[picodrive.git] / cpu / drc / emit_arm.c

// Basic macros to emit ARM instructions and some utils

// (c) Copyright 2008-2009, Grazvydas "notaz" Ignotas
// Free for non-commercial use.

#define CONTEXT_REG 7

// XXX: tcache_ptr type for SVP and SH2 compilers differs..
#define EMIT_PTR(ptr, x) \
	do { \
		*(u32 *)ptr = x; \
		ptr = (void *)((u8 *)ptr + sizeof(u32)); \
		COUNT_OP; \
	} while (0)

#define EMIT(x) EMIT_PTR(tcache_ptr, x)

#define A_R4M  (1 << 4)
#define A_R5M  (1 << 5)
#define A_R6M  (1 << 6)
#define A_R7M  (1 << 7)
#define A_R8M  (1 << 8)
#define A_R9M  (1 << 9)
#define A_R10M (1 << 10)
#define A_R11M (1 << 11)
#define A_R14M (1 << 14)

#define A_COND_AL 0xe
#define A_COND_EQ 0x0
#define A_COND_NE 0x1
#define A_COND_HS 0x2
#define A_COND_LO 0x3
#define A_COND_MI 0x4
#define A_COND_PL 0x5
#define A_COND_VS 0x6
#define A_COND_VC 0x7
#define A_COND_HI 0x8
#define A_COND_LS 0x9
#define A_COND_GE 0xa
#define A_COND_LT 0xb
#define A_COND_GT 0xc
#define A_COND_LE 0xd
#define A_COND_CS A_COND_HS
#define A_COND_CC A_COND_LO

/* unified conditions */
#define DCOND_EQ A_COND_EQ
#define DCOND_NE A_COND_NE
#define DCOND_MI A_COND_MI
#define DCOND_PL A_COND_PL
#define DCOND_HI A_COND_HI
#define DCOND_HS A_COND_HS
#define DCOND_LO A_COND_LO
#define DCOND_GE A_COND_GE
#define DCOND_GT A_COND_GT
#define DCOND_LT A_COND_LT
#define DCOND_LS A_COND_LS
#define DCOND_LE A_COND_LE
#define DCOND_VS A_COND_VS
#define DCOND_VC A_COND_VC

/* addressing mode 1 */
#define A_AM1_LSL 0
#define A_AM1_LSR 1
#define A_AM1_ASR 2
#define A_AM1_ROR 3

#define A_AM1_IMM(ror2,imm8)                  (((ror2)<<8) | (imm8) | 0x02000000)
#define A_AM1_REG_XIMM(shift_imm,shift_op,rm) (((shift_imm)<<7) | ((shift_op)<<5) | (rm))
#define A_AM1_REG_XREG(rs,shift_op,rm)        (((rs)<<8) | ((shift_op)<<5) | 0x10 | (rm))

/* data processing op */
#define A_OP_AND 0x0
#define A_OP_EOR 0x1
#define A_OP_SUB 0x2
#define A_OP_RSB 0x3
#define A_OP_ADD 0x4
#define A_OP_ADC 0x5
#define A_OP_SBC 0x6
#define A_OP_RSC 0x7
#define A_OP_TST 0x8
#define A_OP_TEQ 0x9
#define A_OP_CMP 0xa
#define A_OP_ORR 0xc
#define A_OP_MOV 0xd
#define A_OP_BIC 0xe
#define A_OP_MVN 0xf

#define EOP_C_DOP_X(cond,op,s,rn,rd,shifter_op) \
	EMIT(((cond)<<28) | ((op)<< 21) | ((s)<<20) | ((rn)<<16) | ((rd)<<12) | (shifter_op))

#define EOP_C_DOP_IMM(     cond,op,s,rn,rd,ror2,imm8)             EOP_C_DOP_X(cond,op,s,rn,rd,A_AM1_IMM(ror2,imm8))
#define EOP_C_DOP_REG_XIMM(cond,op,s,rn,rd,shift_imm,shift_op,rm) EOP_C_DOP_X(cond,op,s,rn,rd,A_AM1_REG_XIMM(shift_imm,shift_op,rm))
#define EOP_C_DOP_REG_XREG(cond,op,s,rn,rd,rs,       shift_op,rm) EOP_C_DOP_X(cond,op,s,rn,rd,A_AM1_REG_XREG(rs,       shift_op,rm))

#define EOP_MOV_IMM(rd,   ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_MOV,0, 0,rd,ror2,imm8)
#define EOP_MVN_IMM(rd,   ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_MVN,0, 0,rd,ror2,imm8)
#define EOP_ORR_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_ORR,0,rn,rd,ror2,imm8)
#define EOP_EOR_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_EOR,0,rn,rd,ror2,imm8)
#define EOP_ADD_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_ADD,0,rn,rd,ror2,imm8)
#define EOP_BIC_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_BIC,0,rn,rd,ror2,imm8)
#define EOP_AND_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_AND,0,rn,rd,ror2,imm8)
#define EOP_SUB_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_SUB,0,rn,rd,ror2,imm8)
#define EOP_TST_IMM(   rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_TST,1,rn, 0,ror2,imm8)
#define EOP_CMP_IMM(   rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_CMP,1,rn, 0,ror2,imm8)
#define EOP_RSB_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_RSB,0,rn,rd,ror2,imm8)

#define EOP_MOV_IMM_C(cond,rd,   ror2,imm8) EOP_C_DOP_IMM(cond,A_OP_MOV,0, 0,rd,ror2,imm8)
#define EOP_ORR_IMM_C(cond,rd,rn,ror2,imm8) EOP_C_DOP_IMM(cond,A_OP_ORR,0,rn,rd,ror2,imm8)
#define EOP_RSB_IMM_C(cond,rd,rn,ror2,imm8) EOP_C_DOP_IMM(cond,A_OP_RSB,0,rn,rd,ror2,imm8)

#define EOP_MOV_REG(cond,s,rd,   rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_MOV,s, 0,rd,shift_imm,shift_op,rm)
#define EOP_MVN_REG(cond,s,rd,   rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_MVN,s, 0,rd,shift_imm,shift_op,rm)
#define EOP_ORR_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_ORR,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_ADD_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_ADD,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_ADC_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_ADC,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_SUB_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_SUB,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_SBC_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_SBC,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_AND_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_AND,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_EOR_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_EOR,s,rn,rd,shift_imm,shift_op,rm)
#define EOP_CMP_REG(cond,     rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_CMP,1,rn, 0,shift_imm,shift_op,rm)
#define EOP_TST_REG(cond,     rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_TST,1,rn, 0,shift_imm,shift_op,rm)
#define EOP_TEQ_REG(cond,     rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_TEQ,1,rn, 0,shift_imm,shift_op,rm)

#define EOP_MOV_REG2(s,rd,   rm,shift_op,rs) EOP_C_DOP_REG_XREG(A_COND_AL,A_OP_MOV,s, 0,rd,rs,shift_op,rm)
#define EOP_ADD_REG2(s,rd,rn,rm,shift_op,rs) EOP_C_DOP_REG_XREG(A_COND_AL,A_OP_ADD,s,rn,rd,rs,shift_op,rm)
#define EOP_SUB_REG2(s,rd,rn,rm,shift_op,rs) EOP_C_DOP_REG_XREG(A_COND_AL,A_OP_SUB,s,rn,rd,rs,shift_op,rm)

#define EOP_MOV_REG_SIMPLE(rd,rm)           EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_LSL,0)
#define EOP_MOV_REG_LSL(rd,   rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_LSL,shift_imm)
#define EOP_MOV_REG_LSR(rd,   rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_LSR,shift_imm)
#define EOP_MOV_REG_ASR(rd,   rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_ASR,shift_imm)
#define EOP_MOV_REG_ROR(rd,   rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_ROR,shift_imm)

