#include "cpu.h"
#include "exec/gdbstub.h"
#include "helper.h"
#include "qemu/host-utils.h"
#include "sysemu/sysemu.h"
#include "qemu/bitops.h"

#ifndef CONFIG_USER_ONLY
static inline int get_phys_addr(CPUARMState *env, uint32_t address,
                                int access_type, int is_user,
                                hwaddr *phys_ptr, int *prot,
                                target_ulong *page_size);
#endif

static int vfp_gdb_get_reg(CPUARMState *env, uint8_t *buf, int reg)
{
    int nregs;

    /* VFP data registers are always little-endian.  */
    nregs = arm_feature(env, ARM_FEATURE_VFP3) ? 32 : 16;
    if (reg < nregs) {
        stfq_le_p(buf, env->vfp.regs[reg]);
        return 8;
    }
    if (arm_feature(env, ARM_FEATURE_NEON)) {
        /* Aliases for Q regs.  */
        nregs += 16;
        if (reg < nregs) {
            stfq_le_p(buf, env->vfp.regs[(reg - 32) * 2]);
            stfq_le_p(buf + 8, env->vfp.regs[(reg - 32) * 2 + 1]);
            return 16;
        }
    }
    switch (reg - nregs) {
    case 0: stl_p(buf, env->vfp.xregs[ARM_VFP_FPSID]); return 4;
    case 1: stl_p(buf, env->vfp.xregs[ARM_VFP_FPSCR]); return 4;
    case 2: stl_p(buf, env->vfp.xregs[ARM_VFP_FPEXC]); return 4;
    }
    return 0;
}

static int vfp_gdb_set_reg(CPUARMState *env, uint8_t *buf, int reg)
{
    int nregs;

    nregs = arm_feature(env, ARM_FEATURE_VFP3) ? 32 : 16;
    if (reg < nregs) {
        env->vfp.regs[reg] = ldfq_le_p(buf);
        return 8;
    }
    if (arm_feature(env, ARM_FEATURE_NEON)) {
        nregs += 16;
        if (reg < nregs) {
            env->vfp.regs[(reg - 32) * 2] = ldfq_le_p(buf);
            env->vfp.regs[(reg - 32) * 2 + 1] = ldfq_le_p(buf + 8);
            return 16;
        }
    }
    switch (reg - nregs) {
    case 0: env->vfp.xregs[ARM_VFP_FPSID] = ldl_p(buf); return 4;
    case 1: env->vfp.xregs[ARM_VFP_FPSCR] = ldl_p(buf); return 4;
    case 2: env->vfp.xregs[ARM_VFP_FPEXC] = ldl_p(buf) & (1 << 30); return 4;
    }
    return 0;
}

static int dacr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    env->cp15.c3 = value;
    tlb_flush(env, 1); /* Flush TLB as domain not tracked in TLB */
    return 0;
}

static int fcse_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    if (env->cp15.c13_fcse != value) {
        /* Unlike real hardware the qemu TLB uses virtual addresses,
         * not modified virtual addresses, so this causes a TLB flush.
         */
        tlb_flush(env, 1);
        env->cp15.c13_fcse = value;
    }
    return 0;
}
static int contextidr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    if (env->cp15.c13_context != value && !arm_feature(env, ARM_FEATURE_MPU)) {
        /* For VMSA (when not using the LPAE long descriptor page table
         * format) this register includes the ASID, so do a TLB flush.
         * For PMSA it is purely a process ID and no action is needed.
         */
        tlb_flush(env, 1);
    }
    env->cp15.c13_context = value;
    return 0;
}

static int tlbiall_write(CPUARMState *env, const ARMCPRegInfo *ri,
                         uint64_t value)
{
    /* Invalidate all (TLBIALL) */
    tlb_flush(env, 1);
    return 0;
}

static int tlbimva_write(CPUARMState *env, const ARMCPRegInfo *ri,
                         uint64_t value)
{
    /* Invalidate single TLB entry by MVA and ASID (TLBIMVA) */
    tlb_flush_page(env, value & TARGET_PAGE_MASK);
    return 0;
}

static int tlbiasid_write(CPUARMState *env, const ARMCPRegInfo *ri,
                          uint64_t value)
{
    /* Invalidate by ASID (TLBIASID) */
    tlb_flush(env, value == 0);
    return 0;
}

static int tlbimvaa_write(CPUARMState *env, const ARMCPRegInfo *ri,
                          uint64_t value)
{
    /* Invalidate single entry by MVA, all ASIDs (TLBIMVAA) */
    tlb_flush_page(env, value & TARGET_PAGE_MASK);
    return 0;
}

static const ARMCPRegInfo cp_reginfo[] = {
    /* DBGDIDR: just RAZ. In particular this means the "debug architecture
     * version" bits will read as a reserved value, which should cause
     * Linux to not try to use the debug hardware.
     */
    { .name = "DBGDIDR", .cp = 14, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    /* MMU Domain access control / MPU write buffer control */
    { .name = "DACR", .cp = 15,
      .crn = 3, .crm = CP_ANY, .opc1 = CP_ANY, .opc2 = CP_ANY,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c3),
      .resetvalue = 0, .writefn = dacr_write },
    { .name = "FCSEIDR", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c13_fcse),
      .resetvalue = 0, .writefn = fcse_write },
    { .name = "CONTEXTIDR", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c13_fcse),
      .resetvalue = 0, .writefn = contextidr_write },
    /* ??? This covers not just the impdef TLB lockdown registers but also
     * some v7VMSA registers relating to TEX remap, so it is overly broad.
     */
    { .name = "TLB_LOCKDOWN", .cp = 15, .crn = 10, .crm = CP_ANY,
      .opc1 = CP_ANY, .opc2 = CP_ANY, .access = PL1_RW, .type = ARM_CP_NOP },
    /* MMU TLB control. Note that the wildcarding means we cover not just
     * the unified TLB ops but also the dside/iside/inner-shareable variants.
     */
    { .name = "TLBIALL", .cp = 15, .crn = 8, .crm = CP_ANY,
      .opc1 = CP_ANY, .opc2 = 0, .access = PL1_W, .writefn = tlbiall_write, },
    { .name = "TLBIMVA", .cp = 15, .crn = 8, .crm = CP_ANY,
      .opc1 = CP_ANY, .opc2 = 1, .access = PL1_W, .writefn = tlbimva_write, },
    { .name = "TLBIASID", .cp = 15, .crn = 8, .crm = CP_ANY,
      .opc1 = CP_ANY, .opc2 = 2, .access = PL1_W, .writefn = tlbiasid_write, },
    { .name = "TLBIMVAA", .cp = 15, .crn = 8, .crm = CP_ANY,
      .opc1 = CP_ANY, .opc2 = 3, .access = PL1_W, .writefn = tlbimvaa_write, },
    /* Cache maintenance ops; some of this space may be overridden later. */
    { .name = "CACHEMAINT", .cp = 15, .crn = 7, .crm = CP_ANY,
      .opc1 = 0, .opc2 = CP_ANY, .access = PL1_W,
      .type = ARM_CP_NOP | ARM_CP_OVERRIDE },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo not_v6_cp_reginfo[] = {
    /* Not all pre-v6 cores implemented this WFI, so this is slightly
     * over-broad.
     */
    { .name = "WFI_v5", .cp = 15, .crn = 7, .crm = 8, .opc1 = 0, .opc2 = 2,
      .access = PL1_W, .type = ARM_CP_WFI },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo not_v7_cp_reginfo[] = {
    /* Standard v6 WFI (also used in some pre-v6 cores); not in v7 (which
     * is UNPREDICTABLE; we choose to NOP as most implementations do).
     */
    { .name = "WFI_v6", .cp = 15, .crn = 7, .crm = 0, .opc1 = 0, .opc2 = 4,
      .access = PL1_W, .type = ARM_CP_WFI },
    /* L1 cache lockdown. Not architectural in v6 and earlier but in practice
     * implemented in 926, 946, 1026, 1136, 1176 and 11MPCore. StrongARM and
     * OMAPCP will override this space.
     */
    { .name = "DLOCKDOWN", .cp = 15, .crn = 9, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c9_data),
      .resetvalue = 0 },
    { .name = "ILOCKDOWN", .cp = 15, .crn = 9, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c9_insn),
      .resetvalue = 0 },
    /* v6 doesn't have the cache ID registers but Linux reads them anyway */
    { .name = "DUMMY", .cp = 15, .crn = 0, .crm = 0, .opc1 = 1, .opc2 = CP_ANY,
      .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    REGINFO_SENTINEL
};

static int cpacr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    if (env->cp15.c1_coproc != value) {
        env->cp15.c1_coproc = value;
        /* ??? Is this safe when called from within a TB?  */
        tb_flush(env);
    }
    return 0;
}

static const ARMCPRegInfo v6_cp_reginfo[] = {
    /* prefetch by MVA in v6, NOP in v7 */
    { .name = "MVA_prefetch",
      .cp = 15, .crn = 7, .crm = 13, .opc1 = 0, .opc2 = 1,
      .access = PL1_W, .type = ARM_CP_NOP },
    { .name = "ISB", .cp = 15, .crn = 7, .crm = 5, .opc1 = 0, .opc2 = 4,
      .access = PL0_W, .type = ARM_CP_NOP },
    { .name = "DSB", .cp = 15, .crn = 7, .crm = 10, .opc1 = 0, .opc2 = 4,
      .access = PL0_W, .type = ARM_CP_NOP },
    { .name = "DMB", .cp = 15, .crn = 7, .crm = 10, .opc1 = 0, .opc2 = 5,
      .access = PL0_W, .type = ARM_CP_NOP },
    { .name = "IFAR", .cp = 15, .crn = 6, .crm = 0, .opc1 = 0, .opc2 = 2,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c6_insn),
      .resetvalue = 0, },
    /* Watchpoint Fault Address Register : should actually only be present
     * for 1136, 1176, 11MPCore.
     */
    { .name = "WFAR", .cp = 15, .crn = 6, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0, },
    { .name = "CPACR", .cp = 15, .crn = 1, .crm = 0, .opc1 = 0, .opc2 = 2,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c1_coproc),
      .resetvalue = 0, .writefn = cpacr_write },
    REGINFO_SENTINEL
};

static int pmreg_read(CPUARMState *env, const ARMCPRegInfo *ri,
                      uint64_t *value)
{
    /* Generic performance monitor register read function for where
     * user access may be allowed by PMUSERENR.
     */
    if (arm_current_pl(env) == 0 && !env->cp15.c9_pmuserenr) {
        return EXCP_UDEF;
    }
    *value = CPREG_FIELD32(env, ri);
    return 0;
}

static int pmcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                      uint64_t value)
{
    if (arm_current_pl(env) == 0 && !env->cp15.c9_pmuserenr) {
        return EXCP_UDEF;
    }
    /* only the DP, X, D and E bits are writable */
    env->cp15.c9_pmcr &= ~0x39;
    env->cp15.c9_pmcr |= (value & 0x39);
    return 0;
}

static int pmcntenset_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    if (arm_current_pl(env) == 0 && !env->cp15.c9_pmuserenr) {
        return EXCP_UDEF;
    }
    value &= (1 << 31);
    env->cp15.c9_pmcnten |= value;
    return 0;
}

static int pmcntenclr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    if (arm_current_pl(env) == 0 && !env->cp15.c9_pmuserenr) {
        return EXCP_UDEF;
    }
    value &= (1 << 31);
    env->cp15.c9_pmcnten &= ~value;
    return 0;
}

static int pmovsr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t value)
{
    if (arm_current_pl(env) == 0 && !env->cp15.c9_pmuserenr) {
        return EXCP_UDEF;
    }
    env->cp15.c9_pmovsr &= ~value;
    return 0;
}

static int pmxevtyper_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    if (arm_current_pl(env) == 0 && !env->cp15.c9_pmuserenr) {
        return EXCP_UDEF;
    }
    env->cp15.c9_pmxevtyper = value & 0xff;
    return 0;
}

static int pmuserenr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    env->cp15.c9_pmuserenr = value & 1;
    return 0;
}

static int pmintenset_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    /* We have no event counters so only the C bit can be changed */
    value &= (1 << 31);
    env->cp15.c9_pminten |= value;
    return 0;
}

static int pmintenclr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    value &= (1 << 31);
    env->cp15.c9_pminten &= ~value;
    return 0;
}

static int ccsidr_read(CPUARMState *env, const ARMCPRegInfo *ri,
                       uint64_t *value)
{
    ARMCPU *cpu = arm_env_get_cpu(env);
    *value = cpu->ccsidr[env->cp15.c0_cssel];
    return 0;
}

static int csselr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t value)
{
    env->cp15.c0_cssel = value & 0xf;
    return 0;
}

static const ARMCPRegInfo v7_cp_reginfo[] = {
    /* DBGDRAR, DBGDSAR: always RAZ since we don't implement memory mapped
     * debug components
     */
    { .name = "DBGDRAR", .cp = 14, .crn = 1, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    { .name = "DBGDSAR", .cp = 14, .crn = 2, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    /* the old v6 WFI, UNPREDICTABLE in v7 but we choose to NOP */
    { .name = "NOP", .cp = 15, .crn = 7, .crm = 0, .opc1 = 0, .opc2 = 4,
      .access = PL1_W, .type = ARM_CP_NOP },
    /* Performance monitors are implementation defined in v7,
     * but with an ARM recommended set of registers, which we
     * follow (although we don't actually implement any counters)
     *
     * Performance registers fall into three categories:
     *  (a) always UNDEF in PL0, RW in PL1 (PMINTENSET, PMINTENCLR)
     *  (b) RO in PL0 (ie UNDEF on write), RW in PL1 (PMUSERENR)
     *  (c) UNDEF in PL0 if PMUSERENR.EN==0, otherwise accessible (all others)
     * For the cases controlled by PMUSERENR we must set .access to PL0_RW
     * or PL0_RO as appropriate and then check PMUSERENR in the helper fn.
     */
    { .name = "PMCNTENSET", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 1,
      .access = PL0_RW, .resetvalue = 0,
      .fieldoffset = offsetof(CPUARMState, cp15.c9_pmcnten),
      .readfn = pmreg_read, .writefn = pmcntenset_write },
    { .name = "PMCNTENCLR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 2,
      .access = PL0_RW, .fieldoffset = offsetof(CPUARMState, cp15.c9_pmcnten),
      .readfn = pmreg_read, .writefn = pmcntenclr_write },
    { .name = "PMOVSR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 3,
      .access = PL0_RW, .fieldoffset = offsetof(CPUARMState, cp15.c9_pmovsr),
      .readfn = pmreg_read, .writefn = pmovsr_write },
    /* Unimplemented so WI. Strictly speaking write accesses in PL0 should
     * respect PMUSERENR.
     */
    { .name = "PMSWINC", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 4,
      .access = PL0_W, .type = ARM_CP_NOP },
    /* Since we don't implement any events, writing to PMSELR is UNPREDICTABLE.
     * We choose to RAZ/WI. XXX should respect PMUSERENR.
     */
    { .name = "PMSELR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 5,
      .access = PL0_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
    /* Unimplemented, RAZ/WI. XXX PMUSERENR */
    { .name = "PMCCNTR", .cp = 15, .crn = 9, .crm = 13, .opc1 = 0, .opc2 = 0,
      .access = PL0_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
    { .name = "PMXEVTYPER", .cp = 15, .crn = 9, .crm = 13, .opc1 = 0, .opc2 = 1,
      .access = PL0_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c9_pmxevtyper),
      .readfn = pmreg_read, .writefn = pmxevtyper_write },
    /* Unimplemented, RAZ/WI. XXX PMUSERENR */
    { .name = "PMXEVCNTR", .cp = 15, .crn = 9, .crm = 13, .opc1 = 0, .opc2 = 2,
      .access = PL0_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
    { .name = "PMUSERENR", .cp = 15, .crn = 9, .crm = 14, .opc1 = 0, .opc2 = 0,
      .access = PL0_R | PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c9_pmuserenr),
      .resetvalue = 0,
      .writefn = pmuserenr_write },
    { .name = "PMINTENSET", .cp = 15, .crn = 9, .crm = 14, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c9_pminten),
      .resetvalue = 0,
      .writefn = pmintenset_write },
    { .name = "PMINTENCLR", .cp = 15, .crn = 9, .crm = 14, .opc1 = 0, .opc2 = 2,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c9_pminten),
      .resetvalue = 0,
      .writefn = pmintenclr_write },
    { .name = "SCR", .cp = 15, .crn = 1, .crm = 1, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c1_scr),
      .resetvalue = 0, },
    { .name = "CCSIDR", .cp = 15, .crn = 0, .crm = 0, .opc1 = 1, .opc2 = 0,
      .access = PL1_R, .readfn = ccsidr_read },
    { .name = "CSSELR", .cp = 15, .crn = 0, .crm = 0, .opc1 = 2, .opc2 = 0,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c0_cssel),
      .writefn = csselr_write, .resetvalue = 0 },
    /* Auxiliary ID register: this actually has an IMPDEF value but for now
     * just RAZ for all cores:
     */
    { .name = "AIDR", .cp = 15, .crn = 0, .crm = 0, .opc1 = 1, .opc2 = 7,
      .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    REGINFO_SENTINEL
};

