arch_init.c 34 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
/*
 * QEMU System Emulator
 *
 * Copyright (c) 2003-2008 Fabrice Bellard
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */
#include <stdint.h>
#include <stdarg.h>
26
#include <stdlib.h>
27
#ifndef _WIN32
28
#include <sys/types.h>
29 30 31
#include <sys/mman.h>
#endif
#include "config.h"
32
#include "monitor/monitor.h"
33
#include "sysemu/sysemu.h"
34 35
#include "qemu/bitops.h"
#include "qemu/bitmap.h"
36
#include "sysemu/arch_init.h"
37
#include "audio/audio.h"
Paolo Bonzini's avatar
Paolo Bonzini committed
38
#include "hw/i386/pc.h"
39
#include "hw/pci/pci.h"
Paolo Bonzini's avatar
Paolo Bonzini committed
40
#include "hw/audio/audio.h"
41
#include "sysemu/kvm.h"
42
#include "migration/migration.h"
Paolo Bonzini's avatar
Paolo Bonzini committed
43
#include "hw/i386/smbios.h"
44
#include "exec/address-spaces.h"
Paolo Bonzini's avatar
Paolo Bonzini committed
45
#include "hw/audio/pcspk.h"
46
#include "migration/page_cache.h"
47
#include "qemu/config-file.h"
48
#include "qmp-commands.h"
49
#include "trace.h"
50
#include "exec/cpu-all.h"
51
#include "exec/ram_addr.h"
52
#include "hw/acpi/acpi.h"
53
#include "qemu/host-utils.h"
54

Orit Wasserman's avatar
Orit Wasserman committed
55 56 57 58 59 60 61 62
#ifdef DEBUG_ARCH_INIT
#define DPRINTF(fmt, ...) \
    do { fprintf(stdout, "arch_init: " fmt, ## __VA_ARGS__); } while (0)
#else
#define DPRINTF(fmt, ...) \
    do { } while (0)
#endif

63 64 65 66 67 68 69
#ifdef TARGET_SPARC
int graphic_width = 1024;
int graphic_height = 768;
int graphic_depth = 8;
#else
int graphic_width = 800;
int graphic_height = 600;
70
int graphic_depth = 32;
71 72 73 74 75 76 77 78 79 80 81 82 83
#endif


#if defined(TARGET_ALPHA)
#define QEMU_ARCH QEMU_ARCH_ALPHA
#elif defined(TARGET_ARM)
#define QEMU_ARCH QEMU_ARCH_ARM
#elif defined(TARGET_CRIS)
#define QEMU_ARCH QEMU_ARCH_CRIS
#elif defined(TARGET_I386)
#define QEMU_ARCH QEMU_ARCH_I386
#elif defined(TARGET_M68K)
#define QEMU_ARCH QEMU_ARCH_M68K
Michael Walle's avatar
Michael Walle committed
84 85
#elif defined(TARGET_LM32)
#define QEMU_ARCH QEMU_ARCH_LM32
86 87 88 89
#elif defined(TARGET_MICROBLAZE)
#define QEMU_ARCH QEMU_ARCH_MICROBLAZE
#elif defined(TARGET_MIPS)
#define QEMU_ARCH QEMU_ARCH_MIPS
90 91
#elif defined(TARGET_MOXIE)
#define QEMU_ARCH QEMU_ARCH_MOXIE
92 93
#elif defined(TARGET_OPENRISC)
#define QEMU_ARCH QEMU_ARCH_OPENRISC
94 95 96 97 98 99 100 101
#elif defined(TARGET_PPC)
#define QEMU_ARCH QEMU_ARCH_PPC
#elif defined(TARGET_S390X)
#define QEMU_ARCH QEMU_ARCH_S390X
#elif defined(TARGET_SH4)
#define QEMU_ARCH QEMU_ARCH_SH4
#elif defined(TARGET_SPARC)
#define QEMU_ARCH QEMU_ARCH_SPARC
Max Filippov's avatar
Max Filippov committed
102 103
#elif defined(TARGET_XTENSA)
#define QEMU_ARCH QEMU_ARCH_XTENSA
104 105
#elif defined(TARGET_UNICORE32)
#define QEMU_ARCH QEMU_ARCH_UNICORE32
106 107 108
#endif

const uint32_t arch_type = QEMU_ARCH;
109 110 111
static bool mig_throttle_on;
static int dirty_rate_high_cnt;
static void check_guest_throttling(void);
112 113 114 115

/***********************************************************/
/* ram save/restore */

116 117 118 119 120 121
#define RAM_SAVE_FLAG_FULL     0x01 /* Obsolete, not used anymore */
#define RAM_SAVE_FLAG_COMPRESS 0x02
#define RAM_SAVE_FLAG_MEM_SIZE 0x04
#define RAM_SAVE_FLAG_PAGE     0x08
#define RAM_SAVE_FLAG_EOS      0x10
#define RAM_SAVE_FLAG_CONTINUE 0x20
122
#define RAM_SAVE_FLAG_XBZRLE   0x40
Michael R. Hines's avatar
Michael R. Hines committed
123
/* 0x80 is reserved in migration.h start with 0x100 next */
124

