[qemu.git] / block-migration.c

/*
 * QEMU live block migration
 *
 * Copyright IBM, Corp. 2009
 *
 * Authors:
 *  Liran Schour   <lirans@il.ibm.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2.  See
 * the COPYING file in the top-level directory.
 *
 */

#include "qemu-common.h"
#include "block_int.h"
#include "hw/hw.h"
#include "block-migration.h"
#include <assert.h>

#define BLOCK_SIZE (BDRV_SECTORS_PER_DIRTY_CHUNK << BDRV_SECTOR_BITS)

#define BLK_MIG_FLAG_DEVICE_BLOCK       0x01
#define BLK_MIG_FLAG_EOS                0x02

#define MAX_IS_ALLOCATED_SEARCH 65536
#define MAX_BLOCKS_READ 10000
#define BLOCKS_READ_CHANGE 100
#define INITIAL_BLOCKS_READ 100

//#define DEBUG_BLK_MIGRATION

#ifdef DEBUG_BLK_MIGRATION
#define dprintf(fmt, ...) \
    do { printf("blk_migration: " fmt, ## __VA_ARGS__); } while (0)
#else
#define dprintf(fmt, ...) \
    do { } while (0)
#endif

typedef struct BlkMigDevState {
    BlockDriverState *bs;
    int bulk_completed;
    int shared_base;
    struct BlkMigDevState *next;
    int64_t cur_sector;
    int64_t total_sectors;
    int64_t dirty;
} BlkMigDevState;

typedef struct BlkMigBlock {
    uint8_t *buf;
    BlkMigDevState *bmds;
    int64_t sector;
    struct iovec iov;
    QEMUIOVector qiov;
    BlockDriverAIOCB *aiocb;
    int ret;
    struct BlkMigBlock *next;
} BlkMigBlock;

typedef struct BlkMigState {
    int bulk_completed;
    int blk_enable;
    int shared_base;
    int no_dirty;
    QEMUFile *load_file;
    BlkMigDevState *bmds_first;
    BlkMigBlock *first_blk;
    BlkMigBlock *last_blk;
    int submitted;
    int read_done;
    int transferred;
    int64_t print_completion;
} BlkMigState;

static BlkMigState *block_mig_state = NULL;

static void blk_mig_read_cb(void *opaque, int ret)
{
    BlkMigBlock *blk = opaque;

    blk->ret = ret;

    /* insert at the end */
    if (block_mig_state->last_blk == NULL) {
        block_mig_state->first_blk = blk;
        block_mig_state->last_blk = blk;
    } else {
        block_mig_state->last_blk->next = blk;
        block_mig_state->last_blk = blk;
    }

    block_mig_state->submitted--;
    block_mig_state->read_done++;
    assert(block_mig_state->submitted >= 0);
}

static int mig_read_device_bulk(QEMUFile *f, BlkMigDevState *bms)
{
    int nr_sectors;
    int64_t total_sectors, cur_sector = 0;
    BlockDriverState *bs = bms->bs;
    BlkMigBlock *blk;

    blk = qemu_malloc(sizeof(BlkMigBlock));
    blk->buf = qemu_malloc(BLOCK_SIZE);

    cur_sector = bms->cur_sector;
    total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;

    if (bms->shared_base) {
        while (cur_sector < bms->total_sectors &&
               !bdrv_is_allocated(bms->bs, cur_sector,
                                  MAX_IS_ALLOCATED_SEARCH, &nr_sectors)) {
            cur_sector += nr_sectors;
        }
    }

    if (cur_sector >= total_sectors) {
        bms->cur_sector = total_sectors;
        qemu_free(blk->buf);
        qemu_free(blk);
        return 1;
    }

    if (cur_sector >= block_mig_state->print_completion) {
        printf("Completed %" PRId64 " %%\r", cur_sector * 100 / total_sectors);
        fflush(stdout);
        block_mig_state->print_completion +=
            (BDRV_SECTORS_PER_DIRTY_CHUNK * 10000);
    }

    /* we are going to transfer a full block even if it is not allocated */
    nr_sectors = BDRV_SECTORS_PER_DIRTY_CHUNK;

    cur_sector &= ~((int64_t)BDRV_SECTORS_PER_DIRTY_CHUNK - 1);

    if (total_sectors - cur_sector < BDRV_SECTORS_PER_DIRTY_CHUNK) {
        nr_sectors = (total_sectors - cur_sector);
    }

    bms->cur_sector = cur_sector + nr_sectors;
    blk->sector = cur_sector;
    blk->bmds = bms;
    blk->next = NULL;

    blk->iov.iov_base = blk->buf;
    blk->iov.iov_len = nr_sectors * BDRV_SECTOR_SIZE;
    qemu_iovec_init_external(&blk->qiov, &blk->iov, 1);

    blk->aiocb = bdrv_aio_readv(bs, cur_sector, &blk->qiov,
                                nr_sectors, blk_mig_read_cb, blk);