static int teecr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    value &= 1;
    env->teecr = value;
    return 0;
}

static int teehbr_read(CPUARMState *env, const ARMCPRegInfo *ri,
                       uint64_t *value)
{
    /* This is a helper function because the user access rights
     * depend on the value of the TEECR.
     */
    if (arm_current_pl(env) == 0 && (env->teecr & 1)) {
        return EXCP_UDEF;
    }
    *value = env->teehbr;
    return 0;
}

static int teehbr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t value)
{
    if (arm_current_pl(env) == 0 && (env->teecr & 1)) {
        return EXCP_UDEF;
    }
    env->teehbr = value;
    return 0;
}

static const ARMCPRegInfo t2ee_cp_reginfo[] = {
    { .name = "TEECR", .cp = 14, .crn = 0, .crm = 0, .opc1 = 6, .opc2 = 0,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, teecr),
      .resetvalue = 0,
      .writefn = teecr_write },
    { .name = "TEEHBR", .cp = 14, .crn = 1, .crm = 0, .opc1 = 6, .opc2 = 0,
      .access = PL0_RW, .fieldoffset = offsetof(CPUARMState, teehbr),
      .resetvalue = 0,
      .readfn = teehbr_read, .writefn = teehbr_write },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo v6k_cp_reginfo[] = {
    { .name = "TPIDRURW", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 2,
      .access = PL0_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c13_tls1),
      .resetvalue = 0 },
    { .name = "TPIDRURO", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 3,
      .access = PL0_R|PL1_W,
      .fieldoffset = offsetof(CPUARMState, cp15.c13_tls2),
      .resetvalue = 0 },
    { .name = "TPIDRPRW", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 4,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c13_tls3),
      .resetvalue = 0 },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo generic_timer_cp_reginfo[] = {
    /* Dummy implementation: RAZ/WI the whole crn=14 space */
    { .name = "GENERIC_TIMER", .cp = 15, .crn = 14,
      .crm = CP_ANY, .opc1 = CP_ANY, .opc2 = CP_ANY,
      .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
    REGINFO_SENTINEL
};

static int par_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    if (arm_feature(env, ARM_FEATURE_LPAE)) {
        env->cp15.c7_par = value;
    } else if (arm_feature(env, ARM_FEATURE_V7)) {
        env->cp15.c7_par = value & 0xfffff6ff;
    } else {
        env->cp15.c7_par = value & 0xfffff1ff;
    }
    return 0;
}

#ifndef CONFIG_USER_ONLY
/* get_phys_addr() isn't present for user-mode-only targets */

/* Return true if extended addresses are enabled, ie this is an
 * LPAE implementation and we are using the long-descriptor translation
 * table format because the TTBCR EAE bit is set.
 */
static inline bool extended_addresses_enabled(CPUARMState *env)
{
    return arm_feature(env, ARM_FEATURE_LPAE)
        && (env->cp15.c2_control & (1 << 31));
}

static int ats_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    hwaddr phys_addr;
    target_ulong page_size;
    int prot;
    int ret, is_user = ri->opc2 & 2;
    int access_type = ri->opc2 & 1;

    if (ri->opc2 & 4) {
        /* Other states are only available with TrustZone */
        return EXCP_UDEF;
    }
    ret = get_phys_addr(env, value, access_type, is_user,
                        &phys_addr, &prot, &page_size);
    if (extended_addresses_enabled(env)) {
        /* ret is a DFSR/IFSR value for the long descriptor
         * translation table format, but with WnR always clear.
         * Convert it to a 64-bit PAR.
         */
        uint64_t par64 = (1 << 11); /* LPAE bit always set */
        if (ret == 0) {
            par64 |= phys_addr & ~0xfffULL;
            /* We don't set the ATTR or SH fields in the PAR. */
        } else {
            par64 |= 1; /* F */
            par64 |= (ret & 0x3f) << 1; /* FS */
            /* Note that S2WLK and FSTAGE are always zero, because we don't
             * implement virtualization and therefore there can't be a stage 2
             * fault.
             */
        }
        env->cp15.c7_par = par64;
        env->cp15.c7_par_hi = par64 >> 32;
    } else {
        /* ret is a DFSR/IFSR value for the short descriptor
         * translation table format (with WnR always clear).
         * Convert it to a 32-bit PAR.
         */
        if (ret == 0) {
            /* We do not set any attribute bits in the PAR */
            if (page_size == (1 << 24)
                && arm_feature(env, ARM_FEATURE_V7)) {
                env->cp15.c7_par = (phys_addr & 0xff000000) | 1 << 1;
            } else {
                env->cp15.c7_par = phys_addr & 0xfffff000;
            }
        } else {
            env->cp15.c7_par = ((ret & (10 << 1)) >> 5) |
                ((ret & (12 << 1)) >> 6) |
                ((ret & 0xf) << 1) | 1;
        }
        env->cp15.c7_par_hi = 0;
    }
    return 0;
}
#endif

static const ARMCPRegInfo vapa_cp_reginfo[] = {
    { .name = "PAR", .cp = 15, .crn = 7, .crm = 4, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .resetvalue = 0,
      .fieldoffset = offsetof(CPUARMState, cp15.c7_par),
      .writefn = par_write },
#ifndef CONFIG_USER_ONLY
    { .name = "ATS", .cp = 15, .crn = 7, .crm = 8, .opc1 = 0, .opc2 = CP_ANY,
      .access = PL1_W, .writefn = ats_write },
#endif
    REGINFO_SENTINEL
};

/* Return basic MPU access permission bits.  */
static uint32_t simple_mpu_ap_bits(uint32_t val)
{
    uint32_t ret;
    uint32_t mask;
    int i;
    ret = 0;
    mask = 3;
    for (i = 0; i < 16; i += 2) {
        ret |= (val >> i) & mask;
        mask <<= 2;
    }
    return ret;
}

/* Pad basic MPU access permission bits to extended format.  */
static uint32_t extended_mpu_ap_bits(uint32_t val)
{
    uint32_t ret;
    uint32_t mask;
    int i;
    ret = 0;
    mask = 3;
    for (i = 0; i < 16; i += 2) {
        ret |= (val & mask) << i;
        mask <<= 2;
    }
    return ret;
}

static int pmsav5_data_ap_write(CPUARMState *env, const ARMCPRegInfo *ri,
                                uint64_t value)
{
    env->cp15.c5_data = extended_mpu_ap_bits(value);
    return 0;
}

static int pmsav5_data_ap_read(CPUARMState *env, const ARMCPRegInfo *ri,
                               uint64_t *value)
{
    *value = simple_mpu_ap_bits(env->cp15.c5_data);
    return 0;
}

static int pmsav5_insn_ap_write(CPUARMState *env, const ARMCPRegInfo *ri,
                                uint64_t value)
{
    env->cp15.c5_insn = extended_mpu_ap_bits(value);
    return 0;
}

static int pmsav5_insn_ap_read(CPUARMState *env, const ARMCPRegInfo *ri,
                               uint64_t *value)
{
    *value = simple_mpu_ap_bits(env->cp15.c5_insn);
    return 0;
}

static int arm946_prbs_read(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t *value)
{
    if (ri->crm >= 8) {
        return EXCP_UDEF;
    }
    *value = env->cp15.c6_region[ri->crm];
    return 0;
}

static int arm946_prbs_write(CPUARMState *env, const ARMCPRegInfo *ri,
                             uint64_t value)
{
    if (ri->crm >= 8) {
        return EXCP_UDEF;
    }
    env->cp15.c6_region[ri->crm] = value;
    return 0;
}

static const ARMCPRegInfo pmsav5_cp_reginfo[] = {
    { .name = "DATA_AP", .cp = 15, .crn = 5, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_data), .resetvalue = 0,
      .readfn = pmsav5_data_ap_read, .writefn = pmsav5_data_ap_write, },
    { .name = "INSN_AP", .cp = 15, .crn = 5, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_insn), .resetvalue = 0,
      .readfn = pmsav5_insn_ap_read, .writefn = pmsav5_insn_ap_write, },
    { .name = "DATA_EXT_AP", .cp = 15, .crn = 5, .crm = 0, .opc1 = 0, .opc2 = 2,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_data), .resetvalue = 0, },
    { .name = "INSN_EXT_AP", .cp = 15, .crn = 5, .crm = 0, .opc1 = 0, .opc2 = 3,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_insn), .resetvalue = 0, },
    { .name = "DCACHE_CFG", .cp = 15, .crn = 2, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c2_data), .resetvalue = 0, },
    { .name = "ICACHE_CFG", .cp = 15, .crn = 2, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c2_insn), .resetvalue = 0, },
    /* Protection region base and size registers */
    { .name = "946_PRBS", .cp = 15, .crn = 6, .crm = CP_ANY, .opc1 = 0,
      .opc2 = CP_ANY, .access = PL1_RW,
      .readfn = arm946_prbs_read, .writefn = arm946_prbs_write, },
    REGINFO_SENTINEL
};

static int vmsa_ttbcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
                            uint64_t value)
{
    if (arm_feature(env, ARM_FEATURE_LPAE)) {
        value &= ~((7 << 19) | (3 << 14) | (0xf << 3));
        /* With LPAE the TTBCR could result in a change of ASID
         * via the TTBCR.A1 bit, so do a TLB flush.
         */
        tlb_flush(env, 1);
    } else {
        value &= 7;
    }
    /* Note that we always calculate c2_mask and c2_base_mask, but
     * they are only used for short-descriptor tables (ie if EAE is 0);
     * for long-descriptor tables the TTBCR fields are used differently
     * and the c2_mask and c2_base_mask values are meaningless.
     */
    env->cp15.c2_control = value;
    env->cp15.c2_mask = ~(((uint32_t)0xffffffffu) >> value);
    env->cp15.c2_base_mask = ~((uint32_t)0x3fffu >> value);
    return 0;
}

static void vmsa_ttbcr_reset(CPUARMState *env, const ARMCPRegInfo *ri)
{
    env->cp15.c2_base_mask = 0xffffc000u;
    env->cp15.c2_control = 0;
    env->cp15.c2_mask = 0;
}

static const ARMCPRegInfo vmsa_cp_reginfo[] = {
    { .name = "DFSR", .cp = 15, .crn = 5, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_data), .resetvalue = 0, },
    { .name = "IFSR", .cp = 15, .crn = 5, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_insn), .resetvalue = 0, },
    { .name = "TTBR0", .cp = 15, .crn = 2, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c2_base0), .resetvalue = 0, },
    { .name = "TTBR1", .cp = 15, .crn = 2, .crm = 0, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c2_base1), .resetvalue = 0, },
    { .name = "TTBCR", .cp = 15, .crn = 2, .crm = 0, .opc1 = 0, .opc2 = 2,
      .access = PL1_RW, .writefn = vmsa_ttbcr_write,
      .resetfn = vmsa_ttbcr_reset,
      .fieldoffset = offsetof(CPUARMState, cp15.c2_control) },
    { .name = "DFAR", .cp = 15, .crn = 6, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c6_data),
      .resetvalue = 0, },
    REGINFO_SENTINEL
};

static int omap_ticonfig_write(CPUARMState *env, const ARMCPRegInfo *ri,
                               uint64_t value)
{
    env->cp15.c15_ticonfig = value & 0xe7;
    /* The OS_TYPE bit in this register changes the reported CPUID! */
    env->cp15.c0_cpuid = (value & (1 << 5)) ?
        ARM_CPUID_TI915T : ARM_CPUID_TI925T;
    return 0;
}

static int omap_threadid_write(CPUARMState *env, const ARMCPRegInfo *ri,
                               uint64_t value)
{
    env->cp15.c15_threadid = value & 0xffff;
    return 0;
}

static int omap_wfi_write(CPUARMState *env, const ARMCPRegInfo *ri,
                          uint64_t value)
{
    /* Wait-for-interrupt (deprecated) */
    cpu_interrupt(env, CPU_INTERRUPT_HALT);
    return 0;
}

static int omap_cachemaint_write(CPUARMState *env, const ARMCPRegInfo *ri,
                                 uint64_t value)
{
    /* On OMAP there are registers indicating the max/min index of dcache lines
     * containing a dirty line; cache flush operations have to reset these.
     */
    env->cp15.c15_i_max = 0x000;
    env->cp15.c15_i_min = 0xff0;
    return 0;
}

static const ARMCPRegInfo omap_cp_reginfo[] = {
    { .name = "DFSR", .cp = 15, .crn = 5, .crm = CP_ANY,
      .opc1 = CP_ANY, .opc2 = CP_ANY, .access = PL1_RW, .type = ARM_CP_OVERRIDE,
      .fieldoffset = offsetof(CPUARMState, cp15.c5_data), .resetvalue = 0, },
    { .name = "", .cp = 15, .crn = 15, .crm = 0, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .type = ARM_CP_NOP },
    { .name = "TICONFIG", .cp = 15, .crn = 15, .crm = 1, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c15_ticonfig), .resetvalue = 0,
      .writefn = omap_ticonfig_write },
    { .name = "IMAX", .cp = 15, .crn = 15, .crm = 2, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c15_i_max), .resetvalue = 0, },
    { .name = "IMIN", .cp = 15, .crn = 15, .crm = 3, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .resetvalue = 0xff0,
      .fieldoffset = offsetof(CPUARMState, cp15.c15_i_min) },
    { .name = "THREADID", .cp = 15, .crn = 15, .crm = 4, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c15_threadid), .resetvalue = 0,
      .writefn = omap_threadid_write },
    { .name = "TI925T_STATUS", .cp = 15, .crn = 15,
      .crm = 8, .opc1 = 0, .opc2 = 0, .access = PL1_RW,
      .readfn = arm_cp_read_zero, .writefn = omap_wfi_write, },
    /* TODO: Peripheral port remap register:
     * On OMAP2 mcr p15, 0, rn, c15, c2, 4 sets up the interrupt controller
     * base address at $rn & ~0xfff and map size of 0x200 << ($rn & 0xfff),
     * when MMU is off.
     */
    { .name = "OMAP_CACHEMAINT", .cp = 15, .crn = 7, .crm = CP_ANY,
      .opc1 = 0, .opc2 = CP_ANY, .access = PL1_W, .type = ARM_CP_OVERRIDE,
      .writefn = omap_cachemaint_write },
    { .name = "C9", .cp = 15, .crn = 9,
      .crm = CP_ANY, .opc1 = CP_ANY, .opc2 = CP_ANY, .access = PL1_RW,
      .type = ARM_CP_CONST | ARM_CP_OVERRIDE, .resetvalue = 0 },
    REGINFO_SENTINEL
};

