aboutsummaryrefslogtreecommitdiff
path: root/target-ppc/op_helper.c
diff options
context:
space:
mode:
Diffstat (limited to 'target-ppc/op_helper.c')
-rw-r--r--target-ppc/op_helper.c166
1 files changed, 83 insertions, 83 deletions
diff --git a/target-ppc/op_helper.c b/target-ppc/op_helper.c
index e6cec76..e3bd29c 100644
--- a/target-ppc/op_helper.c
+++ b/target-ppc/op_helper.c
@@ -1782,7 +1782,7 @@ target_ulong helper_divso (target_ulong arg1, target_ulong arg2)
#if !defined (CONFIG_USER_ONLY)
target_ulong helper_rac (target_ulong addr)
{
- a_mmu_ctx ctx;
+ mmu_ctx_t ctx;
int nb_BATs;
target_ulong ret = 0;
@@ -1996,7 +1996,7 @@ SATCVT(sd, uw, int64_t, uint32_t, 0, UINT32_MAX, 1, 1)
#undef SATCVT
#define LVE(name, access, swap, element) \
- void helper_##name (union ppc_avr *r, target_ulong addr) \
+ void helper_##name (ppc_avr_t *r, target_ulong addr) \
{ \
size_t n_elems = ARRAY_SIZE(r->element); \
int adjust = HI_IDX*(n_elems-1); \
@@ -2015,7 +2015,7 @@ LVE(lvewx, ldl, bswap32, u32)
#undef I
#undef LVE
-void helper_lvsl (union ppc_avr *r, target_ulong sh)
+void helper_lvsl (ppc_avr_t *r, target_ulong sh)
{
int i, j = (sh & 0xf);
@@ -2024,7 +2024,7 @@ void helper_lvsl (union ppc_avr *r, target_ulong sh)
}
}
-void helper_lvsr (union ppc_avr *r, target_ulong sh)
+void helper_lvsr (ppc_avr_t *r, target_ulong sh)
{
int i, j = 0x10 - (sh & 0xf);
@@ -2034,7 +2034,7 @@ void helper_lvsr (union ppc_avr *r, target_ulong sh)
}
#define STVE(name, access, swap, element) \
- void helper_##name (union ppc_avr *r, target_ulong addr) \
+ void helper_##name (ppc_avr_t *r, target_ulong addr) \
{ \
size_t n_elems = ARRAY_SIZE(r->element); \
int adjust = HI_IDX*(n_elems-1); \
@@ -2053,7 +2053,7 @@ STVE(stvewx, stl, bswap32, u32)
#undef I
#undef LVE
-void helper_mtvscr (union ppc_avr *r)
+void helper_mtvscr (ppc_avr_t *r)
{
#if defined(HOST_WORDS_BIGENDIAN)
env->vscr = r->u32[3];
@@ -2063,7 +2063,7 @@ void helper_mtvscr (union ppc_avr *r)
set_flush_to_zero(vscr_nj, &env->vec_status);
}
-void helper_vaddcuw (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vaddcuw (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int i;
for (i = 0; i < ARRAY_SIZE(r->u32); i++) {
@@ -2071,13 +2071,13 @@ void helper_vaddcuw (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
}
-#define VARITH_DO(name, op, element) \
-void helper_v##name (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
-{ \
- int i; \
- for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
- r->element[i] = a->element[i] op b->element[i]; \
- } \
+#define VARITH_DO(name, op, element) \
+void helper_v##name (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
+{ \
+ int i; \
+ for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
+ r->element[i] = a->element[i] op b->element[i]; \
+ } \
}
#define VARITH(suffix, element) \
VARITH_DO(add##suffix, +, element) \
@@ -2089,7 +2089,7 @@ VARITH(uwm, u32)
#undef VARITH
#define VARITHFP(suffix, func) \
- void helper_v##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)\
+ void helper_v##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->f); i++) { \
@@ -2109,7 +2109,7 @@ VARITHFP(subfp, float32_sub)
}
#define VARITHSAT_DO(name, op, optype, cvt, element) \
- void helper_v##name (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_v##name (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int sat = 0; \
int i; \
@@ -2142,7 +2142,7 @@ VARITHSAT_UNSIGNED(w, u32, uint64_t, cvtsduw)
#undef VARITHSAT_UNSIGNED
#define VAVG_DO(name, element, etype) \
- void helper_v##name (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)\
+ void helper_v##name (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
@@ -2161,7 +2161,7 @@ VAVG(w, s32, int64_t, u32, uint64_t)
#undef VAVG
#define VCF(suffix, cvt, element) \
- void helper_vcf##suffix (union ppc_avr *r, union ppc_avr *b, uint32_t uim) \