    if (!blk->aiocb) {
        printf("Error reading sector %" PRId64 "\n", cur_sector);
        qemu_free(blk->buf);
        qemu_free(blk);
        return 0;
    }

    bdrv_reset_dirty(bms->bs, cur_sector, nr_sectors);
    block_mig_state->submitted++;

    return (bms->cur_sector >= total_sectors);
}

static int mig_save_device_bulk(QEMUFile *f, BlkMigDevState *bmds)
{
    int len, nr_sectors;
    int64_t total_sectors = bmds->total_sectors, cur_sector = 0;
    uint8_t *tmp_buf = NULL;
    BlockDriverState *bs = bmds->bs;

    tmp_buf = qemu_malloc(BLOCK_SIZE);

    cur_sector = bmds->cur_sector;

    if (bmds->shared_base) {
        while (cur_sector < bmds->total_sectors &&
               !bdrv_is_allocated(bmds->bs, cur_sector,
                                  MAX_IS_ALLOCATED_SEARCH, &nr_sectors)) {
            cur_sector += nr_sectors;
        }
    }

    if (cur_sector >= total_sectors) {
        bmds->cur_sector = total_sectors;
        qemu_free(tmp_buf);
        return 1;
    }

    if (cur_sector >= block_mig_state->print_completion) {
        printf("Completed %" PRId64 " %%\r", cur_sector * 100 / total_sectors);
        fflush(stdout);
        block_mig_state->print_completion +=
            (BDRV_SECTORS_PER_DIRTY_CHUNK * 10000);
    }

    cur_sector &= ~((int64_t)BDRV_SECTORS_PER_DIRTY_CHUNK - 1);

    /* we are going to transfer a full block even if it is not allocated */
    nr_sectors = BDRV_SECTORS_PER_DIRTY_CHUNK;

    if (total_sectors - cur_sector < BDRV_SECTORS_PER_DIRTY_CHUNK) {
        nr_sectors = (total_sectors - cur_sector);
    }

    if (bdrv_read(bs, cur_sector, tmp_buf, nr_sectors) < 0) {
        printf("Error reading sector %" PRId64 "\n", cur_sector);
    }

    bdrv_reset_dirty(bs, cur_sector, nr_sectors);

    /* sector number and flags */
    qemu_put_be64(f, (cur_sector << BDRV_SECTOR_BITS)
                     | BLK_MIG_FLAG_DEVICE_BLOCK);

    /* device name */
    len = strlen(bs->device_name);
    qemu_put_byte(f, len);
    qemu_put_buffer(f, (uint8_t *)bs->device_name, len);

    qemu_put_buffer(f, tmp_buf, BLOCK_SIZE);

    bmds->cur_sector = cur_sector + BDRV_SECTORS_PER_DIRTY_CHUNK;

    qemu_free(tmp_buf);

    return (bmds->cur_sector >= total_sectors);
}

static void send_blk(QEMUFile *f, BlkMigBlock * blk)
{
    int len;

    /* sector number and flags */
    qemu_put_be64(f, (blk->sector << BDRV_SECTOR_BITS)
                     | BLK_MIG_FLAG_DEVICE_BLOCK);

    /* device name */
    len = strlen(blk->bmds->bs->device_name);
    qemu_put_byte(f, len);
    qemu_put_buffer(f, (uint8_t *)blk->bmds->bs->device_name, len);

    qemu_put_buffer(f, blk->buf, BLOCK_SIZE);
}

static void blk_mig_save_dev_info(QEMUFile *f, BlkMigDevState *bmds)
{
}

static void set_dirty_tracking(int enable)
{
    BlkMigDevState *bmds;
    for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
        bdrv_set_dirty_tracking(bmds->bs, enable);
    }
}

static void init_blk_migration(QEMUFile *f)
{
    BlkMigDevState **pbmds, *bmds;
    BlockDriverState *bs;

    for (bs = bdrv_first; bs != NULL; bs = bs->next) {
        if (bs->type == BDRV_TYPE_HD) {
            bmds = qemu_mallocz(sizeof(BlkMigDevState));
            bmds->bs = bs;
            bmds->bulk_completed = 0;
            bmds->total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
            bmds->shared_base = block_mig_state->shared_base;

            if (bmds->shared_base) {
                printf("Start migration for %s with shared base image\n",
                       bs->device_name);
            } else {
                printf("Start full migration for %s\n", bs->device_name);
            }

            /* insert at the end */
            pbmds = &block_mig_state->bmds_first;
            while (*pbmds != NULL) {
                pbmds = &(*pbmds)->next;
            }
            *pbmds = bmds;

            blk_mig_save_dev_info(f, bmds);
        }
    }
}

static int blk_mig_save_bulked_block(QEMUFile *f, int is_async)
{
    BlkMigDevState *bmds;

    for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
        if (bmds->bulk_completed == 0) {
            if (is_async) {
                if (mig_read_device_bulk(f, bmds) == 1) {
                    /* completed bulk section for this device */
                    bmds->bulk_completed = 1;
                }
            } else {
                if (mig_save_device_bulk(f, bmds) == 1) {
                    /* completed bulk section for this device */
                    bmds->bulk_completed = 1;
                }
            }
            return 1;
        }
    }

