* Copyright (c) 2007, 2008 CodeSourcery.
* Written by Paul Brook
*
- * This code is licenced under the GNU GPL v2.
+ * This code is licensed under the GNU GPL v2.
*/
#include <stdlib.h>
#include <stdio.h>
#include "cpu.h"
#include "exec-all.h"
-#include "helpers.h"
+#include "helper.h"
#define SIGNBIT (uint32_t)0x80000000
#define SIGNBIT64 ((uint64_t)1 << 63)
#define SET_QC() env->vfp.xregs[ARM_VFP_FPSCR] = CPSR_Q
-static float_status neon_float_status;
-#define NFS &neon_float_status
-
#define NEON_TYPE1(name, type) \
typedef struct \
{ \
NEON_VOP_BODY(vtype, n)
#define NEON_VOP_ENV(name, vtype, n) \
-uint32_t HELPER(glue(neon_,name))(CPUState *env, uint32_t arg1, uint32_t arg2) \
+uint32_t HELPER(glue(neon_,name))(CPUARMState *env, uint32_t arg1, uint32_t arg2) \
NEON_VOP_BODY(vtype, n)
/* Pairwise operations. */
#undef NEON_FN
#undef NEON_USAT
-uint32_t HELPER(neon_qadd_u32)(CPUState *env, uint32_t a, uint32_t b)
+uint32_t HELPER(neon_qadd_u32)(CPUARMState *env, uint32_t a, uint32_t b)
{
uint32_t res = a + b;
if (res < a) {
return res;
}
-uint64_t HELPER(neon_qadd_u64)(CPUState *env, uint64_t src1, uint64_t src2)
+uint64_t HELPER(neon_qadd_u64)(CPUARMState *env, uint64_t src1, uint64_t src2)
{
uint64_t res;
#undef NEON_FN
#undef NEON_SSAT
-uint32_t HELPER(neon_qadd_s32)(CPUState *env, uint32_t a, uint32_t b)
+uint32_t HELPER(neon_qadd_s32)(CPUARMState *env, uint32_t a, uint32_t b)
{
uint32_t res = a + b;
if (((res ^ a) & SIGNBIT) && !((a ^ b) & SIGNBIT)) {
return res;
}
-uint64_t HELPER(neon_qadd_s64)(CPUState *env, uint64_t src1, uint64_t src2)
+uint64_t HELPER(neon_qadd_s64)(CPUARMState *env, uint64_t src1, uint64_t src2)
{
uint64_t res;
#undef NEON_FN
#undef NEON_USAT
-uint32_t HELPER(neon_qsub_u32)(CPUState *env, uint32_t a, uint32_t b)
+uint32_t HELPER(neon_qsub_u32)(CPUARMState *env, uint32_t a, uint32_t b)
{
uint32_t res = a - b;
if (res > a) {
return res;
}
-uint64_t HELPER(neon_qsub_u64)(CPUState *env, uint64_t src1, uint64_t src2)
+uint64_t HELPER(neon_qsub_u64)(CPUARMState *env, uint64_t src1, uint64_t src2)
{
uint64_t res;
#undef NEON_FN
#undef NEON_SSAT
-uint32_t HELPER(neon_qsub_s32)(CPUState *env, uint32_t a, uint32_t b)
+uint32_t HELPER(neon_qsub_s32)(CPUARMState *env, uint32_t a, uint32_t b)
{
uint32_t res = a - b;
if (((res ^ a) & SIGNBIT) && ((a ^ b) & SIGNBIT)) {
return res;
}
-uint64_t HELPER(neon_qsub_s64)(CPUState *env, uint64_t src1, uint64_t src2)
+uint64_t HELPER(neon_qsub_s64)(CPUARMState *env, uint64_t src1, uint64_t src2)
{
uint64_t res;
NEON_VOP_ENV(qshl_u32, neon_u32, 1)
#undef NEON_FN
-uint64_t HELPER(neon_qshl_u64)(CPUState *env, uint64_t val, uint64_t shiftop)
+uint64_t HELPER(neon_qshl_u64)(CPUARMState *env, uint64_t val, uint64_t shiftop)
{
int8_t shift = (int8_t)shiftop;
if (shift >= 64) {
NEON_VOP_ENV(qshl_s32, neon_s32, 1)
#undef NEON_FN
-uint64_t HELPER(neon_qshl_s64)(CPUState *env, uint64_t valop, uint64_t shiftop)
+uint64_t HELPER(neon_qshl_s64)(CPUARMState *env, uint64_t valop, uint64_t shiftop)
{
int8_t shift = (uint8_t)shiftop;
int64_t val = valop;
NEON_VOP_ENV(qshlu_s16, neon_u16, 2)
#undef NEON_FN
-uint32_t HELPER(neon_qshlu_s32)(CPUState *env, uint32_t valop, uint32_t shiftop)
+uint32_t HELPER(neon_qshlu_s32)(CPUARMState *env, uint32_t valop, uint32_t shiftop)
{
if ((int32_t)valop < 0) {
SET_QC();
return helper_neon_qshl_u32(env, valop, shiftop);
}
-uint64_t HELPER(neon_qshlu_s64)(CPUState *env, uint64_t valop, uint64_t shiftop)
+uint64_t HELPER(neon_qshlu_s64)(CPUARMState *env, uint64_t valop, uint64_t shiftop)
{
if ((int64_t)valop < 0) {
SET_QC();
/* The addition of the rounding constant may overflow, so we use an
