-    u1.ll = op1;
-    u2.ll = op2;
-    return float64_eq(u1.d, u2.d, &env->spe_status) ? 4 : 0;
-}
 #endif
diff --git a/target-ppc/translate.c b/target-ppc/translate.c
index e9ed627..aaec6d8 100644
--- a/target-ppc/translate.c
+++ b/target-ppc/translate.c
@@ -6861,81 +6861,261 @@ GEN_SPE(speundef,       evmwsmfan,     0x0D, 0x17, 0x00000000, PPC_SPE);
 #endif
 
 /***                      SPE floating-point extension                     ***/
-#define GEN_SPEFPUOP_CONV(name)                                               \
+#if defined(TARGET_PPC64)
+#define GEN_SPEFPUOP_CONV_32_32(name)                                         \
 static always_inline void gen_##name (DisasContext *ctx)                      \
 {                                                                             \
-    gen_load_gpr64(cpu_T64[0], rB(ctx->opcode));                              \
-    gen_op_##name();                                                          \
-    gen_store_gpr64(rD(ctx->opcode), cpu_T64[0]);                             \
+    TCGv_i32 t0;                                                              \
+    TCGv t1;                                                                  \
+    t0 = tcg_temp_new_i32();                                                  \
+    tcg_gen_trunc_tl_i32(t0, cpu_gpr[rB(ctx->opcode)]);                       \
+    gen_helper_##name(t0, t0);                                                \
+    t1 = tcg_temp_new();                                                      \
+    tcg_gen_extu_i32_tl(t1, t0);                                              \
+    tcg_temp_free_i32(t0);                                                    \
+    tcg_gen_andi_tl(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rD(ctx->opcode)],       \
+                    0xFFFFFFFF00000000ULL);                                   \
+    tcg_gen_or_tl(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rD(ctx->opcode)], t1);    \
+    tcg_temp_free(t1);                                                        \
 }
-
-#define GEN_SPEFPUOP_ARITH1(name)                                             \
+#define GEN_SPEFPUOP_CONV_32_64(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i32 t0;                                                              \
+    TCGv t1;                                                                  \
+    t0 = tcg_temp_new_i32();                                                  \
+    gen_helper_##name(t0, cpu_gpr[rB(ctx->opcode)]);                          \
+    t1 = tcg_temp_new();                                                      \
+    tcg_gen_extu_i32_tl(t1, t0);                                              \
+    tcg_temp_free_i32(t0);                                                    \
+    tcg_gen_andi_tl(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rD(ctx->opcode)],       \
+                    0xFFFFFFFF00000000ULL);                                   \
+    tcg_gen_or_tl(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rD(ctx->opcode)], t1);    \
+    tcg_temp_free(t1);                                                        \
+}
+#define GEN_SPEFPUOP_CONV_64_32(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i32 t0 = tcg_temp_new_i32();                                         \
+    tcg_gen_trunc_tl_i32(t0, cpu_gpr[rB(ctx->opcode)]);                       \
+    gen_helper_##name(cpu_gpr[rD(ctx->opcode)], t0);                          \
+    tcg_temp_free_i32(t0);                                                    \
+}
+#define GEN_SPEFPUOP_CONV_64_64(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    gen_helper_##name(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rB(ctx->opcode)]);    \
+}
+#define GEN_SPEFPUOP_ARITH2_32_32(name)                                       \
