[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 SECTOR_BITS 9
#define SECTOR_SIZE (1 << SECTOR_BITS)
#define SECTOR_MASK ~(SECTOR_SIZE - 1);

#define BLOCK_SIZE (block_mig_state->sectors_per_block << 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;
    int sectors_per_block;
    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) >> 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 +=
            (block_mig_state->sectors_per_block * 10000);
    }

    /* we going to transfder BLOCK_SIZE any way even if it is not allocated */
    nr_sectors = block_mig_state->sectors_per_block;

    cur_sector &= ~((int64_t)block_mig_state->sectors_per_block -1);

    if (total_sectors - cur_sector < block_mig_state->sectors_per_block) {
        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 * 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 +=
            (block_mig_state->sectors_per_block * 10000);
    }

    cur_sector &= ~((int64_t)block_mig_state->sectors_per_block -1);

    /* we going to transfer BLOCK_SIZE any way even if it is not allocated */
    nr_sectors = block_mig_state->sectors_per_block;

    if (total_sectors - cur_sector < block_mig_state->sectors_per_block) {
        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 << 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 + block_mig_state->sectors_per_block;

    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 << 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) >> 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);
        }
    }

    block_mig_state->sectors_per_block = bdrv_get_sectors_per_chunk();
}

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,
                              block_mig_state->sectors_per_block) < 0) {
                    /* FIXME: add error handling */
                }

                /* sector number and flags */
                qemu_put_be64(f, (sector << 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_mig_state->sectors_per_block *
                                 SECTOR_SIZE));

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

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;

    block_mig_state->sectors_per_block = bdrv_get_sectors_per_chunk();
    buf = qemu_malloc(BLOCK_SIZE);

    do {
        addr = qemu_get_be64(f);

        flags = addr & ~SECTOR_MASK;
        addr &= SECTOR_MASK;