static int xscale_cpar_write(CPUARMState *env, const ARMCPRegInfo *ri,
                             uint64_t value)
{
    value &= 0x3fff;
    if (env->cp15.c15_cpar != value) {
        /* Changes cp0 to cp13 behavior, so needs a TB flush.  */
        tb_flush(env);
        env->cp15.c15_cpar = value;
    }
    return 0;
}

static const ARMCPRegInfo xscale_cp_reginfo[] = {
    { .name = "XSCALE_CPAR",
      .cp = 15, .crn = 15, .crm = 1, .opc1 = 0, .opc2 = 0, .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c15_cpar), .resetvalue = 0,
      .writefn = xscale_cpar_write, },
    { .name = "XSCALE_AUXCR",
      .cp = 15, .crn = 1, .crm = 0, .opc1 = 0, .opc2 = 1, .access = PL1_RW,
      .fieldoffset = offsetof(CPUARMState, cp15.c1_xscaleauxcr),
      .resetvalue = 0, },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo dummy_c15_cp_reginfo[] = {
    /* RAZ/WI the whole crn=15 space, when we don't have a more specific
     * implementation of this implementation-defined space.
     * Ideally this should eventually disappear in favour of actually
     * implementing the correct behaviour for all cores.
     */
    { .name = "C15_IMPDEF", .cp = 15, .crn = 15,
      .crm = CP_ANY, .opc1 = CP_ANY, .opc2 = CP_ANY,
      .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo cache_dirty_status_cp_reginfo[] = {
    /* Cache status: RAZ because we have no cache so it's always clean */
    { .name = "CDSR", .cp = 15, .crn = 7, .crm = 10, .opc1 = 0, .opc2 = 6,
      .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo cache_block_ops_cp_reginfo[] = {
    /* We never have a a block transfer operation in progress */
    { .name = "BXSR", .cp = 15, .crn = 7, .crm = 12, .opc1 = 0, .opc2 = 4,
      .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = 0 },
    /* The cache ops themselves: these all NOP for QEMU */
    { .name = "IICR", .cp = 15, .crm = 5, .opc1 = 0,
      .access = PL1_W, .type = ARM_CP_NOP|ARM_CP_64BIT },
    { .name = "IDCR", .cp = 15, .crm = 6, .opc1 = 0,
      .access = PL1_W, .type = ARM_CP_NOP|ARM_CP_64BIT },
    { .name = "CDCR", .cp = 15, .crm = 12, .opc1 = 0,
      .access = PL0_W, .type = ARM_CP_NOP|ARM_CP_64BIT },
    { .name = "PIR", .cp = 15, .crm = 12, .opc1 = 1,
      .access = PL0_W, .type = ARM_CP_NOP|ARM_CP_64BIT },
    { .name = "PDR", .cp = 15, .crm = 12, .opc1 = 2,
      .access = PL0_W, .type = ARM_CP_NOP|ARM_CP_64BIT },
    { .name = "CIDCR", .cp = 15, .crm = 14, .opc1 = 0,
      .access = PL1_W, .type = ARM_CP_NOP|ARM_CP_64BIT },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo cache_test_clean_cp_reginfo[] = {
    /* The cache test-and-clean instructions always return (1 << 30)
     * to indicate that there are no dirty cache lines.
     */
    { .name = "TC_DCACHE", .cp = 15, .crn = 7, .crm = 10, .opc1 = 0, .opc2 = 3,
      .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = (1 << 30) },
    { .name = "TCI_DCACHE", .cp = 15, .crn = 7, .crm = 14, .opc1 = 0, .opc2 = 3,
      .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = (1 << 30) },
    REGINFO_SENTINEL
};

static const ARMCPRegInfo strongarm_cp_reginfo[] = {
    /* Ignore ReadBuffer accesses */
    { .name = "C9_READBUFFER", .cp = 15, .crn = 9,
      .crm = CP_ANY, .opc1 = CP_ANY, .opc2 = CP_ANY,
      .access = PL1_RW, .type = ARM_CP_CONST | ARM_CP_OVERRIDE,
      .resetvalue = 0 },
    REGINFO_SENTINEL
};

static int mpidr_read(CPUARMState *env, const ARMCPRegInfo *ri,
                      uint64_t *value)
{
    CPUState *cs = CPU(arm_env_get_cpu(env));
    uint32_t mpidr = cs->cpu_index;
    /* We don't support setting cluster ID ([8..11])
     * so these bits always RAZ.
     */
    if (arm_feature(env, ARM_FEATURE_V7MP)) {
        mpidr |= (1 << 31);
        /* Cores which are uniprocessor (non-coherent)
         * but still implement the MP extensions set
         * bit 30. (For instance, A9UP.) However we do
         * not currently model any of those cores.
         */
    }
    *value = mpidr;
    return 0;
}

static const ARMCPRegInfo mpidr_cp_reginfo[] = {
    { .name = "MPIDR", .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 5,
      .access = PL1_R, .readfn = mpidr_read },
    REGINFO_SENTINEL
};

static int par64_read(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t *value)
{
    *value = ((uint64_t)env->cp15.c7_par_hi << 32) | env->cp15.c7_par;
    return 0;
}

static int par64_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    env->cp15.c7_par_hi = value >> 32;
    env->cp15.c7_par = value;
    return 0;
}

static void par64_reset(CPUARMState *env, const ARMCPRegInfo *ri)
{
    env->cp15.c7_par_hi = 0;
    env->cp15.c7_par = 0;
}

static int ttbr064_read(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t *value)
{
    *value = ((uint64_t)env->cp15.c2_base0_hi << 32) | env->cp15.c2_base0;
    return 0;
}

static int ttbr064_write(CPUARMState *env, const ARMCPRegInfo *ri,
                         uint64_t value)
{
    env->cp15.c2_base0_hi = value >> 32;
    env->cp15.c2_base0 = value;
    /* Writes to the 64 bit format TTBRs may change the ASID */
    tlb_flush(env, 1);
    return 0;
}

static void ttbr064_reset(CPUARMState *env, const ARMCPRegInfo *ri)
{
    env->cp15.c2_base0_hi = 0;
    env->cp15.c2_base0 = 0;
}

static int ttbr164_read(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t *value)
{
    *value = ((uint64_t)env->cp15.c2_base1_hi << 32) | env->cp15.c2_base1;
    return 0;
}

static int ttbr164_write(CPUARMState *env, const ARMCPRegInfo *ri,
                         uint64_t value)
{
    env->cp15.c2_base1_hi = value >> 32;
    env->cp15.c2_base1 = value;
    return 0;
}

static void ttbr164_reset(CPUARMState *env, const ARMCPRegInfo *ri)
{
    env->cp15.c2_base1_hi = 0;
    env->cp15.c2_base1 = 0;
}

static const ARMCPRegInfo lpae_cp_reginfo[] = {
    /* NOP AMAIR0/1: the override is because these clash with the rather
     * broadly specified TLB_LOCKDOWN entry in the generic cp_reginfo.
     */
    { .name = "AMAIR0", .cp = 15, .crn = 10, .crm = 3, .opc1 = 0, .opc2 = 0,
      .access = PL1_RW, .type = ARM_CP_CONST | ARM_CP_OVERRIDE,
      .resetvalue = 0 },
    { .name = "AMAIR1", .cp = 15, .crn = 10, .crm = 3, .opc1 = 0, .opc2 = 1,
      .access = PL1_RW, .type = ARM_CP_CONST | ARM_CP_OVERRIDE,
      .resetvalue = 0 },
    /* 64 bit access versions of the (dummy) debug registers */
    { .name = "DBGDRAR", .cp = 14, .crm = 1, .opc1 = 0,
      .access = PL0_R, .type = ARM_CP_CONST|ARM_CP_64BIT, .resetvalue = 0 },
    { .name = "DBGDSAR", .cp = 14, .crm = 2, .opc1 = 0,
      .access = PL0_R, .type = ARM_CP_CONST|ARM_CP_64BIT, .resetvalue = 0 },
    { .name = "PAR", .cp = 15, .crm = 7, .opc1 = 0,
      .access = PL1_RW, .type = ARM_CP_64BIT,
      .readfn = par64_read, .writefn = par64_write, .resetfn = par64_reset },
    { .name = "TTBR0", .cp = 15, .crm = 2, .opc1 = 0,
      .access = PL1_RW, .type = ARM_CP_64BIT, .readfn = ttbr064_read,
      .writefn = ttbr064_write, .resetfn = ttbr064_reset },
    { .name = "TTBR1", .cp = 15, .crm = 2, .opc1 = 1,
      .access = PL1_RW, .type = ARM_CP_64BIT, .readfn = ttbr164_read,
      .writefn = ttbr164_write, .resetfn = ttbr164_reset },
    REGINFO_SENTINEL
};

static int sctlr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value)
{
    env->cp15.c1_sys = value;
    /* ??? Lots of these bits are not implemented.  */
    /* This may enable/disable the MMU, so do a TLB flush.  */
    tlb_flush(env, 1);
    return 0;
}

void register_cp_regs_for_features(ARMCPU *cpu)
{
    /* Register all the coprocessor registers based on feature bits */
    CPUARMState *env = &cpu->env;
    if (arm_feature(env, ARM_FEATURE_M)) {
        /* M profile has no coprocessor registers */
        return;
    }

    define_arm_cp_regs(cpu, cp_reginfo);
    if (arm_feature(env, ARM_FEATURE_V6)) {
        /* The ID registers all have impdef reset values */
        ARMCPRegInfo v6_idregs[] = {
            { .name = "ID_PFR0", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 0, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_pfr0 },
            { .name = "ID_PFR1", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 1, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_pfr1 },
            { .name = "ID_DFR0", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 2, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_dfr0 },
            { .name = "ID_AFR0", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 3, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_afr0 },
            { .name = "ID_MMFR0", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 4, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_mmfr0 },
            { .name = "ID_MMFR1", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 5, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_mmfr1 },
            { .name = "ID_MMFR2", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 6, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_mmfr2 },
            { .name = "ID_MMFR3", .cp = 15, .crn = 0, .crm = 1,
              .opc1 = 0, .opc2 = 7, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_mmfr3 },
            { .name = "ID_ISAR0", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 0, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_isar0 },
            { .name = "ID_ISAR1", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 1, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_isar1 },
            { .name = "ID_ISAR2", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 2, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_isar2 },
            { .name = "ID_ISAR3", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 3, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_isar3 },
            { .name = "ID_ISAR4", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 4, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_isar4 },
            { .name = "ID_ISAR5", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 5, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = cpu->id_isar5 },
            /* 6..7 are as yet unallocated and must RAZ */
            { .name = "ID_ISAR6", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 6, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = 0 },
            { .name = "ID_ISAR7", .cp = 15, .crn = 0, .crm = 2,
              .opc1 = 0, .opc2 = 7, .access = PL1_R, .type = ARM_CP_CONST,
              .resetvalue = 0 },
            REGINFO_SENTINEL
        };
        define_arm_cp_regs(cpu, v6_idregs);
        define_arm_cp_regs(cpu, v6_cp_reginfo);
    } else {
        define_arm_cp_regs(cpu, not_v6_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_V6K)) {
        define_arm_cp_regs(cpu, v6k_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_V7)) {
        /* v7 performance monitor control register: same implementor
         * field as main ID register, and we implement no event counters.
         */
        ARMCPRegInfo pmcr = {
            .name = "PMCR", .cp = 15, .crn = 9, .crm = 12, .opc1 = 0, .opc2 = 0,
            .access = PL0_RW, .resetvalue = cpu->midr & 0xff000000,
            .fieldoffset = offsetof(CPUARMState, cp15.c9_pmcr),
            .readfn = pmreg_read, .writefn = pmcr_write
        };
        ARMCPRegInfo clidr = {
            .name = "CLIDR", .cp = 15, .crn = 0, .crm = 0, .opc1 = 1, .opc2 = 1,
            .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = cpu->clidr
        };
        define_one_arm_cp_reg(cpu, &pmcr);
        define_one_arm_cp_reg(cpu, &clidr);
        define_arm_cp_regs(cpu, v7_cp_reginfo);
    } else {
        define_arm_cp_regs(cpu, not_v7_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_MPU)) {
        /* These are the MPU registers prior to PMSAv6. Any new
         * PMSA core later than the ARM946 will require that we
         * implement the PMSAv6 or PMSAv7 registers, which are
         * completely different.
         */
        assert(!arm_feature(env, ARM_FEATURE_V6));
        define_arm_cp_regs(cpu, pmsav5_cp_reginfo);
    } else {
        define_arm_cp_regs(cpu, vmsa_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_THUMB2EE)) {
        define_arm_cp_regs(cpu, t2ee_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_GENERIC_TIMER)) {
        define_arm_cp_regs(cpu, generic_timer_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_VAPA)) {
        define_arm_cp_regs(cpu, vapa_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_CACHE_TEST_CLEAN)) {
        define_arm_cp_regs(cpu, cache_test_clean_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_CACHE_DIRTY_REG)) {
        define_arm_cp_regs(cpu, cache_dirty_status_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_CACHE_BLOCK_OPS)) {
        define_arm_cp_regs(cpu, cache_block_ops_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_OMAPCP)) {
        define_arm_cp_regs(cpu, omap_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_STRONGARM)) {
        define_arm_cp_regs(cpu, strongarm_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_XSCALE)) {
        define_arm_cp_regs(cpu, xscale_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_DUMMY_C15_REGS)) {
        define_arm_cp_regs(cpu, dummy_c15_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_MPIDR)) {
        define_arm_cp_regs(cpu, mpidr_cp_reginfo);
    }
    if (arm_feature(env, ARM_FEATURE_LPAE)) {
        define_arm_cp_regs(cpu, lpae_cp_reginfo);
    }
    /* Slightly awkwardly, the OMAP and StrongARM cores need all of
     * cp15 crn=0 to be writes-ignored, whereas for other cores they should
     * be read-only (ie write causes UNDEF exception).
     */
    {
        ARMCPRegInfo id_cp_reginfo[] = {
            /* Note that the MIDR isn't a simple constant register because
             * of the TI925 behaviour where writes to another register can
             * cause the MIDR value to change.
             */
            { .name = "MIDR",
              .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 0,
              .access = PL1_R, .resetvalue = cpu->midr,
              .writefn = arm_cp_write_ignore,
              .fieldoffset = offsetof(CPUARMState, cp15.c0_cpuid) },
            { .name = "CTR",
              .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 1,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = cpu->ctr },
            { .name = "TCMTR",
              .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 2,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            { .name = "TLBTR",
              .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 3,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            /* crn = 0 op1 = 0 crm = 3..7 : currently unassigned; we RAZ. */
            { .name = "DUMMY",
              .cp = 15, .crn = 0, .crm = 3, .opc1 = 0, .opc2 = CP_ANY,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            { .name = "DUMMY",
              .cp = 15, .crn = 0, .crm = 4, .opc1 = 0, .opc2 = CP_ANY,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            { .name = "DUMMY",
              .cp = 15, .crn = 0, .crm = 5, .opc1 = 0, .opc2 = CP_ANY,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            { .name = "DUMMY",
              .cp = 15, .crn = 0, .crm = 6, .opc1 = 0, .opc2 = CP_ANY,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            { .name = "DUMMY",
              .cp = 15, .crn = 0, .crm = 7, .opc1 = 0, .opc2 = CP_ANY,
              .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = 0 },
            REGINFO_SENTINEL
        };
        ARMCPRegInfo crn0_wi_reginfo = {
            .name = "CRN0_WI", .cp = 15, .crn = 0, .crm = CP_ANY,
            .opc1 = CP_ANY, .opc2 = CP_ANY, .access = PL1_W,
            .type = ARM_CP_NOP | ARM_CP_OVERRIDE
        };
        if (arm_feature(env, ARM_FEATURE_OMAPCP) ||
            arm_feature(env, ARM_FEATURE_STRONGARM)) {
            ARMCPRegInfo *r;
            /* Register the blanket "writes ignored" value first to cover the
             * whole space. Then define the specific ID registers, but update
             * their access field to allow write access, so that they ignore
             * writes rather than causing them to UNDEF.
             */
            define_one_arm_cp_reg(cpu, &crn0_wi_reginfo);
            for (r = id_cp_reginfo; r->type != ARM_CP_SENTINEL; r++) {
                r->access = PL1_RW;
                define_one_arm_cp_reg(cpu, r);
            }
        } else {
            /* Just register the standard ID registers (read-only, meaning
             * that writes will UNDEF).
             */
            define_arm_cp_regs(cpu, id_cp_reginfo);
        }
    }

    if (arm_feature(env, ARM_FEATURE_AUXCR)) {
        ARMCPRegInfo auxcr = {
            .name = "AUXCR", .cp = 15, .crn = 1, .crm = 0, .opc1 = 0, .opc2 = 1,
            .access = PL1_RW, .type = ARM_CP_CONST,
            .resetvalue = cpu->reset_auxcr
        };
        define_one_arm_cp_reg(cpu, &auxcr);
    }