 static always_inline void gen_##name (DisasContext *ctx)                      \
 {                                                                             \
+    TCGv_i32 t0, t1;                                                          \
+    TCGv_i64 t2;                                                              \
     if (unlikely(!ctx->spe_enabled)) {                                        \
         GEN_EXCP_NO_AP(ctx);                                                  \
         return;                                                               \
     }                                                                         \
-    gen_load_gpr64(cpu_T64[0], rA(ctx->opcode));                              \
-    gen_op_##name();                                                          \
-    gen_store_gpr64(rD(ctx->opcode), cpu_T64[0]);                             \
+    t0 = tcg_temp_new_i32();                                                  \
+    t1 = tcg_temp_new_i32();                                                  \
+    tcg_gen_trunc_tl_i32(t0, cpu_gpr[rA(ctx->opcode)]);                       \
+    tcg_gen_trunc_tl_i32(t1, cpu_gpr[rB(ctx->opcode)]);                       \
+    gen_helper_##name(t0, t0, t1);                                            \
+    tcg_temp_free_i32(t1);                                                    \
+    t2 = tcg_temp_new();                                                      \
+    tcg_gen_extu_i32_tl(t2, t0);                                              \
+    tcg_temp_free_i32(t0);                                                    \
+    tcg_gen_andi_tl(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rD(ctx->opcode)],       \
+                    0xFFFFFFFF00000000ULL);                                   \
+    tcg_gen_or_tl(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rD(ctx->opcode)], t2);    \
+    tcg_temp_free(t2);                                                        \
 }
-
-#define GEN_SPEFPUOP_ARITH2(name)                                             \
+#define GEN_SPEFPUOP_ARITH2_64_64(name)                                       \
 static always_inline void gen_##name (DisasContext *ctx)                      \
 {                                                                             \
     if (unlikely(!ctx->spe_enabled)) {                                        \
         GEN_EXCP_NO_AP(ctx);                                                  \
         return;                                                               \
     }                                                                         \
-    gen_load_gpr64(cpu_T64[0], rA(ctx->opcode));                              \
-    gen_load_gpr64(cpu_T64[1], rB(ctx->opcode));                              \
-    gen_op_##name();                                                          \
-    gen_store_gpr64(rD(ctx->opcode), cpu_T64[0]);                             \
+    gen_helper_##name(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rA(ctx->opcode)],     \
+                      cpu_gpr[rB(ctx->opcode)]);                              \
 }
-
-#define GEN_SPEFPUOP_COMP(name)                                               \
+#define GEN_SPEFPUOP_COMP_32(name)                                            \
 static always_inline void gen_##name (DisasContext *ctx)                      \
 {                                                                             \
-    TCGv_i32 crf = cpu_crf[crfD(ctx->opcode)];                                \
+    TCGv_i32 t0, t1;                                                          \
     if (unlikely(!ctx->spe_enabled)) {                                        \
         GEN_EXCP_NO_AP(ctx);                                                  \
