/*
 *  Virtual page mapping
 *
 *  Copyright (c) 2003 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, see <http://www.gnu.org/licenses/>.
 */

#include "qemu/osdep.h"
#include "qemu-common.h"
#include "qapi/error.h"

#include "qemu/cutils.h"
#include "cpu.h"
#include "exec/exec-all.h"
#include "exec/target_page.h"
#include "tcg.h"
#include "hw/qdev-core.h"
#include "hw/qdev-properties.h"
#if !defined(CONFIG_USER_ONLY)
#include "hw/boards.h"
#include "hw/xen/xen.h"
#endif
#include "sysemu/kvm.h"
#include "sysemu/sysemu.h"
#include "sysemu/tcg.h"
#include "qemu/timer.h"
#include "qemu/config-file.h"
#include "qemu/error-report.h"
#include "qemu/qemu-print.h"
#if defined(CONFIG_USER_ONLY)
#include "qemu.h"
#else /* !CONFIG_USER_ONLY */
#include "hw/hw.h"
#include "exec/memory.h"
#include "exec/ioport.h"
#include "sysemu/dma.h"
#include "sysemu/numa.h"
#include "sysemu/hw_accel.h"
#include "exec/address-spaces.h"
#include "sysemu/xen-mapcache.h"
#include "trace-root.h"

#ifdef CONFIG_FALLOCATE_PUNCH_HOLE
#include <linux/falloc.h>
#endif

#endif
#include "qemu/rcu_queue.h"
#include "qemu/main-loop.h"
#include "translate-all.h"
#include "sysemu/replay.h"

#include "exec/memory-internal.h"
#include "exec/ram_addr.h"
#include "exec/log.h"

#include "migration/vmstate.h"

#include "qemu/range.h"
#ifndef _WIN32
#include "qemu/mmap-alloc.h"
#endif

#include "monitor/monitor.h"

//#define DEBUG_SUBPAGE

#if !defined(CONFIG_USER_ONLY)
/* ram_list is read under rcu_read_lock()/rcu_read_unlock().  Writes
 * are protected by the ramlist lock.
 */
RAMList ram_list = { .blocks = QLIST_HEAD_INITIALIZER(ram_list.blocks) };

static MemoryRegion *system_memory;
static MemoryRegion *system_io;

AddressSpace address_space_io;
AddressSpace address_space_memory;

MemoryRegion io_mem_rom, io_mem_notdirty;
static MemoryRegion io_mem_unassigned;
#endif

#ifdef TARGET_PAGE_BITS_VARY
int target_page_bits;
bool target_page_bits_decided;
#endif

CPUTailQ cpus = QTAILQ_HEAD_INITIALIZER(cpus);

/* current CPU in the current thread. It is only valid inside
   cpu_exec() */
__thread CPUState *current_cpu;
/* 0 = Do not count executed instructions.
   1 = Precise instruction counting.
   2 = Adaptive rate instruction counting.  */
int use_icount;

uintptr_t qemu_host_page_size;
intptr_t qemu_host_page_mask;

bool set_preferred_target_page_bits(int bits)
{
    /* The target page size is the lowest common denominator for all
     * the CPUs in the system, so we can only make it smaller, never
     * larger. And we can't make it smaller once we've committed to
     * a particular size.
     */
#ifdef TARGET_PAGE_BITS_VARY
    assert(bits >= TARGET_PAGE_BITS_MIN);
    if (target_page_bits == 0 || target_page_bits > bits) {
        if (target_page_bits_decided) {
            return false;
        }
        target_page_bits = bits;
    }
#endif
    return true;
}

#if !defined(CONFIG_USER_ONLY)

static void finalize_target_page_bits(void)
{
#ifdef TARGET_PAGE_BITS_VARY
    if (target_page_bits == 0) {
        target_page_bits = TARGET_PAGE_BITS_MIN;
    }
    target_page_bits_decided = true;
#endif
}

typedef struct PhysPageEntry PhysPageEntry;

struct PhysPageEntry {
    /* How many bits skip to next level (in units of L2_SIZE). 0 for a leaf. */
    uint32_t skip : 6;
     /* index into phys_sections (!skip) or phys_map_nodes (skip) */
    uint32_t ptr : 26;
};

#define PHYS_MAP_NODE_NIL (((uint32_t)~0) >> 6)

/* Size of the L2 (and L3, etc) page tables.  */
#define ADDR_SPACE_BITS 64

#define P_L2_BITS 9
#define P_L2_SIZE (1 << P_L2_BITS)

#define P_L2_LEVELS (((ADDR_SPACE_BITS - TARGET_PAGE_BITS - 1) / P_L2_BITS) + 1)

typedef PhysPageEntry Node[P_L2_SIZE];

typedef struct PhysPageMap {
    struct rcu_head rcu;

    unsigned sections_nb;
    unsigned sections_nb_alloc;
    unsigned nodes_nb;
    unsigned nodes_nb_alloc;
    Node *nodes;
    MemoryRegionSection *sections;
} PhysPageMap;

struct AddressSpaceDispatch {
    MemoryRegionSection *mru_section;
    /* This is a multi-level map on the physical address space.
     * The bottom level has pointers to MemoryRegionSections.
     */
    PhysPageEntry phys_map;
    PhysPageMap map;
};

#define SUBPAGE_IDX(addr) ((addr) & ~TARGET_PAGE_MASK)
typedef struct subpage_t {
    MemoryRegion iomem;
    FlatView *fv;
    hwaddr base;
    uint16_t sub_section[];
} subpage_t;

#define PHYS_SECTION_UNASSIGNED 0
#define PHYS_SECTION_NOTDIRTY 1
#define PHYS_SECTION_ROM 2
#define PHYS_SECTION_WATCH 3

static void io_mem_init(void);
static void memory_map_init(void);
static void tcg_commit(MemoryListener *listener);

static MemoryRegion io_mem_watch;

/**
 * CPUAddressSpace: all the information a CPU needs about an AddressSpace
 * @cpu: the CPU whose AddressSpace this is
 * @as: the AddressSpace itself
 * @memory_dispatch: its dispatch pointer (cached, RCU protected)
 * @tcg_as_listener: listener for tracking changes to the AddressSpace
 */
struct CPUAddressSpace {
    CPUState *cpu;
    AddressSpace *as;
    struct AddressSpaceDispatch *memory_dispatch;
    MemoryListener tcg_as_listener;
};

struct DirtyBitmapSnapshot {
    ram_addr_t start;
    ram_addr_t end;
    unsigned long dirty[];
};

#endif

#if !defined(CONFIG_USER_ONLY)

static void phys_map_node_reserve(PhysPageMap *map, unsigned nodes)
{
    static unsigned alloc_hint = 16;
    if (map->nodes_nb + nodes > map->nodes_nb_alloc) {
        map->nodes_nb_alloc = MAX(map->nodes_nb_alloc, alloc_hint);
        map->nodes_nb_alloc = MAX(map->nodes_nb_alloc, map->nodes_nb + nodes);
        map->nodes = g_renew(Node, map->nodes, map->nodes_nb_alloc);
        alloc_hint = map->nodes_nb_alloc;
    }
}

static uint32_t phys_map_node_alloc(PhysPageMap *map, bool leaf)
{
    unsigned i;
    uint32_t ret;
    PhysPageEntry e;
    PhysPageEntry *p;

    ret = map->nodes_nb++;
    p = map->nodes[ret];
    assert(ret != PHYS_MAP_NODE_NIL);
    assert(ret != map->nodes_nb_alloc);

    e.skip = leaf ? 0 : 1;
    e.ptr = leaf ? PHYS_SECTION_UNASSIGNED : PHYS_MAP_NODE_NIL;
    for (i = 0; i < P_L2_SIZE; ++i) {
        memcpy(&p[i], &e, sizeof(e));
    }
    return ret;
}

static void phys_page_set_level(PhysPageMap *map, PhysPageEntry *lp,
                                hwaddr *index, hwaddr *nb, uint16_t leaf,
                                int level)
{
    PhysPageEntry *p;
    hwaddr step = (hwaddr)1 << (level * P_L2_BITS);

    if (lp->skip && lp->ptr == PHYS_MAP_NODE_NIL) {
        lp->ptr = phys_map_node_alloc(map, level == 0);
    }
    p = map->nodes[lp->ptr];
    lp = &p[(*index >> (level * P_L2_BITS)) & (P_L2_SIZE - 1)];

    while (*nb && lp < &p[P_L2_SIZE]) {
        if ((*index & (step - 1)) == 0 && *nb >= step) {
            lp->skip = 0;
            lp->ptr = leaf;
            *index += step;
            *nb -= step;
        } else {
            phys_page_set_level(map, lp, index, nb, leaf, level - 1);
        }
        ++lp;
    }
}

static void phys_page_set(AddressSpaceDispatch *d,
                          hwaddr index, hwaddr nb,
                          uint16_t leaf)
{
    /* Wildly overreserve - it doesn't matter much. */
    phys_map_node_reserve(&d->map, 3 * P_L2_LEVELS);

    phys_page_set_level(&d->map, &d->phys_map, &index, &nb, leaf, P_L2_LEVELS - 1);
}

/* Compact a non leaf page entry. Simply detect that the entry has a single child,
 * and update our entry so we can skip it and go directly to the destination.
 */
static void phys_page_compact(PhysPageEntry *lp, Node *nodes)
{
    unsigned valid_ptr = P_L2_SIZE;
    int valid = 0;
    PhysPageEntry *p;
    int i;

    if (lp->ptr == PHYS_MAP_NODE_NIL) {
        return;
    }

    p = nodes[lp->ptr];
    for (i = 0; i < P_L2_SIZE; i++) {
        if (p[i].ptr == PHYS_MAP_NODE_NIL) {
            continue;
        }

        valid_ptr = i;
        valid++;
        if (p[i].skip) {
            phys_page_compact(&p[i], nodes);
        }
    }

    /* We can only compress if there's only one child. */
    if (valid != 1) {
        return;
    }

    assert(valid_ptr < P_L2_SIZE);

    /* Don't compress if it won't fit in the # of bits we have. */
    if (lp->skip + p[valid_ptr].skip >= (1 << 3)) {
        return;
    }

    lp->ptr = p[valid_ptr].ptr;
    if (!p[valid_ptr].skip) {
        /* If our only child is a leaf, make this a leaf. */
        /* By design, we should have made this node a leaf to begin with so we
         * should never reach here.
         * But since it's so simple to handle this, let's do it just in case we
         * change this rule.
         */
        lp->skip = 0;
    } else {
        lp->skip += p[valid_ptr].skip;
    }
}

void address_space_dispatch_compact(AddressSpaceDispatch *d)
{
    if (d->phys_map.skip) {
        phys_page_compact(&d->phys_map, d->map.nodes);
    }
}

static inline bool section_covers_addr(const MemoryRegionSection *section,
                                       hwaddr addr)
{
    /* Memory topology clips a memory region to [0, 2^64); size.hi > 0 means
     * the section must cover the entire address space.
     */
    return int128_gethi(section->size) ||
           range_covers_byte(section->offset_within_address_space,
                             int128_getlo(section->size), addr);
}

static MemoryRegionSection *phys_page_find(AddressSpaceDispatch *d, hwaddr addr)
{
    PhysPageEntry lp = d->phys_map, *p;
    Node *nodes = d->map.nodes;
    MemoryRegionSection *sections = d->map.sections;
    hwaddr index = addr >> TARGET_PAGE_BITS;
    int i;

    for (i = P_L2_LEVELS; lp.skip && (i -= lp.skip) >= 0;) {
        if (lp.ptr == PHYS_MAP_NODE_NIL) {
            return &sections[PHYS_SECTION_UNASSIGNED];
        }
        p = nodes[lp.ptr];
        lp = p[(index >> (i * P_L2_BITS)) & (P_L2_SIZE - 1)];
    }

    if (section_covers_addr(&sections[lp.ptr], addr)) {
        return &sections[lp.ptr];
    } else {
        return &sections[PHYS_SECTION_UNASSIGNED];
    }
}

