qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Qemu-devel] [PATCH v2 09/11] block/backup: teach TOP to never copy unal


From: John Snow
Subject: [Qemu-devel] [PATCH v2 09/11] block/backup: teach TOP to never copy unallocated regions
Date: Mon, 15 Jul 2019 20:01:15 -0400

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 <address@hidden>
---
 block/backup.c     | 78 ++++++++++++++++++++++++++++++++++++++++------
 block/trace-events |  1 +
 2 files changed, 70 insertions(+), 9 deletions(-)

diff --git a/block/backup.c b/block/backup.c
index b407d57954..e28fd23f6a 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 skip_bytes;
 
     qemu_co_rwlock_rdlock(&job->flush_rwlock);
 
@@ -254,6 +285,15 @@ static int coroutine_fn backup_do_cow(BackupBlockJob *job,
             continue; /* already copied */
         }
 
+        if (job->initializing_bitmap) {
+            ret = backup_bitmap_reset_unallocated(job, start, &skip_bytes);
+            if (ret == 0) {
+                trace_backup_do_cow_skip_range(job, start, skip_bytes);
+                start += skip_bytes;
+                continue;
+            }
+        }
+
         trace_backup_do_cow_process(job, start);
 
         if (job->use_copy_range) {
@@ -436,18 +476,9 @@ 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;
@@ -478,6 +509,13 @@ 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);
     }
 
@@ -499,6 +537,26 @@ 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. */
@@ -507,10 +565,12 @@ static int coroutine_fn backup_run(Job *job, Error **errp)
              * notify callback service CoW requests. */
             job_yield(job);
         }
+        ret = -ECANCELED;
     } else {
         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 d724df0117..04209f058d 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"
-- 
2.21.0




reply via email to

[Prev in Thread] Current Thread [Next in Thread]