         return;                                                               \
     }                                                                         \
-    gen_load_gpr64(cpu_T64[0], rA(ctx->opcode));                              \
-    gen_load_gpr64(cpu_T64[1], rB(ctx->opcode));                              \
-    gen_op_##name();                                                          \
-    tcg_gen_trunc_tl_i32(crf, cpu_T[0]);                                      \
-    tcg_gen_andi_i32(crf, crf, 0xf);                                          \
+    t0 = tcg_temp_new_i32();                                                  \
+    t1 = tcg_temp_new_i32();                                                  \
+    tcg_gen_trunc_tl_i32(t0, cpu_gpr[rA(ctx->opcode)]);                       \
+    tcg_gen_trunc_tl_i32(t1, cpu_gpr[rB(ctx->opcode)]);                       \
+    gen_helper_##name(cpu_crf[crfD(ctx->opcode)], t0, t1);                    \
+    tcg_temp_free_i32(t0);                                                    \
+    tcg_temp_free_i32(t1);                                                    \
+}
+#define GEN_SPEFPUOP_COMP_64(name)                                            \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    if (unlikely(!ctx->spe_enabled)) {                                        \
+        GEN_EXCP_NO_AP(ctx);                                                  \
+        return;                                                               \
+    }                                                                         \
+    gen_helper_##name(cpu_crf[crfD(ctx->opcode)],                             \
+                      cpu_gpr[rA(ctx->opcode)], cpu_gpr[rB(ctx->opcode)]);    \
+}
+#else
+#define GEN_SPEFPUOP_CONV_32_32(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    gen_helper_##name(cpu_gpr[rD(ctx->opcode)], cpu_gpr[rB(ctx->opcode)]);    \
 }
+#define GEN_SPEFPUOP_CONV_32_64(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i64 t0 = tcg_temp_new_i64();                                         \
+    gen_load_gpr64(t0, rB(ctx->opcode));                                      \
+    gen_helper_##name(cpu_gpr[rD(ctx->opcode)], t0);                          \
+    tcg_temp_free_i64(t0);                                                    \
+}
+#define GEN_SPEFPUOP_CONV_64_32(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i64 t0 = tcg_temp_new_i64();                                         \
+    gen_helper_##name(t0, cpu_gpr[rB(ctx->opcode)]);                          \
+    gen_store_gpr64(rD(ctx->opcode), t0);                                     \
+    tcg_temp_free_i64(t0);                                                    \
+}
+#define GEN_SPEFPUOP_CONV_64_64(name)                                         \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i64 t0 = tcg_temp_new_i64();                                         \
+    gen_load_gpr64(t0, rB(ctx->opcode));                                      \
+    gen_helper_##name(t0, t0);                                                \
+    gen_store_gpr64(rD(ctx->opcode), t0);                                     \
+    tcg_temp_free_i64(t0);                                                    \
+}
+#define GEN_SPEFPUOP_ARITH2_32_32(name)                                       \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    if (unlikely(!ctx->spe_enabled)) {                                        \
+        GEN_EXCP_NO_AP(ctx);                                                  \
+        return;                                                               \
+    }                                                                         \
