qemu/block/backup.c
Fam Zheng 61408b250e block: Don't wait serialising for non-COR read requests
The assertion problem was noticed in 06c3916b35, but it wasn't
completely fixed, because even though the req is not marked as
serialising, it still gets serialised by wait_serialising_requests
against other serialising requests, which could lead to the same
assertion failure.

Fix it by even more explicitly skipping the serialising for this
specific case.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-id: 1448962590-2842-2-git-send-email-famz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2015-12-03 11:08:07 +08:00

587 lines
18 KiB
C

/*
* QEMU backup
*
* Copyright (C) 2013 Proxmox Server Solutions
*
* Authors:
* Dietmar Maurer (dietmar@proxmox.com)
*
* This work is licensed under the terms of the GNU GPL, version 2 or later.
* See the COPYING file in the top-level directory.
*
*/
#include <stdio.h>
#include <errno.h>
#include <unistd.h>
#include "trace.h"
#include "block/block.h"
#include "block/block_int.h"
#include "block/blockjob.h"
#include "qapi/qmp/qerror.h"
#include "qemu/ratelimit.h"
#include "sysemu/block-backend.h"
#define BACKUP_CLUSTER_BITS 16
#define BACKUP_CLUSTER_SIZE (1 << BACKUP_CLUSTER_BITS)
#define BACKUP_SECTORS_PER_CLUSTER (BACKUP_CLUSTER_SIZE / BDRV_SECTOR_SIZE)
#define SLICE_TIME 100000000ULL /* ns */
typedef struct CowRequest {
int64_t start;
int64_t end;
QLIST_ENTRY(CowRequest) list;
CoQueue wait_queue; /* coroutines blocked on this request */
} CowRequest;
typedef struct BackupBlockJob {
BlockJob common;
BlockDriverState *target;
/* bitmap for sync=incremental */
BdrvDirtyBitmap *sync_bitmap;
MirrorSyncMode sync_mode;
RateLimit limit;
BlockdevOnError on_source_error;
BlockdevOnError on_target_error;
CoRwlock flush_rwlock;
uint64_t sectors_read;
HBitmap *bitmap;
QLIST_HEAD(, CowRequest) inflight_reqs;
} BackupBlockJob;
/* See if in-flight requests overlap and wait for them to complete */
static void coroutine_fn wait_for_overlapping_requests(BackupBlockJob *job,
int64_t start,
int64_t end)
{
CowRequest *req;
bool retry;
do {
retry = false;
QLIST_FOREACH(req, &job->inflight_reqs, list) {
if (end > req->start && start < req->end) {
qemu_co_queue_wait(&req->wait_queue);
retry = true;
break;
}
}
} while (retry);
}
/* Keep track of an in-flight request */
static void cow_request_begin(CowRequest *req, BackupBlockJob *job,
int64_t start, int64_t end)
{
req->start = start;
req->end = end;
qemu_co_queue_init(&req->wait_queue);
QLIST_INSERT_HEAD(&job->inflight_reqs, req, list);
}
/* Forget about a completed request */
static void cow_request_end(CowRequest *req)
{
QLIST_REMOVE(req, list);
qemu_co_queue_restart_all(&req->wait_queue);
}
static int coroutine_fn backup_do_cow(BlockDriverState *bs,
int64_t sector_num, int nb_sectors,
bool *error_is_read,
bool is_write_notifier)
{
BackupBlockJob *job = (BackupBlockJob *)bs->job;
CowRequest cow_request;
struct iovec iov;
QEMUIOVector bounce_qiov;
void *bounce_buffer = NULL;
int ret = 0;
int64_t start, end;
int n;
qemu_co_rwlock_rdlock(&job->flush_rwlock);
start = sector_num / BACKUP_SECTORS_PER_CLUSTER;
end = DIV_ROUND_UP(sector_num + nb_sectors, BACKUP_SECTORS_PER_CLUSTER);
trace_backup_do_cow_enter(job, start, sector_num, nb_sectors);
wait_for_overlapping_requests(job, start, end);
cow_request_begin(&cow_request, job, start, end);
for (; start < end; start++) {
if (hbitmap_get(job->bitmap, start)) {
trace_backup_do_cow_skip(job, start);
continue; /* already copied */
}
trace_backup_do_cow_process(job, start);
n = MIN(BACKUP_SECTORS_PER_CLUSTER,
job->common.len / BDRV_SECTOR_SIZE -
start * BACKUP_SECTORS_PER_CLUSTER);
if (!bounce_buffer) {
bounce_buffer = qemu_blockalign(bs, BACKUP_CLUSTER_SIZE);
}
iov.iov_base = bounce_buffer;
iov.iov_len = n * BDRV_SECTOR_SIZE;
qemu_iovec_init_external(&bounce_qiov, &iov, 1);
if (is_write_notifier) {
ret = bdrv_co_readv_no_serialising(bs,
start * BACKUP_SECTORS_PER_CLUSTER,
n, &bounce_qiov);
} else {
ret = bdrv_co_readv(bs, start * BACKUP_SECTORS_PER_CLUSTER, n,
&bounce_qiov);
}
if (ret < 0) {
trace_backup_do_cow_read_fail(job, start, ret);
if (error_is_read) {
*error_is_read = true;
}
goto out;
}
if (buffer_is_zero(iov.iov_base, iov.iov_len)) {
ret = bdrv_co_write_zeroes(job->target,
start * BACKUP_SECTORS_PER_CLUSTER,
n, BDRV_REQ_MAY_UNMAP);
} else {
ret = bdrv_co_writev(job->target,
start * BACKUP_SECTORS_PER_CLUSTER, n,
&bounce_qiov);
}
if (ret < 0) {
trace_backup_do_cow_write_fail(job, start, ret);
if (error_is_read) {
*error_is_read = false;
}
goto out;
}
hbitmap_set(job->bitmap, start, 1);
/* Publish progress, guest I/O counts as progress too. Note that the
* offset field is an opaque progress value, it is not a disk offset.
