blob: a87b8bb2813b6e3ad600a4f31e37b80eaf3280d4 [file] [log] [blame]
/*
* CRISv10 emulation for qemu: main translation routines.
*
* Copyright (c) 2010 AXIS Communications AB
* Written by Edgar E. Iglesias.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, see <http://www.gnu.org/licenses/>.
*/
#include "qemu/osdep.h"
#include "crisv10-decode.h"
static const char *regnames_v10[] =
{
"$r0", "$r1", "$r2", "$r3",
"$r4", "$r5", "$r6", "$r7",
"$r8", "$r9", "$r10", "$r11",
"$r12", "$r13", "$sp", "$pc",
};
static const char *pregnames_v10[] =
{
"$bz", "$vr", "$p2", "$p3",
"$wz", "$ccr", "$p6-prefix", "$mof",
"$dz", "$ibr", "$irp", "$srp",
"$bar", "$dccr", "$brp", "$usp",
};
/* We need this table to handle preg-moves with implicit width. */
static int preg_sizes_v10[] = {
1, /* bz. */
1, /* vr. */
1, /* pid. */
1, /* srs. */
2, /* wz. */
2, 2, 4,
4, 4, 4, 4,
4, 4, 4, 4,
};
static inline int dec10_size(unsigned int size)
{
size++;
if (size == 3)
size++;
return size;
}
static inline void cris_illegal_insn(DisasContext *dc)
{
qemu_log_mask(LOG_GUEST_ERROR, "illegal insn at pc=%x\n", dc->pc);
t_gen_raise_exception(EXCP_BREAK);
}
static void gen_store_v10_conditional(DisasContext *dc, TCGv addr, TCGv val,
unsigned int size, int mem_index)
{
TCGLabel *l1 = gen_new_label();
TCGv taddr = tcg_temp_local_new();
TCGv tval = tcg_temp_local_new();
TCGv t1 = tcg_temp_local_new();
dc->postinc = 0;
cris_evaluate_flags(dc);
tcg_gen_mov_tl(taddr, addr);
tcg_gen_mov_tl(tval, val);
/* Store only if F flag isn't set */
tcg_gen_andi_tl(t1, cpu_PR[PR_CCS], F_FLAG_V10);
tcg_gen_brcondi_tl(TCG_COND_NE, t1, 0, l1);
if (size == 1) {
tcg_gen_qemu_st8(tval, taddr, mem_index);
} else if (size == 2) {
tcg_gen_qemu_st16(tval, taddr, mem_index);
} else {
tcg_gen_qemu_st32(tval, taddr, mem_index);
}
gen_set_label(l1);
tcg_gen_shri_tl(t1, t1, 1); /* shift F to P position */
tcg_gen_or_tl(cpu_PR[PR_CCS], cpu_PR[PR_CCS], t1); /*P=F*/
tcg_temp_free(t1);
tcg_temp_free(tval);
tcg_temp_free(taddr);
}
static void gen_store_v10(DisasContext *dc, TCGv addr, TCGv val,
unsigned int size)
{
int mem_index = cpu_mmu_index(&dc->cpu->env, false);
/* If we get a fault on a delayslot we must keep the jmp state in
the cpu-state to be able to re-execute the jmp. */
if (dc->delayed_branch == 1) {
cris_store_direct_jmp(dc);
}
/* Conditional writes. We only support the kind were X is known
at translation time. */
if (dc->flagx_known && dc->flags_x) {
gen_store_v10_conditional(dc, addr, val, size, mem_index);
return;
}
if (size == 1) {
tcg_gen_qemu_st8(val, addr, mem_index);
} else if (size == 2) {
tcg_gen_qemu_st16(val, addr, mem_index);
} else {
tcg_gen_qemu_st32(val, addr, mem_index);
}
}
/* Prefix flag and register are used to handle the more complex
addressing modes. */
static void cris_set_prefix(DisasContext *dc)
{
dc->clear_prefix = 0;
dc->tb_flags |= PFIX_FLAG;
tcg_gen_ori_tl(cpu_PR[PR_CCS], cpu_PR[PR_CCS], PFIX_FLAG);
/* prefix insns don't clear the x flag. */
dc->clear_x = 0;
cris_lock_irq(dc);
}
static void crisv10_prepare_memaddr(DisasContext *dc,
TCGv addr, unsigned int size)
{
if (dc->tb_flags & PFIX_FLAG) {
tcg_gen_mov_tl(addr, cpu_PR[PR_PREFIX]);
} else {
tcg_gen_mov_tl(addr, cpu_R[dc->src]);
}
}
static unsigned int crisv10_post_memaddr(DisasContext *dc, unsigned int size)
{
unsigned int insn_len = 0;
if (dc->tb_flags & PFIX_FLAG) {
if (dc->mode == CRISV10_MODE_AUTOINC) {
tcg_gen_mov_tl(cpu_R[dc->src], cpu_PR[PR_PREFIX]);
}
} else {
if (dc->mode == CRISV10_MODE_AUTOINC) {
if (dc->src == 15) {
insn_len += size & ~1;
} else {
tcg_gen_addi_tl(cpu_R[dc->src], cpu_R[dc->src], size);
}
}
}
return insn_len;
}
static int dec10_prep_move_m(CPUCRISState *env, DisasContext *dc,
int s_ext, int memsize, TCGv dst)
{
unsigned int rs;
uint32_t imm;
int is_imm;
int insn_len = 0;
rs = dc->src;
is_imm = rs == 15 && !(dc->tb_flags & PFIX_FLAG);
