aboutsummaryrefslogtreecommitdiff
path: root/block
diff options
context:
space:
mode:
authorJohn Snow <jsnow@redhat.com>2019-07-29 16:35:55 -0400
committerJohn Snow <jsnow@redhat.com>2019-08-16 16:28:03 -0400
commit7e30dd618ebfe3ab1ed54f2e98ba75d799c0be20 (patch)
tree8f6c4d6398603583e13417c75229c76758ee13d6 /block
parentdba8700f16ebda0632977c303f66021407971081 (diff)
downloadqemu-7e30dd618ebfe3ab1ed54f2e98ba75d799c0be20.zip
qemu-7e30dd618ebfe3ab1ed54f2e98ba75d799c0be20.tar.gz
qemu-7e30dd618ebfe3ab1ed54f2e98ba75d799c0be20.tar.bz2
block/backup: teach TOP to never copy unallocated regions
Presently, If sync=TOP is selected, we mark the entire bitmap as dirty. In the write notifier handler, we dutifully copy out such regions. Fix this in three parts: 1. Mark the bitmap as being initialized before the first yield. 2. After the first yield but before the backup loop, interrogate the allocation status asynchronously and initialize the bitmap. 3. Teach the write notifier to interrogate allocation status if it is invoked during bitmap initialization. As an effect of this patch, the job progress for TOP backups now behaves like this: - total progress starts at bdrv_length. - As allocation status is interrogated, total progress decreases. - As blocks are copied, current progress increases. Taken together, the floor and ceiling move to meet each other. Signed-off-by: John Snow <jsnow@redhat.com> Message-id: 20190716000117.25219-10-jsnow@redhat.com [Remove ret = -ECANCELED change. --js] [Squash in conflict resolution based on Max's patch --js] Message-id: c8b0ab36-79c8-0b4b-3193-4e12ed8c848b@redhat.com Reviewed-by: Max Reitz <mreitz@redhat.com> Signed-off-by: John Snow <jsnow@redhat.com>
Diffstat (limited to 'block')
-rw-r--r--block/backup.c79
-rw-r--r--block/trace-events1
2 files changed, 71 insertions, 9 deletions
diff --git a/block/backup.c b/block/backup.c
index f6bf32c..9e1382e 100644
--- a/block/backup.c
+++ b/block/backup.c
@@ -58,6 +58,7 @@ typedef struct BackupBlockJob {
int64_t copy_range_size;
bool serialize_target_writes;
+ bool initializing_bitmap;
} BackupBlockJob;
static const BlockJobDriver backup_job_driver;
@@ -227,6 +228,35 @@ static int backup_is_cluster_allocated(BackupBlockJob *s, int64_t offset,
}
}
+/**
+ * Reset bits in copy_bitmap starting at offset if they represent unallocated
+ * data in the image. May reset subsequent contiguous bits.
+ * @return 0 when the cluster at @offset was unallocated,
+ * 1 otherwise, and -ret on error.