    /* Generic registers whose values depend on the implementation */
    {
        ARMCPRegInfo sctlr = {
            .name = "SCTLR", .cp = 15, .crn = 1, .crm = 0, .opc1 = 0, .opc2 = 0,
            .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c1_sys),
            .writefn = sctlr_write, .resetvalue = cpu->reset_sctlr
        };
        if (arm_feature(env, ARM_FEATURE_XSCALE)) {
            /* Normally we would always end the TB on an SCTLR write, but Linux
             * arch/arm/mach-pxa/sleep.S expects two instructions following
             * an MMU enable to execute from cache.  Imitate this behaviour.
             */
            sctlr.type |= ARM_CP_SUPPRESS_TB_END;
        }
        define_one_arm_cp_reg(cpu, &sctlr);
    }
}

ARMCPU *cpu_arm_init(const char *cpu_model)
{
    ARMCPU *cpu;
    CPUARMState *env;
    ObjectClass *oc;
    static int inited = 0;

    oc = cpu_class_by_name(TYPE_ARM_CPU, cpu_model);
    if (!oc) {
        return NULL;
    }
    cpu = ARM_CPU(object_new(object_class_get_name(oc)));
    env = &cpu->env;
    env->cpu_model_str = cpu_model;
    arm_cpu_realize(cpu);

    if (tcg_enabled() && !inited) {
        inited = 1;
        arm_translate_init();
    }

    cpu_reset(CPU(cpu));
    if (arm_feature(env, ARM_FEATURE_NEON)) {
        gdb_register_coprocessor(env, vfp_gdb_get_reg, vfp_gdb_set_reg,
                                 51, "arm-neon.xml", 0);
    } else if (arm_feature(env, ARM_FEATURE_VFP3)) {
        gdb_register_coprocessor(env, vfp_gdb_get_reg, vfp_gdb_set_reg,
                                 35, "arm-vfp3.xml", 0);
    } else if (arm_feature(env, ARM_FEATURE_VFP)) {
        gdb_register_coprocessor(env, vfp_gdb_get_reg, vfp_gdb_set_reg,
                                 19, "arm-vfp.xml", 0);
    }
    qemu_init_vcpu(env);
    return cpu;
}

/* Sort alphabetically by type name, except for "any". */
static gint arm_cpu_list_compare(gconstpointer a, gconstpointer b)
{
    ObjectClass *class_a = (ObjectClass *)a;
    ObjectClass *class_b = (ObjectClass *)b;
    const char *name_a, *name_b;

    name_a = object_class_get_name(class_a);
    name_b = object_class_get_name(class_b);
    if (strcmp(name_a, "any") == 0) {
        return 1;
    } else if (strcmp(name_b, "any") == 0) {
        return -1;
    } else {
        return strcmp(name_a, name_b);
    }
}

static void arm_cpu_list_entry(gpointer data, gpointer user_data)
{
    ObjectClass *oc = data;
    CPUListState *s = user_data;

    (*s->cpu_fprintf)(s->file, "  %s\n",
                      object_class_get_name(oc));
}

void arm_cpu_list(FILE *f, fprintf_function cpu_fprintf)
{
    CPUListState s = {
        .file = f,
        .cpu_fprintf = cpu_fprintf,
    };
    GSList *list;

    list = object_class_get_list(TYPE_ARM_CPU, false);
    list = g_slist_sort(list, arm_cpu_list_compare);
    (*cpu_fprintf)(f, "Available CPUs:\n");
    g_slist_foreach(list, arm_cpu_list_entry, &s);
    g_slist_free(list);
}

void define_one_arm_cp_reg_with_opaque(ARMCPU *cpu,
                                       const ARMCPRegInfo *r, void *opaque)
{
    /* Define implementations of coprocessor registers.
     * We store these in a hashtable because typically
     * there are less than 150 registers in a space which
     * is 16*16*16*8*8 = 262144 in size.
     * Wildcarding is supported for the crm, opc1 and opc2 fields.
     * If a register is defined twice then the second definition is
     * used, so this can be used to define some generic registers and
     * then override them with implementation specific variations.
     * At least one of the original and the second definition should
     * include ARM_CP_OVERRIDE in its type bits -- this is just a guard
     * against accidental use.
     */
    int crm, opc1, opc2;
    int crmmin = (r->crm == CP_ANY) ? 0 : r->crm;
    int crmmax = (r->crm == CP_ANY) ? 15 : r->crm;
    int opc1min = (r->opc1 == CP_ANY) ? 0 : r->opc1;
    int opc1max = (r->opc1 == CP_ANY) ? 7 : r->opc1;
    int opc2min = (r->opc2 == CP_ANY) ? 0 : r->opc2;
    int opc2max = (r->opc2 == CP_ANY) ? 7 : r->opc2;
    /* 64 bit registers have only CRm and Opc1 fields */
    assert(!((r->type & ARM_CP_64BIT) && (r->opc2 || r->crn)));
    /* Check that the register definition has enough info to handle
     * reads and writes if they are permitted.
     */
    if (!(r->type & (ARM_CP_SPECIAL|ARM_CP_CONST))) {
        if (r->access & PL3_R) {
            assert(r->fieldoffset || r->readfn);
        }
        if (r->access & PL3_W) {
            assert(r->fieldoffset || r->writefn);
        }
    }
    /* Bad type field probably means missing sentinel at end of reg list */
    assert(cptype_valid(r->type));
    for (crm = crmmin; crm <= crmmax; crm++) {
        for (opc1 = opc1min; opc1 <= opc1max; opc1++) {
            for (opc2 = opc2min; opc2 <= opc2max; opc2++) {
                uint32_t *key = g_new(uint32_t, 1);
                ARMCPRegInfo *r2 = g_memdup(r, sizeof(ARMCPRegInfo));
                int is64 = (r->type & ARM_CP_64BIT) ? 1 : 0;
                *key = ENCODE_CP_REG(r->cp, is64, r->crn, crm, opc1, opc2);
                r2->opaque = opaque;
                /* Make sure reginfo passed to helpers for wildcarded regs
                 * has the correct crm/opc1/opc2 for this reg, not CP_ANY:
                 */
                r2->crm = crm;
                r2->opc1 = opc1;
                r2->opc2 = opc2;
                /* Overriding of an existing definition must be explicitly
                 * requested.
                 */
                if (!(r->type & ARM_CP_OVERRIDE)) {
                    ARMCPRegInfo *oldreg;
                    oldreg = g_hash_table_lookup(cpu->cp_regs, key);
                    if (oldreg && !(oldreg->type & ARM_CP_OVERRIDE)) {
                        fprintf(stderr, "Register redefined: cp=%d %d bit "
                                "crn=%d crm=%d opc1=%d opc2=%d, "
                                "was %s, now %s\n", r2->cp, 32 + 32 * is64,
                                r2->crn, r2->crm, r2->opc1, r2->opc2,
                                oldreg->name, r2->name);
                        assert(0);
                    }
                }
                g_hash_table_insert(cpu->cp_regs, key, r2);
            }
        }
    }
}

void define_arm_cp_regs_with_opaque(ARMCPU *cpu,
                                    const ARMCPRegInfo *regs, void *opaque)
{
    /* Define a whole list of registers */
    const ARMCPRegInfo *r;
    for (r = regs; r->type != ARM_CP_SENTINEL; r++) {
        define_one_arm_cp_reg_with_opaque(cpu, r, opaque);
    }
}

const ARMCPRegInfo *get_arm_cp_reginfo(ARMCPU *cpu, uint32_t encoded_cp)
{
    return g_hash_table_lookup(cpu->cp_regs, &encoded_cp);
}

int arm_cp_write_ignore(CPUARMState *env, const ARMCPRegInfo *ri,
                        uint64_t value)
{
    /* Helper coprocessor write function for write-ignore registers */
    return 0;
}

int arm_cp_read_zero(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t *value)
{
    /* Helper coprocessor write function for read-as-zero registers */
    *value = 0;
    return 0;
}

static int bad_mode_switch(CPUARMState *env, int mode)
{
    /* Return true if it is not valid for us to switch to
     * this CPU mode (ie all the UNPREDICTABLE cases in
     * the ARM ARM CPSRWriteByInstr pseudocode).
     */
    switch (mode) {
    case ARM_CPU_MODE_USR:
    case ARM_CPU_MODE_SYS:
    case ARM_CPU_MODE_SVC:
    case ARM_CPU_MODE_ABT:
    case ARM_CPU_MODE_UND:
    case ARM_CPU_MODE_IRQ:
    case ARM_CPU_MODE_FIQ:
        return 0;
    default:
        return 1;
    }
}

uint32_t cpsr_read(CPUARMState *env)
{
    int ZF;
    ZF = (env->ZF == 0);
    return env->uncached_cpsr | (env->NF & 0x80000000) | (ZF << 30) |
        (env->CF << 29) | ((env->VF & 0x80000000) >> 3) | (env->QF << 27)
        | (env->thumb << 5) | ((env->condexec_bits & 3) << 25)
        | ((env->condexec_bits & 0xfc) << 8)
        | (env->GE << 16);
}

void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask)
{
    if (mask & CPSR_NZCV) {
        env->ZF = (~val) & CPSR_Z;
        env->NF = val;
        env->CF = (val >> 29) & 1;
        env->VF = (val << 3) & 0x80000000;
    }
    if (mask & CPSR_Q)
        env->QF = ((val & CPSR_Q) != 0);
    if (mask & CPSR_T)
        env->thumb = ((val & CPSR_T) != 0);
    if (mask & CPSR_IT_0_1) {
        env->condexec_bits &= ~3;
        env->condexec_bits |= (val >> 25) & 3;
    }
    if (mask & CPSR_IT_2_7) {
        env->condexec_bits &= 3;
        env->condexec_bits |= (val >> 8) & 0xfc;
    }
    if (mask & CPSR_GE) {
        env->GE = (val >> 16) & 0xf;
    }

    if ((env->uncached_cpsr ^ val) & mask & CPSR_M) {
        if (bad_mode_switch(env, val & CPSR_M)) {
            /* Attempt to switch to an invalid mode: this is UNPREDICTABLE.
             * We choose to ignore the attempt and leave the CPSR M field
             * untouched.
             */
            mask &= ~CPSR_M;
        } else {
            switch_mode(env, val & CPSR_M);
        }
    }
    mask &= ~CACHED_CPSR_BITS;
    env->uncached_cpsr = (env->uncached_cpsr & ~mask) | (val & mask);
}

/* Sign/zero extend */
uint32_t HELPER(sxtb16)(uint32_t x)
{
    uint32_t res;
    res = (uint16_t)(int8_t)x;
    res |= (uint32_t)(int8_t)(x >> 16) << 16;
    return res;
}

uint32_t HELPER(uxtb16)(uint32_t x)
{
    uint32_t res;
    res = (uint16_t)(uint8_t)x;
    res |= (uint32_t)(uint8_t)(x >> 16) << 16;
    return res;
}

uint32_t HELPER(clz)(uint32_t x)
{
    return clz32(x);
}

int32_t HELPER(sdiv)(int32_t num, int32_t den)
{
    if (den == 0)
      return 0;
    if (num == INT_MIN && den == -1)
      return INT_MIN;
    return num / den;
}

uint32_t HELPER(udiv)(uint32_t num, uint32_t den)
{
    if (den == 0)
      return 0;
    return num / den;
}

uint32_t HELPER(rbit)(uint32_t x)
{
    x =  ((x & 0xff000000) >> 24)
       | ((x & 0x00ff0000) >> 8)
       | ((x & 0x0000ff00) << 8)
       | ((x & 0x000000ff) << 24);
    x =  ((x & 0xf0f0f0f0) >> 4)
       | ((x & 0x0f0f0f0f) << 4);
    x =  ((x & 0x88888888) >> 3)
       | ((x & 0x44444444) >> 1)
       | ((x & 0x22222222) << 1)
       | ((x & 0x11111111) << 3);
    return x;
}

#if defined(CONFIG_USER_ONLY)

void do_interrupt (CPUARMState *env)
{
    env->exception_index = -1;
}

int cpu_arm_handle_mmu_fault (CPUARMState *env, target_ulong address, int rw,
                              int mmu_idx)
{
    if (rw == 2) {
        env->exception_index = EXCP_PREFETCH_ABORT;
        env->cp15.c6_insn = address;
    } else {
        env->exception_index = EXCP_DATA_ABORT;
        env->cp15.c6_data = address;
    }
    return 1;
}

/* These should probably raise undefined insn exceptions.  */
void HELPER(v7m_msr)(CPUARMState *env, uint32_t reg, uint32_t val)
{
    cpu_abort(env, "v7m_mrs %d\n", reg);
}

uint32_t HELPER(v7m_mrs)(CPUARMState *env, uint32_t reg)
{
    cpu_abort(env, "v7m_mrs %d\n", reg);
    return 0;
}

void switch_mode(CPUARMState *env, int mode)
{
    if (mode != ARM_CPU_MODE_USR)
        cpu_abort(env, "Tried to switch out of user mode\n");
}

void HELPER(set_r13_banked)(CPUARMState *env, uint32_t mode, uint32_t val)
{
    cpu_abort(env, "banked r13 write\n");
}

uint32_t HELPER(get_r13_banked)(CPUARMState *env, uint32_t mode)
{
    cpu_abort(env, "banked r13 read\n");
    return 0;
}

