static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
{
BlockDriverState *source = s->common.bs;
int nb_sectors, sectors_per_chunk, nb_chunks;
int64_t end, sector_num, next_chunk, next_sector, hbitmap_next_sector;
uint64_t delay_ns = 0;
MirrorOp *op;
int pnum;
int64_t ret;
s->sector_num = hbitmap_iter_next(&s->hbi);
if (s->sector_num < 0) {
bdrv_dirty_iter_init(s->dirty_bitmap, &s->hbi);
s->sector_num = hbitmap_iter_next(&s->hbi);
trace_mirror_restart_iter(s, bdrv_get_dirty_count(s->dirty_bitmap));
assert(s->sector_num >= 0);
}
hbitmap_next_sector = s->sector_num;
sector_num = s->sector_num;
sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
end = s->bdev_length / BDRV_SECTOR_SIZE;
/* Extend the QEMUIOVector to include all adjacent blocks that will
* be copied in this operation.
*
* 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.
*/
nb_chunks = 0;
nb_sectors = 0;
next_sector = sector_num;
next_chunk = sector_num / sectors_per_chunk;
/* Wait for I/O to this cluster (from a previous iteration) to be done. */
while (test_bit(next_chunk, s->in_flight_bitmap)) {
trace_mirror_yield_in_flight(s, sector_num, s->in_flight);
s->waiting_for_io = true;
qemu_coroutine_yield();
s->waiting_for_io = false;
}
do {
int added_sectors, added_chunks;
if (!bdrv_get_dirty(source, s->dirty_bitmap, next_sector) ||
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);
s->waiting_for_io = true;
qemu_coroutine_yield();
s->waiting_for_io = false;
}
if (s->buf_free_count < nb_chunks + added_chunks) {
trace_mirror_break_buf_busy(s, nb_chunks, s->in_flight);
break;
}
if (IOV_MAX < nb_chunks + added_chunks) {
trace_mirror_break_iov_max(s, nb_chunks, added_chunks);
break;
}
/* We have enough free space to copy these sectors. */
bitmap_set(s->in_flight_bitmap, next_chunk, added_chunks);
nb_sectors += added_sectors;
nb_chunks += added_chunks;
next_sector += added_sectors;
next_chunk += added_chunks;
if (!s->synced && s->common.speed) {
delay_ns = ratelimit_calculate_delay(&s->limit, added_sectors);
}
} while (delay_ns == 0 && next_sector < end);
/* Allocate a MirrorOp that is used as an AIO callback. */
op = g_new(MirrorOp, 1);
op->s = s;
op->sector_num = sector_num;
op->nb_sectors = nb_sectors;
/* 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);
size_t remaining = (nb_sectors * BDRV_SECTOR_SIZE) - op->qiov.size;
QSIMPLEQ_REMOVE_HEAD(&s->buf_free, next);
s->buf_free_count--;
qemu_iovec_add(&op->qiov, buf, MIN(s->granularity, remaining));
/* Advance the HBitmapIter in parallel, so that we do not examine
* the same sector twice.