    /* we reached here means bulk is completed */
    block_mig_state->bulk_completed = 1;

    return 0;
}

#define MAX_NUM_BLOCKS 4

static void blk_mig_save_dirty_blocks(QEMUFile *f)
{
    BlkMigDevState *bmds;
    uint8_t buf[BLOCK_SIZE];
    int64_t sector;
    int len;

    for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
        for (sector = 0; sector < bmds->cur_sector;) {
            if (bdrv_get_dirty(bmds->bs, sector)) {
                if (bdrv_read(bmds->bs, sector, buf,
                              BDRV_SECTORS_PER_DIRTY_CHUNK) < 0) {
                    /* FIXME: add error handling */
                }

                /* sector number and flags */
                qemu_put_be64(f, (sector << BDRV_SECTOR_BITS)
                                 | BLK_MIG_FLAG_DEVICE_BLOCK);

                /* device name */
                len = strlen(bmds->bs->device_name);
                qemu_put_byte(f, len);
                qemu_put_buffer(f, (uint8_t *)bmds->bs->device_name, len);

                qemu_put_buffer(f, buf, BLOCK_SIZE);

                bdrv_reset_dirty(bmds->bs, sector,
                                 BDRV_SECTORS_PER_DIRTY_CHUNK);
            }
            sector += BDRV_SECTORS_PER_DIRTY_CHUNK;
        }
    }
}

static void flush_blks(QEMUFile* f)
{
    BlkMigBlock *blk, *next;

    dprintf("%s Enter submitted %d read_done %d transfered\n", __FUNCTION__,
            submitted, read_done, transfered);

    for (blk = block_mig_state->first_blk;
         blk != NULL && !qemu_file_rate_limit(f);
         blk = next) {
        send_blk(f, blk);

        next = blk->next;
        qemu_free(blk->buf);
        qemu_free(blk);

        block_mig_state->read_done--;
        block_mig_state->transferred++;
        assert(block_mig_state->read_done >= 0);
    }
    block_mig_state->first_blk = blk;

    if (block_mig_state->first_blk == NULL) {
        block_mig_state->last_blk = NULL;
    }

    dprintf("%s Exit submitted %d read_done %d transferred%d\n", __FUNCTION__,
            block_mig_state->submitted, block_mig_state->read_done,
            block_mig_state->transferred);
}

static int is_stage2_completed(void)
{
    BlkMigDevState *bmds;

    if (block_mig_state->submitted > 0) {
        return 0;
    }

    for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
        if (bmds->bulk_completed == 0) {
            return 0;
        }
    }

    return 1;
}

static int block_save_live(QEMUFile *f, int stage, void *opaque)
{
    dprintf("Enter save live stage %d submitted %d transferred %d\n", stage,
            submitted, transferred);

    if (block_mig_state->blk_enable != 1) {
        /* no need to migrate storage */
        qemu_put_be64(f, BLK_MIG_FLAG_EOS);
        return 1;
    }

    if (stage == 1) {
        init_blk_migration(f);

        /* start track dirty blocks */
        set_dirty_tracking(1);
    }

    flush_blks(f);

    /* control the rate of transfer */
    while ((block_mig_state->submitted +
            block_mig_state->read_done) * BLOCK_SIZE <
           qemu_file_get_rate_limit(f)) {
        if (blk_mig_save_bulked_block(f, 1) == 0) {
            /* no more bulk blocks for now */
            break;
        }
    }

    flush_blks(f);

    if (stage == 3) {
        while (blk_mig_save_bulked_block(f, 0) != 0) {
            /* empty */
        }

        blk_mig_save_dirty_blocks(f);

        /* stop track dirty blocks */
        set_dirty_tracking(0);

        printf("\nBlock migration completed\n");
    }

    qemu_put_be64(f, BLK_MIG_FLAG_EOS);

    return ((stage == 2) && is_stage2_completed());
}

static int block_load(QEMUFile *f, void *opaque, int version_id)
{
    int len, flags;
    char device_name[256];
    int64_t addr;
    BlockDriverState *bs;
    uint8_t *buf;

    buf = qemu_malloc(BLOCK_SIZE);

    do {
        addr = qemu_get_be64(f);

        flags = addr & ~BDRV_SECTOR_MASK;
        addr >>= BDRV_SECTOR_BITS;

        if (flags & BLK_MIG_FLAG_DEVICE_BLOCK) {
            /* get device name */
            len = qemu_get_byte(f);

            qemu_get_buffer(f, (uint8_t *)device_name, len);
            device_name[len] = '\0';

            bs = bdrv_find(device_name);

            qemu_get_buffer(f, buf, BLOCK_SIZE);
            if (bs != NULL) {
                bdrv_write(bs, addr, buf, BDRV_SECTORS_PER_DIRTY_CHUNK);
            } else {
                printf("Error unknown block device %s\n", device_name);
                /* FIXME: add error handling */
            }
        } else if (!(flags & BLK_MIG_FLAG_EOS)) {
            printf("Unknown flags\n");
            /* FIXME: add error handling */
        }
    } while (!(flags & BLK_MIG_FLAG_EOS));

    qemu_free(buf);

    return 0;
}

static void block_set_params(int blk_enable, int shared_base, void *opaque)
{
    assert(opaque == block_mig_state);

    block_mig_state->blk_enable = blk_enable;
    block_mig_state->shared_base = shared_base;

