2012-10-18 18:49:23 +04:00
|
|
|
/*
|
|
|
|
* Image mirroring
|
|
|
|
*
|
|
|
|
* Copyright Red Hat, Inc. 2012
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Paolo Bonzini <pbonzini@redhat.com>
|
|
|
|
*
|
|
|
|
* This work is licensed under the terms of the GNU LGPL, version 2 or later.
|
|
|
|
* See the COPYING.LIB file in the top-level directory.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "trace.h"
|
2012-12-17 21:19:44 +04:00
|
|
|
#include "block/blockjob.h"
|
|
|
|
#include "block/block_int.h"
|
2012-10-18 18:49:23 +04:00
|
|
|
#include "qemu/ratelimit.h"
|
2013-01-21 20:09:43 +04:00
|
|
|
#include "qemu/bitmap.h"
|
2012-10-18 18:49:23 +04:00
|
|
|
|
2013-01-22 12:03:14 +04:00
|
|
|
#define SLICE_TIME 100000000ULL /* ns */
|
|
|
|
#define MAX_IN_FLIGHT 16
|
|
|
|
|
|
|
|
/* The mirroring buffer is a list of granularity-sized chunks.
|
|
|
|
* Free chunks are organized in a list.
|
|
|
|
*/
|
|
|
|
typedef struct MirrorBuffer {
|
|
|
|
QSIMPLEQ_ENTRY(MirrorBuffer) next;
|
|
|
|
} MirrorBuffer;
|
2012-10-18 18:49:23 +04:00
|
|
|
|
|
|
|
typedef struct MirrorBlockJob {
|
|
|
|
BlockJob common;
|
|
|
|
RateLimit limit;
|
|
|
|
BlockDriverState *target;
|
2013-12-16 10:45:29 +04:00
|
|
|
BlockDriverState *base;
|
2013-12-16 10:45:30 +04:00
|
|
|
bool is_none_mode;
|
2012-10-18 18:49:28 +04:00
|
|
|
BlockdevOnError on_source_error, on_target_error;
|
2012-10-18 18:49:25 +04:00
|
|
|
bool synced;
|
|
|
|
bool should_complete;
|
2012-10-18 18:49:23 +04:00
|
|
|
int64_t sector_num;
|
2013-01-21 20:09:46 +04:00
|
|
|
int64_t granularity;
|
2013-01-21 20:09:43 +04:00
|
|
|
size_t buf_size;
|
|
|
|
unsigned long *cow_bitmap;
|
2013-11-13 14:29:43 +04:00
|
|
|
BdrvDirtyBitmap *dirty_bitmap;
|
2013-01-21 20:09:41 +04:00
|
|
|
HBitmapIter hbi;
|
2012-10-18 18:49:23 +04:00
|
|
|
uint8_t *buf;
|
2013-01-22 12:03:14 +04:00
|
|
|
QSIMPLEQ_HEAD(, MirrorBuffer) buf_free;
|
|
|
|
int buf_free_count;
|
2013-01-22 12:03:12 +04:00
|
|
|
|
2013-01-22 12:03:14 +04:00
|
|
|
unsigned long *in_flight_bitmap;
|
2013-01-22 12:03:12 +04:00
|
|
|
int in_flight;
|
|
|
|
int ret;
|
2012-10-18 18:49:23 +04:00
|
|
|
} MirrorBlockJob;
|
|
|
|
|
2013-01-22 12:03:12 +04:00
|
|
|
typedef struct MirrorOp {
|
|
|
|
MirrorBlockJob *s;
|
|
|
|
QEMUIOVector qiov;
|
|
|
|
int64_t sector_num;
|
|
|
|
int nb_sectors;
|
|
|
|
} MirrorOp;
|
|
|
|
|
2012-10-18 18:49:28 +04:00
|
|
|
static BlockErrorAction mirror_error_action(MirrorBlockJob *s, bool read,
|
|
|
|
int error)
|
|
|
|
{
|
|
|
|
s->synced = false;
|
|
|
|
if (read) {
|
|
|
|
return block_job_error_action(&s->common, s->common.bs,
|
|
|
|
s->on_source_error, true, error);
|
|
|
|
} else {
|
|
|
|
return block_job_error_action(&s->common, s->target,
|
|
|
|
s->on_target_error, false, error);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-01-22 12:03:12 +04:00
|
|
|
static void mirror_iteration_done(MirrorOp *op, int ret)
|
|
|
|
{
|
|
|
|
MirrorBlockJob *s = op->s;
|
2013-01-22 12:03:14 +04:00
|
|
|
struct iovec *iov;
|
2013-01-22 12:03:12 +04:00
|
|
|
int64_t chunk_num;
|
2013-01-22 12:03:14 +04:00
|
|
|
int i, nb_chunks, sectors_per_chunk;
|
2013-01-22 12:03:12 +04:00
|
|
|
|
|
|
|
trace_mirror_iteration_done(s, op->sector_num, op->nb_sectors, ret);
|
|
|
|
|
|
|
|
s->in_flight--;
|
2013-01-22 12:03:14 +04:00
|
|
|
iov = op->qiov.iov;
|
|
|
|
for (i = 0; i < op->qiov.niov; i++) {