/* Called from RCU critical section */
static MemoryRegionSection *address_space_lookup_region(AddressSpaceDispatch *d,
                                                        hwaddr addr,
                                                        bool resolve_subpage)
{
    MemoryRegionSection *section = atomic_read(&d->mru_section);
    subpage_t *subpage;

    if (!section || section == &d->map.sections[PHYS_SECTION_UNASSIGNED] ||
        !section_covers_addr(section, addr)) {
        section = phys_page_find(d, addr);
        atomic_set(&d->mru_section, section);
    }
    if (resolve_subpage && section->mr->subpage) {
        subpage = container_of(section->mr, subpage_t, iomem);
        section = &d->map.sections[subpage->sub_section[SUBPAGE_IDX(addr)]];
    }
    return section;
}

/* Called from RCU critical section */
static MemoryRegionSection *
address_space_translate_internal(AddressSpaceDispatch *d, hwaddr addr, hwaddr *xlat,
                                 hwaddr *plen, bool resolve_subpage)
{
    MemoryRegionSection *section;
    MemoryRegion *mr;
    Int128 diff;

    section = address_space_lookup_region(d, addr, resolve_subpage);
    /* Compute offset within MemoryRegionSection */
    addr -= section->offset_within_address_space;

    /* Compute offset within MemoryRegion */
    *xlat = addr + section->offset_within_region;

    mr = section->mr;

    /* MMIO registers can be expected to perform full-width accesses based only
     * on their address, without considering adjacent registers that could
     * decode to completely different MemoryRegions.  When such registers
     * exist (e.g. I/O ports 0xcf8 and 0xcf9 on most PC chipsets), MMIO
     * regions overlap wildly.  For this reason we cannot clamp the accesses
     * here.
     *
     * If the length is small (as is the case for address_space_ldl/stl),
     * everything works fine.  If the incoming length is large, however,
     * the caller really has to do the clamping through memory_access_size.
     */
    if (memory_region_is_ram(mr)) {
        diff = int128_sub(section->size, int128_make64(addr));
        *plen = int128_get64(int128_min(diff, int128_make64(*plen)));
    }
    return section;
}

/**
 * address_space_translate_iommu - translate an address through an IOMMU
 * memory region and then through the target address space.
 *
 * @iommu_mr: the IOMMU memory region that we start the translation from
 * @addr: the address to be translated through the MMU
 * @xlat: the translated address offset within the destination memory region.
 *        It cannot be %NULL.
 * @plen_out: valid read/write length of the translated address. It
 *            cannot be %NULL.
 * @page_mask_out: page mask for the translated address. This
 *            should only be meaningful for IOMMU translated
 *            addresses, since there may be huge pages that this bit
 *            would tell. It can be %NULL if we don't care about it.
 * @is_write: whether the translation operation is for write
 * @is_mmio: whether this can be MMIO, set true if it can
 * @target_as: the address space targeted by the IOMMU
 * @attrs: transaction attributes
 *
 * This function is called from RCU critical section.  It is the common
 * part of flatview_do_translate and address_space_translate_cached.
 */
static MemoryRegionSection address_space_translate_iommu(IOMMUMemoryRegion *iommu_mr,
                                                         hwaddr *xlat,
                                                         hwaddr *plen_out,
                                                         hwaddr *page_mask_out,
                                                         bool is_write,
                                                         bool is_mmio,
                                                         AddressSpace **target_as,
                                                         MemTxAttrs attrs)
{
    MemoryRegionSection *section;
    hwaddr page_mask = (hwaddr)-1;

    do {
        hwaddr addr = *xlat;
        IOMMUMemoryRegionClass *imrc = memory_region_get_iommu_class_nocheck(iommu_mr);
        int iommu_idx = 0;
        IOMMUTLBEntry iotlb;

        if (imrc->attrs_to_index) {
            iommu_idx = imrc->attrs_to_index(iommu_mr, attrs);
        }

        iotlb = imrc->translate(iommu_mr, addr, is_write ?
                                IOMMU_WO : IOMMU_RO, iommu_idx);

        if (!(iotlb.perm & (1 << is_write))) {
            goto unassigned;
        }

        addr = ((iotlb.translated_addr & ~iotlb.addr_mask)
                | (addr & iotlb.addr_mask));
        page_mask &= iotlb.addr_mask;
        *plen_out = MIN(*plen_out, (addr | iotlb.addr_mask) - addr + 1);
        *target_as = iotlb.target_as;

        section = address_space_translate_internal(
                address_space_to_dispatch(iotlb.target_as), addr, xlat,
                plen_out, is_mmio);

        iommu_mr = memory_region_get_iommu(section->mr);
    } while (unlikely(iommu_mr));

    if (page_mask_out) {
        *page_mask_out = page_mask;
    }
    return *section;

unassigned:
    return (MemoryRegionSection) { .mr = &io_mem_unassigned };
}

/**
 * flatview_do_translate - translate an address in FlatView
 *
 * @fv: the flat view that we want to translate on
 * @addr: the address to be translated in above address space
 * @xlat: the translated address offset within memory region. It
 *        cannot be @NULL.
 * @plen_out: valid read/write length of the translated address. It
 *            can be @NULL when we don't care about it.
 * @page_mask_out: page mask for the translated address. This
 *            should only be meaningful for IOMMU translated
 *            addresses, since there may be huge pages that this bit
 *            would tell. It can be @NULL if we don't care about it.
 * @is_write: whether the translation operation is for write
 * @is_mmio: whether this can be MMIO, set true if it can
 * @target_as: the address space targeted by the IOMMU
 * @attrs: memory transaction attributes
 *
 * This function is called from RCU critical section
 */
static MemoryRegionSection flatview_do_translate(FlatView *fv,
                                                 hwaddr addr,
                                                 hwaddr *xlat,
                                                 hwaddr *plen_out,
                                                 hwaddr *page_mask_out,
                                                 bool is_write,
                                                 bool is_mmio,
                                                 AddressSpace **target_as,
                                                 MemTxAttrs attrs)
{
    MemoryRegionSection *section;
    IOMMUMemoryRegion *iommu_mr;
    hwaddr plen = (hwaddr)(-1);

    if (!plen_out) {
        plen_out = &plen;
    }

    section = address_space_translate_internal(
            flatview_to_dispatch(fv), addr, xlat,
            plen_out, is_mmio);

    iommu_mr = memory_region_get_iommu(section->mr);
    if (unlikely(iommu_mr)) {
        return address_space_translate_iommu(iommu_mr, xlat,
                                             plen_out, page_mask_out,
                                             is_write, is_mmio,
                                             target_as, attrs);
    }
    if (page_mask_out) {
        /* Not behind an IOMMU, use default page size. */
        *page_mask_out = ~TARGET_PAGE_MASK;
    }

    return *section;
}

/* Called from RCU critical section */
IOMMUTLBEntry address_space_get_iotlb_entry(AddressSpace *as, hwaddr addr,
                                            bool is_write, MemTxAttrs attrs)
{
    MemoryRegionSection section;
    hwaddr xlat, page_mask;

    /*
     * This can never be MMIO, and we don't really care about plen,
     * but page mask.
     */
    section = flatview_do_translate(address_space_to_flatview(as), addr, &xlat,
                                    NULL, &page_mask, is_write, false, &as,
                                    attrs);

    /* Illegal translation */
    if (section.mr == &io_mem_unassigned) {
        goto iotlb_fail;
    }

    /* Convert memory region offset into address space offset */
    xlat += section.offset_within_address_space -
        section.offset_within_region;

    return (IOMMUTLBEntry) {
        .target_as = as,
        .iova = addr & ~page_mask,
        .translated_addr = xlat & ~page_mask,
        .addr_mask = page_mask,
        /* IOTLBs are for DMAs, and DMA only allows on RAMs. */
        .perm = IOMMU_RW,
    };

iotlb_fail:
    return (IOMMUTLBEntry) {0};
}

/* Called from RCU critical section */
MemoryRegion *flatview_translate(FlatView *fv, hwaddr addr, hwaddr *xlat,
                                 hwaddr *plen, bool is_write,
                                 MemTxAttrs attrs)
{
    MemoryRegion *mr;
    MemoryRegionSection section;
    AddressSpace *as = NULL;

    /* This can be MMIO, so setup MMIO bit. */
    section = flatview_do_translate(fv, addr, xlat, plen, NULL,
                                    is_write, true, &as, attrs);
    mr = section.mr;

    if (xen_enabled() && memory_access_is_direct(mr, is_write)) {
        hwaddr page = ((addr & TARGET_PAGE_MASK) + TARGET_PAGE_SIZE) - addr;
        *plen = MIN(page, *plen);
    }

    return mr;
}

typedef struct TCGIOMMUNotifier {
    IOMMUNotifier n;
    MemoryRegion *mr;
    CPUState *cpu;
    int iommu_idx;
    bool active;
} TCGIOMMUNotifier;

static void tcg_iommu_unmap_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
{
    TCGIOMMUNotifier *notifier = container_of(n, TCGIOMMUNotifier, n);

    if (!notifier->active) {
        return;
    }
    tlb_flush(notifier->cpu);
    notifier->active = false;
    /* We leave the notifier struct on the list to avoid reallocating it later.
     * Generally the number of IOMMUs a CPU deals with will be small.
     * In any case we can't unregister the iommu notifier from a notify
     * callback.
     */
}

static void tcg_register_iommu_notifier(CPUState *cpu,
                                        IOMMUMemoryRegion *iommu_mr,
                                        int iommu_idx)
{
    /* Make sure this CPU has an IOMMU notifier registered for this
     * IOMMU/IOMMU index combination, so that we can flush its TLB
     * when the IOMMU tells us the mappings we've cached have changed.
     */
    MemoryRegion *mr = MEMORY_REGION(iommu_mr);
    TCGIOMMUNotifier *notifier;
    int i;

    for (i = 0; i < cpu->iommu_notifiers->len; i++) {
        notifier = g_array_index(cpu->iommu_notifiers, TCGIOMMUNotifier *, i);
        if (notifier->mr == mr && notifier->iommu_idx == iommu_idx) {
            break;
        }
    }
    if (i == cpu->iommu_notifiers->len) {
        /* Not found, add a new entry at the end of the array */
        cpu->iommu_notifiers = g_array_set_size(cpu->iommu_notifiers, i + 1);
        notifier = g_new0(TCGIOMMUNotifier, 1);
        g_array_index(cpu->iommu_notifiers, TCGIOMMUNotifier *, i) = notifier;

        notifier->mr = mr;
        notifier->iommu_idx = iommu_idx;
        notifier->cpu = cpu;
        /* Rather than trying to register interest in the specific part
         * of the iommu's address space that we've accessed and then
         * expand it later as subsequent accesses touch more of it, we
         * just register interest in the whole thing, on the assumption
         * that iommu reconfiguration will be rare.
         */
        iommu_notifier_init(&notifier->n,
                            tcg_iommu_unmap_notify,
                            IOMMU_NOTIFIER_UNMAP,
                            0,
                            HWADDR_MAX,
                            iommu_idx);
        memory_region_register_iommu_notifier(notifier->mr, &notifier->n);
    }

    if (!notifier->active) {
        notifier->active = true;
    }
}

static void tcg_iommu_free_notifier_list(CPUState *cpu)
{
    /* Destroy the CPU's notifier list */
    int i;
    TCGIOMMUNotifier *notifier;

    for (i = 0; i < cpu->iommu_notifiers->len; i++) {
        notifier = g_array_index(cpu->iommu_notifiers, TCGIOMMUNotifier *, i);
        memory_region_unregister_iommu_notifier(notifier->mr, &notifier->n);
        g_free(notifier);
    }
    g_array_free(cpu->iommu_notifiers, true);
}