* intermediate 64 bits accumulator. */
-uint32_t HELPER(neon_qrshl_u32)(CPUState *env, uint32_t val, uint32_t shiftop)
+uint32_t HELPER(neon_qrshl_u32)(CPUARMState *env, uint32_t val, uint32_t shiftop)
{
uint32_t dest;
int8_t shift = (int8_t)shiftop;
/* Handling addition overflow with 64 bits inputs values is more
* tricky than with 32 bits values. */
-uint64_t HELPER(neon_qrshl_u64)(CPUState *env, uint64_t val, uint64_t shiftop)
+uint64_t HELPER(neon_qrshl_u64)(CPUARMState *env, uint64_t val, uint64_t shiftop)
{
int8_t shift = (int8_t)shiftop;
if (shift >= 64) {
/* The addition of the rounding constant may overflow, so we use an
* intermediate 64 bits accumulator. */
-uint32_t HELPER(neon_qrshl_s32)(CPUState *env, uint32_t valop, uint32_t shiftop)
+uint32_t HELPER(neon_qrshl_s32)(CPUARMState *env, uint32_t valop, uint32_t shiftop)
{
int32_t dest;
int32_t val = (int32_t)valop;
/* Handling addition overflow with 64 bits inputs values is more
* tricky than with 32 bits values. */
-uint64_t HELPER(neon_qrshl_s64)(CPUState *env, uint64_t valop, uint64_t shiftop)
+uint64_t HELPER(neon_qrshl_s64)(CPUARMState *env, uint64_t valop, uint64_t shiftop)
{
int8_t shift = (uint8_t)shiftop;
int64_t val = valop;
return ((x >> 16) & 0xffff) | ((x >> 32) & 0xffff0000);
}
-uint32_t HELPER(neon_unarrow_sat8)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_unarrow_sat8)(CPUARMState *env, uint64_t x)
{
uint16_t s;
uint8_t d;
return res;
}
-uint32_t HELPER(neon_narrow_sat_u8)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_narrow_sat_u8)(CPUARMState *env, uint64_t x)
{
uint16_t s;
uint8_t d;
return res;
}
-uint32_t HELPER(neon_narrow_sat_s8)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_narrow_sat_s8)(CPUARMState *env, uint64_t x)
{
int16_t s;
uint8_t d;
return res;
}
-uint32_t HELPER(neon_unarrow_sat16)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_unarrow_sat16)(CPUARMState *env, uint64_t x)
{
uint32_t high;
uint32_t low;
return low | (high << 16);
}
-uint32_t HELPER(neon_narrow_sat_u16)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_narrow_sat_u16)(CPUARMState *env, uint64_t x)
{
uint32_t high;
uint32_t low;
return low | (high << 16);
}
-uint32_t HELPER(neon_narrow_sat_s16)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_narrow_sat_s16)(CPUARMState *env, uint64_t x)
{
int32_t low;
int32_t high;
return (uint16_t)low | (high << 16);
}
-uint32_t HELPER(neon_unarrow_sat32)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_unarrow_sat32)(CPUARMState *env, uint64_t x)
{
if (x & 0x8000000000000000ull) {
SET_QC();
return x;
}
-uint32_t HELPER(neon_narrow_sat_u32)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_narrow_sat_u32)(CPUARMState *env, uint64_t x)
{
if (x > 0xffffffffu) {
SET_QC();
return x;
}
-uint32_t HELPER(neon_narrow_sat_s32)(CPUState *env, uint64_t x)
+uint32_t HELPER(neon_narrow_sat_s32)(CPUARMState *env, uint64_t x)
{
if ((int64_t)x != (int32_t)x) {
SET_QC();
return (a - b) ^ mask;
}
-uint64_t HELPER(neon_addl_saturate_s32)(CPUState *env, uint64_t a, uint64_t b)
+uint64_t HELPER(neon_addl_saturate_s32)(CPUARMState *env, uint64_t a, uint64_t b)
{
uint32_t x, y;
uint32_t low, high;
return low | ((uint64_t)high << 32);
}
-uint64_t HELPER(neon_addl_saturate_s64)(CPUState *env, uint64_t a, uint64_t b)
+uint64_t HELPER(neon_addl_saturate_s64)(CPUARMState *env, uint64_t a, uint64_t b)
{
uint64_t result;
return result;
}
-#define DO_ABD(dest, x, y, type) do { \
- type tmp_x = x; \
- type tmp_y = y; \
+/* We have to do the arithmetic in a larger type than
+ * the input type, because for example with a signed 32 bit
+ * op the absolute difference can overflow a signed 32 bit value.