*/
if (next_sector > hbitmap_next_sector
&& bdrv_get_dirty(source, s->dirty_bitmap, next_sector)) {
hbitmap_next_sector = hbitmap_iter_next(&s->hbi);
}
next_sector += sectors_per_chunk;
}
bdrv_reset_dirty_bitmap(s->dirty_bitmap, sector_num, nb_sectors);
/* Copy the dirty cluster. */
s->in_flight++;
s->sectors_in_flight += nb_sectors;
trace_mirror_one_iteration(s, sector_num, nb_sectors);
ret = bdrv_get_block_status_above(source, NULL, sector_num,
nb_sectors, &pnum);
if (ret < 0 || pnum < nb_sectors ||
(ret & BDRV_BLOCK_DATA && !(ret & BDRV_BLOCK_ZERO))) {
bdrv_aio_readv(source, sector_num, &op->qiov, nb_sectors,
mirror_read_complete, op);
} else if (ret & BDRV_BLOCK_ZERO) {
bdrv_aio_write_zeroes(s->target, sector_num, op->nb_sectors,
s->unmap ? BDRV_REQ_MAY_UNMAP : 0,
mirror_write_complete, op);
} else {
assert(!(ret & BDRV_BLOCK_DATA));
bdrv_aio_discard(s->target, sector_num, op->nb_sectors,
mirror_write_complete, op);
}
return delay_ns;
}
static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
{
BlockDriverState *source = s->source;
int64_t offset, first_chunk;
uint64_t delay_ns = 0;
/* At least the first dirty chunk is mirrored in one iteration. */
int nb_chunks = 1;
bool write_zeroes_ok = bdrv_can_write_zeroes_with_unmap(blk_bs(s->target));
int max_io_bytes = MAX(s->buf_size / MAX_IN_FLIGHT, MAX_IO_BYTES);
bdrv_dirty_bitmap_lock(s->dirty_bitmap);
offset = bdrv_dirty_iter_next(s->dbi);
if (offset < 0) {
bdrv_set_dirty_iter(s->dbi, 0);
offset = bdrv_dirty_iter_next(s->dbi);
trace_mirror_restart_iter(s, bdrv_get_dirty_count(s->dirty_bitmap));
assert(offset >= 0);
}
bdrv_dirty_bitmap_unlock(s->dirty_bitmap);
first_chunk = offset / s->granularity;
while (test_bit(first_chunk, s->in_flight_bitmap)) {
trace_mirror_yield_in_flight(s, offset, s->in_flight);
mirror_wait_for_io(s);
}
block_job_pause_point(&s->common);
/* Find the number of consective dirty chunks following the first dirty
* one, and wait for in flight requests in them. */
bdrv_dirty_bitmap_lock(s->dirty_bitmap);
while (nb_chunks * s->granularity < s->buf_size) {
int64_t next_dirty;
int64_t next_offset = offset + nb_chunks * s->granularity;
int64_t next_chunk = next_offset / s->granularity;
if (next_offset >= s->bdev_length ||
!bdrv_get_dirty_locked(source, s->dirty_bitmap, next_offset)) {
break;
}
if (test_bit(next_chunk, s->in_flight_bitmap)) {
break;
}
next_dirty = bdrv_dirty_iter_next(s->dbi);
if (next_dirty > next_offset || next_dirty < 0) {
/* The bitmap iterator's cache is stale, refresh it */
bdrv_set_dirty_iter(s->dbi, next_offset);
next_dirty = bdrv_dirty_iter_next(s->dbi);
}
assert(next_dirty == next_offset);
nb_chunks++;
}
/* Clear dirty bits before querying the block status, because
* calling bdrv_block_status_above could yield - if some blocks are
* marked dirty in this window, we need to know.
*/
bdrv_reset_dirty_bitmap_locked(s->dirty_bitmap, offset,
nb_chunks * s->granularity);
bdrv_dirty_bitmap_unlock(s->dirty_bitmap);
bitmap_set(s->in_flight_bitmap, offset / s->granularity, nb_chunks);
while (nb_chunks > 0 && offset < s->bdev_length) {
int ret;
int64_t io_bytes;
int64_t io_bytes_acct;
enum MirrorMethod {
MIRROR_METHOD_COPY,
MIRROR_METHOD_ZERO,
MIRROR_METHOD_DISCARD
} mirror_method = MIRROR_METHOD_COPY;
assert(!(offset % s->granularity));
ret = bdrv_block_status_above(source, NULL, offset,
nb_chunks * s->granularity,
&io_bytes, NULL, NULL);
if (ret < 0) {
io_bytes = MIN(nb_chunks * s->granularity, max_io_bytes);
} else if (ret & BDRV_BLOCK_DATA) {
io_bytes = MIN(io_bytes, max_io_bytes);
}
io_bytes -= io_bytes % s->granularity;
if (io_bytes < s->granularity) {
io_bytes = s->granularity;
} else if (ret >= 0 && !(ret & BDRV_BLOCK_DATA)) {
int64_t target_offset;
int64_t target_bytes;
bdrv_round_to_clusters(blk_bs(s->target), offset, io_bytes,
&target_offset, &target_bytes);
if (target_offset == offset &&
target_bytes == io_bytes) {
mirror_method = ret & BDRV_BLOCK_ZERO ?