+ void helper_vcf##suffix (ppc_avr_t *r, ppc_avr_t *b, uint32_t uim) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->f); i++) { \
@@ -2174,7 +2174,7 @@ VCF(sx, int32_to_float32, s32)
#undef VCF
#define VCMP_DO(suffix, compare, element, record) \
- void helper_vcmp##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vcmp##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
uint32_t ones = (uint32_t)-1; \
uint32_t all = ones; \
@@ -2210,7 +2210,7 @@ VCMP(gtsw, >, s32)
#undef VCMP
#define VCMPFP_DO(suffix, compare, order, record) \
- void helper_vcmp##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vcmp##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
uint32_t ones = (uint32_t)-1; \
uint32_t all = ones; \
@@ -2243,7 +2243,7 @@ VCMPFP(gtfp, ==, float_relation_greater)
#undef VCMPFP_DO
#undef VCMPFP
-static inline void vcmpbfp_internal(union ppc_avr *r, union ppc_avr *a, union ppc_avr *b,
+static inline void vcmpbfp_internal(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b,
int record)
{
int i;
@@ -2267,18 +2267,18 @@ static inline void vcmpbfp_internal(union ppc_avr *r, union ppc_avr *a, union pp
}
}
-void helper_vcmpbfp (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vcmpbfp (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
vcmpbfp_internal(r, a, b, 0);
}
-void helper_vcmpbfp_dot (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vcmpbfp_dot (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
vcmpbfp_internal(r, a, b, 1);
}
#define VCT(suffix, satcvt, element) \
- void helper_vct##suffix (union ppc_avr *r, union ppc_avr *b, uint32_t uim) \
+ void helper_vct##suffix (ppc_avr_t *r, ppc_avr_t *b, uint32_t uim) \
{ \
int i; \
int sat = 0; \
@@ -2304,7 +2304,7 @@ VCT(uxs, cvtsduw, u32)
VCT(sxs, cvtsdsw, s32)
#undef VCT
-void helper_vmaddfp (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+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++) {
@@ -2322,7 +2322,7 @@ void helper_vmaddfp (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union
}
}
-void helper_vmhaddshs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmhaddshs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
int sat = 0;
int i;
@@ -2338,7 +2338,7 @@ void helper_vmhaddshs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, uni
}
}
-void helper_vmhraddshs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmhraddshs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
int sat = 0;
int i;
@@ -2355,7 +2355,7 @@ void helper_vmhraddshs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, un
}
#define VMINMAX_DO(name, compare, element) \
- void helper_v##name (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_v##name (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
@@ -2379,7 +2379,7 @@ VMINMAX(uw, u32)
#undef VMINMAX
#define VMINMAXFP(suffix, rT, rF) \
- void helper_v##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_v##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->f); i++) { \
@@ -2396,7 +2396,7 @@ VMINMAXFP(minfp, a, b)
VMINMAXFP(maxfp, b, a)
#undef VMINMAXFP
-void helper_vmladduhm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmladduhm (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->s16); i++) {
@@ -2406,9 +2406,9 @@ void helper_vmladduhm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, uni
}
#define VMRG_DO(name, element, highp) \
- void helper_v##name (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_v##name (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
- union ppc_avr result; \
+ ppc_avr_t result; \
int i; \
size_t n_elems = ARRAY_SIZE(r->element); \
for (i = 0; i < n_elems/2; i++) { \
@@ -2440,7 +2440,7 @@ VMRG(w, u32)
#undef MRGHI
#undef MRGLO
-void helper_vmsummbm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmsummbm (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
int32_t prod[16];
int i;