*/
job->sectors_read += n;
job->common.offset += n * BDRV_SECTOR_SIZE;
}
out:
if (bounce_buffer) {
qemu_vfree(bounce_buffer);
}
cow_request_end(&cow_request);
trace_backup_do_cow_return(job, sector_num, nb_sectors, ret);
qemu_co_rwlock_unlock(&job->flush_rwlock);
return ret;
}
static int coroutine_fn backup_before_write_notify(
NotifierWithReturn *notifier,
void *opaque)
{
BdrvTrackedRequest *req = opaque;
int64_t sector_num = req->offset >> BDRV_SECTOR_BITS;
int nb_sectors = req->bytes >> BDRV_SECTOR_BITS;
assert((req->offset & (BDRV_SECTOR_SIZE - 1)) == 0);
assert((req->bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
return backup_do_cow(req->bs, sector_num, nb_sectors, NULL, true);
}
static void backup_set_speed(BlockJob *job, int64_t speed, Error **errp)
{
BackupBlockJob *s = container_of(job, BackupBlockJob, common);
if (speed < 0) {
error_setg(errp, QERR_INVALID_PARAMETER, "speed");
return;
}
ratelimit_set_speed(&s->limit, speed / BDRV_SECTOR_SIZE, SLICE_TIME);
}
static void backup_iostatus_reset(BlockJob *job)
{
BackupBlockJob *s = container_of(job, BackupBlockJob, common);
if (s->target->blk) {
blk_iostatus_reset(s->target->blk);
}
}
static void backup_cleanup_sync_bitmap(BackupBlockJob *job, int ret)
{
BdrvDirtyBitmap *bm;
BlockDriverState *bs = job->common.bs;
if (ret < 0 || block_job_is_cancelled(&job->common)) {
/* Merge the successor back into the parent, delete nothing. */
bm = bdrv_reclaim_dirty_bitmap(bs, job->sync_bitmap, NULL);
assert(bm);
} else {
/* Everything is fine, delete this bitmap and install the backup. */
bm = bdrv_dirty_bitmap_abdicate(bs, job->sync_bitmap, NULL);
assert(bm);
}
}
static void backup_commit(BlockJob *job)
{
BackupBlockJob *s = container_of(job, BackupBlockJob, common);
if (s->sync_bitmap) {
backup_cleanup_sync_bitmap(s, 0);
}
}
static void backup_abort(BlockJob *job)
{
BackupBlockJob *s = container_of(job, BackupBlockJob, common);
if (s->sync_bitmap) {
backup_cleanup_sync_bitmap(s, -1);
}
}
static const BlockJobDriver backup_job_driver = {
.instance_size = sizeof(BackupBlockJob),
.job_type = BLOCK_JOB_TYPE_BACKUP,
.set_speed = backup_set_speed,
.iostatus_reset = backup_iostatus_reset,
.commit = backup_commit,
.abort = backup_abort,
};
static BlockErrorAction backup_error_action(BackupBlockJob *job,
bool read, int error)
{
if (read) {
return block_job_error_action(&job->common, job->common.bs,
job->on_source_error, true, error);
} else {
return block_job_error_action(&job->common, job->target,
job->on_target_error, false, error);
}
}
typedef struct {
int ret;
} BackupCompleteData;
static void backup_complete(BlockJob *job, void *opaque)
{
BackupBlockJob *s = container_of(job, BackupBlockJob, common);
BackupCompleteData *data = opaque;
bdrv_unref(s->target);
block_job_completed(job, data->ret);
g_free(data);
}
static bool coroutine_fn yield_and_check(BackupBlockJob *job)
{
if (block_job_is_cancelled(&job->common)) {
return true;
}
/* we need to yield so that bdrv_drain_all() returns.