LOG_DIS("rs=%d rd=%d is_imm=%d mode=%d pfix=%d\n",
rs, dc->dst, is_imm, dc->mode, dc->tb_flags & PFIX_FLAG);
/* Load [$rs] onto T1. */
if (is_imm) {
if (memsize != 4) {
if (s_ext) {
if (memsize == 1)
imm = cpu_ldsb_code(env, dc->pc + 2);
else
imm = cpu_ldsw_code(env, dc->pc + 2);
} else {
if (memsize == 1)
imm = cpu_ldub_code(env, dc->pc + 2);
else
imm = cpu_lduw_code(env, dc->pc + 2);
}
} else
imm = cpu_ldl_code(env, dc->pc + 2);
tcg_gen_movi_tl(dst, imm);
if (dc->mode == CRISV10_MODE_AUTOINC) {
insn_len += memsize;
if (memsize == 1)
insn_len++;
tcg_gen_addi_tl(cpu_R[15], cpu_R[15], insn_len);
}
} else {
TCGv addr;
addr = tcg_temp_new();
cris_flush_cc_state(dc);
crisv10_prepare_memaddr(dc, addr, memsize);
gen_load(dc, dst, addr, memsize, 0);
if (s_ext)
t_gen_sext(dst, dst, memsize);
else
t_gen_zext(dst, dst, memsize);
insn_len += crisv10_post_memaddr(dc, memsize);
tcg_temp_free(addr);
}
if (dc->mode == CRISV10_MODE_INDIRECT && (dc->tb_flags & PFIX_FLAG)) {
dc->dst = dc->src;
}
return insn_len;
}
static unsigned int dec10_quick_imm(DisasContext *dc)
{
int32_t imm, simm;
int op;
/* sign extend. */
imm = dc->ir & ((1 << 6) - 1);
simm = (int8_t) (imm << 2);
simm >>= 2;
switch (dc->opcode) {
case CRISV10_QIMM_BDAP_R0:
case CRISV10_QIMM_BDAP_R1:
case CRISV10_QIMM_BDAP_R2:
case CRISV10_QIMM_BDAP_R3:
simm = (int8_t)dc->ir;
LOG_DIS("bdap %d $r%d\n", simm, dc->dst);
LOG_DIS("pc=%x mode=%x quickimm %d r%d r%d\n",
dc->pc, dc->mode, dc->opcode, dc->src, dc->dst);
cris_set_prefix(dc);
if (dc->dst == 15) {
tcg_gen_movi_tl(cpu_PR[PR_PREFIX], dc->pc + 2 + simm);
} else {
tcg_gen_addi_tl(cpu_PR[PR_PREFIX], cpu_R[dc->dst], simm);
}
break;
case CRISV10_QIMM_MOVEQ:
LOG_DIS("moveq %d, $r%d\n", simm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_MOVE, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(simm), 4);
break;
case CRISV10_QIMM_CMPQ:
LOG_DIS("cmpq %d, $r%d\n", simm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_CMP, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(simm), 4);
break;
case CRISV10_QIMM_ADDQ:
LOG_DIS("addq %d, $r%d\n", imm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_ADD, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(imm), 4);
break;
case CRISV10_QIMM_ANDQ:
LOG_DIS("andq %d, $r%d\n", simm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_AND, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(simm), 4);
break;
case CRISV10_QIMM_ASHQ:
LOG_DIS("ashq %d, $r%d\n", simm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
op = imm & (1 << 5);
imm &= 0x1f;
if (op) {
cris_alu(dc, CC_OP_ASR, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(imm), 4);
} else {
/* BTST */
cris_update_cc_op(dc, CC_OP_FLAGS, 4);
gen_helper_btst(cpu_PR[PR_CCS], cpu_env, cpu_R[dc->dst],
tcg_const_tl(imm), cpu_PR[PR_CCS]);
}
break;
case CRISV10_QIMM_LSHQ:
LOG_DIS("lshq %d, $r%d\n", simm, dc->dst);
op = CC_OP_LSL;
if (imm & (1 << 5)) {
op = CC_OP_LSR;
}
imm &= 0x1f;
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, op, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(imm), 4);
break;
case CRISV10_QIMM_SUBQ:
LOG_DIS("subq %d, $r%d\n", imm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_SUB, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(imm), 4);
break;
case CRISV10_QIMM_ORQ:
LOG_DIS("andq %d, $r%d\n", simm, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_OR, cpu_R[dc->dst],
cpu_R[dc->dst], tcg_const_tl(simm), 4);
break;
case CRISV10_QIMM_BCC_R0:
case CRISV10_QIMM_BCC_R1:
case CRISV10_QIMM_BCC_R2:
case CRISV10_QIMM_BCC_R3:
imm = dc->ir & 0xff;
/* bit 0 is a sign bit. */
if (imm & 1) {
imm |= 0xffffff00; /* sign extend. */
imm &= ~1; /* get rid of the sign bit. */
}
imm += 2;
LOG_DIS("b%s %d\n", cc_name(dc->cond), imm);
cris_cc_mask(dc, 0);
cris_prepare_cc_branch(dc, imm, dc->cond);
break;
default:
LOG_DIS("pc=%x mode=%x quickimm %d r%d r%d\n",
dc->pc, dc->mode, dc->opcode, dc->src, dc->dst);
cpu_abort(CPU(dc->cpu), "Unhandled quickimm\n");
break;
}
return 2;
}
static unsigned int dec10_setclrf(DisasContext *dc)
{
uint32_t flags;
unsigned int set = ~dc->opcode & 1;
flags = EXTRACT_FIELD(dc->ir, 0, 3)
| (EXTRACT_FIELD(dc->ir, 12, 15) << 4);
LOG_DIS("%s set=%d flags=%x\n", __func__, set, flags);
if (flags & X_FLAG) {
dc->flagx_known = 1;
if (set)
dc->flags_x = X_FLAG;
else
dc->flags_x = 0;
}
cris_evaluate_flags (dc);
cris_update_cc_op(dc, CC_OP_FLAGS, 4);
cris_update_cc_x(dc);
tcg_gen_movi_tl(cc_op, dc->cc_op);
if (set) {
tcg_gen_ori_tl(cpu_PR[PR_CCS], cpu_PR[PR_CCS], flags);
} else {
tcg_gen_andi_tl(cpu_PR[PR_CCS], cpu_PR[PR_CCS],
~(flags|F_FLAG_V10|P_FLAG_V10));
}
dc->flags_uptodate = 1;
dc->clear_x = 0;
cris_lock_irq(dc);
return 2;
}
static inline void dec10_reg_prep_sext(DisasContext *dc, int size, int sext,
TCGv dd, TCGv ds, TCGv sd, TCGv ss)
{
if (sext) {
t_gen_sext(dd, sd, size);
t_gen_sext(ds, ss, size);
} else {
t_gen_zext(dd, sd, size);
t_gen_zext(ds, ss, size);
}
}
static void dec10_reg_alu(DisasContext *dc, int op, int size, int sext)
{
TCGv t[2];
t[0] = tcg_temp_new();
t[1] = tcg_temp_new();
dec10_reg_prep_sext(dc, size, sext,
t[0], t[1], cpu_R[dc->dst], cpu_R[dc->src]);
if (op == CC_OP_LSL || op == CC_OP_LSR || op == CC_OP_ASR) {
tcg_gen_andi_tl(t[1], t[1], 63);
}
assert(dc->dst != 15);
cris_alu(dc, op, cpu_R[dc->dst], t[0], t[1], size);
tcg_temp_free(t[0]);
tcg_temp_free(t[1]);
}
static void dec10_reg_bound(DisasContext *dc, int size)
{
TCGv t;
t = tcg_temp_local_new();
t_gen_zext(t, cpu_R[dc->src], size);
cris_alu(dc, CC_OP_BOUND, cpu_R[dc->dst], cpu_R[dc->dst], t, 4);
tcg_temp_free(t);
}
static void dec10_reg_mul(DisasContext *dc, int size, int sext)
{
int op = sext ? CC_OP_MULS : CC_OP_MULU;
TCGv t[2];
t[0] = tcg_temp_new();
t[1] = tcg_temp_new();
dec10_reg_prep_sext(dc, size, sext,
t[0], t[1], cpu_R[dc->dst], cpu_R[dc->src]);
cris_alu(dc, op, cpu_R[dc->dst], t[0], t[1], 4);
tcg_temp_free(t[0]);
tcg_temp_free(t[1]);
}
static void dec10_reg_movs(DisasContext *dc)
{
int size = (dc->size & 1) + 1;
TCGv t;
LOG_DIS("movx.%d $r%d, $r%d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
t = tcg_temp_new();
if (dc->ir & 32)
t_gen_sext(t, cpu_R[dc->src], size);
else
t_gen_zext(t, cpu_R[dc->src], size);
cris_alu(dc, CC_OP_MOVE, cpu_R[dc->dst], cpu_R[dc->dst], t, 4);
tcg_temp_free(t);
}
static void dec10_reg_alux(DisasContext *dc, int op)
{
int size = (dc->size & 1) + 1;
TCGv t;
LOG_DIS("movx.%d $r%d, $r%d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
t = tcg_temp_new();
if (dc->ir & 32)
t_gen_sext(t, cpu_R[dc->src], size);
else
t_gen_zext(t, cpu_R[dc->src], size);
cris_alu(dc, op, cpu_R[dc->dst], cpu_R[dc->dst], t, 4);
tcg_temp_free(t);
}
static void dec10_reg_mov_pr(DisasContext *dc)
{
LOG_DIS("move p%d r%d sz=%d\n", dc->dst, dc->src, preg_sizes_v10[dc->dst]);
cris_lock_irq(dc);
if (dc->src == 15) {
tcg_gen_mov_tl(env_btarget, cpu_PR[dc->dst]);
cris_prepare_jmp(dc, JMP_INDIRECT);
return;
}
if (dc->dst == PR_CCS) {
cris_evaluate_flags(dc);
}
cris_alu(dc, CC_OP_MOVE, cpu_R[dc->src],
cpu_R[dc->src], cpu_PR[dc->dst], preg_sizes_v10[dc->dst]);
}
static void dec10_reg_abs(DisasContext *dc)
{
TCGv t0;
LOG_DIS("abs $r%u, $r%u\n", dc->src, dc->dst);
assert(dc->dst != 15);
t0 = tcg_temp_new();
tcg_gen_sari_tl(t0, cpu_R[dc->src], 31);
tcg_gen_xor_tl(cpu_R[dc->dst], cpu_R[dc->src], t0);
tcg_gen_sub_tl(t0, cpu_R[dc->dst], t0);
cris_alu(dc, CC_OP_MOVE, cpu_R[dc->dst], cpu_R[dc->dst], t0, 4);
tcg_temp_free(t0);
}
static void dec10_reg_swap(DisasContext *dc)
{
TCGv t0;
LOG_DIS("not $r%d, $r%d\n", dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
t0 = tcg_temp_new();
tcg_gen_mov_tl(t0, cpu_R[dc->src]);
if (dc->dst & 8)
tcg_gen_not_tl(t0, t0);
if (dc->dst & 4)
t_gen_swapw(t0, t0);