+    gen_helper_##name(cpu_gpr[rD(ctx->opcode)],                               \
+                      cpu_gpr[rA(ctx->opcode)], cpu_gpr[rB(ctx->opcode)]);    \
+}
+#define GEN_SPEFPUOP_ARITH2_64_64(name)                                       \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i64 t0, t1;                                                          \
+    if (unlikely(!ctx->spe_enabled)) {                                        \
+        GEN_EXCP_NO_AP(ctx);                                                  \
+        return;                                                               \
+    }                                                                         \
+    t0 = tcg_temp_new_i64();                                                  \
+    t1 = tcg_temp_new_i64();                                                  \
+    gen_load_gpr64(t0, rA(ctx->opcode));                                      \
+    gen_load_gpr64(t1, rB(ctx->opcode));                                      \
+    gen_helper_##name(t0, t0, t1);                                            \
+    gen_store_gpr64(rD(ctx->opcode), t0);                                     \
+    tcg_temp_free_i64(t0);                                                    \
+    tcg_temp_free_i64(t1);                                                    \
+}
+#define GEN_SPEFPUOP_COMP_32(name)                                            \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    if (unlikely(!ctx->spe_enabled)) {                                        \
+        GEN_EXCP_NO_AP(ctx);                                                  \
+        return;                                                               \
+    }                                                                         \
+    gen_helper_##name(cpu_crf[crfD(ctx->opcode)],                             \
+                      cpu_gpr[rA(ctx->opcode)], cpu_gpr[rB(ctx->opcode)]);    \
+}
+#define GEN_SPEFPUOP_COMP_64(name)                                            \
+static always_inline void gen_##name (DisasContext *ctx)                      \
+{                                                                             \
+    TCGv_i64 t0, t1;                                                          \
+    if (unlikely(!ctx->spe_enabled)) {                                        \
+        GEN_EXCP_NO_AP(ctx);                                                  \
+        return;                                                               \
+    }                                                                         \
+    t0 = tcg_temp_new_i64();                                                  \
+    t1 = tcg_temp_new_i64();                                                  \
+    gen_load_gpr64(t0, rA(ctx->opcode));                                      \
+    gen_load_gpr64(t1, rB(ctx->opcode));                                      \
+    gen_helper_##name(cpu_crf[crfD(ctx->opcode)], t0, t1);                    \
+    tcg_temp_free_i64(t0);                                                    \
+    tcg_temp_free_i64(t1);                                                    \
+}
+#endif
 
 /* Single precision floating-point vectors operations */
 /* Arithmetic */
-GEN_SPEFPUOP_ARITH2(evfsadd);
-GEN_SPEFPUOP_ARITH2(evfssub);
-GEN_SPEFPUOP_ARITH2(evfsmul);
-GEN_SPEFPUOP_ARITH2(evfsdiv);
-GEN_SPEFPUOP_ARITH1(evfsabs);
-GEN_SPEFPUOP_ARITH1(evfsnabs);
-GEN_SPEFPUOP_ARITH1(evfsneg);
+GEN_SPEFPUOP_ARITH2_64_64(evfsadd);
+GEN_SPEFPUOP_ARITH2_64_64(evfssub);
+GEN_SPEFPUOP_ARITH2_64_64(evfsmul);
+GEN_SPEFPUOP_ARITH2_64_64(evfsdiv);
+static always_inline void gen_evfsabs (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+#if defined(TARGET_PPC64)
+    tcg_gen_andi_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], ~0x8000000080000000LL);