/* Called from RCU critical section */
MemoryRegionSection *
address_space_translate_for_iotlb(CPUState *cpu, int asidx, hwaddr addr,
                                  hwaddr *xlat, hwaddr *plen,
                                  MemTxAttrs attrs, int *prot)
{
    MemoryRegionSection *section;
    IOMMUMemoryRegion *iommu_mr;
    IOMMUMemoryRegionClass *imrc;
    IOMMUTLBEntry iotlb;
    int iommu_idx;
    AddressSpaceDispatch *d = atomic_rcu_read(&cpu->cpu_ases[asidx].memory_dispatch);

    for (;;) {
        section = address_space_translate_internal(d, addr, &addr, plen, false);

        iommu_mr = memory_region_get_iommu(section->mr);
        if (!iommu_mr) {
            break;
        }

        imrc = memory_region_get_iommu_class_nocheck(iommu_mr);

        iommu_idx = imrc->attrs_to_index(iommu_mr, attrs);
        tcg_register_iommu_notifier(cpu, iommu_mr, iommu_idx);
        /* We need all the permissions, so pass IOMMU_NONE so the IOMMU
         * doesn't short-cut its translation table walk.
         */
        iotlb = imrc->translate(iommu_mr, addr, IOMMU_NONE, iommu_idx);
        addr = ((iotlb.translated_addr & ~iotlb.addr_mask)
                | (addr & iotlb.addr_mask));
        /* Update the caller's prot bits to remove permissions the IOMMU
         * is giving us a failure response for. If we get down to no
         * permissions left at all we can give up now.
         */
        if (!(iotlb.perm & IOMMU_RO)) {
            *prot &= ~(PAGE_READ | PAGE_EXEC);
        }
        if (!(iotlb.perm & IOMMU_WO)) {
            *prot &= ~PAGE_WRITE;
        }

        if (!*prot) {
            goto translate_fail;
        }

        d = flatview_to_dispatch(address_space_to_flatview(iotlb.target_as));
    }

    assert(!memory_region_is_iommu(section->mr));
    *xlat = addr;
    return section;

translate_fail:
    return &d->map.sections[PHYS_SECTION_UNASSIGNED];
}
#endif

#if !defined(CONFIG_USER_ONLY)

static int cpu_common_post_load(void *opaque, int version_id)
{
    CPUState *cpu = opaque;

    /* 0x01 was CPU_INTERRUPT_EXIT. This line can be removed when the
       version_id is increased. */
    cpu->interrupt_request &= ~0x01;
    tlb_flush(cpu);

    /* loadvm has just updated the content of RAM, bypassing the
     * usual mechanisms that ensure we flush TBs for writes to
     * memory we've translated code from. So we must flush all TBs,
     * which will now be stale.
     */
    tb_flush(cpu);

    return 0;
}

static int cpu_common_pre_load(void *opaque)
{
    CPUState *cpu = opaque;

    cpu->exception_index = -1;

    return 0;
}

static bool cpu_common_exception_index_needed(void *opaque)
{
    CPUState *cpu = opaque;

    return tcg_enabled() && cpu->exception_index != -1;
}

static const VMStateDescription vmstate_cpu_common_exception_index = {
    .name = "cpu_common/exception_index",
    .version_id = 1,
    .minimum_version_id = 1,
    .needed = cpu_common_exception_index_needed,
    .fields = (VMStateField[]) {
        VMSTATE_INT32(exception_index, CPUState),
        VMSTATE_END_OF_LIST()
    }
};

static bool cpu_common_crash_occurred_needed(void *opaque)
{
    CPUState *cpu = opaque;

    return cpu->crash_occurred;
}

static const VMStateDescription vmstate_cpu_common_crash_occurred = {
    .name = "cpu_common/crash_occurred",
    .version_id = 1,
    .minimum_version_id = 1,
    .needed = cpu_common_crash_occurred_needed,
    .fields = (VMStateField[]) {
        VMSTATE_BOOL(crash_occurred, CPUState),
        VMSTATE_END_OF_LIST()
    }
};

const VMStateDescription vmstate_cpu_common = {
    .name = "cpu_common",
    .version_id = 1,
    .minimum_version_id = 1,
    .pre_load = cpu_common_pre_load,
    .post_load = cpu_common_post_load,
    .fields = (VMStateField[]) {
        VMSTATE_UINT32(halted, CPUState),
        VMSTATE_UINT32(interrupt_request, CPUState),
        VMSTATE_END_OF_LIST()
    },
    .subsections = (const VMStateDescription*[]) {
        &vmstate_cpu_common_exception_index,
        &vmstate_cpu_common_crash_occurred,
        NULL
    }
};

#endif

CPUState *qemu_get_cpu(int index)
{
    CPUState *cpu;

    CPU_FOREACH(cpu) {
        if (cpu->cpu_index == index) {
            return cpu;
        }
    }

    return NULL;
}

#if !defined(CONFIG_USER_ONLY)
void cpu_address_space_init(CPUState *cpu, int asidx,
                            const char *prefix, MemoryRegion *mr)
{
    CPUAddressSpace *newas;
    AddressSpace *as = g_new0(AddressSpace, 1);
    char *as_name;

    assert(mr);
    as_name = g_strdup_printf("%s-%d", prefix, cpu->cpu_index);
    address_space_init(as, mr, as_name);
    g_free(as_name);

    /* Target code should have set num_ases before calling us */
    assert(asidx < cpu->num_ases);

    if (asidx == 0) {
        /* address space 0 gets the convenience alias */
        cpu->as = as;
    }

    /* KVM cannot currently support multiple address spaces. */
    assert(asidx == 0 || !kvm_enabled());

    if (!cpu->cpu_ases) {
        cpu->cpu_ases = g_new0(CPUAddressSpace, cpu->num_ases);
    }

    newas = &cpu->cpu_ases[asidx];
    newas->cpu = cpu;
    newas->as = as;
    if (tcg_enabled()) {
        newas->tcg_as_listener.commit = tcg_commit;
        memory_listener_register(&newas->tcg_as_listener, as);
    }
}

AddressSpace *cpu_get_address_space(CPUState *cpu, int asidx)
{
    /* Return the AddressSpace corresponding to the specified index */
    return cpu->cpu_ases[asidx].as;
}
#endif

void cpu_exec_unrealizefn(CPUState *cpu)
{
    CPUClass *cc = CPU_GET_CLASS(cpu);

    cpu_list_remove(cpu);

    if (cc->vmsd != NULL) {
        vmstate_unregister(NULL, cc->vmsd, cpu);
    }
    if (qdev_get_vmsd(DEVICE(cpu)) == NULL) {
        vmstate_unregister(NULL, &vmstate_cpu_common, cpu);
    }
#ifndef CONFIG_USER_ONLY
    tcg_iommu_free_notifier_list(cpu);
#endif
}

Property cpu_common_props[] = {
#ifndef CONFIG_USER_ONLY
    /* Create a memory property for softmmu CPU object,
     * so users can wire up its memory. (This can't go in qom/cpu.c
     * because that file is compiled only once for both user-mode
     * and system builds.) The default if no link is set up is to use
     * the system address space.
     */
    DEFINE_PROP_LINK("memory", CPUState, memory, TYPE_MEMORY_REGION,
                     MemoryRegion *),
#endif
    DEFINE_PROP_END_OF_LIST(),
};

void cpu_exec_initfn(CPUState *cpu)
{
    cpu->as = NULL;
    cpu->num_ases = 0;

#ifndef CONFIG_USER_ONLY
    cpu->thread_id = qemu_get_thread_id();
    cpu->memory = system_memory;
    object_ref(OBJECT(cpu->memory));
#endif
}

void cpu_exec_realizefn(CPUState *cpu, Error **errp)
{
    CPUClass *cc = CPU_GET_CLASS(cpu);
    static bool tcg_target_initialized;

    cpu_list_add(cpu);

    if (tcg_enabled() && !tcg_target_initialized) {
        tcg_target_initialized = true;
        cc->tcg_initialize();
    }
    tlb_init(cpu);

#ifndef CONFIG_USER_ONLY
    if (qdev_get_vmsd(DEVICE(cpu)) == NULL) {
        vmstate_register(NULL, cpu->cpu_index, &vmstate_cpu_common, cpu);
    }
    if (cc->vmsd != NULL) {
        vmstate_register(NULL, cpu->cpu_index, cc->vmsd, cpu);
    }

    cpu->iommu_notifiers = g_array_new(false, true, sizeof(TCGIOMMUNotifier *));
#endif
}

const char *parse_cpu_option(const char *cpu_option)
{
    ObjectClass *oc;
    CPUClass *cc;
    gchar **model_pieces;
    const char *cpu_type;

    model_pieces = g_strsplit(cpu_option, ",", 2);
    if (!model_pieces[0]) {
        error_report("-cpu option cannot be empty");
        exit(1);
    }

    oc = cpu_class_by_name(CPU_RESOLVING_TYPE, model_pieces[0]);
    if (oc == NULL) {
        error_report("unable to find CPU model '%s'", model_pieces[0]);
        g_strfreev(model_pieces);
        exit(EXIT_FAILURE);
    }

    cpu_type = object_class_get_name(oc);
    cc = CPU_CLASS(oc);
    cc->parse_features(cpu_type, model_pieces[1], &error_fatal);
    g_strfreev(model_pieces);
    return cpu_type;
}

#if defined(CONFIG_USER_ONLY)
void tb_invalidate_phys_addr(target_ulong addr)
{
    mmap_lock();
    tb_invalidate_phys_page_range(addr, addr + 1, 0);
    mmap_unlock();
}

static void breakpoint_invalidate(CPUState *cpu, target_ulong pc)
{
    tb_invalidate_phys_addr(pc);
}
#else
void tb_invalidate_phys_addr(AddressSpace *as, hwaddr addr, MemTxAttrs attrs)
{
    ram_addr_t ram_addr;
    MemoryRegion *mr;
    hwaddr l = 1;

    if (!tcg_enabled()) {
        return;
    }

    rcu_read_lock();
    mr = address_space_translate(as, addr, &addr, &l, false, attrs);
    if (!(memory_region_is_ram(mr)
          || memory_region_is_romd(mr))) {
        rcu_read_unlock();
        return;
    }
    ram_addr = memory_region_get_ram_addr(mr) + addr;
    tb_invalidate_phys_page_range(ram_addr, ram_addr + 1, 0);
    rcu_read_unlock();
}

static void breakpoint_invalidate(CPUState *cpu, target_ulong pc)
{
    MemTxAttrs attrs;
    hwaddr phys = cpu_get_phys_page_attrs_debug(cpu, pc, &attrs);
    int asidx = cpu_asidx_from_attrs(cpu, attrs);
    if (phys != -1) {
        /* Locks grabbed by tb_invalidate_phys_addr */
        tb_invalidate_phys_addr(cpu->cpu_ases[asidx].as,
                                phys | (pc & ~TARGET_PAGE_MASK), attrs);
    }
}
#endif

#if defined(CONFIG_USER_ONLY)
void cpu_watchpoint_remove_all(CPUState *cpu, int mask)

{
}

int cpu_watchpoint_remove(CPUState *cpu, vaddr addr, vaddr len,
                          int flags)
{
    return -ENOSYS;
}

void cpu_watchpoint_remove_by_ref(CPUState *cpu, CPUWatchpoint *watchpoint)
{
}

int cpu_watchpoint_insert(CPUState *cpu, vaddr addr, vaddr len,
                          int flags, CPUWatchpoint **watchpoint)
{
    return -ENOSYS;
}
#else
/* Add a watchpoint.  */
int cpu_watchpoint_insert(CPUState *cpu, vaddr addr, vaddr len,
                          int flags, CPUWatchpoint **watchpoint)
{
    CPUWatchpoint *wp;

    /* forbid ranges which are empty or run off the end of the address space */
    if (len == 0 || (addr + len - 1) < addr) {
        error_report("tried to set invalid watchpoint at %"
                     VADDR_PRIx ", len=%" VADDR_PRIu, addr, len);
        return -EINVAL;
    }
    wp = g_malloc(sizeof(*wp));

    wp->vaddr = addr;
    wp->len = len;
    wp->flags = flags;