125

126 127
static struct defconfig_file {
    const char *filename;
128 129
    /* Indicates it is an user config file (disabled by -no-user-config) */
    bool userconfig;
130
} default_config_files[] = {
131
    { CONFIG_QEMU_CONFDIR "/qemu.conf",                   true },
132
    { CONFIG_QEMU_CONFDIR "/target-" TARGET_NAME ".conf", true },
133 134 135 136
    { NULL }, /* end of list */
};


137
int qemu_read_default_config_files(bool userconfig)
138 139
{
    int ret;
140
    struct defconfig_file *f;
141

142
    for (f = default_config_files; f->filename; f++) {
143 144 145
        if (!userconfig && f->userconfig) {
            continue;
        }
146 147 148 149
        ret = qemu_read_config_file(f->filename);
        if (ret < 0 && ret != -ENOENT) {
            return ret;
        }
150
    }
Laszlo Ersek's avatar
Laszlo Ersek committed
151

152 153 154
    return 0;
}

155
static inline bool is_zero_range(uint8_t *p, uint64_t size)
156
{
157
    return buffer_find_nonzero_offset(p, size) == size;
158 159
}

160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
/* struct contains XBZRLE cache and a static page
   used by the compression */
static struct {
    /* buffer used for XBZRLE encoding */
    uint8_t *encoded_buf;
    /* buffer for storing page content */
    uint8_t *current_buf;
    /* buffer used for XBZRLE decoding */
    uint8_t *decoded_buf;
    /* Cache for XBZRLE */
    PageCache *cache;
} XBZRLE = {
    .encoded_buf = NULL,
    .current_buf = NULL,
    .decoded_buf = NULL,
    .cache = NULL,
};

178 179 180

int64_t xbzrle_cache_resize(int64_t new_size)
{
181 182 183 184
    if (new_size < TARGET_PAGE_SIZE) {
        return -1;
    }

185 186 187 188 189 190 191
    if (XBZRLE.cache != NULL) {
        return cache_resize(XBZRLE.cache, new_size / TARGET_PAGE_SIZE) *
            TARGET_PAGE_SIZE;
    }
    return pow2floor(new_size);
}

192 193 194
/* accounting for migration statistics */
typedef struct AccountingInfo {
    uint64_t dup_pages;
195
    uint64_t skipped_pages;
196 197
    uint64_t norm_pages;
    uint64_t iterations;
Orit Wasserman's avatar
Orit Wasserman committed
198 199 200 201
    uint64_t xbzrle_bytes;
    uint64_t xbzrle_pages;
    uint64_t xbzrle_cache_miss;
    uint64_t xbzrle_overflows;
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
} AccountingInfo;

static AccountingInfo acct_info;

static void acct_clear(void)
{
    memset(&acct_info, 0, sizeof(acct_info));
}

uint64_t dup_mig_bytes_transferred(void)
{
    return acct_info.dup_pages * TARGET_PAGE_SIZE;
}

uint64_t dup_mig_pages_transferred(void)
{
    return acct_info.dup_pages;
}

221 222 223 224 225 226 227 228 229 230
uint64_t skipped_mig_bytes_transferred(void)
{
    return acct_info.skipped_pages * TARGET_PAGE_SIZE;
}

uint64_t skipped_mig_pages_transferred(void)
{
    return acct_info.skipped_pages;
}

231 232 233 234 235 236 237 238 239 240
uint64_t norm_mig_bytes_transferred(void)
{
    return acct_info.norm_pages * TARGET_PAGE_SIZE;
}

uint64_t norm_mig_pages_transferred(void)
{
    return acct_info.norm_pages;
}

Orit Wasserman's avatar
Orit Wasserman committed
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260
uint64_t xbzrle_mig_bytes_transferred(void)
{
    return acct_info.xbzrle_bytes;
}

uint64_t xbzrle_mig_pages_transferred(void)
{
    return acct_info.xbzrle_pages;
}

uint64_t xbzrle_mig_pages_cache_miss(void)
{
    return acct_info.xbzrle_cache_miss;
}

uint64_t xbzrle_mig_pages_overflow(void)
{
    return acct_info.xbzrle_overflows;
}

261 262
static size_t save_block_hdr(QEMUFile *f, RAMBlock *block, ram_addr_t offset,
                             int cont, int flag)
Orit Wasserman's avatar
Orit Wasserman committed
263
{
264 265 266 267
    size_t size;

    qemu_put_be64(f, offset | cont | flag);
    size = 8;
Orit Wasserman's avatar
Orit Wasserman committed
268

269 270 271 272 273 274 275
    if (!cont) {
        qemu_put_byte(f, strlen(block->idstr));
        qemu_put_buffer(f, (uint8_t *)block->idstr,
                        strlen(block->idstr));
        size += 1 + strlen(block->idstr);
    }
    return size;
Orit Wasserman's avatar
Orit Wasserman committed
276 277
}