+ */
+static int64_t backup_bitmap_reset_unallocated(BackupBlockJob *s,
+ int64_t offset, int64_t *count)
+{
+ int ret;
+ int64_t clusters, bytes, estimate;
+
+ ret = backup_is_cluster_allocated(s, offset, &clusters);
+ if (ret < 0) {
+ return ret;
+ }
+
+ bytes = clusters * s->cluster_size;
+
+ if (!ret) {
+ bdrv_reset_dirty_bitmap(s->copy_bitmap, offset, bytes);
+ estimate = bdrv_get_dirty_count(s->copy_bitmap);
+ job_progress_set_remaining(&s->common.job, estimate);
+ }
+
+ *count = bytes;
+ return ret;
+}
+
static int coroutine_fn backup_do_cow(BackupBlockJob *job,
int64_t offset, uint64_t bytes,
bool *error_is_read,
@@ -236,6 +266,7 @@ static int coroutine_fn backup_do_cow(BackupBlockJob *job,
int ret = 0;
int64_t start, end; /* bytes */
void *bounce_buffer = NULL;
+ int64_t status_bytes;
qemu_co_rwlock_rdlock(&job->flush_rwlock);
@@ -262,6 +293,17 @@ static int coroutine_fn backup_do_cow(BackupBlockJob *job,
dirty_end = end;
}
+ if (job->initializing_bitmap) {
+ ret = backup_bitmap_reset_unallocated(job, start, &status_bytes);
+ if (ret == 0) {
+ trace_backup_do_cow_skip_range(job, start, status_bytes);
+ start += status_bytes;
+ continue;
+ }
+ /* Clamp to known allocated region */
+ dirty_end = MIN(dirty_end, start + status_bytes);
+ }
+
trace_backup_do_cow_process(job, start);
if (job->use_copy_range) {
@@ -446,18 +488,9 @@ static int coroutine_fn backup_loop(BackupBlockJob *job)
int64_t offset;
BdrvDirtyBitmapIter *bdbi;
int ret = 0;
- int64_t dummy;
bdbi = bdrv_dirty_iter_new(job->copy_bitmap);
while ((offset = bdrv_dirty_iter_next(bdbi)) != -1) {
- if (job->sync_mode == MIRROR_SYNC_MODE_TOP &&
- !backup_is_cluster_allocated(job, offset, &dummy))
- {
- bdrv_reset_dirty_bitmap(job->copy_bitmap, offset,
- job->cluster_size);
- continue;
- }
-
do {
if (yield_and_check(job)) {
goto out;
@@ -488,6 +521,13 @@ static void backup_init_copy_bitmap(BackupBlockJob *job)
NULL, true);
assert(ret);
} else {
+ if (job->sync_mode == MIRROR_SYNC_MODE_TOP) {
+ /*
+ * We can't hog the coroutine to initialize this thoroughly.
+ * Set a flag and resume work when we are able to yield safely.
+ */
+ job->initializing_bitmap = true;
+ }
bdrv_set_dirty_bitmap(job->copy_bitmap, 0, job->len);
}
@@ -509,6 +549,26 @@ static int coroutine_fn backup_run(Job *job, Error **errp)
s->before_write.notify = backup_before_write_notify;
bdrv_add_before_write_notifier(bs, &s->before_write);
+ if (s->sync_mode == MIRROR_SYNC_MODE_TOP) {
+ int64_t offset = 0;
+ int64_t count;
+
+ for (offset = 0; offset < s->len; ) {
+ if (yield_and_check(s)) {
+ ret = -ECANCELED;
+ goto out;
+ }
+
+ ret = backup_bitmap_reset_unallocated(s, offset, &count);
+ if (ret < 0) {
+ goto out;
+ }
+
+ offset += count;
+ }
+ s->initializing_bitmap = false;
+ }
+
if (s->sync_mode == MIRROR_SYNC_MODE_NONE) {
/* All bits are set in copy_bitmap to allow any cluster to be copied.
* This does not actually require them to be copied. */
@@ -521,6 +581,7 @@ static int coroutine_fn backup_run(Job *job, Error **errp)
ret = backup_loop(s);
}
+ out:
notifier_with_return_remove(&s->before_write);
/* wait until pending backup_do_cow() calls have completed */
diff --git a/block/trace-events b/block/trace-events
index d724df0..04209f0 100644
--- a/block/trace-events
+++ b/block/trace-events
@@ -41,6 +41,7 @@ mirror_yield_in_flight(void *s, int64_t offset, int in_flight) "s %p offset %" P
backup_do_cow_enter(void *job, int64_t start, int64_t offset, uint64_t bytes) "job %p start %" PRId64 " offset %" PRId64 " bytes %" PRIu64
backup_do_cow_return(void *job, int64_t offset, uint64_t bytes, int ret) "job %p offset %" PRId64 " bytes %" PRIu64 " ret %d"
backup_do_cow_skip(void *job, int64_t start) "job %p start %"PRId64
+backup_do_cow_skip_range(void *job, int64_t start, uint64_t bytes) "job %p start %"PRId64" bytes %"PRId64
backup_do_cow_process(void *job, int64_t start) "job %p start %"PRId64
backup_do_cow_read_fail(void *job, int64_t start, int ret) "job %p start %"PRId64" ret %d"
backup_do_cow_write_fail(void *job, int64_t start, int ret) "job %p start %"PRId64" ret %d"