#define EOP_ORR_REG_SIMPLE(rd,rm)           EOP_ORR_REG(A_COND_AL,0,rd,rd,rm,A_AM1_LSL,0)
#define EOP_ORR_REG_LSL(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSL,shift_imm)
#define EOP_ORR_REG_LSR(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSR,shift_imm)
#define EOP_ORR_REG_ASR(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_ASR,shift_imm)
#define EOP_ORR_REG_ROR(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_ROR,shift_imm)

#define EOP_ADD_REG_SIMPLE(rd,rm)           EOP_ADD_REG(A_COND_AL,0,rd,rd,rm,A_AM1_LSL,0)
#define EOP_ADD_REG_LSL(rd,rn,rm,shift_imm) EOP_ADD_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSL,shift_imm)
#define EOP_ADD_REG_LSR(rd,rn,rm,shift_imm) EOP_ADD_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSR,shift_imm)

#define EOP_TST_REG_SIMPLE(rn,rm)           EOP_TST_REG(A_COND_AL,  rn,   0,A_AM1_LSL,rm)

#define EOP_MOV_REG2_LSL(rd,   rm,rs)       EOP_MOV_REG2(0,rd,   rm,A_AM1_LSL,rs)
#define EOP_MOV_REG2_ROR(rd,   rm,rs)       EOP_MOV_REG2(0,rd,   rm,A_AM1_ROR,rs)
#define EOP_ADD_REG2_LSL(rd,rn,rm,rs)       EOP_ADD_REG2(0,rd,rn,rm,A_AM1_LSL,rs)
#define EOP_SUB_REG2_LSL(rd,rn,rm,rs)       EOP_SUB_REG2(0,rd,rn,rm,A_AM1_LSL,rs)

/* addressing mode 2 */
#define EOP_C_AM2_IMM(cond,u,b,l,rn,rd,offset_12) \
	EMIT(((cond)<<28) | 0x05000000 | ((u)<<23) | ((b)<<22) | ((l)<<20) | ((rn)<<16) | ((rd)<<12) | (offset_12))

/* addressing mode 3 */
#define EOP_C_AM3(cond,u,r,l,rn,rd,s,h,immed_reg) \
	EMIT(((cond)<<28) | 0x01000090 | ((u)<<23) | ((r)<<22) | ((l)<<20) | ((rn)<<16) | ((rd)<<12) | \
			((s)<<6) | ((h)<<5) | (immed_reg))

#define EOP_C_AM3_IMM(cond,u,l,rn,rd,s,h,offset_8) EOP_C_AM3(cond,u,1,l,rn,rd,s,h,(((offset_8)&0xf0)<<4)|((offset_8)&0xf))

#define EOP_C_AM3_REG(cond,u,l,rn,rd,s,h,rm)       EOP_C_AM3(cond,u,0,l,rn,rd,s,h,rm)

/* ldr and str */
#define EOP_LDR_IMM(   rd,rn,offset_12) EOP_C_AM2_IMM(A_COND_AL,1,0,1,rn,rd,offset_12)
#define EOP_LDR_NEGIMM(rd,rn,offset_12) EOP_C_AM2_IMM(A_COND_AL,0,0,1,rn,rd,offset_12)
#define EOP_LDR_SIMPLE(rd,rn)           EOP_C_AM2_IMM(A_COND_AL,1,0,1,rn,rd,0)
#define EOP_STR_IMM(   rd,rn,offset_12) EOP_C_AM2_IMM(A_COND_AL,1,0,0,rn,rd,offset_12)
#define EOP_STR_SIMPLE(rd,rn)           EOP_C_AM2_IMM(A_COND_AL,1,0,0,rn,rd,0)

#define EOP_LDRH_IMM(   rd,rn,offset_8)  EOP_C_AM3_IMM(A_COND_AL,1,1,rn,rd,0,1,offset_8)
#define EOP_LDRH_SIMPLE(rd,rn)           EOP_C_AM3_IMM(A_COND_AL,1,1,rn,rd,0,1,0)
#define EOP_LDRH_REG(   rd,rn,rm)        EOP_C_AM3_REG(A_COND_AL,1,1,rn,rd,0,1,rm)
#define EOP_STRH_IMM(   rd,rn,offset_8)  EOP_C_AM3_IMM(A_COND_AL,1,0,rn,rd,0,1,offset_8)
#define EOP_STRH_SIMPLE(rd,rn)           EOP_C_AM3_IMM(A_COND_AL,1,0,rn,rd,0,1,0)
#define EOP_STRH_REG(   rd,rn,rm)        EOP_C_AM3_REG(A_COND_AL,1,0,rn,rd,0,1,rm)

/* ldm and stm */
#define EOP_XXM(cond,p,u,s,w,l,rn,list) \
	EMIT(((cond)<<28) | (1<<27) | ((p)<<24) | ((u)<<23) | ((s)<<22) | ((w)<<21) | ((l)<<20) | ((rn)<<16) | (list))

#define EOP_STMFD_ST(list) EOP_XXM(A_COND_AL,1,0,0,1,0,13,list)
#define EOP_LDMFD_ST(list) EOP_XXM(A_COND_AL,0,1,0,1,1,13,list)

/* branches */
#define EOP_C_BX(cond,rm) \
	EMIT(((cond)<<28) | 0x012fff10 | (rm))

#define EOP_BX(rm) EOP_C_BX(A_COND_AL,rm)

#define EOP_C_B(cond,l,signed_immed_24) \
	EMIT(((cond)<<28) | 0x0a000000 | ((l)<<24) | (signed_immed_24))

#define EOP_B( signed_immed_24) EOP_C_B(A_COND_AL,0,signed_immed_24)
#define EOP_BL(signed_immed_24) EOP_C_B(A_COND_AL,1,signed_immed_24)

/* misc */
#define EOP_C_MUL(cond,s,rd,rs,rm) \
	EMIT(((cond)<<28) | ((s)<<20) | ((rd)<<16) | ((rs)<<8) | 0x90 | (rm))

#define EOP_C_UMULL(cond,s,rdhi,rdlo,rs,rm) \
	EMIT(((cond)<<28) | 0x00800000 | ((s)<<20) | ((rdhi)<<16) | ((rdlo)<<12) | ((rs)<<8) | 0x90 | (rm))

#define EOP_C_SMULL(cond,s,rdhi,rdlo,rs,rm) \
	EMIT(((cond)<<28) | 0x00c00000 | ((s)<<20) | ((rdhi)<<16) | ((rdlo)<<12) | ((rs)<<8) | 0x90 | (rm))

#define EOP_MUL(rd,rm,rs) EOP_C_MUL(A_COND_AL,0,rd,rs,rm) // note: rd != rm

#define EOP_C_MRS(cond,rd) \
	EMIT(((cond)<<28) | 0x010f0000 | ((rd)<<12))

#define EOP_C_MSR_IMM(cond,ror2,imm) \
	EMIT(((cond)<<28) | 0x0328f000 | ((ror2)<<8) | (imm)) // cpsr_f

#define EOP_C_MSR_REG(cond,rm) \
	EMIT(((cond)<<28) | 0x0128f000 | (rm)) // cpsr_f

#define EOP_MRS(rd)           EOP_C_MRS(A_COND_AL,rd)
#define EOP_MSR_IMM(ror2,imm) EOP_C_MSR_IMM(A_COND_AL,ror2,imm)
#define EOP_MSR_REG(rm)       EOP_C_MSR_REG(A_COND_AL,rm)


static void emith_op_imm2(int cond, int s, int op, int rd, int rn, unsigned int imm)
{
	int ror2;
	u32 v;

	if (op == A_OP_MOV)
		rn = 0;
	else if (imm == 0)
		return;

	for (v = imm, ror2 = 0; v != 0 || op == A_OP_MOV; v >>= 8, ror2 -= 8/2) {
		/* shift down to get 'best' rot2 */
		for (; v && !(v & 3); v >>= 2)
			ror2--;