if (dc->dst & 2)
t_gen_swapb(t0, t0);
if (dc->dst & 1)
t_gen_swapr(t0, t0);
cris_alu(dc, CC_OP_MOVE, cpu_R[dc->src], cpu_R[dc->src], t0, 4);
tcg_temp_free(t0);
}
static void dec10_reg_scc(DisasContext *dc)
{
int cond = dc->dst;
LOG_DIS("s%s $r%u\n", cc_name(cond), dc->src);
gen_tst_cc(dc, cpu_R[dc->src], cond);
tcg_gen_setcondi_tl(TCG_COND_NE, cpu_R[dc->src], cpu_R[dc->src], 0);
cris_cc_mask(dc, 0);
}
static unsigned int dec10_reg(DisasContext *dc)
{
TCGv t;
unsigned int insn_len = 2;
unsigned int size = dec10_size(dc->size);
unsigned int tmp;
if (dc->size != 3) {
switch (dc->opcode) {
case CRISV10_REG_MOVE_R:
LOG_DIS("move.%d $r%d, $r%d\n", dc->size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_MOVE, size, 0);
if (dc->dst == 15) {
tcg_gen_mov_tl(env_btarget, cpu_R[dc->dst]);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch = 1;
}
break;
case CRISV10_REG_MOVX:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_movs(dc);
break;
case CRISV10_REG_ADDX:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alux(dc, CC_OP_ADD);
break;
case CRISV10_REG_SUBX:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alux(dc, CC_OP_SUB);
break;
case CRISV10_REG_ADD:
LOG_DIS("add $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_ADD, size, 0);
break;
case CRISV10_REG_SUB:
LOG_DIS("sub $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_SUB, size, 0);
break;
case CRISV10_REG_CMP:
LOG_DIS("cmp $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_CMP, size, 0);
break;
case CRISV10_REG_BOUND:
LOG_DIS("bound $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_bound(dc, size);
break;
case CRISV10_REG_AND:
LOG_DIS("and $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_AND, size, 0);
break;
case CRISV10_REG_ADDI:
if (dc->src == 15) {
/* nop. */
return 2;
}
t = tcg_temp_new();
LOG_DIS("addi r%d r%d size=%d\n", dc->src, dc->dst, dc->size);
tcg_gen_shli_tl(t, cpu_R[dc->dst], dc->size & 3);
tcg_gen_add_tl(cpu_R[dc->src], cpu_R[dc->src], t);
tcg_temp_free(t);
break;
case CRISV10_REG_LSL:
LOG_DIS("lsl $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_LSL, size, 0);
break;
case CRISV10_REG_LSR:
LOG_DIS("lsr $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_LSR, size, 0);
break;
case CRISV10_REG_ASR:
LOG_DIS("asr $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_ASR, size, 1);
break;
case CRISV10_REG_OR:
LOG_DIS("or $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_OR, size, 0);
break;
case CRISV10_REG_NEG:
LOG_DIS("neg $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_NEG, size, 0);
break;
case CRISV10_REG_BIAP:
LOG_DIS("BIAP pc=%x reg %d r%d r%d size=%d\n", dc->pc,
dc->opcode, dc->src, dc->dst, size);
switch (size) {
case 4: tmp = 2; break;
case 2: tmp = 1; break;
case 1: tmp = 0; break;
default:
cpu_abort(CPU(dc->cpu), "Unhandled BIAP");
break;
}
t = tcg_temp_new();
tcg_gen_shli_tl(t, cpu_R[dc->dst], tmp);
if (dc->src == 15) {
tcg_gen_addi_tl(cpu_PR[PR_PREFIX], t, ((dc->pc +2)| 1) + 1);
} else {
tcg_gen_add_tl(cpu_PR[PR_PREFIX], cpu_R[dc->src], t);
}
tcg_temp_free(t);
cris_set_prefix(dc);
break;
default:
LOG_DIS("pc=%x reg %d r%d r%d\n", dc->pc,
dc->opcode, dc->src, dc->dst);
cpu_abort(CPU(dc->cpu), "Unhandled opcode");
break;
}
} else {
switch (dc->opcode) {
case CRISV10_REG_MOVX:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_movs(dc);
break;
case CRISV10_REG_ADDX:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alux(dc, CC_OP_ADD);
break;
case CRISV10_REG_SUBX:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alux(dc, CC_OP_SUB);
break;
case CRISV10_REG_MOVE_SPR_R:
cris_evaluate_flags(dc);
cris_cc_mask(dc, 0);
dec10_reg_mov_pr(dc);
break;
case CRISV10_REG_MOVE_R_SPR:
LOG_DIS("move r%d p%d\n", dc->src, dc->dst);
cris_evaluate_flags(dc);
if (dc->src != 11) /* fast for srp. */
dc->cpustate_changed = 1;
