mirror of
https://github.com/ptitSeb/box86.git
synced 2024-11-27 11:54:50 +08:00
[DYNAREC] Fix issue when code use lots of FPU code with a few jump (i.e. x87 with loops), on dynarec pass 1 (helps Hotline Miami, and probably others)
This commit is contained in:
ptitSeb
parent
79ada49fdc
commit
0a04ef44d4
@ -386,13 +386,16 @@ void emit_unlock(dynarec_arm_t* dyn, uintptr_t addr, int ninst)
|
||||
// x87 stuffs
|
||||
static void x87_reset(dynarec_arm_t* dyn, int ninst)
|
||||
{
|
||||
#if STEP > 1
|
||||
for (int i=0; i<8; ++i)
|
||||
dyn->x87cache[i] = -1;
|
||||
dyn->x87stack = 0;
|
||||
#endif
|
||||
}
|
||||
|
||||
void x87_stackcount(dynarec_arm_t* dyn, int ninst, int scratch)
|
||||
{
|
||||
#if STEP > 1
|
||||
if(!dyn->x87stack)
|
||||
return;
|
||||
MESSAGE(LOG_DUMP, "\tSynch x87 Stackcount (%d)\n", dyn->x87stack);
|
||||
@ -417,10 +420,12 @@ void x87_stackcount(dynarec_arm_t* dyn, int ninst, int scratch)
|
||||
// reset x87stack
|
||||
dyn->x87stack = 0;
|
||||
MESSAGE(LOG_DUMP, "\t------x87 Stackcount\n");
|
||||
#endif
|
||||
}
|
||||
|
||||
int x87_do_push(dynarec_arm_t* dyn, int ninst)
|
||||
{
|
||||
#if STEP > 1
|
||||
dyn->x87stack+=1;
|
||||
// move all regs in cache, and find a free one
|
||||
int ret = -1;
|
||||
@ -432,9 +437,13 @@ int x87_do_push(dynarec_arm_t* dyn, int ninst)
|
||||
ret=dyn->x87reg[i]=fpu_get_reg_double(dyn);
|
||||
}
|
||||
return ret;
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
void x87_do_push_empty(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
dyn->x87stack+=1;
|
||||
// move all regs in cache
|
||||
for(int i=0; i<8; ++i)
|
||||
@ -442,9 +451,11 @@ void x87_do_push_empty(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
++dyn->x87cache[i];
|
||||
if(s1)
|
||||
x87_stackcount(dyn, ninst, s1);
|
||||
#endif
|
||||
}
|
||||
void x87_do_pop(dynarec_arm_t* dyn, int ninst)
|
||||
{
|
||||
#if STEP > 1
|
||||
dyn->x87stack-=1;
|
||||
// move all regs in cache, poping ST0
|
||||
for(int i=0; i<8; ++i)
|
||||
@ -455,10 +466,12 @@ void x87_do_pop(dynarec_arm_t* dyn, int ninst)
|
||||
dyn->x87reg[i] = -1;
|
||||
}
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
||||
static void x87_purgecache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3)
|
||||
{
|
||||
#if STEP > 1
|
||||
int ret = 0;
|
||||
for (int i=0; i<8 && !ret; ++i)
|
||||
if(dyn->x87cache[i] != -1)
|
||||
@ -524,11 +537,13 @@ static void x87_purgecache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3
|
||||
dyn->x87cache[i] = -1;
|
||||
}
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifdef HAVE_TRACE
|
||||
static void x87_reflectcache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3)
|
||||
{
|
||||
#if STEP > 1
|
||||
x87_stackcount(dyn, ninst, s1);
|
||||
int ret = 0;
|
||||
for (int i=0; (i<8) && (!ret); ++i)
|
||||
@ -549,11 +564,13 @@ static void x87_reflectcache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int
|
||||
ADD_REG_LSL_IMM5(s3, s1, s3, 3); // fpu[(emu->top+i)&7] lsl 3 because fpu are double, so 8 bytes
|
||||
VSTR_64(dyn->x87reg[i], s3, 0); // save the value
|
||||
}
|
||||
#endif
|
||||
}
|
||||
#endif
|
||||
|
||||