		EOP_C_DOP_IMM(cond, op, s, rn, rd, ror2 & 0x0f, v & 0xff);

		if (op == A_OP_MOV) {
			op = A_OP_ORR;
			rn = rd;
		}
	}
}

#define emith_op_imm(cond, s, op, r, imm) \
	emith_op_imm2(cond, s, op, r, r, imm)

// test op
#define emith_top_imm(cond, op, r, imm) { \
	u32 ror2, v; \
	for (ror2 = 0, v = imm; v && !(v & 3); v >>= 2) \
		ror2--; \
	EOP_C_DOP_IMM(cond, op, 1, r, 0, ror2 & 0x0f, v & 0xff); \
}

#define is_offset_24(val) \
	((val) >= (int)0xff000000 && (val) <= 0x00ffffff)

static int emith_xbranch(int cond, void *target, int is_call)
{
	int val = (u32 *)target - (u32 *)tcache_ptr - 2;
	int direct = is_offset_24(val);
	u32 *start_ptr = (u32 *)tcache_ptr;

	if (direct)
	{
		EOP_C_B(cond,is_call,val & 0xffffff);		// b, bl target
	}
	else
	{
#ifdef __EPOC32__
//		elprintf(EL_SVP, "emitting indirect jmp %08x->%08x", tcache_ptr, target);
		if (is_call)
			EOP_ADD_IMM(14,15,0,8);			// add lr,pc,#8
		EOP_C_AM2_IMM(cond,1,0,1,15,15,0);		// ldrcc pc,[pc]
		EOP_MOV_REG_SIMPLE(15,15);			// mov pc, pc
		EMIT((u32)target);
#else
		// should never happen
		elprintf(EL_STATUS|EL_SVP|EL_ANOMALY, "indirect jmp %08x->%08x", target, tcache_ptr);
		exit(1);
#endif
	}

	return (u32 *)tcache_ptr - start_ptr;
}


// fake "simple" or "short" jump - using cond insns instead
#define EMITH_SJMP_START(cond) \
	(void)(cond)

#define EMITH_SJMP_END(cond) \
	(void)(cond)

#define emith_move_r_r(d, s) \
	EOP_MOV_REG_SIMPLE(d, s)

#define emith_mvn_r_r(d, s) \
	EOP_MVN_REG(A_COND_AL,0,d,s,A_AM1_LSL,0)

#define emith_or_r_r_r_lsl(d, s1, s2, lslimm) \
	EOP_ORR_REG(A_COND_AL,0,d,s1,s2,A_AM1_LSL,lslimm)

#define emith_eor_r_r_r_lsl(d, s1, s2, lslimm) \
	EOP_EOR_REG(A_COND_AL,0,d,s1,s2,A_AM1_LSL,lslimm)

#define emith_or_r_r_r(d, s1, s2) \
	emith_or_r_r_r_lsl(d, s1, s2, 0)

#define emith_eor_r_r_r(d, s1, s2) \
	emith_eor_r_r_r_lsl(d, s1, s2, 0)

#define emith_add_r_r(d, s) \
	EOP_ADD_REG(A_COND_AL,0,d,d,s,A_AM1_LSL,0)

#define emith_sub_r_r(d, s) \
	EOP_SUB_REG(A_COND_AL,0,d,d,s,A_AM1_LSL,0)

#define emith_and_r_r(d, s) \
	EOP_AND_REG(A_COND_AL,0,d,d,s,A_AM1_LSL,0)

#define emith_or_r_r(d, s) \
	emith_or_r_r_r(d, d, s)

#define emith_eor_r_r(d, s) \
	emith_eor_r_r_r(d, d, s)

#define emith_tst_r_r(d, s) \
	EOP_TST_REG(A_COND_AL,d,s,A_AM1_LSL,0)

#define emith_teq_r_r(d, s) \
	EOP_TEQ_REG(A_COND_AL,d,s,A_AM1_LSL,0)

#define emith_cmp_r_r(d, s) \
	EOP_CMP_REG(A_COND_AL,d,s,A_AM1_LSL,0)

#define emith_addf_r_r(d, s) \
	EOP_ADD_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)

#define emith_subf_r_r(d, s) \
	EOP_SUB_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)

#define emith_adcf_r_r(d, s) \
	EOP_ADC_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)

#define emith_sbcf_r_r(d, s) \
	EOP_SBC_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)

#define emith_move_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_MOV, r, imm)

#define emith_add_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_ADD, r, imm)

#define emith_sub_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_SUB, r, imm)

#define emith_bic_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_BIC, r, imm)

#define emith_and_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_AND, r, imm)

#define emith_or_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_ORR, r, imm)

#define emith_eor_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 0, A_OP_EOR, r, imm)

// note: only use 8bit imm for these
#define emith_tst_r_imm(r, imm) \
	emith_top_imm(A_COND_AL, A_OP_TST, r, imm)

#define emith_cmp_r_imm(r, imm) \
	emith_top_imm(A_COND_AL, A_OP_CMP, r, imm)

#define emith_subf_r_imm(r, imm) \
	emith_op_imm(A_COND_AL, 1, A_OP_SUB, r, imm)

#define emith_add_r_imm_c(cond, r, imm) \
	emith_op_imm(cond, 0, A_OP_ADD, r, imm)

#define emith_sub_r_imm_c(cond, r, imm) \
	emith_op_imm(cond, 0, A_OP_SUB, r, imm)

#define emith_or_r_imm_c(cond, r, imm) \
	emith_op_imm(cond, 0, A_OP_ORR, r, imm)

#define emith_bic_r_imm_c(cond, r, imm) \
	emith_op_imm(cond, 0, A_OP_BIC, r, imm)

#define emith_move_r_imm_s8(r, imm) { \
	if ((imm) & 0x80) \
		EOP_MVN_IMM(r, 0, ((imm) ^ 0xff)); \
	else \
		EOP_MOV_IMM(r, 0, imm); \
}

#define emith_and_r_r_imm(d, s, imm) \
	emith_op_imm2(A_COND_AL, 0, A_OP_AND, d, s, imm)

#define emith_neg_r_r(d, s) \
	EOP_RSB_IMM(d, s, 0, 0)

#define emith_lsl(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_LSL,cnt)

#define emith_lsr(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_LSR,cnt)

#define emith_ror(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_ROR,cnt)

#define emith_rol(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_ROR,32-(cnt)); \

#define emith_lslf(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_LSL,cnt)

#define emith_lsrf(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_LSR,cnt)

#define emith_asrf(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_ASR,cnt)

// note: only C flag updated correctly
#define emith_rolf(d, s, cnt) { \
	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_ROR,32-(cnt)); \
	/* we don't have ROL so we shift to get the right carry */ \
	EOP_TST_REG(A_COND_AL,d,d,A_AM1_LSR,1); \
}

#define emith_rorf(d, s, cnt) \
	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_ROR,cnt)

#define emith_rolcf(d) \
	emith_adcf_r_r(d, d)

#define emith_rorcf(d) \
	EOP_MOV_REG(A_COND_AL,1,d,d,A_AM1_ROR,0) /* ROR #0 -> RRX */

#define emith_negcf_r_r(d, s) \
	EOP_C_DOP_IMM(A_COND_AL,A_OP_RSC,1,s,d,0,0)

#define emith_mul(d, s1, s2) { \
	if ((d) != (s1)) /* rd != rm limitation */ \
		EOP_MUL(d, s1, s2); \
	else \
		EOP_MUL(d, s2, s1); \
}

#define emith_mul_u64(dlo, dhi, s1, s2) \
	EOP_C_UMULL(A_COND_AL,0,dhi,dlo,s1,s2)

#define emith_mul_s64(dlo, dhi, s1, s2) \
	EOP_C_SMULL(A_COND_AL,0,dhi,dlo,s1,s2)

