memory.c 48.4 KB
Newer Older
A
Avi Kivity 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 * Physical memory management
 *
 * Copyright 2011 Red Hat, Inc. and/or its affiliates
 *
 * Authors:
 *  Avi Kivity <avi@redhat.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2.  See
 * the COPYING file in the top-level directory.
 *
 */

#include "memory.h"
A
Avi Kivity 已提交
15
#include "exec-memory.h"
A
Avi Kivity 已提交
16
#include "ioport.h"
17
#include "bitops.h"
A
Avi Kivity 已提交
18
#include "kvm.h"
A
Avi Kivity 已提交
19 20
#include <assert.h>

21 22 23
#define WANT_EXEC_OBSOLETE
#include "exec-obsolete.h"

A
Avi Kivity 已提交
24
unsigned memory_region_transaction_depth = 0;
25
static bool memory_region_update_pending = false;
26 27 28 29
static bool global_dirty_log = false;

static QLIST_HEAD(, MemoryListener) memory_listeners
    = QLIST_HEAD_INITIALIZER(memory_listeners);
A
Avi Kivity 已提交
30

A
Avi Kivity 已提交
31 32
typedef struct AddrRange AddrRange;

A
Avi Kivity 已提交
33 34 35 36 37
/*
 * Note using signed integers limits us to physical addresses at most
 * 63 bits wide.  They are needed for negative offsetting in aliases
 * (large MemoryRegion::alias_offset).
 */
A
Avi Kivity 已提交
38
struct AddrRange {
39 40
    Int128 start;
    Int128 size;
A
Avi Kivity 已提交
41 42
};

43
static AddrRange addrrange_make(Int128 start, Int128 size)
A
Avi Kivity 已提交
44 45 46 47 48 49
{
    return (AddrRange) { start, size };
}

static bool addrrange_equal(AddrRange r1, AddrRange r2)
{
50
    return int128_eq(r1.start, r2.start) && int128_eq(r1.size, r2.size);
A
Avi Kivity 已提交
51 52
}

53
static Int128 addrrange_end(AddrRange r)
A
Avi Kivity 已提交
54
{
55
    return int128_add(r.start, r.size);
A
Avi Kivity 已提交
56 57
}

58
static AddrRange addrrange_shift(AddrRange range, Int128 delta)
A
Avi Kivity 已提交
59
{
60
    int128_addto(&range.start, delta);
A
Avi Kivity 已提交
61 62 63
    return range;
}

64 65 66 67 68 69
static bool addrrange_contains(AddrRange range, Int128 addr)
{
    return int128_ge(addr, range.start)
        && int128_lt(addr, addrrange_end(range));
}

A
Avi Kivity 已提交
70 71
static bool addrrange_intersects(AddrRange r1, AddrRange r2)
{
72 73
    return addrrange_contains(r1, r2.start)
        || addrrange_contains(r2, r1.start);
A
Avi Kivity 已提交
74 75 76 77
}

static AddrRange addrrange_intersection(AddrRange r1, AddrRange r2)
{
78 79 80
    Int128 start = int128_max(r1.start, r2.start);
    Int128 end = int128_min(addrrange_end(r1), addrrange_end(r2));
    return addrrange_make(start, int128_sub(end, start));
A
Avi Kivity 已提交
81 82 83 84 85 86 87
}

struct CoalescedMemoryRange {
    AddrRange addr;
    QTAILQ_ENTRY(CoalescedMemoryRange) link;
};

A
Avi Kivity 已提交
88 89 90 91 92 93 94 95 96 97
struct MemoryRegionIoeventfd {
    AddrRange addr;
    bool match_data;
    uint64_t data;
    int fd;
};

static bool memory_region_ioeventfd_before(MemoryRegionIoeventfd a,
                                           MemoryRegionIoeventfd b)
{
98
    if (int128_lt(a.addr.start, b.addr.start)) {
A
Avi Kivity 已提交
99
        return true;
100
    } else if (int128_gt(a.addr.start, b.addr.start)) {
A
Avi Kivity 已提交
101
        return false;
102
    } else if (int128_lt(a.addr.size, b.addr.size)) {
A
Avi Kivity 已提交
103
        return true;
104
    } else if (int128_gt(a.addr.size, b.addr.size)) {
A
Avi Kivity 已提交
105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
        return false;
    } else if (a.match_data < b.match_data) {
        return true;
    } else  if (a.match_data > b.match_data) {
        return false;
    } else if (a.match_data) {
        if (a.data < b.data) {
            return true;
        } else if (a.data > b.data) {
            return false;
        }
    }
    if (a.fd < b.fd) {
        return true;
    } else if (a.fd > b.fd) {
        return false;
    }
    return false;
}

static bool memory_region_ioeventfd_equal(MemoryRegionIoeventfd a,
                                          MemoryRegionIoeventfd b)
{
    return !memory_region_ioeventfd_before(a, b)
        && !memory_region_ioeventfd_before(b, a);
}

A
Avi Kivity 已提交
132 133 134 135 136 137 138 139
typedef struct FlatRange FlatRange;
typedef struct FlatView FlatView;

/* Range of memory in the global map.  Addresses are absolute. */
struct FlatRange {
    MemoryRegion *mr;
    target_phys_addr_t offset_in_region;
    AddrRange addr;
A
Avi Kivity 已提交
140
    uint8_t dirty_log_mask;
141
    bool readable;
142
    bool readonly;
A
Avi Kivity 已提交
143 144 145 146 147 148 149 150 151 152 153
};

/* Flattened global view of current active memory hierarchy.  Kept in sorted
 * order.
 */
struct FlatView {
    FlatRange *ranges;
    unsigned nr;
    unsigned nr_allocated;
};

154 155 156 157 158 159 160 161
typedef struct AddressSpace AddressSpace;
typedef struct AddressSpaceOps AddressSpaceOps;

/* A system address space - I/O, memory, etc. */
struct AddressSpace {
    const AddressSpaceOps *ops;
    MemoryRegion *root;
    FlatView current_map;
A
Avi Kivity 已提交
162 163
    int ioeventfd_nb;
    MemoryRegionIoeventfd *ioeventfds;
164 165 166 167 168 169 170
};

struct AddressSpaceOps {
    void (*range_add)(AddressSpace *as, FlatRange *fr);
    void (*range_del)(AddressSpace *as, FlatRange *fr);
    void (*log_start)(AddressSpace *as, FlatRange *fr);
    void (*log_stop)(AddressSpace *as, FlatRange *fr);
A
Avi Kivity 已提交
171 172
    void (*ioeventfd_add)(AddressSpace *as, MemoryRegionIoeventfd *fd);
    void (*ioeventfd_del)(AddressSpace *as, MemoryRegionIoeventfd *fd);
173 174
};

A
Avi Kivity 已提交
175 176 177 178 179 180 181
#define FOR_EACH_FLAT_RANGE(var, view)          \
    for (var = (view)->ranges; var < (view)->ranges + (view)->nr; ++var)

static bool flatrange_equal(FlatRange *a, FlatRange *b)
{
    return a->mr == b->mr
        && addrrange_equal(a->addr, b->addr)
182
        && a->offset_in_region == b->offset_in_region
183 184
        && a->readable == b->readable
        && a->readonly == b->readonly;
A
Avi Kivity 已提交
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
}

static void flatview_init(FlatView *view)
{
    view->ranges = NULL;
    view->nr = 0;
    view->nr_allocated = 0;
}

/* Insert a range into a given position.  Caller is responsible for maintaining
 * sorting order.
 */
static void flatview_insert(FlatView *view, unsigned pos, FlatRange *range)
{
    if (view->nr == view->nr_allocated) {
        view->nr_allocated = MAX(2 * view->nr, 10);
201
        view->ranges = g_realloc(view->ranges,
A
Avi Kivity 已提交
202 203 204 205 206 207 208 209 210 211
                                    view->nr_allocated * sizeof(*view->ranges));
    }
    memmove(view->ranges + pos + 1, view->ranges + pos,
            (view->nr - pos) * sizeof(FlatRange));
    view->ranges[pos] = *range;
    ++view->nr;
}

static void flatview_destroy(FlatView *view)
{
212
    g_free(view->ranges);
A
Avi Kivity 已提交
213 214
}