#else

/* Map CPU modes onto saved register banks.  */
static inline int bank_number(CPUARMState *env, int mode)
{
    switch (mode) {
    case ARM_CPU_MODE_USR:
    case ARM_CPU_MODE_SYS:
        return 0;
    case ARM_CPU_MODE_SVC:
        return 1;
    case ARM_CPU_MODE_ABT:
        return 2;
    case ARM_CPU_MODE_UND:
        return 3;
    case ARM_CPU_MODE_IRQ:
        return 4;
    case ARM_CPU_MODE_FIQ:
        return 5;
    }
    cpu_abort(env, "Bad mode %x\n", mode);
    return -1;
}

void switch_mode(CPUARMState *env, int mode)
{
    int old_mode;
    int i;

    old_mode = env->uncached_cpsr & CPSR_M;
    if (mode == old_mode)
        return;

    if (old_mode == ARM_CPU_MODE_FIQ) {
        memcpy (env->fiq_regs, env->regs + 8, 5 * sizeof(uint32_t));
        memcpy (env->regs + 8, env->usr_regs, 5 * sizeof(uint32_t));
    } else if (mode == ARM_CPU_MODE_FIQ) {
        memcpy (env->usr_regs, env->regs + 8, 5 * sizeof(uint32_t));
        memcpy (env->regs + 8, env->fiq_regs, 5 * sizeof(uint32_t));
    }

    i = bank_number(env, old_mode);
    env->banked_r13[i] = env->regs[13];
    env->banked_r14[i] = env->regs[14];
    env->banked_spsr[i] = env->spsr;

    i = bank_number(env, mode);
    env->regs[13] = env->banked_r13[i];
    env->regs[14] = env->banked_r14[i];
    env->spsr = env->banked_spsr[i];
}

static void v7m_push(CPUARMState *env, uint32_t val)
{
    env->regs[13] -= 4;
    stl_phys(env->regs[13], val);
}

static uint32_t v7m_pop(CPUARMState *env)
{
    uint32_t val;
    val = ldl_phys(env->regs[13]);
    env->regs[13] += 4;
    return val;
}

/* Switch to V7M main or process stack pointer.  */
static void switch_v7m_sp(CPUARMState *env, int process)
{
    uint32_t tmp;
    if (env->v7m.current_sp != process) {
        tmp = env->v7m.other_sp;
        env->v7m.other_sp = env->regs[13];
        env->regs[13] = tmp;
        env->v7m.current_sp = process;
    }
}

static void do_v7m_exception_exit(CPUARMState *env)
{
    uint32_t type;
    uint32_t xpsr;

    type = env->regs[15];
    if (env->v7m.exception != 0)
        armv7m_nvic_complete_irq(env->nvic, env->v7m.exception);

    /* Switch to the target stack.  */
    switch_v7m_sp(env, (type & 4) != 0);
    /* Pop registers.  */
    env->regs[0] = v7m_pop(env);
    env->regs[1] = v7m_pop(env);
    env->regs[2] = v7m_pop(env);
    env->regs[3] = v7m_pop(env);
    env->regs[12] = v7m_pop(env);
    env->regs[14] = v7m_pop(env);
    env->regs[15] = v7m_pop(env);
    xpsr = v7m_pop(env);
    xpsr_write(env, xpsr, 0xfffffdff);
    /* Undo stack alignment.  */
    if (xpsr & 0x200)
        env->regs[13] |= 4;
    /* ??? The exception return type specifies Thread/Handler mode.  However
       this is also implied by the xPSR value. Not sure what to do
       if there is a mismatch.  */
    /* ??? Likewise for mismatches between the CONTROL register and the stack
       pointer.  */
}

static void do_interrupt_v7m(CPUARMState *env)
{
    uint32_t xpsr = xpsr_read(env);
    uint32_t lr;
    uint32_t addr;

    lr = 0xfffffff1;
    if (env->v7m.current_sp)
        lr |= 4;
    if (env->v7m.exception == 0)
        lr |= 8;

    /* For exceptions we just mark as pending on the NVIC, and let that
       handle it.  */
    /* TODO: Need to escalate if the current priority is higher than the
       one we're raising.  */
    switch (env->exception_index) {
    case EXCP_UDEF:
        armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_USAGE);
        return;
    case EXCP_SWI:
        /* The PC already points to the next instruction.  */
        armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_SVC);
        return;
    case EXCP_PREFETCH_ABORT:
    case EXCP_DATA_ABORT:
        armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_MEM);
        return;
    case EXCP_BKPT:
        if (semihosting_enabled) {
            int nr;
            nr = arm_lduw_code(env, env->regs[15], env->bswap_code) & 0xff;
            if (nr == 0xab) {
                env->regs[15] += 2;
                env->regs[0] = do_arm_semihosting(env);
                return;
            }
        }
        armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_DEBUG);
        return;
    case EXCP_IRQ:
        env->v7m.exception = armv7m_nvic_acknowledge_irq(env->nvic);
        break;
    case EXCP_EXCEPTION_EXIT:
        do_v7m_exception_exit(env);
        return;
    default:
        cpu_abort(env, "Unhandled exception 0x%x\n", env->exception_index);
        return; /* Never happens.  Keep compiler happy.  */
    }

    /* Align stack pointer.  */
    /* ??? Should only do this if Configuration Control Register
       STACKALIGN bit is set.  */
    if (env->regs[13] & 4) {
        env->regs[13] -= 4;
        xpsr |= 0x200;
    }
    /* Switch to the handler mode.  */
    v7m_push(env, xpsr);
    v7m_push(env, env->regs[15]);
    v7m_push(env, env->regs[14]);
    v7m_push(env, env->regs[12]);
    v7m_push(env, env->regs[3]);
    v7m_push(env, env->regs[2]);
    v7m_push(env, env->regs[1]);
    v7m_push(env, env->regs[0]);
    switch_v7m_sp(env, 0);
    /* Clear IT bits */
    env->condexec_bits = 0;
    env->regs[14] = lr;
    addr = ldl_phys(env->v7m.vecbase + env->v7m.exception * 4);
    env->regs[15] = addr & 0xfffffffe;
    env->thumb = addr & 1;
}

/* Handle a CPU exception.  */
void do_interrupt(CPUARMState *env)
{
    uint32_t addr;
    uint32_t mask;
    int new_mode;
    uint32_t offset;

    if (IS_M(env)) {
        do_interrupt_v7m(env);
        return;
    }
    /* TODO: Vectored interrupt controller.  */
    switch (env->exception_index) {
    case EXCP_UDEF:
        new_mode = ARM_CPU_MODE_UND;
        addr = 0x04;
        mask = CPSR_I;
        if (env->thumb)
            offset = 2;
        else
            offset = 4;
        break;
    case EXCP_SWI:
        if (semihosting_enabled) {
            /* Check for semihosting interrupt.  */
            if (env->thumb) {
                mask = arm_lduw_code(env, env->regs[15] - 2, env->bswap_code)
                    & 0xff;
            } else {
                mask = arm_ldl_code(env, env->regs[15] - 4, env->bswap_code)
                    & 0xffffff;
            }
            /* Only intercept calls from privileged modes, to provide some
               semblance of security.  */
            if (((mask == 0x123456 && !env->thumb)
                    || (mask == 0xab && env->thumb))
                  && (env->uncached_cpsr & CPSR_M) != ARM_CPU_MODE_USR) {
                env->regs[0] = do_arm_semihosting(env);
                return;
            }
        }
        new_mode = ARM_CPU_MODE_SVC;
        addr = 0x08;
        mask = CPSR_I;
        /* The PC already points to the next instruction.  */
        offset = 0;
        break;
    case EXCP_BKPT:
        /* See if this is a semihosting syscall.  */
        if (env->thumb && semihosting_enabled) {
            mask = arm_lduw_code(env, env->regs[15], env->bswap_code) & 0xff;
            if (mask == 0xab
                  && (env->uncached_cpsr & CPSR_M) != ARM_CPU_MODE_USR) {
                env->regs[15] += 2;
                env->regs[0] = do_arm_semihosting(env);
                return;
            }
        }
        env->cp15.c5_insn = 2;
        /* Fall through to prefetch abort.  */
    case EXCP_PREFETCH_ABORT:
        new_mode = ARM_CPU_MODE_ABT;
        addr = 0x0c;
        mask = CPSR_A | CPSR_I;
        offset = 4;
        break;
    case EXCP_DATA_ABORT:
        new_mode = ARM_CPU_MODE_ABT;
        addr = 0x10;
        mask = CPSR_A | CPSR_I;
        offset = 8;
        break;
    case EXCP_IRQ:
        new_mode = ARM_CPU_MODE_IRQ;
        addr = 0x18;
        /* Disable IRQ and imprecise data aborts.  */
        mask = CPSR_A | CPSR_I;
        offset = 4;
        break;
    case EXCP_FIQ:
        new_mode = ARM_CPU_MODE_FIQ;
        addr = 0x1c;
        /* Disable FIQ, IRQ and imprecise data aborts.  */
        mask = CPSR_A | CPSR_I | CPSR_F;
        offset = 4;
        break;
    default:
        cpu_abort(env, "Unhandled exception 0x%x\n", env->exception_index);
        return; /* Never happens.  Keep compiler happy.  */
    }
    /* High vectors.  */
    if (env->cp15.c1_sys & (1 << 13)) {
        addr += 0xffff0000;
    }
    switch_mode (env, new_mode);
    env->spsr = cpsr_read(env);
    /* Clear IT bits.  */
    env->condexec_bits = 0;
    /* Switch to the new mode, and to the correct instruction set.  */
    env->uncached_cpsr = (env->uncached_cpsr & ~CPSR_M) | new_mode;
    env->uncached_cpsr |= mask;
    /* this is a lie, as the was no c1_sys on V4T/V5, but who cares
     * and we should just guard the thumb mode on V4 */
    if (arm_feature(env, ARM_FEATURE_V4T)) {
        env->thumb = (env->cp15.c1_sys & (1 << 30)) != 0;
    }
    env->regs[14] = env->regs[15] + offset;
    env->regs[15] = addr;
    env->interrupt_request |= CPU_INTERRUPT_EXITTB;
}

/* Check section/page access permissions.
   Returns the page protection flags, or zero if the access is not
   permitted.  */
static inline int check_ap(CPUARMState *env, int ap, int domain_prot,
                           int access_type, int is_user)
{
  int prot_ro;

  if (domain_prot == 3) {
    return PAGE_READ | PAGE_WRITE;
  }

  if (access_type == 1)
      prot_ro = 0;
  else
      prot_ro = PAGE_READ;

  switch (ap) {
  case 0:
      if (access_type == 1)
          return 0;
      switch ((env->cp15.c1_sys >> 8) & 3) {
      case 1:
          return is_user ? 0 : PAGE_READ;
      case 2:
          return PAGE_READ;
      default:
          return 0;
      }
  case 1:
      return is_user ? 0 : PAGE_READ | PAGE_WRITE;
  case 2:
      if (is_user)
          return prot_ro;
      else
          return PAGE_READ | PAGE_WRITE;
  case 3:
      return PAGE_READ | PAGE_WRITE;
  case 4: /* Reserved.  */
      return 0;
  case 5:
      return is_user ? 0 : prot_ro;
  case 6:
      return prot_ro;
  case 7:
      if (!arm_feature (env, ARM_FEATURE_V6K))
          return 0;
      return prot_ro;
  default:
      abort();
  }
}

static uint32_t get_level1_table_address(CPUARMState *env, uint32_t address)
{
    uint32_t table;

    if (address & env->cp15.c2_mask)
        table = env->cp15.c2_base1 & 0xffffc000;
    else
        table = env->cp15.c2_base0 & env->cp15.c2_base_mask;

    table |= (address >> 18) & 0x3ffc;
    return table;
}

static int get_phys_addr_v5(CPUARMState *env, uint32_t address, int access_type,
                            int is_user, hwaddr *phys_ptr,
                            int *prot, target_ulong *page_size)
{
    int code;
    uint32_t table;
    uint32_t desc;
    int type;
    int ap;
    int domain;
    int domain_prot;
    hwaddr phys_addr;

    /* Pagetable walk.  */
    /* Lookup l1 descriptor.  */
    table = get_level1_table_address(env, address);
    desc = ldl_phys(table);
    type = (desc & 3);
    domain = (desc >> 5) & 0x0f;
    domain_prot = (env->cp15.c3 >> (domain * 2)) & 3;
    if (type == 0) {
        /* Section translation fault.  */
        code = 5;
        goto do_fault;
    }
    if (domain_prot == 0 || domain_prot == 2) {
        if (type == 2)
            code = 9; /* Section domain fault.  */
        else
            code = 11; /* Page domain fault.  */
        goto do_fault;
    }
    if (type == 2) {
        /* 1Mb section.  */
        phys_addr = (desc & 0xfff00000) | (address & 0x000fffff);
        ap = (desc >> 10) & 3;
        code = 13;
        *page_size = 1024 * 1024;
    } else {
        /* Lookup l2 entry.  */
	if (type == 1) {
	    /* Coarse pagetable.  */
	    table = (desc & 0xfffffc00) | ((address >> 10) & 0x3fc);
	} else {
	    /* Fine pagetable.  */
	    table = (desc & 0xfffff000) | ((address >> 8) & 0xffc);
	}
        desc = ldl_phys(table);
        switch (desc & 3) {
        case 0: /* Page translation fault.  */
            code = 7;
            goto do_fault;
        case 1: /* 64k page.  */
            phys_addr = (desc & 0xffff0000) | (address & 0xffff);
            ap = (desc >> (4 + ((address >> 13) & 6))) & 3;
            *page_size = 0x10000;
            break;
        case 2: /* 4k page.  */
            phys_addr = (desc & 0xfffff000) | (address & 0xfff);
            ap = (desc >> (4 + ((address >> 13) & 6))) & 3;
            *page_size = 0x1000;
            break;
        case 3: /* 1k page.  */
	    if (type == 1) {
		if (arm_feature(env, ARM_FEATURE_XSCALE)) {
		    phys_addr = (desc & 0xfffff000) | (address & 0xfff);
		} else {
		    /* Page translation fault.  */
		    code = 7;
		    goto do_fault;
		}
	    } else {
		phys_addr = (desc & 0xfffffc00) | (address & 0x3ff);
	    }
            ap = (desc >> 4) & 3;
            *page_size = 0x400;
            break;
        default:
            /* Never happens, but compiler isn't smart enough to tell.  */
            abort();
        }
        code = 15;
    }
    *prot = check_ap(env, ap, domain_prot, access_type, is_user);
    if (!*prot) {
        /* Access permission fault.  */
        goto do_fault;
    }
    *prot |= PAGE_EXEC;
    *phys_ptr = phys_addr;
    return 0;
do_fault:
    return code | (domain << 4);
}

static int get_phys_addr_v6(CPUARMState *env, uint32_t address, int access_type,
                            int is_user, hwaddr *phys_ptr,
                            int *prot, target_ulong *page_size)
{
    int code;
    uint32_t table;
    uint32_t desc;
    uint32_t xn;
    uint32_t pxn = 0;
    int type;
    int ap;
    int domain = 0;
    int domain_prot;
    hwaddr phys_addr;