@@ -2454,7 +2454,7 @@ void helper_vmsummbm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
}
-void helper_vmsumshm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmsumshm (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
int32_t prod[8];
int i;
@@ -2468,7 +2468,7 @@ void helper_vmsumshm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
}
-void helper_vmsumshs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmsumshs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
int32_t prod[8];
int i;
@@ -2488,7 +2488,7 @@ void helper_vmsumshs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
}
-void helper_vmsumubm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmsumubm (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
uint16_t prod[16];
int i;
@@ -2502,7 +2502,7 @@ void helper_vmsumubm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
}
-void helper_vmsumuhm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmsumuhm (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
uint32_t prod[8];
int i;
@@ -2516,7 +2516,7 @@ void helper_vmsumuhm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
}
-void helper_vmsumuhs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vmsumuhs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
uint32_t prod[8];
int i;
@@ -2537,7 +2537,7 @@ void helper_vmsumuhs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
#define VMUL_DO(name, mul_element, prod_element, evenp) \
- void helper_v##name (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_v##name (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
VECTOR_FOR_INORDER_I(i, prod_element) { \
@@ -2558,7 +2558,7 @@ VMUL(uh, u16, u32)
#undef VMUL_DO
#undef VMUL
-void helper_vnmsubfp (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+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++) {
@@ -2577,9 +2577,9 @@ void helper_vnmsubfp (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, unio
}
}
-void helper_vperm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vperm (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
- union ppc_avr result;
+ ppc_avr_t result;
int i;
VECTOR_FOR_INORDER_I (i, u8) {
int s = c->u8[i] & 0x1f;
@@ -2602,14 +2602,14 @@ void helper_vperm (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union p
#else
#define PKBIG 0
#endif
-void helper_vpkpx (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vpkpx (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int i, j;
- union ppc_avr result;
+ ppc_avr_t result;
#if defined(HOST_WORDS_BIGENDIAN)
- const union ppc_avr *x[2] = { a, b };
+ const ppc_avr_t *x[2] = { a, b };
#else
- const union ppc_avr *x[2] = { b, a };
+ const ppc_avr_t *x[2] = { b, a };
#endif
VECTOR_FOR_INORDER_I (i, u64) {
@@ -2624,13 +2624,13 @@ void helper_vpkpx (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
#define VPK(suffix, from, to, cvt, dosat) \
- void helper_vpk##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vpk##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
int sat = 0; \
- union ppc_avr result; \
- union ppc_avr *a0 = PKBIG ? a : b; \
- union ppc_avr *a1 = PKBIG ? b : a; \
+ ppc_avr_t result; \
+ ppc_avr_t *a0 = PKBIG ? a : b; \
+ ppc_avr_t *a1 = PKBIG ? b : a; \
VECTOR_FOR_INORDER_I (i, from) { \
result.to[i] = cvt(a0->from[i], &sat); \
result.to[i+ARRAY_SIZE(r->from)] = cvt(a1->from[i], &sat); \
@@ -2653,7 +2653,7 @@ VPK(uwum, u32, u16, I, 0)
#undef VPK
#undef PKBIG
-void helper_vrefp (union ppc_avr *r, union ppc_avr *b)
+void helper_vrefp (ppc_avr_t *r, ppc_avr_t *b)
{
int i;
for (i = 0; i < ARRAY_SIZE(r->f); i++) {
@@ -2664,7 +2664,7 @@ void helper_vrefp (union ppc_avr *r, union ppc_avr *b)
}
#define VRFI(suffix, rounding) \
- void helper_vrfi##suffix (union ppc_avr *r, union ppc_avr *b) \
+ void helper_vrfi##suffix (ppc_avr_t *r, ppc_avr_t *b) \
{ \
int i; \
float_status s = env->vec_status; \