215 216
static bool can_merge(FlatRange *r1, FlatRange *r2)
{
217
    return int128_eq(addrrange_end(r1->addr), r2->addr.start)
218
        && r1->mr == r2->mr
219 220 221
        && int128_eq(int128_add(int128_make64(r1->offset_in_region),
                                r1->addr.size),
                     int128_make64(r2->offset_in_region))
222
        && r1->dirty_log_mask == r2->dirty_log_mask
223 224
        && r1->readable == r2->readable
        && r1->readonly == r2->readonly;
225 226 227 228 229 230 231 232 233 234 235 236
}

/* Attempt to simplify a view by merging ajacent ranges */
static void flatview_simplify(FlatView *view)
{
    unsigned i, j;

    i = 0;
    while (i < view->nr) {
        j = i + 1;
        while (j < view->nr
               && can_merge(&view->ranges[j-1], &view->ranges[j])) {
237
            int128_addto(&view->ranges[i].addr.size, view->ranges[j].addr.size);
238 239 240 241 242 243 244 245 246
            ++j;
        }
        ++i;
        memmove(&view->ranges[i], &view->ranges[j],
                (view->nr - j) * sizeof(view->ranges[j]));
        view->nr -= j - i;
    }
}

247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
static void memory_region_read_accessor(void *opaque,
                                        target_phys_addr_t addr,
                                        uint64_t *value,
                                        unsigned size,
                                        unsigned shift,
                                        uint64_t mask)
{
    MemoryRegion *mr = opaque;
    uint64_t tmp;

    tmp = mr->ops->read(mr->opaque, addr, size);
    *value |= (tmp & mask) << shift;
}

static void memory_region_write_accessor(void *opaque,
                                         target_phys_addr_t addr,
                                         uint64_t *value,
                                         unsigned size,
                                         unsigned shift,
                                         uint64_t mask)
{
    MemoryRegion *mr = opaque;
    uint64_t tmp;

    tmp = (*value >> shift) & mask;
    mr->ops->write(mr->opaque, addr, tmp, size);
}

static void access_with_adjusted_size(target_phys_addr_t addr,
                                      uint64_t *value,
                                      unsigned size,
                                      unsigned access_size_min,
                                      unsigned access_size_max,
                                      void (*access)(void *opaque,
                                                     target_phys_addr_t addr,
                                                     uint64_t *value,
                                                     unsigned size,
                                                     unsigned shift,
                                                     uint64_t mask),
                                      void *opaque)
{
    uint64_t access_mask;
    unsigned access_size;
    unsigned i;

    if (!access_size_min) {
        access_size_min = 1;
    }
    if (!access_size_max) {
        access_size_max = 4;
    }
    access_size = MAX(MIN(size, access_size_max), access_size_min);
    access_mask = -1ULL >> (64 - access_size * 8);
    for (i = 0; i < size; i += access_size) {
        /* FIXME: big-endian support */
        access(opaque, addr + i, value, access_size, i * 8, access_mask);
    }
}

306 307
static void as_memory_range_add(AddressSpace *as, FlatRange *fr)
{
308 309 310 311 312 313 314 315
    MemoryRegionSection section = {
        .mr = fr->mr,
        .offset_within_address_space = int128_get64(fr->addr.start),
        .offset_within_region = fr->offset_in_region,
        .size = int128_get64(fr->addr.size),
    };

    cpu_register_physical_memory_log(&section, fr->readable, fr->readonly);
316 317 318 319
}

static void as_memory_range_del(AddressSpace *as, FlatRange *fr)
{
320 321 322 323 324 325 326 327
    MemoryRegionSection section = {
        .mr = &io_mem_unassigned,
        .offset_within_address_space = int128_get64(fr->addr.start),
        .offset_within_region = int128_get64(fr->addr.start),
        .size = int128_get64(fr->addr.size),
    };

    cpu_register_physical_memory_log(&section, true, false);
328 329 330 331 332 333 334 335 336 337
}

static void as_memory_log_start(AddressSpace *as, FlatRange *fr)
{
}

static void as_memory_log_stop(AddressSpace *as, FlatRange *fr)
{
}

A
Avi Kivity 已提交
338 339 340 341
static void as_memory_ioeventfd_add(AddressSpace *as, MemoryRegionIoeventfd *fd)
{
    int r;

342
    assert(fd->match_data && int128_get64(fd->addr.size) == 4);
A
Avi Kivity 已提交
343

344 345
    r = kvm_set_ioeventfd_mmio_long(fd->fd, int128_get64(fd->addr.start),
                                    fd->data, true);
A
Avi Kivity 已提交
346 347 348 349 350 351 352 353 354
    if (r < 0) {
        abort();
    }
}

static void as_memory_ioeventfd_del(AddressSpace *as, MemoryRegionIoeventfd *fd)
{
    int r;

355 356
    r = kvm_set_ioeventfd_mmio_long(fd->fd, int128_get64(fd->addr.start),
                                    fd->data, false);
A
Avi Kivity 已提交
357 358 359 360 361
    if (r < 0) {
        abort();
    }
}

362 363 364 365 366
static const AddressSpaceOps address_space_ops_memory = {
    .range_add = as_memory_range_add,
    .range_del = as_memory_range_del,
    .log_start = as_memory_log_start,
    .log_stop = as_memory_log_stop,
A
Avi Kivity 已提交
367 368
    .ioeventfd_add = as_memory_ioeventfd_add,
    .ioeventfd_del = as_memory_ioeventfd_del,
369 370 371 372 373 374
};

static AddressSpace address_space_memory = {
    .ops = &address_space_ops_memory,
};

375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
static const MemoryRegionPortio *find_portio(MemoryRegion *mr, uint64_t offset,
                                             unsigned width, bool write)
{
    const MemoryRegionPortio *mrp;

    for (mrp = mr->ops->old_portio; mrp->size; ++mrp) {
        if (offset >= mrp->offset && offset < mrp->offset + mrp->len
            && width == mrp->size
            && (write ? (bool)mrp->write : (bool)mrp->read)) {
            return mrp;
        }
    }
    return NULL;
}

A
Avi Kivity 已提交
390 391 392 393 394 395 396
static void memory_region_iorange_read(IORange *iorange,
                                       uint64_t offset,
                                       unsigned width,
                                       uint64_t *data)
{
    MemoryRegion *mr = container_of(iorange, MemoryRegion, iorange);

397 398 399 400 401
    if (mr->ops->old_portio) {
        const MemoryRegionPortio *mrp = find_portio(mr, offset, width, false);

        *data = ((uint64_t)1 << (width * 8)) - 1;
        if (mrp) {
A
Avi Kivity 已提交
402
            *data = mrp->read(mr->opaque, offset + mr->offset);
J
Jan Kiszka 已提交
403 404 405 406 407
        } else if (width == 2) {
            mrp = find_portio(mr, offset, 1, false);
            assert(mrp);
            *data = mrp->read(mr->opaque, offset + mr->offset) |
                    (mrp->read(mr->opaque, offset + mr->offset + 1) << 8);
408 409 410
        }
        return;
    }
411
    *data = 0;
A
Avi Kivity 已提交
412
    access_with_adjusted_size(offset + mr->offset, data, width,
413 414 415
                              mr->ops->impl.min_access_size,
                              mr->ops->impl.max_access_size,
                              memory_region_read_accessor, mr);
A
Avi Kivity 已提交
416 417 418 419 420 421 422 423 424
}

static void memory_region_iorange_write(IORange *iorange,
                                        uint64_t offset,
                                        unsigned width,
                                        uint64_t data)
{
    MemoryRegion *mr = container_of(iorange, MemoryRegion, iorange);