    /* Pagetable walk.  */
    /* Lookup l1 descriptor.  */
    table = get_level1_table_address(env, address);
    desc = ldl_phys(table);
    type = (desc & 3);
    if (type == 0 || (type == 3 && !arm_feature(env, ARM_FEATURE_PXN))) {
        /* Section translation fault, or attempt to use the encoding
         * which is Reserved on implementations without PXN.
         */
        code = 5;
        goto do_fault;
    }
    if ((type == 1) || !(desc & (1 << 18))) {
        /* Page or Section.  */
        domain = (desc >> 5) & 0x0f;
    }
    domain_prot = (env->cp15.c3 >> (domain * 2)) & 3;
    if (domain_prot == 0 || domain_prot == 2) {
        if (type != 1) {
            code = 9; /* Section domain fault.  */
        } else {
            code = 11; /* Page domain fault.  */
        }
        goto do_fault;
    }
    if (type != 1) {
        if (desc & (1 << 18)) {
            /* Supersection.  */
            phys_addr = (desc & 0xff000000) | (address & 0x00ffffff);
            *page_size = 0x1000000;
        } else {
            /* Section.  */
            phys_addr = (desc & 0xfff00000) | (address & 0x000fffff);
            *page_size = 0x100000;
        }
        ap = ((desc >> 10) & 3) | ((desc >> 13) & 4);
        xn = desc & (1 << 4);
        pxn = desc & 1;
        code = 13;
    } else {
        if (arm_feature(env, ARM_FEATURE_PXN)) {
            pxn = (desc >> 2) & 1;
        }
        /* Lookup l2 entry.  */
        table = (desc & 0xfffffc00) | ((address >> 10) & 0x3fc);
        desc = ldl_phys(table);
        ap = ((desc >> 4) & 3) | ((desc >> 7) & 4);
        switch (desc & 3) {
        case 0: /* Page translation fault.  */
            code = 7;
            goto do_fault;
        case 1: /* 64k page.  */
            phys_addr = (desc & 0xffff0000) | (address & 0xffff);
            xn = desc & (1 << 15);
            *page_size = 0x10000;
            break;
        case 2: case 3: /* 4k page.  */
            phys_addr = (desc & 0xfffff000) | (address & 0xfff);
            xn = desc & 1;
            *page_size = 0x1000;
            break;
        default:
            /* Never happens, but compiler isn't smart enough to tell.  */
            abort();
        }
        code = 15;
    }
    if (domain_prot == 3) {
        *prot = PAGE_READ | PAGE_WRITE | PAGE_EXEC;
    } else {
        if (pxn && !is_user) {
            xn = 1;
        }
        if (xn && access_type == 2)
            goto do_fault;

        /* The simplified model uses AP[0] as an access control bit.  */
        if ((env->cp15.c1_sys & (1 << 29)) && (ap & 1) == 0) {
            /* Access flag fault.  */
            code = (code == 15) ? 6 : 3;
            goto do_fault;
        }
        *prot = check_ap(env, ap, domain_prot, access_type, is_user);
        if (!*prot) {
            /* Access permission fault.  */
            goto do_fault;
        }
        if (!xn) {
            *prot |= PAGE_EXEC;
        }
    }
    *phys_ptr = phys_addr;
    return 0;
do_fault:
    return code | (domain << 4);
}

/* Fault type for long-descriptor MMU fault reporting; this corresponds
 * to bits [5..2] in the STATUS field in long-format DFSR/IFSR.
 */
typedef enum {
    translation_fault = 1,
    access_fault = 2,
    permission_fault = 3,
} MMUFaultType;

static int get_phys_addr_lpae(CPUARMState *env, uint32_t address,
                              int access_type, int is_user,
                              hwaddr *phys_ptr, int *prot,
                              target_ulong *page_size_ptr)
{
    /* Read an LPAE long-descriptor translation table. */
    MMUFaultType fault_type = translation_fault;
    uint32_t level = 1;
    uint32_t epd;
    uint32_t tsz;
    uint64_t ttbr;
    int ttbr_select;
    int n;
    hwaddr descaddr;
    uint32_t tableattrs;
    target_ulong page_size;
    uint32_t attrs;

    /* Determine whether this address is in the region controlled by
     * TTBR0 or TTBR1 (or if it is in neither region and should fault).
     * This is a Non-secure PL0/1 stage 1 translation, so controlled by
     * TTBCR/TTBR0/TTBR1 in accordance with ARM ARM DDI0406C table B-32:
     */
    uint32_t t0sz = extract32(env->cp15.c2_control, 0, 3);
    uint32_t t1sz = extract32(env->cp15.c2_control, 16, 3);
    if (t0sz && !extract32(address, 32 - t0sz, t0sz)) {
        /* there is a ttbr0 region and we are in it (high bits all zero) */
        ttbr_select = 0;
    } else if (t1sz && !extract32(~address, 32 - t1sz, t1sz)) {
        /* there is a ttbr1 region and we are in it (high bits all one) */
        ttbr_select = 1;
    } else if (!t0sz) {
        /* ttbr0 region is "everything not in the ttbr1 region" */
        ttbr_select = 0;
    } else if (!t1sz) {
        /* ttbr1 region is "everything not in the ttbr0 region" */
        ttbr_select = 1;
    } else {
        /* in the gap between the two regions, this is a Translation fault */
        fault_type = translation_fault;
        goto do_fault;
    }

    /* Note that QEMU ignores shareability and cacheability attributes,
     * so we don't need to do anything with the SH, ORGN, IRGN fields
     * in the TTBCR.  Similarly, TTBCR:A1 selects whether we get the
     * ASID from TTBR0 or TTBR1, but QEMU's TLB doesn't currently
     * implement any ASID-like capability so we can ignore it (instead
     * we will always flush the TLB any time the ASID is changed).
     */
    if (ttbr_select == 0) {
        ttbr = ((uint64_t)env->cp15.c2_base0_hi << 32) | env->cp15.c2_base0;
        epd = extract32(env->cp15.c2_control, 7, 1);
        tsz = t0sz;
    } else {
        ttbr = ((uint64_t)env->cp15.c2_base1_hi << 32) | env->cp15.c2_base1;
        epd = extract32(env->cp15.c2_control, 23, 1);
        tsz = t1sz;
    }

    if (epd) {
        /* Translation table walk disabled => Translation fault on TLB miss */
        goto do_fault;
    }

    /* If the region is small enough we will skip straight to a 2nd level
     * lookup. This affects the number of bits of the address used in
     * combination with the TTBR to find the first descriptor. ('n' here
     * matches the usage in the ARM ARM sB3.6.6, where bits [39..n] are
     * from the TTBR, [n-1..3] from the vaddr, and [2..0] always zero).
     */
    if (tsz > 1) {
        level = 2;
        n = 14 - tsz;
    } else {
        n = 5 - tsz;
    }

    /* Clear the vaddr bits which aren't part of the within-region address,
     * so that we don't have to special case things when calculating the
     * first descriptor address.
     */
    address &= (0xffffffffU >> tsz);

    /* Now we can extract the actual base address from the TTBR */
    descaddr = extract64(ttbr, 0, 40);
    descaddr &= ~((1ULL << n) - 1);

    tableattrs = 0;
    for (;;) {
        uint64_t descriptor;

        descaddr |= ((address >> (9 * (4 - level))) & 0xff8);
        descriptor = ldq_phys(descaddr);
        if (!(descriptor & 1) ||
            (!(descriptor & 2) && (level == 3))) {
            /* Invalid, or the Reserved level 3 encoding */
            goto do_fault;
        }
        descaddr = descriptor & 0xfffffff000ULL;

        if ((descriptor & 2) && (level < 3)) {
            /* Table entry. The top five bits are attributes which  may
             * propagate down through lower levels of the table (and
             * which are all arranged so that 0 means "no effect", so
             * we can gather them up by ORing in the bits at each level).
             */
            tableattrs |= extract64(descriptor, 59, 5);
            level++;
            continue;
        }
        /* Block entry at level 1 or 2, or page entry at level 3.
         * These are basically the same thing, although the number
         * of bits we pull in from the vaddr varies.
         */
        page_size = (1 << (39 - (9 * level)));
        descaddr |= (address & (page_size - 1));
        /* Extract attributes from the descriptor and merge with table attrs */
        attrs = extract64(descriptor, 2, 10)
            | (extract64(descriptor, 52, 12) << 10);
        attrs |= extract32(tableattrs, 0, 2) << 11; /* XN, PXN */
        attrs |= extract32(tableattrs, 3, 1) << 5; /* APTable[1] => AP[2] */
        /* The sense of AP[1] vs APTable[0] is reversed, as APTable[0] == 1
         * means "force PL1 access only", which means forcing AP[1] to 0.
         */
        if (extract32(tableattrs, 2, 1)) {
            attrs &= ~(1 << 4);
        }
        /* Since we're always in the Non-secure state, NSTable is ignored. */
        break;
    }
    /* Here descaddr is the final physical address, and attributes
     * are all in attrs.
     */
    fault_type = access_fault;
    if ((attrs & (1 << 8)) == 0) {
        /* Access flag */
        goto do_fault;
    }
    fault_type = permission_fault;
    if (is_user && !(attrs & (1 << 4))) {
        /* Unprivileged access not enabled */
        goto do_fault;
    }
    *prot = PAGE_READ | PAGE_WRITE | PAGE_EXEC;
    if (attrs & (1 << 12) || (!is_user && (attrs & (1 << 11)))) {
        /* XN or PXN */
        if (access_type == 2) {
            goto do_fault;
        }
        *prot &= ~PAGE_EXEC;
    }
    if (attrs & (1 << 5)) {
        /* Write access forbidden */
        if (access_type == 1) {
            goto do_fault;
        }
        *prot &= ~PAGE_WRITE;
    }

    *phys_ptr = descaddr;
    *page_size_ptr = page_size;
    return 0;

do_fault:
    /* Long-descriptor format IFSR/DFSR value */
    return (1 << 9) | (fault_type << 2) | level;
}

static int get_phys_addr_mpu(CPUARMState *env, uint32_t address,
                             int access_type, int is_user,
                             hwaddr *phys_ptr, int *prot)
{
    int n;
    uint32_t mask;
    uint32_t base;

    *phys_ptr = address;
    for (n = 7; n >= 0; n--) {
	base = env->cp15.c6_region[n];
	if ((base & 1) == 0)
	    continue;
	mask = 1 << ((base >> 1) & 0x1f);
	/* Keep this shift separate from the above to avoid an
	   (undefined) << 32.  */
	mask = (mask << 1) - 1;
	if (((base ^ address) & ~mask) == 0)
	    break;
    }
    if (n < 0)
	return 2;

    if (access_type == 2) {
	mask = env->cp15.c5_insn;
    } else {
	mask = env->cp15.c5_data;
    }
    mask = (mask >> (n * 4)) & 0xf;
    switch (mask) {
    case 0:
	return 1;
    case 1:
	if (is_user)
	  return 1;
	*prot = PAGE_READ | PAGE_WRITE;
	break;
    case 2:
	*prot = PAGE_READ;
	if (!is_user)
	    *prot |= PAGE_WRITE;
	break;
    case 3:
	*prot = PAGE_READ | PAGE_WRITE;
	break;
    case 5:
	if (is_user)
	    return 1;
	*prot = PAGE_READ;
	break;
    case 6:
	*prot = PAGE_READ;
	break;
    default:
	/* Bad permission.  */
	return 1;
    }
    *prot |= PAGE_EXEC;
    return 0;
}

/* get_phys_addr - get the physical address for this virtual address
 *
 * Find the physical address corresponding to the given virtual address,
 * by doing a translation table walk on MMU based systems or using the
 * MPU state on MPU based systems.
 *
 * Returns 0 if the translation was successful. Otherwise, phys_ptr,
 * prot and page_size are not filled in, and the return value provides
 * information on why the translation aborted, in the format of a
 * DFSR/IFSR fault register, with the following caveats:
 *  * we honour the short vs long DFSR format differences.
 *  * the WnR bit is never set (the caller must do this).
 *  * for MPU based systems we don't bother to return a full FSR format
 *    value.
 *
 * @env: CPUARMState
 * @address: virtual address to get physical address for
 * @access_type: 0 for read, 1 for write, 2 for execute
 * @is_user: 0 for privileged access, 1 for user
 * @phys_ptr: set to the physical address corresponding to the virtual address
 * @prot: set to the permissions for the page containing phys_ptr
 * @page_size: set to the size of the page containing phys_ptr
 */
static inline int get_phys_addr(CPUARMState *env, uint32_t address,
                                int access_type, int is_user,
                                hwaddr *phys_ptr, int *prot,
                                target_ulong *page_size)
{
    /* Fast Context Switch Extension.  */
    if (address < 0x02000000)
        address += env->cp15.c13_fcse;

    if ((env->cp15.c1_sys & 1) == 0) {
        /* MMU/MPU disabled.  */
        *phys_ptr = address;
        *prot = PAGE_READ | PAGE_WRITE | PAGE_EXEC;
        *page_size = TARGET_PAGE_SIZE;
        return 0;
    } else if (arm_feature(env, ARM_FEATURE_MPU)) {
        *page_size = TARGET_PAGE_SIZE;
	return get_phys_addr_mpu(env, address, access_type, is_user, phys_ptr,
				 prot);
    } else if (extended_addresses_enabled(env)) {
        return get_phys_addr_lpae(env, address, access_type, is_user, phys_ptr,
                                  prot, page_size);
    } else if (env->cp15.c1_sys & (1 << 23)) {
        return get_phys_addr_v6(env, address, access_type, is_user, phys_ptr,
                                prot, page_size);
    } else {
        return get_phys_addr_v5(env, address, access_type, is_user, phys_ptr,
                                prot, page_size);
    }
}

int cpu_arm_handle_mmu_fault (CPUARMState *env, target_ulong address,
                              int access_type, int mmu_idx)
{
    hwaddr phys_addr;
    target_ulong page_size;
    int prot;
    int ret, is_user;

    is_user = mmu_idx == MMU_USER_IDX;
    ret = get_phys_addr(env, address, access_type, is_user, &phys_addr, &prot,
                        &page_size);
    if (ret == 0) {
        /* Map a single [sub]page.  */
        phys_addr &= ~(hwaddr)0x3ff;
        address &= ~(uint32_t)0x3ff;
        tlb_set_page (env, address, phys_addr, prot, mmu_idx, page_size);
        return 0;
    }

    if (access_type == 2) {
        env->cp15.c5_insn = ret;
        env->cp15.c6_insn = address;
        env->exception_index = EXCP_PREFETCH_ABORT;
    } else {
        env->cp15.c5_data = ret;
        if (access_type == 1 && arm_feature(env, ARM_FEATURE_V6))
            env->cp15.c5_data |= (1 << 11);
        env->cp15.c6_data = address;
        env->exception_index = EXCP_DATA_ABORT;
    }
    return 1;
}

hwaddr cpu_get_phys_page_debug(CPUARMState *env, target_ulong addr)
{
    hwaddr phys_addr;
    target_ulong page_size;
    int prot;
    int ret;

    ret = get_phys_addr(env, addr, 0, 0, &phys_addr, &prot, &page_size);

    if (ret != 0)
        return -1;

    return phys_addr;
}

void HELPER(set_r13_banked)(CPUARMState *env, uint32_t mode, uint32_t val)
{
    if ((env->uncached_cpsr & CPSR_M) == mode) {
        env->regs[13] = val;
    } else {
        env->banked_r13[bank_number(env, mode)] = val;
    }
}

uint32_t HELPER(get_r13_banked)(CPUARMState *env, uint32_t mode)
{
    if ((env->uncached_cpsr & CPSR_M) == mode) {
        return env->regs[13];
    } else {
        return env->banked_r13[bank_number(env, mode)];
    }
}

uint32_t HELPER(v7m_mrs)(CPUARMState *env, uint32_t reg)
{
    switch (reg) {
    case 0: /* APSR */
        return xpsr_read(env) & 0xf8000000;
    case 1: /* IAPSR */
        return xpsr_read(env) & 0xf80001ff;
    case 2: /* EAPSR */
        return xpsr_read(env) & 0xff00fc00;
    case 3: /* xPSR */
        return xpsr_read(env) & 0xff00fdff;
    case 5: /* IPSR */
        return xpsr_read(env) & 0x000001ff;
    case 6: /* EPSR */
        return xpsr_read(env) & 0x0700fc00;
    case 7: /* IEPSR */
        return xpsr_read(env) & 0x0700edff;
    case 8: /* MSP */
        return env->v7m.current_sp ? env->v7m.other_sp : env->regs[13];
    case 9: /* PSP */
        return env->v7m.current_sp ? env->regs[13] : env->v7m.other_sp;
    case 16: /* PRIMASK */
        return (env->uncached_cpsr & CPSR_I) != 0;
    case 17: /* BASEPRI */
    case 18: /* BASEPRI_MAX */
        return env->v7m.basepri;
    case 19: /* FAULTMASK */
        return (env->uncached_cpsr & CPSR_F) != 0;
    case 20: /* CONTROL */
        return env->v7m.control;
    default:
        /* ??? For debugging only.  */
        cpu_abort(env, "Unimplemented system register read (%d)\n", reg);
        return 0;
    }
}