// misc
#define emith_ctx_read(r, offs) \
	EOP_LDR_IMM(r, CONTEXT_REG, offs)

#define emith_ctx_write(r, offs) \
	EOP_STR_IMM(r, CONTEXT_REG, offs)

#define emith_clear_msb(d, s, count) { \
	u32 t; \
	if ((count) <= 8) { \
		t = (count) - 8; \
		t = (0xff << t) & 0xff; \
		EOP_BIC_IMM(d,s,8/2,t); \
	} else if ((count) >= 24) { \
		t = (count) - 24; \
		t = 0xff >> t; \
		EOP_AND_IMM(d,s,0,t); \
	} else { \
		EOP_MOV_REG_LSL(d,s,count); \
		EOP_MOV_REG_LSR(d,d,count); \
	} \
}

#define emith_sext(d, s, bits) { \
	EOP_MOV_REG_LSL(d,s,32 - (bits)); \
	EOP_MOV_REG_ASR(d,d,32 - (bits)); \
}

// put bit0 of r0 to carry
#define emith_set_carry(r0) \
	EOP_TST_REG(A_COND_AL,r0,r0,A_AM1_LSR,1) /* shift out to carry */ \

// put bit0 of r0 to carry (for subtraction, inverted on ARM)
#define emith_set_carry_sub(r0) { \
	int t = rcache_get_tmp(); \
	EOP_EOR_IMM(t,r0,0,1); /* invert */ \
	EOP_MOV_REG(A_COND_AL,1,t,t,A_AM1_LSR,1); /* shift out to carry */ \
	rcache_free_tmp(t); \
}

#define host_arg2reg(rd, arg) \
	rd = arg

// upto 4 args
#define emith_pass_arg_r(arg, reg) \
	EOP_MOV_REG_SIMPLE(arg, reg)

#define emith_pass_arg_imm(arg, imm) \
	emith_move_r_imm(arg, imm)

#define emith_call_cond(cond, target) \
	emith_xbranch(cond, target, 1)

#define emith_jump_cond(cond, target) \
	emith_xbranch(cond, target, 0)

#define emith_call(target) \
	emith_call_cond(A_COND_AL, target)

#define emith_jump(target) \
	emith_jump_cond(A_COND_AL, target)

/* SH2 drc specific */
#define emith_sh2_test_t() { \
	int r = rcache_get_reg(SHR_SR, RC_GR_READ); \
	EOP_TST_IMM(r, 0, 1); \
}

#define emith_sh2_dtbf_loop() { \
	int cr, rn;                                                          \
	int tmp_ = rcache_get_tmp();                                         \
	cr = rcache_get_reg(SHR_SR, RC_GR_RMW);                              \
	rn = rcache_get_reg((op >> 8) & 0x0f, RC_GR_RMW);                    \
	emith_sub_r_imm(rn, 1);                /* sub rn, #1 */              \
	emith_bic_r_imm(cr, 1);                /* bic cr, #1 */              \
	emith_sub_r_imm(cr, (cycles+1) << 12); /* sub cr, #(cycles+1)<<12 */ \
	cycles = 0;                                                          \
	emith_asrf(tmp_, cr, 2+12);            /* movs tmp_, cr, asr #2+12 */\
	EOP_MOV_IMM_C(A_COND_MI,tmp_,0,0);     /* movmi tmp_, #0 */          \
	emith_lsl(cr, cr, 20);                 /* mov cr, cr, lsl #20 */     \
	emith_lsr(cr, cr, 20);                 /* mov cr, cr, lsr #20 */     \
	emith_subf_r_r(rn, tmp_);              /* subs rn, tmp_ */           \
	EOP_RSB_IMM_C(A_COND_LS,tmp_,rn,0,0);  /* rsbls tmp_, rn, #0 */      \
	EOP_ORR_REG(A_COND_LS,0,cr,cr,tmp_,A_AM1_LSL,12+2); /* orrls cr,tmp_,lsl #12+2 */\
	EOP_ORR_IMM_C(A_COND_LS,cr,cr,0,1);    /* orrls cr, #1 */            \
	EOP_MOV_IMM_C(A_COND_LS,rn,0,0);       /* movls rn, #0 */            \
	rcache_free_tmp(tmp_);                                               \
}

#define emith_write_sr(srcr) { \
	int srr = rcache_get_reg(SHR_SR, RC_GR_RMW); \
	emith_lsr(srr, srr, 12); \
	emith_or_r_r_r_lsl(srr, srr, srcr, 20); \
	emith_ror(srr, srr, 20); \
}