        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 >> SECTOR_BITS),
                           buf, block_mig_state->sectors_per_block);
            } 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
	20	#define SECTOR_BITS 9
	21	#define SECTOR_SIZE (1 << SECTOR_BITS)
	22	#define SECTOR_MASK ~(SECTOR_SIZE - 1);
	23
a55eb92c	24	#define BLOCK_SIZE (block_mig_state->sectors_per_block << SECTOR_BITS)
c163b5ca	25
	26	#define BLK_MIG_FLAG_DEVICE_BLOCK 0x01
	27	#define BLK_MIG_FLAG_EOS 0x02
	28
	29	#define MAX_IS_ALLOCATED_SEARCH 65536
	30	#define MAX_BLOCKS_READ 10000
	31	#define BLOCKS_READ_CHANGE 100
	32	#define INITIAL_BLOCKS_READ 100
	33
	34	//#define DEBUG_BLK_MIGRATION
	35
	36	#ifdef DEBUG_BLK_MIGRATION
a55eb92c	37	#define dprintf(fmt, ...) \
c163b5ca	38	do { printf("blk_migration: " fmt, ## __VA_ARGS__); } while (0)
c163b5ca	39	#else
a55eb92c	40	#define dprintf(fmt, ...) \
c163b5ca	41	do { } while (0)
	42	#endif
	43
a55eb92c JK	44	typedef struct BlkMigDevState {
	45	BlockDriverState *bs;
	46	int bulk_completed;
	47	int shared_base;
	48	struct BlkMigDevState *next;
	49	int64_t cur_sector;
	50	int64_t total_sectors;
	51	int64_t dirty;
	52	} BlkMigDevState;
	53
c163b5ca	54	typedef struct BlkMigBlock {
	55	uint8_t *buf;
	56	BlkMigDevState *bmds;
	57	int64_t sector;
	58	struct iovec iov;
	59	QEMUIOVector qiov;
	60	BlockDriverAIOCB *aiocb;
	61	int ret;
	62	struct BlkMigBlock *next;
	63	} BlkMigBlock;
	64
	65	typedef struct BlkMigState {
	66	int bulk_completed;
	67	int blk_enable;
	68	int shared_base;
	69	int no_dirty;
	70	QEMUFile *load_file;
	71	BlkMigDevState *bmds_first;
	72	int sectors_per_block;
	73	BlkMigBlock *first_blk;
	74	BlkMigBlock *last_blk;
	75	int submitted;
	76	int read_done;
	77	int transferred;
	78	int64_t print_completion;
	79	} BlkMigState;
	80
a55eb92c	81	static BlkMigState *block_mig_state = NULL;
c163b5ca	82
	83	static void blk_mig_read_cb(void *opaque, int ret)
	84	{
	85	BlkMigBlock *blk = opaque;
a55eb92c	86
c163b5ca	87	blk->ret = ret;
a55eb92c	88
c163b5ca	89	/* insert at the end */
a55eb92c	90	if (block_mig_state->last_blk == NULL) {
c163b5ca	91	block_mig_state->first_blk = blk;
	92	block_mig_state->last_blk = blk;
	93	} else {
	94	block_mig_state->last_blk->next = blk;
	95	block_mig_state->last_blk = blk;
	96	}
a55eb92c	97
c163b5ca	98	block_mig_state->submitted--;
	99	block_mig_state->read_done++;
	100	assert(block_mig_state->submitted >= 0);
c163b5ca	101	}
	102
	103	static int mig_read_device_bulk(QEMUFile f, BlkMigDevState bms)
a55eb92c	104	{
c163b5ca	105	int nr_sectors;
	106	int64_t total_sectors, cur_sector = 0;
	107	BlockDriverState *bs = bms->bs;
	108	BlkMigBlock *blk;
a55eb92c	109
c163b5ca	110	blk = qemu_malloc(sizeof(BlkMigBlock));
c163b5ca	111	blk->buf = qemu_malloc(BLOCK_SIZE);
a55eb92c	112
c163b5ca	113	cur_sector = bms->cur_sector;
c163b5ca	114	total_sectors = bdrv_getlength(bs) >> SECTOR_BITS;
a55eb92c JK	115
	116	if (bms->shared_base) {
	117	while (cur_sector < bms->total_sectors &&
	118	!bdrv_is_allocated(bms->bs, cur_sector,
	119	MAX_IS_ALLOCATED_SEARCH, &nr_sectors)) {
c163b5ca	120	cur_sector += nr_sectors;
	121	}
	122	}
a55eb92c JK	123
a55eb92c JK	124	if (cur_sector >= total_sectors) {
c163b5ca	125	bms->cur_sector = total_sectors;
	126	qemu_free(blk->buf);