|
|
|
|
MirrorBuffer *buf = (MirrorBuffer *) iov[i].iov_base;
|
|
|
|
QSIMPLEQ_INSERT_TAIL(&s->buf_free, buf, next);
|
|
|
|
s->buf_free_count++;
|
|
|
|
}
|
|
|
|
|
2013-01-22 12:03:12 +04:00
|
|
|
sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
|
|
|
|
chunk_num = op->sector_num / sectors_per_chunk;
|
|
|
|
nb_chunks = op->nb_sectors / sectors_per_chunk;
|
2013-01-22 12:03:14 +04:00
|
|
|
bitmap_clear(s->in_flight_bitmap, chunk_num, nb_chunks);
|
2013-01-22 12:03:12 +04:00
|
|
|
if (s->cow_bitmap && ret >= 0) {
|
|
|
|
bitmap_set(s->cow_bitmap, chunk_num, nb_chunks);
|
|
|
|
}
|
|
|
|
|
|
|
|
g_slice_free(MirrorOp, op);
|
|
|
|
qemu_coroutine_enter(s->common.co, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mirror_write_complete(void *opaque, int ret)
|
|
|
|
{
|
|
|
|
MirrorOp *op = opaque;
|
|
|
|
MirrorBlockJob *s = op->s;
|
|
|
|
if (ret < 0) {
|
|
|
|
BlockDriverState *source = s->common.bs;
|
|
|
|
BlockErrorAction action;
|
|
|
|
|
|
|
|
bdrv_set_dirty(source, op->sector_num, op->nb_sectors);
|
|
|
|
action = mirror_error_action(s, false, -ret);
|
|
|
|
if (action == BDRV_ACTION_REPORT && s->ret >= 0) {
|
|
|
|
s->ret = ret;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
mirror_iteration_done(op, ret);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mirror_read_complete(void *opaque, int ret)
|
|
|
|
{
|
|
|
|
MirrorOp *op = opaque;
|
|
|
|
MirrorBlockJob *s = op->s;
|
|
|
|
if (ret < 0) {
|
|
|
|
BlockDriverState *source = s->common.bs;
|
|
|
|
BlockErrorAction action;
|
|
|
|
|
|
|
|
bdrv_set_dirty(source, op->sector_num, op->nb_sectors);
|
|
|
|
action = mirror_error_action(s, true, -ret);
|
|
|
|
if (action == BDRV_ACTION_REPORT && s->ret >= 0) {
|
|
|
|
s->ret = ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
mirror_iteration_done(op, ret);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
bdrv_aio_writev(s->target, op->sector_num, &op->qiov, op->nb_sectors,
|
|
|
|
mirror_write_complete, op);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void coroutine_fn mirror_iteration(MirrorBlockJob *s)
|
2012-10-18 18:49:23 +04:00
|
|
|
{
|
|
|
|
BlockDriverState *source = s->common.bs;
|
2013-01-22 12:03:14 +04:00
|
|
|
int nb_sectors, sectors_per_chunk, nb_chunks;
|
2013-01-22 12:03:15 +04:00
|
|
|
int64_t end, sector_num, next_chunk, next_sector, hbitmap_next_sector;
|
2013-01-22 12:03:12 +04:00
|
|
|
MirrorOp *op;
|
2012-10-18 18:49:23 +04:00
|
|
|
|
2013-01-21 20:09:41 +04:00
|
|
|
s->sector_num = hbitmap_iter_next(&s->hbi);
|
|
|
|
if (s->sector_num < 0) {
|
2013-11-13 14:29:43 +04:00
|
|
|
bdrv_dirty_iter_init(source, s->dirty_bitmap, &s->hbi);
|
2013-01-21 20:09:41 +04:00
|
|
|
s->sector_num = hbitmap_iter_next(&s->hbi);
|
2013-11-13 14:29:43 +04:00
|
|
|
trace_mirror_restart_iter(s,
|
|
|
|
bdrv_get_dirty_count(source, s->dirty_bitmap));
|
2013-01-21 20:09:41 +04:00
|
|
|
assert(s->sector_num >= 0);
|
|
|
|
}
|
|
|
|
|
2013-01-22 12:03:14 +04:00
|
|
|
hbitmap_next_sector = s->sector_num;
|
2013-01-22 12:03:15 +04:00
|
|
|
sector_num = s->sector_num;
|
|
|
|
sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
|
|
|
|
end = s->common.len >> BDRV_SECTOR_BITS;
|
2013-01-22 12:03:14 +04:00
|
|
|
|
2013-01-22 12:03:15 +04:00
|
|
|
/* Extend the QEMUIOVector to include all adjacent blocks that will
|
|
|
|
* be copied in this operation.