@@ -2682,7 +2682,7 @@ VRFI(z, float_round_to_zero)
#undef VRFI
#define VROTATE(suffix, element) \
- void helper_vrl##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vrl##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
@@ -2696,7 +2696,7 @@ VROTATE(h, u16)
VROTATE(w, u32)
#undef VROTATE
-void helper_vrsqrtefp (union ppc_avr *r, union ppc_avr *b)
+void helper_vrsqrtefp (ppc_avr_t *r, ppc_avr_t *b)
{
int i;
for (i = 0; i < ARRAY_SIZE(r->f); i++) {
@@ -2707,13 +2707,13 @@ void helper_vrsqrtefp (union ppc_avr *r, union ppc_avr *b)
}
}
-void helper_vsel (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, union ppc_avr *c)
+void helper_vsel (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, ppc_avr_t *c)
{
r->u64[0] = (a->u64[0] & ~c->u64[0]) | (b->u64[0] & c->u64[0]);
r->u64[1] = (a->u64[1] & ~c->u64[1]) | (b->u64[1] & c->u64[1]);
}
-void helper_vlogefp (union ppc_avr *r, union ppc_avr *b)
+void helper_vlogefp (ppc_avr_t *r, ppc_avr_t *b)
{
int i;
for (i = 0; i < ARRAY_SIZE(r->f); i++) {
@@ -2734,7 +2734,7 @@ void helper_vlogefp (union ppc_avr *r, union ppc_avr *b)
* shift counts are not identical. We check to make sure that they are
* to conform to what real hardware appears to do. */
#define VSHIFT(suffix, leftp) \
- void helper_vs##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vs##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int shift = b->u8[LO_IDX*15] & 0x7; \
int doit = 1; \
@@ -2763,7 +2763,7 @@ VSHIFT(r, RIGHT)
#undef RIGHT
#define VSL(suffix, element) \
- void helper_vsl##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vsl##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
@@ -2777,11 +2777,11 @@ VSL(h, u16)
VSL(w, u32)
#undef VSL
-void helper_vsldoi (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, uint32_t shift)
+void helper_vsldoi (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b, uint32_t shift)
{
int sh = shift & 0xf;
int i;
- union ppc_avr result;
+ ppc_avr_t result;
#if defined(HOST_WORDS_BIGENDIAN)
for (i = 0; i < ARRAY_SIZE(r->u8); i++) {
@@ -2805,7 +2805,7 @@ void helper_vsldoi (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b, uint32
*r = result;
}
-void helper_vslo (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vslo (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int sh = (b->u8[LO_IDX*0xf] >> 3) & 0xf;
@@ -2826,7 +2826,7 @@ void helper_vslo (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
#define SPLAT_ELEMENT(element) (ARRAY_SIZE(r->element)-1 - _SPLAT_MASKED(element))
#endif
#define VSPLT(suffix, element) \
- void helper_vsplt##suffix (union ppc_avr *r, union ppc_avr *b, uint32_t splat) \
+ void helper_vsplt##suffix (ppc_avr_t *r, ppc_avr_t *b, uint32_t splat) \
{ \
uint32_t s = b->element[SPLAT_ELEMENT(element)]; \
int i; \
@@ -2842,7 +2842,7 @@ VSPLT(w, u32)
#undef _SPLAT_MASKED
#define VSPLTI(suffix, element, splat_type) \
- void helper_vspltis##suffix (union ppc_avr *r, uint32_t splat) \
+ void helper_vspltis##suffix (ppc_avr_t *r, uint32_t splat) \
{ \
splat_type x = (int8_t)(splat << 3) >> 3; \
int i; \
@@ -2856,7 +2856,7 @@ VSPLTI(w, s32, int32_t)
#undef VSPLTI
#define VSR(suffix, element) \
- void helper_vsr##suffix (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b) \
+ void helper_vsr##suffix (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b) \
{ \
int i; \
for (i = 0; i < ARRAY_SIZE(r->element); i++) { \
@@ -2873,7 +2873,7 @@ VSR(h, u16)
VSR(w, u32)
#undef VSR
-void helper_vsro (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsro (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int sh = (b->u8[LO_IDX*0xf] >> 3) & 0xf;
@@ -2886,7 +2886,7 @@ void helper_vsro (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
#endif
}
-void helper_vsubcuw (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsubcuw (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int i;
for (i = 0; i < ARRAY_SIZE(r->u32); i++) {