	127	qemu_free(blk);
	128	return 1;
	129	}
a55eb92c JK	130
a55eb92c JK	131	if (cur_sector >= block_mig_state->print_completion) {
c163b5ca	132	printf("Completed %" PRId64 " %%\r", cur_sector * 100 / total_sectors);
c163b5ca	133	fflush(stdout);
a55eb92c	134	block_mig_state->print_completion +=
c163b5ca	135	(block_mig_state->sectors_per_block * 10000);
c163b5ca	136	}
a55eb92c	137
c163b5ca	138	/* we going to transfder BLOCK_SIZE any way even if it is not allocated */
	139	nr_sectors = block_mig_state->sectors_per_block;
	140
	141	cur_sector &= ~((int64_t)block_mig_state->sectors_per_block -1);
a55eb92c JK	142
a55eb92c JK	143	if (total_sectors - cur_sector < block_mig_state->sectors_per_block) {
c163b5ca	144	nr_sectors = (total_sectors - cur_sector);
c163b5ca	145	}
a55eb92c	146
c163b5ca	147	bms->cur_sector = cur_sector + nr_sectors;
	148	blk->sector = cur_sector;
	149	blk->bmds = bms;
	150	blk->next = NULL;
a55eb92c	151
c163b5ca	152	blk->iov.iov_base = blk->buf;
	153	blk->iov.iov_len = nr_sectors * SECTOR_SIZE;
	154	qemu_iovec_init_external(&blk->qiov, &blk->iov, 1);
a55eb92c	155
c163b5ca	156	blk->aiocb = bdrv_aio_readv(bs, cur_sector, &blk->qiov,
c163b5ca	157	nr_sectors, blk_mig_read_cb, blk);
a55eb92c JK	158
a55eb92c JK	159	if (!blk->aiocb) {
c163b5ca	160	printf("Error reading sector %" PRId64 "\n", cur_sector);
	161	qemu_free(blk->buf);
	162	qemu_free(blk);
	163	return 0;
	164	}
	165
	166	bdrv_reset_dirty(bms->bs, cur_sector, nr_sectors);
	167	block_mig_state->submitted++;
a55eb92c	168
c163b5ca	169	return (bms->cur_sector >= total_sectors);
	170	}
	171
	172	static int mig_save_device_bulk(QEMUFile f, BlkMigDevState bmds)
a55eb92c	173	{
c163b5ca	174	int len, nr_sectors;
	175	int64_t total_sectors = bmds->total_sectors, cur_sector = 0;
	176	uint8_t *tmp_buf = NULL;
	177	BlockDriverState *bs = bmds->bs;
	178
	179	tmp_buf = qemu_malloc(BLOCK_SIZE);
a55eb92c	180
c163b5ca	181	cur_sector = bmds->cur_sector;
a55eb92c JK	182
	183	if (bmds->shared_base) {
	184	while (cur_sector < bmds->total_sectors &&
	185	!bdrv_is_allocated(bmds->bs, cur_sector,
	186	MAX_IS_ALLOCATED_SEARCH, &nr_sectors)) {
c163b5ca	187	cur_sector += nr_sectors;
	188	}
	189	}
a55eb92c JK	190
a55eb92c JK	191	if (cur_sector >= total_sectors) {
c163b5ca	192	bmds->cur_sector = total_sectors;
	193	qemu_free(tmp_buf);
	194	return 1;
	195	}
a55eb92c JK	196
a55eb92c JK	197	if (cur_sector >= block_mig_state->print_completion) {
c163b5ca	198	printf("Completed %" PRId64 " %%\r", cur_sector * 100 / total_sectors);
c163b5ca	199	fflush(stdout);
a55eb92c	200	block_mig_state->print_completion +=
c163b5ca	201	(block_mig_state->sectors_per_block * 10000);
c163b5ca	202	}
a55eb92c	203
c163b5ca	204	cur_sector &= ~((int64_t)block_mig_state->sectors_per_block -1);
a55eb92c JK	205
a55eb92c JK	206	/* we going to transfer BLOCK_SIZE any way even if it is not allocated */
c163b5ca	207	nr_sectors = block_mig_state->sectors_per_block;
a55eb92c JK	208
a55eb92c JK	209	if (total_sectors - cur_sector < block_mig_state->sectors_per_block) {
c163b5ca	210	nr_sectors = (total_sectors - cur_sector);
c163b5ca	211	}
a55eb92c JK	212
a55eb92c JK	213	if (bdrv_read(bs, cur_sector, tmp_buf, nr_sectors) < 0) {
c163b5ca	214	printf("Error reading sector %" PRId64 "\n", cur_sector);
	215	}
	216