MIRROR_METHOD_ZERO :
MIRROR_METHOD_DISCARD;
}
}
while (s->in_flight >= MAX_IN_FLIGHT) {
trace_mirror_yield_in_flight(s, offset, s->in_flight);
mirror_wait_for_io(s);
}
if (s->ret < 0) {
return 0;
}
io_bytes = mirror_clip_bytes(s, offset, io_bytes);
switch (mirror_method) {
case MIRROR_METHOD_COPY:
io_bytes = io_bytes_acct = mirror_do_read(s, offset, io_bytes);
break;
case MIRROR_METHOD_ZERO:
case MIRROR_METHOD_DISCARD:
mirror_do_zero_or_discard(s, offset, io_bytes,
mirror_method == MIRROR_METHOD_DISCARD);
if (write_zeroes_ok) {
io_bytes_acct = 0;
} else {
io_bytes_acct = io_bytes;
}
break;
default:
abort();
}
assert(io_bytes);
offset += io_bytes;
nb_chunks -= DIV_ROUND_UP(io_bytes, s->granularity);
if (s->common.speed) {
delay_ns = ratelimit_calculate_delay(&s->limit, io_bytes_acct);
}
}
return delay_ns;
}
static void coroutine_fn mirror_run(void *opaque)
{
MirrorBlockJob *s = opaque;
BlockDriverState *bs = s->common.bs;
int64_t sector_num, end, sectors_per_chunk, length;
uint64_t last_pause_ns;
BlockDriverInfo bdi;
char backing_filename[1024];
int ret = 0;
int n;
if (block_job_is_cancelled(&s->common)) {
goto immediate_exit;
}
s->common.len = bdrv_getlength(bs);
if (s->common.len <= 0) {
block_job_completed(&s->common, s->common.len);
return;
}
length = (bdrv_getlength(bs) + s->granularity - 1) / s->granularity;
s->in_flight_bitmap = bitmap_new(length);
/* 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);
if (s->granularity < bdi.cluster_size) {
s->buf_size = MAX(s->buf_size, bdi.cluster_size);
s->cow_bitmap = bitmap_new(length);
}
}
end = s->common.len >> BDRV_SECTOR_BITS;
s->buf = qemu_blockalign(bs, s->buf_size);
sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
mirror_free_init(s);
if (s->mode != MIRROR_SYNC_MODE_NONE) {
/* First part, loop on the sectors and initialize the dirty bitmap. */
BlockDriverState *base;
base = s->mode == MIRROR_SYNC_MODE_FULL ? NULL : bs->backing_hd;
for (sector_num = 0; sector_num < end; ) {
int64_t next = (sector_num | (sectors_per_chunk - 1)) + 1;
ret = bdrv_is_allocated_above(bs, base,
sector_num, next - sector_num, &n);
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;
}
}
}
bdrv_dirty_iter_init(bs, &s->hbi);
last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
for (;;) {
uint64_t delay_ns;
int64_t cnt;
bool should_complete;
if (s->ret < 0) {
ret = s->ret;
goto immediate_exit;
}
cnt = bdrv_get_dirty_count(bs);
/* 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.