|
2013-01-21 20:09:43 +04:00
|
|
|
*
|
2013-01-22 12:03:15 +04:00
|
|
|
* We have to do this if we have no backing file yet in the destination,
|
|
|
|
* and the cluster size is very large. Then we need to do COW ourselves.
|
|
|
|
* The first time a cluster is copied, copy it entirely. Note that,
|
|
|
|
* because both the granularity and the cluster size are powers of two,
|
|
|
|
* the number of sectors to copy cannot exceed one cluster.
|
|
|
|
*
|
|
|
|
* We also want to extend the QEMUIOVector to include more adjacent
|
|
|
|
* dirty blocks if possible, to limit the number of I/O operations and
|
|
|
|
* run efficiently even with a small granularity.
|
2013-01-21 20:09:43 +04:00
|
|
|
*/
|
2013-01-22 12:03:15 +04:00
|
|
|
nb_chunks = 0;
|
|
|
|
nb_sectors = 0;
|
|
|
|
next_sector = sector_num;
|
|
|
|
next_chunk = sector_num / sectors_per_chunk;
|
2013-01-22 12:03:14 +04:00
|
|
|
|
|
|
|
/* Wait for I/O to this cluster (from a previous iteration) to be done. */
|
2013-01-22 12:03:15 +04:00
|
|
|
while (test_bit(next_chunk, s->in_flight_bitmap)) {
|
2013-01-22 12:03:14 +04:00
|
|
|
trace_mirror_yield_in_flight(s, sector_num, s->in_flight);
|
|
|
|
qemu_coroutine_yield();
|
2013-01-21 20:09:43 +04:00
|
|
|
}
|
|
|
|
|
2013-01-22 12:03:15 +04:00
|
|
|
do {
|
|
|
|
int added_sectors, added_chunks;
|
|
|
|
|
2013-11-13 14:29:43 +04:00
|
|
|
if (!bdrv_get_dirty(source, s->dirty_bitmap, next_sector) ||
|
2013-01-22 12:03:15 +04:00
|
|
|
test_bit(next_chunk, s->in_flight_bitmap)) {
|
|
|
|
assert(nb_sectors > 0);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
added_sectors = sectors_per_chunk;
|
|
|
|
if (s->cow_bitmap && !test_bit(next_chunk, s->cow_bitmap)) {
|
|
|
|
bdrv_round_to_clusters(s->target,
|
|
|
|
next_sector, added_sectors,
|
|
|
|
&next_sector, &added_sectors);
|
|
|
|
|
|
|
|
/* On the first iteration, the rounding may make us copy
|
|
|
|
* sectors before the first dirty one.
|
|
|
|
*/
|
|
|
|
if (next_sector < sector_num) {
|
|
|
|
assert(nb_sectors == 0);
|
|
|
|
sector_num = next_sector;
|
|
|
|
next_chunk = next_sector / sectors_per_chunk;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
added_sectors = MIN(added_sectors, end - (sector_num + nb_sectors));
|
|
|
|
added_chunks = (added_sectors + sectors_per_chunk - 1) / sectors_per_chunk;
|
|
|
|
|
|
|
|
/* When doing COW, it may happen that there is not enough space for
|
|
|
|
* a full cluster. Wait if that is the case.
|
|
|
|
*/
|
|
|
|
while (nb_chunks == 0 && s->buf_free_count < added_chunks) {
|
|
|
|
trace_mirror_yield_buf_busy(s, nb_chunks, s->in_flight);
|
|
|
|
qemu_coroutine_yield();
|
|
|
|
}
|
|
|
|
if (s->buf_free_count < nb_chunks + added_chunks) {
|
|
|
|
trace_mirror_break_buf_busy(s, nb_chunks, s->in_flight);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* We have enough free space to copy these sectors. */
|
|
|
|
bitmap_set(s->in_flight_bitmap, next_chunk, added_chunks);
|
2013-01-22 12:03:14 +04:00
|
|
|
|
2013-01-22 12:03:15 +04:00
|
|
|
nb_sectors += added_sectors;
|
|
|
|
nb_chunks += added_chunks;
|
|
|
|
next_sector += added_sectors;
|
|
|
|
next_chunk += added_chunks;
|
|
|
|
} while (next_sector < end);
|
2013-01-22 12:03:12 +04:00
|
|
|
|
|
|
|
/* Allocate a MirrorOp that is used as an AIO callback. */
|
|
|
|
op = g_slice_new(MirrorOp);
|
|
|
|
op->s = s;
|
|
|
|
op->sector_num = sector_num;
|
|
|
|
op->nb_sectors = nb_sectors;
|
2013-01-22 12:03:14 +04:00
|
|
|
|
|
|
|
/* Now make a QEMUIOVector taking enough granularity-sized chunks
|
|
|
|
* from s->buf_free.