    /* shared base means that blk_enable = 1 */
    block_mig_state->blk_enable |= shared_base;
}

void blk_mig_info(void)
{
    BlockDriverState *bs;

    for (bs = bdrv_first; bs != NULL; bs = bs->next) {
        printf("Device %s\n", bs->device_name);
        if (bs->type == BDRV_TYPE_HD) {
            printf("device %s format %s\n",
                   bs->device_name, bs->drv->format_name);
        }
    }
}

void blk_mig_init(void)
{
    block_mig_state = qemu_mallocz(sizeof(BlkMigState));

    register_savevm_live("block", 0, 1, block_set_params, block_save_live,
                         NULL, block_load, block_mig_state);
}
Commit	Line	Data
c163b5ca	1	/*
	2	* QEMU live block migration
	3	*
	4	* Copyright IBM, Corp. 2009
	5	*
	6	* Authors:
	7	* Liran Schour <lirans@il.ibm.com>
	8	*
	9	* This work is licensed under the terms of the GNU GPL, version 2. See
	10	* the COPYING file in the top-level directory.
	11	*
	12	*/
	13
	14	#include "qemu-common.h"
	15	#include "block_int.h"
	16	#include "hw/hw.h"
	17	#include "block-migration.h"
	18	#include <assert.h>
c163b5ca	19
6ea44308	20	#define BLOCK_SIZE (BDRV_SECTORS_PER_DIRTY_CHUNK << BDRV_SECTOR_BITS)
c163b5ca	21
	22	#define BLK_MIG_FLAG_DEVICE_BLOCK 0x01
	23	#define BLK_MIG_FLAG_EOS 0x02
	24
	25	#define MAX_IS_ALLOCATED_SEARCH 65536
	26	#define MAX_BLOCKS_READ 10000
	27	#define BLOCKS_READ_CHANGE 100
	28	#define INITIAL_BLOCKS_READ 100
	29
	30	//#define DEBUG_BLK_MIGRATION
	31
	32	#ifdef DEBUG_BLK_MIGRATION
a55eb92c	33	#define dprintf(fmt, ...) \
c163b5ca	34	do { printf("blk_migration: " fmt, ## __VA_ARGS__); } while (0)
c163b5ca	35	#else
a55eb92c	36	#define dprintf(fmt, ...) \
c163b5ca	37	do { } while (0)
	38	#endif
	39
a55eb92c JK	40	typedef struct BlkMigDevState {
	41	BlockDriverState *bs;
	42	int bulk_completed;
	43	int shared_base;
	44	struct BlkMigDevState *next;
	45	int64_t cur_sector;
	46	int64_t total_sectors;
	47	int64_t dirty;
	48	} BlkMigDevState;
	49
c163b5ca	50	typedef struct BlkMigBlock {
	51	uint8_t *buf;
	52	BlkMigDevState *bmds;
	53	int64_t sector;
	54	struct iovec iov;
	55	QEMUIOVector qiov;
	56	BlockDriverAIOCB *aiocb;
	57	int ret;
	58	struct BlkMigBlock *next;
	59	} BlkMigBlock;
	60
	61	typedef struct BlkMigState {
	62	int bulk_completed;
	63	int blk_enable;
	64	int shared_base;
	65	int no_dirty;
	66	QEMUFile *load_file;
	67	BlkMigDevState *bmds_first;
c163b5ca	68	BlkMigBlock *first_blk;
	69	BlkMigBlock *last_blk;
	70	int submitted;
	71	int read_done;
	72	int transferred;
	73	int64_t print_completion;
	74	} BlkMigState;
	75
a55eb92c	76	static BlkMigState *block_mig_state = NULL;
c163b5ca	77
	78	static void blk_mig_read_cb(void *opaque, int ret)
	79	{
	80	BlkMigBlock *blk = opaque;
a55eb92c	81
c163b5ca	82	blk->ret = ret;
a55eb92c	83
c163b5ca	84	/* insert at the end */
a55eb92c	85	if (block_mig_state->last_blk == NULL) {
c163b5ca	86	block_mig_state->first_blk = blk;
	87	block_mig_state->last_blk = blk;
	88	} else {
	89	block_mig_state->last_blk->next = blk;
	90	block_mig_state->last_blk = blk;
	91	}
a55eb92c	92
c163b5ca	93	block_mig_state->submitted--;
	94	block_mig_state->read_done++;
	95	assert(block_mig_state->submitted >= 0);
c163b5ca	96	}
	97
	98	static int mig_read_device_bulk(QEMUFile f, BlkMigDevState bms)
a55eb92c	99	{
c163b5ca	100	int nr_sectors;
	101	int64_t total_sectors, cur_sector = 0;
	102	BlockDriverState *bs = bms->bs;
	103	BlkMigBlock *blk;
a55eb92c	104