278 279 280 281
#define ENCODING_FLAG_XBZRLE 0x1

static int save_xbzrle_page(QEMUFile *f, uint8_t *current_data,
                            ram_addr_t current_addr, RAMBlock *block,
282
                            ram_addr_t offset, int cont, bool last_stage)
283 284 285 286 287
{
    int encoded_len = 0, bytes_sent = -1;
    uint8_t *prev_cached_page;

    if (!cache_is_cached(XBZRLE.cache, current_addr)) {
288
        if (!last_stage) {
289
            cache_insert(XBZRLE.cache, current_addr, current_data);
290
        }
Orit Wasserman's avatar
Orit Wasserman committed
291
        acct_info.xbzrle_cache_miss++;
292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
        return -1;
    }

    prev_cached_page = get_cached_data(XBZRLE.cache, current_addr);

    /* save current buffer into memory */
    memcpy(XBZRLE.current_buf, current_data, TARGET_PAGE_SIZE);

    /* XBZRLE encoding (if there is no overflow) */
    encoded_len = xbzrle_encode_buffer(prev_cached_page, XBZRLE.current_buf,
                                       TARGET_PAGE_SIZE, XBZRLE.encoded_buf,
                                       TARGET_PAGE_SIZE);
    if (encoded_len == 0) {
        DPRINTF("Skipping unmodified page\n");
        return 0;
    } else if (encoded_len == -1) {
        DPRINTF("Overflow\n");
Orit Wasserman's avatar
Orit Wasserman committed
309
        acct_info.xbzrle_overflows++;
310 311 312 313 314 315
        /* update data in the cache */
        memcpy(prev_cached_page, current_data, TARGET_PAGE_SIZE);
        return -1;
    }

    /* we need to update the data in the cache, in order to get the same data */
316 317 318
    if (!last_stage) {
        memcpy(prev_cached_page, XBZRLE.current_buf, TARGET_PAGE_SIZE);
    }
319 320

    /* Send XBZRLE based compressed page */
321
    bytes_sent = save_block_hdr(f, block, offset, cont, RAM_SAVE_FLAG_XBZRLE);
322 323 324
    qemu_put_byte(f, ENCODING_FLAG_XBZRLE);
    qemu_put_be16(f, encoded_len);
    qemu_put_buffer(f, XBZRLE.encoded_buf, encoded_len);
325
    bytes_sent += encoded_len + 1 + 2;
Orit Wasserman's avatar
Orit Wasserman committed
326 327
    acct_info.xbzrle_pages++;
    acct_info.xbzrle_bytes += bytes_sent;
328 329 330 331

    return bytes_sent;
}

332 333 334 335

/* This is the last block that we have visited serching for dirty pages
 */
static RAMBlock *last_seen_block;
Juan Quintela's avatar
Juan Quintela committed
336 337
/* This is the last block from where we have sent data */
static RAMBlock *last_sent_block;
338
static ram_addr_t last_offset;
Juan Quintela's avatar
Juan Quintela committed
339 340
static unsigned long *migration_bitmap;
static uint64_t migration_dirty_pages;
341
static uint32_t last_version;
342
static bool ram_bulk_stage;
343

344 345 346
static inline
ram_addr_t migration_bitmap_find_and_reset_dirty(MemoryRegion *mr,
                                                 ram_addr_t start)
347
{
348 349
    unsigned long base = mr->ram_addr >> TARGET_PAGE_BITS;
    unsigned long nr = base + (start >> TARGET_PAGE_BITS);
350 351
    uint64_t mr_size = TARGET_PAGE_ALIGN(memory_region_size(mr));
    unsigned long size = base + (mr_size >> TARGET_PAGE_BITS);
Juan Quintela's avatar
Juan Quintela committed
352

353 354 355 356 357 358 359
    unsigned long next;

    if (ram_bulk_stage && nr > base) {
        next = nr + 1;
    } else {
        next = find_next_bit(migration_bitmap, size, nr);
    }
360

361 362
    if (next < size) {
        clear_bit(next, migration_bitmap);
Juan Quintela's avatar
Juan Quintela committed
363
        migration_dirty_pages--;
364
    }
365
    return (next - base) << TARGET_PAGE_BITS;
366 367
}

368
static inline bool migration_bitmap_set_dirty(ram_addr_t addr)
369
{
Juan Quintela's avatar
Juan Quintela committed
370
    bool ret;
371
    int nr = addr >> TARGET_PAGE_BITS;
372

Juan Quintela's avatar
Juan Quintela committed
373 374 375 376
    ret = test_and_set_bit(nr, migration_bitmap);

    if (!ret) {
        migration_dirty_pages++;
377
    }
Juan Quintela's avatar
Juan Quintela committed
378
    return ret;
379 380
}

381 382 383
static void migration_bitmap_sync_range(ram_addr_t start, ram_addr_t length)
{
    ram_addr_t addr;
384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
    unsigned long page = BIT_WORD(start >> TARGET_PAGE_BITS);