+ */
+#define DO_ABD(dest, x, y, intype, arithtype) do { \
+ arithtype tmp_x = (intype)(x); \
+ arithtype tmp_y = (intype)(y); \
dest = ((tmp_x > tmp_y) ? tmp_x - tmp_y : tmp_y - tmp_x); \
} while(0)
{
uint64_t tmp;
uint64_t result;
- DO_ABD(result, a, b, uint8_t);
- DO_ABD(tmp, a >> 8, b >> 8, uint8_t);
+ DO_ABD(result, a, b, uint8_t, uint32_t);
+ DO_ABD(tmp, a >> 8, b >> 8, uint8_t, uint32_t);
result |= tmp << 16;
- DO_ABD(tmp, a >> 16, b >> 16, uint8_t);
+ DO_ABD(tmp, a >> 16, b >> 16, uint8_t, uint32_t);
result |= tmp << 32;
- DO_ABD(tmp, a >> 24, b >> 24, uint8_t);
+ DO_ABD(tmp, a >> 24, b >> 24, uint8_t, uint32_t);
result |= tmp << 48;
return result;
}
{
uint64_t tmp;
uint64_t result;
- DO_ABD(result, a, b, int8_t);
- DO_ABD(tmp, a >> 8, b >> 8, int8_t);
+ DO_ABD(result, a, b, int8_t, int32_t);
+ DO_ABD(tmp, a >> 8, b >> 8, int8_t, int32_t);
result |= tmp << 16;
- DO_ABD(tmp, a >> 16, b >> 16, int8_t);
+ DO_ABD(tmp, a >> 16, b >> 16, int8_t, int32_t);
result |= tmp << 32;
- DO_ABD(tmp, a >> 24, b >> 24, int8_t);
+ DO_ABD(tmp, a >> 24, b >> 24, int8_t, int32_t);
result |= tmp << 48;
return result;
}
{
uint64_t tmp;
uint64_t result;
- DO_ABD(result, a, b, uint16_t);
- DO_ABD(tmp, a >> 16, b >> 16, uint16_t);
+ DO_ABD(result, a, b, uint16_t, uint32_t);
+ DO_ABD(tmp, a >> 16, b >> 16, uint16_t, uint32_t);
return result | (tmp << 32);
}
{
uint64_t tmp;
uint64_t result;
- DO_ABD(result, a, b, int16_t);
- DO_ABD(tmp, a >> 16, b >> 16, int16_t);
+ DO_ABD(result, a, b, int16_t, int32_t);
+ DO_ABD(tmp, a >> 16, b >> 16, int16_t, int32_t);
return result | (tmp << 32);
}
uint64_t HELPER(neon_abdl_u64)(uint32_t a, uint32_t b)
{
uint64_t result;
- DO_ABD(result, a, b, uint32_t);
+ DO_ABD(result, a, b, uint32_t, uint64_t);
return result;
}
uint64_t HELPER(neon_abdl_s64)(uint32_t a, uint32_t b)
{
uint64_t result;
- DO_ABD(result, a, b, int32_t);
+ DO_ABD(result, a, b, int32_t, int64_t);
return result;
}
#undef DO_ABD
} else if (x < 0) { \
x = -x; \
}} while (0)
-uint32_t HELPER(neon_qabs_s8)(CPUState *env, uint32_t x)
+uint32_t HELPER(neon_qabs_s8)(CPUARMState *env, uint32_t x)
{
neon_s8 vec;
NEON_UNPACK(neon_s8, vec, x);
} else { \
x = -x; \
}} while (0)
-uint32_t HELPER(neon_qneg_s8)(CPUState *env, uint32_t x)