c163b5ca	105	blk = qemu_malloc(sizeof(BlkMigBlock));
c163b5ca	106	blk->buf = qemu_malloc(BLOCK_SIZE);
a55eb92c	107
c163b5ca	108	cur_sector = bms->cur_sector;
6ea44308	109	total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
a55eb92c JK	110
	111	if (bms->shared_base) {
	112	while (cur_sector < bms->total_sectors &&
	113	!bdrv_is_allocated(bms->bs, cur_sector,
	114	MAX_IS_ALLOCATED_SEARCH, &nr_sectors)) {
c163b5ca	115	cur_sector += nr_sectors;
	116	}
	117	}
a55eb92c JK	118
a55eb92c JK	119	if (cur_sector >= total_sectors) {
c163b5ca	120	bms->cur_sector = total_sectors;
	121	qemu_free(blk->buf);
	122	qemu_free(blk);
	123	return 1;
	124	}
a55eb92c JK	125
a55eb92c JK	126	if (cur_sector >= block_mig_state->print_completion) {
c163b5ca	127	printf("Completed %" PRId64 " %%\r", cur_sector * 100 / total_sectors);
c163b5ca	128	fflush(stdout);
a55eb92c	129	block_mig_state->print_completion +=
6ea44308	130	(BDRV_SECTORS_PER_DIRTY_CHUNK * 10000);
c163b5ca	131	}
a55eb92c	132
6ea44308 JK	133	/* we are going to transfer a full block even if it is not allocated */
6ea44308 JK	134	nr_sectors = BDRV_SECTORS_PER_DIRTY_CHUNK;
c163b5ca	135
6ea44308	136	cur_sector &= ~((int64_t)BDRV_SECTORS_PER_DIRTY_CHUNK - 1);
a55eb92c	137
6ea44308	138	if (total_sectors - cur_sector < BDRV_SECTORS_PER_DIRTY_CHUNK) {
c163b5ca	139	nr_sectors = (total_sectors - cur_sector);
c163b5ca	140	}
a55eb92c	141
c163b5ca	142	bms->cur_sector = cur_sector + nr_sectors;
	143	blk->sector = cur_sector;
	144	blk->bmds = bms;
	145	blk->next = NULL;
a55eb92c	146
c163b5ca	147	blk->iov.iov_base = blk->buf;
6ea44308	148	blk->iov.iov_len = nr_sectors * BDRV_SECTOR_SIZE;
c163b5ca	149	qemu_iovec_init_external(&blk->qiov, &blk->iov, 1);
a55eb92c	150
c163b5ca	151	blk->aiocb = bdrv_aio_readv(bs, cur_sector, &blk->qiov,
c163b5ca	152	nr_sectors, blk_mig_read_cb, blk);
a55eb92c JK	153
a55eb92c JK	154	if (!blk->aiocb) {
c163b5ca	155	printf("Error reading sector %" PRId64 "\n", cur_sector);
	156	qemu_free(blk->buf);
	157	qemu_free(blk);
	158	return 0;
	159	}
	160
	161	bdrv_reset_dirty(bms->bs, cur_sector, nr_sectors);
	162	block_mig_state->submitted++;
a55eb92c	163
c163b5ca	164	return (bms->cur_sector >= total_sectors);
	165	}
	166
	167	static int mig_save_device_bulk(QEMUFile f, BlkMigDevState bmds)
a55eb92c	168	{
c163b5ca	169	int len, nr_sectors;
	170	int64_t total_sectors = bmds->total_sectors, cur_sector = 0;
	171	uint8_t *tmp_buf = NULL;
	172	BlockDriverState *bs = bmds->bs;
	173
	174	tmp_buf = qemu_malloc(BLOCK_SIZE);
a55eb92c	175
c163b5ca	176	cur_sector = bmds->cur_sector;
a55eb92c JK	177
	178	if (bmds->shared_base) {
	179	while (cur_sector < bmds->total_sectors &&
	180	!bdrv_is_allocated(bmds->bs, cur_sector,
	181	MAX_IS_ALLOCATED_SEARCH, &nr_sectors)) {
c163b5ca	182	cur_sector += nr_sectors;
	183	}
	184	}
a55eb92c JK	185
a55eb92c JK	186	if (cur_sector >= total_sectors) {
c163b5ca	187	bmds->cur_sector = total_sectors;
	188	qemu_free(tmp_buf);
	189	return 1;
	190	}
a55eb92c JK	191
a55eb92c JK	192	if (cur_sector >= block_mig_state->print_completion) {
c163b5ca	193	printf("Completed %" PRId64 " %%\r", cur_sector * 100 / total_sectors);
c163b5ca	194	fflush(stdout);
a55eb92c	195	block_mig_state->print_completion +=
6ea44308	196	(BDRV_SECTORS_PER_DIRTY_CHUNK * 10000);
c163b5ca	197	}
a55eb92c	198
6ea44308	199	cur_sector &= ~((int64_t)BDRV_SECTORS_PER_DIRTY_CHUNK - 1);
a55eb92c	200