	217	bdrv_reset_dirty(bs, cur_sector, nr_sectors);
a55eb92c JK	218
	219	/* sector number and flags */
	220	qemu_put_be64(f, (cur_sector << SECTOR_BITS) \| BLK_MIG_FLAG_DEVICE_BLOCK);
	221
	222	/* device name */
c163b5ca	223	len = strlen(bs->device_name);
	224	qemu_put_byte(f, len);
	225	qemu_put_buffer(f, (uint8_t *)bs->device_name, len);
a55eb92c JK	226
	227	qemu_put_buffer(f, tmp_buf, BLOCK_SIZE);
	228
c163b5ca	229	bmds->cur_sector = cur_sector + block_mig_state->sectors_per_block;
a55eb92c	230
c163b5ca	231	qemu_free(tmp_buf);
a55eb92c	232
c163b5ca	233	return (bmds->cur_sector >= total_sectors);
	234	}
	235
	236	static void send_blk(QEMUFile f, BlkMigBlock blk)
	237	{
	238	int len;
a55eb92c JK	239
	240	/* sector number and flags */
	241	qemu_put_be64(f, (blk->sector << SECTOR_BITS) \| BLK_MIG_FLAG_DEVICE_BLOCK);
	242
	243	/* device name */
c163b5ca	244	len = strlen(blk->bmds->bs->device_name);
	245	qemu_put_byte(f, len);
	246	qemu_put_buffer(f, (uint8_t *)blk->bmds->bs->device_name, len);
a55eb92c JK	247
a55eb92c JK	248	qemu_put_buffer(f, blk->buf, BLOCK_SIZE);
c163b5ca	249	}
	250
	251	static void blk_mig_save_dev_info(QEMUFile f, BlkMigDevState bmds)
	252	{
	253	}
	254
	255	static void set_dirty_tracking(int enable)
	256	{
	257	BlkMigDevState *bmds;
a55eb92c JK	258	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
a55eb92c JK	259	bdrv_set_dirty_tracking(bmds->bs, enable);
c163b5ca	260	}
c163b5ca	261	}
	262
	263	static void init_blk_migration(QEMUFile *f)
	264	{
	265	BlkMigDevState *pbmds, bmds;
	266	BlockDriverState *bs;
a55eb92c	267
c163b5ca	268	for (bs = bdrv_first; bs != NULL; bs = bs->next) {
a55eb92c	269	if (bs->type == BDRV_TYPE_HD) {
c163b5ca	270	bmds = qemu_mallocz(sizeof(BlkMigDevState));
	271	bmds->bs = bs;
	272	bmds->bulk_completed = 0;
	273	bmds->total_sectors = bdrv_getlength(bs) >> SECTOR_BITS;
	274	bmds->shared_base = block_mig_state->shared_base;
a55eb92c JK	275
	276	if (bmds->shared_base) {
	277	printf("Start migration for %s with shared base image\n",
c163b5ca	278	bs->device_name);
	279	} else {
	280	printf("Start full migration for %s\n", bs->device_name);
	281	}
a55eb92c	282
c163b5ca	283	/* insert at the end */
c163b5ca	284	pbmds = &block_mig_state->bmds_first;
a55eb92c	285	while (*pbmds != NULL) {
c163b5ca	286	pbmds = &(*pbmds)->next;
a55eb92c	287	}
c163b5ca	288	*pbmds = bmds;
a55eb92c	289
c163b5ca	290	blk_mig_save_dev_info(f, bmds);
c163b5ca	291	}
a55eb92c JK	292	}
a55eb92c JK	293
c163b5ca	294	block_mig_state->sectors_per_block = bdrv_get_sectors_per_chunk();
c163b5ca	295	}
	296
	297	static int blk_mig_save_bulked_block(QEMUFile *f, int is_async)
	298	{
	299	BlkMigDevState *bmds;
	300
	301	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
a55eb92c JK	302	if (bmds->bulk_completed == 0) {
	303	if (is_async) {
	304	if (mig_read_device_bulk(f, bmds) == 1) {
c163b5ca	305	/* completed bulk section for this device */
	306	bmds->bulk_completed = 1;
	307	}
	308	} else {
a55eb92c	309	if (mig_save_device_bulk(f, bmds) == 1) {
c163b5ca	310	/* completed bulk section for this device */
	311	bmds->bulk_completed = 1;
	312	}
	313	}
	314	return 1;
	315	}
	316	}
a55eb92c	317
c163b5ca	318	/* we reached here means bulk is completed */
c163b5ca	319	block_mig_state->bulk_completed = 1;
a55eb92c	320
c163b5ca	321	return 0;