* (without, VM does not reboot)
*/
if (job->common.speed) {
uint64_t delay_ns = ratelimit_calculate_delay(&job->limit,
job->sectors_read);
job->sectors_read = 0;
block_job_sleep_ns(&job->common, QEMU_CLOCK_REALTIME, delay_ns);
} else {
block_job_sleep_ns(&job->common, QEMU_CLOCK_REALTIME, 0);
}
if (block_job_is_cancelled(&job->common)) {
return true;
}
return false;
}
static int coroutine_fn backup_run_incremental(BackupBlockJob *job)
{
bool error_is_read;
int ret = 0;
int clusters_per_iter;
uint32_t granularity;
int64_t sector;
int64_t cluster;
int64_t end;
int64_t last_cluster = -1;
BlockDriverState *bs = job->common.bs;
HBitmapIter hbi;
granularity = bdrv_dirty_bitmap_granularity(job->sync_bitmap);
clusters_per_iter = MAX((granularity / BACKUP_CLUSTER_SIZE), 1);
bdrv_dirty_iter_init(job->sync_bitmap, &hbi);
/* Find the next dirty sector(s) */
while ((sector = hbitmap_iter_next(&hbi)) != -1) {
cluster = sector / BACKUP_SECTORS_PER_CLUSTER;
/* Fake progress updates for any clusters we skipped */
if (cluster != last_cluster + 1) {
job->common.offset += ((cluster - last_cluster - 1) *
BACKUP_CLUSTER_SIZE);
}
for (end = cluster + clusters_per_iter; cluster < end; cluster++) {
do {
if (yield_and_check(job)) {
return ret;
}
ret = backup_do_cow(bs, cluster * BACKUP_SECTORS_PER_CLUSTER,
BACKUP_SECTORS_PER_CLUSTER, &error_is_read,
false);
if ((ret < 0) &&
backup_error_action(job, error_is_read, -ret) ==
BLOCK_ERROR_ACTION_REPORT) {
return ret;
}
} while (ret < 0);
}
/* If the bitmap granularity is smaller than the backup granularity,
* we need to advance the iterator pointer to the next cluster. */
if (granularity < BACKUP_CLUSTER_SIZE) {
bdrv_set_dirty_iter(&hbi, cluster * BACKUP_SECTORS_PER_CLUSTER);
}
last_cluster = cluster - 1;
}
/* Play some final catchup with the progress meter */
end = DIV_ROUND_UP(job->common.len, BACKUP_CLUSTER_SIZE);
if (last_cluster + 1 < end) {
job->common.offset += ((end - last_cluster - 1) * BACKUP_CLUSTER_SIZE);
}
return ret;
}
static void coroutine_fn backup_run(void *opaque)
{
BackupBlockJob *job = opaque;
BackupCompleteData *data;
BlockDriverState *bs = job->common.bs;
BlockDriverState *target = job->target;
BlockdevOnError on_target_error = job->on_target_error;
NotifierWithReturn before_write = {
.notify = backup_before_write_notify,
};
int64_t start, end;
int ret = 0;
QLIST_INIT(&job->inflight_reqs);
qemu_co_rwlock_init(&job->flush_rwlock);
start = 0;
end = DIV_ROUND_UP(job->common.len, BACKUP_CLUSTER_SIZE);
job->bitmap = hbitmap_alloc(end, 0);
bdrv_set_enable_write_cache(target, true);
if (target->blk) {
blk_set_on_error(target->blk, on_target_error, on_target_error);
blk_iostatus_enable(target->blk);
}
bdrv_add_before_write_notifier(bs, &before_write);
if (job->sync_mode == MIRROR_SYNC_MODE_NONE) {
while (!block_job_is_cancelled(&job->common)) {
/* Yield until the job is cancelled. We just let our before_write
* notify callback service CoW requests. */
job->common.busy = false;
qemu_coroutine_yield();
job->common.busy = true;
}
} else if (job->sync_mode == MIRROR_SYNC_MODE_INCREMENTAL) {
ret = backup_run_incremental(job);
} else {
/* Both FULL and TOP SYNC_MODE's require copying.. */
for (; start < end; start++) {
bool error_is_read;
if (yield_and_check(job)) {
break;
}
if (job->sync_mode == MIRROR_SYNC_MODE_TOP) {
int i, n;
int alloced = 0;
/* Check to see if these blocks are already in the
* backing file. */
for (i = 0; i < BACKUP_SECTORS_PER_CLUSTER;) {
/* bdrv_is_allocated() only returns true/false based
* on the first set of sectors it comes across that
* are are all in the same state.