+#else
+    tcg_gen_andi_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], ~0x80000000);
+    tcg_gen_andi_tl(cpu_gprh[rA(ctx->opcode)], cpu_gprh[rA(ctx->opcode)], ~0x80000000);
+#endif
+}
+static always_inline void gen_evfsnabs (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+#if defined(TARGET_PPC64)
+    tcg_gen_ori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x8000000080000000LL);
+#else
+    tcg_gen_ori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x80000000);
+    tcg_gen_ori_tl(cpu_gprh[rA(ctx->opcode)], cpu_gprh[rA(ctx->opcode)], 0x80000000);
+#endif
+}
+static always_inline void gen_evfsneg (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+#if defined(TARGET_PPC64)
+    tcg_gen_xori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x8000000080000000LL);
+#else
+    tcg_gen_xori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x80000000);
+    tcg_gen_xori_tl(cpu_gprh[rA(ctx->opcode)], cpu_gprh[rA(ctx->opcode)], 0x80000000);
+#endif
+}
+
 /* Conversion */
-GEN_SPEFPUOP_CONV(evfscfui);
-GEN_SPEFPUOP_CONV(evfscfsi);
-GEN_SPEFPUOP_CONV(evfscfuf);
-GEN_SPEFPUOP_CONV(evfscfsf);
-GEN_SPEFPUOP_CONV(evfsctui);
-GEN_SPEFPUOP_CONV(evfsctsi);
-GEN_SPEFPUOP_CONV(evfsctuf);
-GEN_SPEFPUOP_CONV(evfsctsf);
-GEN_SPEFPUOP_CONV(evfsctuiz);
-GEN_SPEFPUOP_CONV(evfsctsiz);
+GEN_SPEFPUOP_CONV_64_64(evfscfui);
+GEN_SPEFPUOP_CONV_64_64(evfscfsi);
+GEN_SPEFPUOP_CONV_64_64(evfscfuf);
+GEN_SPEFPUOP_CONV_64_64(evfscfsf);
+GEN_SPEFPUOP_CONV_64_64(evfsctui);
+GEN_SPEFPUOP_CONV_64_64(evfsctsi);
+GEN_SPEFPUOP_CONV_64_64(evfsctuf);
+GEN_SPEFPUOP_CONV_64_64(evfsctsf);
+GEN_SPEFPUOP_CONV_64_64(evfsctuiz);
+GEN_SPEFPUOP_CONV_64_64(evfsctsiz);
+
 /* Comparison */
-GEN_SPEFPUOP_COMP(evfscmpgt);
-GEN_SPEFPUOP_COMP(evfscmplt);
-GEN_SPEFPUOP_COMP(evfscmpeq);
-GEN_SPEFPUOP_COMP(evfststgt);
-GEN_SPEFPUOP_COMP(evfststlt);
-GEN_SPEFPUOP_COMP(evfststeq);
+GEN_SPEFPUOP_COMP_64(evfscmpgt);
+GEN_SPEFPUOP_COMP_64(evfscmplt);
+GEN_SPEFPUOP_COMP_64(evfscmpeq);
+GEN_SPEFPUOP_COMP_64(evfststgt);
+GEN_SPEFPUOP_COMP_64(evfststlt);
+GEN_SPEFPUOP_COMP_64(evfststeq);
 
 /* Opcodes definitions */
 GEN_SPE(evfsadd,        evfssub,       0x00, 0x0A, 0x00000000, PPC_SPEFPU); //
@@ -6955,32 +7135,55 @@ GEN_SPE(evfststeq,      speundef,      0x0F, 0x0A, 0x00600000, PPC_SPEFPU); //
 
 /* Single precision floating-point operations */
 /* Arithmetic */
-GEN_SPEFPUOP_ARITH2(efsadd);
-GEN_SPEFPUOP_ARITH2(efssub);
-GEN_SPEFPUOP_ARITH2(efsmul);
-GEN_SPEFPUOP_ARITH2(efsdiv);
-GEN_SPEFPUOP_ARITH1(efsabs);
-GEN_SPEFPUOP_ARITH1(efsnabs);
-GEN_SPEFPUOP_ARITH1(efsneg);
+GEN_SPEFPUOP_ARITH2_32_32(efsadd);
+GEN_SPEFPUOP_ARITH2_32_32(efssub);
+GEN_SPEFPUOP_ARITH2_32_32(efsmul);
+GEN_SPEFPUOP_ARITH2_32_32(efsdiv);
+static always_inline void gen_efsabs (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+    tcg_gen_andi_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], (target_long)~0x80000000LL);
+}
+static always_inline void gen_efsnabs (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+    tcg_gen_ori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x80000000);
+}
+static always_inline void gen_efsneg (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+    tcg_gen_xori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x80000000);
+}
+
 /* Conversion */
-GEN_SPEFPUOP_CONV(efscfui);
-GEN_SPEFPUOP_CONV(efscfsi);
-GEN_SPEFPUOP_CONV(efscfuf);
-GEN_SPEFPUOP_CONV(efscfsf);
-GEN_SPEFPUOP_CONV(efsctui);
-GEN_SPEFPUOP_CONV(efsctsi);
-GEN_SPEFPUOP_CONV(efsctuf);
-GEN_SPEFPUOP_CONV(efsctsf);
-GEN_SPEFPUOP_CONV(efsctuiz);
-GEN_SPEFPUOP_CONV(efsctsiz);
-GEN_SPEFPUOP_CONV(efscfd);
+GEN_SPEFPUOP_CONV_32_32(efscfui);
+GEN_SPEFPUOP_CONV_32_32(efscfsi);
+GEN_SPEFPUOP_CONV_32_32(efscfuf);
+GEN_SPEFPUOP_CONV_32_32(efscfsf);
