forked from suyu/suyu
dyncom: Remove static keyword from header functions
This commit is contained in:
parent
1ea0702eaa
commit
56e22e6aac
3 changed files with 19 additions and 19 deletions
|
@ -30,7 +30,7 @@
|
||||||
* @return If the PC is being read, then the word-aligned PC value is returned.
|
* @return If the PC is being read, then the word-aligned PC value is returned.
|
||||||
* If the PC is not being read, then the value stored in the register is returned.
|
* If the PC is not being read, then the value stored in the register is returned.
|
||||||
*/
|
*/
|
||||||
static inline u32 CHECK_READ_REG15_WA(const ARMul_State* cpu, int Rn) {
|
inline u32 CHECK_READ_REG15_WA(const ARMul_State* cpu, int Rn) {
|
||||||
return (Rn == 15) ? ((cpu->Reg[15] & ~0x3) + cpu->GetInstructionSize() * 2) : cpu->Reg[Rn];
|
return (Rn == 15) ? ((cpu->Reg[15] & ~0x3) + cpu->GetInstructionSize() * 2) : cpu->Reg[Rn];
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -43,6 +43,6 @@ static inline u32 CHECK_READ_REG15_WA(const ARMul_State* cpu, int Rn) {
|
||||||
* @return If the PC is being read, then the incremented PC value is returned.
|
* @return If the PC is being read, then the incremented PC value is returned.
|
||||||
* If the PC is not being read, then the values stored in the register is returned.
|
* If the PC is not being read, then the values stored in the register is returned.
|
||||||
*/
|
*/
|
||||||
static inline u32 CHECK_READ_REG15(const ARMul_State* cpu, int Rn) {
|
inline u32 CHECK_READ_REG15(const ARMul_State* cpu, int Rn) {
|
||||||
return (Rn == 15) ? ((cpu->Reg[15] & ~0x1) + cpu->GetInstructionSize() * 2) : cpu->Reg[Rn];
|
return (Rn == 15) ? ((cpu->Reg[15] & ~0x1) + cpu->GetInstructionSize() * 2) : cpu->Reg[Rn];
|
||||||
}
|
}
|
||||||
|
|
|
@ -38,7 +38,7 @@ enum class ThumbDecodeStatus {
|
||||||
// Translates a Thumb mode instruction into its ARM equivalent.
|
// Translates a Thumb mode instruction into its ARM equivalent.
|
||||||
ThumbDecodeStatus TranslateThumbInstruction(u32 addr, u32 instr, u32* ainstr, u32* inst_size);
|
ThumbDecodeStatus TranslateThumbInstruction(u32 addr, u32 instr, u32* ainstr, u32* inst_size);
|
||||||
|
|
||||||
static inline u32 GetThumbInstruction(u32 instr, u32 address) {
|
inline u32 GetThumbInstruction(u32 instr, u32 address) {
|
||||||
// Normally you would need to handle instruction endianness,
|
// Normally you would need to handle instruction endianness,
|
||||||
// however, it is fixed to little-endian on the MPCore, so
|
// however, it is fixed to little-endian on the MPCore, so
|
||||||
// there's no need to check for this beforehand.
|
// there's no need to check for this beforehand.