t_gen_mov_preg_TN(dc, dc->dst, cpu_R[dc->src]);
break;
case CRISV10_REG_SETF:
case CRISV10_REG_CLEARF:
dec10_setclrf(dc);
break;
case CRISV10_REG_SWAP:
dec10_reg_swap(dc);
break;
case CRISV10_REG_ABS:
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_abs(dc);
break;
case CRISV10_REG_LZ:
LOG_DIS("lz $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_LZ, 4, 0);
break;
case CRISV10_REG_XOR:
LOG_DIS("xor $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_alu(dc, CC_OP_XOR, 4, 0);
break;
case CRISV10_REG_BTST:
LOG_DIS("btst $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_update_cc_op(dc, CC_OP_FLAGS, 4);
gen_helper_btst(cpu_PR[PR_CCS], cpu_env, cpu_R[dc->dst],
cpu_R[dc->src], cpu_PR[PR_CCS]);
break;
case CRISV10_REG_DSTEP:
LOG_DIS("dstep $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_DSTEP, cpu_R[dc->dst],
cpu_R[dc->dst], cpu_R[dc->src], 4);
break;
case CRISV10_REG_MSTEP:
LOG_DIS("mstep $r%d, $r%d sz=%d\n", dc->src, dc->dst, size);
cris_evaluate_flags(dc);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu(dc, CC_OP_MSTEP, cpu_R[dc->dst],
cpu_R[dc->dst], cpu_R[dc->src], 4);
break;
case CRISV10_REG_SCC:
dec10_reg_scc(dc);
break;
default:
LOG_DIS("pc=%x reg %d r%d r%d\n", dc->pc,
dc->opcode, dc->src, dc->dst);
cpu_abort(CPU(dc->cpu), "Unhandled opcode");
break;
}
}
return insn_len;
}
static unsigned int dec10_ind_move_m_r(CPUCRISState *env, DisasContext *dc,
unsigned int size)
{
unsigned int insn_len = 2;
TCGv t;
LOG_DIS("%s: move.%d [$r%d], $r%d\n", __func__,
size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
t = tcg_temp_new();
insn_len += dec10_prep_move_m(env, dc, 0, size, t);
cris_alu(dc, CC_OP_MOVE, cpu_R[dc->dst], cpu_R[dc->dst], t, size);
if (dc->dst == 15) {
tcg_gen_mov_tl(env_btarget, cpu_R[dc->dst]);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch = 1;
return insn_len;
}
tcg_temp_free(t);
return insn_len;
}
static unsigned int dec10_ind_move_r_m(DisasContext *dc, unsigned int size)
{
unsigned int insn_len = 2;
TCGv addr;
LOG_DIS("move.%d $r%d, [$r%d]\n", dc->size, dc->src, dc->dst);
addr = tcg_temp_new();
crisv10_prepare_memaddr(dc, addr, size);
gen_store_v10(dc, addr, cpu_R[dc->dst], size);
insn_len += crisv10_post_memaddr(dc, size);
return insn_len;
}
static unsigned int dec10_ind_move_m_pr(CPUCRISState *env, DisasContext *dc)
{
unsigned int insn_len = 2, rd = dc->dst;
TCGv t, addr;
LOG_DIS("move.%d $p%d, [$r%d]\n", dc->size, dc->dst, dc->src);
cris_lock_irq(dc);
addr = tcg_temp_new();
t = tcg_temp_new();
insn_len += dec10_prep_move_m(env, dc, 0, 4, t);
if (rd == 15) {
tcg_gen_mov_tl(env_btarget, t);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch = 1;
return insn_len;
}
tcg_gen_mov_tl(cpu_PR[rd], t);
dc->cpustate_changed = 1;
tcg_temp_free(addr);
tcg_temp_free(t);
return insn_len;
}
static unsigned int dec10_ind_move_pr_m(DisasContext *dc)
{
unsigned int insn_len = 2, size = preg_sizes_v10[dc->dst];
TCGv addr, t0;
LOG_DIS("move.%d $p%d, [$r%d]\n", dc->size, dc->dst, dc->src);
addr = tcg_temp_new();
crisv10_prepare_memaddr(dc, addr, size);
if (dc->dst == PR_CCS) {
t0 = tcg_temp_new();
cris_evaluate_flags(dc);
tcg_gen_andi_tl(t0, cpu_PR[PR_CCS], ~PFIX_FLAG);
gen_store_v10(dc, addr, t0, size);
tcg_temp_free(t0);
} else {
gen_store_v10(dc, addr, cpu_PR[dc->dst], size);
}
t0 = tcg_temp_new();
insn_len += crisv10_post_memaddr(dc, size);
cris_lock_irq(dc);
return insn_len;
}
static void dec10_movem_r_m(DisasContext *dc)
{
int i, pfix = dc->tb_flags & PFIX_FLAG;
TCGv addr, t0;
LOG_DIS("%s r%d, [r%d] pi=%d ir=%x\n", __func__,
dc->dst, dc->src, dc->postinc, dc->ir);
addr = tcg_temp_new();
t0 = tcg_temp_new();
crisv10_prepare_memaddr(dc, addr, 4);
tcg_gen_mov_tl(t0, addr);
for (i = dc->dst; i >= 0; i--) {
if ((pfix && dc->mode == CRISV10_MODE_AUTOINC) && dc->src == i) {
gen_store_v10(dc, addr, t0, 4);
} else {
gen_store_v10(dc, addr, cpu_R[i], 4);