425 426 427 428
    if (mr->ops->old_portio) {
        const MemoryRegionPortio *mrp = find_portio(mr, offset, width, true);

        if (mrp) {
A
Avi Kivity 已提交
429
            mrp->write(mr->opaque, offset + mr->offset, data);
J
Jan Kiszka 已提交
430 431 432 433 434
        } else if (width == 2) {
            mrp = find_portio(mr, offset, 1, false);
            assert(mrp);
            mrp->write(mr->opaque, offset + mr->offset, data & 0xff);
            mrp->write(mr->opaque, offset + mr->offset + 1, data >> 8);
435 436 437
        }
        return;
    }
A
Avi Kivity 已提交
438
    access_with_adjusted_size(offset + mr->offset, &data, width,
439 440 441
                              mr->ops->impl.min_access_size,
                              mr->ops->impl.max_access_size,
                              memory_region_write_accessor, mr);
A
Avi Kivity 已提交
442 443 444 445 446 447 448 449 450 451
}

static const IORangeOps memory_region_iorange_ops = {
    .read = memory_region_iorange_read,
    .write = memory_region_iorange_write,
};

static void as_io_range_add(AddressSpace *as, FlatRange *fr)
{
    iorange_init(&fr->mr->iorange, &memory_region_iorange_ops,
452
                 int128_get64(fr->addr.start), int128_get64(fr->addr.size));
A
Avi Kivity 已提交
453 454 455 456 457
    ioport_register(&fr->mr->iorange);
}

static void as_io_range_del(AddressSpace *as, FlatRange *fr)
{
458 459
    isa_unassign_ioport(int128_get64(fr->addr.start),
                        int128_get64(fr->addr.size));
A
Avi Kivity 已提交
460 461
}

A
Avi Kivity 已提交
462 463 464 465
static void as_io_ioeventfd_add(AddressSpace *as, MemoryRegionIoeventfd *fd)
{
    int r;

466
    assert(fd->match_data && int128_get64(fd->addr.size) == 2);
A
Avi Kivity 已提交
467

468 469
    r = kvm_set_ioeventfd_pio_word(fd->fd, int128_get64(fd->addr.start),
                                   fd->data, true);
A
Avi Kivity 已提交
470 471 472 473 474 475 476 477 478
    if (r < 0) {
        abort();
    }
}

static void as_io_ioeventfd_del(AddressSpace *as, MemoryRegionIoeventfd *fd)
{
    int r;

479 480
    r = kvm_set_ioeventfd_pio_word(fd->fd, int128_get64(fd->addr.start),
                                   fd->data, false);
A
Avi Kivity 已提交
481 482 483 484 485
    if (r < 0) {
        abort();
    }
}

A
Avi Kivity 已提交
486 487 488
static const AddressSpaceOps address_space_ops_io = {
    .range_add = as_io_range_add,
    .range_del = as_io_range_del,
A
Avi Kivity 已提交
489 490
    .ioeventfd_add = as_io_ioeventfd_add,
    .ioeventfd_del = as_io_ioeventfd_del,
A
Avi Kivity 已提交
491 492 493 494 495 496
};

static AddressSpace address_space_io = {
    .ops = &address_space_ops_io,
};

497 498 499 500 501 502 503 504 505 506 507 508 509 510
static AddressSpace *memory_region_to_address_space(MemoryRegion *mr)
{
    while (mr->parent) {
        mr = mr->parent;
    }
    if (mr == address_space_memory.root) {
        return &address_space_memory;
    }
    if (mr == address_space_io.root) {
        return &address_space_io;
    }
    abort();
}

A
Avi Kivity 已提交
511 512 513 514 515
/* Render a memory region into the global view.  Ranges in @view obscure
 * ranges in @mr.
 */
static void render_memory_region(FlatView *view,
                                 MemoryRegion *mr,
516
                                 Int128 base,
517 518
                                 AddrRange clip,
                                 bool readonly)
A
Avi Kivity 已提交
519 520 521 522
{
    MemoryRegion *subregion;
    unsigned i;
    target_phys_addr_t offset_in_region;
523 524
    Int128 remain;
    Int128 now;
A
Avi Kivity 已提交
525 526 527
    FlatRange fr;
    AddrRange tmp;

528 529 530 531
    if (!mr->enabled) {
        return;
    }

532
    int128_addto(&base, int128_make64(mr->addr));
533
    readonly |= mr->readonly;
A
Avi Kivity 已提交
534 535 536 537 538 539 540 541 542 543

    tmp = addrrange_make(base, mr->size);

    if (!addrrange_intersects(tmp, clip)) {
        return;
    }

    clip = addrrange_intersection(tmp, clip);

    if (mr->alias) {
544 545
        int128_subfrom(&base, int128_make64(mr->alias->addr));
        int128_subfrom(&base, int128_make64(mr->alias_offset));
546
        render_memory_region(view, mr->alias, base, clip, readonly);
A
Avi Kivity 已提交
547 548 549 550 551
        return;
    }

    /* Render subregions in priority order. */
    QTAILQ_FOREACH(subregion, &mr->subregions, subregions_link) {
552
        render_memory_region(view, subregion, base, clip, readonly);
A
Avi Kivity 已提交
553 554
    }

555
    if (!mr->terminates) {
A
Avi Kivity 已提交
556 557 558
        return;
    }

559
    offset_in_region = int128_get64(int128_sub(clip.start, base));
A
Avi Kivity 已提交
560 561 562 563
    base = clip.start;
    remain = clip.size;

    /* Render the region itself into any gaps left by the current view. */
564 565
    for (i = 0; i < view->nr && int128_nz(remain); ++i) {
        if (int128_ge(base, addrrange_end(view->ranges[i].addr))) {
A
Avi Kivity 已提交
566 567
            continue;
        }
568 569 570
        if (int128_lt(base, view->ranges[i].addr.start)) {
            now = int128_min(remain,
                             int128_sub(view->ranges[i].addr.start, base));
A
Avi Kivity 已提交
571 572 573
            fr.mr = mr;
            fr.offset_in_region = offset_in_region;
            fr.addr = addrrange_make(base, now);
A
Avi Kivity 已提交
574
            fr.dirty_log_mask = mr->dirty_log_mask;
575
            fr.readable = mr->readable;
576
            fr.readonly = readonly;
A
Avi Kivity 已提交
577 578
            flatview_insert(view, i, &fr);
            ++i;
579 580 581
            int128_addto(&base, now);
            offset_in_region += int128_get64(now);
            int128_subfrom(&remain, now);
A
Avi Kivity 已提交
582
        }
583 584 585 586 587
        if (int128_eq(base, view->ranges[i].addr.start)) {
            now = int128_min(remain, view->ranges[i].addr.size);
            int128_addto(&base, now);
            offset_in_region += int128_get64(now);
            int128_subfrom(&remain, now);
A
Avi Kivity 已提交
588 589
        }
    }
590
    if (int128_nz(remain)) {
A
Avi Kivity 已提交
591 592 593
        fr.mr = mr;
        fr.offset_in_region = offset_in_region;
        fr.addr = addrrange_make(base, remain);
A
Avi Kivity 已提交
594
        fr.dirty_log_mask = mr->dirty_log_mask;
595
        fr.readable = mr->readable;
596
        fr.readonly = readonly;
A
Avi Kivity 已提交
597 598 599 600 601 602 603 604 605 606 607
        flatview_insert(view, i, &fr);
    }
}

/* Render a memory topology into a list of disjoint absolute ranges. */
static FlatView generate_memory_topology(MemoryRegion *mr)
{
    FlatView view;

    flatview_init(&view);

608 609
    render_memory_region(&view, mr, int128_zero(),
                         addrrange_make(int128_zero(), int128_2_64()), false);
610
    flatview_simplify(&view);
A
Avi Kivity 已提交
611 612 613 614

    return view;
}

A
Avi Kivity 已提交
615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658
static void address_space_add_del_ioeventfds(AddressSpace *as,
                                             MemoryRegionIoeventfd *fds_new,
                                             unsigned fds_new_nb,
                                             MemoryRegionIoeventfd *fds_old,
                                             unsigned fds_old_nb)
{
    unsigned iold, inew;