+uint32_t HELPER(neon_qneg_s8)(CPUARMState *env, uint32_t x)
{
neon_s8 vec;
NEON_UNPACK(neon_s8, vec, x);
} else if (x < 0) { \
x = -x; \
}} while (0)
-uint32_t HELPER(neon_qabs_s16)(CPUState *env, uint32_t x)
+uint32_t HELPER(neon_qabs_s16)(CPUARMState *env, uint32_t x)
{
neon_s16 vec;
NEON_UNPACK(neon_s16, vec, x);
} else { \
x = -x; \
}} while (0)
-uint32_t HELPER(neon_qneg_s16)(CPUState *env, uint32_t x)
+uint32_t HELPER(neon_qneg_s16)(CPUARMState *env, uint32_t x)
{
neon_s16 vec;
NEON_UNPACK(neon_s16, vec, x);
}
#undef DO_QNEG16
-uint32_t HELPER(neon_qabs_s32)(CPUState *env, uint32_t x)
+uint32_t HELPER(neon_qabs_s32)(CPUARMState *env, uint32_t x)
{
if (x == SIGNBIT) {
SET_QC();
return x;
}
-uint32_t HELPER(neon_qneg_s32)(CPUState *env, uint32_t x)
+uint32_t HELPER(neon_qneg_s32)(CPUARMState *env, uint32_t x)
{
if (x == SIGNBIT) {
SET_QC();
}
/* NEON Float helpers. */
-uint32_t HELPER(neon_min_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_min_f32)(uint32_t a, uint32_t b, void *fpstp)
{
- float32 f0 = make_float32(a);
- float32 f1 = make_float32(b);
- return (float32_compare_quiet(f0, f1, NFS) == -1) ? a : b;
+ float_status *fpst = fpstp;
+ return float32_val(float32_min(make_float32(a), make_float32(b), fpst));
}
-uint32_t HELPER(neon_max_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_max_f32)(uint32_t a, uint32_t b, void *fpstp)
{
- float32 f0 = make_float32(a);
- float32 f1 = make_float32(b);
- return (float32_compare_quiet(f0, f1, NFS) == 1) ? a : b;
+ float_status *fpst = fpstp;
+ return float32_val(float32_max(make_float32(a), make_float32(b), fpst));
}
-uint32_t HELPER(neon_abd_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_abd_f32)(uint32_t a, uint32_t b, void *fpstp)
{
+ float_status *fpst = fpstp;
float32 f0 = make_float32(a);
float32 f1 = make_float32(b);
- return float32_val((float32_compare_quiet(f0, f1, NFS) == 1)
- ? float32_sub(f0, f1, NFS)
- : float32_sub(f1, f0, NFS));
+ return float32_val(float32_abs(float32_sub(f0, f1, fpst)));
}
-uint32_t HELPER(neon_add_f32)(uint32_t a, uint32_t b)
+/* Floating point comparisons produce an integer result.
+ * Note that EQ doesn't signal InvalidOp for QNaNs but GE and GT do.
+ * Softfloat routines return 0/1, which we convert to the 0/-1 Neon requires.