int x87_get_cache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
{
|
||||
#if STEP > 1
|
||||
// search in cache first
|
||||
for (int i=0; i<8; ++i)
|
||||
if(dyn->x87cache[i]==st)
|
||||
@ -582,16 +599,24 @@ int x87_get_cache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
MESSAGE(LOG_DUMP, "\t-------x87 Cache for ST%d\n", st);
|
||||
|
||||
return ret;
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
|
||||
int x87_get_st(dynarec_arm_t* dyn, int ninst, int s1, int s2, int a)
|
||||
{
|
||||
#if STEP > 1
|
||||
return dyn->x87reg[x87_get_cache(dyn, ninst, s1, s2, a)];
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
|
||||
|
||||
void x87_refresh(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
{
|
||||
#if STEP > 1
|
||||
x87_stackcount(dyn, ninst, s1);
|
||||
int ret = -1;
|
||||
for (int i=0; (i<8) && (ret==-1); ++i)
|
||||
@ -613,10 +638,12 @@ void x87_refresh(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
ADD_REG_LSL_IMM5(s2, s1, s2, 3); // fpu[(emu->top+i)&7] lsl 3 because fpu are double, so 8 bytes
|
||||
VSTR_64(dyn->x87reg[ret], s2, 0); // save the value
|
||||
MESSAGE(LOG_DUMP, "\t--------x87 Cache for ST%d\n", st);
|
||||
#endif
|
||||
}
|
||||
|
||||
void x87_forget(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
{
|
||||
#if STEP > 1
|
||||
x87_stackcount(dyn, ninst, s1);
|
||||
int ret = -1;
|
||||
for (int i=0; (i<8) && (ret==-1); ++i)
|
||||
@ -642,10 +669,12 @@ void x87_forget(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
fpu_free_reg_double(dyn, dyn->x87reg[ret]);
|
||||
dyn->x87cache[ret] = -1;
|
||||
dyn->x87reg[ret] = -1;
|
||||
#endif
|
||||
}
|
||||
|
||||
void x87_reget_st(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
{
|
||||
#if STEP > 1
|
||||
// search in cache first
|
||||
for (int i=0; i<8; ++i)
|
||||
if(dyn->x87cache[i]==st) {
|
||||
@ -690,8 +719,7 @@ void x87_reget_st(dynarec_arm_t* dyn, int ninst, int s1, int s2, int st)
|
||||
ADD_REG_LSL_IMM5(s2, s1, s2, 3);
|
||||
VLDR_64(dyn->x87reg[ret], s2, 0);
|
||||
MESSAGE(LOG_DUMP, "\t-------x87 Cache for ST%d\n", st);
|
||||
|
||||
return;
|
||||
#endif
|
||||
}
|
||||
|
||||
static int round_map[] = {0, 2, 1, 3}; // map x86 -> arm round flag
|
||||
@ -719,12 +747,15 @@ void x87_restoreround(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
// MMX helpers
|
||||
static void mmx_reset(dynarec_arm_t* dyn, int ninst)
|
||||
{
|
||||
#if STEP > 1
|
||||
for (int i=0; i<8; ++i)
|
||||
dyn->mmxcache[i] = -1;
|
||||
#endif
|
||||
}
|
||||
// get neon register for a MMX reg, create the entry if needed
|
||||
int mmx_get_reg(dynarec_arm_t* dyn, int ninst, int s1, int a)
|
||||
{
|
||||
#if STEP > 1
|
||||
if(dyn->mmxcache[a]!=-1)
|
||||
return dyn->mmxcache[a];
|
||||
int ret = dyn->mmxcache[a] = fpu_get_reg_double(dyn);
|
||||
@ -732,18 +763,24 @@ int mmx_get_reg(dynarec_arm_t* dyn, int ninst, int s1, int a)
|
||||
ADD_REG_LSL_IMM5(s1, xEmu, s1, 0);
|
||||
VLD1_32(ret, s1);
|
||||
return ret;
|
||||
#endif
|
||||
}
|
||||
// get neon register for a MMX reg, but don't try to synch it if it needed to be created
|
||||
int mmx_get_reg_empty(dynarec_arm_t* dyn, int ninst, int s1, int a)
|
||||
{
|
||||