    /* Generate a symmetric difference of the old and new fd sets, adding
     * and deleting as necessary.
     */

    iold = inew = 0;
    while (iold < fds_old_nb || inew < fds_new_nb) {
        if (iold < fds_old_nb
            && (inew == fds_new_nb
                || memory_region_ioeventfd_before(fds_old[iold],
                                                  fds_new[inew]))) {
            as->ops->ioeventfd_del(as, &fds_old[iold]);
            ++iold;
        } else if (inew < fds_new_nb
                   && (iold == fds_old_nb
                       || memory_region_ioeventfd_before(fds_new[inew],
                                                         fds_old[iold]))) {
            as->ops->ioeventfd_add(as, &fds_new[inew]);
            ++inew;
        } else {
            ++iold;
            ++inew;
        }
    }
}

static void address_space_update_ioeventfds(AddressSpace *as)
{
    FlatRange *fr;
    unsigned ioeventfd_nb = 0;
    MemoryRegionIoeventfd *ioeventfds = NULL;
    AddrRange tmp;
    unsigned i;

    FOR_EACH_FLAT_RANGE(fr, &as->current_map) {
        for (i = 0; i < fr->mr->ioeventfd_nb; ++i) {
            tmp = addrrange_shift(fr->mr->ioeventfds[i].addr,
659 660
                                  int128_sub(fr->addr.start,
                                             int128_make64(fr->offset_in_region)));
A
Avi Kivity 已提交
661 662
            if (addrrange_intersects(fr->addr, tmp)) {
                ++ioeventfd_nb;
663
                ioeventfds = g_realloc(ioeventfds,
A
Avi Kivity 已提交
664 665 666 667 668 669 670 671 672 673
                                          ioeventfd_nb * sizeof(*ioeventfds));
                ioeventfds[ioeventfd_nb-1] = fr->mr->ioeventfds[i];
                ioeventfds[ioeventfd_nb-1].addr = tmp;
            }
        }
    }

    address_space_add_del_ioeventfds(as, ioeventfds, ioeventfd_nb,
                                     as->ioeventfds, as->ioeventfd_nb);

674
    g_free(as->ioeventfds);
A
Avi Kivity 已提交
675 676 677 678
    as->ioeventfds = ioeventfds;
    as->ioeventfd_nb = ioeventfd_nb;
}

679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
typedef void ListenerCallback(MemoryListener *listener,
                              MemoryRegionSection *mrs);

/* Want "void (&MemoryListener::*callback)(const MemoryRegionSection& s)" */
static void memory_listener_update_region(FlatRange *fr, AddressSpace *as,
                                          size_t callback_offset)
{
    MemoryRegionSection section = {
        .mr = fr->mr,
        .address_space = as->root,
        .offset_within_region = fr->offset_in_region,
        .size = int128_get64(fr->addr.size),
        .offset_within_address_space = int128_get64(fr->addr.start),
    };
    MemoryListener *listener;

    QLIST_FOREACH(listener, &memory_listeners, link) {
        ListenerCallback *callback
            = *(ListenerCallback **)((void *)listener + callback_offset);
        callback(listener, &section);
    }
}

#define MEMORY_LISTENER_UPDATE_REGION(fr, as, callback) \
    memory_listener_update_region(fr, as, offsetof(MemoryListener, callback))

705 706 707 708
static void address_space_update_topology_pass(AddressSpace *as,
                                               FlatView old_view,
                                               FlatView new_view,
                                               bool adding)
A
Avi Kivity 已提交
709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
{
    unsigned iold, inew;
    FlatRange *frold, *frnew;

    /* Generate a symmetric difference of the old and new memory maps.
     * Kill ranges in the old map, and instantiate ranges in the new map.
     */
    iold = inew = 0;
    while (iold < old_view.nr || inew < new_view.nr) {
        if (iold < old_view.nr) {
            frold = &old_view.ranges[iold];
        } else {
            frold = NULL;
        }
        if (inew < new_view.nr) {
            frnew = &new_view.ranges[inew];
        } else {
            frnew = NULL;
        }

        if (frold
            && (!frnew
731 732
                || int128_lt(frold->addr.start, frnew->addr.start)
                || (int128_eq(frold->addr.start, frnew->addr.start)
A
Avi Kivity 已提交
733 734 735
                    && !flatrange_equal(frold, frnew)))) {
            /* In old, but (not in new, or in new but attributes changed). */

736
            if (!adding) {
737
                MEMORY_LISTENER_UPDATE_REGION(frold, as, region_del);
738 739 740
                as->ops->range_del(as, frold);
            }

A
Avi Kivity 已提交
741 742 743 744
            ++iold;
        } else if (frold && frnew && flatrange_equal(frold, frnew)) {
            /* In both (logging may have changed) */

745 746
            if (adding) {
                if (frold->dirty_log_mask && !frnew->dirty_log_mask) {
747
                    MEMORY_LISTENER_UPDATE_REGION(frnew, as, log_stop);
748 749 750
                    as->ops->log_stop(as, frnew);
                } else if (frnew->dirty_log_mask && !frold->dirty_log_mask) {
                    as->ops->log_start(as, frnew);
751
                    MEMORY_LISTENER_UPDATE_REGION(frnew, as, log_start);
752
                }
A
Avi Kivity 已提交
753 754
            }

A
Avi Kivity 已提交
755 756 757 758 759
            ++iold;
            ++inew;
        } else {
            /* In new */

760 761
            if (adding) {
                as->ops->range_add(as, frnew);
762
                MEMORY_LISTENER_UPDATE_REGION(frnew, as, region_add);
763 764
            }

A
Avi Kivity 已提交
765 766 767
            ++inew;
        }
    }
768 769 770 771 772 773 774 775 776 777 778
}


static void address_space_update_topology(AddressSpace *as)
{
    FlatView old_view = as->current_map;
    FlatView new_view = generate_memory_topology(as->root);

    address_space_update_topology_pass(as, old_view, new_view, false);
    address_space_update_topology_pass(as, old_view, new_view, true);

779
    as->current_map = new_view;
A
Avi Kivity 已提交
780
    flatview_destroy(&old_view);
A
Avi Kivity 已提交
781
    address_space_update_ioeventfds(as);
A
Avi Kivity 已提交
782 783
}

784
static void memory_region_update_topology(MemoryRegion *mr)
785
{
A
Avi Kivity 已提交
786
    if (memory_region_transaction_depth) {
787
        memory_region_update_pending |= !mr || mr->enabled;
A
Avi Kivity 已提交
788 789 790
        return;
    }

791 792 793 794
    if (mr && !mr->enabled) {
        return;
    }

A
Avi Kivity 已提交
795 796 797 798 799 800
    if (address_space_memory.root) {
        address_space_update_topology(&address_space_memory);
    }
    if (address_space_io.root) {
        address_space_update_topology(&address_space_io);
    }
801 802

    memory_region_update_pending = false;
803 804
}

A
Avi Kivity 已提交
805 806 807 808 809 810 811 812 813
void memory_region_transaction_begin(void)
{
    ++memory_region_transaction_depth;
}

void memory_region_transaction_commit(void)
{
    assert(memory_region_transaction_depth);
    --memory_region_transaction_depth;
814 815 816
    if (!memory_region_transaction_depth && memory_region_update_pending) {
        memory_region_update_topology(NULL);
    }
A
Avi Kivity 已提交
817 818
}

819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837
static void memory_region_destructor_none(MemoryRegion *mr)
{
}

static void memory_region_destructor_ram(MemoryRegion *mr)
{
    qemu_ram_free(mr->ram_addr);
}

static void memory_region_destructor_ram_from_ptr(MemoryRegion *mr)
{
    qemu_ram_free_from_ptr(mr->ram_addr);
}

static void memory_region_destructor_iomem(MemoryRegion *mr)
{
    cpu_unregister_io_memory(mr->ram_addr);
}