+ */
+uint32_t HELPER(neon_ceq_f32)(uint32_t a, uint32_t b, void *fpstp)
{
- return float32_val(float32_add(make_float32(a), make_float32(b), NFS));
+ float_status *fpst = fpstp;
+ return -float32_eq_quiet(make_float32(a), make_float32(b), fpst);
}
-uint32_t HELPER(neon_sub_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_cge_f32)(uint32_t a, uint32_t b, void *fpstp)
{
- return float32_val(float32_sub(make_float32(a), make_float32(b), NFS));
+ float_status *fpst = fpstp;
+ return -float32_le(make_float32(b), make_float32(a), fpst);
}
-uint32_t HELPER(neon_mul_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_cgt_f32)(uint32_t a, uint32_t b, void *fpstp)
{
- return float32_val(float32_mul(make_float32(a), make_float32(b), NFS));
+ float_status *fpst = fpstp;
+ return -float32_lt(make_float32(b), make_float32(a), fpst);
}
-/* Floating point comparisons produce an integer result. */
-#define NEON_VOP_FCMP(name, cmp) \
-uint32_t HELPER(neon_##name)(uint32_t a, uint32_t b) \
-{ \
- if (float32_compare_quiet(make_float32(a), make_float32(b), NFS) cmp 0) { \
- return ~0; \
- } else { \
- return 0; \
- } \
-}
-
-NEON_VOP_FCMP(ceq_f32, ==)
-NEON_VOP_FCMP(cge_f32, >=)
-NEON_VOP_FCMP(cgt_f32, >)
-
-uint32_t HELPER(neon_acge_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_acge_f32)(uint32_t a, uint32_t b, void *fpstp)
{
+ float_status *fpst = fpstp;
float32 f0 = float32_abs(make_float32(a));
float32 f1 = float32_abs(make_float32(b));
- return (float32_compare_quiet(f0, f1,NFS) >= 0) ? ~0 : 0;
+ return -float32_le(f1, f0, fpst);
}
-uint32_t HELPER(neon_acgt_f32)(uint32_t a, uint32_t b)
+uint32_t HELPER(neon_acgt_f32)(uint32_t a, uint32_t b, void *fpstp)
{
+ float_status *fpst = fpstp;
float32 f0 = float32_abs(make_float32(a));
float32 f1 = float32_abs(make_float32(b));
- return (float32_compare_quiet(f0, f1, NFS) > 0) ? ~0 : 0;
+ return -float32_lt(f1, f0, fpst);
}
#define ELEM(V, N, SIZE) (((V) >> ((N) * (SIZE))) & ((1ull << (SIZE)) - 1))
-void HELPER(neon_qunzip8)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_qunzip8)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm0 = float64_val(env->vfp.regs[rm]);
uint64_t zm1 = float64_val(env->vfp.regs[rm + 1]);
env->vfp.regs[rd + 1] = make_float64(d1);
}
-void HELPER(neon_qunzip16)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_qunzip16)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm0 = float64_val(env->vfp.regs[rm]);
uint64_t zm1 = float64_val(env->vfp.regs[rm + 1]);
env->vfp.regs[rd + 1] = make_float64(d1);
}
-void HELPER(neon_qunzip32)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_qunzip32)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm0 = float64_val(env->vfp.regs[rm]);
uint64_t zm1 = float64_val(env->vfp.regs[rm + 1]);
env->vfp.regs[rd + 1] = make_float64(d1);
}
-void HELPER(neon_unzip8)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_unzip8)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm = float64_val(env->vfp.regs[rm]);
uint64_t zd = float64_val(env->vfp.regs[rd]);
env->vfp.regs[rd] = make_float64(d0);
}
-void HELPER(neon_unzip16)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_unzip16)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm = float64_val(env->vfp.regs[rm]);
uint64_t zd = float64_val(env->vfp.regs[rd]);
env->vfp.regs[rd] = make_float64(d0);
}
-void HELPER(neon_qzip8)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_qzip8)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm0 = float64_val(env->vfp.regs[rm]);
uint64_t zm1 = float64_val(env->vfp.regs[rm + 1]);
env->vfp.regs[rd + 1] = make_float64(d1);
}
-void HELPER(neon_qzip16)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_qzip16)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm0 = float64_val(env->vfp.regs[rm]);
uint64_t zm1 = float64_val(env->vfp.regs[rm + 1]);
env->vfp.regs[rd + 1] = make_float64(d1);
}
-void HELPER(neon_qzip32)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_qzip32)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm0 = float64_val(env->vfp.regs[rm]);
uint64_t zm1 = float64_val(env->vfp.regs[rm + 1]);
env->vfp.regs[rd + 1] = make_float64(d1);
}
-void HELPER(neon_zip8)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_zip8)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm = float64_val(env->vfp.regs[rm]);
uint64_t zd = float64_val(env->vfp.regs[rd]);
env->vfp.regs[rd] = make_float64(d0);
}
-void HELPER(neon_zip16)(CPUState *env, uint32_t rd, uint32_t rm)
+void HELPER(neon_zip16)(CPUARMState *env, uint32_t rd, uint32_t rm)
{
uint64_t zm = float64_val(env->vfp.regs[rm]);
uint64_t zd = float64_val(env->vfp.regs[rd]);
projects / qemu.git / blobdiff