    /* start address is aligned at the start of a word? */
    if (((page * BITS_PER_LONG) << TARGET_PAGE_BITS) == start) {
        int k;
        int nr = BITS_TO_LONGS(length >> TARGET_PAGE_BITS);
        unsigned long *src = ram_list.dirty_memory[DIRTY_MEMORY_MIGRATION];

        for (k = page; k < page + nr; k++) {
            if (src[k]) {
                unsigned long new_dirty;
                new_dirty = ~migration_bitmap[k];
                migration_bitmap[k] |= src[k];
                new_dirty &= src[k];
                migration_dirty_pages += ctpopl(new_dirty);
                src[k] = 0;
            }
        }
    } else {
        for (addr = 0; addr < length; addr += TARGET_PAGE_SIZE) {
            if (cpu_physical_memory_get_dirty(start + addr,
                                              TARGET_PAGE_SIZE,
                                              DIRTY_MEMORY_MIGRATION)) {
                cpu_physical_memory_reset_dirty(start + addr,
                                                TARGET_PAGE_SIZE,
                                                DIRTY_MEMORY_MIGRATION);
                migration_bitmap_set_dirty(start + addr);
            }
412 413 414 415 416
        }
    }
}


417 418
/* Needs iothread lock! */

419 420
static void migration_bitmap_sync(void)
{
Juan Quintela's avatar
Juan Quintela committed
421 422
    RAMBlock *block;
    uint64_t num_dirty_pages_init = migration_dirty_pages;
423 424
    MigrationState *s = migrate_get_current();
    static int64_t start_time;
425
    static int64_t bytes_xfer_prev;
426 427
    static int64_t num_dirty_pages_period;
    int64_t end_time;
428 429 430 431 432
    int64_t bytes_xfer_now;

    if (!bytes_xfer_prev) {
        bytes_xfer_prev = ram_bytes_transferred();
    }
433 434

    if (!start_time) {
435
        start_time = qemu_clock_get_ms(QEMU_CLOCK_REALTIME);
436
    }
437 438

    trace_migration_bitmap_sync_start();
439
    address_space_sync_dirty_bitmap(&address_space_memory);
Juan Quintela's avatar
Juan Quintela committed
440

441
    QTAILQ_FOREACH(block, &ram_list.blocks, next) {
442
        migration_bitmap_sync_range(block->mr->ram_addr, block->length);
Juan Quintela's avatar
Juan Quintela committed
443 444
    }
    trace_migration_bitmap_sync_end(migration_dirty_pages
445
                                    - num_dirty_pages_init);
446
    num_dirty_pages_period += migration_dirty_pages - num_dirty_pages_init;
447
    end_time = qemu_clock_get_ms(QEMU_CLOCK_REALTIME);
448 449 450

    /* more than 1 second = 1000 millisecons */
    if (end_time > start_time + 1000) {
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
        if (migrate_auto_converge()) {
            /* The following detection logic can be refined later. For now:
               Check to see if the dirtied bytes is 50% more than the approx.
               amount of bytes that just got transferred since the last time we
               were in this routine. If that happens >N times (for now N==4)
               we turn on the throttle down logic */
            bytes_xfer_now = ram_bytes_transferred();
            if (s->dirty_pages_rate &&
               (num_dirty_pages_period * TARGET_PAGE_SIZE >
                   (bytes_xfer_now - bytes_xfer_prev)/2) &&
               (dirty_rate_high_cnt++ > 4)) {
                    trace_migration_throttle();
                    mig_throttle_on = true;
                    dirty_rate_high_cnt = 0;
             }
             bytes_xfer_prev = bytes_xfer_now;
        } else {
             mig_throttle_on = false;
        }
470 471
        s->dirty_pages_rate = num_dirty_pages_period * 1000
            / (end_time - start_time);
472
        s->dirty_bytes_rate = s->dirty_pages_rate * TARGET_PAGE_SIZE;
473 474 475
        start_time = end_time;
        num_dirty_pages_period = 0;
    }
476 477
}

478 479 480
/*
 * ram_save_block: Writes a page of memory to the stream f
 *
481 482
 * Returns:  The number of bytes written.
 *           0 means no dirty pages
483 484
 */

485
static int ram_save_block(QEMUFile *f, bool last_stage)
486
{
487
    RAMBlock *block = last_seen_block;
488
    ram_addr_t offset = last_offset;
489
    bool complete_round = false;
490
    int bytes_sent = 0;
Avi Kivity's avatar
Avi Kivity committed
491
    MemoryRegion *mr;
492
    ram_addr_t current_addr;
493

494
    if (!block)
495
        block = QTAILQ_FIRST(&ram_list.blocks);
496

497
    while (true) {
Avi Kivity's avatar
Avi Kivity committed
498
        mr = block->mr;
499 500 501 502 503 504 505 506 507 508 509
        offset = migration_bitmap_find_and_reset_dirty(mr, offset);
        if (complete_round && block == last_seen_block &&
            offset >= last_offset) {
            break;
        }
        if (offset >= block->length) {
            offset = 0;
            block = QTAILQ_NEXT(block, next);
            if (!block) {
                block = QTAILQ_FIRST(&ram_list.blocks);
                complete_round = true;
510
                ram_bulk_stage = false;
511 512
            }
        } else {
Michael R. Hines's avatar
Michael R. Hines committed
513
            int ret;
514
            uint8_t *p;
Juan Quintela's avatar
Juan Quintela committed
515
            int cont = (block == last_sent_block) ?
516
                RAM_SAVE_FLAG_CONTINUE : 0;
517