838 839 840
static void memory_region_destructor_rom_device(MemoryRegion *mr)
{
    qemu_ram_free(mr->ram_addr & TARGET_PAGE_MASK);
A
Avi Kivity 已提交
841
    cpu_unregister_io_memory(mr->ram_addr & ~TARGET_PAGE_MASK);
842 843
}

844 845 846 847 848 849 850 851 852
static bool memory_region_wrong_endianness(MemoryRegion *mr)
{
#ifdef TARGET_BIG_ENDIAN
    return mr->ops->endianness == DEVICE_LITTLE_ENDIAN;
#else
    return mr->ops->endianness == DEVICE_BIG_ENDIAN;
#endif
}

A
Avi Kivity 已提交
853 854 855 856 857 858
void memory_region_init(MemoryRegion *mr,
                        const char *name,
                        uint64_t size)
{
    mr->ops = NULL;
    mr->parent = NULL;
859 860 861 862
    mr->size = int128_make64(size);
    if (size == UINT64_MAX) {
        mr->size = int128_2_64();
    }
A
Avi Kivity 已提交
863 864
    mr->addr = 0;
    mr->offset = 0;
A
Avi Kivity 已提交
865
    mr->subpage = false;
866
    mr->enabled = true;
867
    mr->terminates = false;
A
Avi Kivity 已提交
868
    mr->ram = false;
869
    mr->readable = true;
870
    mr->readonly = false;
A
Avi Kivity 已提交
871
    mr->rom_device = false;
872
    mr->destructor = memory_region_destructor_none;
A
Avi Kivity 已提交
873 874 875 876 877 878
    mr->priority = 0;
    mr->may_overlap = false;
    mr->alias = NULL;
    QTAILQ_INIT(&mr->subregions);
    memset(&mr->subregions_link, 0, sizeof mr->subregions_link);
    QTAILQ_INIT(&mr->coalesced);
879
    mr->name = g_strdup(name);
A
Avi Kivity 已提交
880
    mr->dirty_log_mask = 0;
A
Avi Kivity 已提交
881 882
    mr->ioeventfd_nb = 0;
    mr->ioeventfds = NULL;
A
Avi Kivity 已提交
883 884 885 886
}

static bool memory_region_access_valid(MemoryRegion *mr,
                                       target_phys_addr_t addr,
887 888
                                       unsigned size,
                                       bool is_write)
A
Avi Kivity 已提交
889
{
890 891 892 893 894
    if (mr->ops->valid.accepts
        && !mr->ops->valid.accepts(mr->opaque, addr, size, is_write)) {
        return false;
    }

A
Avi Kivity 已提交
895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910
    if (!mr->ops->valid.unaligned && (addr & (size - 1))) {
        return false;
    }

    /* Treat zero as compatibility all valid */
    if (!mr->ops->valid.max_access_size) {
        return true;
    }

    if (size > mr->ops->valid.max_access_size
        || size < mr->ops->valid.min_access_size) {
        return false;
    }
    return true;
}

911 912 913
static uint64_t memory_region_dispatch_read1(MemoryRegion *mr,
                                             target_phys_addr_t addr,
                                             unsigned size)
A
Avi Kivity 已提交
914
{
915
    uint64_t data = 0;
A
Avi Kivity 已提交
916

917
    if (!memory_region_access_valid(mr, addr, size, false)) {
A
Avi Kivity 已提交
918 919 920
        return -1U; /* FIXME: better signalling */
    }

921 922 923 924
    if (!mr->ops->read) {
        return mr->ops->old_mmio.read[bitops_ffsl(size)](mr->opaque, addr);
    }

A
Avi Kivity 已提交
925
    /* FIXME: support unaligned access */
926 927 928 929
    access_with_adjusted_size(addr + mr->offset, &data, size,
                              mr->ops->impl.min_access_size,
                              mr->ops->impl.max_access_size,
                              memory_region_read_accessor, mr);
A
Avi Kivity 已提交
930 931 932 933

    return data;
}

934
static void adjust_endianness(MemoryRegion *mr, uint64_t *data, unsigned size)
A
Avi Kivity 已提交
935
{
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960
    if (memory_region_wrong_endianness(mr)) {
        switch (size) {
        case 1:
            break;
        case 2:
            *data = bswap16(*data);
            break;
        case 4:
            *data = bswap32(*data);
        default:
            abort();
        }
    }
}

static uint64_t memory_region_dispatch_read(MemoryRegion *mr,
                                            target_phys_addr_t addr,
                                            unsigned size)
{
    uint64_t ret;

    ret = memory_region_dispatch_read1(mr, addr, size);
    adjust_endianness(mr, &ret, size);
    return ret;
}
A
Avi Kivity 已提交
961

962 963 964 965 966
static void memory_region_dispatch_write(MemoryRegion *mr,
                                         target_phys_addr_t addr,
                                         uint64_t data,
                                         unsigned size)
{
967
    if (!memory_region_access_valid(mr, addr, size, true)) {
A
Avi Kivity 已提交
968 969 970
        return; /* FIXME: better signalling */
    }

971 972
    adjust_endianness(mr, &data, size);

973 974 975 976 977
    if (!mr->ops->write) {
        mr->ops->old_mmio.write[bitops_ffsl(size)](mr->opaque, addr, data);
        return;
    }

A
Avi Kivity 已提交
978
    /* FIXME: support unaligned access */
979 980 981 982
    access_with_adjusted_size(addr + mr->offset, &data, size,
                              mr->ops->impl.min_access_size,
                              mr->ops->impl.max_access_size,
                              memory_region_write_accessor, mr);
A
Avi Kivity 已提交
983 984 985 986 987 988 989 990 991 992 993
}

void memory_region_init_io(MemoryRegion *mr,
                           const MemoryRegionOps *ops,
                           void *opaque,
                           const char *name,
                           uint64_t size)
{
    memory_region_init(mr, name, size);
    mr->ops = ops;
    mr->opaque = opaque;
994
    mr->terminates = true;
995
    mr->destructor = memory_region_destructor_iomem;
996
    mr->ram_addr = cpu_register_io_memory(mr);
A
Avi Kivity 已提交
997 998 999 1000 1001 1002 1003
}

void memory_region_init_ram(MemoryRegion *mr,
                            const char *name,
                            uint64_t size)
{
    memory_region_init(mr, name, size);
A
Avi Kivity 已提交
1004
    mr->ram = true;
1005
    mr->terminates = true;
1006
    mr->destructor = memory_region_destructor_ram;
1007
    mr->ram_addr = qemu_ram_alloc(size, mr);
A
Avi Kivity 已提交
1008 1009 1010 1011 1012 1013 1014 1015
}

void memory_region_init_ram_ptr(MemoryRegion *mr,
                                const char *name,
                                uint64_t size,
                                void *ptr)
{
    memory_region_init(mr, name, size);
A
Avi Kivity 已提交
1016
    mr->ram = true;
1017
    mr->terminates = true;
1018
    mr->destructor = memory_region_destructor_ram_from_ptr;
1019
    mr->ram_addr = qemu_ram_alloc_from_ptr(size, ptr, mr);
A
Avi Kivity 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
}

void memory_region_init_alias(MemoryRegion *mr,
                              const char *name,
                              MemoryRegion *orig,
                              target_phys_addr_t offset,
                              uint64_t size)
{
    memory_region_init(mr, name, size);
    mr->alias = orig;
    mr->alias_offset = offset;
}

1033 1034
void memory_region_init_rom_device(MemoryRegion *mr,
                                   const MemoryRegionOps *ops,
1035
                                   void *opaque,
1036 1037 1038 1039
                                   const char *name,
                                   uint64_t size)
{
    memory_region_init(mr, name, size);
1040
    mr->ops = ops;
1041
    mr->opaque = opaque;
1042
    mr->terminates = true;
A
Avi Kivity 已提交
1043
    mr->rom_device = true;
1044
    mr->destructor = memory_region_destructor_rom_device;
1045
    mr->ram_addr = qemu_ram_alloc(size, mr);
1046
    mr->ram_addr |= cpu_register_io_memory(mr);
1047 1048
}