    /* keep all GDB-injected watchpoints in front */
    if (flags & BP_GDB) {
        QTAILQ_INSERT_HEAD(&cpu->watchpoints, wp, entry);
    } else {
        QTAILQ_INSERT_TAIL(&cpu->watchpoints, wp, entry);
    }

    tlb_flush_page(cpu, addr);

    if (watchpoint)
        *watchpoint = wp;
    return 0;
}

/* Remove a specific watchpoint.  */
int cpu_watchpoint_remove(CPUState *cpu, vaddr addr, vaddr len,
                          int flags)
{
    CPUWatchpoint *wp;

    QTAILQ_FOREACH(wp, &cpu->watchpoints, entry) {
        if (addr == wp->vaddr && len == wp->len
                && flags == (wp->flags & ~BP_WATCHPOINT_HIT)) {
            cpu_watchpoint_remove_by_ref(cpu, wp);
            return 0;
        }
    }
    return -ENOENT;
}

/* Remove a specific watchpoint by reference.  */
void cpu_watchpoint_remove_by_ref(CPUState *cpu, CPUWatchpoint *watchpoint)
{
    QTAILQ_REMOVE(&cpu->watchpoints, watchpoint, entry);

    tlb_flush_page(cpu, watchpoint->vaddr);

    g_free(watchpoint);
}

/* Remove all matching watchpoints.  */
void cpu_watchpoint_remove_all(CPUState *cpu, int mask)
{
    CPUWatchpoint *wp, *next;

    QTAILQ_FOREACH_SAFE(wp, &cpu->watchpoints, entry, next) {
        if (wp->flags & mask) {
            cpu_watchpoint_remove_by_ref(cpu, wp);
        }
    }
}

/* Return true if this watchpoint address matches the specified
 * access (ie the address range covered by the watchpoint overlaps
 * partially or completely with the address range covered by the
 * access).
 */
static inline bool cpu_watchpoint_address_matches(CPUWatchpoint *wp,
                                                  vaddr addr,
                                                  vaddr len)
{
    /* We know the lengths are non-zero, but a little caution is
     * required to avoid errors in the case where the range ends
     * exactly at the top of the address space and so addr + len
     * wraps round to zero.
     */
    vaddr wpend = wp->vaddr + wp->len - 1;
    vaddr addrend = addr + len - 1;

    return !(addr > wpend || wp->vaddr > addrend);
}

#endif

/* Add a breakpoint.  */
int cpu_breakpoint_insert(CPUState *cpu, vaddr pc, int flags,
                          CPUBreakpoint **breakpoint)
{
    CPUBreakpoint *bp;

    bp = g_malloc(sizeof(*bp));

    bp->pc = pc;
    bp->flags = flags;

    /* keep all GDB-injected breakpoints in front */
    if (flags & BP_GDB) {
        QTAILQ_INSERT_HEAD(&cpu->breakpoints, bp, entry);
    } else {
        QTAILQ_INSERT_TAIL(&cpu->breakpoints, bp, entry);
    }

    breakpoint_invalidate(cpu, pc);

    if (breakpoint) {
        *breakpoint = bp;
    }
    return 0;
}

/* Remove a specific breakpoint.  */
int cpu_breakpoint_remove(CPUState *cpu, vaddr pc, int flags)
{
    CPUBreakpoint *bp;

    QTAILQ_FOREACH(bp, &cpu->breakpoints, entry) {
        if (bp->pc == pc && bp->flags == flags) {
            cpu_breakpoint_remove_by_ref(cpu, bp);
            return 0;
        }
    }
    return -ENOENT;
}

/* Remove a specific breakpoint by reference.  */
void cpu_breakpoint_remove_by_ref(CPUState *cpu, CPUBreakpoint *breakpoint)
{
    QTAILQ_REMOVE(&cpu->breakpoints, breakpoint, entry);

    breakpoint_invalidate(cpu, breakpoint->pc);

    g_free(breakpoint);
}

/* Remove all matching breakpoints. */
void cpu_breakpoint_remove_all(CPUState *cpu, int mask)
{
    CPUBreakpoint *bp, *next;

    QTAILQ_FOREACH_SAFE(bp, &cpu->breakpoints, entry, next) {
        if (bp->flags & mask) {
            cpu_breakpoint_remove_by_ref(cpu, bp);
        }
    }
}

/* enable or disable single step mode. EXCP_DEBUG is returned by the
   CPU loop after each instruction */
void cpu_single_step(CPUState *cpu, int enabled)
{
    if (cpu->singlestep_enabled != enabled) {
        cpu->singlestep_enabled = enabled;
        if (kvm_enabled()) {
            kvm_update_guest_debug(cpu, 0);
        } else {
            /* must flush all the translated code to avoid inconsistencies */
            /* XXX: only flush what is necessary */
            tb_flush(cpu);
        }
    }
}

void cpu_abort(CPUState *cpu, const char *fmt, ...)
{
    va_list ap;
    va_list ap2;

    va_start(ap, fmt);
    va_copy(ap2, ap);
    fprintf(stderr, "qemu: fatal: ");
    vfprintf(stderr, fmt, ap);
    fprintf(stderr, "\n");
    cpu_dump_state(cpu, stderr, CPU_DUMP_FPU | CPU_DUMP_CCOP);
    if (qemu_log_separate()) {
        qemu_log_lock();
        qemu_log("qemu: fatal: ");
        qemu_log_vprintf(fmt, ap2);
        qemu_log("\n");
        log_cpu_state(cpu, CPU_DUMP_FPU | CPU_DUMP_CCOP);
        qemu_log_flush();
        qemu_log_unlock();
        qemu_log_close();
    }
    va_end(ap2);
    va_end(ap);
    replay_finish();
#if defined(CONFIG_USER_ONLY)
    {
        struct sigaction act;
        sigfillset(&act.sa_mask);
        act.sa_handler = SIG_DFL;
        act.sa_flags = 0;
        sigaction(SIGABRT, &act, NULL);
    }
#endif
    abort();
}

#if !defined(CONFIG_USER_ONLY)
/* Called from RCU critical section */
static RAMBlock *qemu_get_ram_block(ram_addr_t addr)
{
    RAMBlock *block;

    block = atomic_rcu_read(&ram_list.mru_block);
    if (block && addr - block->offset < block->max_length) {
        return block;
    }
    RAMBLOCK_FOREACH(block) {
        if (addr - block->offset < block->max_length) {
            goto found;
        }
    }

    fprintf(stderr, "Bad ram offset %" PRIx64 "\n", (uint64_t)addr);
    abort();

found:
    /* It is safe to write mru_block outside the iothread lock.  This
     * is what happens:
     *
     *     mru_block = xxx
     *     rcu_read_unlock()
     *                                        xxx removed from list
     *                  rcu_read_lock()
     *                  read mru_block
     *                                        mru_block = NULL;
     *                                        call_rcu(reclaim_ramblock, xxx);
     *                  rcu_read_unlock()
     *
     * atomic_rcu_set is not needed here.  The block was already published
     * when it was placed into the list.  Here we're just making an extra
     * copy of the pointer.
     */
    ram_list.mru_block = block;
    return block;
}

static void tlb_reset_dirty_range_all(ram_addr_t start, ram_addr_t length)
{
    CPUState *cpu;
    ram_addr_t start1;
    RAMBlock *block;
    ram_addr_t end;

    assert(tcg_enabled());
    end = TARGET_PAGE_ALIGN(start + length);
    start &= TARGET_PAGE_MASK;

    rcu_read_lock();
    block = qemu_get_ram_block(start);
    assert(block == qemu_get_ram_block(end - 1));
    start1 = (uintptr_t)ramblock_ptr(block, start - block->offset);
    CPU_FOREACH(cpu) {
        tlb_reset_dirty(cpu, start1, length);
    }
    rcu_read_unlock();
}

/* Note: start and end must be within the same ram block.  */
bool cpu_physical_memory_test_and_clear_dirty(ram_addr_t start,
                                              ram_addr_t length,
                                              unsigned client)
{
    DirtyMemoryBlocks *blocks;
    unsigned long end, page;
    bool dirty = false;
    RAMBlock *ramblock;
    uint64_t mr_offset, mr_size;

    if (length == 0) {
        return false;
    }

    end = TARGET_PAGE_ALIGN(start + length) >> TARGET_PAGE_BITS;
    page = start >> TARGET_PAGE_BITS;

    rcu_read_lock();

    blocks = atomic_rcu_read(&ram_list.dirty_memory[client]);
    ramblock = qemu_get_ram_block(start);
    /* Range sanity check on the ramblock */
    assert(start >= ramblock->offset &&
           start + length <= ramblock->offset + ramblock->used_length);

    while (page < end) {
        unsigned long idx = page / DIRTY_MEMORY_BLOCK_SIZE;
        unsigned long offset = page % DIRTY_MEMORY_BLOCK_SIZE;
        unsigned long num = MIN(end - page, DIRTY_MEMORY_BLOCK_SIZE - offset);

        dirty |= bitmap_test_and_clear_atomic(blocks->blocks[idx],
                                              offset, num);
        page += num;
    }

    mr_offset = (ram_addr_t)(page << TARGET_PAGE_BITS) - ramblock->offset;
    mr_size = (end - page) << TARGET_PAGE_BITS;
    memory_region_clear_dirty_bitmap(ramblock->mr, mr_offset, mr_size);

    rcu_read_unlock();

    if (dirty && tcg_enabled()) {
        tlb_reset_dirty_range_all(start, length);
    }

    return dirty;
}

DirtyBitmapSnapshot *cpu_physical_memory_snapshot_and_clear_dirty
    (MemoryRegion *mr, hwaddr offset, hwaddr length, unsigned client)
{
    DirtyMemoryBlocks *blocks;
    ram_addr_t start = memory_region_get_ram_addr(mr) + offset;
    unsigned long align = 1UL << (TARGET_PAGE_BITS + BITS_PER_LEVEL);
    ram_addr_t first = QEMU_ALIGN_DOWN(start, align);
    ram_addr_t last  = QEMU_ALIGN_UP(start + length, align);
    DirtyBitmapSnapshot *snap;
    unsigned long page, end, dest;

    snap = g_malloc0(sizeof(*snap) +
                     ((last - first) >> (TARGET_PAGE_BITS + 3)));
    snap->start = first;
    snap->end   = last;

    page = first >> TARGET_PAGE_BITS;
    end  = last  >> TARGET_PAGE_BITS;
    dest = 0;

    rcu_read_lock();

    blocks = atomic_rcu_read(&ram_list.dirty_memory[client]);

    while (page < end) {
        unsigned long idx = page / DIRTY_MEMORY_BLOCK_SIZE;
        unsigned long offset = page % DIRTY_MEMORY_BLOCK_SIZE;
        unsigned long num = MIN(end - page, DIRTY_MEMORY_BLOCK_SIZE - offset);

        assert(QEMU_IS_ALIGNED(offset, (1 << BITS_PER_LEVEL)));
        assert(QEMU_IS_ALIGNED(num,    (1 << BITS_PER_LEVEL)));
        offset >>= BITS_PER_LEVEL;

        bitmap_copy_and_clear_atomic(snap->dirty + dest,
                                     blocks->blocks[idx] + offset,
                                     num);
        page += num;
        dest += num >> BITS_PER_LEVEL;
    }

    rcu_read_unlock();

    if (tcg_enabled()) {
        tlb_reset_dirty_range_all(start, length);
    }

    memory_region_clear_dirty_bitmap(mr, offset, length);

    return snap;
}

bool cpu_physical_memory_snapshot_get_dirty(DirtyBitmapSnapshot *snap,
                                            ram_addr_t start,
                                            ram_addr_t length)
{
    unsigned long page, end;

    assert(start >= snap->start);
    assert(start + length <= snap->end);

    end = TARGET_PAGE_ALIGN(start + length - snap->start) >> TARGET_PAGE_BITS;
    page = (start - snap->start) >> TARGET_PAGE_BITS;

    while (page < end) {
        if (test_bit(page, snap->dirty)) {
            return true;
        }
        page++;
    }
    return false;
}