+GEN_SPEFPUOP_CONV_32_32(efsctui);
+GEN_SPEFPUOP_CONV_32_32(efsctsi);
+GEN_SPEFPUOP_CONV_32_32(efsctuf);
+GEN_SPEFPUOP_CONV_32_32(efsctsf);
+GEN_SPEFPUOP_CONV_32_32(efsctuiz);
+GEN_SPEFPUOP_CONV_32_32(efsctsiz);
+GEN_SPEFPUOP_CONV_32_64(efscfd);
+
 /* Comparison */
-GEN_SPEFPUOP_COMP(efscmpgt);
-GEN_SPEFPUOP_COMP(efscmplt);
-GEN_SPEFPUOP_COMP(efscmpeq);
-GEN_SPEFPUOP_COMP(efststgt);
-GEN_SPEFPUOP_COMP(efststlt);
-GEN_SPEFPUOP_COMP(efststeq);
+GEN_SPEFPUOP_COMP_32(efscmpgt);
+GEN_SPEFPUOP_COMP_32(efscmplt);
+GEN_SPEFPUOP_COMP_32(efscmpeq);
+GEN_SPEFPUOP_COMP_32(efststgt);
+GEN_SPEFPUOP_COMP_32(efststlt);
+GEN_SPEFPUOP_COMP_32(efststeq);
 
 /* Opcodes definitions */
 GEN_SPE(efsadd,         efssub,        0x00, 0x0B, 0x00000000, PPC_SPEFPU); //
@@ -7000,37 +7203,71 @@ GEN_SPE(efststeq,       speundef,      0x0F, 0x0B, 0x00600000, PPC_SPEFPU); //
 
 /* Double precision floating-point operations */
 /* Arithmetic */
-GEN_SPEFPUOP_ARITH2(efdadd);
-GEN_SPEFPUOP_ARITH2(efdsub);
-GEN_SPEFPUOP_ARITH2(efdmul);
-GEN_SPEFPUOP_ARITH2(efddiv);
-GEN_SPEFPUOP_ARITH1(efdabs);
-GEN_SPEFPUOP_ARITH1(efdnabs);
-GEN_SPEFPUOP_ARITH1(efdneg);
+GEN_SPEFPUOP_ARITH2_64_64(efdadd);
+GEN_SPEFPUOP_ARITH2_64_64(efdsub);
+GEN_SPEFPUOP_ARITH2_64_64(efdmul);
+GEN_SPEFPUOP_ARITH2_64_64(efddiv);
+static always_inline void gen_efdabs (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+#if defined(TARGET_PPC64)
+    tcg_gen_andi_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], ~0x8000000000000000LL);
+#else
+    tcg_gen_andi_tl(cpu_gprh[rA(ctx->opcode)], cpu_gprh[rA(ctx->opcode)], ~0x80000000);
+#endif
+}
+static always_inline void gen_efdnabs (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+#if defined(TARGET_PPC64)
+    tcg_gen_ori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x8000000000000000LL);
+#else
+    tcg_gen_ori_tl(cpu_gprh[rA(ctx->opcode)], cpu_gprh[rA(ctx->opcode)], 0x80000000);
+#endif
+}
+static always_inline void gen_efdneg (DisasContext *ctx)
+{
+    if (unlikely(!ctx->spe_enabled)) {
+        GEN_EXCP_NO_AP(ctx);
+        return;
+    }
+#if defined(TARGET_PPC64)
+    tcg_gen_xori_tl(cpu_gpr[rA(ctx->opcode)], cpu_gpr[rA(ctx->opcode)], 0x8000000000000000LL);
+#else
+    tcg_gen_xori_tl(cpu_gprh[rA(ctx->opcode)], cpu_gprh[rA(ctx->opcode)], 0x80000000);
+#endif
+}
+
 /* Conversion */
+GEN_SPEFPUOP_CONV_64_32(efdcfui);
+GEN_SPEFPUOP_CONV_64_32(efdcfsi);
+GEN_SPEFPUOP_CONV_64_32(efdcfuf);
+GEN_SPEFPUOP_CONV_64_32(efdcfsf);
+GEN_SPEFPUOP_CONV_32_64(efdctui);
+GEN_SPEFPUOP_CONV_32_64(efdctsi);
+GEN_SPEFPUOP_CONV_32_64(efdctuf);
+GEN_SPEFPUOP_CONV_32_64(efdctsf);
+GEN_SPEFPUOP_CONV_32_64(efdctuiz);
+GEN_SPEFPUOP_CONV_32_64(efdctsiz);
+GEN_SPEFPUOP_CONV_64_32(efdcfs);
+GEN_SPEFPUOP_CONV_64_64(efdcfuid);
+GEN_SPEFPUOP_CONV_64_64(efdcfsid);
+GEN_SPEFPUOP_CONV_64_64(efdctuidz);
+GEN_SPEFPUOP_CONV_64_64(efdctsidz);
 
-GEN_SPEFPUOP_CONV(efdcfui);
-GEN_SPEFPUOP_CONV(efdcfsi);
-GEN_SPEFPUOP_CONV(efdcfuf);
-GEN_SPEFPUOP_CONV(efdcfsf);
-GEN_SPEFPUOP_CONV(efdctui);
-GEN_SPEFPUOP_CONV(efdctsi);
-GEN_SPEFPUOP_CONV(efdctuf);
-GEN_SPEFPUOP_CONV(efdctsf);
-GEN_SPEFPUOP_CONV(efdctuiz);
-GEN_SPEFPUOP_CONV(efdctsiz);
-GEN_SPEFPUOP_CONV(efdcfs);
-GEN_SPEFPUOP_CONV(efdcfuid);
-GEN_SPEFPUOP_CONV(efdcfsid);
-GEN_SPEFPUOP_CONV(efdctuidz);
-GEN_SPEFPUOP_CONV(efdctsidz);
 /* Comparison */
-GEN_SPEFPUOP_COMP(efdcmpgt);
-GEN_SPEFPUOP_COMP(efdcmplt);
-GEN_SPEFPUOP_COMP(efdcmpeq);
-GEN_SPEFPUOP_COMP(efdtstgt);
-GEN_SPEFPUOP_COMP(efdtstlt);
-GEN_SPEFPUOP_COMP(efdtsteq);
+GEN_SPEFPUOP_COMP_64(efdcmpgt);
+GEN_SPEFPUOP_COMP_64(efdcmplt);
+GEN_SPEFPUOP_COMP_64(efdcmpeq);
+GEN_SPEFPUOP_COMP_64(efdtstgt);
+GEN_SPEFPUOP_COMP_64(efdtstlt);
+GEN_SPEFPUOP_COMP_64(efdtsteq);
 
 /* Opcodes definitions */
 GEN_SPE(efdadd,         efdsub,        0x10, 0x0B, 0x00000000, PPC_SPEFPU); //