#define emith_carry_to_t(srr, is_sub) { \
	if (is_sub) { /* has inverted C on ARM */ \
		emith_or_r_imm_c(A_COND_CC, srr, 1); \
		emith_bic_r_imm_c(A_COND_CS, srr, 1); \
	} else { \
		emith_or_r_imm_c(A_COND_CS, srr, 1); \
		emith_bic_r_imm_c(A_COND_CC, srr, 1); \
	} \
}
Commit	Line	Data
	1	// Basic macros to emit ARM instructions and some utils
	2
	3	// (c) Copyright 2008-2009, Grazvydas "notaz" Ignotas
	4	// Free for non-commercial use.
	5
	6	#define CONTEXT_REG 7
	7
	8	// XXX: tcache_ptr type for SVP and SH2 compilers differs..
	9	#define EMIT_PTR(ptr, x) \
	10	do { \
	11	(u32 )ptr = x; \
	12	ptr = (void )((u8 )ptr + sizeof(u32)); \
	13	COUNT_OP; \
	14	} while (0)
	15
	16	#define EMIT(x) EMIT_PTR(tcache_ptr, x)
	17
	18	#define A_R4M (1 << 4)
	19	#define A_R5M (1 << 5)
	20	#define A_R6M (1 << 6)
	21	#define A_R7M (1 << 7)
	22	#define A_R8M (1 << 8)
	23	#define A_R9M (1 << 9)
	24	#define A_R10M (1 << 10)
	25	#define A_R11M (1 << 11)
	26	#define A_R14M (1 << 14)
	27
	28	#define A_COND_AL 0xe
	29	#define A_COND_EQ 0x0
	30	#define A_COND_NE 0x1
	31	#define A_COND_HS 0x2
	32	#define A_COND_LO 0x3
	33	#define A_COND_MI 0x4
	34	#define A_COND_PL 0x5
	35	#define A_COND_VS 0x6
	36	#define A_COND_VC 0x7
	37	#define A_COND_HI 0x8
	38	#define A_COND_LS 0x9
	39	#define A_COND_GE 0xa
	40	#define A_COND_LT 0xb
	41	#define A_COND_GT 0xc
	42	#define A_COND_LE 0xd
	43	#define A_COND_CS A_COND_HS
	44	#define A_COND_CC A_COND_LO
	45
	46	/* unified conditions */
	47	#define DCOND_EQ A_COND_EQ
	48	#define DCOND_NE A_COND_NE
	49	#define DCOND_MI A_COND_MI
	50	#define DCOND_PL A_COND_PL
	51	#define DCOND_HI A_COND_HI
	52	#define DCOND_HS A_COND_HS
	53	#define DCOND_LO A_COND_LO
	54	#define DCOND_GE A_COND_GE
	55	#define DCOND_GT A_COND_GT
	56	#define DCOND_LT A_COND_LT
	57	#define DCOND_LS A_COND_LS
	58	#define DCOND_LE A_COND_LE
	59	#define DCOND_VS A_COND_VS
	60	#define DCOND_VC A_COND_VC
	61
	62	/* addressing mode 1 */
	63	#define A_AM1_LSL 0
	64	#define A_AM1_LSR 1
	65	#define A_AM1_ASR 2
	66	#define A_AM1_ROR 3
	67
	68	#define A_AM1_IMM(ror2,imm8) (((ror2)<<8) \| (imm8) \| 0x02000000)
	69	#define A_AM1_REG_XIMM(shift_imm,shift_op,rm) (((shift_imm)<<7) \| ((shift_op)<<5) \| (rm))
	70	#define A_AM1_REG_XREG(rs,shift_op,rm) (((rs)<<8) \| ((shift_op)<<5) \| 0x10 \| (rm))
	71
	72	/* data processing op */
	73	#define A_OP_AND 0x0
	74	#define A_OP_EOR 0x1
	75	#define A_OP_SUB 0x2
	76	#define A_OP_RSB 0x3
	77	#define A_OP_ADD 0x4
	78	#define A_OP_ADC 0x5
	79	#define A_OP_SBC 0x6
	80	#define A_OP_RSC 0x7
	81	#define A_OP_TST 0x8
	82	#define A_OP_TEQ 0x9
	83	#define A_OP_CMP 0xa
	84	#define A_OP_ORR 0xc
	85	#define A_OP_MOV 0xd
	86	#define A_OP_BIC 0xe
	87	#define A_OP_MVN 0xf
	88
	89	#define EOP_C_DOP_X(cond,op,s,rn,rd,shifter_op) \
	90	EMIT(((cond)<<28) \| ((op)<< 21) \| ((s)<<20) \| ((rn)<<16) \| ((rd)<<12) \| (shifter_op))
	91
	92	#define EOP_C_DOP_IMM( cond,op,s,rn,rd,ror2,imm8) EOP_C_DOP_X(cond,op,s,rn,rd,A_AM1_IMM(ror2,imm8))
	93	#define EOP_C_DOP_REG_XIMM(cond,op,s,rn,rd,shift_imm,shift_op,rm) EOP_C_DOP_X(cond,op,s,rn,rd,A_AM1_REG_XIMM(shift_imm,shift_op,rm))
	94	#define EOP_C_DOP_REG_XREG(cond,op,s,rn,rd,rs, shift_op,rm) EOP_C_DOP_X(cond,op,s,rn,rd,A_AM1_REG_XREG(rs, shift_op,rm))
	95
	96	#define EOP_MOV_IMM(rd, ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_MOV,0, 0,rd,ror2,imm8)
	97	#define EOP_MVN_IMM(rd, ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_MVN,0, 0,rd,ror2,imm8)
	98	#define EOP_ORR_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_ORR,0,rn,rd,ror2,imm8)
	99	#define EOP_EOR_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_EOR,0,rn,rd,ror2,imm8)
	100	#define EOP_ADD_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_ADD,0,rn,rd,ror2,imm8)
	101	#define EOP_BIC_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_BIC,0,rn,rd,ror2,imm8)
	102	#define EOP_AND_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_AND,0,rn,rd,ror2,imm8)
	103	#define EOP_SUB_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_SUB,0,rn,rd,ror2,imm8)
	104	#define EOP_TST_IMM( rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_TST,1,rn, 0,ror2,imm8)
	105	#define EOP_CMP_IMM( rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_CMP,1,rn, 0,ror2,imm8)
	106	#define EOP_RSB_IMM(rd,rn,ror2,imm8) EOP_C_DOP_IMM(A_COND_AL,A_OP_RSB,0,rn,rd,ror2,imm8)
	107
	108	#define EOP_MOV_IMM_C(cond,rd, ror2,imm8) EOP_C_DOP_IMM(cond,A_OP_MOV,0, 0,rd,ror2,imm8)
	109	#define EOP_ORR_IMM_C(cond,rd,rn,ror2,imm8) EOP_C_DOP_IMM(cond,A_OP_ORR,0,rn,rd,ror2,imm8)
	110	#define EOP_RSB_IMM_C(cond,rd,rn,ror2,imm8) EOP_C_DOP_IMM(cond,A_OP_RSB,0,rn,rd,ror2,imm8)
	111
	112	#define EOP_MOV_REG(cond,s,rd, rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_MOV,s, 0,rd,shift_imm,shift_op,rm)
	113	#define EOP_MVN_REG(cond,s,rd, rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_MVN,s, 0,rd,shift_imm,shift_op,rm)
	114	#define EOP_ORR_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_ORR,s,rn,rd,shift_imm,shift_op,rm)
	115	#define EOP_ADD_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_ADD,s,rn,rd,shift_imm,shift_op,rm)
	116	#define EOP_ADC_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_ADC,s,rn,rd,shift_imm,shift_op,rm)
	117	#define EOP_SUB_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_SUB,s,rn,rd,shift_imm,shift_op,rm)
	118	#define EOP_SBC_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_SBC,s,rn,rd,shift_imm,shift_op,rm)
	119	#define EOP_AND_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_AND,s,rn,rd,shift_imm,shift_op,rm)