c163b5ca	322	}
	323
	324	#define MAX_NUM_BLOCKS 4
	325
	326	static void blk_mig_save_dirty_blocks(QEMUFile *f)
	327	{
	328	BlkMigDevState *bmds;
	329	uint8_t buf[BLOCK_SIZE];
	330	int64_t sector;
	331	int len;
a55eb92c JK	332
	333	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
	334	for (sector = 0; sector < bmds->cur_sector;) {
	335	if (bdrv_get_dirty(bmds->bs, sector)) {
	336	if (bdrv_read(bmds->bs, sector, buf,
	337	block_mig_state->sectors_per_block) < 0) {
	338	/* FIXME: add error handling */
c163b5ca	339	}
a55eb92c JK	340
	341	/* sector number and flags */
	342	qemu_put_be64(f, (sector << SECTOR_BITS)
	343	\| BLK_MIG_FLAG_DEVICE_BLOCK);
	344
c163b5ca	345	/* device name */
c163b5ca	346	len = strlen(bmds->bs->device_name);
c163b5ca	347	qemu_put_byte(f, len);
c163b5ca	348	qemu_put_buffer(f, (uint8_t *)bmds->bs->device_name, len);
a55eb92c JK	349
	350	qemu_put_buffer(f, buf,
	351	(block_mig_state->sectors_per_block *
c163b5ca	352	SECTOR_SIZE));
a55eb92c JK	353
a55eb92c JK	354	bdrv_reset_dirty(bmds->bs, sector,
c163b5ca	355	block_mig_state->sectors_per_block);
a55eb92c JK	356	}
a55eb92c JK	357	sector += block_mig_state->sectors_per_block;
c163b5ca	358	}
c163b5ca	359	}
c163b5ca	360	}
	361
	362	static void flush_blks(QEMUFile* f)
	363	{
a55eb92c JK	364	BlkMigBlock blk, next;
	365
	366	dprintf("%s Enter submitted %d read_done %d transfered\n", __FUNCTION__,
c163b5ca	367	submitted, read_done, transfered);
a55eb92c JK	368
	369	for (blk = block_mig_state->first_blk;
	370	blk != NULL && !qemu_file_rate_limit(f);
	371	blk = next) {
c163b5ca	372	send_blk(f, blk);
a55eb92c JK	373
a55eb92c JK	374	next = blk->next;
c163b5ca	375	qemu_free(blk->buf);
c163b5ca	376	qemu_free(blk);
a55eb92c	377
c163b5ca	378	block_mig_state->read_done--;
	379	block_mig_state->transferred++;
	380	assert(block_mig_state->read_done >= 0);
	381	}
	382	block_mig_state->first_blk = blk;
a55eb92c JK	383
a55eb92c JK	384	if (block_mig_state->first_blk == NULL) {
c163b5ca	385	block_mig_state->last_blk = NULL;
	386	}
	387
a55eb92c JK	388	dprintf("%s Exit submitted %d read_done %d transferred%d\n", __FUNCTION__,
a55eb92c JK	389	block_mig_state->submitted, block_mig_state->read_done,
c163b5ca	390	block_mig_state->transferred);
c163b5ca	391	}
	392
	393	static int is_stage2_completed(void)
	394	{
	395	BlkMigDevState *bmds;
a55eb92c JK	396
a55eb92c JK	397	if (block_mig_state->submitted > 0) {
c163b5ca	398	return 0;
c163b5ca	399	}
a55eb92c	400
c163b5ca	401	for (bmds = block_mig_state->bmds_first; bmds != NULL; bmds = bmds->next) {
a55eb92c	402	if (bmds->bulk_completed == 0) {
c163b5ca	403	return 0;
	404	}
	405	}
a55eb92c	406
c163b5ca	407	return 1;
	408	}
	409
	410	static int block_save_live(QEMUFile f, int stage, void opaque)
	411	{
a55eb92c	412	dprintf("Enter save live stage %d submitted %d transferred %d\n", stage,
c163b5ca	413	submitted, transferred);
a55eb92c JK	414
a55eb92c JK	415	if (block_mig_state->blk_enable != 1) {
c163b5ca	416	/* no need to migrate storage */
a55eb92c	417	qemu_put_be64(f, BLK_MIG_FLAG_EOS);
c163b5ca	418	return 1;
c163b5ca	419	}
a55eb92c JK	420
a55eb92c JK	421	if (stage == 1) {
c163b5ca	422	init_blk_migration(f);
a55eb92c	423
c163b5ca	424	/* start track dirty blocks */
c163b5ca	425	set_dirty_tracking(1);
c163b5ca	426	}
	427