A
Avi Kivity 已提交
1049 1050 1051
void memory_region_destroy(MemoryRegion *mr)
{
    assert(QTAILQ_EMPTY(&mr->subregions));
1052
    mr->destructor(mr);
A
Avi Kivity 已提交
1053
    memory_region_clear_coalescing(mr);
1054 1055
    g_free((char *)mr->name);
    g_free(mr->ioeventfds);
A
Avi Kivity 已提交
1056 1057 1058 1059
}

uint64_t memory_region_size(MemoryRegion *mr)
{
1060 1061 1062 1063
    if (int128_eq(mr->size, int128_2_64())) {
        return UINT64_MAX;
    }
    return int128_get64(mr->size);
A
Avi Kivity 已提交
1064 1065
}

1066 1067 1068 1069 1070
const char *memory_region_name(MemoryRegion *mr)
{
    return mr->name;
}

A
Avi Kivity 已提交
1071 1072 1073 1074 1075
bool memory_region_is_ram(MemoryRegion *mr)
{
    return mr->ram;
}

1076 1077 1078 1079 1080
bool memory_region_is_logging(MemoryRegion *mr)
{
    return mr->dirty_log_mask;
}

A
Avi Kivity 已提交
1081 1082 1083 1084 1085
bool memory_region_is_rom(MemoryRegion *mr)
{
    return mr->ram && mr->readonly;
}

A
Avi Kivity 已提交
1086 1087 1088 1089 1090 1091 1092
void memory_region_set_offset(MemoryRegion *mr, target_phys_addr_t offset)
{
    mr->offset = offset;
}

void memory_region_set_log(MemoryRegion *mr, bool log, unsigned client)
{
A
Avi Kivity 已提交
1093 1094 1095
    uint8_t mask = 1 << client;

    mr->dirty_log_mask = (mr->dirty_log_mask & ~mask) | (log * mask);
1096
    memory_region_update_topology(mr);
A
Avi Kivity 已提交
1097 1098 1099 1100 1101
}

bool memory_region_get_dirty(MemoryRegion *mr, target_phys_addr_t addr,
                             unsigned client)
{
1102
    assert(mr->terminates);
A
Avi Kivity 已提交
1103
    return cpu_physical_memory_get_dirty(mr->ram_addr + addr, 1 << client);
A
Avi Kivity 已提交
1104 1105 1106 1107
}

void memory_region_set_dirty(MemoryRegion *mr, target_phys_addr_t addr)
{
1108
    assert(mr->terminates);
A
Avi Kivity 已提交
1109
    return cpu_physical_memory_set_dirty(mr->ram_addr + addr);
A
Avi Kivity 已提交
1110 1111 1112 1113
}

void memory_region_sync_dirty_bitmap(MemoryRegion *mr)
{
A
Avi Kivity 已提交
1114 1115
    FlatRange *fr;

1116
    FOR_EACH_FLAT_RANGE(fr, &address_space_memory.current_map) {
A
Avi Kivity 已提交
1117
        if (fr->mr == mr) {
1118
            MEMORY_LISTENER_UPDATE_REGION(fr, &address_space_memory, log_sync);
A
Avi Kivity 已提交
1119 1120
        }
    }
A
Avi Kivity 已提交
1121 1122 1123 1124
}

void memory_region_set_readonly(MemoryRegion *mr, bool readonly)
{
1125 1126
    if (mr->readonly != readonly) {
        mr->readonly = readonly;
1127
        memory_region_update_topology(mr);
1128
    }
A
Avi Kivity 已提交
1129 1130
}

1131 1132 1133 1134
void memory_region_rom_device_set_readable(MemoryRegion *mr, bool readable)
{
    if (mr->readable != readable) {
        mr->readable = readable;
1135
        memory_region_update_topology(mr);
1136 1137 1138
    }
}

A
Avi Kivity 已提交
1139 1140 1141
void memory_region_reset_dirty(MemoryRegion *mr, target_phys_addr_t addr,
                               target_phys_addr_t size, unsigned client)
{
1142
    assert(mr->terminates);
A
Avi Kivity 已提交
1143 1144 1145
    cpu_physical_memory_reset_dirty(mr->ram_addr + addr,
                                    mr->ram_addr + addr + size,
                                    1 << client);
A
Avi Kivity 已提交
1146 1147 1148 1149 1150 1151 1152 1153
}

void *memory_region_get_ram_ptr(MemoryRegion *mr)
{
    if (mr->alias) {
        return memory_region_get_ram_ptr(mr->alias) + mr->alias_offset;
    }

1154
    assert(mr->terminates);
A
Avi Kivity 已提交
1155

1156
    return qemu_get_ram_ptr(mr->ram_addr & TARGET_PAGE_MASK);
A
Avi Kivity 已提交
1157 1158 1159 1160 1161 1162 1163 1164
}

static void memory_region_update_coalesced_range(MemoryRegion *mr)
{
    FlatRange *fr;
    CoalescedMemoryRange *cmr;
    AddrRange tmp;

1165
    FOR_EACH_FLAT_RANGE(fr, &address_space_memory.current_map) {
A
Avi Kivity 已提交
1166
        if (fr->mr == mr) {
1167 1168
            qemu_unregister_coalesced_mmio(int128_get64(fr->addr.start),
                                           int128_get64(fr->addr.size));
A
Avi Kivity 已提交
1169 1170
            QTAILQ_FOREACH(cmr, &mr->coalesced, link) {
                tmp = addrrange_shift(cmr->addr,
1171 1172
                                      int128_sub(fr->addr.start,
                                                 int128_make64(fr->offset_in_region)));
A
Avi Kivity 已提交
1173 1174 1175 1176
                if (!addrrange_intersects(tmp, fr->addr)) {
                    continue;
                }
                tmp = addrrange_intersection(tmp, fr->addr);
1177 1178
                qemu_register_coalesced_mmio(int128_get64(tmp.start),
                                             int128_get64(tmp.size));
A
Avi Kivity 已提交
1179 1180 1181 1182 1183 1184 1185 1186
            }
        }
    }
}

void memory_region_set_coalescing(MemoryRegion *mr)
{
    memory_region_clear_coalescing(mr);
1187
    memory_region_add_coalescing(mr, 0, int128_get64(mr->size));
A
Avi Kivity 已提交
1188 1189 1190 1191 1192 1193
}

void memory_region_add_coalescing(MemoryRegion *mr,
                                  target_phys_addr_t offset,
                                  uint64_t size)
{
1194
    CoalescedMemoryRange *cmr = g_malloc(sizeof(*cmr));
A
Avi Kivity 已提交
1195

1196
    cmr->addr = addrrange_make(int128_make64(offset), int128_make64(size));
A
Avi Kivity 已提交
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
    QTAILQ_INSERT_TAIL(&mr->coalesced, cmr, link);
    memory_region_update_coalesced_range(mr);
}

void memory_region_clear_coalescing(MemoryRegion *mr)
{
    CoalescedMemoryRange *cmr;

    while (!QTAILQ_EMPTY(&mr->coalesced)) {
        cmr = QTAILQ_FIRST(&mr->coalesced);
        QTAILQ_REMOVE(&mr->coalesced, cmr, link);
1208
        g_free(cmr);
A
Avi Kivity 已提交
1209 1210 1211 1212
    }
    memory_region_update_coalesced_range(mr);
}