*/
if (qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - last_pause_ns < SLICE_TIME &&
s->common.iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
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);
qemu_coroutine_yield();
continue;
} else if (cnt != 0) {
mirror_iteration(s);
continue;
}
}
should_complete = false;
if (s->in_flight == 0 && cnt == 0) {
trace_mirror_before_flush(s);
ret = bdrv_flush(s->target);
if (ret < 0) {
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);
cnt = bdrv_get_dirty_count(bs);
}
}
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();
cnt = bdrv_get_dirty_count(bs);
}
ret = 0;
trace_mirror_before_sleep(s, cnt, s->synced);
if (!s->synced) {
/* Publish progress */
s->common.offset = (end - cnt) * BDRV_SECTOR_SIZE;
if (s->common.speed) {
delay_ns = ratelimit_calculate_delay(&s->limit, sectors_per_chunk);
} else {
delay_ns = 0;
}
block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
if (block_job_is_cancelled(&s->common)) {
break;
}
} else if (!should_complete) {
delay_ns = (s->in_flight == 0 && cnt == 0 ? SLICE_TIME : 0);
block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
} 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;
}
last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
}
immediate_exit:
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);
qemu_vfree(s->buf);
g_free(s->cow_bitmap);
g_free(s->in_flight_bitmap);
bdrv_set_dirty_tracking(bs, 0);
bdrv_iostatus_disable(s->target);
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);
}
bdrv_close(s->target);
bdrv_unref(s->target);
block_job_completed(&s->common, ret);
}
static void coroutine_fn commit_run(void *opaque)
{
CommitBlockJob *s = opaque;
BlockDriverState *active = s->active;
BlockDriverState *top = s->top;
BlockDriverState *base = s->base;
BlockDriverState *overlay_bs = NULL;
int64_t sector_num, end;
int ret = 0;
int n = 0;
void *buf;
int bytes_written = 0;
int64_t base_len;
ret = s->common.len = bdrv_getlength(top);
if (s->common.len < 0) {
goto exit_restore_reopen;
}
ret = base_len = bdrv_getlength(base);
if (base_len < 0) {
goto exit_restore_reopen;
}
if (base_len < s->common.len) {
ret = bdrv_truncate(base, s->common.len);
if (ret) {
goto exit_restore_reopen;
}
}
overlay_bs = bdrv_find_overlay(active, top);
end = s->common.len >> BDRV_SECTOR_BITS;
buf = qemu_blockalign(top, COMMIT_BUFFER_SIZE);
for (sector_num = 0; sector_num < end; sector_num += n) {
uint64_t delay_ms = 0;
bool copy;
wait:
/* Note that even when no rate limit is applied we need to yield
* with no pending I/O here so that qemu_aio_flush() returns.
*/
block_job_sleep(&s->common, rt_clock, delay_ms);
if (block_job_is_cancelled(&s->common)) {
break;
}
/* Copy if allocated above the base */
ret = bdrv_co_is_allocated_above(top, base, sector_num,
COMMIT_BUFFER_SIZE / BDRV_SECTOR_SIZE,
&n);
copy = (ret == 1);
trace_commit_one_iteration(s, sector_num, n, ret);
if (copy) {
if (s->common.speed) {
delay_ms = ratelimit_calculate_delay(&s->limit, n);
if (delay_ms > 0) {
goto wait;
}
}
ret = commit_populate(top, base, sector_num, n, buf);
bytes_written += n * BDRV_SECTOR_SIZE;
}
if (ret < 0) {
if (s->on_error == BLOCK_ERR_STOP_ANY ||
s->on_error == BLOCK_ERR_REPORT ||
(s->on_error == BLOCK_ERR_STOP_ENOSPC && ret == -ENOSPC)) {
goto exit_free_buf;
} else {
n = 0;
continue;
}
}
/* Publish progress */
s->common.offset += n * BDRV_SECTOR_SIZE;
}
ret = 0;
if (!block_job_is_cancelled(&s->common) && sector_num == end) {
/* success */
ret = bdrv_drop_intermediate(active, top, base);
}
exit_free_buf:
qemu_vfree(buf);
exit_restore_reopen:
/* restore base open flags here if appropriate (e.g., change the base back
* to r/o). These reopens do not need to be atomic, since we won't abort
* even on failure here */
if (s->base_flags != bdrv_get_flags(base)) {
bdrv_reopen(base, s->base_flags, NULL);
}
if (s->orig_overlay_flags != bdrv_get_flags(overlay_bs)) {
bdrv_reopen(overlay_bs, s->orig_overlay_flags, NULL);
}
block_job_complete(&s->common, ret);
}