/* Called from RCU critical section */
hwaddr memory_region_section_get_iotlb(CPUState *cpu,
                                       MemoryRegionSection *section,
                                       target_ulong vaddr,
                                       hwaddr paddr, hwaddr xlat,
                                       int prot,
                                       target_ulong *address)
{
    hwaddr iotlb;
    CPUWatchpoint *wp;

    if (memory_region_is_ram(section->mr)) {
        /* Normal RAM.  */
        iotlb = memory_region_get_ram_addr(section->mr) + xlat;
        if (!section->readonly) {
            iotlb |= PHYS_SECTION_NOTDIRTY;
        } else {
            iotlb |= PHYS_SECTION_ROM;
        }
    } else {
        AddressSpaceDispatch *d;

        d = flatview_to_dispatch(section->fv);
        iotlb = section - d->map.sections;
        iotlb += xlat;
    }

    /* Make accesses to pages with watchpoints go via the
       watchpoint trap routines.  */
    QTAILQ_FOREACH(wp, &cpu->watchpoints, entry) {
        if (cpu_watchpoint_address_matches(wp, vaddr, TARGET_PAGE_SIZE)) {
            /* Avoid trapping reads of pages with a write breakpoint. */
            if ((prot & PAGE_WRITE) || (wp->flags & BP_MEM_READ)) {
                iotlb = PHYS_SECTION_WATCH + paddr;
                *address |= TLB_MMIO;
                break;
            }
        }
    }

    return iotlb;
}
#endif /* defined(CONFIG_USER_ONLY) */

#if !defined(CONFIG_USER_ONLY)

static int subpage_register (subpage_t *mmio, uint32_t start, uint32_t end,
                             uint16_t section);
static subpage_t *subpage_init(FlatView *fv, hwaddr base);

static void *(*phys_mem_alloc)(size_t size, uint64_t *align, bool shared) =
                               qemu_anon_ram_alloc;

/*
 * Set a custom physical guest memory alloator.
 * Accelerators with unusual needs may need this.  Hopefully, we can
 * get rid of it eventually.
 */
void phys_mem_set_alloc(void *(*alloc)(size_t, uint64_t *align, bool shared))
{
    phys_mem_alloc = alloc;
}

static uint16_t phys_section_add(PhysPageMap *map,
                                 MemoryRegionSection *section)
{
    /* The physical section number is ORed with a page-aligned
     * pointer to produce the iotlb entries.  Thus it should
     * never overflow into the page-aligned value.
     */
    assert(map->sections_nb < TARGET_PAGE_SIZE);

    if (map->sections_nb == map->sections_nb_alloc) {
        map->sections_nb_alloc = MAX(map->sections_nb_alloc * 2, 16);
        map->sections = g_renew(MemoryRegionSection, map->sections,
                                map->sections_nb_alloc);
    }
    map->sections[map->sections_nb] = *section;
    memory_region_ref(section->mr);
    return map->sections_nb++;
}

static void phys_section_destroy(MemoryRegion *mr)
{
    bool have_sub_page = mr->subpage;

    memory_region_unref(mr);

    if (have_sub_page) {
        subpage_t *subpage = container_of(mr, subpage_t, iomem);
        object_unref(OBJECT(&subpage->iomem));
        g_free(subpage);
    }
}

static void phys_sections_free(PhysPageMap *map)
{
    while (map->sections_nb > 0) {
        MemoryRegionSection *section = &map->sections[--map->sections_nb];
        phys_section_destroy(section->mr);
    }
    g_free(map->sections);
    g_free(map->nodes);
}

static void register_subpage(FlatView *fv, MemoryRegionSection *section)
{
    AddressSpaceDispatch *d = flatview_to_dispatch(fv);
    subpage_t *subpage;
    hwaddr base = section->offset_within_address_space
        & TARGET_PAGE_MASK;
    MemoryRegionSection *existing = phys_page_find(d, base);
    MemoryRegionSection subsection = {
        .offset_within_address_space = base,
        .size = int128_make64(TARGET_PAGE_SIZE),
    };
    hwaddr start, end;

    assert(existing->mr->subpage || existing->mr == &io_mem_unassigned);

    if (!(existing->mr->subpage)) {
        subpage = subpage_init(fv, base);
        subsection.fv = fv;
        subsection.mr = &subpage->iomem;
        phys_page_set(d, base >> TARGET_PAGE_BITS, 1,
                      phys_section_add(&d->map, &subsection));
    } else {
        subpage = container_of(existing->mr, subpage_t, iomem);
    }
    start = section->offset_within_address_space & ~TARGET_PAGE_MASK;
    end = start + int128_get64(section->size) - 1;
    subpage_register(subpage, start, end,
                     phys_section_add(&d->map, section));
}


static void register_multipage(FlatView *fv,
                               MemoryRegionSection *section)
{
    AddressSpaceDispatch *d = flatview_to_dispatch(fv);
    hwaddr start_addr = section->offset_within_address_space;
    uint16_t section_index = phys_section_add(&d->map, section);
    uint64_t num_pages = int128_get64(int128_rshift(section->size,
                                                    TARGET_PAGE_BITS));

    assert(num_pages);
    phys_page_set(d, start_addr >> TARGET_PAGE_BITS, num_pages, section_index);
}

/*
 * The range in *section* may look like this:
 *
 *      |s|PPPPPPP|s|
 *
 * where s stands for subpage and P for page.
 */
void flatview_add_to_dispatch(FlatView *fv, MemoryRegionSection *section)
{
    MemoryRegionSection remain = *section;
    Int128 page_size = int128_make64(TARGET_PAGE_SIZE);

    /* register first subpage */
    if (remain.offset_within_address_space & ~TARGET_PAGE_MASK) {
        uint64_t left = TARGET_PAGE_ALIGN(remain.offset_within_address_space)
                        - remain.offset_within_address_space;

        MemoryRegionSection now = remain;
        now.size = int128_min(int128_make64(left), now.size);
        register_subpage(fv, &now);
        if (int128_eq(remain.size, now.size)) {
            return;
        }
        remain.size = int128_sub(remain.size, now.size);
        remain.offset_within_address_space += int128_get64(now.size);
        remain.offset_within_region += int128_get64(now.size);
    }

    /* register whole pages */
    if (int128_ge(remain.size, page_size)) {
        MemoryRegionSection now = remain;
        now.size = int128_and(now.size, int128_neg(page_size));
        register_multipage(fv, &now);
        if (int128_eq(remain.size, now.size)) {
            return;
        }
        remain.size = int128_sub(remain.size, now.size);
        remain.offset_within_address_space += int128_get64(now.size);
        remain.offset_within_region += int128_get64(now.size);
    }

    /* register last subpage */
    register_subpage(fv, &remain);
}

void qemu_flush_coalesced_mmio_buffer(void)
{
    if (kvm_enabled())
        kvm_flush_coalesced_mmio_buffer();
}

void qemu_mutex_lock_ramlist(void)
{
    qemu_mutex_lock(&ram_list.mutex);
}

void qemu_mutex_unlock_ramlist(void)
{
    qemu_mutex_unlock(&ram_list.mutex);
}

void ram_block_dump(Monitor *mon)
{
    RAMBlock *block;
    char *psize;

    rcu_read_lock();
    monitor_printf(mon, "%24s %8s  %18s %18s %18s\n",
                   "Block Name", "PSize", "Offset", "Used", "Total");
    RAMBLOCK_FOREACH(block) {
        psize = size_to_str(block->page_size);
        monitor_printf(mon, "%24s %8s  0x%016" PRIx64 " 0x%016" PRIx64
                       " 0x%016" PRIx64 "\n", block->idstr, psize,
                       (uint64_t)block->offset,
                       (uint64_t)block->used_length,
                       (uint64_t)block->max_length);
        g_free(psize);
    }
    rcu_read_unlock();
}

#ifdef __linux__
/*
 * FIXME TOCTTOU: this iterates over memory backends' mem-path, which
 * may or may not name the same files / on the same filesystem now as
 * when we actually open and map them.  Iterate over the file
 * descriptors instead, and use qemu_fd_getpagesize().
 */
static int find_min_backend_pagesize(Object *obj, void *opaque)
{
    long *hpsize_min = opaque;

    if (object_dynamic_cast(obj, TYPE_MEMORY_BACKEND)) {
        HostMemoryBackend *backend = MEMORY_BACKEND(obj);
        long hpsize = host_memory_backend_pagesize(backend);

        if (host_memory_backend_is_mapped(backend) && (hpsize < *hpsize_min)) {
            *hpsize_min = hpsize;
        }
    }

    return 0;
}

static int find_max_backend_pagesize(Object *obj, void *opaque)
{
    long *hpsize_max = opaque;

    if (object_dynamic_cast(obj, TYPE_MEMORY_BACKEND)) {
        HostMemoryBackend *backend = MEMORY_BACKEND(obj);
        long hpsize = host_memory_backend_pagesize(backend);

        if (host_memory_backend_is_mapped(backend) && (hpsize > *hpsize_max)) {
            *hpsize_max = hpsize;
        }
    }

    return 0;
}

/*
 * TODO: We assume right now that all mapped host memory backends are
 * used as RAM, however some might be used for different purposes.
 */
long qemu_minrampagesize(void)
{
    long hpsize = LONG_MAX;
    long mainrampagesize;
    Object *memdev_root;

    mainrampagesize = qemu_mempath_getpagesize(mem_path);

    /* it's possible we have memory-backend objects with
     * hugepage-backed RAM. these may get mapped into system
     * address space via -numa parameters or memory hotplug
     * hooks. we want to take these into account, but we
     * also want to make sure these supported hugepage
     * sizes are applicable across the entire range of memory
     * we may boot from, so we take the min across all
     * backends, and assume normal pages in cases where a
     * backend isn't backed by hugepages.
     */
    memdev_root = object_resolve_path("/objects", NULL);
    if (memdev_root) {
        object_child_foreach(memdev_root, find_min_backend_pagesize, &hpsize);
    }
    if (hpsize == LONG_MAX) {
        /* No additional memory regions found ==> Report main RAM page size */
        return mainrampagesize;
    }

    /* If NUMA is disabled or the NUMA nodes are not backed with a
     * memory-backend, then there is at least one node using "normal" RAM,
     * so if its page size is smaller we have got to report that size instead.
     */
    if (hpsize > mainrampagesize &&
        (nb_numa_nodes == 0 || numa_info[0].node_memdev == NULL)) {
        static bool warned;
        if (!warned) {
            error_report("Huge page support disabled (n/a for main memory).");
            warned = true;
        }
        return mainrampagesize;
    }

    return hpsize;
}

long qemu_maxrampagesize(void)
{
    long pagesize = qemu_mempath_getpagesize(mem_path);
    Object *memdev_root = object_resolve_path("/objects", NULL);

    if (memdev_root) {
        object_child_foreach(memdev_root, find_max_backend_pagesize,
                             &pagesize);
    }
    return pagesize;
}
#else
long qemu_minrampagesize(void)
{
    return getpagesize();
}
long qemu_maxrampagesize(void)
{
    return getpagesize();
}
#endif

#ifdef CONFIG_POSIX
static int64_t get_file_size(int fd)
{
    int64_t size = lseek(fd, 0, SEEK_END);
    if (size < 0) {
        return -errno;
    }
    return size;
}

static int file_ram_open(const char *path,
                         const char *region_name,
                         bool *created,
                         Error **errp)
{
    char *filename;
    char *sanitized_name;
    char *c;
    int fd = -1;