}
tcg_gen_addi_tl(addr, addr, 4);
}
if (pfix && dc->mode == CRISV10_MODE_AUTOINC) {
tcg_gen_mov_tl(cpu_R[dc->src], t0);
}
if (!pfix && dc->mode == CRISV10_MODE_AUTOINC) {
tcg_gen_mov_tl(cpu_R[dc->src], addr);
}
tcg_temp_free(addr);
tcg_temp_free(t0);
}
static void dec10_movem_m_r(DisasContext *dc)
{
int i, pfix = dc->tb_flags & PFIX_FLAG;
TCGv addr, t0;
LOG_DIS("%s [r%d], r%d pi=%d ir=%x\n", __func__,
dc->src, dc->dst, dc->postinc, dc->ir);
addr = tcg_temp_new();
t0 = tcg_temp_new();
crisv10_prepare_memaddr(dc, addr, 4);
tcg_gen_mov_tl(t0, addr);
for (i = dc->dst; i >= 0; i--) {
gen_load(dc, cpu_R[i], addr, 4, 0);
tcg_gen_addi_tl(addr, addr, 4);
}
if (pfix && dc->mode == CRISV10_MODE_AUTOINC) {
tcg_gen_mov_tl(cpu_R[dc->src], t0);
}
if (!pfix && dc->mode == CRISV10_MODE_AUTOINC) {
tcg_gen_mov_tl(cpu_R[dc->src], addr);
}
tcg_temp_free(addr);
tcg_temp_free(t0);
}
static int dec10_ind_alu(CPUCRISState *env, DisasContext *dc,
int op, unsigned int size)
{
int insn_len = 0;
int rd = dc->dst;
TCGv t[2];
cris_alu_m_alloc_temps(t);
insn_len += dec10_prep_move_m(env, dc, 0, size, t[0]);
cris_alu(dc, op, cpu_R[dc->dst], cpu_R[rd], t[0], size);
if (dc->dst == 15) {
tcg_gen_mov_tl(env_btarget, cpu_R[dc->dst]);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch = 1;
return insn_len;
}
cris_alu_m_free_temps(t);
return insn_len;
}
static int dec10_ind_bound(CPUCRISState *env, DisasContext *dc,
unsigned int size)
{
int insn_len = 0;
int rd = dc->dst;
TCGv t;
t = tcg_temp_local_new();
insn_len += dec10_prep_move_m(env, dc, 0, size, t);
cris_alu(dc, CC_OP_BOUND, cpu_R[dc->dst], cpu_R[rd], t, 4);
if (dc->dst == 15) {
tcg_gen_mov_tl(env_btarget, cpu_R[dc->dst]);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch = 1;
return insn_len;
}
tcg_temp_free(t);
return insn_len;
}
static int dec10_alux_m(CPUCRISState *env, DisasContext *dc, int op)
{
unsigned int size = (dc->size & 1) ? 2 : 1;
unsigned int sx = !!(dc->size & 2);
int insn_len = 2;
int rd = dc->dst;
TCGv t;
LOG_DIS("addx size=%d sx=%d op=%d %d\n", size, sx, dc->src, dc->dst);
t = tcg_temp_new();
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_prep_move_m(env, dc, sx, size, t);
cris_alu(dc, op, cpu_R[dc->dst], cpu_R[rd], t, 4);
if (dc->dst == 15) {
tcg_gen_mov_tl(env_btarget, cpu_R[dc->dst]);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch = 1;
return insn_len;
}
tcg_temp_free(t);
return insn_len;
}
static int dec10_dip(CPUCRISState *env, DisasContext *dc)
{
int insn_len = 2;
uint32_t imm;
LOG_DIS("dip pc=%x opcode=%d r%d r%d\n",
dc->pc, dc->opcode, dc->src, dc->dst);
if (dc->src == 15) {
imm = cpu_ldl_code(env, dc->pc + 2);
tcg_gen_movi_tl(cpu_PR[PR_PREFIX], imm);
if (dc->postinc)
insn_len += 4;
tcg_gen_addi_tl(cpu_R[15], cpu_R[15], insn_len - 2);
} else {
gen_load(dc, cpu_PR[PR_PREFIX], cpu_R[dc->src], 4, 0);
if (dc->postinc)
tcg_gen_addi_tl(cpu_R[dc->src], cpu_R[dc->src], 4);
}
cris_set_prefix(dc);
return insn_len;
}
static int dec10_bdap_m(CPUCRISState *env, DisasContext *dc, int size)
{
int insn_len = 2;
int rd = dc->dst;
LOG_DIS("bdap_m pc=%x opcode=%d r%d r%d sz=%d\n",
dc->pc, dc->opcode, dc->src, dc->dst, size);
assert(dc->dst != 15);
#if 0
/* 8bit embedded offset? */
if (!dc->postinc && (dc->ir & (1 << 11))) {
int simm = dc->ir & 0xff;
/* cpu_abort(CPU(dc->cpu), "Unhandled opcode"); */
/* sign extended. */
simm = (int8_t)simm;
tcg_gen_addi_tl(cpu_PR[PR_PREFIX], cpu_R[dc->dst], simm);
cris_set_prefix(dc);
return insn_len;
}
#endif
/* Now the rest of the modes are truly indirect. */
insn_len += dec10_prep_move_m(env, dc, 1, size, cpu_PR[PR_PREFIX]);
tcg_gen_add_tl(cpu_PR[PR_PREFIX], cpu_PR[PR_PREFIX], cpu_R[rd]);
cris_set_prefix(dc);
return insn_len;
}
static unsigned int dec10_ind(CPUCRISState *env, DisasContext *dc)
{
unsigned int insn_len = 2;
unsigned int size = dec10_size(dc->size);
uint32_t imm;