Avi Kivity's avatar
Avi Kivity committed
518
            p = memory_region_get_ram_ptr(mr) + offset;
519

520 521
            /* In doubt sent page as normal */
            bytes_sent = -1;
Michael R. Hines's avatar
Michael R. Hines committed
522 523 524 525 526 527 528 529 530 531 532
            ret = ram_control_save_page(f, block->offset,
                               offset, TARGET_PAGE_SIZE, &bytes_sent);

            if (ret != RAM_SAVE_CONTROL_NOT_SUPP) {
                if (ret != RAM_SAVE_CONTROL_DELAYED) {
                    if (bytes_sent > 0) {
                        acct_info.norm_pages++;
                    } else if (bytes_sent == 0) {
                        acct_info.dup_pages++;
                    }
                }
533
            } else if (is_zero_range(p, TARGET_PAGE_SIZE)) {
534
                acct_info.dup_pages++;
535 536 537 538
                bytes_sent = save_block_hdr(f, block, offset, cont,
                                            RAM_SAVE_FLAG_COMPRESS);
                qemu_put_byte(f, 0);
                bytes_sent++;
539
            } else if (!ram_bulk_stage && migrate_use_xbzrle()) {
540 541
                current_addr = block->offset + offset;
                bytes_sent = save_xbzrle_page(f, p, current_addr, block,
542 543 544 545
                                              offset, cont, last_stage);
                if (!last_stage) {
                    p = get_cached_data(XBZRLE.cache, current_addr);
                }
546 547
            }

548
            /* XBZRLE overflow or normal page */
549
            if (bytes_sent == -1) {
550
                bytes_sent = save_block_hdr(f, block, offset, cont, RAM_SAVE_FLAG_PAGE);
551
                qemu_put_buffer_async(f, p, TARGET_PAGE_SIZE);
552
                bytes_sent += TARGET_PAGE_SIZE;
553
                acct_info.norm_pages++;
554 555
            }

556
            /* if page is unmodified, continue to the next */
557
            if (bytes_sent > 0) {
Juan Quintela's avatar
Juan Quintela committed
558
                last_sent_block = block;
559 560
                break;
            }
561
        }
562
    }
563
    last_seen_block = block;
564
    last_offset = offset;
565

566
    return bytes_sent;
567 568 569 570
}

static uint64_t bytes_transferred;

571 572 573 574 575 576 577 578 579 580 581 582
void acct_update_position(QEMUFile *f, size_t size, bool zero)
{
    uint64_t pages = size / TARGET_PAGE_SIZE;
    if (zero) {
        acct_info.dup_pages += pages;
    } else {
        acct_info.norm_pages += pages;
        bytes_transferred += size;
        qemu_update_position(f, size);
    }
}

583 584
static ram_addr_t ram_save_remaining(void)
{
Juan Quintela's avatar
Juan Quintela committed
585
    return migration_dirty_pages;
586 587 588 589 590 591 592 593 594 595 596 597 598 599
}

uint64_t ram_bytes_remaining(void)
{
    return ram_save_remaining() * TARGET_PAGE_SIZE;
}

uint64_t ram_bytes_transferred(void)
{
    return bytes_transferred;
}

uint64_t ram_bytes_total(void)
{
600 601 602
    RAMBlock *block;
    uint64_t total = 0;

603
    QTAILQ_FOREACH(block, &ram_list.blocks, next)
604 605 606
        total += block->length;

    return total;
607 608
}

Orit Wasserman's avatar
Orit Wasserman committed
609 610
static void migration_end(void)
{
611 612 613 614 615
    if (migration_bitmap) {
        memory_global_dirty_log_stop();
        g_free(migration_bitmap);
        migration_bitmap = NULL;
    }
616

617
    if (XBZRLE.cache) {
618 619 620 621 622 623
        cache_fini(XBZRLE.cache);
        g_free(XBZRLE.cache);
        g_free(XBZRLE.encoded_buf);
        g_free(XBZRLE.current_buf);
        g_free(XBZRLE.decoded_buf);
        XBZRLE.cache = NULL;
624 625 626
        XBZRLE.encoded_buf = NULL;
        XBZRLE.current_buf = NULL;
        XBZRLE.decoded_buf = NULL;
627
    }
Orit Wasserman's avatar
Orit Wasserman committed
628 629
}

630 631 632 633 634
static void ram_migration_cancel(void *opaque)
{
    migration_end();
}

635 636
static void reset_ram_globals(void)
{
637
    last_seen_block = NULL;
Juan Quintela's avatar
Juan Quintela committed
638
    last_sent_block = NULL;
639
    last_offset = 0;
640
    last_version = ram_list.version;
641
    ram_bulk_stage = true;
642 643
}

644 645
#define MAX_WAIT 50 /* ms, half buffered_file limit */

646
static int ram_save_setup(QEMUFile *f, void *opaque)
647
{
648
    RAMBlock *block;
Juan Quintela's avatar
Juan Quintela committed
649 650 651
    int64_t ram_pages = last_ram_offset() >> TARGET_PAGE_BITS;

    migration_bitmap = bitmap_new(ram_pages);
652
    bitmap_set(migration_bitmap, 0, ram_pages);
Juan Quintela's avatar
Juan Quintela committed
653
    migration_dirty_pages = ram_pages;
654 655
    mig_throttle_on = false;
    dirty_rate_high_cnt = 0;
656