|
||||||
|
|
|
@ -85,7 +85,7 @@ enum : u32 {
|
||||||
|
|
||||||
#define vfp_single(inst) (((inst) & 0x0000f00) == 0xa00)
|
#define vfp_single(inst) (((inst) & 0x0000f00) == 0xa00)
|
||||||
|
|
||||||
static inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift)
|
inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift)
|
||||||
{
|
{
|
||||||
if (shift) {
|
if (shift) {
|
||||||
if (shift < 32)
|
if (shift < 32)
|
||||||
|
@ -96,7 +96,7 @@ static inline u32 vfp_shiftright32jamming(u32 val, unsigned int shift)
|
||||||
return val;
|
return val;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift)
|
inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift)
|
||||||
{
|
{
|
||||||
if (shift) {
|
if (shift) {
|
||||||
if (shift < 64)
|
if (shift < 64)
|
||||||
|
@ -107,7 +107,7 @@ static inline u64 vfp_shiftright64jamming(u64 val, unsigned int shift)
|
||||||
return val;
|
return val;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline u32 vfp_hi64to32jamming(u64 val)
|
inline u32 vfp_hi64to32jamming(u64 val)
|
||||||
{
|
{
|
||||||
u32 v;
|
u32 v;
|
||||||
u32 highval = val >> 32;
|
u32 highval = val >> 32;
|
||||||
|
@ -121,7 +121,7 @@ static inline u32 vfp_hi64to32jamming(u64 val)
|
||||||
return v;
|
return v;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
||||||
{
|
{
|
||||||
*resl = nl + ml;
|
*resl = nl + ml;
|
||||||
*resh = nh + mh;
|
*resh = nh + mh;
|
||||||
|
@ -129,7 +129,7 @@ static inline void add128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
||||||
*resh += 1;
|
*resh += 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
||||||
{
|
{
|
||||||
*resl = nl - ml;
|
*resl = nl - ml;
|
||||||
*resh = nh - mh;
|
*resh = nh - mh;
|
||||||
|
@ -137,7 +137,7 @@ static inline void sub128(u64* resh, u64* resl, u64 nh, u64 nl, u64 mh, u64 ml)
|
||||||
*resh -= 1;
|
*resh -= 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m)
|
inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m)
|
||||||
{
|
{
|
||||||
u32 nh, nl, mh, ml;
|
u32 nh, nl, mh, ml;
|
||||||
u64 rh, rma, rmb, rl;
|
u64 rh, rma, rmb, rl;
|
||||||
|
@ -164,20 +164,20 @@ static inline void mul64to128(u64* resh, u64* resl, u64 n, u64 m)
|
||||||
*resh = rh;
|
*resh = rh;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline void shift64left(u64* resh, u64* resl, u64 n)
|
inline void shift64left(u64* resh, u64* resl, u64 n)
|
||||||
{
|
{
|
||||||
*resh = n >> 63;
|
*resh = n >> 63;
|
||||||
*resl = n << 1;
|
*resl = n << 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline u64 vfp_hi64multiply64(u64 n, u64 m)
|
inline u64 vfp_hi64multiply64(u64 n, u64 m)
|
||||||
{
|
{
|
||||||
u64 rh, rl;
|
u64 rh, rl;
|
||||||
mul64to128(&rh, &rl, n, m);
|
mul64to128(&rh, &rl, n, m);
|
||||||
return rh | (rl != 0);
|
return rh | (rl != 0);
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline u64 vfp_estimate_div128to64(u64 nh, u64 nl, u64 m)
|
inline u64 vfp_estimate_div128to64(u64 nh, u64 nl, u64 m)
|
||||||
{
|
{
|
||||||
u64 mh, ml, remh, reml, termh, terml, z;
|
u64 mh, ml, remh, reml, termh, terml, z;
|
||||||
|
|
||||||
|
@ -249,7 +249,7 @@ enum : u32 {
|
||||||
VFP_SNAN = (VFP_NAN|VFP_NAN_SIGNAL)
|
VFP_SNAN = (VFP_NAN|VFP_NAN_SIGNAL)
|
||||||
};
|
};
|
||||||
|
|
||||||
static inline int vfp_single_type(const vfp_single* s)
|
inline int vfp_single_type(const vfp_single* s)
|
||||||
{
|
{
|
||||||
int type = VFP_NUMBER;
|
int type = VFP_NUMBER;
|
||||||
if (s->exponent == 255) {
|
if (s->exponent == 255) {
|
||||||
|
@ -271,7 +271,7 @@ static inline int vfp_single_type(const vfp_single* s)
|
||||||
// Unpack a single-precision float. Note that this returns the magnitude
|
// Unpack a single-precision float. Note that this returns the magnitude
|
||||||
// of the single-precision float mantissa with the 1. if necessary,
|
// of the single-precision float mantissa with the 1. if necessary,
|
||||||
// aligned to bit 30.
|
// aligned to bit 30.