void HELPER(v7m_msr)(CPUARMState *env, uint32_t reg, uint32_t val)
{
    switch (reg) {
    case 0: /* APSR */
        xpsr_write(env, val, 0xf8000000);
        break;
    case 1: /* IAPSR */
        xpsr_write(env, val, 0xf8000000);
        break;
    case 2: /* EAPSR */
        xpsr_write(env, val, 0xfe00fc00);
        break;
    case 3: /* xPSR */
        xpsr_write(env, val, 0xfe00fc00);
        break;
    case 5: /* IPSR */
        /* IPSR bits are readonly.  */
        break;
    case 6: /* EPSR */
        xpsr_write(env, val, 0x0600fc00);
        break;
    case 7: /* IEPSR */
        xpsr_write(env, val, 0x0600fc00);
        break;
    case 8: /* MSP */
        if (env->v7m.current_sp)
            env->v7m.other_sp = val;
        else
            env->regs[13] = val;
        break;
    case 9: /* PSP */
        if (env->v7m.current_sp)
            env->regs[13] = val;
        else
            env->v7m.other_sp = val;
        break;
    case 16: /* PRIMASK */
        if (val & 1)
            env->uncached_cpsr |= CPSR_I;
        else
            env->uncached_cpsr &= ~CPSR_I;
        break;
    case 17: /* BASEPRI */
        env->v7m.basepri = val & 0xff;
        break;
    case 18: /* BASEPRI_MAX */
        val &= 0xff;
        if (val != 0 && (val < env->v7m.basepri || env->v7m.basepri == 0))
            env->v7m.basepri = val;
        break;
    case 19: /* FAULTMASK */
        if (val & 1)
            env->uncached_cpsr |= CPSR_F;
        else
            env->uncached_cpsr &= ~CPSR_F;
        break;
    case 20: /* CONTROL */
        env->v7m.control = val & 3;
        switch_v7m_sp(env, (val & 2) != 0);
        break;
    default:
        /* ??? For debugging only.  */
        cpu_abort(env, "Unimplemented system register write (%d)\n", reg);
        return;
    }
}

#endif

/* Note that signed overflow is undefined in C.  The following routines are
   careful to use unsigned types where modulo arithmetic is required.
   Failure to do so _will_ break on newer gcc.  */

/* Signed saturating arithmetic.  */

/* Perform 16-bit signed saturating addition.  */
static inline uint16_t add16_sat(uint16_t a, uint16_t b)
{
    uint16_t res;

    res = a + b;
    if (((res ^ a) & 0x8000) && !((a ^ b) & 0x8000)) {
        if (a & 0x8000)
            res = 0x8000;
        else
            res = 0x7fff;
    }
    return res;
}

/* Perform 8-bit signed saturating addition.  */
static inline uint8_t add8_sat(uint8_t a, uint8_t b)
{
    uint8_t res;

    res = a + b;
    if (((res ^ a) & 0x80) && !((a ^ b) & 0x80)) {
        if (a & 0x80)
            res = 0x80;
        else
            res = 0x7f;
    }
    return res;
}

/* Perform 16-bit signed saturating subtraction.  */
static inline uint16_t sub16_sat(uint16_t a, uint16_t b)
{
    uint16_t res;

    res = a - b;
    if (((res ^ a) & 0x8000) && ((a ^ b) & 0x8000)) {
        if (a & 0x8000)
            res = 0x8000;
        else
            res = 0x7fff;
    }
    return res;
}

/* Perform 8-bit signed saturating subtraction.  */
static inline uint8_t sub8_sat(uint8_t a, uint8_t b)
{
    uint8_t res;

    res = a - b;
    if (((res ^ a) & 0x80) && ((a ^ b) & 0x80)) {
        if (a & 0x80)
            res = 0x80;
        else
            res = 0x7f;
    }
    return res;
}

#define ADD16(a, b, n) RESULT(add16_sat(a, b), n, 16);
#define SUB16(a, b, n) RESULT(sub16_sat(a, b), n, 16);
#define ADD8(a, b, n)  RESULT(add8_sat(a, b), n, 8);
#define SUB8(a, b, n)  RESULT(sub8_sat(a, b), n, 8);
#define PFX q

#include "op_addsub.h"

/* Unsigned saturating arithmetic.  */
static inline uint16_t add16_usat(uint16_t a, uint16_t b)
{
    uint16_t res;
    res = a + b;
    if (res < a)
        res = 0xffff;
    return res;
}

static inline uint16_t sub16_usat(uint16_t a, uint16_t b)
{
    if (a > b)
        return a - b;
    else
        return 0;
}

static inline uint8_t add8_usat(uint8_t a, uint8_t b)
{
    uint8_t res;
    res = a + b;
    if (res < a)
        res = 0xff;
    return res;
}

static inline uint8_t sub8_usat(uint8_t a, uint8_t b)
{
    if (a > b)
        return a - b;
    else
        return 0;
}

#define ADD16(a, b, n) RESULT(add16_usat(a, b), n, 16);
#define SUB16(a, b, n) RESULT(sub16_usat(a, b), n, 16);
#define ADD8(a, b, n)  RESULT(add8_usat(a, b), n, 8);
#define SUB8(a, b, n)  RESULT(sub8_usat(a, b), n, 8);
#define PFX uq

#include "op_addsub.h"

/* Signed modulo arithmetic.  */
#define SARITH16(a, b, n, op) do { \
    int32_t sum; \
    sum = (int32_t)(int16_t)(a) op (int32_t)(int16_t)(b); \
    RESULT(sum, n, 16); \
    if (sum >= 0) \
        ge |= 3 << (n * 2); \
    } while(0)

#define SARITH8(a, b, n, op) do { \
    int32_t sum; \
    sum = (int32_t)(int8_t)(a) op (int32_t)(int8_t)(b); \
    RESULT(sum, n, 8); \
    if (sum >= 0) \
        ge |= 1 << n; \
    } while(0)


#define ADD16(a, b, n) SARITH16(a, b, n, +)
#define SUB16(a, b, n) SARITH16(a, b, n, -)
#define ADD8(a, b, n)  SARITH8(a, b, n, +)
#define SUB8(a, b, n)  SARITH8(a, b, n, -)
#define PFX s
#define ARITH_GE

#include "op_addsub.h"

/* Unsigned modulo arithmetic.  */
#define ADD16(a, b, n) do { \
    uint32_t sum; \
    sum = (uint32_t)(uint16_t)(a) + (uint32_t)(uint16_t)(b); \
    RESULT(sum, n, 16); \
    if ((sum >> 16) == 1) \
        ge |= 3 << (n * 2); \
    } while(0)

#define ADD8(a, b, n) do { \
    uint32_t sum; \
    sum = (uint32_t)(uint8_t)(a) + (uint32_t)(uint8_t)(b); \
    RESULT(sum, n, 8); \
    if ((sum >> 8) == 1) \
        ge |= 1 << n; \
    } while(0)

#define SUB16(a, b, n) do { \
    uint32_t sum; \
    sum = (uint32_t)(uint16_t)(a) - (uint32_t)(uint16_t)(b); \
    RESULT(sum, n, 16); \
    if ((sum >> 16) == 0) \
        ge |= 3 << (n * 2); \
    } while(0)

#define SUB8(a, b, n) do { \
    uint32_t sum; \
    sum = (uint32_t)(uint8_t)(a) - (uint32_t)(uint8_t)(b); \
    RESULT(sum, n, 8); \
    if ((sum >> 8) == 0) \
        ge |= 1 << n; \
    } while(0)

#define PFX u
#define ARITH_GE

#include "op_addsub.h"

/* Halved signed arithmetic.  */
#define ADD16(a, b, n) \
  RESULT(((int32_t)(int16_t)(a) + (int32_t)(int16_t)(b)) >> 1, n, 16)
#define SUB16(a, b, n) \
  RESULT(((int32_t)(int16_t)(a) - (int32_t)(int16_t)(b)) >> 1, n, 16)
#define ADD8(a, b, n) \
  RESULT(((int32_t)(int8_t)(a) + (int32_t)(int8_t)(b)) >> 1, n, 8)
#define SUB8(a, b, n) \
  RESULT(((int32_t)(int8_t)(a) - (int32_t)(int8_t)(b)) >> 1, n, 8)
#define PFX sh

#include "op_addsub.h"

/* Halved unsigned arithmetic.  */
#define ADD16(a, b, n) \
  RESULT(((uint32_t)(uint16_t)(a) + (uint32_t)(uint16_t)(b)) >> 1, n, 16)
#define SUB16(a, b, n) \
  RESULT(((uint32_t)(uint16_t)(a) - (uint32_t)(uint16_t)(b)) >> 1, n, 16)
#define ADD8(a, b, n) \
  RESULT(((uint32_t)(uint8_t)(a) + (uint32_t)(uint8_t)(b)) >> 1, n, 8)
#define SUB8(a, b, n) \
  RESULT(((uint32_t)(uint8_t)(a) - (uint32_t)(uint8_t)(b)) >> 1, n, 8)
#define PFX uh

#include "op_addsub.h"

static inline uint8_t do_usad(uint8_t a, uint8_t b)
{
    if (a > b)
        return a - b;
    else
        return b - a;
}

/* Unsigned sum of absolute byte differences.  */
uint32_t HELPER(usad8)(uint32_t a, uint32_t b)
{
    uint32_t sum;
    sum = do_usad(a, b);
    sum += do_usad(a >> 8, b >> 8);
    sum += do_usad(a >> 16, b >>16);
    sum += do_usad(a >> 24, b >> 24);
    return sum;
}

/* For ARMv6 SEL instruction.  */
uint32_t HELPER(sel_flags)(uint32_t flags, uint32_t a, uint32_t b)
{
    uint32_t mask;

    mask = 0;
    if (flags & 1)
        mask |= 0xff;
    if (flags & 2)
        mask |= 0xff00;
    if (flags & 4)
        mask |= 0xff0000;
    if (flags & 8)
        mask |= 0xff000000;
    return (a & mask) | (b & ~mask);
}

uint32_t HELPER(logicq_cc)(uint64_t val)
{
    return (val >> 32) | (val != 0);
}

/* VFP support.  We follow the convention used for VFP instructions:
   Single precision routines have a "s" suffix, double precision a
   "d" suffix.  */

/* Convert host exception flags to vfp form.  */
static inline int vfp_exceptbits_from_host(int host_bits)
{
    int target_bits = 0;

    if (host_bits & float_flag_invalid)
        target_bits |= 1;
    if (host_bits & float_flag_divbyzero)
        target_bits |= 2;
    if (host_bits & float_flag_overflow)
        target_bits |= 4;
    if (host_bits & (float_flag_underflow | float_flag_output_denormal))
        target_bits |= 8;
    if (host_bits & float_flag_inexact)
        target_bits |= 0x10;
    if (host_bits & float_flag_input_denormal)
        target_bits |= 0x80;
    return target_bits;
}

uint32_t HELPER(vfp_get_fpscr)(CPUARMState *env)
{
    int i;
    uint32_t fpscr;

    fpscr = (env->vfp.xregs[ARM_VFP_FPSCR] & 0xffc8ffff)
            | (env->vfp.vec_len << 16)
            | (env->vfp.vec_stride << 20);
    i = get_float_exception_flags(&env->vfp.fp_status);
    i |= get_float_exception_flags(&env->vfp.standard_fp_status);
    fpscr |= vfp_exceptbits_from_host(i);
    return fpscr;
}

uint32_t vfp_get_fpscr(CPUARMState *env)
{
    return HELPER(vfp_get_fpscr)(env);
}

/* Convert vfp exception flags to target form.  */
static inline int vfp_exceptbits_to_host(int target_bits)
{
    int host_bits = 0;

    if (target_bits & 1)
        host_bits |= float_flag_invalid;
    if (target_bits & 2)
        host_bits |= float_flag_divbyzero;
    if (target_bits & 4)
        host_bits |= float_flag_overflow;
    if (target_bits & 8)
        host_bits |= float_flag_underflow;
    if (target_bits & 0x10)
        host_bits |= float_flag_inexact;
    if (target_bits & 0x80)
        host_bits |= float_flag_input_denormal;
    return host_bits;
}

void HELPER(vfp_set_fpscr)(CPUARMState *env, uint32_t val)
{
    int i;
    uint32_t changed;

    changed = env->vfp.xregs[ARM_VFP_FPSCR];
    env->vfp.xregs[ARM_VFP_FPSCR] = (val & 0xffc8ffff);
    env->vfp.vec_len = (val >> 16) & 7;
    env->vfp.vec_stride = (val >> 20) & 3;

    changed ^= val;
    if (changed & (3 << 22)) {
        i = (val >> 22) & 3;
        switch (i) {
        case 0:
            i = float_round_nearest_even;
            break;
        case 1:
            i = float_round_up;
            break;
        case 2:
            i = float_round_down;
            break;
        case 3:
            i = float_round_to_zero;
            break;
        }
        set_float_rounding_mode(i, &env->vfp.fp_status);
    }
    if (changed & (1 << 24)) {
        set_flush_to_zero((val & (1 << 24)) != 0, &env->vfp.fp_status);
        set_flush_inputs_to_zero((val & (1 << 24)) != 0, &env->vfp.fp_status);
    }
    if (changed & (1 << 25))
        set_default_nan_mode((val & (1 << 25)) != 0, &env->vfp.fp_status);

    i = vfp_exceptbits_to_host(val);
    set_float_exception_flags(i, &env->vfp.fp_status);
    set_float_exception_flags(0, &env->vfp.standard_fp_status);
}

void vfp_set_fpscr(CPUARMState *env, uint32_t val)
{
    HELPER(vfp_set_fpscr)(env, val);
}

#define VFP_HELPER(name, p) HELPER(glue(glue(vfp_,name),p))

#define VFP_BINOP(name) \
float32 VFP_HELPER(name, s)(float32 a, float32 b, void *fpstp) \
{ \
    float_status *fpst = fpstp; \
    return float32_ ## name(a, b, fpst); \
} \
float64 VFP_HELPER(name, d)(float64 a, float64 b, void *fpstp) \
{ \
    float_status *fpst = fpstp; \
    return float64_ ## name(a, b, fpst); \
}
VFP_BINOP(add)
VFP_BINOP(sub)
VFP_BINOP(mul)
VFP_BINOP(div)
#undef VFP_BINOP

float32 VFP_HELPER(neg, s)(float32 a)
{
    return float32_chs(a);
}

float64 VFP_HELPER(neg, d)(float64 a)
{
    return float64_chs(a);
}

float32 VFP_HELPER(abs, s)(float32 a)
{
    return float32_abs(a);
}

float64 VFP_HELPER(abs, d)(float64 a)
{
    return float64_abs(a);
}

float32 VFP_HELPER(sqrt, s)(float32 a, CPUARMState *env)
{
    return float32_sqrt(a, &env->vfp.fp_status);
}

float64 VFP_HELPER(sqrt, d)(float64 a, CPUARMState *env)
{
    return float64_sqrt(a, &env->vfp.fp_status);
}