6ea44308 JK	201	/* we are going to transfer a full block even if it is not allocated */
6ea44308 JK	202	nr_sectors = BDRV_SECTORS_PER_DIRTY_CHUNK;
a55eb92c	203
6ea44308	204	if (total_sectors - cur_sector < BDRV_SECTORS_PER_DIRTY_CHUNK) {
c163b5ca	205	nr_sectors = (total_sectors - cur_sector);
c163b5ca	206	}
a55eb92c JK	207
a55eb92c JK	208	if (bdrv_read(bs, cur_sector, tmp_buf, nr_sectors) < 0) {
c163b5ca	209	printf("Error reading sector %" PRId64 "\n", cur_sector);
	210	}
	211
	212	bdrv_reset_dirty(bs, cur_sector, nr_sectors);
a55eb92c JK	213
a55eb92c JK	214	/* sector number and flags */
6ea44308 JK	215	qemu_put_be64(f, (cur_sector << BDRV_SECTOR_BITS)
6ea44308 JK	216	\| BLK_MIG_FLAG_DEVICE_BLOCK);
a55eb92c JK	217
a55eb92c JK	218	/* device name */
c163b5ca	219	len = strlen(bs->device_name);
	220	qemu_put_byte(f, len);
	221	qemu_put_buffer(f, (uint8_t *)bs->device_name, len);
a55eb92c JK	222
	223	qemu_put_buffer(f, tmp_buf, BLOCK_SIZE);
	224
6ea44308	225	bmds->cur_sector = cur_sector + BDRV_SECTORS_PER_DIRTY_CHUNK;
a55eb92c	226
c163b5ca	227	qemu_free(tmp_buf);
a55eb92c	228
c163b5ca	229	return (bmds->cur_sector >= total_sectors);
	230	}
	231
	232	static void send_blk(QEMUFile f, BlkMigBlock blk)
	233	{
	234	int len;
a55eb92c JK	235
a55eb92c JK	236	/* sector number and flags */
6ea44308 JK	237	qemu_put_be64(f, (blk->sector << BDRV_SECTOR_BITS)
6ea44308 JK	238	\| BLK_MIG_FLAG_DEVICE_BLOCK);
a55eb92c JK	239
a55eb92c JK	240	/* device name */
c163b5ca	241	len = strlen(blk->bmds->bs->device_name);
	242	qemu_put_byte(f, len);
	243	qemu_put_buffer(f, (uint8_t *)blk->bmds->bs->device_name, len);
a55eb92c JK	244
a55eb92c JK	245	qemu_put_buffer(f, blk->buf, BLOCK_SIZE);
c163b5ca	246	}
	247
	248	static void blk_mig_save_dev_info(QEMUFile f, BlkMigDevState bmds)
	249	{
	250	}
	251
	252	static void set_dirty_tracking(int enable)
	253	{
	254	BlkMigDevState *bmds;
a55eb92c JK	255	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
a55eb92c JK	256	bdrv_set_dirty_tracking(bmds->bs, enable);
c163b5ca	257	}
c163b5ca	258	}
	259
	260	static void init_blk_migration(QEMUFile *f)
	261	{
	262	BlkMigDevState *pbmds, bmds;
	263	BlockDriverState *bs;
a55eb92c	264
c163b5ca	265	for (bs = bdrv_first; bs != NULL; bs = bs->next) {
a55eb92c	266	if (bs->type == BDRV_TYPE_HD) {
c163b5ca	267	bmds = qemu_mallocz(sizeof(BlkMigDevState));
	268	bmds->bs = bs;
	269	bmds->bulk_completed = 0;
6ea44308	270	bmds->total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
c163b5ca	271	bmds->shared_base = block_mig_state->shared_base;
a55eb92c JK	272
	273	if (bmds->shared_base) {
	274	printf("Start migration for %s with shared base image\n",
c163b5ca	275	bs->device_name);
	276	} else {
	277	printf("Start full migration for %s\n", bs->device_name);
	278	}
a55eb92c	279
c163b5ca	280	/* insert at the end */
c163b5ca	281	pbmds = &block_mig_state->bmds_first;
a55eb92c	282	while (*pbmds != NULL) {
c163b5ca	283	pbmds = &(*pbmds)->next;
a55eb92c	284	}
c163b5ca	285	*pbmds = bmds;
a55eb92c	286
c163b5ca	287	blk_mig_save_dev_info(f, bmds);
c163b5ca	288	}
a55eb92c	289	}
c163b5ca	290	}
	291
	292	static int blk_mig_save_bulked_block(QEMUFile *f, int is_async)
	293	{
	294	BlkMigDevState *bmds;
	295
	296	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
a55eb92c JK	297	if (bmds->bulk_completed == 0) {
	298	if (is_async) {
	299	if (mig_read_device_bulk(f, bmds) == 1) {
c163b5ca	300	/* completed bulk section for this device */