    *created = false;
    for (;;) {
        fd = open(path, O_RDWR);
        if (fd >= 0) {
            /* @path names an existing file, use it */
            break;
        }
        if (errno == ENOENT) {
            /* @path names a file that doesn't exist, create it */
            fd = open(path, O_RDWR | O_CREAT | O_EXCL, 0644);
            if (fd >= 0) {
                *created = true;
                break;
            }
        } else if (errno == EISDIR) {
            /* @path names a directory, create a file there */
            /* Make name safe to use with mkstemp by replacing '/' with '_'. */
            sanitized_name = g_strdup(region_name);
            for (c = sanitized_name; *c != '\0'; c++) {
                if (*c == '/') {
                    *c = '_';
                }
            }

            filename = g_strdup_printf("%s/qemu_back_mem.%s.XXXXXX", path,
                                       sanitized_name);
            g_free(sanitized_name);

            fd = mkstemp(filename);
            if (fd >= 0) {
                unlink(filename);
                g_free(filename);
                break;
            }
            g_free(filename);
        }
        if (errno != EEXIST && errno != EINTR) {
            error_setg_errno(errp, errno,
                             "can't open backing store %s for guest RAM",
                             path);
            return -1;
        }
        /*
         * Try again on EINTR and EEXIST.  The latter happens when
         * something else creates the file between our two open().
         */
    }

    return fd;
}

static void *file_ram_alloc(RAMBlock *block,
                            ram_addr_t memory,
                            int fd,
                            bool truncate,
                            Error **errp)
{
    MachineState *ms = MACHINE(qdev_get_machine());
    void *area;

    block->page_size = qemu_fd_getpagesize(fd);
    if (block->mr->align % block->page_size) {
        error_setg(errp, "alignment 0x%" PRIx64
                   " must be multiples of page size 0x%zx",
                   block->mr->align, block->page_size);
        return NULL;
    } else if (block->mr->align && !is_power_of_2(block->mr->align)) {
        error_setg(errp, "alignment 0x%" PRIx64
                   " must be a power of two", block->mr->align);
        return NULL;
    }
    block->mr->align = MAX(block->page_size, block->mr->align);
#if defined(__s390x__)
    if (kvm_enabled()) {
        block->mr->align = MAX(block->mr->align, QEMU_VMALLOC_ALIGN);
    }
#endif

    if (memory < block->page_size) {
        error_setg(errp, "memory size 0x" RAM_ADDR_FMT " must be equal to "
                   "or larger than page size 0x%zx",
                   memory, block->page_size);
        return NULL;
    }

    memory = ROUND_UP(memory, block->page_size);

    /*
     * ftruncate is not supported by hugetlbfs in older
     * hosts, so don't bother bailing out on errors.
     * If anything goes wrong with it under other filesystems,
     * mmap will fail.
     *
     * Do not truncate the non-empty backend file to avoid corrupting
     * the existing data in the file. Disabling shrinking is not
     * enough. For example, the current vNVDIMM implementation stores
     * the guest NVDIMM labels at the end of the backend file. If the
     * backend file is later extended, QEMU will not be able to find
     * those labels. Therefore, extending the non-empty backend file
     * is disabled as well.
     */
    if (truncate && ftruncate(fd, memory)) {
        perror("ftruncate");
    }

    area = qemu_ram_mmap(fd, memory, block->mr->align,
                         block->flags & RAM_SHARED, block->flags & RAM_PMEM);
    if (area == MAP_FAILED) {
        error_setg_errno(errp, errno,
                         "unable to map backing store for guest RAM");
        return NULL;
    }

    if (mem_prealloc) {
        os_mem_prealloc(fd, area, memory, ms->smp.cpus, errp);
        if (errp && *errp) {
            qemu_ram_munmap(fd, area, memory);
            return NULL;
        }
    }

    block->fd = fd;
    return area;
}
#endif

/* Allocate space within the ram_addr_t space that governs the
 * dirty bitmaps.
 * Called with the ramlist lock held.
 */
static ram_addr_t find_ram_offset(ram_addr_t size)
{
    RAMBlock *block, *next_block;
    ram_addr_t offset = RAM_ADDR_MAX, mingap = RAM_ADDR_MAX;

    assert(size != 0); /* it would hand out same offset multiple times */

    if (QLIST_EMPTY_RCU(&ram_list.blocks)) {
        return 0;
    }

    RAMBLOCK_FOREACH(block) {
        ram_addr_t candidate, next = RAM_ADDR_MAX;

        /* Align blocks to start on a 'long' in the bitmap
         * which makes the bitmap sync'ing take the fast path.
         */
        candidate = block->offset + block->max_length;
        candidate = ROUND_UP(candidate, BITS_PER_LONG << TARGET_PAGE_BITS);

        /* Search for the closest following block
         * and find the gap.
         */
        RAMBLOCK_FOREACH(next_block) {
            if (next_block->offset >= candidate) {
                next = MIN(next, next_block->offset);
            }
        }

        /* If it fits remember our place and remember the size
         * of gap, but keep going so that we might find a smaller
         * gap to fill so avoiding fragmentation.
         */
        if (next - candidate >= size && next - candidate < mingap) {
            offset = candidate;
            mingap = next - candidate;
        }

        trace_find_ram_offset_loop(size, candidate, offset, next, mingap);
    }

    if (offset == RAM_ADDR_MAX) {
        fprintf(stderr, "Failed to find gap of requested size: %" PRIu64 "\n",
                (uint64_t)size);
        abort();
    }

    trace_find_ram_offset(size, offset);

    return offset;
}

static unsigned long last_ram_page(void)
{
    RAMBlock *block;
    ram_addr_t last = 0;

    rcu_read_lock();
    RAMBLOCK_FOREACH(block) {
        last = MAX(last, block->offset + block->max_length);
    }
    rcu_read_unlock();
    return last >> TARGET_PAGE_BITS;
}

static void qemu_ram_setup_dump(void *addr, ram_addr_t size)
{
    int ret;

    /* Use MADV_DONTDUMP, if user doesn't want the guest memory in the core */
    if (!machine_dump_guest_core(current_machine)) {
        ret = qemu_madvise(addr, size, QEMU_MADV_DONTDUMP);
        if (ret) {
            perror("qemu_madvise");
            fprintf(stderr, "madvise doesn't support MADV_DONTDUMP, "
                            "but dump_guest_core=off specified\n");
        }
    }
}

const char *qemu_ram_get_idstr(RAMBlock *rb)
{
    return rb->idstr;
}

void *qemu_ram_get_host_addr(RAMBlock *rb)
{
    return rb->host;
}

ram_addr_t qemu_ram_get_offset(RAMBlock *rb)
{
    return rb->offset;
}

ram_addr_t qemu_ram_get_used_length(RAMBlock *rb)
{
    return rb->used_length;
}

bool qemu_ram_is_shared(RAMBlock *rb)
{
    return rb->flags & RAM_SHARED;
}

/* Note: Only set at the start of postcopy */
bool qemu_ram_is_uf_zeroable(RAMBlock *rb)
{
    return rb->flags & RAM_UF_ZEROPAGE;
}

void qemu_ram_set_uf_zeroable(RAMBlock *rb)
{
    rb->flags |= RAM_UF_ZEROPAGE;
}

bool qemu_ram_is_migratable(RAMBlock *rb)
{
    return rb->flags & RAM_MIGRATABLE;
}

void qemu_ram_set_migratable(RAMBlock *rb)
{
    rb->flags |= RAM_MIGRATABLE;
}

void qemu_ram_unset_migratable(RAMBlock *rb)
{
    rb->flags &= ~RAM_MIGRATABLE;
}

/* Called with iothread lock held.  */
void qemu_ram_set_idstr(RAMBlock *new_block, const char *name, DeviceState *dev)
{
    RAMBlock *block;

    assert(new_block);
    assert(!new_block->idstr[0]);

    if (dev) {
        char *id = qdev_get_dev_path(dev);
        if (id) {
            snprintf(new_block->idstr, sizeof(new_block->idstr), "%s/", id);
            g_free(id);
        }
    }
    pstrcat(new_block->idstr, sizeof(new_block->idstr), name);

    rcu_read_lock();
    RAMBLOCK_FOREACH(block) {
        if (block != new_block &&
            !strcmp(block->idstr, new_block->idstr)) {
            fprintf(stderr, "RAMBlock \"%s\" already registered, abort!\n",
                    new_block->idstr);
            abort();
        }
    }
    rcu_read_unlock();
}

/* Called with iothread lock held.  */
void qemu_ram_unset_idstr(RAMBlock *block)
{
    /* FIXME: arch_init.c assumes that this is not called throughout
     * migration.  Ignore the problem since hot-unplug during migration
     * does not work anyway.
     */
    if (block) {
        memset(block->idstr, 0, sizeof(block->idstr));
    }
}

size_t qemu_ram_pagesize(RAMBlock *rb)
{
    return rb->page_size;
}

/* Returns the largest size of page in use */
size_t qemu_ram_pagesize_largest(void)
{
    RAMBlock *block;
    size_t largest = 0;

    RAMBLOCK_FOREACH(block) {
        largest = MAX(largest, qemu_ram_pagesize(block));
    }

    return largest;
}

static int memory_try_enable_merging(void *addr, size_t len)
{
    if (!machine_mem_merge(current_machine)) {
        /* disabled by the user */
        return 0;
    }

    return qemu_madvise(addr, len, QEMU_MADV_MERGEABLE);
}

/* Only legal before guest might have detected the memory size: e.g. on
 * incoming migration, or right after reset.
 *
 * As memory core doesn't know how is memory accessed, it is up to
 * resize callback to update device state and/or add assertions to detect
 * misuse, if necessary.
 */
int qemu_ram_resize(RAMBlock *block, ram_addr_t newsize, Error **errp)
{
    assert(block);

    newsize = HOST_PAGE_ALIGN(newsize);

    if (block->used_length == newsize) {
        return 0;
    }

    if (!(block->flags & RAM_RESIZEABLE)) {
        error_setg_errno(errp, EINVAL,
                         "Length mismatch: %s: 0x" RAM_ADDR_FMT
                         " in != 0x" RAM_ADDR_FMT, block->idstr,
                         newsize, block->used_length);
        return -EINVAL;
    }

    if (block->max_length < newsize) {
        error_setg_errno(errp, EINVAL,
                         "Length too large: %s: 0x" RAM_ADDR_FMT
                         " > 0x" RAM_ADDR_FMT, block->idstr,
                         newsize, block->max_length);
        return -EINVAL;
    }

    cpu_physical_memory_clear_dirty_range(block->offset, block->used_length);
    block->used_length = newsize;
    cpu_physical_memory_set_dirty_range(block->offset, block->used_length,
                                        DIRTY_CLIENTS_ALL);
    memory_region_set_size(block->mr, newsize);
    if (block->resized) {
        block->resized(block->idstr, newsize, block->host);
    }
    return 0;
}

/* Called with ram_list.mutex held */
static void dirty_memory_extend(ram_addr_t old_ram_size,
                                ram_addr_t new_ram_size)
{
    ram_addr_t old_num_blocks = DIV_ROUND_UP(old_ram_size,
                                             DIRTY_MEMORY_BLOCK_SIZE);
    ram_addr_t new_num_blocks = DIV_ROUND_UP(new_ram_size,
                                             DIRTY_MEMORY_BLOCK_SIZE);
    int i;