int32_t simm;
TCGv t[2];
if (dc->size != 3) {
switch (dc->opcode) {
case CRISV10_IND_MOVE_M_R:
return dec10_ind_move_m_r(env, dc, size);
break;
case CRISV10_IND_MOVE_R_M:
return dec10_ind_move_r_m(dc, size);
break;
case CRISV10_IND_CMP:
LOG_DIS("cmp size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_alu(env, dc, CC_OP_CMP, size);
break;
case CRISV10_IND_TEST:
LOG_DIS("test size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_evaluate_flags(dc);
cris_cc_mask(dc, CC_MASK_NZVC);
cris_alu_m_alloc_temps(t);
insn_len += dec10_prep_move_m(env, dc, 0, size, t[0]);
tcg_gen_andi_tl(cpu_PR[PR_CCS], cpu_PR[PR_CCS], ~3);
cris_alu(dc, CC_OP_CMP, cpu_R[dc->dst],
t[0], tcg_const_tl(0), size);
cris_alu_m_free_temps(t);
break;
case CRISV10_IND_ADD:
LOG_DIS("add size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_alu(env, dc, CC_OP_ADD, size);
break;
case CRISV10_IND_SUB:
LOG_DIS("sub size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_alu(env, dc, CC_OP_SUB, size);
break;
case CRISV10_IND_BOUND:
LOG_DIS("bound size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_bound(env, dc, size);
break;
case CRISV10_IND_AND:
LOG_DIS("and size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_alu(env, dc, CC_OP_AND, size);
break;
case CRISV10_IND_OR:
LOG_DIS("or size=%d op=%d %d\n", size, dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_alu(env, dc, CC_OP_OR, size);
break;
case CRISV10_IND_MOVX:
insn_len = dec10_alux_m(env, dc, CC_OP_MOVE);
break;
case CRISV10_IND_ADDX:
insn_len = dec10_alux_m(env, dc, CC_OP_ADD);
break;
case CRISV10_IND_SUBX:
insn_len = dec10_alux_m(env, dc, CC_OP_SUB);
break;
case CRISV10_IND_CMPX:
insn_len = dec10_alux_m(env, dc, CC_OP_CMP);
break;
case CRISV10_IND_MUL:
/* This is a reg insn coded in the mem indir space. */
LOG_DIS("mul pc=%x opcode=%d\n", dc->pc, dc->opcode);
cris_cc_mask(dc, CC_MASK_NZVC);
dec10_reg_mul(dc, size, dc->ir & (1 << 10));
break;
case CRISV10_IND_BDAP_M:
insn_len = dec10_bdap_m(env, dc, size);
break;
default:
/*
* ADDC for v17:
*
* Instruction format: ADDC [Rs],Rd
*
* +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+-+
* |Destination(Rd)| 1 0 0 1 1 0 1 0 | Source(Rs)|
* +---+---+---+---+---+---+---+---+---+---+---+---+---+---+--+--+
*
* Instruction format: ADDC [Rs+],Rd
*
* +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+-+
* |Destination(Rd)| 1 1 0 1 1 0 1 0 | Source(Rs)|
* +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+-+
*/
if (dc->opcode == CRISV17_IND_ADDC && dc->size == 2 &&
env->pregs[PR_VR] == 17) {
LOG_DIS("addc op=%d %d\n", dc->src, dc->dst);
cris_cc_mask(dc, CC_MASK_NZVC);
insn_len += dec10_ind_alu(env, dc, CC_OP_ADDC, size);
break;
}
LOG_DIS("pc=%x var-ind.%d %d r%d r%d\n",
dc->pc, size, dc->opcode, dc->src, dc->dst);
cpu_abort(CPU(dc->cpu), "Unhandled opcode");
break;
}
return insn_len;
}
switch (dc->opcode) {
case CRISV10_IND_MOVE_M_SPR:
insn_len = dec10_ind_move_m_pr(env, dc);
break;
case CRISV10_IND_MOVE_SPR_M:
insn_len = dec10_ind_move_pr_m(dc);
break;
case CRISV10_IND_JUMP_M:
if (dc->src == 15) {
LOG_DIS("jump.%d %d r%d r%d direct\n", size,
dc->opcode, dc->src, dc->dst);
imm = cpu_ldl_code(env, dc->pc + 2);
if (dc->mode == CRISV10_MODE_AUTOINC)
insn_len += size;
t_gen_mov_preg_TN(dc, dc->dst, tcg_const_tl(dc->pc + insn_len));
dc->jmp_pc = imm;
cris_prepare_jmp(dc, JMP_DIRECT);
dc->delayed_branch--; /* v10 has no dslot here. */
} else {
if (dc->dst == 14) {
LOG_DIS("break %d\n", dc->src);
cris_evaluate_flags(dc);
tcg_gen_movi_tl(env_pc, dc->pc + 2);
t_gen_mov_env_TN(trap_vector, tcg_const_tl(dc->src + 2));
t_gen_raise_exception(EXCP_BREAK);
dc->is_jmp = DISAS_UPDATE;
return insn_len;
}
LOG_DIS("%d: jump.%d %d r%d r%d\n", __LINE__, size,
dc->opcode, dc->src, dc->dst);
t[0] = tcg_temp_new();
t_gen_mov_preg_TN(dc, dc->dst, tcg_const_tl(dc->pc + insn_len));
