QEMU-Devel Archive on lore.kernel.org
 help / color / Atom feed
From: John Snow <jsnow@redhat.com>
To: qemu-devel@nongnu.org
Cc: peter.maydell@linaro.org, jsnow@redhat.com,
	qemu-stable@nongnu.org, qemu-block@nongnu.org,
	Max Reitz <mreitz@redhat.com>
Subject: [Qemu-devel] [PULL 30/36] block/backup: teach TOP to never copy unallocated regions
Date: Fri, 16 Aug 2019 19:13:12 -0400
Message-ID: <20190816231318.8650-31-jsnow@redhat.com> (raw)
In-Reply-To: <20190816231318.8650-1-jsnow@redhat.com>

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>
---
 block/backup.c     | 79 ++++++++++++++++++++++++++++++++++++++++------
 block/trace-events |  1 +
 2 files changed, 71 insertions(+), 9 deletions(-)

diff --git a/block/backup.c b/block/backup.c
index f6bf32c9438..9e1382ec5c6 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 d724df0117d..04209f058d4 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



  parent reply index

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-16 23:12 [Qemu-devel] [PULL 00/36] Bitmaps patches John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 01/36] qapi/block-core: Introduce BackupCommon John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 02/36] drive-backup: create do_backup_common John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 03/36] blockdev-backup: utilize do_backup_common John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 04/36] qapi: add BitmapSyncMode enum John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 05/36] block/backup: Add mirror sync mode 'bitmap' John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 06/36] block/backup: add 'never' policy to bitmap sync mode John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 07/36] hbitmap: Fix merge when b is empty, and result is not an alias of a John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 08/36] hbitmap: enable merging across granularities John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 09/36] block/dirty-bitmap: add bdrv_dirty_bitmap_merge_internal John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 10/36] block/dirty-bitmap: add bdrv_dirty_bitmap_get John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 11/36] block/backup: upgrade copy_bitmap to BdrvDirtyBitmap John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 12/36] block/backup: add 'always' bitmap sync policy John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 13/36] iotests: add testing shim for script-style python tests John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 14/36] iotests: teach run_job to cancel pending jobs John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 15/36] iotests: teach FilePath to produce multiple paths John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 16/36] iotests: Add virtio-scsi device helper John Snow
2019-08-16 23:12 ` [Qemu-devel] [PULL 17/36] iotests: add test 257 for bitmap-mode backups John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 18/36] block/backup: loosen restriction on readonly bitmaps John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 19/36] blockdev: reduce aio_context locked sections in bitmap add/remove John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 20/36] qapi: implement block-dirty-bitmap-remove transaction action John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 21/36] iotests: test bitmap moving inside 254 John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 22/36] iotests/257: add Pattern class John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 23/36] iotests/257: add EmulatedBitmap class John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 24/36] iotests/257: Refactor backup helpers John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 25/36] block/backup: hoist bitmap check into QMP interface John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 26/36] iotests/257: test API failures John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 27/36] block/backup: improve sync=bitmap work estimates John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 28/36] block/backup: centralize copy_bitmap initialization John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 29/36] block/backup: add backup_is_cluster_allocated John Snow
2019-08-16 23:13 ` John Snow [this message]
2019-08-16 23:13 ` [Qemu-devel] [PULL 31/36] block/backup: support bitmap sync modes for non-bitmap backups John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 32/36] iotests/257: test traditional sync modes John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 33/36] qapi: add dirty-bitmaps to query-named-block-nodes result John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 34/36] block/backup: deal with zero detection John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 35/36] block/backup: refactor write_flags John Snow
2019-08-16 23:13 ` [Qemu-devel] [PULL 36/36] tests/test-hbitmap: test next_zero and _next_dirty_area after truncate John Snow
2019-08-19 11:32 ` [Qemu-devel] [PULL 00/36] Bitmaps patches Peter Maydell

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190816231318.8650-31-jsnow@redhat.com \
    --to=jsnow@redhat.com \
    --cc=mreitz@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-stable@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

QEMU-Devel Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/qemu-devel/0 qemu-devel/git/0.git
	git clone --mirror https://lore.kernel.org/qemu-devel/1 qemu-devel/git/1.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 qemu-devel qemu-devel/ https://lore.kernel.org/qemu-devel \
		qemu-devel@nongnu.org
	public-inbox-index qemu-devel

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.nongnu.qemu-devel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git