	120	#define EOP_EOR_REG(cond,s,rd,rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_EOR,s,rn,rd,shift_imm,shift_op,rm)
	121	#define EOP_CMP_REG(cond, rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_CMP,1,rn, 0,shift_imm,shift_op,rm)
	122	#define EOP_TST_REG(cond, rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_TST,1,rn, 0,shift_imm,shift_op,rm)
	123	#define EOP_TEQ_REG(cond, rn,rm,shift_op,shift_imm) EOP_C_DOP_REG_XIMM(cond,A_OP_TEQ,1,rn, 0,shift_imm,shift_op,rm)
	124
	125	#define EOP_MOV_REG2(s,rd, rm,shift_op,rs) EOP_C_DOP_REG_XREG(A_COND_AL,A_OP_MOV,s, 0,rd,rs,shift_op,rm)
	126	#define EOP_ADD_REG2(s,rd,rn,rm,shift_op,rs) EOP_C_DOP_REG_XREG(A_COND_AL,A_OP_ADD,s,rn,rd,rs,shift_op,rm)
	127	#define EOP_SUB_REG2(s,rd,rn,rm,shift_op,rs) EOP_C_DOP_REG_XREG(A_COND_AL,A_OP_SUB,s,rn,rd,rs,shift_op,rm)
	128
	129	#define EOP_MOV_REG_SIMPLE(rd,rm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_LSL,0)
	130	#define EOP_MOV_REG_LSL(rd, rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_LSL,shift_imm)
	131	#define EOP_MOV_REG_LSR(rd, rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_LSR,shift_imm)
	132	#define EOP_MOV_REG_ASR(rd, rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_ASR,shift_imm)
	133	#define EOP_MOV_REG_ROR(rd, rm,shift_imm) EOP_MOV_REG(A_COND_AL,0,rd,rm,A_AM1_ROR,shift_imm)
	134
	135	#define EOP_ORR_REG_SIMPLE(rd,rm) EOP_ORR_REG(A_COND_AL,0,rd,rd,rm,A_AM1_LSL,0)
	136	#define EOP_ORR_REG_LSL(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSL,shift_imm)
	137	#define EOP_ORR_REG_LSR(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSR,shift_imm)
	138	#define EOP_ORR_REG_ASR(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_ASR,shift_imm)
	139	#define EOP_ORR_REG_ROR(rd,rn,rm,shift_imm) EOP_ORR_REG(A_COND_AL,0,rd,rn,rm,A_AM1_ROR,shift_imm)
	140
	141	#define EOP_ADD_REG_SIMPLE(rd,rm) EOP_ADD_REG(A_COND_AL,0,rd,rd,rm,A_AM1_LSL,0)
	142	#define EOP_ADD_REG_LSL(rd,rn,rm,shift_imm) EOP_ADD_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSL,shift_imm)
	143	#define EOP_ADD_REG_LSR(rd,rn,rm,shift_imm) EOP_ADD_REG(A_COND_AL,0,rd,rn,rm,A_AM1_LSR,shift_imm)
	144
	145	#define EOP_TST_REG_SIMPLE(rn,rm) EOP_TST_REG(A_COND_AL, rn, 0,A_AM1_LSL,rm)
	146
	147	#define EOP_MOV_REG2_LSL(rd, rm,rs) EOP_MOV_REG2(0,rd, rm,A_AM1_LSL,rs)
	148	#define EOP_MOV_REG2_ROR(rd, rm,rs) EOP_MOV_REG2(0,rd, rm,A_AM1_ROR,rs)
	149	#define EOP_ADD_REG2_LSL(rd,rn,rm,rs) EOP_ADD_REG2(0,rd,rn,rm,A_AM1_LSL,rs)
	150	#define EOP_SUB_REG2_LSL(rd,rn,rm,rs) EOP_SUB_REG2(0,rd,rn,rm,A_AM1_LSL,rs)
	151
	152	/* addressing mode 2 */
	153	#define EOP_C_AM2_IMM(cond,u,b,l,rn,rd,offset_12) \
	154	EMIT(((cond)<<28) \| 0x05000000 \| ((u)<<23) \| ((b)<<22) \| ((l)<<20) \| ((rn)<<16) \| ((rd)<<12) \| (offset_12))
	155
	156	/* addressing mode 3 */
	157	#define EOP_C_AM3(cond,u,r,l,rn,rd,s,h,immed_reg) \
	158	EMIT(((cond)<<28) \| 0x01000090 \| ((u)<<23) \| ((r)<<22) \| ((l)<<20) \| ((rn)<<16) \| ((rd)<<12) \| \
	159	((s)<<6) \| ((h)<<5) \| (immed_reg))
	160
	161	#define EOP_C_AM3_IMM(cond,u,l,rn,rd,s,h,offset_8) EOP_C_AM3(cond,u,1,l,rn,rd,s,h,(((offset_8)&0xf0)<<4)\|((offset_8)&0xf))
	162
	163	#define EOP_C_AM3_REG(cond,u,l,rn,rd,s,h,rm) EOP_C_AM3(cond,u,0,l,rn,rd,s,h,rm)
	164
	165	/* ldr and str */
	166	#define EOP_LDR_IMM( rd,rn,offset_12) EOP_C_AM2_IMM(A_COND_AL,1,0,1,rn,rd,offset_12)
	167	#define EOP_LDR_NEGIMM(rd,rn,offset_12) EOP_C_AM2_IMM(A_COND_AL,0,0,1,rn,rd,offset_12)
	168	#define EOP_LDR_SIMPLE(rd,rn) EOP_C_AM2_IMM(A_COND_AL,1,0,1,rn,rd,0)
	169	#define EOP_STR_IMM( rd,rn,offset_12) EOP_C_AM2_IMM(A_COND_AL,1,0,0,rn,rd,offset_12)
	170	#define EOP_STR_SIMPLE(rd,rn) EOP_C_AM2_IMM(A_COND_AL,1,0,0,rn,rd,0)
	171
	172	#define EOP_LDRH_IMM( rd,rn,offset_8) EOP_C_AM3_IMM(A_COND_AL,1,1,rn,rd,0,1,offset_8)
	173	#define EOP_LDRH_SIMPLE(rd,rn) EOP_C_AM3_IMM(A_COND_AL,1,1,rn,rd,0,1,0)
	174	#define EOP_LDRH_REG( rd,rn,rm) EOP_C_AM3_REG(A_COND_AL,1,1,rn,rd,0,1,rm)
	175	#define EOP_STRH_IMM( rd,rn,offset_8) EOP_C_AM3_IMM(A_COND_AL,1,0,rn,rd,0,1,offset_8)
	176	#define EOP_STRH_SIMPLE(rd,rn) EOP_C_AM3_IMM(A_COND_AL,1,0,rn,rd,0,1,0)
	177	#define EOP_STRH_REG( rd,rn,rm) EOP_C_AM3_REG(A_COND_AL,1,0,rn,rd,0,1,rm)
	178
	179	/* ldm and stm */
	180	#define EOP_XXM(cond,p,u,s,w,l,rn,list) \
	181	EMIT(((cond)<<28) \| (1<<27) \| ((p)<<24) \| ((u)<<23) \| ((s)<<22) \| ((w)<<21) \| ((l)<<20) \| ((rn)<<16) \| (list))
	182
	183	#define EOP_STMFD_ST(list) EOP_XXM(A_COND_AL,1,0,0,1,0,13,list)
	184	#define EOP_LDMFD_ST(list) EOP_XXM(A_COND_AL,0,1,0,1,1,13,list)
	185
	186	/* branches */
	187	#define EOP_C_BX(cond,rm) \
	188	EMIT(((cond)<<28) \| 0x012fff10 \| (rm))
	189
	190	#define EOP_BX(rm) EOP_C_BX(A_COND_AL,rm)
	191
	192	#define EOP_C_B(cond,l,signed_immed_24) \
	193	EMIT(((cond)<<28) \| 0x0a000000 \| ((l)<<24) \| (signed_immed_24))
	194
	195	#define EOP_B( signed_immed_24) EOP_C_B(A_COND_AL,0,signed_immed_24)
	196	#define EOP_BL(signed_immed_24) EOP_C_B(A_COND_AL,1,signed_immed_24)
	197
	198	/* misc */
	199	#define EOP_C_MUL(cond,s,rd,rs,rm) \
	200	EMIT(((cond)<<28) \| ((s)<<20) \| ((rd)<<16) \| ((rs)<<8) \| 0x90 \| (rm))
	201
	202	#define EOP_C_UMULL(cond,s,rdhi,rdlo,rs,rm) \
	203	EMIT(((cond)<<28) \| 0x00800000 \| ((s)<<20) \| ((rdhi)<<16) \| ((rdlo)<<12) \| ((rs)<<8) \| 0x90 \| (rm))
	204
	205	#define EOP_C_SMULL(cond,s,rdhi,rdlo,rs,rm) \
	206	EMIT(((cond)<<28) \| 0x00c00000 \| ((s)<<20) \| ((rdhi)<<16) \| ((rdlo)<<12) \| ((rs)<<8) \| 0x90 \| (rm))
	207