    /* Only need to extend if block count increased */
    if (new_num_blocks <= old_num_blocks) {
        return;
    }

    for (i = 0; i < DIRTY_MEMORY_NUM; i++) {
        DirtyMemoryBlocks *old_blocks;
        DirtyMemoryBlocks *new_blocks;
        int j;

        old_blocks = atomic_rcu_read(&ram_list.dirty_memory[i]);
        new_blocks = g_malloc(sizeof(*new_blocks) +
                              sizeof(new_blocks->blocks[0]) * new_num_blocks);

        if (old_num_blocks) {
            memcpy(new_blocks->blocks, old_blocks->blocks,
                   old_num_blocks * sizeof(old_blocks->blocks[0]));
        }

        for (j = old_num_blocks; j < new_num_blocks; j++) {
            new_blocks->blocks[j] = bitmap_new(DIRTY_MEMORY_BLOCK_SIZE);
        }

        atomic_rcu_set(&ram_list.dirty_memory[i], new_blocks);

        if (old_blocks) {
            g_free_rcu(old_blocks, rcu);
        }
    }
}

static void ram_block_add(RAMBlock *new_block, Error **errp, bool shared)
{
    RAMBlock *block;
    RAMBlock *last_block = NULL;
    ram_addr_t old_ram_size, new_ram_size;
    Error *err = NULL;

    old_ram_size = last_ram_page();

    qemu_mutex_lock_ramlist();
    new_block->offset = find_ram_offset(new_block->max_length);

    if (!new_block->host) {
        if (xen_enabled()) {
            xen_ram_alloc(new_block->offset, new_block->max_length,
                          new_block->mr, &err);
            if (err) {
                error_propagate(errp, err);
                qemu_mutex_unlock_ramlist();
                return;
            }
        } else {
            new_block->host = phys_mem_alloc(new_block->max_length,
                                             &new_block->mr->align, shared);
            if (!new_block->host) {
                error_setg_errno(errp, errno,
                                 "cannot set up guest memory '%s'",
                                 memory_region_name(new_block->mr));
                qemu_mutex_unlock_ramlist();
                return;
            }
            memory_try_enable_merging(new_block->host, new_block->max_length);
        }
    }

    new_ram_size = MAX(old_ram_size,
              (new_block->offset + new_block->max_length) >> TARGET_PAGE_BITS);
    if (new_ram_size > old_ram_size) {
        dirty_memory_extend(old_ram_size, new_ram_size);
    }
    /* Keep the list sorted from biggest to smallest block.  Unlike QTAILQ,
     * QLIST (which has an RCU-friendly variant) does not have insertion at
     * tail, so save the last element in last_block.
     */
    RAMBLOCK_FOREACH(block) {
        last_block = block;
        if (block->max_length < new_block->max_length) {
            break;
        }
    }
    if (block) {
        QLIST_INSERT_BEFORE_RCU(block, new_block, next);
    } else if (last_block) {
        QLIST_INSERT_AFTER_RCU(last_block, new_block, next);
    } else { /* list is empty */
        QLIST_INSERT_HEAD_RCU(&ram_list.blocks, new_block, next);
    }
    ram_list.mru_block = NULL;

    /* Write list before version */
    smp_wmb();
    ram_list.version++;
    qemu_mutex_unlock_ramlist();

    cpu_physical_memory_set_dirty_range(new_block->offset,
                                        new_block->used_length,
                                        DIRTY_CLIENTS_ALL);

    if (new_block->host) {
        qemu_ram_setup_dump(new_block->host, new_block->max_length);
        qemu_madvise(new_block->host, new_block->max_length, QEMU_MADV_HUGEPAGE);
        /* MADV_DONTFORK is also needed by KVM in absence of synchronous MMU */
        qemu_madvise(new_block->host, new_block->max_length, QEMU_MADV_DONTFORK);
        ram_block_notify_add(new_block->host, new_block->max_length);
    }
}

#ifdef CONFIG_POSIX
RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
                                 uint32_t ram_flags, int fd,
                                 Error **errp)
{
    RAMBlock *new_block;
    Error *local_err = NULL;
    int64_t file_size;

    /* Just support these ram flags by now. */
    assert((ram_flags & ~(RAM_SHARED | RAM_PMEM)) == 0);

    if (xen_enabled()) {
        error_setg(errp, "-mem-path not supported with Xen");
        return NULL;
    }

    if (kvm_enabled() && !kvm_has_sync_mmu()) {
        error_setg(errp,
                   "host lacks kvm mmu notifiers, -mem-path unsupported");
        return NULL;
    }

    if (phys_mem_alloc != qemu_anon_ram_alloc) {
        /*
         * file_ram_alloc() needs to allocate just like
         * phys_mem_alloc, but we haven't bothered to provide
         * a hook there.
         */
        error_setg(errp,
                   "-mem-path not supported with this accelerator");
        return NULL;
    }

    size = HOST_PAGE_ALIGN(size);
    file_size = get_file_size(fd);
    if (file_size > 0 && file_size < size) {
        error_setg(errp, "backing store %s size 0x%" PRIx64
                   " does not match 'size' option 0x" RAM_ADDR_FMT,
                   mem_path, file_size, size);
        return NULL;
    }

    new_block = g_malloc0(sizeof(*new_block));
    new_block->mr = mr;
    new_block->used_length = size;
    new_block->max_length = size;
    new_block->flags = ram_flags;
    new_block->host = file_ram_alloc(new_block, size, fd, !file_size, errp);
    if (!new_block->host) {
        g_free(new_block);
        return NULL;
    }

    ram_block_add(new_block, &local_err, ram_flags & RAM_SHARED);
    if (local_err) {
        g_free(new_block);
        error_propagate(errp, local_err);
        return NULL;
    }
    return new_block;

}


RAMBlock *qemu_ram_alloc_from_file(ram_addr_t size, MemoryRegion *mr,
                                   uint32_t ram_flags, const char *mem_path,
                                   Error **errp)
{
    int fd;
    bool created;
    RAMBlock *block;

    fd = file_ram_open(mem_path, memory_region_name(mr), &created, errp);
    if (fd < 0) {
        return NULL;
    }

    block = qemu_ram_alloc_from_fd(size, mr, ram_flags, fd, errp);
    if (!block) {
        if (created) {
            unlink(mem_path);
        }
        close(fd);
        return NULL;
    }

    return block;
}
#endif

static
RAMBlock *qemu_ram_alloc_internal(ram_addr_t size, ram_addr_t max_size,
                                  void (*resized)(const char*,
                                                  uint64_t length,
                                                  void *host),
                                  void *host, bool resizeable, bool share,
                                  MemoryRegion *mr, Error **errp)
{
    RAMBlock *new_block;
    Error *local_err = NULL;

    size = HOST_PAGE_ALIGN(size);
    max_size = HOST_PAGE_ALIGN(max_size);
    new_block = g_malloc0(sizeof(*new_block));
    new_block->mr = mr;
    new_block->resized = resized;
    new_block->used_length = size;
    new_block->max_length = max_size;
    assert(max_size >= size);
    new_block->fd = -1;
    new_block->page_size = getpagesize();
    new_block->host = host;
    if (host) {
        new_block->flags |= RAM_PREALLOC;
    }
    if (resizeable) {
        new_block->flags |= RAM_RESIZEABLE;
    }
    ram_block_add(new_block, &local_err, share);
    if (local_err) {
        g_free(new_block);
        error_propagate(errp, local_err);
        return NULL;
    }
    return new_block;
}

RAMBlock *qemu_ram_alloc_from_ptr(ram_addr_t size, void *host,
                                   MemoryRegion *mr, Error **errp)
{
    return qemu_ram_alloc_internal(size, size, NULL, host, false,
                                   false, mr, errp);
}

RAMBlock *qemu_ram_alloc(ram_addr_t size, bool share,
                         MemoryRegion *mr, Error **errp)
{
    return qemu_ram_alloc_internal(size, size, NULL, NULL, false,
                                   share, mr, errp);
}

RAMBlock *qemu_ram_alloc_resizeable(ram_addr_t size, ram_addr_t maxsz,
                                     void (*resized)(const char*,
                                                     uint64_t length,
                                                     void *host),
                                     MemoryRegion *mr, Error **errp)
{
    return qemu_ram_alloc_internal(size, maxsz, resized, NULL, true,
                                   false, mr, errp);
}

static void reclaim_ramblock(RAMBlock *block)
{
    if (block->flags & RAM_PREALLOC) {
        ;
    } else if (xen_enabled()) {
        xen_invalidate_map_cache_entry(block->host);
#ifndef _WIN32
    } else if (block->fd >= 0) {
        qemu_ram_munmap(block->fd, block->host, block->max_length);
        close(block->fd);
#endif
    } else {
        qemu_anon_ram_free(block->host, block->max_length);
    }
    g_free(block);
}

void qemu_ram_free(RAMBlock *block)
{
    if (!block) {
        return;
    }

    if (block->host) {
        ram_block_notify_remove(block->host, block->max_length);
    }

    qemu_mutex_lock_ramlist();
    QLIST_REMOVE_RCU(block, next);
    ram_list.mru_block = NULL;
    /* Write list before version */
    smp_wmb();
    ram_list.version++;
    call_rcu(block, reclaim_ramblock, rcu);
    qemu_mutex_unlock_ramlist();
}

#ifndef _WIN32
void qemu_ram_remap(ram_addr_t addr, ram_addr_t length)
{
    RAMBlock *block;
    ram_addr_t offset;
    int flags;
    void *area, *vaddr;

    RAMBLOCK_FOREACH(block) {
        offset = addr - block->offset;
        if (offset < block->max_length) {
            vaddr = ramblock_ptr(block, offset);
            if (block->flags & RAM_PREALLOC) {
                ;
            } else if (xen_enabled()) {
                abort();
            } else {
                flags = MAP_FIXED;
                if (block->fd >= 0) {
                    flags |= (block->flags & RAM_SHARED ?
                              MAP_SHARED : MAP_PRIVATE);
                    area = mmap(vaddr, length, PROT_READ | PROT_WRITE,
                                flags, block->fd, offset);
                } else {
                    /*
                     * Remap needs to match alloc.  Accelerators that
                     * set phys_mem_alloc never remap.  If they did,
                     * we'd need a remap hook here.
                     */
                    assert(phys_mem_alloc == qemu_anon_ram_alloc);

                    flags |= MAP_PRIVATE | MAP_ANONYMOUS;
                    area = mmap(vaddr, length, PROT_READ | PROT_WRITE,
                                flags, -1, 0);
                }
                if (area != vaddr) {
                    error_report("Could not remap addr: "
                                 RAM_ADDR_FMT "@" RAM_ADDR_FMT "",
                                 length, addr);
                    exit(1);
                }
                memory_try_enable_merging(vaddr, length);
                qemu_ram_setup_dump(vaddr, length);
            }
        }
    }
}
#endif /* !_WIN32 */

/* Return a host pointer to ram allocated with qemu_ram_alloc.
 * This should not be used for general purpose DMA.  Use address_space_map
 * or address_space_rw instead. For local memory (e.g. video ram) that the
 * device owns, use memory_region_get_ram_ptr.
 *
 * Called within RCU critical section.
 */
void *qemu_map_ram_ptr(RAMBlock *ram_block, ram_addr_t addr)
{
    RAMBlock *block = ram_block;

    if (block == NULL) {
        block = qemu_get_ram_block(addr);
        addr -= block->offset;
    }

    if (xen_enabled() && block->host == NULL) {
        /* We need to check if the requested address is in the RAM
         * because we don't want to map the entire memory in QEMU.
         * In that case just map until the end of the page.
         */
        if (block->offset == 0) {
            return xen_map_cache(addr, 0, 0, false);
        }

        block->host = xen_map_cache(block->offset, block->max_length, 1, false);
    }
    return ramblock_ptr(block, addr);
}