657 658 659 660 661 662 663 664 665 666
    if (migrate_use_xbzrle()) {
        XBZRLE.cache = cache_init(migrate_xbzrle_cache_size() /
                                  TARGET_PAGE_SIZE,
                                  TARGET_PAGE_SIZE);
        if (!XBZRLE.cache) {
            DPRINTF("Error creating cache\n");
            return -1;
        }
        XBZRLE.encoded_buf = g_malloc0(TARGET_PAGE_SIZE);
        XBZRLE.current_buf = g_malloc(TARGET_PAGE_SIZE);
667
        acct_clear();
668 669
    }

670 671 672 673 674
    qemu_mutex_lock_iothread();
    qemu_mutex_lock_ramlist();
    bytes_transferred = 0;
    reset_ram_globals();

675
    memory_global_dirty_log_start();
Juan Quintela's avatar
Juan Quintela committed
676
    migration_bitmap_sync();
677
    qemu_mutex_unlock_iothread();
678

679
    qemu_put_be64(f, ram_bytes_total() | RAM_SAVE_FLAG_MEM_SIZE);
680

681
    QTAILQ_FOREACH(block, &ram_list.blocks, next) {
682 683 684
        qemu_put_byte(f, strlen(block->idstr));
        qemu_put_buffer(f, (uint8_t *)block->idstr, strlen(block->idstr));
        qemu_put_be64(f, block->length);
685 686
    }

687
    qemu_mutex_unlock_ramlist();
Michael R. Hines's avatar
Michael R. Hines committed
688 689 690 691

    ram_control_before_iterate(f, RAM_CONTROL_SETUP);
    ram_control_after_iterate(f, RAM_CONTROL_SETUP);

692 693 694 695 696
    qemu_put_be64(f, RAM_SAVE_FLAG_EOS);

    return 0;
}

697
static int ram_save_iterate(QEMUFile *f, void *opaque)
698 699 700
{
    int ret;
    int i;
701
    int64_t t0;
702
    int total_sent = 0;
703

704 705
    qemu_mutex_lock_ramlist();

706 707 708 709
    if (ram_list.version != last_version) {
        reset_ram_globals();
    }

Michael R. Hines's avatar
Michael R. Hines committed
710 711
    ram_control_before_iterate(f, RAM_CONTROL_ROUND);

712
    t0 = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
713
    i = 0;
714
    while ((ret = qemu_file_rate_limit(f)) == 0) {
715
        int bytes_sent;
716

717
        bytes_sent = ram_save_block(f, false);
718
        /* no more blocks to sent */
719
        if (bytes_sent == 0) {
720 721
            break;
        }
722
        total_sent += bytes_sent;
723
        acct_info.iterations++;
724
        check_guest_throttling();
725 726 727 728 729 730
        /* we want to check in the 1st loop, just in case it was the 1st time
           and we had to sync the dirty bitmap.
           qemu_get_clock_ns() is a bit expensive, so we only check each some
           iterations
        */
        if ((i & 63) == 0) {
731
            uint64_t t1 = (qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - t0) / 1000000;
732
            if (t1 > MAX_WAIT) {
733
                DPRINTF("big wait: %" PRIu64 " milliseconds, %d iterations\n",
734 735 736 737 738
                        t1, i);
                break;
            }
        }
        i++;
739 740
    }

741 742
    qemu_mutex_unlock_ramlist();

Michael R. Hines's avatar
Michael R. Hines committed
743 744 745 746 747 748
    /*
     * Must occur before EOS (or any QEMUFile operation)
     * because of RDMA protocol.
     */
    ram_control_after_iterate(f, RAM_CONTROL_ROUND);

749 750 751 752 753 754 755 756 757 758
    bytes_transferred += total_sent;

    /*
     * Do not count these 8 bytes into total_sent, so that we can
     * return 0 if no page had been dirtied.
     */
    qemu_put_be64(f, RAM_SAVE_FLAG_EOS);
    bytes_transferred += 8;

    ret = qemu_file_get_error(f);
759 760 761 762
    if (ret < 0) {
        return ret;
    }

763
    return total_sent;
764 765 766 767
}

static int ram_save_complete(QEMUFile *f, void *opaque)
{
768
    qemu_mutex_lock_ramlist();
769
    migration_bitmap_sync();
770

Michael R. Hines's avatar
Michael R. Hines committed
771 772
    ram_control_before_iterate(f, RAM_CONTROL_FINISH);

773
    /* try transferring iterative blocks of memory */
Orit Wasserman's avatar
Orit Wasserman committed
774

775
    /* flush all remaining blocks regardless of rate limiting */
776
    while (true) {
777 778
        int bytes_sent;