	208	#define EOP_MUL(rd,rm,rs) EOP_C_MUL(A_COND_AL,0,rd,rs,rm) // note: rd != rm
	209
	210	#define EOP_C_MRS(cond,rd) \
	211	EMIT(((cond)<<28) \| 0x010f0000 \| ((rd)<<12))
	212
	213	#define EOP_C_MSR_IMM(cond,ror2,imm) \
	214	EMIT(((cond)<<28) \| 0x0328f000 \| ((ror2)<<8) \| (imm)) // cpsr_f
	215
	216	#define EOP_C_MSR_REG(cond,rm) \
	217	EMIT(((cond)<<28) \| 0x0128f000 \| (rm)) // cpsr_f
	218
	219	#define EOP_MRS(rd) EOP_C_MRS(A_COND_AL,rd)
	220	#define EOP_MSR_IMM(ror2,imm) EOP_C_MSR_IMM(A_COND_AL,ror2,imm)
	221	#define EOP_MSR_REG(rm) EOP_C_MSR_REG(A_COND_AL,rm)
	222
	223
	224	static void emith_op_imm2(int cond, int s, int op, int rd, int rn, unsigned int imm)
	225	{
	226	int ror2;
	227	u32 v;
	228
	229	if (op == A_OP_MOV)
	230	rn = 0;
	231	else if (imm == 0)
	232	return;
	233
	234	for (v = imm, ror2 = 0; v != 0 \|\| op == A_OP_MOV; v >>= 8, ror2 -= 8/2) {
	235	/* shift down to get 'best' rot2 */
	236	for (; v && !(v & 3); v >>= 2)
	237	ror2--;
	238
	239	EOP_C_DOP_IMM(cond, op, s, rn, rd, ror2 & 0x0f, v & 0xff);
	240
	241	if (op == A_OP_MOV) {
	242	op = A_OP_ORR;
	243	rn = rd;
	244	}
	245	}
	246	}
	247
	248	#define emith_op_imm(cond, s, op, r, imm) \
	249	emith_op_imm2(cond, s, op, r, r, imm)
	250
	251	// test op
	252	#define emith_top_imm(cond, op, r, imm) { \
	253	u32 ror2, v; \
	254	for (ror2 = 0, v = imm; v && !(v & 3); v >>= 2) \
	255	ror2--; \
	256	EOP_C_DOP_IMM(cond, op, 1, r, 0, ror2 & 0x0f, v & 0xff); \
	257	}
	258
	259	#define is_offset_24(val) \
	260	((val) >= (int)0xff000000 && (val) <= 0x00ffffff)
	261
	262	static int emith_xbranch(int cond, void *target, int is_call)
	263	{
	264	int val = (u32 )target - (u32 )tcache_ptr - 2;
	265	int direct = is_offset_24(val);
	266	u32 start_ptr = (u32 )tcache_ptr;
	267
	268	if (direct)
	269	{
	270	EOP_C_B(cond,is_call,val & 0xffffff); // b, bl target
	271	}
	272	else
	273	{
	274	#ifdef __EPOC32__
	275	// elprintf(EL_SVP, "emitting indirect jmp %08x->%08x", tcache_ptr, target);
	276	if (is_call)
	277	EOP_ADD_IMM(14,15,0,8); // add lr,pc,#8
	278	EOP_C_AM2_IMM(cond,1,0,1,15,15,0); // ldrcc pc,[pc]
	279	EOP_MOV_REG_SIMPLE(15,15); // mov pc, pc
	280	EMIT((u32)target);
	281	#else
	282	// should never happen
	283	elprintf(EL_STATUS\|EL_SVP\|EL_ANOMALY, "indirect jmp %08x->%08x", target, tcache_ptr);
	284	exit(1);
	285	#endif
	286	}
	287
	288	return (u32 *)tcache_ptr - start_ptr;
	289	}
	290
	291
	292	// fake "simple" or "short" jump - using cond insns instead
	293	#define EMITH_SJMP_START(cond) \
	294	(void)(cond)
	295
	296	#define EMITH_SJMP_END(cond) \
	297	(void)(cond)
	298
	299	#define emith_move_r_r(d, s) \
	300	EOP_MOV_REG_SIMPLE(d, s)
	301
	302	#define emith_mvn_r_r(d, s) \
	303	EOP_MVN_REG(A_COND_AL,0,d,s,A_AM1_LSL,0)
	304
	305	#define emith_or_r_r_r_lsl(d, s1, s2, lslimm) \
	306	EOP_ORR_REG(A_COND_AL,0,d,s1,s2,A_AM1_LSL,lslimm)
	307
	308	#define emith_eor_r_r_r_lsl(d, s1, s2, lslimm) \
	309	EOP_EOR_REG(A_COND_AL,0,d,s1,s2,A_AM1_LSL,lslimm)
	310
	311	#define emith_or_r_r_r(d, s1, s2) \
	312	emith_or_r_r_r_lsl(d, s1, s2, 0)
	313
	314	#define emith_eor_r_r_r(d, s1, s2) \
	315	emith_eor_r_r_r_lsl(d, s1, s2, 0)
	316
	317	#define emith_add_r_r(d, s) \
	318	EOP_ADD_REG(A_COND_AL,0,d,d,s,A_AM1_LSL,0)
	319
	320	#define emith_sub_r_r(d, s) \
	321	EOP_SUB_REG(A_COND_AL,0,d,d,s,A_AM1_LSL,0)
	322
	323	#define emith_and_r_r(d, s) \
	324	EOP_AND_REG(A_COND_AL,0,d,d,s,A_AM1_LSL,0)
	325
	326	#define emith_or_r_r(d, s) \
	327	emith_or_r_r_r(d, d, s)
	328
	329	#define emith_eor_r_r(d, s) \
	330	emith_eor_r_r_r(d, d, s)
	331
	332	#define emith_tst_r_r(d, s) \
	333	EOP_TST_REG(A_COND_AL,d,s,A_AM1_LSL,0)
	334
	335	#define emith_teq_r_r(d, s) \
	336	EOP_TEQ_REG(A_COND_AL,d,s,A_AM1_LSL,0)
	337
	338	#define emith_cmp_r_r(d, s) \
	339	EOP_CMP_REG(A_COND_AL,d,s,A_AM1_LSL,0)
	340
	341	#define emith_addf_r_r(d, s) \
	342	EOP_ADD_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)
	343
	344	#define emith_subf_r_r(d, s) \
	345	EOP_SUB_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)
	346
	347	#define emith_adcf_r_r(d, s) \
	348	EOP_ADC_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)
	349
	350	#define emith_sbcf_r_r(d, s) \
	351	EOP_SBC_REG(A_COND_AL,1,d,d,s,A_AM1_LSL,0)
	352
	353	#define emith_move_r_imm(r, imm) \
	354	emith_op_imm(A_COND_AL, 0, A_OP_MOV, r, imm)
	355
	356	#define emith_add_r_imm(r, imm) \
	357	emith_op_imm(A_COND_AL, 0, A_OP_ADD, r, imm)
	358
	359	#define emith_sub_r_imm(r, imm) \
	360	emith_op_imm(A_COND_AL, 0, A_OP_SUB, r, imm)
	361
	362	#define emith_bic_r_imm(r, imm) \
	363	emith_op_imm(A_COND_AL, 0, A_OP_BIC, r, imm)
	364
	365	#define emith_and_r_imm(r, imm) \
	366	emith_op_imm(A_COND_AL, 0, A_OP_AND, r, imm)
	367
	368	#define emith_or_r_imm(r, imm) \
	369	emith_op_imm(A_COND_AL, 0, A_OP_ORR, r, imm)
	370
	371	#define emith_eor_r_imm(r, imm) \
	372	emith_op_imm(A_COND_AL, 0, A_OP_EOR, r, imm)
	373
	374	// note: only use 8bit imm for these
	375	#define emith_tst_r_imm(r, imm) \
	376	emith_top_imm(A_COND_AL, A_OP_TST, r, imm)
	377
	378	#define emith_cmp_r_imm(r, imm) \
	379	emith_top_imm(A_COND_AL, A_OP_CMP, r, imm)
	380
	381	#define emith_subf_r_imm(r, imm) \
	382	emith_op_imm(A_COND_AL, 1, A_OP_SUB, r, imm)
	383
	384	#define emith_add_r_imm_c(cond, r, imm) \
	385	emith_op_imm(cond, 0, A_OP_ADD, r, imm)
	386
	387	#define emith_sub_r_imm_c(cond, r, imm) \
	388	emith_op_imm(cond, 0, A_OP_SUB, r, imm)
	389
	390	#define emith_or_r_imm_c(cond, r, imm) \
	391	emith_op_imm(cond, 0, A_OP_ORR, r, imm)
	392
	393	#define emith_bic_r_imm_c(cond, r, imm) \
	394	emith_op_imm(cond, 0, A_OP_BIC, r, imm)