#if STEP > 1
|
||||
if(dyn->mmxcache[a]!=-1)
|
||||
return dyn->mmxcache[a];
|
||||
int ret = dyn->mmxcache[a] = fpu_get_reg_double(dyn);
|
||||
return ret;
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
// purge the MMX cache only(needs 3 scratch registers)
|
||||
static void mmx_purgecache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
int old = -1;
|
||||
for (int i=0; i<8; ++i)
|
||||
if(dyn->mmxcache[i]!=-1) {
|
||||
@ -765,10 +802,12 @@ static void mmx_purgecache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
if(old!=-1) {
|
||||
MESSAGE(LOG_DUMP, "\t------ Purge MMX Cache\n");
|
||||
}
|
||||
#endif
|
||||
}
|
||||
#ifdef HAVE_TRACE
|
||||
static void mmx_reflectcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
int old = -1;
|
||||
for (int i=0; i<8; ++i)
|
||||
if(dyn->mmxcache[i]!=-1) {
|
||||
@ -784,6 +823,7 @@ static void mmx_reflectcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
}
|
||||
VST1_32_W(dyn->mmxcache[i], s1);
|
||||
}
|
||||
#endif
|
||||
}
|
||||
#endif
|
||||
|
||||
@ -791,12 +831,15 @@ static void mmx_reflectcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
// SSE / SSE2 helpers
|
||||
static void sse_reset(dynarec_arm_t* dyn, int ninst)
|
||||
{
|
||||
#if STEP > 1
|
||||
for (int i=0; i<8; ++i)
|
||||
dyn->ssecache[i] = -1;
|
||||
#endif
|
||||
}
|
||||
// get neon register for a SSE reg, create the entry if needed
|
||||
int sse_get_reg(dynarec_arm_t* dyn, int ninst, int s1, int a)
|
||||
{
|
||||
#if STEP > 1
|
||||
if(dyn->ssecache[a]!=-1)
|
||||
return dyn->ssecache[a];
|
||||
int ret = dyn->ssecache[a] = fpu_get_reg_quad(dyn);
|
||||
@ -804,18 +847,26 @@ int sse_get_reg(dynarec_arm_t* dyn, int ninst, int s1, int a)
|
||||
ADD_REG_LSL_IMM5(s1, xEmu, s1, 0);
|
||||
VLD1Q_32(ret, s1);
|
||||
return ret;
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
// get neon register for a SSE reg, but don't try to synch it if it needed to be created
|
||||
int sse_get_reg_empty(dynarec_arm_t* dyn, int ninst, int s1, int a)
|
||||
{
|
||||
#if STEP > 1
|
||||
if(dyn->ssecache[a]!=-1)
|
||||
return dyn->ssecache[a];
|
||||
int ret = dyn->ssecache[a] = fpu_get_reg_quad(dyn);
|
||||
return ret;
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
// purge the SSE cache only(needs 3 scratch registers)
|
||||
static void sse_purgecache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
int old = -1;
|
||||
for (int i=0; i<8; ++i)
|
||||
if(dyn->ssecache[i]!=-1) {
|
||||
@ -837,10 +888,12 @@ static void sse_purgecache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
if(old!=-1) {
|
||||
MESSAGE(LOG_DUMP, "\t------ Purge SSE Cache\n");
|
||||
}
|
||||
#endif
|
||||
}
|
||||
#ifdef HAVE_TRACE
|
||||
static void sse_reflectcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
int old = -1;
|
||||
for (int i=0; i<8; ++i)
|
||||
if(dyn->ssecache[i]!=-1) {
|
||||
@ -856,11 +909,13 @@ static void sse_reflectcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
}
|
||||
VST1Q_32_W(dyn->ssecache[i], s1);
|
||||
}
|
||||
#endif
|
||||
}
|
||||
#endif
|
||||
|
||||
void fpu_pushcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
// only need to push 16-31...