|
|
|
|
*/
|
|
|
|
qemu_iovec_init(&op->qiov, nb_chunks);
|
|
|
|
next_sector = sector_num;
|
|
|
|
while (nb_chunks-- > 0) {
|
|
|
|
MirrorBuffer *buf = QSIMPLEQ_FIRST(&s->buf_free);
|
|
|
|
QSIMPLEQ_REMOVE_HEAD(&s->buf_free, next);
|
|
|
|
s->buf_free_count--;
|
|
|
|
qemu_iovec_add(&op->qiov, buf, s->granularity);
|
|
|
|
|
|
|
|
/* Advance the HBitmapIter in parallel, so that we do not examine
|
|
|
|
* the same sector twice.
|
|
|
|
*/
|
2013-11-13 14:29:43 +04:00
|
|
|
if (next_sector > hbitmap_next_sector
|
|
|
|
&& bdrv_get_dirty(source, s->dirty_bitmap, next_sector)) {
|
2013-01-22 12:03:14 +04:00
|
|
|
hbitmap_next_sector = hbitmap_iter_next(&s->hbi);
|
|
|
|
}
|
|
|
|
|
|
|
|
next_sector += sectors_per_chunk;
|
|
|
|
}
|
2013-01-22 12:03:12 +04:00
|
|
|
|
2013-01-21 20:09:43 +04:00
|
|
|
bdrv_reset_dirty(source, sector_num, nb_sectors);
|
2012-10-18 18:49:23 +04:00
|
|
|
|
|
|
|
/* Copy the dirty cluster. */
|
2013-01-22 12:03:12 +04:00
|
|
|
s->in_flight++;
|
2013-01-21 20:09:43 +04:00
|
|
|
trace_mirror_one_iteration(s, sector_num, nb_sectors);
|
2013-01-22 12:03:12 +04:00
|
|
|
bdrv_aio_readv(source, sector_num, &op->qiov, nb_sectors,
|
|
|
|
mirror_read_complete, op);
|
|
|
|
}
|
2012-10-18 18:49:28 +04:00
|
|
|
|
2013-01-22 12:03:14 +04:00
|
|
|
static void mirror_free_init(MirrorBlockJob *s)
|
|
|
|
{
|
|
|
|
int granularity = s->granularity;
|
|
|
|
size_t buf_size = s->buf_size;
|
|
|
|
uint8_t *buf = s->buf;
|
|
|
|
|
|
|
|
assert(s->buf_free_count == 0);
|
|
|
|
QSIMPLEQ_INIT(&s->buf_free);
|
|
|
|
while (buf_size != 0) {
|
|
|
|
MirrorBuffer *cur = (MirrorBuffer *)buf;
|
|
|
|
QSIMPLEQ_INSERT_TAIL(&s->buf_free, cur, next);
|
|
|
|
s->buf_free_count++;
|
|
|
|
buf_size -= granularity;
|
|
|
|
buf += granularity;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-01-22 12:03:12 +04:00
|
|
|
static void mirror_drain(MirrorBlockJob *s)
|
|
|
|
{
|
|
|
|
while (s->in_flight > 0) {
|
|
|
|
qemu_coroutine_yield();
|
|
|
|
}
|
2012-10-18 18:49:23 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
static void coroutine_fn mirror_run(void *opaque)
|
|
|
|
{
|
|
|
|
MirrorBlockJob *s = opaque;
|
|
|
|
BlockDriverState *bs = s->common.bs;
|
2013-01-21 20:09:46 +04:00
|
|
|
int64_t sector_num, end, sectors_per_chunk, length;
|
2013-01-22 12:03:12 +04:00
|
|
|
uint64_t last_pause_ns;
|
2013-01-21 20:09:43 +04:00
|
|
|
BlockDriverInfo bdi;
|
|
|
|
char backing_filename[1024];
|
2012-10-18 18:49:23 +04:00
|
|
|
int ret = 0;
|
|
|
|
int n;
|
|
|
|
|
|
|
|
if (block_job_is_cancelled(&s->common)) {
|
|
|
|
goto immediate_exit;
|
|
|
|
}
|
|
|
|
|
|
|
|
s->common.len = bdrv_getlength(bs);
|
2013-01-22 18:01:11 +04:00
|
|
|
if (s->common.len <= 0) {
|
2012-10-18 18:49:23 +04:00
|
|
|
block_job_completed(&s->common, s->common.len);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2013-01-22 12:03:14 +04:00
|
|
|
length = (bdrv_getlength(bs) + s->granularity - 1) / s->granularity;
|
|
|
|
s->in_flight_bitmap = bitmap_new(length);
|
|
|
|
|
2013-01-21 20:09:43 +04:00
|
|
|
/* If we have no backing file yet in the destination, we cannot let
|
|
|
|
* the destination do COW. Instead, we copy sectors around the
|
|
|
|
* dirty data if needed. We need a bitmap to do that.