* For that reason we must verify each sector in the
* backup cluster length. We end up copying more than
* needed but at some point that is always the case. */
alloced =
bdrv_is_allocated(bs,
start * BACKUP_SECTORS_PER_CLUSTER + i,
BACKUP_SECTORS_PER_CLUSTER - i, &n);
i += n;
if (alloced == 1 || n == 0) {
break;
}
}
/* If the above loop never found any sectors that are in
* the topmost image, skip this backup. */
if (alloced == 0) {
continue;
}
}
/* FULL sync mode we copy the whole drive. */
ret = backup_do_cow(bs, start * BACKUP_SECTORS_PER_CLUSTER,
BACKUP_SECTORS_PER_CLUSTER, &error_is_read, false);
if (ret < 0) {
/* Depending on error action, fail now or retry cluster */
BlockErrorAction action =
backup_error_action(job, error_is_read, -ret);
if (action == BLOCK_ERROR_ACTION_REPORT) {
break;
} else {
start--;
continue;
}
}
}
}
notifier_with_return_remove(&before_write);
/* wait until pending backup_do_cow() calls have completed */
qemu_co_rwlock_wrlock(&job->flush_rwlock);
qemu_co_rwlock_unlock(&job->flush_rwlock);
hbitmap_free(job->bitmap);
if (target->blk) {
blk_iostatus_disable(target->blk);
}
bdrv_op_unblock_all(target, job->common.blocker);
data = g_malloc(sizeof(*data));
data->ret = ret;
block_job_defer_to_main_loop(&job->common, backup_complete, data);
}
void backup_start(BlockDriverState *bs, BlockDriverState *target,
int64_t speed, MirrorSyncMode sync_mode,
BdrvDirtyBitmap *sync_bitmap,
BlockdevOnError on_source_error,
BlockdevOnError on_target_error,
BlockCompletionFunc *cb, void *opaque,
BlockJobTxn *txn, Error **errp)
{
int64_t len;
assert(bs);
assert(target);
assert(cb);
if (bs == target) {
error_setg(errp, "Source and target cannot be the same");
return;
}
if ((on_source_error == BLOCKDEV_ON_ERROR_STOP ||
on_source_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
(!bs->blk || !blk_iostatus_is_enabled(bs->blk))) {
error_setg(errp, QERR_INVALID_PARAMETER, "on-source-error");
return;
}
if (!bdrv_is_inserted(bs)) {
error_setg(errp, "Device is not inserted: %s",
bdrv_get_device_name(bs));
return;
}
if (!bdrv_is_inserted(target)) {
error_setg(errp, "Device is not inserted: %s",
bdrv_get_device_name(target));
return;
}
if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_BACKUP_SOURCE, errp)) {
return;
}
if (bdrv_op_is_blocked(target, BLOCK_OP_TYPE_BACKUP_TARGET, errp)) {
return;
}
if (sync_mode == MIRROR_SYNC_MODE_INCREMENTAL) {
if (!sync_bitmap) {
error_setg(errp, "must provide a valid bitmap name for "
"\"incremental\" sync mode");
return;
}
/* Create a new bitmap, and freeze/disable this one. */
if (bdrv_dirty_bitmap_create_successor(bs, sync_bitmap, errp) < 0) {
return;
}
} else if (sync_bitmap) {
error_setg(errp,
"a sync_bitmap was provided to backup_run, "
"but received an incompatible sync_mode (%s)",
MirrorSyncMode_lookup[sync_mode]);
return;
}
len = bdrv_getlength(bs);
if (len < 0) {
error_setg_errno(errp, -len, "unable to get length for '%s'",
bdrv_get_device_name(bs));
goto error;
}
BackupBlockJob *job = block_job_create(&backup_job_driver, bs, speed,
cb, opaque, errp);
if (!job) {
goto error;
}
bdrv_op_block_all(target, job->common.blocker);
job->on_source_error = on_source_error;
job->on_target_error = on_target_error;
job->target = target;
job->sync_mode = sync_mode;
job->sync_bitmap = sync_mode == MIRROR_SYNC_MODE_INCREMENTAL ?
sync_bitmap : NULL;
job->common.len = len;
job->common.co = qemu_coroutine_create(backup_run);
block_job_txn_add_job(txn, &job->common);
qemu_coroutine_enter(job->common.co, job);
return;
error:
if (sync_bitmap) {
bdrv_reclaim_dirty_bitmap(bs, sync_bitmap, NULL);
}
}