|
||||
int n=0;
|
||||
for (int i=8; i<24; i++)
|
||||
@ -883,10 +938,12 @@ void fpu_pushcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
}
|
||||
}
|
||||
MESSAGE(LOG_DUMP, "\t------- Push FPU Cache (%d)\n", n);
|
||||
#endif
|
||||
}
|
||||
|
||||
void fpu_popcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
{
|
||||
#if STEP > 1
|
||||
// only need to push 16-31...
|
||||
int n=0;
|
||||
for (int i=8; i<24; i++)
|
||||
@ -909,6 +966,7 @@ void fpu_popcache(dynarec_arm_t* dyn, int ninst, int s1)
|
||||
ADD_IMM8(xSP, xSP, n*8);
|
||||
}
|
||||
MESSAGE(LOG_DUMP, "\t------- Pop FPU Cache (%d)\n", n);
|
||||
#endif
|
||||
}
|
||||
|
||||
void fpu_purgecache(dynarec_arm_t* dyn, int ninst, int s1, int s2, int s3)
|
||||
@ -941,11 +999,15 @@ void fpu_reset(dynarec_arm_t* dyn, int ninst)
|
||||
// get the single reg that from the double "reg" (so Dx[idx])
|
||||
int fpu_get_single_reg(dynarec_arm_t* dyn, int ninst, int reg, int idx)
|
||||
{
|
||||
#if STEP > 1
|
||||
if(reg<16)
|
||||
return reg*2+idx;
|
||||
int a = fpu_get_scratch_double(dyn);
|
||||
VMOV_64(a, reg);
|
||||
return a*2+idx;
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
// put back (if needed) the single reg in place
|
||||
void fpu_putback_single_reg(dynarec_arm_t* dyn, int ninst, int reg, int idx, int s)
|
||||
|
||||
@ -48,6 +48,7 @@ void arm_pass(dynarec_arm_t* dyn, uintptr_t addr)
|
||||
// ok, go now
|
||||
INIT;
|
||||
while(ok) {
|
||||
if(dyn->insts && (ninst>dyn->size)) {dynarec_log(LOG_NONE, "Warning, too many inst treated (%d / %d)\n",ninst, dyn->size);}
|
||||
ip = addr;
|
||||
NEW_INST;
|
||||
fpu_reset_scratch(dyn);
|
||||
@ -82,7 +83,7 @@ void arm_pass(dynarec_arm_t* dyn, uintptr_t addr)
|
||||
if(!ok && !need_epilog && !dyn->insts) { // check if need to continue
|
||||
uintptr_t next = get_closest_next(dyn, addr);
|
||||
if(next && ((next-addr)<15) && is_nops(dyn, addr, next-addr)) {
|
||||
dynarec_log(LOG_DEBUG, "Extend block, %p -> %p\n", (void*)addr, (void*)next);
|
||||
dynarec_log(LOG_DEBUG, "Extend block %p, %p -> %p (ninst=%d)\n", dyn, (void*)addr, (void*)next, ninst);
|
||||
ok = 1;
|
||||
}
|
||||
}
|
||||
|
||||
@ -11,4 +11,4 @@
|
||||
dyn->insts[ninst].x86.addr = ip; \
|
||||
if(ninst) dyn->insts[ninst-1].x86.size = dyn->insts[ninst].x86.addr - dyn->insts[ninst-1].x86.addr;
|
||||
#define INST_EPILOG
|
||||
#define INST_NAME(name)
|
||||
#define INST_NAME(name)
|
||||
|
||||
Loading…
Reference in New Issue
Block a user