	428	flush_blks(f);
a55eb92c	429
c163b5ca	430	/* control the rate of transfer */
a55eb92c JK	431	while ((block_mig_state->submitted +
	432	block_mig_state->read_done) * BLOCK_SIZE <
	433	qemu_file_get_rate_limit(f)) {
	434	if (blk_mig_save_bulked_block(f, 1) == 0) {
	435	/* no more bulk blocks for now */
c163b5ca	436	break;
a55eb92c	437	}
c163b5ca	438	}
a55eb92c	439
c163b5ca	440	flush_blks(f);
a55eb92c JK	441
	442	if (stage == 3) {
	443	while (blk_mig_save_bulked_block(f, 0) != 0) {
	444	/* empty */
	445	}
	446
c163b5ca	447	blk_mig_save_dirty_blocks(f);
a55eb92c	448
c163b5ca	449	/* stop track dirty blocks */
a55eb92c JK	450	set_dirty_tracking(0);
	451
	452	printf("\nBlock migration completed\n");
c163b5ca	453	}
a55eb92c JK	454
	455	qemu_put_be64(f, BLK_MIG_FLAG_EOS);
	456
c163b5ca	457	return ((stage == 2) && is_stage2_completed());
	458	}
	459
	460	static int block_load(QEMUFile f, void opaque, int version_id)
	461	{
	462	int len, flags;
	463	char device_name[256];
	464	int64_t addr;
	465	BlockDriverState *bs;
	466	uint8_t *buf;
a55eb92c	467
c163b5ca	468	block_mig_state->sectors_per_block = bdrv_get_sectors_per_chunk();
c163b5ca	469	buf = qemu_malloc(BLOCK_SIZE);
a55eb92c	470
c163b5ca	471	do {
c163b5ca	472	addr = qemu_get_be64(f);
a55eb92c	473
c163b5ca	474	flags = addr & ~SECTOR_MASK;
c163b5ca	475	addr &= SECTOR_MASK;
a55eb92c JK	476
a55eb92c JK	477	if (flags & BLK_MIG_FLAG_DEVICE_BLOCK) {
c163b5ca	478	/* get device name */
c163b5ca	479	len = qemu_get_byte(f);
a55eb92c	480
c163b5ca	481	qemu_get_buffer(f, (uint8_t *)device_name, len);
c163b5ca	482	device_name[len] = '\0';
a55eb92c	483
c163b5ca	484	bs = bdrv_find(device_name);
a55eb92c JK	485
	486	qemu_get_buffer(f, buf, BLOCK_SIZE);
	487	if (bs != NULL) {
	488	bdrv_write(bs, (addr >> SECTOR_BITS),
c163b5ca	489	buf, block_mig_state->sectors_per_block);
	490	} else {
	491	printf("Error unknown block device %s\n", device_name);
a55eb92c	492	/* FIXME: add error handling */
c163b5ca	493	}
a55eb92c	494	} else if (!(flags & BLK_MIG_FLAG_EOS)) {
c163b5ca	495	printf("Unknown flags\n");
a55eb92c	496	/* FIXME: add error handling */
c163b5ca	497	}
a55eb92c JK	498	} while (!(flags & BLK_MIG_FLAG_EOS));
a55eb92c JK	499
c163b5ca	500	qemu_free(buf);
	501
	502	return 0;
	503	}
	504
	505	static void block_set_params(int blk_enable, int shared_base, void *opaque)
	506	{
	507	assert(opaque == block_mig_state);
	508
	509	block_mig_state->blk_enable = blk_enable;
	510	block_mig_state->shared_base = shared_base;
a55eb92c	511
c163b5ca	512	/* shared base means that blk_enable = 1 */
c163b5ca	513	block_mig_state->blk_enable \|= shared_base;
c163b5ca	514	}
	515
	516	void blk_mig_info(void)
	517	{
	518	BlockDriverState *bs;
a55eb92c	519
c163b5ca	520	for (bs = bdrv_first; bs != NULL; bs = bs->next) {
c163b5ca	521	printf("Device %s\n", bs->device_name);
a55eb92c JK	522	if (bs->type == BDRV_TYPE_HD) {
a55eb92c JK	523	printf("device %s format %s\n",
c163b5ca	524	bs->device_name, bs->drv->format_name);
	525	}
	526	}
	527	}
	528
	529	void blk_mig_init(void)
a55eb92c	530	{
c163b5ca	531	block_mig_state = qemu_mallocz(sizeof(BlkMigState));
c163b5ca	532
a55eb92c JK	533	register_savevm_live("block", 0, 1, block_set_params, block_save_live,
a55eb92c JK	534	NULL, block_load, block_mig_state);
c163b5ca	535	}