@@ -2894,11 +2894,11 @@ void helper_vsubcuw (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
}
-void helper_vsumsws (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsumsws (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int64_t t;
int i, upper;
- union ppc_avr result;
+ ppc_avr_t result;
int sat = 0;
#if defined(HOST_WORDS_BIGENDIAN)
@@ -2919,10 +2919,10 @@ void helper_vsumsws (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
}
-void helper_vsum2sws (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsum2sws (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int i, j, upper;
- union ppc_avr result;
+ ppc_avr_t result;
int sat = 0;
#if defined(HOST_WORDS_BIGENDIAN)
@@ -2945,7 +2945,7 @@ void helper_vsum2sws (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
}
-void helper_vsum4sbs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsum4sbs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int i, j;
int sat = 0;
@@ -2963,7 +2963,7 @@ void helper_vsum4sbs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
}
-void helper_vsum4shs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsum4shs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int sat = 0;
int i;
@@ -2979,7 +2979,7 @@ void helper_vsum4shs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
}
}
-void helper_vsum4ubs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
+void helper_vsum4ubs (ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
{
int i, j;
int sat = 0;
@@ -3005,10 +3005,10 @@ void helper_vsum4ubs (union ppc_avr *r, union ppc_avr *a, union ppc_avr *b)
#define UPKLO 1
#endif
#define VUPKPX(suffix, hi) \
- void helper_vupk##suffix (union ppc_avr *r, union ppc_avr *b) \
+ void helper_vupk##suffix (ppc_avr_t *r, ppc_avr_t *b) \
{ \
int i; \
- union ppc_avr result; \
+ ppc_avr_t result; \
for (i = 0; i < ARRAY_SIZE(r->u32); i++) { \
uint16_t e = b->u16[hi ? i : i+4]; \
uint8_t a = (e >> 15) ? 0xff : 0; \
@@ -3024,10 +3024,10 @@ VUPKPX(hpx, UPKHI)
#undef VUPKPX
#define VUPK(suffix, unpacked, packee, hi) \
- void helper_vupk##suffix (union ppc_avr *r, union ppc_avr *b) \
+ void helper_vupk##suffix (ppc_avr_t *r, ppc_avr_t *b) \
{ \
int i; \
- union ppc_avr result; \
+ ppc_avr_t result; \
if (hi) { \
for (i = 0; i < ARRAY_SIZE(r->unpacked); i++) { \
result.unpacked[i] = b->packee[i]; \
@@ -3921,7 +3921,7 @@ static inline int booke_page_size_to_tlb(target_ulong page_size)
/* Helpers for 4xx TLB management */
target_ulong helper_4xx_tlbre_lo (target_ulong entry)
{
- a_ppcemb_tlb *tlb;
+ ppcemb_tlb_t *tlb;
target_ulong ret;
int size;
@@ -3940,7 +3940,7 @@ target_ulong helper_4xx_tlbre_lo (target_ulong entry)
target_ulong helper_4xx_tlbre_hi (target_ulong entry)
{
- a_ppcemb_tlb *tlb;
+ ppcemb_tlb_t *tlb;
target_ulong ret;
entry &= 0x3F;
@@ -3955,7 +3955,7 @@ target_ulong helper_4xx_tlbre_hi (target_ulong entry)
void helper_4xx_tlbwe_hi (target_ulong entry, target_ulong val)
{
- a_ppcemb_tlb *tlb;
+ ppcemb_tlb_t *tlb;
target_ulong page, end;
LOG_SWTLB("%s entry %d val " TARGET_FMT_lx "\n", __func__, (int)entry,
@@ -4010,7 +4010,7 @@ void helper_4xx_tlbwe_hi (target_ulong entry, target_ulong val)
void helper_4xx_tlbwe_lo (target_ulong entry, target_ulong val)
{
- a_ppcemb_tlb *tlb;
+ ppcemb_tlb_t *tlb;
LOG_SWTLB("%s entry %i val " TARGET_FMT_lx "\n", __func__, (int)entry,
val);
@@ -4039,7 +4039,7 @@ target_ulong helper_4xx_tlbsx (target_ulong address)
/* PowerPC 440 TLB management */
void helper_440_tlbwe (uint32_t word, target_ulong entry, target_ulong value)
{
- a_ppcemb_tlb *tlb;
+ ppcemb_tlb_t *tlb;
target_ulong EPN, RPN, size;
int do_flush_tlbs;
@@ -4101,7 +4101,7 @@ void helper_440_tlbwe (uint32_t word, target_ulong entry, target_ulong value)
target_ulong helper_440_tlbre (uint32_t word, target_ulong entry)
{
- a_ppcemb_tlb *tlb;
+ ppcemb_tlb_t *tlb;
target_ulong ret;
int size;