|
|
|
|
*/
|
|
|
|
bdrv_get_backing_filename(s->target, backing_filename,
|
|
|
|
sizeof(backing_filename));
|
|
|
|
if (backing_filename[0] && !s->target->backing_hd) {
|
|
|
|
bdrv_get_info(s->target, &bdi);
|
2013-01-21 20:09:46 +04:00
|
|
|
if (s->granularity < bdi.cluster_size) {
|
2013-01-22 12:03:13 +04:00
|
|
|
s->buf_size = MAX(s->buf_size, bdi.cluster_size);
|
2013-01-21 20:09:43 +04:00
|
|
|
s->cow_bitmap = bitmap_new(length);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-10-18 18:49:23 +04:00
|
|
|
end = s->common.len >> BDRV_SECTOR_BITS;
|
2013-01-21 20:09:43 +04:00
|
|
|
s->buf = qemu_blockalign(bs, s->buf_size);
|
2013-01-21 20:09:46 +04:00
|
|
|
sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
|
2013-01-22 12:03:14 +04:00
|
|
|
mirror_free_init(s);
|
2012-10-18 18:49:23 +04:00
|
|
|
|
2013-12-16 10:45:30 +04:00
|
|
|
if (!s->is_none_mode) {
|
2012-10-18 18:49:23 +04:00
|
|
|
/* First part, loop on the sectors and initialize the dirty bitmap. */
|
2013-12-16 10:45:29 +04:00
|
|
|
BlockDriverState *base = s->base;
|
2012-10-18 18:49:23 +04:00
|
|
|
for (sector_num = 0; sector_num < end; ) {
|
2013-01-21 20:09:46 +04:00
|
|
|
int64_t next = (sector_num | (sectors_per_chunk - 1)) + 1;
|
2013-09-04 21:00:24 +04:00
|
|
|
ret = bdrv_is_allocated_above(bs, base,
|
|
|
|
sector_num, next - sector_num, &n);
|
2012-10-18 18:49:23 +04:00
|
|
|
|
|
|
|
if (ret < 0) {
|
|
|
|
goto immediate_exit;
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(n > 0);
|
|
|
|
if (ret == 1) {
|
|
|
|
bdrv_set_dirty(bs, sector_num, n);
|
|
|
|
sector_num = next;
|
|
|
|
} else {
|
|
|
|
sector_num += n;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-11-13 14:29:43 +04:00
|
|
|
bdrv_dirty_iter_init(bs, s->dirty_bitmap, &s->hbi);
|
2013-08-21 19:03:08 +04:00
|
|
|
last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
|
2012-10-18 18:49:23 +04:00
|
|
|
for (;;) {
|
|
|
|
uint64_t delay_ns;
|
|
|
|
int64_t cnt;
|
|
|
|
bool should_complete;
|
|
|
|
|
2013-01-22 12:03:12 +04:00
|
|
|
if (s->ret < 0) {
|
|
|
|
ret = s->ret;
|
|
|
|
goto immediate_exit;
|
|
|
|
}
|
|
|
|
|
2013-11-13 14:29:43 +04:00
|
|
|
cnt = bdrv_get_dirty_count(bs, s->dirty_bitmap);
|
2013-01-22 12:03:12 +04:00
|
|
|
|
|
|
|
/* Note that even when no rate limit is applied we need to yield
|
|
|
|
* periodically with no pending I/O so that qemu_aio_flush() returns.
|
|
|
|
* We do so every SLICE_TIME nanoseconds, or when there is an error,
|
|
|
|
* or when the source is clean, whichever comes first.
|
|
|
|
*/
|
2013-08-21 19:03:08 +04:00
|
|
|
if (qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - last_pause_ns < SLICE_TIME &&
|
2013-01-22 12:03:12 +04:00
|
|
|
s->common.iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
|
2013-01-22 12:03:14 +04:00
|
|
|
if (s->in_flight == MAX_IN_FLIGHT || s->buf_free_count == 0 ||
|
|
|
|
(cnt == 0 && s->in_flight > 0)) {
|
|
|
|
trace_mirror_yield(s, s->in_flight, s->buf_free_count, cnt);
|
2013-01-22 12:03:12 +04:00
|
|
|
qemu_coroutine_yield();
|
|
|
|
continue;
|
|
|
|
} else if (cnt != 0) {
|
|
|
|
mirror_iteration(s);
|
|
|
|
continue;
|
2012-10-18 18:49:23 +04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
should_complete = false;
|
2013-01-22 12:03:12 +04:00
|
|
|
if (s->in_flight == 0 && cnt == 0) {
|
2012-10-18 18:49:23 +04:00
|
|
|
trace_mirror_before_flush(s);
|
|
|
|
ret = bdrv_flush(s->target);
|
|
|
|
if (ret < 0) {
|
2012-10-18 18:49:28 +04:00
|
|
|
if (mirror_error_action(s, false, -ret) == BDRV_ACTION_REPORT) {
|
|
|
|
goto immediate_exit;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
/* We're out of the streaming phase. From now on, if the job
|
|
|
|
* is cancelled we will actually complete all pending I/O and
|
|
|
|
* report completion. This way, block-job-cancel will leave
|
|
|
|
* the target in a consistent state.
