adjust CLOCK_CYCLES and related macros to handle fpu/iu concurrency

add CLOCK_CYCLES_FPU, which does exactly what CLOCK_CYCLES already did.

add CONCURRENCY_CYCLES, which sets fpu_cycles, which is the number of
available concurrent execution cycles that the integer unit can do
"free" work in while the fpu is executing.

adjust CLOCK_CYCLES so that if there are fpu_cycles, the cycle count is
subtracted from fpu_cycles instead of cycles, emulating the behavior of
these concurrent cycles being "free" as on real hardware.

src/cpu/386.c

@@ -81,7 +81,20 @@ x386_log(const char *fmt, ...)
 
 #define OP_TABLE(name) ops_ ## name
 
-#define CLOCK_CYCLES(c) cycles -= (c)
+#define CLOCK_CYCLES(c) \
+	{\
+		if (fpu_cycles > 0) {\
+			fpu_cycles -= (c);\
+			if (fpu_cycles < 0) {\
+				cycles += fpu_cycles;\
+			}\
+		} else {\
+			cycles -= (c);\
+		}\
+	}
+
+#define CLOCK_CYCLES_FPU(c) cycles -= (c)
+#define CONCURRENCY_CYCLES(c) fpu_cycles = (c)
 #define CLOCK_CYCLES_ALWAYS(c) cycles -= (c)
 
 #include "x86_ops.h"

src/cpu/386_dynarec.c

@@ -267,9 +267,22 @@ static void prefetch_flush()
 
 
 #define OP_TABLE(name) ops_ ## name
-#define CLOCK_CYCLES(c) cycles -= (c)
+#define CLOCK_CYCLES(c) \
+	{\
+		if (fpu_cycles > 0) {\
+			fpu_cycles -= (c);\
+			if (fpu_cycles < 0) {\
+				cycles += fpu_cycles;\
+			}\
+		} else {\
+			cycles -= (c);\
+		}\
+	}
+#define CLOCK_CYCLES_FPU(c) cycles -= (c)
+#define CONCURRENCY_CYCLES(c) fpu_cycles = (c)
 #define CLOCK_CYCLES_ALWAYS(c) cycles -= (c)
 
+
 #include "386_ops.h"
 
 

src/cpu/386_dynarec_ops.c

@@ -65,6 +65,8 @@ static __inline void fetch_ea_16_long(uint32_t rmdat)
 #define OP_TABLE(name) dynarec_ops_ ## name
 
 #define CLOCK_CYCLES(c)
+#define CLOCK_CYCLES_FPU(c)
+#define CONCURRENCY_CYCLES(c) fpu_cycles = (c)
 #define CLOCK_CYCLES_ALWAYS(c) cycles -= (c)
 
 #include "386_ops.h"

src/cpu/808x.c

@@ -89,11 +89,32 @@ static int refresh = 0, cycdiff;
 		wait(val, 0);	\
 	}
 
-#define CLOCK_CYCLES(val)		\
+#define CLOCK_CYCLES_ALWAYS(val)		\
 	{			\
 		wait(val, 0);	\
 	}
 
+#define CLOCK_CYCLES_FPU(val)		\
+	{			\
+		wait(val, 0);	\
+	}
+
+
+#define CLOCK_CYCLES(val) \
+	{	\
+		if (fpu_cycles > 0) {	\
+			fpu_cycles -= (val);	\
+			if (fpu_cycles < 0) {	\
+				wait(val, 0);	\
+			}	\
+		} else {	\
+			wait(val, 0);	\
+		}	\
+	}
+
+#define CONCURRENCY_CYCLES(c) fpu_cycles = (c)
+
+
 typedef	int (*OpFn)(uint32_t fetchdat);