A
Avi Kivity 已提交
1213 1214 1215 1216 1217 1218 1219 1220
void memory_region_add_eventfd(MemoryRegion *mr,
                               target_phys_addr_t addr,
                               unsigned size,
                               bool match_data,
                               uint64_t data,
                               int fd)
{
    MemoryRegionIoeventfd mrfd = {
1221 1222
        .addr.start = int128_make64(addr),
        .addr.size = int128_make64(size),
A
Avi Kivity 已提交
1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234
        .match_data = match_data,
        .data = data,
        .fd = fd,
    };
    unsigned i;

    for (i = 0; i < mr->ioeventfd_nb; ++i) {
        if (memory_region_ioeventfd_before(mrfd, mr->ioeventfds[i])) {
            break;
        }
    }
    ++mr->ioeventfd_nb;
1235
    mr->ioeventfds = g_realloc(mr->ioeventfds,
A
Avi Kivity 已提交
1236 1237 1238 1239
                                  sizeof(*mr->ioeventfds) * mr->ioeventfd_nb);
    memmove(&mr->ioeventfds[i+1], &mr->ioeventfds[i],
            sizeof(*mr->ioeventfds) * (mr->ioeventfd_nb-1 - i));
    mr->ioeventfds[i] = mrfd;
1240
    memory_region_update_topology(mr);
A
Avi Kivity 已提交
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
}

void memory_region_del_eventfd(MemoryRegion *mr,
                               target_phys_addr_t addr,
                               unsigned size,
                               bool match_data,
                               uint64_t data,
                               int fd)
{
    MemoryRegionIoeventfd mrfd = {
1251 1252
        .addr.start = int128_make64(addr),
        .addr.size = int128_make64(size),
A
Avi Kivity 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267
        .match_data = match_data,
        .data = data,
        .fd = fd,
    };
    unsigned i;

    for (i = 0; i < mr->ioeventfd_nb; ++i) {
        if (memory_region_ioeventfd_equal(mrfd, mr->ioeventfds[i])) {
            break;
        }
    }
    assert(i != mr->ioeventfd_nb);
    memmove(&mr->ioeventfds[i], &mr->ioeventfds[i+1],
            sizeof(*mr->ioeventfds) * (mr->ioeventfd_nb - (i+1)));
    --mr->ioeventfd_nb;
1268
    mr->ioeventfds = g_realloc(mr->ioeventfds,
A
Avi Kivity 已提交
1269
                                  sizeof(*mr->ioeventfds)*mr->ioeventfd_nb + 1);
1270
    memory_region_update_topology(mr);
A
Avi Kivity 已提交
1271 1272
}

A
Avi Kivity 已提交
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
static void memory_region_add_subregion_common(MemoryRegion *mr,
                                               target_phys_addr_t offset,
                                               MemoryRegion *subregion)
{
    MemoryRegion *other;

    assert(!subregion->parent);
    subregion->parent = mr;
    subregion->addr = offset;
    QTAILQ_FOREACH(other, &mr->subregions, subregions_link) {
        if (subregion->may_overlap || other->may_overlap) {
            continue;
        }
1286 1287 1288 1289
        if (int128_gt(int128_make64(offset),
                      int128_add(int128_make64(other->addr), other->size))
            || int128_le(int128_add(int128_make64(offset), subregion->size),
                         int128_make64(other->addr))) {
A
Avi Kivity 已提交
1290 1291
            continue;
        }
1292
#if 0
1293 1294
        printf("warning: subregion collision %llx/%llx (%s) "
               "vs %llx/%llx (%s)\n",
A
Avi Kivity 已提交
1295
               (unsigned long long)offset,
1296
               (unsigned long long)int128_get64(subregion->size),
1297 1298
               subregion->name,
               (unsigned long long)other->addr,
1299
               (unsigned long long)int128_get64(other->size),
1300
               other->name);
1301
#endif
A
Avi Kivity 已提交
1302 1303 1304 1305 1306 1307 1308 1309 1310
    }
    QTAILQ_FOREACH(other, &mr->subregions, subregions_link) {
        if (subregion->priority >= other->priority) {
            QTAILQ_INSERT_BEFORE(other, subregion, subregions_link);
            goto done;
        }
    }
    QTAILQ_INSERT_TAIL(&mr->subregions, subregion, subregions_link);
done:
1311
    memory_region_update_topology(mr);
A
Avi Kivity 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339
}


void memory_region_add_subregion(MemoryRegion *mr,
                                 target_phys_addr_t offset,
                                 MemoryRegion *subregion)
{
    subregion->may_overlap = false;
    subregion->priority = 0;
    memory_region_add_subregion_common(mr, offset, subregion);
}

void memory_region_add_subregion_overlap(MemoryRegion *mr,
                                         target_phys_addr_t offset,
                                         MemoryRegion *subregion,
                                         unsigned priority)
{
    subregion->may_overlap = true;
    subregion->priority = priority;
    memory_region_add_subregion_common(mr, offset, subregion);
}

void memory_region_del_subregion(MemoryRegion *mr,
                                 MemoryRegion *subregion)
{
    assert(subregion->parent == mr);
    subregion->parent = NULL;
    QTAILQ_REMOVE(&mr->subregions, subregion, subregions_link);
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349
    memory_region_update_topology(mr);
}

void memory_region_set_enabled(MemoryRegion *mr, bool enabled)
{
    if (enabled == mr->enabled) {
        return;
    }
    mr->enabled = enabled;
    memory_region_update_topology(NULL);
A
Avi Kivity 已提交
1350
}
A
Avi Kivity 已提交
1351

1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372
void memory_region_set_address(MemoryRegion *mr, target_phys_addr_t addr)
{
    MemoryRegion *parent = mr->parent;
    unsigned priority = mr->priority;
    bool may_overlap = mr->may_overlap;

    if (addr == mr->addr || !parent) {
        mr->addr = addr;
        return;
    }

    memory_region_transaction_begin();
    memory_region_del_subregion(parent, mr);
    if (may_overlap) {
        memory_region_add_subregion_overlap(parent, addr, mr, priority);
    } else {
        memory_region_add_subregion(parent, addr, mr);
    }
    memory_region_transaction_commit();
}

1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386
void memory_region_set_alias_offset(MemoryRegion *mr, target_phys_addr_t offset)
{
    target_phys_addr_t old_offset = mr->alias_offset;

    assert(mr->alias);
    mr->alias_offset = offset;

    if (offset == old_offset || !mr->parent) {
        return;
    }

    memory_region_update_topology(mr);
}

1387 1388 1389 1390 1391
ram_addr_t memory_region_get_ram_addr(MemoryRegion *mr)
{
    return mr->ram_addr;
}

1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438
static int cmp_flatrange_addr(const void *addr_, const void *fr_)
{
    const AddrRange *addr = addr_;
    const FlatRange *fr = fr_;

    if (int128_le(addrrange_end(*addr), fr->addr.start)) {
        return -1;
    } else if (int128_ge(addr->start, addrrange_end(fr->addr))) {
        return 1;
    }
    return 0;
}

static FlatRange *address_space_lookup(AddressSpace *as, AddrRange addr)
{
    return bsearch(&addr, as->current_map.ranges, as->current_map.nr,
                   sizeof(FlatRange), cmp_flatrange_addr);
}

MemoryRegionSection memory_region_find(MemoryRegion *address_space,
                                       target_phys_addr_t addr, uint64_t size)
{
    AddressSpace *as = memory_region_to_address_space(address_space);
    AddrRange range = addrrange_make(int128_make64(addr),
                                     int128_make64(size));
    FlatRange *fr = address_space_lookup(as, range);
    MemoryRegionSection ret = { .mr = NULL, .size = 0 };

    if (!fr) {
        return ret;
    }

    while (fr > as->current_map.ranges
           && addrrange_intersects(fr[-1].addr, range)) {
        --fr;
    }

    ret.mr = fr->mr;
    range = addrrange_intersection(range, fr->addr);
    ret.offset_within_region = fr->offset_in_region;
    ret.offset_within_region += int128_get64(int128_sub(range.start,
                                                        fr->addr.start));
    ret.size = int128_get64(range.size);
    ret.offset_within_address_space = int128_get64(range.start);
    return ret;
}

1439 1440
void memory_global_sync_dirty_bitmap(MemoryRegion *address_space)
{
1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452
    AddressSpace *as = memory_region_to_address_space(address_space);
    FlatRange *fr;

    FOR_EACH_FLAT_RANGE(fr, &as->current_map) {
        MEMORY_LISTENER_UPDATE_REGION(fr, as, log_sync);
    }
}

void memory_global_dirty_log_start(void)
{
    MemoryListener *listener;