crisv10_prepare_memaddr(dc, t[0], size);
gen_load(dc, env_btarget, t[0], 4, 0);
insn_len += crisv10_post_memaddr(dc, size);
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch--; /* v10 has no dslot here. */
tcg_temp_free(t[0]);
}
break;
case CRISV10_IND_MOVEM_R_M:
LOG_DIS("movem_r_m pc=%x opcode=%d r%d r%d\n",
dc->pc, dc->opcode, dc->dst, dc->src);
dec10_movem_r_m(dc);
break;
case CRISV10_IND_MOVEM_M_R:
LOG_DIS("movem_m_r pc=%x opcode=%d\n", dc->pc, dc->opcode);
dec10_movem_m_r(dc);
break;
case CRISV10_IND_JUMP_R:
LOG_DIS("jmp pc=%x opcode=%d r%d r%d\n",
dc->pc, dc->opcode, dc->dst, dc->src);
tcg_gen_mov_tl(env_btarget, cpu_R[dc->src]);
t_gen_mov_preg_TN(dc, dc->dst, tcg_const_tl(dc->pc + insn_len));
cris_prepare_jmp(dc, JMP_INDIRECT);
dc->delayed_branch--; /* v10 has no dslot here. */
break;
case CRISV10_IND_MOVX:
insn_len = dec10_alux_m(env, dc, CC_OP_MOVE);
break;
case CRISV10_IND_ADDX:
insn_len = dec10_alux_m(env, dc, CC_OP_ADD);
break;
case CRISV10_IND_SUBX:
insn_len = dec10_alux_m(env, dc, CC_OP_SUB);
break;
case CRISV10_IND_CMPX:
insn_len = dec10_alux_m(env, dc, CC_OP_CMP);
break;
case CRISV10_IND_DIP:
insn_len = dec10_dip(env, dc);
break;
case CRISV10_IND_BCC_M:
cris_cc_mask(dc, 0);
imm = cpu_ldsw_code(env, dc->pc + 2);
simm = (int16_t)imm;
simm += 4;
LOG_DIS("bcc_m: b%s %x\n", cc_name(dc->cond), dc->pc + simm);
cris_prepare_cc_branch(dc, simm, dc->cond);
insn_len = 4;
break;
default:
LOG_DIS("ERROR pc=%x opcode=%d\n", dc->pc, dc->opcode);
cpu_abort(CPU(dc->cpu), "Unhandled opcode");
break;
}
return insn_len;
}
static unsigned int crisv10_decoder(CPUCRISState *env, DisasContext *dc)
{
unsigned int insn_len = 2;
/* Load a halfword onto the instruction register. */
dc->ir = cpu_lduw_code(env, dc->pc);
/* Now decode it. */
dc->opcode = EXTRACT_FIELD(dc->ir, 6, 9);
dc->mode = EXTRACT_FIELD(dc->ir, 10, 11);
dc->src = EXTRACT_FIELD(dc->ir, 0, 3);
dc->size = EXTRACT_FIELD(dc->ir, 4, 5);
dc->cond = dc->dst = EXTRACT_FIELD(dc->ir, 12, 15);
dc->postinc = EXTRACT_FIELD(dc->ir, 10, 10);
dc->clear_prefix = 1;
/* FIXME: What if this insn insn't 2 in length?? */
if (dc->src == 15 || dc->dst == 15)
tcg_gen_movi_tl(cpu_R[15], dc->pc + 2);
switch (dc->mode) {
case CRISV10_MODE_QIMMEDIATE:
insn_len = dec10_quick_imm(dc);
break;
case CRISV10_MODE_REG:
insn_len = dec10_reg(dc);
break;
case CRISV10_MODE_AUTOINC:
case CRISV10_MODE_INDIRECT:
insn_len = dec10_ind(env, dc);
break;
}
if (dc->clear_prefix && dc->tb_flags & PFIX_FLAG) {
dc->tb_flags &= ~PFIX_FLAG;
tcg_gen_andi_tl(cpu_PR[PR_CCS], cpu_PR[PR_CCS], ~PFIX_FLAG);
if (dc->tb_flags != dc->tb->flags) {
dc->cpustate_changed = 1;
}
}
/* CRISv10 locks out interrupts on dslots. */
if (dc->delayed_branch == 2) {
cris_lock_irq(dc);
}
return insn_len;
}
void cris_initialize_crisv10_tcg(void)
{
int i;
cc_x = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_x), "cc_x");
cc_src = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_src), "cc_src");
cc_dest = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_dest),
"cc_dest");
cc_result = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_result),
"cc_result");
cc_op = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_op), "cc_op");
cc_size = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_size),
"cc_size");
cc_mask = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, cc_mask),
"cc_mask");
env_pc = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, pc),
"pc");
env_btarget = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, btarget),
"btarget");
env_btaken = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, btaken),
"btaken");
for (i = 0; i < 16; i++) {
cpu_R[i] = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, regs[i]),
regnames_v10[i]);
}
for (i = 0; i < 16; i++) {
cpu_PR[i] = tcg_global_mem_new(cpu_env,
offsetof(CPUCRISState, pregs[i]),
pregnames_v10[i]);
}
}