/* XXX: check quiet/signaling case */
#define DO_VFP_cmp(p, type) \
void VFP_HELPER(cmp, p)(type a, type b, CPUARMState *env)  \
{ \
    uint32_t flags; \
    switch(type ## _compare_quiet(a, b, &env->vfp.fp_status)) { \
    case 0: flags = 0x6; break; \
    case -1: flags = 0x8; break; \
    case 1: flags = 0x2; break; \
    default: case 2: flags = 0x3; break; \
    } \
    env->vfp.xregs[ARM_VFP_FPSCR] = (flags << 28) \
        | (env->vfp.xregs[ARM_VFP_FPSCR] & 0x0fffffff); \
} \
void VFP_HELPER(cmpe, p)(type a, type b, CPUARMState *env) \
{ \
    uint32_t flags; \
    switch(type ## _compare(a, b, &env->vfp.fp_status)) { \
    case 0: flags = 0x6; break; \
    case -1: flags = 0x8; break; \
    case 1: flags = 0x2; break; \
    default: case 2: flags = 0x3; break; \
    } \
    env->vfp.xregs[ARM_VFP_FPSCR] = (flags << 28) \
        | (env->vfp.xregs[ARM_VFP_FPSCR] & 0x0fffffff); \
}
DO_VFP_cmp(s, float32)
DO_VFP_cmp(d, float64)
#undef DO_VFP_cmp

/* Integer to float and float to integer conversions */

#define CONV_ITOF(name, fsz, sign) \
    float##fsz HELPER(name)(uint32_t x, void *fpstp) \
{ \
    float_status *fpst = fpstp; \
    return sign##int32_to_##float##fsz((sign##int32_t)x, fpst); \
}

#define CONV_FTOI(name, fsz, sign, round) \
uint32_t HELPER(name)(float##fsz x, void *fpstp) \
{ \
    float_status *fpst = fpstp; \
    if (float##fsz##_is_any_nan(x)) { \
        float_raise(float_flag_invalid, fpst); \
        return 0; \
    } \
    return float##fsz##_to_##sign##int32##round(x, fpst); \
}

#define FLOAT_CONVS(name, p, fsz, sign) \
CONV_ITOF(vfp_##name##to##p, fsz, sign) \
CONV_FTOI(vfp_to##name##p, fsz, sign, ) \
CONV_FTOI(vfp_to##name##z##p, fsz, sign, _round_to_zero)

FLOAT_CONVS(si, s, 32, )
FLOAT_CONVS(si, d, 64, )
FLOAT_CONVS(ui, s, 32, u)
FLOAT_CONVS(ui, d, 64, u)

#undef CONV_ITOF
#undef CONV_FTOI
#undef FLOAT_CONVS

/* floating point conversion */
float64 VFP_HELPER(fcvtd, s)(float32 x, CPUARMState *env)
{
    float64 r = float32_to_float64(x, &env->vfp.fp_status);
    /* ARM requires that S<->D conversion of any kind of NaN generates
     * a quiet NaN by forcing the most significant frac bit to 1.
     */
    return float64_maybe_silence_nan(r);
}

float32 VFP_HELPER(fcvts, d)(float64 x, CPUARMState *env)
{
    float32 r =  float64_to_float32(x, &env->vfp.fp_status);
    /* ARM requires that S<->D conversion of any kind of NaN generates
     * a quiet NaN by forcing the most significant frac bit to 1.
     */
    return float32_maybe_silence_nan(r);
}

/* VFP3 fixed point conversion.  */
#define VFP_CONV_FIX(name, p, fsz, itype, sign) \
float##fsz HELPER(vfp_##name##to##p)(uint##fsz##_t  x, uint32_t shift, \
                                    void *fpstp) \
{ \
    float_status *fpst = fpstp; \
    float##fsz tmp; \
    tmp = sign##int32_to_##float##fsz((itype##_t)x, fpst); \
    return float##fsz##_scalbn(tmp, -(int)shift, fpst); \
} \
uint##fsz##_t HELPER(vfp_to##name##p)(float##fsz x, uint32_t shift, \
                                       void *fpstp) \
{ \
    float_status *fpst = fpstp; \
    float##fsz tmp; \
    if (float##fsz##_is_any_nan(x)) { \
        float_raise(float_flag_invalid, fpst); \
        return 0; \
    } \
    tmp = float##fsz##_scalbn(x, shift, fpst); \
    return float##fsz##_to_##itype##_round_to_zero(tmp, fpst); \
}

VFP_CONV_FIX(sh, d, 64, int16, )
VFP_CONV_FIX(sl, d, 64, int32, )
VFP_CONV_FIX(uh, d, 64, uint16, u)
VFP_CONV_FIX(ul, d, 64, uint32, u)
VFP_CONV_FIX(sh, s, 32, int16, )
VFP_CONV_FIX(sl, s, 32, int32, )
VFP_CONV_FIX(uh, s, 32, uint16, u)
VFP_CONV_FIX(ul, s, 32, uint32, u)
#undef VFP_CONV_FIX

/* Half precision conversions.  */
static float32 do_fcvt_f16_to_f32(uint32_t a, CPUARMState *env, float_status *s)
{
    int ieee = (env->vfp.xregs[ARM_VFP_FPSCR] & (1 << 26)) == 0;
    float32 r = float16_to_float32(make_float16(a), ieee, s);
    if (ieee) {
        return float32_maybe_silence_nan(r);
    }
    return r;
}

static uint32_t do_fcvt_f32_to_f16(float32 a, CPUARMState *env, float_status *s)
{
    int ieee = (env->vfp.xregs[ARM_VFP_FPSCR] & (1 << 26)) == 0;
    float16 r = float32_to_float16(a, ieee, s);
    if (ieee) {
        r = float16_maybe_silence_nan(r);
    }
    return float16_val(r);
}

float32 HELPER(neon_fcvt_f16_to_f32)(uint32_t a, CPUARMState *env)
{
    return do_fcvt_f16_to_f32(a, env, &env->vfp.standard_fp_status);
}

uint32_t HELPER(neon_fcvt_f32_to_f16)(float32 a, CPUARMState *env)
{
    return do_fcvt_f32_to_f16(a, env, &env->vfp.standard_fp_status);
}

float32 HELPER(vfp_fcvt_f16_to_f32)(uint32_t a, CPUARMState *env)
{
    return do_fcvt_f16_to_f32(a, env, &env->vfp.fp_status);
}

uint32_t HELPER(vfp_fcvt_f32_to_f16)(float32 a, CPUARMState *env)
{
    return do_fcvt_f32_to_f16(a, env, &env->vfp.fp_status);
}

#define float32_two make_float32(0x40000000)
#define float32_three make_float32(0x40400000)
#define float32_one_point_five make_float32(0x3fc00000)

float32 HELPER(recps_f32)(float32 a, float32 b, CPUARMState *env)
{
    float_status *s = &env->vfp.standard_fp_status;
    if ((float32_is_infinity(a) && float32_is_zero_or_denormal(b)) ||
        (float32_is_infinity(b) && float32_is_zero_or_denormal(a))) {
        if (!(float32_is_zero(a) || float32_is_zero(b))) {
            float_raise(float_flag_input_denormal, s);
        }
        return float32_two;
    }
    return float32_sub(float32_two, float32_mul(a, b, s), s);
}

float32 HELPER(rsqrts_f32)(float32 a, float32 b, CPUARMState *env)
{
    float_status *s = &env->vfp.standard_fp_status;
    float32 product;
    if ((float32_is_infinity(a) && float32_is_zero_or_denormal(b)) ||
        (float32_is_infinity(b) && float32_is_zero_or_denormal(a))) {
        if (!(float32_is_zero(a) || float32_is_zero(b))) {
            float_raise(float_flag_input_denormal, s);
        }
        return float32_one_point_five;
    }
    product = float32_mul(a, b, s);
    return float32_div(float32_sub(float32_three, product, s), float32_two, s);
}

/* NEON helpers.  */

/* Constants 256 and 512 are used in some helpers; we avoid relying on
 * int->float conversions at run-time.  */
#define float64_256 make_float64(0x4070000000000000LL)
#define float64_512 make_float64(0x4080000000000000LL)

/* The algorithm that must be used to calculate the estimate
 * is specified by the ARM ARM.
 */
static float64 recip_estimate(float64 a, CPUARMState *env)
{
    /* These calculations mustn't set any fp exception flags,
     * so we use a local copy of the fp_status.
     */
    float_status dummy_status = env->vfp.standard_fp_status;
    float_status *s = &dummy_status;
    /* q = (int)(a * 512.0) */
    float64 q = float64_mul(float64_512, a, s);
    int64_t q_int = float64_to_int64_round_to_zero(q, s);

    /* r = 1.0 / (((double)q + 0.5) / 512.0) */
    q = int64_to_float64(q_int, s);
    q = float64_add(q, float64_half, s);
    q = float64_div(q, float64_512, s);
    q = float64_div(float64_one, q, s);

    /* s = (int)(256.0 * r + 0.5) */
    q = float64_mul(q, float64_256, s);
    q = float64_add(q, float64_half, s);
    q_int = float64_to_int64_round_to_zero(q, s);

    /* return (double)s / 256.0 */
    return float64_div(int64_to_float64(q_int, s), float64_256, s);
}

float32 HELPER(recpe_f32)(float32 a, CPUARMState *env)
{
    float_status *s = &env->vfp.standard_fp_status;
    float64 f64;
    uint32_t val32 = float32_val(a);

    int result_exp;
    int a_exp = (val32  & 0x7f800000) >> 23;
    int sign = val32 & 0x80000000;

    if (float32_is_any_nan(a)) {
        if (float32_is_signaling_nan(a)) {
            float_raise(float_flag_invalid, s);
        }
        return float32_default_nan;
    } else if (float32_is_infinity(a)) {
        return float32_set_sign(float32_zero, float32_is_neg(a));
    } else if (float32_is_zero_or_denormal(a)) {
        if (!float32_is_zero(a)) {
            float_raise(float_flag_input_denormal, s);
        }
        float_raise(float_flag_divbyzero, s);
        return float32_set_sign(float32_infinity, float32_is_neg(a));
    } else if (a_exp >= 253) {
        float_raise(float_flag_underflow, s);
        return float32_set_sign(float32_zero, float32_is_neg(a));
    }

    f64 = make_float64((0x3feULL << 52)
                       | ((int64_t)(val32 & 0x7fffff) << 29));

    result_exp = 253 - a_exp;

    f64 = recip_estimate(f64, env);

    val32 = sign
        | ((result_exp & 0xff) << 23)
        | ((float64_val(f64) >> 29) & 0x7fffff);
    return make_float32(val32);
}

/* The algorithm that must be used to calculate the estimate
 * is specified by the ARM ARM.
 */
static float64 recip_sqrt_estimate(float64 a, CPUARMState *env)
{
    /* These calculations mustn't set any fp exception flags,
     * so we use a local copy of the fp_status.
     */
    float_status dummy_status = env->vfp.standard_fp_status;
    float_status *s = &dummy_status;
    float64 q;
    int64_t q_int;

    if (float64_lt(a, float64_half, s)) {
        /* range 0.25 <= a < 0.5 */

        /* a in units of 1/512 rounded down */
        /* q0 = (int)(a * 512.0);  */
        q = float64_mul(float64_512, a, s);
        q_int = float64_to_int64_round_to_zero(q, s);

        /* reciprocal root r */
        /* r = 1.0 / sqrt(((double)q0 + 0.5) / 512.0);  */
        q = int64_to_float64(q_int, s);
        q = float64_add(q, float64_half, s);
        q = float64_div(q, float64_512, s);
        q = float64_sqrt(q, s);
        q = float64_div(float64_one, q, s);
    } else {
        /* range 0.5 <= a < 1.0 */

        /* a in units of 1/256 rounded down */
        /* q1 = (int)(a * 256.0); */
        q = float64_mul(float64_256, a, s);
        int64_t q_int = float64_to_int64_round_to_zero(q, s);

        /* reciprocal root r */
        /* r = 1.0 /sqrt(((double)q1 + 0.5) / 256); */
        q = int64_to_float64(q_int, s);
        q = float64_add(q, float64_half, s);
        q = float64_div(q, float64_256, s);
        q = float64_sqrt(q, s);
        q = float64_div(float64_one, q, s);
    }
    /* r in units of 1/256 rounded to nearest */
    /* s = (int)(256.0 * r + 0.5); */

    q = float64_mul(q, float64_256,s );
    q = float64_add(q, float64_half, s);
    q_int = float64_to_int64_round_to_zero(q, s);

    /* return (double)s / 256.0;*/
    return float64_div(int64_to_float64(q_int, s), float64_256, s);
}

float32 HELPER(rsqrte_f32)(float32 a, CPUARMState *env)
{
    float_status *s = &env->vfp.standard_fp_status;
    int result_exp;
    float64 f64;
    uint32_t val;
    uint64_t val64;

    val = float32_val(a);

    if (float32_is_any_nan(a)) {
        if (float32_is_signaling_nan(a)) {
            float_raise(float_flag_invalid, s);
        }
        return float32_default_nan;
    } else if (float32_is_zero_or_denormal(a)) {
        if (!float32_is_zero(a)) {
            float_raise(float_flag_input_denormal, s);
        }
        float_raise(float_flag_divbyzero, s);
        return float32_set_sign(float32_infinity, float32_is_neg(a));
    } else if (float32_is_neg(a)) {
        float_raise(float_flag_invalid, s);
        return float32_default_nan;
    } else if (float32_is_infinity(a)) {
        return float32_zero;
    }

    /* Normalize to a double-precision value between 0.25 and 1.0,
     * preserving the parity of the exponent.  */
    if ((val & 0x800000) == 0) {
        f64 = make_float64(((uint64_t)(val & 0x80000000) << 32)
                           | (0x3feULL << 52)
                           | ((uint64_t)(val & 0x7fffff) << 29));
    } else {
        f64 = make_float64(((uint64_t)(val & 0x80000000) << 32)
                           | (0x3fdULL << 52)
                           | ((uint64_t)(val & 0x7fffff) << 29));
    }

    result_exp = (380 - ((val & 0x7f800000) >> 23)) / 2;

    f64 = recip_sqrt_estimate(f64, env);

    val64 = float64_val(f64);

    val = ((result_exp & 0xff) << 23)
        | ((val64 >> 29)  & 0x7fffff);
    return make_float32(val);
}

uint32_t HELPER(recpe_u32)(uint32_t a, CPUARMState *env)
{
    float64 f64;

    if ((a & 0x80000000) == 0) {
        return 0xffffffff;
    }

    f64 = make_float64((0x3feULL << 52)
                       | ((int64_t)(a & 0x7fffffff) << 21));

    f64 = recip_estimate (f64, env);

    return 0x80000000 | ((float64_val(f64) >> 21) & 0x7fffffff);
}

uint32_t HELPER(rsqrte_u32)(uint32_t a, CPUARMState *env)
{
    float64 f64;

    if ((a & 0xc0000000) == 0) {
        return 0xffffffff;
    }

    if (a & 0x80000000) {
        f64 = make_float64((0x3feULL << 52)
                           | ((uint64_t)(a & 0x7fffffff) << 21));
    } else { /* bits 31-30 == '01' */
        f64 = make_float64((0x3fdULL << 52)
                           | ((uint64_t)(a & 0x3fffffff) << 22));
    }

    f64 = recip_sqrt_estimate(f64, env);

    return 0x80000000 | ((float64_val(f64) >> 21) & 0x7fffffff);
}

/* VFPv4 fused multiply-accumulate */
float32 VFP_HELPER(muladd, s)(float32 a, float32 b, float32 c, void *fpstp)
{
    float_status *fpst = fpstp;
    return float32_muladd(a, b, c, 0, fpst);
}

float64 VFP_HELPER(muladd, d)(float64 a, float64 b, float64 c, void *fpstp)
{
    float_status *fpst = fpstp;
    return float64_muladd(a, b, c, 0, fpst);
}