1453
    cpu_physical_memory_set_dirty_tracking(1);
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467
    global_dirty_log = true;
    QLIST_FOREACH(listener, &memory_listeners, link) {
        listener->log_global_start(listener);
    }
}

void memory_global_dirty_log_stop(void)
{
    MemoryListener *listener;

    global_dirty_log = false;
    QLIST_FOREACH(listener, &memory_listeners, link) {
        listener->log_global_stop(listener);
    }
1468
    cpu_physical_memory_set_dirty_tracking(0);
1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500
}

static void listener_add_address_space(MemoryListener *listener,
                                       AddressSpace *as)
{
    FlatRange *fr;

    if (global_dirty_log) {
        listener->log_global_start(listener);
    }
    FOR_EACH_FLAT_RANGE(fr, &as->current_map) {
        MemoryRegionSection section = {
            .mr = fr->mr,
            .address_space = as->root,
            .offset_within_region = fr->offset_in_region,
            .size = int128_get64(fr->addr.size),
            .offset_within_address_space = int128_get64(fr->addr.start),
        };
        listener->region_add(listener, &section);
    }
}

void memory_listener_register(MemoryListener *listener)
{
    QLIST_INSERT_HEAD(&memory_listeners, listener, link);
    listener_add_address_space(listener, &address_space_memory);
    listener_add_address_space(listener, &address_space_io);
}

void memory_listener_unregister(MemoryListener *listener)
{
    QLIST_REMOVE(listener, link);
1501
}
1502

A
Avi Kivity 已提交
1503 1504
void set_system_memory_map(MemoryRegion *mr)
{
1505
    address_space_memory.root = mr;
1506
    memory_region_update_topology(NULL);
A
Avi Kivity 已提交
1507
}
A
Avi Kivity 已提交
1508 1509 1510 1511

void set_system_io_map(MemoryRegion *mr)
{
    address_space_io.root = mr;
1512
    memory_region_update_topology(NULL);
A
Avi Kivity 已提交
1513
}
B
Blue Swirl 已提交
1514

1515 1516
uint64_t io_mem_read(int io_index, target_phys_addr_t addr, unsigned size)
{
1517
    return memory_region_dispatch_read(io_mem_region[io_index], addr, size);
1518 1519 1520 1521 1522
}

void io_mem_write(int io_index, target_phys_addr_t addr,
                  uint64_t val, unsigned size)
{
1523
    memory_region_dispatch_write(io_mem_region[io_index], addr, val, size);
1524 1525
}

B
Blue Swirl 已提交
1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538
typedef struct MemoryRegionList MemoryRegionList;

struct MemoryRegionList {
    const MemoryRegion *mr;
    bool printed;
    QTAILQ_ENTRY(MemoryRegionList) queue;
};

typedef QTAILQ_HEAD(queue, MemoryRegionList) MemoryRegionListHead;

static void mtree_print_mr(fprintf_function mon_printf, void *f,
                           const MemoryRegion *mr, unsigned int level,
                           target_phys_addr_t base,
1539
                           MemoryRegionListHead *alias_print_queue)
B
Blue Swirl 已提交
1540
{
1541 1542
    MemoryRegionList *new_ml, *ml, *next_ml;
    MemoryRegionListHead submr_print_queue;
B
Blue Swirl 已提交
1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558
    const MemoryRegion *submr;
    unsigned int i;

    if (!mr) {
        return;
    }

    for (i = 0; i < level; i++) {
        mon_printf(f, "  ");
    }

    if (mr->alias) {
        MemoryRegionList *ml;
        bool found = false;

        /* check if the alias is already in the queue */
1559
        QTAILQ_FOREACH(ml, alias_print_queue, queue) {
B
Blue Swirl 已提交
1560 1561 1562 1563 1564 1565 1566 1567 1568
            if (ml->mr == mr->alias && !ml->printed) {
                found = true;
            }
        }

        if (!found) {
            ml = g_new(MemoryRegionList, 1);
            ml->mr = mr->alias;
            ml->printed = false;
1569
            QTAILQ_INSERT_TAIL(alias_print_queue, ml, queue);
B
Blue Swirl 已提交
1570
        }
J
Jan Kiszka 已提交
1571
        mon_printf(f, TARGET_FMT_plx "-" TARGET_FMT_plx " (prio %d): alias %s @%s "
B
Blue Swirl 已提交
1572 1573
                   TARGET_FMT_plx "-" TARGET_FMT_plx "\n",
                   base + mr->addr,
1574 1575
                   base + mr->addr
                   + (target_phys_addr_t)int128_get64(mr->size) - 1,
J
Jan Kiszka 已提交
1576
                   mr->priority,
B
Blue Swirl 已提交
1577 1578 1579
                   mr->name,
                   mr->alias->name,
                   mr->alias_offset,
1580 1581
                   mr->alias_offset
                   + (target_phys_addr_t)int128_get64(mr->size) - 1);
B
Blue Swirl 已提交
1582
    } else {
J
Jan Kiszka 已提交
1583
        mon_printf(f, TARGET_FMT_plx "-" TARGET_FMT_plx " (prio %d): %s\n",
B
Blue Swirl 已提交
1584
                   base + mr->addr,
1585 1586
                   base + mr->addr
                   + (target_phys_addr_t)int128_get64(mr->size) - 1,
J
Jan Kiszka 已提交
1587
                   mr->priority,
B
Blue Swirl 已提交
1588 1589
                   mr->name);
    }
1590 1591 1592

    QTAILQ_INIT(&submr_print_queue);

B
Blue Swirl 已提交
1593
    QTAILQ_FOREACH(submr, &mr->subregions, subregions_link) {
1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614
        new_ml = g_new(MemoryRegionList, 1);
        new_ml->mr = submr;
        QTAILQ_FOREACH(ml, &submr_print_queue, queue) {
            if (new_ml->mr->addr < ml->mr->addr ||
                (new_ml->mr->addr == ml->mr->addr &&
                 new_ml->mr->priority > ml->mr->priority)) {
                QTAILQ_INSERT_BEFORE(ml, new_ml, queue);
                new_ml = NULL;
                break;
            }
        }
        if (new_ml) {
            QTAILQ_INSERT_TAIL(&submr_print_queue, new_ml, queue);
        }
    }

    QTAILQ_FOREACH(ml, &submr_print_queue, queue) {
        mtree_print_mr(mon_printf, f, ml->mr, level + 1, base + mr->addr,
                       alias_print_queue);
    }

A
Avi Kivity 已提交
1615
    QTAILQ_FOREACH_SAFE(ml, &submr_print_queue, queue, next_ml) {
1616
        g_free(ml);
B
Blue Swirl 已提交
1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638
    }
}

void mtree_info(fprintf_function mon_printf, void *f)
{
    MemoryRegionListHead ml_head;
    MemoryRegionList *ml, *ml2;

    QTAILQ_INIT(&ml_head);

    mon_printf(f, "memory\n");
    mtree_print_mr(mon_printf, f, address_space_memory.root, 0, 0, &ml_head);

    /* print aliased regions */
    QTAILQ_FOREACH(ml, &ml_head, queue) {
        if (!ml->printed) {
            mon_printf(f, "%s\n", ml->mr->name);
            mtree_print_mr(mon_printf, f, ml->mr, 0, 0, &ml_head);
        }
    }

    QTAILQ_FOREACH_SAFE(ml, &ml_head, queue, ml2) {
A
Avi Kivity 已提交
1639
        g_free(ml);
B
Blue Swirl 已提交
1640 1641
    }

J
Jan Kiszka 已提交
1642 1643 1644 1645 1646 1647
    if (address_space_io.root &&
        !QTAILQ_EMPTY(&address_space_io.root->subregions)) {
        QTAILQ_INIT(&ml_head);
        mon_printf(f, "I/O\n");
        mtree_print_mr(mon_printf, f, address_space_io.root, 0, 0, &ml_head);
    }
B
Blue Swirl 已提交
1648
}