|
|
|
|
*/
|
|
|
|
s->common.offset = end * BDRV_SECTOR_SIZE;
|
|
|
|
if (!s->synced) {
|
|
|
|
block_job_ready(&s->common);
|
|
|
|
s->synced = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
should_complete = s->should_complete ||
|
|
|
|
block_job_is_cancelled(&s->common);
|
2013-11-13 14:29:43 +04:00
|
|
|
cnt = bdrv_get_dirty_count(bs, s->dirty_bitmap);
|
2012-10-18 18:49:25 +04:00
|
|
|
}
|
2012-10-18 18:49:23 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
if (cnt == 0 && should_complete) {
|
|
|
|
/* The dirty bitmap is not updated while operations are pending.
|
|
|
|
* If we're about to exit, wait for pending operations before
|
|
|
|
* calling bdrv_get_dirty_count(bs), or we may exit while the
|
|
|
|
* source has dirty data to copy!
|
|
|
|
*
|
|
|
|
* Note that I/O can be submitted by the guest while
|
|
|
|
* mirror_populate runs.
|
|
|
|
*/
|
|
|
|
trace_mirror_before_drain(s, cnt);
|
|
|
|
bdrv_drain_all();
|
2013-11-13 14:29:43 +04:00
|
|
|
cnt = bdrv_get_dirty_count(bs, s->dirty_bitmap);
|
2012-10-18 18:49:23 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
2012-10-18 18:49:25 +04:00
|
|
|
trace_mirror_before_sleep(s, cnt, s->synced);
|
|
|
|
if (!s->synced) {
|
2012-10-18 18:49:23 +04:00
|
|
|
/* Publish progress */
|
2013-01-21 20:09:44 +04:00
|
|
|
s->common.offset = (end - cnt) * BDRV_SECTOR_SIZE;
|
2012-10-18 18:49:23 +04:00
|
|
|
|
|
|
|
if (s->common.speed) {
|
2013-01-21 20:09:46 +04:00
|
|
|
delay_ns = ratelimit_calculate_delay(&s->limit, sectors_per_chunk);
|
2012-10-18 18:49:23 +04:00
|
|
|
} else {
|
|
|
|
delay_ns = 0;
|
|
|
|
}
|
|
|
|
|
2013-08-21 19:03:05 +04:00
|
|
|
block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
|
2012-10-18 18:49:23 +04:00
|
|
|
if (block_job_is_cancelled(&s->common)) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else if (!should_complete) {
|
2013-01-22 12:03:12 +04:00
|
|
|
delay_ns = (s->in_flight == 0 && cnt == 0 ? SLICE_TIME : 0);
|
2013-08-21 19:03:05 +04:00
|
|
|
block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
|
2012-10-18 18:49:23 +04:00
|
|
|
} else if (cnt == 0) {
|
|
|
|
/* The two disks are in sync. Exit and report successful
|
|
|
|
* completion.
|
|
|
|
*/
|
|
|
|
assert(QLIST_EMPTY(&bs->tracked_requests));
|
|
|
|
s->common.cancelled = false;
|
|
|
|
break;
|
|
|
|
}
|
2013-08-21 19:03:08 +04:00
|
|
|
last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
|
2012-10-18 18:49:23 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
immediate_exit:
|
2013-01-22 12:03:12 +04:00
|
|
|
if (s->in_flight > 0) {
|
|
|
|
/* We get here only if something went wrong. Either the job failed,
|
|
|
|
* or it was cancelled prematurely so that we do not guarantee that
|
|
|
|
* the target is a copy of the source.