	301	bmds->bulk_completed = 1;
	302	}
	303	} else {
a55eb92c	304	if (mig_save_device_bulk(f, bmds) == 1) {
c163b5ca	305	/* completed bulk section for this device */
	306	bmds->bulk_completed = 1;
	307	}
	308	}
	309	return 1;
	310	}
	311	}
a55eb92c	312
c163b5ca	313	/* we reached here means bulk is completed */
c163b5ca	314	block_mig_state->bulk_completed = 1;
a55eb92c	315
c163b5ca	316	return 0;
c163b5ca	317	}
	318
	319	#define MAX_NUM_BLOCKS 4
	320
	321	static void blk_mig_save_dirty_blocks(QEMUFile *f)
	322	{
	323	BlkMigDevState *bmds;
	324	uint8_t buf[BLOCK_SIZE];
	325	int64_t sector;
	326	int len;
a55eb92c JK	327
	328	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
	329	for (sector = 0; sector < bmds->cur_sector;) {
	330	if (bdrv_get_dirty(bmds->bs, sector)) {
	331	if (bdrv_read(bmds->bs, sector, buf,
6ea44308	332	BDRV_SECTORS_PER_DIRTY_CHUNK) < 0) {
a55eb92c	333	/* FIXME: add error handling */
c163b5ca	334	}
a55eb92c JK	335
a55eb92c JK	336	/* sector number and flags */
6ea44308	337	qemu_put_be64(f, (sector << BDRV_SECTOR_BITS)
a55eb92c JK	338	\| BLK_MIG_FLAG_DEVICE_BLOCK);
a55eb92c JK	339
c163b5ca	340	/* device name */
c163b5ca	341	len = strlen(bmds->bs->device_name);
c163b5ca	342	qemu_put_byte(f, len);
c163b5ca	343	qemu_put_buffer(f, (uint8_t *)bmds->bs->device_name, len);
a55eb92c	344
6ea44308	345	qemu_put_buffer(f, buf, BLOCK_SIZE);
a55eb92c JK	346
a55eb92c JK	347	bdrv_reset_dirty(bmds->bs, sector,
6ea44308	348	BDRV_SECTORS_PER_DIRTY_CHUNK);
a55eb92c	349	}
6ea44308	350	sector += BDRV_SECTORS_PER_DIRTY_CHUNK;
c163b5ca	351	}
c163b5ca	352	}
c163b5ca	353	}
	354
	355	static void flush_blks(QEMUFile* f)
	356	{
a55eb92c JK	357	BlkMigBlock blk, next;
	358
	359	dprintf("%s Enter submitted %d read_done %d transfered\n", __FUNCTION__,
c163b5ca	360	submitted, read_done, transfered);
a55eb92c JK	361
	362	for (blk = block_mig_state->first_blk;
	363	blk != NULL && !qemu_file_rate_limit(f);
	364	blk = next) {
c163b5ca	365	send_blk(f, blk);
a55eb92c JK	366
a55eb92c JK	367	next = blk->next;
c163b5ca	368	qemu_free(blk->buf);
c163b5ca	369	qemu_free(blk);
a55eb92c	370
c163b5ca	371	block_mig_state->read_done--;
	372	block_mig_state->transferred++;
	373	assert(block_mig_state->read_done >= 0);
	374	}
	375	block_mig_state->first_blk = blk;
a55eb92c JK	376
a55eb92c JK	377	if (block_mig_state->first_blk == NULL) {
c163b5ca	378	block_mig_state->last_blk = NULL;
	379	}
	380
a55eb92c JK	381	dprintf("%s Exit submitted %d read_done %d transferred%d\n", __FUNCTION__,
a55eb92c JK	382	block_mig_state->submitted, block_mig_state->read_done,
c163b5ca	383	block_mig_state->transferred);
c163b5ca	384	}
	385
	386	static int is_stage2_completed(void)
	387	{
	388	BlkMigDevState *bmds;
a55eb92c JK	389
a55eb92c JK	390	if (block_mig_state->submitted > 0) {
c163b5ca	391	return 0;
c163b5ca	392	}
a55eb92c	393
c163b5ca	394	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
a55eb92c	395	if (bmds->bulk_completed == 0) {
c163b5ca	396	return 0;
	397	}
	398	}
a55eb92c	399
c163b5ca	400	return 1;
	401	}
	402
	403	static int block_save_live(QEMUFile f, int stage, void opaque)
	404	{
a55eb92c	405	dprintf("Enter save live stage %d submitted %d transferred %d\n", stage,
c163b5ca	406	submitted, transferred);
a55eb92c JK	407
a55eb92c JK	408	if (block_mig_state->blk_enable != 1) {
c163b5ca	409	/* no need to migrate storage */
a55eb92c	410	qemu_put_be64(f, BLK_MIG_FLAG_EOS);
c163b5ca	411	return 1;
c163b5ca	412	}