779
        bytes_sent = ram_save_block(f, true);
780
        /* no more blocks to sent */
781
        if (bytes_sent == 0) {
782
            break;
783
        }
784
        bytes_transferred += bytes_sent;
785
    }
Michael R. Hines's avatar
Michael R. Hines committed
786 787

    ram_control_after_iterate(f, RAM_CONTROL_FINISH);
788
    migration_end();
789

790
    qemu_mutex_unlock_ramlist();
791 792
    qemu_put_be64(f, RAM_SAVE_FLAG_EOS);

793
    return 0;
794 795
}

796 797 798 799 800 801 802
static uint64_t ram_save_pending(QEMUFile *f, void *opaque, uint64_t max_size)
{
    uint64_t remaining_size;

    remaining_size = ram_save_remaining() * TARGET_PAGE_SIZE;

    if (remaining_size < max_size) {
803
        qemu_mutex_lock_iothread();
804
        migration_bitmap_sync();
805
        qemu_mutex_unlock_iothread();
806 807 808 809 810
        remaining_size = ram_save_remaining() * TARGET_PAGE_SIZE;
    }
    return remaining_size;
}

811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851
static int load_xbzrle(QEMUFile *f, ram_addr_t addr, void *host)
{
    int ret, rc = 0;
    unsigned int xh_len;
    int xh_flags;

    if (!XBZRLE.decoded_buf) {
        XBZRLE.decoded_buf = g_malloc(TARGET_PAGE_SIZE);
    }

    /* extract RLE header */
    xh_flags = qemu_get_byte(f);
    xh_len = qemu_get_be16(f);

    if (xh_flags != ENCODING_FLAG_XBZRLE) {
        fprintf(stderr, "Failed to load XBZRLE page - wrong compression!\n");
        return -1;
    }

    if (xh_len > TARGET_PAGE_SIZE) {
        fprintf(stderr, "Failed to load XBZRLE page - len overflow!\n");
        return -1;
    }
    /* load data and decode */
    qemu_get_buffer(f, XBZRLE.decoded_buf, xh_len);

    /* decode RLE */
    ret = xbzrle_decode_buffer(XBZRLE.decoded_buf, xh_len, host,
                               TARGET_PAGE_SIZE);
    if (ret == -1) {
        fprintf(stderr, "Failed to load XBZRLE page - decode error!\n");
        rc = -1;
    } else  if (ret > TARGET_PAGE_SIZE) {
        fprintf(stderr, "Failed to load XBZRLE page - size %d exceeds %d!\n",
                ret, TARGET_PAGE_SIZE);
        abort();
    }

    return rc;
}

852 853 854 855 856 857 858 859 860 861 862 863 864 865
static inline void *host_from_stream_offset(QEMUFile *f,
                                            ram_addr_t offset,
                                            int flags)
{
    static RAMBlock *block = NULL;
    char id[256];
    uint8_t len;

    if (flags & RAM_SAVE_FLAG_CONTINUE) {
        if (!block) {
            fprintf(stderr, "Ack, bad migration stream!\n");
            return NULL;
        }

866
        return memory_region_get_ram_ptr(block->mr) + offset;
867 868 869 870 871 872
    }

    len = qemu_get_byte(f);
    qemu_get_buffer(f, (uint8_t *)id, len);
    id[len] = 0;

873
    QTAILQ_FOREACH(block, &ram_list.blocks, next) {
874
        if (!strncmp(id, block->idstr, sizeof(id)))
875
            return memory_region_get_ram_ptr(block->mr) + offset;
876 877 878 879 880 881
    }

    fprintf(stderr, "Can't find block %s!\n", id);
    return NULL;
}

882 883 884 885 886 887
/*
 * If a page (or a whole RDMA chunk) has been
 * determined to be zero, then zap it.
 */
void ram_handle_compressed(void *host, uint8_t ch, uint64_t size)
{
888
    if (ch != 0 || !is_zero_range(host, size)) {
889 890 891 892
        memset(host, ch, size);
    }
}

893
static int ram_load(QEMUFile *f, void *opaque, int version_id)
894 895
{
    ram_addr_t addr;
Orit Wasserman's avatar
Orit Wasserman committed
896
    int flags, ret = 0;
897
    int error;
Orit Wasserman's avatar
Orit Wasserman committed
898 899 900
    static uint64_t seq_iter;

    seq_iter++;
901

902
    if (version_id < 4 || version_id > 4) {
903 904 905 906 907 908 909 910 911 912
        return -EINVAL;
    }

    do {
        addr = qemu_get_be64(f);

        flags = addr & ~TARGET_PAGE_MASK;
        addr &= TARGET_PAGE_MASK;

        if (flags & RAM_SAVE_FLAG_MEM_SIZE) {
913
            if (version_id == 4) {
914 915 916 917 918 919 920 921 922 923 924 925 926 927
                /* Synchronize RAM block list */
                char id[256];
                ram_addr_t length;
                ram_addr_t total_ram_bytes = addr;

                while (total_ram_bytes) {
                    RAMBlock *block;
                    uint8_t len;

                    len = qemu_get_byte(f);
                    qemu_get_buffer(f, (uint8_t *)id, len);
                    id[len] = 0;
                    length = qemu_get_be64(f);