|
|
|
|
*/
|
|
|
|
assert(ret < 0 || (!s->synced && block_job_is_cancelled(&s->common)));
|
|
|
|
mirror_drain(s);
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(s->in_flight == 0);
|
2013-01-15 18:29:10 +04:00
|
|
|
qemu_vfree(s->buf);
|
2013-01-21 20:09:43 +04:00
|
|
|
g_free(s->cow_bitmap);
|
2013-01-22 12:03:14 +04:00
|
|
|
g_free(s->in_flight_bitmap);
|
2013-11-13 14:29:43 +04:00
|
|
|
bdrv_release_dirty_bitmap(bs, s->dirty_bitmap);
|
2012-10-18 18:49:28 +04:00
|
|
|
bdrv_iostatus_disable(s->target);
|
2012-10-18 18:49:25 +04:00
|
|
|
if (s->should_complete && ret == 0) {
|
|
|
|
if (bdrv_get_flags(s->target) != bdrv_get_flags(s->common.bs)) {
|
|
|
|
bdrv_reopen(s->target, bdrv_get_flags(s->common.bs), NULL);
|
|
|
|
}
|
|
|
|
bdrv_swap(s->target, s->common.bs);
|
2013-12-16 10:45:31 +04:00
|
|
|
if (s->common.driver->job_type == BLOCK_JOB_TYPE_COMMIT) {
|
|
|
|
/* drop the bs loop chain formed by the swap: break the loop then
|
|
|
|
* trigger the unref from the top one */
|
|
|
|
BlockDriverState *p = s->base->backing_hd;
|
|
|
|
s->base->backing_hd = NULL;
|
|
|
|
bdrv_unref(p);
|
|
|
|
}
|
2012-10-18 18:49:25 +04:00
|
|
|
}
|
2013-08-23 05:14:47 +04:00
|
|
|
bdrv_unref(s->target);
|
2012-10-18 18:49:23 +04:00
|
|
|
block_job_completed(&s->common, ret);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mirror_set_speed(BlockJob *job, int64_t speed, Error **errp)
|
|
|
|
{
|
|
|
|
MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
|
|
|
|
|
|
|
|
if (speed < 0) {
|
|
|
|
error_set(errp, QERR_INVALID_PARAMETER, "speed");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
ratelimit_set_speed(&s->limit, speed / BDRV_SECTOR_SIZE, SLICE_TIME);
|
|
|
|
}
|
|
|
|
|
2012-10-18 18:49:28 +04:00
|
|
|
static void mirror_iostatus_reset(BlockJob *job)
|
|
|
|
{
|
|
|
|
MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
|
|
|
|
|
|
|
|
bdrv_iostatus_reset(s->target);
|
|
|
|
}
|
|
|
|
|
2012-10-18 18:49:25 +04:00
|
|
|
static void mirror_complete(BlockJob *job, Error **errp)
|
|
|
|
{
|
|
|
|
MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
|
2013-09-05 16:45:29 +04:00
|
|
|
Error *local_err = NULL;
|
2012-10-18 18:49:25 +04:00
|
|
|
int ret;
|
|
|
|
|
2013-09-05 16:45:29 +04:00
|
|
|
ret = bdrv_open_backing_file(s->target, NULL, &local_err);
|
2012-10-18 18:49:25 +04:00
|
|
|
if (ret < 0) {
|
|
|
|
char backing_filename[PATH_MAX];
|
|
|
|
bdrv_get_full_backing_filename(s->target, backing_filename,
|
|
|
|
sizeof(backing_filename));
|
2013-09-05 16:45:29 +04:00
|
|
|
error_propagate(errp, local_err);
|
2012-10-18 18:49:25 +04:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
if (!s->synced) {
|
|
|
|
error_set(errp, QERR_BLOCK_JOB_NOT_READY, job->bs->device_name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
s->should_complete = true;
|
|
|
|
block_job_resume(job);
|
|
|
|
}
|
|
|
|
|
2013-10-08 13:29:38 +04:00
|
|
|
static const BlockJobDriver mirror_job_driver = {
|
2012-10-18 18:49:23 +04:00
|
|
|
.instance_size = sizeof(MirrorBlockJob),
|
2013-10-08 13:29:40 +04:00
|
|
|
.job_type = BLOCK_JOB_TYPE_MIRROR,
|
2012-10-18 18:49:23 +04:00
|
|
|
.set_speed = mirror_set_speed,
|
2012-10-18 18:49:28 +04:00
|
|
|
.iostatus_reset= mirror_iostatus_reset,
|
2012-10-18 18:49:25 +04:00
|
|
|
.complete = mirror_complete,
|
2012-10-18 18:49:23 +04:00
|
|
|
};
|
|
|
|
|
2013-12-16 10:45:30 +04:00
|
|
|
static const BlockJobDriver commit_active_job_driver = {
|
|
|
|
.instance_size = sizeof(MirrorBlockJob),
|
|
|
|
.job_type = BLOCK_JOB_TYPE_COMMIT,
|
|
|
|
.set_speed = mirror_set_speed,
|
|
|
|
.iostatus_reset
|
|
|
|
= mirror_iostatus_reset,
|
|
|
|
.complete = mirror_complete,
|
|
|
|
};
|
|
|
|
|
|
|
|
static void mirror_start_job(BlockDriverState *bs, BlockDriverState *target,