|
||||||
static inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr)
|
inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr)
|
||||||
{
|
{
|
||||||
s->sign = vfp_single_packed_sign(val) >> 16,
|
s->sign = vfp_single_packed_sign(val) >> 16,
|
||||||
s->exponent = vfp_single_packed_exponent(val);
|
s->exponent = vfp_single_packed_exponent(val);
|
||||||
|
@ -293,7 +293,7 @@ static inline void vfp_single_unpack(vfp_single* s, s32 val, u32* fpscr)
|
||||||
|
|
||||||
// Re-pack a single-precision float. This assumes that the float is
|
// Re-pack a single-precision float. This assumes that the float is
|
||||||
// already normalised such that the MSB is bit 30, _not_ bit 31.
|
// already normalised such that the MSB is bit 30, _not_ bit 31.
|
||||||
static inline s32 vfp_single_pack(const vfp_single* s)
|
inline s32 vfp_single_pack(const vfp_single* s)
|
||||||
{
|
{
|
||||||
u32 val = (s->sign << 16) +
|
u32 val = (s->sign << 16) +
|
||||||
(s->exponent << VFP_SINGLE_MANTISSA_BITS) +
|
(s->exponent << VFP_SINGLE_MANTISSA_BITS) +
|
||||||
|
@ -335,7 +335,7 @@ struct vfp_double {
|
||||||
#define vfp_double_packed_exponent(v) (((v) >> VFP_DOUBLE_MANTISSA_BITS) & ((1 << VFP_DOUBLE_EXPONENT_BITS) - 1))
|
#define vfp_double_packed_exponent(v) (((v) >> VFP_DOUBLE_MANTISSA_BITS) & ((1 << VFP_DOUBLE_EXPONENT_BITS) - 1))
|
||||||
#define vfp_double_packed_mantissa(v) ((v) & ((1ULL << VFP_DOUBLE_MANTISSA_BITS) - 1))
|
#define vfp_double_packed_mantissa(v) ((v) & ((1ULL << VFP_DOUBLE_MANTISSA_BITS) - 1))
|
||||||
|
|
||||||
static inline int vfp_double_type(const vfp_double* s)
|
inline int vfp_double_type(const vfp_double* s)
|
||||||
{
|
{
|
||||||
int type = VFP_NUMBER;
|
int type = VFP_NUMBER;
|
||||||
if (s->exponent == 2047) {
|
if (s->exponent == 2047) {
|
||||||
|
@ -357,7 +357,7 @@ static inline int vfp_double_type(const vfp_double* s)
|
||||||
// Unpack a double-precision float. Note that this returns the magnitude
|
// Unpack a double-precision float. Note that this returns the magnitude
|
||||||
// of the double-precision float mantissa with the 1. if necessary,
|
// of the double-precision float mantissa with the 1. if necessary,
|
||||||
// aligned to bit 62.
|
// aligned to bit 62.
|
||||||
static inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr)
|
inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr)
|
||||||
{
|
{
|
||||||
s->sign = vfp_double_packed_sign(val) >> 48;
|
s->sign = vfp_double_packed_sign(val) >> 48;
|
||||||
s->exponent = vfp_double_packed_exponent(val);
|
s->exponent = vfp_double_packed_exponent(val);
|
||||||
|
@ -379,7 +379,7 @@ static inline void vfp_double_unpack(vfp_double* s, s64 val, u32* fpscr)
|
||||||
|
|
||||||
// Re-pack a double-precision float. This assumes that the float is
|
// Re-pack a double-precision float. This assumes that the float is
|
||||||
// already normalised such that the MSB is bit 30, _not_ bit 31.
|
// already normalised such that the MSB is bit 30, _not_ bit 31.
|
||||||
static inline s64 vfp_double_pack(const vfp_double* s)
|
inline s64 vfp_double_pack(const vfp_double* s)
|
||||||
{
|
{
|
||||||
u64 val = ((u64)s->sign << 48) +
|
u64 val = ((u64)s->sign << 48) +
|
||||||
((u64)s->exponent << VFP_DOUBLE_MANTISSA_BITS) +
|
((u64)s->exponent << VFP_DOUBLE_MANTISSA_BITS) +
|
||||||
|
@ -415,7 +415,7 @@ struct op {
|
||||||
u32 flags;
|
u32 flags;
|
||||||
};
|
};
|
||||||
|
|
||||||
static inline u32 fls(u32 x)
|
inline u32 fls(u32 x)
|
||||||
{
|
{
|
||||||
int r = 32;
|
int r = 32;
|
||||||
|
|
||||||
|
|
Loading…
Reference in a new issue