928
                    QTAILQ_FOREACH(block, &ram_list.blocks, next) {
929
                        if (!strncmp(id, block->idstr, sizeof(id))) {
Orit Wasserman's avatar
Orit Wasserman committed
930
                            if (block->length != length) {
931 932 933
                                fprintf(stderr,
                                        "Length mismatch: %s: " RAM_ADDR_FMT
                                        " in != " RAM_ADDR_FMT "\n", id, length,
934
                                        block->length);
Orit Wasserman's avatar
Orit Wasserman committed
935 936 937
                                ret =  -EINVAL;
                                goto done;
                            }
938 939 940 941 942
                            break;
                        }
                    }

                    if (!block) {
943 944
                        fprintf(stderr, "Unknown ramblock \"%s\", cannot "
                                "accept migration\n", id);
Orit Wasserman's avatar
Orit Wasserman committed
945 946
                        ret = -EINVAL;
                        goto done;
947 948 949 950
                    }

                    total_ram_bytes -= length;
                }
951 952 953 954
            }
        }

        if (flags & RAM_SAVE_FLAG_COMPRESS) {
955 956 957
            void *host;
            uint8_t ch;

958
            host = host_from_stream_offset(f, addr, flags);
959 960 961
            if (!host) {
                return -EINVAL;
            }
962 963

            ch = qemu_get_byte(f);
964
            ram_handle_compressed(host, ch, TARGET_PAGE_SIZE);
965
        } else if (flags & RAM_SAVE_FLAG_PAGE) {
966 967
            void *host;

968
            host = host_from_stream_offset(f, addr, flags);
969 970 971
            if (!host) {
                return -EINVAL;
            }
972 973

            qemu_get_buffer(f, host, TARGET_PAGE_SIZE);
974 975 976 977 978 979 980 981 982 983
        } else if (flags & RAM_SAVE_FLAG_XBZRLE) {
            void *host = host_from_stream_offset(f, addr, flags);
            if (!host) {
                return -EINVAL;
            }

            if (load_xbzrle(f, addr, host) < 0) {
                ret = -EINVAL;
                goto done;
            }
Michael R. Hines's avatar
Michael R. Hines committed
984 985
        } else if (flags & RAM_SAVE_FLAG_HOOK) {
            ram_control_load_hook(f, flags);
986
        }
987 988
        error = qemu_file_get_error(f);
        if (error) {
Orit Wasserman's avatar
Orit Wasserman committed
989 990
            ret = error;
            goto done;
991 992 993
        }
    } while (!(flags & RAM_SAVE_FLAG_EOS));

Orit Wasserman's avatar
Orit Wasserman committed
994
done:
995 996
    DPRINTF("Completed load of VM with exit code %d seq iteration "
            "%" PRIu64 "\n", ret, seq_iter);
Orit Wasserman's avatar
Orit Wasserman committed
997
    return ret;
998 999
}

1000
SaveVMHandlers savevm_ram_handlers = {
1001
    .save_live_setup = ram_save_setup,
1002 1003
    .save_live_iterate = ram_save_iterate,
    .save_live_complete = ram_save_complete,
1004
    .save_live_pending = ram_save_pending,
1005
    .load_state = ram_load,
1006
    .cancel = ram_migration_cancel,
1007 1008
};

1009 1010 1011 1012 1013 1014
struct soundhw {
    const char *name;
    const char *descr;
    int enabled;
    int isa;
    union {
1015
        int (*init_isa) (ISABus *bus);
1016 1017 1018 1019
        int (*init_pci) (PCIBus *bus);
    } init;
};

1020 1021
static struct soundhw soundhw[9];
static int soundhw_count;
1022

1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
void isa_register_soundhw(const char *name, const char *descr,
                          int (*init_isa)(ISABus *bus))
{
    assert(soundhw_count < ARRAY_SIZE(soundhw) - 1);
    soundhw[soundhw_count].name = name;
    soundhw[soundhw_count].descr = descr;
    soundhw[soundhw_count].isa = 1;
    soundhw[soundhw_count].init.init_isa = init_isa;
    soundhw_count++;
}
1033

1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
void pci_register_soundhw(const char *name, const char *descr,
                          int (*init_pci)(PCIBus *bus))
{
    assert(soundhw_count < ARRAY_SIZE(soundhw) - 1);
    soundhw[soundhw_count].name = name;
    soundhw[soundhw_count].descr = descr;
    soundhw[soundhw_count].isa = 0;
    soundhw[soundhw_count].init.init_pci = init_pci;
    soundhw_count++;
}
1044 1045 1046 1047 1048

void select_soundhw(const char *optarg)
{
    struct soundhw *c;

1049
    if (is_help_option(optarg)) {
1050 1051
    show_valid_cards:

1052 1053 1054 1055 1056 1057 1058 1059 1060
        if (soundhw_count) {
             printf("Valid sound card names (comma separated):\n");
             for (c = soundhw; c->name; ++c) {
                 printf ("%-11s %s\n", c->name, c->descr);
             }
             printf("\n-soundhw all will enable all of the above\n");
        } else {
             printf("Machine has no user-selectable audio hardware "
                    "(it may or may not have always-present audio hardware).\n");
1061
        }
1062
        exit(!is_help_option(optarg));
1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087