/* Return a host pointer to guest's ram. Similar to qemu_map_ram_ptr
 * but takes a size argument.
 *
 * Called within RCU critical section.
 */
static void *qemu_ram_ptr_length(RAMBlock *ram_block, ram_addr_t addr,
                                 hwaddr *size, bool lock)
{
    RAMBlock *block = ram_block;
    if (*size == 0) {
        return NULL;
    }

    if (block == NULL) {
        block = qemu_get_ram_block(addr);
        addr -= block->offset;
    }
    *size = MIN(*size, block->max_length - addr);

    if (xen_enabled() && block->host == NULL) {
        /* We need to check if the requested address is in the RAM
         * because we don't want to map the entire memory in QEMU.
         * In that case just map the requested area.
         */
        if (block->offset == 0) {
            return xen_map_cache(addr, *size, lock, lock);
        }

        block->host = xen_map_cache(block->offset, block->max_length, 1, lock);
    }

    return ramblock_ptr(block, addr);
}

/* Return the offset of a hostpointer within a ramblock */
ram_addr_t qemu_ram_block_host_offset(RAMBlock *rb, void *host)
{
    ram_addr_t res = (uint8_t *)host - (uint8_t *)rb->host;
    assert((uintptr_t)host >= (uintptr_t)rb->host);
    assert(res < rb->max_length);

    return res;
}

/*
 * Translates a host ptr back to a RAMBlock, a ram_addr and an offset
 * in that RAMBlock.
 *
 * ptr: Host pointer to look up
 * round_offset: If true round the result offset down to a page boundary
 * *ram_addr: set to result ram_addr
 * *offset: set to result offset within the RAMBlock
 *
 * Returns: RAMBlock (or NULL if not found)
 *
 * By the time this function returns, the returned pointer is not protected
 * by RCU anymore.  If the caller is not within an RCU critical section and
 * does not hold the iothread lock, it must have other means of protecting the
 * pointer, such as a reference to the region that includes the incoming
 * ram_addr_t.
 */
RAMBlock *qemu_ram_block_from_host(void *ptr, bool round_offset,
                                   ram_addr_t *offset)
{
    RAMBlock *block;
    uint8_t *host = ptr;

    if (xen_enabled()) {
        ram_addr_t ram_addr;
        rcu_read_lock();
        ram_addr = xen_ram_addr_from_mapcache(ptr);
        block = qemu_get_ram_block(ram_addr);
        if (block) {
            *offset = ram_addr - block->offset;
        }
        rcu_read_unlock();
        return block;
    }

    rcu_read_lock();
    block = atomic_rcu_read(&ram_list.mru_block);
    if (block && block->host && host - block->host < block->max_length) {
        goto found;
    }

    RAMBLOCK_FOREACH(block) {
        /* This case append when the block is not mapped. */
        if (block->host == NULL) {
            continue;
        }
        if (host - block->host < block->max_length) {
            goto found;
        }
    }

    rcu_read_unlock();
    return NULL;

found:
    *offset = (host - block->host);
    if (round_offset) {
        *offset &= TARGET_PAGE_MASK;
    }
    rcu_read_unlock();
    return block;
}

/*
 * Finds the named RAMBlock
 *
 * name: The name of RAMBlock to find
 *
 * Returns: RAMBlock (or NULL if not found)
 */
RAMBlock *qemu_ram_block_by_name(const char *name)
{
    RAMBlock *block;

    RAMBLOCK_FOREACH(block) {
        if (!strcmp(name, block->idstr)) {
            return block;
        }
    }

    return NULL;
}

/* Some of the softmmu routines need to translate from a host pointer
   (typically a TLB entry) back to a ram offset.  */
ram_addr_t qemu_ram_addr_from_host(void *ptr)
{
    RAMBlock *block;
    ram_addr_t offset;

    block = qemu_ram_block_from_host(ptr, false, &offset);
    if (!block) {
        return RAM_ADDR_INVALID;
    }

    return block->offset + offset;
}

/* Called within RCU critical section. */
void memory_notdirty_write_prepare(NotDirtyInfo *ndi,
                          CPUState *cpu,
                          vaddr mem_vaddr,
                          ram_addr_t ram_addr,
                          unsigned size)
{
    ndi->cpu = cpu;
    ndi->ram_addr = ram_addr;
    ndi->mem_vaddr = mem_vaddr;
    ndi->size = size;
    ndi->pages = NULL;

    assert(tcg_enabled());
    if (!cpu_physical_memory_get_dirty_flag(ram_addr, DIRTY_MEMORY_CODE)) {
        ndi->pages = page_collection_lock(ram_addr, ram_addr + size);
        tb_invalidate_phys_page_fast(ndi->pages, ram_addr, size);
    }
}

/* Called within RCU critical section. */
void memory_notdirty_write_complete(NotDirtyInfo *ndi)
{
    if (ndi->pages) {
        assert(tcg_enabled());
        page_collection_unlock(ndi->pages);
        ndi->pages = NULL;
    }

    /* Set both VGA and migration bits for simplicity and to remove
     * the notdirty callback faster.
     */
    cpu_physical_memory_set_dirty_range(ndi->ram_addr, ndi->size,
                                        DIRTY_CLIENTS_NOCODE);
    /* we remove the notdirty callback only if the code has been
       flushed */
    if (!cpu_physical_memory_is_clean(ndi->ram_addr)) {
        tlb_set_dirty(ndi->cpu, ndi->mem_vaddr);
    }
}

/* Called within RCU critical section.  */
static void notdirty_mem_write(void *opaque, hwaddr ram_addr,
                               uint64_t val, unsigned size)
{
    NotDirtyInfo ndi;

    memory_notdirty_write_prepare(&ndi, current_cpu, current_cpu->mem_io_vaddr,
                         ram_addr, size);

    stn_p(qemu_map_ram_ptr(NULL, ram_addr), size, val);
    memory_notdirty_write_complete(&ndi);
}

static bool notdirty_mem_accepts(void *opaque, hwaddr addr,
                                 unsigned size, bool is_write,
                                 MemTxAttrs attrs)
{
    return is_write;
}

static const MemoryRegionOps notdirty_mem_ops = {
    .write = notdirty_mem_write,
    .valid.accepts = notdirty_mem_accepts,
    .endianness = DEVICE_NATIVE_ENDIAN,
    .valid = {
        .min_access_size = 1,
        .max_access_size = 8,
        .unaligned = false,
    },
    .impl = {
        .min_access_size = 1,
        .max_access_size = 8,
        .unaligned = false,
    },
};

/* Generate a debug exception if a watchpoint has been hit.  */
static void check_watchpoint(int offset, int len, MemTxAttrs attrs, int flags)
{
    CPUState *cpu = current_cpu;
    CPUClass *cc = CPU_GET_CLASS(cpu);
    target_ulong vaddr;
    CPUWatchpoint *wp;

    assert(tcg_enabled());
    if (cpu->watchpoint_hit) {
        /* We re-entered the check after replacing the TB. Now raise
         * the debug interrupt so that is will trigger after the
         * current instruction. */
        cpu_interrupt(cpu, CPU_INTERRUPT_DEBUG);
        return;
    }
    vaddr = (cpu->mem_io_vaddr & TARGET_PAGE_MASK) + offset;
    vaddr = cc->adjust_watchpoint_address(cpu, vaddr, len);
    QTAILQ_FOREACH(wp, &cpu->watchpoints, entry) {
        if (cpu_watchpoint_address_matches(wp, vaddr, len)
            && (wp->flags & flags)) {
            if (flags == BP_MEM_READ) {
                wp->flags |= BP_WATCHPOINT_HIT_READ;
            } else {
                wp->flags |= BP_WATCHPOINT_HIT_WRITE;
            }
            wp->hitaddr = vaddr;
            wp->hitattrs = attrs;
            if (!cpu->watchpoint_hit) {
                if (wp->flags & BP_CPU &&
                    !cc->debug_check_watchpoint(cpu, wp)) {
                    wp->flags &= ~BP_WATCHPOINT_HIT;
                    continue;
                }
                cpu->watchpoint_hit = wp;

                mmap_lock();
                tb_check_watchpoint(cpu);
                if (wp->flags & BP_STOP_BEFORE_ACCESS) {
                    cpu->exception_index = EXCP_DEBUG;
                    mmap_unlock();
                    cpu_loop_exit(cpu);
                } else {
                    /* Force execution of one insn next time.  */
                    cpu->cflags_next_tb = 1 | curr_cflags();
                    mmap_unlock();
                    cpu_loop_exit_noexc(cpu);
                }
            }
        } else {
            wp->flags &= ~BP_WATCHPOINT_HIT;
        }
    }
}

/* Watchpoint access routines.  Watchpoints are inserted using TLB tricks,
   so these check for a hit then pass through to the normal out-of-line
   phys routines.  */
static MemTxResult watch_mem_read(void *opaque, hwaddr addr, uint64_t *pdata,
                                  unsigned size, MemTxAttrs attrs)
{
    MemTxResult res;
    uint64_t data;
    int asidx = cpu_asidx_from_attrs(current_cpu, attrs);
    AddressSpace *as = current_cpu->cpu_ases[asidx].as;

    check_watchpoint(addr & ~TARGET_PAGE_MASK, size, attrs, BP_MEM_READ);
    switch (size) {
    case 1:
        data = address_space_ldub(as, addr, attrs, &res);
        break;
    case 2:
        data = address_space_lduw(as, addr, attrs, &res);
        break;
    case 4:
        data = address_space_ldl(as, addr, attrs, &res);
        break;
    case 8:
        data = address_space_ldq(as, addr, attrs, &res);
        break;
    default: abort();
    }
    *pdata = data;
    return res;
}

static MemTxResult watch_mem_write(void *opaque, hwaddr addr,
                                   uint64_t val, unsigned size,
                                   MemTxAttrs attrs)
{
    MemTxResult res;
    int asidx = cpu_asidx_from_attrs(current_cpu, attrs);
    AddressSpace *as = current_cpu->cpu_ases[asidx].as;

    check_watchpoint(addr & ~TARGET_PAGE_MASK, size, attrs, BP_MEM_WRITE);
    switch (size) {
    case 1:
        address_space_stb(as, addr, val, attrs, &res);
        break;
    case 2:
        address_space_stw(as, addr, val, attrs, &res);
        break;
    case 4:
        address_space_stl(as, addr, val, attrs, &res);
        break;
    case 8:
        address_space_stq(as, addr, val, attrs, &res);
        break;
    default: abort();
    }
    return res;
}

static const MemoryRegionOps watch_mem_ops = {
    .read_with_attrs = watch_mem_read,
    .write_with_attrs = watch_mem_write,
    .endianness = DEVICE_NATIVE_ENDIAN,
    .valid = {
        .min_access_size = 1,
        .max_access_size = 8,
        .unaligned = false,
    },
    .impl = {
        .min_access_size = 1,
        .max_access_size = 8,
        .unaligned = false,
    },
};

static MemTxResult flatview_read(FlatView *fv, hwaddr addr,
                                 MemTxAttrs attrs, uint8_t *buf, hwaddr len);
static MemTxResult flatview_write(FlatView *fv, hwaddr addr, MemTxAttrs attrs,
                                  const uint8_t *buf, hwaddr len);
static bool flatview_access_valid(FlatView *fv, hwaddr addr, hwaddr len,
                                  bool is_write, MemTxAttrs attrs);

static MemTxResult subpage_read(void *opaque, hwaddr addr, uint64_t *data,
                                unsigned len, MemTxAttrs attrs)
{
    subpage_t *subpage = opaque;
    uint8_t buf[8];
    MemTxResult res;

#if defined(DEBUG_SUBPAGE)
    printf("%s: subpage %p len %u addr " TARGET_FMT_plx "\n", __func__,
           subpage, len, addr);
#endif
    res = flatview_read(subpage->fv, addr + subpage->base, attrs, buf, len);
    if (res) {
        return res;
    }
    *data = ldn_p(buf, len);
    return MEMTX_OK;
}

static MemTxResult subpage_write(void *opaque, hwaddr addr,
                                 uint64_t value, unsigned len, MemTxAttrs attrs)
{
    subpage_t *subpage = opaque;
    uint8_t buf[8];

#if defined(DEBUG_SUBPAGE)
    printf("%s: subpage %p len %u addr " TARGET_FMT_plx
           " value %"PRIx64"\n",
           __func__, subpage, len, addr, value);
#endif
    stn_p(buf, len, value);
    return flatview_write(subpage->fv, addr + subpage->base, attrs, buf, len);
}

static bool subpage_accepts(void *opaque, hwaddr addr,
                            unsigned len, bool is_write,
                            MemTxAttrs attrs)
{
    subpage_t *subpage = opaque;
#if defined(DEBUG_SUBPAGE)
    printf("%s: subpage %p %c len %u addr " TARGET_FMT_plx "\n",
           __func__, subpage, is_write ? 'w' : 'r', len, addr);
#endif

    return flatview_access_valid(subpage->fv, addr + subpage->base,
                                 len, is_write, attrs);
}

static const MemoryRegionOps subpage_ops = {
    .read_with_attrs = subpage_read,
    .write_with_attrs = subpage_write,
    .impl.min_access_size = 1,
    .impl.max_access_size = 8,