a55eb92c JK	413
a55eb92c JK	414	if (stage == 1) {
c163b5ca	415	init_blk_migration(f);
a55eb92c	416
c163b5ca	417	/* start track dirty blocks */
c163b5ca	418	set_dirty_tracking(1);
c163b5ca	419	}
	420
	421	flush_blks(f);
a55eb92c	422
c163b5ca	423	/* control the rate of transfer */
a55eb92c JK	424	while ((block_mig_state->submitted +
	425	block_mig_state->read_done) * BLOCK_SIZE <
	426	qemu_file_get_rate_limit(f)) {
	427	if (blk_mig_save_bulked_block(f, 1) == 0) {
	428	/* no more bulk blocks for now */
c163b5ca	429	break;
a55eb92c	430	}
c163b5ca	431	}
a55eb92c	432
c163b5ca	433	flush_blks(f);
a55eb92c JK	434
	435	if (stage == 3) {
	436	while (blk_mig_save_bulked_block(f, 0) != 0) {
	437	/* empty */
	438	}
	439
c163b5ca	440	blk_mig_save_dirty_blocks(f);
a55eb92c	441
c163b5ca	442	/* stop track dirty blocks */
a55eb92c JK	443	set_dirty_tracking(0);
	444
	445	printf("\nBlock migration completed\n");
c163b5ca	446	}
a55eb92c JK	447
	448	qemu_put_be64(f, BLK_MIG_FLAG_EOS);
	449
c163b5ca	450	return ((stage == 2) && is_stage2_completed());
	451	}
	452
	453	static int block_load(QEMUFile f, void opaque, int version_id)
	454	{
	455	int len, flags;
	456	char device_name[256];
	457	int64_t addr;
	458	BlockDriverState *bs;
	459	uint8_t *buf;
a55eb92c	460
c163b5ca	461	buf = qemu_malloc(BLOCK_SIZE);
a55eb92c	462
c163b5ca	463	do {
c163b5ca	464	addr = qemu_get_be64(f);
a55eb92c	465
6ea44308 JK	466	flags = addr & ~BDRV_SECTOR_MASK;
6ea44308 JK	467	addr >>= BDRV_SECTOR_BITS;
a55eb92c JK	468
a55eb92c JK	469	if (flags & BLK_MIG_FLAG_DEVICE_BLOCK) {
c163b5ca	470	/* get device name */
c163b5ca	471	len = qemu_get_byte(f);
a55eb92c	472
c163b5ca	473	qemu_get_buffer(f, (uint8_t *)device_name, len);
c163b5ca	474	device_name[len] = '\0';
a55eb92c	475
c163b5ca	476	bs = bdrv_find(device_name);
a55eb92c JK	477
	478	qemu_get_buffer(f, buf, BLOCK_SIZE);
	479	if (bs != NULL) {
6ea44308	480	bdrv_write(bs, addr, buf, BDRV_SECTORS_PER_DIRTY_CHUNK);
c163b5ca	481	} else {
c163b5ca	482	printf("Error unknown block device %s\n", device_name);
a55eb92c	483	/* FIXME: add error handling */
c163b5ca	484	}
a55eb92c	485	} else if (!(flags & BLK_MIG_FLAG_EOS)) {
c163b5ca	486	printf("Unknown flags\n");
a55eb92c	487	/* FIXME: add error handling */
c163b5ca	488	}
a55eb92c JK	489	} while (!(flags & BLK_MIG_FLAG_EOS));
a55eb92c JK	490
c163b5ca	491	qemu_free(buf);
	492
	493	return 0;
	494	}
	495
	496	static void block_set_params(int blk_enable, int shared_base, void *opaque)
	497	{
	498	assert(opaque == block_mig_state);
	499
	500	block_mig_state->blk_enable = blk_enable;
	501	block_mig_state->shared_base = shared_base;
a55eb92c	502
c163b5ca	503	/* shared base means that blk_enable = 1 */
c163b5ca	504	block_mig_state->blk_enable \|= shared_base;
c163b5ca	505	}
	506
	507	void blk_mig_info(void)
	508	{
	509	BlockDriverState *bs;
a55eb92c	510
c163b5ca	511	for (bs = bdrv_first; bs != NULL; bs = bs->next) {
c163b5ca	512	printf("Device %s\n", bs->device_name);
a55eb92c JK	513	if (bs->type == BDRV_TYPE_HD) {
a55eb92c JK	514	printf("device %s format %s\n",
c163b5ca	515	bs->device_name, bs->drv->format_name);
	516	}
	517	}
	518	}
	519
	520	void blk_mig_init(void)
a55eb92c	521	{
c163b5ca	522	block_mig_state = qemu_mallocz(sizeof(BlkMigState));
c163b5ca	523
a55eb92c JK	524	register_savevm_live("block", 0, 1, block_set_params, block_save_live,
a55eb92c JK	525	NULL, block_load, block_mig_state);
c163b5ca	526	}