	395
	396	#define emith_move_r_imm_s8(r, imm) { \
	397	if ((imm) & 0x80) \
	398	EOP_MVN_IMM(r, 0, ((imm) ^ 0xff)); \
	399	else \
	400	EOP_MOV_IMM(r, 0, imm); \
	401	}
	402
	403	#define emith_and_r_r_imm(d, s, imm) \
	404	emith_op_imm2(A_COND_AL, 0, A_OP_AND, d, s, imm)
	405
	406	#define emith_neg_r_r(d, s) \
	407	EOP_RSB_IMM(d, s, 0, 0)
	408
	409	#define emith_lsl(d, s, cnt) \
	410	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_LSL,cnt)
	411
	412	#define emith_lsr(d, s, cnt) \
	413	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_LSR,cnt)
	414
	415	#define emith_ror(d, s, cnt) \
	416	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_ROR,cnt)
	417
	418	#define emith_rol(d, s, cnt) \
	419	EOP_MOV_REG(A_COND_AL,0,d,s,A_AM1_ROR,32-(cnt)); \
	420
	421	#define emith_lslf(d, s, cnt) \
	422	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_LSL,cnt)
	423
	424	#define emith_lsrf(d, s, cnt) \
	425	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_LSR,cnt)
	426
	427	#define emith_asrf(d, s, cnt) \
	428	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_ASR,cnt)
	429
	430	// note: only C flag updated correctly
	431	#define emith_rolf(d, s, cnt) { \
	432	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_ROR,32-(cnt)); \
	433	/* we don't have ROL so we shift to get the right carry */ \
	434	EOP_TST_REG(A_COND_AL,d,d,A_AM1_LSR,1); \
	435	}
	436
	437	#define emith_rorf(d, s, cnt) \
	438	EOP_MOV_REG(A_COND_AL,1,d,s,A_AM1_ROR,cnt)
	439
	440	#define emith_rolcf(d) \
	441	emith_adcf_r_r(d, d)
	442
	443	#define emith_rorcf(d) \
	444	EOP_MOV_REG(A_COND_AL,1,d,d,A_AM1_ROR,0) /* ROR #0 -> RRX */
	445
	446	#define emith_negcf_r_r(d, s) \
	447	EOP_C_DOP_IMM(A_COND_AL,A_OP_RSC,1,s,d,0,0)
	448
	449	#define emith_mul(d, s1, s2) { \
	450	if ((d) != (s1)) /* rd != rm limitation */ \
	451	EOP_MUL(d, s1, s2); \
	452	else \
	453	EOP_MUL(d, s2, s1); \
	454	}
	455
	456	#define emith_mul_u64(dlo, dhi, s1, s2) \
	457	EOP_C_UMULL(A_COND_AL,0,dhi,dlo,s1,s2)
	458
	459	#define emith_mul_s64(dlo, dhi, s1, s2) \
	460	EOP_C_SMULL(A_COND_AL,0,dhi,dlo,s1,s2)
	461
	462	// misc
	463	#define emith_ctx_read(r, offs) \
	464	EOP_LDR_IMM(r, CONTEXT_REG, offs)
	465
	466	#define emith_ctx_write(r, offs) \
	467	EOP_STR_IMM(r, CONTEXT_REG, offs)
	468
	469	#define emith_clear_msb(d, s, count) { \
	470	u32 t; \
	471	if ((count) <= 8) { \
	472	t = (count) - 8; \
	473	t = (0xff << t) & 0xff; \
	474	EOP_BIC_IMM(d,s,8/2,t); \
	475	} else if ((count) >= 24) { \
	476	t = (count) - 24; \
	477	t = 0xff >> t; \
	478	EOP_AND_IMM(d,s,0,t); \
	479	} else { \
	480	EOP_MOV_REG_LSL(d,s,count); \
	481	EOP_MOV_REG_LSR(d,d,count); \
	482	} \
	483	}
	484
	485	#define emith_sext(d, s, bits) { \
	486	EOP_MOV_REG_LSL(d,s,32 - (bits)); \
	487	EOP_MOV_REG_ASR(d,d,32 - (bits)); \
	488	}
	489
	490	// put bit0 of r0 to carry
	491	#define emith_set_carry(r0) \
	492	EOP_TST_REG(A_COND_AL,r0,r0,A_AM1_LSR,1) /* shift out to carry */ \
	493
	494	// put bit0 of r0 to carry (for subtraction, inverted on ARM)
	495	#define emith_set_carry_sub(r0) { \
	496	int t = rcache_get_tmp(); \
	497	EOP_EOR_IMM(t,r0,0,1); /* invert */ \
	498	EOP_MOV_REG(A_COND_AL,1,t,t,A_AM1_LSR,1); /* shift out to carry */ \
	499	rcache_free_tmp(t); \
	500	}
	501
	502	#define host_arg2reg(rd, arg) \
	503	rd = arg
	504
	505	// upto 4 args
	506	#define emith_pass_arg_r(arg, reg) \
	507	EOP_MOV_REG_SIMPLE(arg, reg)
	508
	509	#define emith_pass_arg_imm(arg, imm) \
	510	emith_move_r_imm(arg, imm)
	511
	512	#define emith_call_cond(cond, target) \
	513	emith_xbranch(cond, target, 1)
	514
	515	#define emith_jump_cond(cond, target) \
	516	emith_xbranch(cond, target, 0)
	517
	518	#define emith_call(target) \
	519	emith_call_cond(A_COND_AL, target)
	520
	521	#define emith_jump(target) \
	522	emith_jump_cond(A_COND_AL, target)
	523
	524	/* SH2 drc specific */
	525	#define emith_sh2_test_t() { \
	526	int r = rcache_get_reg(SHR_SR, RC_GR_READ); \
	527	EOP_TST_IMM(r, 0, 1); \
	528	}
	529
	530	#define emith_sh2_dtbf_loop() { \
	531	int cr, rn; \
	532	int tmp_ = rcache_get_tmp(); \
	533	cr = rcache_get_reg(SHR_SR, RC_GR_RMW); \
	534	rn = rcache_get_reg((op >> 8) & 0x0f, RC_GR_RMW); \
	535	emith_sub_r_imm(rn, 1); /* sub rn, #1 */ \
	536	emith_bic_r_imm(cr, 1); /* bic cr, #1 */ \
	537	emith_sub_r_imm(cr, (cycles+1) << 12); /* sub cr, #(cycles+1)<<12 */ \
	538	cycles = 0; \
	539	emith_asrf(tmp_, cr, 2+12); /* movs tmp_, cr, asr #2+12 */\
	540	EOP_MOV_IMM_C(A_COND_MI,tmp_,0,0); /* movmi tmp_, #0 */ \
	541	emith_lsl(cr, cr, 20); /* mov cr, cr, lsl #20 */ \
	542	emith_lsr(cr, cr, 20); /* mov cr, cr, lsr #20 */ \
	543	emith_subf_r_r(rn, tmp_); /* subs rn, tmp_ */ \
	544	EOP_RSB_IMM_C(A_COND_LS,tmp_,rn,0,0); /* rsbls tmp_, rn, #0 */ \
	545	EOP_ORR_REG(A_COND_LS,0,cr,cr,tmp_,A_AM1_LSL,12+2); /* orrls cr,tmp_,lsl #12+2 */\
	546	EOP_ORR_IMM_C(A_COND_LS,cr,cr,0,1); /* orrls cr, #1 */ \
	547	EOP_MOV_IMM_C(A_COND_LS,rn,0,0); /* movls rn, #0 */ \
	548	rcache_free_tmp(tmp_); \
	549	}
	550
	551	#define emith_write_sr(srcr) { \
	552	int srr = rcache_get_reg(SHR_SR, RC_GR_RMW); \
	553	emith_lsr(srr, srr, 12); \
	554	emith_or_r_r_r_lsl(srr, srr, srcr, 20); \
	555	emith_ror(srr, srr, 20); \
	556	}
	557
	558	#define emith_carry_to_t(srr, is_sub) { \
	559	if (is_sub) { /* has inverted C on ARM */ \
	560	emith_or_r_imm_c(A_COND_CC, srr, 1); \
	561	emith_bic_r_imm_c(A_COND_CS, srr, 1); \
	562	} else { \
	563	emith_or_r_imm_c(A_COND_CS, srr, 1); \
	564	emith_bic_r_imm_c(A_COND_CC, srr, 1); \
	565	} \
	566	}