|
|
|
|
int64_t speed, int64_t granularity,
|
|
|
|
int64_t buf_size,
|
|
|
|
BlockdevOnError on_source_error,
|
|
|
|
BlockdevOnError on_target_error,
|
|
|
|
BlockDriverCompletionFunc *cb,
|
|
|
|
void *opaque, Error **errp,
|
|
|
|
const BlockJobDriver *driver,
|
|
|
|
bool is_none_mode, BlockDriverState *base)
|
2012-10-18 18:49:23 +04:00
|
|
|
{
|
|
|
|
MirrorBlockJob *s;
|
|
|
|
|
2013-01-21 20:09:46 +04:00
|
|
|
if (granularity == 0) {
|
|
|
|
/* Choose the default granularity based on the target file's cluster
|
|
|
|
* size, clamped between 4k and 64k. */
|
|
|
|
BlockDriverInfo bdi;
|
|
|
|
if (bdrv_get_info(target, &bdi) >= 0 && bdi.cluster_size != 0) {
|
|
|
|
granularity = MAX(4096, bdi.cluster_size);
|
|
|
|
granularity = MIN(65536, granularity);
|
|
|
|
} else {
|
|
|
|
granularity = 65536;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
assert ((granularity & (granularity - 1)) == 0);
|
|
|
|
|
2012-10-18 18:49:28 +04:00
|
|
|
if ((on_source_error == BLOCKDEV_ON_ERROR_STOP ||
|
|
|
|
on_source_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
|
|
|
|
!bdrv_iostatus_is_enabled(bs)) {
|
|
|
|
error_set(errp, QERR_INVALID_PARAMETER, "on-source-error");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2013-12-16 10:45:29 +04:00
|
|
|
|
2013-12-16 10:45:30 +04:00
|
|
|
s = block_job_create(driver, bs, speed, cb, opaque, errp);
|
2012-10-18 18:49:23 +04:00
|
|
|
if (!s) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2012-10-18 18:49:28 +04:00
|
|
|
s->on_source_error = on_source_error;
|
|
|
|
s->on_target_error = on_target_error;
|
2012-10-18 18:49:23 +04:00
|
|
|
s->target = target;
|
2013-12-16 10:45:30 +04:00
|
|
|
s->is_none_mode = is_none_mode;
|
2013-12-16 10:45:29 +04:00
|
|
|
s->base = base;
|
2013-01-21 20:09:46 +04:00
|
|
|
s->granularity = granularity;
|
2013-01-22 12:03:13 +04:00
|
|
|
s->buf_size = MAX(buf_size, granularity);
|
2013-01-21 20:09:43 +04:00
|
|
|
|
2013-11-13 14:29:43 +04:00
|
|
|
s->dirty_bitmap = bdrv_create_dirty_bitmap(bs, granularity);
|
2012-10-18 18:49:23 +04:00
|
|
|
bdrv_set_enable_write_cache(s->target, true);
|
2012-10-18 18:49:28 +04:00
|
|
|
bdrv_set_on_error(s->target, on_target_error, on_target_error);
|
|
|
|
bdrv_iostatus_enable(s->target);
|
2012-10-18 18:49:23 +04:00
|
|
|
s->common.co = qemu_coroutine_create(mirror_run);
|
|
|
|
trace_mirror_start(bs, s, s->common.co, opaque);
|
|
|
|
qemu_coroutine_enter(s->common.co, s);
|
|
|
|
}
|
2013-12-16 10:45:30 +04:00
|
|
|
|
|
|
|
void mirror_start(BlockDriverState *bs, BlockDriverState *target,
|
|
|
|
int64_t speed, int64_t granularity, int64_t buf_size,
|
|
|
|
MirrorSyncMode mode, BlockdevOnError on_source_error,
|
|
|
|
BlockdevOnError on_target_error,
|
|
|
|
BlockDriverCompletionFunc *cb,
|
|
|
|
void *opaque, Error **errp)
|
|
|
|
{
|
|
|
|
bool is_none_mode;
|
|
|
|
BlockDriverState *base;
|
|
|
|
|
|
|
|
is_none_mode = mode == MIRROR_SYNC_MODE_NONE;
|
|
|
|
base = mode == MIRROR_SYNC_MODE_TOP ? bs->backing_hd : NULL;
|
|
|
|
mirror_start_job(bs, target, speed, granularity, buf_size,
|
|
|
|
on_source_error, on_target_error, cb, opaque, errp,
|
|
|
|
&mirror_job_driver, is_none_mode, base);
|
|
|
|
}
|
|
|
|
|
|
|
|
void commit_active_start(BlockDriverState *bs, BlockDriverState *base,
|
|
|
|
int64_t speed,
|
|
|
|
BlockdevOnError on_error,
|
|
|
|
BlockDriverCompletionFunc *cb,
|
|
|
|
void *opaque, Error **errp)
|
|
|
|
{
|
2013-12-16 10:45:31 +04:00
|
|
|
if (bdrv_reopen(base, bs->open_flags, errp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
bdrv_ref(base);
|
2013-12-16 10:45:30 +04:00
|
|
|
mirror_start_job(bs, base, speed, 0, 0,
|
|
|
|
on_error, on_error, cb, opaque, errp,
|
|
|
|
&commit_active_job_driver, false, base);
|
|
|
|
}
|